159538M PDF

UNESCO Publication as part of “Ecological Research for
Sustaining the Environment in China (ERSEC)” under funding

from the German Federal Ministry of Education and Research
(BMBF). Th9is conference was organized to communicate on a
multilateral platform the results of the Sino-German Coal Fire
Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China” under
ERSEC umbrella.
Disclaimer ERSEC Project Secretariat

The designations employed and the presentation of material throughout this Waijiaogongyu 5-15-3
publication do not imply the expression of any opinion whatsoever on the Jianguomenwai Compound
part of UNESCO concerning the legal status of any country, territory, city 100600 Beijing, P. R. China
or area of its authorities, or concerning the delimitation of its frontiers or Phone: +86-10-6532 7683
boundaries. Fax: +86-10-6532 4854
E-mail: beijing.sc@unesco.org
© UNESCO 2008 URL: http://www.unescobeijing.org
Foreword
Spontaneous combustion has resulted in coal fires since prehistoric times. Starting
with the industrial revolution, however, human activities have expanded the scale
and severity of this phenomenon. The effects have been devastating. Coal fires not
only destroy valuable coal resources that could be easily exploited – shallow-depth
coal seams for example; they also lead to blockage and devaluation of neighboring
deposits. While this in itself represents a serious economic setback, coal fires,
especially when occurring in underground mine systems, also impair worker safety
and often send mining costs spiraling. Cave-ins, roof collapses, and gas explosions
in affected workings figure behind the tragic casualties that have made coal
mining one of the most dangerous and potentially lethal occupations worldwide –
particularly so in China.
China’s economy is presently growing at a staggering pace. The country’s real gross
domestic product is estimated to have risen by 10.9 % in 2007. Inflows of foreign
direct investment totaled USD 86.1 billion in 2005 – a new record and roughly
twice the amount of 2001. Massive economic expansion, however, also makes
for increasing energy consumption. Currently, 69 % of China’s total primary
energy supply is met by coal. China is the world’s largest consumer of coal, and
also boasts the world’s largest production rates. Needless to say, it is the country
most afflicted by coal fires. According to reports of the Beijing Remote Sensing
Corporation (BRSC) and the Aerophotogrammetry and Remote Sensing Bureau of
China Coal (ARSC), 10 to 20 million t of Chinese coal are burned away annually
in uncontrolled coal fires; this equals a direct economic loss of USD 125 to 250
million.

Beyond their dire implications for mine output and safety, coal fires are of course
a grave environmental concern. Mining companies may deplore cuts in profits;
the people working and living next to coal fires, however, bear the brunt of heavy
pollution and chronic ailments. Incomplete combustion of coal – a result of
oxygen deficiency underground – releases an acrid mix of aerosols and toxic gases.
Streaming upward in convective flow through cracks and fissures, these gases,
together with particulate matter, infest the air over coal fires zones; they are often
carried away with the wind to locations far from the original fire site. Condensates
of combustion products contaminate soil and groundwater. Coal fire environment,
in short, is inhospitable to human, animal, and plant life.
In the ongoing debate on climate change, coal fires have attracted increasing
attention as a source of greenhouse gases – carbon dioxide and methane in
particular. Despite their very palpable local effects, it is these emissions that link coal
fires to the global arena. Coal fire mitigation and prevention should indeed be the
subject of international scientific and political cooperation. With this conviction,
the German Federal Ministry of Education and Research (BMBF) and the Ministry
of Science and Technology of the People’s Republic of China (MoST) launched
in 2003 the Sino-German Coal Fire Research Initiative “Innovative Technologies
for Exploration, Extinction, and Monitoring of Coal Fires in North China” – the
project behind the conference that generated this book.
The “International Conference on Coal Fire Research – ERSEC Conference on

Understanding, Extinction, and Prevention of Spontaneous Coal Seam Fires” was

held in Beijing, P.R. China from 29 November to 1 December 2005, marking the
end of the first three-year phase of the initiative (Phase A). It was staged within
the framework of “Ecological Research for Sustaining the Environment in China
(ERSEC),” an umbrella project for bilateral environmental research projects, also
funded by BMBF. Along with renowned Chinese and German experts on the
subject, more than 50 international scientists participated, from France, India,
Kazakhstan, Mongolia, New Zealand, South Africa, and the U.S.A. – virtually a
global cross-section of countries affected by spontaneous coal seam fires.
It is part of ERSEC’s agenda to publish the conference recommendations in a book

series intended for policy-makers, academia, and the general public. The book
in your hands is the forth volume of this endeavor published by UNESCO, the
ERSEC Ecological Book Series – 4. I am positive that it gives a concise overview
of state-of-the-art coal fire research, and aptly describes all the different methods
used in exploration, extinction, and monitoring of this natural – and yet man-
made – disaster. With its emphasis on the regional and global consequences of
coal fires, moreover, this book is sure to encourage further research into sustainable
approaches to coal fire mitigation – and thus to bluer skies in China and worldwide.
Prof. Dr. Walter R. Erdelen

Assistant Director-General for Natural Sciences
UNESCO

Preface
China is by far the biggest producer of coal; at the same time, it is the stage for
one of the world’s most ravenous coal-consuming phenomena – 10 to 20 million t
of coal are burned away in uncontrolled coal seam fires in China annually. What
is more, approximately ten times this amount, about 100 to 200 million t, is lost
for mining as access for vehicles and machines is barred through the fires, or the
rock-mechanical changes they entail. Toxic and greenhouse gas emissions – carbon
dioxide and monoxide, as well as methane to name only the most prominent –
pollute the atmosphere and threaten flora and fauna in the fire areas. Extinguishing
uncontrolled coal seam fires is an extremely difficult, time-consuming, and costly
enterprise. Even large-scale efforts often fail since they lack a thorough scientific
understanding of the processes leading up to and controlling coal fire development.
The Sino-German Coal Fire Research Initiative “Innovative Technologies for
Exploration, Extinction, and Monitoring of Coal Fires in North China” was
launched to deepen the knowledge and understanding of these processes. Under its
umbrella, experts from different technological and scientific backgrounds conduct
experimental studies; geological, geographical, and geophysical field campaigns;
as well as workshops, symposia, and training courses; with a view to fostering
efficient structures and policies for fire mitigation, extinction, and prevention in
China and worldwide.
Coal fire fighting and prevention, in many countries, is still considered a minor
task; it is relegated to local mines and engineers, who at best work under central
or federal regulations. Internationally concerted approaches, standards, or policy
frameworks do not exist. The very slow and partly unsuccessful course of action that

coal fire fighting has taken so far can be explained by three significant shortcomings,
all human in origin:
1) Missing awareness of the ecological impact of coal fires,

2) Missing economic interest in coal deposits threatened by coal fires, and
3) Missing willingness to go to the great expense of extinguishing and preventing
coal fires.
Sufficient funding, appropriate mining and fire fighting techniques and

regulations, as well as national and best international policies need to be put in
place to tackle this triple dilemma. The Clean Development Mechanism (CDM),
as foreseen in the Kyoto Protocol, provides an interesting and powerful tool to
render coal fire extinction more profitable. As a link to international trading
markets of Certified Emission Reductions (CERs), it has the potential to attract
international investors to the coal fire problem and, in the long run, establish a
win-win situation for the owners of affected coal deposits and those bearing the
expense of fire fighting.
The “International Conference on Coal Fire Research – ERSEC Conference on

Understanding, Extinction, and Prevention of Spontaneous Coal Seam Fires,” held
in Beijing, P.R. China from 29 November to 1 December 2005, was deliberately
set at a phase change in the Sino-German Coal Fire Research Initiative. Summing
up the results of the first three years of the project, it also paved the way for a (still
ongoing) second phase. More than 200 participants attended the conference; a

total of 53 papers were presented in oral and poster sessions. These papers were
arranged around the following foci:
• Spontaneous coal seam fires as a worldwide challenge for resource conservation
and climate protection;
• Understanding the processes of ignition, combustion, and inertization of
spontaneous coal seam fires;
• Remote sensing and monitoring of coal fires;
• Geology, geophysics, and modeling; and
• Extinction and prevention of spontaneous coal seam fires.
A final session on “challenges, opportunities, and the way ahead” was to provide
a synthesis of all presentations and ensuing discussions, as well as to formulate
recommendations for the direction of further research work – in the then upcoming
Phase B of the project, but also for coal fire assessment in general. A panel
discussion, at the very end of the conference, tried to resume and explicate the most
contentious scientific and political issues that had arisen over the preceding days.
The central outcome of the conference, without doubt, is a common understanding

that coal fire extinction and prevention is a very complex task and involves a
whole range of disciplines in its four stages of mapping, modeling, mitigation,
and monitoring. Being of great economic and environmental impact, it concerns
individual enterprises as much as whole industries, local and national governments as
much as international organizations; it even touches upon international agreements
and treaties. Only if all parties make best use of the existing scientific, technological,

economic, and political tools and frameworks, and collaborate in a synergistic
fashion, can the coal fire problem be tackled in the long run. Any shortcoming in
either of these domains – science, technology, economy, and politics – is bound to
weaken the other domains’ capacity to mitigate coal fire impact. For instance, an
excellent fire mapping and monitoring system can only be applied successfully if
adequate financial support is given. The best fire extinction and prevention guidelines
remain on paper only, if the right policies to apply them are missing. Consequently,
it is very important to raise awareness for coal fires in all domains and support a rigid
and instantaneous implementation of fire fighting and prevention.
International conventions and policy frameworks that address the coal fire issue do
not exist today. They would however be a perfect vehicle to standardize fire fighting
and prevention worldwide and link the different local and/or national approaches to
the problem. With regard to funding, the Clean Development Mechanism (CDM) of
the Kyoto Protocol, through the generation of Certified Emission Reductions (CERs),
has the potential to make coal fire extinction and prevention a highly profitable
undertaking. Obstacles to such an international network for coal fire mitigation are
luckily being removed step by step by international capacity and knowledge building,
by new scientific and technological developments, and finally by a number of political
milestones, such as the recent ratification of the Kyoto Protocol.
With an explicit focus on the political, cross-border dimension of the problem, the
International Conference on Coal Fire Research was meant to contribute to this
development toward greater international coordination and cooperation. Its impact
10
as an outstanding platform for networking and exchange among the global coal fire
research community is sure to last, both in China and worldwide. We would like
to thank all decision-makers, scientists, and engineers who attended the conference,
listened and learned from the presentations, or presented work of their own. The
same goes to the contributors of this book, most of whom were scientists of the
Sino-German Coal Fire Research Initiative, and hence relied for their work on the
generous funding from the German Federal Ministry of Education and Research
(BMBF), the Ministry of Science and Technology of the People’s Republic of China
(MoST), Shenhua Group Corporation Limited, the Ministry of Land and Resources
of the People’s Republic of China (MLR), and several other entities – we sincerely
thank all of them for their support. The conference and its written legacy, this
book, were organized under “Ecological Research for Sustaining the Environment
in China,” the ERSEC Project of UNESCO Office Beijing. We are much obliged
to all reviewers, who, in a voluntary effort, offered their expertise to improve the
papers presented in the following pages. Special thanks, at last, go to Geoffrey
Schöning for his endless patience, enthusiasm, and perseverance in preparing the
conference and coordinating the editing process, which has led, eventually, to this
very milestone of coal fire research.
Dr. Stefan Voigt

Prof. Dr. Horst Rüter
Dr. Li Jiahong
Prof. Li Jing
Dr. R. Jayakumar
11
Acknowledgements
Many thanks to the following people:
 Review Panel
Referees
• Dr. Winfried Kessels, Germany
• Prof. Dr. Horst Rüter, Germany
• Dr. Martin Schmidt, Germany
• Dr. Stefan Voigt, Germany
Reviewers
• Prof. Dr. Uli Barth, Germany • Dr. Jolanta Kus, Germany
• Dr. Christian Fischer, Germany • Prof. Li Jing, P.R. China
• Dr. Ralf Fritschen, Germany • Dr. Christian Lohrer, Germany
• Dr. Hartwig Gielisch, Germany • Dr. Rupert Müller, Germany
• Dr. Siegfried Greinwald, Germany • Dr. Uwe Meyer, Germany
• Dr. Guo Shan, P.R. China • Prof. Dr. Dieter Oertel, Germany
• Dr. Kurt Günther, Germany • Dr. Anupma Prakash, U.S.A.
• Dr. Andreas Hirner, Germany • Dr. Peter Reinartz, Germany
• Ms. Han Jing, P.R. China • Mr. Wolfram Rühaak, Germany
• Mr. Ralf Junker, Germany • Dr. Stefan Schlömer, Germany
• Mr. Manfred Keil, Germany • Dr. Katrin Schwalenberg, Germany
• Dr. Ann G. Kim, U.S.A. • Mrs. Tang Ping, P.R. China
• Mr. Alexander Korten, Germany • Dr. Volker Tank, Germany
• Dr. Ulrich Krause, Germany • Dr. Stefan Wessling, Germany
• Mr. Thomas Krauss, Germany • Dr. Jens Wiegand, Germany
• Ms. Sonja Kuhlmann, Germany • Mr. Thomas Wündrich, Germany
12
• Dr. Manfred W. Wuttke, Germany
• Prof. Zhang Xinhai, P.R. China
• Dr. Boris Zhukov, Germany
 Summary Report
• Prof. Dr. Wolfgang Busch, Germany
• Prof. Dr. Cornelia Gläßer, Germany
• Prof. Li Jing, P.R. China
 Chinese Translations (non-Chinese authors)

• Dr. Zhang Jianmin, P.R. China
 Project Staff
• Dr. R. Jayakumar
• Ms. Han Xiaoting
• Mr. Liu Ke
• Ms. Liu Yi
• Mr. Geoffrey Schöning
• Ms. Elena Stecca
13
Organizations Involved
Project Name
Coal Fire Research – A Sino-German Initiative: “Innovative Technologies for
Exploration, Extinction, and Monitoring of Coal Fires in North China”
Project Partners – China

 National Remote Sensing Center of China (NRSCC)
15 Fuxing Street
Haidian District
100862 Beijing
P.R. China
 Shenhua Group Corporation Limited

3 Dongzhimen Nandajie
Dongcheng District
100007 Beijing
P.R. China
 China Geological Survey (CGS)

24 Huangsi Dajie
Xicheng District
100011 Beijing
P.R. China
 China National Administration of Coal Geology (CNACG)

299 Dianchang
Fengtai District
100039 Beijing
P.R. China
14
 Beijing Remote Sensing Corporation (BRSC), Shenhua Group
1 Shangdi Sijie
Haidian District
100085 Beijing
P.R. China
 Aerophotogrammetry and Remote Sensing Bureau of China Coal (ARSC)

3 Jianxi Street
Xi’an, Shaanxi Province
P.R. China
 Aero Geophysical Survey and Remote Sensing Center for Land and Resources
(AGRS)
29 Xueyuan Road
Xicheng District
100083 Beijing
P.R. China
 Beijing Normal University (BNU)

19 Xinjiekou Waidajie
Haidian District
100875 Beijing
P.R. China
 Beijing Ouhualian Science and Technologie Ltd. (BOST)

B Chengyuan Mansion, Rooms 301-305
3 Xisan Qijiancaicheng Zhonglu
Haidian District
100096 Beijing
P.R. China
15
 Research Institute for Coal Geophysical Exploration (RICGE)
56 Fanyang Xilu
072750 Zhuozhou, Hebei Province
P.R. China
 Xi’an Branch of China Coal Research Institute (CCRI)

52 Yanta Beilu
710054 Xi’an, Shaanxi Province
P.R. China
 Wuda Coal Mining Corporation, Shenhua Group

2 Bayinsai Xijie,
Wuda District
016000 Wuhai, Inner Mongolia Autonomous Region
P.R. China
 Ningxia Coal Corporation, Shenhua Group

217 Xinhuadongjie
750004 Yinchuang, Ningxia Autonomous Region
P.R. China
 State Coal Industry Bureau (SCIB), Inner Mongolia (IMBIC)

1 Cilechuan Street
010000 Huhehaote, Inner Mongolia Autonomous Region
P.R. China
Project Partners – Germany

 Federal Institute for Materials Research and Testing (BAM)
Unter den Eichen 87
12205 Berlin
Germany
16
 Federal Institute for Geosciences and Natural Resources (BGR)
Geozentrum Hannover
Stilleweg 2
30655 Hanover
Germany
 German Aerospace Center (DLR)

German Remote Sensing Data Center (DFD)
Oberpfaffenhofen
82234 Weßling
Germany
 Leibniz Institute for Applied Geosciences (GGA)

Stilleweg 2
30655 Hannover
Germany
 HarbourDom GmbH
Schürbankstraße 20a
44287 Dortmund
Germany
 Freiberg University of Mining and Technology

Institut für Bergbau und Spezialtiefbau
Zeunerstraße 1a
09596 Freiberg
Germany
17
 Bergische Universität Wuppertal
Gaußstraße 20
42119 Wuppertal
Germany
Project Partners – Associated Members

 Deutsche Steinkohle AG (DSK)
Shamrockring 1
44623 Herne
Germany
 International Institute for Geo-Information Science and Earth Observation (ITC)

Hengelosestraat 99
P.O. Box 6
7500 AA Enschede
Netherlands
 Deutsche Montan Technologie GmbH (DMT)

Am Technologiepark 1
45307 Essen
Germany
Conference Coordinator
 National Remote Sensing Center of China (NRSCC)
15 Fuxing Street
Haidian District
100862 Beijing
P.R. China
18
Extraordinary Sponsor
 Shenhua Group Corporation Limited
3 Dongzhimen Nandajie
Dongcheng District
100007 Beijing
P.R. China
Funding Agency
 German Federal Ministry of Education and Research (BMBF)
Division 623: Sustainability Concepts for Production and Consumption
53175 Bonn
Germany
Executive Agencies
 UNESCO Office Beijing
Waijiaogongyu 5-13-5
Jianguomenwai Compound
100600 Beijing
P.R. China
 Chinese National Commission for UNESCO

Division of Science and Culture
37 Damucang Hutong, Xidan
100816 Beijing
P.R. China
 Ministry of Education of the People’s Republic of China (MoE)

Department of Basic Research
37 Damucang Hutong, Xidan
100816 Beijing
P.R. China
19
Table of Contents
Foreword···························································································· 4
Preface································································································ 7
Acknowledgements············································································ 12
Organizations Involved····································································· 14

Part I
Statements
Mr. Zhang Guocheng / 张国成先生················································· 28 / 30

Executive Director General, National Remote Sensing Center of China (NRSCC), Beijing, P.R. China /
中国北京市，国家遥感中心主任
Dr. Renate Loskill / 罗瑞娜博士······················································ 32 / 35

Head of Division, Division of Sustainability Concepts for Production and Consumption, German Federal
Ministry of Education and Research, Berlin, Germany / 德国柏林，德国教育研究部生产与消费的持续性理
念部门负责人
Dr. Yasuyuki Aoshima / 青岛泰之博士· ············································ 37 / 39

Director and Representative, UNESCO Office Beijing, Office of the Representative to Democratic People’s
Republic of Korea, Japan, Mongolia, People’s Republic of China, and Republic of Korea, Beijing, P.R. China / 中国
北京市，联合国教科文组织驻朝鲜民主主义人民共和国，日本国，蒙古国，中华人民共和国，大韩民
国代表处主任和代表
20
Part II
Keynote Speeches
A. Sinha and V.K. Singh

Spontaneous Coal Seam Fires: A Global Phenomenon·································· 42
Guan Haiyan
Spontaneous Coal Seam Fires: The Chinese Perspective· ······························· 67
Stefan Voigt and Horst Rüter

Spontaneous Coal Seam Fires: The Sino-German Coal Fire Research Initiative·· 82
Part III
A Global Phenomenon
J. Denis N. Pone, Kim A.A. Hein, Glenn B. Stracher, Robert B. Finkelman, and Harold J. Annegarn
Potential Environmental and Health Impacts of Burning Coal in Witbank
Coalfield, South Africa··········································································· 94
M. Naze-Nancy Masalehdani, Jean-Luc Potdevin, Fabrice Cazier, and Dominique Courcot

Emission of Toxic Gases and Condensation Minerals from Burning Coal Waste
Heaps in Northern France: Measurements and Analyses······························ 103
Claudia Künzer, Zhang Jianzhong, Andreas Hirner, Jia Yaorong, and Sun Yulin
Multi-Temporal In-Situ Mapping of the Wuda Coal Fires from 2000 to 2005:
Assessing Coal Fire Dynamics································································ 132
Liu Bokun, Guan Haiyan, Ma Jianwei, Kong Bing, Wang Mei, Wu Xiaoying, and Xiang Yaling
General Investigations of Wuda Coal Fire Area· ········································ 149
21
Shi Xiaolei, Cao Daiyong, Fan Xinjie, and Wu Chacha
Environmental Impact Assessment of Spontaneous Coal Seam Fires in China···· 161
Kathryn A. Norman
Coal Fires as a Social Problem: An Anthropological Perspective···················· 171
Ümüt Halik, Friedrich-Karl Bandelow, and Jörg Schulz

Socio-Economic Aspects of the Local Mining Sector and Alternative Sources of
Income in the Helan Shan···································································· 193
Part IV
Basic Processes
and Investigations
Christian Lohrer, Martin Schmidt, and Ulrich Krause

Influence of Environmental Parameters on the Self-Ignition Behavior of Coal···· 210
Jolanta Kus, Werner Hiltmann, and Aenne Balke

Researching Coal Spontaneous Combustion: Micropetrography of
Coal Oxidation and Carbonization························································· 233
Yu Minggao, Jia Hailin, Zuo Qiuling, and Xu Jun

Research on Gas Indicators of Coal Spontaneous Combustion in
Wuda Coal Mining Area· ····································································· 249
Manfred P. Hochstein and Christopher J. Bromley

Heat Flux Measurement of Hot Ground·················································· 259
Stefan Schlömer, Manfred Teschner, Jürgen Poggenburg, and Christian Seeger

Gas and Temperature Monitoring of a Spontaneous Coal Seam Fire in
Wuda Coal Mining Area· ····································································· 277
22
Tom Litschke, Jens Wiegand, Stefan Schlömer, Hartwig Gielisch1, and Friedrich-Karl Bandelow
Detailed Mapping of Coal Fires in Combination with In-Situ Gas Flow
Measurements to Estimate Mass Flow Balance and Fire Development· ·········· 306

Part V
Remote Sensing
and Monitoring
Gerlinde Schaumann, Bernhard Siemon, and Yu Changchun

Geophysical Investigation of Wuda Coal Mining Area, Inner Mongolia:
Electromagnetics and Magnetics for Coal Fire Detection····························· 336
Prasun K. Gangopadhyay, Freek van der Meer, and Paul van Dijk
Monitoring Coal Fires Using Remote Sensing·················································· 351
Claudia Künzer, Zhang Jianzhong, Anke Tetzlaff, Stefan Voigt, and Wolfgang Wagner
Automated Demarcation, Detection, and Quantification of Coal Fires
in China Using Remote Sensing Data· ··················································· 362
Yang Bo, Li Jing, Chen Yunhao, Zhang Jianzhong, and Claudia Künzer
Automated Detection and Extraction of Surface Cracks from
High-Resolution Quickbird Imagery······················································· 381
Zhang Jianmin, Xiang Yaling, Wang Mei, and Huan Zhongdan

4-Level High-Resolution Detection and Monitoring – Applications in
Coal Fire Detection and Management····················································· 390
Zhang Jianmin, Huan Zhongdan, Wang Mei, and Xiang Yaling

3D Detection and Visualization of Underground Coal Fires· ······················· 410
23
Part VI
Data Management,
Visualization, and Modeling
Wang Mei, Zhang Jianmin, Wang Yufei, Gao Yunze, Li Guangliang, and Xiao Haihong
3D Visualization of Underground Coal Fires············································ 428
Li Guangliang, Wang Mei, and Xiao Haihong

Information Visualization in Coal Fire Research········································ 443
Huan Zhongdan, Tian Yuan, Sun Yujing, and Zhang Jianmin

Mathematical Simulations and Methods of Inversion on
Underground Coal Fires········································································ 451
Winfried Kessels, Manfred W. Wuttke, Stefan Wessling, and Li Xuan

Coal Fires Between Self-Ignition and Fire Fighting: Numerical Modeling
and Basic Geophysical Measurements······················································ 467
Ulrich Krause, Martin Schmidt, and Christian Lohrer

Computation of Real-Scale Coal Fire Scenarios Based on Linking Lab-Scale
Experiments with a Numerical Heat and Mass Transfer Model····················· 485
Stefan Wessling, Tom Litschke, Jens Wiegand, Stefan Schlömer, and Winfried Kessels
Simulating Dynamic Subsurface Coal Fires and Its Applications··················· 513
24
Part VII
Fire Fighting
and Prevention
Zhang Xinhai, Xu Jianzhong, Xi Guang, Deng Jun, and Wen Hu

Model and Control of High-Temperature Coal Fire··································· 544
D.D. Tripathi
Control and Prevention of Coal Fires in Sealed-Off Mine Areas
Using Different Means of Inertization····················································· 554
D.D. Tripathi
Control of Shallow-Depth Underground Coal Fires in Jharia Coalfield·········· 567
Uli Barth, Birte Grashorn, and Friedrich-Karl Bandelow

Sustainable Control of Coal Fires Through Systematic Decision Processes and
Cyclic Management············································································· 576
Friedrich-Karl Bandelow, Hartwig Gielisch, and Jörg Schulz

CER-Trading as a Means of Funding Coal Fire Fighting in China················· 585
Part VIII
Summary
and Conclusions
Toward a Better Understanding of Coal Fires for Sustainable Fire Control and
Prevention: Conclusions and Recommendations /
进一步提高面向可持续发展的煤火控制与预防的认识：
结论与建议··············································································· 596/ 602
25
Part I
Statements
Mr. Zhang Guocheng
Executive Director General, National Remote Sensing Center of China (NRSCC), Beijing, P.R. China
Distinguished guests, ladies and gentleman,
Good morning!
It is my extraordinary pleasure to be attending today the “International Conference on

Coal Fire Research – ERSEC Conference on Understanding, Extinction and Prevention
of Spontaneous Coal Seam Fires.” Please allow me to congratulate you on behalf of the
Ministry of Science and Technology of the People’s Republic of China to the opening of
this conference. My warmest greetings go out to the scientists and delegates from Germany,
the U.S.A., France, India, Mongolia, Kazakhstan, New Zealand, and China, and to the
representatives of international organizations that have gathered here today!
Ladies and gentlemen,

Spontaneous coal seam fires are a problem of global proportions. The coal fires raging in
China’s north are becoming ever more serious; they do not only destroy coal resources and
pollute the environment, but also seriously threaten mining safety and production. Each
year the Chinese coal mining areas mobilize a great amount of manpower, equipment, and
capital to extinguish and prevent coal fires. Improving fire fighting through scientific and
technological cooperation is a keen interest of the mining industry, but it also figures high
on the agenda of coal fire research.
The Chinese and German governments are very concerned about the coal fires in China’s north;
they share this concern with UNESCO. In September 2003, both governments launched
the scientific cooperative project “Innovative Technologies for Exploration, Extinction, and
Monitoring of Coal Fires in North China.” This Sino-German Coal Fire Research Initiative is
jointly managed by the National Remote Sensing Center of China (NRSCC) and the German
Aerospace Center (DLR). It has also been placed under the umbrella of a UNESCO project –
“Ecological Research for Sustaining the Environment in China (ERSEC).”
28
Spontaneous Coal Seam Fires: Mitigating a Global Disaster
Thanks to the special attention that the initiative has received from officials both in
the Chinese Ministry of Science and Technology and the German Federal Ministry of
Education and Research, but also to the devotion with which the scientists from China
and German went about their work, the first phase of the Sino-German Coal Fire Research
Initiative has already been completed successfully. Its purpose was to create a comprehensive
detection and analysis system based on spaceborne remote sensing, airborne surveys and
ground monitoring so as to provide the foundation for large-scale underground coal fire
exploration and fire fighting in a second phase. At present, this second phase is being
planned by the Chinese and German project partners.
Today’s conference serves to summarize and present the key technological achievements
in coal fire exploration, monitoring, and extinction made over the last years, as well as
their modes of application; it also functions as a discussion forum for innovative research
methods and theories, for latest trends and developments in coal fire fighting and research
in China and abroad. At the same time, the conference is meant to facilitate a more
thorough-going understanding of the first results of the Sino-German Coal Fire Research
Initiative and offers ample opportunity to discuss all scientific and organizational matters
pertaining to the second phase of the project. For officials from the two involved ministries
and state departments under whose authority fire fighting is organized, it will sure be a
source of insight into new mitigation strategies and measures.

I can reassure you that the Ministry of Science and Technology of the People’s Republic of
China will continue its support for the Sino-German Coal Fire Research Initiative！
At last, I wish this conference all success, and every delegate a splendid stay in Beijing!
29
张国成先生
中国北京市，国家遥感中心主任
尊敬的各位来宾，女士们，先生们：
上午好！
我非常高兴能够出席这次“国际煤火研讨会——暨中国环境可持续发展生态研
究（ERSEC）框架下煤层自燃探测、灭火及预防研讨会”。请允许我代表中华
人民共和国科学技术部，对研讨会的召开表示热烈的祝贺。同时向来自德国、
美国、法国、印度、蒙古、哈萨克斯坦、新西兰、中国等8个国家以及国际组织
的专家及参会人员表示诚挚的问候！
女士们，先生们：
煤层自燃是一个全球的问题。中国北方地下煤田火灾越来越严重，不仅破坏
煤炭资源和生态环境，也严重威胁煤矿的安全生产。中国各大矿区每年投入
大量人力、物力和财力用于防、灭火工程。通过科技合作，促进矿区防火、
灭火工作，一方面是矿区生产单位的迫切希望，同时也是煤田灭火研究领域
的关键内容。
中德两国政府以及联合国教科文组织都对中国北方地下煤田自燃十分关注。在2003
年9月，中德政府间科技合作项目“中国北方煤火探测、灭火与监测新技术研究”
正式启动，该项目由中国国家遥感中心和德国宇航中心共同主持，同时该项目被纳
入联合国教科文组织的“中国环境可持续发展生态研究（ERSEC）”框架。
在中国科技部和德国教研部有关领导的亲切关怀下、在中德双方专家的共同努
力下，该项目目前第一期已经顺利结束，初步形成了一套包括卫星遥感、航空
物探及地面观测组成的综合探测与分析方法，为第二期全面探测地下煤火和进
行灭火提供了基础。目前中德双方正在推进该项目的第二期的合作。
此次研讨会的召开，将重点总结、交流国内外近年来煤火探测、监测与灭火新
技术成果和重大应用，研讨具有创新意义的研究方法和理论，以及国内外有关
30
技术的前沿动态和发展趋势。同时对中德政府间科技合作“中国北方煤火探
测、灭火与监测新技术研究”项目第一期合作研究的成果进行深入的交流，对
该合作项目的第二期的合作内容、方向等诸多方面进行深入讨论。同时还将为
相关的政府决策部门提出煤火防治新思路和建议措施。
中华人民共和国科学技术部还会一如既往地对“中国北方煤火探测、灭火与监
测新技术研究”项目给予支持！
最后，祝本次研讨会取得圆满成功，祝各位代表在北京生活愉快！
31
Dr. Renate Loskill
Head of Division, Division of Sustainability Concepts for Production and Consumption,
German Federal Ministry of Education and Research, Berlin, Germany
Dear Mr. Zhang Guocheng, Mr. Lei Chaozi, Mr. Liu Jiang, Dr. Aoshima,
I am delighted to welcome you on behalf of the German Federal Ministry of Education

and Research (BMBF) to the “International Conference on Coal Fire Research – ERSEC
Conference on Understanding, Extinction, and Prevention of Spontaneous Coal Seam
Fires.” I would like to pass on greetings from Mr. Riehl, Head of Division at the German
Federal Ministry of Education and Research and responsible for Sino-German cooperation
in the field of environmental technology and ecology. Mr. Riehl would have very much
liked to be here today, but, unfortunately, had to cancel his trip to Beijing at short notice
in order to attend to affairs in connection with the formation of the new government in
Germany.
This conference is a milestone in Sino-German cooperation in research and education. Its

subject is the development of concrete methods to prevent spontaneous coal seam fires and
to reduce their environmental and economic impact. An interdisciplinary collaborative
research project is addressing this issue. This project is coordinated on the German side
by the German Aerospace Center (DLR). The scope of its work demands that numerous
research institutions work together with state authorities and representatives of the
industry. Seven German and thirteen Chinese partners are involved in the project. By
way of example, I would like to mention here: Shenhua Group Corporation Limited and
Deutsche Montan Technologie GmbH (DMT), the National Remote Sensing Center
of China (NRSCC), Beijing Normal University, and Freiberg University of Mining and
Technology.
This project can make an effective contribution to climate protection and resource
conservation. I do not intend to go into detail about the approaches and objectives of
Sino-German research cooperation in this area or the successes which have already been
32
achieved. I am happy to leave this to subsequent speakers, the scientists who have actually
performed this exciting research work.
However, I would like to take this opportunity to emphasize that this research will bring us
a step closer to our goal – achieving sustainability. This is the accepted ideal of responsible
political action. On 30 June 2004, the German Federal Government introduced a new
framework program entitled “Research for Sustainability” as part of Germany’s national
strategy for sustainability. The program aims to promote new technologies and plans of
action for sustainability – and Sino-German cooperation in research and education is
making an important contribution to it.
I would like to thank the Chinese side and both ministries involved, the Ministry of Education
(MoE) and the Ministry of Science and Technology (MoST) of the People’s Republic
of China, for the long years of successful cooperation that link our two countries.
The present conference must be seen within the framework of this cooperation –
a cooperation which essentially pursues three important objectives:
First of all, it is essential to determine the status quo and to define future opportunities
and challenges for interdisciplinary, collaborative research, and to discuss these with our
Chinese partners and with international experts. The joint MoST/BMBF research project,
which will be expanded even further in the year to come, can look back on years of very
successful work. In this context, I would like to refer in particular to the close cooperation
between DLR and NRSCC, which received an important stimulus from the visit of the
Minister of Science and Technology of the People’s Republic of China in June 2002.
Secondly, in line with the objectives of the German “Research for Sustainability” framework
program, it is also important to make pertinent knowledge available to decision-makers.
The ERSEC umbrella project “Ecological Research for Sustaining the Environment
33
ERSEC Ecological Book Series - 4
in China,” a joint initiative by MoE, BMBF, and UNESCO Office Beijing, offers a
perfect framework for this. ERSEC does not conduct research of its own, but encourages
transdisciplinary exchange on selected topics and provides information tailored for specific
user groups. UNESCO Office Beijing, as the executive agency of ERSEC, has already
organized a number of conferences and other transfer activities within the framework of
this umbrella project.
Thirdly, this conference is meant to contribute to capacity-building in China and stimulate

cross-border scientific communication. One particular highlight will be the ERSEC Young
Scientist Award ceremony at the end of this session. At this point, I would like to sincerely
congratulate the prize winners.

I have been able to attend two ERSEC conferences on different topics in the course of the
last few years. I have profited greatly from each of these conferences – thanks to the quality
of the presentations, the international expertise, and the outstanding hospitality of both the
Chinese partners and UNESCO. I am looking forward to the upcoming sessions and will
follow the discussions with keen interest.
Thank you for your attention.
34
罗瑞娜博士

德国柏林，德国教育研究部生产与消费的持续性理念部门负责人
尊敬的张国成先生，雷朝滋先生，刘疆先生，青岛泰之博士，
我仅代表德国教育研究部十分荣幸的欢迎各位出席“国际煤火研讨会——暨中
国环境可持续发展生态研究（ERSEC）框架下煤层自燃探测、灭火及预防研讨
会”。在此，我特意转达德国教育与研究部部门负责人并主管中德环境科学与
生态领域合作的Riehl先生的问候，Riehl先生非常愿意参加今天的会议，但很遗
憾因需参加与德国新政府成立有关事务而取消了北京之行。
本次会议是中德研究与教育领域合作的重要里程碑，会议主题是开发预防煤层
自燃和降低对环境和经济影响的具体方法，多学科合作研究项目正是突出了
该主题。中德煤火项目是由德国宇航中心协调，其工作范围需要众多研究机构
与国家政府和工业代表的共同合作，德方七个单位和中方13个单位参与了该项
目。我想提及的是神华集团、德国蒙坦技术公司、中国国家遥感中心、北京师
范大学和德国福莱希贝尔格技术大学。
该项目为环境及资源保护做出了重要贡献。在此，我不想就中德研究合作途径
与目标以及已取得的成果赘述，愿留给具体开展这项挑战性研究工作的科学家
们详细叙述。
然而，借此机会我要强调的是该研究使我们逐步接近于我们提出的可持续性发
展目标，这也是有责任政府作用的共识。2004年6月30日，德国联邦政府提出了
题为“可持续性研究”新框架计划作为德国可持续性发展战略的部分内容，该
计划旨在提出促进可持续性发展的新技术和行动计划，而且中德研究与教育方
面的合作正在为此做出重要的贡献。
在此我想感谢中方和参与部门，中华人民共和国教育部和科技部多年来为两国
合作做出的成功努力。在合作框架下的本次会议可见合作基本上把握了三个重
要目标：
35
首先，最基本的是确定跨学科合作研究的现状、未来机遇和挑战，并需要与中
方合作伙伴和国际专家进行讨论。中国科技部和德国科教部之间的合作研究项
目在近几年已非常成功的基础上将在未来几年内进一步扩大。我要特别提到的
是德国宇航中心与中国国家遥感中心的合作，在2002年6月中国科技部长访问德
国中获得了重大推进。
其次，与德方“可持续性研究”框架计划目标一致，为决策者了解有关知识也
是相当重要的。中国教育部与德国教育研究部以及联合国教科文组织北京办事
处联合发起的ERSEC框架下中国支撑环境的生态学研究项目就为提供系统的知
识创造了完美的框架。ERSEC并不是孤立地开展研究，而是在选择主题时就鼓
励跨学科之间的交流，并为特殊的用户提供所需信息。联合国教科文组织北京
办事处作为ERSEC的执行机构在该框架下已经组织了许多会议及交流活动。
第三，本会议有助于中国扩展研究能力，激励交叉边缘学科的交流。其中最突
出的是在本次会议结束时将进行ERSEC青年科学家授奖仪式。在此，我想向获
奖者表示衷心祝贺！
各位女士和先生：
我在过去几年中已经参加两次有关不同主题的ERSEC会议。每次我从参加会议
的精彩演示、国际知识、中方合作者和联合国教科文组织的热情款待都获益匪
浅。我期待接下来的会议内容，并将热切关注讨论。
感谢各位出席会议！
36
Dr. Yasuyuki Aoshima
Director and Representative, UNESCO Office Beijing,
Office of the Representative to Democratic People’s Republic of Korea, Japan, Mongolia,
People’s Republic of China, and Republic of Korea, Beijing, P.R. China
Respected Dr. Renate Loskill,

Dear Mr. Zhang Guocheng, Mr. Lei Chaozi, and Mr. Liu Jiang,
Distinguished guests, ladies and gentlemen,
On behalf of the United Nations Educational, Scientific and Cultural Organization

(UNESCO), it is my great pleasure to welcome you all to the “International Conference
on Coal Fire Research” under ERSEC umbrella – ERSEC, that is “Ecological Research for
Sustaining the Environment in China.”
We would like to express our sincere gratitude to the German Federal Ministry of Education
and Research for providing the necessary funds to the ERSEC initiative and thus making
this conference possible. We would also like to thank the Ministry of Education of the
People’s Republic of China, the Chinese National Commission for UNESCO, and all other
bilateral project partners from China and Germany – without their continued support, this
series of international conferences could have never been implemented. Our special thanks,
at last, goes to the Ministry of Science and Technology of the People’s Republic of China,
the National Remote Sensing Center of China, and Shenhua Group Corporation Limited
for their contribution, both financial and in kind, to this conference. Likewise, the support
of the Sino-German Center for Research Promotion should be duly acknowledged at this
point. Thank you very much for providing these excellent conference facilities!
The “International Conference on Coal Fire Research” is the fourth of a series of

international ERSEC conferences. These conferences serve as a platform for the
different projects under ERSEC umbrella; they are meant to help them engage in
interdisciplinary exchange with other scientists and institutional experts, and to present
and discuss their findings. The results of these three-day scientific encounters – research
syntheses and recommendations – are published in the ERSEC Ecological Book Series,
directed at planners, policy-makers and the general public. Book no. 1 left the printing
37
presses this September and no. 2 is expected to follow suite at the end of December; the
remaining books no. 3 and 4 will be published soon.
At this conference, we will discuss one of the most serious environmental and socio-
economic problems in China – coal seam fires. Far from being endemic to this country, the
phenomenon is known all over the globe – in India, the U.S.A., in Australia, Indonesia, and
South Africa. Coal seam fires do not only destroy valuable coal resources, they also impact
heavily on their immediate inanimate and living environment. Whereas large amounts of
aerosols and toxic gases affect local flora and fauna, and of course the people living next to
coal fire areas, greenhouse gases – carbon dioxide and methane, for example – take their toll
on a regional and global level. Coal fires then, are a global issue and should be the subject of
international scientific cooperation – in fact, they are, as shows this conference.
We are glad to inform you that international experts from India, the U.S.A., South Africa,
New Zealand, the Netherlands, France, Mongolia, and Kazakhstan are participating this
morning. They are going to join, over the next two days, the experts of the Sino-German
Coal Fire Research Initiative to discuss their research findings and add valuable experience
from their home countries. The last half-day, in the morning of 1 December, is the most
crucial part of the conference. I hope you will all participate in the panel discussion,
and together we will achieve this conference’s foremost objective: recommendations for
implementing socio-economically viable and environmentally friendly policy measures to
combat coal seam fires in China and worldwide.
Let me take this opportunity to once again express our sincere gratitude to all of you for
coming and contributing to this conference. I wish you the most productive sessions and
hope they will result in the most knowledgeable recommendations for the international
problem that are coal seam fires.
Thank you.
38
青岛泰之博士
中国北京市，联合国教科文组织
驻朝鲜民主主义人民共和国，日本国，蒙古国，中华人民共和国，大韩民国
代表处主任和代表
尊敬的罗瑞娜博士，
尊敬的张国成先生，雷朝滋先生，刘疆先生，
尊敬的各位贵宾，女士们，先生们：
我代表联合国教科文组织热烈欢迎各位参加“中国环境可持续发展生态研究”
（ERSEC）框架下的“国际煤火研究会议”。
在此我想对德国联邦教育与研究部为ERSEC倡议并使本次会议成功召开所提供
的必要资助表示诚挚的感谢。同时也感谢中国教育部、联合国教科文组织中国
国家委员会，以及所有参与双边项目的中德双方合作伙伴，如果没有他们的支
持，这一系列的国际会议将无法实现。最后，还要特别感谢中国科技部，中国
国家遥感中心，神华集团为本会议提供的资金及各方面的支持。同时还要提及
的是中德研究促进中心，感谢他们为本次会议提供的完备设施。
“国际煤火研究会议”是一系列ERSEC国际会议的第四次会议，这些会议都为
ERSEC框架下不同项目提供了一个平台，旨在帮助各个项目与其他的科学家和
学科专家进行跨学科交流，共同展示并讨论他们的成果。本次会议三天的学科
碰撞——即研究成果和建议将在面向规划者、政策制定者和公众的ERSEC生态
系列丛书中发表。
本次会议上，我们将讨论在中国最严重的环境与社会经济问题之一 —— 煤火。
煤火问题远不止在中国，而且在世界很多地方都存在，如印度、美国、澳大利
亚、印尼和南非。煤火不仅损害了宝贵的煤炭资源，而且严重影响周围无生命
和有生命的环境。大量浮尘和有毒气体不仅会影响当地的动植物群，自燃也影
响了火区附近居住的居民，诸如二氧化碳温室气体和甲烷都在区域和全球范围
产生影响。因此，煤火是一个世界性的问题，事实上应该如本次会议所见是一
个国际科学合作研究的主题。
39
今天早晨，我们很高兴获悉来自印度、美国、南非、新西兰、荷兰、法国、蒙
古、哈萨克斯坦等国专家参加了本次会议，将在接下来两天会议中与中德煤火
研究项目专家共同讨论取得的成果并进一步补充各国有价值的经验。12月1日上
午最后半天会议也是本次会议最重要的部分，我希望到时各位都能参加小组讨
论，共同努力实现本次会议最重要目标：为实施社会经济上可行和环境友好的
政策措施来治理中国及世界范围的煤火问题而献计献策。
请允许我借此机会再次对各位与会者的出席及为此会议所给予的支持表示诚挚
的感谢，祝愿您们在本次会议取得积极的成效，并希望这些成果对煤火国际问
题提出更精辟的建议。
谢谢！
40
Part II
Keynote Speeches
Spontaneous Coal Seam Fires:
A Global Phenomenon
A. Sinha and V.K. Singh
Central Institute of Mining and Fuel Research (CIMFR)

(Council of Scientific and Industrial Research, CSRI), Dhanbad, Jharkhand, India
Abstract
Coal fires are of great concern to scientists, politicians, and the general public worldwide. Almost all coal-
producing nations are reporting coal fires – Australia, China, Germany, India, Indonesia, South Africa, and
the U.S.A. have been faced with the problem for decades. Besides effecting local life and property, coal
fires are a problem of truly global proportions: they contribute to global warming, pollute the environment
in various degrees and on various levels, and destroy valuable non-renewable resources. Given this
enormous impact, it is imperative for the scientific community to strive for answers to the fundamental
questions that arise from the topic: How do coal fires start? What hazards do they cause? How can they be
detected and their extent assessed? What technology is required for their extinction, or better: prevention?
Undoubtedly, scientists have not stood idle in the face of the problem – R&D on coal fires started with
the first official fire reports. Quite a number of issues, however, remain unresolved and technical solutions
still show room for improvement. The present paper deals with general questions on understanding,
assessment, extinction, and prevention of coal fires on a global scale.
摘要
煤火是世界上科学家、政治家和公众都关心的问题。几乎所有的煤炭生产国家都有煤火方面的报
道，澳大利亚、中国、德国、印度、印度尼西亚、南非和美国几十年来一直都面临着煤火问题。
煤火除影响当地的居民和财产外，还是一个全球性的问题，在不同程度和不同层次上加剧了全球
变暖和污染环境、以及破坏宝贵不可再生的煤炭资源。针对煤火的巨大影响，科学界有必要给出
产生煤火问题的基本答案：煤火是如何开始的？煤火会引发什么样的灾害？如何探测煤火和估计
煤火范围？灭火及其预防都需要什么技术？当然，科学家们面临这种挑战并非无奈，煤火研发从
报道煤火现象起的正式报告就已经开始。然而，还有相当多未解决的问题，并且已有的技术方案
仍有进一步改进的空间。本文就讨论了煤火的认识、评价、灭火和预防等问题，并介绍了全球性
的煤火相关问题。
42
1 Introduction
Nature has provided mankind with a virtual gift in the form of coal – a plentiful resource,
able to supply a large share of our daily energy demand. Spontaneous fires in coal,
though, put coal users and scientists at a serious risk. Those fires are the result of heat
accumulation in oxidation processes, set off when coal is exposed to atmospheric oxygen
(O2). Here, the term “spontaneous” signifies that the burning process starts independent of
any recognizable external source.
Spontaneous fires in coal may occur naturally or be triggered by other “external” causes.
Yet, if a coal seam ignites and countermeasures fail to be taken at an early stage, the seam
may continue to burn for tens to hundreds of years; this time depending primarily on the
availability of O2. Fires occurring in natural course are generally referred to as coal fires.
Coal fires include surface coal seam fires, underground coal seam fires, fires in coal and
carbonaceous dump material, as well as fires in stored coal or coal that is being transported.
About 70 % of the coal fires in India are the result of spontaneous heating (Dhar 1996;
Zutshi et al. 2001). Coal fires have been reported from Australia, China, the Czech
Republic, Germany, India, Indonesia, Pakistan, Poland, Spain, Venezuela, and the U.S.A.
Among these, China, India, and the U.S.A are particularly affected. Coal fires have become
a challenge for researchers and engineers throughout the world (Figure 1).
Figure 1: World map of coal fire regions

Data source: DLR (2005)
43
2 Spontaneous Combustion
Spontaneous combustion in coal is a result of heating through oxidation processes, which

may be accelerated by catalytic compounds (e.g. water, pyrite, etc.). Processes like
adsorption and desorption of water also generate heat owing to the difference between in-
situ and equilibrium moisture in coal and air.
Coal oxidation starts with exothermic chemical reactions (Kok 1981; Schmal 1989). It
can be described as a process of three sequential steps. These are: (i) physical adsorption;
(ii) chemical adsorption or chemisorption, resulting in coal-oxygen complexes; and (iii)
chemical reaction. The chemical reaction breaks down less stable coal-oxygen complexes
to form gaseous products such as carbon monoxide (CO), carbon dioxide (CO2), and water
(H2O); it can be simplified as in equation (1) (Rosema et. al. 1999):
(1) Coal + O2 →CO2+ ∆H↑
The above reaction takes place in several steps and pathways, which, in turn, depend on
the presence of substances like water and pyrite. The complete oxidation of coal may be
described as in equation (2) (Kok 1981; Schmal 1989):
(2) C100H74O11 + 113O2 → 100CO2 + 37H2O (+ 4.2 ×108J kmol-1 O2)
The initial step of this reaction is the chemical adsorption or chemisorption of O2 at the
coal surface; it is presented in equation (3) (Kok 1981; Schmal 1989):
(3) C100H74O11 + 17.5O2 → C100H74O46 (+ 2.5 ×108J kmol-1 O2)
In the presence of catalytic substances like pyrite, heat-generating reactions as shown in

equations (4) and (5) take place (Wilke 1966; Schmal 1987a):
(4) 2FeS2 + 7O2 + 2H2O → 2H2SO4 + 2FeSO4 (+ 0.37 ×108J kmol-1 O2)
(5) 4FeS2 + 15O2 + 2H2O → 2H2SO4 + 2Fe(SO4)3 (+ 0.56 ×108J kmol-1 O2)
Equation (6) can be used to calculate the total quantity of O2 required for combustion of
1 kg of coal. Assuming C2H to be its basic component, about 11.1 kg of O2 are required for
complete combustion (Rosema et al. 1999):
(6) 4C2H + 9O2 → 8CO2 + 2H2O + ∆H
44
3 Parameters Affecting Spontaneous Combustion
There are several parameters, namely mining, as well as geological and chemical
conditions, which influence the process of spontaneous heating and combustion in coal.
Mining parameters include panel dimension; ratio of coal extraction; residual coal in goafs,
pillar stocks (parted pillars), partings, and mine roof and floor; as well as deformations
caused by underground mining activities. In opencast mining, micro- and macrocracks in
coal benches, outcrops, and loose coal at the surface of coal benches are main parameters.
Geological parameters are tectonic disturbances, faults, joints, cracks and fissures, as well
as organic minerals. The chemical content of coal is decisive for spontaneous combustion
too. A list of parameters affecting coal spontaneous combustion is given in Table 1.
Table 1: Parameters affecting coal spontaneous combustion

Intrinsic Factors Extrinsic Factors
(nature of coal) (atmospheric, geological, mining conditions)
Pyrites Temperature
Moisture Moisture
Particle size and surface area Barometric pressure
Rank and petrographical O2 concentration
constituents
Bacteria
Chemical constituents
Coal seam and surrounding strata
Mineral matter
Mining Method
Ventilation system and airflow rate
Support system
Data source: Guney (1968); Kaymakci & Didari (2002)
4 External Causes
External causes also contribute significantly to spontaneous fires in coal mines; these
are: conveyor belt frictions, electrical short circuits, and explosions, as well as Bantulsi
conflagration, dumping of hot ash, and illicit distillation. At some places, unprofessional/
improper mining has led to mine fires too.
45
5 Hazards Associated with Coal Fires
Coal fires have serious detrimental effects on coal mining, both in terms of commercial
exploitation and environmental safety. Apart from consuming valuable non-renewable
resources, they pose considerable operational difficulties and may increase mining costs
drastically. Another problem is land subsidence, which threatens surface and underground
infrastructure. Greenhouse gases are emitted in huge quantities, CO 2 being the most
prominent among them. At present, the Chinese coal fires alone are the source of 2 to 3 %
of the world's total annual emission of CO2.
The environmental effects of coal fires are observable on a local and global level. Heat and
noxious gases affect the fires’ immediate environment, making it inhospitable for plant,
animal, and human life. Smoke from coal fires contains poisonous gases (carbon, nitrogen,
and sulfuric oxides), which along with particulate matter are the main cause of lung and
skin diseases, e.g. chronic bronchitis and asthma. Extensive surface cracking is another
adverse effect of coal fires. Buildings, roads, and railways may be damaged in ground
movements and subsidence.
The major commercial setback resulting from coal fires is the loss of large amounts of coal. It
is not only the burned coal that becomes useless for economic purposes, but also neighboring
coal reserves, as they are barred from extraction or thermally altered through the fire.
6 Detection and Assessment
Since the beginning of the twentieth century, a number of fire indices and ratios have
been put forward to help assess the development stage of coal fires in underground mines.
Most of them work with gases produced during coal combustion and are premised on the
following two assumptions:
1) The (atmospheric) air which enters the mine contains 20.93 % O2, 79.04 % nitrogen
(N2), and 0.03 % CO2 (other gases excepted).
2) N2 is neither added to nor taken from the air during oxidation.
CO concentration, in particular, is used as an indicator for spontaneous heating/combustion in

combination with the Graham’s Ratio (Graham 1920-21). Sometimes, the production of hydrogen
(H2) is interpreted as a signal of fire emergence or explosion risk. The presence of H2 and ethylene
46
(C2H4) is a good indicator of advanced heating. A brief description of different fire indices and
(C2H4)used
ratios is aworldwide
good indicator of advanced
for detection heating. A of
and assessment brief description
mine of different
fires is given fire indices and
in Table 2.
ratios used worldwide for detection and assessment of mine fires is given in Table 2.
Table 2: Fire indices/ratios used worldwide for detection and assessment of coal fires
Table 2: Fire indices/ratios
Fire Index/Ratio used worldwide for detection
Significance/Interpretation Merits and assessment of coal fires
Demerits
Graham’s Ratio (GR)
Fire Index/Ratio < 0.5 Normal
Significance/Interpretation Remains
Merits unaffected Does not yield fire extent,
Demerits
by dilution through i.e. amount of coal
Graham’s Ratio (GR)
100CO / (0.265N - O2) > 0.5 <- <0.5
1 Normal
Heating Remains
air or CHunaffected
. Does not yield fire extent,
involved.
2
by dilution 4
through i.e. amount of coal
100CO / (1914-15)
Graham (0.265N2 - O2) > >0.51 -- <
< 21 Heating stage
Advanced air or CH
Allows for. early involved.
Disappearance of CO
4
fire detection does not mean complete
Graham (1914-15) > 1 - <> 23 Advanced
Active fire stage Allows for early
if fire increases Disappearance of CO
extinction of fire.
fire detection
continuously. does not mean complete
> 73 Active
Fire fire
ablaze if fire increases extinction of fire.
continuously.
Assesses fire status.
>7 Fire ablaze
Young’s Ratio (YR) < 25 Normal Assesses
CO foundfire status.
where fire CO2 from extraneous
2
reaches advanced sources and its solubility
Young’s 2Ratio
100(CO - 0.03)(YR)
/ 25-50
< 25 Superficial
Normal heating CO
stage.
2
found where fire CO frommay
in water
2
extraneous
lead to
(0.265N2 - O2) reaches advanced sources and its solubility
misinterpretations.
100(CO2 - 0.03) / > 50
25-50 High-intensity fire
Superficial heating stage. in water may lead to
(0.265N2&- O
Storrow )
Graham misinterpretations.
(1925)
2
> 50 High-intensity fire
Storrow & Graham
(1925)and Trickett Ratio
Jones < 0.4 No fire Can identify nature/ Does not indicate fire
type of fire. status.
Jones
(CO 2
+and Trickett
0.75CO - Ratio 0.4-0.5
< 0.4 CH burning
No 4fire Can identify nature/ Does not indicate fire
0.25H2) / (0.265N2 - O2) type of fire. status.not allow for early
Does
(CO2 + 0.75CO - 0.4-0.5
0.5-1 CH4 burning
Coal, oil, conveyor fire detection.
0.25Hand
Jones 2
) / (0.265N
Trickett2 - O2) belt burning Does not allow for early
(1954) 0.5-1 Coal, oil, conveyor fire detection.
Jones and Trickett 0.9-1.6 Timber burning
belt burning
(1954)
0.9-1.6
> 1.6 Timber
Error inburning
analysis
> 1.6 Error in analysis
Willets Ratio Figures vary from seam to seam. Allows for early Does not indicate fire
Ratio increases with increasing fire detection if CO status.
CO / (excess
Willets Ratio N2 + CO2 Figures vary from seam to seam. Allows for early Does not indicate fire
temperature; also depending on produced.
+ combustibles) excess Ratio increases with increasing fire detection if CO status.
If CO disappears,
CO magnitude and extent of fire.
(N2 +/ (excess N2 + CO
Ar) = sample (N22 temperature; also depending on produced. prediction of fire status
+ combustibles)
+ Ar) excess
- (3.8 sample O) If COdifficult.
very disappears,
magnitude and extent of fire.
(N2 + Ar) = sample (N22 prediction of fire status
+ Ar) -(1951-52)
Willet (3.8 sample O2) very difficult.
C / H Ratio
Willet (1951-52) 0-2 Normal Can indicate fire
intensity and extent.
C / H 2Ratio
6(CO + CO + CH4 + 0-2
3-4 Normal
Superficial heating Can indicate fire
2C2H4) / 2(0.265N2 - O2 intensity
Can and extent.
distinguish
6(CO
- CO22++CH CO + + CH
C2H4 4++ 3-4
>5 Superficial
Active fire heating between coal and
2C
H2 2-HCO)
4
) / 2(0.265N
4
2
- O2 Can distinguish
wood fires.
- CO2 + CH4 + C2H4 + >> 20
5 Active
Fire fire
ablaze between coal and
H2 - CO)
Ghosh and Banerjee wood fires.
Large range of
> 20 Fire ablaze
(1967) values provides
Ghosh and Banerjee Large rangeforoffire
better tool
(1967) values provides
characterization.
better tool for fire
Morris Ratio Figures vary from seam to seam. Very useful for single
characterization. Does not indicate fire
Ratio increases with increasing seam assessment. status.
Morris
N 2
/ (CORatio
+ CO2) Figures vary from seam to seam.
temperature. Very useful for single Does not indicate fire
Ratio increases with increasing seam assessment. status.varies randomly
Ratio
N2 / (CO(1988)
Morris + CO2) temperature. with high temperatures,
Ratio variesonrandomly
depending production
Morris (1988) with
of COhigh
andtemperatures,
CO2.
depending on production
of CO and CO2.
47
47
47
ERSEC Ecological Book Series - 4 (continued)
ERSEC Ecological Book Series -Significance/Interpretation
Fire Index/Ratio 4 Merits Demerits (continued)
(continued)
Fire Index/Ratio Significance/Interpretation Merits Demerits
Litton’s Ratio (R)
Fire Index/Ratio ≤1 Equilibrium for 30
Significance/Interpretation Provides good information
Merits Does (continued)
not indicate
Demerits fire
Litton’s Ratio ≤1 days until ambient
Equilibrium for 30 for sealed-off
goodareas. status.
Fire –3/2 8 (R)
Index/Ratio Significance/Interpretation Provides
Merits information Does not indicate fire
Demerits
1/3COs R O-1/2 temperature reached
Litton’s Ratio g
(R)-1/2 ≤1 days until ambient
Equilibrium for 30 for sealed-off
Provides areas.
goodfor
Very useful information
re- status.
Does
Does not
not indicate
allow forfire
early
1/3CO
R = 100 R-g 4.774O
–3/2 8
O ≤1 temperature
days
Equilibrium reached
until ambient
for 30 for sealed-off areas. status.
Litton’s s Ratio (R) - CH4 - >1 Smoldering combustion Provides
opening good
of information
sealed-off Doesdetection.
fire not indicate fire
1/3CO
g
Rg–3/2 8 O-1/22 temperature reached Very useful for re- Does not allow for early
C
R2gH=2100 s
- –3/2
4.774O - CH4 - days until ambient for sealed-off areas.
areas. status.
>1 or above ambient
Smoldering opening of sealed-off
Very useful for re- fire
Doesdetection.
not allow for early
1/3COs Rg 8 O-1/22
Rg2H
C =2100 - 4.774O2 - CH4 - temperaturecombustion
temperature reached areas.
opening of sealed-off fire
COs = CO in ppm >1 or above ambient
Smoldering combustion Very useful for re- Doesdetection.
not allow for early
C
R2gH=2100 - 4.774O2 - CH4 - (oxidation)
temperature
or above ambient areas.
opening of sealed-off fire detection.
CO = CO in ppm >1 Smoldering combustion
Litton
C2Hs 2 (1986) (oxidation)
temperature
or above ambient areas.
COs = CO in ppm
Litton (1986)
Desorbed (oxidation)
For bituminoustemperature
COs = CO Hydrocarbon
in ppm coal: Assesses fire status. Does not allow for early
Litton (1986)
Ratio (RI)
Desorbed Hydrocarbon For bituminous(oxidation)
coal: Assesses fire status.
Provides good results
fire
Doesdetection.
not allow for early
Litton(RI)(1986) 0-50 Normal
Desorbed
Ratio Hydrocarbon For bituminous coal: Assesses fire status. Doesdetection.
fire not allow for early
{1.01(CH4 + C2H2 + C2H4) - 0-50 Normal for
Provides goodareas
sealed-off resultsif Without hydrocarbons,
Ratio
Desorbed (RI) Hydrocarbon
CH4 / (CH4++CC2HH2++CC2HH4))- For 50-100
bituminous Possible
coal: source of Assesses
sampling fire status. in
performed
fire
Does detection.
value not allow
is zero andforfires
early
{1.01(CH 0-50 Normal
heating source of Provides
for good
sealed-off results
areas if Without hydrocarbons,
Ratio
+ 0.01} (RI) 4
1000
2 2 2 4
50-100 Possible sufficient numbers. fire detection.
remain undetected.
{1.01(CH
CH / (CH ++CC22HH22++CC22HH44))- 0-50 Normal
heating for sealed-off
sampling
Provides areas
performed
good resultsifin Without
value hydrocarbons,
is zero and fires
4 44
50-100
> 100 Possible
Hot zone source of
CH
+ / (CH
0.01}
{1.01(CH 1000 ++CC22HH22++CC22HH44))- sampling
sufficient
for performed
numbers. ifin value
remain isundetected.
zero and fires
Kim4 (1978, 44 1974)
50-100 heating source of
Possible Cansealed-off
identify areas
fire location Without hydrocarbons,
+
CH 0.01}
/ (CH 1000 + C2H2 + C2H4) > 100 Hot zone sufficient
sampling numbers. in
performed remain
value isundetected.
zero and fires
Kim 4
(1978, 4 heating if sampling
Can identifypoints
fire location
+ 0.01} 10001974) > 100 Hot zone sufficient
evenly numbers.
distributed. remain undetected.
Kim (1978, 1974) > 100 Hot zone Can
if identify
sampling fire location
points
Kim (1978,
Oxides 1974) Ratio
of Carbon <2 Normal if
Cansampling
evenly
Allows points
distributed.
identify fire location
for early fire CO2 from extraneous
evenly
if distributed.
sampling
detection points
Oxides
CO / CO2 of Carbon Ratio <2
2-13 Normal
Superficial heating Allows forif early
fire increases
fire sources
CO2 from and its solubility
extraneous
Oxides of Carbon Ratio evenly
Allowsdistributed.
continuously.
forif early
detection fire
fire increases in
COwater
sourcesfrommay
and itslead
extraneousto
solubility
CO / CO2 <2 Normal
2-13 Superficial
Fire ablaze heating
2
misinterpretations.
Mitchell
Oxides of(1989) Carbon Ratio ><132 detection
Allows if early
continuously. fire increases sources
in
COwater and
may its solubility
lead to
CO / CO2 2-13 Normal
Superficial heating Assessesforfire fire
status. 2
from extraneous
Mitchell (1989) > 13 Fire ablaze continuously.
detection if fire increases in waterand
sources mayitslead
misinterpretations. to
solubility
CO / CO2 2-13 Superficial heating Assesses fire status. misinterpretations.
Mitchell (1989) > 13 Fire ablaze continuously. in water may lead to
H2 / CH4 Ratio increasesFire withablaze
increasing Assesses fire status.
Provides good Does not allow for early
Mitchell (1989) > 13 misinterpretations.
H / CH4 (1990) temperature;
Ratio increases also depending
with increasing on Assesses
informationfireifstatus.
Provides good fire fire
Doesdetection.
not allow for early
Mitchell
2
magnitude and extent of fire. on in advanced stage
H2 / CH4 temperature;
Ratio increases also depending
with increasing Provides
informationgoodif fire or Doesdetection.
Mitchell (1990) magnitude
temperature; and extent
also of fire. on
depending ablaze.
information if fire or fire
H2 / CH4 Ratio increases with increasing in advanced
Provides good stage Doesdetection.
not allow for early
Mitchell (1990) magnitude and extent of fire. in advanced stage
C2H2 / C2H4 temperature;
1:1 also
Early stages ofonheat-
depending ablaze.
information
Provides goodif fire or fire
Doesdetection.
not allow for early
Mitchell (1990) magnitude and ing; extent of fire. ablaze.
in advanced if stage or
C H / C2H4/ ethylene) 1:1 temp.
Early stages > of
50 heat-
°C information
Provides good fire fire
Doesdetection.
not allow for early
(acetylene
2 2
ablaze.
in advanced stage
C2H2 / C2H4 1:1 ing; temp.
Early stages> of
50 heat-
°C Provides
informationgoodif fire or Doesdetection.
(acetylene / ethylene) ablaze.
C2H2 / C2H4
(acetylene / ethylene) 11 :: 12 Advanced
ing; temp.
Early stages
stages> of °Cof
50 heat- information
in advanced
Provides goodif fire or
stage fire
Doesdetection.
not allow for early
1:2 heating;
ing; temp.
Advanced temp.
> 50 °Cof
stages in advanced if
ablaze.
information stage
fire or fire detection.
(acetylene / ethylene) 1:2 > 100 °Ctemp.
heating;
Advanced stages of ablaze.
in advanced stage or
1 : 2 heating;
Advanced °Ctemp.
stages ablaze.
1:3 > 100ablaze;
Fire temp.of
heating;
> 100 °Ctemp.
1:3 > 300ablaze;
Fire °C temp.
1:3 > 100 °C temp.
Fire
> 300ablaze;
°C
1:3 Fire
> 300ablaze;
°C temp.
CO2 / CO < 0.5 % Normal Allows for early fire CO2 from extraneous
> 300 °C detection
CO2 / CO < 0.5 % % Normal Allows forif early
fire increases
fire sources
CO2 from and its solubility
extraneous
CO2 / CO 0.5-1.5 Smoldering effect continuously.
Allows forif early
detection fire
fire increases in
COwater
from
sources mayitslead
extraneous
and to
solubility
< 0.5 % Normal 2
CO2 / CO 0.5-1.5 % Smoldering effect detection
Allows if early
continuously.
for fire increases
fire misinterpretations.
sources
in
CO water and
frommayits solubility
lead
extraneousto
< 0.5 %
1.5-10 Normalfire
Active Assesses fire status. 2
0.5-1.5 % Smoldering effect continuously.
detection if fire increases in waterand
sources mayitslead
misinterpretations. to
solubility
1.5-10
0.5-1.5
> 10 % % Active
Smoldering
Fire fire effect
ablaze Assesses fire status. misinterpretations.
continuously. in water may lead to
1.5-10 % Active fire Assesses fire status.
> 10 % Fire ablaze misinterpretations.
1.5-10 % Active fire Assesses fire status.
> 10 % Fire ablaze
RI / RE Ratio Increase
> 10 % in RI/RE
Firevalue
ablazesignifies that
Provides good results Does not indicate fire
RI / RE Ratio fire is ascending
Increase in RI/REtoward
in value sampling
fire localization
Provides if
signifies that
good results status.
Does not indicate fire
Relative Efficiency (RE) = location; converse applying,
sampling too. that
extensive.
RI / RE
(0.265N Ratio
- 3.83O2) /(RE) (CO =+ fire is ascending
Increase in RI/REtoward
in fire
value
Provides sampling
localization
signifies
good if
results status.
Does
Does not
not indicate
allow forfire
early
Relative2Efficiency location;
fire is converse
ascending applying,
sampling
toward
in too. that
extensive.
sampling
fire localization if status.
RI / )RE Ratio
CO Increase in RI/RE value signifies
Provides good results Doesdetection.
fire not indicate fire
(0.265N
Relative
2 - 3.83O2) /(RE)
2Efficiency
(CO =+ location; converse applying,
sampling too.
extensive.if Does not allow for early
fire is ascending toward sampling
in fire localization status.
CO ) 2Intensity
(0.265N
Relative - 3.83O2)(RI)
Efficiency /(RE)
(CO
= =+ fire
Doesdetection.
not allow for early
2 location; converse applying, too.
sampling extensive.
CO
{(1 )
(0.265N
2- 3.83O - 3.83O
) / N )
} / (CO
[CO /+ fire
Doesdetection.
not allow for early
Relative2Intensity
2 22 (RI) =
CO2) - O ]
0.265N fire detection.
{(1 - 3.83O
Relative 2) / N }(RI)
2 Intensity
2 2
/ [CO
= /
0.265N
{(1
Relative
Mitchell -(1984)
- 3.83O O22)] / N2}(RI)
2 Intensity / [CO
= /
0.265N
{(1 - 3.83O- O2)] / N } / [CO /
Mitchell (1984)
2 2 2
Data
0.265N source:
- O2] Litton (1984); Singh (1990); Timko & Derick (1993); Tripathi & Sen (1996)
Mitchell2 (1984)
Data
Mitchellsource:
(1984)Litton (1984); Singh (1990); Timko & Derick (1993); Tripathi & Sen (1996)
Data source: Litton (1984); Singh (1990); Timko & Derick (1993); Tripathi & Sen (1996)
Data source: Litton (1984); Singh (1990); Timko &48
Derick (1993); Tripathi & Sen (1996)
48
48
48
48
Borehole temperature measurements were the principal tool for detection of subsurface
coal fires until the 1960s. Their advantage lay in the fact that they could be performed in
close proximity to combustion centers. In the early 1960s, increasing availability of air-
and spaceborne thermal scanners made remote sensing a better tool for coal fire detection
and monitoring. In-depth studies using Landsat TM and/or airborne thermal data were
performed in various countries. These studies applied an average emissivity value (0.96)
to represent all landcover. Geographical information systems were used for storage and
analysis of the data generated.
Since the very beginning of coal fire research, fire models as well as digital computer and
information technologies have contributed a great deal to simulating, understanding, and
mitigating coal fires. Some of the works are listed in Table 3.
Table 3: Groundbreaking works on coal fires using IT and digitalization

Author/Scientist Year Technique/Device Function
Stefanove et al. 1984 Computer program Solve complex simulations of mine ventilation
(VENT-4) networks during open fires.
Deliac et al. 1985 PC fire module Calculate effects of mine fires in tunnels on
temperature, mass flow, and pressure variation.
Schmal 1987 TNO model Model one-dimensional convective and
diffusive transport, including chemisorption,
oxygen depletion in piles, heat generation, and
cooling and heating processes.
Dziurzynski 1988 Mathematical model Display behavior of ventilation systems for
et al. emergency planning.
Chang et al. 1990 MFIR computer Perform normal ventilation network planning
simulation program calculations and dynamic transient state
simulations of ventilation networks under
various conditions, including fires.
Zhu et al. 1991 Mathematical model Calculate air flow and self-heating of goafs in
early stages.
Denby & Ren 1992 Knowledge-based Control risk of spontaneous combustion in
expert system (EESH) underground coal mines.
Lea 1994 CFD model Demonstrate potential for simulating near-fire
flow.
Prakash et al. 1995 Remote sensing & GIS Estimate temperature at surface level above
techniques underground coal fires.
Smith et al. 1995 SPONCOM Assess spontaneous combustion risk of
underground mining operations.
Saghafi et al. 1995 Mathematical model Model air, water, and heat flow in porous
media; used for prediction of self-heating
behavior of longwall goafs.
49
(continued)
Author/Scientist Year Technique/Device Function
Panigrahi & 1996 VENTSYS software Analyze ventilation network and underground
Singh environment.
Vekerdy & 1999 Information system Used for monitoring and fire fighting planning.
Genderen (Coalman)
Fierro et al. 2001 Modified TNO model Predict spontaneous heating of stockpiled coal.
Akgun & 2001 2D unsteady model Predict hot spots in stockpiles.
Essenhigh
Sahay et al. 2003 Computer program Assess explosibility of air mixtures in post-
(EXPLO) campaign monitoring.
Humphreys 2005 Numerical model Assess potential for and detection of
spontaneous combustion in coal mines.
7 Monitoring
Gas monitoring is commonly performed with a combination of sensors and tele-monitoring

systems. Parameters such as temperature, humidity, and air quantity are monitored, too.
Monitoring points are placed strategically (in terms of total number and location) so that they
can adequately represent the gas mix in a mine. Samples should be collected in view of gas
layering and response time. Gas monitoring is used to indicate abnormal conditions rather
than active fires or heating. Continuous monitoring is generally performed outside of sealed-
off areas to assess mine body air. Sealed-off areas are monitored at intervals as needed; the gas
is collected from the sealed-off area via boreholes. Various instruments (Haldane apparatus,
Graham Lawrence apparatus, different electronic sensors) are used for analysis of mine gases
such as CO2, O2, CH4, H2, CO, and other hydrocarbons. The Graham Lawrence apparatus
measures CO; the Haldane apparatus O2, CO2, CH4, and H2. Electronic sensor-based equipment
is used for all gases. The Haldane and Graham Lawrence apparatuses are premised on
adsorption and combustion processes. Electronic-based sensors operate with catalytic oxidation,
electrochemical cells, and infrared technology (Banerjee 1985).
8 Prevention and Control
A number of methods and techniques have been developed throughout the world to
extinguish coal fires and prevent (re-)ignition. Though some of these methods are very
efficient in dealing with the problem, they only apply to specific geo-mining conditions.
Their performance rate also varies from case to case. The methods currently in use may be
summarized as follows:
50
• Trench cutting and filling;

• Hydro-pneumatic/fly ash stowing;
• Blanketing;
• Digging out;
• Sealing through isolation stopping;
• Dynamic pressure balancing;
• Arrangement of multi-zonal ventilation systems;
• Chamber ventilation;
• Application of sealant/surface coating material;
• Injection of cement/hydrogel in coal mass;
• Hard foam for isolation of fire areas;
• Inertization processes:
o Liquid/gaseous N2,
o High-pressure high-stability foam,
o CO2;
• Application of water mist;
• Application of fire-fighting chemicals.
Examples of successful implementation of the above methods in Indian coal mines are
given in Table 4.
Table 4: Successful examples of coal fire fighting in Indian coal mines

Name of Mine Year Working Method Developed & Applied by CMRI
Method for Fire Control
Sayal’D Colliery CCL 1992 Board & pillar Dynamic pressure balancing
Churcha East SECL 1992 Board & pillar Dynamic pressure balancing
Naba Kajora Colliery 1993 Board & pillar Dynamic pressure balancing
ECL
North Searsole 1993 Board & pillar Dynamic pressure balancing and injection
Colliery of hydrogel
Sijua Colliery TISCO 1994 Board & pillar Dynamic pressure balancing and injection
of nitrogen
Ningah Colliery ECL 1995 Board & pillar Dynamic pressure balancing
Hingiram Pur Colliery 1997 Board & pillar Dynamic pressure balancing
Kottadih Colliery ECL 1997 Longwall Dynamic pressure balancing and injection
of nitrogen
Jhingurdah NCL 1998 Opencast Sealant application
Kachhi Balihari 2000 Board & pillar Dynamic pressure balancing
Colliery BCCL
51
(continued)
Name of Mine Year Working Method Developed & Applied by CMRI
Method for Fire Control
Jhanjra Colliery ECL 2000 Longwall Chamber ventilation and
injection of high-pressure high-stability
nitrogen foam
VK-7 Incline (BG-1) 2001 Blasting gallery Sealant; pressure control;
Panel Mine SCCL injection of foam, N2, and CO2
Sudamdih Shaft Mine 2001 Horizon mining Dynamic pressure balancing and injection
of foam
GDK-10 Incline Mine 2002 Blasting gallery Sealant; pressure control; injection of
SCCL foam, N2, and CO2
Sendra Bansjora BCCL 2003 Board & pillar Dynamic pressure balancing and injection
of foam
Haripur ECL 2003 Board & pillar Application of CO2, detection of leakage
paths, repair of stoppings
Lodna Colliery BCCL 2004 Fire in shaft Dynamic pressure balancing, injection of
pillars foam, stowing and strengthening of pillars
Shatabdi BCCL 2004 Opencast Application of chemicals
West Bokaro TISCO 2005 Opencast Injection of hydrogel
9 Global Overview
Coal fires occur in almost every part of the world. Although the nature and magnitude of
the problem differs from country to country, they are of particular concern in Australia,
China, Germany, India, Indonesia, South Africa, and the U.S.A. A brief description of the
situation in some of these countries is given below.
9.1 India
India’s major coal fires are located in Jharia, Raniganj, Singrauli and Singareni Coalfields.
By 1997, about 160 mine fires had been detected. Out of these, more than 70 were reported
from Jharia Coalfield (JCF) alone. The main impact of these fires on a regional level are
sinkholes and land subsidence. The total area affected extends over 6,300 ha. Subsurface
fires and land subsidence threaten the township of Jharia (DLR 2005).
It is estimated that 75 % of India’s coal fires result from mining activities. Surface mining,
in particular, has been proven to be a cause of coal fires, mainly owing to the prolonged
exposure of coal to atmospheric oxygen in this method. The fires in JCF are due, for the
most part, to unprofessional mining and coal extraction in the past (Figure 2).
52
53

Figure 2: Coal fire zones in Jharia Coalfield (not to scale)
Based on barometric pressure, underground coal fire development in JCF may be

categorized into four stages:
Stage I: Cracks and fissures allow air to enter the ground and reach abandoned mine
workings or sealed-off areas. Breathing starts as a result of changing barometric pressure
and residual coal is supplied with sufficient oxygen to ignite. The process of spontaneous
combustion has begun (Figure 3).
Stage II: Once sufficient draught has been created by the spreading fire, areas of exclusive
in- and out-breathing develop. A continuous air circulation loop forms that enables the fire
to progress steadily (Figure 4).
Stage III: In multi-seam workings, caved goafs in upper seams may be supplied with
oxygen through fans installed for mine ventilation below (Figure 5). This also facilitates
spontaneous heating.
Stage IV: A mature fire creates its own draught and pathways for oxygen uptake and flue
gas discharge; it starts descending toward lower seams (Figure 6).
Figure 3: Jharia coal fire development Figure 4: Jharia coal fire development
- Stage I - Stage II
Figure 5: Jharia coal fire development Figure 6: Jharia coal fire development
- Stage III - Stage IV
54
Coal fire peculiarities as observed in JCF after 1980 may be summarized as follows
(Bhattacharjee & Tikader 2001):
• Fires are located either in opencast workings or in shallow-depth underground workings

of different seams.
• Most fires started three to four decades ago.
• Recent fires often resulted from fires descending from surface to underground workings
or from one part of a mine to another.
• In the majority of cases, fires occurred as a consequence of unplanned and/or
unprofessional mining.
• In the greatest number of cases, heating and/or fires were only detected and reported
when smoke or flames were already visible at surface level.
• Amalgamation of smaller mining units into large workings removed inter-mine barrier
and promoted fire propagation.
In April 2004, the Central Mining Research Institute (CMRI; now joined with the Central
Fuel Research Institute (CFRI) to the Central Institute of Mining and Fuel Research
(CIMFR)) was awarded by the Government of India through the Council of Scientific
and Industrial Research (CSIR), a mission mode network project worth INR 160 million
under the 10th Five Year Plan (2002-07). The project had been drawn up to develop
comprehensive technologies for disaster prevention and management in JCF. Apart from
CMRI, executive agency of the project, a host of other institutes were participating,
namely: the National Geophysical Research Institute (NGRI), the Structural Engineering
Research Institute (SERC), the Center for Disaster Mitigation and Management
(CDMM), the Chennai and Central Electrical and Electronics Research Institute
(CEERI), Bharat Coking Coal Limited (BCCL), and the Directorate General of Mines
Safety (DGMS). For experimental research on assessment, isolation, and extinction
of coal fires, three sites were chosen, each representing specific mine conditions. The
project is ongoing but is sure to provide improved, technologically advanced methods to
tackle the coal fire problem in JCF.
9.2 U.S.A.
Approximately 20 % of all underground coal mine fires in the U.S.A. were caused by
spontaneous combustion. This proportion is expected to increase with rising consumption
and utilization of lower rank coal, increasing depth of mines, and the steady spread of
longwall mining (Smith et. al. 1991).
55
In 1962, the U.S. Bureau of Mines (USBM) reported 223 coal fires (Slavecki 1964).
Waste heap and subsurface coal fires in Pennsylvania – at that time infamous for its large
number of active fires – were studied using an airborne ‘Reconofax’ thermal scanner by
Slavecki (1964), Knuth (1968), and Greene et al. (1969). The major fires then were the
Centralia Mine Fire (anthracite coal), the Red Ash Mine Fire, the Powderly Culm Bank
Mine Fire, and the Beever Brook Mine Fire. In Centralia, a subsurface coal fire had been
wreaking havoc since 1962. Although a series of measures were taken to stop the fire,
the town had to be evacuated at last for risk of land subsidence, air pollution, and water
contamination.
The Federal Office of Surface Mining (OSM) has developed a database of mine fires
(Abandoned Mine Land Inventory System, AMLIS), listing 150 entries up to the year
1999. Besides Pennsylvania, coal fires are also recorded for Kentucky, West Virginia (east
of the Appalachian Coalfield), Colorado (Rocky Mountains), Utah, Illinois, and Wyoming
(DLR 2005).
In Utah, most coal fires are the result of mining activities and explosions. An underground
coal fire in an old coal mine near Craig, Colorado is said to have been burning for over 50
years at a depth of 19.8 m. In Wyoming, natural coal fires are reported along coal outcrops
in the Powder River Basin (Heffern & Coates 2004). Spontaneous combustion is thought
to be the cause of some of the fires there; the majority, however, was sparked by wildfires.
9.3 Europe and Russia
In Britain’s mining regions, fires often developed in coal waste heaps. However, once
mining stopped, these fires were excavated and the coal waste heaps re-vegetated. In
Germany, seven coal waste heaps were reported to harbor small-scale fires. These fires
were assessed using airborne thermal scanners and controlled by injection of cement and/
or anhydrite grout. In the north-east of France, coal with a high potential for spontaneous
combustion is still being mined. Here, specialized ventilation and mining methods were
adopted to minimize oxygen exposure. Some fires have also been reported from Eastern
Europe (Czech Republic, Poland, Ukraine). The Ukrainian coal industry has deposited coal
oddments totaling 2,000 million t in over 2,100 waste heaps. 140 of them are assumed to
be burning. In 1998, 74 coal fires were reported from Russia, in Kuzbass, and the Pechora
and Donetsk Basins. Fires at the latter location show similarities with those in the Ukraine
(DLR 2005).
56
Studies from Romania proved that all spontaneous combustion there occurred in seams
predisposed to self-ignition. They contained high concentrations of volatile matter
and were strongly fractured due to tectonic disturbance. The majority of coal fires was
sparked by residual coal in goafs or partial mining of seams in thinning-out areas (Babut
& Goldan 1995).
9.4 South Africa
Africa’s largest coal deposits lie in its southern states (Botswana, Mozambique, South
Africa, Zambia, and Zimbabwe). There, huge piles of refused coal have accumulated over
years of intensified export activity – a virtual hotbed for coal fires. However, in the wake of
a study conducted in 1985, storage techniques were improved and the number and extent
of coal fires dropped significantly (DLR 2005).
9.5 Australia
Australia is home to one of the world’s few naturally burning coal seams. The seam
extends over a length of 6 km at 30 m underground in the Hunter Valley Coalfields, New
South Wales. It is moving southward at only 1 m yr-1, mainly due to the lack of oxygen in
this depth. The fire is estimated to be about 5,500 years old (Kingsley 2003).
9.6 China
China is the largest coal-producer and consumer in the world. Vast amounts of coal are
lost each year in rampant coal fires; they are reported from more than 750 localities.
Approximately 20 million t yr-1are burned directly and more than 100 million t yr-1 are
lost as mineable reserves due to mechanical and thermal destruction. Fires occur in an
area extending over 5,000 km from east to west and 750 km from north to south (DLR
2005).
9.7 Other regions
Coal fires are also found in the southern part of Sumatra, where coal is abundant at surface
level. Under these conditions, soil erosion or surface mining are sufficient to lay bare coal
seams and create a climate favorable to coal fires. In July/August 1997, the Indonesian
Forest Fire Prevention and Control Project (FFPCP) reported coal seam fires which had
been smoldering for four years (DLR 2005).
57
10 Status of Research and Development
10.1 India
A sustained R&D effort by CMRI has resulted in the development of a new foam for fire
quenching. This technique is relatively simple and its application in mines has already
been tested successfully (Singh & Dixit 2001).
A fire-protective coating has been developed by CMRI for prevention of spontaneous

combustion in benches of opencast/underground mines. It is based on a bitumen emulsion
with additives and can be broadcasted using a simple spraying system. CMRI, together
with the Indian Institute of Technology (IIT), has also developed a system to assess fire
status, quantify sealing efficiency, and administer injections to halt fire progress. For time-
and cost-efficiency in weed control as part of surface coal fire prevention, CMPDI devised
a new method of chemical spraying (Singh & Dixit 2001). Dynamic pressure balancing
is another technique to control underground coal fires and has been used frequently in
India. Spaceborne remote sensing and airborne thermal infrared surveys were deployed to
determine fire propagation in JCF. Geological modeling served to evaluate the impact of
fires on coal reserves (CMPDI 1995-96; Loveson & Dhar 1995).
A model for prediction of spontaneous combustion risk in mine panels was developed by
CMRI in 2000 (CMRI S&T Report 2000). It is based on a set of combustion parameters
grouped into three categories, viz. panel specifics, as well as environmental and coal seam
characteristics.
In 2001, a mine fire model gallery was opened at CMRI, allowing for simulation
studies of open fires in coal mines. It is meant to contribute to the continuous
improvement of fire and explosion suppression techniques. The gallery is equipped
with state-of-the-art instruments, including sensors for online monitoring of various
parameters such as gas concentration, temperature, pressure, velocity, heat flux,
and smoke density. Three trials with liquid N 2, high-pressure high-stability nitrogen
foam, and water mist have been completed successfully; their results are promising.
Water mist, which also helps to reduce backlash in full-fledged fires, was found to
be most suitable for dealing with open fires. However, this technology is still at an
experimental stage and needs to undergo further testing under varied conditions and
with different samples before release.
58
10.2 China
A “Mine Fire Rescue Command Decision Support System” developed in China is able
to detect mine fires and their locations underground, assess and display ventilation
mechanisms in the main fire phase, and forecast the effect of ventilation control
measures (such as air flow and air reversal regulation). It also displays pertinent
information for rescue missions, for example location of mine personnel underground,
distribution of communication facilities, fire fighting equipment, etc. (Li & Jiang 1997;
Li 1998).
10.3 United Kingdom
Current practice in British coal mines includes a wide array of devices for fire detection,
based on an equally wide array of combustion products. Sensors are mainly: types G,
AA, TL1 TL1A, Figaro TGS11; as well as standard electro-chemical types, which are
cross-sensitive to H2, CO, and CH4 (Hambleton 1996). Experiments have provided a
detailed picture of combustion products distribution downstream of a fire source, which is
necessary for developing and evaluating computational models and fluid dynamics (Zhou
1996).
10.4 Romania
New studies on self-heating and spontaneous combustion in Romanian coal mines have
widened the range of means and methods for fire fighting by use of anorganic inhibitors
(Ion et al. 1997). Among 14 different types, 5 % MnCl2 proved most efficient, with 88 to
95 % inhibiting effect on coal oxidation.
10.5 Japan
Ohga and Higuchi (1995) developed a new smell-sensor-based system for early detection
of spontaneous combustion of coal and other materials in underground mine environment.
The sensor component uses a synthetic bi-layer membranous structure closely resembling
the human lipid membrane. A piezoelectric crystal device quantifies the amount of odorants
absorbed by the membrane. This device is an oscillator, which vibrates at a rate of several
million times per second. When the membrane covering the oscillator absorbs odorants,
its weight increases, and the change in vibration frequency enables the identification of the
odorants absorbed.
59
10.6 Poland
A new sealing technology using the effect of light chemical foam plugs, particularly
krylamin foams, on CH4 explosion behavior has been developed in Poland (GIG 1995).
The plugs attenuate the pressure wave and break flame propagation.
10.7 Czech Republic
The most effective technique to prevent and/or suppress spontaneous combustion is

injecting mechanical nitrogen foam into a mined-out area. This is widely used in the Czech
Republic, which also produces the equipment necessary for this technique. Fire fighting
methods have been optimized recently by gas analyses, mathematical modeling, and
detection of leakage using tracer gases -Kr85, SF6 (Adamus et. al. 1995; Voracek 1997).
10.8 France
At Houilleres du Bassin de Lorraine (HBL), France, stoppings fitted with fast closing,
remote-controlled doors at the end of laneways serving areas of longwall faces with sub-
level recovery were developed and have proven very effective. They were used on many
occasions in the last ten years for closure of districts in which serious fire accidents had
occurred, including two gas fires with considerable release of toxic gases and smoke. The
system allowed workings to re-open and resume operation.
The stoppings system was supplemented by: (i) a reliable and effective warning system;
(ii) a proven self-rescuer; (iii) a network of pressurized shelters; (iv) a nitrogen distribution
system linked directly to a high-capacity production unit (12,000 m3 h-1), injection pumps,
and stocks of resins and other equipments; as well as (v) an advanced ventilation computer
system (Amartin & Josien 1997).
10.9 South Africa
One of the major risks in mechanized coal mining is the ignition of CH4 through frictional
heat or sparks generated when the cutter hits on quartz or pyrite. Philip (1996) identified
and introduced methods to reduce this risk in continuous mining in South African coal
mines.
60
11 Scope of Research
Although research has progressed considerably worldwide over the last decades, coal fires
remain a serious problem that, in part, still awaits full scientific exploration. R&D efforts
that have the potential to provide new insights into the subject are listed below.
• Development and standardization of methods for assessing weathering of coal and its
effect on long-standing coal pillars – for both highly fire-prone and poorly susceptible
coals.
• Reassessment of biotechnology to hinder self-oxidation (aerial) of pyrites in coal.
• Studies on computer-based smart monitoring systems for early detection/prediction of
fires/explosions in mines.
• Studies on removal of heat from strata of long-burning fires.
• Development of technologies for effective fan reversal for fire control.
• Use of remote sealing techniques for separation of old underground fire areas from the
surface.
• Application of new technologies such as remote sensing, thermal imaging, GIS, aerial
approaches, ground penetrating radar, etc. for fire detection and localization.
• Improvement of fire risk prediction models for coal barred behind fires.
• Development of surface mining methods with minimum coal face exposure, faster
progress rates, and less production of coal fines.
• Temperature and fire sensing devices, quick seals, effective extinguishers for large fires,
better sealants and protective coatings.
12 Conclusion
This paper, through a most complete depiction and review of all aspects related to the
phenomenon, went to prove the hazardous potential of spontaneous coal seam fires. Today,
coal fires have grown to proportions where they compel governments, researchers, and
industries to think in global terms and join forces for effective control. Substantial joint
efforts in policy-making and research are needed to mitigate this global disaster.
61
Acknowledgements
The authors are thankful to UNESCO Office Beijing and the organizers of the International
Conference on Coal Fire Research for their invitation to present to an international
audience of scientific expertise a keynote paper on such a “burning topic” as coal fires.
References
Adamus, A., Hojek, L., and Posta, V., 1995. A review of experiences on the use of nitrogen in Czech coal
mines. In Proceedings of the 7th US Mine Ventilation Symposium, University of Kentucky,
Lexington, Kentucky, U.S.A., 5-7 June 1995, 237-241.
Akgun, F. and Essenhigh, R.H., 2001. Self-ignition characteristics of coal stockpiles: Theoretical
prediction from a two-dimensional unsteady state model. Fuel 80: 409-415.
Amartin, J.P. and Josien, J.P., 1997. Preventive facilities and emergency operation in case of fire in CdF
coal mines. In Proceedings of the 27th International Conference of Safety in Mines, New
Delhi, India, 20-22 February 1997, 443-456.
Babut, G. and Goldan, T., 1995. Research on the assessment of spontaneous combustion causes in the
mines of Valea Jiului. In Proceedings of an International Scientific Conference on the present
state and future of Czech and world mining, Ostrava, Czech Republic, 12-16 September 1995,
71-76.
Banerjee, S.C., 1985. Spontaneous combustion of coal and mine fires. New Delhi, Oxford, Rotterdam:
IBH Publishing.
Bhattacharjee, R.M. and Tikader, M., 2001. Fires in Jharia Coalfield vis-à-vis legislative provisions.
In Proceedings on mine environment and ventilation. Indian School of Mines, Dhanbad,
Jharkhand, India.
Chang, X., Laage, L.W., and Greuer, R.E., 1990. A user’s manual for MFIR: A computer simulation
program for mine ventilation and fire modeling. Information Circular 9245. U.S. Bureau of
Mines (USBM), Washington, D.C., U.S.A.
CMPDI, 1995-1996. Annual Report. Coal Science and Technology, Central Mine Planning and Design
Institute Limited (CMPDI), Ranchi, Bihar, India.
CMPDI, 1994. Compendium of completed research projects Coal Science and Technology Programme.
Central Mine Planning and Design Institute Limited (CMPDI), Ranchi, Bihar, India.
CMRI, 2002. Study for early detection of the occurrences of spontaneous heating in the Blasting Gallery
method and to evaluate suitable measures for prevention and control of spontaneous heating
in thick coal seams. Science and Technology Report. Central Mining Research Institute
(CMRI), Dhanbad, Jharkhand, India.
Deliac, E.P., Chorosz, G., and Albrand, N.D., 1985. Development of ventilation software on personal
computers in France and application to the simulation of mine fires. In Proceeding of the
2nd US Mine Ventilation Symposium, University of Nevada, Reno, Nevada, U.S.A., 23-25
September 1985, 19-27.
62
Denby, B. and Ren, T.X., 1992. A knowledge-based decision support system for spontaneous combustion
control. The Mining Engineer 151(366): 253-258.
Dhar, B.B., 1996. Keynote address on status of mine fires – Trends and challenges. In Proceedings of the
Conference on Prevention and Control of Mine and Industrial Fires - Trends and Challenges,
Calcutta, India, 21-22 December 1996, 1-8.
DLR, 2005. Coal Fire Research – A Sino-German Initiative. World map of coal fires. German Aerospace
Center (DLR), Wessling, Germany. 10 November 2005 http://www.coalfire.caf.dlr.de/
projectareas/world_wide_distribution_en.html
Dziurzynski, W., Tracsz, J., and Trutwin, W., 1988. Simulation of mine fires. In Gillies, A.D.S. (Ed.),
Proceedings of the 4th International Mine Ventilation Congress, Brisbane, Australia, 3-6 July
1988, 357-363.
Fierro, V., Miranda, J.L., Romero, C., Andrés, J.M., Arriaga, A., and Schmal, D., 2001. Model predictions
and experimental results on self-heating prevention of stockpiled coals. Fuel 80: 125-134.
Ghosh, A.K. and Banerjee, B.D., 1967. Use of the carbon-hydrogen ratio as an index in the investigation
of explosions and underground fires. Journal of Mines, Metals, and Fuels, 15(11): 334 -340.
GIG, 1995. On hazards in coal mines. Yearly report. Central Mining Institute (GIG), Katowice, Poland.
Graham, J.I., 1920-21. The normal production of carbon monoxide in coal mines. Transaction of the
Institution of Mining Engineers 60: 222-234.
Graham, J.I., 1914-15. The absorption of oxygen by coal. Part V: The influence on temperature on rates
of absorption by different parts of the Barsley Bed. Transaction of the Institution of Mining
Engineers 48: 521-549.
Greene, G.W., Moxham, R.M., and Harvey, A.H., 1969. Aerial infrared surveys and borehole temperature
measurements of coal mine fires in Pennsylvania. In Proceedings of the 6th Symposium on
Remote Sensing of Environment, Institute of Science and Technology, University of Michigan,
Flint, Michigan, U.S.A.,13-16 October 1969, 517-525.
Guney, M., 1968. Certain factors affecting the oxidation and the spontaneous combustion of coal. Min.
Soc. Mag. 20: 71-90.
Hambleton, R., 1996. Placement of fire detection equipment in coal mines, incipient fire detection
experiments in an underground roadway. HSL Section paper IR/LFR/95/10. Health and Safety
Laboratory, Buxton, Derbyshire, United Kingdom.
Heffern, E.L. and Coates, D.A., 2004. Geologic history of natural coal bed fires, Powder River Basin,
U.S.A. International Journal of Coal Geology 59: 25-47.
Humphreys, D., 2005. The application of numerical modeling to the assessment of the potential for, and
the detection of, spontaneous combustion in coal mines. In Proceedings of the 8th International
Mine Ventilation Congress, Brisbane, Australia, 6-8 July 2005, 361-369.
Ion, T., Gilgor, C., Cioclea, D., and Jurca, L., 1997. New researches in diminishing self-heating/self-
combustion phenomenon in the Jiu Valley, Romania coal mines by use of inorganic inhibitors.
In Proceedings of the 27th International Conference of Safety in Mines, New Delhi, India,
20-22 February 1997, 555-558.
Jones, J.H. and Trickett, J.C., 1955. Some observations on the examination of gases resulting from
explosions in collieries. Transaction of the Institution of Mining Engineers 114(9):
768-791.
63
Kaymakci, E. and Didari, V., 2002. Technical note on relations between coal properties and spontaneous
combustion parameters. Turkish J. Eng. Env. Sci. 26: 59-64.
Kim, A.G., 1978. Experimental studies on the origin and accumulation of coal bed gas. Report of
Investigations 8317. U.S. Bureau of Mines (USBM), Washington, D.C., U.S.A.
Kim, A.G., 1974. Low temperature evolution of hydrocarbon gases from coal. Report of Investigations
7965. U.S. Bureau of Mines (USBM), Washington, D.C., U.S.A.
Kingsley, D., 2003. Underground coal fires – A looming catastrophe. ABC Science Online, Wednesday,
19 February 2003. 10 November 2005 http://www.abc.net.au/science/news/stories/s786127.html
Knuth, W.M., 1968. Using an airborne infrared imaging system to locate subsurface coal fires in culm
banks. Proceedings of the Pennsylvania Academy of Science 42: 152-156.
Kok, A., 1981. Kolenbroei. Report WSK/2780-55. KEMA, Arnhem, Netherlands.
Lea, C.J., 1994. Computational modeling of mine fire. The Mining Engineer 153(394): 17-21.
Li, X.D., 1998. Mine fire rescue command decision support system. PhD thesis, Central South University
of Technology, Shandong, China.
Li, X.D. and Jiang, G., 1997. Mine fire rescue command decision support system. In Proceedings of the
27th International Conference of Safety in Mines, New Delhi, India, 20-22 February 1997,
431-435.
Litton, C.D., 1986. Gas equilibrium in sealed coal mines. Report of Investigations 9031. U.S. Bureau of
Litton, C.D., 1984. Gas equilibrium in sealed coal mines. Report of Investigations 8474. U.S. Bureau of
Loveson, V.J. and Dhar, B.B., 1995. Satellite remote sensing survey over some environmentally critical
areas in coalfields of Damodar Basin, India. In Proceedings of the First World Mining
Environment Congress, New Delhi, India, 11-14 December 1995, 223-234.
Mitchell, D.W., 1990. Recommended detection, mine fires. Chicago: Maclean Hunter Publ. Co.
Mitchell, D.W., 1984. Understanding a fire – Case studies. In Proceedings of workshops on combating
mine fires, U.S. Bureau of Mines (USBM), Pittsburgh Research Center, Pittsburgh,
Pennsylvania, U.S.A., 20-21 September 1984, 14-19.
Morris, R., 1988. A new fire ratio for determining conditions in sealed areas. Mining Engineer 147(317):
369-375.
Ohga, K. and Higuchi, K., 1995. Early detection of mine fire in underground by using smell detectors. In
Proceedings of the 7th US Mine Ventilation Symposium, University of Kentucky, Lexington,
Kentucky, U.S.A., 5-7 June 1995, 231-236.
Panigrahi, D.C. and Singh, M.K., 1996. Ventsys – A software for the design of underground mine
environmental systems. The Indian Mining and Engineering Journal 35(7): 19-29.
Philip, H.R., 1996. Identify methods to reduce the risk of explosions and fire caused by frictional ignition
hazards. Final report of SIMRAC Project, Col 226.
Prakash, A., Sastry, R.G.S., Gupta, R. P., and Saraf, A.K., 1995. Estimating the depth of buried hot
features from thermal IR remote sensing data: A conceptual approach. International Journal
of Remote Sensing 16(13): 2503-2510.
64
Rosema, A., Guan, H.Y., Veld, H., Vekerdy, Z., Ten Katen, A.M., and Prakash, A., 1999. Manual of coal
fire detection and monitoring. Netherlands Institute of Applied Geosciences (TNO), Utrecht,
Netherlands.
Saghafi, A., Bainbridge, N.W., and Carras, J.N., 1995. Modeling of spontaneous heating in a longwall
goaf. In Proceedings of the 7th US Mine Ventilation Symposium, University of Kentucky,
Lexington, Kentucky, U.S.A., 5-7 June 1995, 167-177.
Sahay, N., Verma, N.K., Ray, S.K., Narayanan, A., Singh, V.K., and Misra, D.D., 2003. Computer
application for assessment of intensity of fire for providing safety in underground mines. In
Proceedings of National Seminar on RIT-2003, Central Mining Research Institute (CMRI),
Dhanbad, Jharkhand, India, 1-3 February 2003, 287-293.
Schmal, D., 1989. Spontaneous heating of stored coal. In Nelson, C.R. (Ed.), Chemistry of coal
weathering. Amsterdam: Elsevier, 133-215.
Schmal, D., 1987a. A model for the spontaneous heating of stored coal. PhD thesis, Delft University of
Technology, Delft, Netherlands.
Schmal, D., 1987b. Verification of the TNO model for the spontaneous heating of stored coal. MT-TNO
Report R 87/46. Netherlands Institute of Applied Geosciences (TNO), Utrecht, Netherlands.
Singh, B., 1990. Assessment of status and control of underground coal mine fire. Report. Central Mining
Research Institute (CMRI), Dhanbad, Jharkhand, India.
Singh, V.K. and Dixit, J.P., 2001. Recent trends and developments in mine fire prevention with special
reference to Indian coal mines. The Indian Mining and Engineering Journal 40(12): 31-34.
Slavecki, R.J., 1964. Detection and location of subsurface coal fires. In Proceedings of the
3rd International Symposium on Remote Sensing of Environment, University of Michigan, Ann
Arbor, Michigan, U.S.A., 14-16 October 1964, 537-547.
Smith, A.C., Rumancik, W.P., and Lazzara, C.P., 1995. SPONCOM – A computer program for the
prediction of the spontaneous combustion potential of an underground coal mine. In
Proceedings of the 26th International Conference of Safety in Mines Research Institutes,
Katowice, Poland, 4-8 September 1995. Vol. IV, 39-51. (http://www.cdc.gov/niosh/mining/
pubs)
Smith, A.C., Miron, Y., and C.P. Lazzara, 1991. Large-scale studies of spontaneous combustion of coal.
Report of Investigations 9346. U.S. Bureau of Mines (USBM), Washington, D.C., U.S.A.
Stefanov, T.P., Arsenyan, E.E., and Vlasseva, E.D., 1984. Unsteady-state process during an open
fire in a ventilation network. In Hoves, M.J. and Jones, M.J. (Eds.), Proceedings of the
3rd International Mine Ventilation Congress, Harrogate, England, 13-19 June 1984, 417-420.
Storrow, J.T. and Graham, J.I., 1925. The application of gas analysis to the detection of gob fires.
Transaction of the Institution of Mining Engineers 68: 408-429.
Timko, R.J. and Derick, R.L., 1993. Applying atmospheric status equations to data collected from a
sealed mine, post-fire atmosphere. Report of Investigations 9362. U.S. Bureau of Mines
(USBM), Washington, D.C., U.S.A.
Tripathi, D.D. and Sen, S.K., 1996. Prediction of fire status – A new approach to an old problem. Journal
of Coal International 244(3): 118-120.
Vekerdy, Z. and van Genderen, J.L., 1999. Coalman – Information system for the monitoring of
subsurface coal fires and management of fire fighting in coal mining areas. In Proceedings of
65
the Geoinformatics: Beyond 2000 Conference, Indian Institute of Remote Sensing, Dehradun,
India, 9-11 March 1999, 179-184.
Voracek, V, 1997. Current planning procedures and mine practice in the field of prevention and
suppression of spontaneous combustion in deep coal mines of the Czech part of upper Silosia
Coal Basin. In Proceedings of the 27th International Conference of Safety in Mines, New
Delhi, India, 20-22 February 1997, 437-441.
Wilke, K., 1966. Die Selbstentzündung der Kohle und ihre Einflußfaktoren, die Feststellung der Neigung
der Kohle zur Selbstentzündung und die meßtechnischen Verfahren der Früherkennung. PhD
thesis, Aachen University of Technology, Aachen, Germany.
Willet, H.L., 1951-52. The interpretation of CO samples from behind stoppings with a view to reopening.
Transaction of the Institution of Mining Engineers 3: 629-651.
Zhou, X.X., 1996. Placement of fire detection equipment in coal mines: Computational fluid dynamics
simulations of fires and CO dispersion. HSL Section paper, IR/L/FR/96/4, May. Health and
Saftey Laboratory (HSL), Derbyshire, United Kingdom.
Zhu, M., Xu, Y., and Wu, J., 1991. Computer simulation of spontaneous combustion in goaf. In
Proceedings of the 5th US Mine Ventilation Symposium, Morgantown, West Virginia, U.S.A.,
3-5 June 1991, 88-93.
Zutshi, A., Ray, S.K., and Bhowmick, B.C., 2001. Indian coal vis-à-vis spontaneous heating problems.
Journal of Mines, Metals, and Fuels 44(5): 123-128.
66
The Chinese Perspective
Guan Haiyan
Beijing Remote Sensing Corporation (BRSC), Shenhua Group, Beijing, P.R. China
Abstract
China is the world’s largest producer of coal and the country most afflicted by spontaneous coal seam
fires. Recent studies estimate that an area of 720 km2 is affected, with 62 coalfields and/or coal mines
burning in a broad strip extending from west to east across northern China, the so-called North China
Coal Fire Belt. Coal fires seriously threaten mine safety, destroy valuable non-renewable resources, and
pollute the environment and atmosphere. Through large-scale research efforts over the last half-century,
Chinese scientists were able to delineate China’s coal fires, determine combustion centers and systems,
develop a series of combustion stage models, and describe in great detail the physical, chemical, and
thermal processes related to the phenomenon. Research specifically addressed coal fire transmission
zones, as well as combustion kinetic and thermodynamic systems. The results of this pioneer work form
the core of today’s knowledge on spontaneous coal seam fires in China. With the 4-Level High-Resolution
Detection and Monitoring System for coal fire management, integrating space- and airborne, as well as
ground and underground investigations, and the 3D Detection and Visualization System for underground
coal fires, China has created a comprehensive tool kit for coal fire detection and monitoring, which has
proven effective in fire fighting and ecological restoration in the north-western parts of the country. For
the harmonious development of coal mining and the environment in China, especially in its remote west,
the author puts forward a number of measures designed to broaden the application of remote sensing,
refine and refocus fire fighting activities, and improve mining and resource management in the Chinese
coal fire areas.
摘要
中国煤炭生产消费量居世界第一，也是世界上煤炭自燃灾害最严重的国家。据调查煤炭自燃面
积已达720平方公里，正在燃烧的煤田和矿区有62处，形成中国北方煤火燃烧带，危害煤矿生产
安全，破坏资源环境，污染地球大气层。经过半个世纪大范围的研究，确定了煤田火区分类、
燃烧中心、燃烧系统、燃烧地质模型和燃烧的理化热作用；进行了自燃导火带、燃烧动力系统
和热动力系统研究，形成了中国煤火的核心知识。通过航天、航空、地面、地下四层空间同步
遥感探测煤火试验，地下煤火三维可视化系统的开发，形成了中国煤火遥感探测、监测技术体
67
系。在西部灭火与环境治理中形成了一套有效方法。在西部实施煤炭资源开发与生态环境保护
协调发展中，提出了遥感监测、分批灭火治理、改进采煤工艺和煤田火区资源开采利用的有效
对策。
1 Introduction
The gross production and consumption of coal in China exceeded 1.9 billion t in 2006,
ranking first in the world. It is estimated that coal will remain China’s main energy source
until 2050. However, along with the steady increase in production, hazards of spontaneous
coal combustion have also multiplied, making China the world’s most prominent stage for
coal fires.
Coal fires are an open system determined by the constant flux of mass and energy. This
state of change brings forth spectral, thermal, magnetic, and electromagnetic anomalies,
which can be used for coal fire detection and monitoring in remote sensing. Coal fires,
although detrimental and dangerous where they occur, may well be used as a source
of energy – the principles and processes of spontaneous combustion are comparable to
those of in-situ coal gasification – and gasification can be controlled and exploited on an
industrial scale (Xu 2003).
This paper aims to give a broad picture of coal fire exploration, extinction, and monitoring
in China – it presents basic insights of Chinese research, main results of recent field and
laboratory investigations, key technologies such as remote sensing and other methods in
fire fighting and ecological restoration of fire areas, as well as efforts aimed at harmonizing
coal exploitation and environmental protection. International cooperation on coal fire
research will also be discussed.
2 Coal Fires in China
2.1 Classification and current distribution
Investigations in the 1990s led to a threefold classification of Chinese coal fires into:
(i) coalfield fires, (ii) coal mine fires, and (iii) storage/transport fires. Coalfield fires fall
within the responsibility of the national government. Coal mine fires and storage-transport
fires ought to be controlled by the corporate entity owning the mine or the coal that is
being stored/transported.
68
1) In coalfield fires, coal underground ignites spontaneously in a heat-generating reaction

of atmosphere and lithosphere, when oxygen enters the ground and reacts with the
chemical elements contained in coal, mostly carbon. Smoke which is emitted through
cracks and vents at the surface signals this naturally occurring combustion. The main
regions in China affected by coalfield fires are: Xinjiang, Gansu, Qinghai, Ningxia,
Shaanxi, Shanxi, Inner Mongolia, Sichuan, Chongqing, and Fujian. A center of fire
activity exists in the desert zone of northern China, where coal fires spread over an area
of 5,000 km from east to west, and 150 to 350 km from north to south, totaling about
720 km2. 62 fire zones with a burning surface of altogether 17 to 20 km2 have been
identified there. Xinjiang Autonomous Region alone reports 40 fire zones extending
over 10 to 13 km2. 10 fire zones are reported from Inner Mongolia (5 km2), three from
Gansu (0.5 km2), two from Qinghai and Shaanxi (both 0.5 km2), and one from Ningxia
(2.21 km2) and Shanxi (0.012 to 0.5 km2). Fire zones in Sichuan, Chongqing, and Fujian
(one each) are seasonal phenomena only.
2) Coal mine fires are coal fires occurring in underground or opencast coal mines; fires in
mine dumps and waste heaps also belong to this category. Coal mine fires are distributed
over 11 provinces in northern China, affecting 48 collieries; ten are burning in Xinjiang;
six in Shaanxi, Shanxi, and Liaoning; five in Jilin; four in Heilongjiang; three in Ningxia
and Inner Mongolia; two in Hebei and Gansu; and one in Qinghai. Coal mine fires are
reported occasionally from Henan, Hebei, Shandong, Anhui, and Jiangsu.
3) In storage/transport fires, coal self-ignites during storage and transport. Fires of this
kind are burning in the Datong Mine Coal Storage Field, and in Qin Huangdao and
Huanghua Port.
2.2 Hazard potential
Coal fires are serious hazards in at least four respects:
1) Coal fires obstruct coal mining. Methane and coal dust explosions, roof collapse, and
shaft flooding in the wake of coal fires threaten the lives of miners, destroy mining
facilities, and may even force a mine to close down (Tan 2000).
2) Coal fires destroy coal resources. Occurring for the most part in highly combustible,
shallow-depth low-ash coal seams of great thickness, coal fires eat up the most valuable
(and easy-to-exploit) of coal deposits. According to a survey commissioned by the
69
Chinese government in the years from 1990 to 1993 and estimates by Guan et al.
(1998), as much as 20 million t of coal are burned away in Chinese coal fires each year;
and with them another 200 million t are lost for mining due to the fires’ after-effects
(subsidence, mechanical obstruction, heat-induced changes in coal contents).
3) Coal fires pollute the environment. The enormous heat produced in coal combustion
bakes the overlying bedrock, and damages or kills surface vegetation. Direct heating,
noxious gases, and large-scale subsidence combine to expel fauna and flora from coal
fire areas. The pollutants emitted contaminate surface water and shallow groundwater.
Site sample analysis at different locations in China found that the concentration of
harmful substances around coal fires exceeded the applicable standards by 3 to 7 times
(Guan & Zhang 1999).
4) Coal fires contribute to atmospheric pollution. Measurements for the North China Coal
Fire Belt yielded a staggering amount of gas emissions per year: 490,200 t of carbon
monoxide, 514,700 t of sulfur dioxide, 300,000 t of nitrogen dioxide, 112,000 t of dust,
as well as other harmful gases such as large quantities of the greenhouse gas carbon
dioxide, and hydrogen sulfide. Along with these emissions, thermal heat in the order of
3.09 × 1017 J is given off to the atmosphere (NBS 2004; Guan et al. 1998).
3 Coal Fire Research in China
In the last 50 years, China has carried out field investigations over an area of
5 million km2, north of 35° N latitude, from the Pamir to the western slopes of the Great
Xingan Mountains. Together with data from coal fire areas in Yaojie of Gansu Province,
Liuhuanggou of Xinjiang Autonomous Region, Rujigou of Ningxia Autonomous Region,
and Wuda of Inner Mongolia Autonomous Region, the results of these investigations form
the core of Chinese knowledge on spontaneous coal seam fires today.
3.1 Cause and place of combustion
Based on direct cause and place of combustion, coal fires can be divided into three
different classes: (i) coal fires in coal seam outcrops, (ii) coal fires in small-scale coal
mines, and (iii) coal fires in residual coal in mined-out workings of large-scale coal
mines.
70
1) Coal fires in coal seams outcrops: Found mostly in the parched plains of northern China,
these natural fires are visible through smoke rising from coal protruding at an outcrop.
Outcrop fires have a long history in China, occurring as early as the Paleocene (e.g. in
Shenfu, Shanbei), the early Quaternary (Yaojie, Gansu), and the Quaternary Glaciations
(Liuhuanggou, Xinjiang). Today, fires of this category occur for example in Da Hinggan
Ling, Inner Mongolia.
2) Coal fires in small-scale coal mines: In contrast to outcrop fires, coal fires in small-
scale mines are principally anthropogenic. Occurring mostly in the surface-near parts
of abandoned coal mines, they sometimes set whole tunnel systems on fire. Smoke
as a visual marker is released from underground combustion centers. Small-scale
mine fires were first described in historical records from the Song dynasty (more than
1,000 years ago), and kept recurring in Ming and Qing annals (500 to 200 years ago).
Their modern heyday is marked by two important dates – 1958, when private coal
pits mushroomed during the “Great Leap Forward”; and 1984, when these pits were
officially prohibited.
3) Coal fires in residual coal in mined-out workings of large-scale coal mines: In 2004,
Quickbird satellite data helped discover a coal fire in residual coal in a mined-out shaft
of Wuda Coalmine, Inner Mongolia. Oxygen had reached the coal through collapse
cracks and mining-induced fissures. Smoke was being emitted through the same
pathways. The affected shaft was built in 1958; it had been burning since 1976.
3.2 Coal fire development
Coal fire development can be described in four successive stages: (i) spontaneous
combustion, (ii) formation of a combustion center, (iii) formation of a combustion system,
and (iv) temperature drop and extinction.
1) Spontaneous combustion: Coal spontaneous combustion has been a focus

of scientific research for more than 200 years, and there exists a plethora of
models today (> 100). Based on long-term large-scale multi-temporal and multi-
circumstance field research as well as analytical laboratory testing, it has been
established beyond doubt that oxygen adsorption on coal surface in the absence
of cooling gas motion is the major impetus for spontaneous combustion. The
availability of oxygen, in fact, controls the whole process from oxidation to heat
accumulation to spontaneous combustion. This has direct consequences on the
71
self-ignition behavior of coal: in the lignite and anthracite coal seams of northern
China, temperatures of up to 400 ºC and extensive reaction times are required for
spontaneous combustion. Outcrops, on the other hand, may self-ignite at 100 ºC
and below. At the western slopes of the Great Xingan Mountains, a few hours of
exposure to free air in good weather suffice to start a fire. The same process would
take years in an abandoned shaft of Wuda Coal Mine.
2) Formation of a combustion center: When a coal seam ignites, a pressure difference

builds up underground due to the difference in temperature and density of air within
the burning coal seam, its surrounding bedrock, and the surface. As a result, cold
ambient air is drawn into the ground, while hot flue gases push to leave the ground
through vents and cracks. These two counter-directional flows eventually give way to
an aerodynamic system of gas circulation. Together with the burning seam, they form
a combustion kinetic system, in which gas is continuously sucked to and released from
the combustion center. Parts of the coal seam may be consumed in low-temperature
pyrolysis. Combustion centers can range from several to hundreds of cubic meters in
size.
3) Formation of a combustion system: After a first combustion center has formed, the
conditions which originally led to spontaneous combustion may change as temperatures
continue to rise and more cracks develop. Multiple combustion centers may form around
the first center. These separate centers are connected through aeration channels and
may interact with each other in a way that further accelerates combustion. Parts of the
burning coal at this stage undergo a process of gelatinization, carbonization (coking),
and graphitization. Combustion systems are the basic unit for coal fire fighting and
monitoring. A combustion system can measure hundreds to thousands of cubic meters.
4) Temperature drop and extinction: The extreme temperatures in underground coal fires
tend to induce substantial structural changes in overlying rock strata. When the seam
has been exhausted and temperatures begin to fall, pyrometamorphism produces very
distinct minerals and compounds, namely: corundum (2,400 °C; Yue 1987); lava breccia
(1,600 to 1,200 °C); burned dandered coal (1,000 °C); vitrified and vitreous burned
shale (800 °C); burned sandstone (600 °C); baked loess (300 °C); as well as mirabilite
and sulfur crystallization druses (150 to 110 °C). During cooling, the rock visibly
degrades and eventually collapses, creating the large-scale collapse zones and burned
fissure clusters typical of coal fire areas.
72
3.3 Combustion stages and physical, chemical, and thermal processes
In what is a more detailed version of the 4-stage coal fire development model, coal
combustion can be described in 12 graded stages:
1) Weathering,
2) Oxidation,
3) Spontaneous combustion,
4) Formation of a combustion center,
5) Formation of a combustion system,
6) Complete combustion,
7) Gasification,
8) Continuous combustion,
9) Formation of metamorphic rock,
10) Precipitation of minerals,
11) Dormancy, and
12) Extinction.
The physical, chemical, and thermal processes governing coal fire development, as
well as the changes they effect in over- and underlying rock strata and at the surface,
were investigated in Wuda through spaceborne and airborne optical, thermal, and
electromagnetic surveys at different altitudes. These experiments where meant to
improve remote sensing parameters and techniques in coal fire detection and monitoring
(Guan et al. 1998).
3.4 Coal fire transmission zones and combustion kinetic and

thermodynamic systems
Research into modern mining methods and their impact on spontaneous combustion
has spawned a series of new coal fire models in China, describing the phenomenon
from different angles. Coal fire transmission zones, for example, specifically refer to the
geological and mining setting that a coal fire occurs in. The combustion kinetic system
focuses on combustion processes and the way they determine the extent and propagation
of a fire. The thermodynamic system, at last, puts temperatures and coal fire development
into context.
73
4 Coal Fire Detection and Monitoring Using Remote Sensing
From 1980 onwards, remote sensing tests combining space- and airborne, as well as
ground and underground investigations have been carried out continually in Taiyuan,
Shenmu, Changji, and Rujigou Coal Fire Areas. They served to confirm parameters and
bandwidths to be used in remote sensing-based coal fire detection and monitoring. The
information management system Coalman (Rosema 1999) and the 3D Detection and
Visualization System for underground coal fires (Zhang & Guan 2005; Zhang et al. 2008a
in this volume) were developed as a result of this research.
4.1 Investigation platforms
1) Spaceborne remote sensing works with data from a variety of radar and high-resolution
satellites such as NOAA, Landsat TM/ETM, SPOT, CBERS, IKONOS, and Quickbird,
as well as a number of Indian and Russian satellites.
2) Airborne remote sensing mostly employs black/white and color photogrammetry, as

well as multi-spectral photography. Single, double, and four-channel scanners are used
in thermal infrared scanning in medium-sized investigations. High-spectral remote
sensing (OMISI, 124 channels) falls into this category too.
3) Ground level remote sensing basically refers to thermal infrared temperature

measurements, imaging, and photography on site.
4) Underground exploration encompasses two different approaches depending on

location: for coalfields without mining operations, it mostly includes borehole surveys,
geophysical prospecting, and geochemical analysis. The “beam tube system” is put to
use in coal mines; it measures temperature and gases along tunnel systems (Song 2002).
4.2 Workflows
1) Regional monitoring in northern China covers entire coalfields and generally uses
remotely sensed data of lower resolution. Figure 1 shows the workflow of the North
China Coal Fire Information System, a typical example of regional monitoring (Guan et
al. 1998).
74
Figure 1: Workflow of North China Coal Fire Information System
2) Coalfield monitoring in northern China focuses on individual coalfields and uses mid-
to high-resolution remote sensing data for this purpose. Figure 2 shows the workflow of
Coalman, an information system used for coalfield exploration and monitoring (Guan
1999).
Figure 2: Workflow of Coalman system
3) Coal fire monitoring is applied directly in fire fighting operations. Exploring individual
fire sites, it uses high-resolution data, thus enabling a detailed view onto (and into the
mechanisms of) a single coal fire. Figure 3 shows the workflow of the 3D Detection
and Visualization System for underground coal fires (Zhang & Guan 2005; Zhang et al.
2008a in this volume):
75
Figure 3: Workflow of 3D Detection and Visualization System for underground coal fires
4.3 Key applications
1) For effective cola fire detection and monitoring, parameters and best-possible
combinations of bandwidths in remote sensing should be chosen in field trials according
to the 4-Level High-Resolution Detection and Monitoring System for coal fire
management (Zhang et al. 2008b in this volume).
2) Models such as the 4-stage coal fire development model, or those describing the
physical, chemical, and thermal processes that govern coal fire development can
provide a basic framework for interpretation of various detection data and thus facilitate
scientific understanding of coal fires. Combustion centers as the smallest unit of a coal
fire system (and thus of coal fire analysis) provide a powerful new concept in very
much the same way. High-resolution Quickbird data can be used to identify combustion
centers and monitor their evolution from ignition to extinction (described according to
the graded 4- or 12-stage coal fire development models).
3) Coal fire transmission zones as well as the combustion kinetic and thermodynamic
systems are new basic geological models for interpretation of coal fire spatial
development. Coal fire information extracted from remote sensing images and 3D
digital models using the 3D Detection and Visualization System for underground coal
fires is immediately relevant to coal fire fighting and prevention.
76
5 Coal Fire Fighting and Ecological Restoration in China
Over the last 30 years, coal fire fighting and ecological restoration of coal fire areas have
advanced considerably in China.
5.1 Coal fire fighting
Targeting the combustion kinetic system of a coal fire, fire fighting measures can aim either
at full extinction, cooling, or prevention of further expansion. The most common methods
applied in China are:
1) Drilling and water injection,

2) Excavation of fire source in shallow depth,
3) Drilling and grout injection,
4) Ground compaction and soil coverage,
5) Isolation walls (barriers) where mine production is at risk,
6) Inert gas (nitrogen) injection, or
7) Closing of small-scale mines where possible.
Most of the times, the above methods are used in combination: (1) and (2), for instance,
in case of shallow coal fires; (3) and (4) in case of deep underground coal fires. A
combination of (4), (5), and (6) is chosen when a coal fire is raging in close proximity to
an active mine and threatens ongoing mining activities.
According to the national fire fighting regulations issued by the Chinese Ministry of
Energy in 1995, a coal fire can be regarded as extinguished if temperatures on site remain
below 70 °C over a period of 6 to 12 months after extinction, and concentrations of carbon
monoxide and oxygen do not exceed 5 and 6 %, respectively.
5.2 Ecological restoration
Once a coal fire has been put out, a number of measures are taken to restore the fire
environment as close as possible to its pre-fire status – soil is refilled where necessary,
dams are built to store rain water for tree plantations, and wherever tree growth is
inhibited, grass is sown.
77
6 Coal Fire Policies in China
6.1 Historical background
The Chinese government is aware of the coal fire problem and has tried to provide a favorable
political framework for fire mitigation since the early days of the People’s Republic of
China. In the 1950s, Prime Minister Zhou Enlai made significant instructions for
coal fire management in Xinjiang Autonomous Region. Later ministers were to
follow his example. In the 1980s, the Chinese government launched a great Xinjiang
fire fighting project, which, in 1991, resulted in a set of fire fighting regulations
for the whole of China. Only three years later, the State Planning Committee (now
NDRC) listed coal fire monitoring and management as an item on “China’s Agenda
21” (ACCA21 1994).
Eight state-funded fire fighting projects in significant coal fire areas have been carried out
to date, including Hami South Lake, Ai Wei Gou, Tiechang Gou, Qitai North Mountain,
Fu Kang Baiyang River, Xiao Longkou, Xiao Huangshan and Liu Huanggou in Xinjiang.
Three fire fighting projects are ongoing in Ke Erjian, Xinjiang; Rujigou Coalfield, Ningxia;
and Wuda, Inner Mongolia.
6.2 How to harmonize coal mining and environmental protection?
1) Carry out truthfully “China’s Agenda 21” and set up a remote sensing-based coal fire
detection and monitoring system for northern China.
2) Define fire fighting criteria and manage coal fires by stages and in batches, so that all
fire fighting projects can be completed within 15 to 20 years.
3) Continue research on appropriate technologies for processing coal tar, coke, and gas
generated in the different stages of coal fire development as proven by the combustion
stage model and the models of coal fire physical, chemic, and thermal processes.
4) Innovate mining techniques, and ensure through project work that coal mining, coal fire
prevention, and environmental protection are developed harmoniously.
5) Strengthen government control and investment – about RMB 5 billion are needed for
fire fighting and RMB 1.5 billion for protection of coal mines against spontaneous fires.
78
7 International Cooperation
China has cooperated with a good many countries on different aspects and levels of coal
fire research. Main projects so far were:
• 1992-1997: “Environment Monitoring of Coal Fires in North China,” with EU support;

• 1994-1999: “Coal Fire Environmental Investigations in Rujigou Coalfield of Ningxia
Autonomous Region, P.R. China,” with the International Institute for Geo-Information
Science and Earth Observation (ITC), Enschede, Netherlands;
• 2002-2008, “Innovative Technologies for Exploration, Extinction, and Monitoring of
Coal Fires in North China,” with a number of German partners. (The Sino-German Coal
Fire Research Initiative was to present their findings at the International Conference on
Coal Fire Research and is thus the raison d’être of this book).
In the future, emphasis in international cooperative efforts will be placed on:
1) Setting up a coal fire monitoring system for northern China; and developing new
methods and technologies for coal fire fighting and ecological restoration of coal fire
areas in China.
2) Implementing the Kyoto Protocol (e.g. Clean Development Mechanism), raising capital
for fire fighting, and launching new coal fire and ecological restoration projects.
3) Establishing an international center for coal fire research or other international

institutions working on the topic.
Acknowledgements
We wish to acknowledge the valuable contributions to Chinese coal fire research made
by Zheng Lizhong of the Ministry of Science and Technology of the People’s Republic of
China (MoST); Ni Bin of the China National Administration of Coal Geology (CNACG);
Fu Jianhua of the State Administration of Coal Mine Safety; as well as John L. van
Genderen, Andries Rosema, and the German research team of the Sino-German Coal Fire
Research Initiative “Innovative Technologies for Exploration, Extinction, and Monitoring
of Coal Fires in North China.”
79
We are also much obliged to the Ministry of Science and Technology (MoST) and the
National Development Reform Committee (NDRC) of the People’s Republic of China
for the attention and support they have awarded coal fire monitoring and management in
China.
At last, we would like to thank the EU as well as the German and Dutch governments for
their long-term financial and technical assistance in Chinese coal fire research.
References
ACCA21, 1994. China’s Agenda 21 - White paper on China’s population, environment, and development
in the 21st century. Administrative Center for China’s Agenda 21 (ACCA21), Beijing, P.R.
China.
Cao, D.Y. et al., 2005. Analysis of internal factors for spontaneous combustion and coal fire types in Wuda
Coalfield, Inner Mongolia. Journal of China Coal Society 30(3): 288-292. (in Chinese)
Guan, H.Y. and Zhang, J.M., 1999. Project report, as a contribution to the Sino-Dutch Coal Fire
Research Initiative “Coal Fire Environmental Investigations in Rujigou Coalfield of Ningxia
Autonomous Region, P.R. China.” Beijing Remote Sensing Corporation (BRSC), Beijing, P.R.
China and Shenhua Group Corporation Limited, Beijing, P.R. China.
Guan, H.Y., van Genderen, J.L., Tan, Y.J., Kang, G.F., and Wang, Y.Q., 1998. Report on environmental
monitoring of spontaneous combustion in the north China coalfields. Beijing: Coal Industry
Publishing Company. (in Chinese)
NBS, 2004. Statistical communiqué on the 2004 national economic and social development of China.
National Bureau of Statistics of China (NBS), Beijing, P.R. China. (in Chinese)
Song, Y.J., 2002. Grading fire prevention in coal mines. Beijing: Coal Industry Publishing House. (in
Chinese)
Tan, Y.J., 2000. Disaster hazards and control of spontaneous combustion in coalfields, China. Coalfield
Geology and Prospecting 28(6): 8-10. (in Chinese)
Xu, Y.Q., 2003. Mining. Xuzhou, Jiangsu Province: China University of Mining and Technology Press.
(in Chinese)
Yue, C.S., 1987. Petrology. Beijing: Geological Publishing House. (in Chinese)
Zhang, J.M. and Guan, H.Y., 2005. Project report, as a contribution to the National Program No. 863
“Research on Key Techniques Using Remote Sensing for Detection of Underground Coal
Fires.” Shenhua Group Corporation Limited, Beijing, P.R. China and Beijing Remote Sensing
Corporation (BRSC), Beijing, P.R. China. (in Chinese)
Zhang, J.M., Huan, Z.D., Wang, M., and Xiang, Y.L., 2008a. 3D detection and visualization of
underground coal fires. In UNESCO Office Beijing (Ed.), Spontaneous coal seam fires:
Mitigating a global disaster. International Research for sustainable control and management.
ERSEC Ecological Book Series – 4. Beijing: Tsinghua University Press and Springer Verlag,
410-426. (this volume)
80
Zhang, J.M., Xiang, Y.L., Wang, M., and Huan, Z.D., 2008b. 4-Level high-resolution detection and
monitoring – Applications in coal fire detection and management. In UNESCO Office Beijing
(Ed.), Spontaneous coal seam fires: Mitigating a global disaster. International Research for
sustainable control and management. ERSEC Ecological Book Series – 4. Beijing: Tsinghua
University Press and Springer Verlag, 390-409. (this volume)
81
The Sino-German Coal Fire
Research Initiative
Stefan Voigt1 and Horst Rüter2
1
German Aerospace Center (DLR), Wessling, Germany
2
HarbourDom GmbH, Dortmund, Germany
Abstract
China is by far the biggest producer of coal; at the same time, it is the stage for one of the world’s most
ravenous coal-consuming phenomena – 10 to 20 million t of coal are burned away in uncontrolled coal seam
fires in China annually. What is more, approximately ten times this amount, about 100 to 200 million t, is lost
for mining as access for vehicles and machines is barred through the fires, or the rock-mechanical changes
they entail. Toxic and greenhouse gas emissions – CO2 and CO, as well as CH4, to name only the most
prominent – pollute the atmosphere and threaten flora and fauna in the fire areas. Extinguishing uncontrolled
coal seam fires is an extremely difficult, time-consuming, and costly enterprise. Even large-scale efforts often
fail since they lack a thorough scientific understanding of the processes leading up to and controlling coal
fire development. The Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China” was launched to deepen the knowledge and
understanding of these processes. Under its umbrella, experts from different technological and scientific
backgrounds conduct experimental studies; geological, geographical, and geophysical field campaigns; as
well as workshops, symposia, and training courses; with a view to fostering efficient structures and policies
for fire mitigation, extinction, and prevention in China and worldwide.
Coal fire fighting and prevention, in many countries, is still considered a minor task; it is relegated to
local mines and engineers, who at best work under central or federal regulations. Internationally concerted
approaches, standards, or policy frameworks do not exist.
The very slow and partly unsuccessful course of action that coal fire fighting has taken so far can be
explained by three significant shortcomings, all human in origin:
82
(i) Missing awareness of the ecological impact of coal fires,

(ii) Missing economic interest in coal deposits threatened by coal fires, and
(iii) Missing willingness to go to the great expense of extinguishing and preventing coal fires.
Sufficient funding, appropriate mining and fire fighting techniques and regulations, as well as national and
best international policies need to be put in place to tackle this triple dilemma. The Clean Development
Mechanism (CDM), as foreseen in the Kyoto Protocol, provides an interesting and powerful tool to
render coal fire extinction more profitable. As a link to international trading markets of Certified Emission
Reductions (CERs), it has the potential to attract international investors to the coal fire problem and, in
the long run, establish a win-win situation for the owners of affected coal deposits and those bearing the
expense of fire fighting.
摘要
目前，中国是一个最大的煤炭生产国，同时煤炭资源消耗现象也处在最严重阶段——每年由于
无法控制的煤火烧掉大约1 000~2 000万吨煤，大约烧掉资源的10倍，约1~2亿吨的煤炭资源，
由于火区周围巷道采掘机械和设备无法接近而不能开采。火区喷出的有毒气体和温室气体——
CH 4 、CO以及最主要的CO 2 气体污染了大气环境，威胁着宝贵不可再生的资源。挑战不可控
的煤火是一项极其艰巨、耗时和耗资的任务。即使大规模的投入也常因为对煤火产生和控制发
展的特殊过程缺乏全面的科学认识而失败。中德煤火研究项目“中国北方煤火探测、灭火和监
测新技术”计划就是要深化煤火发展过程的认识和理解。在该计划下，来自不同技术和科学领
域的专家进行了理论与试验研究，开展了地质、地理和地球物理方面的现场工作，以及学术交
流、研讨会和培训等工作，以期通过采用适当的组织形式和政策提高在中国和世界上火区减
灾、灭火和预防中的效率。
近来，煤火灭火和防治在许多国家仍被视为一项无关紧要的工作；据地方煤矿和工程技术人员了
解，充其量是按国家和联邦政府的要求开展工作；还没有国际上协调方法、标准和灭火体系。
自人类起源起，下列三个方面的不足阻碍了灭火工作，导致进程非常缓慢并且还未能成功实施：
(1) 缺乏煤火对生态影响作用的了解，
(2) 缺乏对煤火威胁的煤炭资源的经济上兴趣，
(3) 缺少投入大量实施防灭火工程的费用积极性。
充足的资金，合理的开采和灭火技术与规范，以及国家和良好的国际政策需要到位以解决这三个
问题。正如在京都协议书所见，清洁发展机制（CDM）为使实施灭火工程更具经济效益提供了令
人感兴趣和有效的途径。与认定减排量（CERs）的国际贸易市场相联系，CDM还会吸引国际投
资家们进入灭火领域，并且从长远看为受影响煤矿的所有者和承担灭火费的人建立了一种双赢的
机制。
1 Introduction
Uncontrolled coal seam fires are a serious problem worldwide. The phenomenon is
known all over the globe – from the U.S.A., the states of the former U.S.S.R., South
83
Africa, and Venezuela to India, the Indonesian archipelago, and China. Approximately
70 % of the People’s Republic’s energy demand is met by coal. China is by far the
biggest producer of coal in the world (c. 37 % of global production in 2003), with an
annual output of over 1,500 million t of raw coal. In fact, coal production in China
tripled over the last 30 years (IEA 2004). According to the 10th Five Year Plan for the
Chinese coal industry (2001 to 2005) by the State Economic and Trade Commission,
coal will remain China’s major energy source in the years to come, its impact even
increasing if the country’s economy continues to grow at its current pace of 7 to 8 %.
Despite this evident economic importance, 10 to 20 million t of coal are burned in
China each year in uncontrolled coal seam fires. This is equivalent to the annual coal
production of Germany. What is more, approximately ten times this amount, about
100 to 200 million t, is lost for mining as access for vehicles and machines is barred
through the fires, or the rock-mechanical changes they entail. Large, but generally not
precisely monitored toxic and greenhouse gas (GHG) emissions, such as carbon dioxide
(CO2), methane (CH4), and carbon monoxide (CO), are released from coal fires into
the atmosphere. Beyond their dire impact on mining and mine safety, coal fires pose a
serious threat to local, regional, and global environment; and to the health and safety of
people working and/or living in their vicinity.
Coal fire fighting is an extremely difficult, time-consuming, and costly enterprise. Even
large-scale efforts often fail since they lack a thorough scientific understanding of the
processes leading up to and controlling coal fire development (Walter 1999). The size
and remoteness of affected areas, as well as the many technical challenges in detecting,
analyzing, and extinguishing coal fires add significantly to the problem. The Sino-German
Coal Fire Research Initiative “Innovative Technologies for Exploration, Extinction, and
Monitoring of Coal Fires in North China” was launched in the fall of 2003 to deepen the
knowledge and understanding of the processes involved in coal spontaneous combustion.
By probing new methods of coal fire analysis and establishing a new sound scientific
basis for monitoring, control, and prevention of fires, the project tries to gear China’s coal
mining sector toward a more sustainable and environmentally friendly future.
2 A Better Understanding of Fire Processes
It is evident today that most coal seam fires occur in the context of mining operations – in
mines small or large, abandoned or active. In almost all cases, oxygen supply to (mined)
coal deposits is the precondition for and the cause of self-heating and combustion.
However, the situations and reaction chains that give rise to uncontrolled coal seam fires
84
(as well as those that keep them ablaze) are extremely varied and complex. In contrast
to the relatively simple concept of spontaneous combustion, understanding, controlling,
and preventing a fire in a given setting is a very difficult undertaking. Immense efforts
have been made all over the world to extinguish coal fires: millions of tons of clay were
moved and broadcast over fire areas to suffocate the fires beneath, water injections over
several years were meant to cool temperatures underground. In most cases, however,
the desired results failed to materialize. It became obvious that improved scientific
analysis and research work was required to tackle the problem successfully – a better
understanding of the complex processes that govern coal fire development for a better
technological response. It takes indeed considerable time and effort to determine the
size and exact location of an underground coal seam fire, let alone a whole coal fire
zone; to understand mine ventilation systems and the dynamics of gas emissions; or to
devise more efficient, more economical methods of fire extinction and prevention. Given
these challenges and the very limited knowledge we currently have on preconditions,
processes, geometry, and dynamics of coal fires, it was clear to everyone involved in the
planning process of the Sino-German Coal Fire Research Initiative that only a large and
interdisciplinary research team would be able to generate novel and enhanced scientific
knowledge on the subject.
Consequently, experts from a great number of disciplines and technical fields were
invited to participate in the project and are now determining its scope. They carry out
experimental studies; explore the terrain in geological, geographical, and geophysical field
campaigns; and impart their insights at workshops, symposia, and training courses. This
Sino-German research consortium consists of professionals and scientists from mining,
petrography, geology, geophysics, geography, remote sensing, material studies, and
resource management. Their research work focuses on the four major stages of the “coal
fire analysis and control cycle” (Figure 1):
1) Mapping, observing, and analyzing existing coal fires and coal fire risk areas by means
of geological, petrographical, and topographical investigations at different scales; as
well as mining-related and socio-economic features of coal fire areas.
2) Modeling and understanding the different chemical, physical, geological, fluid-dynamic,

and rock-mechanical processes and preconditions of coal fires and their dependence on
time at various spatial scales.
3) Devising innovative and improved extinction and prevention methods for mitigation of
coal fires.
85
4) Monitoring coal fire dynamics, extinction efforts, and coal fire risk areas with a view
to fire prevention, by using space- and airborne, as well as ground- and underground
measuring techniques.
Figure 1: 4M – Coal fire analysis and control cycle
All elements of the “4M cycle” are equally important in achieving a better understanding
of coal fire processes and preconditions, and developing and improving methods of coal
fire containment and mitigation.
The mapping stage comprises all analyses conducted in a coal fire area with respect to
geological, stratigraphical, and petrographical features. It also covers ground and borehole
geophysical fire localization and quantification, as well as aerial geophysical and satellite-
based analysis of coal fire-related rock alterations and thermal anomalies, coal seam
outcrops and mining activities (Voigt et al. 2004).
The next stage, modeling, tries to harmonize and refine the conceptual and theoretical
understanding of coal combustion on the basis of measurement and mapping results.
The three main approaches here are: (i) rock-mechanical modeling to derive stress
and crack field patterns resulting from a fire, and the stratigraphical conditions of its
setting; (ii) fluid-dynamic modeling of airflow, heat transport, and fire extinction through
fluid injection to study the temporal development of coal fires, convection cells, and
86
ventilation regimes; and (iii) physico-chemical modeling of coal combustion based on

finite difference models. All these approaches, alone or in combination, help to optimize
the understanding and accuracy of concepts of fire dynamics as they allow for validation
against in-situ observations and measurements. Probability-based modeling, recently,
has served to incorporate different types of coal fire risk indicators such as thermal
anomalies, cracks, gas emissions, and coal surface exposure into an integrated coal fire
risk map.
Mapping and modeling of coal fires is of use and benefit only if it contributes, in some way
or another, to more effective fire fighting and prevention. Hence, the third stage of the “coal
fire analysis and control cycle” – mitigation. It is at this stage that new and improved fire
fighting techniques and strategies (fuel blocking, disruption of oxygen supply, cooling) are
studied and enhanced. Fire zone rehabilitation is of relevance here too.
The fourth and final stage of the “4M cycle” stresses the temporal aspect of coal fire
assessment and prevention. Only those measures or methods that guarantee permanent
extinction and/or prevention can be considered effective and therefore useful. Every
technique has to be assessed for its suitability as a monitoring tool of coal fires and/or
coal fire risk areas (i.e. for its ability to yield reliable and reproducible mapping and
analysis results). The method to select eventually depends on the spatial scale for a
given time series of observations – it may range from local ground or underground
data by means of geophysical surveys, gas- or temperature logging, or thermographic
mapping; to regional or continental monitoring based on satellite data on thermal
or morphological anomalies, as well as fire spectral features and fire-related surface
characteristics.
The main task in all coal fire research and management is the coordinated handling,
synthesizing, and archiving of relevant data obtained during the four stages of the “coal
fire analysis and control cycle.” Modern information technology can be used to establish
coal fire data warehouses and information systems, for example by applying GIS, database
management tools, and web mapping. A data warehouse can function as a free-access
repository for engineers, scientists, and coal fire experts, providing them with precise and
bespoke data sets. A less advanced user may resort to more general information systems,
which allow an aggregated and processed view of relevant data through synthesized maps,
statistical charts, and area-focused reports.
87
3 Toward Improved Frameworks and Policies for Coal Fire

Mitigation, Extinction, and Prevention
In many countries, coal fire fighting and prevention is still considered a task in the
responsibility of local mines and engineers, guided at best by central or federal regulations.
Internationally coordinated approaches, standards, and extinction frameworks are
missing in most places. A common argument for this parochialism is that fire types and
situations depend on the local setting (geology, mine structure, etc.) and are hence site-
specific. However, a close cross-border look at coal fires reveals an astonishing similarity
in mapping, modeling, mitigating, and monitoring techniques and approaches. Where is
the fire source? What thermal anomalies do occur? Do old workings ventilate the fire?
Is surface sealing an option for fire extinction? How can the fire zone be cooled down?
– Evidently, no matter what kind of a fire and where, the same questions of localization,
surveillance, containment and after-care arise. It is thus all the more astonishing to see
what little international coordination on coal fires exists to date. Besides the technical
and scientific intricacies of fire fighting, funding is a major obstacle. Often, fires have
been burning for years before the funds necessary for extinction can be raised from local,
regional, or central government authorities and/or mining corporations. These periods
of inaction allow the fires to grow in size and intensity, and raise the costs for extinction
measures even more – a vicious circle found wherever coal fires occur – in China, India,
South Africa, and the U.S.A.!
The very slow and partly unsuccessful course of action that coal fire fighting has taken so
far can be explained by three significant shortcomings, all human in origin:
1) Missing awareness of the ecological impact of coal fires,

2) Missing economic interest in coal deposits threatened by coal fires, and
3) Missing willingness to go to the great expense of extinguishing and preventing coal fires.
Sufficient funding, appropriate mining and fire fighting techniques and regulations, as
well as national and best international policies need to be put in place to tackle this triple
dilemma.
With respect to funding, it is important that financial and technical support be given
rapidly and flexibly so that fires can be contained at an early stage. Both can only be
achieved if there exist clear and well-defined responsibilities and regulations, and an open
budget independent in time and space of any given local fire. Extinction efforts are often
eschewed because they are thought to be economically unviable. The Kyoto Protocol
88
and emission reduction treaties are a promising avenue in this respect. They would enable
developing countries to use and commercialize the large amounts of greenhouse gases
that coal fires release. As Certified Emission Reductions (CER) from Clean Development
Mechanism (CDM) projects for example, these gases could be traded on a number of
growing international markets and turn fire fighting into a very profitable business, able to
attract international investors (World Bank 2004). In the long run, a win-win situation would
result, both for the owners of affected coal deposits and for those bearing the expense of fire
fighting. First feasibility studies of CDM projects are being carried out in India and China.
However, two main problems – one technical, one political in nature – need to be solved
before coal fire related CERs can be generated under the provisions of the Kyoto Protocol:
1) A scientifically sound method for determining coal fire emission baselines has to be
devised. It must allow for proper quantification of emission reductions through fire
extinction in a CDM project. Currently, science teams (including the Sino-German Coal
Fire Research Initiative) are working on this issue.
2) The ratifying nations of the Kyoto Protocol have to acknowledge (at least for those fires
for which a CDM project has been proposed) that uncontrolled coal seam fires are a
man-made source of emissions and not a result of natural processes.
With respect to mining techniques/practices, it is clear from a good number of studies that
correct sealing of abandoned workings and laneways is the key to coal fire prevention. As
mentioned earlier, oxygen supply to residual coal in improperly mined coal seams is the
main cause of spontaneous combustion in coal mines. Although this by now should be a
well-known fact, practice in many places still deviates from theory. Good implementation
rests on strict mining regulations and their proper execution at all levels, industrial and
governmental. In places where oxygen supply can not be disrupted, the only way to
prevent fires is the complete removal of residual coal. This applies to underground mining
as well as to opencast mining. In complex geological settings where very thin or unusually
thick coal seams obstruct complete underground extraction, a shift toward opencast mining
seams advisable. Such a structural shift would have the additional advantage of tapping
new funding resources and allowing new perspectives in mine planning. It is important
to note at this point that all technical approaches toward coal fire prevention can in fact
be reduced to these two elementary principles: control of oxygen supply and/or complete
removal of fuel!
Funding and the technical aspects of coal fire fighting and prevention are obviously linked
through policy frameworks and regulations at local/regional, national, and international
89
levels. In countries with long-standing mining tradition – China, Germany, India, and the
U.S.A., for example – such regulations are already in place. Common standards and fire
fighting guidelines on an international level, however, are missing; the same applies to
global funding structures as foreseen, for example, in the Kyoto Protocol. If an international
framework or convention for spontaneous coal seam fire extinction and prevention could
be established, the global community would be given a powerful tool to mitigate the grave
effects of this phenomenon and save valuable non-renewable resources.
4 Conclusion and Outlook
Coal fire extinction and prevention is a very complex task and involves a whole range of
disciplines in its four stages of mapping, modeling, mitigation, and monitoring. Being of
great economic and environmental impact, it concerns individual enterprises as much as
whole industries, local and national governments as much as international organizations;
it even touches upon international agreements and treaties. Only if all parties make best
use of the existing scientific, technological, economic, and political tools and frameworks,
and collaborate in a synergistic fashion, can the coal fire problem be tackled in the long
run. Any shortcoming in either of these domains – science, technology, economy, and
politics – is bound to weaken the other domains’ capacity to mitigate coal fire impact. For
instance, an excellent fire mapping and monitoring system can only be applied successfully
if adequate financial support is given. The best fire extinction and prevention guidelines
remain on paper only, if the right policies to apply them are missing. Consequently, it is
very important to raise awareness for coal fires in all domains and support a rigid and
instantaneous implementation of fire fighting and prevention.
International conventions and policy frameworks that address the coal fire issue do not
exist today. They would however be a perfect vehicle to standardize fire fighting and
prevention worldwide and link the different local and/or national approaches to the
problem. With regard to funding, the Clean Development Mechanism (CDM) of the
Kyoto Protocol, through the generation of Certified Emission Reductions (CERs), has
the potential to make coal fire extinction and prevention a highly profitable undertaking.
Obstacles to such an international network for coal fire mitigation are luckily being
removed step by step by international capacity and knowledge building, by new scientific
and technological developments, and finally by a number of political milestones, such as
the recent ratification of the Kyoto Protocol.
90
Acknowledgements
The authors wish to thank all scientists and technicians of the Sino-German Coal Fire
Research Initiative for the very fruitful cooperation in this interdisciplinary international
research effort. The work was funded by the German Federal Ministry of Education and
Research (BMBF), the Ministry of Science and Technology of the People’s Republic of
China (MoST), Shenhua Group Corporation Limited, the Ministry of Land and Resources
of the People’s Republic of China (MLR), and several other entities – we sincerely thank
all of them for their support. We would also like to thank Andreas Hirner of DLR for
carefully reviewing this manuscript and discussing its content with us.
References
IEA, 2004: Key World Energy Statistics 2004. International Energy Agency (IEA), Paris, France. March
2006 http://www.iea.org
Voigt, S. et al., 2004. Integrating satellite remote sensing techniques for detection and analysis of
uncontrolled coal seam fires in north China. International Journal of Coal Geology 59:
121-136.
Walker, S., 1999. Uncontrolled fires in coal and coal wastes. International Energy Agency (IEA), London,
United Kingdom.
World Bank, 2004. Clean Development Mechanism in China - Taking a proactive and sustainable
approach. 2nd edition. The International Bank for Reconstruction and Development/The World
Bank, Washington, D.C., U.S.A.
91
Part III
A Global Phenomenon
Potential Environmental and Health
Impacts of Burning Coal in Witbank
Coalfield, South Africa
J. Denis N. Pone1, Kim A.A. Hein1, Glenn B. Stracher2,

Robert B. Finkelman3, and Harold J. Annegarn4
1
School of Geosciences, University of Witwatersrand, Johannesburg, South Africa
2
Division of Science and Mathematics, East Georgia College, Swainsboro, Georgia, U.S.A.
3
U.S. Geological Survey, Reston, Virginia, U.S.A.
4
Department of Geography and Environmental Management, University of Johannesburg,
Auckland Park, South Africa.
Abstract
Spontaneous combustion of coal in Witbank Coalfield, South Africa was studied in order to determine the
extent of toxic chemical elements released into the environment. Samples of gas emissions from vents
were analyzed to investigate the presence of these elements. Relatively high concentrations of benzene,
toluene, ethylbenzene, and xylene (BTEX) were found in the samples. Analogous concentrations of
carbon monoxide, carbon dioxide, and methane, in addition to a variety of hydrocarbons, were detected.
BTEX are the elements with major, potentially hazardous, impact on human health and the environment
around coal fire sites. Fires caused by spontaneous combustion of coal are pervasive in Witbank Coalfield.
Health problems because of exposure to hazardous elements are a serious risk. It would be prudent for
the South African government to conduct an environmental impact assessment of these elements in order
to determine the magnitude of their impact on mineworkers’ health and the environment in Witbank
Coalfield.
摘要
本文针对南非韦特班克煤田的煤火燃烧问题进行了研究，以便确定煤火释放到环境中的有害化学
元素。对由裂隙中释放的气体进行现场采样分析并研究了这些有害元素。样品中发现了较高含量
的苯、甲苯、乙苯和二甲苯（BTEX）。除大量碳氢化合物气体外，还检测到一氧化碳、二氧化
碳、甲烷气体。而BTEX是一种对火区周围的人群健康和环境具有潜在灾害影响的主要元素。韦
94
特班克煤田煤自燃引发的煤火是普遍发生的，由于有害元素的存在影响人们的健康已是一个严重
的问题。对南非政府来说进行环境影响评价研究要务须慎重，以便确定韦特班克煤田中有害元素
对矿工和环境的影响程度。
1 Introduction
Coal is South Africa’s primary source of energy; it has been a major stimulus for economic
growth and a significant factor in the country’s industrialization. South Africa is the fourth
largest coal producer in the world, with an annual production of about 300 million t (EIA
2005). The exploitable coal reserves in South Africa amount to 55 billion t (Falcon et
al. 2001), which is equivalent to almost 11 % of the world’s total coal resources. Coal
deposits are hosted by Permian age sediments of the Karoo Sequence, which forms a cover
sequence for a large portion of South Africa. The country’s nineteen coalfields extend over
an area of 700 km from north to south, and about 500 km form east to west (Figure 1).
Their coal is usually classified as bituminous and occurs in seams of up to several meters
in thickness.
Figure 1: Location of Witbank Coalfield, South Africa
The coal mining industry in South Africa runs the whole gamut of mining scales,
from small operations with production limited to a few thousand tons per annum, to
Secunda, the world’s largest underground coal mining operation, boasting an annual
production of c. 35 million t (Lloyd 2002). With power generation demands increasing
95
in South Africa and its neighboring countries, and an economy projected to grow by
3 to 5 % each year, it is likely that coal will continue to play an important role in the
Cape’s economic development. Unfortunately, coal, as the country’s primary fossil
fuel, is also heavily threatened by uncontrolled fires. Estimates about how much coal
is burned each year in coal fires go as high as 3 million t and beyond – a financial
loss of millions of dollars! Besides this economic rampage, the environmental and
human health impacts of coal fires, both on a local and regional scale, cannot be
underestimated. According to Lloyd (2002), c. 170 million t of carbon dioxide are
emitted from coal fires in South Africa annually. Methane and other volatile gases
from these fires are now considered a significant contributor to the global deterioration
of biosphere, and represent a threat to miners and communities in the fires’ vicinity
(Stracher 2003; Finkelman 2004).
This paper outlines the potential human health and environmental impact of toxic gas
emissions from coal fires in Witbank Coalfield, South Africa.
2 Study Site
Witbank Coalfield is situated on the irregularly eroded northern margin of the Karoo Basin,
about 130 km east of the city of Johannesburg. It is about 40 km in length and 180 km in
width, stretching from the towns of Belfast in the east to Springs in the west.
The Vryheid Formation (Ecca Group) of the Karoo Sequence, which is present throughout
the Witbank area, reaches a maximum thickness of about 140 m. The Ecca sediments
overlay a formation commonly known as the Dwyka tillite. Tectonically, the Karoo
sediments are relatively undisturbed and disruptions of the strata are rare. However,
fractures and dolerite dykes are common in compact rock such as sandstone and shale.
Stratigraphical studies of Witbank Coalfield have proven that five main coal seams exist in
the area; they are thought to have been deposited in glaciofluvial deltaic and coastal plain
settings (Cairncross et al. 1988; Cairncross 2001).
Witbank Coalfield is the focus of coal mining in South Africa and currently accounts for
about 53 % of the country’s coal production, including supplies for generation of 41 % of
South Africa’s electricity (Uludag 2001). Mining operations in the coalfield commenced in
the 1890s – during those days, mostly in the form of board and pillar mining characterized
by a low recovery ratio leaving considerable amounts of coal behind in floor and roof
96
pillars. Collapsing board areas and multiple pillar failures in old workings between 1930
and 1940 due to the Pillars Extraction Program (Bell et al. 2001) have led to the formation
of crown-holes and extensive surface fractures above the workings. In 1947, spontaneous
combustion of coal was reported, and, despite extinction efforts, fires have been burning in
the workings ever since (Bell et al. 2001).
Extensive opencast mining operations in Witbank Coalfield began in 1996, in areas

previously mined underground. These activities allowed oxygen to enter old workings,
causing oxidation of roof and floor coal in spontaneous combustion. Today, particulate
matter and gases generated by coal fires make the mine environment hazardous in
terms of health and safety for miners and communities living in Witbank (Emalahleni)
Township.
3 Methodology
3.1 Sampling techniques
Five gas samples were collected from Witbank Coalfield. For cost efficiency, a hand-
operated pump with Teflon inlet and exhaust tubes was used. The gas was pumped into
an inert and impermeable container and brought to a laboratory for analysis. The vent
temperature was measured using a digital thermocouple connected to a 1 m platinum
thermoprobe designed for this purpose.
3.2 Analysis
Gas chromatographic analysis was performed according to Colman et al. (2001) and
Barletta et al. (2002). After cryogenic pre-concentration with liquid nitrogen (-196 °C),
the air samples were re-vaporized in a hot water bath and channeled into five different
detectors, connected themselves to three separate Hewlett-Packard (HP) 6890 gas
chromatographs (GC). Two electron capture detectors (ECD), a mass spectrometer detector
(MSD), and two flame ionization detectors (FID) were used for analysis of hydrocarbons,
halocarbons, and a variety of halogenated compounds. Carbon monoxide and methane
were analyzed in a GC (HP 5890) equipped with FID. For carbon dioxide analysis, an HP
5890 gas chromatograph with a thermal conductivity detector was used.
97
4 Results and Discussion
The gases from the sampled coal fire vents of Witbank Coalfield consisted of a complex
mixture of hydrocarbons, halocarbons, and greenhouse gases (GHGs). Seven compounds
were predominant; all of them are commonly associated with environmental and human
health hazards (Terblanche et al. 1992, 1994; Finkelman 2004). These compounds are
benzene, toluene, ethylbenzene, xylene, methane, carbon dioxide, and carbon monoxide.
The concentrations found are listed in Table 1.
Table 1: Predominant compounds from coal fire gas discharge in Witbank Coalfield, South
Africa
Compound Concentration Range
Benzene (C6H6) 1,260-11,710 ppb
Toluene (C7H8) 315-39,736 ppb
Ethylbenzene (C8H10) 10-21,140 ppb
Xylene (C8H10) 14-75,880 ppb
Methane (CH4) 165-11,710 ppm
Carbon dioxide (CO2) 4,288-89,612 ppm
Carbon monoxide (CO) 1-27,500 ppm
The presence of these gases in the samples is a result of combustion-induced thermal

degradation of biopolymers contained in coal (Puttmann et al. 1991). Stracher et al. (in
press) stress that exchange reactions between gas, rocks, and solutions influence the
chemistry of gas en route to the surface. The high concentrations of carbon monoxide
and carbon dioxide detected in the samples were in part anticipated by Annegarn et al.
(1996a,b), who found that gas levels above the recommended World Health Organization
(WHO) standards and South African national guidelines sometimes occurred in the Vaal
Triangle.
Spontaneous combustion of coal releases large quantities of toxic gases into the
atmosphere, impacting heavily on local and regional air quality. Data for Witbank
Coalfield confirm that high amounts of BTEX are released from coal fires: benzene 1,260
to 11,710 ppb, toluene 315 to 39,736 ppb, xylene 14 to 75,880 ppb, and smaller quantities
of ethylbenzene (10 to 21,140 ppb) (Figure 2). Volatile organic compounds (VOCs) are
very reactive; once emitted, they react swiftly with other chemicals to form secondary
pollutants.
98
Figure 2: Variation of BTEX in gas samples from Witbank Coalfield, South Africa
Carbon monoxide, the main gas released throughout the initial stages of coal combustion
(incomplete combustion), was also detected (1 to 27,500 ppm). Depending on the
combustion regime (temperatures varying between 80 and 700 °C) in the coal seam,
carbon dioxide was produced in substantial concentrations (4,288 to 89,612 ppm). Other
elements released were nitrogen oxide, nitrogen dioxide, sulfur dioxide, and methane (165
to 11,710 ppm). The concentrations of carbon dioxide, methane, and carbon monoxide as
found in the samples are presented in Figure 3.
Figure 3: Variation of GHGs and carbon monoxide in gas samples from Witbank
Coalfield, South Africa
As a consequence of chemical reactions and gas-to-particle conversions, secondary

pollutants may be produced in coal fires. Ozone is the main secondary gaseous
99
emission from coal combustion, but considerable volumes of highly reactive secondary
particulate pollutants are also formed after oxidation of sulfur and nitrogen oxides
(Querol et al. 1998).
Trace gases, i.e. elements and compounds such as water vapor, carbon dioxide, methane,
nitrous oxide, etc., which contribute to the greenhouse effect but make up less than 1 % of
the earth’s atmosphere, have relatively long atmospheric lifetimes. They can be transported
over long distances when released from coal fires. The result is wide-spread dust and
air pollution, even far off the original source. Transport of pollutants, particularly of
visible and odorous compounds, causes problems between coal mines and neighboring
communities. Concerned residents often complain about dust generation; they are worried
about the health impacts of such emissions (Evans & Bennett 1998).
Spontaneous combustion of coal has been recognized by the Inter-Governmental Panel

for Climate Change (IPCC) as a potential source of GHG emissions. However, it has been
excluded so far from greenhouse inventories (Australian Coal Research 2003), since there
is no acceptable method for estimating emissions. This is partly due to a lack of accurate
data on the extent of fires for different areas.
Coal fires had developed into a severe problem in the Witbank area by the mid 1980s.
Measures were taken then to bring them under control – with hardly or no effect. The
fires continued, releasing ever more pollutants into the atmosphere. Overall, this study
provided important information about emissions from spontaneous combustion in
coal mines in South Africa. The data have significantly enhanced our understanding
of this phenomenon and its potential contribution to local and regional degradation
of atmosphere and human health. Given both the international and local experience
with air pollution and health problems arising from coal fires, as well as the generally
compromised health status of the South African populace due to diseases and epidemics,
it is necessary to foster understanding of coal spontaneous combustion in South Africa.
Determining the impact of coal mining and coal fires on air quality is essential for
devising effective control measures. The necessary shift in South Africa’s air quality
legislation from a source-based approach to an ambient air quality paradigm requires the
identification and assessment of individual sources of pollution in a given area. Although
guidelines for most toxic combustion compounds are currently absent in the South Africa
Air Quality Standards, an environmental impact assessment seems useful considering
the extent of spontaneous combustion of coal in Witbank Coalfield.
100
5 Conclusion
The results of our gas analyses indicate that coal fires in Witbank Coalfield release
substantial concentrations of toxic gases, including benzene, toluene, ethylbenzene,
xylene, methane, and carbon oxides. The BTEX compounds in particular are important
from a health perspective. The concentrations detected in Witbank Coalfield were higher
than local and international standards, and this definitely calls for further investigations
to determine more precisely the impact of these compounds on human health and the
environment. Collecting individual samples from burning areas may provide useful
information; however, an increased number of samples seems more appropriate for
determining overall mine emissions. Concentration measurements at the source as in our
study provided useful data; continuous emission measurements over a specific burning
area would possibly enhance these data even further. They would provide information
on the exact concentrations emitted from a site, before any transformations in or into the
atmosphere have occurred.
References
Annegarn, H.J., Turner, C.R., Helas, G., Tosen, G.R., and Rorich, R.P. 1996a. Gaseous pollutants. In
Held, G., Gore, B.J., Surridge, A.D., Tosen, G.R., Turner, C.R., and Walmsley, R.D. (Eds.),
Air pollution and its impacts on the South African Highveld. Cleveland, Ohio: Environmental
Scientific Association, 25-33.
Annegarn, H.J., Terblanche, A.P.S., Sithole, J.S., Rorich, R.P., and Turner, C.R., 1996b. Residential air
pollution. In Held, G., Gore, B.J., Surridge, A.D., Tosen, G.R. , Turner, C.R., and Walmsley,
R.D. (Eds.), Air pollution and its impacts on the South African Highveld. Cleveland, Ohio:
Environmental Scientific Association, 47-57.
Australian Coal Research, 2003. Greenhouse gas emissions from spontaneous combustion at open cut
coal mines: An investigation of air quality modeling and inverse techniques. CSIRO Energy
Technology, PMB7, Bangor, New South Wales, Australia, 2234.
Barletta, B., Meinardi, S., Simpson, I.J., Khwaja, H.A., Blake, D.R., and Rowland, F.S., 2002. Mixing
ratios of volatile organic compounds (VOCs) in the atmosphere of Karachi, Pakistan.
Atmospheric Environment 36: 3429-3443.
Bell, F.G., Bullock, S.E.T., Halbich, T.F.J., and Lindsay, P., 2001. Environmental impacts associated with
an abandoned mine in the Witbank Coalfield, South Africa. International Journal of Coal
Geology 45: 195-216.
Cairncross, B., 2001. An overview of the Permian (Karoo) coal deposits of southern Africa. Journal of
African Earth Sciences 33(3-4): 529-562.
Cairncross, B. and Cadle, A.B., 1988. Palaeoenvironmental control on coal formation, distribution and
quality in the Permian Vryheid Formation, East Witbank Coalfield, South Africa. International
Journal of Coal Geology 9(4): 343-370.
101
Colman, J.J., Swanson, A.L., Meinardi, S., Sive, B.C., Blake, D.R., and Rowland, F.S., 2001. Description
of the analysis of a wide range of volatile organic compounds in whole air samples collected
during PEM-Tropics A and B. Analytical Chemistry 73: 3723–3731.
EIA, 2005. South Africa: Energy and environmental issues. Energy Information Administration (EIA),
Washington, D.C., U.S.A. 31 October 2005 http://www.eia.doe.gov/emeu/cabs/safrenv.html
Evans, M. and Bennett, A., 1998. Air quality: Environmental exposure to pollutants. Health Evidence
Bulletin. 12 November 2005 http://hebw.cf.ac.uk/healthyenvironments
Falcon, R.M.S. and Keyser, M.J., 2001. The status of coal in South Africa: Production, utilization,
characterization. In Final Proceedings of the International Conference on Coal Quality:
Global Priority, Reston, Virginia, U.S.A., September 2001. US Geological Survey Circular
1272. (CD-ROM)
Finkelman, R.B., 2004. Potential health impacts of burning coal beds and waste banks. International
Journal of Coal Geology 59: 19-24.
Lloyd, P.J., 2002. Coal mining and the environment. In Proceedings of the International Conference on
Coal and Environment, Durban, South Africa, October 2002. International Bar Association
(IBA), London, United Kingdom, 8-9.
Puttmann, W., Steffens, K., and Gobel., W., 1991. Analysis of aromatic hydrocarbons in overburden from
coal mines: Assessment of the environmental impact. In Peters, D.C. (Ed.), Geology in coal
resource utilization. Fairfax, Virginia: TechBooks, 483-498.
Querol, X., Alastuey, A., Puicercus, J.A., Mantilla, E., Miro, J.V., Lopez, A., Plana, F., and Artinano, B.,
1998. Seasonal evolution of suspended particles: Particulate levels and sources. Atmospheric
Environment 32(11): 1963-1978.
Stracher, G.B., 2003. Coal mine fire – Gas and condensation products: Collection techniques for
laboratory analysis. Energeia 14(5): 1-6.
Stracher, G.B. and Taylor, T.P., 2004. Coal fires burning out of control around the world: Thermodynamic
recipe for environmental catastrophe. International Journal of Coal Geology 59: 7-17.
Stracher, G.B., Prakash, A., Schroeder, P., McCormack, J., Zhang, X., Dijk, P.V., and Blake, D. New
mineral occurrences and mineralization processes: Wuda coal fire gas vents of Inner
Mongolia. American Mineralogy. (in press)
Terblanche, A.P.S., Nel, R., and Golding, T., 1994. Household energy sources in South Africa: An
overview of the impact of air pollution on human health. CSIR Environmental Services,
Department of Mineral and Energy Affairs & EMSA, Pretoria, South Africa.
Terblanche, A.P.S., Opperman, L., Nel, C.M.E., Reinach, S.G., Tosen, G., and Cadman, A., 1992.
Preliminary results of exposure measurements and health effects of the Vaal Triangle air
pollution health study. South African Medical Journal 81: 550-556.
Uludag, S., 2001. Assessing spontaneous combustion risk in a South African coal mine using a GIS tool.
MSc thesis, University of the Witwatersrand, Johannesburg, South Africa.
102
Emission of Toxic Gases and Condensation
Minerals from Burning Coal Waste Heaps
in Northern France:
Measurements and Analyses
M. Naze-Nancy Masalehdani1, Jean-Luc Potdevin1,

Fabrice Cazier2, and Dominique Courcot3
1
UMR CNRS 8110 “Processus et Bilans des Domaines Sédimentaires,” Université des Sciences et
Technologies de Lille, Villeneuve d’Ascq, France
2
Centre Commun de Mesures, MREID, Université du Littoral, Dunkerque, France
3
Laboratoire de Catalyse et Environnement, Université du Littoral, Dunkerque, France
Abstract
The Nord-Pas-de-Calais region in northern France is well known for its great number of coal
waste heaps. One of the problems associated with these residues from coal mining is spontaneous
combustion. Spontaneous combustion poses significant safety, geo-technical, and environmental
hazards; it is at the origin of gas spots (vents), ground subsidence, surface cavities, noxious gases,
secondary minerals, and thermal metamorphism of rock debris. Gas spots were localized at one
site based on field observations and aerial infrared thermography. In-situ gas measurements and
analyses were carried out around active spots at two burning coal waste heaps. The gases measured
were greenhouse (CO 2 and CH 4) as well as acidic gases (SO 2 and NO x), CO, O 2, and H 2O. The
concentration of CO 2 (7 %), in particular, was higher than expected. O 2 concentrations ranged
between < 1 and 19 %. The ambient air at some spots had to be classified as contaminated or
poisonous. Secondary salt efflorescence of single minerals or mineral compounds was observed on
carbonaceous shale around gas vents. To a large part, these minerals were deposited as encrustations
on rock debris at temperatures of less than 300 °C; they included sulfates (thenardite, bloedite,
konyaite, glauberite, aphthitalite, langbeinite, arcanite, gypsum, epsomite, hexahydrite, alunogen,
kalinite, pickeringite, halorichite, tschermigite, mascagnite, lecontite), halides (salammoniac and
halite), native sulfur, and two possibly new unknown minerals. Most of the sulfate minerals were
documented for the first time in connection with coal fire emissions; the same applies to fluid
inclusions in thenardite crystals. Major chemical components detected in fumarole discharge were
103
SO4, NH4, H2O, Cl, Mg, K, Na, Al, Ca, Fe, Mn, and S. A multitude of different approaches – field
observations, optical microscopic examinations, X-ray diffraction, scanning electron microscope
observations, energy-dispersive spectrometry, wavelength-dispersive spectroscopy, and total
chemical analysis (ICP-MS) – were combined to study the occurrence of secondary minerals.
摘要
法国北部的Nord-Pas-de-Calais地区以其大量的煤矸石堆而著名。煤自燃是与这些煤矿开采废弃
物有关的重要问题之一。煤自燃引起了安全、地质和环境灾害，正是煤自燃引发了气体排放、地
表沉陷、地表洞穴、有害气体、次生矿物和岩石碎屑热变质作用。根据野外观察和航空热红外测
量圈定了气体排放位置，还在两个正在燃烧的矸石堆周围气体排放处进行了现场气体测量与分
析。所测气体包括了温室气体（CO2和CH4），及酸性气体（SO2 和NOx）、CO、O2和H2O。特
别是CO2气体浓度的比例（7%）远高于预期值。O2含量介于1到19%。在某些观测点周围空气可归
类为污染气体或者有害气体。在气体排放孔周围的碳质页岩中可见单矿物或复和矿物的次生盐积
物。这些矿物大部分在低于300℃时沉积形成岩石碎屑的盖层，包括硫化矿物（芒硝、无水芒硝、
白钠镁钒、konyaite、钙芒硝、钾芒硝、无水钾镁矾、单钾芒硝、石膏、泻利盐、六水氢化物、毛
矾、纤维钾明矾、镁明矾、铁明矾、铵明矾、铵矾、钠铵矾），卤化矿物（氯化铵和岩盐），天
然硫磺，可能还有两种未知的矿物。大部分硫酸盐矿物在煤火排放中属于第一次有记载，同样在
无水芒硝晶体中还发现了流体包体。在矸石堆煤火喷气孔探测到的主要化学成分是SO4、NH4、
H 2O、Cl、Mg、K、Na、Al、Ca、Fe、Mn和S。研究中采用多种不同测试方法（野外观测、光
学测试、X光衍射分析、电子显微镜扫描、能量扩散光谱仪、波长分散光谱仪和全部化学分析
（ICP-MS）等方法）相结合进一步研究发现的次生矿物。
1 Introduction
Expanding energy consumption and the discovery of coal in the 17th century had the
coal mining industry in northern France thrive over centuries. Underground coal mining,
however, also produced large amounts of coal waste, which was deposited in heaps
(about 600) near coal mines, towns, and roads, with little or no environmental concern.
The composition of these coal waste heaps was largely dependent on the mining method
and type of equipment used in their construction, and the sequence of tipping. The coal
waste heaps in the Nord-Pas-de-Calais region vary in size and shape. The oldest heaps
are elongated; those of the early 20th century are mostly cone-shaped (Figure 1). They
consist of shale, sandstone, and coal, deposited in sedimentary basins during the Upper
Carboniferous 335 to 295 million years ago. The predominant geological material in most
heaps is carbonaceous shale, often constituting over 85 % of all waste. Its mineralogical
composition may differ, but pyrite occurs in almost all waste material.
104
Figure 1: Different types of coal waste heaps. a) Elongated coal waste heap
“Pinchonvalles,” Avion. b) Cone-shaped coal waste heaps “Loos-en-Gohelle,”
Pas-de-Calais, around 1960 (Paquette 2003).
Some of the coal waste heaps in the Nord-Pas-de-Calais were later removed for civil works or
transformed into leisure areas. Others could not be put to immediate use and continue to spoil
the landscape, or worse – pollute the environment. Some of them have undergone spontaneous
combustion. Today there remain about 300 coal waste heaps in northern France. Spontaneous
combustion is most likely to occur in loosely tipped and hence well aerated heaps with high
carbonaceous and pyrite content. Moisture content and grading are important too. At low
temperatures, free moisture increases the rate of spontaneous combustion (Cook 1990).
Coal waste heaps are complex structures, made up of a mixture of materials with different
reactivities toward oxygen and different broad particle size distribution. Where particles
105
are large, air flux can easily dissipate heat; in fine materials, air stagnates and allows heat
to build up gradually. However, stagnant air also means that fires caused by accumulative
heating will stop as soon as the oxygen in the material is consumed. Waste distribution in
coal waste heaps depends both on the layering of the mined soil as well as on the dumping
method, each resulting in a different overall layering and mixing of materials in the heap.
For example, dragline mining will produce a sequence of layers different from that of
shovel mining (Carras et al. 1999). Particles of different sizes may be segregated over time
as surface material erodes down the heap.
Broadly speaking, coal waste heaps are structures of reactive material which allow gas and
heat transport. Oxygen (O2) and water (H2O) vapor are the decisive factors for spontaneous
combustion. Fires caused by self-heating are indeed a moderately common occurrence in
coal mining. They received considerable attention in the past from the perspective of mine
safety, but are now coming under closer scrutiny as a source of greenhouse gases (GHGs).
Coal and carbonaceous rock debris react with atmospheric O2 in a heat-generating process.
Under certain conditions, the heat accumulates and raises the temperature inside the material.
This rise in temperature increases the rate of chemical reactions and thus, again, the rate
at which heat is generated. Should the heating remain unchecked, it is bound to result in
combustion.
Spontaneous combustion poses significant safety, geo-technical and environmental hazards;

it is at the origin of gas spots (vents), ground subsidence, surface cavities, noxious gases,
secondary minerals, and thermal metamorphism of rock debris (Masalehdani & Potdevin
2004; Mroueh et al. 2005). In places where gas and water vapor are discharged along open
cracks and vents, solid minerals condensate on unbaked and baked shale debris.
This paper presents methods and results of in-situ gas measurements and analyses at two
burning coal waste heaps in the Nord-Pas-de-Calais region. At one site, gas vents (hot
spots) were localized based on field observations and aerial infrared thermography. At
both sites, samples were collected and laboratory-tested. Efflorescence salt minerals were
identified, too. Texture, composition, and mode of occurrence of these minerals will be
presented in another publication.
2 Study Sites
The burning coal waste heaps selected for this study were coal waste heaps nos. 76 and
83, numbered according to the Charbonnages de France for the Nord-Pas-de-Calais
region. Heap no. 76 is situated 1.3 km south of the city of Avion. Before road works led
to a large-scale clearing in the years following 1993, it used to be cone-like in shape,
106
rising up to 85 m above ground level over an area of 96,000 m2 (Figure 2). At that time,
heap no. 76 was estimated to contain 500,000 t of mixed waste materials. Spontaneous
combustion here occurred in 1987.
Figure 2: Coal waste heap no. 76, Avion. a) Location map 1:60,000 and aerial photo of
site during combustion in 1987 (photo courtesy of S.A. Terrils, 1993). b) Visible
gas discharge, 2002.
107
Coal waste heap no. 83 borders the national motorway A21 and lies northeast of the city of
Fouquières-les-Lens. It is elongated (tabular), with a ridge up to 10 m above ground level,
running discontinuously over about 82 ha (Figure 3). After a short period of debris clearing
from 1992 to 1999, the heap was largely abandoned.
Figure 3: Coal waste heap no. 83, Fouquières-les-Lens. a) Location map 1:25,000 and
aerial photo of site 1:5,000 (Paquette 2002). b) Horizontal fissure and gas vent
(35 cm wide, 3 m long), 2005.
108
3 Detection of Gas Spots (Vents)
Basic visual monitoring at regular intervals is imperative for coal waste heaps with high
combustibles content. Fumaroles can be detected in wet weather and gas spots in snowy
seasons; changes in vegetation and noxious gas emissions have to be surveyed too.
Aerial infrared thermography is a common method for surveying coal waste heaps. The
advantage of extensive spatial coverage and clear visualization of risk areas, however,
is offset by rather high application costs, a pronounced sensitivity toward atmospheric
conditions, and the impossibility to follow up on combustion reactions. Field investigations
must complement infrared surveys in order to localize gas spots (vents) in burning coal
waste heaps in a most efficient and exhaustive manner. Investigations at heap no. 83
combined both approaches (Figure 4a).
For thermographic detection of infrared radiation (3-5 and 8-14 µm), an airborne multi-
spectral numeric scanner of the DEDALUS type was used. Measurements were performed
before dawn, at a height of 400 m to ensure a ground resolution of 1 m. Thermal resolution
at ground level was 0.25 °C. Images were color-treated to bring out combustion zones in
greater depth and different vegetation zones. It showed that coal waste heap no. 83 hosts
a number of combustion zones (gas spots) in its flanks. These zones are located upslope;
particularly along the heap’s east flanks near the purification station, along the west flank
of the Marais de Lengles sector, and at the tip of the southwest flank (sector of Le Tierce)
(Figures 4b and 4c).
109
Figure 4: Coal waste heap no. 83, Fouquières-les-Lens. a) Map of gas spots and location
of gas measurements. b) and c) Aerial thermographic infrared analysis 1:5,000
(after Paquette 2002): b) north sector, c) south sector.
110
4 In-Situ Gas Measurements: Methods of Analysis

As site conditions at heap no. 83 were not optimal (the heap is covered with shrubs and
small trees, and contains a number of dangerous vents), gas measurements and analyses
at this site were carried out only twice, from a single gas spot in the east flank near the
purification station (Figure 4a). Due to technical problems, carbon dioxide (CO 2) and
carbon monoxide (CO) concentrations could not be measured for either site in the first
measurements. At coal waste heap no. 76, gas analyses were carried out at different
selected spots as shown in Figure 5c.
A special technique was devised to measure gas emissions in situ at both heaps. For
this purpose, a stainless steel chamber was placed over a gas spot (vent), and the gas
streaming up from the surface directed into the sampling apparatus through an orifice
in the chamber’s casing (Figure 5a). Since water vapor in the gas mix would disturb
infrared measurements and harm the high-tech analyzer, it was eliminated by a tangential
condenser placed between gas chamber and analyzer. Tangential condensers are very
common in these set-ups; they trap water vapor by freezing on dry ice. Upstream to the
water trap, a fraction of the gas was channeled into bubblers in order to bind hydrogen
sulfide (H2S). Water vapor here was no hazard. At the exit of the water trap, through a
bifurcated cold transfer line, the then water-free gas was pumped into the analyzer bay
placed in a specially equipped vehicle (a mobile laboratory) for analysis of CO2, CO,
hydrocarbons, sulfur dioxide (SO2), nitrogen oxides (NOx), and O2. Again, bubblers
served to trap remnant H2S (Figure 5b).
111
Figure 5: Coal waste heap no. 76, Avion. a) In-situ gas measurements. b) Outline of gas
sampling device. c) Gas spots selected for gas measurements.
112
H2S was trapped with a solution of sodium hydroxide (NaOH; 0.5 mol l-1), to which
N,N-dimenthyl-p-phenylenediamine oxalate was later added for examination by UV
spectroscopy (absorbance 665 nm). The coloring (methylene blue) caused by this reagent
is proportional to the amount of sulfides (S2-) in the host solution. Other gases such as CO,
CO2, methane (CH4), SO2, nitrogen dioxide (NO2), and nitric oxide (NO) were measured
with a complex system composed of the following elements (Figure 6):
• A membrane pump with heated head,

• A heated dust filter for elimination of particles,
• A cooling apparatus for trapping of residual moisture,
• A methane and non-methane hydrocarbon analyzer
(FID-model Cosma Graphite 355),
• An oxygen analyzer (based on paramagnetic detection), and
• A non-dispersive infrared analyzer (Hartmann and Braun, Model URAS 100) for CO/
CO2, NO/NO2, and SO2.
Figure 6: Outline of gas analysis system
113
5 Gas Mineralization: Analytical Methodology
Secondary minerals are a frequent sight along open cracks and fissures with hot gas
and/or water discharge; they are reaction products of subsurface combustion of coal and
carbonaceous rock debris (Figure 7). Specimens of these minerals were collected at both
study sites and placed into containers after 5 minutes of cooling in fresh air. The containers
were then sealed and brought to a laboratory for analysis. All samples were kept at ambient
temperature in a shelter throughout testing. A first series of analyses was performed within
a week after collection.
Figure 7: Coal waste heap no. 76, Avion. a) Sulfate mineral crystallizations. b) Gas vent;
nipper for indication of scale: 15 cm.
114
The standard method of sample preparation in whole-sample mineralogy is to grind

samples with a mortar and spread the powder obtained on a glass slide. This method
was also used in the present study. X-ray diffraction (XRD) data were generated using
a Phillips PW1710 diffractometer, with an accelerating voltage of 40 kV and a current
intensity of 25 mA. Diffraction patterns were collected over 3-60°2θ using CuKα radiation
at 1°2θ min-1. Peak search for minerals within each x-ray pattern was performed with
MACDIFF software version 4.2.2.
Some of the samples provided enough condensate for preparation of thin sections
amenable to examination of mineralogical, textural, and morphological features, as well as
quantitative compositional analysis. Special precautions were taken in sample preparation
to account for the salt crystals’ brittleness and high solubility, which, for instance,
precluded water use. Polished uncovered sections in a range allowing for both transmitted
light viewing and fluid inclusion analysis (60 to 100 and 50 µm) were prepared. The
thickness of these sections depended on the granulometry of the samples (fan-shaped).
They were placed in purpose-built boxes (15 × 15 × 15 mm), which were then filled with
epoxy. After hardening, the faces of the boxes were smoothed down with an abrasive until
perfectly flat; they were then coated and left to dry at room temperature for about 24 hours.
After drying, the upper faces were again rectified by means of a series of coarse to very
fine grain abrasives (sizes: 1200 to 2400 µm). It is imperative to impregnate the samples
during this process. After polishing, the samples were fixed on frosted glass wafers,
where sections of interest were leveled with petrol. Final rectification was performed with
abrasive paper, and a last polish given with petrol and diamond paste.
Optical studies were carried out using an Olympus BX60 transmitted and reflected light
microscope, equipped with software and camera. This special device allowed detection of
fluid inclusions in the samples as well as micro-thermometric analysis. The composition
of trapped fluids is a direct link to the composition of the vapor which has led to the
formation of the crystals in the first place. The results will be presented elsewhere.
Environmental scanning electron microscopy (ESEM) analysis of bulk samples was

carried out with a Quanta 200 plus EDS (energy-dispersive spectrometry microanalysis
device) to confirm XRD mineral identification results and trace minerals contained in only
small concentrations in the sample material. Accelerating voltage was 15 kV. Samples
for mineral phase identification, elemental analysis, and photomicrographs by EDS were
carbon-coated; they were used as either uncovered thin sections or crystal chips.
115
Quantitative electron microprobe (EMP) and wavelength-dispersive spectroscopy (WDS)

determination of chemical composition was performed with a Cameca SX50 and SX100 at
the CAMPARIS laboratory, Department of Geology, Jussieu, Université Paris VI. Due to
the instability and fragility of the minerals, beam power was minimized to 15 kV, current
intensity to 5-10 nA, and spot size to 10 µm.
Some efflorescence samples were additionally tested for soluble sulfate condensate
composition, which XRD mineralogy alone was not able to determine. Some of them
contained trace elements such as As, Pb, Zn, and Cu.
Secondary salt efflorescence of single minerals or mineral compounds was observed on

carbonaceous shale around gas vents. To a large part, these minerals were deposited as
encrustations on rock debris at temperatures of less than 300 °C; they included sulfates
(thenardite, bloedite, konyaite, glauberite, aphthitalite, langbeinite, arcanite, gypsum,
epsomite, hexahydrite, alunogen, kalinite, pickeringite, halorichite, tschermigite,
mascagnite, lecontite), halides (salammoniac and halite), native sulfur, and two possibly
new unknown minerals.
Some of the identified minerals are generally held to be the product of volcanic sublimates
or evaporates. They have never been described in connection with coal seam fires or
burning coal waste heaps.
6 Results of Gas Measurements and Analyses
Gas temperatures and measurement results for H2S, SO2, NOx, CO, CO2, CH4, O2, as well
as H2O are presented in Table 1 and Figures 8 to 12. Measurements at coal waste heap
no. 83 (gas spot E1) showed that gas composition can change within minutes to seconds.
During a period of 1 h 45 min, SO2, CO, CO2, CH4, NOx, and O2 were found to vary
between 15 and 25 ppm (SO2), 20 and 25 ppm (CO), 3 and 7 % (CO2), 110 and 260 ppm
(CH4), 15 and 25 ppm (NOx), and 13 and 18 % (O2), respectively. Gas emissions at this
spot were evidently pulsating – a phenomenon directly rooted in or influenced by the zone
of combustion and gasification. A strong heterogeneity of temperatures around gas vents
was observed at both coal waste heaps.
The temperatures at which the measured gases first appear is unknown, but at the active
gas spots (vents) of both coal waste heaps, near surface temperatures were measured in
a range from 56 to 255.7 °C and 74.3 to 77.8 °C, respectively (Table 1). The gases could
116
not have been generated at such low temperatures and must undoubtedly stem from hotter
internal zones in the coal waste heaps.
Table 1: Gas temperatures and measurement results from coal waste heaps nos. 76 and 83
Site, Gas T CO CO2 CH4 SO2 NOx O2 H2Ob

Spot (°C) (ppm) (%) (ppm) (ppm) (ppm) (%) (ppm)
76, S1 100.5 ND ND 80 <5 15-26 ND 100
76, S2 225.7 ND ND 47 11 15-26 18-19 ND
76, S3 99.7 110 7 ND 9.6 26 15-18 3550
76, S4 56 22 6.13 191 19.8 7.87 14-15 30
83, E1 74.3 ND ND 96 20 25 <1 ND
83, E1a 77.8 20-25 3-7 100-260 15-25 15-25 13-18 68

a
Second measurement
b
Condensate water collected
ND = no data
Figure 8: Gas analysis at gas spot S1 of coal waste heap no. 76, Avion
117
118
Figure 11: Gas analyses at gas spot E1 (first measurement) of coal waste heap no. 83,
Fouquières-les-Lens
Figure 12: Gas analyses at gas spot E1 (second measurement) of coal waste heap no. 83,
Fouquières-les-Lens
119
6.1 Hydrogen sulfide (H2S)
The NaOH bubbler solutions obtained during gas measurements at spots S3 of coal waste
heap no. 76 and E1 of coal waste heap no. 83 were analyzed for their UV/visible spectrum
as described in Chapter 4. Standard sulfide (S2-) solutions have their peak of absorption at
665 nm; the two tested solutions, however, failed to show such a peak (Figure 13). This
confirms that H2S was absent, both in the solutions and in the emitted gases. To make
sure traces of the gas had not been trapped in the apparatus upstream to the bubblers, the
condensate water collected in the cold traps was analyzed too. Test results here were also
negative.
Figure 13: Verification of H2S at 665 nm wavelength, coal waste heaps nos. 76 and 83
Although H2S or “stink damp” was not found in gas analysis, it should be noted that,
on several occasions during field investigations at both sites, the typical “rotten egg
smell” indicative of H2S hovered over gas spots. It is probable that H2S in the emission
gases had oxidized and precipitated at temperatures higher than what was measurable
with our devices, or had escaped from the gases already below surface. H2S is a reaction
product of combustion of elemental sulfur, usually at temperatures above pyrolysis level
(Mondragón et al. 2002), provided oxygen is not sufficient for complete oxidation (Bell
et al. 2001).
H 2S analysis for spots S3 and S1 revealed that NO x and SO 2 had also partly been
trapped in the form of nitrate (NO3-), nitrite (NO2-), and sulfate (SO42-). At gas spot S3,
120
25 ppm NOx and 13.4 ppm SO2 were directly measured by the analyzer; 1 ppm (NO3-)
and 0.61 ppm (SO42-) were detected in the condensates. Likewise, at gas spot S1, 0.91 ppm
NO3- and NO2-, and 0.05 ppm SO42- were found in the water trap.
6.2 Carbon monoxide (CO)
CO or “white damp” is a colorless toxic gas, whose hazardous nature is aggravated by

the fact that it cannot be detected either by taste, smell, or irritation (Bell 1996). CO
nonetheless is a good indicator of coal fires because it is released sequentially with
increasing temperature during heating (Stracher & Taylor 2004).
The reaction of coal with oxygen is an exothermic process, leading in a first step to
the formation of a “coal oxygen complex.” As a chemical phenomenon it has received
particular scientific attention (e.g. Schmidt 1945; Sondreall & Ellman 1974; Nordon
et al. 1979; Cudmore & Sanders 1984; Nelson 1989; Khan et al. 1990; Chen & Scott
1993; Carras & Young 1994; Krishnaswamy et al. 1996). CO is known to be a result of
incomplete combustion of carbon contained in coal. Kim (1973) discovered that CO is not
only released from coal but also from carbonaceous shale at relatively high temperatures
(≥ 100°C). CO content in gas discharge from coal fires is above all a function of
temperature; this clearly shows in Table 1. The higher the combustion temperature, the
higher the CO content (Wang et al. 2002).
CO concentrations at gas spot S3 measured up to 110 ppm, which is significantly

higher than the Threshold Limit Value (TLV; see Table 2). All other measurement
locations ranked lower (Tables 1 and 2). CO emissions of any concentration must be
considered highly dangerous as CO readily combines with hemoglobin in human blood,
where it impedes oxygen uptake. The highest concentration of CO to which man may
be exposed continually without adverse effect is 50 ppm (TLV). Above this level it
will cause symptoms such as headache, fatigue, or dizziness in healthy individuals if
breathed (Grossman et al. 1994). Higher concentrations can of course also contribute to
combustion hazards (DEP 2001).
121
Table 2: Effects of noxious gases

Gas Concentration by Effects
Volume in Air
(%) (ppm)
NO2 0.005 50 Threshold Limit Value

0.062 620 Least amount causing immediate irritation to throat
0.077 770 Rapidly fatal after short exposure
CO 0.005 50 Threshold Limit Value

0.02 200 Headache after c. 7 h at rest or after 2 h of exertion
0.04 400 Headache and discomfort with possibility of collapse
after 2 h at rest or 45 min of exertion
0.12 1200 Palpitation after 30 min at rest or 10 min of exertion
0.20 2000 Unconsciousness after 30 min at rest or 10 min of
exertion
CO2 0.5 5000 Threshold Limit Value

– Lung ventilation slightly increased
5.0 Breathing labored
9.0 Depression of breathing commences
SO2 0.0001-0.0005 1-5 Can be detected by taste at lower levels and by smell at
upper levels
0.0005 5 Threshold Limit Value
– Onset of irritation to nose and throat
0.002 20 Irritation to eyes
0.04 400 Immediately dangerous to life
Notes:
1) Some gases have a synergistic effect, that is their presence augments the effect of others and lowers the
concentration at which symptoms normally occur. Also, a gas which is not toxic itself may increase
the toxicity of another, for example by increasing the rate of respiration; strenuous work has a similar
effect.
2) Of the gases listed, only CO is likely to pose a danger to life; it is the most common and cannot
be detected by the human senses,while all other gases become intolerably unpleasant already in
concenrations far below the danger level.
Data source: Sumi and Tsuchiya (1971); NCB (1973)
6.3 Carbon dioxide (CO2) and methane (CH4)
CO2 is another product of coal oxidation; it is liberated from carboxyl groups which
form in a dehydroxylation process (Walker 1999). The complete oxidation of carbon to
122
CO2 is an exothermic reaction (1). The heat generated in this reaction is, under normal
circumstances, released into the atmosphere (as pollutant); under these conditions, the
ignition temperature for coal lies between 420 and 480 °C. However, under adiabatic
conditions, where the heat generated cannot dissipate, the minimum temperature at which
coal ignites drops dramatically to between 35 and 140 °C, depending on coal rank (Sujanti
1999).
(1) C + O2 → CO2↑ + 394 kJ mol-1
CH4 was the principal hydrocarbon emitted from the gas spots (vents) at both study
sites (cf. profiles of total hydrocarbon (THC) and CH4 in Figures 8 and 11). Here too,
a significant variation in concentration was observed at different spots. At coal waste
heap no. 76, variance strongly depended on location, ranging from 47 over 80 to 191 %
(see Table 1; gas spots S2, S1, and S4). At coal waste heap no. 83, one and the same gas
spot (E1) showed different levels of CH4 concentration at different O2 and temperature
levels (see Table 1). Emissions at the latter site were in fact all higher than at the former.
Laboratory oxidation tests revealed that CH4 is not emitted from carbonaceous shales but
exclusively from coal, at temperatures ≥ 120 °C. Therefore, all CH4 measured in situ must
have resulted from coal combustion.
CO 2 and CH 4 are greenhouse gases (GHGs). Global CO 2 emissions are expected to

rise to 29.575 million t yr -1 by 2010 and 36.102 by 2020. Geo-natural hazards such as
spontaneous combustion of coal seams or coal waste heaps are thus not only a significant
source of large quantities of toxic gases, but also of GHGs. CO2 cannot be eliminated from
the combustion process, but it can be reduced significantly (IEA 2000). If the combustion
is incomplete, more CH4 is released. According to Havenga (1993), gas emissions from
burning coal heaps increase with temperature; if the fire remains unchecked and heat is
allowed to build up, the content of CO2 and CH4 in the gas discharge can reach up to 20
and 2.5 %, respectively!
Since the adoption of the Kyoto Protocol to the United the Nations Framework Convention
on Climate Change (UNFCCC) in 1997, much attention has been directed at quantifying
GHG emissions from a variety of sources, including coal mining and coal fires. Scientists
have pointed out that controlling these man-made disasters might be a very cost-effective
way to meet national emission reduction targets under the Kyoto Protocol. CH4 (mine
methane gas) is of greater concern in this respect, since its Global Warming Potential is
more than 20 times higher than that of CO2 (Carras et al. 1999; Yoshimura 1999).
123
Our measurements indicate that CO2 is indeed produced in high quantities at both coal
waste heaps; in fact, it is predominant among the emitted gases. CO2 concentrations as high
as 7 % were detected – this is much higher than the TLV of 5,000 ppm (0.5 %) (see Table 2).
Apart from their environmental impact, high CO2 concentrations also affect human health
if inhaled by people living in the vicinity of burning coal waste heaps. Inhalation of CO2
stimulates respiration and this, in turn, increases inhalation of both oxygen and toxic gases/
vapors from the fires. Respiration is pronounced at a concentration of 5 %; exposure for
30 minutes will produce signs of intoxication. At more than 7 %, unconsciousness results
after only a few minutes (ACGIH 1971; Onargan et al. 2003).
Sampled gas emissions from the two study sites varied considerably in CH4 concentration.
CH4 is the most easily combustible gas among the emissions from coal waste heaps nos. 76
and 83. 10 years ago, there seems to have been a gas explosion at coal waste heap no. 83 (oral
confirmation); this phenomenon may actually reproduce itself around the gas spots (vents)
when CH4 exceeds a concentration of 5 to 15 % (Yalcin & Gurgen 1995; DEP 2001). This
concentration is called Lower Explosion Limit (LEL); it is a function of O2 availability/
concentration and ambient temperature. According to Onargana et al. (2003), CH4 at a
concentration of 5.5 % may trigger off an explosion at as low as 100 °C. When concentration
falls to 5 %, the temperature necessary for explosion rises to 175 °C. At 4.5 % it is 250 °C;
and at 3 % 600 °C. At CH4 concentrations below 5 % and temperatures below 100 °C, there
will be no explosion, but simple gas combustion (US EPA 1993). If the CH4 concentration in
air is > 14 % and O2 concentration < 12 %, an explosion is not possible either, due to the high
specific heat of CH4 (Cp = 2.254 kJ kg-1) (Solid Waste Landfill Guidance 1999). During the
study, CH4 levels at both sites remained below the LEL (≤ 260 ppm); they were capped by
insufficient O2 and low gas temperatures (Table 1).
6.4 Sulfur dioxide (SO2)
SO2 is a colorless, water-soluble toxic gas. It can be detected by taste at concentrations of

0.35 to 1.05 ppm, and has an immediately pungent and irritating odor at concentrations
above 3.5 ppm (Rabinovitch et al. 1989).
XRD, petrography, EDS, and SEM observations confirmed the presence of pyrite (FeS2) in
some of the carbonaceous shale samples. This mineral is also found in coal. Oxidation tests
proved that SO2 was emitted from both coal and carbonaceous shale. This clearly points to
a low-temperature oxidation of pyrite (see equations (2) and (3)) – an exothermic process,
which is catalyzed in the presence of thiobacillus ferrooxidans (Harries & Ritchie 1981;
124
Bannerjee 1985; Speight 1994). Direct oxidation of FeS2 to SO2 gas (equation 4) produces
an even greater amount of heat (Chunli 1997):
(2) 2FeS2 + 7O2 + 16H2O → 2H2SO4 + 2FeSO4 + 7H2O + 1324 kJ mol-1
(3) 2FeS2 + 7O2 + 2H2O → 2H2SO4 + 2FeSO4 + 260 kJ mol-1
(4) 4FeS2 + 11O2 → 2Fe2O3 + SO2↑+ 3412 kJ mol-1
As shown in Table 1, in-situ gas measurements yielded varying concentrations of SO2,

depending on location. They ranged from the detection limit (< 5 ppm) to 19.8 ppm in coal
waste heap no. 76, and 15 to 25 ppm in coal waste heap no. 83 (see also Figures 8, 9, 11,
and 12). At gas spot S1, concentrations were especially low, and secondary soluble sulfate
minerals were observed on adjacent rock debris (Masalehdani et al. 2004). They may be a
result of oxidation due to water vapor. In the presence of water, SO2 can dissolve into an
acid aqueous solution (e.g. H2SO4). This solution reacts with the clay minerals in the debris
to form the said secondary sulfate-bearing minerals (Masalehdani et al. 2005). According
to Rabinovitch et al. (1989), SO2 can react with O2 at ambient temperature to form sulfur
trioxide SO3 (equation 5), which then reacts with water (on moist surfaces) to form sulfuric
acid (H2SO4) (equation 6).
(5) 2SO2 + O2 ↔ 2SO3↑
(6) SO3 + H2O ↔ H2SO4
SO2 is a toxic gas, and exposure to concentrations as low as 15 to 25 ppm has detrimental
effects on human health (see Table 2). At this range, SO2 may cause distress to persons
with respiratory ailments (Bell 1996). It may even lead to pulmonary edema when in
contact with mucous membranes and the respiratory tract, where it reacts to H 2SO4
(ACGIH 1971).
SO2 is also an atmospheric pollutant with profound impact on the environment as the
major cause of acid rain. When deposited, SO2 can contribute to the acidification of
soils and surface waters (de Kluizenaar et al. 2001). Recent peaks in global sulfur
emissions have been linked to coal seam fires, for example in China (Lefohn et al. 1999).
Uncontrolled emissions of SO2 from burning coal waste heaps may contribute just as
well.
125
6.5 Nitrogen oxides (NOx)
The emission of NOx (or fuel-NOx) from coal waste heaps no. 76 and 83 as observed
in our gas measurements (see Table 1) stems from nitrogen (N) bound in coal (coal-N).
Coal-N exists in the form of pyridinic-, pyrrolic- and quaternary-N and is released
during coal pyrolysis (Kelemen et al. 1994). Xu and Kumagai (2002) showed that during
hydropyrolysis of coal, the predominant nitrogen gaseous species produced is ammonia
(NH 3), which evolves from slow-heating pyrolysis; together with small amounts of
hydrogen cyanide (HCN), released from decomposition of N-containing rings in coal tar.
Most HCN is converted to NH3 in secondary reactions. Xu and Kumagai also indicated that
nitrogen evolution during pyrolysis depends on coal rank and nitrogen content. Bassilakis
et al. (1993) suggested three possible pathways for NH3 formation:
(7) (C, H, N) in coal → HCN + NH3
(8) HCN + H in coal → HCN + NH3
(9) 2HCN + H2 → HCN + NH3 + C
Kambara et al. (1993) demonstrated that NH 3 is derived from decomposition of

quaternary-N in coal, whereas HCN is a product of pyridinic- and pyrrolic-N break-up.
In-situ measurements and analyses confirmed NO x presence in the gas discharge of

both study sites, in the form of NO2 and NO. Investigation of secondary minerals on
rock debris around gas vents at coal waste heap no. 76, however, found mostly NH4+-
bearing efflorescent salt minerals (salammoniac, mascagnite, lecontite, and tschermigite)
(Masalehdani et al. 2004). It is widely held that NH3 gas reacts with atmospheric oxygen
-
to form NOx. The NO3- and NO2 anions in the condensate water of the analyzer bay
mentioned earlier may serve as evidence in this respect. Migration of NO3- and NO2- into
the soil and ground water is a source of environmental pollution (Panov et al. 1999).
NOx were detected at both coal waste heaps in a range from 7.8 to 26 ppm (see Table 1),
which is clearly above the TLV (5 ppm; Table 2). NO2 is a very toxic gas produced in
the combustion of cellulose nitrate. NO does not exist in atmospheric air because it is
immediately converted to NO2 in the presence of O2. Both are strong irritants, particularly
to mucous membrane, and will damage tissues in the respiratory tract by reacting with
moisture to nitrous (HNO2) and nitric (HNO3) acids if inhaled (Sumi & Tsuchiya 1971).
126
NOx are considered one of the major causes of acid rain worldwide. Their various derivates
are also categorized as GHGs, both in view of their phytotoxic effect and their role in
the formation of ozone process in the stratosphere. Studies indicate that NOx increase
susceptibility to lung bacterial infections. Extended exposure to concentrations of 1 ppm
causes symptoms in pulmonary alveoli similar to pulmonary emphysema. The principal
function of NO is that of precursor to NO2, which is insoluble and able to penetrate deep
into the respiratory system. NO2 inhibits the pulmonary defenses and has a phytotoxic
effect (Teixeira & Lora 2004).
7 Conclusion
Gas emissions from the two burning coal waste heaps nos. 76 and 83 in the French region
of Nord-Pas-de-Calais can be described, in very broad terms, as similar to those from
large-scale coal seam fires. The main emission gas is CO2 (7 %), with smaller quantities
of NOx and SO2. Products of incomplete coal combustion, namely CO and CH4, are also
present. Other hydrocarbons were not detected.
Carbonaceous shales are the main component of coal waste heaps. They contain pyrite
and organic carbon, which contribute to SO2, CO2, and CO formation during combustion.
Rock debris is thus a significant source of pollutant gas; it also needs to be considered in
calculations of CO2 emissions for climate change scenarios.
NH3 released from fire sites reacts with O2 to form NOx (NO and NO2) gases, which may
precipitate as NO3- and NO2-. Secondary minerals resulting from chemical reactions of
NH3 derivates, such as SO42-, NH4+, and NH4+-/SO42--bearing minerals, must be studied in
more detail, since they, too, can act as environmental (air, water, soil) pollutants.
O2 concentrations in a range from < 1 to 19 % were measured at both sites. This is below
the stipulations of the International Mining Law, which requires a minimum proportion of
19.5 % oxygen for safe working (Sumi & Tsuchiya 1971). The air above and around some
gas spots (vents) had to be classified as contaminated and/or poisonous – burning coal
waste heaps, in consequence, should be fenced off to prevent accidents and health hazards.
Various gases emitted from the heaps may impact directly on the well-being of people.
As the total amount of GHGs in the atmosphere accumulates and the global climate heats
up ever more noticeably, it becomes imperative to carry out direct and regular field gas
127
measurements at coal fire sites (i.e. burning coal seams and coal waste heaps) to identify
gas composition; quantify emission gases, especially GHGs; and determine the location
and extent of active combustion zones.
Precautions must be taken for those working – as scientists or miners – on or close to

burning coal waste heaps. The danger of accident and poisoning is omnipresent; suitable
working attire is an absolute must (e.g. gas masks).
Further research is required to measure the heat flux in CO2 discharge from burning coal
waste heap no. 83 in Fouquières-les-Lens. Under the Geneva Convention on Long-Range
Transboundary Air Pollution (CLRTAP), the parties of the United Nations Economic
Commission for Europe (UNECE) have been called upon to report emission data annually
to the Steering Body to the Cooperative Program for Monitoring and Evaluation of the
Long-Range Transmission of Air Pollutants in Europe (EMEP). Concordantly, the authors,
through this article, are reporting the release of noxious pollutant gases from the two
burning coal waste heaps nos. 76 and 83 in northern France.
Acknowledgements
The authors wish to gratefully acknowledge the assistance of Mr. Yves Paquette of the
Institut National de l’Environnement Industriel et des Risques (INERIS), who provided
the thermography report and photos used in this paper. Dr. Valérie Nollet of the Ecole
Supérieure de Chimie of the Université de Lille is acknowledged for advising on various
gas measurement techniques. Also, the authors would like to thank the laboratory staff of
the Université du Littoral Côte d’Opale-CCM – Ms. H. Nouali, Mrs. D.Dewaele, and Ms. A.
Vasseur – as well as Pascale Ciurlik of the Geology Department of the Université de Lille
for their help and assistance during gas measurements and data preparation. Finally, the
authors extend their thanks to Mr. Jean Maurice Planchenault, Charbonnages de France,
Service des Sites Arrêtés et de l’Environnement du Nord-Pas-de-Calais for supplying data
for coal waste heap no. 83.
128
References
ACGIH, 1971. Documentation of the Threshold Limit Values. 3rd edition. American Conference of
Governmental Industrial Hygienists (ACGIH), Cincinnati, Ohio, U.S.A.
Bell, F.G., 1996. Dereliction: Colliery spoil heaps and their rehabilitation. Environmental and Engineering
Geosciences 2(1): 85-96.
Bell, F.G., Bullock, S.E.T., Halbich, T.F.J., and Lindsay, P., 2001. Environmental impacts associated with
an abandoned mine in the Witbank Coalfield, South Africa. International Journal of Coal
Geology 45: 195-216.
Carras, J.N. and Young, B.C., 1994. Self-heating of coal and related materials: Models, applications, and
test methods. Prog. Energy Combustion Science 20: 1-15.
Carras, J.N. et al., 1999. Spontaneous combustion in spoil piles from open cut coal mines and the
emission of greenhouse gases. In Proceedings of the 2nd International Symposium on the
Operationalization of Remote Sensing, International Institute for Geo-Information Science
and Earth Observation (ITC), Enschede, Netherlands, 16-20 August 1999, 8.
Cook, B.J., 1999. Coal discard rehabilitation of a burning heap. In Rainbow, A.K.M. (Ed.), Reclamation,
treatment, and utilization of coal mining wastes. Rotterdam: A.A. Balkema, 223-230.
Cudmore, J.F. and Sanders, R.H., 1984. Spontaneous combustion of coal, mine fires, and interpretation of
analyses of mine gases. A literature review. Report no. 10. Australian Coal Industry Research
Laboratories Ltd. (ACIRL), Queensland, Australia.
De Kluizenaar, Y. et al., 2001. Modeling the spatial distribution of SO2 and NOx emissions in Ireland.
Environmental Pollution 112: 171-182.
DEP, 2001. Questions and answers on mine rescue and recovery operations following coal mine fires
and explosions. Pennsylvania Department of Environmental Protection (DEP), Harrisburg,
Pennsylvania, U.S.A. August 2001 http://www.depweb.state.pa.us/deepminesafety/cwp/view.
asp?a=1249&q=447954
Harries, J.R. and Ritchie, A.I.M., 1981. The use of temperature profiles to estimate the pyretic oxidation
role in a waste rock dump from an open cut mine. Water, Air and Soil Pollution 15: 405-423.
Havenga, W.J., 1993. Rapid gas analyses for the investigation of spontaneous combustion using capillary
gas chromatography. Talanta 40(2): 213-219.
IEA, 2000. World Energy Outlook 2000. International Energy Agency (IEA), Paris, France.
Grossman, S.L. et al., 1994. Emission of toxic and hazardous gases from open air coal stockpiles. Fuel
73(7): 1184-1188.
Kambara, S., 1993. Relation between functional forms of coal nitrogen and formation of NO. Energy
Fuels 7: 1013-1020.
Kelemen, S.R. et al., 1994. Quantification of nitrogen forms in Argonne premium coals. Energy Fuels 8:
896-906.
Khan, R.M., 1990. Modeling of oxygen chemisorption on coal char. Combustion and Flame 80: 83-93.
Kim, A.G., 1973. The composition of coal bed gas. U.S. Bureau of Mines Report of Investigations 7762.
U.S. Department of the Interior, Bureau of Mines and Safety Research Center, Pittsburgh,
Pennsylvania, U.S.A.
129
Krishnaswamy, S., Agarwal, P.K., and Gunn, R.D., 1996. Low-temperature oxidation of coal. Part I: A
single particle reaction-diffusion model. Fuel 75: 333-343.
Lefohn, A.S. et al., 1999. Estimating historical anthropogenic global sulfur emission patterns for the
period 1850-1990. Atmospheric Environmental 33: 3435-3444.
Masalehdani, M.N.N. and Potdevin, J.L., 2004. Problèmes géo-environnementaux liés à la combustion
des terrils dans les zones miniers de la région Nord-Pas-de-Calais, France. In Proceedings of
the Journées Nationales de Géotechnique et de Géologie de l’Ingénieur “Géotechnique et
Aménagement,” Lille, France, 28-30 June 2004, 209-218.
Masalehdani, M.N.N. et al., 2004. Products of oxidation and combustion within the burning coal waste
heaps in the Nord-Pas-de-Calais Coalfield, northern France. In Proceedings of the Joint Earth
Sciences Meeting (RST), Strasburg, France, 20-25 September 2004, 39.
Mondragón, F. et al., 2002. Effect of early stages of coal oxidation on its reaction with elemental sulfur.
Fuel 81: 381-388.
Mroueh, H. et al., 2005. Analysis of the stability of coal tips subjected to self-combustion. In Proceedings
of the International Conference on Urban Engineering, Lille, France, 12-13 October 2005,
299-305.
NCB, 1973. Spoil heaps and lagoons. Technical handbook. National Coal Board (NCB), London, United
Kingdom.
Nelson, C.R., 1989. Coal weathering: Chemical processes and pathways. In Nelson, C.R. (Ed.), Chemistry
of coal weathering. New York: Elsevier Science, 1-32.
Nordon, P. et al., 1979. The rate of oxidation of char and coal in relation to their tendency to self-heat.
Fuel 58 (6): 443-446.
Onargan, T. et al., 2003. An investigation of the presence of methane and other gases at the Uzundere-
Izmir solid waste disposal site, Izmir, Turkey. Waste Management 23: 741-747.
Panov, B.S. et al., 1999. On pollution of the biosphere in industrial areas: The example of the Donets Coal
Basin. International Journal of Coal Geology 40: 199-210.
Paquette, Y., 2002. In Concession de Courrières, Terrils NO. 83-100-230, de Pas-de-Calais, 3 and
annexes.
Paquette, Y., 2003. Guide du détenteur de terrils. Pulnoy, France: Les Fascicules de l’Industrie Minérale.
Rabinovitch, P.S. et al., 1989. Clinical and laboratory features of acute sulfur dioxide inhalation poisoning:
Two-years follow-up (case report). American Review of Respiratory Disease 139(2): 556-558.
Schmidt, L.D., 1945. Chemistry of coal utilization. New York: John Wiley and Sons.
Solid Waste Landfill Guidance, 1999. Landfill gas monitoring and mitigation. Technical section, Division
of Solid Waste Management, Solid Waste Management Program, Tennessee Division of Solid
Waste Management, 3 March.
Sondreal, E.A. and Ellman, R.C., 1974. Laboratory determination of factors affecting storage of North
Dakota lignite. Report no. 7887. U.S. Bureau of Mines (USBM), Washington, D.C., U.S.A.
Speight, J.C., 1994. The chemistry and technology of coal. New York: Dekker.
130
Sujanti, W., 1999. A laboratory study of spontaneous combustion: On the influence of inorganic matter
and reactor size. Fuel 78: 549-556.
Sumi, K. and Tsuchiya, Y., 1971. Toxic gases and vapors produced at fires. BRN 76. National Research
Council Canada, Division of Building Research, Ottawa, Ontario, Canada.
Teixeira, F.N. and Lora, E.S., 2004. Experimental and analytical evaluation of NOx emissions in bagasse
boilers. Biomass and Bioenergy 26: 571-577.
US EPA, 1993. Solid waste disposal facility criteria: Technical manual. EPA 530-R-93-017. United States
Environmental Protection Agency (US EPA), Office of Solid Waste and Emergency Response
(OSWER), Washington, D.C., U.S.A.
United Kingdom.
Wang, H., 2002. Thermal decomposition of solid oxygenated complexes formed by coal oxidation at low
temperatures. Fuel 81: 1913-1923.
Xu, W.C. and Kumagai, M., 2002. Nitrogen evolution during rapid hydropyrolysis of coal. Fuel 81:
2325-2334.
Yalcin, E. and Gurgen, S., 1995. Mine ventilation. Publication No. 251. Dokuz Eylul University, Faculty
of Engineering, Izmir, Turkey. (in Turkish)
Yoshimura, Y., 1999. Energy policy and support for overseas resource development. Coal and Safety
Journal 15: 1-4.
131
Multi-Temporal In-Situ Mapping of the
Wuda Coal Fires from 2000 to 2005:
Assessing Coal Fire Dynamics
Claudia Künzer1, Zhang Jianzhong1, Andreas Hirner2,

Jia Yaorong3, and Sun Yulin3
1
Institute of Photogrammetry and Remote Sensing, Vienna University of Technology,
Vienna, Austria
2
German Remote Sensing Data Center (DFD) of the German Aerospace Center (DLR),
Wessling, Germany
3
Wuda Mine Bureau, Wuhai, Inner Mongolia Autonomous Region, P.R. China
Abstract
This paper presents the results of multi-temporal in-situ coal fire mapping between the years 2000 and
2005 for all major coal seam fires in Wuda Coalfield, north-central China. A four-year mapping time series
and local information collected in 2000 and 2001 allowed detailed and quantitative statements on the
dynamics of each of the over 20 coal fire zones currently identified for Wuda. Based on these dynamics
and our field observations, we propose to classify coal fires into five different classes: (i) newly ignited
fires, (ii) accelerating fires, (iii) consistently burning fires, (iv) slowly burning out fires, and (v) extinct
fires. Mapping techniques combined wide-spread and numerous in-situ thermal point measurements,
crack and fissure mapping, vegetation density mapping, and integration of additional parameters such as
exhalation minerals, pyrometamorphic rock, ash layers, and underground mine workings information. The
annual changes in coal fire dynamics observed in between the field visits were much larger than expected.
For protection of valuable coal resources in the syncline, fast and prioritized extinction activities are
imperative.
摘要
本文介绍了中国北方地区中部的乌达煤田中主要煤田火区从2000年到2005年的多时相现场煤火测
绘方法。根据连续四年的测绘以及2000年和2001年采集到的现场数据，对目前已识别的乌达煤田
20多个火区带中每个火区都做了年动态变化的详细定量描述。基于这些动态信息和我们的野外试
验，我们认为可以把火区分为五种不同的类型：（1）新生火区,（2）加速燃烧火区，（3）持续
132
燃烧火区，（4）缓慢熄灭火区和（5）熄灭火区。煤火测绘方法集成了广泛和大量的现场热点测
量、裂缝和裂隙测绘、植被密度测绘图，以及其他诸如喷发物、热变质岩、灰层以及矿井开采信
息等参数。正像在野外考察期间所见，煤火的年动态变化比预想的要显著。为保护地下宝贵的煤
炭资源，快速和有效的灭火方法是十分必要的。
1 Coal Fires in Wuda Coalfield
The north-south striking Wuda Coal Mining Syncline, located 5 km west of Wuda City,
has a spatial extent of 35 km² and holds an expected total of 630 million t of coal. The
coal layers originate from the Upper Carboniferous and Lower Permian (Dai et al. 2002).
Mineable reserves are stated to be 27 million t, with a seam thickness varying from 1.5
to 12 m. The coal layers are embedded in different layers of coarse to fine-grained, white
to dark grey or yellowish sandstone and grayish-, brown-, or green-yellowish shale with
varying organic content, which also originate from the Carboniferous and Permian.
Occurring formations are the Taiyuan, Shanxi, Xiashihezi, and Shangshihezi Formations;
the Taiyuan bearing the major part of the coal (Dai et al. 2002). The strata dip to the
east with an average angle of 6 to 10°, but steeper dipping angles may occur locally.
Quaternary alluvial layers of silt, sand, and gravel irregularly cover small parts of Permian
and Carboniferous outcrops (Künzer 2005).
In the syncline, coal has been mined under the authority of the Wuda Mine Bureau
in three coal mines operating since 1958 – Wuhushan in the south, Huangbaici in the
east, and Suhaitu in the northwest. In these three mines, three different qualities of
coal – fat coal, cooking coal, and steam coal – are extracted from the 17 mineable of
the area’s 24 coal seams. The upper seven coal seams from Permian times belong to
the Shanxi Group; they have an overall thickness of 88 m and are characterized by a
low sulfur and high ash content. All lower seams from the Carboniferous belong to the
Thairen Group and have a high sulfur but low ash content (Jia 2002; Sun 2003; Künzer
2005).
From the beginning of mining operations to the year 2000, 120 million t of coal have
been extracted in the Wuda syncline. Mechanized longwall mining continues to be the
preferred method. The average mining depth of state-run or commercial mines is around
100 m, but can reach up to 200 to 300 m. The first coal fires in the area were discovered
in 1961. Before 1989, most fires were isolated and scattered in different locations. These
fires started to connect gradually between 1989 and 1995. The process is accelerating from
year to year. The amount of coal burned in the Wuda fires is estimated at 200,000 t yr-1.
133
It is assumed that so far over 1 million t of coal have been destroyed or made inaccessible.
In 2002, 8.8 % of the three coal mines in the syncline were affected by coal fires, over an
area of altogether 3.07 km² (Jia 2002).
The Wuda fires are monitored by mine engineers and mine administrators, and have been
numbered for common reference. In 2000, the Wuda Mine carried out a general assessment
of the boundaries of the coal fires through satellite data, GPS field measurements, and
geological analysis. Thermal measurements and borehole drillings supported these
activities. The result was a map of Wuda syncline with 16 demarcated coal fire zones
(FZs).
Figure 1: Underground coal fires in Wuda Coalfield. Heat is transferred through vents
and cracks in the overlying bedrock. Sudden collapse of bedrock due to volume
loss underground is common.
2 Coal Fire Mapping Techniques
Coal fire mapping in Wuda is a complex task. Most coal fires are underground fires – that
is the object of interest cannot be mapped directly. Indirect parameters, acting as indicators
of the situation underground, need to be determined. A coal fire map can therefore
never precisely reflect the fire outline; it can only approximate the burning conditions
underground. The following parameters were incorporated in the multi-parameter coal fire
mapping approach over the last five years.
First coal fire mapping of 2000

The first coal fire mapping conducted by Wuda Mine in 2000 served as the basis for
mapping efforts in the following years. This initial mapping combined the knowledge of
134
many local experts, who had been working in Wuda Coal Mining Area for decades. The
advantage of such “indigenous” knowledge is that it includes information not accessible
to the non-local researcher – information about very precise low-scale maps for example,
or insights from underground mine observations, interviews with local miners, and those
people working in the Wuda syncline every day.
Temperature measurements at vents and cracks

A thermal radiometer, “Raytek MX-2,” was used to measure the temperature of hot gases
emitted from vents and cracks in the surface bedrock. Temperatures for every crack along
the fire outline, for all cracks within the FZ, and for cracks outside the known area of a
fire were recorded. This way, a net of temperature measurements could be woven along
the fire outline, beyond the FZ, and into new FZs. Depending on the size of the coal fire,
temperature measurements were conducted for three to several hundred cracks. In 2005,
the gas temperatures at the measured cracks varied between 45 °C (therefore hardly
distinguishable from the background temperature of sun-heated bedrock) and 890 °C. If
cracks with hot emission were observed outside known FZs, they were recorded and the
adjacent area was also investigated. This mapping technique, leading from one hot vent/
crack to the next, made it possible to detect completely new fires or strong changes in a
given fire’s extent.
Temperature measurements – Conductive heat

In addition to crack temperatures, surface temperatures were measured in regular
grids. Some coal fires – especially when located under heavily weathered substrate
like shale – will not create distinct cracks in the surface bedrock. Also, very deep
coal fires may not necessarily be visible through cracks or hot gas emissions at the
surface. For such fires, regular temperature measurements of the surface bedrock can
give an indication of the situation underground. Temperatures slightly higher than the
surrounding bedrock may be a sign of conductive heat stream – an observation which
can easily be confirmed by digging a few centimeters into the ground. Under normal
conditions, at daytime in summer, the soil temperature at shallow depth will be lower
than at the surface. If the temperature is higher, this is a clear indication of a fire burning
underground.
135
Figure 2: In-situ thermal camera image superimposed on a panoramic picture of FZ 7.

Areas of unchanged background temperature are cast in blue. White, reddish,
and yellowish hues indicate regions where hot gasses are emitted. Slightly
elevated temperatures are marked in green; they are indicative of conductive
heat streams at the surface.
Crack and subsidence mapping

In Wuda Coalfield, cracking usually occurs as a result of mining or coal fire-related
volume loss underground. When the coal has burned out completely, an ash layer of
smaller thickness remains underground. This leads to cracks in the overburden bedrock.
Sometimes bedrock also collapses in troughs over large underground caves of burned-
out coal or along former mining shafts. Such cracks and subsidence phenomena were not
completely mapped for the Wuda area.
Crack fields are usually mapped based on high-resolution satellite data. Every year, a
Quickbird satellite image is acquired for the Wuda syncline. Images of this sensor have
a spatial resolution of 60 cm per pixel. Crack fields above coal fires and their temporal
development can be mapped on screen by manual digitization supported by edge detection
filters. Major cracks usually develop perpendicular to a fire’s moving direction. This leads
to a pattern of more or less parallel cracks, which, through its annual changes, can give an
indication of coal fire spatial movement.
Fumarolic minerals and pyrometamorphic rock observations

A local phenomenon resulting from coal fire-induced heating is pyrometamorphic rock.
These are rock strata that have been baked or heated by coal fires and, in consequence,
changed their texture and/or color in oxidation and dehydration processes. In the Wuda
syncline, pyrometamorphic rocks occur only scattered and over very small areas.
136
Another indicator of coal fire activities are fumarolic minerals along cracks and vents –
salmiac, sulfur, gypsum, and calcite, for example. On rare occasions, selenium; oxides
like hematite, arsenolite, cassiterite, and downeyite; as well as halides like mascagnite or
tschermigite may be found (Lapham et al. 1980; Zhang 1996; Guan & van Genderen 1997;
Zhang 1998; Zhang et al. 2004).
Figure 3: Fumarolic minerals (salmiac) formed at an exhalation hole in soft bedrock

(left). Pyrometamorphic rock altered in color and texture by coal fire heating
(right).
Vegetation density observations

During the first field trip to Wuda Coalfield, it was plain for everyone to see that
underground coal fires have a decisive influence on vegetation growth. Elevated
temperatures below surface and especially the highly toxic fumes released from
fires and discharged through fractures and cracks lead to plant damage and death.
To quantify these observations, vegetation density was mapped above three coal
fires (FZs 6, 8, and 12) as well as in unaffected adjacent areas in 2002 and 2003. In
reaction to an announcement of the Wuda Fire Fighting Team and mining engineers
that FZ 8 had extended to the north between September 2002 and September 2003,
the vegetation grid above this fire was remapped in September 2003. Already during
mapping, it was apparent that the area with degraded and dead vegetation had extended
northward too. Details of this study are presented in Künzer and Voigt (2003).
137
Figure 4: Degradation of vegetation due to toxic gas and heat discharge from coal fires
Coal fire mapping, after all, is a highly interactive process. Mapping accuracy strongly
depends on the experience of the scientist and the access he or she has to information
such as high-resolution satellite data, underground maps, and knowledge of local miners.
The combination of these factors in in-situ mapping leads to individual coal fire outlines,
displayed as so-called coal fire polygons in common GIS software. Nevertheless, it is
well-nigh impossible to map the situation underground precisely – the fires themselves are
inaccessible. Thermal camera images taken in situ are not sufficient either, for they merely
show the heat pattern at the surface. The same applies to thermal remote sensing data,
which are much less accurate than those from in-situ thermal camera imaging and can only
assist in locating the hottest coal fire areas. This clearly shows in Figure 5.
138
Figure 5: Wuda syncline as seen in panchromatic (black and white) 60 cm high-
resolution Quickbird satellite imagery. Dark areas indicate coal accumulations,
outcropping coal seams, and roads covered with coal dust. The outskirts of the
city of Wuda can be seen in the very east. In the northwest, the first dunes of
the Badain Jaran desert are discernable. The bright surface in the center of the
image is a plateau consisting of white sandstone. The superimposed polygons
in blue and yellow show individual coal fire outlines as mapped in 2003. Blue
regions represent temperatures below 150 °C; yellow hot spot areas are well
above 150 °C, soaring as high as 900 °C in some places. Note the dashed red
and white objects: these are thermal anomalies extracted from 60 m resolution
thermal Landsat satellite data. Only the hottest (yellow) areas of coal fire
activity can be relocated in these thermal satellite images of low resolution.
139
3 Results: Coal Fire Dynamics in Wuda Coalfield
The changes in coal fire dynamics observed between the annual field visits were much
larger than expected. To group the dynamics of the numerous Wuda fires, we propose a
classification system based on burning intensity. Doing so requires some knowledge about
the fires’ burning states in previous years. We propose to classify the coal fires into:
1) Newly ignited fires (NF),

2) Accelerating fires (AF),
3) Consistently burning fires (CF),
4) Slowly burning out fires (SF), and
5) Extinct fires (EF).
This classification represents the typical life cycle of a coal fire. Of course, variations may
occur, e.g. when a slowly burning out fire re-accelerates due to ignition of a seam close by.
Also, some fires were observed to fluctuate in regular intervals between the accelerating
and consistently burning states.
Newly ignited fires are characterized by medium to high temperatures (ca. 80 to 300 °C)
accompanied by steam development on a freshly cut or exposed coal seam with smoldering
surface. Indicators like fumarolic minerals, tar, pyrometamorphic rock, or vast areas of
deteriorated vegetation do not exist. Also, subsidence has not occurred yet. The fires’
extent is small, and they can still be controlled relatively easily by appropriate immediate
action. Newly ignited fires do not pose major dangers to mapping staff.
Accelerating fires show high to very high temperatures (200 to > 800°C); smoldering and
steam development are intensified. Fumarolic minerals begin to form around vents and
cracks. Vegetation shows first signs of degradation, and subsidence phenomena may occur.
Accelerating fires are usually easy to locate through their gas emissions and “visible”
streams of hot air from the surface. They can be mapped without great difficulty, but high
temperatures on the ground require caution.
Consistently burning fires usually show all surface indicators described in Chapter 2. Various
exhalation minerals have formed around cracks and vents, subsidence has occurred in many
places, and all vegetation above the fires has died. Depending on a given fire’s temperature,
pyrometamorphic rock may form. Consistently burning fires are very dangerous in terms
of mapping because large-scale subsidence may occur unexpectedly over burned-out parts
140
of the seam. Some cracks may discharge gases well over 500 °C, and surface temperatures
may be elevated due to conductive heat.
Slowly burning out fires pose the greatest threat to mapping staff. While temperatures and
steam development at the surface are moderate (50 to 200 °C), the situation underground is
marked by extensive caves and tunnels, as well as fluffy ash layers, which render the surface
very unstable. The weight of one person may be enough to cause the roof over a seam to
collapse – a seam which has turned into powdery and yet very hot ash during combustion!
Naturally extinct coal fires cannot be traced by elevated temperatures at the surface. While
fumarolic minerals, pyrometamorphic rocks, and surface sinks and collapses remain as
visual markers, vegetation density is on the rise again. Sudden subsidence, however, is still
possible. What is more, extinct fires may only be dormant. If only a tiny part of the seam
is smoldering or if enough heat is stored in the neighboring rocks, residual coal may flare
up or ignite adjacent seams. The development of the current 20 coal fires in Wuda over the
study period is presented in Table 1.
Table 1: Coal fire dynamics in Wuda Coalfield over the past five years
Fires/ Yeara
FZs 2001 2002 2003 2004 2005
1 CF SF SF EF EF
2 CF CF SF EF EF
3 CF CF SF SF CF
4 CF CF SF EF EF
5 CF CF SF AF EF
6 CF CF AF CF CF
7 CF CF CF CF SF
8 AF AF AF AF CF
9 SF SF EF EF EF
10 SF SF SF SF SF
11 SF SF CF SF AF
12 CF AF CF AF CF
13 CF CF CF AF CF
14 SF SF SF AF AF
15 SF SF EF EF EF
16 SF SF SF EF EF
17 Not existent Not existent Not existent NF AF
18 Not existent Not existent NF AF CF
19 Not existent Not existent Not existent NF AF
20 Not existent Not existent NF AF AF
a
NF = Newly ignited fire; AF = Accelerating fire; CF = Consistently burning fire; SF = slowly burning
out fire; EF = Extinct fire
The information on the fire situation in 2001 was sourced from Wuda Mine; it was only afterwards that
the DFD-DLR field team started their field investigations.
141
From the five-year mapping effort, the following conclusions can be drawn. FZ 1 – still
active in 2001 – burned out and showed no activity from 2004 onwards. The area around
the fire is not expected to re-ignite. The same applies to FZ 2, even though two new fires
– FZs 19 and 20 – developed northwest of this fire in 2005. Since 2002, FZ 3 has been
decreasing in burning intensity; it is shifting to the southeast, as can also be seen in Figure
6. While its overall spatial extent is shrinking, the burning intensity of the hottest areas
remains stable.
Figure 6: Development of FZ 3. The fire is shrinking, with its hot spots stable at high
temperatures. A new hot spot occurred southwest of the fire in 2004. The fire’s
intensity slightly decreased in 2005. FZ 1 (still burning in 2002) is also present
in the right image (dark orange); it was detected in thermal satellite imagery
as represented by the dashed red and white objects indicating thermally
anomalous pixels.
FZ 4 slowly burned out in 2003, and was neither detected nor mapped in the following
two years. Likewise, FZ 5 failed to be detected in 2005, while it still showed increasing
intensity and some spatial advances in 2004. It is rather untypical of a coal fire to burn
out not in a slow, regular process but with strong activities shortly before extinction.
FZ 6 was consistently burning during the observation period with only slight changes in
intensity. The fire is located under thick sandstone and shale deposits relatively deep below
the surface. Hence, surface temperatures are not very high and mapping required very
142
close observation. Increasing temperatures could only be measured by digging into the
weathered surface layer. Especially in the southwest of the fire, new hot spots developed;
they are progressing as indicated in Figure 7.
Figure 7: Development of FZ 6. The fire decreased in extent from 2001 onwards.
Currently, its extent is stable. However, new hot spots have developed in the
southeast. Note in the lower right picture how the hot spot extends in size from
2003 to 2004 and how the fire moves southeastward along the underground
seam.
FZ 7 has been burning consistently over the past five years and only in 2005 showed signs
of decreasing intensity. The fire, ablaze underneath a highly fractured surface, is expected
to burn out completely within the next months. FZ 8 is the largest fire in the Wuda syncline
and also one of the hottest and fastest changing fires. Stretching over 5 km from northwest
to southeast (see Figure 5), it has advanced north-northeastward recently. Here, a whole
coal seam and its neighboring underlying seam are burning. The fire was the target of
a fire fighting campaign, but is still not under control. FZs 9 and 10 are smaller fires in
decline. FZ 9 was not detected before 2004; the overall activity of FZ 10 is moderate. At
the opposite end is FZ 11; its intensity increased greatly during the last year, and the local
143
miners are fighting hard to control the fire by digging trenches, trying to separate burning
from unaffected parts of the coal seam. The fire has meanwhile extended to the east.
FZs 12, 13-1 and 13-2 – the latter two merging in 2002 – have been burning consistently.
Their intensity is slightly decreasing. However, especially in the area of FZ 13, the
ground is extremely prone to collapse and subsidence, this mainly due to underground
burning, unstable soft overlying bedrock, and strongly undulating terrain. FZ 14 can be
considered the strongest fire in the Wuda syncline for the moment. The fire is eating its
way to the northeast and has doubled in length between 2001 and 2005. Closely located
mine workings and a very thick coal seam make this fire a number one candidate for rapid
extinction. Temperatures measured in 2005 well exceeded 700 °C in several cracks and
vents. While FZs 15 and 16 are small fires not showing any activity in the past years, a
new fire, FZ 17, developed west of FZ 14. This fire is still small in extent but shows an
accelerating burning intensity. We expect it to merge soon with FZ 14, thus making the
extinction of FZ 14 an ever more urgent task! The very hot fire detected in 2003, named
FZ 18, caused the collapse of underground mining shafts and bedrock strata, leaving the
surface completely fractured. In 2005, this fire’s burning intensity was observed to decline
– the coal reserves here are almost exhausted. The rapid expansion of FZ 18 from 2003 to
2004 is shown in Figure 8.
Figure 8: Development of FZ 18. This fire did not exist either in 2001 or 2002; it was
detected by our mapping team in 2003. FZ 18 showed a very strong burning
intensity in the first year, which led to heavy collapses as can be seen in the left
satellite image. Note the mining tunnel which has collapsed in the northern part
of the fire (white circle).
144
The following table (Table 2) contains information on the local setting of the Wuda coal
fires. The FZs were rated according to estimated danger of land subsidence and sudden
collapse, as well as accessibility. Based on these parameters, extinction priorities were
determined. All coal fires in the Wuda syncline are located next to mine workings, mine
entrances, or other infrastructure. This leads us to believe that purely natural coal fires,
independent of mining activities, do not exist in Wuda.
The danger of land subsidence was classified into low, medium, and high. In areas of
low subsidence danger, the bedrock has either already subsided or local conditions
do not allow for sudden subsidence. Areas of high subsidence danger are usually
located above accelerating, burning, and especially slowly burning out fires, where
volume loss underground has still not come to a halt or is actually increasing.
Accessibility was ranked from poor to fair to good. Poorly accessible fires can
hardly be reached by car and require hikes through difficult terrain. Fairly accessible
fires can be reached by car with some difficulties. Coal fires ranked good basically
allow for parking next to the fire.
Table 2: General information on Wuda fires and extinction priorities as per 2005.
Accessibility and collapse danger may be subject to change in the future.
Extinction priority was classified according to burning intensity, closeness to
active mine workings, and knowledge sourced from interviews with miners.
Fires/ Adjacent Danger of Land Accessibility Extinction
FZs Mine Workings and Subsidence and Collapse Priority
Seams
1 Yes Medium Poor Extinct
2 Yes Medium Poor Extinct
3 Yes Medium Fair Low
4 Yes Low Poor Extinct
5 Yes Low Poor Extinct
6 Yes Low Good Medium
7 Yes Medium Fair Medium
8 Yes Medium Good High
9 Yes Low Fair Extinct
10 Yes Medium Fair Extinct
11 Yes High Fair High
12 Yes Medium Good High
13 Yes High Good High
14 Yes High Good Very high
17 Yes Medium Fair Medium
18 Yes High Fair Medium
19 Yes High Poor Medium
20 Yes High Poor High
145
Based on the time series and four years of experience in mapping the Wuda syncline, we
would currently (as of March 2006, based on the last visit in June 2005) give FZs 8, 11, 12,
13, and particularly FZ 14 highest priority for extinction. While especially FZs 8 and 11
are closely monitored by the local Mine Bureau, FZ 14 has not received a lot of attention
so far and hence could spread continuously in the last years. Here, burning intensity was
highest of all fires in 2005.
4 Conclusion
Mapping coal fires is a highly complex process as it means tracing a phenomenon below
surface. Fire outlines can only be retrieved indirectly through the combination of thermal
point measurements, in-situ thermal camera imaging, and mapping of coal fire-induced
phenomena such as land subsidence, vegetation density, and mineral exhalations. Local
geological maps, maps of mine workings, and especially the experience of locals are of
great value to coal fire mapping. For reasons of consistency it is very important to map
with similar instruments, a common database, and possibly with experienced people, who
have been or will be mapping the coal fires over several years. Sending a different field
team each year will invariably lead to non-comparable, inconsistent mapping results. The
impact of interpretation on coal fire mapping is such that the possible influence of an error
committed by an (inexperienced) analyst can only be minimized by keeping the mapping
team as stable as possible.
All coal fires in the Wuda syncline displayed a very dynamic behavior over the past five
years. Detailed reports on each fire’s development have been supplied to the Wuda Mine, the
mining authorities in Inner Mongolia Autonomous Region, and the project partners of the
Sino-German Coal Fire Research Initiative. From four years of continuous mapping, we can
conclude that coal fires can advance by several tenths to hundreds of meters per year. While
some fires show a “stable” burning behavior over time, others ignite suddenly, burn at a very
high intensity, and die down abruptly – within one or two years after first signs of decline.
Burning velocity is mainly influenced by oxygen supply, itself a function of nearby mine
ventilation systems as well as cracks and vents in the overlying bedrock. Fires with only
few pathways for air (e.g. FZ 6) seem to burn rather consistently, while fires with sufficient
ventilation (e.g. FZ 18) burn at a high intensity for shorter periods of time.
We experienced the Wuda syncline to be an environment of constant change. Large

amounts of bedrock are moved from one place to another in the course of mining.
146
Extinction campaigns take off, then pause or stop due to lack of funding or staff. New
private wild mines spread in the neighborhood of controlled state-run or corporate mining.
Every week, new parts of coal seams are exposed to the air, and it is only a question of
time when these will ignite. The current fires, however, are by far the greatest threat to
the syncline’s unaffected coal seams. Fast and prioritized extinction activities here are of
utmost importance for resource conservation.
Acknowledgements
The authors would like to thank the local authorities and mining staff in Wuda for
answering numerous coal fire-related questions, as well as for providing information on
mining operations. Further thanks go to the Chinese and German project partners of the
Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China,” who contributed to our
understanding of the coal fire problem and the local conditions in Wuda.
References
Dai, S.F., Ren, D.Y., Tang, Y., Shao, L., and Li, S., 2002. Distribution, isotopic variation, and origin of
sulfur in coals in Wuda Coalfield, Inner Mongolia, China. International Journal of Coal
Geology 57: 237-250.
Guan, H.Y. and van Genderen, J.L., 1997. Report on environmental monitoring of spontaneous
combustion in the north China coalfields. Aerophotogrammetry and Remote Sensing Bureau
of China Coal (ARSC), Xi’an, P.R. China and International Institute for Geo-Information
Science and Earth Observation (ITC), Enschede, Netherlands.
Jia, Y.R., 2002. Personal interview / Oral communication during the field campaign in September 2002.
(Jia Yaorong is the chief engineer of the Wuda Coal Fire Fighting Team.)
Künzer, C., 2005. Demarcating coal fire risk areas based on spectral test sequences and partial unmixing
using multi-sensor remote sensing data. PhD thesis, Vienna University of Technology, Vienna,
Austria.
Künzer, C. and Voigt, S., 2003. Vegetationsdichte als möglicher Indikator für Kohleflözbrände?
Untersuchung mittels Fernerkundung und GIS. In Strobl, J., Blaschke, T., and Griesebner, G.
(Eds.), Angewandte Geographische Informationsverarbeitung XV. Beiträge zum 15. AGIT-
Symposium Salzburg 2003. Heidelberg, Germany: Wichmann, 256-261.
Lapham, D.M., Barnes, J.H., Downey, W.F., and Finkelman, R., 1980. Mineralogy associated with
burning anthracite deposits of eastern Pennsylvania. Mineral Resource Report 78.
Pennsylvania Geological Survey, Middletown, Pennsylvania, U.S.A.
147
Sun, Y.L. 2003: Personal interview / Oral communication during the field campaign in September 2003.
(Sun Yulin is an engineer in the Wuda Coal Fire Fighting Department and leads the team on
coal geology.)
Zhang, J.Z., 1996. SWIR spectra of rocks in areas affected by coal fires, Xinjiang Autonomous Region,
P.R. China. MSc thesis, International Institute for Geo-Information Science and Earth
Observation (ITC), Enschede, Netherlands.
Zhang, J.Z., Wagner, W., Prakash, A., Mehl, H., and Voigt, S., 2004. Detecting coal fires using remote
sensing techniques. International Journal of Remote Sensing 25 (16): 3193-3220.
Zhang, X.M., 1998. Coal fires in northwest China – Detection, monitoring, and prediction using remote
sensing data. PhD thesis, International Institute for Geo-Information Science and Earth
148
General Investigations of
Wuda Coal Fire Area
Liu Bokun, Guan Haiyan, Ma Jianwei, Kong Bing, Wang Mei,

Wu Xiaoying, and Xiang Yaling
Abstract
Wuda Coalfield (Inner Mongolia Autonomous Region, P.R. China) has an extension of about 35 km2.
Large-scale coal mining here started in 1958, with continuous production from 1962 onwards. Coal
output in 2004 was 4.8 million t. Spontaneous coal seam fires were reported for the first time in 1974
and spread to form entire fire zones in 1978. Remote sensing investigations revealed that there are
currently 16 major coal fire zones in Wuda Coalfield, covering about 3.5 km2. Many fires were found
to be caused by residual coal in mined-out workings. Research regarding the impact of mine roof and
floor conditions on spontaneous combustion has led to a distinction of coal fires into: (i) coal fires in
residual coal and (ii) coal fires in coal pillars in mined-out workings of large-scale coal mines.
摘要
乌达煤田面积35平方公里，1958年开始建设大型煤矿，1962年投产，2004年产煤480万吨。从1974
年开始煤层自燃，1978年形成煤田火区。经2004年遥感调查，共有16个火区，燃烧面积达3.5平方
公里。遥感调查发现大型煤矿采空区遗煤自燃形成大面积煤田地面火区，经煤矿井上下对照，大
煤矿煤田火区有两种，包括：（1）井下采空区工作面遗煤自燃形成的煤田地面火区和（2）井下
采空区煤柱自燃形成的煤田地面火区。
1 Introduction
Wuda Coalfield is the focal study area of the Sino-German Coal Fire Research Initiative
“Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires
in North China.” It is located near the city of Wuhai in Inner Mongolia Autonomous
Region, on the west bank of the Yellow River, in the northern part of the Helan Shan,
149
just south of the Ulan Buh Desert. Hilly and dry, the landscape is typical of these
latitudes; it is shaped by a droughty desert climate. Wuda Coalfield lies in the basin-like
Wuda syncline at the western edge of the Ordos Basin and belongs to the Carboniferous
and Permian of northern China. The syncline extends 12 km from east to west and 3 to
5 km from north to south (Figure 1).
Figure 1: Map of Wuda Coalfield
150
The main coal-bearing strata in Wuda Coalfield are the Taiyuan (C 3t) and Shanxi (P1s)
Formations. They include 27 (identified and numbered) coal seams with an overall
thickness of 42.8 m, including 19 exploitable coal seams with an overall thickness of
33.6 m. The coal of these seams is mostly medium volatile bituminous coal. Coal seams
nos. 9 and 10 have a very high sulfur content. After centuries of small-scale private surface
mining and manual coal digging in Wuda, modern coal mining on an industrial scale
started in 1958, and since 1962 has been running continuously. Three major coal mines are
operational in the area – the Wuhushan, Suhaitu, and Huangbaici Mines; together they had
an output of 4.8 million t of coal in 2004.
The goafs of Suhaitu Mine started to burn in 1974, propagating into six major coal fire
zones by 1978. With the number of small-scale coal pits soaring in the decade from 1985
to 1995, another 10 coal fires emerged in this and the other two mines. According to
remote sensing investigations in Wuda, the total area affected by coal fires amounted to
2 million m2 in 1995. This area increased to 3.07 million m2 in 2002 (Figure 2), reaching
3.5 million m2 in 2004 (Ma et al. 2004).
Figure 2: Wuda coal fires in 2002
2 Remote Sensing Detection of Wuda Coal Fires
Analysis of high-resolution Quickbird satellite data revealed that most of the larger
coal fires in Wuda Coalfield have been sparked off by spontaneous combustion in coal
remaining in goafs after extraction. Remote sensing investigations here heavily rely on
high-sulfur emissions and burned (coked) coal seams as surface indicators of underground
coal fires (Figure 3).
151
Figure 3: Remote sensing exploration in Wuda Coalfield. Quickbird image of mined-out

area (top left); high-sulfur coal combustion (top right); natural coke as a result
of coal fire heating/burning (bottom).
152
3 Development of Wuda Coal Fires
Coal fire development can generally be described in four successive phases:
1) Spontaneous combustion: The temperature inside the coal seam begins to increase as a
result of exothermal oxidation. Multiple combustion spots flare up when the coal’s self-
ignition temperature is reached at isolated points.
2) Formation of a combustion center: Along with the rising temperature, a compact
combustion center based on pressure-induced gas circulation (oxygen uptake and flue
gas emission) develops from the isolated combustion spots.
3) Formation of a combustion system: Propelled by increasing gas circulation and
temperature, several combustion centers combine to form a combustion system.
4) Temperature drop and extinction: After the coal in the seam has been exhausted, the
coal fire dies out gradually, leaving in its wake pyrometamorphic rock and collapsed
ground.
4 Classification of Wuda Coal Fires
In-situ observations of Wuda coal fire formation, aided by practical mining knowledge, led
to the following classification of coal fires in Wuda (Guan et al. 2004):
1) Coal fires in residual coal in mined-out workings of large-scale coal mines,

2) Coal fires in coal pillars in mined-out workings of large-scale coal mines,
3) Coal fires in small-scale coal mines,
4) Coal fires in coal seam outcrops.
153
5 Examples of Wuda Coal Fires
5.1 Coal fires in residual coal in mined-out workings of large-scale coal

mines
A good example of this type of coal fire is fire zone (FZ) 10 in Wuhushan Mine. High-
resolution Quickbird imagery revealed the existence of an extensive zone of cracks
running from east to west at the northern top border of the FZ. The direction in which the
cracks had developed was identical with a goaf of the mine system underneath; it is thus
highly probable that they were a result of spontaneous combustion in residual coal and
ensuing roof collapse. The affected coal seam no. 4 crops out at the southern end of the FZ.
The total combustion system spans over 21,000 m2. Eight burning lines and 21 combustion
centers were identified for the system, with temperatures ranging between 106 and 298 °C.
Borehole measurements yielded a seam thickness of 4.74 m. The coal fire developed along
a coal face in an abandoned working, which lies in the northern part of the center panel
of Wuhushan Mine. Mining there stopped in 1980; the fire started in 1994 (Figures 4,5,6)
(Guan 2004).
Figure 4: Quickbird image of FZ 10 in Wuhushan Mine
154
Figure 5: Dip cross-section of FZ 10 in Wuhushan Mine
155
156
Figure 6: Strike cross-section of FZ 10 in Wuhushan Mine

5.2 Coal fires in coal pillars in mined-out workings of large-scale coal

mines
A good example of this type of coal fire is FZ 8 in Wuhushan Mine. Quickbird imagery
highlights the shape of the combustion system – thin crack lines arranged in a cluster.
Altogether 10 combustion spots and eight combustion centers have been identified for this
FZ, spanning over a total area of 7,537 m2. Temperatures underground range between 110 and
800 °C. The combustion system includes coal seams nos. 9 and 10. Borehole measurements
yielded a thickness of 4.6 m for coal seam no. 9 and 3.01 m for coal seam no. 10; the
distance between the two seams is 3.62 m. The coal fire developed along coal pillars
between the coal faces of two abandoned workings. Mining there had stopped between
1977 and 1978; spontaneous combustion in the pillars began in 1994 (Figures 7,8,9)
(Zhang et al. 2004).
Figure 7: FZ 8 in Wuhushan Mine. Quickbird image (top); ground photograph (bottom).
157
158
Figure 8: Strike cross-section of FZ 8 in Wuhushan Mine

Figure 9: Dip cross-section of FZ 8 in Wuhushan Mine
159
Acknowledgements
The research behind this paper was realized as part of the Sino-German Coal Fire Research
Initiative “Innovative Technologies for Exploration, Extinction, and Monitoring of Coal
Fires in North China” and the National Program No. 863 “Research on Key Techniques
Using Remote Sensing for Detection of Underground Coal Fires.”
References
Ma, J.W. et al., 2004. Wuda Coalfield supplementary detailed investigation report. Beijing Remote
Sensing Corporation (BRSC), Beijing, P.R. China. (in Chinese)
Zhang, J.M. et al., 2004. Wuda Coal Fire Area in Inner Mongolia research report, as a contribution to
the Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China.” Beijing Remote Sensing
Guan, H.Y., Wu, C.C., and Xiang Y.L., 2004. Remote sensing detection methods for Wuda Coal Fire
Area, as a contribution to the National Program No. 863 “Research on Key Techniques
Using Remote Sensing for Detection of Underground Coal Fires.” Beijing Remote Sensing
160
Environmental Impact Assessment of
Spontaneous Coal Seam Fires in China
Shi Xiaolei, Cao Daiyong, Fan Xinjie, and Wu Chacha
Laboratory of the Ministry of Education, Coal Resources Lab,

China University of Mining and Technology, Beijing, P.R. China
Abstract
Spontaneous coal seam fires have come to pose a major problem in coal production. They not only destroy
large amounts of coal, but also cause serious environmental problems. The hazards related to coal fires
in China are listed in “China’s Agenda 21.” The present paper analyses the effects of coal spontaneous
combustion on geology, atmosphere, water, soil, ecology, and human health. At the end of the paper,
suggestions on how to mitigate the impact of coal fires are given.
摘要
煤层自燃已成为煤炭生产的主要灾害之一。它不仅浪费了大量的煤炭资源，还带来了严重的环境
问题，中国已将煤层自燃造成的环境问题列入“中国21世纪议程”。本文结合实际，就煤层自燃
灾害带来的环境效应从地质、大气、水、土壤、生态和人体健康等方面进行分析，最后提出了减
缓煤火对环境影响的建议。
1 Introduction
China has the most abundant coal reserves in the world and ranks first in coal production.
The spatial distribution of coal resources, however, is uneven; about 80 % are found in
northern China, in semi-arid to arid areas, where solar radiation is strong, coal seams are
thick, and the overburden very shallow. Exploitation through collieries there is intense,
leading to wide-spread coal fires at surface and subsurface levels where coal is exposed
to air and left to oxidize and heat up to spontaneous combustion. Typical fire areas are
coal outcrops in Xinjiang, small mine shafts in Ningxia, and mined-out workings of large-
scale coal mines in Inner Mongolia. By 1998, the number of coal fires in northern China
161
had risen to 56, covering an area of 720 km2. Coal fires eat away 10 to 13.6 million t of
coal each year in China; the total loss so far amounts to 4.2 billion t (Guan et al. 1998).
Spontaneous coal seam fires, though, have not only led to massive destruction of coal
resources; they also emit a host of harmful substances, threatening both ecological integrity
and human health in fire zones. Pollutants from coal fires have become a global problem
(Stracher & Taylor 2004). The Chinese government has listed the northern China coal fires
and related hazards in its 1994 “China’s Agenda 21” (ACCA21 1994). In April 1994, the
Chinese State Science and Technology Commission entered a new phase of cooperation
with the European Economic Community’s Scientific and Technological Development
Department when it launched the project “Environmental Monitoring of Spontaneous
Combustion of Coal Seams in Northern China” (Guan et al. 1998).
2 Environmental Impact of Spontaneous Coal Seam Fires
2.1 Impact on geological environment
The main geological impact of spontaneous coal seam fires is the generation of extensive
zones of burned rock (Figure 1). Spontaneous combustion not only burns up residual coal in
mine shafts, coal wedges, and coal seams, leaving but large empty burned-out areas below
surface; it also heats up extensively the area above the coal, where it changes the physico-
chemical properties of the overlying bedrock. Burned rock is riddled with fissures and
holds water very well – it is a perfect aquifer. These qualities are instrumental in collecting
groundwater during mining, adsorbing water and reducing perpendicular infiltration from
rock layers above adjacent mine shafts, and drying out water-retentive coal wedges for
maximum exploitation. In the Daliuta and Huojitu Mines, for instance, wells drilled into
burned rock deliver 2,000 to 3,000 m3 of water each day for use in mining works. High
permeability and retention, however, can also be a source of calamities: if the water-bearing
burned rock stratum or the fissure and fault zones connecting areas of burned rock are hit in
the course of mining, water may gush into mines and flood shafts (Hu et al. 2001).
Caves and hollows that occur underground in the wake of fires strongly destabilize the
coal seam. When exposed to external pressure or through its own gravity, their roof may
subside and collapse to form collapse pits and burned fissure clusters. In Shuijingqu
Mine, spontaneous combustion of coal resulted in four oval-shaped subsidence holes with
diameters ranging from 20 to 84 m, and depths of 10 to 30 m. Figure 1 (right) shows one
such collapse zone with burned fissure clusters in Wuda Coalfield, Inner Mongolia. Creating
ever new openings for oxygen to enter the soil and react with coal, holes and fissures
162
perpetuate the vicious circle of “combustion-collapse-combustion.” The surface structure,

moreover, becomes more complex, which affects regional microclimate. The cracked surface
accelerates processes such as wind and water erosion, soil desertification, and soil and water
loss. Under heavy rainfall, coal fire areas can be the stage for hazardous debris flows.
b
b
Figure 1: Burned rock (a) and burned fissure clusters (b) as a result of spontaneous
combustion in Wuda Coalfield
163
2.2 Impact on atmospheric environment
Spontaneous coal seam fires produce great quantities of emission gases, which leave
the surface through fissures, collapse holes, and small mine shaft pitheads as a result of
convective air flow (Figure 2). According to Guan et al. (1998), coal fires in northern China
discharge into the atmosphere each year: 490,200 t of carbon monoxide (CO), 514,700 t
of sulfur dioxide (SO2), 300,000 t of nitrogen dioxide (NO2), 112,000 t of dust, as well as
other harmful gases such as carbon dioxide (CO2) and hydrogen sulfide (H2S). Liuhuanggou
Coal Fire Area in Xinjiang alone sets free about 1,080,000 m3 of various harmful gases each
1
year. With an air aggregative pollution index of 5.905, it ranges among the country’s most
heavily polluted areas. A steadily blowing wind from southwest drives the smoke and dust
to Urumqi, the capital of Xinjiang Autonomous Region, where the national standards for
suspended particles, SO2, and nitrogen oxides (NOx) are constantly exceeded in consequence.
During up to sixteen weeks in winter and spring, pollution hovers about grade IV of the air
aggregative pollution index, and may even reach grade V at peak times (Liu 2002). In Wuda
Coal Fire Area, the air concentrations of SO2, NO2, and CO in the immediate fire zones are
3.1, 4.1, and 5.1 times higher than the national standard; they are 1.5, 2.2, and 5.3 times
higher in residential areas of the same region (Table 1).
Table 1: Gas discharge from Wuda coal fires

Species SO2 NO2 CO
(mg m-3 d-1) (mg m-3 d-1) (mg m-3 d-1)
Site Std. Meas. Diff. Std. Meas. Diff. Std. Meas. Diff.
Fire area 0.25 0.78 3.10 0.15 0.62 4.10 0.0006 0.03 5.l
Residential area 0.25 0.38 1.52 0.15 0.33 2.20 0.0006 0.0032 5.3
Data source: Ma (2002)
1
The air aggregative pollution index is the exponential sum of the single-item factors of various air
pollutants and serves to appraise the overall condition and the year border change of a given city’s air
quality. It can also be used for comparison of inter-urban air pollution. The higher the index, the worse
the air quality. The air aggregative pollution index (P) is calculated using the following equation:
n Ci
(1) P = ∑ pi
pi pi =
Si
,
i =1
where pi is the subindex of a pollutant i; Ci the daily average concentration per year of this pollutant in
the whole city or downtown area, or at the sample point; and Si its standard.
The grading system is as follows:

Air Aggregative >6 5-6 4-5 3-4 2-3 <2
Pollution Index
Pollution Rating Serious Heavy Medium Light Better Good
pollution pollution pollution pollution
(VI) (V) (IV) (III) (II) (I)
164
Figure 2: Fume released from collapse pits (left) and fissures (right) in Wuda Coalfield
Fume and smoke released from coal fires also influence visibility. In fire zone (FZ) 14
of Wuda Coal Fire Area, walls and columns of rising smoke can be seen everywhere;
visibility is below 5 m, which makes it very dangerous to walk the area by foot. If the
harmful gases emitted from coal fires cannot dissipate in time, they may contribute to
secondary pollution. SO2 can transform into sulfur trioxide (SO3) through photochemical
oxidation and heterogeneous catalysis reactions, and change further into toxic sulfuric
acid (H2SO4) and sulfate (SO42-). In the presence of hydrocarbons and nitrogen oxides,
the photochemical oxidation velocity of SO2 increases tenfold (Hao & Wang 2001). In
addition, burning coal produces large amounts of suspended particulate matter. Floating
dust (PM10), in particular, has harmful effects on human health and can cause lasting
damage when inhaled.
2.3 Impact on water environment
Spontaneous coal seam fires change the geological conditions of surface water and shallow
groundwater flow. As noted earlier, underground coal combustion produces extensive
areas of burned rock, collapse pits, and burned fissure clusters. These changes in rock/
surface structure cut off pathways for water infiltration and disturb soil water balance.
Groundwater in a combustion site may flow to and concentrate near a fire in a process
called “gathering.” This reduces the groundwater available in other areas, lowers the
groundwater table there, and negatively affects vegetation growth. The greatest impact
spontaneous combustion may develop on water cycles is when a river flows through a
fire zone. Fissures and the generally lowered water table around a coal fire reduce water
flux. For example, four months of continuous observation in Baiyanghe Mining Area,
Xinjiang, revealed that the Baiyang River lost up to 0.578 m3 s-1 or 18 % of its total flux
when passing over burned rock (Hu et al. 2001). Acid and alkaline compounds produced
during combustion of metallic and non-metallic elements in coal can pollute surface water
165
and shallow layers of groundwater; they turn sweet water with heavy carbonic acid content
into bitter water saturated with sulfides and/or chlorides.
2.4 Impact on soil environment
Soil is generally made up of mineral and organic matter, microorganisms, soil moisture,
and soil air. Mineral and organic matter are important indicators of soil fertility.
Microorganisms feed on organic residues in the soil; they fix nitrogen and offer nutrients
to plants. Soil moisture and air have a great impact on plant growth. Coal fires change the
physico-chemical properties of soil. Yellow burned soil is omnipresent in coal fire areas
such as Wuda. It is fragile and prone to crumbling, with very low organic matter content
and only small numbers of microorganisms. As a result, nutrient recycling is virtually
inhibited and fertility drops, making it more difficult for plants to grow. In addition, sulfur,
mirabilite, and coal tar – by-products of the combustion process – are deposited in and on
the soil, and aggravate its pollution (Figure 3). The ubiquitous fissures increase surface
evaporation, so that upper soil layers in fire areas are often very dry, and thus vulnerable to
wind erosion and desertification. Wuda Coal Fire Area, for instance, has already entered a
state of semi-desertification.
Figure 3: Sulfur (left), mirabilite (middle) and coal tar (right) on the surface above coal
fires in Wuda Coalfield
2.5 Impact on ecological environment
High temperatures and various kinds of pollutants wreak havoc on the environment in
coal fire areas. Underground coal combustion in northern China releases 7.385 × 1013 kJ
of thermal energy each year (Guan et al. 1998). This energy raises the surface temperature
and forms extensive “heat islands,” which can destroy vegetation cover, e.g. by forest or
meadow fires. The surface temperature of Liuhuanggou Fire Area is in many parts above
100 ºC; the color of the rock has changed to brick-red there; and the grasses formerly
covering the area have already burned over 2,200,000 m2.
166
Coal seam fires also reduce biodiversity (Whitehouse & Mulyana 2004). In comparison
to areas in Wuda Coalfield that have so far been unaffected by spontaneous combustion,
the landscape “on fire” has degraded to a semi-desert or desert (Figure. 4). Noxious
substances released during combustion take their toll on the environment around the fire
zones. The content of benzopyrene in poplar leaves of Liuhuanggou Coal Fire Area is 130
times higher than in unaffected areas; dust, sulfur-, and nitrosamines block respiration and
photosynthesis over the leaves (Liu 2002). Pollutant delivery has expanded the area of acid
rain and ecological deterioration.
Figure 4: Vegetation cover in fire area (left) and unaffected area (right) in Wuda Coalfield
2.6 Impact on human health
Pollution and environmental degradation in coal fire areas seriously threaten the health and
life of local residents. Visitors to Wuda Coal Fire Area will find the air hazy and infested
with a sulfurous smell of rotten eggs – a smell which the wind easily carries over into
adjacent urban areas. Gas monitoring performed by the local Environmental Protection
Department went to show that the concentrations of H2S, SO2, and other noxious gases
in urban areas of Wuda are twice as high as the national standard. Long-term exposure to
these gases may severely damage the respiratory system. The number of people suffering
from respiratory diseases has more than doubled in Wuda Coal Mining Area recently; the
proportion of residents who suffer from rhinitis is as high as 70 to 80 % near Liuhuanggou
Coal Fire Area. Coal tar is a respiratory carcinogen identified by the International Cancer
Research Institution of the World Health Organization (Wei 2001). Arsenic contained
in coal fire discharge can cause chronic intoxication with after-effects such as pigment
deficiency, over-pigmentation, keratosis, and skin cancer. Fluorine can lead to fluorine
osteopathy, and even impair people’s ability to work (Finkelman 2004). As a last resort,
people have to be relocated to unaffected areas. Abandoned houses are a frequent sight
around Wuda Coal Fire Area.
167
3 Environmental Countermeasures
The spread of spontaneous coal seam fires is massively destructive, both in view of coal
resource management as well as environmental protection. This makes fire control an
even more urgent task in China. With due regard to the country’s present condition, some
suggestions for environmental improvement are proposed in the following:
1) Strengthen mine management and set reasonable coal production targets. Tightly
control the number of small minings and restrict where possible. Prohibit excessive and
arbitrary coal digging. Increase coal recovery. Prolong mine service life and apply the
existent Laws on Mineral Resources Conservation and Environmental Protection more
conscientiously.
2) Pay more attention to fire prevention. Assess the potential of coal for spontaneous
combustion. During pre-mining, mining, and post-mining works, do not be negligent
about precautionary measures, i.e. building windbreaks, filling goafs with inert gas, or
spraying fire-retardants. These measures can help prevent coal spontaneous combustion;
if a fire sparks off nonetheless, initiate immediate efforts for extinction.
3) Stimulate and foster technological research on fire extinction, e.g. on complex colloid
material, organic high-molecular polymer filling, and inert gas injection (Hu et al. 2003;
Tian 2004). Set up a coal fire information system for monitoring fire prevention and
extinction campaigns, as well as destruction and recovery of resources in real time (Mao
et al. 1997; Zhang et al. 1998).
4) Restore the environment and reduce fire impact in coal fire areas. Mines and their
management ought to be urged to develop forestation and water supply plans, as well as
water-saving irrigation techniques. Vegetation recovery and biodiversity protection need
to be enhanced. Further ecological deterioration must be avoided at all costs!
5) Strengthen international exchange and cooperation on coal fires to broaden the financial
base for fire fighting and introduce advanced technologies to China: this would double
the effectiveness of measures.
168
Acknowledgements
The research behind this paper was supported by the National High Technology Research
and Development Program of China (No. 2003AA131100-02). The authors wish to thank
the following individuals for providing documents and information necessary to complete
this paper: Zhang Jianmin of Shenhua Group Corporation Limited; Guan Haiyan, Ma
Jianwei, and Xiang Yaling of Beijing Remote Sensing Corporation (BRSC). Special thanks
go to the Department of Prevention of Coal Fires of Wuda Mine Ltd., Shenhua Group for
providing information and data on geology, coal fires, and mining engineering in Wuda
Coalfield.
References
ACCA21, 1994. China’s Agenda 21 – White paper on China’s population, environment, and development
in the 21st century. Administrative Center for China’s Agenda 21 (ACCA21), Beijing, P.R.
China.
Finkelman, R.B., 2004. Potential health impacts of burning coal beds and waste banks. International
monitoring of spontaneous combustion in the North China coalfields. Beijing: Coal Industry
Hao, Q.J. and Wang, Q.C., 2001. Effects of sulfuric pollutants from mining and combustion of coal on the
soil environment in China. Environmental Protection of Xinjiang 23(2): 34-38. (in Chinese)
Hu, S.J., Yang, J.B., and Li, Z.M., 2003. Science and technology in the prevention of coal seam fires. Coal
12(2): 40-41. (in Chinese)
Hu, S.R., Jiang, D.C., Li, Z.G., and Jia, Y.P., 2001. Fires in coalfields and coal mines – Environmental
effects and prevention. Journal of Geological Hazards and Environment Preservation 12(1):
21-23. (in Chinese)
Liu, L.C., 2002. The impact of coal seam fires and how to control them. Coal Geology of China 14(2):
83-85. (in Chinese)
Ma, J.W., 2002. Detailed survey report of Wuda Coalfield in Inner Mongolia Autonomous Region. Beijing
Remote Sensing Corporation (BRSC), Beijing, P.R. China. (in Chinese)
Mao, Y.B., Peng, W.X., Wang, Y.Q., Kang, G.F., Wu, J.H., Ma, H.P., and Lei, X.W., 1997. Coal fire
monitoring information system in Xinjiang, China. Remote Sensing for Land and Resources 1:
37-43. (in Chinese)
recipe for environmental catastrophe. International Journal of Coal Geology 59: 7-17
Tian, R.Y., 2004. Calamities caused by spontaneous combustion of shallow coal seams and their control
in Wuda. Coal of Shanxi 2: 11-12. (in Chinese)
169
Wei, J.G., 2001. Mine environment engineering. Beijing: Metallurgy Industry Publishing Company. (in
Chinese)
Whitehouse, A.E. and Mulyana, A.A.S., 2004. Coal fires in Indonesia. International Journal of Coal
Geology 59: 91-97.
Zhang, J.M., Ning, S.N., and Cao, Y., 1998. Study of environmental effects and measures to prevent
spontaneous coal seam fires in northern China. Natural Disaster Reduction in China 8(1):
34-38. (in Chinese)
170
Coal Fires as a Social Problem:
An Anthropological Perspective
Kathryn A. Norman
Bodega Bay Consulting, San Francisco, California, U.S.A.
Abstract
Underground coal fires are recognized to be a global problem. Yet few people are aware of the extent
and the long-term effects of these fires. Spontaneous combustion and other natural ignition sources
have resulted in coal fires since prehistoric times. Since the industrial revolution, human activities have
proliferated their scale and severity. The effects have been devastating.
In addition to depleting energy resources, coal fires release large quantities of greenhouse gases.
Pollutants vary from fire to fire, but overall the aerosols and toxic gases released are calamitous. These
have disastrous effects not only on the atmosphere, but on soil and water, and cause human diseases,
physical and mental malfunctioning, and death. Human beings not only are part of the cause of coal fires,
they are directly affected. They must contend with the effects of coal fires – social, physical, and mental
health issues, as well as a damaged and destroyed ecosystem.
Almost always governments and industries focus on the physical nature of underground coal fires ignoring
the interrelated human issues. Financial resources are directed toward designing scientific and engineering
tools to map and extinguish fires. The human dimension is recognized when natural disasters, such as
tsunamis, hurricanes, and earthquakes, have immediate emergency impact. Over a longer time scale, coal
fires have similar negative effects on human mental and physical health, life styles, and well-being. Yet
worldwide, government and corporate research and innovation policies related to underground fires focus
on technological issues, not on related social and psychological concerns.
This paper looks at one underground fire in the United States. Community discord and distress resulted
when the socio-psychological issues were more or less ignored by government policy and planning.
In contrast, the study of the Chinese coal fire regions of Ningxia and Inner Mongolia is reviewed as an
example of a project in which local residents have been included as part of gathering data. This paper
is not designed to offer a solution toward the problems of extinguishing fires. Rather it calls attention to
related human issues. One avenue that has been shown to be of benefit is for officials and researchers to
incorporate local residents in research, planning, and implementing programs related to underground coal
171
fires. Residents may not understand the technical aspects of managing fires, but they are well aware of
their own behavior, values, needs, and concerns.
摘要
地下煤火已经被公认为是全球性问题，然而，却很少人知道它的范围和长期的影响。自史前时
代，煤自燃和其他天然火源就产生了煤火。自从工业革命以后，人类的活动扩大了煤火的规模和
严重程度，这种影响是具有破坏性的。
除了消耗大量的资源外，煤火还释放了大量的温室气体。煤火污染各地是不一样的，通常火区的
烟雾和释放的有毒气体都是有害的。煤火不仅对大气环境，而且对水和土壤会产生严重的影响，
引起人们的疾病、身体和精神上的异常和导致死亡。人类不仅是引起煤火的部分原因，也直接受
到煤火影响。因此，人类必须要与煤火对社会、人类身体健康和精神健康以及破坏和毁灭生态系
统等方面的影响做斗争。
政府和工业界几乎总是关注于地下煤火本身，而忽视了与煤火相关的人类影响问题。他们的资金
主要用于设计科学的工程方法进行煤火测绘和灭火工程。只有当海啸、飓风和地震等对于人们有
直接突发作用的灾害影响时，才关注到对人类的影响。从长期影响考虑，煤火对于人们的精神和
身体健康、生活方式和寿命都有类似的负面影响。然而，全球与地下煤火有关的政府、合作研究
和改革方面的政策都注重于技术问题，而忽略了相关的社会与心理方面的内容。
本文研究了一个美国的地下煤火实例。当政府的政策和规划中几乎无视社会心理问题时，引发了
社区的不满和危机。相反，在中国宁夏和内蒙火区的研究示范项目中，数据采集则包括了当地居
民的情况。本文并不有意提出一个灭火问题的解决方案，而是要引起对相关人类问题的关注。政
府和研究人员要在与地下煤火有关的研究、规划和实施计划中与当地的居民和企业相结合，是一
种有益的途径。他们可能对于煤火管理中技术方面的问题不了解，但对于他们自己的行为、价
值、需要和利害关系是十分清楚的。
1 Introduction
“Coal fires are a global catastrophe. For most people who don’t live near one of these
fires, it never reaches them. There may be a little clip in the newspaper, but most
people aren’t aware of the extent of the problems involved in these fires.”
Glenn Stracher, geologist, East Georgia University; quoted in Business Day article
on 2003 Meetings of the American Association for the Advancement of Science
Geologic evidence has shown that spontaneous combustion and other natural ignition
sources, such as lightening and forest fires, have produced coal seam fires since prehistoric
times (e.g. Coates & Heffern 2003). Since the industrial revolution, human actions have
proliferated the scale and severity of underground fires. With the desire for increased
energy and a demand for more coal, humans developed opencast and deep coal mining,
and the coal mining industry developed on a large scale. Mining accidents, drilling and
172
exposing of new seams, deforestation, and planned agricultural fires have been the genesis
of underground fires. Today underground coal fires are actively burning around the world.
Whether triggered by spontaneous combustion, other natural causes, or human actions,

underground coal fires are devastating. These fires create air pollution that is toxic to
people and produces hazards of cave-ins. In addition, they consume large amounts of
resources with no economic return and add to atmospheric CO2. Because underground coal
fires have a significant effect on resources, environment, and human ecology and health,
they have a large social and societal impact.
The effects of underground coal fires are global, but many people are unaware of the
magnitude and effects of fires they cannot see. However, as the following examples show,
those who do live near such fires are well aware of the local negative impact.
1.1 An example from South Africa
The children of Mpumalanga, South Africa have a daily routine filled with fear. As they
walk to school they hold cloths to their noses to avoid breathing in toxic dust and gases
from the underground fires over which they walk. They pick a path across land that was
abandoned by a coal mining company. Subsidence has resulted in land cave-ins that have
been known to swallow up children. Roads nearby have swallowed cars and buses in cave-
ins (Bauer 2003).
1.2 An example from India
In one of India’s most important coalfields, the Jharia mining complex, the walls of
opencast mines have been on fire since 1916. They now “hiss and flow like lava flows”
(Revkin 2003). This is in a densely populated region. The 150,000 miners, truck drivers,
train loaders, and others stoically work with the disturbing constant backdrop of orange
flames and brown smoke. In 1995, the walls of one complex collapsed, flooding pits
and tunnels with fire and killing more than 60 miners. Residents and miners know the
danger. In 2003, Andrew Revkin reported: “[The] fires there still burn, and residents and
mineworkers continue to adapt. In places where the ground cracks and slumps and smokes,
people simply dismantle brick homes and move them somewhere else.” Whole villages
have been relocated and re-relocated. The low-sulfur coal that is used to produce coke for
steel mills continues to be mined. Yet little attention is given to the effects on the locals
that hinder the social fabric.
173
1.3 Current technological focus
Since the industrial revolution, governments and industries have focused on achieving
technological solutions to problems. As Stephen R. Couch (1996) wrote: “Our customary
response tends to be: ‘If it’s broken, fix it; if it’s contaminated, clean it; if it’s burning,
douse it.’” Yet, as we have seen in South Africa and India, human issues commonly are
not addressed; in fact, they are generally disregarded while emphasis is on implementing
scientific and engineering technology. Indeed, as is the case in mine fires worldwide,
technology can contribute to human distress. Oddly enough, though human issues are
overlooked, human well-being, needs, and desires are at the core of why problems of
underground fires are being addressed.
Up until now, the best researched impact of coal fires on human beings has been pollution.
The nature of the pollutants varies from fire to fire, but the effects on human beings are
devastating in many ways. The long-term overall effects are well recognized. A 1999
report by the Clean Coal Center of the International Energy Agency concluded that the
biggest coal fires – China has the most fires and India has the greatest concentration –
“make a significant global impact” (Revkin 2003). Greenhouse gases are released in large
quantities. Aerosols and toxic gases including carbon dioxide, carbon monoxide, sulfur
oxides, mercury, fluorine, arsenic, selenium, thallium, and organic compounds are part of
coal fire pollutants. They have disastrous effects not only on the atmosphere, but on soil
and water, and cause human diseases, physical and mental malfunctioning, and death.
Human beings’ health and psychological well-being is directly affected, and they must
contend with a damaged and destroyed ecosystem.
Extinguishing underground fires is a difficult and complex task. Technological solutions

are needed. But the problem extends beyond the physical, scientific, and engineering level,
and even beyond the medical level. Just as with earthquakes, typhoons, and volcanoes,
the social and psychological effects of coal fires can be long-term and devastating. They
need to be addressed. In the United States, the Centralia fire, in the Commonwealth of
Pennsylvania, is an example of what can happen when technological solutions, but not
human needs are addressed by government planners.
174
2 Centralia Fire (1962 to Present): An Example of Deficient

Policy
Today, approximately 48 underground coal fires are burning under 486 ha across the
anthracite rich Appalachian Mountains in the Commonwealth of Pennsylvania (Kerkstra
2002). Many of these fires are the legacy of two centuries of primitive mining practices.
Seams were worked by carving out rooms supported by pillars of coal. When the mines
were abandoned, mine workings were left unsealed. Laws that existed were not enforced.
As David Philbin has said: “These coal fires are like animals: They live in holes, hide from
predators like us, and want to eat all the time” (Kerkstra 2002).
Centralia is the town and fire that has gained the most notoriety in the area. 1 Of
Pennsylvania’s active coal fires, the Centralia fire is the largest (Perkins 2003). Officials
working on the Centralia fire focused on the technological “fix-its,” disregarding the
human dimension. Focus was on extinguishing the fire, or at least restraining it, not on
community conflicts that obstructed and delayed resolution of contamination problems.
Local culture and attitudes were transformed while local individuals and families had to
contend with the psychological issues of distress, anxiety, fear, depression, grief, anger,
and post-traumatic stress disorder (PTSD). 2 Technological “fix-its” and society issues were
in contention, not collaboration.
At one point, the residents were in severe internal and external social conflict related to
recovery or transformation from the fire disaster. People came to hold a different view
of their town and the outside world. New subcultures and norms evolved to deal with
insecurity. New groups formed and were institutionalized. As we shall see, one faction
sought and received outside assistance of national environmental groups. From the outside
people they learned that national news media reports could bring direct political action.
This proved to be a successful confrontational tactic; hence, the town and the fire were
well publicized. The results have been government actions that led to Centralia being
mostly evacuated and demolished. Twelve people continue to live there and pay taxes on
what became government-owned, condemned housing 21 years ago. Media coverage is
now infrequent, and people outside the immediate region are no longer aware of the effects
of the fire on the community. The fire has continued to burn while the town and community
of Centralia could not rebound. They died.
1
For a general overview of the fire’s history, see DEP (1996).
2
The novel “Those Who Favor Fire” (Wolk 1998) is based on the Centralia fire. It describes the
psychological impact on local residents over time. See Schiraldi (2000) for more information on stress
disorders.
175
2.1 Centralia’s history
Centralia was established as a mining community. In the 1850s and 1860s, after the
discovery of coal in the region, mine companies evolved and drew miners.3 Centralia and
other mining towns developed in the area. The first building in Centralia was a tavern, the
Bull’s Head Tavern. By the time the fire started in 1962, Centralia residents were third- and
fourth-generation descendants of immigrant Polish, Welsh, Ukrainian, Irish, and Russian
miners. In the latter part of the 19th century, Centralia had become a company town with
periods of instability and rises and falls in population. Miners were mobile, and community
ties were weak. Power rested in the external company agents. Mining companies in the
region resisted union organization until 1902, although before 1902 residents did take
collective action in the form of strikes and trade union activities. However, the companies
found it to their interest to work to reduce the power of locally elected officials. These
officials generally supported the miners against the companies. Overall, collective
decision-making by the local people in their own self-interest did not evolve as part of
Centralia community culture. It was a culture based on class and incorporated distrust
toward the coal companies.
Beginning in the 1920s and intensifying in the 1950s, alternative fuels came into use in the
United States, and anthracite coal production declined. By the 1960s, active mining had
ceased in the Centralia mine. While the general practice of coal mining companies was to
retain subsurface mineral rights when they abandoned a mine, the borough of Centralia
had acquired these rights in the early 20th century. By 1962, Centralia’s population had
declined to around 1,100 largely working-class residents. A high percentage was elderly
and retired. Most had lived in Centralia all their lives. This was their place, a physical
part of their identity and sense of self. Contamination of Centralia was sensed to be a
contamination of individuals and families, as well as the physical components of the town.
There were approximately 400 “two-story brick row houses, each twelve feet wide, that
rose up the hillsides in strict parallel lines […], five churches, three Catholic and two
Russian Orthodox, with plump, gilded domes […], [and] a polka dance hall” (Tietz 2004).
New industries were not interested in locating in a declining one-industry town, but a
tailor, a shoemaker, an electrician, a druggist, and a general medical practitioner continued
to work there. 4
3
For more information on Centralia’s pre-fire history, see Caudill (1963), Aurand (1971), DeKok (2000),
Couch (1996), and Johnson (2005).
4
See Robert Keller Fetterman’s “Centralia Pennsylvania Remembered” for a view of living in Centralia
in the 1920s. www.offroaders.com/album/centralia/memories.htm
176
2.2 Genesis of the fire and human action
The genesis of the Centralia fire is speculated to be related to an above-ground fire set in
a dump located in an abandoned stripping pit. The literature is at odds as to whether this
was a municipal dump or an unsanctioned community dump. However, it is accepted that
the burning of debris was intended to reduce garbage volume and control rodents (Nolter
2004). The clean-up fire was assumed to be controllable as other such fires in the region
had been.
Initially, little action was taken. The dump was located a half mile away from the town,
a distance thought sufficient to protect town residents. Workers attempted to smother the
fire with fly-ash. After a month the fire had not died out. The burning waste was scraped
back and a hole was found in the ground. Unsuccessful attempts were made to dig a
trench to isolate the burn, but the fire had already moved further than expected. Remnant
underground coal was later recognized to have been ignited. The new underground fire
crawled through the unfilled mouths of the deep-mine workings in the exhausted mine-
tunnel.
As Stephen R. Couch (1996) relates: “Despite the prominence of coal companies in the
town’s past, the fire was not due to corporate malfeasance, because there was no active
mining where the fire started.” The local borough now owned the underground mineral
rights. Local regional governments argued over who was responsible for fighting the fire.
Finally, a local government official called the state agency and they in turn contacted
the Federal Bureau of Mines. While the problem was seen to be serious, action was
further delayed by mandatory protocols and procedures to solicit proposals and bids for
extinguishing.
Government agencies viewed it as prohibitively expensive to extinguish the fire, which in

their eyes was too remote to affect local people. Underground coal fires have been common
in Pennsylvania. The dangers of poisonous gas and subsidence from most of these ongoing
underground fires do not have dramatic human effects. Neither did the one in Centralia –
until three months after it had started.
By August 1962, the fire had reached a network of old mine tunnels beneath the town. The
first effects evoked curiosity rather than calamity. For example, home gardens with their
heated soils continued to produce crops such as tomatoes on into winter. Plowing snow off
pathways became easier.
177
2.3 Evolution of social issues
Gradually, over the months and years, the picture changed as social, emotional, and health
issues surfaced.
By the end of the 1970s, people all over town were aware of problems. Local, state, and
federal government officials and agencies failed to take effective action. Sporadic attempts
to control and extinguish the fire through numerous engineering projects had been made
without success in the 1960s and 70s. Officials blamed the lack of sufficient funding,
government agencies blamed lack of responsibility. From 1962 to 1978, local, state,
and federal agencies spent more than US$ 3.3 million to control the fire – with marginal
success (DEP 1996).
In 1969, the Commonwealth assisted moving three families whose houses were
endangered by the presence of toxic gases. The following year, the federal government
purchased seven properties that were estimated to be unsafe. Gas alarms were installed
in all houses. Seven years later, highly lethal carbon monoxide concentrations were
found to be emanating from a borehole near a residence. Three years after that, a
gasoline station was ordered to close when its underground tanks heated up and spewed
out sour smelling liquid when uncapped. The process was slow, but people were
becoming increasingly aware of potential dangers. Basement walls in some houses were
hot up to 82 ºC. The owner of a motorcycle shop found that carbon monoxide from
ambient coal combustion had displaced most of the oxygen in his basement, thereby
preventing the lighting of his coal furnace. Headaches were common afflictions and
some residents passed out from carbon monoxide leaking into their homes. Water ran hot
from cold-water taps. The main road into town dropped 8 ft, and steam started to spurt
out of cracks in pavements.
Conflict developed between government bodies and the community that sought aid. As
Caudill (1963) explains, the local attitude toward the local, state, and federal governments
was suspicion. For decades, Centralia residents had been exploited and then ignored by
outsiders. Some of the local people carried the fatalistic attitude over to their view of all
government institutions. Indeed, year after year during this time, political and bureaucratic
hassles led to studies of the scientific and technical aspects of the fire rather than the
applied action sought by the community.
178
2.4 Community discord
Discord developed within the community. In general, three attitudes had emerged. One
group of residents saw the fire as a distant hazard, but not something to panic over.
Another group, the older, long-time residents, many of whom had lived in Centralia
all their lives, did not see the fire as a problem. Many of these were miners who knew
the topography of the region’s anthracite veins, the bridges and valleys along the beds,
and the water table. They felt that the fire would naturally be kept from contaminating
Centralia. The third group was composed of those who lived in the “impact zone” – the
area directly affected by the fire. They saw the fire as an impending disaster. This last
group, composed mostly of younger adults and their children, represented about one
third of the population. Each of these groups had their own “fire subcultures.” That is, as
time went on, each bonded and supported one another with their own values and actions.
Tension was high, but the community was politically weak and unable to dispel the
growing conflict.
Finally, on 14 February 1981, after a year without government aid, Centralia was prompted
to increase their efforts for aid. On that day, the ground collapsed under a twelve-year-old
boy as he walked across his grandmother’s yard to investigate smoke arising. By catching
onto a tree root, he was saved from falling further into the 57 ºC carbon monoxide filled
sink hole. His cousin pulled him to safety, but two months later his mother reported she
was caring for his post traumatic stress syndrome. A few weeks later, an elderly man was
overcome by carbon monoxide in his home.
These serious near-death incidents aroused those who viewed impending disaster to
formally organize “Concerned Citizens Against the Centralia Mine Fire.” The goal of this
group was to obtain government aid for protection from the fire. They were not opposed
to being relocated if necessary. Letter writing, and demonstrations and marches, brought
them to the attention of the news media. They learned from environmental interest groups
that media use was a good avenue to government involvement. Hence, the general public
rapidly became aware of Centralia’s fire and the effect it was having on the people.
However, the people in Centralia who were not as fearful of disaster were adamant that
relocation was not necessary. They did not like the media’s attention. For two years,
arguments and discord intensified between the two sides. The strength of Centralia as a
cohesive social community was drained. Environmental contamination had led to social
pollution.
179
2.5 Relocation
In 1983, the U.S. Office of Surface Mining (OSM), part of the Department of the Interior,
released a study showing that the fire was worse than known. They estimated it might
take another US$ 663 million to extinguish the fire. Buying the town and condemning and
demolishing the buildings were a far less costly alternative. Authorities opted to have the
National Congress appropriate about US$ 42 million for voluntary acquisition and relocation
of businesses and residences impacted by the dangers caused by subsidence and noxious
gases.
“Concerned Citizens Against the Centralia Mine Fire” had achieved their goal. The
majority of citizens voted to accept relocation. But as noted earlier, some opted to stay and
have yet to move. With the availability of relocation resources, two new citizen groups
were formed. Those who opposed relocation formed the “Citizens to Save Our Borough”
in an attempt to persevere with a corporate community. The “Homeowners’ Association”
was instituted to help residents obtain a fair settlement price.
Between 1985 and 1991, a total of 545 residences and businesses were acquired by the
Commonwealth of Pennsylvania in the borough of Centralia, the town of Byrnesville, and
scattered housings outside these areas. The residents moved, but throughout there was
tension between the “Homeowners’ Association” and the “Citizens to Save Our Borough.”
Neither of the two citizen groups wanted media attention, so the earlier publicity dwindled.
Today, articles published about Centralia are rare. 5
2.6 Centralia today
The fire continues to spread at varying rates and levels about 23 m yr-1 along separate arms.
Only general estimates can be made, but it may well be another 250 years before the fire
runs out of fuel (Alteres 2005).
The drama of the Centralia fire has ended. The OSM maintains observation, and
scientists come to study the minerals forming in cracks as well as the heat-loving
bacteria. Tourists arrive hoping to see plumes of smoke and steam near bleached tree
stumps. Twelve of the original 400 houses remain standing. The others have been
demolished as they were vacated. About 12 older residents are tenacious and continue
to reside in and pay taxes on condemned houses the state government now owns. Over
5
The most recent is Krajick (2005).
180
time, the number dwindles from death or departure. None of those who relocated have
moved back, but the number of previous residents in town varies as relatives and old
neighbors visit.
Like those living in the nearby towns of Mount Carmel and Locust Gap, people who
continue to reside in Centralia do not feel in danger. They are accustomed to the
environment and content to maintain continuity of location. In fact, they have drawn
together and formed a “new community” of diehards. Eight of the 12 are local office
holders working to maintain the streets and houses. They unsuccessfully crusaded to have
their postcode renewed when the U.S. Postal Service finally revoked it. Meetings are held
and some former residents help to set up festive holiday decorations.
There are mixed attitudes among those who left Centralia. Some say they have better
housing, but regret the higher cost of living. New, larger houses have higher mortgages.
Some are reported to be “heart-broken by their exile” (Tietz 2004). Those who relocated
in the early years are angered that the Commonwealth did not reimburse them to the
extent it did others who relocated five years later after the federal government supplied the
Commonwealth with funds.
Stephen R. Couch (1995) concluded that industrial contamination in such situations as

Centralia results in “destructive social and psychological consequences.” He argues that
these are the results of psychological and emotional experience of contamination itself,
the assignment of blame, problems of technical controllability, ambiguity of cause and
effect, social stigma, and time. In his terms, social change in such situations is too great for
“recovery. ” “Transformation” – a change to something new – can occur, but not “recovery”
– the resurrection of what was lost.
In the Centralia fire, government agencies and local people saw the issues from differing
perspectives. One gave precedence to technology to “fix it” while the other strove for social
solutions. Working in contention, not collaboration, resulted in Centralia’s being transformed
into a community that is in the final stages of dying while the fire continues to burn.
3 The Value of Local Residents Being Included
Our experience in defining and meeting environmental problems in countries worldwide

has demonstrated that government and business planners generally are not directly involved
181
at the local level. All too often they are unaware that they overlook, and do not take into
account, the needs of individuals and groups of people in the problem area. Centralia
residents were not heard until they went to the news media. Even then, officials sought a
quick and easy solution by relocation without researching alternate social solutions. The
cultural, psychological, and emotional needs of the people were never addressed.
Often, the people best informed on the problems, and the possible solutions, are those
most intimately involved – the local residents. Officials and planners at different levels
may accurately comprehend the general problem, be highly informed and skilled in related
technological solutions, but they are unfamiliar with the day-to-day social setting. As we
have seen in Centralia, this can lead to a narrow perspective that misreads the picture.
In Centralia, the social anthropological approach would have been to work from a broad,
holistic perspective. That is, anthropologists not only want to understand the general
organization, values, and related behavior, they strive to know what is happening behind
the public mirror. Did the conflicting groups have any mutual interests that could have
been used to hold the bonds of community together? What was needed to have unified
leadership emerge in the politically weak town? What local social and cultural information
did the government agencies need in order to incorporate the human perspective?
For scientific, engineering, and government planners to understand the human framework of
current problems and the context in which projects will be implemented, they need to know:
• Local demographic information – including age structure, family size, birth and death
rates, education levels, income levels, history, and demographic changes;
• Local culture, including beliefs and learned behaviors;
• Local attitudes, emotions, and related behavioral patterns; 6
• Local social organization, activities, and interactions with other groups in the region
and on to the higher levels of government.
Obtaining this information involves the active participation of local and regional levels.
This involves a paradigm shift from central planning to involvement of local people.
Problems are better solved at local level with people who are educated in the problem and
are encouraged to contribute to the solution.
6
See Vaught et al. (2000) for an in-depth review of the individual and group experience (perceptions,
attitudes, and behavior) of an underground coal mine fire.
182
Local residents have participated in some situations as informants in qualitative

ethnographic research.7 In others, they have initiated the planning, been included as part of
the planning team, and participated as local project managers.
4 An Example of Incorporating the Human Dimension:

Ningxia and Inner Mongolia Autonomous Regions, P.R.
China
A study of Chinese coal fire regions in Ningxia and Inner Mongolia Autonomous Regions
was undertaken by Professor Dr. Ümüt Halik and his colleagues in 2003. This was
sponsored by the Sino-German Coal Fire Research Initiative “Innovative Technologies
for Exploration, Extinction, and Monitoring of Coal Fires in North China.” The
following discussion reviews their working paper. We shall see how they focus on the
human dimension as the base for the region’s infrastructure and economic situation by
incorporating residents of Ningxia and Inner Mongolia in narrative interviews to obtain the
necessary data.
The primary focus of the project was land use development in the two autonomous
regions. The study was “to find strategies for the exploration, extinction, and monitoring
of the coal fires in north-west China” (Halik et al. 2003). Part of the Sino-German Coal
Fire Research Initiative was to stop the impact of the fires on the residents – and logically,
the only way would be to extinguish them. However, the effect on the environment and
population would have to be taken into account when considering any possible method of
extinguishing. This goal required data on the socio-economic situation in the fire areas.
4.1 Setting
Long cold winters and very hot summers with huge sandstorms set the stage in the two
grassland and desert regions studied. Coal is the major economic support of the region, and
fires there are a grave problem. China’s largest and most dangerous fire zone is in this region; of
China’s 56 documented active fires, 16 burn here (Meyer 2005). One fire in Wuda Coalfield in
southern Inner Mongolia stretches across a barren area of nearly 39 km2. The Wuda Coal Mine
official in charge of underground blazes says this is China’s most destructive fire, putting it in
the running for the world’s worst ecological disaster (Meyer 2005).
7
For information on qualitative ethnographic research, see Ellen (1984), Fetterman (1997), Silverman
(2000), and Maxwell (2004).
183
Ningxia and Inner Mongolia are inhabited by a relatively small population with low
income and a low standard of living. They reside in dense air pollution with relatively
limited economic resources. River water is available to those close to the Yellow River,
but other regions struggle to meet their water needs (Han & Zhao 2005). Many struggle to
survive by whatever means possible. One practice of rural Chinese residents in coal areas
is to hand-dig their household coal from hundreds of thousands of surface locations. When
the hollowed small pits become too deep to access they are abandoned. As a result, the
highly combustible loose coal chunks and coal dust inside the pits are exposed to air and
thus vulnerable to spontaneous combustion.
The working report notes that “the human dimension of the coal fire problem is of great
importance” (Halik et al. 2003). Research was conducted to provide a general description
of the main human activities in the region. The methods included web and literature
review, meetings of the researchers, and a three-week field trip with fourteen open, non-
structured, narrative interviews.
One of the goals of the researchers’ fieldwork was to “explore the investigation area under
spatial planning criteria and to accomplish observations” (Halik et al. 2003). By choosing
to do fieldwork in three diverse topographic and ecological areas, they obtained a wide
range of data to draw a more complete overall picture. Ethnographic research, a major
tool in social anthropology, incorporates participation in, as well as observation of, the
society or group studied. Here, the researchers could only participate as outsiders, but by
using only a few pre-formulated questions and allowing the interviews to be unstructured
narratives, they were able to gain substantial qualitative data, such as patterns of overtime
work and lack of overtime pay. In ethnographic research, open, unstructured discussions
can reveal relevant topics and data known to local residents. Outside researchers, being
unaware of local patterns, values, and culture cannot always formulate relevant questions.
In this study, the professions of those interviewed varied, providing a range of information
from local officials and planners, technicians, coal engineers, a farmer, the owner of a
small coal pit, a miner, and the owner of a local restaurant.
4.2 Background of the study
Coal mining has a 3000-year history in China. However, until the 19th century, primitive
techniques remained in use. With the Industrial Revolution, westerners with modern
methods were drawn to opportunities (Ottens 2005). Today, 75 % of the energy used in
China is supplied by coal. The People’s Republic of China is the largest coal producer and
184
exporter in the world. To achieve this status, the economy relies upon the vast coalfields
that lie mainly in the northern provinces, including Ningxia and Inner Mongolia. Stephan
Voigt of the German Aerospace Center (DLR) reports that China has the most severe
underground coal fires of any country in the world (Meyer 2005). This is true in both
number and scale. 20 to 30 million t of coal burn underground there annually, exceeding
the annual amount exported.
It is widely recognized that the numerous surface vents pollute the air with toxic
substances, directly affecting local environment and health. This effluent, added to the
heavy pollution from China’s coal-fired power plants, produces a national problem that
has expanded to become an global disaster. The World Bank estimates that 400,000 people
in China die each year from air pollution-related illnesses, mainly lung and heart diseases
(AFP English 2004). Energy specialists in the Asian Development Bank estimate that as
much as 40 % of the air pollution in Japan and South Korea originates in China (AFP
English 2004). These countries are now demanding reforms. There is growing evidence
that pollution from China has reached the United States.
In addition to air pollution, in part caused by underground fires, China must contend with
a high rate of death in the coal industries. 8 World wide coal mining is the most dangerous
and potentially lethal job in the mining industry. Indeed, by far the most dangerous
occupation in China is mining. Year round, on average, 14 miners are killed every day in
fires, cave-ins, and other accidents. News releases perpetually report mine deaths. 9 Victims’
families are compensated, but the industry and government are facing severe challenges in
attempts to upgrade mine safety. 10
4.3 Population data obtained
The following information is from the working report (Halik et. al. 2003). It does not give
all the social information needed by outside planners – cultural values, behavior, attitudes,
social organization. However it demonstrates that at this stage, the researchers have collected
information of value in further planning related to environmental issues such as coal fires.
8
Yang Li’s 2004 film “Blind Shaft” is an international film festival award winner that gives a realistic
portrayal of the gruelling day-to-day existence of the ordinary men and women who are miners.
9
For examples see Wade (2002), Johnson (2004), Lambrecht (2005), AP Online (2005a,b), AP
Worldstream (2005), Xinhua News Agency (2005).
10
See Jerry Tien (1997) for a slightly outdated, but nonetheless insightful article on the structure and
conditions in China’s coal industry.
185
4.3.1 Demography
Due to varied geographical conditions, the population in the Helan Shan in Ningxia
and Inner Mongolia is unevenly distributed. Higher dry-land areas with colder climate
are sparsely inhabited. Expanses with closer proximity to water in the Yellow River are
favored. The largest city is the capital, Yinchuan, with approximately 954,000 people. The
18 urban centers, each with more than 100,000 residents, are reported to have retained a
rural character.
Field research by Halik’s team included Wuda, a town of 30,000 residents in the county
of the same name, with a population of 130,000. Water from the Yellow River meets the
needs of people here and allows for some agriculture. The second area, in Ningxia, was
in a mountainous region. The field work focused on the town of Rujigou, which is about
the same size as Wuda, and is in a county of 110,000 inhabitants. The third area lay also in
Inner Mongolia. The mountain town of Gulaben has only 2,428 residents and is accessible
only in good weather conditions.
4.3.2 Family structure

The size of households was reported to be shrinking. Three-generation extended families
are no longer the standard as they were only a few decades ago. Now, most households
(71.8 %) are nuclear, two-generation or single generation stem units (8.2 %). Hence,
kinship interaction across generations has declined in frequency. Interaction within
generations is also decreasing. Pressure for a couple to have only one child has resulted in
loss of siblings and the roles of aunt and uncle.
4.3.3 Population age

The government family planning program has resulted in a significant age structure
change. In the terms used in the report, before initiation of the family planning program
“the age structure” was “young.” By 1990, in Ningxia, “the population in urban areas had
entered the “adult pattern,” while in rural areas the structure remained as the “children
pattern” (Halik et al. 2003). In the same year, the average life expectancy here was 68.6
years for males and 67.8 years for females, rates that are lower than mean levels for the
whole country. Long life and health is not easy to achieve in the Helan Shan.
While the urban areas in Ningxia are beginning to show an older population percentage,
the average age in the province is younger than the average in China as a whole. This
implies that in general, rural families here are less likely to adhere to the one-child family
planning policy. Indeed, the mountainous areas had a lower acceptance of family planning
(73.8 %) than did the plains (90.7 %).
186
4.3.4 Education
The levels of education and literacy rates in Ningxia are below the national average,
although they have improved since 1982. In 1996, the province was awarded an
international illiteracy-elimination prize by the United Nations. Yet nearly a third of
Ningxia’s population currently is illiterate or only partly literate. Only about a quarter
of the residents have graduated from primary school and only a third have gone on to
graduate from middle school. The largest university is in the provincial capital, but less
than one percent of the population has a university degree. In the town of Wuda, there is a
vocational school and the coal office has its own training center, where training is available
for welders, mining workers, and power station workers.
The report states: “In Ningxia, the difference in the education level can be found through
all ethnic groups. The Hui nationality has the lowest education level, the Han’s is higher
than Hui.” Studying the causes and long-term effects of these differences on the social
relations and community unity would be of value to planners.
The education and literacy rates in Inner Mongolia appear to be significantly higher. Here,
almost all (99.5 %) “school age” children were attending school, there is no illiteracy in
this age range, and the number of years of schooling has been increasing annually. There
are 18 institutions of “higher education” (the definition of “higher education” is not given),
that have helped the region to have 500,000 specialists, over a third of whom have high-
grade professional titles. Although the education level is increasing in this region, this
is separate from the coal mining industry where the benefits of education are not being
applied to coal mining problems.
The report of Halik et al. (2003) is a working paper, thus not a conclusive study. However,
as discussed, this study already has included local residents in collecting data. Sufficient
information is presented for planners to continue incorporating local people in further
research, planning, and implementing of coal fire programs in the region.
The data and information in the report offer insight into the human component of the
region and can enhance strategies for the exploration, extinction, and monitoring of coal
fires toward better land use. We hope sufficient information was obtained to enable the
researchers to provide planners with a more complete and in-depth understanding of the
psychological, cultural, and behavioral components of the region.
The report mentions the existence of varying levels of education, income, and status, as well
as multiple national ethnic groups in the two autonomous regions. What are the differences
187
and similarities of these groups? Are there community bonds between them or is conflict
likely to develop? What are the attitudes and emotions of these people regarding their current
situation and possible change? What are local behavioral patterns (such as the hand-digging
of small pits that was mentioned earlier) that might be detrimental in the long run? What is
the local political situation? Could a sudden, unanticipated fire danger be managed, or would
a contentious internal community conflict develop as it did in Centralia?
Answers to all of these questions would provide planners the opportunity to develop and
implement successful strategic policies.
5 Summary
As noted earlier, human beings both cause and are affected by underground fires. In
both the Centralia and the China examples, human action – i.e. education, behavior, law
enforcement – could have helped prevent the fires. In Centralia, abandoned mine workings
were left unsealed, existing laws were not enforced, responsibility for the mines changed
from companies to the borough, and measures to prevent fires were not undertaken. In China,
the coal industry is known to have had numerous challenges in many forms (Tien 1997). In
both cases, the underground fires were not perceived to be of imminent effect. Yet in both
cases, human action, properly applied, could have helped prevent self-ignition of coal.
In the Centralia case, government agencies became involved when the fire had already
been burning for some time and had become a political issue. Government agencies came
on the scene in a “fix-it” mode. Following long custom, “fix-it” for issues related to fires
meant using technological solutions. The human dimension was ignored until people
forced the issue.
The Ningxia and Inner Mongolia study came from a different direction. Here, the
underground fires had been burning for some time and were known to be detrimental but
were not considered to be an imminent threat. Hence, this study was designed to approach
the area’s problems, some of which are directly related to the fires, by involving local
people in gathering data toward social solutions.
Underground coal fire projects require long-term studies combining social and
technological strategies. The “fix-it” approach and technological tools are essential
for successful actual fire fighting. However, as discussed in other papers presented by
engineers and scientists at the International Conference on Coal Fire Research, there
188
are a variety of reasons why technological solutions have been unable to quench a
large number of coal fires. Technological research is ongoing and needed, but until
underground fires – from natural or human causes – are easily extinguished, social issues
related to causes and effects will arise and need to be addressed. Unfortunately, as was
seen in Centralia, bureaucrats and officials can overlook, avoid, or lack a sense of urgency
even in prime social issues.
Some underground fire issues are clearly in the realm of specialists – be they experts
in the physical sciences, engineering, or social sciences – but many problems could be
better resolved with input from multiple disciplines. Here, two issues arise. First, while
interdisciplinary teams instituted by governments, international agencies, and academic
groups have been successful, to be successful, members of such teams must be willing to
cooperate with experts coming from different perspectives and approaches. Second, no
matter how successful teams are in gathering data and making recommendations, the weak
point are administrative or political hurdles when instituting recommendations. Often, as in
Centralia, plans and projects are not implemented because of a lack of financial resources.
Somehow money only becomes available when there is a high level of social awareness
about the need (and the relatively greater need than that of other projects).
6 Points to Ponder
1) Few people are aware of the global and catastrophic nature of underground coal fires.
Few realize that underground coal fires may be caused by either natural or human
actions. Methods are available to reduce pollution. Is it not time that efforts were made
to educate the world?
2) Underground coal fires are detrimental and destructive socially as well as ecologically.
3) The common response to these fires has been for policy-makers to employ technological
solutions, addressing environmental problems without addressing the concurrent social
problems. Policy-makers need to address both technological and social issues related to
underground coal fires.
4) Much research related to underground coal fires needs to be performed by

interdisciplinary teams of engineers, physical scientists, and social scientists. Some
issues are clearly in the realm of specialists, but interdisciplinary teams can better
resolve issues with both social and physical impact.
189
5) A paradigm shift is needed from wholly central administrative planning to incorporation

of local residents.
6) The most persistent hurdle in overcoming underground coal fire issues lies not in the
domain of engineering and social sciences, but in the political and administrative realm.
References
AFP English, 2004. China’s dependence on coal for energy causing pollution at home and abroad.
Agence France Press (AFP) English, 29 July 2004. October 2005 http://www.highbeam.com/
library/doc3.asp?DOCID=1P1:96914008&num=35&ctrlInfo
Alteres, T., 2005. Personal interview / Oral communication. (Timothy Alteres works at the U.S. Office of
Surface Mining (OSM), Washington, D.C., U.S.A.)
AP Online, 2005a. 203 killed in China coal mine explosion. AP Online, 15 February 2005. October 2005
http://www.highbeam.com/library/doc3.asp?DOCID=1P1:105411626&num=37&ctrlInfo
AP Online, 2005b. China says coal mining deaths up in 2005. AP Online, 15 April 2005. October 2005
http://www.highbeam.com/library/doc3.asp?DOCID=1P1:107088807&num=30&ctrlInfo
AP Worldstream, 2005. China says coal mining deaths up 21 percent this year despite safety crackdown.
AP Worldstream, 5 April 2005. October 2005 http://www.highbeam.com/library/doc3.asp
DOCID=1P1:107066413&num=36&ctrlInfo
Aurand, H., 1971. From the Molly Maguires to the United Mine Workers: The social ecology of an
industrial union, 1869-1897. Philadelphia: Temple University Press.
Bauer, D. 2003. An alliance for Africa. Penn State Online Research. October 2005 http://www.rps.psu.
edu/0309/africa.html
Business Day, 2003. Coal seam fires pose a major threat. Business Day, 31 March 2003. October
2005 http://www.c-kit.com/Right%20Index/News%20Letter/Newsletter/2003/April/
coal.htm
Caudill, H., 1963. Night comes to the Cumberlands. Boston: Little, Brown.
Couch, S.R., 1996. Environmental contamination, community transformation, and the Centralia Mine
Fire. In Mitchell, J.K. (Ed.), The long road to recovery: Community responses to industrial
disorder. New York: United Nations University Press.
DeKok, D., 2000. Unseen danger: A tragedy of people, government, and the unseen dangers of the
Centralia Mine Fire. New York: toExcell.
DEP, 1996. A brief history of the Centralia Mine Fire (Borough of Centralia, Columbia County).
Pennsylvania Department of Environmental Protection (DEP), Harrisburg, Pennsylvania,
U.S.A. October 2005 http://www.dep.state.pa.us/dep/DEPUTATE/MINRES/CENTBRF.htm
Ellen, R.F. (Ed.) 1990. Ethnographic research: A guide to general coduct. London: Academic Press.
Harcourt Brace Jovanvich Publ.
Fetterman, D.M., 1997. Ethnography: Step-by-step. London: Sage Publ.
190
Fetterman, R.K., 2005. Centralia Pennsylvania remembered. October 2005 http://www.offroaders.com/

album/centralia/memories.htm
Halik, Ü., Bandelow, F.K., and Schulz, J., 2003. Land use and socio-economic aspects of the Helan
Shan in Ningxia and Inner Mongolia, China. Field report for Work Package 2140-2150, as a
contribution to the Sino-German Coal Fire Research Initiative “Innovative Technologies for
Exploration, Extinction, and Monitoring of Coal Fires in North China.” Xinjiang University,
Urumqi, Xinjiang Autonomous Region, P.R. China; Deutsche Montan Technologie GmbH
(DMT), Essen, Germany; and Dortmund University, Dortmund, Germany.
Han, H. and Zhao, L., 2005. Rural poverty in western China is water poverty. China & World Economy
13(5): 76-88.
Heffern, E.L. and Coates, D.A., 2004. Geologic history of natural coal bed fires, Powder River Basin,
U.S.A. International Journal of Coal Geology 59: 25-47.
Johnson, D.B., 2005. Centralia (Images of America: Pennsylvania). Charleston, South Carolina: Arcadia
Publishing.
Johnson, T., 2004. 14 miners die every day hauling coal for China’s masses. Knight Rider, Tribune News
Service, 20 August 2004. October 2005 http://www.highbeam.com/library/doc3.asp?DOCID=
1G1:120942089&num=4ctrlInfo
Kerkstra, P., 2002. Mine fires: A global catastrophe. The Philadelphia Inquirer. October 2005
http://www.highbeam.com/library/doc3.asp?DOCID=1G1:120215639&num=6&ctrlInfo
Krajick, K., 2005. Fire in the hole. Smithsonian 36(2): 52-57, 59-61.
Lambrecht, B., 2005. China signals new urgency in cutting coal-mining risks. St. Louis Post-Dispatch,
3 March 2005. October 2005 http://www.highbeam.com/library/doc3.asp?DOCID=1P1:10598
9317&num=8&ctrlInfo
Li, Y., 2003. Blind shaft. Li Yang & Tang Splendor Films Ltd. New York: Kino Intl. Corp.
Maxwell, J.A. 2004. Qualitative research design: An interactive approach. London: Sage Publ.
Meyer, M., 2005. Flaming dragon. Smithsonian 36(2): 58.
Nolter, M.A., 2004. Changes in the Centralia coal fire over forty years. In Proceedings of the Denver
Annual Meeting of the Geological Society of America, Denver, Colorado, U.S.A., 7-10
November 2004, paper no. 15-13.
Ottens, B., 2005. Mining in China: A 3000-year tradition (Industry review). The Mineralogical Record
36(1): 5-11.
Perkins, S., 2003. The fires below. Science News 163(19): 296-298.
Revkin, A., 2003. Sunken fires menace land and climate. The New York Times, 15 January 2003. October
2005 http://www4.nationalacademies.org/news.nsf/isbn/s10062003g?OpemDocument
Schiraldi, G., 2000. The post traumatic stress disorder sourcebook: A guide to healing, recovery, and
growth. Los Angeles: Lowell House.
Silverman, D., 2000. Doing qualitative research: A practical handbook. London: Sage Publ.
Tien, J. 1997. China’s coal industry – Past, present, and challenges (industry overview). Engineering and
Mining Journal: 16D-16D.
191
Tietz, J., 2004. The great Centralia coal fire: How one small mining town went up in smoke. Letter from
Pennsylvania. Harper’s Magazine, February 2004. October 2005 http://www.highbeam.com/
library/doc3.asp?DOCID=1G1:19900809&num=1&ctrlInfo
Vaught, C. et al., 2004. Behavioral and organizational dimensions of underground mine fires. Information
Circular 9450. U.S. Department of Health and Human Services, Pittsburgh, Pennsylvania,
U.S.A.
Wade, C., 2002. Deadly accidents plague China’s coal mines. United Press International, 6 October 2002.
October 2005 http://www.highbeam.com/library/doc3.asp?DOCID=1P1:53529244&num=29
&ctrlInfo
Wolk, L., 1998. Those who favor fire: A novel. New Work: Random House.
Xinhua News Agency, 2005. Coal mine victims to be compensated for 200,000 yuan in NE China. Xinhua
News Agency, 14 September 2005. October 2005 http://www.highbeam.com/library/doc3.
asp?DOCID=1G1:136136117&num=100&ctrlnf
192
Socio-Economic Aspects of the Local
Mining Sector and Alternative Sources of
Income in the Helan Shan
Ümüt Halik1, Friedrich-Karl Bandelow2, and Jörg Schulz2,3
1
Postdoctoral Fellow Station of Theoretical Economics, College for Resources and Environmental
Research, Xinjiang University, Urumqi, Xinjiang Autonomous Region, P.R. China
2
Deutsche Montan Technologie GmbH (DMT), Essen, Germany
3
Faculty of Spatial Planning, Dortmund University, Dortmund, Germany
Abstract
The economy of the Helan Shan coalfields (Inner Mongolia and Ningxia Autonomous Regions,
P.R. China) is heavily dependent on large coal mining enterprises as well as private pits scattered along
seam outcrops. This study estimates the influence of the coal industry on the region’s economy and
shows possible employment alternatives. It first takes a look at the economic development of the city
of Wuda and tries to identify other developable and prospering industries there. The results are then
compared to the situation in the other two areas under investigation, Rujigou and Gulaben. Are there
fundamental differences between them? Information on important structural and environmental problems
in the test areas is given too. One of the major objectives of this study is to find a path toward sustainable
development in the region. Therefore, future strategies will also be presented. They are capped by the
author’s own suggestions for improvement.
摘要
贺兰山煤田的经济主要依赖于重点煤矿与分散在煤层出露区的个体采矿者。本研究就煤炭开采对
整个区域经济的影响进行了评估，提出了其他可能的发展途径。首先，通过研究乌达城市的经济
发展，试图确定当地其他可发展和有前景的产业。然后，将研究结果与另外两个试验区汝箕沟和
古拉本进行比较。它们之间有重要的区别吗？同时，还列举了试验区内建筑和环境方面问题。本
研究的主要目标之一就是寻求一条该区可持续发展的途径。因此，还提出了未来发展采取的措
施、以及作者本人有关发展方面的建议。
193
Motivation
The Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China” has formulated a set of detailed
work packages. Besides technical and geological questions, the economic and social
impact of coal fires has been highlighted as being of great importance. Technology-based
solutions aimed at preventing coal fires might run counter to well-established mining
procedures and their implementation thus impact on the socio-economic well-being of
local people. Solutions which simply destroy local livelihoods have no prospects. Instead,
the results gathered within the framework of the project so far point toward sustainable
development in the region.
1 General Overview
“Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires in

North China” has its geographical focus in the mountainous region of the Helan Shan,
located in Ningxia and Inner Mongolia Autonomous Regions in northern China.
Wuda is a coal city on the Yellow River at the northern tip of the Helan Shan in Inner
Mongolia. It belongs to the administrative region of Wuhai, forming one of its three
districts. Whereas Wuda District numbers 130,000 inhabitants over an area of about
216 km², the city itself covers only 30 km² with about 30,000 citizens. Wuda is enclosed
by mountains and the Gobi Desert; the Yellow River, though, provides enough water for
the population and for some additional agriculture.
Rujigou is located in the west of Pingluo County in Ningxia, near the border to Inner
Mongolia, under the administration of the city of Shizuishan. It lies in the alpine central
region of the Helan Shan, at an average altitude of around 2,000 m. Pingluo County has
110,000 inhabitants, 30,000 of whom live in Rujigou. There are two other important coal
towns in Rujigou Coalfield – Baijigou and Dafeng. The whole area is 54 km² in size.
Gulaben lies just about 10 km west of Rujigou, across the provincial border in Inner
Mongolia. It belongs to Alashan Zuoqi County, which itself is larger than Ningxia
Autonomous Region, totaling about 80,000 km². The next city is Alashan, 45 km west of
Gulaben in the Tengger Desert. Gulaben is relatively small with only 2,500 inhabitants,
also at an average altitude of 2,000 m.
194
1.1 Economic classification
The major part of this study deals with the city of Wuda. Contact with the authorities for
interviews was much easier here than in the other areas under investigation. Besides, the
socio-economic situation in Wuda is more complex.
Two economic indicators are presented in the following – average income and gross
domestic product (GDP). These figures will help visualize the present economic situation
in the Helan Shan coalfields.
Table 1: Comparison of average income in Germany and China in 2003

Country Average Income (RMB)
Germany c. 270,000
P.R. China 14,040

Ningxia 12,981
Inner Mongolia 11,279
Mining sector in Inner Mongolia 9,760
Wuda urban-rural (2003) 5,865-2,994
Wuda urban-rural (2004) 7,328-4,139
Data source: Statistisches Bundesamt (2005); Zhang (2005); National Bureau of Statistics of China (2006)
While the difference in income between Germany and China clearly marks the latter
as a developing country, the difference between China’s provinces is also striking. The
economic indices of the eastern provinces, however, lie well above the national average,
the people in the western provinces, however, earn much less. Huge differences do also
exist on the economic sublevels, that is within separate sectors of the economy, as well
as for different spatial units. An inhabitant of Wuda’s rural areas earns only 30 % of the
national average (Table 1). On account of these values, the difficult economic situation of
the investigation area is not surprising. Even in view of China’s recent economic growth
(cf. gross domestic product (GDP) in Table 2), the structural deficits of the Helan Shan
coalfields remain remarkable.
195
Table 2: GDP growth in Germany and China in 2004

Country GDP Growth Rate (%)
Germany 1.7
P.R. China 9.5
Ningxia 12.2
Inner Mongolia 42.39

Data source: Statistisches Bundesamt (2005); Zhang (2005); National Bureau of Statistics of China (2006)
Statistics from China, economic data in particular, should generally be taken with a pinch
of salt. According to the Chinese National Bureau of Statistics, the national GDP amounted
to 5.877 trillion RMB in the first six months of 2004, growing by 9.7 % that year. However,
if one adds up the GDP’s released by all 32 local Chinese governments, the result should
be a national GDP of 7.027 trillion RMB at a growth rate of 13.4 %. This difference is an
indication of the huge political pressure under which the local and provincial governments
are working to fulfill the economic targets that Beijing has set for them. According to Chan
(2004), slightly higher economic development rates are reported to the next administrative
level in order to be competitive with other regions.
Despite the general inconsistency of figures, those for Wuda seem to be reasonable (Table 2).
Considering the transformation of the city during the past two years, with the Wuda
Industrial Park as its epitome, one cannot help but acknowledge an impressive economic
growth. Nevertheless, coal still reigns supreme as energy source, and the city’s economic
dependence on it is omnipresent.
1.2 Sustainable development in China
China is confronted with two important tasks – economic development and environmental
protection. Judging by official political communiqués and statements, the protection of the
environment is a national concern, particularly so since the beginning of the reform era in
the early 1980s. While ensuring that the rapid modernization of the country does not lose
momentum, Beijing has made the fight against pollution and wastage of natural resources
a high-ranking item on their political agenda. Active environmental protection is said to be
a primary concern.
After China’s participation in the UN Conference on Environment and Development in Rio

de Janeiro in 1992, the Chinese government adopted in March 1994 “China’s Agenda 21,”
196
where environmental and resource protection, and sustainable development are defined
as “basic political principles of China” (基本国策 = jƯbČnguócè ). xƯbù
Since then, a set of laws,
dàkƗifƗ
guidelines, and measures has been specified to support the harmonious development of
economy, society, and environment.
Just before the turn of the century, the Chinese government made a strategic decision for
the “development of west China” (西部大开发 jƯbČnguócè= xƯbù dàkƗifƗ ). The main objectives of this
large project are infrastructural development, environmental stabilization, and rational use
of natural resources – each of them meant to help overcome the general backwardness of
China’s west.
One of the region’s most daunting problems are coal fires, especially in the autonomous
regions of Xinjiang, Ningxia, and Inner Mongolia. As a source of substantial pollution,
they are of grave environmental concern. However, since they consume vast quantities of
precious resources, coal fires are also a serious economic setback. Fire fighting has been
practiced for more than 40 years in China. Yet modern technical equipment and efficient
management to tackle this difficult task are often lacking.
Despite all political vows to set China on the path of sustainable development, there
remain a good number of restrictions in the area under investigation. In the isolated
coal districts, but also in urban areas and in those industries with high levels of energy
consumption, a continuous and clean supply of energy is far from guaranteed. To make
up for the old power stations’ frequent break-downs, many new coal-fired stations
are built, which – in terms of technical standards – are hardly more likely to steer the
country toward a sustainable future.
The following statement summarizes the polarity which mars sustainable development in
the Helan Shan:
Economic development vs. environmental protection

Daily-life problems vs. ecological awareness
Especially in the field of environmental protection, both financial and technical support are
necessary. This mainly includes financial backing from the Chinese government, but also
foreign capital.
197
2 Socio-Economy of the Helan Shan Region
The entire study area in the Helan Shan is dependent on coal mining. While Wuda is a
growing provincial city, whose main but not exclusive economic base is the coal industry,
Rujigou and Gulaben are completely reliant on coal. Agriculture here is impossible and
industrial enterprises and employers other than those of the coal mining industry do not
exist. To give a most comprehensive picture of economic possibilities in the region, this
study thus focuses on Wuda.
In Wuda, all economic sectors have been expanding in recent years. While the primary
sector grows at a comparatively slow pace, high growth rates have been achieved in the
industries (cf. Chapter 2.1) – 5 to 10 % on average. The agricultural sector grew by 15 %,
the industry sector by an impressive 70 %. Coal mining accounted for approximately 25 %
of Wuda’s GDP in 2004 (Zhang 2005).
Figure 1: Workers’ dwellings in front of a coal mining complex in Wuda
The central government in Beijing has established a special office to inspect the quality
of coal countrywide. Coal mining is organized in the national Five Year Plan, i.e. Beijing
198
specifies the yearly amounts of coal to be extracted and passes these targets down to the
city level. However, the choice of where to mine rests with the local administration. The
city therefore has a certain scope to develop its own strategies as long as it meets the
production targets. The amounts of coal extracted are generally well above the values
from Beijing.
In summary, a considerable growth can be observed for Wuda, even in the Chinese context.
However, the present figures have been released by the local government, which means
they are most likely to be optimistic if not willfully unrealistic. Our own observations for
the years 2003 to 2005, if visual only, do confirm a certain momentum of growth; they can
however not be quantified.
The same problem exists for information on unemployment. Numbers for the whole
of China vary from < 5 % to > 30 %. Sure enough, the rates for the western provinces
and for the Helan Shan are above average. Our own observations and discussions
with local people point toward a higher rate in Wuda than in other places of the Helan
Shan. In addition, Shenhua Group, the major mine owner in Wuda, cut down operation
costs during the time of the study and reduced the number of staff considerably. While
in 2000, 22,000 people were employed in the Wuda mines, this number dropped to
approximately 20,000 in 2002. With the closure of Huangbaici Mine, another 10,000
workplaces were lost (Jia 2005).
Besides active workers, the Shenhua Group is also responsible for about 12,000 pensioners
and their families. On estimate, 8,000 men are currently looking for work or are temporary
workers. These workers live at the lowest level of revenue with a minimum salary of
RMB 238 per month (for Wuda) – this is below subsistence level!
During a field trip in 2005, an overall improvement of infrastructure was observed in

Wuda. The city center is that of a typical Chinese provincial city, with all its accompanying
benefits and challenges. Traffic accessibility is fair; Wuda is connected to national
highways and to the national railway network. Two airports, Yinchuan and Wuhai,
with regular national connections are within short distance (150 km). The road system
connecting Wuda with other local towns and villages has been improved and extended in
recent years; yet most streets off the main highways are still in poor condition and allow
for low speed only. It should be added, though, that this is not a problem specific to Wuda.
It applies to the whole of China.
199
Figure 2: Development in Wuda: The city’s main street
Compared to the rest of the country, the Helan Shan region is still lagging behind socio-
economically. People here have to contend with a host of (daily-life) problems; the
economy, despite all positive signs of growth, should not be portrayed in too bright a light
either. Wuda has developed a lot in the last years, but this growth has to be seen in relation
to where it came from – grinding poverty!
2.1 Situation of the mining industry sector
Coal mining has a long tradition in Wuda, notwithstanding the city’s only very recent
foundation in 1976. Before the founding of the People’s Republic of China (proclaimed in
1949), private small pits were run in the area for supply of farmers with coal for heating
and cooking. In the first years after 1949, these local minings produced up to 100,000 t of
coal per year. Coal fires did not occur at that time. As with the whole country, the “Great
Leap Forward” marks an important period in the Helan Shan. Coal extraction stepped
up and many people settled down in the region or were resettled there. Production in the
years from 1959 to 1978 rose to more than 1 million t yr-1. After the discovery of large coal
deposits, the city of Wuda was founded in 1976 and construction of three mines began
200
(Wuhushan, Suhaitu, and Huangbaici), each with a production capacity of more than
1.2 million t yr-1.
In response to the growing population and its ever growing demand for raw materials and
energy, the number of small private pits with high cumulative production shot up in the
1980s. Ten coal fires were caused by these pits, another eight by the three big coal mines
(Guan 2005).
Figure 3: Small private pit in the Helan Shan coalfields near Wuda
A much bigger problem, however, grew to be coal transportation. In an effort to limit

transportation distances and costs, the government decided to force electrification in the
mining sector and build several new coal-fired power stations. At the same time, other
branches of industry started to settle in Wuda and quickly rose to considerable economic
weight. Coal yet continued to dominate the region’s economy, either in mining or in the
processing industries.
Many small private pits were closed at the end of the 1990s as a consequence of
legislative changes instigated by the central government to counteract the rising
number of accidents in coal mines. Private pits, often inadequately equipped, were
201
made responsible for most of them. A lot indeed has been done over the years by the
Government. But given the immense size of the country, and the basic fact that “Beijing
is far away,” as the saying goes, it seems unlikely that actually all small pits were closed.
According to local reports and our own observations, many pits simply re-opened on
the spot or nearby. Larger coal enterprises may also have an interest in sustaining small
pits because their coal can be bought up at low prices. Obviously, this would explain
why some small mines are equipped with very expensive modern technology that would
normally be beyond their means. Today, coal production in small pits is estimated to
account for less than 10 % of all mining activities in the Helan Shan (interview on site).
Wuda produced about 4.2 million t of coal in 2004. Approximately 5 %, i.e. most high-
quality coal, was exported to Japan. The remainder was used in coal-fired power stations
in Wuda and neighboring counties. In order to comply with national regulations, which
stipulate that the proportion of mineral matter in industrial coal be below 40 %, coal from
other mining areas had to be bought on occasion and blended with Wuda coal for burning.
Wuda is an extremely important location not only for coal mining, but also for the coal
processing industries. The district accommodates a good number of state enterprises as
well as small and medium-sized private companies, all based on coal. Near the big mines,
numerous small enterprises can be found, which produce or trade by-products of coal, for
example fertilizers or raw materials for ceramics and the chemical industry, selling them in
China and all over the world.
“Without local coal mining, Wuda would not do well; however, it could survive,” say the
staff of the Planning Office of the city. Our own observations support such a view. Wuda
has developed rapidly into an industrial city of commercial importance even beyond the
Helan Shan region. Its economic strength rose mainly thanks to four core sectors: (i) the
salt-chemical industry, (ii) the coal-chemical industry, (iii) coal processing, (iv) and electric
power generation.
An important factor in the economic development was and still is the Wuda Industrial
Park opened in 1998. The park is located in the southwest of the city and has a total
area of c. 20 km². So far 72 companies have settled there; total investment amounts to
RMB 7 billion (Wuda Local Government 2004).
The coal fires in Wuda seriously threaten this development, especially if they are not
brought under effective control but continue to expand in the future. Also, the influence of
202
small private pits has to be limited. Without intervention and targeted projects, sustainable
development in Wuda is not feasible.
2.2 Environmental problems and agriculture
Thanks to the Yellow River, agriculture is possible in and around the arid area of Wuda;
it contributes about 5 % to the region’s economy. This is a big and important difference
to Rujigou and Gulaben, which both are pure coal mining settlements. Water is the
limiting factor for economic development in these cities; it needs to be transported to the
mountains mostly by tank trucks. Water, however, also plays a significant role in Wuda.
The Yellow River may be an invaluable resource for the city, but water use/wastage (also
of groundwater) has a grave effect on the environment.
All industrial enterprises in Wuda need water from the Yellow River for their production.
They pay RMB 0.6 per m3 (c. EUR 0.06), private households pay RMB 1; for agricultural
purposes, water is free of charge. This regulation is not acceptable from an environmental
point of view since it invites careless use of resources, which is particularly harmful in
Wuda, a fragile urban oasis amidst the desert. The greatest concern of the city’s Water
Management Office consequently is to take care that water in the district is used in an
efficient and sustainable way. There exists an obligatory water management plan, and the
office also controls the use of groundwater. Unfortunately, the responsibilities for drinking
water and wastewater are with other offices.
A severe problem in Wuda is the lack of wastewater treatment. Since the very beginning
of industrial activity in the city, sewage has been discharged directly into the Yellow River,
treated only in a circulation pond. At present, c. 70 % of the wastewater seeps away where
it is produced, 30 % is discharged into the Yellow River. According to local interview
partners, 57 % of the wells in Wuhai County are polluted. Although every industrial
enterprise must pay an annual environmental protection fee, it is not clear to what extent
and for which purposes this fee is used. Apparently, a chemical purification plant is under
construction and will, in two steps over the next few years, reach a capacity of 30,000 m³ d-1.
RMB 65.7 million have already been invested to lay more than 51 km of sewage pipes in
Wuda District. In terms of environmental management and protection, there is obviously
still a very long way to go before the Helan Shan region can enter a phase of sustainable
development.
203
3 Future Development of the Helan Shan Region
One important factor in the future development of the Helan Shan are its natural resources;
in fact, they curtail prospects from the outset. Neither in the short, nor in the long run
will the region be able to evolve into an area of outstanding economic performance. The
living standards of the majority of the population are still well below the Chinese average,
and the environment has been scourged by coal mining to such an extent that a rapid
improvement is unlikely.
Most parts of the Helan Shan, remote as they are, lack basic means of infrastructure;
their prospects for development are bleak. If the coal fires here are not brought under
effective control, the economy of the whole region will turn to ashes along with the coal.
An increasingly over-used and polluted Yellow River is sure to become a determining
factor in the economic development of the region. Environmental regulations and
standards must be given political precedence in the future; water use must be controlled
more strictly.
For a planner making projections about the future development of the Helan Shan, it
is essential to assume that the coal fires cannot be extinguished within the next years,
and thus will continue to greatly impact on population and environment. One important
aspect is the high risk that miners and workers are taking when extracting or processing
coal in fire areas (which is still very common!). Also, the population is and will be
affected by the immense air pollution from the fires. The soil and its vegetation cover are
being destroyed.
Often, the political will to extinguish coal fires only develops in those areas where:
(i) easily exploitable resources are endangered, or (ii) a direct threat to human life exists.
From an economic point of view, this behavior might be acceptable, but its ecological
implications are dire.
Political measures in favor of sustainable development and their implementation in

the region are barely underway and tend to be very selective. Local politicians feel
bound to prefer economic growth over sustainable development because their careers
within the party apparatus depend on it. Who will be advanced after all? – Someone
presenting growth rates of only 5 % in his term, or someone with an impressive 11 %
in the same time?
204
With regard to daily life in the Helan Shan and the basic needs of the population, it is
difficult to make drastic decisions – decisions which environmental protection would
otherwise necessitate. How can sustainable development in the region take place? Is there
a way to reach a reasonable standard of living for the population by way of sustainability?
And what must measures to this effect look like?
Sustainable development in the Helan Shan will not occur by chance or without external
help. It needs input from the Government as well as from international projects, which can
provide capital and (scientific) expertise. China is confronted with two important tasks –
economic development and environmental protection. The consequences of high economic
growth and increasing prosperity, so far, have been massive problems in the environment,
especially water and air pollution.
With this in mind, our attention must invariably turn back to the politicians and to
what they practice and preach. Of course, this is not solely a Chinese phenomenon. For
sustainable development in the Helan Shan region to be achieved, it is indispensable that
higher technical standards in the mining sector and the processing industries be enforced.
These changes have to go hand in hand with the global implementation of Environmental
Impact Assessment and the development of instruments such as the Clean Development
Mechanism. Admittedly, the standards of Chinese power plants have improved of late,
but there are still a lot of old and outdated facilities, which, from an ecological point of
view, should have long been shut down. On the same note it should be added that this
improvement in technical standards is only relative, i.e. relative to what was in use in
China before. There still is a great gap between Chinese and international standards,
which needs to be overcome by investments and scientific exchange. A huge problem
is to secure compliance with technical and safety regulations. As said earlier, local
implementation can hardly be controlled on a permanent basis. A few weeks or months
after their shut-down, illegal pits will re-open, even if official statistics state the contrary.
Single landmark projects are welcome sources of income for local politicians eager to
attract foreign capital to their fiefs. Chinese environmental and economic politics should
not be presented too negatively though. As a matter of fact, China’s western provinces
have to manoeuver their economy from the 19th century right into the 3rd millennium. This
task cannot be completed without friction. China, after all, is about to choose an acceptable
way, even if there still is a huge backlog to clear before the country can arrive at a status
similar to that of the West.
205
Two things seem to be important for a positive development in the future. First, China
has to strengthen and evolve its bi- and multilateral relations. Co-operations, joint
ventures, and especially the import of western know-how and industrial technology
are means to this end. They have a huge potential for the Chinese economy. Second,
considering its unprecedented economic growth, China has to deal with its emissions.
The Kyoto Protocol is a promising tool in this respect. Today, developing countries like
China and India are exempted from its regulations – however, it is these countries that
already account for a large share of global emissions (with upward trend). Even if they
commit themselves to emission reductions, a global contract or protocol for control is
needed. The Kyoto Protocol, nevertheless, is a milestone for future politics, from which
both China (plus other developing countries) and the developed world can profit.
References
Chan, F., 2004. Lies, damn lies and Chinese statistics. Comment on Chinese statistical data. Asia Times Online,
October 2004. 12 November 2005 http://www.atimes.com/atimes/China/FJ23Ad01.html
Guan, H.Y., 2005. Personal interview / Oral communication. (Prof. Guan Haiyan is a renowned Chinese
expert on coal fires and has over years of research accumulated an impressive knowledge on
Wuda Coal Fire Area.)
Halik, Ü., Bandelow, F.K., and Schulz, J., 2003. Land use and socio-economic aspects of the Helan
Shan in Ningxia and Inner Mongolia, China. Field report for Work Package 2140-2150, as a
Exploration, Extinction, and Monitoring of Coal Fires in North China.” Xinjiang University,
Urumqi, Xinjiang Autonomous Region, P.R. China; Deutsche Montan Technologie GmbH
(DMT), Essen, Germany; and Dortmund University, Dortmund, Germany.
Halik, Ü., Bandelow, F.K., and Schulz, J., 2005. Socio-economic aspects of the local mining sector and
alternative sources of income of the Helan Shan region in Ningxia and Inner Mongolia,
China. Field report for Work Package 2150, as a contribution to the Sino-German Coal Fire
Research Initiative “Innovative Technologies for Exploration, Extinction, and Monitoring of
Coal Fires in North China.” Xingjiang University, Urumqi, Xinjiang Autonomous Region,
P.R. China; Deutsche Montan Technologie GmbH (DMT), Essen, Germany; and Dortmund
University, Dortmund, Germany.
Jia, Y.R., 2005. Personal interview / Oral communication. (Jia Yaorong is the chief engineer of the Wuda
Coal Fire Fighting Team and affiliated to the Wuda Reform and Planning Office.)
NBS, 2006. Chinese statistical data. National Bureau of Statistics of China (NBS), Beijing, P.R. China.
January 2006 http://www.stats.gov.cn/english/statisticaldata/index.htm
Statistisches Bundesamt, 2005. Deutschland im Internationalen Vergleich 2005. Statistisches Bundesamt,

Wiesbaden, Germany. April 2005 http://www.destatis.de/download/d/veroe/dimintvergleich.pdf
Wuda Local Government 2004. Wuda District – A new industrial city. Marketing brochure to attract
investors. Wuda Local Government, Party Committee Wuda, Wuda, Inner Mongolia
Autonomous Region, P.R. China.
206
Zhang, B.L. 2005. Offizieller Bericht über die Durchführung des Plans zu GDP und ökonomischer
Entwicklung 2004 und der vorläufigen Planung für 2005. Internes Dokument des Reform- und
Entwicklungsamtes, Wuda, Inner Mongolia Autonomous Region, P.R. China. (in Germen)
207
Part IV
Basic Processes and

Investigations
Influence of Environmental Parameters on
the Self-Ignition Behavior of Coal
Christian Lohrer1, Martin Schmidt2, and Ulrich Krause3
Federal Institute for Materials Research and Testing (BAM), Berlin, Germany
1
Division II.1 “Gases, Gas Plants”
2
Division II.2 “Reactive Substances and Systems”
3
Division VII.3 “Fire Engineering”
Abstract
Self-ignition of coal is a major cause of coal fires. Coal fires occur in coal mines, virgin natural coal
seams, and coal deposits of all varieties. Environmental parameters influence the self-ignition behavior
of coal. These parameters include ambient temperature, oxygen content (in the fuel and its immediate
vicinity), direct contact with water (“wetting”), relative humidity, as well as convection of air (wind).
This paper presents experimental investigations of the influence of environmental parameters on the self-
ignition behavior of coal. Experiments were carried out on samples of Chinese black coal and German
lignite coal. In preliminary tests, material properties such as thermal conductivity, density, calorific value,
moisture content, elemental composition, ash content, porosity, and specific surface were determined.
Subsequent experiments revealed that reduced levels of oxygen (“inerting”) significantly increase the
self-ignition temperature and combustion time of coal. Combustion temperature decreases with reduced
oxygen content due to less reactivity. Moisture content of coal and humidity of ambient air also affect
self-ignition. If coal is exposed to water vapor, e.g. dried coal piles in moist atmosphere, heat is generated
and transported into the material by condensation effects and water adsorption. If under these conditions,
coal is stored only slightly below its self-ignition temperature, it may ignite. Direct contact with water
over the surface of stored coal has the same effect – the water is adsorbed by the coal particles in a heat-
generating process, which may lead to ignition. Convection of air may also alter self-ignition behavior.
Experiments showed that a uniform surrounding air flow slightly increases the self-ignition temperature
of coal.
A numerical model was developed to describe the effects of self-heating, ignition, and extinction of coal,
as well as moisture transport. Experimental results from the above investigations – kinetic parameters, for
example – were used as input for the model. Lab-scale experiments are important to validate simulations
of this kind and guarantee a realistic scale-up to fit coal fire scenarios.
210
摘要
煤的自燃是煤火的主要原因。煤火发生在煤矿、未开采的自燃煤层和各种各样的煤矿中。环境参
数影响煤的自燃行为。这些参数包括了环境温度、氧含量（燃烧体及其周围）与水的直接接触
（“润湿”）、相对湿度以及空气的流通（风）。本研究介绍了环境参数对煤自燃行为影响的试验
研究。试验基于中国的黑煤和德国的褐煤样品进行。试验前，测定了样品的各种特性，如热传导
率、密度、发热值、水分、化学成分、含灰量、孔隙度和表面类型等。随后的试验显示氧气水平
降低（“惰性”）增加了自燃温度和显著延长煤的燃烧时间。反之，由于反应过程弱而氧气成分减
少，燃烧温度降低。煤的水分和环境空气的湿度也影响煤的自燃。当煤暴露在水蒸气下，如潮湿
环境下的干燥煤堆，热量是通过冷凝效应和水分吸收生成和传输的。如果在这些条件下，煤在稍
低于自燃温度条件下储存，煤就可能变得接近自燃到自燃状态。储存煤表面与水直接接触具有相
同的效果——煤在发热过程中煤颗粒吸收了水分，导致煤自燃。最后，空气对流也可以改变自燃
行为。试验表明均匀的环境空气流动略微增加了煤自燃的温度。
数值模型描述了煤的自热、发火和熄灭效果以及水分传输。上述的研究试验结果（如动力参数）
用于数值模拟的输入参数。试验室的各种实验对于确认数值模拟结果和确保现实条件下适用的煤
火解决方案是非常重要的。
1 Introduction
Numerous studies on the self-ignition behavior of combustible bulk materials have

been conducted over the last decades. Early pioneers were Semjonov (1928) and Frank-
Kamenetzkii (1959), who developed the thermal explosion theory for flammable systems.
The theory holds that if the active molecule sides of flammable materials are exposed to
air (such as in porous coal), oxygen will be adsorbed into the material and induce a series
of heterogeneous surface reactions, whose rates rise exponentially with temperature. A key
parameter in this process is the volume-to-surface ratio (V/A ratio) of the bulk material. Heat
generation as a result of chemical oxidation is related to volume; heat transfer is dependent
on the surface of the reactive system. According to the thermal explosion theory, self-ignition
occurs when the ambient temperature rises above a critical value (self-ignition temperature,
SIT) and the heat generation within the material exceeds the heat loss to the surrounding air.
Besides geometrical parameters, material properties such as porosity and particle size also
influence the self-ignition behavior of coal. Hensel et al. (2000) found that coal deposits
with a wide range of particles had much lower SITs than deposits with a narrow range,
small particles in the former having the double effect of increasing bulk density and
specific surface area. Whereas increased bulk density raises the SIT through increased
thermal conductivity (= increased heat dissipation), an enlarged specific surface helps
building up “stored energy” in the system and thus lowers the SIT. The latter, apparently,
is the dominating effect here. Ren et al. (1999) and Küçük et al. (2003) went to show that
indeed the potential for self-ignition of materials increases with decreasing particle size
211
– a fact which they attributed to the increased specific surface of materials with smaller
sized particles. Wang et al. (1999) carried out investigations of the influence of porosity
on the oxidation rates of different coals. They found that oxidation reactions in deposits
consisting of high-porosity coal are kinetically controlled.
The thermal conductivity of a coal seam and its surrounding bedrock affect the temperature
field and the oxidation processes of a coal fire. Two major effects counteract each other
here: on the one hand, increased thermal conductivity speeds up the heating process
within the coal seam. However, as soon as the temperature in the seam exceeds that of
the surrounding bedrock, the same increased thermal conductivity will facilitate heat
transfer to the extreme ends of the seam and beyond, thus entailing a significant cooling
effect. Numerical simulations performed by the authors of the present study (Lohrer
2006) confirmed that the SIT of coal rises with increasing thermal conductivity (all other
parameters being kept constant).
The environmental – outside – parameters which influence the self-ignition behavior of

coal are ambient temperature, oxygen content, direct contact with water (“wetting”; e.g.
rain), relative humidity, and convection of air (wind).
Wiemann and Scholl (1985) and Schmidt et al. (2003) measured SITs as a function of
oxygen content in the immediate vicinity of coal deposits. Their investigations were carried
out on German lignite coal. When the oxygen content was lowered (“inerting”), the SIT
increased. Even at extremely low oxygen contents of 1.3 to 3 %, self-ignition occurred,
albeit at lower combustion temperatures and extended combustion times.
Water (liquid and/or vapor) affects the self-ignition behavior of porous materials like
coal. Depending on the stored material’s specific surface, water is adsorbed by its active
molecule sides in a heat-generating process (frequently termed “wetting”). This heat is
transported into the colder parts of the bulk phase. The heat of “wetting” will increase with
a lower initial water content, there being more space for adsorption of water molecules
(McIntosh et al. 1994; Chong & Chen 1999). Similarly, a humid environment makes self-
ignition more likely due to condensation followed by “wetting” – both heat-generating
processes (Stott 1960; Ren et al. 1999). Gray et al. (2002) demonstrated the effect by
pouring water onto a coal stockpile. With the released adsorption heat, the temperature
inside the pile rose by about 25 K. Ignition, however, did not occur.
Convection of air (wind) outside a coal deposit has two effects similar to those of increased
thermal conductivity. On the one head, increased air flow enhances the oxygen supply to the
reaction zone, thus leading to higher combustion velocity. On the other hand, heat transfer
212
away from the combustion zone is also amplified, making for a substantial cooling effect.
Fierro et al. (1999, 2001) investigated self-ignition in stockpiles of Spanish black coal. They
reasoned that the stock would not be in danger of self-ignition if convection outside was
either sufficiently low (= limited oxygen supply) or high (= cooling effect dominating).
The present paper discusses experimental investigations and corresponding numerical

simulations of the influence of environmental parameters on the self-ignition behavior of coal.
These experiments were performed on samples of Chinese black coal and German lignite coal.
2 Experimental Set-Up
Table 1 in the Appendix lists the origin and general characteristics of the coal samples
used in the experiments presented in this paper. Other properties and the methods used
to determine them are given in Table 2. All samples consisted mainly of coal dust and a
smaller fraction of larger particles. The experiments were performed on coal as taken from
the site of sampling (i.e. no mechanical processing after sampling).
The SITs of the coal samples as a function of V/A ratio and oxygen content in ambient air
were determined according to German VDI Guideline 2263 (VDI 1990), a standard test
method for this purpose. The experimental set-up is presented in Figure 1.
Figure 1: Experimental set-up to determine SIT of coal as a function of V/A ratio and
oxygen content in ambient air
213
The set-up consisted of a laboratory oven with natural convection (i.e. without fan),
commonly used for hot storage tests. To adjust the oxygen content in the immediate
vicinity of the coal samples, an inner steel chamber was inserted. This chamber was flushed
with an oxygen/nitrogen mix of desired composition, itself preheated with a copper pipe
coil and pumped into the chamber at 100 dm3 h-1. Oxygen content inside the apparatus was
monitored continuously during experiments with an oxygen analyzer. Before placement,
the samples were evacuated in a vacuum vessel and flushed with the oxygen/nitrogen mix.
Their weight was measured before and after the experiments with a standard scale.
The temperature in the center of the dust samples and in the oven (here: the inner chamber)
was measured at intervals with thermocouples. Cylinders of wire gauze with volumes of
31, 100, 200, 400, 800, 1,600, 3,200, 6,400, and 12,800 cm3 were used as dust containers;
they had a consistent length-to-diameter ratio of 1 (equidistant cylinders). The SITs of bulk
materials are generally determined in an iterative procedure. Once ignition occurs, a new trial
is started at slightly reduced oven temperature, and the procedure replicated until no ignition
occurs. The maximum difference in oven temperature between a test with ignition and a test
without ignition was set to be 4 K. According to VDI Guideline 2263, the SIT of a flammable
material is defined as the highest oven temperature at which no ignition occurs.
Experiments to determine the influence of “wetting” on coal SIT were carried out for
lignite coal only, with equidistant wire gauze cylinders of 3,200 and 6,400 cm3. The
samples were placed in a hot storage oven at temperatures below their SIT. As soon as the
temperature inside the samples reached the oven temperature, the oven was opened and
1,000 cm3 of water (at oven temperature) were poured onto the samples. This procedure
was replicated for various oven temperatures.
A different experimental set-up was used to investigate the influence of water vapor on the
self-ignition behavior of lignite coal (Figure 2).
As in the “wetting” experiments, equidistant wire gauze cylinders of 3,200 and 6,400 cm3
were placed in a preheated hot storage oven. The temperature development in the samples
and in the oven was recorded with thermocouples. When the temperature in the samples
had reached the oven temperature, saturated water vapor from an evaporator outside the
oven was sprayed onto the samples through a perforated pipe system at a mass flow rate of
1.6 × 10-4 kg s-1. The samples were stored at different oven temperatures below their SIT.
In another series of experiments, the influence of forced air convection on the self-ignition
and combustion behavior of coal was investigated. Figure 3 shows the set-up to determine
the influence of uniform surrounding air flow.
214
Figure 2: Experimental set-up to determine influence of water vapor on self-ignition

behavior of lignite coal
Figure 3: Experimental set-up to determine influence of uniform surrounding air flow on

self-ignition behavior of lignite coal
215
Samples of lignite coal (400 cm3) in wire gauze cylinders were placed in a preheated oven,
surrounded by a concentric system of nozzles. Pressurized air, preheated in a copper pipe
coil, was pumped around the coal. The flow velocities at the nozzles were measured with
an impeller anemometer. The SITs of the samples were determined as a function of various
air flow rates (0, 3,000, 5,000, and 6,000 dm3 h-1).
Fire propagation in coal can take two different directions. “Reverse smoldering” occurs
when coal self-ignites internally. The smoldering reaction starts in the center of the coal,
and propagates toward the surface, whence oxygen enters the coal. In contrast, “forward
smoldering” involves a parallel movement of reaction front and oxygen in the coal. This
occurs, for example, when a fully developed fire moves through a seam. The unburned
coal in front of the fire is influenced by the heat generated in the reaction. Pre-reactions
(pyrolysis) will take place and heat-induced cracks develop in the contact area. To
investigate the burning behavior of coal under conditions of “forward smoldering,” the
following experimental set-up was chosen (Figure 4).
Figure 4: Experimental set-up to determine combustion velocity of coal in “forward smoldering”
The set-up consisted of a reaction tube filled with coal, several thermocouples placed at
its longitudinal axis to measure the temperature in the sample, and a hot air blower at
one end to heat the coal in the tube. The temperature of the air blower was kept constant
at 500 °C, at a flow velocity of 4 m s-1. The distance between the blower and the coal was
1 to 2 cm. Two different coal fractions of WU2 were investigated: The first consisted of
particles > 10 mm (without dust), the second of dust and gravel < 10 mm.
216
3 Experimental Results
3.1 V/A ratio
The influence of V/A ratio on the self-ignition behavior of Chinese black coal and German
lignite coal as measured by the standard hot storage procedure described in Chapter 2
(Figure 1) is presented in Figure 5 (samples in air; 21 % oxygen).
Figure 5: SITs of different Chinese black coals and German lignite coal as a function of V/A ratio
Evidently, the SITs decreased with increasing V/A ratio. The relatively smaller surface
(compared to volume) in the larger samples allowed less heat to be transported to the
surface and given off to the surrounding air. From the above graph, drawn according to
VDI Guideline 2263, a linear dependence of SIT and V/A ratio can be extrapolated.
The SITs of the samples of Chinese black coal and German lignite coal differed
significantly. The Chinese samples had much higher SITs due to the higher amount of
volatiles in lignite coal.
The effect of “aging” is apparent in the coals from Wuda and Rujigou. In both cases, the
weathered coals (WU1 and RU1) showed higher SITs than the unaffected coals (WU2,
WU3, WU4, and RU2) from inside the seam.
217
3.2 Oxygen content
The effect of “inerting” is displayed in Figure 6. Here, the oxygen content in the air around
the coal samples was reduced from 21 % to 6 %.
Figure 6: SITs of different Chinese black coal samples as a function of oxygen content in
ambient air
It shows clearly that the SITs increased with reduced oxygen content in ambient air. The
lowered oxygen also led to much slower reaction rates and lower combustion temperatures.
3.3 Direct contact with water (“wetting”)
The influence of “wetting” on the self-ignition behavior of lignite coal is illustrated in

Figure 7. The SIT of the 3,200 cm3 lignite coal samples had been determined at 103 °C, for
an oxygen content of 21 % in ambient air prior to the “wetting” experiments. The storage
temperatures of 98 °C and 92 °C were thus at subcritical levels, which under normal
circumstances should not have led to self-ignition. The sample kept at 98 °C, however,
ignited after “wetting.” Here, the heat generated during absorption was sufficient to pass
the threshold from subcritical to supercritical behavior (“wetting”-induced ignition). In
replications of the experiment at lower temperatures, ignition only occurred at temperatures
above 92 °C, which means that the SIT of this volume of lignite coal was lowered by 11 K
from 103 to 92 °C as a result of “wetting.”
218
Figure 7: Temperature development inside 3,200 cm 3 sample of lignite coal at two

different oven temperatures with addition of water (“wetting”)
The SIT of samples of 6,400 cm3 was determined at 95 °C, for an oxygen content of 21 %.
“Wetting”-induced ignition here occurred at an oven temperature of 89 °C, i.e. 6 K below
the normal SIT.
3.4 Relative humidity
Figure 8 shows the results of the ignition experiments with water vapor. Next to adsorption
of water molecules into the coal, condensation was a preeminent source of heat here.
Figure 8: Temperature development inside different lignite coal samples at a constant oven
temperature of 98 °C with addition of water vapor
219
The storage temperature of 98 °C was subcritical for all volumes (400, 800, 1,600, and
3,200 cm3). 15 min after spraying, the temperature inside the samples started to rise as a
result of condensation and “wetting.” Heat generation and heat release to the surrounding
air finally lost their equilibrium and caused self-ignition.
The same effect can be seen in Figure 9 for a larger volume of lignite coal (6,400 cm3);
temperature development inside two samples at different (subcritical) oven temperatures is
shown. After the sample at 82 °C was sprayed, ignition behavior changed from subcritical
to supercritical. The maximum temperature at which no ignition occurred after spaying
was determined at 72 °C (23 K lower than the normal SIT).
Figure 9: Temperature development inside 6,400 cm3 sample of lignite coal at different
oven temperatures with and without addition of water vapor
3.5 Convection of air (wind)
Forced convection outside bulk materials affects self-ignition behavior and combustion
velocity through changed heat transfer to the outside. Following Bähr and Stephan (1998),
the flow velocity of the air outside the coal cylinders in the experiments was converted into
the median external heat transfer coefficient αm using equations (1) to (5):
(1) ,
(2) ,
220
(3) ,
(4) ,
(5) .
In these equations, L is the characteristic length of the cylinder (diameter dcylinder π/2),

λair the thermal conductivity of air (W m-1 K-1), v the middle flow velocity (m s-1), νair the
viscosity of air (m2 s-1), ρair the density of air (kg m-3), and cP,air the specific heat capacity
of air (J kg-1 K-1). Nu, Re, and Pr are dimensionless numbers; they describe the ratio of
convective to diffusive heat transfer (Nu) as a function of the ratio of internal to viscous
forces (Re) and the ratio of viscous to thermal diffusion rate (Pr). Equations (1) to (3) are
valid in the range of 10 < Re < 107 and 0.6 < Pr < 103.
Figure 10 depicts the SITs of a 400 cm3 lignite coal sample as a function of the external
heat transfer coefficient.
Figure 10: SITs of 400 cm3 sample of lignite coal as a function of external heat transfer
coefficient
There is almost no correlation with outside convection, the SIT rising but very slightly with
increasing heat transfer coefficient. This effect is due to the simultaneously improved heat
transfer to the surrounding air (cooling). It was not possible to increase convection over
a certain limit (heat transfer coefficients over 18 W m-2 K-1) as dust particles would have
been blown away through the wire gauze walls of the sample cylinders. The convection
experiments prove that self-ignition of coal is kinetically controlled and not depended on
oxygen supply.
221
Nevertheless, enhanced convection had a significant influence on the combustion behavior

in the experiments. This can be seen in Figure 11, where the temperature curves for 400 cm3
lignite coal samples at different oven temperatures and heat transfer coefficients are plotted.
Figure 11: Temperature development inside 400 cm3 samples of lignite coal at different
oven temperatures and heat transfer coefficients
The heating period shortened with increasing outside convection as heat transfer into the
coal samples was enhanced. The increased availability of oxygen at the reaction zone
increased the combustion temperature, and shortened the combustion time.
3.6 “Forward smoldering”
The results of the “forward smoldering” experiments are presented in the following two
diagrams. Figure 12 shows the temperature curves in the combustion tube for Chinese
black coal from WU2 with a particle fraction of > 10 mm. At the thermocouple located
farthest from the heater (40 cm), the maximum temperature (970 °C) was reached
c. 85 min after ignition. Thus, combustion must have propagated at a velocity of c.
0.47 cm min-1 (2,470 m yr-1).
Figure 13 shows the temperature curve in the combustion tube for coal from WU2 with
a particle fraction of < 10 mm. The maximum reaction temperature in this sample was
600 °C; temperatures being generally lower than for the coarser coal. The reaction front
reached the last measuring point at 40 cm distance approximately 2,000 min after ignition.
This amounts to a combustion velocity of 0.02 cm min-1 (105 m yr-1). A possible reason for
the pronounced difference in combustion behavior is the varying distance between coal
222
particles. Air can pass coarse coal with low resistance. Thus, a large amount of oxygen is
transported to the reaction zone. If, however, the sample consists of smaller particles, the
flow resistance is comparably high. Consequently, air flow velocity decreases. This, in
turn, leads to lower combustion temperatures and propagation velocities.
1200
d ista n ce fro m h e a te r
1000
0 cm (e d g e o f tu b e )
1 0 cm
Temperature (°C)
800
2 0 cm
4 0 cm
600
400
200
0
0 50 100 150 200 250 300 350 400
Time (min)
Figure 12: Temperature development in combustion tube filled with Chinese black coal
from WU2, particle fraction > 10 mm
Figure 13: Temperature development in combustion tube filled with Chinese black coal
from WU2, particle fraction < 10 mm
223
4 Numerical Simulation
The numerical model used in this work is based mainly on the investigations carried out
by Schmidt (2001), Krause et al. (2005), and Lohrer et al. (2005). The aim here is not to
describe the model in minute detail but to give an overview and present possibilities for
application to the coal fire problem.
For the model, the combustion of coal was assumed to be a four-step reaction,
incorporating coal decomposition, pyrolysis, as well as gas and char combustion. The
combustion process is illustrated in Figure 14.
Figure 14: Four-step reaction model of coal combustion
In three main reactions, coal reacts to form the intermediate products – hydrogen (H2),
carbon monoxide (CO), and carbon dioxide (CO2). In side reactions, H2 and CO react with
oxygen (O2) to form water (H2O) and CO2.
A fictitious “fuel molecule” was assumed for the reaction, based on the mass fractions of C,
H, N, and O contained in the fuel. Also, the composition of gaseous reaction products must
be known; this, however, depends mainly on the reaction temperature and the availability
of oxygen. A reasonable approach to obtain input data on gas composition during the
different stages of combustion (smoldering, glowing fire, flame) for modeling purposes
is to apply the method of Fourier Transform Infrared Spectroscopy (FTIR) at different
reaction temperatures during fuel conversion. To match the effect of “inerting,” N was
considered too. N does not react with O2.
Moisture transport affects the self-ignition and combustion behavior of coal. For this
reason, the coupled effects of evaporation, condensation, desorption and adsorption were
also considered in the model.
224
Temperature and concentration fields were solved in equations (6) and (7):
(6) ,
(7) .
Each species was converted at its specific conversion rate during each reaction. Solid fuel,
for instance, exhibits an Arrhenius-type reaction rate when decomposed; it was converted
in equation (8):
(8)
where k0 is the pre-exponential factor, E the apparent activation energy specific for each
reaction, and R the universal gas constant. The reaction rate is of first order since it was
assumed that the decomposition reaction depends on the concentration of the fuel only
(besides temperature).
For the rates of the species produced in this reaction, the following equation applies:
(9)
where υ is the stoichiometric coefficient and M the molecular weight. Note that the
stoichiometric coefficients are always negative for educts and positive for products.
The source term for liquid water and vapor can be calculated using the empirical equation (10):
(10) .
The first part represents the evaporation/desorption of liquid water, the second part the
condensation/adsorption of water vapor. In equation (10), Ew is the evaporation enthalpy;
kevap. and kcond. (s-1) are constants and have to be determined experimentally.
The source term for the temperature field equation comprises the sum of all reaction
enthalpies (equation (11)):
(11)
225
The model eventually consists of 11 coupled time-dependent PDEs solving the temperature
field and 10 concentration fields (coal, C, O2, CO2, CO, H2, N2, H2O [reaction], H2O [bound,
liquid], and H2O [vapor]).
For solving the system of equations numerically, the commercial Finite-Element-Code

FEMLAB was used.
An application of this numerical model (2D) is presented in Figure 15.
Figure 15: Geometrical simulation of 3,200 cm3 mesh wire cylinder filled with lignite coal
For calculation, the surrounding temperature was assumed to be at a constant 104 °C.

The initial temperature of the coal was set at 20 °C. The temperature curve and
concentration curves of water and water vapor inside the simulated coal deposit are
presented in Figure 16.
226
300 0 ,3
250 0 ,2 5
Temperature (°C) and water
tem perature
concentration (kg/m )
3
Water vapor concentration

200 0 ,2
150 temperature (without moisture) 0 ,1 5
(kg/m )
3
100 0 ,1
w ater
w ater vapor
50 0 ,0 5
0 0
0 200 400 600 800 1000
Time (h)
Figure 16: Simulated temperature development and concentration development of water
and water vapor in center of lignite coal cylinder (3,200 cm3) at constant
storage temperature of 104 °C
Under the impact of the ambient temperature (104 ºC), the temperature inside the coal
rises until self-ignition. Starting from an initial value of 20 % (or 114.8 kg m-3, based
on the bulk density of lignite coal), the water concentration inside the coal increases
shortly as water vapor from the surrounding air diffuses into the coal and condenses,
adding to its contents. At the same time, water continuously evaporates and decreases
in the direction of heat transfer, i.e. from the very surface of the sample toward
its center. After the initial rise in temperature, evaporation displaces condensation
as the dominating process in the center of the coal. From this point onward, water
concentration inside the coal decreases effectively. The same holds for water vapor –
after an early maximum, following the evaporation of the surface-near water which
allows vapor to diffuse into the coal and partly condense, the water in the center starts
evaporating and diffuses in the opposite direction, leading to a drop of total vapor
concentration inside. At the time of ignition, temperature and vapor concentration
in the center shoot up briefly, while water concentration decreases down to zero as a
result of complete evaporation.
The calculation was replicated without consideration of moisture content and transport
(dotted line in Figure 16). Here, the time to ignition was reduced since no water
evaporation occured.
227
The concentration profiles for coal, C, and CO in the sample for the above calculations are
presented in Figure 17.
500 0,25
coal CO
Concentration of coal and C
400 0,2
Concentration of CO (kg/m )
3
300 0,15
(kg/m )
3
200 C 0,1
100 0,05
0 0
0 200 400 600 800 1000
Time (h)
Figure 17: Simulated concentration development of coal, C, and CO in the center of

lignite coal cylinder (3,200 cm3) at constant storage temperature of 104 °C
At the time of ignition, the concentrations of CO and C shoot up, while coal concentration
drops dramatically. After a maximum, CO and C also decrease as they react with O2 to
form CO2. These reactions continue until the combustion front has moved through the
sample and all coal and its intermediates have been fully consumed. The temperature in the
cylinder then approximates storage temperature.
5 Conclusion
The experiments presented in this paper clearly demonstrate that environmental parameters
affect the self-ignition and combustion behavior of coal. Investigations were carried out on
Chinese black coal (several samples) and German lignite coal (one sample). The SITs of
all samples decreased with increasing V/A ratio – an observation which can be attributed to
the relatively smaller specific surface over which heat can be released at higher V/A ratios.
The reduction of oxygen content in the immediate vicinity of samples led to higher SITs
compared to normal ambient air. Self-ignition occurred at oxygen levels as low as 6 %.
Pouring liquid water (“wetting,” e.g. rain) onto coal generates heat in adsorption processes.
It was shown that a subcritical deposit of coal can become supercritical if water molecules
are adsorbed over its surface. “Wetting” here lowers the critical surrounding temperature.
228
Besides liquid water, water vapor also influences the self-ignition behavior of coal.
Condensation followed by adsorption of water molecules into the coal made temperatures
in samples rise. It was possible to change the status from subcritical to supercritical by
exposing coal samples to an elevated level of humidity. Convection outside coal slightly
increases SIT. It was shown that the self-ignition process is kinetically controlled and not
dependent on oxygen supply. Combustion velocity strongly depends on the number and
size of void volumes in coal. Combustion was much more pronounced (faster and at higher
reaction temperatures) for coal samples with large void volumes. This is due to lowered
flow resistance and enhanced oxygen supply.
A numerical model incorporating processes such as heating, ignition, propagation, and

moisture transport was established to simulate coal combustion. Simulation results were in
good accordance with lab-scale experiments. They allow a reliable and fast assessment of
novel extinguishing methods.
Appendix
Table 1: Origin and characteristics of coal samples used in experiments
Origin Coal Type Region Conditiona dParticles Name
China Black coal Wuda (FZ 3-2) Weathered < 10 mm WU1
Wuda (FZ 3-2) Unaffected < 10 mm WU2
Wuda (FZ 11) Unaffected < 10 mm WU3
Wuda (FZ 8) Unaffected < 10 mm WU4
Rujigou Weathered < 10 mm RU1
Rujigou Unaffected < 10 mm RU2
Shizuishan Unaffected < 10 mm SH
Germany Lignite coal Lausitz Unaffected < 20 mm LC
a
Unaffected = coal from inside the seam; weathered = coal from outcrop surface
229
Table 2: Material properties of coal samples used in experiments
Name Thermal Density Density Porosity Calorific Specific Specific C/H/N H2O Ash
Conducti- ρbulkb ρsolidc εbulkd Valuee Surface ASf Pore Contenth Contenti Content
(J kg )
-1
vity (kg m-3) (kg m-3) (%) (m2 g-1) Volume (% / % / %) (%) (waf)j
λbulka V Sg (%)
-1
(W m K )
-1
(mm3 g-1)
WU1 0.1218 907 1520 40.3 2.1897E+07 1 2 59.3/3.9/1.1 14.30 6.33
WU2 0.1228 905 1531 40.9 2.2982E+07 1.1 1.9 62.7/3.9/0.9 14.08 4.85
WU3 0.1385 991 1637 40.8 1.8965E+07 2.8 5.9 51.9/3.7/0.8 13.50 21.71
WU4 0.1250 959 1487 35.5 2.4212E+07 0.8 1.8 63.5/4/1.2 10.47 6.18
RU1 0.1619 992 1696 41.5 2.2820E+07 3.4 / 3.7 4.3 / 4.3 63.3/2.8/0.5 9.16 19.01
230
RU2 0.1346 927 1615 42.6 2.3415E+07 11.7 / 11.6 3.1 / 3.0 66.3/3.3/0.5 13.55 7.61
SH … 910 1587 42.7 2.4885E+07 2 4.1 68/3.1/0.9 10.06 5

52.9/0.1/
LC 0.1100 574 ~1200 52.2 2.2035E+07 … … 20 5.89
a
Measured at 50 °C
b
ρbulk = massbulk / volumebulk
c
DIN 66137-2
d
εbulk = 100 [1-(ρbulk / ρsolid)]
e
DIN 51900-3
f
BET
g
Cranston-Inkley
h
“Coal method”
i
Karl-Fischer (coulometric)
j
DIN 51719
References
Bähr, H.D. and Stephan, K., 1998. Wärme- und Stoffübertragung. 3rd edition. Berlin, Heidelberg:
Springer-Verlag.
Chong, L.V. and Chen, X.D., 1999. A mathematical model of the self-heating of spray-dried food powders
containing fat, protein, sugar, and moisture. Chemical Engineering Science 54: 4165-4178.
Fierro, V., Miranda, J.L., Romero, C., Andrés, J.M., Arriaga, A., Schmal, D., and Visser, G.H., 1999.
Prevention of spontaneous combustion in coal stockpiles – Experimental results in coal
storage yard. Fuel Processing Technology 59: 23-34.
Fierro, V., Miranda, J.L., Romero, C., Andrés, J.M., Arriaga, A., and Schmal, D., 2001. Model predictions
and experimental results on self-heating prevention of stockpiled coals. Fuel 80: 125-134.
Frank-Kamenetzkii, D.A., 1959. Stoff- und Wärmeübertragung in der chemischen Kinetik. Berlin,
Göttingen, Heidelberg: Springer Verlag.
Gray, B.F., Sexton, M.J., Halliburton, B., and Macaskill, C., 2002. Wetting-induced ignition in cellulosic
materials. Fire Safety Journal 37: 465-479.
Hensel, W., Krause, U., and Löffler, U., 2004. Self-ignition of solid materials (including dusts). In
Hattwig, M. and Steen, H. (Eds.), Handbook of explosion prevention and protection.
Weinheim: Wiley-vch Verlag GmbH, 227-256.
Krause, U., Schmidt, M., and Lohrer, C., 2006. A numerical model to simulate fires in bulk materials and
dust deposits. Journal of Loss Prevention in the Process Industries 19: 218-226.
Küçük, A., Kadioğlu, Y., and Gülaboğu, M.Ş., 2003. A study of spontaneous combustion characteristics of
a Turkish lignite: Particle size, moisture of coal, humidity of air. Combustion and Flame 133:
255-261.
Lohrer, C., 2006. Einflussgrößen auf die Selbstentzündung von Schüttgütern und Stäuben - Experimentelle
Untersuchungen und numerische Simulationen. PhD thesis, Technical University of Berlin,
Berlin, Germany.
Lohrer, C., Schmidt, M., and Krause, U., 2005. A study on the influence of liquid water and water vapor
on the self-ignition of lignite coal - Experiments and numerical simulations. Journal of Loss
Prevention in the Process Industries 18: 167-177.
McIntosh, A.C., Gray, B.F., and Wake, G.C., 1994. The ignition of combustible material in the presence of
a damp atmosphere. Physics Letters A 191: 61-70.
Ren, T.X., Edwards, J.S., and Clarke, D., 1999. Adiabatic oxidation study on the propensity of pulverized
coals to spontaneous combustion. Fuel 78: 1611-1620.
Schmidt, M., 2001. Untersuchungen zu Schwelbränden in Feststoffschüttungen verursacht durch

Glimmnester und eingeschüttete heiße Gegenstände. PhD thesis, University of Magdeburg,
Magdeburg, Germany.
Schmidt, M., Lohrer, C., and Krause, U., 2003. Self-ignition of dusts at reduced volume fractions of
ambient oxygen. Journal of Loss Prevention in the Process Industries 16: 141-147.
Semjonov, N.N., 1928. Zur Theorie des Verbrennungsprozesses. Zeitschrift für Physik 48: 571-582.
Stott, J.B., 1960. Influence of moisture on the spontaneous heating of coal. Nature 188: 54-55.
231
VDI, 1990. Untersuchungsmethoden zur Ermittlung von sicherheitstechnischen Kenngrößen von Stäuben.
Guideline 2263. Verein Deutscher Ingenieure (VDI), Düsseldorf, Germany.
Wang, H., Dlugogorski, B.Z., and Kennedy, E.M., 1999. Theoretical analysis of reaction regimes in low-
temperature oxidation of coal. Fuel 78: 1073-1081.
Wiemann, W. and Scholl, W.W., 1985. Selbstentzündung von Braunkohlenstaub bei vermindertem
Sauerstoffgehalt. Staub – Reinhaltung der Luft 45: 147-150.
232
Researching Coal Spontaneous
Combustion: Micropetrography of
Coal Oxidation and Carbonization
Jolanta Kus1, Werner Hiltmann2, and Aenne Balke1
1
Federal Institute for Geosciences and Natural Resources (BGR), Hanover, Germany
2
Schneppenhorststraße 14, Hanover, Germany
Abstract
As part of Phase A of the Sino-German Coal Fire Research Initiative “Innovative Technologies for
Exploration, Extinction, and Monitoring of Coal Fires in North China,” micropetrographical investigations
of coal samples were performed to assess the prerequisites and effects of coal spontaneous combustion in
coal seams.
Numerous coal samples collected from coalfields in Inner Mongolia (Wuda, Gulaben) and Ningxia
Autonomous Regions (Rujigou, and Shizuishan) in north China were affected by coal fires. Thermally
altered coal particles were found in high volatile bituminous coals from Wuda and in high-rank coals
from Gulaben and Rujigou. Heating through coal fires had led to marked changes in coal optical
properties, reflected by distinct anisotropy effects and elevated coalification degrees. Optical anisotropy
patterns were particularly well developed in contact zones of preserved thermally altered vitrinite
particles and coke-like components. The degree of coalification varied between the different coal fire
zones of Wuda; reaching in some parts a maximum vitrinite reflectance Rmax of 1.60 %. Heating through
coal fires had also entailed changes in coal microscopic structure and organic composition, ranging
from shrinkage cracks associated with coal oxidation (weathering) and drying due to heat flux, to
rounded maceral outlines.
Although the pyrite content of the Wuda coals was extremely low, the existence of preserved roasted and/
or oxidized products of pyrite in the coals suggests much higher primary levels in the past and points to a
possible contribution to spontaneous combustion. Coal rank or volatile matter content correlates with self-
ignition temperature and thus exerts a major influence on spontaneous combustion of coal.
233
摘要
为评价煤层自燃的环境条件和影响，作为中德煤火研究项目“中国北方煤火探测、灭火和监测新
技术”第一阶段研究的组成部分，本文进行了煤样的岩相学研究，用以评估煤层中煤自燃的必要
条件及其影响。
在中国北方的内蒙古乌达矿区和古拉本矿区、宁夏的汝箕沟矿区和石嘴山矿区，我们采集了大量
受煤火影响的煤样。在乌达矿区高挥发分煤样与古拉本和汝箕沟矿区优质无烟煤中发现了热变质
微粒，煤火加热后导致煤光学特性发生明显变化，反映出显著各向异性效应和煤化程度提高。在
残留的热变质镜质煤岩微粒与似焦状成分接触带上特别研究了光学各向异性模式，发现乌达矿区
不同火区的煤化程度是变化的，一些样品最大镜质煤岩反射率（Rmax）达到了1.60%。煤火热作用
影响到煤微观结构和有机组分，变化范围从与煤氧化作用（风化作用）和热流干燥有关的煤体收
缩裂隙到圆形的煤组分轮廓。
尽管乌达研究区煤样的黄铁矿含量非常低，但煤中残留的黄铁矿焙烧物和氧化物显示出过去黄铁
矿含量较高并对煤自燃具有重要的作用。煤的种类或挥发物含量与自燃温度是相关的，它们在煤
自燃过程中具有重要的影响作用。
1 Introduction
Thermal alterations of coal due to natural combustion (as opposed to industrial

combustion) have been investigated extensively (Clegg 1955; Berkowitz 1967; Jones &
Creaney 1977; Bostick 1979; Goodarzi & Jerzykiewicz 1986; Goodarzi 1987; Goodarzi
et al. 1988b; Raymond & Murchison 1989). Such combustion occurs in coal outcrops,
in underground coal mines, as well as in coal waste heaps and other coal storage and
treatment facilities; it can also affect coal that is being transported.
A number of triggers of natural coal combustion have been identified so far –

anthropogenic causes, emplacement of intrusive bodies, forest fires, and lightning followed
by spontaneous combustion.
Spontaneous combustion is a process in which coal starts to burn, i.e. self-ignites (and
subsequently develops into a full-blown coal fire) in the absence of any exterior ignition
source. Self-ignition is the climax of a series of exothermic chemical reactions which lead
to a substantial temperature rise in the combustible material.
The impact of coal fires caused by spontaneous combustion is calamitous – it ranges from
irrecoverable loss of non-renewable energy resources and rising mining costs to land
subsidence and climate change.
234
Within the framework of Phase A of the Sino-German Coal Fire Research Initiative
“Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires in
North China,” spontaneous coal fires were investigated in Inner Mongolia and Ningxia
Autonomous Regions, P.R. China (Figure 1).
Figure 1: Study area of the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in
North China”
Among the numerous scientific approaches that were employed as part of the initiative,
coal petrography was to investigate possible processes leading to coal spontaneous
combustion and their parameters, among them:
1) Pyrite content. Pyrite (FeS2) is a mineral which swells when heated and may accelerate
spontaneous combustion as it leads to disintegration and reduction of coal particle size
(Lyman & Volkmer 2001).
2) Coal rank. Coal rank is the degree of coalification of coal organic matter. It classifies
coal according to moisture content, calorific value, vitrinite reflectance, and volatile
235
matter content. The tendency to self-ignition increases with decreasing coal rank; i.e.
lignite is more susceptible to self-heating than bituminous coal.
Industrial coal combustion involves a number of distinct processes which profoundly alter the
optical properties and microscopic structure of coal, viz. oxidation and carbonization (as in
the formation of industrial coke). It is widely accepted today that coal oxidation is a primary
source of heat, which may even lead to spontaneous combustion. Reactions involved in low-
temperature oxidation are still not well understood; a broad outline was given by Schmal
(1987). Coal oxidation takes place when coal is exposed to air. The resulting changes in the
organic and inorganic composition and physical structure of coal are of great importance
to industrial combustion (Nandi et al. 1977; Mahajan et al. 1980; Lee et al. 1983). These
changes are identifiable under the microscope in incident light as “oxidation rims” along coal
grain boundaries, or fissures and fractures. Shrinkage cracks, rounded edges, oxidized pyrite,
swollen clay minerals, and secondary mineral coatings on edges also point to oxidation
processes (Benedict et al. 1968; Gray 1982; Teichmüller et al. 1998a,c).
Apart from coal oxidation, industrial coal combustion also involves carbonization, where
coal is transformed into coke with specific anisotropic textures.
Coal petrographical investigations of industrial coal combustion provide a suitable platform

for the assessment of natural coal combustion (coal fires) and its products, e.g. natural coke.
For the present study, numerous coal samples collected from various coal fire zones in the
study area of the Sino-German Coal Fire Research Initiative (Figure 1) were subjected to
coal petrographical and fuel chemistry investigations. This paper presents some of the coal
fire-induced petrographical changes in coal samples from Wuda, Rujigou, Gulaben, and
Shizuishan Coalfields. It also discusses the susceptibility of these coals to spontaneous
combustion with respect to coal rank and pyrite content.
2 Materials
During two summer campaigns in 2004 and 2005, fresh coal channel samples were
collected along various coal seams in coal fire zones of Wuda, Rujigou, Gulaben und
Shizuishan Coalfields. Samples were sealed in polyethylene bags after excavation. The
previous time-temperature history of these samples remains unknown. In Wuda, the
samples were collected from coal seams nos. 9 and 10 in fire zones (FZs) 3-2 and 8 in the
western-central and northern part of the Wuda syncline (Figure 2).
236
Figure 2: Sampling locations (red) in Wuda FZs 3-2 (left) and 8 (right). Gas sampling
locations are marked as green and yellow dots (left). Orange polygons indi-
cate hot areas with temperatures above 150 ºC as mapped by the German
Aerospace Center (DLR) in the summer campaign of 2005.
3 Methods
After transfer to a BGR laboratory, all coal samples were subjected to a standardized
German sample preparation procedure (DIN 51701-3). For coal petrographical
investigations, polished coal grain mounts were prepared in accordance with German
standard methods for microscopic analysis in incident light (DIN 22020). These mounts
were analyzed using a Leica DMRX microscope photometer (and other devices) to
determine coal rank (measurement of mean vitrinite reflectance; DIN 22020, part 5).
Results with respect to coal material and mineral matter were recorded in detail. Moisture,
ash, volatile matter, and total sulfur content were determined in solid fuel chemistry
investigations according to DIN 51718, 51719, 51724-1, and 51720.
237
4.1 Pyrite
The investigation of coal pyrite content based on maceral group analysis in samples from
Wuda coal seam no. 9 yielded extremely low values (consistently below 0.1 vol. %),
with one exception – 1.1 vol. % at W3.2-17. Pyrite minerals in the samples occurred
as weathered isolated or aggregated framboids or infills of micro-fissures and tectonic
fractures (Figures 3 and 4).
Figure 3: Coal samples W2-6-1 (left) and W2-6-2 (right): Oxidized or rusted framboidal
pyrite in incident light
Figure 4: Coal sample W2-8-1 (left): Oxidized framboidal pyrite in incident light.
Coal sample W2-8-2 (right): Massive pyrite as infills of micro-fissures.
These findings agree well with results from investigations by Yang et al. (1996), who
determined the pyrite content of coal from Wuda coal seam no. 9 at 0.2 vol. %. They
disagree, however, with results of Dai et al. (2000), i.e. a pyrite content ranging between
238
0.2 and 7.3 vol. %, with a mean of 2.5 %. Petrographical investigations in Rujigou,
Gulaben, and Shizuishan also yielded very low values of 1.0 vol. % and below.
One possible explanation for these observations is that pyrite that was originally contained in
the coals had oxidized to form iron hydroxide as a result of exposure to oxygen and water:
(1) 4FeS2(s) + 14H2O(l) → 4Fe2+(aq) + 8SO42-(aq) + 8H+
The ferrous ions oxidized to form ferric ions:
(2) 4Fe2+(aq) + O2(g) + 4H+(aq) → 4Fe3+(aq) + 2H2O(l)
The ferric ions hydrolyzed to form insoluble yellow-orange precipitate:
(3) 4Fe3+(aq) + 12H2O(l) → 4Fe(OH)3(s) + 12H+(aq)
This yellow-orange precipitate, the so-called “yellow boy,” was observed on the surface of
coal seam no. 10 in FZ 3-2 at several places, e.g. at W3.2-16, -17, -19, and -21, but also at
coal seam no. 9, e.g. at W8-33 and -35 (Figure 5). It is therefore conceivable that extensive
oxidation is the reason for the presently very low pyrite content in the coal samples
collected from a depth of ca. 30 to 50 cm in the seam.
Figure 5: Coal seams nos. 10 (FZ 3.2; left) and 9 (FZ 8; right) of Wuda Coalfield:
Rusty coatings and bands indicate pyrite or siderite oxidation products, e.g.
iron oxides, possibly hematite
239
Another possible explanation for the very low pyrite content in most samples is
thermal degradation. In open-system experiments at temperatures of up to 470 ºC
(Golden 2004), thermal degradation of pyrite produced pyrrhotite minerals, which
subsequently oxidized to iron oxides, such as hematite and magnetite. The thermally
affected coal samples of this study have not yet been analyzed for pyrrhotite content.
However, based on results from coalification studies of coal samples W2-6 and -8, an
extensive long-term thermal alteration leading to increased coalification and thus to
thermal degradation can be ruled out. Petrographical analyses support this assumption
as they revealed a lack of natural coke particles and devolatilization features in the
samples. Instead, extensive fractures, i.e. shrinkage fractures crosscutting nearly all
macerals, were found; they rather point toward coal oxidation and shrinkage due to
drying, i.e. heat flux.
Oxidation or transformation of coal pyrite represents one of the potential factors

contributing to coal spontaneous combustion. It leads to fracture development through
volume expansion, and generates substantial amounts of heat, which may support a general
temperature rise in coal. It is to be assumed that a fine dispersed pyrite content in order
of or greater than 5 to 10 vol. % will lead to noticeable effects of pyrite oxidation on the
self-ignition behavior of coal. Given the figures by Dai et al. (2000), pyrite content might
indeed have been a significant contributor to coal spontaneous combustion in at least some
localities.
4.2 Coal rank
Thermally unaltered coals from Wuda Coalfield can be classified as high volatile
bituminous A coals, whereas coals from Gulaben correspond to a stage of anthracite
coal. The rank of coals from both Rujigou and Shizuishan ranges between low volatile
bituminous and anthracite coals. The correlation of coal rank with volatile matter and
maximum vitrinite reflectance according to ASTM classification allows for a direct
comparison of these coal properties with self-ignition temperature (SIT). The latter was
successfully determined for different coals from the test sites by the Federal Institute for
Materials Research and Testing (BAM) (Schmidt et al. 2005; see also Lohrer et al. 2008
in this volume).
Plotting the SITs of the selected coal samples (particle size < 10 mm; different specific
volumes) against coal rank or volatile matter content clearly illustrates the relationship
between these two parameters (Figure 6).
240
Figure 6: Cross-correlation of SIT and volatile matter content, averaged for coal samples
from Shizuishan, Rujigou, and Wuda FZs 8, 11, and 3-2
The SIT of a given volume of coal appears to be dependent on its volatile matter content,
and therefore on coal rank, i.e. the lower the SIT, the higher the volatile matter content and
the lower the coalification degree. This tendency is greatest in the samples of 800 ml, but
lowest in the samples of 31 ml – a phenomenon related to the specific surface area of coal.
The lowest high volatile bituminous A coal from Wuda FZ 3-2 shows the lowest SIT
of 167 °C, whereas the semi-anthracite coal from Shizuishan ignites at a much higher
temperature of 184 °C.
The interrelatedness of coal rank and susceptibility to self-heating is a widely reported

phenomenon (Kim 1977; Walters 1996). It is generally accepted that low-rank coals are
more susceptible to self-heating than high-rank coals (Karcz 1980; Kuchta et al. 1980;
Chen et al. 1984). Krause (2005) offers a good literature review on self-ignition of porous
combustible materials.
4.3 Optical properties
Optical anisotropy is a distinct property of coal. It derives from a progressive orientation

of aromatic nuclei in the bedding plane, caused by pressure load and/or carbonization.
Optical anisotropy can even be observed in sub-bituminous coals, but becomes noticeable
241
only in medium volatile bituminous coals, i.e. at a random vitrinite reflectance R r of c. 1,3 %
(Teichmüller et al. 1998a).
Optical anisotropy was observed in both thermally unaffected and affected coal samples.
Within the first group of samples, distinct anisotropy effects were detected in vitrinite
particles of high-rank coals from Rujigou, Gulaben, and Shizuishan. Here, optical
anisotropy was a result of static load pressure caused by subsidence and tectonically
induced stress during Helan Shan orogeny. Thermally altered vitrinite particles in coal
from Wuda Coalfield displayed similar optical textures; these however derived from
heating (Figure 7; left).
Consistent with these findings, vitrinite reflectance measurements on the thermally

affected coal sample W8-36 from Wuda FZ 8 yielded a mean maximum reflectance
R max of 1.06 %, a mean minimum ref lectance R min of 0.98 %, and mean difference
ref lectance R dif of 0.08 %. Wuda semi-coke and coke particles displayed textures
ranging from isotropic to coarse-grained mosaics with a domain size diameter of
up to 5.0 μm (Figure 7; right). Thermal alteration of low volatile bituminous and
anthracite coals from Gulaben had resulted in fine-grained mosaics and lenticular
textures (Figure 8).
Figure 7: Coal sample W8-47b_23 (left): Thermally affected and strongly fragmented
vitrinite particle with isotropic texture and silicates filling pore spaces. Coal
sample W8-47b_25 (right): Thermally affected vitrinite particle with isotropic
texture, laminations of fine-grained pyrolitic carbon (PC), devolatilization
vacuoles (V), and undifferentiated carbonized material in incident light.
242
Figure 8: Coal sample S-83_1 (left): Thermally affected coal particle with fine-grained
mosaic texture (F) among residual material (R) in incident plane polarized
light. Coal sample S-83_2 (right): Thermally altered vitrinite particle changed
to natural coke characterized by fine-grained lenticular anisotropic texture (F)
(Rr 5.09 %) in incident plane polarized light.
Experimental investigations of coal carbonization exhibited complex changes in the

physical structure of coal, e.g. in the form of coal components containing optically-
anisotropic units. Their appearance, form, and size depend on the rank of the carbonized
coal (Brooks et al. 1968). Several schemes have been developed in order to classify the
textural composition of carbonized coal (Grint and Marsh 1981; Moreland et al. 1988;
Patrick, et al. 1988). Optical anisotropy is thought to be caused by a loss of volatile matter,
variations in viscosity of plastic mass, and the distortion of ordered phases by pressure
from evolving gases (Patrick et al. 1979).
4.4 Coalification degree
The average background random vitrinite reflectance Rr of the coals from Wuda FZs 3.2
and 8 was 0.95 %. However, some thermally affected samples collected from several coal
fire subzones of FZ 8 in 2005 had an elevated coalification degree amounting to a random
vitrinite reflectance Rr of 1.1, 1.2, and 1.8 % (Figure 9). The coal fire sub-zones correspond
macroscopically to zones of heavily altered coal, reflected by volume loss, discoloration
and/or coke-like texture (Figure 10).
243
Figure 9: Coalification profile along coal seam no. 9 in Wuda FZ 8 as sampled in 2005 from
north (left) to south (right)
Figure 10: Coal seam no. 9 of Wuda Coalfield. One of the numerous subzones documented
for the northern part of FZ 8. The area displayed corresponds to coal sample
locations W8-11, -12, and -13. Mr. Men Yuefei supported the sample collection.
244
Similar results regarding coalification degree and optical anisotropic patterns were reported
by Goodarzi and Gentzis (1990) from Britisch Columbia, Canada. In thermally altered
coals with a maximum vitrinite reflectance Rmax between 1.2 and 2.0 %, they found relicts
of former liptinite macerals, i.e., hydrogen-rich coal components of fine-grained mosaic
texture. Analogous examples of thermally altered liptinite are relatively limited due to their
generally low to very low content ≤ 1.0 vol. %.
Coal seam no.10 in Wuda FZ 3.2 did not show any marked increase in vitrinite reflectance.
Instead, anomalously low values of down to 0.85 % were recorded (Figure 11).
Vitrinite reflectance (% Rr)
1,00
0,95
0,90
0,85
0,80
W W W W W W W W W W W W W W W W W
3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2- 3.2-
22 21 20 19 18 17 64a 16 66a 67 65 23 27 28 24 25 26
Coal channel samples

Figure 11: Coalification profile along coal seam no. 10 in Wuda FZ 3.2 as sampled in
2005 from northeast (left) to southwest (right)
Possible explanations for these low values of random vitrinite reflectance may be type
variation within vitrinite macerals; nature and amount of other organic components,
especially liptinite; expulsion of aliphatic substances during early stages of geochemical
coalification; fluctuation in thermal conductivities of stratigraphical sequences (Murchison
1991); depositional facies (Correia da Silva & Wolf 1980); maceral subtype and mineral
matrix (Goodarzi et al. 1988a); as well as pH conditions (Mastalerz 1991).
Coal sample W3.2-64a of Wuda FZ 3-2 was collected directly below a thin (ca. 10 cm
deep) silty sandstone layer. Marked by a strong yellow hue, this layer seemed to contain
iron oxide, most likely an oxidation product of pyrite. The coal below, however, did not
have any anomalous features, apart from extensive slickensides, subparallel to parallel to
the bedding. The last coal sample W3.2-23 was collected in a fault gauge crosscutting the
coal seam that exhibited extensive jointing in a north-northwesterly to south-southeasterly
direction. It is therefore probable that the anomalous coalification values were either the
results of weathering (oxidation) or of tectonically induced stress.
245
5 Conclusion
Micropetrographical investigations on coal samples from coal fires zones in Wuda

Coalfield support the assumption that pyrite minerals act as catalysts in spontaneous
combustion. Moreover, as SITs correlated negatively with volatile matter content, and
thus with coal rank, it was confirmed that coal susceptibility to spontaneous combustion is
dependent on coal rank.
The use of reflected light microscopy allowed the identification of coal particles that had
undergone thermal alteration as a result of coal fire-induced heating. Low-temperature
oxidation (heat drying) of coal (and pyrite) due to continuous heat flux had most likely
given rise to extensive formation of iron oxides, some of which precipitated in coal fissures
and fractures at the surface or in the coal seam. Shrinkage cracks and coal defragmentation
were a result of this process too.
The analysis of thermally affected and oxidized coal revealed the formation of coke and
semi-coke particles as well as an expected increase in coalification degree, especially in
those coal fire sub-zones marked macroscopically by distinctive discoloration, volume
loss, coal seam degradation and distortion. Distinctive optical anisotropy patterns of coke
particles were recorded.
References
Benedict, L.G., Thompson, R.R., Shigo-III, J.J., and Aikmann, R.P., 1968. Pseudovitrinite in Appalachian
coals. Fuel 47: 493-501.
Berkowitz, N., 1967. The coal-carbon transformation: Basic mechanism. In Proceedings of the Symposium
on the Science and Technology of Coal, Ottawa, Canada, March 1967. Department of Energy
Mines and Resources, Ottawa, Canada, 149-155.
Bostick, N.H., 1979. Microscopic measurement of the level of catagenesis of solid organic matter
in sedimentary rocks to aid exploration for petroleum and to determine former burial
temperatures – A review. In Scholle, P.A. and Schluger, P.A. (Eds.), Aspects of digenesis.
Society of Economic Palaeontology and Mineralogy, Special Publication 26. Tulsa, Oklahoma,
17-43.
Chen, M.R., Fan, L.S., and Essenhigh, R.H., 1984. Prediction and measurement of ignition temperature of
coal particles. In Proceedings of the 20th International Symposium on Combustion, Ann Arbor,
Michigan, U.S.A., 12-17 August 1984, 1513-1519.
Clegg, K.E., 1995. Metamorphism of coal by periodotite dikes in southern Illinois. Report of Investigation
178. Illinois State Geological Survey, Champaign, Illinois, U.S.A.
246
Correia da Silva, Z.C. and Wolf, M., 1980. O poder reflector como parâmetro pará determinação do grao
de Carbonificação dos carvões Gondwanicos do Brasil. Pesquisas, Instituto de Geosciências
UFRGS 13: 35-42.
Dai, S.F., Ren, D.Y., Ai, T.J., Zhou, Q., Mao, H.L., Ma, F.X., and Li, B.C., 2000. Influence of geological
factors on evaluating washability of coal in Wuda Coalfield, Inner Mongolia. Journal of China
University of Mining and Technology 29(3): 249-252.
Golden, D.C., Ming, D.W., Lauer, H.V. Jr., and Morris, R.V., 2004. Thermal decomposition of siderite-
pyrite assemblages: Implications for sulfide mineralogy in Martian Meteorite ALH84001
Carbonate Globules. In Proceedings of the Lunar and Planetary Science Conference, Houston,
Texas, U.S.A., 15-19 March 2004, 1396.
Goodarzi, F., 1987. Reflectance and petrology of a burning bituminous coal seam. Fuel 66: 1073-1078.
Goodarzi, F. and Gentzis, T., 1990. The lateral and vertical reflectance and petrological variation of a heat-
affected bituminous coal seam from southeastern British Columbia, Canada. International
Goodarzi, F. and Jerzykiewicz, T., 1986. Petrology of a burning bituminous coal seam at Coalspur,
Alberta. In Current Research, Part B. Geological Survey of Canada Papers 86-1B, 421-427.
Goodarzi, F., Feinstein, T., and Snowdon, L., 1988a. Effect of maceral subtypes and mineral matrix on
measured reflectance of subbituminous coal and dispersed organic matter. International
Goodarzi, F., Gentzis, T., and Bustin, R.M., 1988b. Reflectance and petrology profile of a partially
combusted and cooked bituminous coal seam from British Columbia. Fuel 67: 1218-1222.
Gray, R.J., 1982. A petrologic method of analysis of non-maceral microstructures in coal. International
Journal of Coal Petrology 2: 79-97.
Grint, A. and Marsh, H., 1981. Carbonization and liquid-crystal (mesophase) development 20: Co-
carbonization of a high-volatile caking coal with several petroleum pitches. Fuel 60: 513-518.
Jones, J.M., and Creaney, S., 1977. Optical character of thermally metamorphosed coals of Northern
England. Journal of Microscopy 109(1): 105-118.
Karcz, H., Kordylewski, W., Rybak, W., 1980. Evaluation of kinetic parameters of. coal ignition. Fuel 59:
799–802.
Kim, A.G., 1977. Laboratory studies on spontaneous heating of coal. A summary of information in the
literature. Information Circular 8756. U.S. Bureau of Mines (USBM), Washington, D.C.,
U.S.A.
Kuchta, J.M., Rowe, V.R., and Burgess, D.S., 1980. Spontaneous combustion susceptibility of U.S. coals.
Report of Investigations 8474. U.S. Bureau of Mines (USBM), Washington, D.C., U.S.A.
Lee, G.K. and Whaley, H., 1983. Modification of combustion and fly-ash characteristics by coal blending.
Journal of the Institute of Energy 56: 190-197.
Lohrer, C., Schmidt, M., and Krause, U., 2008. Influence of environmental parameters on the self-ignition
behaviour of coal. In UNESCO Office Beijing (Ed.), Spontaneous coal seam fires: Mitigating
a global disaster. International Research for sustainable control and management. ERSEC
Ecological Book Series – 4. Beijing: Tsinghua University Press and Springer Verlag, 210-232.
(this volume)
247
Lyman R.M. and Volkmer, J.E., 2001. Pyrophoricity (spontaneous combustion) of Powder River Basin
coals – Considerations for coalbed methane development. Coal Report CR 01-1. Wyoming
State Geological Survey, Laramie, Wyoming, U.S.A.
Mahajan, O.P., Komatsu, M., and Walker, P.L. Jr., 1980. Low-temperature air oxidation of caking coals 1:
Effect on subsequent reactivity of chars produced. Fuel 59: 3-10.
Mastalerz, M., 1991. Variation of vitrinite reflectance in a vertical seam section, an example from the
Intrasudetic Basin, SW Poland. Bull. Soc. Geol. France 162(2): 39-45.
Moreland, A., Patrick, J.W., and Walker, A., 1988. Optical anisotropy in cokes from high-rank coals. Fuel
67: 730-732.
Murchison, D.G., Pearson, J., and Raymond, A., 1991. Anomalies in vitrinite reflectance gradients. Bull.
Soc. Geol., France 162(2): 183-191.
Nandi, B.N., Brown, T.D., and Lee, G.K., 1977. Inert coal macerals in combustion. Fuel 56: 125-130.
Patrick, J.W., Green, P.D, Thomas, K.M., and Walker, A., 1988. The influence of pressure on the
development of optical anisotropy during carbonization of coal. Fuel 68: 149-154.
Patrick, J.W., Malcom, J.R., and Shaw, F.H., 1979. Optical anisotropy of carbonized coking and caking-
coal vitrains. Fuel 58: 501-509.
Raymond, A.C. and Murchison, D.G., 1989. Organic maturation and its timing in a Carboniferous
sequence in the central Midland Valley of Scotland: Comparisons with northern England. Fuel
68: 328-335.
Schmal, D., 1987. A model for the spontaneous heating of stored coal. PhD thesis, Delft University of
Schmidt, M., Lohrer, C., and Krause, U., 2005. Results of laboratory investigations: Self-ignition and
burning behavior of coal samples. Project report for Work Package 2400: Spontaneous
ignition / inertization, as a contribution to the Sino-German Coal Fire Research Initiative
“Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North
China.” Federal Institute for Materials Research and Testing (BAM), Berlin, Germany.
Teichmüller, M., Taylor, G.H., Littke, R., and Robert, P., 1998a. Coalification and maturation process,
petrographical changes. In Taylor et al. (Eds.), Organic petrology. Berlin, Stuttgart: Gebrüder
Bornträger, 116-125.
Teichmüller, M., Littke, R., and Robert, P., 1998b. Thermal alteration and the formation of natural coke.
In Taylor et al. (Eds.), Organic petrology. Berlin, Stuttgart: Gebrüder Bornträger, 162-174.
Teichmüller, M., Taylor, G.H., Littke, R., and Robert, P., 1998c. Oxidized organic matter in coal. In Taylor
et al. (Eds.), Organic petrology. Berlin, Stuttgart: Gebrüder Bornträger, 199-201.
Walters, A.D., 1996. Joseph Conrad and the spontaneous combustion of coal. Part I. Coal Prep. 17:
147-165.
Yang, Y.K. (Ed.), 1996. Atlas for coal petrography of China. Beijing: China University of Mining and
Technology Press.
248
Research on Gas Indicators
of Coal Spontaneous Combustion
in Wuda Coal Mining Area
Yu Minggao1, Jia Hailin1, Zuo Qiuling2, and Xu Jun1
1
Department of Safety Science and Engineering, School of Safety Science and Engineering,
Henan Polytechnic University, Jiaozuo, Henan Province, P.R. China
2
Department of Safety Engineering, Henan Engineering Institute, Zhengzhou,
Henan Province, P.R. China
Abstract
In order to predict coal spontaneous combustion in Wuda Coal Mining Area, six coal samples from Wuda
coal fire zones 6, 7, and 8 were brought to calefactive oxidation in a hot storage oven, in temperature
steps from 20 to 400 °C. The amount and composition of the gaseous products from this process were
analyzed by gas chromatography. Results were plotted against temperature. It was found that the volume
of gases from coal oxidation grows exponentially with rising coal temperature. Also, the important role of
CO, C3H8, as well as CnH2n+2 ratios in signaling the stage of coal spontaneous combustion was confirmed
(indicator gases).
摘要
为解决乌达矿区煤自燃的早期预测，采集乌达矿区6、7、8号火区6种煤样进行煤的绝热升温氧化
实验，煤样的升温氧化实验是在20~400℃的范围内进行的，氧化过程中产生的水蒸气被干燥剂吸
收，各气体组分的发生量由专人记录。最后实验数据经过软件处理后以图例的形式给出。分析图
例可知：气体产物的释放量与温度大致呈指数递增关系；且CO、C3H8和CnH2n+2比率均可作为煤
自燃进程的早期预报指标参数。
249
1 Introduction
Coal fires in Xinjiang, Ningxia, and Inner Mongolia Autonomous Regions not only impact
heavily on the environment, but also consume valuable resources. Wuda Coal Mining
Area, ranking third in size among China’s coalfields, has developed 26 fire zones extending
over an area of 3.076 km2. Apart from air pollution and land subsidence, coal fires bring
prodigious hidden troubles, especially for mine safety. Early detection and prevention are
extremely important. The earlier a fire is detected, the earlier measures can be taken to
control it and reduce coal loss. Recent studies on coal fires in China have made the gaseous
products of coal oxidation their focus; they probe the capability of combustion gases to
indicate fire status and development (Wang & Zhang 1990; Ren et al. 1999; Rosema et al.
2001; Fan et al. 2002; Wang et al. 2002; He et al. 2003; Luo & Qian 2003). In the present
study, experiments on coal samples from Wuda Coal Mining Area were carried out in a
hot storage oven, using two gas chromatographs for analysis. The samples were brought
to calefactive oxidation in temperature steps from 20 to 400 °C. The resulting gases were
examined for their usefulness in signaling the stage of coal spontaneous combustion.
2 Preparation of Samples
Six coal samples from different fire zones in Wuda Coal Mining Area were selected for
this study. The exact sampling dates and locations are shown in Table 1.
Table 1: Sampling dates and locations of samples used in this study

Sample No. Date Fire Zone (FZ) Coal Seam No.
1 9 October 2004 FZ 6 2
2 10 October 2004 FZ 6 4
3 12 October 2004 FZ 7 9
4 12 October 2004 FZ 7 9
5 15 October 2004 FZ 8 9
6 15 October 2004 FZ 8 9
The raw coal was crushed and sealed in plastic bags. To limit the size of the granules to
0.1 mm and below, the material was sieved through a standard screen. Table 2 features the
proximate and ultimate values yielded by elemental analysis of the six samples.
250
Table 2: Proximate and ultimate results of elemental analysis of coal samples

Sample Proximate Analysis Ultimate Analysis
No. (%) (%)
M A V C H O N S
1 0.99 22.27 31.49 87.57 5.46 4.93 1.21 0.83
2 0.82 20.33 29.95 87.15 5.58 5.15 1.14 0.98
3 0.75 14.19 31.42 86.49 4.85 4.32 1.08 3.26
4 0.95 17.08 32.05 87.30 5.06 3.95 1.52 2.17
5 0.89 18.23 34.57 86.12 5.21 5.22 1.07 2.38
6 0.94 19.14 33.04 85.84 5.38 5.37 0.97 2.44
Legend: M = Moisture; A = Ash; V = Volatile
3 Experimental Set-Up and Method
A schematic diagram of the experimental apparatus is presented in Figure 1. It consisted of

two central parts – a heating system and a system for gas analysis. Heating was controlled
by pre-set temperature steps of 2.4 °C min-1 in a range from 20 to 400 °C. A measurement
system served to monitor the temperature change in the coal samples. After extraction of
water with a drier absorber, the oxidation gases were channeled through a system of joint
pipelines to two gas chromatographs, where gas composition and volumes were logged
together with corresponding temperature values.
2 4
10
4 6 15
2 11 12 13 13
5
3
14
2
7 7
1 8 9 8 9
1 – air source; 2 – surge damping valve; 3 – barometer; 4 – flow meter; 5 – heating system; 6 – reactor;
7 – thermocouple; 8 – recording apparatus; 9 – silicon controller; 10 – triple valve; 11 – dehumidifier;
12 – dust filter; 13 – gas chromatograph; 14 – iodine-starch apparatus; 15 – sulfur dioxide gauge
Figure 1: Experimental apparatus for coal calefactive oxidation
251
4 Experimental Results
Coal calefactive oxidation experiments have been discussed at length in Yu et al. (2005),
who examined gaseous products of coal oxidation under similar conditions. Indicator
gases in their study were determined based on the amount of a gas component produced
during oxidation and the lowest temperature at which this component appeared (“critical
temperature”). The present study can be seen as a direct continuation of Yu et al.’s
research; it uses the same experimental set-up and, indeed, broadly replicates its results
with regard to product gases and volumes. The average proportion of these gases in the
total gas discharge is listed in Table 3; it is plotted against coal temperature in Figure 2.
Table 3: Average proportion of gas products from oxidation of Wuda coal samples
T O2 CO2 CO CH4 C2H6 C3H8 i-C4H10 n-C4H10 C2H4 C3H6
(°C) (%) (%) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm) (ppm)
20 20.00 0.003 1.0 5.7 1.2 … … … … …
50 19.91 0.010 1.5 1.6 5.3 0.1 … … … …
70 19.82 0.018 3.2 35.6 14.0 0.6 … … … …
100 19.71 0.020 6.7 61.5 28.2 0.8 … … … …
130 19.57 0.024 13.3 51.4 28.2 1.0 … … … …
160 19.37 0.066 36.1 25.5 22.9 1.3 … 0.2 0.7 …
190 18.89 0.121 221.0 18.6 18.0 1.8 0.1 0.3 2.7 0.3
220 17.57 0.449 1322.8 61.7 29.1 2.1 0.1 0.5 15.2 1.9
250 16.30 0.912 2463.5 69.6 32.9 3.0 0.2 0.8 18.6 3.9
280 11.81 2.832 5611.2 95.5 33.5 5.2 0.5 1.5 29.4 8.8
310 6.84 3.908 8023.5 110.5 27.3 5.3 0.8 2.1 33.8 11.2
340 4.19 6.094 15342.5 158.2 20.4 9.1 5.7 3.7 55.4 16.2
370 2.69 8.590 22262.0 298.0 45.9 55.3 20.3 13.3 75.6 61.3
400 1.52 10.570 23285.2 560.2 132.9 115.4 23.6 25.4 97.4 80.7
252
Figure 2: Average proportion of gas products from oxidation of Wuda coal samples
versus coal temperature
253
5 Discussion and Selection of Indicator Gases
5.1 Carbon dioxide (CO2)
When plotted against coal temperature (Figure 2a), it shows clearly that the quantity of
CO2 emitted during combustion grows exponentially with rising temperature, 200 °C
marking a sharp increase. However, in Wuda Coal Mining Area, CO2 concentrations much
higher than in the present experiments were measured in air flow from burning coal seams,
reaching as high as 0.4 %. CO2 may be absorbed into a coal seam from circumfluent air
with a higher CO2 content, especially so in case of moist coal. CO2 hence cannot function
as a proper indicator of spontaneous combustion, its proportion in the overall gas discharge
from a coal fire being highly variable and dependent on various external factors that need
not necessarily be related to the combustion process.
5.2 Carbon monoxide (CO)
Until well into the 1980s, CO was used as the main indicator to distinguish stages of coal
spontaneous combustion (He et al. 1994); it is still being used in many coal mines as a
very effective and representative means of coal fire detection. Figure 2b shows that the
temperature at which CO is initially produced (“critical temperature”) is very low, and the
absolute quantity generated during oxidation is higher than that of all other combustion
gases (except CO2). CO is released throughout the oxidation process and its output
grows exponentially with rising temperature. Therefore, it can rightly be considered an
appropriate indicator of coal spontaneous combustion and its different stages.
5.3 Alkanes (CnH2n+2)
Numerous alkanes such as methane (CH4), ethane (C2H6), propane (C3H8), and butane
(C4H10) are generated in coal combustion. During slow oxidation, CH4, C2H6, and C3H8
are yielded in gradual succession, both in time of occurrence and in quantity; their
actual volume, though, is minimal. When reaching the stage of accelerated oxidation,
quantities rise significantly, and even little amounts of isobutane (i-C 4H10) and normal
butane (n-C4H10) are released, again in temporal and quantitative succession. During
drastic oxidation, large quantities are produced, their output growing exponentially with
coal temperature. CH4 is discharged along the whole process of oxidation (Figure 2c).
Its usefulness as gas indicator is limited though, since it is also contained in mine air.
C2H6, with the same continued production throughout the process cannot be used either as
volumes evidently fluctuate (two production peaks before drastic oxidation; Figure 2d).
254
i-C4H10 and n-C4H10 do not qualify because they do not appear until accelerated oxidation
(Figures 2f and 2g).
In contrast, C 3H8 occurs already at low temperatures and is released throughout the
oxidation process (Figure 2e). As such, it is an appropriate indicator of early coal
spontaneous combustion.
5.4 Alkenes (CnH2n)
Alkenes such as ethylene (C 2H 4) and propylene (C 3H 6) are also released during the
combustion process; they are used throughout the world, e.g. in China, the United
Kingdom, and the U.S.A., as indicators of coal spontaneous combustion (Zhang et al.
2000). In the Chinese coal mines of Xinji (Huainan Mining Group Co. Ltd.) and Chaili
(Zaozhuang Mining Group Co. Ltd.), C2H4 has proven a very effective indicator gas.
Mine technical personnel in Chaili even calculated coal temperature on the basis of alkene
production (He et al. 1994; Lu & Song 1995).
Figure 2h shows that no C2H4 is produced during slow oxidation. Very small quantities
appear during accelerated oxidation. Output then grows exponentially with rising coal
temperature. In this respect, the occurrence of C2H4 can be said to mark the onset of
accelerated oxidation; it can, however, not be used to signal coal spontaneous combustion
in its early stages and thus is only of limited use in coal fire detection.
5.5 Alkane ratio
Alkanes such as CH 4, C 2H 6, C 3H 8, and C 4H 10 are generated at different rates during

coal oxidation. Many coal mines use this incongruity to determine the progress of
combustion by comparing their ratios. For example, Japan has made good experiences
with C 3 H 8 / C 2 H 6 and C 2 H 6 / CH 4 ratios (Wang et al. 2002). In experiments at
Chaili Coal Mine, alkane ratios were found to be sensitive indicators of early coal
spontaneous combustion in high-methane environment. The authors of the present
study can confirm this: C3H8 / C2H6 and i-C4H10 / C2H6 ratios in the present experiments
clearly signal the different combustion stages when plotted against coal temperature
(Figures 3 and 4). I.e. at a C3H8 / C2H6 ratio of 0.01 to 0.05, coal is in the process of
slow oxidation; at a ratio of 0.06 to 0.1 (and/or 0.001 to 0.005 for i-C4H10 / C2H6), it has
reached accelerated oxidation; beyond a ratio of 0.1 (and/or 0.005 for i-C4H10 / C2H6), it
is in the stage of drastic oxidation.
255
Figure 3: C3H8 / C2H6 ratio versus coal temperature
Figure 4: i-C4H10 / C2H6 ratio versus coal temperature
6 Conclusion
The analysis of gaseous reaction products from coal oxidation depending on temperature
as performed in the present study leads to the following conclusions:
1) The temperature at which CO occurs is relatively low, its absolute occurrence volume,
however, relatively high; it hence makes for an effective indicator of coal spontaneous
combustion.
2) The temperature at which C3H8 occurs is also relatively low; the quantity produced
grows exponentially with rising coal temperature, throughout the oxidation process.
Therefore, C3H8 can also be used to indicate coal spontaneous combustion.
256
3) The ratios of C3H8 to C2H6 and i-C4H10 to C2H6 develop sequentially along the three
stages of coal spontaneous combustion (slow – accelerated – drastic oxidation), and
thus can be applied as sensitive indicators of these stages.
In view of the variety and complexity of the conditions in coal mines, it seems advisable
to diversify the indicator gases currently used for coal fire detection, and to enhance their
application. Also, one should try to combine different indicator gases such as CO, C3H8,
and ratios of C3H8 to C2H6 and i-C4H10 to C2H6, in an effort to improve the accurateness of
predictions.
Acknowledgements
The authors are thankful for the support of this work under the National Program No. 863
Fires” (2003AA131100-02-06) and the National Nature Science Foundation of China
(50274061).
References
Fan, W.T. et al., 2002. China coal industry encyclopedia. Safety volume. Beijing: China Coal Industry
Publishing House. (in Chinese)
He, P. et al., 1994. Characteristics of gases produced in coal oxidation and selection of indicator gases
for prediction of spontaneous combustion. Journal of China Coal Society 19(6): 635-643. (in
Chinese)
He, Q.L. et al., 2003. Critical role of CO in forecasting coal spontaneous combustion. Journal of China
University of Mining and Technology 13(2): 121-125. (in Chinese)
Lu, S.S. and Song, Y.F., 1995. Forecasting and predicting coal spontaneous combustion in Chaili Coal
Mine. Safety in Coal Mines 8:12-13. (in Chinese)
Luo, H.Z. and Qian, G.Y., 2003. Indicator gases for different coal ranks. Safety in Coal Mines 34(B09):
86-89. (in Chinese)
Ren, T.X., Edwards, J.S., and Clarke, D., 1999. Adiabatic oxidation study on the propensity of pulverized
coals to spontaneous combustion. Fuel 78:1611-1620.
Rosema, A., Guan, H.Y., and Veld, H., 2001. Simulation of spontaneous combustion to study the causes of
coal fires in the Rujigou Basin. Fuel. 80(1): 7-16.
Wang, X.S. and Zhang, G.S., 1990. Prevention and mitigation of coal mine fires. Xuzhou: China
University of Mining and Technology Press. (in Chinese)
257
Wang, X.Z. et al., 2002. New technologies for coal mine safety. Beijing: China Coal Industry Publishing
House. (in Chinese)
Yu, M.G. et al. 2005. Study on the usefulness of indicator gases for predicting coal spontaneous
combustion in Wuda Coal Mining Area. Journal of Henan Polytechnic University 24(2):
89-94. (in Chinese)
Zhang, G.S. et al., 2000. Ventilation safety science. Journal of China University of Mining 24(2): 89-94.
(in Chinese)
258
Heat Flux Measurement of Hot Ground
Manfred P. Hochstein1 and Christopher J. Bromley2
1
Geology Department, University of Auckland, Auckland, New Zealand
2
Institute of Geological and Nuclear Sciences, Wairakei, New Zealand
Abstract
The total heat transfer at the surface of hot ground is characterized by the magnitude of heat energy
transferred through various surface discharge features. Diffusive heat discharge is typical for large areas
of warm and hot ground surrounding manifestations with high, focused heat discharge. This applies to
hot ground over high-temperature geothermal systems and to thermal ground associated with burning
coal seams. Measurement of the heat output of warm and hot ground surrounding high temperature
manifestations is described here with reference to examples from the well studied geothermal area of
Karapiti (Wairakei Field, New Zealand).
Heat flux measurement at the surface of hot ground over geothermal systems involves the assessment of
convective, conductive, and radiation fluxes. The total flux and most flux components can be measured
with a water-filled ground calorimeter. The conductive component can be assessed independently using
near-surface soil parameters. All fluxes are affected by daily and seasonal variations. Over hot ground the
total heat flux, when measured with a ground calorimeter, contains convective, conductive, and radiation
components. By lifting the calorimeter slightly above ground, heat transfer by conduction can be eliminated;
this shows that the observed heat flux contains a small radiation component (c. 10 to 20 % of total flux in
our field studies). The convective component becomes significant once the total flux is > 200 W m -2. The
magnitude of total flux measured was between 50 and 1,500 W m-2 and correlates well with the boiling
point depth; this points to a power-law relationship which can be used to infer the total flux for any site
with a known soil temperature profile. Analysis of soil parameters and temperature sections point to a “heat
pipe” transfer mechanism that can maintain a high conductive transfer in a thin surface layer where sub-
surface steam condensation is enhanced.
摘要
在热地表面的总的热传输用各种地表热排放传输的热能大小来确定。周围显示有高强度集中热量
排放的大面积温热地面，热扩散排放是典型的。这应用于覆盖在高温地热系统上的热地面和与地
下煤层燃烧有关的热地面。本文以具有良好研究的新西兰怀拉基地热田卡拉比蒂地热区为例阐述
环绕高温地热显示的温热地面的热量测量。
259
在地热系统顶上热地表面的热通量测量包括对流、传导和辐射通量，其总通量和大部分通量组分
能用充水的地面热量计测定。传导组分用近地表的土壤参数独立评价。所有的通量受日变和季节
变化影响。热地面上的总的热通量当用地面热量计测定时包括对流、传导和辐射组分。从地面上
略微拿起该热量计，传导的传输可以减少；这显示热通量含少量辐射组分（我们的野外研究约为
总热通量的10~20%）。当总热通量大于 200Wm-2 时对流组分变为重要。测得的总热通量的大小
是50~1 500Wm-2 ，它与沸点的深度相关性很好；这给出了可用于推断在已知土壤温度剖面任何地
点的总热通量的功率定律关系。土壤参数和温度剖面分析表明了在地下蒸汽冷凝增强的地表薄层
条件下保持高传导的“热管”传输机制。
1 Introduction
Heat discharged by thermal manifestations, such as hot springs, hot lakes, mud pools,
and steam vents that occur over high temperature hydrothermal reservoirs, was assessed
by methods which had been developed in the early 1960s (Dawson 1964; Thompson
et al. 1964). The magnitude of the inferred total heat output was used to rank New
Zealand’s geothermal prospects. By 1970 field methods had been developed which
allow zoning of thermal ground surrounding thermal manifestations by using patterns
of (reflective) infrared photos (Hochstein & Dickinson 1970) and thermally stressed
vegetation (Dawson & Dickinson 1970). Thermal ground was subdivided into warm
ground and “steaming” ground. It was assumed that in warm ground heat is transferred
by conduction only, which caused no significant increase in surface temperature. Some
“steaming” ground exhibits elevated surface temperatures which were thought to result
from minor steam diffusion (Banwell et al. 1957), although steam discharge is usually
not visible at the surface of “steaming” ground except when cold air and high humidity
prevail. Some of the earlier methods of assessing manifestations with high heat discharge
rates have been revised, for example the assessment of large natural steam discharges
(Hochstein & Bromley 2001).
High-resolution, video-based, aerial infrared surveys have been used since 1989 to
monitor the extent and changes in discharge intensity of hot ground. Attempts were
made to interpret infrared surveys of thermal ground in terms of apparent surface
temperatures and to translate the images into heat loss maps, which, however, produced
only inconsistent results (Allis et al. 1999; Mongillo & Graham 1999; Bromley &
Hochstein 2005). The concealing effects of vegetation and wafting steam and the lack of
representative ground calibration data contributed to the failure. Similar and additional
atmospheric effects also cause difficulties in the quantitative interpretation of satellite
data recorded over hot ground in terms of surface heat fluxes (Mongillo et al. 1993;
Eymard & Taconet 1995).
260
Renewed efforts were made in the late 1990s to develop and test methods which allow
reproducible measurements of heat flux at the surface of thermal ground, especially of
“steaming” ground over geothermal prospects in New Zealand (Hochstein & Bromley
2002). The results of these studies are summarized in the following section.
2 Terminology
Our heat flow studies were made over two separate large areas of thermal ground of the
greater Wairakei system in New Zealand, whose reservoir has been exploited since 1958.
Earlier measurements had shown that several heat flux modes transfer heat through and
away from the surface (Hochstein & Bromley 2002). These are shown in Figure 1 and
include the following components:
1) Convective transfer of minor steam through the surface that is associated with the heat
flux component qconv.
2) Conductive transfer qcond through a thin, near-surface layer; this flux contains a time-
variable component (∆qdy) resulting from daily variations of near-surface temperature
and near-surface soil temperature gradient (∆T / ∆z0); the depth z is positive downward.
3) Heat transfer by radiation (qrad) to the atmosphere is controlled by the (absolute) surface
temperature T0 and the surface emittance. With reference to measurements from air
(space), a flux difference ∆qrad between hot ground and surrounding “cold” ground
can be defined. A smaller fraction ∆’qrad enters measurements made with a ground
calorimeter and is controlled by the temperature and acceptance of the calorimeter
bottom. Both ∆qrad and ∆’qrad contain a variable component depending on T0.
4) An episodic flux component qfilt is caused by surface water infiltrating the ground after
a period of rainfall. This component can be neglected during dry periods (> 5 dry days,
from our experience).
5) A seasonal component ∆qyr causes in colder seasons a decrease in ground temperatures

and increases the depth below which boiling temperatures prevail. This component is
smallest during the summer months when the total heat flux of hot ground reaches a
maximum.
261
Figure 1: Sketch showing heat flux components at the surface of hot ground and inferred
fluid movement and phase changes at shallow depths beneath hot ground
All heat flux modes and flux components listed refer to a specific heat flux (unit: W m-2).
All surface heat flux components (including the total flux qtot) are time-variable parameters.
The effect of time-variable disturbances is reduced if measurements are taken during dry
periods in the summer and during the same daily period under overcast conditions. Total
flux measurements can then be made on the surface (z = 0) using a calorimeter with the
total flux given by:
(1a) qtot(0) = qconv + (qcond + ∆’qrad) .
The conductive component qcond can be eliminated if flux measurements are taken at a small
height (-h) above the plane surface, in which case the radiation fraction ∆’qrad becomes
noticeable although the observed flux also contains a fraction (∆) of qconv. For this setting:
(1b) qobs(-h) = ∆qconv + ∆’qrad .
Measurement of the various flux components are described below with particular reference
to ground surveys of the well-studied Karapiti thermal ground (part of the greater Wairakei
geothermal system, New Zealand).
262
3 Measurement of Total Flux qtot
The ground flux of several New Zealand geothermal prospects was first measured
almost 50 years ago by Banwell et al. (1957), who tested calorimeters with closed and
open bottoms. The open bottom version (differential psychrometer type) was refined by
Benseman (1959), who obtained some average flux values although data reproducibility
was poor and conductive and convective components could not be separated. Calorimeter
studies of hot ground in New Zealand were discontinued after 1965.
A few years ago, we constructed a new ground calorimeter with a closed bottom, whose
design has been presented elsewhere (Hochstein & Bromley 2002) and which yields
reproducible data. The calorimeter consists of a flat, cylindrical vessel with a bottom
diameter of 240 mm and a liquid capacity of 1.5 l. It has a very thin steel bottom (0.9 mm
thick); a stirrer provides good mixing when the vessel is filled with water (usually close to
ambient air temperature) and placed on thermal ground. Conductive losses or gains through
mantle and lid are reduced by a double layer of insulating material. The temperature
change of the water inside the calorimeter is monitored by several mini-bead thermistors.
The temperature of lid and mantle is also monitored. The heat flux entering the vessel
through the bottom (0.045 m2) can be assessed from the rate of heating of the water mass
after reducing calorimeter drift. Calibration tests in the laboratory showed that the thermal
resistance of the bottom plate is small and can be neglected. For field measurements the
surface of each site is leveled; stable ground temperatures are usually reached after one
day with the site remaining shielded from the sun. With the calorimeter on the ground at a
prepared stable site, its rate of heating is then monitored using a data logger, usually at 15 s
intervals. Shallow soil temperatures at five depths (between 0.01 and 0.2 m) are measured
before and after the heat flux measurements.
The heat flux qtot(0) entering the calorimeter can be obtained by using a time-based,
temperature gradient reduction. The recording procedure adopted was to occupy a site
for 5 min periods separated by an “off-the-ground” period of 2 min to check temperature
drift. This can be seen from the sample record shown in Figure 2. The temperature
drift rate (∆Tdr / ∆t) of the liquid fill can be predicted from characteristics of the lid-
temperature curve. Denoting the recorded steep temperature rise during a recording time
interval by (∆Trec / ∆t), the heat flux is given by:
(2) qtot(0) = [mc{(∆Trec / ∆t) – (∆Tdr / ∆t)}] / A ,
where m is the water mass, c its specific heat, and A the ground area covered. The
uncertainty in computing heating rates can be reduced by using a least squares linear fit
263
applied to segments of the record. The heating rate in the example of Figure 2 shows a
quasi-exponential decline which is mainly caused by a decline of the convective flux. Its
effect can be reduced when computing the undisturbed flux at the beginning of the record
(see Figure 2). At other sites, heating rates are rather constant. The short- and long-term
reproducibility of qtot(0) was checked by re-occupying the same site several days and a few
months later. During summer months, the repeat values differed on average by c. 7 % from
the means. All observed total flux data contain the radiation flux component ∆’ qrad relative
’
to ambient air temperature and are subsequently referred to (with a prime) as q’tot .
Figure 2: a) Calorimeter data recorded over hot ground; the soil surface temperature
beneath the calorimeter was 40 °C.
b) Plot of recording time versus reduced heating rate of the water mass in the
calorimeter (taken from the record shown above).
4 Measurement of Convective Flux qconv
During the first field tests, it was found that condensation of diffuse steam usually occurs
at the bottom of the calorimeter when measuring qtot. Collecting all condensates (mass mc)
with dry tissue paper showed that the rate of condensation (∆mc / ∆t) is almost the same for
repeat measurements at the same site a day apart. The associated heat transfer rate (from
latent heat) is given by:
(3) qconv = [(∆mc / ∆t) / A] (hg - hc) + L ,
where A is the area of the calorimeter bottom plate, hg the specific enthalpy of steam
escaping at the surface with temperature T0, hc the enthalpy of the condensate droplets, and
L an unknown small loss component (droplets adhering to the ground).
A plot of qconv versus q’tot of our test data in Figure 3 (taken from Hochstein & Bromley
264
2005) shows a clear positive correlation. At sites with repeat measurements, the normalized
standard deviation of qconv lies between 7 and 30 %. The graph shows that convective
transfer can account for up to 50 % of the heat transfer of “steaming” ground although no
visible steam discharge can usually be observed.
Figure 3: Plot of observed convective flux q conv versus observed total flux q’tot at
calibration sites
5 Assessment of Radiation Flux Components ∆q’rad and ∆qrad
When the calorimeter is raised at a constant level above the ground (placing it for example
on a ring of height -h) and measurements are conducted in the same way as described in
Chapter 3, the observed flux qobs(-h) contains still a fraction of the convective flux (∆qconv)
observed on the ground and the radiation flux component ∆’qrad. Subtracting ∆qconv from
qobs(-h), the component ∆’qrad can be obtained. This flux can be assessed by using the Stefan-
Boltzmann law for heat transfer by radiation between two plates with:
265
(4a) ∆’qrad / F = ε1 σ T04 – α12 σ T 4 (with T0 > T) ,
where ε1 is the “emittance” of the hot ground, α12 the “absorptance” of the bottom of the
calorimeter, T0 the surface temperature of the hot ground (in K), and T the temperature of
the calorimeter bottom (in K); σ is the Stefan-Boltzmann constant (0.567 × 10-9 W m-2 K-4),
and F the so-called “direct view factor” (Perry & Chilton 1973). Retaining α12 as parameter
and assuming that ε1 = 0.95 and F = 0.9 for (-h) = 0.02 m, we found agreement between
computed and observed values of ∆’qrad when α12 is close to 0.9. When the calorimeter is
placed on the ground, the effect ∆’qrad can be computed for each site using equation (4a)
assuming that F = 1. It was found that ∆’qrad in our measurements is commonly between 10
to 20 % of q’tot(0). Since the calorimeter temperatures were within ± 5 °C of ambient (air)
temperature, the observed ∆’qrad component is equivalent to the radiation flux as measured
with a sensor at mean air temperature.
When measurements from an aircraft are available (images from an airborne infrared
survey, for example), the flux difference ∆qrad between a given (unit) test area with surface
temperature T0 with respect to a unit area of surrounding “cold” ground with temperature T
is given by:
(4b) ∆qrad = ε1 σ T04 – ε2 σ T 4 (with T0 > T) ,
where ε1 and ε2 are the emittance of hot and cold ground, respectively.
6 Assessment of Surface Conductive Heat Flux qcond
The conductive heat flux at the surface is obtained from the difference of the two fluxes
qtot and qconv, namely: qcond = qtot - qconv. However, the observed flux q’tot contains the small
radiation flux component ∆’qrad which we had neglected in our earlier studies. Hence, the
difference (q’tot - qconv) stands for an apparent conductive heat flow q’cond which is greater
than qcond. During earlier tests we tried to obtain an independent estimate of the (true)
conductive heat flux at the surface (z = 0) using the relation:
(5) qcond = (∆T / ∆z)0 λ(z0) ,
where (∆T / ∆z)0 is the vertical temperature gradient and λ(z0) the thermal conductivity
of the surface (contact) layer. In thermal ground, the temperature gradient decreases with
266
depth and contains a variable component which reflects the effect of the variable daily
surface flux (∆qdy). Temperature measurements were made at each calorimeter site at depths
of 0.01 and 0.05 m at several (up to 5) points, which provided a representative near surface
T gradient. The surface of most geothermal areas in New Zealand is covered by weathered
pumice soils which exhibit high porosity (between 60 to 80 %). The thermal conductivity
was computed by using the relationship:
(6) λ(z) = α ceff σw ,
where α is the thermal diffusivity [m2 s-1], ceff the effective thermal capacity per unit mass
[J kg-1 K-1], and σw the bulk (“wet”) density [kg m-3] of a soil sample from the surface layer
beneath the calorimeter site. The parameters σw and ceff were obtained from soil sample
measurements, which yielded porosity Φ and the liquid saturation fraction Sl for each
sample. Using an auto-correlation approach to check the relation between observed and
computed conductive flux values (Hochstein & Bromley 2005), it was found that for high
porosity soils with significant moisture contents (0.2 < (ΦSl) < 0.6) the best fit thermal
diffusivity α is given by the empirical relation:
(7) α = [-0.25 (ΦSl) + 0.4] × 10-6 [m2 s-1] .
With the observed near-surface gradient (∆T / ∆z)0 and the computed site-specific
thermal conductivity λ(z), the conductive flux qcond can be obtained. It was found that best
fit thermal conductivity values λ(z) of porous wet soils are rather low (all between 0.4
and 0.9 W m-1 K-1). Conductive losses have often been assessed in older studies by using
an “average” thermal conductivity taken from data published in handbooks or tables of
(geo-)physical parameters. That approach can introduce systematic errors since thermal
conductivity of soils is not a constant but a variable, controlled by the degree of water
saturation as our studies have shown.
7 Constructing a Heat Flux Map of Hot Ground
7.1 Use of boiling depth zBP to predict heat flux components
Measurement of the various heat flux components across a large area of hot ground using the
calorimeter method is time-consuming. We searched therefore for a characteristic parameter
which allows assessment of flux components at stations located in between the calorimeter
sites. Tests showed that the boiling depth zBP beneath a calorimeter station can be correlated
267
with q’tot and q’cond. At stations where some, even minor, anomalous conductive transfer
occurred, soil temperature Tz versus depth z profiles exhibit a typical quasi-exponential shape
which can be linearized by using appropriate empirical functions (Hochstein & Bromley 2005).
Hence, zBP data can be obtained from any set of Tz data. For the observed total flux q’tot and its
conductive component q’cond, a power-law relationship of the form:
(8) q = a [1 / (zBP / z0)b]
was found where (zBP / z0) is a dimensionless depth with z0 = unit depth (1 m), and a
and b are empirical constants. Figure 4 shows the relationships for q’tot and q’cond based
on data from the Karapiti area. The values of constant a listed in Figure 4 indicate that
the conductive flux at the surface is almost half the total flux, which implies that the
conductive flux almost equals the convective flux (see also results shown in Figure 3). The
constants will be affected by the average permeability of the soil which controls subsurface
steam flux. Additional calibration tests are required if equation (8) is used to assess the
heat discharge of hot ground in other thermal areas.
Figure 4: Relationship between (a) boiling depth (zBP) and q’tot and (b) boiling depth zBP
and q’cond
7.2 Thermal vegetation zoning
When all flux data from calorimeter stations (n = 20) and data derived from stations with
known zBP values (n = 85) were plotted on a map covering the c. 0.35 km2 thermal area of
Karapiti, it was found that additional information was still required to obtain representative
heat flux contours for the whole area – for this we used thermal vegetation zoning.
The vegetation pattern over thermal ground is related to boiling depth since the average
height of plants and shrubs which grow on thermal soil is controlled by the vertical extent of
their roots, which, in turn, is restricted by boiling depth. Height and type of plants growing on
268
thermal ground are also affected by other parameters, such as availability of mineral nutrients
and soil pH value, although temperature at the deepest root level is the most important
parameter (Given 1980). Zoning of thermally stressed plants at Karapiti was originally used
by Thompson et al. (1964) to outline coherent areas with high ground temperatures. We
used the more robust parameter of boiling depth zBP for zoning. For this a “vegetation index”
was assigned to each site with known or inferred heat flux. A plot of log zBP versus adopted
vegetation index is shown in Figure 5. Areas with the same index were assessed from vertical
air photos (in color) and composite aerial infrared images (Bromley & Hochstein 2005).
7.3 Heat flux at the surface of Karapiti thermal ground
The area around each occupied site was assigned to one of the seven zones shown in Figure 5,
which are closely linked to the vegetation index and the depth range of zBP values. The extent
of the inner zones (1 and 2) is defined by the anomalous surface temperature patterns on
the aerial infrared images; the adjacent zone 3 (vegetation index 4) can be identified by its
subdued infrared anomalies. Zones 4 and 5 were identified from field observations of the
height of shrubs. The extent of thermal ground in zones 6 and 7 and the surrounding cold
ground (high standing pine trees and native bush) was also taken from air photos.
Figure 5: Vegetation index and boiling depth zBP (log scale) at sites with known heat flux
(Karapiti thermal area)
269
The approximate aerial extent Ai of the seven zones within the Karapiti thermal ground
is shown in Figure 6; the average heat output (Qi = qtot Ai) of each zone i can be assessed
from the range of boiling depth zBP values and using the power law relationships shown in
Figure 4. Isolated anomalies based on the value from a single site were disregarded. The
area Ai of each thermal zone was obtained by integration; subtotals of discharged heat
are listed in Table 1. The area of warm ground of zones 6 and 7 could not be separated so
their outputs are combined. Zone 1 is confined to large steaming craters exhibiting bare
ground. Here, concentrated heat discharges via fumaroles, steam vents, and hot pools.
These manifestations transfer c. 90 MW of heat in zone 1 (Hochstein & Bromley 2001)
and c. 17 MW in zones 2 and 3 (see Table 1). In contrast, wide-spread diffuse discharge of
heat through warm and hot ground amounts to a sub-total of 141 MW (approximately 15 %
of this subtotal is probably due to radiation losses with respect to ambient air temperature);
all heat is discharged over an area of 355,000 m2. The convective losses can also be
expressed in terms of surface steam flux rates; iso-flux contours are shown in Figure 6,
which gives a good overview of the whole setting of this type of thermal ground.
Table 1: Heat discharge of thermal ground and thermal manifestations (Karapiti)

Zone Vegetation Range Average Area Heat Heat
i Index zBP (m) Flux Ai(m2) Diffuse Direct
qi tot (W m-2) Qia(MW) Qdb(MW)
1 1 < 0.1 c. 1350 c. 5,300 c. 7 90

2 2&3 0.1-0.2 810 ± 50 63,000 51.5 15
3 4 0.2-0.4 480 76,000 36.5 c. 2
4 5 0.4-0.8 275 135,000 37 0
5 6 0.8-1.5 170 43,700 7.5 0
6 7 1.5-3.0 65 ± 10 31,200 2 0
7 7 3.0-10
Sum 355,000 141.5c 107
a
Diffuse heat discharge due to convective and conductive heat flux
b
Concentrated heat discharge (dominantly by steam) through manifestations such as fumaroles, steam
vents, small hot- and mud pools
c
Radiation losses with respect to calorimeter temperatures (similar to average ambient air temperature)
are included in the subtotal; these account for an estimated 15 % (i.e. c. 20 MW) of the diffusive heat.
270
Figure 6: Map of Karapiti thermal area showing zones with the same range of total heat
flux and boiling depth (from a 2003/4 summer survey). Heat flux sites are
shown by a small cross, the figure at each site lists boiling depth zBP (rounded)
in m; also shown are contours of daily steam flux rates (in kg m-2 d-1) which
were derived from convective flux qconv data.
271
7.4 Aerial infrared surveys
When compiling the heat flux-related data shown in Figure 6, images from an aerial
infrared survey were also used to outline areas of similar apparent radiation pattern
which can be interpreted in terms of apparent surface temperatures (Figure 7; taken from
Bromley & Hochstein 2005). Airborne infrared surveys have been conducted over several
geothermal prospects in New Zealand since 1989 to monitor changes in the extent of
surface manifestations. The helicopter-borne infrared survey shown in Figure 7 was made
at night on 1 May 2000 (c. 600 m above ground) to minimize solar effects. Emissive
infrared radiation in the wave band of 8 to 12 μm was recorded. Temperature calibration
of the imagery was from water surfaces at measured temperatures. The original data
were recorded as 8 bit (0-255) grey-scale values covering a range of apparent surface
temperatures from 11.5 to 63.5 °C. The composite infrared image shown in Figure 7 was
compiled from the 2000 infrared survey (Bromley & Hochstein 2000), using a mosaic of
images digitally captured from videotape and overlaid on an enlarged air photo.
Figure 7: Thermal infrared map of Karapiti area obtained from a mosaic of images from
a 2000 infrared survey; images were fitted to an enlarged air photo
272
Any quantitative interpretation of infrared imagery data in terms of assessing heat losses
from hot and steaming ground is confounded by the difficulties of obtaining representative
surface calibration temperatures (at the time of the survey) and the masking effects of
vegetation and wafting steam (Bromley & Hochstein 2005). The apparent temperature at
the top of steam clouds over fumaroles and steam vents, for example, cannot be used to
assess the heat discharge of these manifestations. However, information in Figure 7 was
used to define the hot, bare ground areas of zone 2 shown in Figure 6 where zBP occurs
at 0.1 to 0.2 m depth and to outline zone 3 where 0.2 m < zBP < 0.4 m. A quantitative
interpretation of the data in Figure 7 is still outstanding.
8 Steam Flux and CO2 Emission
Steam derived from subsurface boiling usually contains some carbon dioxide (CO2) gas. In
thermal areas where the CO2 / steam ratio is rather constant, the diffuse steam flux through
hot ground has been assessed from CO2 soil gas surveys (Chiodini et al. 2001). Fumarole
gases discharged over the Karapiti area had a mean CO2 / steam ratio of 190 ± 26 mmol
CO2 / 100 mol H2O [1 mmol = 1 millimole] during the 1990s (Glover et al. 1999). Since
rapid measurement of CO2 fluxes can now be made using, for example, accumulation
chamber methods, we measured the CO2 flux at most sites where the convective flux qconv
had been observed or could be inferred from zBP data. The qconv data, in turn, were converted
into specific steam mass flux data for each site, which could be contoured; some of the
contours are shown in Figure 6. A plot of steam flux values versus observed CO2 fluxes,
however, showed poor positive correlation (R2 = 0.35), barely significant due to the large
data scatter (Werner et al. 2004). The mean of the CO2 / steam ratio was 150 ± 110 mmol
CO2 / 100 mol H2O (for 80 stations covering c. 300,000 m2 of the Karapiti area). The large
standard deviation of this mean reflects the data scatter which points to yet unknown CO2
enrichment and depletion processes at very shallow depths.
The gas emission study shows that diffusive steam flux data, and hence convective flux
(qconv) values, cannot be predicted with any accuracy from escaping CO2 soil gas data at
Karapiti. The data, however, were adequate to deduce a CO2 emission rate of 6 to 7 t CO2
per day for the thermal area, a rate which is less than half that of the 16 t of CO2 discharged
per day by all fumaroles and steam vents located inside the steaming craters shown in
Figure 6 (Werner et al. 2004).
273
9 Explanatory Heat Transfer Model
Separate soil samples at 0.05 to 0.1 m depth intervals were taken down to boiling depth
at several calorimeter sites in the Karapiti test area, together with temperature profiles.
Soil parameters, such as specific density (dry, saturated, and particle density), porosity,
and liquid saturation, were measured in the laboratory (Hochstein & Bromley 2005). Plots
of these parameters versus depth showed that enhanced condensation of steam occurs in
a thin (usually < 0.1 m) shallow layer. Most of the conductive heat transfer is driven by
this condensation and the subsequent liberation of latent heat, which heats this layer. The
process maintains a steep temperature gradient above the condensate layer. The liquid
saturation Sl rarely exceeds 70 % in the condensate layer. Thus some vapor can always
pass through this layer to the surface. The ascending vapor, together with some evaporation
from the condensate layer, maintains the convective heat transfer at the surface. The liquid
saturation of pumice soils below the condensate layer down to boiling depth, and below
it, is between 50 and 60 % and points to the existence of a natural two-phase flow with
an upflow of vapor and a counter flow of liquid droplets from the condensate layer. If the
liquid droplets evaporate in the two-phase layer, this constitutes a “heat pipe” setting as it
is known from laboratory studies (Bau & Torrance 1982) and which is known to be a very
effective heat transfer system. Capillary forces are also involved, as shown by numerical
modeling of the temperature-depth curve at one of our sites (Newson & O’Sullivan 2002).
An explanatory model of the inferred fluid movements at Karapiti is shown in Figure 1
(from Hochstein & Bromley 2005). Processes leading to local enrichment and depletion of
CO2 have still to be incorporated into the model.
10 Summary
The total heat flux and the total energy transferred through the surface of hot ground,
away from manifestations with high heat discharge, consists of a convective, a subsurface
conductive, and a surface radiation flux component. These fluxes can be measured on
the ground with a calorimeter at selected calibration sites. The conductive flux can be
obtained independently from measurement of near-surface soil parameters and temperature
gradients. All three flux components are affected by daily and seasonal variations, which
can be reduced in part by using repeat measurements during the same daily and seasonal
periods. Once a calibrated calorimeter network of stations has been established, convective
and conductive heat flux values can also be inferred from soil temperature profiles, using
boiling depth as a characteristic parameter. Zonation of vegetation and infrared patterns
274
over thermal ground by using air photos and infrared imagery can define local zones
with similar heat transfer characteristics. Tests show that the convective component can
not be inferred from soil CO2 measurements since the CO2 / steam ratio exhibits large,
local variations pointing to unknown shallow enrichment and absorption processes. The
calorimeter method described in this paper allows the assessment of the total energy
transfer over hot ground and could be applied, in principle, to quantify the energy release
over areas with burning coal seams, thus providing additional calibration data for the
quantitative interpretation of satellite data.
Acknowledgements
The initiative of Prof. Horst Rüter of HarbourDom GmbH is acknowledged in inviting

us to present a summary of our heat flow studies over hot ground at the International
Conference on Coal Fire Research (Beijing, 29 November to 1 December 2005).
Permission from the Associate Editor of Geothermics to include here previously published
Figures 1, 2, 3, and 4 is gratefully acknowledged.
References
Allis, R.G., Nash, G., and Johnson, S., 1999. Conversion of thermal infrared surveys to heat flow:
Comparisons from Dixie Valley, Nevada, and Wairakei, New Zealand. Geothermal Resources
Council Transactions 23: 501-507.
Banwell, C.J., Cooper, E.R., Thompson, G.E.K., and McCree, K.J., 1957. Physics of the New Zealand
thermal areas. Bulletin 123. New Zealand Department of Scientific and Industrial Research,
Wellington, New Zealand, 30-36.
Bau, H.H. and Torrance, K.E., 1982. Boiling in low permeability porous materials. International Journal
of Heat and Mass Transfer 25: 45-55.
Bensemann, R.F., 1959. The calorimetry of steaming ground in thermal areas. Journal of Geophysical
Research 64: 123-126.
Bromley, C.J. and Hochstein, M.P., 2005. Heat discharge of steaming ground at Karapiti (Wairakei), New
Zealand. In Proceedings of World Geothermal Congress 2005, Antalya, Turkey, April 2005,
paper 0727. (on CD only)
Bromley, C.J. and Hochstein, M.P., 2000. Heat transfer of the Karapiti fumarole field (1946-2000).
In Proceedings of the 22nd New Zealand Geothermal Workshop, University of Auckland,
Auckland, New Zealand, 8-10 November 2000, 87-92.
Chiodini, G., Frondini, F., Cardellini, C., Granieri, D., Marini, L., and Ventura, G., 2001. CO2 degassing
and energy release at Solfatara Volcano, Campi Flegrei, Italy. Journal of Geophysical
Research 106 (B8): 16213-16221.
275
Dawson, G.B., 1964. The nature and assessment of heat flow from hydrothermal areas. New Zealand
Journal of Geology and Geophysics 7: 144-154.
Dawson, G.B. and Dickinson, D.J., 1970. Heat flow studies in thermal areas of the North Island of New
Zealand. Geothermics, Spec. Issue 2: 466-473.
Eymard, L. and Taconet, O., 1995. The methods for inferring surface fluxes from satellite data, and their
use for atmosphere model validation. International Journal of Remote Sensing 16: 1907-1930.
Given, D.R., 1980. Vegetation on heated soils at Karapiti, Central North Island, New Zealand, and its
relation to ground temperature. New Zealand Journal of Botany 18: 1-13.
Glover, R.B., Mroczek, E.K., and Finlayson, J.B., 1999. Changes in major gas concentrations in the
Karapiti thermal area in response to development of Wairakei. In Proceedings of the 21st New
Zealand Geothermal Workshop, University of Auckland, Auckland, New Zealand, 10-12
November 1999, 7-13.
Hochstein, M.P. and Bromley, C.J., 2005. Measurement of heat flux from steaming ground. Geothermics
34: 131-158.
Hochstein, M.P. and Bromley, C.J., 2002. Assessment of heat losses of steaming ground by calorimetry.
In Proceedings of the 24th New Zealand Geothermal Workshop, University of Auckland,
Hochstein, M.P. and Bromley, C.J., 2001. Steam cloud characteristics and heat output of fumaroles.
Geothermics 30: 547-559.
Hochstein, M.P. and Dickinson, D.J., 1970. Infrared remote sensing of thermal ground in the Taupo
Region, New Zealand. Geothermics, Spec. Issue 2: 420-423, 428-429.
Mongillo, M.A., Browne, P.R.L., Cochrane, G.R., Derroin, J.P., 1993. Satellite studies of Craters of
the Moon geothermal area. In Proceedings of the 15th New Zealand Geothermal Workshop,
University of Auckland, Auckland, New Zealand, 11-13 November 1993, 87-92.
Mongillo, M.A. and Graham, D.J., 1999. Quantitative evaluation of airborne video TIR survey imagery.
In Proceedings of the 21st New Zealand Geothermal Workshop, University of Auckland,
Newson, J.A. and O’Sullivan, M.J., 2002. Computer modeling of heat and mass flow in warm ground,
Karapiti. In Proceedings of the 24th New Zealand Geothermal Workshop, University of
Auckland, Auckland, New Zealand, 13-15 November 2004, 273-278.
Perry, R.H. and Chilton, C.H., 1973. Chemical Engineers Handbook. Section 10: Heat transmission by
radiation. International Student Edition. Kogakusha, Japan: McGraw Hill, 48-64.
Thompson, G.E.K., Banwell, C.J., Dawson, G.B., and Dickinson, D.J., 1964. Prospecting of hydrothermal
areas by surface thermal survey. In Proceedings of the UN Conference on New Sources of
Energy, Rome, Italy, August 1961. Vol. II, 386-401.
Werner, C.A., Hochstein, M.P., and Bromley, C.J., 2004. CO 2 flux of steaming ground at Karapiti,
Wairakei. In Proceedings of the 26th New Zealand Geothermal Workshop, University of
Auckland, Auckland, New Zealand, 6-9 December 2004, 59-64.
276
Gas and Temperature Monitoring
of a Spontaneous Coal Seam Fire
in Wuda Coal Mining Area
Stefan Schlömer, Manfred Teschner, Jürgen Poggenburg, and Christian Seeger
Abstract
This study was conducted within the framework of the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North China.” Since 2003 this
extensive joint research program has been investigating spontaneous coal seam fires in China. Near the
township of Wuda (Inner Mongolia Autonomous Region, P.R. China), an area of 280,000 m2 is affected
by subsurface coal fires, most of which can be traced to small-scale mining operations. The analysis of
combustion gases of two fire zones in 2004 revealed that a variety of combustion processes take place at
the same time in a coal fire. The occurrence of CH4 and CO is associated with coal pyrolysis processes
(i.e. the chemical decomposition of organic materials by heating in the absence of O2 or other reagents);
H2 results from coking, i.e. the generation of volatile constituents (water and coal gas) or coal gasification
processes (conversion of carbonaceous materials into CO and H2).
For the first time, combustion gases and temperatures of spontaneous coal seam fires were measured in-situ
over an extended period of time. At different sites close to the main combustion zone (c. 25 m) of fire zone 8,
one of the largest fire zones in the Wuda area, temperatures at three vents and gas composition at one
vent were measured from June to October 2005. Data were logged every 30 seconds and complemented
by meteorological data (wind direction and velocity, barometric pressure, and ambient temperature).
Temperature measurements at vents with gas discharge showed intense fluctuations of temperature (up
to ± 100 °C within 48 h), which could not be related altogether to changes in meteorological conditions.
These short-term fluctuations were superimposed by long-term temperature trends. Variations over the
entire measuring period ranged from +100 to -30 °C per month. They are thought to be a result of the
advancing fire front – a conclusion given further credence by the change in gas composition. Over the
measuring period of c. 5.5 months, the CO / CO2 ratio significantly decreased. At the same time, the
temperature at one sampled vent increased by 35 °C (from 80 to 115 °C). Both findings indicate that the
main combustion zone was approaching. However, the observed trends are smaller than expected and
point to a very slow fire progression (< 10 m yr-1).
277
During October 2005, water flooding experiments were carried out by the project partner Deutsche
Montan Technologie GmbH (DMT). Up to 15 m3 of water were pumped below surface at three different
locations, and the impact on temperature and gas composition was measured by our group (Federal
Institute for Geosciences and Natural Resources, BGR). In one experiment close to the gas measuring
site (where about 4.5 m3 of water were injected), a significant change in gas composition was observed.
Due to the limited period of monitoring that followed, this change could only be attributed to dilution
through steam. A direct effect on the combustion process could not be verified. The injection of 15 m3 of
water close to the main combustion zone in another experiment led to intense steam production and a
temperature drop from 290 to 260 °C within a short period of time in a nearby fracture. The temperature
continued to decrease almost linearly for several days. This suggests that the flooding had significantly
influenced the combustion process, the temperature decline resulting from a subsequent cooling of the
bedrock.
摘要
本研究是在中德煤火研究工作意向框架下执行的。2003年以来，中德双方在中国内蒙古自治区的
乌达矿区进行了煤火调查。该区地下煤火影响的面积达到280 000平方米，其中大部分火区的形成
与小煤窑开采有关。2004年，通过对两个火区带的燃烧气体分析，表明有大量的燃烧过程正在同
时进行中。煤的热解过程中（如在缺氧或其他气体条件下加热时有机物的化学分解），CH4和CO
气体出现；焦化作用中产生了H2，如产生挥发性组分（水和煤气）或者煤的气化作用过程（碳化
物转换成CO和H2）。
在现场，首次进行了煤火燃烧的气体和温度的长时间观测。从2005年6月到10月，在乌达矿区最大
的火区之一8号火区中，距主要燃烧带约25米远裂隙上，布置了三个温度测量点和一个气体测量
点，每30秒采集一次数据，并补充气象数据（风向、风速、大气压、环境温度）。采用气体排放
法的温度测量结果显示出强烈的温度变化，在48小时内变化幅度达到±100℃，这种变化总的说不
会受气象条件变化影响。长期温度变化趋势与短期波动变化相互叠加在一起，在整个观测期间内
每个月的变化范围从+100到-30℃。推测该观测位置处于火区前缘，同时采用气体成分观测的结
果也证明了该推测结果。在五个半月中，观测的CO / CO2比率显著下降，同时，在一个观测点的
温度增加了35℃（从80到115℃）。两者都显示正在观测的位置靠近主要的燃烧带。然而，观测结
果表明煤火发展趋势较预期的要慢，而煤火传播缓慢到每年发展距离小于10米。
在2005年10月，由德国DMT进行了注水试验。从地表三个不同的位置将15立方米的水注入地
下，同时BGR测量了对于温度和气体成分的影响。在靠近一个气体测量位置，将约4.50立方米
的水注入地下，观测到气体成分有显著变化。由于随后的监测时间有限，只观测到了产生的水
蒸气逐步达到稀释，尚未发现对煤自燃的直接影响作用。另外一项试验是将15立方米的水注入
到主要燃烧带，产生了强烈的蒸汽，在短时间内附近的裂隙中温度由290℃下降到260℃。随后
的几天，观测的温度几乎呈线性持续下降。这表明注水对于燃烧过程具有显著的影响作用，温
度下降导致附近的基岩降温。
278
1 Introduction
Outcropping coal seams and coal waste heaps are burning around the world, leading
to severe environmental and economic damage (Stracher & Taylor 2004). Apart from
depleting exploitable coal reserves, they cause land subsidence and gravely affect human
health in their vicinity. Enormous quantities of noxious gases (SO2, NO, CO2, CO, CH4)
and particulate matter (ash) are emitted into the atmosphere; their condensation products
pollute water and soil (Stracher 1995, 2003, 2004; Stracher & Taylor 2004).
Since 2003, the Sino-German Coal Fire Research Initiative has been investigating the
coal fire problem in China. Aiming to develop innovative technologies for exploration,
extinction, and monitoring of spontaneous coal seam fires, it assembles under its umbrella
various partners from different geo-scientific backgrounds. As a prerequisite for successful
extinction, the first phase of the project gave special attention to the processes leading up
to coal fires (i.e. spontaneous combustion) and those controlling their spatial and temporal
development. One major aim was to build static and – added time-related information –
dynamic 3D models of combustion processes. Many coal fires were detected during field
work; they had been burning for years but gone unnoticed for want of continuous and early
combustion gas/smoke detection systems (de Rosa 2004). Detailed (static and continuous)
gas measurements now allow for identification of underground combustion processes and
their temporal changes. Such data are an important foundation for emission monitoring,
dynamic modeling of coal fires (cf. Wessling et al. 2008 in this volume), and surveillance
of potentially hazardous areas.
In-situ one-time gas and temperature measurements have been performed by different
working groups (Federal Institute for Geosciences and Resources, BGR; Deutsche
Montan Technologie GmbH, DMT). These measurements helped to identify fire
zones and provided indicators for prevailing combustion processes (see below); they
also resulted in a classification and mapping of cracks and vents, and their spatial
distribution, as well as an estimation of total gas flows (Litschke et al. 2008 in this
volume). However, from the sparse gas-geochemical data available for underground coal
fires and our own experience, we knew that subsurface coal fires are highly dynamic
systems in time and space. Therefore, it was deemed necessary to carry out long-term
measurements of gas emissions.
Gas-geochemical sampling typically requires that samples be taken directly from vents
or from soil with gas discharge. This can be risky and might only be carried out on a
279
weekly or even monthly basis for remote locations. Continuous monitoring based on
optical and/or chemical sensing techniques coupled with temperature measurements is
preferable. Permanent multi-parameter measuring stations offer a wide range of potential
applications and have been used by BGR for several years in the field of volcano and
mud volcano monitoring (Faber et al. 2003; Seidl et al. 2003; Delisle et al. 2005;
Teschner et al. 2005), earthquake prediction (Bräuer et. al. (submitted)), and CO2 storage
(Brune et al. 2003).
BGR developed a continuous monitoring system of gas emissions, which was put into
operation for over five months in the study area of the Research Initiative, yielding
a considerable amount of data on combustion gas composition, vent temperature,
and weather conditions (i.e. barometric pressure, temperature, wind direction and
velocity).
2 Geological Setting of Fire Zone 8 (Wuda Coalfield)
The study was conducted in fire zone (FZ) 8 of Wuda Coalfield, located in Inner Mongolia
Autonomous Region. FZ 8 lies in the western part of the Wuda syncline. It is one of
Wuda’s largest fire zones, with seams no. 9 and 10 burning intensely. The fires have
caused massive landslides and up to 1 m wide cracks, resulting in extensive collapses of
overburden rock.
Wuda Coalfield represents an isolated outcrop of Permo-Carboniferous strata in the

northern part of the Helan Shan, a 300 km long mountain range at the border between
Ningxia and Inner Mongolia Autonomous Regions (Gielisch & Kahlen 2003). It is part of
a southward-dipping syncline with an extension of about 10 km in north-south direction
and up to 5 km in east-west direction (Gangopadhyay 2003). The syncline is confined to
the north and west by the flat basin of the Gobi Desert; to the east lie the wide valley of the
Yellow River and the Ordos Massive.
According to Diaz et al. (1983) as well as the IUGS 2000 report (cf. Jones 1995), the
Permo-Carboniferous coal strata belong entirely to the Permian. The Permian sedimentary
sequence (about 325 to 350 m in total thickness) is made up of clastic sediments, ranging
from claystone to conglomeratic sandstones, and bears 18 coal seams (numbering starts
with the stratigraphically youngest formations). The sequence belongs to the Lower
Permian Yanghogou (C2y) and Taiyuan (C3t) Formations, and Middle to Upper Permian
280
Shanxi (P1), and Shihexi (P2) Formations (Gielisch & Kahlen 2003); it displays the typical
cyclic sedimentation from coal to claystone to silt-/sandstone and back to claystone and
coal. Partly conglomeratic coarse sandstones and occasional conglomerates of two coal
seams were deposited during the sedimentary phase of each cycle.
The strata forming the lower part of the Upper Taiyuan Formation include seams no.
18 to 9. At FZ 8, the double seam nos. 9 and 10 is underlain by a thick basal sandstone
unit (16 m), and separated from seam no. 11 by a 3 m thick mudstone sequence.
Between seams nos. 9 and 10, there exists another mudstone layer with siderite
concretions. Seam no. 9 represents the boundary between the Lower Upper Taiyuan
(3t 2/1) and the Uppermost Taiyuan (3 t 2/2) Formations. It is overlain by an 8 m thick
sequence of siltstones with plant debris, interrupted by thin sandstone layers. Above
this sequence is an alternated bedding of sandstones (partly including bed drift wood
and iron concretions) and mudstones. The uppermost outcropping rock is made up of
sandstone (Gielisch & Kahlen 2003).
The strata generally slope down in easterly direction, with angles between 10° and 20° . Due
to massive land slides in the area, these angles might be a result of post-fire events and
do not necessarily represent natural bedding conditions (Gielisch & Kahlen 2003).
3 Combustion Gases of Wuda Coal Fires
Gases deriving from coal conversion under high temperatures (e.g. coke production,
coal gasification) are a result of different processes (e.g. Stillman 1979; Dikeç et al.
1994). Although most of the scientific data on these processes stem from industrial
gasification research, they can as well be used to explain the gases discharged by
spontaneous coal seam fires. A detailed discussion of the potential processes will be
published elsewhere.
As spontaneous coal seam fires basically liken an out-of-control underground

gasification reactor, it is not surprising that the same gas components as in laboratory-
scale coal gasification are found in vents and cracks at the surface above such fires.
Herman & McAteer (1999) measured combustion gases in boreholes close to the
Loveridge Mine Fire (Fairmont, West Virginia, U.S.A.) and found up to 18 % methane
(CH4) and 11,500 ppm hydrogen (H2). For an Illinois mine fire, they reported up to 7 %
carbon dioxide (CO2), 2.6 % CH4, 77.8 % nitrogen (N2), 0.93 % argon (Ar), 4,343 ppm
281
carbon monoxide (CO), and 211 ppm H2. Close to a mine fire at Somerset, Colorado (West
Elk Mine), they measured up to 2 % CH4 and 4,600 ppm H2.
Extensive measurements of coal combustion products were carried out in Wuda

Coalfield during 2004 (Schlömer 2005). Concentrations of combustion gases of
FZ 8’s main fire zone, which had been chosen as installation site for the permanent
monitoring station, were as follows: CO2 from 0.03 to 7.8 %, oxygen (O2) from 3.4 %
to 20.9 %, CO from 0 to 17,500 ppm, hydrogen sulfide (H2S) from 0 to 900 ppm, and
CH4 up to 9 %. These data suggest that there existed close to the main ventilation zone a
small area where unusually high CO and CH4 concentrations occurred (see also Litschke
et al. 2008 in this volume).
4 Permanent Monitoring Station:

Location, Initial Conditions, and Technical Design
FZ 8 was chosen jointly by DMT and BGR as the main measurement site because it
offered the most suitable conditions compared to FZs 3.2 and 11, which had also been
selected for in-depth investigations. At FZ 8 no active fire fighting with water/mud
injections and/or artificial sealing had taken place or was planned; the geology and
the geometry of the underground coal fire was fairly simple; and abundant geological
information was available from mapping and drilling. Also, extensive temperature and
chemical measurements had been carried out at earlier stages of the project. The long-
term gas and temperature monitoring station was set up through May and June 2005. In
proximity to this installation, DMT operated a permanent micro-acoustic survey, with
two seismometers in boreholes (20 and 10 m deep) and two at the surface (Figure 1).
At the end of October, after the DMT flooding experiments and the accompanying gas
and temperature measurements of our group, the gas monitoring unit was de-installed in
compliance with local regulations.
The system used was a modification of the standard BGR gas monitoring system, which
has been improved continuously over the last years. Special emphasis was placed on the
design and development of an adapter for reproducible sampling of hot (T ≤ 110 °C) gases
with a high humidity and corrosive sulfur species content. The system has been operated
successfully for extended periods of time (i.e. months and years), e.g. at Galeras Volcano
(Colombia) and at Nisyros Hydrothermal Vents (Greece); similar systems are still in use
for monitoring of earthquake precursors. Software for instrument control, data acquisition
282
with cycle times of seconds, data storage, and online telemetry have been developed
additionally.
The Wuda coal fire permanent multi-parameter station consisted of two units. The central
processing and data acquisition unit was based in a newly built shack; the remote (outdoor
installation) unit was located close to the main fire zone (c. 15 m).
The central unit consisted of a mains power supply, a control computer (PC/104), a data
logger, and installations for meteorological measurements. Power supply at remote locations
is usually secured by solar panels, but at the chosen location in FZ 8 electric power was
made available by Wuda Coal Mine. The system, operated at 24 V, was backed up by two
12 V serially connected constantly charged batteries, allowing for a maximum downtime of
the mains 220 V power supply of 10 hours. The data logger was a Gantner Instruments ISM
111 (4 analog inputs 16 bit ADC and 4 digital in/outputs) used for A/D conversion of signals
from the wind vane and wind anemometer, for ambient temperature, barometric pressure, and
battery voltage. The accuracy of the wind anemometer was ± 0.5 % (min. 0.6 m s-1), that of
the wind direction sensor 7° at 0.6 m s-1. Barometric pressure was measured with a Cerabar T
PMC 131 pressure sensor (long-term stability ± 0.15 % yr-1; accuracy ± 0.5 %). The control
computer was equipped with a GSM modem for daily data transmission to Germany (thus
providing for real-time monitoring), and special software to control all components from
remote locations. All data from the ISM 111 logger were stored on the computer’s hard disk.
The distance between the remote unit and the shack was about 70 m. The communication
lines running from and to the remote unit (Teflon® tubing for gas and control lines) were
placed in plastic tubes (5 cm in diameter) for protection against external damage. The
remote unit consisted of two Gantner IDL100 data loggers (8 analog inputs 16 bit ADC,
6 digital in/outputs, 256 KB internal RAM, and an 8 MB PCMCIA Flash memory card),
mass flowmeters (TSI, accuracy ± 2 %), miniature diaphragm air pumps (maximum
flow rate 1.2 l min -1), infrared gas sensors, and a thermocouple for measurement of
temperature in the equipment box. All loggers and flow rate regulators (2 channel D/
A converter) were connected to the control computer in the central unit via a RS 485
fieldbus interface. All sensors, loggers, flowmeters and pumps were contained in water-
and dust-protected cases (IP 67). Three different fracture locations were monitored by
the remote unit (Figure 1).
283
Figure 1: Location of the central unit (shack), the remote unit, and the three different
measurement sites in FZ 8 (projection UTM 48 S (WGS84))
4.1 Location 1
Location 1 was situated close to the main combustion zone, 20 m off the remote unit. The
Type K thermocouples (50 cm long, ceramic protective tube) were placed in two 15 cm
wide cracks, forming an almost perfect 90° angle with each other (hence TL1.N-S & TL1.
E-W; Figure 2). White precipitation of calcite or gypsum/anhydrite was observed at these
fractures, but no traces of elemental sulfur were found. The thermocouples were placed
in a depth of c. 50 cm, well protected from direct wind influence. They were connected
to the data logger by a thermocouple compensating line. In June, the temperature at TL1.
N-S was 350 °C, at TL1.E-W 400 °C (see below). About 3 m southward, temperatures
in excess of 750 °C were measured, and in another nearby large crack, only a few meters
284
from the surface, red-hot bedrock was visible to the naked eye. It must have been a result
of secondary combustion processes since the coal seam is much deeper at this location.
Figure 2: Thermocouples at location 1 installed in two north-south and east-west oriented

fractures, in close vicinity to the main combustion zone
4.2 Location 2
Location 2 represents a site of low-temperature gas discharge (75 °C; Figure 3); it is about
the same distance from the main combustion zone as location 3. A Teflon®-insulated Pt-100
thermocouple was installed at the beginning of the monitoring period (June 2005) at site
TL2.1, a second one on 18 October at site TL2.2.
285
Figure 3: Location 2 of low-temperature thermocouples (TL2.1 June to October 2005,

TL2.2 October 2005)
The thermocouples were inserted into fractures to a depth of about 75 cm. Location
TL2.1 was a small vent (several centimeters in diameter), but seems to have been part of
a large discontinuous fracture. Around this vent, typical condensation products (i.e. tar
and waxy hydrocarbons) were staining the surface. The vent gases at this site comprised
c. 7.7 % H2, 61 % N2, 25.5 % CO2, 1 % CO, 2 % CH4, 1000 ppm ethane (C2H6), plus traces
of higher hydrocarbons (saturated and unsaturated compounds). TL2.2 was part of a larger
and deeper fracture, which could be traced continuously to the main combustion zone.
Gas composition was not determined at this site. Although the thermocouple here was also
placed in a depth of c. 75 cm, the temperature profile measured from 19 to 26 October
showed a clear dependence on wind direction (Figure 4). This variation is not an indication
of changing combustion efficiency, but has to be considered a result of fresh air entering
the fracture vertically.
286
Figure 4: Temperature profile of TL2.1, TL2.2, and TL3.3 from 19 to 26 October 2005,
showing a well-defined correlation between wind direction and temperature for
TL2.2 (red). TL2.1 (green) and TL3.3 (light blue) were not influenced.
4.3 Location 3
At location 3 (Figure 5), gas composition (CO2, CO, CH4) and temperature were measured.
The emanating gas was diluted by a two-step system, consisting of a primary membrane
in the vent and a secondary exchanger/diluter – a modification of a patented system
by Poggenburg and Faber (2004). The combustion gases were pumped to a secondary
exchanger (Figure 5). After dilution, they were channeled to the infrared sensors. At
this stage, the steam content of the combustion gases had largely been removed through
condensation, and concentrations were well within the measuring range of the sensors. The
dilution factor was determined during installation at a value of about 5. All results shown
are uncorrected values as measured without due compensation for dilution. The exact
calibration is part of ongoing work.
287
Figure 5: Overview of location 3 with 3 thermocouples (TL3.1 and TL3.2 June to

October 2005, TL3.3 October 2005), tubing, and secondary exchanger
Pewatron Carbondio and Gas Module infrared sensors were used for the gas
measurements. The measuring range of the CO2 and CH4 sensors was 0 to 10 %, that
of the CO sensor 0 to 30,000 ppm (accuracy ± 2 % for all sensors). Originally, it was
also planned measure molecular H 2, with a semiconductor sensor. However, even
after dilution the sensor signal was still close to the maximum measuring range of
10,000 ppm, so that the H2 detector was eventually removed. The sensor showed cross-
sensitivity to other combustible gases (e.g. CH 4, C2H6), which evidently skewed the
288
results. The flow rates were controlled by two miniature diaphragm gas pumps and
constantly verified by mass flowmeters.
The sampling location was part of a long fracture extending from east to west, close to
the main combustion zone (c. 15 m). Temperatures varied strongly in this fracture. At
measuring site TL3.1, where gas concentrations were measured too, the initial temperature
in June 2005 was 78 °C. The gas composition (measured at standard P-T laboratory
conditions) at that time was c. 10 % H2, 59 % N2, 22 % CO2, 3.8 % CO, 3.2 % CH4, 1,400
ppm C2H6,, plus traces of higher hydrocarbons (saturated and unsaturated compounds).
At TL3.2 the initial temperature was much higher (150 °C), although this site belonged
to the same fracture and was very close to location TL3.1 (distance less than 1.5 m).
The different outlet temperatures were also reflected by different precipitation patterns
around the vents. The lower temperature vent TL3.1 showed more dark colored organic
condensation products with only minor amounts of sulfur, whereas TL3.2, as well as TL3.3
(temperature measurement only during the flooding experiments in October 2005), were
clearly dominated by sulfur-rich deposits (Figure 5).
In total, 15 different parameters were recorded at intervals of 30 sec. All data were stored
on the control computer; the loggings of the remote unit were additionally stored on the
two 8 MB PCMCIA Flash memory cards. The system was checked on a daily basis by
remote connection via GSM modem to download the data of the preceding 24 hours.
5.1 Temperature and gas measurements from June to October 2005
During the first days of the installation and monitoring period from 8 to 14 June, no
condensation of water was observed in the system, that is neither in the tubing nor in the
built-in water trap. However, as the water trap caused a certain lag time in gas measurements
(up to 10 min, depending on operating conditions), it was soon removed. After seven weeks
of smooth operation, the values of the infrared sensors at location 3 started to fluctuate (see
below). Inspection of the sensors on 17 October revealed that water had condensed in the
feed tubing and measuring cells. This may have been caused by simple condensation of water
vapor in the system due to day/night variance of ambient and measuring box temperatures
(usually 20 °C; not shown). Another explanation is the fouling of the gas transfer membrane
(see Figure 6). Although the membrane had been used successfully in volcanic environments,
289
a long-term application in combustion gas vents had never been tested before. The inspection
of the membrane, after operation for five months, revealed heavy fouling with hydrocarbons
of high molecular weight. This may have degraded the diluting efficiency of the membrane
and ultimately led to a drastic increase of water vapor in the measuring gas.
Figure 6: Example of the original gas transfer membrane on 14 June 2005 (left) and
after removal on 16 October (right). The membrane is heavily fouled with
condensates (It was torn from the Teflon tubing during removal from the vent.).
Due to a failure of the operating system on 25 August 2005 (hang-up during re-boot
sequence), the remote connection was lost and weather data were not recorded until
17 October 2005.
Temperature measurements showed remarkable fluctuations over relatively short periods

of time (Figures 7, 8, 10). These variations were more pronounced and more frequent
at locations TL1.N-S and TL1.E-W, where temperature drops in excess of 200 °C were
observed within one or two days (Temperatures, however, rose back to their original levels
nearly as fast following such drastic declines.). Temperature variations at locations TL2
and TL3 were less pronounced but still above the statistically significant level; they were
lower not only in terms of absolute values – which was to be expected – but also in terms
of relative changes. This might indicate that vents with higher temperature are affected
much stronger by dynamic combustion processes and turbulent gas flows than those with
lower temperature, where the flow regime is more stable due to the larger distance from
the main combustion zone.
Figure 7 illustrates typical temperature variations at location 1. At TL1.E-W, the

temperature dropped in two phases. The first decrease by around 100 °C (from 360 to
260 °C) occurred during c. 7 hours on 13 June, starting at 13:15; the second phase with a
decline of 120 °C (from 260 to 140 °C) started on 14 June at 15:00 and lasted for about six
hours. At the same time (14 June, 11:30), the temperature at location TL1.N-S increased
290
by 60 °C (from 340 to 400 °C) within 10 hours. At first glance, it appeared that these
temperature variations were related to emerging gale force winds (wind speed higher than
40 m s-1; Figure 7). However, a closer look at the data revealed that the drastic change in
wind conditions occurred not until 15 June, 14:30; it could thus not have influenced the
variations in temperature.
Figure 7: Temperature variations at location 1 (TL1.N-S, TL1.E-W) from 14 June to 24

July 2005 and corresponding meteorological data (wind velocity and direction,
barometric pressure)
At the same time (14 July, 18:47), similar changes in chemical composition of combustion
gases and temperature were observed at TL3.2 (Figure 8). Temperature dropped by roughly
40 °C (from 165 to 125 °C); concentration of CO decreased by 70 % (from 10,000 to
3,000 ppm1), CO2 by 60 % (from 3.74 to 1.4 %), and CH4 by 65 % (from 1.8 to 0.6 %)
within several minutes (see also Figure 9). Although there was a similar temperature
drop at location TL3.1 at that time, this is no proof of a connection to the same external
event, since this change was not significantly different from other temperature variations
commonly observed at the location.
1
Concentrations have not been corrected for dilution (dilution factor about 5). Actual values are proxies only.
291
Figure 8: Gas composition and temperature at location 3 (TL3.1, TL3.2), and temperature
at location 1 (TL1.N-S, TL1.E-W) from 21 June to 24 July 2005
Figure 9: Detailed view of data (gas composition and temperature) of 14 and 19 July,
illustrating two changes in gas composition at location 3 (TL3.1, TL3.2)
292
It is possible that the sharp decrease in gas concentrations is a first effect of the sensor
failure. However, two features argue against this assumption: the first is the synchronous
occurrence of other variations (e.g. temperature changes) on 14 July; the second is the real-
time history of the decrease. A detailed view of the relevant data of 14 and 19 June clearly
shows the difference: the sensor failure on 19 July is identified by an abrupt decrease
within one interval of digitization (30 sec), whereas the effect on 14 June was a continuous
decrease over several minutes (Figure 9).
Weather and temperature data were analyzed extensively, yet no clear relationship between
them could be established (with the exception of location TL2.2; cf. Figure 4). At times,
it seemed that temperatures varied with wind direction (see e.g. Figure 13), at others,
there was no correlation whatsoever. Several reasons might account for this observation
(Technical effects on temperature measurements can be excluded.): Most obviously, a
20 m deep subsurface coal fire represents a chaotic system, which is unaffected by wind
direction and velocity. Such a conclusion, though, is of course very simplistic, especially so
since the particular features of the coal fire under investigation (e.g. a large collapse zone
toward its south, with meter-wide, deep fractures) point to a certain causal relationship.
It seems more probable that the amount of local weather data at hand was not sufficient,
mainly due to the failure of the operation system after 10 weeks of operation. The data
were unable to reveal the potential relations between combustion processes (which are the
ultimate reason for the temperatures observed) and external effects such as wind direction
and velocity, or barometric pressure.
The temperature variations could also originate from rain entering through fractures.
Although the regional climate is commonly defined as semi-arid, heavy rainfalls (e.g.
thunderstorms) may occur at times. Unfortunately, we do not dispose of local or regional
precipitation data to confirm or deny this hypothesis. The only data available supporting
the influence of rainfall on temperature fluctuations are from Weather Online.2 For
Yinchuan (Ningxia Autonomous Region, c. 100 km south of Wuda), precipitation was
reported for five consecutive days from 17 to 21 July; and for Otog Qi (Inner Mongolia
Autonomous Region, c. 100 km east of Wuda) for four days from 18 to 21 July.
It is quite conceivable that the way in which a complex system such as a spontaneous
coal seam fire responds to external weather influences is rather slow and that other factors
mask potential effects. Nevertheless, it is important to obtain local, i.e. on-site data rather
than meteorological reports from far-off weather stations (e.g. several kilometers); this
correlation should be further investigated.
2
http://www.weatheronline.co.uk/Asia.htm; last visited in June 2006.
293
Of course, there are other parameters influencing the combustion process and, through it,
the measured temperature. In particular, unique phenomena such as collapsing roof walls
can result in abrupt modifications (e.g. jamming) of migration pathways for combustion
gases and thus disturb the instable equilibrium between subsurface combustion and surface
measurements. Normal fluctuations based on recurring changes in boundary conditions
(e.g. barometric pressure or other unidentified effects) can also cause variations. However,
a first FFT frequency and wavelet analysis of temperature data did not reveal any
systematic short- (i.e. daily or diurnal) or long-term oscillations. It is unlikely that other
factors like coal chemical composition or humidity are responsible for the observed short-
term fluctuations.
Besides these changes, long-term trends superimposing the temperature profiles were
also observed (Figure 10). Temperatures at location TL1 generally decreased with time.
At TL1.N-S, they decreased by around 110 °C (from 350 to 240 °C) with an interim rise
in mid July. The overall rate of temperature decline was c. -30 °C per month. A trend
for TL1.E-W, although more speculative because of more short-term fluctuations, was
detectable too; here, temperatures decreased gradually by 110 °C within five months
(rate: c. -20 °C per month). Temperature decreases at location TL2.1 were less distinct
but clearly identifiable (-10 °C in five months), despite short-term variations in mid
October.
Reverse trends were observed for gas temperatures at location 3. TL3.1 showed an
overall temperature increase by 35 °C (from 80 °C to 115 °C), occurring at two different
rates (Figure 10). At TL3.2, a more complex development was observed. Temperatures
here increased smoothly during the first ten weeks (+20 °C), then declined momentarily,
increased again sharply from mid September to mid October (+150 °C in c. six weeks),
and finally showed a very sharp decline during the last two weeks in October (-90 °C
in one week). Most likely, these trends are a result of the advancing fire front. The front
must move in a north-easterly rather than a northerly direction, as is evidenced by the
temperature trends at location 3: TL3.1 (the westernmost site) showed first a temperature
increase, then a decrease; TL3.2 (situated more to the east) showed a continuous increase
in temperature. These two developments have to be seen in the context of a slight decrease
at location 2. As the temperature changes were moderate, we conclude that fire propagation
in this area is relatively slow, much slower than the occasionally supposed rates in excess
of 50 m yr -1.
294
Figure 10: Observed long-term temperature variances at measuring locations. Numbers

indicate temperature decrease and increase of selected time intervals
The gas composition measurements are consistent with this assumption. Upon arrival
at the measuring site on 16 October, we replaced the defective CO and CO2 infrared
sensors and the primary membrane in the vent at location TL3.1. The composition of
the gas discharge was then measured continuously until 26 October for subsequent
correlation with the results from the DMT flooding experiments (see below).
Figure 11 shows the long-term variations of combustion gases for two time intervals. The
sinusoidal variations in the concentrations are a technical artifact (concentration varies
with the strong ambient temperature undulations during summer time). In particular,
it was impossible to protect the CO2 sensor from extreme temperature changes in the
remote measuring unit. Therefore, even the CO / CO2 ratio (ppm / %) varied strongly
with diurnal frequency. These fluctuations vanished in October when the differences in
day- and nighttime temperature became much smaller (Figure 11).
295
The major finding with regard to gas composition is the decrease of the CO / CO2 ratio
between June and October 2005. During June and July, the ratio went as high as 1,800,
even 1,900 (determined at the low points of the concentration profile). During the second
half of October, the ratio was significantly lower, around 1,400. These results were
confirmed by gas chromatographic measurements of gas samples in our laboratory, one
sample taken on 13 July, the other on 22 October (see Figure 11). All in all, the change
in chemical composition supports the assumption of an advancing fire front responsible
for the long-term temperature changes: with decreasing distance to the combustion zone,
the fraction of combustion gases increases (compared to the products of gasification
processes), hence more CO2 than CO.
Figure 11: Long-term change in gas composition at location 3 (TL3.1)
5.2 Effects of flooding experiments
During October 2005, DMT performed several flooding tests to induce microcracks
close to their micro-acoustic monitoring station. The overall objective was to verify
that crack development can be measured and quantified (in terms of location and
296
orientation). Three experiments were performed at several locations with different

amounts of water.
The first experiment was carried out close to location 1. 4.5 m3 of fresh water were
discharged from a truck through a fire-fighting hose into a large crack 5 m downhill of
the installed thermocouples TL1.N-S and TL1.E-W (Figure 12). The water flow lasted
for c. 75 min (from 10:30 to 11:45; local time). Temperature profiles of TL1.N-S and
TL1.E-W are presented in Figures 13 and 14. Temperature at TL1.N-S dropped from 310
to 270 °C in a smooth fall over two hours, while TL1.E-W showed a sharp decrease by
about 50 °C within 10 minutes. However, these temperature variations coincided with
a significant change in wind direction (from south to west) and wind velocity (from 2
to up to 15 m s-1; Figure 14). It cannot be ruled out that they were the result of these
meteorological changes rather than the flooding – an explanation even more probable
given that steam was only emitted downhill of the point of water injection, not uphill.
Figure 12: 1st flooding test at location 1 (23 October 2005; about 4.5 m³ of water injected).
See Figure 2 for a detailed view of thermocouple positions.
297
Figure 13: Temperature and wind data at location 1 (TL1.N-S, TL1.E-W) from 18 to 26
October 2005
Figure 14: Detailed view of temperature profiles during flooding experiments 1 and 2 at
location 1 (TL1.N-S, TL1.E-W)
298
The second flooding experiment was performed close to measuring location 3.

Approximately 3 m west of this location, 4.5 m3 of water were injected into a small
fracture, which appears to have been an extension of the main east-west rupture zone (see
Figures 5 and 15). Shortly after the flooding had begun (about 5 min), a massive steam
emission at TL3.1 was observed (Figure 16). This had a strong effect on gas composition
(Figure 17); CO and CO2 concentration decreased sharply as a consequence of the steam’s
diluting effect. However, the CO2 / CO ratio did not change (not shown).
Figure 15: 2nd flooding test at location 3 (23 October 2005; about 4.5 m³ of water injected)
299
Figure 16: Steam eruption at TL3.1, shortly after flooding started
The temperature effects at the three measuring locations are complex. Corresponding to
the change in gas composition, the temperature at location TL3.1 decreased sharply from
117 to 109 °C, then increased again for a few minutes, and from there constantly declined
until the very end of the flooding experiment. The temperature drop at TL3.1 at about
10:30 is most likely an effect of the changing wind conditions and not a result of the first
flooding experiment, which simply happened at too great a distance. The response of the
TL3.2 temperature logger (about 2 m east in the same fracture, and thus even more distant)
was delayed by around 25 min counting from the start of the experiment. Similar to TL3.1,
the temperature here increased shortly (< 5 °C) and then showed a smooth and constant
decline from 215 to 195 °C. Interestingly enough, the temperature at location TL3.3, just
installed then, almost immediately fell after the experiment had begun, despite the fact that
this location was the most distant from the point of water injection (about 5 m; Figure 5).
In summary, the effects of flooding on the temperatures measured were relatively small,
much smaller than expected (10 to 15 °C). They were unlikely a result of the changing
wind conditions (unless there would have been a lag time of at least 2 hours between wind
change and temperature response).
The third flooding experiment was performed close to injection location 1. Around 15
m3 of water were delivered by two trucks and discharged simultaneously into a large
fracture (0.5 m width; Figure 2) close to the main combustion zone, displaying red-hot
bedrock (not shown). Shortly after the start of the experiment, heavy steam burst out of
300
Figure 17: Gas composition and temperature variations in consequence of flooding near
monitoring location 3
the fracture. The temperature record of TL1.N-S showed a two-phase development. The
first phase, a decline from 295 to 260 °C, occurred before the flooding (Figure 18). This
decline was followed by a sharp increase in temperature. Both developments might be
a result of wind changes which occurred at the same time (Figure 18). The second and
ultimate drop in temperature at location TL1.N-S occurred shortly after the flooding with
concomitant steam eruption. The temperature decline had an exponential form until six
hours after the experiment. Then, temperatures dropped almost linearly for the next 15
hours until the thermocouple had to be removed. Temperatures at TL1.E-W showed a
more or less vertical drop by 50 °C (from 270 to 220 °C), followed by a short and less
significant increase. One hour after the water injection, temperatures started to decrease
almost linearly (Figure 18). Although the measuring period and the data obtained after
the third experiment are very limited (due to early de-installation), we conclude that the
third flooding experiment had a strong effect on the combustion processes at this part of
the coal fire. Presumably, the combustion ceased and the temperature decline reflects the
cooling of the overlying rock strata.
301
Figure 18: Temperature profiles of TL1.N-S and TL1.E-W during and after flooding
experiment 3
6 Summary
After extensive investigations of combustion gases in FZs 3-2 and 8 of Wuda Coal Mining
Area (Inner Mongolia Autonomous Region, P.R. China) in 2004, BGR and DMT jointly
decided on a site for a long-term gas and temperature monitoring station in the area (FZ 8).
Measurements of gas discharge were carried out from July to October 2005 and yielded a
considerable amount of data on combustion products (minute scale), gas temperature, and
metereological parameters (wind direction and velocity, barometric pressure, and ambient
temperature); they revealed short- and long-term changes in the dynamics of combustion
processes.
For gas measurements, a location with relatively low temperatures (< 100 °C) and
higher-than-average amounts of non-combustion products (H2, CO, CH4) was selected.
At this site, signs of the approaching fire front were expected to be best perceived. At
two additional sites, temperature measurements were conducted and meteorological
302
data recorded. The data from long-term surveillance of FZ 8 showed that temperatures
measured at gas-emanating vents were highly variable (short-term fluctuations); the same
applies to chemical composition of gas discharge. The interrelation of these variations
with meteorological conditions (barometric pressure, wind direction and velocity) proved
very complex and remains unresolved. Different rates of temperature decrease/increase at
distinct locations and changes in the CO / CO2 ratio, however, point to a slow progression
of the FZ 8 fire front in a northeasterly direction (< 10 m yr -1).
For a follow-up project (application pending), a similar but simplified continuous

monitoring station is planned. It is meant to help determine a baseline for gas emissions,
survey corrective measures (i.e. fire extinction campaigns), and enable an eventual transfer
to long-term, best permanent, and autonomously operating surveillance of extinguished
FZs. The main objective is to establish an early warning system for coal fire re-ignition in
the long run.
Acknowledgements
The Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction and Monitoring of Coal Fires in North China” is funded by the German Federal
Ministry of Education and Research (BMBF), Grant No. 0330490D. We gratefully
acknowledge the support given by Dr. Hartwig Gielisch and his colleagues of Deutsche
Montan Technologie GmbH (DMT) for the introduction to the local geology and during
various field trips. Dr. Ümüt Halik of Xinjiang University provided excellent and
irreplaceable assistance in all matters concerning organization in Wuda. We wish to express
our gratitude to the staff of Wuda Coal Mine, our Chinese colleagues from the Ministry
of Science and Technology of the People’s Republic of China (MoST), from the Beijing
Remote Sensing Corporation (BRSC), and Beijing Ouhualian Science and Technology Ltd.
(BOST), as well as to our driver and aide Mr. Xu. Last but not least, we are much obliged
to Mr. and Mrs. Liu who, under very harsh conditions, guarded our equipment at FZ 8
throughout the monitoring period and gave us a hand during the back-braking fieldwork.
303
References
Bräuer, K., Kämpf, H., Faber, E., Koch, U., Nitzsche, H., and Strauch, G. Gas and carbon isotope
composition of methane: Monitoring results in relation to the Vogtland-NW Bohemia
earthquake swarm period 2000. Geochemical Journal. (submitted)
Brune, S., Faber, E., Teschner, M., Poggenburg, J., and Hagendorf, J., 2003. CO2 deposits at Vorderrhön
area (Thuringia) – Gas migration from a deep reservoir to surface? In Proceedings of
the International Conference on Gas Geochemistry – ICGG7, Freiberg, Germany, 22-29
September 2003. Katlenburg-Lindau, Germany: Copernicus GmbH, 37-39.
Delisle, G., Teschner, M., Panahi, B., Iguliev, I., Aliev, C., and Faber, E., 2005. Continuous one-
year record of methane flux from the Dashgil mud volcano/Azerbaijan. In Proceedings of
the General Assembly European Geosciences Union, Vienna, Austria, 24-25 April 2005.
Geophysical Research Abstracts, Vol. VII, 06256.
Diaz, C.M., Wagner, R.H., Winkler Prins, C.F., and Granados, L.F., 1983. The Carboniferous of the world.
Vol. China, Korea, Japan, and S.E. Asia. IUGS Publications No. 16. Available only through
Commerciál General Del Libro y Material Didáctico, S.A., C/Reina No. 15, 28004 Madrid,
Spain.
Dickeç, F., Ateşok, G., and Arslan, C., 1994. Coking of coal. In Kural, O. (Ed.), Coal. Istanbul: Istanbul
Technical University, 309-335.
Faber, E., Móran, C., Poggenburg, J., Gárzon, G., and Teschner, M., 2003. Continuous gas monitoring at
Galeras volcano, Columbia: First evidence. Journal of Volcanology and Geothermal Research
125: 13-23.
Gangopadhyay, P.K., 2003. Coal fire detection and monitoring in Wuda, north China: A multi-spectral
and multi-sensor TIR approach. MSc thesis, International Institute for Geo-Information
Gielisch, H. and Kahlen, E., 2003. Geology of the Helan Shan. Field report. Project report for Work
Package 2210: General geological description of Helan Mountain Area with emphasis on coal
fire areas, as a contribution to the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North China.”
Deutsche Montan Technologie GmbH (DMT), Essen, Germany.
Herman, A. and McAteer, J.D., 1999. Estimating the rate of coal combustion in a mine fire. IR 1249. U.S.
Department of Labor, Mine Safety and Health Administration (MSHA), Arlington, Virginia,
U.S.A.
Jones, P.J., 1995. Carboniferous: Timescales 5. Report 1995/34. Australian Geological Survey
Organization (AGSO), Canberra, Australian Capital Territory, Australia.
Jones, V.T. and Thune, H.W., 1982. Surface detection of retort gases from an underground coal
gasification reactor in steeply dipping beds near Rawlins, Wyoming. SPE 10050. Society of
Petroleum Engineers (SPE), Houston, Texas, U.S.A.
Litschke, T., Wiegand, J., Schlömer, S., Gielisch, H., and Bandelow, F.K., 2008. Detailed mapping of coal
fires in combination with in-situ gas flow measurements to estimate mass flow balance and
fire development. In UNESCO Office Beijing (Ed.), Spontaneous coal seam fires: Mitigating
a global disaster. International Research for sustainable control and management. ERSEC
Ecological Book Series – 4. Beijing: Tsinghua University Press and Springer Verlag, 306-333.
(this volume)
304
Poggenburg, J. and Faber, E., 2004. Offenlegungsschrift DE 10240330A1. Deutsches Patent- und
Markenamt, Munich, Germany.
Onozaki, M., Watanabe, K., Hashimoto, T., Saegusa H., and Katayama, Y., 2006. Hydrogen production by
the partial oxidation and steam reforming of tar from hot coke oven gas. Fuel 85(2): 143-149.
Schlömer, S., 2005. Intermediate project report for Work Packages 3410 and 3420: Gas geochemistry
of combustion gases, as a contribution to the Sino-German Coal Fire Research Initiative
China.” Federal Institute for Geosciences and Natural Resources (BGR), Hanover, Germany.
Schlömer, S. et al. Combustion gases of natural coal fires in Wuda (Inner Mongolia Autonomous Region,
PRC). International Journal of Coal Geology. (to be submitted)
Schobert, H.H., 1987. Coal - The energy source of the past and the future. TP325 S34. American
Chemical Society (ACS), Washington, D.C., U.S.A.
Seidl, D. et al., 2003. The multi-parameter station at Galeras Volcano (Colombia) - Concept and
realization. Journal of Volcanology and Geothermal Research 125: 1-12.
Stillman, R., 1979. Simulating of a moving bed gasifier for a western coal. IBM Journal of Research and
Development 23(3): 240-250.
Stracher, G.B., 2004. Coal fires burning around the world: A global catastrophe. International Journal of
Coal Geology 59: 1-6.
Stracher, G.B. 2003. Coal mine fire – Gas and condensation products: Collection techniques for laboratory
analysis. Energeia 14(5): 1-6.
Stracher, G.B., 1995. The anthracite smokers of eastern Pennsylvania: PS stability diagram by TL
2(g)
analysis. Mathematical Geology 27(4): 499-511.
Teschner, M., Vougioukalakis, G.E., Faber, E., Poggenburg J., and Hatziyannis, G., 2005. Real time
monitoring of gas-geochemical parameters in Nisyros fumaroles. In Fytikas, M. and
Vougioukalakis, G. (Eds.), The South Aegean Volcanic Arc: Present knowledge and future
perspectives. Developments in Volcanology 7. Amsterdam: Elsevier, 247-254.
Teschner, M., Faber, E., Poggenburg, J., Vougioukalakis, G.E., and Hatziyannis, G. Real-time gas-
geochemical surveillance on Nisyros Island (Aegean Sea, Greece) – Installation and
continuous operation of a direct monitoring system in hydrothermal vents. Pure and Applied
Geophysics (PAGEOPH). (submitted)
Wessling, S., Litschke, T., Wiegand, J., Schlömer, S., and Kessels, W., 2008. Simulating dynamic
subsurface coal fires and its applications. In UNESCO Office Beijing (Ed.), Spontaneous coal
seam fires: Mitigating a global disaster. International Research for sustainable control and
management. ERSEC Ecological Book Series – 4. Beijing: Tsinghua University Press and
Springer Verlag, 513-541. (this volume)
Yang, L., Jie, L., and Li, Y., 2003. Clean coal technology – Study on the pilot project experiment of
underground coal gasification. Energy 28(14): 1445-1460.
305
Detailed Mapping of Coal Fires in
Combination with In-Situ Gas Flow
Measurements to Estimate Mass Flow
Balance and Fire Development
Tom Litschke1, Jens Wiegand1, Stefan Schlömer2,

Hartwig Gielisch1, and Friedrich-Karl Bandelow1
1
2
Abstract
Gas concentration and gas flow measurements, combined with detailed mapping of cracks and openings
were conducted in parts of fire zones 3-2 and 8 of Wuda Coalfield (Inner Mongolia Autonomous Region,
P.R. China), to estimate the mass flow balance of gases emitted from the coal fires in these two zones. The
measurements were conducted in two field campaigns in June and November 2004.
After mapping, fire-related openings were classified according to morphological characteristics, and
parallel gas concentration and gas flow measurements were carried out at several representative locations.
The gas flow on undisturbed soil and areas with porous surface was quantified with the flux chamber
method. Distinct openings were measured with a pitot-static tube in combination with a portable gas
analyzer (dynamic pressure method). The two test areas showed similar gas flow characteristics. A
comparison of the two campaigns in June and November corroborated the assumption that the velocity of
fire propagation in both cases must be significantly below 2.5 m per month.
The median values of gas flow of the different opening classes, as well as their frequency and extent,
were used to estimate the CO2 mass flow balance of the test areas, and ultimately of the entire coalfield.
According to this estimation, the fires of Wuda Coalfield emit 90 to 360 thousand t of CO2 and consume
about 70 to 270 thousand t of coal per year.
306
摘要
在中国内蒙古自治区乌达煤田8号火区和3-2火区，与火区局部裂隙和裂缝详细测绘同步，采用
气体浓度和流量测量方法评估了这两个火区由煤火喷出的气体中各种成分总量。现场测量分别在
2004年的6月和11月实施。
裂隙测绘后，确定了一类与煤火有关的裂缝，同时在几个有代表性的位置进行了气体成分和流量
测量。在未受影响的土壤和具有多孔渗水的地表，采用流量计进行了气体涌出量定量测量，对火
区明显裂缝处使用了空速静压测管测量，测管直接连接到便携式气体分析仪（动压法）。两个试
验区的两种测量方法比较显示了相似的气体流动特征，两次野外测试对比印证了我们的假设，
即：在两种情况下火区发展速度明显小于2.5 米/月。
火区中不同类型裂缝的气体涌出量均值，与其涌出频率和范围参数，可用于评估试验区的CO 2
释放总量，最终可估计整个煤田的CO 2释放总量。按照这种评估法，预计乌达煤田火区每年释
放9~36万吨的CO2，耗煤7~27万吨。
1 Introduction
Coal fires are a global phenomenon occurring in almost all coal-mining countries of the
world. They are caused by spontaneous (natural) combustion or wholly anthropogenic
in origin. The fires themselves, their emissions, and the drastic geological changes they
entail (e.g. land subsidence) are a hazard to the environment and the people living and
working in close proximity to coal fire areas; they also pose a serious economic threat.
Greenhouse gases (GHGs) emitted from coal fires have come to occupy a prominent place
in the discussion on global climate change. This study presents a ground-based approach
for estimation of coal fire emissions using a combination of gas concentration and gas flux
measurements as well as detailed mapping.
Until today, estimations of carbon dioxide (CO2) emissions from coal seam fires have
almost always been calculated on the bases of remote sensing data (Rosema et al. 1993).
These estimations use the observed surface temperature above the coal seam as a proxy
for conclusions on burning intensity and emission volumes. The ground-based approach,
on the other hand, extrapolates spot-sample measurements to a whole coalfield. As
both approaches involve uncertainties, a combination of them is likely to lead to higher
accuracy in results.
1.1 Global distribution of coal fires
The problem of spontaneous combustion and uncontrolled fires in coal seams and coal
heaps is not limited to China. It exists throughout the world, in most coal-producing
307
countries (Figure 1). The fires in Centralia, Pennsylvania, U.S.A. and in Jharia Coalfield,
India, as well as those in Ningxia, Xinjiang, and Inner Mongolia (Wuda) Autonomous
Regions, P.R. China are the best documented to date (Stracher & Taylor 2004).
Figure 1: Well investigated and documented coal fires worldwide
Data source: Prakash (2002) (modified)
1.2 Classification of coal fires
Coal fires can be classified in many ways. A very comprehensive classification by Zhang
(2004) is shown in Table 1. In the first class, it distinguishes between fires burning in situ,
i.e. in the coal seam (coal seam fires) and fires that ignite ex situ, i.e. after the coal has
been extracted (coal heap fires). Coal heap fires can be divided into coal stockpile fires and
coal waste fires, depending on the composition of the coal heap. Coal seam fires, on the
other hand, comprise naturally caused fires (natural coal fires) and fires related to mining
activities (coal mine fires). Additional features can be used for further differentiation.
For example, the position of a fire relative to the surface marks it as either surface or
subsurface fire. Another feature is the status of a fire – is it still active, or extinct, or only
dormant? Of interest, too, is whether an extinguished fire was burning recently or some
million years ago. The coal fires investigated in this study were mainly underground or
near-surface; they had been triggered presumably by mining activities.
308
Table 1: Classification of coal fires

First class Second class Further Attributes
Coal seam fire Natural coal fire

Surface/Subsurface or Underground
Coal mine fire
Active/Extinct/Dormant
Coal heap fire Coal waste fire
Paleo/Recent
Coal stockpile fire
Data source: Zhang (2004)
1.3 Study area
The two field campaigns of this study (June and November 2004) were conducted in the Wuda
syncline, a few kilometers west of the city of Wuda. The city lies about 20 km southeast of the
regional capital of Wuhai, in the southern part of Inner Mongolia Autonomous Region. Wuda
and the corresponding coal mining area are nestled between the Helan Shan (Helan Mountains)
to the south and the Ordos Massive to the east. These two roughly north-south striking
mountain ranges are divided by the Yellow River, which is northbound in this part of its course.
Wuda Coalfield is bordered to the north and west by the Gobi Desert. The whole syncline has
an extension of approximately 10 km from north to south and 3 to 5 km from east to west
(Gangopadhyay 2003).
Wuda and the regional capital of Wuhai were founded in 1976 as industrial bases for the
economic exploitation of the coal reserves in the region. Coal mining here started in 1958;
the first coal fires occurred as early as 1961. These fires were isolated and scattered over
the whole mining area; it was only between 1989 and 1995 that they connected gradually.
In 2001, the Beijing Remote Sensing Corporation (BRSC) was asked by the local coal
mining authority to investigate the spreading of coal fires in Wuda. Their first detailed
investigation report in 2002 concluded that 3.07 million m2 of the area were affected by
coal fires; this is 8.8 % of the Wuda syncline (Gangopadhyay 2003).
Wuda Coal Mining Area, indeed, hosts a good number of coal fires (cf. Figure 2). They
have been numbered by the local coal mining authority, and it seemed appropriate to retain
the numbering for this study. The term “fire zone (FZ)” refers to a self-contained coal fire
area, while a “test area” is the part of a FZ where measurements were conducted. Of the
three test areas, those in FZs 3-2 and 8 were given greatest attention; a test area in FZ 11
was later abandoned for safety reasons. FZ 11 is subject to intense land subsidence; even
roads there are riddled with cracks and sinkholes.
309
Figure 2: Overview of test areas and FZs in Wuda Coal Mining Area
1.4 Geological setting
It is common in Chinese stratigraphy to attribute the coal-bearing strata of Wuda Coalfield

to the Upper Carboniferous and Lower Permian. Martinez Diaz et al. (1983) and Jones
(1995), however, hold that all Carboniferous and Lower Permian coal-bearing strata in
China are part of the Cathaysia Floral Province, which developed following a transgression
in the Lower Permian. Therefore, all these strata should be classified as solely Permian
today. The sequences of the coalfield consist of clastic sediments ranging from mudstone
to conglomeratic sandstone, the latter holding a total number of 18 coal seams. The seams
are numbered chronologically from young to old; the oldest being no. 18, the youngest
no. 1. The sedimentary sequence belongs to the Yanghogou (C2y), Taiyuan (C3t), Shanxi
(P1), and Shihexi (P2) Formations and takes the shape of a syncline which is divided in
the center by a fault striking from north to south. The Yanghogou Formation at the outer
rim of the syncline was not investigated as it bears no coal (Gielisch & Kahlen 2003). The
vitrinite reflectance of coal from the area ranges between 0.83 and 1.36 %, with an average
reflectance of 0.91 %. Hence, Wuda coal can be classified as high volatile bituminous (A)
to medium volatile coal. The ash content is 8.22 to 27.21 wt. %, volatile material content
310
33.49 to 36.33 wt. % (of ash free material) (Kus 2005; see also Kus et al. 2008 in this
volume).
2 Methods
2.1 Crack mapping
For evaluation of CO2 mass flow balance, it was necessary to locate and map the crack
systems of the test areas. A map containing all cracks down to the smallest fissure
would hardly have been feasible and exceeded the time frame of the field campaigns.
Consequently, the task was narrowed down to a mapping of all cracks and openings with
gas emissions above ambient temperature (Figure 3). These were mapped with indication
of their general features such as length, width, strike, and position in a first move.
During this procedure, it became obvious that cracks and openings generally occurred
in systems of several tens of meters. Classification hence was based on morphological
characteristics. Large, one-directional openings were classified as “cracks.” Circular
openings were categorized by size and convective activity into bigger (typically more
than 10 cm in diameter) and highly convective “vents” and smaller (typically less than
10 cm in diameter), less convective “funnels.” Within the crack systems, areas of several
meters in diameter were observed where the soil was distinctly marked by increased
porosity and different coloring compared to surrounding soil. These areas were classified
as “sponges” owing to their porous properties. In many cases, circular openings were
observed within sponges, and in some cases, a separate class called “semi-funnels” was
created to account for their transitional state between sponges and funnels. The term
“undisturbed soil” refers to soil without signs of alteration in close vicinity to crack
systems (less than 10 m to the nearest active opening). The crack systems were labeled
according to FZ numbers, and given their own chronological number, separated from
the FZ by a period, i.e. “[number of FZ].[number of crack system].” The points at which
measurements were carried out were referenced by a number added to the crack system
label.
311
Figure 3: Overview of FZ 3-2 test area (solid line = above ambient temperature, dashed
line = ambient temperature) and detailed map of crack system no. 10 as
example
2.2 Gas flow measurements
Gas flow measurements were performed with the dynamic pressure method for cracks
and vents, and with the flux chamber method for the remainder of openings. During
measurements, gas temperature and velocity of air flow above the measurement points
were also recorded.
2.2.1 Flux chamber measurements

Gas flow measurements on undisturbed soil, sponges, and over very small openings
such as semi-funnels were carried out with a flux chamber. The chamber consisted of a
steel cylinder with an open bottom and two valves in its closed top lid. The valves were
connected to the gas in- and output tubes of a landfill gas analyzer. For measurement,
the chamber was gently pressed into the soil to prevent inflow of surrounding air. The
gas volume in the cylinder was pumped continuously into the analyzer and back into the
chamber. Measurements started with ambient air gas concentrations to ensure plausibility.
312
The concentrations of the chemical species in the combustion gas were logged at intervals
adapted to the different openings (e.g. 1 min for sponges, 30 min for undisturbed soil).
The mass flow density m A flux (kg h-1 m-2) of each measured species was calculated by
quantifying the rise in concentration in the chamber with the following equation (modified
after Bergfeld et al. 2001):
 V  ddc
c 
(1) m A flux =    ,
 A  dt
dt 
where V is the volume of the cylinder, A the soil area covered by the chamber, and
dc / dt the rate of change in concentration of the gaseous species in the chamber with
time.
2.2.2 Dynamic pressure measurements

Pitot-static tubes are generally made of welded high-grade steel, which makes them a perfect
tool for gas flow measurements in cracks over burning coal seams, where temperatures can
reach up to 400 °C. A pitot-static tube is composed of two concentric tubes. At its front
end, the outer tube is sealed. At this so-called stagnation port, gas flow is impeded; the
drag coefficient therefore is 1. The pressure exerted on the stagnation port is built up of the
dynamic pressure of the entering gas flow and the static ambient pressure (Figure 4). At
a certain distance from the stagnation port – sufficiently large to eliminate dynamic flow
effects – the pitot-static tube has a couple of small holes perforating the outer tube. These are
called static ports and are usually distributed evenly around the tube. The pressure exerted
on the static ports is the ambient atmospheric pressure only (Figure 4). The rear end of the
pitot-static tube bears two connections, one transferring the pressure applied to the outer
tube (pstatic) and the other that of the inner tube (pstag).
Figure 4: Pitot-static tube
Data source: Brock & Richardson (2001) (modified)
313
With Bernoulli’s equation, the relation between gas flow velocity and pressure difference
can be expressed as follows (Brock & Richardson 2001):
2∆p
(2) v= ,
ρ
where ν is the gas flow velocity, ρ the air density, and Δp the difference of pstag - pstatic. The
variable ρ can be substituted with p / RT, where p is the ambient air pressure, R the gas
constant for dry air (287 J kg-1 K-1), and T the gas temperature in K. This leads to the final
equation:
∆p
(3) v = 22RT
R
T .
p
As R is the gas constant for dry air, humidity will have an effect on gas flow velocity in
this equation. According to Brock & Richardson (2001), however, this effect is less than
1 %. Also, although the pitot-tube has to be placed upright into the gas flow, a tube will
typically tolerate up to 20° of misalignment.
The mass flow density m v ccalculated with the following equation, where ν is the
 A dynamic =was
gas flow velocity and c the concentration of any measured chemical species within the gas:
(4) m A dynamic = v c .
The pitot-static tube used in this study consisted of a straight, 1.5 m long, welded high-
grade steel tube and a precision pressure sensor, manufactured by testo AG, Germany. It
had an accuracy of ± 0.3 Pa and a measuring range of up to 100 Pa. The tube was attached
to an aluminum rack to eliminate measurement errors due to vibration.
2.3 In-situ concentration measurements of chemical species in gas

discharge
The concentration of emission gases in the test areas was measured with the landfill gas
analyzer GA 2000 by Geotechnical Instruments Ltd., United Kingdom. Methane (CH4) and
CO2 were detected via infrared absorption at a frequency of 3.41 and 4.29 µm. Oxygen
(O2) was measured with a newly developed galvanic cell with reduced cross-sensitivity
effects. Additionally, carbon monoxide (CO) and hydrogen sulfide (H2S) were analyzed
with electrochemical cells. The detection limit was 0.1 vol. % for CH4, CO2, and O2, and
1 ppm for CO and H2S (Sanderson 2001). Combustion gases in cracks and openings were
collected at the greatest depth accessible with a 1.5 m high-grade steel pipe. Before being
314
pumped into the landfill gas analyzer, the gases were channeled through a gas-washing
flask and a peltier gas cooler to prevent condensation of corrosive liquids (e.g. H2SO3) in
the interior of the instruments.
In order to quantify the influence of the gas-washing flask on gas composition and
concentration, several measurements with and without flask were carried out at the
beginning of the field campaign in June 2004 (Table 2). It was then found that H2S was
being dissolved in the process, so that its measurement was discarded for the remainder of
the campaign.
Table 2: Gas measurement results with and without gas-washing flask

Measurement Point Gas Composition Comment
CH4 CO2 O2 CO H2S

(vol. %) (vol. %) (vol. %) (ppm) (ppm)
3-2.10.1 0.0 14.4 6.6 82 11 without flask
3-2.10.1 0.0 14.3 6.7 80 … with flask
Corrosion of instruments did not pose a problem where gas composition analysis took
place in combination with flux chamber measurements, as temperatures here, i.e. in
sponges and undisturbed ground, were much lower. It was thus possible to measure gas
concentrations without gas-washing flask, which in turn meant that H2S was included for
results.
2.4 Temperature measurements
Temperature measurements were carried out with a thermocouple sensor of 1 m length,

manufactured by testo AG, Germany. The measuring range of the sensor was -200 to +900 °C,
its tolerance ± 1.5 °C. The sensor was placed directly into cracks and openings to ensure
undisturbed measurements.
315
3 Results
3.1 Crack mapping
Crack mapping was intended to give an overview of the spatial distribution of different
types of coal fire-related openings as well as to determine the location of possible gas flow
measurement points. The frequency and extent of the mapped openings was later combined
with the results of the gas flow measurements to estimate the CO2 mass flow balance for
each of the test areas as well as for the entire coalfield. Exact figures for the FZ 3-2 test
area are given in Table 3, for the FZ 8 test area in Table 4.
Table 3: Frequency and extent of openings in FZ 3-2 test area

Crack System Cracks Vents Funnels Semi-Funnels Sponges
No. (m²) (number) (number) (number) (m²)
2 0.1 4 5 0 1.6
3 0.4 1 5 0 0.0
9 0.2 13 6 10 9.2
10 0.6 1 5 9 8.6
11 1.6 5 12 21 22.4
12 0.0 3 0 2 7.1
13 2.5 7 2 4 23.2
Total 5.3 34 35 46 72.0
Table 4: Frequency and extent of openings in FZ 8 test area

Crack System Cracks Vents Funnels Semi-Funnels Sponges
No. (m²) (number) (number) (number) (m²)
1 4.1 0 0 0 7.9
2 3.7 8 0 0 0.0
3 0.0 1 0 5 6.3
4 0.5 1 0 0 0.0
5 1.0 0 0 0 21.2
6 1.5 3 3 0 43.6
7 0.0 0 1 20 80.1
Total 10.8 13 4 25 159.0
316
The results of the gas flow measurements for the two test areas and for the two field
campaigns are presented separately at the end of this chapter (Tables 5 to 8). The
measurement points were not entirely the same for the two field campaigns as crack
systems had changed in their visible features and some of the openings could not be
retrieved. The exact locations of the measurement points are given in Litschke (2005).
The area of gas flow passage is not given for sponges and undisturbed soil, as these do not
have distinct measurable openings with gas emissions as opposed to vents and funnels.
For similar reasons, no gas flow velocity is given. The flux chamber method produces
values of mass flow density directly, without intermediate data for gas flow velocity and
area of gas flow passage. Emissions at a few measurement points, especially from semi-
funnels with very small circular openings, were easier to analyze with the flux chamber
method although the dynamic pressure method should have been used there. Where such
deviations occurred, area of gas flow passage (but no gas flow velocity) is exceptionally
given in the tables, and the respective measurement point is marked by addition of “flux”
to its opening class (e.g. “vent flux”). Measurements with dynamic pressure method did
not produce values for gas concentration and mass flow density for H2S. This is because
the high temperatures at the openings investigated with this method necessitated the use of
a gas-washing flask, which, while protecting the gas analyzer from corrosive condensate,
also elutriated H2S.
In the dynamic pressure measurements, the gas flow velocities are 3 min mean values. They
were calculated from three separate 1 min means. In the FZ 3-2 test area, the maximum
standard deviation for these three measurements was 0.5 m s-1 for very low velocities, with
an average standard deviation of only 0.1 m s-1. The maximum standard deviation in the test
area of FZ 8 was lower, at 0.4 m s-1, with an average standard of 0.2 m s-1.
317
Table 5: Gas flow measurements in FZ 3-2 test area, June campaign

Gas Concentration Mass Flow Density
Measurement Class Temp. Area Gas Flow CH4 CO2 CO H2S CH4 CO2 CO H2S
Point (°C) (m2) (m s-1) (vol. %) (vol. %) (ppm) (ppm) (kg m-2 h-1) (kg m-2 h-1) (kg m-2 h-1) (kg m-2 h-1)
3-2.2.1 Funnel 183 0.018 1.2 ND 10.2 712 … ND 222.680 0.20460 …
3-2.2.2 Vent 392 0.004 7.3 ND 13.4 2100 … ND 1598.982 2.50587 …
3-2.2.4 Undist. 44 … … ND 0.9 11 ND ND 0.004 0.00001 ND
3-2.3.1 Funnel 259 0.005 1.7 ND 8.9 988 … ND 238.670 0.34968 …
3-2.3.2 Undist. 40 … … ND 1.1 5 ND ND 0.008 0.00001 ND
3-2.6.1 Sponge 149 … … ND 3.4 106 1 ND 0.224 0.00070 0.00001
3-2.6.2 Sponge 154 … … ND 4.0 182 4 ND 0.261 0.00119 0.00003
3-2.6.3 Sponge 167 … … ND 4.0 207 3 ND 0.253 0.00131 0.00002
3-2.6.4 Sponge 154 … … ND 2.8 160 1 ND 0.182 0.00104 0.00001
3-2.6.5 Sponge 203 … … ND 6.9 290 2 ND 0.403 0.00170 0.00001
3-2.9.1 Funnel 323 0.023 0.7 ND 8.5 389 … ND 105.067 0.05139 …
318
3-2.9.2 Undist. 53 … … ND 1.4 325 ND ND 0.007 0.00016 ND

3-2.9.3 Semi-funnel 51 0.005 … ND 5.2 130 ND ND 8.681 0.00217 ND
3-2.9.4 Semi-funnel 49 0.002 … ND 6.7 171 1 ND 28.830 0.00736 0.00004
3-2.10.1 Vent 245 0.049 1.7 ND 14.3 80 … ND 497.238 225.05188 …
3-2.10.2 Undist. 57 … … ND 2.4 103 ND ND 0.020 0.00009 ND
3.2.11.1 Crack 394 0.008 3.0 ND 8.2 81 … ND 395.420 622.06253 …
3-2.11.4 Sponge 184 … … ND 3.0 170 5 ND 0.183 0.00104 …
3-2.11.5 Sponge 178 … … ND 5.3 220 11 ND 0.327 0.00136 0.00003
3-2.11.6 Sponge 172 … … ND 2.7 111 8 ND 0.169 0.00069 0.00007
3-2.11.7 Sponge 281 … … ND 2.8 100 9 ND 0.422 0.00151 0.00005
3-2.11.8 Sponge 227 … … ND 7.9 240 22 ND 1.319 0.00401 0.00014
3-2.11.9 Sponge 202 … … ND 3.0 95 10 ND 0.527 0.00167 0.00037
3-2.11.10 Vent 331 0.004 1.3 ND 19.6 2317 … ND 447.784 100.40880 0.00018
3-2.12.1 Funnel flux 228 0.003 … ND 3.7 9 2 ND 3.707 0.00090 0.00020
3-2.13.1 Vent 225 0.007 1.6 ND 5.4 557 … ND 187.242 563.80517 …
3-2.13.2 Crack 311 0.036 0.7 ND 4.6 514 … ND 59.576 217.38634 …
Legend: ND = not detected
Table 6: Gas flow measurements in FZ 3-2 test area, November campaign
3-2.2.1 Funnel 153 0.018 1.0 ND 9.8 615 … ND 245.250 0.15391 …
3-2.2.2 Vent 350 0.004 6.5 ND 9.1 1197 … ND 1061.268 1.39598 …
Vent 397 0.004 7.4 ND 12.5 1450 … ND 1553.889 1.80251 …
Vent 371 0.004 7.0 ND 12.0 1309 … ND 1439.454 1.57002 …
3-2.2.3 Funnel 277 0.001 4.1 ND 10.6 981 … ND 875.454 0.81021 …
3-2.2.4 Undist. 25 … … ND 1.7 25 ND ND 0.016 0.00002 ND
3-2.3.1 Funnel 278 0.005 1.3 ND 7.5 380 … ND 199.523 0.10109 …
3-2.6.1 Sponge 92 … … ND 0.1 ND ND ND 0.008 ND ND
3-2.6.2 Sponge 147 … … ND 2.2 63 ND ND 0.446 0.00128 ND
3-2.6.3 Sponge 75 … … ND 0.8 12 ND ND 0.196 0.00029 ND
3-2.6.4 Sponge 192 … … ND 0.5 13 ND ND 0.092 0.00024 ND
3-2.6.5 Sponge 180 … … ND 2.2 53 ND ND 0.826 0.00199 ND
3-2.9.1 Funnel 352 0.023 1.0 ND 6.8 25 … ND 114.637 0.00421 …
3-2.9.2 Undist. 33 … … ND 0.6 98 ND ND 0.004 0.00007 ND
319


3.2.10.1 Crack 223 0.049 2.0 ND 8.2 1095 … ND 365.971 0.48870 …
3-2.10.2 Undist. 17 … … ND 0.4 46 ND ND 0.235 0.00270 ND
3-2.10.3 Crack 105 0.150 0.7 ND 6.1 275 … ND 119.914 0.05406 …
3-2.10.4 Funnel flux 36 0.002 … ND 2.7 357 ND ND 18.609 0.02461 ND
3-2.10.5 Funnel flux 39 0.008 … ND 5.2 76 3 ND 5.925 0.00087 0.00003
3-2.10.9 Undist. 17 … … ND 0.4 11 ND ND 0.004 0.00001 ND
3-2.11.1 Vent 385 0.008 2.0 ND 13.0 1305 … ND 439.986 0.44168 …
3-2.11.4 Sponge 130 … … ND 0.4 21 11 ND 0.085 0.00044 0.00023
3-2.11.6 Sponge 156 … … ND 0.4 13 ND ND 0.079 0.00026 ND
3-2.11.9 Sponge 308 … … ND 3.3 54 24 ND 1.452 0.00238 0.00106
3-2.13.1 Vent 249 0.007 1.6 0.2 12.9 2050 … 7 452.801 0.71957 …
3.2.13.2 Crack 249 0.036 1.9 ND 3.8 575 ND … 152.920 0.23139 …
Table 7: Gas flow measurements in FZ 8 test area, June campaign
8.0.1 Undist. 38 … … ND ND ND ND ND ND ND ND
8.0.4 Undist. 34 … … ND ND 4 ND ND ND 0.00001 ND
8.0.5 Undist. 46 … … ND 0.1 12 ND ND 0.001 0.00001 ND
8.1.1 Vent 55 0.006 0.4 3.1 20.8 > 2800 … 41.1 275.841 > 3.71324 …
8.1.2 Sponge 51 … … ND 0.8 525 8 ND 0.068 0.00446 0.00007
8.1.3 Vent flux 48 0.001 … 1.4 6.1 > 2800 54 49.9 217.254 > 0.99723 0.01923
8.1.4 Crack 48 1.827 0.4 1.7 10.0 > 2800 … 25.4 149.657 > 4.19039 …
8.1.5 Vent 407 0.008 2.2 ND 5.8 > 2800 … ND 205.049 > 9.89891 …
320
8.1.6 Vent 161 0.002 0.7 ND 5.6 > 2800 … ND 101.882 > 5.09412 …
8.2.1 Crack 243 0.866 1.2 ND 2.4 218 … ND 61.757 0.56096 …
8.4.1 Crack 71 0.010 0.4 10.1 21.2 > 2800 … 125.8 263.974 > 3.48645 …
8.4.2 Undist. 43 … … ND ND 6 ND ND ND 0.00001 ND
8.4.3 Vent flux 40 0.001 … ND 0.9 584 22 ND 18.491 0.11999 0.00452
Table 8: Gas flow measurements in FZ 8 test area, November campaign
8.0.1 Undist. 18 … … ND 0.1 ND ND ND 0.0004 ND ND
321
8.0.3 Undist. 16 … … ND 0.1 7 ND ND 0.0010 0.00001 ND

8.0.4 Undist. 24 … … ND 0.4 ND ND ND 0.0038 ND ND
8.0.5 Undist. 37 .. … ND 0.2 7 ND ND 0.0018 0.00001 ND
8.1.1 Vent 52 0.006 0.6 4.5 17.2 > 2800 … 96.1 367.2532 > 5.97854 …
8.1.2 Sponge 18 … … ND 0.7 570 ND ND 0.2025 0.01649 ND
8.1.3 Vent flux 34 0.001 … 1.2 9.3 > 2800 15 45.5 352.4313 > 1.06108 0.00568
8.1.4 Crack 43 1.827 0.8 0.5 5.0 > 2800 … 14 139.6729 > 7.82168 …
8.1.5 Vent 320 0.008 3.5 ND 8.8 1701 … ND 570.5600 11.02867 …
8.1.6 Vent 180 0.002 1.6 ND 5.8 1590 … ND 218.1622 5.98065 …
8.2.1 Crack 301 0.866 1.3 ND 3.7 193 … ND 92.8382 0.48426 …
8.2.2 Funnel flux 62 0.004 … ND 0.6 29 ND ND 3.8091 0.00184 ND
8.2.3 Funnel flux 57 0.002 … ND 0.7 257 ND ND 8.8319 0.03243 ND
8.4.1 Crack 61 0.010 1.3 1 13.4 > 2800 … 45.2 606.2670 >12.66827 …
8.4.2 Undist. 14 … … ND 0.3 9 ND ND 0.0020 0.00001 ND
8.4.3 Vent flux 62 0.001 … ND 3.3 532 42 ND 64.0119 0.10319 0.00815
4 Discussion
4.1.1 Comparison between test areas

The most striking difference between the test areas in the two FZs is the almost complete
lack of detectable concentrations of CH4 in FZ 3-2. Only in the November campaign
did measurement point 3-2.13.1 produce a trace above the detection limit (0.2 vol. %).
High concentrations of CH4 are generally associated with the early stages of coal fire
development, when the smoldering fire produces or emits considerable amounts of
hydrocarbons. The coal fire of the FZ 3-2 test area hence must be substantially older than
that of FZ 8. There are indeed records of fire fighting activities in the FZ dating back to
the 1970s (Jia 2004); and various trenches and landfills today bear witness to them. The
younger parts of FZ 3-2 seem to lie southeast of the investigated area (referred to as “uphill”
in the following). Additional measurement points toward this area would have been
necessary to confirm the assumption, but the difference in CH4 concentrations between the
two FZs may suffice to account for the existence of a difference in age.
A similar situation was found with regard to CO concentrations: A total of 10

measurements in FZ 8 exceeded the maximum detectable amount of 2,800 ppm, while
concentrations in the test area of FZ 3-2 never reached this value, neither in June nor
in November. Again this may be attributed to the fact that the early stages of coal fire
development are underrepresented in the investigated parts of FZ 3-2. Another difference
between the two test areas is in the gas flow through rock and soil above the coal seam.
While the test area of FZ 8, especially in June, showed no or very little detectable gas
flow through undisturbed soil, measurements in the test area of FZ 3-2 always produced
detectable concentrations of CO2 and CO. The fire fighting history of FZ 3-2 may provide
the key for this phenomenon: the soil layer spread over the active part of the test area in
an attempt to extinguish the fire has a greater porosity than the silt- and sandstone strata
above the FZ 8 fire. Also, the sandstone below the layer may have a greater porosity than
the siltstone in most of the investigated parts of FZ 8.
Apart from these differences, the test areas share the same classes of openings with similar
characteristics of gas flow.
322
4.1.2 Comparison between campaigns

In both test areas, cracks, vents, and funnels were more constant in terms of gas flow
velocity and gas composition than sponges and semi-funnels. This is consistent with
observations in the field, where sponges were noted to have changed in form and size
between the campaigns. Semi-funnels had sometimes disappeared or moved within a
sponge. Evidently, a coal fire tends to use its discrete emission channels over extended
periods of time, most possibly because the resistance to gas flow here is significantly lower
than in the surrounding bedrock. Sponges and semi-funnels, on the other hand, are more
readily shifted as they are loci of diffuse gas flow.
For similar reasons, the “older” test area in FZ 3-2 showed more constant values of
gas flow velocity, with the exception of 3-2.13.2. In combination with the unusual high
concentrations of CH4 at measurement point 3-2.13.1 in November, this exception suggests
that the western part of crack system no. 13 (near the “main combustion zone”) is still in
an early stage of coal fire development – which even has a certain impact on adjoining
areas of greater maturity (Figure 3). The remaining parts of the test area, except for the
“uphill” section in the southeast, did not show patterns of change in any direction.
The general state of inertia at FZ 3-2 may partly be due to small-scale displacements of
gas channels in the soil. They cause active openings to become dormant and vice versa,
and thus put off the gradual decline in combustion gas concentrations generally associated
with the passage from early combustion to maturity in coal fire development. Although
the measurements in the “uphill” part in November did not produce any values for CH4,
the geological setting (dip and strike of coal seams) suggests that the fire is moving in a
southeasterly direction. For instance, measurement points 3-2.11.1 and 3-2.10.1, which
lie just in front of the “uphill” area, showed a considerable increase in CO concentration
between the campaigns.
The results of gas flow measurements in the test area of FZ 8 show a more coherent
pattern of change between the two campaigns. This is certainly due to the fact that the
area under investigation included more parts in an early stage of fire development. While
gas flow velocities remained fairly constant over the two campaigns, gas concentrations
at most investigated openings declined from June to November. Except for no. 8.4.1 with
its implausible high concentration of CH4 in June, two out of three measurement points
showed a decrease in CH4 concentration. The values for CO in November were less or
equal to those in June. Measurement results at points 8.1.5 and 8.1.6, for example, were
significantly lower than the maximum measurable amount of 2,800 ppm, which they
323
had exceeded only 5 months earlier. Again, the findings for CO2 are less clear, although
some points showed the expected decline in gas concentration. Apparently, the period
between the two campaigns was too short (or the movement of the fire too slow) to
produce a significant decrease in all combustion gases and at all investigated openings.
Nonetheless, the results suggest that the fire is moving in a northwesterly direction. The
points situated in the northwestern corner of the test area were marked by rather constant
CO concentrations, while those in the southeast (8.1.5 and 8.1.6) showed slightly reduced
concentrations. Also remarkable is the increase in combustion gas concentrations in the
measurements on undisturbed soil. While in June, three of six investigated points did not
produce any sign of combustion gases, in November, no. 8.0.2 (with the greatest distance
to the thermally active crack systems) was the only measurement point where no gas
flow was detected. This and the fact that the relatively older test area of FZ 3-2 exhibited
detectable gas flow in this medium throughout the whole study period, is again proof of
heat-induced rock alterations during the aging process of a coal fire. Deformations of this
type eventually enable small-scale diffuse gas flow through undisturbed soil.
4.2 Mass flow balance
The combined results of crack mapping and gas flow measurements were used to estimate
the mass flow balance of the individual crack systems, the test areas, and the entire Wuda
Coalfield. For greatest accuracy, at least one spot-sample of each opening class was
measured per crack system and the mean value of the system calculated on the basis of the
frequency of the measured class. While the frequency of each class per crack system in this
example represents an exact value, mass flow in the following is a median of all measured
crack systems of the campaign.
Although the classification of cracks and openings was based exclusively on morphological
characteristics, the box-plots of CO2 mass flow densities for the different opening classes
(Figures 5 and 6) show considerable differences from class to class – a fact which must be
attributed mainly to variations in gas flow velocity. The values for “vents” and “funnels” of
the field campaign in June were poorly separated and treated as two discrete classes only
after investigations in November had confirmed the need for such a distinction (Figure 6).
“Cracks” were established as a separate class because mass flow here was calculated per
m2 and not per unit as for “vents,” “funnels,” and “semi-funnels.” The frequency/extent of
“sponges” and “undisturbed soil” was also measured in m2.
324
Figure 5: CO2 mass flow density of different opening classes, June campaign (central line =
median; box boundaries = 25 to 75 %, whiskers = 10 to 90 %, points = 5 to 95 %)
Figure 6: CO2 mass flow density of different opening classes, November campaign
(central line = median; box boundaries = 25 to 75 %, whiskers = 10 to 90 %,
points = 5 to 95 %)
325
The medians of CO2 mass flow density in the two campaigns were more or less of the same
order of magnitude, except for “funnels” and “undisturbed soil.” In the June campaign,
only a few specimens of “funnels” were measured, resulting in an overestimation of this
class. The value of the November campaign is more accurate because fourteen specimens
were measured. The difference in mass flow between June and November on “undisturbed
soil” is not a statistical error but the result of an increase in gas flow through the ground in
the test area of FZ 8, owing to the advance of the fire front as discussed in Chapter 4.1.
Estimation of mass flow balance based on crack mapping and gas flow measurements can
be enhanced by adding time as a distinctive parameter since gas concentrations change
over the different stages of coal fire development. Such an addition, however, requires
that the spatial distribution of these stages be investigated prior to gas flow measurements
to allow for placement of an adequate number of measurement points for each phase and
opening class. As the time for field work in the two campaigns was limited, and such an
approach would have indeed multiplied the necessary measurements by a factor of four
at least, the present estimations are based instead on the assumption that the decrease
in CO2 concentrations over the different stages is linear, and that the stages are evenly
distributed. This also explains why no information on CH4, mass flow can be given –
measurable amounts are only encountered during the early stages of fire development, and
measurements for each class in this phase were not sufficient.
An example for mass flow balance estimation is given for crack system no. 10 of the
FZ 3-2 test area. The medians of gas flow values of the different opening classes were
multiplied with the absolute quantity of specimens of these classes in the crack system. As
gas flow is likely to change over time, the medians were calculated separately for the two
campaigns (Tables 9 and 10). The area of undisturbed soil in this calculation is a buffer of
10 m width around the crack system, or around all active crack systems of a test area.
Table 9: CO2 mass flow balance of crack system no. 10 (FZ 3-2 test area), June campaign
Cracks Vents Funnels Semi- Sponges Undist. Soil Total
Funnels
Mass flow density 106a 1.66b 2.36b 0.0566b 0.257a 0.000333a
median
a
(kg h-1 m-2)
b
(kg h-1 unit-1)
Quantity 0.6c 1d 5d 9d 11c 1071c
c
(m2)
d
(unit)
Results 63.4 1.7 11.8 0.5 2.8 0.4 80.6
(kg h-1)
326
Table 10: CO2 mass flow balance of crack system no. 10 (FZ 3-2 test area), November
campaign
Cracks Vents Funnels Semi- Sponges Undist. Soil Total
Funnels
Mass flow 146a 3.36b 0.0368b 0.0781b 0.196a 0.00276a
density
median
a
(kg h-1 m-2)
b
(kg h-1 unit-1)
Quantity 0.6c 1d 5d 9d 11c 1071c
c
(m2)
d
(unit)
Results
87.6 3.4 0.2 0.7 2.7 3.0 97.1
(kg h-1)
4.2.1 CO2 mass flow of FZ 3-2 test area, June campaign

The CO2 mass flow of the FZ 3-2 test area in June reached its peak in crack systems nos.
11 and 13 (Table 11). The overall mass flow of all crack systems was roughly equal to that
of systems 11 and 13 combined. The mass flow of the entire test area was hence mainly a
product of “cracks,” although mass flow density (kg h-1 m2) in “vents” was slightly higher
(see Table 6). As the average diameter of a “vent” is only about 10 cm and “cracks” can
be several meters in length, the slightly lower mass flow density of the latter was offset in
mass flow balance. The overall mass flow of the test area was 729 kg h-1.
As said earlier, the mass flow of “funnels” was overestimated for the June campaign and
should at least be one order of magnitude lower than that of “vents.” However, considering
that the absolute value (82.7 kg h-1) is a mere 10 % of the overall mass flow of the test area
(729 kg h-1), this error is but minimal.
Table 11: CO2 mass flow (kg h-1) of FZ 3-2 test area, June campaign
Crack Cracks Vents Funnels Semi- Sponges Undist. Total
System No. Funnels Soil
2 14 6.6 11.8 0 0.4 0.2 33
3 42 1.7 11.8 0 0 0.4 56
9 21 21.6 14.2 0.6 2.4 0.8 61
10 63 1.7 11.8 0.5 2.2 0.4 80
11 165 8.3 28.4 1.2 5.7 0.9 209
12 0 5 0 0.1 1.8 0.5 7
13 259 11.6 4.7 0.2 5.9 0.6 282
Total 564 56.4 82.7 2.6 18.5 4.6 729
327
4.2.2 CO2 mass flow of FZ 3-2 test area, November campaign

The CO2 mass flow of the FZ 3-2 test area in November did not show much difference
compared to June (Table 12). Again crack systems nos. 11 and 13 contributed the greatest
portion to the overall mass flow, and “cracks” had by far the highest values. The mass flow
of “funnels” in November was more regular than in June, amounting to only 1.3 kg h-1 for
the whole test area.
Table 12: CO2 mass flow (kg h-1) of FZ 3-2 test area, November campaign
2 19 13.4 0.2 0 0.3 1.6 35
3 59 3.4 0.2 0 0 3.6 66
9 29 43.7 0.2 0.8 1.8 6.3 82
10 88 3.4 0.2 0.7 1.7 3 97
11 228 16.8 0.4 1.6 4.4 7.8 259
12 0 10.1 0 0.2 1.4 4.2 16
13 358 23.5 0.1 0.3 4.5 5.3 392
Total 781 114.2 1.3 3.6 14.1 38.2 953
The overall CO2 mass flow of 953 kg h-1 in the test area in November was higher than in
June, because of an increase in mass flow in “cracks” by roughly 50 %. The variation in
“funnels” and “undisturbed soil,” in contrast, did not have any significant effect on the
overall mass flow; they emitted only about 40 kg h-1 altogether.
4.2.3 CO2 mass flow of FZ 8 test area, June campaign

The FZ 8 test area is similar to that of FZ 3-2 in that only two crack systems (nos. 1 and 2)
produced the bulk of mass flow. The remainder contributed only as little as one of the two
former alone (Table 13). Also, “cracks” here provided the greatest portion to the overall
CO2 mass flow too. The absolute amount of CO2 mass flow (1,214 kg h-1) was higher than
that of the FZ 3-2 test area.
328
Table 13: CO2 mass flow (kg h-1) of FZ 8 test area, June campaign
1 436 0 0 0 2 0.7 439
2 391 13.3 0 0 0 0.3 405
3 0 1.7 0 0.3 1.6 0.4 4
4 48 1.7 0 0 0 0.2 49
5 106 0 0 0 5.4 0.2 111
6 159 5 7.1 0 11.2 0.2 182
7 0 0 2.4 1.1 20.6 0.4 24
Total 1139 21.6 9.5 1.4 40.8 1.8 1214
4.2.4 CO2 mass flow of FZ 8 test area, November campaign

The CO 2 mass flow balance of the FZ 8 test area in November exhibited the same
characteristics as in June (Table 14). As with FZ 3-2 in November, the mass flow of “cracks”
here increased by about 50 %, reaching the highest value of the two test areas in both
campaigns with 1,668 kg h-1.
Table 14: CO2 mass flow (kg h-1) of FZ 8 test area, November campaign
1 603 0 0 0 1.5 5.5 610
2 541 26.9 0 0 0 2.6 571
3 0 3.4 0 0.39 1.2 2.9 8
4 66 3.4 0 0 0 1.4 71
5 146 0 0 0 4.1 1.3 152
6 219 10.1 0.11 0 8.5 1.6 240
7 0 0 0.04 1.56 15.7 3 20
Total 1576 43.7 0.15 1.95 31.1 14.7 1668
4.2.5 CO2 mass flow of coal fires in Wuda Coalfield

The CO2 mass flow balance of the entire Wuda Coalfield can only be a rough estimation, as
the mass flow balances of the test areas, which serve as the basis for calculation, are only
approximations themselves. Nevertheless, the overall mass flow balance is important to
evaluate the impact of coal fires in Wuda, even if only the order of magnitude is accurate.
As both test areas represent hot coal fires (mainly > 150 °C) and their results are applicable
to such fires only, a map of hot coal fires by Gielisch & Künzer (2003) (Figure 7) was used
329
for extrapolation of mass flow balance to the whole coalfield (Table 15). The investigated
test areas in FZs 3-2 and 8 cover an area of about 19,000 m2 and 11,000 m2, respectively;
the hot coal fires of the entire coalfield extend over 270,000 m2 in total. It is very probable
that hot coal fires contribute the biggest share to the overall mass flow, which gives this
approach further credence.
Figure 7: Map of hot coal fires in Wuda Coalfield

Data source: Gielisch & Künzer (2003) (modified)
330
Table 15: CO2 mass flow balance of hot coal fires in Wuda Coalfield
Test Areas Only Hot Coal Fires (270,000 m²)
CO2 CO2 CO2 CO2 Coal

(kg h-1) (kg h-1 m-2) (t h-1) (kt yr-1) (kt yr-1)
FZ 3-2 test area June 729 0.04 10 91 69

(19,000 m²) Nov. 953 0.05 14 119 90
FZ 8 test area June 1214 0.11 30 261 199
(11,000 m²) Nov. 1668 0.15 41 359 273
The conversion from CO2 to “coal” was made according to the reaction equation and
measured coal carbon content of 53.82 % presented by Warnecke (2004):
(1) 16C + 12.8824H +0.3111H2O + 16.1457O2 →

10.6498CO2 + 5.0836CO + 0.2668CH4 + 6.2188H2O .
The resulting coefficient for the conversion from CO2 to “coal” is 0.761. According to this
calculation and the results from mass flow balance estimations in this study, the coal fires
of the Wuda syncline emit approximately 90 to 360 thousand t of CO2 each year, equaling
70 to 270 thousand t of coal burned.
5 Conclusions
A comparison between the test areas in FZs 3-2 and 8 of Wuda Coalfield showed that both
share the same classes of cracks and openings, as well as some general characteristics of
gas flow. It seems thus justified to apply the findings and results from these investigations
to other localities and coal fires.
The comparison of the two campaigns from June and November 2004 allowed statements
about the velocity of coal fire propagation in the two test areas. While the combustion
processes in FZ 3-2 were complex and did not yield sufficient data for definite conclusions
on coal fire development, the results from the FZ 8 test area clearly point toward a fire
progression in northwesterly direction. The main combustion zone here is proceeding
from measurement point no. 8.1.5 toward 8.1.4. As these two points are only about 10 m
apart, and the development between June and November revealed only slight changes in
gas composition, the velocity of coal fire propagation in this test area must be significantly
below 10 m per four months or 2.5 m per month.
331
The detailed mapping of the test areas in combination with in-situ gas flow measurements
provided a basis for an estimation of the CO2 mass flow of the entire coalfield. According
to this estimation, the annual CO2 emissions from the Wuda syncline amount to 90 to
360 thousand t yr -1. Although this figure itself appears to be rather small, for example
compared to the 110 million t of CO2 emitted each year by the German traffic, one has to
bear in mind that it only represents the Wuda coal fires – there are about one hundred such
fires in China (Prakash 2002).
The calculated amount of CO 2 indicates an annual coal consumption of about 70

to 270 thousand t yr -1. This estimation is in line with figures issued by the coal mining
authority in Wuda, which set the amount of burned coal at about 200 thousand t yr -1
(Künzer 2003).
References
Bergfeld, D., Goff, F., and Janik, C.J., 2001. Elevated carbon dioxide flux at the Dixie Valley geothermal
field, Nevada. Relations between surface phenomena and the geothermal reservoir. Chemical
Geology 177: 43-66.
Brock, F.V. and Richardson, S.J., 2001. Meteorological measurement systems. Oxford: Oxford University Press.
Gangopadhyay, P.K., 2003. Coal fire detection and monitoring in Wuda, north China: A multi-spectral
and multi-sensor TIR approach. MSc thesis, International Institute for Geo-Information
Gielisch, H. and Künzer, C., 2003. Cracks and crack systems of the test areas. Project report for Work
Package 2200: Structural pattern of the coal deposits (Wuda, Rujigou, Gulaben); analysis of
linear pattern (air photos, satellite images), as a contribution to the Sino-German Coal Fire
Coal Fires in North China.” Deutsche Montan Technologie GmbH (DMT), Essen, Germany
and German Remote Sensing Data Center (DFD) of the German Aerospace Center (DLR),
Wessling, Germany.
Jia, Y.R., 2004. Personal interview / Oral communication during the field campaign in November 2004.
(Jia Yaorong is the chief engineer of the Wuda Coal Fire Fighting Team.)
Jones, P.J., 1995. Carboniferous: Timescales 5. Report 1995/34. Australian Geological Survey
Organization (AGSO), Canberra, Australian Capital Territory, Australia.
Künzer, C., 2003. General description of broader study area. Project Report for Work Package 2110, as
a contribution to the Sino-German Coal Fire Research Initiative “Innovative Technologies
332
for Exploration, Extinction, and Monitoring of Coal Fires in North China.” German Remote
Sensing Data Center (DFD) of the German Aerospace Center (DLR), Wessling, Germany.
Kus, J., 2005. Personal interview / Oral communication about the preliminary results of micro-
petrographical analyses in the context of the Sino-German Coal Fire Research Initiative
China” via Hartmut Gielisch. (Jolanta Kus of the Federal Institute for Geosciences and
Natural Resources (BGR), Hanover, Germany measured vitrinite reflectance; H. Gielisch is
the coordinator of the initiative at the Deutsche Montan Technologie GmbH (DMT), Essen,
Germany.)
Kus, J., Hiltmann, W., and Balke, A., 2008. Researching coal spontaneous combustion: Micropetrography
of coal oxidation and carbonization. In UNESCO Office Beijing (Ed.), Spontaneous coal
Litschke, T., 2005. Detailed mapping of coal fire sites in combination with in-situ flux measurements of
combustion gases to estimate gas flow balance and fire development (Wuda Coalfield, Inner
Mongolia Autonomous Region). Dipl. thesis, University of Duisburg-Essen, Essen, Germany.
Martinez Diaz, C., Wagner, R.H., Winkler-Prins, C.F., and Cranados, L.F., 1983. The Carboniferous
of the world. Vol. I: China, Korea, Japan, and S.E. Asia. IUGS Publication Series. Instituto
Geológico y Minero de España, Madrid, Spain.
Prakash, A., 2002. Coal fires - A natural or man-made hazard? 22 October 2006 http://www.gi.alaska.
edu/~prakash/coalfires/introduction.html
Rosema, A., van Genderen, J.L., and Schalke, H.J.W.G., 1993. Environmental monitoring of coal fires in
north China. BCRS 93-29, Project Identification Mission Report. International Institute for
Geo-Information Science and Earth Observation (ITC), Enschede, Netherlands.
Sanderson, A., 2001. GA2000 gas analyzer general operating manual. Geotechnical Instruments (UK)
Ltd., Leamington Spa, United Kingdom.
Warnecke, C., 2004. Qualitative und quantitative Charakterisierung von Brandgasen aus Schwelbränden
mittels FT-IR-Spektroskopie und Ableitung der hauptsächlichen Reaktionspfade. Dipl. thesis,
Clausthal University of Technology, Clausthal, Germany.
Zhang, Z.J., 2004. Spatial and statistical analysis of thermal satellite imagery for extraction of coal fire-
related anomalies. PhD thesis, Vienna University of Technology, Vienna, Austria.
333
Part V
Remote Sensing
and Monitoring
Geophysical Investigation of Wuda Coal
Mining Area, Inner Mongolia:
Electromagnetics and Magnetics
for Coal Fire Detection
Gerlinde Schaumann1, Bernhard Siemon1, and Yu Changchun2
1
2
China Aero Geophysical Survey and Remote Sensing Center for Land and Resources (AGRS),
Beijing, P.R. China
Abstract
As part of the Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China,” the China Aero Geophysical Survey and
Remote Sensing Center for Land and Resources (AGRS), assisted by the German Federal Institute for
Geosciences and Natural Resources (BGR), carried out in 2004 an airborne electromagnetic and magnetic
survey of Wuda Coal Mining Area (Inner Mongolia Autonomous Region, P.R. China). In the following
year, ground geophysical surveys using transient electromagnetics (BGR) and magnetics (AGRS and
Deutsche Montan Technologie GmbH, DMT) were conducted in selected parts of the area. The objective
of these geophysical surveys was to detect and delineate coal fire areas through physical parameters
obtained over burning and burned coal seams. The electromagnetic surveys served to reveal areas of high
electrical conductivity, the magnetic surveys those with clearly pronounced magnetic field intensity.
摘要
本研究是中德研究项目“中国北方煤火探测、灭火与监测新技术”的一部分。2004年，中国国土
资源航空物探遥感中心（AGRS）在德国联邦地球科学和自然资源研究院（BGR）的协助下在内
蒙乌达矿区进行了航空电磁和航空磁法测量工作。2005年，选择该矿区部分地段分别由BGR采用
瞬变电磁方法与AGRS和DMT采用磁法进行了地面地球物理测量工作。其探测目的是基于正在燃
烧和烧过的煤层上获得的物理参数探测和圈定火区。电磁探测方法旨在揭示低电阻率区，而磁法
测量用于圈定具有显著磁场强度区。
336
1 Introduction
Within the framework of the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North China”
– a project launched in 2002 and funded by the German Federal Ministry of Education
and Research (BMBF) – several Chinese and German research teams from various
geo-scientific backgrounds have investigated Wuda Coal Mining Area in the Chinese
autonomous region of Inner Mongolia (Figure 1). In 2004, the German Federal Institute for
Geosciences and Natural Resources (BGR) assisted the China Aero Geophysical Survey
and Remote Sensing Center for Land and Resources (AGRS) in conducting a helicopter
survey of the area. This first survey was followed by several ground geophysical surveys
in 2005. Wuda Coal Mining Area is covered mainly by sandstones, under which 18 mined
coal seams extend to greater depths, varying from a few meters to several hundreds of
meters below surface.
Figure 1: Location of Wuda (Inner Mongolia Autonomous Region, P.R. China)
Coal combustion under electromagnetic and magnetic aspects has been researched
for decades. In his 1977 and 1983 studies, Duba described the changes in electrical
conductivity (or resistivity for that matter, its reciprocal) of coal samples and their
pyrolysis products he had observed in laboratory experiments. The heating of water-
saturated coal samples increased their conductivity from an initial value of 10-3 Sm-1
(resistivity = 1,000 Ohm*m) at 24 °C to 100 Sm-1 (0.01 Ohm*m) when recovered as char
from pyrolysis at temperatures of 800 °C. After a strong decrease owing to water loss from
337
drying in the temperature range from 24 to 110 °C, the conductivity curve flattened while
temperatures rose continually. At 300 °C, it began to rise, first slowly, and then rapidly
after 515 °C. This large increase in conductivity was attributed to the higher carbon content
of char. Based on his experiments, Duba (1977) suggested to use electrical conductivity
of coal and its pyrolysis products as a means of locating and identifying areas of different
physical properties underground. Because pyrolysis induces enormous changes in coal
conductivity, and is in fact one of the main processes in coal fires, Duba reasoned, it should
be possible to explore underground reaction zones by measuring conductivities off the
surface.
Powell and Schofield (1939) investigated electrical conductivity of carbon and graphite
at high temperatures. They found that electrical conductivity increased with thermal
conductivity and that graphitization had the same effect.
Bartel (1982) described CSAMT (Controlled Source AudioMagnetoTelluric) measurements

over coal fires in abandoned mines. In all surveys, resistivity anomalies were observed
which could clearly be associated with fires. In laboratory experiments, Bartel was able to
show that coal undergoes drastic electrical resistivity changes during combustion. Given
these changes, Bartel argued, electrical and electromagnetic techniques were suitable for
coal fire mapping and monitoring.
King (1987) performed electromagnetic measurements over burned coal seams in

Australia. He found significantly lower resistivities for coal heated to 800 °C and above,
and inferred that burned coal seams must be detectable by resistivity lows on the sounding
curves.
Hooper (1987) investigated magnetic properties of baked rock in Wyoming. These

properties proved quite distinct from those of surrounding sedimentary rock, so he
suggested magnetic surveys be used to locate baked rock. Hooper reported that baked
siltstone had a higher magnetic susceptibility than baked sandstone and shale of similar
iron oxide content and thermal alteration history. While being generally low, magnetic
susceptibility in the baked rock samples showed a rather wide range of values. In most
rocks, magnetic susceptibility is proportional to magnetite content, i.e. high magnetic
susceptibility can be correlated with high magnetite content. Except for some iron oxides
already present before baking, most of the hematites and magnetites in baked rock were
derived from thermal alteration of sedimentary minerals (Hooper 1987).
338
Sternberg and Lippincott (2004) presented magnetic surveys over clinkers and coal seam
fires in the United States. They measured enhanced magnetic susceptibility in sedimentary
rock above burning coal seams, including clinker formations.
Mainly drawing from insights of previous international research as described above,

the BGR/AGRS teams went to investigate the changes in electrical conductivity and
local anomalies of magnetic field intensity of areas related to coal seam fires in Wuda.
The ground surveys were conducted in selected parts of the airborne survey area.
Numerous ground electromagnetic measurements, especially transient electromagnetics
(TEM), were carried out by BGR across several fire zones (FZs) and in adjacent
unaffected areas. A ground magnetic survey by AGRS covered most of FZ 8; it was
later supplemented by the Deutsche Montan Technologie GmbH (DMT). Satellite-
supported maps of the burning areas provided by the German Aerospace Center (DLR)
were used for completion of the survey (Voigt et al. 2004; Künzer et al. 2005). Further
investigations with respect to geological, micropetrographical, thermal, and geochemical
characteristics of Wuda Coal Mining Area are presented by other authors in this volume
(Kus et al. 2008; Schlömer et al. 2008).
2 Geophysical Surveys
The airborne electromagnetic and magnetic surveys of Wuda Coal Mining Area were conducted
in August and September 2004. The electromagnetic system, an IMPULSE 2 developed by
Aeroquest, Canada, used two coil configurations (coplanar and coaxial) at a coil separation
of 6.5 m, each operating at three different frequencies between 870 and 23,250 Hz. The total
magnetic field was measured simultaneously using a CS-3 Cesium sensor by Scintrex, Canada.
The output sampling rate was 30 Hz for electromagnetic signals and 10 Hz for magnetic
signals, resulting in a sampling distance of c. 1.2 m and 3.8 m, respectively, at an average flight
velocity of 140 km h-1. The size of the survey area was c. 120 km2. About 300 survey lines
were flown at a line distance of 50 m from east to west and 250 m for north to south.
In May and June 2005, numerous ground electromagnetic measurements followed across
the FZs and in adjacent unaffected areas. While TEM was the method of choice; a few
survey profiles were measured using other electromagnetic techniques, e.g. Horizontal-
Loop (HL) with an Apex MaxMin. The results of these measurements are not presented in
this paper.
339
TEM uses transmitter and receiver coils of different sizes and configurations. Currents are
shot off intermittently in the transmitter coil so as to change the corresponding magnetic
field. This way, eddy currents are induced in the soil, where they cause a secondary
magnetic field, whose change induces a decaying voltage in the receiving coil. This
decaying voltage, also called the transient, is recorded and gives information about the
resistivity distribution in the ground. The effective exploration depth depends on the local
noise level (King 1987; Greinwald & Schaumann 1997).
Compared to HEM, the TEM system offers the advantage of variable coil size and
extension of recording time, which allows measurements at greater depths. Near-surface
depth resolution is less clear. To investigate subsurface resistivity distribution from near-
surface to depths of several hundred meters it is best to combine both methods.
TEM measurements were conducted across FZs and at single locations of a FZ. A number
of 65 transient electromagnetic soundings on 12 profile lines were performed using the
Geonics PROTEM47 system. Those across the well investigated FZ 8 (Figure 2) – one
of the 20 or so FZs in Wuda Coal Mining Area – are discussed in this paper. The spacing
of the TEM sites was adapted to the terrain. A loop size of 50 m × 50 m was chosen
for most sites in inloop configuration, where the receiver coil is placed in the center of
the transmitter loop. Models for each site were created from the data obtained without
a-priori information, using an iterative Marquardt algorithm (Weidelt 1984) to determine
resistivities (Rho) and depths of model layers.
AGRS carried out its ground magnetic survey of FZ 8 in May 2005. It was supplemented
by DMT in October of the same year (Elsen 2006). Survey lines in selected parts of the
area were narrowed or extended to places of special interest.
3 Investigation Results
Figure 2 shows a satellite image of Wuda Coal Mining Area. The dotted red polygon
marks the area of the helicopter survey, the continuous red rectangles mark the FZ study
areas; FZ 8 is marked in bold. The FZs are outlined in light to dark blue (colder fires of
less than 150 °C) and orange to red to purple (hotter fires of more than 150 °C), depending
on surface temperature. These zones had been delineated by several DLR ground thermal
infrared surveys (Gielisch & Künzer 2003).
340
.
Figure 2: Satellite image of Wuda Coal Mining Area (projection UTM 48S (WGS84))
Data source: Gielisch & Künzer (2003)
3.1 Airborne electromagnetic survey
An apparent resistivity map at a frequency of 4,650 Hz (horizontal coplanar coil

configuration) was drawn after the airborne electromagnetic survey over FZ 8; it is shown
in Figure 3. Apparent resistivities (Rhoa) as well as centroid depth values were calculated
using a homogeneous half-space model (Siemon 2001). The centroid depth depends on
conductivity and varies between 20 and 35 m at the given frequency. Close to the TEM
profiles nos. 6 and 7 (see Figure 3 for exact location), the centroid depth ranged from 20
to 26 m. Several resistivity lows obviously caused by ongoing coal removal or mining
activities are marked by black circles. The conductive area in the left part of the map might
be a result of high iron oxide content in sandstones.
341
Figure 3: Map of apparent resistivity Rhoa (Ohm*m) at a frequency of 4,650 Hz, derived
from airborne survey over FZ 8. The locations of TEM sites crossing the active
parts of the FZ are marked by black squares, including a drilling site close to
TEM site no. 4 (see Chapter 3.2).
3.2 Ground electromagnetic survey
Several TEM ground profiles crossed FZ 8. TEM profile no. 3 ran from west to east in the
northern part of the FZ, profile no. 5 in the southern part. Profiles nos. 6 and 7 ran from
south to north, starting directly north of the inaccessible part of the area.
The areas of low apparent resistivity (Rhoa below 50 Ohm*m) determined by the
helicopter survey coincide with areas showing low resistive layers (Rho below 30
Ohm*m) in the ground TEM survey, i.e. the eastern part of profile no. 3 (Figure 4), the
southern part of profile no. 6 (Figure 5), and the central part of profile no. 5 (Figure 7).
Low resistive layers were found at sites 47, 45, and 44 of profile no. 3 (from west to
east), at sites 1, 4, 2, 3, and 6 of profiles no. 6 (from south to north), and at sites 62 to
57 of profile no. 5 (from west to east). Site 47 of profile no. 3 shows a low resistive
layer (Rho about 26 Ohm*m) below 30 m depth. This location coincides with a
burning coal seam and an anomaly found in the ground magnetic survey (Figure 9).
Sites 46 to 44 display a conductive layer close to the surface. These locations coincide
342
with a low resistive area found in the helicopter survey (Figure 3) and with an anomaly
of the total magnetic field intensity discovered in the airborne magnetic survey (Figure
10). The TEM resistivity-depth section of profile no. 6 displays a northward-dipping
conductor (Figure 5) at a depth of 23 m (site 1) down to about 50 m (site 6). The
southern part of FZ 8 was not accessible on ground due to numerous cracks passing
the area; no data could be obtained here. At a drilling site about 50 m north of site 1 of
profile no. 6, close to TEM site no. 4, the top coal seam was localized at 20 m below
surface. With increasing distance from the cracks (in northerly direction), the depth
of this conductor increases, and so does its resistivity (from 10 to 23 Ohm*m). In the
northern part of profile no. 6 (sites 10 and 11), the resistivity structure underground is
more complicated and one-dimensional modeling may not be sufficient for adequate
explanation of data.
Investigations of gas temperatures at FZ 8 point toward a fire front movement in

northeasterly direction (Schlömer et al. 2008 in this volume). This corresponds to
the continuation of the northward-dipping conductor of TEM profile no. 6 to profile
no. 7 (Figure 6), situated 50 m to the east of the former profile (see Figure 3).
Figure 4: Resistivity-depth section from one-dimensional inversion of TEM data

along profile no. 3 from west to east. The burning zone is marked by a
dashed red line.
343
Figure 5: Resistivity-depth section from one-dimensional inversion of TEM data along

profile no. 6 from south to north. The burning zone is marked by a dashed red
line. The conductor is dipping northward from about 25 m down to 50 m depth.
Figure 6: Resistivity-depth section derived from one-dimensional inversion of TEM data

along profile no. 7 from south to north
344
The resistivity-depth section along profile no. 5 (Figure 7) displays a sufficient spatial
correlation between shallow conductor and burning zone. The eastward-dipping conductive
layer coincides with the thermally affected coal seam at sites 60 to 58. The area of low
apparent resistivities in the airborne electromagnetic survey (Figure 3) corresponds to the
location of the thermally affected coal seam.
Figure 7: Resistivity-depth section derived from one-dimensional inversion of TEM data

along profile no. 5 from west to east. The burning zone is marked by a dashed
red line and correlates with the eastward-dipping shallow conductor.
TEM soundings at sites 35 and 36 were performed over two unaffected coal seams (seams
nos. 6 and 7), about 1 km southeast of FZ 8. The one-dimensional modeling results did
not reveal conductive layers at coal seam depth. Figure 8 shows an example derived from
TEM data; at the depth of coal seams nos. 6 and 7 no resistivity low was observed.
Figure 8: One-dimensional inversion models derived from TEM data outside FZ 8’s burning
zones over unaffected coal seams nos. 6 and 7 about 1 km southeast of FZ 8
345
3.3 Airborne and ground magnetic surveys
The ground magnetic survey performed by AGRS was carried out with a HC-95 ground
helium optically-pumped magnetometer. The line direction was mainly east to west,
and the nominal line spacing and sampling distance was 10 m for most measurements.
The second ground magnetic survey, carried out by DMT, used a GSM-19, a system
manufactured by GEM, Canada (Elsen 2006). The results of both surveys did not differ
significantly, but the latter shows more details due to increased density of survey sites.
Both data sets were merged. The map of anomalies of magnetic field intensity of FZ 8
presented in Figure 9 shows pronounced positive amplitudes over burning and burned
coal seams. The burning zones (purple, red, and blue polygon lines) were derived from
temperature and satellite data (Gielisch & Künzer 2003; Künzer et al. 2008 in this volume).
Figure 9: Anomalies of magnetic field intensity ∆T, derived from ground magnetic
surveys conducted by AGRS and DMT in 2005. Dark or light blue and purple
or red lines outline areas of colder and hotter coal fires.
The map of anomalies of the magnetic field intensity ∆T derived from the airborne
magnetic survey of FZ 8 is shown in Figure 10. Although the regional magnetic
346
field has not yet been deducted, the airborne magnetic anomalies correspond to
those measured on ground. Due to cubic amplitude reduction of magnetic field
intensity with increasing distance from the magnetic source, i.e. the soil, the
amplitudes obtained at a relatively high flight altitude (about 60 m) are smaller and
the anomalies broader than in the ground magnetic surveys. Nonetheless, anomalies
with pronounced positive amplitudes coincide with areas of burning and burned coal
underground.
Figure 10: Anomalies of magnetic field intensity ∆T over FZ 8, derived from magnetic
airborne survey at an average flight altitude of 60 m and a line separation of
about 50 m. Blue and red lines mark areas of colder and hotter coal seam fires.
347
4 Conclusions
The objective of the geophysical surveys in Wuda Coal Mining Area was to detect and
delineate coal fire areas by means of physical parameters, such as electrical conductivity
and magnetic field intensity. Both parameters can point to coal fires – electromagnetic
surveys highlight areas of lower resistivity, and magnetic surveys higher amplitudes of
magnetic field intensity. The investigations presented in this study clearly prove that both
methods can be used effectively as a tool for detection and exploration of burning and
burned coal seams.
As the variance of geophysical parameters over coal fires compared to unaffected areas is
rather small, ground geophysical surveys are the more reliable tool in coal fire detection
and exploration; they allow for higher data sensitivity than airborne surveys. However, coal
fire areas are not always accessible on ground, and airborne surveys are an alternative in
these cases. Because of the weak parameter variance, high-resolution and high-quality data
are imperative in geophysical investigations. It is essential for obtaining meaningful results
to: (i) survey at low flight altitude, (ii) cover the investigated area by narrow survey lines,
and (iii) use high-precision measuring devices, as well as (iv) sophisticated processing and
interpretation tools.
Acknowledgements
We are very much obliged to the German and Chinese research teams, who performed
the airborne geophysical surveys under leadership of AGRS. Special thanks are due in
particular to Chen Bin of AGRS and Hans-Joachim Rehli of BGR, in charge of the survey
operations and field activities. Gernot Reitmayr and Hanno Schmidt of BGR participated
in the ground geophysical survey and contributed largely with their experience, both
in the field and during data interpretation. Wolfgang Voß helped with the airborne data
interpretation. We also wish to express our gratitude toward our German and Chinese
colleagues, who assisted in the field trips and provided many useful comments throughout
the study.
348
References
Bartel, L.C., 1982. Evaluation of the CSAMT geophysical technique to map abandoned coal mine fires.
Contract DE-AC04-76DP00789. Department of Energy, Sandia National Laboratories,
Albuquerque, New Mexico, U.S.A. In Proceedings of the 52nd Annual International Meeting
of the Society of Exploration Geophysicists, Dallas, Texas, U.S.A., 419-421.
Duba, A., 1983. Electrical conductivity of Colorado oil shale to 900 °C. Fuel 62: 966-972.
Duba, A., 1977. Electrical conductivity of coal and coal char. Fuel 56: 441-443.
Elsen, R., 2006. Geophysikalische Messungen zur Erkundung der Brandzonen 8 und 3, Wuda, Magnetik.
Ergebnisbericht Deutsche Montan Technologie GmbH (DMT). Unpublished Report, as a
Exploration, Extinction, and Monitoring of Coal Fires in North China.” Federal Institute for
Geosciences and Natural Resources (BGR), Hanover, Germany.
Wessling, Germany.
Greinwald, S. and Schaumann, G., 1997. Transientelektromagnetik. In Knödel, K., Krummel, H., and
Lange, G. (Eds.), Handbuch zur Erkundung des Untergrundes von Deponien und Altlasten.
Band III: Geophysik. Berlin, Heidelberg: Springer Verlag, 305-318.
Hooper, R.L., 1987. Factors affecting the magnetic susceptibility of baked rocks above a burned coal
seam. International Journal of Coal Geology 9: 157-169.
King, A., 1987. Cindered coal detection using transient electromagnetic methods. Geoexploration 24:
367-379.
Künzer, C., Zhang, J.Z., and Hirner, A., 2005. Multi-temporal coal fire dynamics – Combining
thermal remote sensing analysis and temperature field mappings to assess coal fire
development in Wuda Coal Mining Area. In Proceedings of the 9 th International
Symposium on Physical Measurements and Signatures in Remote Sensing (ISPMSRS),
Beijing, P.R. China, 17-19 October 2005, 805-809.
Künzer, C., Zhang, J.Z., Hirner, A., Jia, Y.R., and Sun, Y.L., 2008. Multi-temporal in-situ mapping of the
Wuda coal fires from 2000 to 2005: Assessing coal fire dynamics. In UNESCO Office Beijing
Kus, J., Hiltmann, W., and Balke, A., 2008. Researching coal spontaneous combustion: Micropetrography
of coal oxidation and carbonization. In UNESCO Office Beijing (Ed.), Spontaneous coal
349
Powell, R.W. and Schofield, F.H., 1939. The thermal and electrical conductivities of carbon and graphite
to high temperatures. Proc. Phys. Soc. 51: 153-172.
Schlömer, S., Teschner, M., Poggenburg, J., and Seeger, C., 2008. Gas and temperature monitoring
of a spontaneous coal seam fire in Wuda Coal Mining Area. In UNESCO Office Beijing
Siemon, B., 2001. Improved and new resistivity-depth profiles for helicopter electromagnetic data.
Journal of Applied Geophysics 46: 65-76.
Sternberg, R. & Lippincott, C., 2004. Magnetic surveys over clinkers and coal seam fires in Western North
Dakota. Presentation at the Denver Annual Meeting of the Geological Society of America,
Denver, Colorado, U.S.A., 7-10 October 2004.
121-136.
Weidelt, P., 1984. Bericht über Inversion transient-elektromagnetischer Messungen über geschichtetem
Untergrund. Archives-no. 96720. Federal Institute for Geosciences and Natural Resources
(BGR), Hanover, Germany.
350
Monitoring Coal Fires Using Remote Sensing
Prasun K. Gangopadhyay, Freek van der Meer, and Paul van Dijk
Department of Earth Systems Analysis, International Institute for Geo-Information Science and Earth
Observation (ITC), Enschede, Netherlands
Abstract
Recent studies on ice cores from Greenland and Antarctica have proven that atmospheric temperature
and CO2 content follow a regular 100,000-year cycle of change, and that these two quantities are closely
related. It was also found that the recent increase of CO2 in the atmosphere exceeds the values extrapolated
from historical data. Other than industrialization and rapid urbanization, natural hazards such as wild fires,
spontaneous coal fires, and leakage from hydro-carbon reservoirs contribute to atmospheric CO2 increase.
Coal fires, both underground and at surface, are a serious problem in most coal-producing countries. In the
1960s, the United States was the first country to use remote sensing as a tool for coal fire detection. A good
number of space- and airborne thermal remote sensing sensors have since come into use. Multi-spectral
spaceborne thermal remote sensing data can be employed to detect coal fire-related hot spots. For a better
understanding of coal fire dynamics, it is necessary to quantify fire front propagation. Multi-spectral/multi-
temporal spaceborne remote sensing (using satellites such as Landsat TM, ETM+, and ASTER) can be used
for this purpose by indicating direction and velocity of propagation. In the present study, different thermal
infrared images were acquired for a fire zone in Wuda Coal Mining Area (Inner Mongolia Autonomous
Region, P.R. China), and processed so as to allow detection and monitoring of coal fire hot spots and their
spatial variation over time. The data sets were geo-corrected and could be overlaid up to pixel level. Fire
locations as predicted in controlled model simulations showed a partial overlap with the hot spots detected in
the area. In a second study, radiative transfer models were used to identify CO2 absorption bands with a view
to facilitating remote sensing-based detection of CO2 emissions from coal fires.
摘要
近年来对格陵兰岛和南极洲的冰芯研究已经证明了大气温度和CO 2浓度基本上按10万年周期变
化，而且大气温度和CO2浓度密切相关。也有人认为：大气中增加的CO2量已超过基于历史数据
的推测结果。除了工业化和快速城市化原因外，野火、自然煤火和来自碳氢化合物储藏装置泄漏
的CO2都增加了大气CO2的浓度。在大多数产煤国，地下煤火和地表煤火都是一个严重问题。在
20世纪60年代，美国是世界上第一个采用遥感技术进行煤火探测的国家，此后许多星载和机载热
红外遥感传感器都相继使用，其中星载多光谱热红外遥感数据可用于探测与煤火相关的热点。为
更好地了解煤火动态情况，定量分析煤火区前沿扩展是非常必要的。多光谱/多时相的星载遥感
351
（Landsat TM和ETM+，ASTER）可达到了解煤火发展方向和扩展速度的目标。本研究采集了内
蒙古乌达矿区中一个火区不同时间的热红外图像，并进行了图像处理以便探测与煤火相关的热点
及其随时间的空间变化。热图像数据集进行了地理校正而且可以达到象素级覆盖，使用控制模型
仿真所预测的煤火位置显示与实际上探测到的热点部分重叠。其次研究中为推进利用遥感探测来
自煤火的CO2排放量，使用了辐射转换模型来识别CO2吸收带。
1 Introduction
Over the past two centuries, anthropogenic emissions of greenhouse gases (GHGs) have
increased to an alarming degree. GHGs in the atmosphere act as a blanket that retains solar
radiation and cause what is generally referred to as the greenhouse effect. Carbon dioxide
(CO2) plays the most important role in this phenomenon. Since pre-industrial times, the
concentration of CO2 in the atmosphere has increased from 280 to 375 ppmv (IPCC 2001).
This increase is only partly anthropogenic (as in industrialization and deforestation); it also
pertains to geo-natural events – coal fires for instance.
Spontaneous coal fires occur in almost all coal-producing countries, including Australia,
China, India, Indonesia, Russia, South Africa, and the U.S.A. (Ellyett & Fleming 1974;
Saraf et al. 1995; van Genderen et al. 1996; Walker 1999; Tetuko et al. 2003; Stracher
& Taylor 2004). Apart from depleting valuable non-renewable resources, coal fires emit
enormous amounts of sulfur dioxide (SO2), nitrogen oxides (NOx), as well as carbon
monoxide (CO) and CO2. Spontaneous combustion of coal depends on its propensity to
oxidation at ambient temperature. The absorption of oxygen at the surface of coal is an
exothermic reaction, which, if dissipation fails, may increase the temperature within the
coal body. If this rise continues until a so-called “threshold” temperature, somewhere
between 80 and 120 °C, is reached, a continuous reaction will result, which produces
gases like CO2 (Banerjee 1985; Schmal 1987). The coal temperature will then continually
increase until, at 230 to 280 °C, the reaction becomes more rapid and strongly exothermic,
i.e. the coal reaches a point of “ignition” or “flashing” and starts to burn. In some cases,
human negligence or (mine) accidents have caused coal fires (Sinha 1986).
Until the 1960s, borehole temperature measurements were the main tool in subsurface
coal fire detection. Though this method allows measurements in close proximity to a
fire, it does not produce sufficient data to cover larger areas in a synoptic view. Starting
in the early 1960s, space- and airborne thermal scanners came to be used for coal fire
detection in the U.S.A. A significant number of studies concentrating on coal fire detection,
monitoring, depth estimation, and thermal modeling based on remotely sensed thermal
data have since been carried out in China, India, Australia, and the U.S.A., (Fisher &
352
Knuth 1968; Ellyett & Fleming 1974; Prakash et al. 1997; Cassells 1998; Gangopadhyay
et al. 2005). Subsequent attempts at coal fire modeling involved estimations of fire extent
and propagation, and required the development of a remote sensing/GIS-based coal fire
detection and monitoring system (Rosema et al. 1999; Prakash & Vekerdy 2004; Voigt et
al. 2004; Zhang et al. 2004).
Multi-temporal thermal infrared data can be used in coal fire monitoring to obtain a better
understanding of fire propagation; this, however, presupposes a detailed knowledge of
the study area. The present study explores possibilities of fire propagation assessment by
lab-scale modeling. Coal fire maps were simulated in IDL (Interactive Data Language)
and validated against thermal satellite data and local geological parameters. To avoid
positional error, the data sets were geometrically corrected. Human interference such as
fire extinction or excavation in between acquisition and processing events of satellite data
could not be excluded and may have contributed to inaccuracies.
As said earlier, CO2 emissions from coal fires have adverse effects on the local and global
environment. Recent coal fire studies in China, the world’s leading producer of coal,
estimate that the country is responsible for 3 % of the world’s total annual output of CO2
from fossil fuels (Rosema et al. 1995). Others put figures at a lower (but still significant)
level – 0.3 % of the world’s total according to Voigt et al. (2004). The problem with these
estimates is that they are based on indirect methods, such as calculations from the amount
of burned coal in a fire area. Direct measurements and figures so far are missing.
CO2 has only a few narrow absorption bands in the visible near infrared region of the
electromagnetic spectrum (see Figure 5). In a second part, this study attempts to identify
these bands and establish a relationship between transmission and CO2 concentration at a
particular wavelength. The method is described in Chapter 3.2.
2 Study Area
The study area, Wuda Coal Mining Area (or Wuda Syncline), is located in Inner Mongolia
Autonomous Region, P.R. China (Figure 1). To the north and west it is bordered by the Gobi
Desert, to the east by the Yellow River, and to the south by the Helan Mountains. Wuda
Coal Mining Area extends over 10 km from north to south, and 3 to 5 km from east to west,
covering a total area of 35 km2. It has been subdivided into three mining zones by the local
mining authority: Wuhushan, Suhaitu, and Huangbaici.
353
Figure 1: Location of Wuda Coal Mining Area
The first coal fires in Wuda were recorded in 1961, in a small pit. Before 1989, these fires
were isolated and scattered in different places, but they gradually connected between 1989
and 1995 and have since been spreading rapidly. According to estimations by the Beijing
Remote Sensing Corporation (BRSC) from 2002, the total area affected by coal fires in
Wuda is 3.07 km2, equaling 8.77 % of the syncline (Wuda Coal Mine 2002).
3 Methodology
3.1 Simulating coal fire propagation
In the electromagnetic spectrum (EMS), the 3 to 14 µm region is known as the thermal

infrared region (Lillesand & Kiefer 2000). Thermal remote sensing uses atmospheric
windows at 3 to 5 µm and 8 to14 µm because these parts of the EMS are less affected
by atmospheric interaction. It exploits the fact that everything above absolute zero
354
(-273.15 °C) emits radiation in the EMS thermal infrared region. The radiation of an object
is controlled mainly by its emissivity, geometry, and temperature. Thermal infrared sensors
record differences in the received infrared radiation.
Thermal infrared data from Landsat 5 TM of 30 August 1991 (daytime) and Landsat
7 ETM of 12 August 1999 (daytime) were used in this study, along with an ASTER-
derived DEM (Digital Elevation Model) of 2002. To reduce the terrain effect, ATCOR3
(Atmospheric and Topographic Correction, version 3) was employed in combination with
the DEM (Richter 2005).
The geo-corrected images of 1991 and 1999 were processed in a common remote sensing
procedure (using Planck’s equation) to create isothermal maps with a spatial resolution of
10 m. Local geological and ground-truth data collected in 2002 served to identify potential
coal fire areas in these maps. To define the threshold temperature on the basis of these data,
the following procedure was applied:
1) First, the temperature difference between the two images (1991 and 1999) at a specific
point of a near blackbody (here the Yellow River), for which ground-truth data had
been collected during satellite overpass time, was determined.
2) This difference was then compensated by adding or subtracting from the threshold value
of 2002 (determined by averaging temperatures of different land covers during satellite
overpass time).
3) After definition of the threshold value, coal fire locations were identified in the
isothermal maps depending on local geological setting.
As expected, few pixels with higher-than-average temperature were found that could be
defined as related to active coal fires. In view of the study area’s complex mixed land cover,
a small plot of 4 km2 was therefore chosen to test fire propagation modeling. The following
data were used as input for the simulation model: (i) the location of (a) coal fire(s) as
identified in the 1991 thermal infrared image, (ii) a general coal fire propagation velocity
of 10 m yr-1 as gleaned from in-situ investigations and discussions with Wuda Coal Mine,
and (iii) the location of the coal seam hosting the fire(s) as known from geological surveys.
The location of the 1991 coal fire served as a starting point for north- and southbound fire
propagation along the coal seam. A time series of simulated images from 1991 to 1999 was
created and compared with the thermal infrared image of 1999.
355
3.2 Possibilities of estimating CO2 emissions based on remote sensing
To identify discrete CO 2 absorption bands for use in remote sensing-based coal fire
detection and monitoring, two types of atmosphere were simulated using FASCOD (Fast
Atmospheric Signature Code). They were modeled after the atmospheric conditions in
the study area. The first type consisted of only CO2, i.e. was void of water vapor, other
atmospheric gases, and aerosols; the second consisted of all constituents except CO2. These
two data sets were re-sampled to a resolution of 1 nm so as to facilitate analysis in the
range from 700 to 2,500 nm covered by most hyperspectral remote sensing sensors. After
positive identification, partial least squares (PLS) regression was run with different CO2
concentrations (330 to 570 ppmv) to estimate the scientific significance of the identified
absorption bands.
4.1 Simulating coal fire propagation
The 1999 image of the simulation times series in IDL (Figure 2) and the processed thermal
infrared image of 1999 (Figure 3) showed only partial overlap. Whereas the simulated
northern “fire 1” coincided with a hot spot detected in the thermal infrared image, the
second, southern “fire 2” bore no relation to any thermal anomaly. This may be attributed
to the surface set-up, which in reality is far more complex than in the simulations. For
example, ventilation in coal seams can greatly change under the impact of external factors
and significantly increase or decrease fire propagation. Despite this discrepancy, the model
offers a promising avenue for further research on coal fire propagation.
Figure 2: Simulation of coal fire propagation. Assumed coal fire starting point in 1991
(red square; left) and position of coal fires along coal seam (white line) in 1999
(red squares; right).
356
Simulated coal fires 1999
Hot spots as detected in processed image of 1999
Coal seam
Figure 3: Overlay of simulated and processed thermal infrared images 1999
4.2 Possibilities of estimating CO2 emissions based on remote sensing
Among GHG emissions from coal fires, CO2 is the most critical gas in terms of quantity.
Estimations as to how much CO2 is released from a particular coal fire may contribute to
efficient planning and implementation of coal fire fighting campaigns. In order to establish
a remote sensing-based model for inversion of transmission and CO2 concentration, it
is important to identify specific CO2 absorption bands in a first step. Figure 4 shows the
results of the FASCOD simulations with CO2-free and CO2-only atmospheres. The whole
EMS range from 2.008 to 2.013 µm, apparently, is not influenced by atmospheric gases
other than CO2, which in this range shows frequent absorption lines. These lines are
particularly pronounced at 2.0083, 2.0086, 2.0095, 2.0102, 2.0101, 2.0115, 2.0122, and
2.0129 µm.
357
Figure 4: CO2 absorption lines (red) with less interference in wavelength range of 2.008
to 2.013 µm. The blue lines represent transmission of water vapor and other
atmospheric gases.
To estimate the significance of the identified CO2 absorption lines, PLS regression was
run with different CO2 concentrations (330 to 570 ppmv) at the same wavelength range.
Among the significant absorption bands, two were chosen for further testing because they
were not influenced by any other atmospheric constituent. These two bands are situated at
2.001 and 2.010 µm (Figure 5).
Figure 5: Distribution of significant CO2 absorption bands
358
The two absorption bands were tested for their sensitivity toward changing CO 2
concentrations. As evident from Figures 6a and b, there is a good correlation between CO2
concentration and transmission. Tests on simulated radiance images proved that the same
exists for CO2 concentration and radiance at sensor, which can be inverted to quantify CO2
concentration per pixel (notwithstanding some uncertainties).
CO2
CO2
Figure 6: Relation between CO2 concentration and transmission at wavelengths 2.001

(a) and 2.010 µm (b) in controlled atmosphere
359
5 Conclusion
Remote sensing plays an important role in detecting and monitoring coal fires and
can help prevent economic loss and environmental disaster. An adequate and well-
adjusted simulation model has the potential to predict the direction and velocity of coal
fire propagation, which may prove useful in fire fighting planning and implementation.
However, external factors that may increase oxygen supply and hence fire propagation
such as air flow and subsidence can not be accounted for in the model.
GHG emissions from coal fires are gaining importance in connection with global
climate change. As described in this study, the relation between transmission and CO2
concentration can be extrapolated to radiance at sensor and CO2 concentration by inversion
(unpredictable parameters such as cirrus cloud and local aerosol loading remain and have
to be ignored in such calculations). Although most hyperspectral sensors operate in visible
and shortwave infrared range, where the absorption bands of CO2 are very narrow, there
is an option to scale down the parameters retrieved from controlled laboratory model
scenarios and fit in the available remote sensing bands so as to allow for estimation of coal
fire-related CO2 emissions.
Acknowledgements
The authors wish to acknowledge the contribution to this study of Mr. Ulanbek
Turdukulov, GIP Department, ITC.
References
Ellyett, C.D. and Fleming, A.W., 1974. Thermal infrared imagery of the Burning Mountain coal fire.
Remote Sensing of Environment 3(1): 79-86.
Fisher, W.J. and Knuth, W.M., 1968. Detection and delineation of subsurface coal fires by aerial infrared
scanning. Geological Society of America Bulletin 115: 67-68.
Gangopadhyay, P.K., Maathuis, B., van Dijk, P., 2005. ASTER-derived emissivity and coal fire-related
surface temperature anomaly: A case study in Wuda, north China. International Journal of
Remote Sensing 26(24): 5555-5751.
IPCC, 2001. Climate Change 2001: The Scientific Basis. Intergovernmental Panel on Climate Change
(IPCC), Geneva, Switzerland. Cambridge: Cambridge University Press.
Lillesand, T.M and Kiefer, R.W., 2000. Remote sensing and image interpretation. 4th edition. New York:
John Wiley and Sons.
360
Prakash, A. and Vekerdy, Z., 2004. Design and implementation of a dedicated prototype GIS for coal fire
investigations in north China. Coal Geology 59: 107-119.
Prakash, A., Gupta, R.P., and Saraf, A.K., 1997. A Landsat TM-based comparative study of surface and
subsurface fires in the Jharia Coalfield, India. International Journal of Remote Sensing 18(11):
2463-2469.
Richter, R., 2005. Atmospheric/topographic correction for satellite imagery. ATCOR 2/3 user guide,
version 6.1, January 2005. German Aerospace Center (DLR), Wessling, Germany.
Netherlands.
Rosema, A., van Genderen, J.L., and Schalke, H.J.W.G., 1995. Environmental monitoring of coal fires in
north China. Project report 4.2/TO-04. Beijing Remote Sensing Corporation (BRSC), Beijing,
P.R. China.
Saraf, A.K., Prakash, A., Sengupta, S., and Gupta, R.P., 1995. Landsat TM data for estimating ground
temperature and depth of subsurface coal fires in the Jharia Coalfield, India. International
Journal of Remote Sensing 16(12): 2111-2124.
Sinha, P.R., 1986. Mine fires in Indian coalfields. Energy 11(12): 1147-1154.
Stracher, G.B. and Taylor, P.T., 2004. Coal fires burning out of control around the world: Thermodynamic
Tetuko, S.S.J., Tateishi, R., and Takeuchi, N., 2003. A physical method to analyze scattered waves from
burned coal seams and its application to estimate thickness of fire scars in central Borneo
using L-Band SAR data. International Journal of Remote Sensing 24(15): 3119-3136.
Van Genderen, J.L., Cassells, C.J.S., and Zhang, X.M., 1996. The synergistic use of remotely sensed data
for the detection of underground coal fires. International Archives of Photogrammetry and
Remote Sensing 31(7): 722-727.
121-136.
United Kingdom.
Wuda Coal Mine, 2002. Personal interview / Oral communication. (Wuda Coal Mine, Inner Mongolia
Autonomous Region, P.R. China)
Zhang, J.M., Wagner, W., Prakash, A., Mehl, H. and Voigt, S., 2004. Detecting coal fires using remote
sensing techniques. International Journal of Remote Sensing 25(16): 3193-3220.
361
Automated Demarcation, Detection, and
Quantification of Coal Fires in China
Using Remote Sensing Data
Claudia Künzer1, Zhang Jianzhong1, Anke Tetzlaff2, Stefan Voigt3, and

Wolfgang Wagner1
1
Vienna, Austria
2
Swedish Meteorological and Hydrological Institute, Norrkoeping, Sweden
3
German Aerospace Center (DLR), Wessling, Germany
Abstract
This paper briefly summarizes the results of three PhD theses completed as part of the Sino-German
Coal Fire Research Initiative “Innovative Technologies for Exploration, Extinction, and Monitoring
of Coal Fires in North China.” The theses had a common topic; the analysis of coal fire areas based on
remote sensing data. The first work focused on the automated demarcation of coal fire risk areas. The
second investigated methods to detect subtle thermal anomalies from thermal remote sensing data. The
third dealt with thermal anomaly quantification of coal fires. Compared to remote sensing-based coal
fire research before 2001, a major challenge in these works was the development of automated analysis
methods, applicable on a large spatial scale so as to provide for a coal fire monitoring system covering
extensive areas in northern China as well as transfer regions. The major sensors for multi-spectral and
thermal data were the common earth-observing satellites Landsat 5 TM, 7 ETM+, the US-Japanese
ASTER satellite, the German experimental satellite BIRD, as well as the US sensor MODIS. The coal
fire risk area demarcation algorithm was based on a multi-spectral knowledge-based test sequence,
which automatically extracted land cover surfaces with increased likelihood of hosting coal fires.
The algorithm for thermal anomaly extraction from thermal data bands employed a moving window
approach, which extracted subtle local thermal anomalies based on filter-window histogram analysis.
The quantification algorithm linked thermal radiance received at the satellite sensor with the energy
release of a pixel through a linear relationship. All three algorithms allowed large-scale data processing
and spatial transfer of results.
362
摘要
本文主要总结了作为中德关于“中国北方地区煤火的探测、灭火和监测新方法”合作研究项目中
三个博士论文的成果。论文的工作重点集中在基于遥感数据的煤田火区分析方法。第一部分重点
放在识别煤火危险区，第二部分重点是基于遥感热红外数据探测微弱热异常的调查方法，第三篇
论文讨论了煤火热异常的定量分析方法。与2001年以前基于遥感数据的煤火研究相比，该项研究
中主要面临的挑战是要开发煤火自动分析方法，该方法必须适用于大空间尺度，从而为中国北方
广大地区煤火实用化监测工作提供坚实的基础。提供多光谱和热红外数据的主要传感器是常见的
美国陆地资源探测卫星5TM和7号卫星的ETM+、美国和日本合作的ASTER卫星、德国试验卫星
BIRD以及美国的传感器MODIS。煤火危险区识别算法依据多光谱信息测试序列，它可以随着包
含煤火的可能性增加而自动提取出地面覆盖区域。从热波段数据中提取热异常的算法是基于滤窗
柱状图分析，采用了移动窗口的方法提取细微的热异常。最后，定量分析算法用线性关系将卫星
传感器上接收的热辐射与每个象素的能量释放相联系。上述三种方法可以进行大规模的数据处理
和成果的空间转换。
1 Remote Sensing of Coal Fires in Past Decades
Documented coal fire research based on remote sensing data started around 1963, with a
producer of airborne thermal cameras testing their equipment above burning coal waste
heaps in Scranton (Pennsylvania, U.S.A.). The thermal imagery reproduced the burning
areas within the heaps as light grey values (Slavecki 1964). Further aerial surveys with
thermal cameras in the 1960s and 70s were conducted by Moxham and Greene (1967),
Greene et al. (1968), Fisher and Knuth (1968), Knuth et al. (1968), Rabchevsky (1972),
and Ellyett and Fleming (1974). Research then focused on coal fires in the U.S.A. on
a very local scale (Zhang et al. 2004); it was dominated by visual interpretation and
localization of well-known fires from imagery. Density slicing was the only method used
to extract fires from image data.
In the early 1980s, coal fires had become the focus of studies in several countries; yet
airborne remote sensing was still the main platform in use. In 1983, multi-spectral scanner
data and thermal infrared scanner data were employed for a first time in the Chinese region
of Taiyuan Xishan in Shanxi Province (Li 1985; Guan 1989). Coal fire research based on
airborne multi-spectral scanner data was continued in India, where Bhattacharya et al.
(1991, 1994) and Prakash et al. (1995) investigated Jharia Coalfield, deriving coal fire
maps from aerial thermal data based on ground-truth knowledge. Details on the use of
airborne thermal scanner data for coal fire detection are given in Zhang. et al. (2004). The
Jharia region was studied again based on satellite data by Mansor et al. (1994), who used
thermal and mid-infrared NOAA-AVHRR and thermal Landsat 5 TM data. Saraf et al.
(1995) as well as Prakash et al. (1997), Prakash and Gupta (1998), and Gupta and Prakash
363
(1999) studied the Jharia coal fires using Landsat 5 TM thermal band data, with a spatial
resolution of 120 m. This data basis was sufficient for detection of surface fires but often
proved inadequate (i.e. too coarse) for very deep or small coal fires (Zhang 1998).
In general, studies based on thermal Landsat 5 TM band 6 data were limited to the
demarcation of hot and/or large-surface coal fires, and, to a minor degree, of hot or large
underground coal fires supported by field measurements. Research practice included
the application of satellite data in different locations, the evaluation of density slicing
techniques for coal fire detection, and temperature retrieval when fires exceeded several
pixels in size. Time series analyses were carried out to describe fire movement. First relief-
based thermal calibration and differentiation between thermal daytime and nighttime data
were frequently used, also in combination. Information sets from Landsat band 6, and band
5 and 7 (1.55 to 1.75 µm and 2.08 to 2.35 µm) were combined in some studies. The size
and temperature of very hot surface coal fires occupying sub-pixel portions was calculated
based on the “dual band method” developed by Matson and Dozier (1981) (Zhang et al.
1997; Zhang 1998; Prakash & Gupta 1999).
Satellite data with coarser spatial resolution than Landsat TM were rarely used for coal fire
research. Though Mansor et al. (1994) were able to spot very hot surface coal fires in Jharia
with NOAA-AVHRR data (1 km), Zhang (1998) proved that the spatial resolution of this
satellite is too low for detection of underground coal fires in northwest China. The same
applies in general to un-enhanced 1 km resolution data from MODIS. Hecker and Künzer
(2006), however, demonstrated that MODIS data – now available as daytime data from
TERRA and as valuable pre-dawn data from AQUA – can indeed support the extraction
of thermal anomalies resulting from surface coal fires, by successfully employing ratio
images from MODIS thermal nighttime data in their detection efforts.
Sensors with an intermediate spatial resolution such as the German BIRD sensor with a
pixel size of 185 m (no pre-dawn overpass) can only detect extremely hot surface coal fires
in mid infrared (3.4 to 4.2 µm).
Remote sensing research in the past decades had a strong focus on interactive analysis,
extracting coal fire thermal anomalies by means of density slicing and thresholding.
Spatially transferable or automated methods were absent. Research based on multi-spectral
reflective remote sensing information was conducted in very rare cases only, and depended
largely on local descriptions of coal fire environment (Guan & van Genderen 1997;
Prakash & Gupta 1998, 1999; Prakash et al. 2001). The three approaches presented in this
364
paper aimed to realize large-scale transferable analyses to enable a China-wide monitoring

of coal fires. These approaches can be modified and extended by additional interactive
analyses if necessary (Künzer 2005).
Figure 1 shows an area of about 20 × 20 km. The green, red, and near-infrared sensor
bands are displayed in blue, green, and red. Since vegetation has its reflection maximum
in the near-infrared region, it appears reddish. In the picture on the right, part of a thermal
image has been superimposed on a multi-spectral satellite image.
Figure 1: False-color satellite image of Wuda Coal Mining Area with part of a color-
coded thermal image superimposed (right); extent: 20 × 20 km
The red (strongly heated) areas on top of the outcropping coal seams in Figure 1 are likely
to be coal fires. Coal fire-related thermal anomalies do usually not exceed 1 km² in size.
However, thermal data need to be normalized prior to analysis to eliminate effects of
uneven solar heating. The thermal detection algorithm presented in Chapter 3 of this paper
is independent of such effects as it is a local investigation filter.
2 Automated Coal Fire Risk Area Demarcation
Automated coal fire risk area demarcation analyzes multi-spectral non-thermal data to
identify areas with high coal fire potential. The delineation of such risk areas can help
eliminate false alarms in thermal anomaly detection. It also highlights areas which are at
risk of spontaneous combustion but have not yet produced thermal anomalies.
365
From field observations and mapping during field campaigns, the authors of this study
inferred that coal fire-related thermal anomalies are always located directly on top of or
adjacent to coal exposed to the air at surface – i.e. outcropping coal seams, coal waste
heaps, coal storage piles, or mine entrances. Coal fires also manifest themselves in the
degradation and death of surface vegetation. Subsurface fires are not found underneath
densely vegetated soil or bedrock (Künzer & Voigt 2003). An additional phenomenon
induced by very hot coal fires is pyrometamorphism of bedrock. If a fire reaches
sufficient temperatures, it can alter the color and texture of overlying and adjacent rock
strata. Depending on its spatial extent, pyrometamorphic rock can be detected by remote
sensing. A knowledge-based spectral test sequence was developed for risk area definition
through automated extraction of three surface classes – coal, dense vegetation, and
pyrometamorphic rock. Coal at the surface is automatically extracted and delineated in
a certain radius. Within these plots, densely vegetated areas are excluded, and pixels of
pyrometamorphic rock highlighted. The algorithm runs fully automatically without user
interaction. A second approach deploying linear partial unmixing to extract surfaces based
on quantitative subpixel fraction estimation was also tested but is not presented here
(Künzer 2005).
A common and very effective approach to classify image data is the use of non-parametric
unsupervised decision tree classifiers (DTC) (Cihlar et al. 2000). “A decision tree is
defined as a classification procedure that recursively partitions a data set into smaller
subdivisions on the basis of a set of tests defined at each branch or node in the tree. A tree
is composed of a root node (formed from all of the data), a set of internal nodes (splits),
and a set of terminal nodes (leaves)” (Friedl & Brodley 1996). In this Boolean scheme, a
pixel is classified through the path which the answers to the questions at each node carve
out for it, from the root node to a leave node (Cortijo & Perez de la Blanca 1997).
Figure 2 demonstrates the single steps and calculations of the algorithm. An ETM+ or
nadir ASTER layer stack of sensor-calibrated and atmospherically corrected surface
reflectance data from different spectral bands is processed into several intermediate data
products. These are spectral mean value channels (SPM) and vegetation indices, which are
subsequently tested for the three surface types described above.
366
Figure 2: Summarized workflow of the knowledge-based test sequence for automated

coal fire risk area demarcation (not every temporary output file shown). After
spectral mean value channels (SPM) calculation, vegetation index data are
generated. Six equations and inequalities are tested for automated extraction of
coal, two equations for extraction of dense vegetation, and nine relationships
for pyrometamorphic rock.
Data source: Künzer (2005)
Surface extraction and coal fire risk area demarcation were performed for a time series
of eight ETM+ scenes covering the main study areas in China – Wuda, Rujigou, and
Gulaben. Applicability was also tested for an ETM+ scene in Baotou, located outside
the known study regions. In addition, transferability was tested successfully for ETM+
scenes in the Jharia and Rây Coalfields in India and Leigh Creek Coal Mining Area in
Australia.
367
Figure 3: Pixels automatically extracted as coal surface in up to eight time steps for the
broader Wuda area (filtered). The color scheme represents how often a pixel
was extracted as coal surface.
Time series analyses as those for Wuda shown in Figure 3 are important in fire fighting
and prevention as they reveal increases in outcropping coal surface and especially the
spread of “wild” private small-scale mining. The seams, waste heaps, and storage piles
in the syncline are correctly extracted by the algorithm. Also, the numerical increase and
spatial expansion of private small-scale mines in the east (on the opposite side of the
Yellow River) is apparent. The same applies to the growth of large storage piles near coal
workings in almost concentric circles (enlarged in the top left corner). Given these constant
changes in coal surface spatial extent due to mining, it is obvious that coal fire risk area
demarcation should be undertaken regularly, e.g. on a yearly basis.
Automated coal fire risk area extraction in the two main study areas of the Sino-German
Coal Fire Research Initiative, Wuda and Rujigou, brought to light an increase in mining
activity in the order of 230 to 350 % during the past 15 years. Surface extraction accuracy
was validated through an extensive set of over 50 in-situ mapped ground truth polygons.
It ranged from 76 to 100 % for all surfaces. Good surface extraction accuracy at 91 % on
368
average was also calculated for the transfer regions, where pyrometamorphic rock did not
exist at detectable scales (Künzer 2005).
Many new coal fire risk areas were discovered next to already known coal fire zones in
the two study regions. They are presented in detailed maps at a spatial scale of 1:100,000
in Künzer (2005). Within two of these risk areas (one off them is shown in Figures 4 and
5) unknown coal fires were detected by a combination of automated coal fire risk area
demarcation and automated thermal anomaly detection as developed by Zhang (2004). The
findings were verified during two field campaigns in 2003 and 2004 (Figure 5). They prove
that unknown coal fires can be detected on the basis of remote sensing data exclusively
(Zhang 2004; Künzer 2005).
Figure 4: New coal fires detected by combined use of coal fire risk area demarcation and
thermal anomaly detection, 25 km southeast of Wuda. The coal fires were not
known except to a few locals. Extracted coal is displayed as a superimposed
polygon vector layer in light blue, thermal anomalies in yellow (inside the
demarcated area) and green (outside the demarcated area), the demarcated
coal fire risk areas are superimposed in red. (Center coordinate: 669169E,
4365414N, UTM, Z48N)
369
a b
Figure 5: New coal fires detected by combined use of coal fire risk area demarcation and
thermal anomaly detection, 25 km southeast of Wuda. Note the coal fire-related
smoke in image a) and the improperly sealed former private mines in image b).
The trench was dug to prevent further propagation.
(Photos: a) C. Künzer; b) J. Zhang, June 2004)
3 Automated Thermal Anomaly Detection
The algorithm for automated thermal anomaly extraction from thermal bands of Landsat 7
ETM+, ASTER, and MODIS uses raw satellite data (DN values) or calibrated and corrected
thermal data as input for sub-image statistical analysis. Within a moving window of varying
size, the histogram of these subsets is investigated for thermally anomalous pixels. To determine
the average statistical “behavior” of coal fires in thermal images, the image histogram statistics
of known coal fire areas were studied in detail. From these, specific features were derived
within the subset histograms (first local minimum after main maximum), which could be
defined as thresholds, allowing for separation of thermally anomalous pixels from background
pixels. As the histograms were taken in a moving window, each pixel within the scene was
sampled repeatedly (> 1,000). Depending on how often a pixel was found to be thermally
anomalous (> 70 % of cases), it was marked a “thermal anomaly” (Zhang 2004).
The advantage of this approach compared to overall threshold definition is that local
thermal peculiarities can be accounted for. This means that, depending on the setting,
thermal anomalies of completely different temperatures can be extracted since their
definition is a mere function of the surrounding background.
The process is shown in Figure 6. The thermal band is investigated by a moving filter
of varying window size (from 19 × 19 to 35 × 35 pixels). The histogram of each sub-
370
window can contain no thermal anomaly or gradually be filled up to 100 % by thermal

anomalies. Zhang (2004) proceeded on the assumption that every sub-window histogram is
a superimposed whole of two different parts – the main/overall histogram representing the
background DNs (temperatures) (B in Figure 6, green) and the thermally anomalous part
(F in Figure 6, red). The first local minimum is defined as the relative threshold to separate
the sub-window from the main histogram maximum. This automated method leads to loss
of thermal anomalies if the sub-window is filled completely by a thermal anomaly or if the
thermally anomalous part of the histogram extends over the threshold (left side). Erroneous
anomalies are extracted if no thermally anomalous area is contained in the image and the
high temperature end of the histogram is regarded as anomalous. However, since every
center pixel is investigated more than 1,000 times and has to be marked as “thermally
anomalous” in at least 70 % of the cases, this last source of error is largely eliminated.
Figure 6: Principle of thermal anomaly extraction. The blue arrow over the small image
subset histograms indicates the first local minimum after the main histogram’s
maximum.
371
The thermally anomalous pixels are clustered by an 8-neighbourhood. Since one pixel is
surrounded by eight other pixels, these are checked for thermal anomalies too; directly
adjacent thermally anomalous pixels are clustered. These clusters are numbered and
statistically investigated for their minimum, maximum, and mean DN, their standard
deviations, and their spatial coverage. For example, coal fire areas do not exceed a certain
size (> 1 km²). If the algorithm picks out an anomalous cluster of several square kilometers,
this is most probably a sun-illuminated slope. Warm water surfaces can be excluded based
on the specific cluster’s DN or temperature variance – water shows a very low temperature
variance, coal fire clusters are the very opposite (Zhang 2004). Following this procedure,
the final output image only contains small-size thermal anomalies, which have an increased
likelihood of being coal fires. The whole workflow is depicted in Figure 7.
Figure 7: Workflow of automated thermal anomaly extraction and post-processing

according to thermal cluster statistics
Despite all precautionary measures, automated statistical analysis of thermal anomalies

can also lead to the extraction of pixels whose thermal anomaly is not related to coal fires.
Such anomalies are smaller sun-illuminated surfaces, which pass the area threshold. They
even occur in nighttime data (e.g. Landsat 7 ETM+ data acquired at around 10:30 p.m.; in
summer – when the sun sets only shortly before this time, the soil is still heated at some
points and can produce slight solar effects). Other thermal anomalies result form heating
of houses, industry, burning of fields and garbage, limestone burning, or even forest or
372
grassland fires. Coal fire risk area demarcation can serve as a spatial corrective to faulty
thermal anomaly extraction (Zhang 2004).
Figure 8 shows the result of nighttime thermal anomaly extraction on top of a daytime
Landsat 7 ETM+ image. Even from nighttime data, numerous anomalies are extracted
which are not related to coal fires. However, the synergies of automated thermal anomaly
detection and coal fire risk area demarcation can help eliminate errors.
Figure 8: Part of a multi-spectral Landsat satellite daytime image (c. 80 × 80 km) of the
northern Chinese coalfields in Inner Mongolia and Ningxia, including the two
major coal fire areas in the region – Wuda in the north and Rujigou in the east
of the subset. Overlaid in yellow are thermal anomalies extracted from thermal
Landsat nighttime data. Note that several yellow clusters are located in coal
fire areas; others stem from industry, domestic burning, households, and other
non-coal fire sources. This applies especially to all anomalies in the vegetated
regions (red) along the Yellow River.
373
4 Synergies of Coal Fire Risk Area Demarcation and Thermal

Anomaly Detection
The successful discovery of unknown coal fires based exclusively on remote sensing data
proves that the combined use of automated coal fire risk area demarcation and thermal
anomaly detection is indeed a very useful approach to coal fire detection and monitoring.
This will support the ongoing set-up of an operational coal fire detection and monitoring
system for large areas of northern China. But the synergies go beyond the mere detection
of new fires. Both algorithms were also tested successfully for transferability in processing
six scenes from different locations.
It was demonstrated that the demarcation of coal fire risk areas can substantially reduce
the spatial extent of thermal investigations. On average, only 10 % of each scene of all
investigated ETM+ input data sets were considered relevant for detection of thermal
anomalies related to coal fires. For the study areas of Wuda and Rujigou, this proportion
could even be reduced to a mere 1.5 %. To show the transferability of the implemented
algorithms, several transfer scenes within China, but also in Australia and India, were
chosen. For a scene of Baotou, China, 0.6 % of the input data set was extracted as coal
fire risk area. For a scene of the Leigh Creek Coal Mining Area, Australia, 1.5 % was
delineated. The same applies to India.
For daytime thermal remote sensing data, an average of 80 % of the automatically extracted
thermal anomalies can be classed as false alarm, while even 50 % in nighttime data bear no
relation to coal fires. These findings reveal that coal fire risk area demarcation can effectively
reduce errors in automated thermal anomaly extraction. Many subtle thermal anomalies
detected by the algorithm are related to industry, settlement, biomass, and limestone
burning, as well as solar effects. If they are located outside the delineated risk areas, there
is only a very small probability of their being coal fires. Thus, automated coal fire risk area
demarcation is a powerful tool for directing thermal anomaly monitoring toward appropriate
and relevant areas, and eliminating false alarms.
5 Thermal Anomaly Quantification
In the third study, an algorithm served to quantify the thermal clusters extracted by the thermal
detection algorithm. For each thermal cluster, an average temperature and energy release in
MW (still the common measuring unit in many geophysical analyses) was computed. The
calculation of this energy release was based on a linear relationship between the spectral
374
radiance observed by the Landsat ETM+ sensor in the thermal band 6 and the total energy
release of the coal fires in the cluster. The relationship is described in the following equation:
(1) CFREETM = 6300 + 185500 (LETM,P – LETM,BG) + 5700 (LETM,P – LETM,BG)2 [W],
where LETM,P is the atmospherically corrected ETM band 6 radiance of a potential coal fire
pixel (in W m-2 sr-1 μm-1), and LETM,BG is the mean atmospherically corrected ETM band 6
radiance of surrounding background pixels (in W m-2 sr-1 μm-1). This relationship had been
extrapolated from 1,000 modeled fire scenarios, with fire areas ranging from 1 to 1,100 m², fire
temperatures ranging from 350 to 600 K, and background temperatures ranging from 273 to
300 K (Tetzlaff 2004; see Figure 9). A polynomial best-fit relationship, overlaid in red, is in
very good correlation with the modeled data within the limits of the thermal infrared band
saturation (dashed line = saturation at a background temperature of 273 K; dotted line =
saturation at a background temperature of 298 K for Landsat ETM+ band 6 low-gain data).
Figure 9: M o d e l e d n e a r- l i n e a r r e l a t i o n s h i p b e t w e e n s p e c t r a l r a d i a n c e ( i n
W m-2 sr-1 μm-1) received at satellite sensor in thermal band 6 and radiative
energy (in W) released by coal fires
Correct quantification of thermal coal fire clusters depends on accurate extraction of

thermally anomalous clusters by the thermal detection algorithm. Only those clusters that
375
have been extracted as thermally anomalous will be quantified. For high-accuracy results
in remote sensing-based coal fire detection (and hence, also in quantification), nighttime
thermal data are most suitable. Thermal daytime data from winter will also yield sufficient
results. Thermal daytime data from summer, however, are less appropriate. This is apparent
in Figure 10 (Tetzlaff 2004). For complete processing, it is best to use daytime data for
coal fire risk area demarcation and thermal nighttime data from the same period (week
or month of the same year) for thermal analysis. Such paired daytime/nighttime data are
available for all mentioned sensors.
Figure 10: Left: Thermal images of Wuda from ETM+ with extracted quantified thermal
anomalies adjacent. Top: Daytime summer scene; middle: Daytime winter
scene; bottom: Nighttime summer scene. Note that the number of thermal
anomalies extracted is highest in the nighttime scene. Coal fires with greater
heat release are found in the southern part of the Wuda syncline.
Right, for comparison: Extraction of thermal anomalies and their quantification
from BIRD data. Much fewer anomalies are extracted from these data of
coarser resolution (> 180 m), which in turn would lead to less accurate
quantification. However, coal fire energy release patterns from both sensors
generally coincide well.
376
Currently, first links between remote sensing-based coal fire quantifications and
geophysical measurements of burning intensity and propagation velocity are being forged.
The final objective is not only to establish a remote sensing-based monitoring system of
coal fires along the North China Coal Fire Belt, but also to derive quantitative statements
about the amount of coal burned in these fires. It is believed that remote sensing is a
powerful tool in investigations of Kyoto-relevant gas emissions and related instruments
such as the Clean Development Mechanism (CDM).
6 Conclusion
Coal surface expansion, as in new mines or new coal waste heaps or stockpiles, increases
coal fire risk. The regular observation of land cover changes may help detect new coal fires
at an early stage. An algorithm for the demarcation of coal fire risk areas also allows for the
extraction of areas with coal fire potential that have not yet developed pronounced thermal
anomalies. Automated detection of this kind can direct thermal anomaly monitoring toward
appropriate and relevant areas. On average, 80 % of the extracted anomalies from daytime
data and 50 % from nighttime data can be identified as false alarms. For demarcation as
well as for detection and quantification, the sensors EMT+ and ASTER do not show major
differences in results (Tetzlaff 2004; Zhang 2004; Künzer 2005). The sensor BIRD is often
too coarse for detection of small coal fires; like ETM+, it is no longer operational. However,
ASTER data are still available. Coal fire detection with MODIS data yielded promising
results as shown by Hecker and Künzer (2006). Even though MODIS has a lower spatial
resolution than e.g. BIRD, it offers more thermal bands, the great advantage of pre-dawn data
available on a daily basis, and an overall better sensor stability. Yet for coal fire quantification
(not detection!) this sensor’s resolution is too coarse. To estimate coal fire energy release,
data of more than 100 m in resolution are most suitable. Here, nighttime data outstrip daytime
data. Winter daytime data give a better contrast for coal fire detection and quantification than
summer data.
From a remote sensing perspective, coal fires are a very difficult object of analysis.
Extensive field validation data (see Künzer et al. 2008 in this volume) have proven that
only the hottest coal fires can be extracted automatically from remote sensing data.
“Colder” or smaller coal fires often go unnoticed. Consequently, the quantification of coal
fires based on automated thermal anomaly extraction is likely to underestimate the energy
released by these fires. This is all the more critical given the magnitude of the problem
– millions of tons of coal are burned away in coal fires each year. How much coal is lost
377
exactly depends on a multitude of factors: the burning volume underground, the burning
intensity of the fire, changing burning conditions such as ventilation and local weather,
etc. Coal fires in a precisely known area are therefore best approximated when all field
data (coal fire mappings, coal rank, combustion processes, etc.) and remote sensing data
are combined. To quantify the amount of burning coal in all Chinese coal fires – and thus
derive the amount of greenhouse gases released – is a very difficult task; with reasonable
accuracy based on scientific results it may be impossible. However, since the strength of
remote sensing is its transferability and wide spatial range covering large areas at low-cost
approximations, this task is not beyond probability.
References
Bhattacharya, A. and Reddy, C.S., 1994. Underground and surface coal mine fire detection in India’s
Jharia Coalfield using airborne thermal infrared data. Asian Pacific Remote Sensing Journal 7:
59-73.
Bhattacharya, A., Reddy, C.S., and Mukherjee, T.K., 1991. Multi-TIR remote sensing data analyses
for coal fire mapping in Jharia Coalfield of Bihar, India. In Proceedings of the 12th Asian
Conference on Remote Sensing, Singapore, 30 October – 5 November 1991. National
University of Singapore, Singapore, 221-226.
Cihlar, J., 2000. Land cover mapping of large areas from satellites: Status and research priorities.
International Journal of Remote Sensing 21(6,7): 1093-1114.
Cortijo, F. and Perez de la Blanca, N., 1997. A comparative study of some non-parametric spectral
classifiers. Applications to problems with high-overlapping training sets. International Journal
of Remote Sensing 18(6): 1259-1275.
Ellyett, C.D. and Fleming, A.W., 1974. Thermal infrared imagery of the Burning Mountain coal fire.
Remote Sensing of Environment 3(1): 79-86.
Fisher, W.J. and Knuth, W.M., 1968. Detection and delineation of subsurface coal fires by aerial infrared
scanning. Geological Society of America Bulletin 115: 67-68.
Friedl, M.A. and Brodley, C.E., 1996. Decision tree classification of land cover from remotely sensed
data. Remote Sensing of Environment 61(3): 399-409.
Greene, G.W., Moxham, R.M., and Harvey, A.H., 1969. Aerial infrared surveys and borehole temperature
measurements of coal mine fires in Pennsylvania. In Proceedings of the 6th International
Symposium on Remote Sensing of Environment, Institute of Science and Technology,
University of Michigan, Flint, Michigan, U.S.A.,13-16 October 1969, 517-525.
Guan, H.Y., 1989. Applications of remote sensing techniques in coal geology. Acta Geologica Sinica 2:
253-269.
of China Coal (ARSC), Xi’an, P.R. China and International Institute for Geo-Information
378
Gupta, R.P. and Prakash, A., 1998. Reflectance aureoles associated with thermal anomalies due to
subsurface mine fires in the Jharia Coalfield, India. International Journal of Remote Sensing
19(14): 2619-2622.
Hecker, C. and Künzer, C., 2006. Remote sensing-based coal fire detection with low-resolution MODIS
data. In Stracher, G. (Ed.), GSA book on coal fires. (accepted for publication in 2007)
Knuth, W.M., Fisher, W. Jr., and Stingelin, R.W., 1968. Detection, delineation, and monitoring of
subsurface coal fires by aerial infrared scanning. In Geographer. State College, Pennsylvania,
U.S.A., 877-881.
Künzer, C., 2005. Demarcating coal fire risk areas based on spectral test sequences and partial unmixing using
multi-sensor remote sensing data. PhD thesis, Vienna University of Technology, Vienna, Austria.
Künzer, C. and Voigt, S., 2003. Vegetationsdichte als möglicher Indikator für Kohleflözbrände.Untersuchung
mittels Fernerkundung und GIS. In Strobl, J., Blaschke, T., and Griesebner, G. (Eds.), Angewandte
Geographische Informationsverarbeitung XV. Beiträge zum 15. AGIT-Symposium Salzburg 2003.
Heidelberg, Germany: Wichmann, 256-261.
Künzer, C., Zhang, J.Z., Hirner, A., Jia, Y.R. and Sun, Y.L., 2008. Multi-temporal in-situ mapping of the
Wuda coal fires from 2000 to 2005: Assessing coal fire dynamics. In UNESCO Office Beijing
Li, D.X., 1985. Applied study of airborne remote sensing in Taiyuan Coalfield, China. Internal Report of
the Remote Sensing Center of the Ministry of Coal Industry, Xi’an, Shaanxi Province, P.R.
China. (in Chinese).
Mansor S.B., Cracknell, A.P., Shilin, B.V., and Gornyi, V.I., 1994. Monitoring of underground coal fires
using thermal infrared data. International Journal of Remote Sensing 15(8): 1675-1685.
Matson, M. and Dozier, J., 1981. Identification of sub-resolution high temperature sources using a thermal
IR sensor. Photogrammetric Engineering and Remote Sensing 47(1): 1311-1318.
Moxham, R.M. and Greene, G.W., 1967. Infrared surveys of coal mine fires in the anthracite and
bituminous fields, Pennsylvania, U.S.A. Geological Survey Inter-Agency Report BM-2. U.S.
Bureau of Mines (USBM), Washington, D.C., U.S.A.
Prakash, A. and Gupta, R.P., 1999. Surface fires in Jharia Coalfield, India – Their distribution and
estimation of area and temperature from TM data. International Journal of Remote Sensing
20(10): 1935-1946.
Prakash, A. and Gupta, R.P., 1998. Land-use mapping and change detection in a coal mining area – A case
study in the Jharia Coalfield, India. International Journal of Remote Sensing 19(3): 391-410.
Prakash, A., Fielding, E.J., Gens, R., van Genderen, J.L., and Evans, D.L., 2001. Data fusion for
investigating land subsidence and coal fire hazards in a coal mining area. International
Prakash, A., Gupta, R.P., and Saraf, A.K., 1997. A Landsat TM based comparative study of surface and
subsurface fires in the Jharia Coalfield, India. International Journal of Remote Sensing 18(11):
2463-2469.
Prakash, A., Saraf, A.K., Gupta, R.P., Dutta, M. and Sundaram, R.M., 1995. Surface thermal anomalies
with underground fires in Jharia Coal Mine, India. International Journal of Remote Sensing
16(2): 2105-2109.
379
Rabchevsky, G.A., 1972. Determination from available satellite and aircraft imagery of the applicability
of remote sensing techniques to the detection of fires burning in abandoned coal mines
and unmined coal deposits located in north-central Wyoming and southern Montana. U.S.
Department of the Interior, Bureau of Mines, Contract No. SO 211087, Final Report No.
8G86-F, prepared by Allied Research Associates Inc.
Saraf, A.K., Prakash, A., Sengupta, S., and Gupta, R.P., 1995. Landsat TM data for estimating ground
temperature and depth of subsurface coal fires in the Jharia Coalfield, India. International
Slavecki, R.J., 1964. Detection and location of subsurface coal fires. In Proceedings of the
3rd International Symposium on Remote Sensing of Environment, Institute of Science and
Technology, University of Michigan, Flint, Michigan, U.S.A., 14-16 October 1964, 537-547.
Tetzlaff, A., 2004. Coal fire quantification using ASTER, ETM and BIRD satellite instrument data. PhD
thesis, Ludwig Maximilians University Munich, Munich, Germany.
Zhang, J.Z., 2004. Spatial and statistical analysis of thermal satellite imagery for extraction of coal fire
related anomalies. PhD thesis, Vienna University of Technology, Vienna, Austria.
Zhang, X.M., 1998. Coal fires in northwest China – Detection, monitoring, and prediction using remote
sensing data. PhD thesis, International Institute for Geo-Information Science and Earth
Zhang, X.M., van Genderen, J.L., and Kroonenberg, S.B., 1997. A method to evaluate the capability of
Landsat 5 TM band 6 data for sub-pixel coal fire detection. International Journal of Remote
Sensing 18(15): 3279-3288.
Zhang, X.M., Zhang, J.Z., Künzer, C., Voigt, S., and Wagner, W., 2004. Capability evaluation of 3-5 µm
and 8-12,5 µm airborne thermal data for underground coal fire detection. International
380
Automated Detection and Extraction of
Surface Cracks from High-Resolution
Quickbird Imagery
Yang Bo1, Li Jing2, Chen Yunhao2, Zhang Jianzhong3, and Claudia Künzer3
1
College of Resources and Environment Science, Hunan Normal University,
Changsha, Hunan Province, P.R. China
2
College of Resources Science and Technology, Beijing Normal University,
Beijing, P.R. China
3
Vienna, Austria
Abstract
Wuda Coalfield in Inner Mongolia Autonomous Region is a typical coal fire area of northern China.
Surface cracks from coal fires and mining operations are ubiquitous there. Using high-resolution
Quickbird satellite imagery with a number of edge detection operators and different convolution kernel
sizes in this study confirmed that Laplacian edge detection with Gaussian filtering (LoG) is the best
method to detect linear features, i.e. surface cracks, in images of coal fire areas. For automated detection
and extraction, the following techniques were used after LoG: (i) image histogram stretching, (ii) raster to
vector conversion, and (iii) removal of disturbances caused by ground objects such as roof edges, streets,
and shadows of sand dunes. In-situ validation proved the usefulness of this method.
摘要
内蒙古乌达煤田是中国北方典型地下煤火区之一，由于地下煤火和采煤活动导致的地表裂隙几乎
随处可见。研究采用高空间分辨率Quickbird卫星影像，在尝试多种边缘检测算子和卷积核大小对
遥感影像进行处理的基础上，发现“拉普拉斯边缘检测+高斯滤波”是最适合本区地表裂隙等线
性要素提取的方法。通过：（1）影像灰度拉伸，（2）栅格转矢量，（3）地物干扰因素（如房屋
边缘、街道和沙丘阴影）去除等处理，最终自动检测和提取出了地表裂隙。实地验证表明该方法
效果较好。
381
1 Introduction
In the last decades, a great number of underground coal seams in northern China have been
affected by coal fires. The pressure from the soil above the burned seams caused gradual
collapse. Countless cracks and faults ranging in width from a few centimeters to about 5 m
developed underground and at the surface. Under certain circumstances, these cracks and
faults led to serious geological hazards, such as collapse, landslide, and land subsidence;
they destroyed traffic lines, mining facilities, and residential areas; and caused land
degradation to a degree unrecoverable for years.
Surface cracks are indicators of underground coal fires and mining activities. High-
resolution satellite imagery is commonly applied for their optical identification.
Manual feature extraction depends on the interpretation skills of the operator and can
be time-consuming, and thus very expensive (Baumgartner et al. 1999). The success
of linear feature extraction is above all a function of spatial resolution (Wang & Zhang
2000). In high-resolution imagery, roads are often displayed as elongated features
(Agouris et al. 2001). Penn and Livo (2002) hypothesized that improvements in spatial
resolution would provide the basis for an automated detection of linear features,
impossible to perform using the digital imagery commercially available to date. Most
research, though, focused on road networks, not on cracks. In this study, a number
of edge detection operators and different convolution kernel sizes were applied to
assist the automated detection and extraction of surface cracks from high-resolution
Quickbird imagery.
2 Study Area
Wuda Coalfield is located in the south of Inner Mongolia, on the west bank of the
Yellow River, north of the Helan Shan mountain range. Its geographical location is
39°27’ to 39°34’ N latitude and 106°34’ to 106°38’ E longitude. The Wuda syncline
is oval in shape and runs about 13 km from north to south, covering an area of about
200 km2, with Suhaitu Coal Mine in the north, Wuhushan in the south, and Huangbaici
in the center (Figure 1). Elevation in this region ranges between 1,122 to 1,350 m
above sea level.
382
Figure 1: Study area in Landsat ETM+ image, Band 741 RGB, 21 September 2002
3 Data
High-resolution Quickbird imagery has five spectral bands. Bands 1 to 4 belong to

the visible and near-infrared regions of the electromagnetic spectrum; in addition, a
panchromatic band is recorded. The technical parameters of the Quickbird data sets used in
this study are listed in Table 1.
383
Table 1: Parameters of Quickbird data sets used in this study

Acquisition Spectral Band Spectral Range Spatial Resolution
Time (μm) (m)
Band 1 0.45-0.52 2.44

Oct. 2003 Band 2 0.52-0.60 2.44
Band 3 0.63-0.69 2.44
Sept. 2004 Band 4 0.76-0.90 2.44
Panchromatic 0.45-0.90 0.61
4 Methods and Steps
Surface cracks in high-resolution satellite imagery appear as threadlike dark shadows on

the computer screen, with edges standing out from the background area. These edges are
always lying between objects and their background, or between different objects, regions,
and colors – a necessary basis for image analysis such as image segmentation. Edge
detection, hence, is a good method for automated extraction of surface cracks from remote
sensing imagery.
Park et al. (2002) summarized common edge detection methods for linear feature
extraction in remote sensing. Edge detection includes four steps: (i) image filtering,
(ii) edge enhancement, (iii) edge detection, and (iv) edge location. The first three steps
are widely applied these days. Edge detection mainly relies on three central technical
parameters/devices: (i) smooth filter image fit, (ii) convolution kernel size, and (iii) edge
detection operators. In the last two decades, a number of edge detection operators were
developed and released, e.g. Laplacian, Roberts, Sobel, and Prewitt.
4.1 Filters and operators
Filters and operators in this study were tested on a small subset that was cut out by
ERDAS8.5 from a high-resolution Quickbird panchromatic image of coal fire zone (FZ)
3 of Suhaitu Coal Mine (Figure 2). In a first trial, edge detection with Sobel and Prewitt
operators was performed on the original image, and on the image after filtering with
a 7 × 7 Gaussian smooth filter. The cracks extracted in both cases were unrealistically long,
so that no further tests were conducted with these two operators.
384
Figure 2: Comparison between Sobel (left column) and Prewitt (right column) edge
detection on Quickbird original image (top) and the same image after filtering
with a 7 × 7 Gaussian smooth filter (bottom). The middle column shows the
image without edge detection.
Marr and Hildreth (1980) ran the Laplacian edge detection operator on a Gaussian filtered
image to sieve off image noise before edge enhancement, thus creating the so-called
LoG operator (Laplacian of Gaussian). Attempts to use this operator for linear feature
identification and extraction on Quickbird imagery yielded poor results. Extraction was
very coarse and brought out only very large cracks (Figure 3). Gunn (1998) described
similar problems with this operator.
Gaussian filtering and Laplacian edge detection nonetheless have been proven to be
good, precise, and useful methods in numerous studies. Consequently, experiments with
different combinational sequences and convolution kernel sizes were carried out after the
first abortive trials to still realize automated linear feature (crack) detection and extraction
based on these two methods.
385
Figure 3: 5 × 5 LoG edge detection on Quickbird unfiltered image
4.2 Convolution kernel sizes
Surface cracks from coal fires and mining operations are limited in length and width, and
hence rather difficult to extract from images of coarse spatial resolution. For the LoG
trials of this study, therefore, only the high-resolution Quickbird panchromatic band and
small convolution kernel sizes were used. Gaussian smooth filtering and Laplacian edge
detection were downsized to 3 × 3. Comparative analyses proved that Laplacian edge
detection before Gaussian filtering can preserve image information for even very slight
edges.
4.3 Steps
Repeated testing led to the following sequence of steps for optimal detection results
(Figure 4):
1) 3 × 3 kernel size Laplacian edge detection;

2) 3 × 3 kernel size Gaussian smooth filtering;
3) Image histogram stretching;
4) Raster to vector (R2V) conversion;
5) Manual removal of pseudo-cracks such as streets, roof edges, and sand dune shadows
by mask treatment.
386
Figure 4: Steps of automated linear feature (crack) detection. 1) Original Quickbird

panchromatic image. 2) 3 × 3 Laplacian edge detection. 3) 3 × 3 Gaussian
smooth filtering. 4) Image histogram stretching. 5) R2V conversion before
stretching and 6) after stretching.
Image histogram stretching is of great importance for R2V conversion. Without stretching,
results are poor and basically useless for further processing (cf. pictures 5 and 6 of Figure 4).
5 Results and Validation
Two methods were used for validation of results in this study: (i) in-situ measurements
and crack mapping, as well as (ii) comparisons with results from optical interpretation.
Different sections of the study area were selected for validation. All in all, the 5-step LoG
method proved effective and useful for linear feature identification in high-resolution
remote sensing images.
387
Figure 5 is a Quickbird panchromatic image (pixel size 61 cm) of a big fault extending
from west to east across the study area. Automated detection and extraction of linear
features here produced high-accuracy results with more than 90 % of cracks identified.
Figure 5: Automated extraction of cracks (red lines; right) from a high-resolution

Quickbird panchromatic image (left)
6 Discussion and Conclusion
Automated detection and extraction of surface cracks from high-resolution Quickbird

imagery as described in this study can speed up image interpretation and save time, cost,
and manpower in coal fire detection. It is suitable for rapid investigation of linear features
such as surface cracks and roads. The resulting vector maps can be used for long-term
monitoring and detection of coal fire propagation. Tests showed that the method can
function as an easy-to-use and very effective tool in or supplement to high-resolution
image interpretation.
Time series of high-resolution remote sensing images are expected to enhance separation
of coal fire-related cracks from other linear features in crack detection. Further testing
is planned to identify more arithmetic operators for edge detection and linear feature
extraction. In general, ongoing research should focus on methodological improvements.
Methods that will allow for a more reliable assessment of accuracy in crack detection and
extraction are needed, especially so with regard to times series analysis.
388
Acknowledgements
This study was supported by the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North
China” (contract no. 2002DFG00032) and the Hi-Tech Research and Development
Program of China (contract nos. 2002AA130020 and 2003AA131100). Our sincere
thanks go to Dr. Stefan Voigt of the German Aerospace Center (DLR) for supplying the
Quickbird data used in this study.
References
Agouris, P., Doucette, P., and Stefanidis, A., 2001. Spatiospectral cluster analysis of elongated regions in
aerial imagery. In Proceedings of the IEEE International Conference on Image Processing,
Thessaloniki, Greece, 07-10 October 2001, 789-792.
Baumgartner, A., Steger, C., Mayer, H., Eckstein, W., and Ebner, H., 1999. Automatic road extraction
based on multi-scale, grouping, and context. Photogrammetric Engineering and Remote
Sensing 65(7): 777-785.
Gunn, S.R., 1998. Edge detection error in the discrete Laplacian of Gaussian. In Proceedings of the IEEE
International Conference on Image Processing, Chicago, Illinois, U.S.A., 4-7 October 1998,
515-519.
Marr, D. and Hildreth, E.C., 1980. Theory of edge detection. In Proceedings of the Royal Society of
London, London, United Kingdom, 207, 187-217.
Park, J.S., Saleh, R.A., and Yeu, Y., 2002. Comprehensive survey of extraction techniques of linear
features from remote sensing imagery for updating road spatial databases. In Proceedings of
the ACSM-ASPRS Annual Conference, Washington, D.C., U.S.A., 21-26 April 2002. American
Society for Photogrammetry and Remote Sensing (ASPRS), Bethesda, Maryland, U.S.A.
(unpaginated CD-ROM)
Penn, B.S., and Livo, K.E., 2002. Using the semi-automated plug-in tool for feature identification,
recognition, and extraction (SPITFIRE) to extract roads and surface material types from
AVIRIS imagery. In Proceedings of the AVIRIS Airborne Geoscience Workshop, Pasadena,
California, U.S.A., September 2005
Wang, J. and Zhang, Q., 2000. Applicability of a gradient profile algorithm for road network extraction:
Sensor, resolution, and background considerations. Canadian Journal of Remote Sensing
26(5): 428-439.
389
4-Level High-Resolution Detection and
Monitoring – Applications in Coal Fire
Detection and Management
Zhang Jianmin1, Xiang Yaling2, Wang Mei2, and Huan Zhongdan3
1
Shenhua Group Corporation Limited, Beijing, P.R. China
2
3
School of Mathematical Sciences, Beijing Normal University, Beijing, P.R. China
Abstract
Great efforts have been made in China over the last years to fully understand and mitigate the ecological
and economic disaster that are coal fires; many significant results have been achieved. With an ever
greater demand being placed on data accuracy, exploration depth, and rapid information flow, it has
become apparent that high-resolution remote sensing and high-precision geophysical methods are most
advantageous for coal fire detection and management. A systematic summary of all existing methods,
integrated into a comprehensive operational system, was long overdue.
Building on state-of-the-art technology, and with due regard to the general objectives of coal fire fighting
and management in China, the present paper establishes a new 4-Level High-Resolution Detection and
Monitoring System (4LHDMS). In a first step, it develops a 4-level detection system based on high-
resolution remote sensing, GIS, and geophysical investigations. This system is then refined by adaptation
of its various methods to the physical characteristics and different stages of coal fire development. In
a second step, 4LHDMS is applied to a new layout of hierarchies in coal fire management, reassigning
duties and functions at the different ranks of the administrative ladder. As such, the paper is of eminent
significance for management, decision, and design processes in coal fire fighting at government level.
摘要
近年来，人们围绕地下煤火引发的环境灾害与经济影响的研究已做了大量的工作并获得了许多值得
借鉴的成果。然而，随着煤火管理与灭火工作对于探测与监测技术和方法的要求日益提高（精度要
求、深度要求、信息提供、灭火工程），高精度遥感技术和高精度地球物理技术在煤田火区的研究
中显示出优势。显然，在研究和实践的基础上进行系统的总结和应用提升是十分必要的。
390
本文针对地下煤火的管理与治理目标，通过总结近年来的煤火探测成果，面向煤火管理目标，建
立了基于煤火发展地质模型、高精度探测方法、三维GIS可视化技术的高精度四层空间探测与监
测体系（4LHDMS）。一是研究了基于高精度方法（遥感、GIS、地球物理技术）的四层空间探
测系统；二是依据地下煤火发生的不同阶段（煤田火区发生阶段、煤田火区发展阶段、熄灭阶
段）的特征，不同管理目标、不同方法技术特点，按照经济、有效的原则建立了“系统”体系下
的适用型解决方案，通过简要实例说明地下煤火的高分辨率四层空间探测体系方法效果；三是基
于政府的煤火管理目标（煤田火区调查阶段、煤田火区治理阶段、煤田火区复垦阶段），结合国
家不同比例尺的灾害调查，提出了针对不同阶段和比例尺的煤火调查方法和适用于政府管理目标
和矿区灭火工程的不同技术方案。本文对于政府管理与决策和灭火工程规划与监测具有重要的指
导意义。
1 Introduction
China has abundant coal resources, but also suffers from the world’s greatest number
of coal fires. Much progress has been made over recent years in coal fire research, and
Chinese fire fighters now have a lot of practical experience at hand to help them tackle
their formidable task. Nonetheless, studies conducted in northern China’s notorious coal
fire areas of Shaanxi (ARSC 1989), Xinjiang (Guan & van Genderen 1997), Ningxia
(Guan & Zhang 1999), and Inner Mongolia (BRSC 2006) all had to contend with the
same dilemma – how establish a high-resolution detection and monitoring system of
underground coal fires combining modern techniques with the hands-on knowledge of
experienced mine personnel? Such a system could be a precursor to early warning systems
in the development of large coalfields, or even serve as the backbone of a fire fighting
project for the whole of China.
The path that coal fire detection has taken so far, the progress it has made with respect to
resolution and methodology, can be described in three successive phases:
1) Surface detection mainly using magnetic and electromagnetic field surveys to identify
boundaries of coal fire zones.
2) 3-level low-resolution (80 to 120 m) detection using multi-spectral scanning, aerial

thermal infrared scanning, and ground magnetic and apparent resistivity surveys to
identify coal fires.
3) 4-level medium-resolution (3 to 30 m) detection using Spot, IRS, and ETM satellite

data, as well as aerial thermal infrared scanning, and ground magnetic and apparent
resistivity surveys for detailed mapping of coal fire areas. A first model representing
this 4-level detection was established by Guan and Zhang (1999) and further developed
by Zhang and Guan (2004).
391
Three geological models of coal fire development are currently applied in Chinese coal
fire research, namely burned rock analysis and interpretation, a four-stage development
model by Guan and Zhang (1999) and a 12-stage development model by Zhang and
Guan (2005).
Coal fire detection in this context means that the geographical extent of a coal fire and its
characteristics (such as combustion center(s) and burning intensity) are identified by means
of in-situ investigations, remote sensing, and lab-scale modeling and simulation. A coal
fire detection system specifically refers to modes of data collection that combine for their
end-users all, or a selection of, methods applied in coal fire detection – a package solution
so to speak. Such a system is central to coal fire research and fighting as it links detection
methods with (geological) models for data interpretation. The system described in this
paper addresses all levels of coal fire management.
2 Chinese Coal Fire Research – The Status Quo
Coal fire research is a very intricate task. It involves a whole gamut of different working
modes and disciplines, ranging from basic research on coal seam combustion, over
detection and monitoring, to fire fighting design and management, both financial and
political. Despite recent advances in technology and R&D, many facets of the problem
have yet not been adequately understood and are awaiting scientific solution. Evidently,
research has to go a long way before coal fires can be controlled effectively.
2.1 Previous achievements
Surveys of coal fires and their environmental impact have been conducted in China
starting with the introduction of remote sensing technology and its strategic development
on a national scale in the early 1980s to late 1990s. Coal fire areas were investigated using
remote sensing in Xishan of Shanxi Province (ARSC 1989), Liuhuanggou of Xinjiang
Autonomous Region (Guan & van Genderen 1997), Rujigou of Ningxia Autonomous
Region (Guan & Zhang 1999; Rosema et al 1999), and Wuda of Inner Mongolia
Autonomous Region (Zhang & Guan 2005). A high-resolution map of active coal fires in
Rujigou Coalfield was obtained through aerial hyperspectral imaging (Wan & Yan 2003).
Dynamic monitoring of coal fires was tested in Wuda (Zhang & Guan 2005). All in all, the
accuracy of coal fire mapping was improved considerably by combining high-resolution
remote sensing techniques with mining engineering.
392
In the wake of these investigations, a GIS–based management system for fire fighting was
established and continuously improved along the progress in GIS technology, i.e. from
2D to 3D technology and from vector data to image-vector data. This system spawned a
number of other comprehensive coal fire-related management schemes, e.g. a LWIS-based
information system for data management in coal fire investigation (Guan & van Genderen
1997), the so-called ILWIS-based Coalman system for data management in coal fire
investigation and interpretation (Guan & Zhang 1999; Rosema et al. 1999), as well as the
Micromine-based Coalminer3D (Zhang & Guan 2005), a system that integrates various data
both for coal fire research and mining activities.
The accuracy and sensitivity of coal fire detection was greatly enhanced by means of high-
resolution spaceborne remote sensing (Zhang & Guan 2005; DLR 2006). The structure
of Quickbird images, for example, allowed analysis of underground combustion systems
in combination with their corresponding surface features, such as cracks and fissures.
12 types of features in Quickbird images were defined and equated with different sub-
stages of coal fire development, which were then ordered into a 12-stage combustion
model (Zhang & Guan 2005).
Thermal data interpretation was improved to facilitate surface temperature acquisition

from thermal images. Thermal anomalies at the surface and thermal fields underground
have always been of particular interest to coal fire researchers. Based on anomaly data
processed from both satellite and aerial thermal infrared images, thermal conductivity
imaging was enhanced by introducing a wavelet algorithm for data interpretation. Thermal
anomaly areas can now be acquired from surface temperature profiles, in a model-based
process called 3D thermal imaging (Zhang 1998, 1999a,b). This further improved thermal
image interpretation for coal fire research and fighting.
Enhanced detection methods also promoted progress in modeling. In the last years, a
geological model of coal fire development and a physical model for data interpretation
and dynamic coal fire monitoring were developed in China. The geological model
evolved progressively from a basic understanding and interpretation of burned rock to
a 4-stage model of coal fire development, and finally to a 12-stage combustion model,
informed by new insights into the physical and chemical processes surrounding coal fire
combustion.
2.2 Current challenges in coal fire management
Coal fire management as a term refers to financial investment, control, and supervision
of coal fire research by the Government or corporate bodies, namely in organizing coal
393
extraction and fire fighting. Coal fire management extends from the general public arena to
the funding and organizing of local extinction campaigns. In order to clarify and streamline
hierarchies and responsibilities between the Government and its corporate counterparts, it
is necessary to place coal fire management within governmental hierarchies and existing
networks of resource protection and disaster mitigation.
The set-up of a technological support system is instrumental in linking up coal fire research
with government management. This system would integrate various methods of coal fire
detection and fire fighting, with a clear focus on technological requirements and efficient
workflows so as to ensure informed decision-making. Recent results in coal fire research
promise to improve the practicability of such a system, allowing for a thorough integration
of geological models and methods of coal fire detection, data interpretation, and fighting
into a comprehensive overriding structure of coal fire management.
2.3 Current challenges in coal fire research and detection
Mainly two factors determine the success and effectiveness of coal fire investigations –
detection accuracy and depth. Detection accuracy represents the level and exactness of
data that are measured/produced on site. The higher the detection accuracy and the larger
the investigation scale, the more detailed the data obtained for a particular coal fire, and
thus the more focused and successful the ensuing fire fighting operations. The resolution
of thermal infrared imagery in Landsat ETM and aerial scanning is 45 m and 2 m,
respectively, while that of Quickbird is 0.6 m.
Detection depth indicates the vertical scope of the technology used; i.e. how far into
the ground signals and sensitivities reach. It is one of the most important parameters for
coal fire monitoring, and yet remains a central problem in fire detection and fighting.
High-accuracy detection data and extended investigation depth are indispensable for 3D
visualization of coal fires. They are the basis for a more comprehensive understanding of
coal fire development and its mechanisms.
3 4-Level High-Resolution Detection and Monitoring
In order to enhance coal fire detection accuracy and depth, and, along with it, coal fire
fighting in the long run, it is crucial to foster R&D on coal fires. Doing so requires a
combination of various methods, based on precise tasks and objectives, and with due
394
regard to technical feasibility and cost, so that the most effective approach and operational
scenario be chosen.
3.1 4-Level Detection and Monitoring System (4LDMS) for coal fire
research
Underground coal fires are dynamic 3D structures, which can be observed in and described
by various physical and chemical parameters and processes. Coal fire detection can
possibly span four different levels (platforms) of investigation, viz. space- and airborne,
as well as ground and underground investigations, each of which has its own detection
methods and specific modes of information.
The 4-Level Detection and Monitoring System(4LDMS) for coal fire research developed
by Zhang and Guan (2004) is a practical technology scheme that integrates different
kinds of coal fire data obtained from the four detection platforms as well as a vast
range of means and methodologies into a GIS-based information system. The space-
or satellite-borne detection platform provides remote sensing information on radiation
(light), including visible, near-infrared, and thermal infrared bands. Allowing for easy
data processing, with a large detection scope, but low accuracy, it is mainly used for
coal fire surveys. The airborne detection platform produces data on visible light, near-
infrared, and thermal infrared radiation, as well as on magnetic and electromagnetic field
activity. It mainly serves to delineate combustion centers and zones, regions of thermal
anomalies, and coal fire areas as opposed to unaffected areas. In particular, aerial
thermal infrared imaging is useful for fine-combing coal fire areas and assessing fire
extinction. The ground detection platform uses various surface detection methods and
technologies such as portable remote sensing, geophysics, and borehole measurements.
Its data cover quantities such as visible light, temperature, as well as magnetic and
electromagnetic field activity. Ground detection methods are employed to gather detailed
information on individual coal fires, e.g. the depth and size/boundaries of a combustion
system. The underground detection platform relies on remote sensing investigations of
mine infrastructure and workings, as well as gas measurements. These give information
on self-ignition temperatures (SITs) and temporal variations of gas emissions, which
allow for detection and monitoring of early spontaneous combustion. 4LDMS integrates
all the above data into a GIS-based system, adding in geological and mining data, so as
to illustrate the spatial relationships that govern a coal fire. Coalman is such a system
(Guan & Zhang 1999). Figure 1 shows the detection methods used on the four different
platforms of 4LDMS.
395
Figure 1: 4-Level Detection and Monitoring System (4LDMS) for coal fire research
3.2 4-Level High-Resolution Detection and Monitoring System

(4LHDMS) for coal fire management
The 4-Level High-Resolution Detection and Monitoring System (4LHDMS) for coal fire
management is an optimized version of 4LDMS, integrating the latest geological models,
several high-resolution detection methods operational in the field, and 3D-visualization
software for data management and interpretation. The system is an optimal combination
of these tools with respect to coal fire management and its requirements. It features the
following important improvements on 4LDMS:
• High-resolution remote sensing: Quickbird data greatly raise the surface resolution
for identifying crack structures. High-resolution ground thermal infrared imaging
enhances the rate and accuracy of monitoring. High-resolution topography helps the
understanding of coal fires and their underground structure. All of these methods raise
detection accuracy, and improve the applicability of 4LDMS.
• Modeling and interpretation: The introduction of concepts such as “combustion

centers,” “combustion systems,” and “coal fire transmission zones” in 4LHDMS
(Zhang & Guan 2005) provides a comprehensive basis for model description of coal
fire development, identification of geophysical methods best suited for effective
detection, and explanation of data obtained in the context of individual fires.
396
• 3D visualization: GIS-based software in 4LHDMS enables viewing and assessment

of coal fires in their immediate environment, i.e. together with coal seams, and mine
infrastructure and workings. Supported by programs such as Coalminer3D, detection
data have been integrated to give fire site managers a clear picture of the situation
underground.
Various methods to detect the geological, geophysical, and geochemical alterations caused
by underground coal fires have been integrated in 4LHDMS; they compliment each other
for effective measurements along a fire’s vertical and horizontal axes. 4LHDMS works
with the following technical parameters:
• Spatial parameters:
o Horizontal resolution: 0.6 m, determined by Quickbird;
o Separability of depth below surface: < 3 m, determined by high-precision
electromagnetic surveying;
• Physical parameters:
o Separability of temperature T: 0.2 °C;
o Magnetic intensity M: < 1 nT;
o Separability of resistivity ρs: 10 Ωm.
3.3 High-resolution detection and its benefits
The high-resolution methods used in 4LHDMS are of better accuracy, performance, price
ratio, and applicability than regular methods in coal fire detection, monitoring, and fighting.
Figure 2 illustrates various system applications from high-resolution coal fire detection and
ground thermal scanning to gas modeling and 3D visualization. Figure 2a is a Quickbird
image of a combustion center highlighting surface features such as rock structures,
lithological classes, coal seam outcrops, ground vegetation, and soil sub-regions. It can be
used for identifying mined-out areas and crack zones, as well as areas of coal seam roof
collapse and land subsidence. Surface features of combustion centers are also discernable.
Figure 2b is an image of ground thermal anomalies obtained from high-resolution aerial
thermal infrared scanning. The fire outline is apparent. Figure 2c depicts the structures of a
ground combustion system, including combustion centers and crack zone. Figure 2d shows
the oxygen (O2) distribution in a mine working derived from gas measurements. Figure
2e is a 3D visualization of a coal fire and its environment; it includes the underground
combustion system as well as mine infrastructure and workings.
397
Figure 2: Main detection methods in 4LHDMS. a) Quickbird image of combustion

center (Guan 2005). b) Combustion zone in aerial thermal infrared imaging
(Xigou-Dafeng, Rujigou Coalfield, 1997). c) Thermal anomaly zone as
obtained from ground thermal imaging (Fire zone 8, Wuda Coalfield, 2004).
d) O2 distribution in mine working (Daliuta Mine, Shaanxi, 2005). e) 3D
visualization of coal fire with various components of mine infrastructure
(Suhaitu Mine, Wuda Coalfield, 2006).
All these methods contribute directly to the understanding of spontaneous combustion in

coal seams. Aerial thermal infrared imaging is a well developed detection and monitoring
method for coal fires; it is mainly used for identifying ground thermal anomalies of low
to high temperatures. A large-scale isothermal map (1:5,000 to 1:500) of a coalfield, in
most cases, is sufficient to spot and delineate coal fire combustion centers and systems,
as well as crack and fissure zones. High-resolution ground thermal infrared imaging can
provide information promptly at any given interval. Continuous gas measurements with
thermal detectors report dynamic changes of coal fire indicator gases in mine systems.
3D visualization with GIS allows for combined analysis and interpretation of ground and
underground information.
398
4 Applications of 4LHDMS
A target-oriented detection and monitoring system of underground coal fires should be

governed by the principles of applicability, practicability, reliability, and economy; it
should offer an array of complementary detection methods easily adaptable to different
settings and users. In the following, two possible applications schemes of 4LHDMS in
respect of coal fire development and management are presented.
4.1 4LHDMS in the different stages of coal fire development
This scheme is based on a reorganization and optimization of detection methods in

4LDMS, adapted to the life cycle of a coal fire and the accompanying changes in its
measurable parameters. The 4-stage model suggests itself here for reasons of detectability
(Guan et al. 2005; Zhang & Guan 2005). Obviously, there exists a distinct relationship
between detectable anomalies of a coal fire and its stage of development. Surface cracks,
vegetation changes, thermal radiation, magnetic field anomalies, gas emissions, and other
parameters were combined in a 3D model for coal fire detection and data interpretation to
indicate these stages (Zhang 1998). Figure 3 shows methods that have proven particularly
efficient for parameter measurement and surveying in 4LHDMS.
Figure 3: Detection methods in 4LHDMS according to coal fire combustion stages
399
The 4-stage coal fire development model provides a good basis for an ordered sequential
description of the dynamic geological and physical features of a coal fire and their
detection in the system:
1) Self-ignition (1st stage)
Weathering is at the beginning of spontaneous combustion in a coal seam. In this stage,

the coal seam starts to oxidize at temperatures lower than 35 °C – in geological terms:
a weathering zone develops with spontaneous combustion tendency. The coal adsorbs
oxygen from the ambient air and oxidizes in a heat-generating process. If the amount of
heat generated exceeds the coal’s capacity for dissipation, heat can build up inside the coal
and temperature rises. As a result, water contained in the coal and its immediate vicinity
starts to evaporate and rises from the seam as white steam. By the time a temperature
level of 110 °C is reached, the coal has dried out completely and pyrogenation accelerates,
which further increases temperatures. At this point, the coal seam starts to emit grey
smoke. At its SIT, the coal ignites with bursting flames or black smoke. Spontaneous
combustion of a coal seam is visible at ground and underground levels, and in the mine
system (if existent). Owing to differences and instabilities in oxygen adsorption and
supply, as well as temperature variations, fire spots along the weathering zone flare up and
die down alternately at the very beginning of spontaneous combustion. With increasing
temperature, the number of these spots increases. However, they remain in a state of
alternation until the critical moment of combustion center formation.
According to field surveys investigating the relationship between coal fire-induced

variations in rock and coal strata and the responses they entail in geophysical (i.e.
magnetic, electrical, dielectric, and thermal) fields, the anomalies that a coal fire
produces in the self-ignition stage appear as scattered weak point-sources, without
much influence on regular fields. The research objective in this stage, hence, must be
to develop ways and methods that can predict a coal seam’s potential for spontaneous
combustion, e.g. based on an analysis of coal general conditions and composites, of
environmental indicators, ventilation, and changes in gas emissions. It is advisable to
use highly sensitive detection methods in this stage.
Surface detection should focus on thermal anomaly monitoring in cracks using

high-precision thermal imaging, and 3D modeling based on thermal conductivity
measurements. This would allow precise statements on the depth, range, and
development of a fire. All data generated can be used in 3D dynamic monitoring and
guide the design of extinction efforts.
400
2) Oxidation and combustion (2nd stage)
The second stage, during which temperatures rise continuously within the seam, is
the actual stage of early fire development. Cracks caused by spontaneous combustion
enlarge and grow toward the surface, thereby improving conditions for combustion
(ventilation). Slowly, combustion centers form and link up to a full-fledged combustion
system. During the oxidation and combustion stage, gas discharge and oxygen uptake
of each combustion center are roughly of the same order, so that in effect a circulative
oxidation/combustion system forms, which further propels fire development. From this
point onwards, the intensity with which gases (such as carbon dioxide monoxide) are
produced and emitted increases drastically.
The oxidation and combustion stage is the primary formation stage of a coal fire.
Burning intensity has not yet reached a measurable scale and separate combustion
systems appear with cracks as their characteristic surface features. The coal fire-induced
thermal, magnetic, and gas anomalies can be measured at surface level by geophysical
and geochemical means. Research here must aim to determine the spatial distribution
and exact stage of the fire. The quickest and most effective way to get to a systematic
understanding of a fire in the dimensions of a whole coalfield is to use high-resolution
Quickbird data to analyze the relationship between fire and surface crack systems, and
high-sensitivity gas detectors to investigate the composition of gas discharge so as to
confirm the fire’s development stage. Thermal anomalies at the surface can be delineated
with high-precision ground thermal imaging, which may also prove helpful in finding
connections and pathways between crack and combustion systems. Gas measurements
can shed light on the location, extent, and development trend of a coal fire in a mine
system.
3) Combustion and expansion (3rd stage)
During the combustion and expansion stage, heat accumulates gradually in the coal
seam. The overlying bedrock is in a state ranging from low-temperature burning to
high-temperature meltdown. The fire slowly eats away the coal, depleting the seam and
leading to roof collapse where the coal body is no longer able to hold the weight of the
overlying strata. Cracks and fissures, as well as burned rock holes, appear in greater
density at the surface, and grow in length and width. They provide effective pathways
for oxygen uptake and gas and heat emission, which in itself intensifies combustion and
drives the fire deeper into the ground, to the very limbs of the coal seam. By this time, a
401
heat-induced difference in pressure between the underground coal body and the surface
has built up; it connects single combustion systems to a whole combustion zone, mainly
through heat convection in cracks. At such a state of maturity, the coal fire is clearly
visible at the surface through smoking cracks and fissures above the combustion centers,
and collapse zones where the coal seam has already been exhausted and caved in.
High-resolution (Quickbird) satellite data should be used at this stage to investigate the
relationship between underground combustion and surface crack systems. Combustion
zones can be delineated using high-precision aerial thermal infrared imaging. The exact
location of a combustion system and its centers can be determined with high-precision
ground thermal imaging. High-precision magnetic methods can serve to determine the
depth of a fire. By adding mine observations and results from spatial investigations
of mine areas to ground surveys and gas measurements, it is possible to evaluate and
project distribution, extent, and development of a coal fire at this stage.
4) Extinction (4th stage)
When a burning coal seam is nearing exhaustion, the coal fires slowly dies down.
Temperatures at the beginning of this process, however, are still very high and can
effectively burn the overlying bedrock from sedimentary to burned rock, a process
called pyrometamorphism. With falling temperatures, pyrometamorphism produces
very distinct minerals and compounds, namely: corundum (2,400 °C); lava breccia
(1,600 to 1,200 °C); burned dandered coal (1,000 °C); vitrified and vitreous burned
shale (800 °C); burned sandstone (600 °C); baked loess (300 °C); as well as mirabilite
and sulfur crystallization druses (150 to 110 °C). Also, the overlying bedrock can
break and collapse, creating large subsidence zones or burned fissure clusters. As these
extensive and at times very drastic shifts and ground movements expand crack zones
and thus air flux to the seam, residual coal may re-ignite, or dormant fire spots flare up
to propagate further down toward the water table. Below this line, coal and bedrock
strata are saturated with water and hence sealed from surface air and combustion.
The extinction stage is the last stage of coal fire development. In the ever-changing
environment of a coal fire, combustion oscillates between states of continuous burning,
blazing, and gradual extinction. The magnetic field of a dying fire tends to become
more stable, and temperature anomalies smooth down until they disappear. Emissions
of noxious gases cease after complete burn-out. Owing to its high-resolution, Quickbird
imagery is the best means for investigation of coal fires in the stage of extinction. High-
precision aerial thermal infrared imaging should be used to assess development trends
402
and re-ignition potential. High-precision ground thermal imaging can serve to establish
the exact degree of extinction, while high-precision magnetic methods give information
on the distribution of burned rock, i.e. the area underground that has been directly
affected by a coal fire.
4.2 Selection of detection methods
For optimal results, detection methods in 4LHDMS need to be chosen according to

technical requirements, which are mainly dictated by the geophysical properties of a coal
fire and their measurable macro-effects during the fire’s different stages of development.
Table 1 lists them in extenso; it shows that:
1) High-resolution Quickbird imagery can be used to interpret the dynamics of an

underground coal fire through investigation of its surface features (e.g. cracks and
fissures). It is an indispensable method for the whole process of coal fire detection and
monitoring.
2) High-precision ground thermal imaging is a very convenient and effective area-based

method for temperature measuring and monitoring. It allows rapid mapping, and is
generally used to delineate large areas of coal fire anomalies.
3) Magnetic field activity is an important response parameter in coal fire research

and detection, usable throughout the life cycle of a coal fire from self-ignition to
extinction. Although generally governed by the sun, i.e. radiation, the total magnetic
field of a survey area may suffer local variations from coal fire impact as chemical
processes during combustion lead to local enrichments of minerals. Magnetic field
surveys allow robust statements on the depth and extent of a fire by indicating strata
of burned rock.
4) Gas measurements are a common method for all stages of coal fire development.
Because emissions from a fire alter in a specific, measurable way during combustion,
they can be used as indicators of combustion stages. Gas measurements are
indispensable in coal fire research and detection.
The technical requirements for coal fire detection as listed in Table 1 are based on Chinese
national criteria for coal fire fighting and detection; they have been ordered according to
the 4-stage coal fire development model.
403
Table 1: Major technical requirements for optimal 4LHDMS detection in the four stages
of coal fire development
Development Geophysical Detection Methodsa Requirements for
Stage Properties Optimal Detectionb
1st stage Weak geo-stress Quickbird S: 0.6 m × 0.6 m

Weak thermal Ground TIR imaging T: 0.2 °C; S: 5 m × 5 m
anomalies TIR temperature T: 0.1 °C
measurements
Extreme gas Gas measurements (CO, C: 2 %
anomalies CO2, etc.)
2nd stage Weak geo-stress Quickbird S: 0.6 m × 0.6 m

Weak thermal Ground TIR imaging T: 0.5 °C; S: 5 m × 5 m
measurements
Medium gas Gas measurements C: 2 %
anomalies (CO, CO2, Rn)
3rd stage Medium geo-stress Quickbird S: 0.6 m × 0.6 m
Strong thermal Aerial TIR scanning T: 1.0 °C; S: 2 m × 2 m

anomalies Ground TIR imaging T: 0.2 °C; S: 1 m × 1 m
TIR temperature T: 0.2 °C
measurements
Strong local magnetic Magnetic surveys M: 1.0 nT
anomalies
Strong gas anomalies Gas measurements C: 2 %
(CO, CO2, Rn)
4th stage Strong geo-stress Quickbird S: 0.6 m × 0.6 m

Weak to no thermal Ground TIR imaging T: 0.2 °C; S: 2 m × 2 m
measurements
Strong local magnetic Magnetic surveys M: < 1.0 nT
anomalies
Weak to no gas Bundle-tube C: 2 %

anomalies measurements (CO, CO2, Rn)
a
TIR = Thermal infrared
b
S = Spatial resolution; T = Temperature resolution; M = Magnetic resolution; C = Relative gas concentration
404
4.3 4LHDMS in coal fire management
In this scheme, the operational and technological architecture of 4LHDMS serves as the
basis for a re-organization of hierarchies and tasks around coal fire management, and for an
adaptation of their workflows to the different stages of coal fire development. The current
management levels in China as far as coal fires are concerned include: (i) the Chinese
government, (ii) related subordinate government departments and authorities, (iii) mining
corporations, and (iv) fire fighting teams. Their main duties and functions are: (i) coal fire
research, surveying, and planning; (ii) detailed investigation of coal fire areas; as well as
(iii) organization, and (iv) implementation of fire fighting campaigns (Figure 4). 4LHDMS
can be optimized by selecting effective, easy-to-operate, and economical methods
according to each level’s management targets.
Figure 4: Coal fire management levels and functions
1) The Chinese government at the top of the coal fire management pyramid decides over
the objective, mission, and operational layout of coal fire research and fighting (e.g.
“China’s Agenda 21”). It defines political guidelines, sources funding, and organizes
general surveys and environmental impact assessments of coal fire areas. Large-scale
planning of fire fighting operations also falls within the government domain. 4LHDMS
here has to focus on coal fire investigations and monitoring on a national or provincial
scale, using detection methods of low resolution on a scale from 1:100,000 to 1:25,000.
2) Subordinate government departments and authorities organize campaigns to clarify

coal fire location and environmental impact. They provide the detailed information that
underlies government decisions on coal fire management. 4LHDMS here focuses on
coalfield investigations and monitoring, using detection methods of medium resolution
on a scale from 1:50,000 to 1:10,000, combined with space- and airborne surveys.
405
3) Mining corporations are responsible for the implementation of coal extraction and
fire fighting plans issued by government bodies. They manage the allocated funds
and undertake all organizational tasks on site. 4LHDMS here also revolves around
coalfield investigations and monitoring, though it uses detection methods of slightly
higher resolution on a scale from 1:5,000 to 1:1,000, combining them with space- and
airborne, as well as ground surveys.
4) Coal fire fighting teams choose the appropriate technology for fire extinction and resource
conservation; they also carry out post-campaign monitoring and environmental restoration.
4LHDMS here has a clear focus on fire fighting (i.e. engineering) and monitoring,
using detection methods with the highest resolution on a scale from 1:500 to 1:1,000.
Geophysical surveys, gas monitoring, and GIS-based 3D visualization assist in the task.
Table 2 lists the research foci, main tasks, deliverables, and detection methods at the
different levels of coal fire management as applicable in 4LHDMS.
The following general conclusions can be drawn from the array of different duties and
functions in coal fire management:
1) Different coal fire management levels have different objectives and different principles
to guide them.
For government officials, coal fire fighting is a means of disaster mitigation driven by
concerns of resource conservation and environmental protection. From a fire fighter’s
perspective, coal fire fighting is above all a question of economic and technical feasibility.
2) Different coal fire management levels place different demands on coal fire research.
Government officials are interested in base data on coal fire distribution on a national or
provincial scale. Spaceborne remote sensing together with ground investigations in low-
resolution is sufficient for their purposes. Coal fire fighting teams, in contrast, have an
immediate interest in getting as exact and minute a picture of the situation underground
as possible. Here, high-resolution detection and monitoring are the methods of choice.
3) Different coal fire management levels have different technical standards.
Government officials carry out disaster surveys and set up the standards and technical
specifications that lower echelons of coal fire management are working with. Coal fire
fighting on all levels must follow national regulations and produce the data they stipulate.
406
Table 2: Coal fire management levels and their duties and functions
Management Research Focus Main Tasks Deliverables Detection Methods
Level
Government General Determination of coal fire extent and Brief report on Landsat ETM, ASTER,
investigation of position/environment coal fire investigations IKONOS, Quickbird
coal fire area (1:25,000 to 1:5,000)
Environmental Impact Assessment Aerial TIR imaging
Subordinate Detailed Determination of conditions for Detailed report on Quickbird

government investigation of coal fire fighting coal fire investigations
Aerial TIR imaging
departments and coal fire area (1:5,000 to 1:2,000)
authorities Ground TIR imaging
Base data for fire fighting
planning
Mining Organizational Investigation of combustion systems Detailed report on Quickbird

corporations planning of coal fire investigations
Determination of conditions for Aerial TIR imaging
coal fire fighting (1:1,000 to 1:500)
407
coal fire fighting

Ground TIR imaging
Methodological design for

Fire fighting planning
fire fighting Electromagnetic and magnetic
measurements, strata imaging, gas
measurements, 3D visualization and
modeling
Fire fighting Technical High-resolution multi-temporal Dynamic assessment of work Ground TIR imaging, ground TIR
teams implementation of detection and monitoring of progress to assure coal fire temperature measurements, borehole
coal fire fighting combustion systems fighting in accordance with measurements of temperature and gas
national regulations emissions
Determination of coal fire
propagation and development trend 3D visualization and modeling
Fire fighting High-resolution multi-temporal Evaluation report on fire

evaluation and detection and monitoring of fighting implementation in
post-campaign combustion systems accordance with national
monitoring regulations
5 Conclusions
4LHDMS integrates a series of detection methods with various functions and resolutions.
High-resolution methods have been introduced to enhance detection accuracy and thus
coal fire fighting effectiveness. The system has significantly improved coal fire research
and its application in China by optimizing existing methods.
Two comprehensive operational schemes of coal fire detection and management have been
developed from 4LHDMS. The detection scheme re-organizes detection methods according
to the four stages of coal fire development, trying to achieve optimal detection results by
matching technological capacities with geological, geophysical, and geochemical realities.
The management scheme re-organizes responsibilities, functions, and duties between the
different echelons of administration concerned with coal fires, guaranteeing management
efficiency by catalyzing technology integration and application.
Acknowledgements
This study was supported by the National Program No. 863 “Research on Key Techniques
Using Remote Sensing for Detection of Underground Coal Fires” and the Sino-German
Monitoring of Coal Fires in North China.” Thanks are also due to the National Remote
Sensing Center of China (NRSCC) and Shenhua Group Corporation Limited for their
budget support.
References
ARSC, 1989. Investigation report on coal fires in Shenfu Coalfield of Shaanxi Province, P.R. China.
Aerophotogrammetry and Remote Sensing Bureau of China Coal (ARSC), Xi’an, Shaanxi
Province, P.R. China. (in Chinese)
BRSC, 2006. Project report, as a contribution to the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction and Monitoring of Coal Fires in North China.”
Beijing Remote Sensing Corporation (BRSC), Beijing, P.R. China.
DLR, 2006. Project report, as a contribution to the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction and Monitoring of Coal Fires in North China”.
German Aerospace Center (DLR), Wessling, Germany.
408
of China Coal (ARSC), Xi’an, Shaanxi Province, P.R. China and International Institute for
Geo-Information Science and Earth Observation (ITC), Enschede, Netherlands.
Guan, H.Y. et al., 2005. Study on detection methods of remote sensing for extinction of coal fires in Wuda
Coalfield. In Proceedings of the 1st Symposium of Shenhua Science and Technology, Beijing,
P.R. China, September 2005. Beijing: China Coal Industry Publishing House, 272-278. (in
Chinese)
Netherlands.
Wan, Y.Q. and Yan, Y.Z., 2003. The extraction of mineral information from hyper-spectral remote sensing
images. Remote Sensing for Land and Resources 56(2): 50-54. (in Chinese)
Zhang, J.M., 1999a. Study on 3D dynamic monitoring system for spontaneous combustion of coal seams.
In Xie, H.P. and Golosiniki, T.S. (Eds.), Mining Science and Technology ’ 99. Rotterdam: A.A.
Balkema, 97-100.
Zhang, J.M., 1999b. Study on 3D imaging method for detection of underground coal fires and application
to Rujigou Coalfield of Ningxia, China. In Proceedings of the 13th International Conference in
Applied Geological Remote Sensing, Vancouver, British Columbia, Canada, 1-3 March 1999.
ERIM Int., Vol. II, 142-149.
Zhang, J.M., 1998. Study on 3D dynamic monitoring of spontaneous combustion of coal seams. PhD
thesis, China University of Mining and Technology, Beijing, P.R. China. (in Chinese)
Fires.” Shenhua Group Corporation Limited, Beijing, P.R. China and Beijing Remote Sensing
Zhang, J.M. and Guan, H.Y., 2004. Study on four-level remote sensing detection system for underground
coal fire monitoring. Remote Sensing for Land and Resources 62(4): 50-53. (in Chinese)
409
3D Detection and Visualization of
Underground Coal Fires
Zhang Jianmin1, Huan Zhongdan2, Wang Mei3, and Xiang Yaling3
1
2
3
Abstract
3D detection and visualization of underground coal fires is a most pressing concern in coal fire fighting.
Based on continuous research on geological modeling and remote sensing detection of coal fires
in northern China, the present paper discusses a new systematic approach to coal fire detection and
visualization under three aspects. First, it analyses the effects of coal combustion on its immediate
environment (bedrock), especially in relation to the different stages of coal fire development. How can
these stages inform strategic coal fire detection? Temperature is the most robust parameter in this respect
as there exists a distinct (and comparatively easy-to-measure) relationship between thermal anomalies and
combustion stages. In a second step, the paper presents a combustion system model that combines surface
geological features (such as cracks, temperature anomalies, and gas discharge) with various elements of
mine underground infrastructure (workings, laneways, etc.) to describe the spatial relationships that govern
coal fire development. The fully developed 3D combustion system appears through point, linear, and layer
features at the surface and punctiform, planar, and body features underground. High-resolution detection
methods are strongly suggested for 3D detection and visualization of this underground phenomenon. In
a third and final step, the combustion system model is made the basis of coal fire monitoring and fighting
in a 3D Detection and Visualization System for underground coal fires. This system heavily draws on
experience from coal fire investigations in Ningxia and Inner Mongolia Autonomous Regions, as well as
Shaanxi Province.
摘要
地下煤火三维探测与动态监测一直是灭火工程中最需要迫切考虑的方法。本文在中国北方地下煤
火的地质模型和遥感探测方法研究基础上，从三方面讨论了煤火探测和可视化的一种新的系统方
法。一是分析了煤火对周围环境（岩床）的影响，尤其是煤火发展不同阶段的影响。这些阶段如
410
何与战略上的煤火探测相关？在这方面温度是最强有力的参数，因为热异常和燃烧阶段之间存在
明显的（相对容易测量的）关系。二是介绍了一种结合地面火区地质特征（裂隙、温度异常、气
体排放等）与矿井地下结构（矿坑、巷道等）各种要素的燃烧系统模型来描述支配煤火发展的空
间关系。这种完全展开的3D燃烧系统模型通过地表的点、线、层和地下的点、面、体特征显示出
来。高分辨的探测方法在地下煤火的3D探测和可视化过程中被重点提出。三是燃烧系统模型成为
所谓的地下煤火3D探测和可视化系统中煤火监测和灭火的基础。这套系统大大地吸收了在宁夏、
内蒙古自治区以及陕西省开展的煤火研究的经验。
1 Introduction
A good number of coal fire studies with a focus on coal fire detection have been
published in China over the last years. Stages of coal fire development, for instance, were
summarized in two studies relating to investigations in Xinjiang Autonomous Region
by Guan and van Genderen (1997) and Zhang (1998). They divided the fire process into
four stages from: (i) weathering and oxidation to (ii) self-ignition, (iii) combustion and
expansion, and (iv) extinction. A 4-Level Detection and Monitoring System (4LDMS)
for coal fire research was developed as a result of investigations in Rujigou Coalfield
of Ningxia Autonomous Region (Zhang & Guan 2004); it specifically referred to
observations in Xigou and Beisan Coal Fire Areas. High-precision magnetic surveys
for coal fire depth estimation were conducted by Zhang (1998). High-resolution
imagery (Quickbird as well as aerial and ground thermal scanning) was employed for
investigations in Wuda Coalfield of Inner Mongolia Autonomous Region by the Beijing
Remote Sensing Corporation (BRSC), within the frameworks of two cooperative projects
with the International Institute for Geo-Information Science and Earth Observation
(ITC), as well as the German Aerospace Center (DLR), Deutsche Montan Technologie
GmbH (DMT), Federal Institute for Geosciences and Natural Resources (BGR), and
Freiberg University of Mining and Technology in the years from 2000 to 2005.
These last projects were mainly drawing on high-resolution data, such as Quickbird
imagery, airborne thermal maps, and ground measurements, which allow for detailed
mapping at different scales. In some investigations, data were used to trace combustion
centers with 3D inversion methods. In order to monitor coal fire real-time variations,
however, it is necessary to combine all technologies available to date and integrate them
into a comprehensive system for coal fire detection and visualization. Based on results
of recent Chinese coal fire research (Zhang & Guan 2005), and with a view to including
detection methods that have already been used successfully in the field, this paper attempts
to develop such a system.
411
2 Coal Fire Development and Its Detectable Features
During spontaneous combustion, a series of physico-chemical processes such as

dehydration, degasification, pyrolysis, combustion, and gasification take place in a coal
seam. Obviously, there exists a distinct relationship between the detectable anomalies of
a coal fire and its stage of development. A 12-stage combustion model was developed
by Zhang and Guan (2005) to describe coal fire development based on the geological,
geophysical, and geochemical changes that occur in the coal seam and its surrounding
bedrock as a result of these processes (see also Guan et al. 2008 in this volume); the model
comprises the following stages:
1) Weathering,
2) Oxidation,
3) Spontaneous combustion,
4) Formation of a combustion center,
5) Formation of a combustion system,
6) Complete combustion,
7) Gasification,
8) Continuous combustion,
9) Formation of pyometamorphic rock,
10) Precipitation of minerals,
11) Dormancy, and
12) Extinction.
Figure 1 illustrates the stage model by means of temperature curves. Temperature

was chosen because it represents a most direct link to phenomena as diverse as
pyrometamorphism, changes in magnetic and conductive properties of rock, and
convective gas discharge from surface cracks – all induced by the heat emitted from a
coal fire. Temperature has the additional advantage of being measurable from the four
levels of coal fire detection, i.e. in space- and airborne, as well as ground and underground
investigations (see Zhang et al. 2008 in this volume).
412
Figure 1: Temperature development at different locations of a coal fire, based on the

12-stage combustion model (see Figure 2 for exact locations). a) Combustion
center; temperatures as indicated by mineralogical, physical, and chemical
variations. b) Inside cracks with gas discharge. c) Temperature difference at
ground above fire zone.
413
Figure 2: Schematic overview of measurement locations Figure 1
As not all parts of a coal fire are directly accessible with measuring devices, especially
so the extremely hot combustion center(s), temperatures representative of the thermal
regime within a coal fire have to be derived indirectly from investigations of minerals,
chemical soil components, or physical and chemical processes such as pyrometamorphism
and gasification. The temperature values displayed in Figure 1a, for example, represent
averaged values from a comparison of simulated temperatures of a combustion center and
unaffected nearby ground based on geological models of coal fire development (Zhang
& Guan 2005). A most remarkable drop in the combustion center takes place in the stage
change from 9 to 10 at a temperature of more than 800 °C. Temperature values inside
cracks with gas discharge are slightly lower and show weak anomalies starting with the
second to third combustion stage (Figure 1b). In the early stages of coal fire development,
detection of thermal anomalies is very difficult. Figure 1c presents the minimum
temperature differences at the ground above a coal fire. The difference is but very slight
and ranges from 5 to 25 K. There are no anomalies in the first three stages, and even in
stages 4 to 6, they are very weak. A comparatively strong difference begins to manifest
itself only with the expansion stages from stage 7 onwards.
Temperature is a direct indicator of coal fire combustion stages and can be used in
mineralogical studies, ground geological surveys, as well as remote sensing detection. For
instance, underground coal fires can be detected through temperature measurements in
cracks as early as the second to third stage of combustion (where thermal anomalies first
appear; see Figure 1b). If fire-related cracks with gas discharge have not been identified
prior to temperature measurements, positive detection at the surface is only possible if the
fourth stage in the combustion process has been reached (see Figure 1c).
414
Despite the evident relationship between temperature and combustion, it is well-nigh

impossible to identify each stage by means of temperature or indicator minerals alone.
For reasons of detectability and practicality, the model is generally reduced to four stages
in field applications, namely: (i) spontaneous combustion, (ii) formation of a combustion
center, (iii) formation of a combustion system, and (iv) temperature drop and extinction
(Guan et al. 2005; Zhang & Guan 2005; see also Zhang et al. 2008 in this volume).
3 Models and Monitoring Requirements
An exact knowledge of the geological conditions underground is a precondition for

detection and monitoring of underground coal fires. Based on the then newly developed
combustion stage theory, Guan and Zhang (1999) developed a simplified coal fire model.
This so-called “combustion system” model is composed of a combustion center – the
spot with the highest temperatures in a coal fire – and an air circulation loop (oxygen
uptake and combustion gas discharge), with cracks linking the burning coal seam to the
surface. The model was subsequently optimized by Zhang and Guan (2005) with a view to
displaying geological structures for coal fire detection, monitoring, and data interpretation.
Combining underground mine structures with results from surface temperature
measurements, crack mapping, and emission gas surveys, the combustion system model
allows for detailed fire representation (see geological cross-section; Figure 3).
Figure 3: Geological cross-section of a coal fire combustion system
415
The combustion system model visualizes in three-dimensional space the structures formed
in a coal seam during spontaneous combustion, among them: punctiform structures in
the initial stages of combustion and, in later stages, planar structures across the burning
coal seam, as well as body structures, e.g. in the development of fissures and cracks.
These three different types are also visible in the field, i.e. at the surface (Figure 4). Point
structures correspond to the early stages, where air enters the ground and triggers off
oxidation processes in the coal seam leading to spontaneous combustion. Linear or layer
structures represent development stages, during which a proper combustion system is set
in place. The planar or body features, eventually, mark a fully developed 3D combustion
system.
Figure 4: Surface features of a coal fire area. FZs 8 (a) and 11 (b), Wuda Coalfield, 2005
A coal fire combustion system possesses a number of geophysical features that change
over the different stages of combustion. Monitoring measurements should hence be
repeated at intervals, best with the same equipment and preferably in high resolution
– up to 0.1 K for thermal investigations, and 1 to 5 m for spatial coverage (compare
Figure 5). In satellite images and airborne thermal maps, the common resolution of
ground measurements is high enough to detect subtle surface anomalies. Addressing
requirements of both fire detection and extinction, the combustion system model, in
its systemic approach, can be used as a basic unit for coal fire fighting and engineering
design (Zhang & Guan 2005).
Surface cracks, vegetation changes, thermal radiation, magnetic field anomalies, gas
emissions, and other parameters were combined in a 3D model for coal fire detection
and data interpretation along the different stages of coal fire development by Zhang
(1998).
416
Figure 5: Resolution in monitoring of underground coal fires, based on the combustion

system model
4 Monitoring Methods and Data Visualization
In fire detection based on the combustion system model, high-resolution monitoring

methods are needed for data collection, interpretation, and visualization. The 4-Level
High-Resolution Detection and Monitoring System for coal fire management (4LHDMS;
see Zhang et al. 2008 in this volume) combines the four levels of investigation: space- and
airborne surveys, as well as ground and underground measurements. This chapter presents
examples of high-resolution, multi-level coal fire detection and monitoring from Ningxia
and Inner Mongolia Autonomous Regions, as well as Shaanxi Province.
4.1 3D thermal imaging for coal fire delineation in Rujigou Coalfield,

Ningxia
Coal fire-related thermal anomalies are generally detected in ground surveys, or inferred
from satellite and aerial thermal infrared image data. Thermal images have first to be
417
transformed into isothermal maps by way of radiation correction and computation of

temperature levels from radiation before thermal anomaly profiles can be drawn.
Suffering the impact of geology, topography, mining, and underground water environment,
the thermal field of an underground coal fire is more complex than theory suggests. Modeling
here helps to conclude from both isothermal images and thermal anomaly profiles about the
actual situation underground. In Rujigou Coalfield, thermal models of punctiform, linear, and
layer sources were developed and analyzed for the relationship between model parameters and
detected thermal anomalies. Thermal conductivity imaging was forwarded by introducing wavelet
algorithms in data interpretation. Thermal anomaly sections, identical to surface temperature
profiles, were established and processed in 3D thermal imaging (Zhang 1998; Zhang 1999a,b).
4.2 Ground thermal coal fire detection in Wuda, Inner Mongolia
Ground thermal detection is a direct and near-distance method used wherever soil
conditions allow for access of fire sites. Its validity and methods of data processing were
studied in Wuda Coalfield of Inner Mongolia (Zhang & Guan 2005). Wuda is located
west of Wuhai City and is a well-known Chinese coalfield, famous as much for its coke
as for its coal fires. It hosts a number of shallow coal seams, which are mostly worked in
longwall mining. This means that coal fires here generally evolve along vertical axes in
mine systems and horizontally in open pits.
Thermographic images of selected FZs in Wuda – a hot crack zone, an underground

combustion center, and a zone of cracks of low temperature – are presented in Figures 6
and 7; they were taken with a DL-700A thermal infrared camera. Temperature values for
each image are listed in Table 1. There are stark differences between the sites – temperature
anomalies are decreasing with increasing distance from the main combustion center.
Table 1: Temperature values for selected thermographic images of different FZs in Wuda
Coalfield, Inner Mongolia, 2004/05
Image Tave Tmin Tmax Std Tmax - Tmin Geological Features
488a
30.46 15.47 42.06 2.9 26.59 Cracks with low-temperature gas discharge
478b
32.32 0.47 180.06 12.05 179.59 Crack line above mine shaft
464c 30.81 15.91 70.02 3.47 54.11 Area of pronounced thermal anomalies
a
Taken at FZ 8 in Wuhushan Mine.
b
Taken above underground conveyer shaft in Suhaitu Mine.
c
Taken at FZ 8 in Wuhushan Mine.
It is common in coal fire detection to delineate coal fire areas on the basis of temperature
differences/anomalies (in combination with geological features). These can be obtained
418
from thermal maps by image processing. From a thermal mapping image, the average
temperature Tave, the minimum temperature Tmin, the maximum temperature Tmax, and the
temperature difference ΔT can be calculated. The applicable procedures are:
1) Conversion of the image to a gray-level image H with the formula ΔT = (Hmax- Hmin) / ΔT,
reducing the temperature to TH = ΔT × H;
2) Analysis of temperature variations and extrapolation of temperature anomalies in a
graph of isothermal curves through T = Tave± Ti, i = 0,1,2,3;
3) Coloring of different temperature ranges for segmentation of the thermal image;
4) Image enhancement by sharpening to highlight boundaries of temperature variations and
adding proper isothermal curves to mark specific regions.
Figure 6 shows two thermal infrared images with clearly marked areas of thermal
anomalies (significantly different from ambient temperature) and their respective field
photos. Linear (surface) features pertaining to underground coal fires are evident.
Anomalies are very helpful in interpreting convective air flow (wind) in cracks.
Figure 6: Ground thermal scanning in Wuda Coalfield, 2004. a) Combustion center in

FZ 8 of Wuhushan Mine (image no. 488). b) Crack zone above underground
conveyor shaft in Suhaitu Mine (image no. 478).
419
An example of thermal image data processing is presented in Figure 7. Figure 7a is the

original thermal image of a combustion center in FZ 8 of Wuhushan Mine. It depicts
thermal anomalies in thresholds of 3 ,10, 20, and 30 °C from average temperature. The
effective range of temperature can be determined directly from those anomalies twice as
high as the average background temperature. The field photo is given for reference because
the thermal image was not geometrically corrected, which makes fusion processing
difficult. However, temperature variations can be detected through replication. Ground
thermal detection is an important tool in fire fighting since it allows delineation of
combustion centers at a resolution of up to 1 × 1 m.
Figure 7: Thermal data processing for Wuda Coalfield, 2005. a) Original thermal image
of FZ 8 in Wuhushan Mine (image no. 464). b) Field photo. c) Temperature
classified by thresholds of 20, 23, 30, and 40 ˚C. d) Temperature classified by
thresholds of 20, 23, 30, and 40 ˚C, based on brightness. e) Color slice image
classified by thresholds of 23, 25, 30, 40, and 50˚C from Tstart = Tave (ambient
temperature).
420
4.3 High-precision resistivity layer detection in Daliuta Mine, Shaanxi
The change of a (primary) electromagnetic field at surface level causes currents in the
subsurface, which induce secondary fields that spread underground and influence the
electromagnetic field. The nature and magnitude of these changes depend on the physical
properties of the underground layers (e.g. rock strata and coal seams) and can be measured.
For example, the electrical conductivity (or resistivity, its reciprocal) of burned and virgin
coal differs significantly, and this difference can be used to trace underground coal fires
by way of electromagnetic surveys. A high-precision electromagnetic detection system
(MES) was developed by Zhang and Guan (2005) for investigations in Daliuta Mine of
Shaanxi Province. Conductivity measurements showed a positive correlation between
virgin coal layers (as found in borehole measurements) and conductivity lows as derived
from one-dimensional inversion of MES data (Figure 8a). In order to separate unaffected
coal layers from non-coal layers, the MES wave spectrum was measured at the eastern
boundaries of Daliuta Mine and converted into apparent resistivity (Ps) profiles. In these
profiles, materials other than coal appear as areas of high resistivity, burned rock and
unaffected coal show comparatively low values (Figure 8b). Resistivity anomalies can
also serve to localize combustion areas, but the layering capacity needs to be enhanced
for this purpose.
Figure 8: Electromagnetic surveys in Daliuta Mine, 2005. a) Coal seam distribution as

per borehole measurements (grey) and one-dimensional inversion of MES data
(near borehole; black) at location C7. b) Apparent resistivity profile of location
B7 showing the boundary between burned coal seam of higher resistivity (left)
and unaffected coal seam of lower resistivity (right).
421
4.4 3D visualization of coal fires in combination with mine elements
3D coal fire visualization based on GIS is helpful in highlighting the spatial relationship
between a coal fire and its immediate environment. A 3D visualization system was
developed by Zhang and Guan (2005) for Daliuta Mine; it was later improved by
integration of a coal fire model. In this system, underground elements such as coal seams,
rock strata, faults, boreholes, active and abandoned mine workings, underground pathways,
etc. can be arranged for combined view. Figure 9 shows an image of Suhaitu Mine in Wuda
Coalfield created with the Micromine-based Coalminer3D. It depicts surface coal fires, an
underground working, and underground mine infrastructure (laneway). Other models can
be inserted into the system for data interpretation, e.g. thermo-dynamic models of coal fire
development.
Figure 9: 3D-visualization of Suhaitu Mine in Wuda Coalfield, based on Coalminer3D
422
5 Fire Fighting-Oriented Monitoring Based on

High-Resolution Detection
For efficient and successful fire fighting, it is essential to integrate geological models with
detection and interpretation tools. Operations can in fact be planned so as to meet the data
and measurement requirements of the models and methods applied. Based on the 4-Level
Detection and Monitoring System for coal fire research (4LDMS) (Zhang & Guan 2004;
see also Zhang et al. 2008 in this volume), the combustion system model, as well as long-
term experience and results from application of high-resolution detection techniques in the
field, the authors of this study propose a scenario for coal fire monitoring, briefly called 3D
Detection and Visualization System for underground coal fires.
The system is designed to meet the following requirements:
1) Data and results must be available to fire fighting teams in engineering scale.
2) S patial resolution of any detection method employed must not be coarser than
3 m × 3 m × 5 m.
3) Detection data must be reproducible at identical time intervals to allow for step-to-step
comparison.
4) Detection data must be normalized and interpreted (e.g. temperature parameters) with
respect to geological setting.
5) Data must be visualized in a combination of geological and mine elements.
The 3D Detection and Visualization System for underground coal fires has a threefold
foundation: (i) it uses latest high-resolution detection technology (Quickbird imagery,
portable thermal scanners) to obtain data from the four levels of coal fire investigation;
(ii) it visualizes these data with special software (3D-GIS applications such as Data
Mining, Micromine, Su-pac); and (iii) it proceeds to interpretation and application in
fire fighting based on the combustion system model. Figure 10 shows the workflow that
should be observed in applying the system; its most important component is monitoring.
1) Planning of monitoring layout: design according to fire fighting requirements and

adapted to geological realities, e.g. selection of detection profiles and locations, time
intervals and methods, as well as required spatial resolution; also models for data
interpretation.
2) High-resolution detection: repeated measurements along a fixed profile or around a
fixed point.
423
3) Data processing: after correction according to space (azimuth, location, and physical
parameters of detection) and time (seasonal and daily variations), transfer into images,
graphs, etc.
4) Temporal and spatial interpretation and comparison: coal fire evaluation in terms
of burning intensity, range, depth, and combustion center(s) based on coal fire
(combustion system) model; data from other fires can be used for comparison.
5) 3D integration and visualization: including background information such as geology,
mine workings, and infrastructure.
6) Output: Final evaluation of coal fire outline and planning of fire fighting.
Figure 10: 3D Detection and Visualization System for underground coal fires
424
6 Conclusion
Coal fire research so far, especially with regard to coal fire development and the
identification of combustion stages, has relied on the detection of weak anomalies of
temperature, magnetism, and conductivity. Data on coal fire depth and heat sources/
combustion centers were added to these parameters for 3D interpretation. In order to
improve detection accuracy and efficiency, any operational system of coal fire detection
and monitoring should work on the basis of the combustion system model, employ high-
resolution methods such as thermal imaging and MES, and use the 3D Detection and
Visualization System for underground coal fires to support data interpretation. The system
can facilitate coal fire fighting by providing pertinent technical scenarios of coal fire
environment, monitoring, and fighting operations.
Acknowledgements
This study was supported by the National Program No. 863 “Research on Key Techniques
Using Remote Sensing for Detection of Underground Coal Fires” and the Sino-German
Monitoring of Coal Fires in North China.” Special thanks for general advice and technical
instruction go to Shenhua Group Corporation Limited and to Prof. Guan Haiyan of BRSC;
Prof. van Genderen of ITC; Dr. Stefan Voigt of DLR; and Dr. Friedrich-Karl Bandelow of
DMT. The same is due to Dr. Yu Minggao of Henan Polytechnic University and Ms. Ma
Jie of Coal Research Institute of China (CRI).
References
BRSC, 2006. Project report, as a contribution to the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction and Monitoring of Coal Fires in North China.”.
Beijing Remote Sensing Corporation (BRSC), Beijing, P.R. China.
Guan, H.Y., 2008. Spontaneous coal seam fires: The Chinese perspective. In UNESCO Office Beijing
425
Guan, H.Y. et al., 2005. Study on detection methods of remote sensing for extinction of coal fires in Wuda
Coalfield. In Proceedings of the 1st Symposium of Shenhua Science and Technology, Beijing,
P.R. China, September 2005. Beijing: China Coal Industry Publishing House, 272-278. (in
Chinese)
Zhang, J.M., 1999a. Study on 3D dynamic monitoring system for spontaneous combustion of coal seams.
In Xie, H.P. and Golosiniki, T.S. (Eds.), Mining Science and Technology ’99. Rotterdam: A.A.
Balkema, 97-100.
Zhang, J.M., 1999b. Study on 3D imaging method for detection of underground coal fires and application
to Rujigou Coalfield of Ningxia, China. In Proceedings of the 13th International Conference in
Applied Geological Remote Sensing, Vancouver, British Columbia, Canada, 1-3 March 1999.
ERIM Int., Vol. II, 142-149.
“Research on Key Techniques Using Remote Sensing Method for Detection of Underground
Coal Fires.” Shenhua Group Corporation Limited, Beijing, P.R. China and Beijing Remote
Sensing Corporation (BRSC), Beijing, P.R. China. (in Chinese)
Zhang, J.M. and Guan, H.Y., 2004. Study on four-level remote sensing detection system for underground
coal fire monitoring. Remote Sensing for Land and Resources 62(4): 50-53. (in Chinese)
Zhang, J.M., Xiang, Y.L., Wang, M., and Huan, Z.D., 2008. 4-level high-resolution detection and
monitoring – Applications in coal fire detection and management. In UNESCO Office Beijing
426
Part VI
Data Management,
Visualization, and Modeling
3D Visualization of Underground Coal Fires
Wang Mei1, Zhang Jianmin2, Wang Yufei3, Gao Yunze3,

Li Guangliang3, and Xiao Haihong1
1
2
3
Abstract
3D visualization of underground coal fires facilitates coal fire research and can improve the efficiency of
coal fire fighting. The research behind this paper was conducted within the framework of two projects,
the National Project No. 863 “Research on Key Techniques Using Remote Sensing for Detection of
Underground Coal Fires” and the Sino-German Coal Fire Research Initiative “Innovative Technologies
for Exploration, Extinction, and Monitoring of Coal Fires in North China.” Wuda Coal Fire Area (Inner
Mongolia Autonomous Region, P.R. China), characterized by a complex geological pattern and a mosaic
of underground coal fires, was selected as study area. In order to visualize coal fire environment (e.g.
geology, coal seams, mine workings and infrastructure) at selected spots, a series of 3D models was
developed from data collected during a field trip in 2004. These models, and the approaches that informed
them, are described in the present paper according to the survey data used for input. The authors discuss
3D visualization of underground coal fires as a means to (potentially re-)direct coal fire fighting, so as to
raise efficiency and economic value.
摘要
地下煤火的三维可视化是提高地下煤火研究水平和煤田火区治理工程效率的基础。本文依托863
攻关项目“地下煤火遥感探测关键技术研究”和中德科技合作开展的“中国北方煤火探测、灭火与
监测”项目，选择地下火区环境条件复杂和地下煤火严重的内蒙古乌达试验区，基于三维软件
Micromine，建立了地下煤火赋存的地层、地质构造、煤层、采空区及其相关联的巷道工程等内容
的三维模型，提出了地下煤火环境三维可视化模型的构造方法。作者从煤火监测数据角度对煤火
的三维建模和可视化方法进行了研究与分析，针对不同类型的基础数据提出了具体的地下煤火建
模方法。基于煤火建模结果，本文提出了指导灭火工程的新方法，使灭火工程实现经济、高效。
428
1 Introduction
Wuda Coal Mining Area is located in the west of Inner Mongolia and covers an area of
200 km2, including Wuhushan, Suhaitu, and Huangbaici Coal Mines. Wuda Coalfield, with
a total area of 35 km2, lies in the northwest of Wuda Coal Mining Area, extending over
10 km from east to west, and 3 to 5 km from north to south. The geological setting in the
north of the coalfield is rather simple – folds are gentle and open, while faults of generally
large angles furrow the area. By contrast, the south is far more complex geologically,
with a greater number of folds and steeper faults. The terrain here is marked by intensive
tectonic stress. Wuda Coal Mining Area has 17 mineable coal seams with an average
thickness of 33.6 m.
From field work and surveys, we know that coal seams nos. 1, 2, 4, 6, 7, 9, 10, and 12 are
currently affected by spontaneous combustion. Based on the mining safety requirements
in effect in Wuda and the propagation velocity that a potential coal fire may develop
there, we selected coal seams nos. 1, 2, 4, 6, 7, 9, 10, 12, 13, 15, 16, and 17 for modeling.
Additionally, four representative faults were selected, which all have significant influence
on the overall coal output of Wuda Coal Mining Area. These were an overthrust fault, a
small outcrop fault, the F29 fault zone, and the Yangquan reversed fault.
As mining in Wuda Coal Mining Area has been underway on an industrial scale for at
least 50 years, mine shafts and workings are in a very bad state and prone to further
deterioration. The main shafts of all three coal mines, as well as some workings of Suhaitu
Mine were chosen for modeling.
2 3D Visualization of Underground Coal Fire Environment
Coal fires are dynamic by nature. For effective representation and analysis on a 3D
platform, it is necessary to build 3D models of their environment, i.e. the geological
setting, including rock strata; the coal seam; as well as mine workings and infrastructure.
Based on common 3D geology modeling and visualization, this paper discusses methods
of modeling underground coal fire environment in 3D space.
2.1 Ground surface
Ground surface simulation can facilitate the understanding of underground mine structures.
It is based on a ground surface model that represents the relative spatial position of
subterranean coal deposits, mine workings, and goafs.
429
Data from DEMs (Digital Elevation Models) and high-resolution remote sensing are
primary sources for ground surface modeling – DEMs represent hypsography very clearly;
while aerial imagery contains more information than common relief maps (Wu et al. 2003).
Figure 1 is a 3D ground surface model of Wuda Coal Mining Area that has been created
by superimposing a Quickbird image on a DEM; the DEM was built from contours of the
ground surface.
Figure 1: 3D visualization of ground surface of Wuda Coal Mining Area
2.2 Coal seams
3D simulation of coal seams is generally based on a 3D rebuild of borehole models. We

used the section-TIN (Triangulation Irregular Net) method for this purpose, selecting a
number of 2D geological sections and interpreting the data they contained for interactive
delineation of strata and borderlines. In a final move, these borderlines were connected
with adjacent sections of the same attributes. The 3D coal seam model thus created is
shown in Figure 2.
430
Figure 2: 3D coal seam model
2.3 Faults
Within the limited scope of Wuda Coal Mining Area, it is extremely difficult to find
enough reference points for fault plane simulation. Consequently, fault outcrop lines, their
pattern and thickness, were used as indirect indicators. Figure 3 is a 3D fault plane model
of Wuda Coal Mining Area.
Figure 3: 3D fault plane model of Wuda Coal Mining Area
431
We applied manual modeling to build fault hanging (upper) walls and footwalls (lower
walls) separately. When closing the outline on each side of a fault, we made sure it crossed
the fault section line. Hanging walls and footwalls were calculated with a Boolean operator
for each fault plane so as to cut out those parts outside the plane.
We divided the faults into two types based on modeling angle – faults traversing the whole
mining area as well as those disappearing within the mining area. Models for these two
types are presented in the following.
Building models of faults traversing the mining area is relatively easy. Figure 4 illustrates
such a model of hanging walls and footwalls, based on Boolean operation on a fault plane.
Figure 4: Model of fault traversing the mining area
For faults disappearing within the mining area, an artificial division line has to be added
where the fault ends. This line divides the model area into three parts, which can be
modeled separately and then re-integrated to form a whole as shown in Figure 5. The red
line here is the fault, the blue line the added division line.
432
Figure 5: Partition of model area for fault disappearing within the mining area
Using this technique, we were able to divide Wuda Coal Mining Area into 11 parts, based
on the area’s four representative faults (Figure 6; see Chapter 1).
Figure 6: Division of Wuda Coal Mining Area based on its four representative faults
2.4 Mine workings and infrastructure
3D visualization of mine workings and infrastructure can assist in mine planning, production,
and safety management. For our 3D models, we used survey data of shaft traverse points, from
which we inferred working outlines intercepting the corresponding coal seams. This method is
a rather easy-to-apply scheme for modeling mine workings and infrastructure (Figure 7).
433
Figure 7: Model of mine workings and infrastructure in Suhaitu Mine
In Micromine, it is possible to integrate various models, such as ground surface and coal
seam models, as well as models of mine workings and boreholes, into a comprehensive 3D
model of coal fire environment. Such integration can substantially improve visualization
and spatial analysis of underground coal fires (Figure 8).
Figure 8: Integrated model of underground coal fire environment in Suhaitu Mine
434
3 3D Visualization of Coal Fires
Advances in remote sensing and physico-geographical technology have spawned numerous

new detection methods in coal fire research (Guan et al. 1998). Table 1 lists the major
methods currently in use for coal fire detection.
Table 1: Methods of coal fire detection

Means Methods Representation
Remote sensing Spaceborne: TM, Quickbird, Aster Image data
Airborne: thermal infrared, color infrared
Physical geography Airborne physical detection: navigation magnetism and 2D data

navigation electricity sectionwise
Ground physical detection: high-precision magnetic and
electromagnetic surveys, earthquake detection
Geochemistry CO, CO2, O2, CH4, Rn gas detection 2D discrete data
Boreholes Contact and no-contact temperature measurements 3D discrete

Sampling gas tests temperature data
Ground geology GPS orientation, geology investigation, mensurating Geological maps

norm section and sections
Ground Measurements with infrared thermodetector 2D discrete
temperature temperature data
Mine detection Temperature and gas measurements in coal mines 2D discrete

spatial data
3D mathematic simu- Positive and inverse models 3D grid data
lation 3D numerical simulation
Proceeding from this review, we designed four types of data sets for use in coal fire
visualization: (i) 2D discrete spatial data, (ii) remote sensing image data, (iii) coal fire
section data (based on map structure), and (iv) 3D grid data. For each type, we applied a
specific method of coal fire modeling.
3.1 2D discrete spatial data
2D discrete spatial data on coal fires include results from ground temperature
measurements as well as geochemistry data. Measured with an infrared thermodetector, the
former can be used to draw temperature isolines in a 3D grid model.
For data acquisition (including geographical coordinates as well as ground and crack
temperatures), Wuda Coal Fire Area was surveyed in the summer of 2004. The collected
data are listed in Table 2.
435
Table 2: Ground coordinates and temperatures as per 2004 survey of Wuda Coal Fire Area
Measurement Ground Coordinates Temperature
Point X Y (°C)
1 36380246.62 4376430.10 68
2 36380383.42 4376419.30 83
3 36380422.42 4376397.10 134
4 36380447.02 4376394.70 117
5 36380517.82 4376397.10 84
6 36380573.02 4376395.30 89
7 36380636.62 4376388.70 116
8 36380669.01 4376392.30 83
9 36380852.30 4376419.36 95
10 36380916.21 4376405.50 67
11 36380926.41 4376411.50 100
12 36381007.41 4376412.10 88
13 36381021.81 4376394.10 126
14 36381062.61 4376413.90 110
15 36381087.81 4376424.10 112
16 36381084.81 4376358.10 110
17 36381081.21 4376307.70 94
18 36381110.01 4376265.70 99
19 36381218.81 4376197.90 108
20 36381090.81 4376484.70 115
21 36381098.01 4376539.31 112
22 36381137.61 4376610.11 87
23 36381225.80 4376539.31 109
24 36381179.01 4376398.30 95
25 36381203.01 4376433.10 103
These data were processed by IDW (Inverse Distance Weighting) interpolation in

Micromine, which yielded the input for a regular grid model presented in Figure 9.
Different colors here indicate different temperatures – from red/high to blue/low. After
interpolation, an isoline map was created denoting regions of intense temperature change
with red isolines, i.e. coal fire combustion centers or crack zones (Figure 10).
Figure 9: Visualization of coal fire temperature grid data in Wuda Coal Fire Area
436
Figure 10: Isoline map of coal fire combustion centers and crack zones in Wuda Coal Fire
Area
3.2 Remote sensing image data
Remote sensing detection is a most direct way of obtaining 3D coal fire information.
A DEM can be overlaid in its entirety with coal fire images from remote sensing, thus
displaying underground coal fires in 3D. Based on TIN, we built a DEM model of Wuda
Coal Mining Area from terrain contour lines; it was subsequently overlaid with coal fire
information derived from remote sensing imagery (Figure 11).
Figure 11: DEM of Wuda Coal Mining Area overlaid with remotely sensed coal fire
information
3.3 Coal fire section data
The main information contained in 2D coal fire sections is a series of geological

borderlines, such as faults and ore bodies. We used triangular facets to connect adjacent
437
coal fire borderlines in a 3D spatial surface model. This model was then filled with defined
block models so as to form a coal fire solid model (Figure 12).
Figure 12: Coal fire solid model derived from geological sections
3.4 3D grid data
Information on coal fires is generally multi-dimensional. Four-dimensional coal fire data

can be created by using a positive model, in which pre-defined initial values and time steps
serve to generate new data sets from basic field survey data. These new sets are processed
by IDW interpolation in Micromine. The result is a 3D grid model as displayed in Figure
13. Red areas here denote areas of high temperature.
Figure 13: Integral display of coal fire grid data
438
By defining slice directions and intervals, coal fire development can be displayed in a time
series, which allows for analysis of coal fire dynamics (Figure 14).
Figure 14: 3D grid model of coal fire development
4 Applications of Coal Fire Modeling
4.1 Spatial orientation of coal fires
Remote sensing detection is an effective means to localize and gauge the extent of a coal
fire; yet it does hardly yield information on coal fire depth. This in fact is a key obstacle
to directly using remote sensing in the planning and implementation of coal fire fighting.
3D modeling and visualization of coal fires, however, has the potential to mitigate this
problem. We took fire zone (FZ) 10 of Wuda Coalfield as an example for coal fire depth
modeling. The top of the model presented in Figure 15 is the ground DEM of FZ 10, its
bottom the soleplate of coal seam no. 4.
439
Figure 15: Coal fire depth model of Wuda FZ 10
Micromine allows for additional coal fire information (e.g. combustion centers and crack
zones extracted from remote sensing imagery) to be entered in the depth model, which in
turn allows for calculation of coal fire depth. Applying this approach, we found that the
maximum combustion depth in FZ 10 is 73.39 m, which is close to the 76 m measured in
traditional ground geological surveys.
440
4.2 3D integrated display of coal fire combustion systems
Coal fires are dynamic 3D phenomena – a fact which is highly relevant to modeling
combustion systems and their evolution in time. Figure 16 is such a combustion model,
built in Micromine from data on the X2 combustion system in FZ 10. The different colors
of the planar structures representing underground mine workings are indicative of different
combustion states. The brown lines above the planes represent combustion crack zones.
The red columns are combustion centers. The model was superimposed by a ground DEM
and covered with a temperature grid. Red color denotes regions of high temperature.
Figure 16: 3D model of X2 combustion system in Wuda FZ 10
Combustion systems modeled as above can be sliced into sections so as to indicate the
direction of fire propagation. Figure 17 shows such a lengthwise section. 3D visualization
models, unquestionably, have the potential to improve the efficiency of coal fire fighting,
for example by displaying combustion centers (and connected air pathways) for targeted
extinction campaigns.
441
Figure 17: Lengthwise section of coal fire combustion system
5 Conclusion
3D visualization of coal fires and their environment is a key to improving coal fire research
and fire fighting efficiency. Using Micromine software and other 3D visualization and
modeling techniques, the authors of this paper were able to build a 3D model of coal fire
environment and present various methods to build coal fire models from different sets
of detection data. Such models offer easy-to-use and very accurate technical support to
underground coal fire fighting and prevention. With the development of ever new detection
techniques, the range of application for coal fire models is widening, yet it remains
imperative to foster research in the area.
References
Guan, H.Y. et al., 1998. Investigation and research on coalfield spontaneous combustion environment in
north China. Beijing: Coal Industry Publishing House. (in Chinese)
Wu, L.X. and Shi, W.Z., 2003. Principles and arithmetics of geographical information systems. Beijing:
Science Publishing House. (in Chinese)
442
Information Visualization in
Coal Fire Research
Li Guangliang1, Wang Mei2, and Xiao Haihong2
1
2
Abstract
Spontaneous combustion in coal seams not only consumes large amounts of resources, but also leads
to serious environmental pollution. Coal fire fighting and prevention is an intricate task that requires
continued investment and monitoring over extended periods of time. The computer-aided display of coal
fire-related information in three-dimensional space is a central concern of many people working on the
phenomenon. In this paper, traditional 3D theory, mathematical equations, and state-of-the-art technology
are used to build 3D models of coal fires.
摘要
煤层自燃对煤矿生产以及环境保护有重要影响，人们一直以来与煤火的斗争从来没有停止过，并
且在多方面取得了重大突破。不过煤火的信息可视化研究还是一个空白，本文试图利用三维可视
化技术，提取有效信息，建立合理模型，综合煤层自燃的多维信息，从而科学地界定煤层自燃的
现状与发展趋势。实现煤田火区的系统仿真，并针对提取的多种地球物探信息，使之科学高效地
反映到可视化系统中。
1 Introduction
Since Laks Arthur started his studies on coal fires in 1907 (Guan et al. 1998), scientists in
many countries, from Australia to the U.S.A., from China to India, have turned to the topic
and advanced research a great deal over the last 100 years. However, as adequate exploration
technologies are missing and exact spatial data have yet to be produced, we are still not able to
study coal fires directly in three-dimensional space (Greuer 1984; Zhang 1998).
443
Only the combined use of all information available on coal fires to date can improve the
theory of spontaneous combustion and propel the development of innovative extinction and
prevention methods. This information can come from sources as diverse as experiments on
the self-ignition behavior of coal, coal fire environmental impact assessments, monitoring
of fire propagation, and estimations of resource loss. The main techniques and disciplines
involved are remote sensing, geophysics, and geology.
The challenge in any of these efforts is to abstract from factual information, build
mathematical models that are able to visualize coal fires multi-dimensionally, and, based
on this visualization, make a sound scientific judgment on fire extent and development.
This then will inform extinction and monitoring operations.
2 Visualizing Coal Fires
2.1 The status quo
In the many years that coal fires have been investigated, a good number of models and
methods for 3D visualization have been put forward. Almost all of these models reworked
and presented coal fire-related information based on traditional 3D visualization techniques
(e.g. the three systematic facial, volumetric, and mixed models; Wu & Shi 2003).
The China Aero Geophysical Survey and Remote Sensing Center for Land and Resources
(AGRS) developed a Coal Fire Information System (CFIS) of a very simple and thus easy-
to-use structure. It integrates source and result data obtained from coal fire research, and
provides a tool to manage attribute data.
The Beijing Remote Sensing Corporation (BRSC) and Netherlands Environmental

Analysis and Remote Sensing (EARS) developed an environmental monitoring and
information management system called Coalman. Coalman integrates information from
geology, mining, and remote sensing. Users can enter questions directly into the system
and get answers from its database (Zhang et al. 1998).
The two information systems described above, however, were in fact only based on two-
dimensional software. For 3D representation, they covered a DEM (Digital Elevation
Model) with remote sensing images from coal fire locations. While this allows a bird’s
view on coal fires, it yet does not represent a thorough-going 3D scientific analysis.
444
2.2 Traditional methods of 3D visualization
For want of tangible new developments in the field, scientists often have to resort to
traditional methods of 3D visualization to build 3D models of coal fires. These approaches
generally model the coalfield in a first step, i.e. terrain, coal seams, rock strata, faults, and
other geological features, and then superimpose a separately built coal fire model in those
spots where fires occurs in reality. If the match between the two models is sufficiently
large, the fire model can replace the part of the coalfield model that it overlaps with (Cui et
al. 1993; Wu & Shi 2003). Figure 1 illustrates this technique: areas of coal fires (modeled
in red) have been superimposed on a 3D model of Wuda Coalfield (Inner Mongolia
Autonomous Region, P.R. China).
Figure 1: Model overlay. Coal fire distribution in Wuda Coalfield.
3 Mathematical Models
In forward modeling, a coal fire model is developed based on the laws of heat transfer and
subsequently calibrated against data from in-situ measurements. Inversion, on the other
hand, is a method where conclusions are directly drawn from measurement data and these
are used as parameters for the model. After a forward model has been set up, fire position
and intensity can be ascertained by real-term temperature data, i.e. by means of inversion.
Because inversion evidently depends on the forward model to provide a framework for
445
data processing, model research generally starts with a simple forward model and then
proceeds to inversion.
Although numerical simulation and inversion are rather novel approaches in coal fire
research, numerous studies have already been conducted on coal fire temperature fields,
including a number of mining workface and coal fire exploration models (Zhang et al.
1998). Based on in-situ investigations in Wuda Coalfield, the authors of the present study
have developed a dynamic mathematical model of temperature evolution and distribution
in coal fires.
It was assumed that a coal fire heat source/combustion center can generally be of three
shapes – punctiform, globular, or rectangular. Simulation data from the model were
checked and calibrated against data measured in situ. After the degree of consistency had
reached a satisfactory level, the model was applied to predictions on coal fire development.
Assuming that parameters such as the specific heat c, the rock density ρ, and the heat
exchange coefficient α are generally constant and that the heat source intensity ω is
independent of temperature, we were able to develop the following simple steady
mathematical model of temperature T:
(1)
where
(2)
where λ is the thermal conductivity.
With input from measurements in Wuda Coalfield, the following result data were obtained
from the model (Table 1).
446
Table 1: Coal fire simulation data from mathematical model

Measurement Point Coordinates Temperature (°C)
X Y Z Ttime1 Ttime2
137 441.474 535.157 1223.06 47 25
136 440.819 491.977 1223.70 37 15
58 787.685 669.733 1200.00 57 35
56 849.379 488.870 1199.71 35 13
48 514.736 481.600 1217.35 49 27
150 492.064 462.154 1218.07 32 10
168 408.872 433.848 1220.72 37 15
161 443.864 377.787 1220.00 35 13
169 471.409 303.330 1214.01 46 24
170 490.521 303.040 1211.92 40 18
176 499.420 417.049 1217.23 45 23
166 440.351 303.801 1218.91 30 8
171 483.494 312.401 1214.50 40 18
172 483.728 327.823 1215.09 42 20
173 460.400 365.196 1220.00 45 23
177 491.316 355.472 1213.05 47 25
These data were used as input for a 3D visualization model, an example of which is given in
Figure 2. The series from left to right depicts the gradual combustion of coal expressed by
an enlarging temperature anomaly field.
Figure 2: Dynamic 3D model of coal fire temperature evolution and distribution
447
4 Coal Fires in 3D
4.1 Visualizing multi-dimensional information
3D visualization based on traditional heat flow modeling alone can no longer meet the
demands of modern coal fire research. Advances in exploration technology make it
possible today to incorporate information on coal fire thermal field effects with respect
to rock mechanics, rock magnetism, and electromagnetic properties of rock, as well as
on radiation and the electromagnetic spectrum of combustion energy (Garland & Hecker
1998). The present research effort attempts to make these physical properties visible in the
computer.
Thermal anomalies are the most important (measurable) indicator of coal fires.
Temperature monitoring has meanwhile reached a very high standard. In the following, a
model for visualization of temperature evolution and distribution on the surface of a 3D
object is presented.
4.2 Algorithms for spatial data interpolation
The precondition for displaying temperature variations in a computer model is a database

that assigns different textures or colors to different temperatures. Temperature data
collected in situ by geophysical means, however, are often not able to cover the surface of
a computer-modeled coal fire in its entirety. Here, algorithms for spatial data interpolation
can help guarantee full coverage. In the present approach, algorithms for 3D hypersurface
fitting and cubic interpolation by Rui (2004) were adopted, namely:
n
(3) T ( x, y, z ) = a0 + a1 x + a2 y + a3 z + ∑ Fi ri 2 ln(ri 2 + εε ) ,
i =1
where T is a point of spatial attribute value, indicating the temperature of a 3D object;

a0, a1, a2, a3, Fi (i = 1,2,...,n) are undetermined coefficients;
(4) ri 2 = ( x − xi ) 2 + ( y − y i ) 2 + ( z − z i ) 2 ,
and ε is the experience parameter relative to the curvature of the 3D object’s (i.e. the coal
fire’s) surface.
448
Temperature values can be interpolated using equation (5):
(5)
The matrix form of this equation is:
(6)
where
and
After interpolation, the previously established texture or color database can be used to
cover the surface of the model according to temperature. Figure 3 shows a 3D coal fire
surface model based on data from Wuda Coalfield.
449
Figure 3: 3D coal fire surface temperature model
References
Cui, Z., Cohn, A.G., Randell, D.A., 1993. Qualitative and topological relationships in spatial databases. In
Proceedings of the 3rd International Symposium on Advances in Spatial Databases, Singapore,
23-25 June 1993. London: Springer Verlag, 296-315.
Garland, M. and Heckbert, P.S., 1998. Simplifying surfaces with color and texture using quadric error
metrics. In Proceedings of the conference on Visualization ’98, Research Triangle Park, North
Carolina, U.S.A., 18-23 October 1998. Los Alamitos, California: IEEE Computer Society
Press, 263-269.
Greuer, R.E, 1984. Transient state simulation of ventilation system in fire condition. In Proceedings of the
3rd International Mine Ventilation Congress, Harrogate, United Kingdom, 13-16 June 1984.
Institution of Mining and Metallurgy (IMM), Leeds, United Kingdom, 44-49.
Rui, X.P., 2004. Research on key technologies for visualization of spatial information. PhD thesis, Remote
Sensing Institute, Chinese Academy of Sciences, Beijing, P.R. China. (in Chinese)
Wu, L.X. and Shi, W.Z., 2003. Spatial modeling technology. In Gold, C. (Ed.), 3D GIS and 3D GMS.
Geography Information Science 19. (in Chinese)
Zhang, J.M. et al., 1998. Research on remote sensing exploration of coal fires. Natural Disaster Reduction
in China 8(1):55-66. (in Chinese)
450
Mathematical Simulations and Methods of

Inversion on Underground Coal Fires
Huan Zhongdan1, Tian Yuan1, Sun Yujing1, and Zhang Jianmin2
1
2
Abstract
Based on results of previous research and in-situ investigations conducted in Wuda Coalfield (Inner
Mongolia Autonomous Region, P.R. China), this paper formulates several 3D dynamic and steady heat
mathematical models of underground coal fire development, and proposes a method for inversion of
heat sources/combustion centers, the so-called multi-point source inversion (MPSI). The models can
be used in a detection-inversion-prediction dynamic coal fire monitoring program: after inversion of a
heat source/combustion center based on its ground thermal field in the steady model, predictions using
the dynamic model can be made on the evolution of this field as well as the underground thermal field.
These predictions operate with general evolution laws and underground geological parameters. The paper
provides relevant numerical algorithms and several numerical simulation results.
摘要
根据现有对中国内蒙古自治区乌达煤田火区调查研究的可利用条件和资料，本文建立了几个地下
煤火的三维空间温度变化的动态和稳态热数学物理模型，并提出了一种反演火源的方法，叫做多
点火源反演法(MSPI)。根据这些模型和方法，本文提出了一套由探测-反演-预测等三部分组成的
煤火动态监测规划，即由地面温度场出发利用稳态模型探求火源、根据动态正演模型和火源变化
规律以及地下地质构造推断未来地面和地下温度场的演变。文中还提供了相关的数值方法和一些
数值模拟结果。
451
1 Background
Underground coals fires cause numerous anomalies in magnetism and electromagnetism, in

temperature and radioactivity, each of which, if measured, provides pertinent information
on a particular fire. At the same time, the lack of proper geophysical models to describe
these anomalies makes it very hard to examine underground coal fires quantitatively (Guan
et al 1998; Zhang 1998). Based on results of previous research and in-situ investigations
conducted in Wuda Coalfield (Inner Mongolia Autonomous Region, P.R. China), we chose
thermal anomalies as the focus of our research. Coal fire-related heat phenomena have
been studied extensively, so there was a plethora of tools and methods to proceed from
(e.g. Lykov 1952; Carslaw & Jaeger 1959; Huan & Huang 2004). In this paper, we will
present some of the model that we have developed and the results from their application;
they show that thermal anomalies are an appropriate choice for underground coal fire
investigations in Wuda.
2 Models and Analyses
Our models were based on the following two assumptions: (i) the heat delivered from a
combustion source in a coal seam is released through different media underground (rock
strata and coal seam(s)); (ii) thermal phenomena (anomalies) are generally detectable
(Figure 1).
Figure 1: Heat flux in an underground coal fire
The tasks that a potential model must be able to perform are: (i) prediction of thermal
anomalies when source, medium, and other environmental factors are known; (ii) inversion
of the heat source/combustion center when thermal anomalies, medium, and other
environmental factors are known.
452
We will not discuss models under very general conditions. Instead, we will formulate
two types of models – dynamic and steady models. Fourier’s law of heat conduction is
assumed to be always valid.
2.1 Dynamic models
Fourier’s law of heat conduction describes the mechanisms of heat flux in a thermal field.
It assigns a heat flux F perpendicular to isothermal layers in this field. The magnitude | F |
of F represents the amount of thermal energy per unit time that is transferred to a lower
temperature layer per unit surface area. If u denotes the thermal field, Fourier’s law has the
form F = −λ ∇u, where λ is the thermal conductivity.
If the medium through which the heat is released is heterogeneous and the thermal
coefficients are thus dependent on the position and/or temperature of u, the conservation of
thermal energy implies that the governing equation is:
∂u 1 3
∂  ∂u 
(1) = ∑ λλ ( x, u )
∂t c( x)ρ ( x) i =1 ∂xi 
 + f ( x, t ) ,
∂xi 
w h e r e λ ( x , u ) ( c a l m - 1 s - 1 d e g - 1 ) , c ( x ) ( c a l g - 1 d e g - 1 ) , a n d ρ ( x ) ( g m - 3 ) a r e t h e
thermal conductivity, specific heat, and density of the medium, respectively;
f(x,t) = ω (x,t) / (c(x) ρ (x)) and ω ( x ,t) (cal m -3 s -1) is the intensity of the heat source/
combustion center.
The dynamic model is useful for prediction of thermal anomalies if the heterogeneity of
the medium is to be accounted for. Its drawback, though, is that exact values of λ , c, and ρ
are required for calculations – they are impossible to obtain in reality!
Models with constant thermal properties can be considered seamwise or piecewise. In

other words, λ , c, and ρ are considered seamwise or piecewise. However, such models are
not practical for investigations in Wuda, as relevant data are lacking. Moreover, they are
very complex mathematically.
Another practical model is the so-called homogenous model:
∂u
(2) = a∆u + f ( x, t ) ,
∂t
where a = λ / (cρ ) and f(x,t) = ω (x,t) / (cρ ). In this case, λ , c, and ρ are constant, or
independent of position.
453
We studied these models on a spatial region Ω = {(x′,x3): x3 > 0} with x′ = (x1, x2).
Heat release from the ground can be modeled with Newton’s law of cooling; here, the
boundary condition of x3 = 0 should be of the Robin type:
 ∂u 
(3)  − +σ u (x′,0, t ) =σσu1 (x′, t ) ,
∂
 3x 
where σ = α / λ , α (cal m -2 s -1 deg -1) is the heat transfer coefficient, u 1 the ambient
temperature, and x′ = (x1, x2). To determine the thermal field underground, the underground
temperature at a specific time has to be known, that is, given an initial condition
u(x,0) = u0(x).
Now dynamic models can be formulated. The homogeneous model, for example, is

 3
ut − a∆u = f ( x, t ), x ∈ R+ , t > 0

(4) u ( x, 0) = u0 ( x), x ∈ R+3 .

 − ∂u + σ u  (x′, 0, t ) = σ u (x′, t ), x′ = (x , x )∈ R 2 , t > 0
 ∂x3 

1 1 2
For a dynamic model to be used for data acquisition on the evolutional behavior of a
thermal field u(x,t), i.e. the spatio-temporal variation of temperature underground, the
thermal conductivity, specific heat, and density of the medium, the intensity of the heat
source, as well as the initial and ambient thermal fields have to be known.
Dynamic models can be used effectively for making predictions on ground thermal
anomaly fields and underground spatio-temporal thermal fields. They are used for these
purposes in the program that we propose.
2.2 Steady models
The second task of coal fire modeling as described above is to inverse heat sources/
combustion centers from thermal anomalies, taking into account all given environmental
factors and media. If we proceeded with dynamic models here, we would have to know the
ground thermal field u(x′,0,t), as well as the ambient temperature u1(x′,0,t), and the initial
underground thermal field u(x,0). To obtain u(x′,0,t) is almost impossible as that would require
knowledge of the ground temperature from time 0 to the present; the same applies to the initial
data of u(x,0). Instead, another model appropriate for inversion needs to be formulated.
454
Underground coal fire development is a rather slow process, and variations of thermal
anomalies in the timeframe covered by a single detection event are rather small. Thus, the
thermal field can be thought of as time-independent, which allows for a steady model to be
formulated as follows:
−a∆u = f ( x), x ∈ R+3

(5)  ∂u 
 − ∂x +σσ u  (x′, 0 ) =σσ u1 (x′ ), x′ = (x1 , x2 )∈ R 2 .
 3 
In this formula, only the ground thermal field u(x′,0) and the ambient temperature u1(x′,0)
are needed for inversion of f(x), which requires much less data than the corresponding
dynamic model.
Steady models are mainly used to inverse heat sources/combustion centers and to calculate
initial underground thermal fields for prediction of thermal anomaly development.
2.3 Numerical simulation of dynamic models
Numerical simulation of 3D models requires a huge amount of work. For the present study,
the standard explicit scheme and the semi-implicit scheme were used. Another scheme
was AOS (Additive Operator Splitting; Weickert et al 1998), which in fact proved more
efficient than the previous two.
For numerical simulation, underground artificial boundary conditions have to be added to

the dynamic model, namely x1 = ±M1, x2 = ±M2, and x3 = H. Here, homogeneous Neumann
conditions were added. We considered the following initial boundary value problem:
ut − a∆u = f ( x, t ), xi < M i , i = 1, 2, 0 < x3 <H, t > 0


u ( x , 0) = u 0 ( x ), x i < M i , i = 1, 2, 0 < x3 <H

(6)  ∂u  ,
 − +σ u  (x′, 0, t ) = σ u1 (x′, t ), xi < M i , i = 1, 2, t > 0
 ∂x3 

 (± M 1 , x2 , x3 ) = 0, ∂u (x1 , ± M 2 , x3 ) = 0, ∂u (x1 , x2 , H ) = 0
∂u
∂
 x1 ∂x2 ∂x3
where Mi and H need to be chosen large enough to ensure that the source f(x,t) is in a
certain distance form the boundary conditions in the time window concerned. The explicit
scheme can be applied without adding artificial boundary conditions, but more (about three
times as many) calculations need to be performed. All three schemes are programmed in
C. They can be applied to general equations, e.g. heterogeneous and anisotropic media.
The heat sources/combustion centers in these simulations are punctiform, globular, or
rectangular in shape.
455
3 Inversion Based on Steady Models
For inversion of heat sources/combustion centers, we first analyzed the homogeneous

steady model and its solution formula. The very intricate task of inversion here is reduced
to an integral equation. Solving this integral equation, however, is still very difficult. The
complexity can be simplified by working with specific types of combustion sources. Based
on the observation that the size of a combustion source is generally much smaller than that
of a coal fire area, we defined such a source as consisting of several point sources. If the
size of the combustion source were large, one would have to adopt other simplified type
sources, for example, globes or rectangles. In this paper, we propose a method of inversion
for multi-point sources (MPSI), where the inversion is reduced to an elementary procedure.
This method will later be validated against measurement data from Wuda Coalfield.
3.1 The homogeneous steady model
The model-governing equation is -a∆u = f(x) with a = λ / (cρ ) and f(x) = ω (x) / (cρ ).
 ∂u 
The boundary condition of this model is  − + σ u  (x′, 0 ) = σ u1 (x′ ), where σ = α / λ .
 ∂x3 
α is the heat transfer coefficient between the medium and the air at the surface, λ the
thermal conductivity of the medium, and u1 the ambient temperature.
The solution of the homogeneous steady model can be split into two parts
u(x) = v(x) + w(x), where v(x) is the anomaly field created by the heat source/combustion
center (ambient temperature u1(x′) = 0), and w(x) is the natural field at ambient temperature
(no source: ƒ(x) = 0).
The anomaly field, i.e. the solution of the model when u1(x′) = 0 is realized through:
v( x) = ∫
R+3
[Γ( x − y) + Γ( x − y )]f ( y )dy
(7) y3
,
−2σ ∫ 3
Γ( x − y ) ∫ f ( y′, s )e −σσ ( y3 − s ) dsdy
R+ 0
11 1 1
where ′ −( xy))==
Γ( xΓ and y′ = ( y1 , y2 ), y = ( y′, − y3 ) .
4π a x x′ − y ′ 2 + y 2
3
y3
fσ ( y ) = f ( y ) − σ ∫ f ( y ′, s)e −σ ( y3 − s )ds
0
456
The natural field, i.e. the solution of the model when ƒ(x) = 0 is realized through:
(8) w( x) = u1 ( x′) + ∫
R+3
[Γ( x − y ) + Γ( x − y )]a∆u1 ( y′)dy.
−2 ∫ R+3
(
−σ yy33
Γ( x − y )a 1 − e-σ )
∆u1 ( y′)dy
If u1(x′) = u1, then w(x) = u1, which suggests that calculating the anomaly field is much
simpler if the ambient temperature is constant. To obtain formulae (7) and (8), we reduced
the problems to calculate the solution of -a∆u = f(x) in the whole space for a properly
chosen f(x), the details of which can be found in Huan & Huang (2004).
3.2 Inversion of heat sources/combustion centers in the homogeneous

steady model
To inverse intensities ω (x) of heat sources/combustion centers, a large amount of data

from in-situ measurements is needed; these are the ground thermal field β (x′) = u(x1,x2,0),
the ambient temperature u1(x′), as well as geophysical parameters – thermal conductivity
λ , specific heat c, density ρ of the overlying medium (underground), and the heat transfer
coefficient α between the medium and the air at the surface.
First, the influence of ambient temperature in the measured data β (x ′ ) needs to be

eliminated. The anomaly field is ϕ (x′) = β (x′) - w(x′,0). The source can be recovered by
solving the integral equation:
(9) ϕ (x′ ) = 2 ∫ Γ (x′ − y )fσσ ( y )dy ,

R+3
where ƒ(x) = ω (x) / (cρ) , and:
11 11
Γ((xx′′ −− yy)) ==
Γ
(10) 4π a
4π 22
xx′′−− yy′ + y3322 .
y3 y3 −σσ ( y3 − s )
fσ (fσy )( y=) =f ( fy() y−)σ−∫σσ ∫0f (fy(′,ys′,)se)−eσ ( y3 − s )dsds
0
Equation (9) is a Fredholm integral equation of the first kind, which is normally solved by
regularization. This, however, has so far been impossible. Instead, we focused on developing
methods for inversion of specific types of sources in equation (9), e.g. point, globe, or
rectangle sources. A method for point sources has been developed successfully.
457
3.3 Multi-point source inversion (MPSI)
MPSI is based on the observation that the size of heat sources/combustion centers in coal
fires is much smaller than the entire coal fire area; it tends to be even smaller that what is
generally taken as the smallest spatial unit of detection. Therefore, it is possible to think of
a coal fire heat source/combustion center as consisting of only a few single-point sources.
A source consisting of only one such single-point source is called a point source.
The underlying assumption of MPSI is that a single-point source in a point source

corresponds to a peak of the (ground) anomaly field and vice versa.
For inversion, we first focused on a point source consisting of one single-point source.
We assumed that the single-point source is at (b, h) of intensity ω . b represents the
ground coordinates (b1,b2), and h the depth of the single-point source. χ = ω / (cρ ) is the
intensity coefficient of the single-point source. a = λ / (cρ ) and σ = α / λ are the thermal
conductivity and the heat transfer coefficient, respectively. The anomaly field created by
the single-point source is:
 
∫
∞
−σ s
χ  1 e-σ
(11) ϕϕ ( x′) = σ
−σ ds 
2π a 
2πα 2
x′ − b + h 2 x′ − b + (s + h )
2 2 
 0 
It can be easily verified that: (i) ϕ reaches its maximum at b; (ii) ϕ increases linearly with
the intensity coefficient χ ; (iii) ϕ (x′) decreases as the heat transfer coefficient σ or the
thermal conduction coefficient a increases; and (iv) ϕ (x′) decreases and its graph flattens as
the depth h increases. Given these interdependencies, the following inversion scheme for
single-point sources was developed:
1) Identify the maximum point b of ϕ (x′) and make it the ground coordinate of the source.
2) Select an adjacent point b different from b and obtain the depth h = 1 / r by solving
j( b ) / j(b) = g(r) for r, where:
 
∫
∞
 1 1  -σ
−σ s
 2
−
2 e ds
r 2 b − b + (rs + 1)
2
0  r2 b − b +1 
(12) g (r ) =   .
∫
∞
−σ s
sre-σ
ds
1 + sr
0
3) Calculate the intensity coefficient χ .
458
For MPSI, the scheme looks as follows:
1) Find the peaks of the anomaly field.

2) Make each peak a single-point source with the same ground coordinates and calculate
its depth h according to the inversion scheme for single-point sources.
3) Obtain the intensity coefficients by optimization.
4) Modify positions of single-point sources.
When the intensities and depths of single-point heat sources of a coal fire are roughly of
the same order and the distance among them is rather large, the peaks of the anomaly field
are very distinct and MPSI works well. If, however, the intensities, depths, and distances
do not match well, MPSI may miss some single-point sources. It can still provide useful
information though.
3.4 Numerical simulation of MPSI
To validate our studies on MPSI, we first formulated three tri-point test groups for
application, and then checked the results against data from fire zone (FZ) 10 of Wuda
Coalfield.
3.4.1 Results of test groups

Test group 1: All three single-point sources had the same intensity coefficient χ = 1000
with loci P1(25,10,15), P2(-25,10,15), and P3(0,0,14). This was a situation where intensities
and depths of single-point sources were roughly of the same order and the distance among
them was large enough to produce three clearly separated local maxima (σ = 0.5 and
2π a = 1; Figure 2).
459
Figure 2: Thermal anomaly field for test group 1
Test group 2: In test group 2, the single-point sources had different intensity coefficients
χ 1 = χ 2 = 400, χ 3 = 1000 with the same loci as group 1 P1(25,10,15), P2(-25,10,15), and
P3(0,0,14). This was a situation where the intensities of the single-point sources were at a
large difference, the depths were roughly of the same order, and the distance among them
was still large enough to produce three clearly separated local maxima. In real detection,
the smaller anomalies might have gone unnoticed due to noise, such as osculation of
ambient temperature or measurement errors (σ = 0.5 and 2π a = 1; Figure 3).
460
Test group 3: In test group 3, all three single-point sources had the same intensity
coefficient χ = 1000 but with much closer loci P1(5,10,15), P2(-5,10,15), and P3(0,0,14). In
this scenario, there was only one local maximum (σ = 0.5 and 2π a = 1¸ Figure 4).
461
The inversion results of test groups 1 and 2 with steps 1 to 3 of the MPSI scheme are
given in Table 1, together with comparable in-situ measurements at FZ 10. We believe
that the results would improve after iterations, which are currently being undertaken.
Table 1: Inversion results of test groups 1 and 2 with steps 1 to 3 of MPSI scheme
Test Parameters Ground Coordinates Depth Intensity
Group (m) (m)
Sources Real MPSI Real MPSI Real MPSI
Group 1 P1 (25,10) (23.6,10) 15.0 16.35 1000 992.57
P2 (-25,10) (-23.6,10) 15.0 16.35 1000 943.96
P3 (0,0) (0,0.6) 14.0 15.97 1000 1178.97
Group 2 P1 (15,10) (16.5,11.5) 15.0 15.17 400 263.28
P2 (-15,10) (-16.5,11.5) 15.0 15.17 400 321.57
P3 (0,0) (0,0.85) 14.0 15.67 1000 1336.12
Error G1+G2 Absolute Error Relative Error Relative Error
1.41 m 7.59 % 16.53 %
3.4.2 Validation of results

The test area from which data for validation of our models were taken is located in Wuda
Coalfield (Inner Mongolia Autonomous Region, P.R. China). Wuda Coalfield covers an
area of 35 km2 in the west of Wuhai City (Figure 5). It harbors 19 mineable coal seams, of
which coal seams nos. 1, 2, 4, 6, 7, and 8 have already been mined out. Seams nos. 9, 10,
13 are currently being mined. 16 FZs have been identified for the area; they extend over
c. 3,496,000 m2. In some parts, the combustion depth reaches as far as 84 m. The area of
active combustion and intensive burning was about 422,000 m2 in 2004, entailing huge
losses for the local coal industry.
FZ 10 of Wuda Coalfield is a typical coal fire area. In July 2004 and January 2005,
wecollected detailed multi-phase geological and ground temperature data in the region.
The data obtained at non-fissure observation points were used as base data in this
study.
462
Figure 5: Map of Wuda Coalfield (Inner Mongolia Autonomous Region, P.R. China)
Figure 6 is a comprehensive high-resolution satellite map of FZ 10 indicating the location

of observation points (Quickbird; pixel size 0.6 × 0.6 m). Spots in the image are identical
with scattered grass. Observation points are marked with a plus +; measurements here
included temperature, fissures and cracks, etc. Points with a red triangle Δ were chosen for
inversion. Ground and ambient temperatures at each point were acquired using a portable
thermal infrared thermometer, at a radiation rate of 0.92.
Figure 6: Quickbird image of FZ 10 indicating observation points (+) and inversion
points (Δ), January 2005
463
Since MSPI is based on the condition that thermal conduction take place in a homogenous
medium, it is important to choose input data which fulfill this condition. Hence, only
observation points with bare rocks as overlying strata in the thermal anomaly zone were
chosen for inversion. These points and their coordinates are listed in Table 2.
Table 2: Observation points in FZ 10 of Wuda Coalfield

Observation X Y Ground Thermal Elevation
Point Coordinate Coordinate Temperature Anomaly (m)
(ºC) (ºC)
48 382460.95 4376426.72 49 27 1217.35
161 382413.18 4376339.50 35 13 1220.00
169 382437.78 4376266.90 46 24 1214.01
166 382407.18 4376268.10 26 4 1218.91
173 382423.38 4376326.30 45 23 1220.00
174 382459.38 4376427.70 47 25 1217.62
Three candidate peaks were selected – points 48, 169, and 173; as well as three adjacent
observation points, namely 174, 166, and 161. The heat transfer coefficient in the area was
σ = 0.03; the thermal conduction coefficient a = 0.5.
Table 3 plots the results of MPSI against the data measured in situ. The real depth was
chosen to be the thickness of the roof above a coal seam, since heat sources/combustion
centers are located within coal seams. It was acquired from a 3D visualized geological
model, itself based on various data from borehole measurements and investigations of coal
seams and mine workings and infrastructure.
Table 3: Inversion results vs. measured data from FZ 10 of Wuda Coalfield
Observation Point Thermal Anomaly Depth Inversed Intensity
(ºC) (m) (ºC h-1)
Real Inversed Real Inversed
48 27 26.045 19.21 18.64 814.99
174(A)a 25 25.857
169 24 24.002 10.76 10.636 369.87
a
166(A) 4 4.547
173 23 21.824 20.36 16.22 948.35
a
161(A) 13 14.750
a
(A) indicates the adjacent point to the inversion point directly above in the column.
464
The results from MPSI calculations proved satisfactorily close to the values measured in
situ. The depth calculated for observation points 48 and 169 was surprisingly accurate. The
error in the depth calculated for point 173 is larger, but still decent.
The temperature values for the observation points are also presented in Table 3. The root-
mean-square error here is 1.032 °C, the relative error is 5.33 %. The root-mean-square
error for depth is 1.39 m, the relative error 8.3 %.
4 Conclusion
From results of previous research and our own in-situ observations in Wuda Coalfield, we
were able to conclude that the mathematical inversion of heat sources/combustion centers
of coal fires can be performed on the basis of ground thermal anomalies in a steady model.
After simplifying geological conditions and heat source structure, we developed a method
to calculate the depth and intensity of coal fire heat sources/combustion centers by means
of inversion (MPSI). In applying this method, we found that besides obtaining relevant
geological parameters, it is important to make topography and background corrections in
ground temperature data. Also, fissure observation data should be removed.
Based on our findings, we propose a detection-inversion-prediction dynamic coal fire

monitoring program. After inversion of a heat source/combustion center based on its
ground thermal field in the steady model, predictions using the dynamic model can
be made on the evolution of this field as well as the underground thermal field. These
predictions operate with general evolution laws and underground geological parameters.
The program is sketched in Figure 7.
Figure 7: Detection-inversion-prediction dynamic coal fire monitoring program
465
References
Carslaw, H.S. and Jaeger, J.C., 1959. Conduction of heat in solids. New York: Oxford University Press.
Huan, Z.D. and Huang, H.Y., 2004. Partial differential equations. Beijing: Higher Education Press. (in
Chinese)
Lykov, A.V., 1952. Theory of heat conductivity. Moscow: State Technological-Theoretical Literature
Publishing House. (in Russian, Chinese translation from 1954)
Weickert, J., ter Haar Romeny, B.M., and Viergever, M.A., 1998. Efficient and reliable schemes for non-
linear diffusion filtering. IEEE Transactions on Image Processing 7(3): 398-410.
466
Coal Fires Between Self-Ignition and
Fire Fighting: Numerical Modeling and
Basic Geophysical Measurements
Winfried Kessels, Manfred W. Wuttke, Stefan Wessling, and Li Xuan
Leibniz Institute for Applied Geosciences (GGA), Hanover, Germany
Abstract
Understanding spontaneous combustion in underground coal seams and coal stockpiles is a precondition
for effective coal fire fighting and prevention. One possible key is the analysis of oxygen transport to
the combustion zone. Gas flow measurements in Wuda Coal Mining Area (Inner Mongolia Autonomous
Region, P.R. China) and a coal heap near Mehrum, Germany confirmed that convection is the dominant
process in oxygen transport. The driving forces behind convective oxygen flow may be weather conditions
such as changing wind direction and velocity, free thermal buoyancy, or differences between atmospheric
and mine ventilation pressure. It was the aim of the GGA contribution to the Sino-German Coal Fire
Research Initiative “Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires
in North China” to quantify coal fire development below surface by scenario simulations using applied
numerical methods. In this paper, we present basic physico-chemical phenomena of underground coal fire
development, as well as a number of numerical simulations. The influence of mine ventilation on the self-
ignition behavior of coal is also illustrated by way of simulation.
摘要
了解在地下煤矿和煤堆的煤的自燃现象是有效的灭火和预防煤火的先决条件。其中一个基本点就
是了解氧气在燃烧区的传送。在中国神华集团乌达煤矿区和德国靠近梅尔姆地区的煤堆进行的气
体流动检测表明，对流是主要的气体传输过程。导致氧气对流的原因有可能是气候原因，比如风
力和风向的变化，自由热能浮动或者是大气和煤矿内流通空气的压力差。通过建立数学模型和情
况模拟来量化地下燃烧发展是GGA研究所在这个中德煤火研究，即“中国北方地区煤火探测、灭
火与监测新技术”项目中所承担的任务。在此文中，我们主要介绍了发生在地下煤火区物理-化学
现象以及大量依据于煤堆监测数据的数学模拟。同时，还通过模拟阐明了煤矿区的通风对煤缝隙
处的自燃现象的影响。
467
1 Modeling in Coal Fire Fighting
Coal fire fighting is of use only if it is able to extinguish a fire and prevent its re-ignition
on a permanent basis, namely through:
• Coverage of the burning area with soil to suffocate the fire;

• Cooling of certain areas to assist and accelerate fire extinction; or
• Optimization of mining infrastructure to minimize fire hazards.
These operations involve the use of heavy machinery, the relocation of substantial amounts
of soil, and a great deal of manpower – they entail large costs, which will be compensated
for only much later in time and on a hardly measurable scale by a reduced or best halted
loss of resources (coal) and environmental benefits such as capped greenhouse gas
emissions. Predictive modeling can assist in controlling these costs in a more sustainable
manner. Figure 1 sketches the principle of coal fire suffocation by soil coverage or sealing.
An uncovered fire would continue to release emission gases through the surface and
supply itself with oxygen from air entering the overlying rock strata. This convection-
driven process depends on the degree of permeability of the materials above and around
the fire. Oxygen supply can be controlled by an impermeable layer on the surface that
is sufficiently thick to isolate the fire from the atmosphere. If the permeabilities of all
materials are known, it is possible to estimate by way of predictive modeling the amount
of soil necessary for permanent fire extinction. This, however, presupposes that oxygen
cannot reach the fire from other directions – underground mining facilities for example.
Figure 1: Coal fire suffocation by soil coverage or sealing
468
Figure 2 gives an idea of the dimensions involved in the process of sealing. It shows
a satellite view of fire zone (FZ) 3-2 of Wuda Coal Mining Area (Inner Mongolia
Autonomous Region, P.R. China). The marked fractures were caused by longwall mining;
they are the main pathways for oxygen and emission gases.
Figure 2: Area (orange) to be covered with soil for post-mining fire prevention
469
Figure 3 illustrates another method of coal fire fighting – cooling of fire environment
with water. The reduced temperature draws stored heat from surrounding bedrock and
thus hinders re-ignition in residual coal. Predictive modeling here can help estimate the
exact amount of water needed and the temperatures resulting from cooling in a given fire
scenario.
Figure 3: Cooling of soil before coal fire coverage
470
Figure 4 points to a crucial problem in coal fire fighting and prevention – ventilation of
mining galleries. Indispensable for mine staff, ventilation may also provide coal fires with
oxygen and propel combustion, especially if barriers between active mining galleries and
abandoned shafts (i.e. those locations where spontaneous combustion may occur unnoticed
and unhindered) are not sealed off properly. This effect was simulated with the numerical
1D transport model code NTRANS. Results for three different air velocities are shown in
Figure 5. Between the surface and the mining gallery, a pressure difference of 10 mbar was
assumed to prevail due to mine ventilation. During the first month, the reaction between
coal and oxygen was dominated by reaction kinetics. Reactivity, however, increased
with increasing temperature (diagrams in middle column), and finally tilted the reaction
behavior toward transport-dominated mechanisms. Coal concentration (diagrams in the left
column) moved with the combustion zone in a rather sharp front.
Figure 4: Ventilation facilities in Wuda Coal Mining Area
471
Figure 5: Simulation of reactive transport in coal combustion with strong air flow
induced by mine ventilation, measured for one year in 20 time steps
(FD Program NTRANS)
Data source: Kessels and Wessling (2006)
472
The coal in these simulations reacted from time step to time step only in the first numerical
elements at the boundary between inflowing air and coal, at the very top of the coal seam.
The development of oxygen concentration in the right columns documents the emergence
of a sharp combustion front after only two weeks.
Figure 6 is a picture of FZ 3 in Wuda Coal Mining Area. While a good example of

the drastic environmental impact of coal fires, it also stresses (through the massive
development of smoke) the important role that convection plays for transport of
reactants and reaction products. Similar examples of combustion processes affecting the
environment and human health are waste disposal, internal combustion techniques for
enhanced oil-field recovery (Brenner et al. 2000; Krause & Schmidt 2001; Perkins &
Sahajwalla 2005), and combustion in power plants (Hjertager 1986; Oran & Boris 2001).
Figure 6: FZ 3 in Wuda Coal Mining Area

Data source: Stracher and Taylor (2004)
If combustion is viewed as a physico-chemical process, the hydraulic fluid is air, and

the reactive air component is oxygen. The reaction between oxygen and coal is strongly
dependent on temperature and can be described by an Arrhenius-type function. Oxygen
transport, on the other hand, depends on a complex flow/flux system. In common reactive
multi-component transport calculations, the concentration of different species is described
473
by a conservation equation in space and time. The source term in this equation is usually
calculated from a functional dependence of the reaction parameters on temperature and
the concentrations of other reactive partners. Model discretization in space and time is
chosen from this functional dependence so as to produce realistic and stable calculations;
it yields very small time steps and very small spatial units for areas with high reactivity.
To avoid the unreasonably long calculation times that simulations of this type normally
entail, we made a distinction in our approach between reaction-dominated and transport-
dominated numerical elements based on the following relationship between relaxation time
τ (depending on reactivity), numerical time step Δt, penetration length L (depending on
reactivity and flow velocity; see below), and element length Δx:
τ < Δt and L < Δx - τ > Δt and L < Δx - τ < Δt and L > Δx - τ>Δt and L > Δx.
To introduce τ and L, we proceeded from the well-known transport equation (1):
∂c
(1) ∇ • j + Φ ⋅ i − Qi = 0 ,
i ∂t
where ji is the flow density of a species i, Φ the porosity, ci the species’ concentration in
the fluid, and Qi the production/consumption term. By ignoring the transport term (i.e.
assuming j = 0) and assuming a first-order reaction for the production term:
(2) Qi = K i ⋅ ci ,
we were able to arrive at equation (3) for a constant reaction rate valid for a starting
concentration c0:
−
( )
K i c i0
⋅t
0 − τ⋅t
(3) ci (t ) = c ⋅ e
i
Φ
= ci0 ⋅ e .
Thus, the relaxation time τ for a reactive process is the time needed to reduce
a given concentration of a reactive species i by a factor 1/e without flow of the
reactive species:
Φ
(4) ττ = .
( )
K i ci0
474
Proceeding from the same equation (1), but ignoring diffusion and dispersion, and instead
assuming stationary concentration, first-order reaction, and constant porosity and velocity,
we arrived at:
−
()
Ki ci0
⋅x
0 Φ⋅v fluid −x
(5) ci ( x) = c ⋅ e
i = ci0 ⋅ e L ,
with a convective penetration length:
Φ ⋅ v fluid
(6) L= = τ ⋅ v fluid .
( )
K i ci0
The reaction between atmospheric oxygen and coal offers itself as an example of a
particular reaction rate. The formula for the production rate of this reaction has been given
by Kessels and Wessling (2006):
β
−
(7) QO2 = k O2 ⋅ e k ⋅T
⋅ cO2 ⋅ ccoal .
Its derivation is based on investigations and interpretations by Schmidt et al. (2005) and
Krause et al. (2008) in this volume, carried out for different types of coal from the study
area of the Sino-German Coal Fire Research Initiative “Innovative Technologies for
Exploration, Extinction, and Monitoring of Coal Fires in North China.” The concentration
of coal here is equivalent to coal bulk density. The oxygen concentration is given as partial
density in the pores. Figure 7 shows temperature-dependent relaxation times derived
from experimental kinetic data (Krause & Schmidt 2001). The corresponding penetration
lengths according to equation (6) are plotted in Figure 8.
Figure 7: Relaxation times for oxygen consumption in first-order reaction with constant
coal concentration (grain size given in round brackets)
Data source: Krause and Schmidt (2001); Kessels et al. (2005)
475
Figure 8: Penetration length L of oxygen concentration in fine-grained coal (< 0.5 mm)

for different air velocities.
Kinetic data for Wuda coal from: Krause and Schmidt (2001)
The pronounced variation of the relaxation times over more than 12 orders of magnitude
at temperatures in a range between 300 and 1,200 K illustrates the general problem in
calculating coal fire development from spontaneous combustion to a fully burning coal
seam. Modeling of temperature distribution in Wuda FZ 3 by Wessling et al. (2008; in this
volume) showed that temperatures may range from 300 to up to 1,300 K over all model
elements. This leads to relaxation times from less then 1 second to over 100 years for all
model elements. To achieve numerical stability, the time step Δt chosen for calculation
should be smaller then the smallest relaxation time of any element (Figure 9) – another
source of prolonged calculation times.
Figure 9: Selection of calculation time steps depending on reaction rate
476
2 New Approaches
The method presented in the following aims to achieve reasonable calculation times
for simulations of fast reaction processes such as in coal fires. Its basis is an element
discretization algorithm that distinguishes between reaction- and transport-dominated
numerical elements.
For reaction-dominated elements with Δt << τ, the reacted mass between t and t + Δt in an
element is much smaller than the reactive mass available at time t. Concentration values at
t + Δt can be extrapolated from t (Figure 10). Negative concentrations for elements will not
occur if the general stability criterion for classic transport calculations is fulfilled.
Figure 10: Particle path vs. discretization length in numerical grid
For transport-dominated elements with Δt > τ the inflow of reactive components into the
numerical element stops immediately when the element boundary is passed. This means
that the conditions c(t) = 0 and c(t + Δt) = 0 are both valid. Reactive mass is determined
by species inflow and not by reactivity. A transport calculation within the element is not
necessary because the amount of entering oxygen is given by the neighboring elements.
Analogous to the Courant criterion, the convective penetration length depends on the
calculated velocity field.
477
Operator-splitting as described above can be used if: (i) the diffusion is large enough, (ii) the
discretization length Δx is greater than the particle path Δs in the time step Δt, and (iii) Δx is
smaller than the convective penetration length L. The calculation results presented in Wessling
et al. (2008; in this volume) are based on operator-splitting; they were obtained with a modified
ROCKFLOW code. The advantage of this method is that it uses an existing procedure, which
has already been applied successfully for chemical reactions in equilibrium.
The newly developed sharp front method distinguishes between reaction-dominated

numerical elements (RE) and transport-dominated numerical elements (TE). For coal fire
calculations, we used this method to compute the transport and one-way reaction processes
for the following chain:
Oxygen → coal → heat → carbon dioxide → ash.
Backward coupling was accounted for by variation of the state conditions for process
acceleration. We calculated the state conditions at time t + Δt on the basis of their status at
time t. Modeling was realized in five steps:
1) Calculate τ for all elements and compare with Δt to distinguish TE and RE.
2) Calculate oxygen concentration for very small time step Δt without chemical reactions.
Based on this value, calculate oxygen production rate (or consumption rate in the case
of coal fires) from oxygen accumulated in TEs.
3) Set oxygen concentration for all TEs at zero and calculate oxygen concentration for all
REs in time step Δt.
4) Calculate production rates of coal → heat → carbon dioxide → ash on the basis of
production rate determined for oxygen.
5) Proceed to next time step.
Because temperature and concentration fields are dynamic, the transport- and reaction-
dominated elements may vary from time step to time step. Step 1, i.e. the calculation of the
relaxation time, may be replaced by linear interpolation. After calculation of full oxygen
concentration for a time step Δt, negative concentrations for other elements may result.
These are then classified as transport-dominated. For all calculations, the classic criterion
of stability has to be solved.
478
3 Monitoring Examples and Model Calibration
Monitoring of coal fire-induced gas circulation in the ground and above generally
comprises pressure and temperature measurements of surface-near atmosphere and
emission gases (Figure 11). Our measurements in FZ 3-2 of Wuda Coal Mining Area
revealed temporal variations for both quantities. Although in-situ observations point to
a distinct relationship between them, diurnal variations and (anti-)correlations on longer
time scales failed to reveal such (see Figure 12). Our own simulations did not produce any
definite proofs either. This divergence might be due to higher gas flow in fractures, which
could only be treated numerically for relatively low flow velocities. Effects of pressure
variation on coal fire burning intensity could not be confirmed because there were no
boreholes in FZ 3-2 at the time of our measurements.
Figure 11: Monitoring in Wuda Coal Mining Area. Maintenance of data loggers.
479
Figure 12: Results of temperature monitoring in FZ 3-2 of Wuda Coal Mining Area
A complicated geological setting as it prevails in the study area of the Sino-German Coal
Fire Research Initiative makes a calibration of a coupled transport model with in-situ
measurement data almost impossible. Calibration experiments, however, are essential for
assessing the accuracy of modeling results. Experimental investigations to validate our
model were therefore performed in Germany, on a coal stockpile near Mehrum. These
measurements confirmed the low thermal conductivity of coal. In the interior of the
heap, in 2 to 3 m depth, a rise in temperature of up to 70 K after 5 months was measured.
Monitoring of temperature and oxygen concentration also proved that convection occurred
within the heap. This was later confirmed by simulations with the FE code ROCKFLOW.
The simulations also enabled us to determine reaction-kinetic parameters of self-heating,
which can serve to upscale laboratory results to real-scale scenarios.
480
Figure 13: Coal stockpile near Mehrum, Germany, used for model validation and
calibration. Simplified geometry, oxygen concentration, and temperature
distribution along profile (bottom).
481
Figure 14: Coal stockpile near Mehrum, Germany. a) Location of measurement points.
b) Temperature evolution with time at measurement points. c) Temperature
evolution with time along profile.
482
4 Conclusion
The most important result of our research is the successful implementation of an element
discretization algorithm that reduces the unreasonably long computing times common in
coal fire simulations to this day. The algorithm is based on a strict separation of spatial
regions, in which the modeled fire is dominated by either reaction kinetics (RE) or the
amount of oxygen supplied, i.e. transport mechanisms (TE). Two slightly different methods
were developed. The first is an extension of operator-splitting and was implemented in the
code ROCKFLOW. The second, called sharp-front-method, was implemented in the code
NTRANS.
Transport modeling underlines the importance for coal fire development of the combined
effect of low thermal conductivity and oxygen supply via pathways in disrupted media.
The natural convection system which provides the fire with oxygen is influenced by
meteorological variations above the surface.
Our numerical models were calibrated with results from large-scale experiments on a coal
stockpile near Mehrum, Germany.
Based on the results of this study and our observations in the field, we feel urged to make
the following recommendations for coal fire prevention:
1) Work the seam properly and extract as much coal as possible; in particular, take care in
removing small-grained residues after mining.
2) Inertize residual coal by powder or fluid injection or sealing to reduce reactivity,
especially at the coal face.
3) Cut off caved areas from ventilation systems of active mining areas.
4) Cover caved areas with material of low permeability above ground.
References
Brenner, G., Pickenäcker, K., Pickenäcker, O., Trimis, D., Wawrzinek, K., and Weber, T., 2000. Numerical
and experimental investigation of matrix-stabilized methane/air combustion in porous inert
media. Combustion and Flame 123: 201-213.
Hjertager, B.H., 1986. Three-dimensional modeling of flow, heat transfer, and combustion. In
Cheremisinoff, N.P. (Ed.), Handbook of heat and mass transfer. Vol. I: Heat transfer
operations. Houston, Texas: Golf Publishing Company, 1303-1350.
483
Kessels, W. and Wessling, S., 2006. Uncertainties in modeling and parameterization of fluid-rock
consumption-production interaction in porous media. In Calibration and reliability in
groundwater modeling: From uncertainty to decision-making. Proceedings of ModelCare
2005, The Hague, Netherlands, June 2005. IAHS Publ. 304, 143-150.
Krause, U. and Schmidt, M., 2001. The influence of initial conditions on the propagation of smoldering
fires in dust accumulations. Journal of Loss Prevention in the Process Industries 14: 527-532.
Krause, U., Schmidt, M., and Lohrer; C., 2008. Computation of real-scale coal fire scenarios based on
linking lab-scale experiments with a numerical heat and mass transfer model. In UNESCO
Office Beijing (Ed.), Spontaneous coal seam fires: Mitigating a global disaster. International
Research for sustainable control and management. ERSEC Ecological Book Series – 4.
Beijing: Tsinghua University Press and Springer Verlag, 485-512. (this volume)
Oran, E.S. and Boris, J.P., 2001. Numerical simulation of reactive flow. 2nd edition. Cambridge: Cambridge
University Press.
Perkins, G. and Sahajwalla, V., 2005. A mathematical model for the chemical reaction of a semi-infinite
block of coal in underground coal gasification. Energy and Fuels 19: 1679-1692.
Wessling, S., Litschke, T., Wiegand, J., Schlömer, S. and Kessels, W., 2008. Simulating dynamic
subsurface coal fires and its applications. In UNESCO Office Beijing (Ed.), Spontaneous coal
484
Computation of Real-Scale Coal Fire
Scenarios Based on Linking Lab-Scale
Experiments with a Numerical Heat and
Mass Transfer Model
Ulrich Krause1, Martin Schmidt2, and Christian Lohrer3
Federal Institute for Materials Research and Testing (BAM), Berlin, Germany
1
Division VII.3 “Fire Engineering”
2
Division II.2 “Reactive Substances and Systems”
3
Division II.1 “Gases, Gas Plants”
Abstract
Accidental fires in only very rare cases can be investigated on the large scale they typically occur on.
Instead, methods have to be developed to simulate fire scenarios on a laboratory or medium scale and to
transfer the data and observations obtained to real-life situations. The present paper describes laboratory
experiments conducted to measure coal properties, characterize predominant chemical reactions, and
provide validation data for numerical simulation of real-scale coal fire scenarios.
A numerical model consisting of a set of partial differential equations for heat and mass transfer of
chemical species was developed to describe the onset of self-ignition and smoldering fire propagation in
deposits of bulk materials and dust accumulations. The chemical reaction sub-model includes solid fuel
decomposition as well as combustion of char, carbon CO, and H2. As an additional feature the role of
moisture can be included too.
The model was validated against data from lab-scale self-ignition and smoldering fire propagation
experiments, and subsequently applied to simulations of fire scenarios in a coal silo and in a coal seam.
The computations allow to trace hot spot formation and reaction front propagation with time; they
highlight the role of O2 and moisture in fire propagation.
485
摘要
大比例尺下煤火现状的研究只有在极偶然情况下才可以进行。所以，有必要在实验室或方法上开
发一套手段模拟煤火现状，并将这些数据和观测结果调整到现场实际情况。本文描述了在实验室
进行煤特性测量的试验研究，突出了主要煤化学反应，并为后续数值模拟实际尺度的煤火情形提
供了可信的数据。
为描述块状煤层体和残煤堆积中的自燃、闷燃和燃烧扩散过程，提出了由一组煤化学组分的热和
质量转换偏微分方程数值模型，化学作用子模型还包括了固体燃烧物的分解和焦化、CO和H2等，
模型将含水量影响作为一种辅助作用纳入。
该模型已经实验室自燃和闷燃试验数据的验证，适用于模拟实体煤和煤层条件下的燃烧状态。计
算表明点燃烧的形成与热传播作用过程与前述试验是一致的。
1 Introduction
Numerous problems in fire science are of large scale, among them the ignition and
propagation of fires in virgin or mined coal seams, storage deposits of coal, or coal waste
heaps. Real-scale scientific experiments on such fires are very rare due to the high costs
involved and the difficulty to control the enormous amounts of heat and pollutants released
from the fires. Schmal (1987) performed experiments on stockpiles of Australian black
coal in volumes from 2,000 to 3,000 m³, measuring temperature development during low-
rate oxidation. Experiments were stopped when the temperature in the stockpiles reached
90 °C; the hot spots were then excavated and the piles left to cool down.
Using Spanish black coal stockpiled at similar volumes, Fierro et al. (2001) investigated
the role of wind-driven convection on self-ignition. They found that signs of self-
ignition appeared first in those parts of the stockpile surface that were exposed to wind.
Unfortunately, temperatures were measured close to the surface only, so that some of the
conclusions may be erroneous.
Self-ignition occurs when the thermal equilibrium between the two counteracting effects
of heat release from oxidation and heat dissipation away from the bulk material is tilted.
If the rate of heat production exceeds that of heat loss, the temperature within the material
rises and further accelerates the oxidation process. This positive feed-back loop ultimately
results in self-ignition.
It is out of the scope of this paper to summarize the vast literature that exists on self-
ignition hazards with regard to experimental methods and the so-called thermal explosion
486
theory. Standard references are e.g. the monograph by Bowes (1984), and the related
chapter in Babrauskas (2003).
This paper focuses on numerical model applications to address on a technical scale the
problem of self-ignition and smoldering fire propagation; it also presents small- and
medium-scale experimental investigations conducted to support the model.
Two decades ago, Ohlemiller (1985) provided a comprehensive summary of the models of
his time (most of them non-dimensional) and the relevant assumptions and simplifications
they contained; he compared them to what he called a benchmark model (itself multi-
dimensional). Ohlemiller concluded that the models were not able to reflect smoldering
fire propagation in a realistic manner and that a general model comprising heat and mass
transfer through porous beds of solid fuels combined with detailed chemical kinetics on
particle scale was still too complex to be realized.
Nonetheless, a number of attempts were made in later years to compute temperature and
concentration fields and their evolution with time in real-scale fire scenarios. These models
assumed that the solid fuel was an isotropic solid body; variations in its structure (pore size
distribution, inhomogeneity) were neglected.
One-dimensional finite element calculations on the self-ignition of coal were published by

Winters and Cliffe (1985) and Liang and Tanaka (1987). Reproducing basket experiments
on German black coal dust by Leuschke (1983), Liang and Tanaka were able to mirror the
spatial movement of the point where self-ignition occurs first in a dust deposit. The point
moved closer to the surface with increasing oven temperature (the deposit being stored at a
temperature above the critical level).
Among the more recent models, that of Krishnaswamy et al. (1996) contains a solution of
the temperature field coupled with the velocity field, the concentration field of oxygen, and
the pressure field. While the latter three are considered constant with time, the temperature
field is transient. The model was applied to coal stockpiles, simulating temperature
evolution with time for different conditions of steady-state flow, oxygen concentration, and
pressure fields.
Hull et al. (1997) used a very similar approach to investigate the role of oxygen diffusion
in self-ignition of coal stockpiles. They computed the time-dependent temperature field and
the steady-state oxygen concentration field. Their main conclusion was that compaction
487
impedes the diffusion of oxygen to the active sites of coal particles in the bulk deposit and
thus leads to a higher level of safety.
Schmidt and Krause (2000) and Schmidt (2001) developed a model consisting of separate
transport equations for five chemical species (solid fuel, oxygen, solid product, gaseous
product, and nitrogen) solved together with the temperature field equation.
Akgün and Essenhigh (2001) presented a two-dimensional time-dependent model that

considers mass conservation for oxygen and moisture, and energy conservation for the gas
and solid phases. Convective heat and gas transport are accounted for by applying Darcy’s
law. The model was applied to investigate the dependence of temperature development on
slope in coal stockpiles.
The influence of moisture on self-ignition was investigated by Chen (1998) and Chong
and Chen (1999), albeit with milk powder as test substance. Their models contained
conservation equations for energy and mass fractions of liquid and vaporized water.
The purpose of the model presented in this paper is to enable computation of real-scale
fire scenarios in coal seams, coal stockpiles, and coal waste heaps. The model is based
on the conservation of energy and mass of eight chemical species. A four-step reaction
mechanism is assumed to govern fuel conversion. The model was validated against data
from lab-scale self-ignition and smoldering fire propagation experiments. It was then
applied to large-scale fire scenarios.
2 Experiments
To provide input and validation data for computation, four classes of experiments were
performed:
1) Determination of coal properties relevant to combustion,

2) Self-ignition experiments with variation of geometry and boundary conditions,
3) Analysis of flue gases during self-ignition,
4) Validation experiments of self-ignition and smoldering fire propagation.
The quantities measured in these experiments are summarized in Table 1.
488
Table 1: Experimental quantities relevant to self-ignition behavior of coal

Coal Reaction Reaction Heat and Mass
Properties Stoichiometry Kinetics Transfer
Particle density Elemental Apparent Diffusion
composition activation energy coefficient
Bulk density
Ash content Apparent Heat transfer
Porosity
frequency factor coefficient
Humidity
Permeability
Mass transfer
Composition of flue gases
Thermal coefficient
conductivity
Specific heat
Calorific value
It is out of the scope of this paper to explain all these experimental methods in detail. This
has been done in the various reports from BAM pertaining to the Sino-German Coal Fire
of Coal Fires in North China,” e.g. Krause et al. (2003) or Schmidt et al. (2005). Only one
example is given here; it allows for simultaneous measurement of several experimental
quantities.
Self-ignition behavior of bulk materials, e.g. of coal, is commonly investigated in so-

called hot storage tests. In these tests, a sample of the material is stored in an oven at a pre-
defined constant temperature. The temperature within the sample is controlled both by the
heat of reaction and by the heat transfer between the sample and the air in the oven. Self-
ignition occurs when the heat of reaction released within the sample exceeds the heat flux
from the sample. For our experiments, we used a 120 l laboratory oven heatable up to
300 °C. Sample sizes ranged from 100 ml to 6 l. Additionally, the oven was equipped with
a thermo-gravimetric scale to observe mass loss of the samples during combustion (When
this feature was used though, sample mass was restricted to 300 g.). Also, representative
quantities of gaseous reaction products were collected and transferred to an FTIR (Fourier
Transform Infrared) spectrometer for analysis of fractions of carbon monoxide (CO),
carbon dioxide (CO2), sulfur dioxide (SO2), methane (CH4), and water (H2O). Together
with the results from elemental analysis of the coal samples (for fractions of C, H, N, S,
and O) conducted prior to heating, as well as the mass loss with time measured during
combustion, the analysis of the flue gases provided the input for a stoichiometric equation
of the combustion reaction.
489
Figure 1 shows the laboratory oven used in the experiments and its peripheral devices.
Figure 1: Experimental set-up for hot storage experiments of combustible solid materials
including thermo-gravimetric scale and FTIR spectrometer for flue gas analysis
Figure 2 gives an example of the data obtained from the experiments. A 400 ml sample of
crushed coal from Wuda Coalfield (Inner Mongolia Autonomous Region, P.R. China) was
stored at a constant temperature of 207 °C over a period of about 1,600 min. Temperature
evolution with time was measured in the center and at half the distance between the center
and the surface of the sample.
490
Figure 2: Temperature evolution with time in the center (T1) and at half the distance
between center and surface (T3) of a 400 ml sample of Wuda coal during hot
storage at 207 °C (T2), as well as evolution of volume fractions of CH4, CO,
CO2, and H2O with time in flue gas from the sample
From the temperature-time curves, together with the volume fraction profiles, it is possible
to broadly extrapolate different stages of coal combustion. The coal sample was heated
from room temperature to storage temperature during the first 290 min of the experiment.
After this, the temperature rose continuously to a level of 670 °C, which remained the
maximum temperature reached. A clear distinction between different stages of combustion
in terms of threshold temperatures cannot be made. Production of CO2 and CO started at
a reaction temperature of about 300 °C. This indicates that the coal was pyrolized at this
stage, i.e. transformed in incomplete combustion to solid and gaseous products, which are
themselves combustible.
The third major species measured was H2O. It had two peaks in its concentration-time
curve. The first resulted from physisorption into the coal; the second was due to water
production during combustion. Additionally, the presence of water vapor enhanced the
formation of CH4 at temperatures above 500 °C.
The last stage of combustion, after about 800 min, was the combustion of solid residues from
the preceding stages, e.g. char and coke. During this stage, temperatures remained at 670 °C,
but the reaction rate decreased continuously. Combustion products were CO, CO2, and ash.
491
Experimental data as those in Figure 2 not only help to differentiate combustion stages, but
also serve to validate mathematical models. For modeling, the quantities listed in Table 1
have to be merged; the workflow for this procedure is shown in Figure 3. Although the
molecular structure of coal can not be established experimentally, a reaction model can be
inferred from the fractions of the relevant elemental components of coal and a fictitious
“coal molecule” created therefrom. In our model, the same procedure was applied to solid
reaction products.
Figure 3: Components of mathematical model for heat and mass transfer during self-
ignition and fire propagation in porous solids, e.g. coal
3 A Multi-Dimensional Time-Dependent Coal Fire Model
Promoted by the rapid development of computer resources, the numerical solution of

coupled heat and mass transfer problems has become a matter of considerable interest
in fire safety engineering over recent years. Numerical simulations offer the following
advantages:
492
• There are no limitations to geometry. Three-dimensional geometries can be computed

as well as “sandwich” materials, i.e. materials consisting of layers with different
properties.
• Various kinds of boundary conditions can be considered (Dirichlet, von Neumann,
Newtonian cooling); in addition, these boundary conditions may vary in space and
time.
• The dependence of temperature on time can be computed such that the entire process of
heating, self-ignition, combustion, and extinction can be traced continuously.
• Information on local distribution of chemical species and their evolution with time is
easy to obtain.
• Kinetic and diffusion-controlled reaction regimes can be included.
The input for a model based on these simulations covers:
• Experimental quantities as given in Table 1,

• Boundary and initial conditions, and
• Geometrical data.
The main simplifications underlying the present model are:
• Heat and mass are transferred throughout the bulk material only by conduction and
diffusion (convection may occur at the outer surface of the deposit; if necessary, it can
be approximated by a variation of the diffusion coefficient).
• The bulk material is a homogeneous and isotropic body (locally uniform thermal
transport coefficients).
• Porosity is implicitly contained in bulk density.
These simplifications were necessary to keep the computational efforts in the current
stage of model development as low as possible. They are not restrictions of the model in
principle and can be omitted in later stages.
Based on the above simplifications, the balance equations were:
∂T
(1) = a ⋅div grad T + S T
∂t
for heat transfer, and:
493
∂C i
(2) = Di ⋅ div grad C i + S Ccii
∂t
for the transport of a chemical species i.
T in equation (1) is the temperature, a the thermal diffusivity, t the time, and ST the heat
source term. Ci in equation (2) is the mass concentration of a species i, Di the diffusion
coefficient of the binary system consisting of the species i and the mixture, and SCi the rate
of its production or consumption.
The most significant of the above simplifications is the neglect of convective heat and mass
transfer. Its impact was partly offset by altering the thermal diffusivity in equation (1) and
the diffusion coefficient in equation (2):
(3) a = a diff + a conv ,
(4) D = Ddiff + Dconv .
The convective portion in equations (3) and (4) was defined as:
(5) Dconv = v ⋅ l conv ,
where v is the velocity of the convective flow and lconv the characteristic of the convective
transport, e.g. the pore distance. One problem with this approach is that the velocity of the
convective flow can only be approximated. Also, the convective flow has automatically the
same direction as the diffusional flow, while in a realistic convection model these flows should
follow the direction of the pressure gradient. This, however, is acceptable given that convective
and diffusional flows do in fact have the same direction in the case of self-sustained smoldering
fire propagation. The advantage of this model hence is computing efficiency; it relieves its user
of the necessity to introduce three additional momentum equations.
For the reaction model, it was assumed that the fuel is converted in the following four
reaction steps:
1) Decomposition to char (C), CO, CO2, and hydrogen (H2) (equation (6));
2) Char combustion (equation (7));
3) Oxidation of CO (equation (8)); and
4) Oxidation of H2 (equation (9)).
494
A fictitious “fuel molecule” was assumed based on elemental analysis of the mass
fractions of C, H, N, and O in the virgin fuel. Table 2 shows as an example the elemental
composition of Chinese black coal from Wuda (water- and ash-free), which provided the
basic values for equation (6).
Table 2: Mass fractions of C, H, N, O, and S in Chinese black coal from Wuda (waf)
Element C H N O S
Mass fraction (%) 59.3 4.0 1.1 32.6 3.0
For the above coal, reaction steps 1) to 4) yield:
(6) C50H20O10 → 44C + 4CO2 + 2CO + 10H2
(7) C + O2 → CO2
(8) 2CO + O2 → 2CO2
(9) 2H2 + O2 → 2H2O
The oxidation of nitrogen (N) and sulfur (S) was neglected in this example.
Proceeding from the above equations, seven different chemical species are involved
and considered for coal combustion in the reaction model. For each of these species,
an equation of the type of equation (2) was incorporated. To account for the effect of
inertization, another such equation was included for nitrogen in the gas phase. The source
term for nitrogen is zero.
Each species is converted at its specific rate during each of the reaction steps. E.g. the solid
fuel when decomposed according to equation (6) exhibits an Arrhenius-type reaction rate
as given in equation (10):
dCf  E 
(10) = − C f ⋅ k 0 ⋅ exp −  ,
dt  RT
RT 
where k0 is the pre-exponential factor, E the apparent activation energy specific to the
reaction given in equation (6), and R the universal gas constant. The reaction rate is of first
order as it is assumed that, besides temperature, decomposition depends exclusively on
fuel concentration.
For the rate at which the species are produced in this reaction, the following equation applies:
495
d C i υυii M i d C f
(11) = ⋅ ⋅ ,
dt υυff M f d t
where the index i refers to the species C, CO, CO2, and H2, υi to the stoichiometric
coefficients, and Mi to the molecular weight. Stoichiometric coefficients are always
negative for educts and positive for products.
The rate equations for all species are analogous. However, the dependence of the reaction
rate of a species (in our model C, CO, and H2) on the concentration of oxygen was included
to reflect the impact of oxygen diffusion on the reaction rate. Table 4 in the Appendix gives
an overview of the reaction rates of the different species in the different reactions.
Table 5 in the Appendix lists the apparent activation energies and pre-exponential factors
for the reactions according to equations (6) to (9) as used in our computations. Other
quantities for input are listed too.
The source term in the temperature field equation was computed as:
1 4
d Ci
(12) ST = ⋅ ∑ ∆H R ,i ⋅ .
ρ.c⋅ c i =1 dt
Initial and boundary conditions are generally selected on a case-to-case basis. Often, the
bulk material undergoing self-ignition has a free surface over which heat is exchanged
with the surrounding air. At such a surface, the heat flux may be calculated as follows:
(13) α ⋅ (TS − Ta ) ,
q = α
where α is the heat transfer coefficient, TS the surface temperature, and Ta the ambient temperature.
The mass flow of a species i into the bulk material (or from it) can be calculated using:
(14) ji = βi (Ci -Ci ) ,

s a
where βi is the mass transfer coefficient, Cis the concentration of the species i at the surface,
and Cia the concentration of the species i in ambient air.
If symmetry axes exist within the bulk material, they can be used as boundaries too. This
reduces the computational effort. At a symmetry axis, the heat and mass flows are zero.
As initial conditions, the local distribution of temperature and concentration of each

species i have to be known.
Thus, the system of equations is mathematically closed. For solving the system of equations
numerically, we used the commercial finite element code COMSOL Multiphysics.
496
4 Computational Examples
4.1 Smoldering fire propagation in silo-stored coal
For an example of real-scale computation, self-ignition in a silo filled with crushed lignite
coal was simulated. The cylindrical part of the silo was assumed to have a height of 10 m
and a diameter of 3 m. The funnel had a height of 3 m, ending in a discharge opening of
0.4 m in diameter. The total volume of the silo was 78 m³.
As the axes of the silo were symmetrical, computation could be reduced to two dimensions.
It was assumed that the silo was filled completely with coal. An adaptive finite element
mesh with comparatively fine meshing in the funnel and along the symmetry axis, and
comparatively coarse meshing near the outer surface was used for computations. The mesh
consisted of about 12,000 elements. In this range, computations were insensitive to grid size.
In the example, the initial temperature of the coal was set at 25 °C, the ambient
temperature at 40 °C. The latter value was selected to reflect the conditions at a location
with a relatively hot climate (Day and night variations in ambient temperature were not
considered.). Free convection at the outer surfaces of the silo, for the sake of simplicity,
was expressed by a uniform heat transfer coefficient of 8 W m-2 K-1.
Figure 4 depicts the temperature evolution with time in the silo.
Figure 4: Temperature evolution with time during smoldering fire in a silo filled with
crushed lignite coal (silo volume 78 m³; ambient temperature 40 °C; walls
permeable for O2)
497
The induction time for self-ignition was about 240 days in this example. As the transport
of oxygen through the silo walls was assumed to be unlimited, two hot spots developed.
The reason for this may be explained as follows:
Suppose the silo consisted of two independent parts one stacked on top of the other.
Neglecting the funnel, the volume of each part would be about 35 m³, and the volume-
to-surface ratio V/A of each 0.57 m. Extrapolating from a number of experimentally
determined self-ignition temperatures of lignite coal, one arrives at a self-ignition
temperature of about 15 °C for this V/A. Hence, at an ambient temperature of 40 °C each
of the two sub-volumes was stored under supercritical conditions so that hot spots must
occur in both of them.
After ignition, a smoldering fire developed and propagated for about 11 days until it
reached the wall of the silo. The computed temperature of about 550 K at the reaction
front was within the range typical of smoldering fire propagation in lignite coal (see
Schmidt 2001). The average propagation velocity was computed at 6.1 mm h-1, which
is in satisfying agreement with the experimental values given by Schmidt (2001).
Figure 5 exhibits the evolution of the oxygen volume fraction with time.
Figure 5: Evolution of oxygen volume fraction with time during smoldering fire in a silo
filled with crushed lignite coal (silo volume 78 m³; ambient temperature 40 °C;
walls permeable for O2)
498
In a next step, self-ignition and smoldering fire propagation were modeled for non-
permeable silo walls. In this scenario, only the oxygen initially contained in the voids
between the particles of the bulk material with a volume fraction of 21 % in the gas phase,
and the oxygen entering the silo through the silo top and the discharge opening at the
funnel bottom were available for combustion. All other conditions were the same as in the
previous example.
The evolution of temperature with time under such limited oxygen supply is shown in
Figure 6. Figure 7 shows the corresponding oxygen volume fraction. In contrast to fire
propagation without limitations in oxygen supply (Figure 4), only one hot spot developed
(Figure 6). This hot spot occurred in the upper part of the silo. In the lower part, oxygen
was insufficient for a significant rise in temperature. In addition, self-ignition occurred
72 days later than under unlimited oxygen supply.
Figure 6: Temperature evolution with time during smoldering fire in a silo filled with
crushed lignite coal (silo volume 78 m³; ambient temperature 40 °C; only silo
top and discharge opening permeable for O2)
499
Figure 7: Evolution of oxygen volume fraction with time during smoldering fire in a silo
filled with crushed lignite coal (silo volume 78 m³; ambient temperature 40 °C;
only silo top and discharge opening permeable for O2)
After ignition, the vertical propagation of the temperature front (in the direction of the
open end) was about twice as fast as the horizontal, while before, under unlimited oxygen
supply, the velocity of temperature propagation had been about the same in both directions
(Buoyancy not taken into account in either case.).
The propagation velocity of the smoldering front was c. 4.7 mm h-1 in upward direction,
which is 77 % of the value computed for unlimited oxygen supply.
Another difference between the two examples shows upon comparison of Figures 5 and 7.
While in Figure 5, oxygen is only depleted behind the reaction front, in Figure 7, depletion
extends over almost the entire silo – this although the reaction zone is limited to the silo’s
upper central part. The model is hence able to reflect the diffusion of oxygen from virgin
parts of the material to the reaction zone.
Although the model’s agreement with results from lab-scale experiments is currently not
better than fair, the computations show plausible patterns of self-ignition and smoldering
fire propagation in real-scale scenarios, which are beyond the means of experimental
investigation.
500
4.2 Simulation of self-ignition in a coal seam
As another example, a 2D computation of self-ignition and fire propagation in a coal seam

was performed.
The geological setting was adopted from investigated parts of Wuda Coalfield. Figure 8
shows the stratigraphy of fire zone (FZ) 8 according to Gielisch and Kus (2005). Coal
seams nos. 9 and 10 of Wuda Coalfield run through the area.
Figure 8: Stratigraphy of FZ 8 of Wuda Coalfield with coal seams nos. 9 and 10.
Numbers at the right indicate altitude (m).
Data source: Gielisch and Kus (2005)
The right-hand edge of the computational domain was modeled as an adiabatic wall.
Consequently, self-ignition started at this location in the computations. The lower edge was
assumed to have a constant temperature of 293 K, while at all other edges, free convection
was the boundary condition. For the sake of simplicity, a uniform and constant heat
transfer coefficient of 8 W m-2 K-1 was adopted for the entire domain.
Figure 9 shows the distribution of temperature after 1,053, 1,351, 1,456, and 1,618 days.
Figure 10 depicts the depletion of virgin coal at the same intervals. While the occurrence
of self-ignition at the right-hand edge followed directly from the set-up, it is interesting
to note that during fire propagation through the seam, a second hot spot developed
independently from the first one in both coal seams. Heat transmission from seam no. 9,
where the fire started, triggered off ignition in seam no. 10.
501
502
Figure 9: Temperature distribution in self-ignition simulation for FZ 8 of Wuda Coalfield

(coal seams nos. 9 and 10) after 1,053 (left top), 1,351 (left bottom), 1,456
(right top), and 1,618 (right bottom) days.
503
504
Figure 10: Mass concentration in self-ignition simulation for FZ 8 of Wuda Coalfield (coal
seams nos. 9 and 10) after 1,053 (left top), 1,351 (left bottom), 1,456 (right top),
and 1,618 (right bottom) days.
505
Figure 11 shows the distribution of CO produced as per equation (6) after 1,053 and 1,456
days. Figure 12 exhibits the same for O2. During coal conversion, evidently, the vicinity of
the reaction zone is emptied of O2 and filled with combustion products (CO).
Figure 11: CO volume fraction in self-ignition simulation for FZ 8 of Wuda Coalfield
(coal seams nos. 9 and 10) after 1,053 (top) and 1,456 (bottom) days
506
Figure 12: O2 volume fraction in self-ignition simulation for FZ 8 of Wuda Coalfield (coal
seams nos. 9 and 10) after 1,053 (top) and 1,456 (bottom) days
507
5 Conclusion
The model presented in this paper covers the entire process of spontaneous combustion in
porous solids. This process is characterized by the following stages:
• Temperature equilibration between combustible material and surroundings (if initially

different),
• Heat accumulation and subsequent temperature rise due to exothermic reactions within
porous solid, and
• Initiation of self-ignition and subsequent propagation of smoldering fire front.
While chemical kinetics determine reaction rates in the first stage of self-ignition, oxygen
diffusion becomes the controlling mechanism in later stages. Both are accounted for in the
model, including a smooth transitional phase.
Examples of how to apply the model to simulations of fire scenarios in stored coal and
coal seams were given in this paper. One of our main findings is that hot spots are rather
unlikely to occur in compact coal. The more presumable scenario is that of hot spots in
parts of coal seams or mines with a sufficient amount of loose coal.
In the Wuda Coalfield scenario, self-ignition was found to start about three years from the
assumed beginning of oxygen-induced heating. This is in good agreement with empirical
observations in situ, which locate the onset of self-ignition in Wuda Coalfield approximately
three years after the beginning of mining activities and exposure to air (see Guan 2005).
The average propagation velocity of the reaction front, however, was computed faster than
in reality (100 m in two years compared to 100 m in seven years). The main reason for this
difference is that the computations assumed a substantial amount of pre-conditioned loose
coal. In reality though, coal seams are very compact structures, where fissures and cracks
have to form under thermal and mechanical stress prior to chemical conversion to allow
for oxygen uptake. These factors delay propagation.
In general, computation results largely depend on the accuracy and quality of input data,
which normally have to be generated in experiments – hence, the overriding importance of
reliable experimental procedures for model computations.
Undoubtedly, advanced computer modeling of self-ignition and smoldering fire

propagation will become an increasingly common tool for assessing fire risk in porous
solids. Current models differ mainly in the number of chemical components – a fact which
508
is directly linked to the number of transport equations considered. The model presented
in this paper includes seven chemical species, the extension of which is only limited by
computer resources. A further refinement of the model, including transport of liquid and
vaporized water, is under development.
Acknowledgements
The contribution of BAM to the Sino-German Coal Fire Research Initiative “Innovative
Technologies for Exploration, Extinction, and Monitoring of Coal Fires in North China”
was funded by the German Ministry for Education and Research (registration no.
0330049C), which is hereby gratefully acknowledged.
Appendix
Table 3: Nomenclature
Symbol Quantity Unit
a Thermal diffusivity m² s-1
c Specific heat capacity J kg-1 K-1
C Mass concentration kg m-³
D Diffusion coefficient m² s-1
E Apparent activation energy J mol-1
HR Heat of reaction J kg-1
j Mass flow kg m-2 s-1
k0 Pre-exponential factor s-1
M Molecular weight g mol-1
q Heat flow J s-1 m-2
R Universal gas constant J kg-1 K-1
SCi Production/consumption rate of species i kg m-3 s-1
ST Heat source term J m-3 s-1
T Time S
T Temperature K
α Heat transfer coefficient J m-2 s-1 K-1
β Mass transfer coefficient m s-1
ρ Bulk density kg m-³
υ Stoichiometric coefficient
Subscripts:
a
ambient
f
fuel
i
species i
s
surface
509
TableTable
4: Reaction ratesrates
4: Reaction

Species Reaction 1 Reaction 2 Reaction 3 Reaction 4
C 241 H 228 O46 o C O2 o CO2 2CO O2 o 2CO2 2 H 2 O2 o 2 H 2 O
213C 18CO2 10CO 114 H 2
Fuel § d Cf · § E · ... ... ...
¨¨ ¸¸ C f k 0,1 exp¨ 1 ¸
© dt ¹1 © RT ¹
C X C ,1 M C § d C f · § d CC · § E · ... ...
¨ ¸¸ ¨¨ ¸¸ CC k 0, 2 exp¨ 2 ¸
X f ,1 M f ¨© d t ¹1 © dt ¹2 © RT ¹
CO X CO ,1 M CO § d Cf · ... § d CCO · § E · ...

¨¨ ¸¸ ¨¨ ¸¸ CCO k 0,3 exp¨ 3 ¸
X f ,1 M f © dt ¹1 © d t ¹3 © RT ¹
CO2 X CO ,1 M CO § d Cf · X CO , 2 M CO § d CC · X CO ,3 M CO § d CCO · ...

510

2 2
¨¨ ¸¸ 2
2
¨¨ ¸¸ 2
2
¨¨ ¸¸
X f ,1 M f © dt ¹1 XC ,2 M C © dt ¹2 X CO ,3 M CO © dt ¹3
H2 X H ,4 M H § d C f · ... ... § d CH2 · § E ·

2
¨ 2
¸¸ ¨¨ ¸¸ C H 2 k 0, 4 exp¨ 4 ¸
X f ,1 M f ¨© d t ¹1 © dt ¹4 © RT ¹
O2 ... XO ,2 M O § d CC · X O ,3 M O § d CC · XO ,4 M O § d CH ·
2 2
¨¨ ¸¸ 2
2
¨¨ ¸¸ 2
¨ 2 2
¸¸
XC ,2 M C © dt ¹2 X C ,3 M C © dt ¹3 X H , 4 M H ¨© d t
2 2 ¹4
H2O ... ... ... X H O,4 M H O § d CH2 ·
2
2
¨¨ ¸¸
X H ,4 M H
2 2 © dt ¹4
N2 ... ... ... ...
101
Table 5: Input data for experimental quantities of Chinese black coal from Wuda
ρ a Di ∆HR,f ∆HR,C ∆HR,CO ∆HR,H 2
-7 -5 7 7 7
476 1.83 ×10 2 × 10 2.2035 × 10 2.7 × 10 1.004 × 10 1.588 × 107
kg m-³ m² s-1 m² s-1 J kg-1 J kg-1 J kg-1 J kg-1
k0,1 k0,2 k0,3 k0,4 E1/R E2/R E3/R E4/R

2.44 × 10 6
2.44 × 10 5
3.981 × 10 * 14
3.311 × 10 * 13
11 658 8 157 20 600* 23 200*
* Data source: Görner (1991); all other data determined experimentally by the authors
References
Akgun, F. and Essenhigh, R.H, 2001. Self-ignition characteristics of coal stockpiles: Theoretical
prediction from a two-dimensional unsteady-state model. Fuel 80: 409-415.
Babrauskas, V., 2003. Ignition Handbook. Chapter 9: Self-heating. Issaquah, Washington: Fire Science
Publishers.
Bowes, P.C., 1984. Self-heating and controlling the hazards. Her Majesty’s Stationary Office, London,
United Kingdom.
Chen, X.D., 1998. On the fundamentals of diffusive self-heating in water-containing combustible

materials. Chemical Engineering and Processing 37: 367-378.
Chong, L.V. and Chen, X.D., 1999. A mathematical model of the self-heating of spray-dried food powders
containing fat, protein, sugar, and moisture. Chemical Engineering Science 54: 4165-4178.
Fierro, V., Miranda, J.L., Romero, C., Andrés, J.M., Arriaga, A., Schmal, D., and Visser, G.H., 1999.
Prevention of spontaneous combustion in coal stockpiles – Experimental results in coal
storage yard. Fuel Processing Technology 59: 23-34.
Gielisch, H. and Kus, J., 2005. An anomalous coalification distribution in the Northern Helan Shan – A possible
key into the geological history of the Helan Shan area. Poster presented at theInternational
Conference on Coal Fire Research – ERSEC Conference on Understanding, Extinction, and
Prevention of Spontaneous Coal Seam Fires, Beijing, P.R. China, 29 November – 1 December 2005.
Guan, H.Y., Fan, X.J., and Wu, C.C., 2005. Study on thermodynamic system and detecting model of coal
fire area. Paper presented at the International Conference on Coal Fire Research – ERSEC
Conference on Understanding, Extinction, and Prevention of Spontaneous Coal Seam Fires,
Beijing, P.R. China, 29 November – 1 December 2005.
Hull, A.S., Lanthier, J.L., and Agarwal, P.K., 1997. The role of the diffusion of oxygen in the ignition of a
coal stockpile in confined storage. Fuel 76(10): 975-983.
Kessels, W., Wessling, S., and Li., X., 2005. Coal Fire Development from self-ignition to fire fighting – Numerical
modeling and measurements for basic investigation. Paper presented at the International
Conference on Coal Fire Research – ERSEC Conference on Understanding, Extinction, and
Prevention of Spontaneous Coal Seam Fires, Beijing, P.R. China, 29 November – 1 December
2005.
511
Krause, U., Schmidt, M., and Lohrer, C., 2003. Report on the activities for the Chinese-German Mutual
Project on Innovative Technologies for Exploration, Extinction, and Monitoring of Coal Fires
in North China. Internal report, as a contribution to the Sino-German Coal Fire Research
Fires in North China.” Federal Institute for Materials Research and Testing (BAM), Berlin,
Germany.
Krishnaswamy, S., Agarwal, P.K., and Gunn, R.D., 1996. Low-temperature oxidation of coal. Part III:
Modeling spontaneous combustion in coal stockpiles. Fuel 75: 353-362.
Leuschke, G., 1983. Experimental investigations on self-ignition of dust deposits in hot environments.
IchemE Symp. Series 68: 2/G: 1-15.
Liang, H. and Tanaka, T., 1987. The spontaneous ignition of dust deposits – Ignition temperature and
induction time. KONA Powder Sc. Technol. in Japan 5: 25-32.
Ohlemiller, T.J., 1985. Modeling of smoldering combustion propagation. Prog. Energy Comb. Sci. 11:
277-310.
Schmidt, M., 2001. Untersuchungen zu Schwelbränden in Feststoffschüttungen verursacht durch

Glimmnester und eingeschüttete heiße Gegenstände. PhD thesis, University of Magdeburg,
Magdeburg, Germany.
Schmidt, M. and Krause, U., 2000. The influence of initial conditions on the propagation of smoldering
fires in dust accumulations. In Proceedings of the 3rd International Symposium on Hazards
Prevention Mitigation of Industrial Explosions, Tsukuba, Japan, 21-27 October 2000,
Tsukuba, Japan, 230-239.
Winters, K.H., and Cliffe, K.A., 1985. The prediction of critical points for thermal explosion in a finite
volume. Comb. Flame 62: 13-20.
512
Simulating Dynamic Subsurface Coal Fires
and Its Applications
Stefan Wessling1, Tom Litschke2, Jens Wiegand2,

Stefan Schlömer3, and Winfried Kessels1
1
Leibniz Institute for Applied Geosciences (GGA), Hanover, Germany
2
3
Abstract
Understanding underground coal fire propagation is essential for developing extinction strategies and
preventing coal mining hazards. The development of coal fires is controlled by complex hydraulic, chemical,
thermal, and mechanical processes. These processes take place in the coal seam and adjacent geological
formations. The temporal development, i.e. the dynamics of underground coal fires with respect to physical
and chemical processes have received little attention so far in research and literature on the topic.
Some aspects of the dynamic behavior of fire zone (FZ) 3-2 in Wuda Coal Mining Area (Inner Mongolia
Autonomous Region, P.R. China) were investigated in a combination of in-situ measurements and
numerical simulations. Surveys on site comprised temperature monitoring and logging, as well as
measurements of oxygen concentration, ventilation, and hydraulics. Simulations were performed based on
a homogenization approach, which assumes gas and energy transport, as well as reactive species transport
to be intrinsic components of coal fire development. In reactive species transport coal combustion is
addressed from the perspective of oxygen transport through adjacent rock formations. Gas flow is a result
of natural thermal convection. Energy is released in the form of heat during combustion. A simplified set-
up was chosen according to the present conditions in FZ 3-2.
The simulation results illustrate the effects of transport phenomena on physical and chemical variables
of coal combustion. A comparison between simulation results and in-situ observations highlights those
processes most likely to occur in FZ 3-2. Three-dimensional calculations are proposed to calibrate
simulations by means of ventilation measurements. An extinction scenario predicts underground coal fire
development after surface sealing with nearly impermeable material. It shows that the temperature of a
downward-moving fire decreases by 40 % after 279 days. Outcrops have an impact on the extinction of
upward-moving fires, because insufficient coverage leads to continued supply with oxygen through the
outcrop.
513
Although field observations show that FZ 3-2 is highly heterogeneous, the homogenization approach
chosen for simulation is still valid as it helps to find explanations for at least some of the in-situ
observations. In particular, the successful simulation of an extinction scenario demonstrates the usefulness
of numerical simulations for coal fire fighting.
摘要
了解地下煤火的传播过程是研究灭火策略和预防煤矿开采灾害的基础。煤火的发展受到了复杂的
水文条件、化学过程、热过程和动力过程的控制。这些过程发生在煤层中和邻近的岩层中。目
前，有关煤火的时序发展，即地下煤火相对于物理过程和化学过程的动态变化，尚未在研究中受
到重视，而且相关文献也极少。
本研究采用现场测量与数值模拟的方法研究了中国内蒙古乌达矿区3-2号火区的动态变化的一些问
题。现场测量包括了温度监测和地球物理测井，以及氧气浓度、通风和水文参数的测量。数值模
拟是基于均质化的方法进行的，即假设气体和热能传输，以及反应过程都是煤火过程的内在组成
部分。反应过程可观察到煤在氧气通过相邻的岩层传输下的燃烧。气体流动导致了热对流，能量
在燃烧期间以热的形式释放。研究中依据3-2号火区带现状选择了简化的参数。
模拟结果表明一些物理和化学参数对于传输现象的影响。模拟与现场结果对比显示煤火过程大部
分类似在3-2号火区发生的情况。建议采用三维计算方法利用通风测量结果校正模拟参数。最后，
灭火解决方案提出了地表采用不可渗透的材料封闭后地下煤火的时序发展过程，表明朝下运动的
火源温度在279天后降低了40％。露头对于扑灭朝上运动的火源似乎有影响，因为覆盖不足导致通
过露头不断的提供氧气。
野外观测证明尽管3-2号火区具有较高的各向异性特点，但模拟所选的均质化方法足以帮助我们至少
解释了现场观测的现象。成功的煤火灭火模拟尤其显示了数值模拟方法对于灭火工程的有效性。
1 Introduction
Subsurface coal fires are one of the most challenging problems in modern-day coal
mining. Apart from their generally large extent, it is their location underground that mars
attempts at precise analysis and complicates fire extinction and prevention. Technology
today allows to investigate underground coal fires by means of geophysics, e.g. by
magnetic measurements or borehole logging. Numerical calculations are useful to
understand the temporal development of subsurface fires with respect to physical and
chemical processes. Additionally, in-situ investigations at the surface (temperature and
ventilation measurements) and/or the interpretation of satellite data can assist in creating
an approximate picture of the situation underground. Every single method on its own,
however, can give only limited information; using but one may result in inadequate or
inefficient extinction strategies.
Many of the above-mentioned techniques are being used in the Sino-German Coal Fire
514
of Coal Fires in North China.” They are complemented by laboratory experiments where
needed. This paper focuses on a combination of numerical calculations and surface
measurements to analyze subsurface coal fires and their dynamic behavior.
Numerical calculations have been applied extensively in investigations of underground

coal gasification (Yang 2004; Perkins & Sahajwalla 2005) and the self-ignition behavior
of various coals (Schmal 1987; Brooks et al. 1988a,b; Rosema et al. 2001; Schmidt et al.
2003). Self-ignition, however, is limited to low-temperature regimes; the simulation of
burning coal seams (= high-temperature coal fires) has not been addressed so far.
Numerical simulations, though, have already been used for investigation of coal fire
phenomena. Klika et al. (2004) modeled the partial burning of three bituminous coal seams
and the resulting temperature distribution in adjacent beds. The model assumed heat transport
from a heat-generating source within the seams and explained the formation of a body of
variegated beds over the course of 2000 years of continued burning. Gas flow coupled with
heat transport was simulated by Huang et al. (2001). In their model, free thermal convection
was assumed to be the driving mechanism for gas flow. Unfortunately, the model was two-
dimensional and schematic only, so their results were fairly general. The study concluded in
an appeal to foster investigations of coal fires as a complex system by way of modeling.
Simulations of complete burn-out so far have been run under the steady-state assumption
(Brenner et al. 2000; Huang et al. 2001), or over time scales relatively shorter than actual
coal fire combustion times (Westbrook & Dryer 1981; Hjertager 1986; Krause & Schmidt
2001; Perkins & Sahajwalla 2005). Simulations aiming at the investigation of underground
coal fires are still outstanding.
This work presents the results of numerical and experimental research and is meant to be a
contribution to the understanding of the dynamic behavior of subsurface coal fires. Special
emphasis is given to transport phenomena through over- and underlying beds of burning coal
seams. These beds are regarded as a link between the surface and the combustion center.
They are the main pathway for oxygen. Likewise, the discharge of gaseous combustion
products takes place through these beds. A numerical model calculates the combustion
process in the seam with respect to gas, energy, and reactive species transport in adjacent
beds. Results from laboratory experiments on combustion processes are also used in
the model. The calculation results are compared with in-situ measurements, including
temperature monitoring and logging, as well as measurements of oxygen concentration,
ventilation, and hydraulics. An extinction scenario is presented as a simulation example
for practical purposes. It assumes surface sealing with nearly impermeable material, which
results in reduced oxygen transport through the surface.
515
2 Study Site
In-situ investigations and measurements were performed in fire zone (FZ) 3-2 of Wuda
Coal Mining Area (Inner Mongolia Autonomous Region, P.R. China) – one of three study
areas of the Sino-German Coal Fire Research Initiative (Figure 1). The geological setting
of this zone has been described by Gielisch and Kahlen (2003). Typical of subsurface coal
fires are strongly disrupted overlying beds, which enable the discharge of combustion
gases through and away from the surface – a result of rock mechanical failure (Buhrow
et al. 2004). Large-scale fractures (Figure 2) typically occur next to small-scale openings
like funnels or sponge-like structures. A more detailed description of cracks and fissures
detected in FZ 3-2 is given in Gielisch and Künzer (2003).
Figure 1: Location of Wuda Coal Mining Area (Inner Mongolia Autonomous Region,
P.R. China)
516
Figure 2: Large-scale fractures commonly observed in Wuda Coal Mining Area
An additional feature of the study area are coal seam outcrops, e.g. at coal seam no. 10.
The seam crops out below a c. 10 m high sandstone scarp (Figure 3). Scarps of this kind
may function as gas inlets due to their relatively high fracture density. Coverage for fire
quenching is very costly.
Figure 3: Outcrop of coal seam no. 10 under sandstone scarp
517
3 In-Situ Measurements
3.1 Ventilation measurements
Ventilation was determined through emission chamber measurements and dynamic

pressure measurements. Emission chamber measurements were performed with a “flux
chamber,” a circular chamber made of steel or temperature-resistant Plexiglas, with
inlet and outlet connectors. It was placed leak-tight onto the soil. The increase of gas
concentration (or decrease in case of gas sink) was plotted versus time. This served as the
basis for total gas flow calculations.
Dynamic pressure measurements were performed with a pitot-static tube. The pitot-static
tube is a joint construction of two concentric tubes. At the front end, the outer tube is
sealed (stagnation port). The pressure on this port consists of the dynamic pressure exerted
by the inflowing gas and the static ambient pressure. The pressure on the static port (i.e.
the port of the outer tube at the back end) is only the ambient atmospheric pressure. The
total gas discharge per unit time can be calculated from the pressure difference between
stagnation and static port.
Ventilation measurements were performed for different surface structures with (visible)
gas discharge. These structures were classified, and the average volume of gas discharge
calculated. Classification was done according to shape, i.e. cracks, vents, funnels, semi-
funnels, sponges, and undisturbed soil (Table 1). The measurements covered an area
of c. 14,000 m². A total volume flow of approximately 37,149 m³ h -1 was measured.
The measurements showed that the specific gas flow in m h-1 varies over four orders of
magnitude between cracks and undisturbed soil. Thus, large-scale heterogeneities of coal
beds effect huge variations in gas discharge at the surface.
Table 1: Classification and results of ventilation measurements

Structure Class Average Area Average Flow Rate Specific Flow Rate
(m2) (m3 h-1) (m3 h-1 m-2 = m h-1)
Crack 5.3 22,456 4,237
Vent … a
2,302
Funnel … 653
Semi-funnel … 75
Sponge 72.0 1,642 22.8
Undisturbed soil 13,868 10,022 0.72
a
Not determined because too small and therefore irrelevant.
518
It should be added here that no surface locations (or outcrops) were observed at which
gas entered the soil, nor could any such gas intake be proven by measurements. However,
supply with oxygen through inflowing gas is essential for combustion. This failure in
observation has to be investigated.
3.2 Measurements of oxygen concentration
The entry and transport of air through overlying rock strata to the combustion center
of a coal fire is of eminent scientific interest for the investigation of coal fire dynamics;
it is a precondition for continued burning. Air circulation at FZ 3-2 was determined
through surface measurements of oxygen concentration. Combustion gas measurements
were carried out with portable gas detectors (Multiwarn II by Dräger Safety AG & Co.
KGaA, Germany), equipped with several chemical sensors and an infrared sensor. Gas
composition was investigated in depth ranges of 0.2 to 1 m (using steel conduits) to avoid
mixing with surface air. Gases were diluted (up to 50:1) to concentrations within the
metering range of the sensors. The oxygen concentrations found in surface gas discharge
are presented in Figure 4. They range from values close to zero to full concentration as in
ambient air (20.9 vol. %; four samples). This observation is surprising: if the gas leaving
the surface contains no combustion products, air must circulate through the rocks without
consumption of oxygen in the burning process.
The data do not allow for a clear separation of regions with different oxygen concentration
which could be related to the main zones of combustion in FZ 3-2. This is for two reasons.
First, the strong heterogeneity of the overlying bedrock might lead to complex flow
patterns, especially when taking into account the big fractures in porous rock. Second, the
sample locations were unevenly distributed.
519
Figure 4: Distribution of measured oxygen concentrations (vol. %) in FZ 3-2 of Wuda

Coal Mining Area (coordinates: UTM, WGS84)
3.3 Temperature logging and hydraulic measurements in boreholes
The hydraulic properties of coal fires seen as a system affect the gas flow through over-
and underlying beds. These properties were investigated through measurements of the
difference between atmospheric pressure and the pressure below surface, performed in
nine boreholes of FZ 3-2, with drillings going as deep as 20 to 25 m. In these depths, coal
seam no. 12 is intersected, and temperature logging pointed to a fire burning underground.
A sample temperature log is presented in Figure 5 (see Figure 6 for logging locations).
520
Figure 5: Sample temperature log
For hydraulic investigations, the boreholes were covered with a steel plate to interrupt air
flow. A pressure difference built up between the atmospheric pressure at the surface and
the pressure prevailing below the casing (in c. 20 m depth). This pressure difference is
generally maintained over large lateral distances and also prevails in the rock formations
around the boreholes.
Measurement results are presented in Figure 6. White lines represent isobars. Maximum
pressure differences varied between -0.25 to -2.2 mbar. The minus indicates that all wells
showed gas influx. This behavior originates from a combination of two physical transport
processes: mining-induced forced convection (sucking mine ventilation) and thermally
driven natural convection (free convection). Presumably, the pressure difference due to free
convection was less eminent than the externally induced difference due to mine ventilation.
The flow directions do not necessarily indicate the type of convection though, for in both
free and forced convection, flow can take place in either direction: whenever gas outflow
takes place, gas inflow must take place at another location in order to guarantee mass
conservation.
521
Figure 6: Position of boreholes for hydraulic measurements, temperature logging,

and temperature monitoring in FZ 3-2. Values give the pressure difference
prevailing between the surface and approx. 20 m below surface.
3.4 Temperature monitoring
Temperature monitoring was conducted over a period of one month at two positions of
FZ 3-2, using a Prodata data logger in combination with PT100 sensors, manufactured
by PHYTEC GmbH, Germany. Their location is marked in Figure 6. Both sensors were
installed c. 0.8 m below surface in small funnels with visible gas discharge. One location
was closed with sand after installation to interrupt the convective energy flow. Mineral-
insulated cables were used because of high temperatures. Monitoring results are presented
in Figure 7. At the covered sensor location, temperatures between 570 and 585 K were
observed; they changed over the monitoring period, but these temporal variations were
rather smooth. At the open sensor location, temporal variations were much more turbulent,
ranging between 715 and 735 K. These variations are presumed to originate from turbulent
convective gas motions, themselves likely to be the result of a superposition of turbulent
free convection and weather-induced air circulation.
522
Figure 7: Temperature monitoring in an open and a covered funnel in FZ 3-2
The in-situ observations presented above are proof of the great spatial and temporal
heterogeneity of gas flux through rock formations overlying burning coal seams. Massive
gas discharge occurs especially through large-scale fractures, which often display
turbulent flow behavior. An exact interpretation of the observations is very difficult, if not
impossible. Numerical simulations can help find explanations for and draw conclusions
from such complex in-situ observations.
4 Numerical Simulation
Numerical simulations are an efficient means to investigate dynamic systems such as coal
fires with respect to physical and chemical processes. For modelling purposes, a set of
macroscopic balance equations with appropriate boundary and initial conditions has to be
solved numerically with a finite elements discretization approach. Although a simplification
of reality, these equations resemble important physical and chemical processes which
take place in the coal seam and adjacent rock formations during a coal fire. The assumed
physico-chemical set-up is illustrated in Figure 8. Formulations and calculations of
mechanical influences on the over- and underlying beds have been neglected.
523
Figure 8: Assumed physico-chemical set-up for dynamic coal fire simulations. Arrows
mark the mechanisms through which the governing processes are coupled.
One major problem for the formulation of macroscopic balance equations are the strong
heterogeneities existing in FZ 3-2 (see above). Large-scale fractures have a great impact
on transport phenomena because they provide effective pathways for gas discharge, but
are impossible to be averaged over representative elementary volumes (see e.g. Yeh 2000
for a detailed derivation of macroscopically averaged balance equations and resulting
scale dependencies). For the present calculations, we used a porous medium formulation
and estimated governing parameters (like porosity and permeability) with a view to
approximating the prevailing heterogeneities. The porous medium formulation assumes a
homogeneous and isotropic nature of the macroscopic medium.
4.1 Mathematical process formulation
Physico-chemical processes as assumed here are described through a set of six differential
equations: one hydraulic equation, one energy conservation equation, and four mass
conservation equations for gas components contributing to the combustion. We will briefly
introduce the set of equations. The symbols are explained in Table 3 in the Appendix.
The mathematical formulation of turbulent-like, non-linear gas motion as a function of

pressure gradients is described by the Forchheimer equation (Nield & Bejan 1999):
524
The mathematical formulation of turbulent-like, non-linear gas motion as a function of
pressure gradients is described by the Forchheimer equation (Nield
Spontaneous Coal Seam Fires: & Bejan
Mitigating 1999):
a Global Disaster
T0 P cF
(1)
(1) U0 g P q U fl q q. .
T k k
It can be reduced to a linear Darcian flow when the last term on the right-hand-side is
neglected.
It can be empirictoForchheimer
Thereduced equation iswhen
a linear Darcian flow frequently used
the last termforon
thethe
consideration of is
right-hand-side
turbulent-like
neglected.fluid
The motions
empiric through discontinuous fractured
Forchheimer equation or porous
is frequently used media
for the(see Withaker of
consideration
1996; Skjetne et al. 1999; Rochette & Clain 2003; Fourar et al. 2004). In equation (1), q
turbulent-like fluid motions through discontinuous fractured or porous media (see
is the specific flow rate, i.e. the flow rate in m3 s-1 per unit area in m2. Thus, the specific
Withaker 1996; Skjetne et al. 1999; Rochette & Clain 2003; Fourar et al. 2004). In
ERSEC Ecological Book Series – 4
volume flow is described in units of velocity (m s-1). The first term in equation (1) describes
equation
thermal (1), q which
buoyancy, is the is
specific flow rate,
the driving i.e. the for
mechanism flow m3 s-1 convection.
rate inthermal
natural in m2.
per unit areaThe
firstThus,
differential equation
the specific is the flow
volume hydraulic equation:in units of velocity (m s-1). The first term in
is described
equation (1) ∂describes
PT thermal buoyancy, which is the driving mechanism for natural
b w fl fl fl
(2)
(3) cS +∇U• (cρp qq)T= 0
ERSEC Ecological Book Series – 4 . x Dij T S H 0 .
∂wt t
thermal convection. The first differential
ERSEC Ecological Book Series – 4 equation is the hydraulic equation:
This equation is coupled to equation (1) through the specific volume flow q. Equations (1)
S H represents the energy produced during combustion (see equation (9)).
and (2) are used to wPexpress gas mass transport as illustrated in Figure 8. Equation (1) is
(2) S fl fl x ( U fl q) 0 .
also Dcoupled O (Tto tTc p D ij qconservation
) canbUwwenergy equation, formulated as: tensor with the thermal
ij (3) U fl c pflisqthe
T macroscopic
x Dij T S H dispersion
0.
b wT wt fl fl
(3) c
This equation
conductivity ∂isTUcoupled
c p qfl Ttoequation
x Dij T(1) Sthrough
0 . the specific volume flow q. Equations
wctbcalculated+ ρ fl cas:
H
(3) p q ∇ T − ∇ • D ij ∇ T + S H = 0.
(1) and (2) are∂used t to express gas mass transport as illustrated in Figure 8. Equation (1)
S H represents the energy produced during combustion (see equation (9)).
SS H represents
isrepresents
also coupled the
the anenergy
to energy energy produced during
conservation during combustion
equation, (see
formulated(see equation (9)).
as: equation (9)).
H
Dij (4)O (T ) O (UTfl)c pflDOij0q Vis T 3produced
. the macroscopic combustion
dispersion tensor with the thermal
Dij O (T ) U fl c pflD ij q isisthethe macroscopic
macroscopic dispersion
dispersion tensor
tensor with with the
the thermal thermal
conductivity
conductivity calculated as:
calculated
conductivity
Equation as:calculated
(4) is formulated as: to incorporate the radiant heat flow at high temperatures
according to Vortmeyer 3(1980). Values for the thermal conductivity O0 are used
(4)
(4) O (T ) O0 VT .. 117
(4) O (T ) O
according to experimental0 V T 3
. findings from Haenel and Zoth (1973), Wessling and Kessels
Equation (4) is formulated
(2003), and Schmidt et al. (2005). to incorporate the radiant heat flow at high temperatures
Equation (4) is formulated to incorporate the radiant heat flowλ0 at high temperatures
according to Vortmeyer (1980). Values for the thermal conductivity are used according
Equation (4) is formulated to incorporate the radiant heat flow at high temperatures
according tofindings
to experimental Vortmeyer (1980).and
from Haenel Values for theWessling
Zoth (1973), thermal and conductivity O0 areandused
Kessels (2003),
Incorporation
according
Schmidt to of
Vortmeyer
et al. (2005). chemical (1980).reactions
Valuesin the
forcoal
theseam in
thermal combination
conductivity withO 0 transport
are usedphe-
according to experimental findings from Haenel and Zoth (1973), Wessling and Kessels
nomenatothrough
according experimentalboth the seam and
findings fromadjacent
Haenel beds was (1973),
and Zoth realizedWessling
through aand reactive trans-
Kessels
(2003), and Schmidt et al. (2005).
Incorporation
port and
(2003), of chemical
formulation
Schmidt for four
et al. reactions
chemical in
(2005). the coal seam
components: in combination
two solids with products)
(fuel and solid transportand
phenomena
two fluids through
(oxygen bothand seam and adjacent
gaseous beds was
combustion realized Combustion,
products). through a reactive
i.e. thetransport
chemical
Incorporation
formulation for four of chemical reactions
components: two solids (fuel and solid products) and twophe-
in the coal seam in combination with transport
process, was
Incorporation incorporated
of chemical according
reactions in the to Schmidt
coal seam inetcombination
al. (2003, 2005) in a single-step
with transport phe-
fluids (oxygen
nomena and gaseous
through both the combustion
seam and products). Combustion,
adjacent beds was realizedi.e. the chemical
through process,
a reactive trans-
reaction:
nomena through both the seam and adjacent beds was realized through a reactive trans-
wasport
incorporated
formulation according
for fourtochemical
Schmidt et al. (2003, 2005)
components: in a single-step
two solids (fuel and reaction:
solid products) and
port formulation for four chemical components: two solids (fuel and solid products) and
two fluids (oxygen and gaseous combustion products). Combustion, i.e. the chemical
I.I. (oxygen
two fluids X f fueland Xgaseous
O2 oxygen o X sp solidproducts).
combustion X gp gaseous products
products Combustion, 'H ,
i.e. the chemical
process, was incorporated according to Schmidt et al. (2003, 2005) in a single-step
process, was incorporated according to Schmidt et al. (2003, 2005) in a single-step
reaction:
reaction:
where X j represents the stoichiometric coefficients for the four participating species
525
I. (f), oxygen
fuel (O2),Xsolid
X f fuel product (sp), and gaseous product (gp). Stoichiometric coef-
O2 oxygen o X sp solid products X gp gaseous products 'H ,
where υ j represents the stoichiometric coefficients for the four participating species fuel
(f), oxygen (O2), solid product (sp), and gaseous product (gp). Stoichiometric coefficients
are defined negative for educts and positive for products. ∆H is the heat released in the
reaction. The consumption of fuel and the production of ash, i.e. the solid components, are
For fluid components, transport through over- and underlying beds was considered in
described through:
combination with reaction kinetics in the coal seam. The resulting reactive species
For fluid components, transport through over- and underlying beds was considered in
∂c j is (see
transport
(5) equation+ S = 0,also Kröhn
( (j = 1991
f,sp) .for over-
a derivation):
For fluid with
combination components,
∂t reaction
j transport
kinetics through
in the coal and underlying
seam. bedsreactive
The resulting was considered
species in
For combination
transport
fluidequation with reaction
w)cis (seetransport
components, j
kinetics
also Kröhn 1991infor
through the coal
and seam.
a derivation):
over- The resulting
underlying beds wasreactive species
considered
(6) qc x ( D c ) S 0, (j O , pg ) ,
transport equation
in combination wwith
t is (see
reactionj
also Kröhn
kinetics
ij j
in1991
the
j
for aseam.
coal derivation):
The
2
resulting reactive species
transport equation
w)c j is (see also Kröhn 1991 for a derivation):
(6) qc x ( D c j ) S j 0, ( j O , pg ) ,
where Dij wt )
w) Dcmolj Dj ij q is the ijmacroscopic dispersion tensor2after Scheidegger (1961).
(6)
(6) qc j x ( Dij c j ) S j 0, ( j O2 , pg ) ,
Incorporation ofwt reaction kinetics was realized through source term S j in equations (5)
whereandDij(6). )
where Fuel D ijchosen
Dmolwas q isis the
thetomacroscopic
macroscopic dispersion
dispersion
be the leading componenttensor
and after
tensor Scheidegger
the reaction (1961).
rate was
Incorporation
where Dij of reaction
) Dkinetics was realized through
ij q is the Arrhenius
Dmola second-order macroscopic source
: dispersion tensor S Scheidegger
termafter in equations (1961).
(5)
formulated
Incorporation through
of reaction kinetics was realizedlaw
through source term S jj inwas
and (6). Fuel was chosen to be the leading component and the reaction rate
equations (5)
formulated
Incorporation
through
and (6). aFuel wasofchosen
second-orderreaction
Arrhenius toEkinetics
thewas
belaw: realized
leading through source
component and theterm S j in equations
reaction rate was (5)

(7) S c c 2 k e RT
.
and (6).through
formulated Fuel
f
was chosen
f O 0
a second-order −E
to Arrhenius
be the leading
law: component and the reaction rate was
(7) S f = −c f cO2 k 0 e RRT
T.
formulated through a second-order Arrhenius law:
The other source terms areE related to S f through stoichiometry:

The(7)
other source
S f terms
c f cO2are
k 0 e related
RT
. to S f through stoichiometry:
E
(7) Sf
c f cO k 0 e RT .
§ X j ·§ M2 j ·
(8)
(8) S ¨ ¸¨ ¸Sf .
¨ X ¸are
The other sourcej terms ¨ ¸ to S f through stoichiometry:
© f ¹© Mrelated
f ¹
The other source terms are related to S f through stoichiometry:

The source term Sj in equations (5) and (6) represents the link between transport processes
and The source reactions.
chemical term Sj in equations
§ X j ·§ M j ·Thus, (5) and (6)
equations (5)represents
and (6) the valid
are link between
within transport
the coal proc-
seam. In
(8)
esses andS ¨ ¸¨ ¸ S .
the over- andjchemical
underlying
¨ X §¸¨X reactions.
·beds,
Thus,
§ ¸ equation
f equations
(5) is(5)
notand (6) areand
solved, validthe
within the coal
transport seam. (6)
equation
© f ¹© M j f ¹M j
·
is solved over- S
(8)without
In the and ¨ ¸
underlying
j source
¨ ¸ S equation
beds, f . the(5) is not production
solved, and the transport equation (see
¨ X ¸¨term. ¸Finally, energy during combustion
© f ¹© M f ¹
equation
(6) is (3)) is linked
solved withouttosource
the fuel consumption
term. by: production during combustion (see
Finally, the energy
The equation
source term(3)) isSjlinked
in equations
to the fuel(5) and (6) represents
consumption by: the link between transport proc-
essesThe(9)
and source S term
chemical ∆SHj in
H =reactions. f , Thus, equations
⋅ S equations (5) and (6)
(5)represents
and (6) arethe linkwithin
valid between
the transport
coal seam.proc-
In theesses
(9)andand
over- chemical
'H reactions.
S f , beds, Thus,
S Hunderlying equations
equation (5) solved,
(5) is not and (6) and
are valid within the
the transport coal seam.
equation
which applies within the coal seam.
(6) isInsolved
the over- and source
without underlyingterm.beds, equation
Finally, (5) isproduction
the energy not solved,during
and the transport equation
combustion (see
(6)
equationis(3))
which solved without
is linked
applies source
withintothe
the fuel
coal term. Finally, the
consumption
seam. by: energy production during combustion (see
equation (3)) is linked to the fuel consumption by:
(9) SH 'H S f ,
(9) SH 'H S f ,
which applies within the coal seam. 119

526
which applies within the coal seam.
4.2 Geometrical and numerical set-up
The geometrical set-up was modeled after the present conditions in FZ 3-2 (Chapter 2).
The numerical set-up including initial and boundary conditions is presented in Figure 9.
Self-ignition was not taken into account. Instead, an initial combustion center was placed
1
into the seam. Grid generation was performed with GINA. Local grid refinement was
employed to avoid numerical instabilities. The scenario calculations were performed with
2
the software tool ROCKFLOW. The program is the combined outcome of various PhD
theses and has therefore been tested repeatedly for the reliability of its calculations (see e.g.
Kröhn 1991; Habbar 2001; Kaiser 2001; Thorenz 2001). An additional reactive transport
module was developed at the Leibniz Institute for Applied Geosciences (GGA). This
module enables an adequate simulation of underground combustion coupled with transport
phenomena in adjacent rocks. It uses an operator-splitting algorithm to distinguish between
slow and fast chemical reaction rates, the latter prevailing at high temperatures. For fast
chemical reaction rates, the algorithm replaces the source terms in equations (7) and (8)
with a source resulting from pure transport of oxygen to the combustion center. Thus,
the overall combustion is controlled by the amount of oxygen per time unit reaching the
combustion center at high temperatures (transport-dominated as opposed to reaction-
dominated; see Kessels et al. 2008 in this volume).
Parameters for the simulation are given in Table 4 in the Appendix. The permeability of
k = 10-9 m² is unusually high for pure porous media. The normal permeability of sandstone,
for example, varies between 10-12 and 10-16 m² (Gueguen & Palciauskas 1994). Numerical
calculations by Huang et al. (2001) were performed with similarly high permeability rates
obtained from underground coal gasification. The authors argued that fractures and/or
higher permeability zones are necessary to sustain a natural convective flow which can
bring air to the coal fire. Permeability variations in the present approach led to a similar
conclusion: permeability rates lower than k = 10 -10 m² are insufficient to supply the
fire with oxygen. After all, the overlying beds in FZ 3-2 are strongly disrupted. These
disruptions represent very effective pathways for gas flow.
1
GINA, developed by H. Kunz, Federal Institute for Geosciences and Resources (BGR), Hanover, Ger-
many; contact: H.Kunz@bgr.de.
2
Institute for Fluid Mechanics and Computer Applications in Civil Engineering (ISEB), University of
Hanover, Hanover, Germany; http://www.hydromech.uni-hannover.de.
527
Figure 9: Two-dimensional geometrical and numerical set-up for scenario calculations.

Constrained boundary conditions differ for in- and outflowing gas. cf_init is the
initial fuel concentration within the coal seam.
Underneath the strongly disrupted overlying beds of FZ 3-2, mining activities may be
responsible for further oxygen flow to the combustion center (see Buhrow et al. 2004
for the impact of mining activities on surrounding rock formations). For the present
simulation, mining activities were not considered.
4.3 Comparison between simulation results and in-situ measurements
This section presents simulation results that can help explain some of the in-situ
observations in FZ 3-2. For the scenario calculations, natural thermal convection was
assumed to be the only driving force of gas flow. Thermal convection is a natural
phenomenon and results from temperature-dependent density differences (also observed in
factory slots or chimneys).
Figure 10 shows the simulated underground temperature distribution as a consequence of

combustion. Isolines mark the initial coal concentration; they circumscribe the unconsumed
coal. Evidently, a high amount of thermal energy is still stored in the overlying rocks
although the fire already moved further through the seam.
528
Figure 10: Simulated temperature distribution as a consequence of combustion below

surface after 246 days
4.3.1 Oxygen concentration at the surface

Simulated results for oxygen distribution are presented for two consecutive dates with a
delay of 340 days in Figures 11. Colored contour plots show the oxygen concentration as a
result of gas flow through over- and underlying beds. Vectors indicate the surface locations
at which gas leaves or enters the surface. In the left simulation of Figure 11, the coal seam is
still ablaze. 340 days later, it has burned out in full thickness and the combustion centers have
spread along the seam in downward and upward direction. An important observation made
in these simulations concerns the oxygen distribution at the surface. As long as the coal seam
has not completely burned out along its vertical axis, zero oxygen concentration is observed
at locations of gas discharge. After the fire breakthrough (Figure 11; right), gas flow is still
observed above the previous combustion center, but oxygen consumption stopped due to
missing fuel. As a result, the oxygen concentration in the gas discharge varies between
zero and the full concentration as prevailing in air (0.29 kg m-3). Oxygen concentrations of
this order were also measured in FZ 3-2 (Figure 4). Thus, in-situ observations can indeed
be explained by means of simulated transport phenomena in rock strata surrounding the
coal fire.
529
Figure 11: Simulated oxygen concentration in adjacent beds during combustion for two
consecutive dates (340 days delay). Left: Coal seam still ablaze. Right: Seam
completely burned out along vertical axis.
4.3.2 Pressure measurements and calculations

Simulated dynamic pressure differences due to pure thermal convection are presented in
Figure 12. The pressure varies from -0.12 to +0.06 mbar – evidently, a direct response to
the temperature distribution described above (Figure 10). In-situ measurements, however,
yielded pressure values varying by one order of magnitude. Pure thermal convection then
is not the only driving force for gas motions in coal seam no. 12. Instead, we believe that
the pressure differences are a result of mining-induced ventilation (forced convection).
This observation is essential for preventing unintended self-ignition. It shows that air
penetrates the ground and supplies the coal fire in seam no. 12 with oxygen. If other
coal seams intersect the penetration path, these are also supplied with oxygen and hence
exposed to the danger of self-ignition.
530
Figure 12: Simulated distribution of dynamic pressure due to pure thermal convection
4.3.3 Relation between in- and outcoming gas flow

Figure 13 shows the absolute value of the specific volume flow of gas. Vectors at the
surface indicate flow direction; if they are directed downward, they illustrate surface
locations where gas enters the subsurface, and vice versa. Although the numerical model
assumes homogeneous media, the specific volume flow displays strong variations at
surface level – one order of magnitude between in- and outflowing gas flow; values
between 0.1 and 3 m d-1. The lower value corresponds to inflowing surface areas and the
higher value to those of gas discharge. Gas flow into the subsurface covers a large area, but
with a low flow rate. In contrast, gas flow out of the surface takes place over small areas,
but with higher flow rates. This result might explain why surface locations through which
gas enters the subsurface could not be observed in situ.
531
Figure 13: Simulated specific volume flow q in meters per day (m d-1)
So far, only two-dimensional numerical simulations have been presented with a view
to finding possible explanations for the in-situ observations described in Chapter 3. A
major problem for a detailed comparison is the incongruence of the highly heterogeneous
overlying rocks in FZ 3-2 with the homogenized simulation set-up.
4.3.4 Comparison of flow rates

This section presents an attempt to compare in-situ measurements in large-scale
heterogeneous systems with simulation results from homogeneous models. For this, a
three-dimensional set-up of FZ 3-2 was used. The set-up (Figure 14) is a simplification
of a static three-dimensional model from Deutsche Montan Technologie GmbH (DMT).
The grid is composed of 60,000 elements, which increases the computation time greatly
in comparison to two-dimensional simulations. Again, a small combustion center with an
initial temperature T = 500 K is assumed for coal seam no. 10.
The simulation results are presented in combined view with the in-situ ventilation
measurements in Figure 15. This comparison can only give an indication of how
532
to calibrate homogeneous numerical models by means of in-situ observations in

heterogeneous media. The total area over which ventilation measurements were performed
is given by a yellow circle of 120 m in diameter. It covers an area of 14,000 m² (cf.
Chapter 3.1) and represents the approximate location of where gas discharge flow rates
were measured. The red circle (35 m in diameter) marks the area of surface gas discharge
according to our simulations. The average specific volume flow can be determined for both
circles. Likewise, the total gas discharge volumes can be estimated and compared with
each other (Table 2). A successful calibration is achieved when both the circles and the
averaged volumes are in agreement within a certain tolerance. If the values do not agree,
hydraulic and/or thermal parameters have to be adjusted. This way, prognostic scenario
calculations can give realistic insights into the development of underground coal fires.
Figure 14: Set-up for three-dimensional calculations, coal seam no. 10 in Wuda Coal
Mining Area. Initial and boundary conditions are identical with two-
dimensional set-up of Figure 9.
533
Table 2: Comparison between results from ventilation measurements and simulation

Approach Area of Gas Total Gas Volume Average Specific
Discharge (m2) (m3 h-1) Volume Flow (m h-1)
Ventilation 14,000 37,149 2.96
measurements
3D simulation 3,000 938 0.33
Figure 15: Top view of three-dimensional simulation block. Circles mark areas of gas
discharge. Colors indicate the magnitude of simulated specific volume flow.
4.4 Simulation of surface sealing
One commonly applied method to reduce oxygen transport in coal fire fighting and
prevention is surface sealing with nearly impermeable material – clay or very fine-grained
sand, for example.
Coal fire extinction by surface sealing was simulated on the basis of the two-dimensional
model displayed in Figure 9. The top of the domain was covered with a layer of very low
-18
permeability = 10 m² (Figure 16). Similar permeability values are observed for clay
(Gueguen & Palciauskas 1994), which is hence a useful material for coverage. All other
parameters remained unchanged (Table 4). The steep outcrop and the surface below seam
level were not covered in accordance with the real conditions in FZ 3-2. Simulations were
performed with the simulation results displayed in Figure 10. Results are presented in
Figure 17. In case of non-coverage (Figure 17; left), the fire propagates up- and downwards
along the coal seam at an average temperature of 1,250 K. The comparison with the
534
temperature distribution in Figure 10 shows that the maximum temperature remains nearly
unchanged after 279 days. In case of surface coverage (Figure 17; right), the maximum fire
temperature drops to 917 K after 279 days at the upward-moving fire. At the downward-
moving fire, it drops to about 750 K. These results stem from a change in gas circulation
pattern. The streamlines in Figure 17 (right) show that the coverage acts as a cap for gas
motion. Gas flow is restricted to the subsurface system, gaseous matter being unable to
leave or enter the surface. The upward-moving fire is still supplied with oxygen as gas
now enters through the scarp. The scarp represents an obstacle for fire quenching because
non-coverage will lead to continued oxygen supply. This, in turn, impedes the effective
reduction of combustion temperature.
Figure 16: Set-up for numerical simulation of extinction effects of surface coverage
Figure 17: Simulated temperature distribution of uncovered (right) and covered (left)
surface after 279 days
535
5 Conclusion and Outlook
Numerical simulations allow for prediction of coal fire propagation as a function of

hydraulic, thermal, and chemical processes. These processes take place in the coal seam
and in adjacent bedrock. Mechanical processes, which have not been accounted for in
the present simulations, also influence fire propagation, and should be included in future
efforts. Nonetheless, the current model helps to understand transport phenomena caused by
subsurface combustion.
Our simulations suggest that full oxygen concentrations in surface gas discharge above a
coal fire may be due to oxygen transport through thermal buoyancy. These concentrations
occur whenever the coal seam has burned out vertically. Gas discharge at the surface takes
place only over small areas, but at high flow rates; whereas fresh air permeates into the
ground over large areas, but at relatively low flow rates. This difference might explain why
locations of air intake are hardly ever found in the field. Finally, the comparison between
measured and simulated pressure also proved that mine ventilation must cause air intake at
coal seam no. 12 in FZ 3-2 of Wuda Coal Mining Area.
A precise comparison between homogeneous models and in-situ measurements is difficult

because of the very heterogeneous nature of overlying rock strata in natural settings.
Comparing and calibrating three-dimensional simulations with surface-near ventilation
measurements may solve this problem. Proper calibration enables reliable prognostic data
about fire development, which are essential for efficient application of extinction strategies
and, thus, for successful prevention of coal mining hazards.
Our extinction scenario proves that numerical models can also be used for prognosis of
fire quenching. In summary, numerical simulations are an effective means for the scientific
investigation of subsurface coal fires and can be used for real-life applications such as the
planning of extinction strategies.
Acknowledgements
We acknowledge financial support of the Sino-German Coal Fire Research Initiative from
the German Federal Ministry of Education and Research (BMBF); Grant No. 0330490B.
We would also like to thank Jia Yaorong and his team for great “in-situ” support in Wuda
Coal Mining Area. We enjoyed his great organization and company in a good many
field trips. Finally, great thanks go to the project partners, who contributed with helpful
discussions to the overall success of the project.
536
Appendix
Table 3: Nomenclature
Symbol
Symbol Description
Description Unit
Unit
ccj j Concentration of component j kg m-3

Concentration of component j kg m-3
cb Bulk heat capacitiy J m-3 K-1
ccb Bulk heat capacitiy
Forchheimer drag coefficient m-3 K-1
J…
F
c fl Specific heat capacitiy of fluid (gas) J kg-1 K-1

cFP Forchheimer drag coefficient …
Dij Macroscopic dispersion tensor m2 s-1
2 -1-1 -1
cDPflmol Specific
Molecularheat capacitiy
diffusion of fluid (gas)
coefficient Jmkg s K
E Activation energy J mol-1

D ij Macroscopic dispersion tensor m2 s-2-1
g 9.81 Gravitational acceleration (positive in z direction) ms
'H Heat of reaction J kg-1
D mol Molecular diffusion coefficient m2 s-1
k Permeability m2
-1
E Activation energy J mol
k0 Pre-exponential factor m3 s-1 kg-1
gM 9.81
j
Gravitational
Molar mass ofacceleration
component(positive
j in z direction) m s-2kg-1
mol
'PH Absolute
Heat fluid pressure
of reaction JPa
kg-1
kR 8.31 Unified gas constant
Permeability mJ 2mol-1 K-1
S Storativity Pa-1
k0 Pre-exponential factor m3 s-1-3kg-1-1
Sj Source term of component j kg m s
S Hj
M Sourcemass
Molar termofofcomponent
energy release
j J m-3kgs-1
mol -1
T Temperature K
P Absolute fluid pressure Pa
T 273 Reference temperature K
R 0 8.31 Unified gas constant J mol-1 K-1
q Specific volume flow m s-1
S Storativity Pa-1
D ij (D l , D t ) Longitudinal/transversal dispersion length M
Sj Source term of component j kg m-3-1 s-1-1
O Thermal conductivity (temperature-dependent) Wm K
SOH0 Thermal conductivity at room temperarture W m-1 K-1

Source term of energy release J m-3 s-1
P Dynamic viscosity
ERSEC Ecological Book Series – 4 kg m-1 s-1
T Temperature K
Xj Stochiometric coefficient of component j …
T0 273 Reference
Porosity temperature K
…
)
-3
qU fl Fluid density
Specific volume flow Kgs-1m-3
m
U 1.29 Reference density of fluid Kg m-3
D ij0 (D l , D t ) Longitudinal/transversal dispersion length M
V Constant W m-1 K-4
O 131
Thermal conductivity (temperature-dependent) W m-1 K-1
Table 4: Parameters for numerical simulation
O0 Thermal conductivity at room 537
temperarture W m-1 K-1
Parameters Value Unit
P
Thermal parameters Dynamic viscosity kg m-1 s-1
U0 1.29 Reference density of fluid Kg m-3
V Ecological BookConstant
ERSEC Series - 4 W m-1 K-4
Table 4: Parameters
Table 4: Parameters for
fornumerical
numericalsimulation
simulation
Parameters Value Unit
Thermal parameters
c pfl 1000 J kg-1 K-1
cb 2.4 × 1066 (solid rock);

2.4 J m-3 K-1
3.06 ×× 10
3.06 1055 (coal)
(coal)
O0 2.0 (solid rock); 0.1 (coal) W m-1 K-1
V 7 × 10-9 W m-1 K-4
Reaction kinetic parameters

E/R 12700a K
'H 2.19 × 107 b J kg-1
k0 4.7 × 106 a m3 kg-1 s-1
M f , M sp 1089, 489b mol kg-1
M O2 , M gp 32, 30.5b mol kg-1
X f ,X sp +1, -1 c …
XO , X gp
2
26, -47b …
Hydraulic parameters
cF 0.5 …
Dmol 10-6 m2 s-1
D ij (D l , D t ) (0.5, 0.1) m2
k 10-9 m2
S 10-6 Pa-1
) 0.2 (solid rock); 0.4 (coal) …
P 1.81 × 10 -5
kg m-1 s-1
aa Schmidt et al.
al. (2003)
(2003)
b
Schmidt et
Lohrer et al. (2004)
132
538
References
Brenner, G., Pickenäcker, K., Pickenäcker, O., Trimis, D., Wawrzinek, K., and Weber, T., 2000. Numerical
and experimental investigation of matrix-stabilized methane/air combustion in porous inert
media. Combustion and Flame 123: 201-213.
Brooks, K., Balakotaiah, V., and Luss, D., 1988a. Effect of natural convection on spontaneous combustion
of coal stockpiles. AIChE Journal 34(3): 353-365.
Brooks, K., Bradshan, S., and Glasser, D., 1988b. Spontaneous combustion of coal stockpiles – An
unusual reaction engineering problem. Chemical Engineering Science 43(8): 2139-2145.
Buhrow, C., Lippmann, G., and Stöttner, M.T., 2004. Kohlebrände in der Volksrepublik China. Glückauf
140(10): 488-494.
Fourar, M., Radilla, G., Lenormand, R., and Moyne, C., 2004. On the non-linear behavior of a laminar
single-phase flow through two- and three-dimensional porous media. Advances in Water
Resources 27: 669-677.
Gielisch, H. and Kahlen, E., 2003. Geology of the Helan Shan. Field report. Project re-port for Work
Technologies for Exploration, Extinc-tion, and Monitoring of Coal Fires in North China.”
Wessling, Germany.
Gueguen, Y. and Palciauskas, V., 1994. Introduction to the physics of rocks. Princeton, New Jersey:
Princeton University Press.
Habbar, A., 2001. Direkte und inverse Modellierung reaktiver Transportprozesse in klüftig-porösen
Medien. PhD thesis, Institute for Fluid Mechanics and Computer Applications in Civil
Engineering (ISEB), University of Hanover, Hanover,Germany.
Haenel, R. and Zoth, G., 1973. Heat flow measurements in Austria and heat flow maps of Central Europe.
Z. Geophys. 39: 425-439.
Hjertager, B.H. (1986). Three-dimensional modeling of flow, heat transfer, and combustion. In
Cheremisinoff, N.P. (Ed.), Handbook of heat and mass transfer. Vol. I: Heat transfer
operations. Houston, Texas: Gulf Publishing Company, 1303-1350.
Huang, J., Bruining, J., and Wolf, K.H.A.A., 2001. Modeling of gas flow and temperature fields in
underground coal fires. Fire Safety Journal 36: 477-489.
Kaiser, R., 2001. Gitteradaption für die Finite-Elemente-Modellierung gekoppelter Prozesse in geklüftet-
porösen Medien. PhD thesis, Institute for Fluid Mechanics and Computer Applications in
Civil Engineering (ISEB), University of Hanover, Hanover, Germany.
Kessels, W., Wuttke, M.W., Wessling, S., and Li, X., 2008. Coal fires between self-ignition and fire
fighting: Numerical modeling and basic geophysical measurements. In UNESCO Office
539
Beijing (Ed.), Spontaneous coal seam fires: Mitigating a global disaster. International
Research for sustainable control and management. ERSEC Ecological Book Series – 4.
Beijing: Tsinghua University Press and Springer Verlag, 467-484. (this volume)
Klika, Z., Kozubek, T., Martinec, P., Kliková, C., and Dostál, Z., 2004. Mathematical modeling of
bituminous coal seams burning contemporaneously with the formation of a variegated beds
body. International Journal of Coal Geology 59: 137-151.
Krause, U. and Schmidt, M., 2001. The influence of initial conditions on the propagation of smoldering
fires in dust accumulations. Journal of Loss Prevention in the Process Industries 14: 527-532.
Kröhn, K.P., 1991. Simulation von Transportvorgängen im klüftigen Gestein mit der Methode der Finiten
Elemente. PhD thesis, Institute for Fluid Mechanics and Computer Applications in Civil
Engineering (ISEB), University of Hanover, Hanover, Germany.
Lohrer, C., Schmidt, M., and Krause, U., 2004. Ein Modell zur Berechnung des gekoppelten Wärme- und
Stofftransportes in porösen, reaktionsfähigen Medien. Project report, as a contribution to the Sino-
German Coal Fire Research Initiative “Innovative Technologies for Exploration, Extinction, and
Monitoring of Coal Fires in North China.” Working Group “Flammable Bulk Materials and Dusts,
Solid Fuels,” Federal Institute for Materials Research and Testing (BAM), Berlin, Germany.
Nield, D.A. and Bejan, A., 1999. Convection in porous media. New York: Springer Verlag.
Perkins, G. and Sahajwalla, V., 2005. A mathematical model for the chemical reaction of a semi-infinite
block of coal in underground coal gasification. Energy and Fuels 19: 1679-1692.
Rochette, D. and Clain, S., 2003. Numerical simulation of Darcy and Forchheimer Force Distribution in a
HBC Fuse. Transport in Porous Media 53: 25-37.
Rosema, A., Guan, H., and Veld, H., 2001. Simulation of spontaneous combustion, to study the causes of
coal fires in the Rujigou Basin. Fuel 80: 7-16.
Scheidegger, A.E., 1961. General theory of dispersion in porous media. J. Geophys. Research 66(10):
3273-3278.
Schmidt, M., Lohrer, C., and Krause, U., 2003. Self-ignition of dust at reduced volume fractions of
ambient oxygen. Journal of Loss Prevention in the Process Industries 16: 141-147.
Skjetne, E., Hansen, A., and Gudmundsson, J.S., 1999. High-velocity flow in a rough fracture. Journal of
Fluid Mechanics 383: 1-28.
Thorenz, C., 2001. Model adaptive simulation of multiphase and density driven flow in fractured and
porous media. PhD thesis, Institute for Fluid Mechanics and Computer Applications in Civil
Engineering (ISEB), University of Hanover, Hanover, Germany.
Vortmeyer, D., 1980. Radiation in packed solids. Ger. Chem. Eng. 3: 124-138.
Wessling, S. and Kessels, W., 2003. Thermal parameters for dry rocks with respect to high-temperature
variations. Project report, as a contribution to the Sino-German Coal Fire Research Initiative
540
China.” Archive No. 012 3542. Leibniz Institute for Applied Geosciences (GGA), Hanover,
Germany.
Westbrook, C.K. and Dryer, F.L., 1981. Simplified reaction mechanism of the oxidation of hydrocarbon
fuels in flames. Combustion Science and Technology 27: 31-43.
Withaker, S., 1996. The Forchheimer equation: A theoretical development. Transport in Porous Media 25:
27-61.
Yang, L., 2004. Study on the model experiment and numerical simulation for underground coal
gasification. Fuel 83: 537-584.
Yeh, G.T., 2000. Computational subsurface hydrology – Reactions, transport, and fate. Dorwell,
Massachusetts: Kluwer Academic Publishers.
541
Part VII
Fire Fighting
and Prevention
Model and Control of
High-Temperature Coal Fires
Zhang Xinhai1,2, Xu Jianzhong1,3, Xi Guang1, Deng Jun2, and Wen Hu2
1
Energy and Power School, Xi’an Jiaotong University, Xi’an, Shaanxi Province, P.R. China
2
Xi’an University of Science and Technology, Xi’an, Shaanxi Province, P.R. China
3
Engineering Thermo-Physics Institute, China Academy of Sciences, Beijing, P.R. China
Abstract
Temperatures in the combustion center of a burning coal seam are always very high. This paper presents
a model depicting chemical reactions and mass transfer in high-temperature fire zones. The model is
premised on the following assumptions: (i) Oxidation and gasification reactions that produce CO and
CO2, and convert coal into coke take place on the surface of high-temperature coal. (ii) The CO produced
in these reactions streams upward, where it mixes with oxygen flowing downward from the surface
through cracks and fissures; this mixture of gases may combust in flames. (iii) High-temperature coal is
reactive with water and CO2 and produces combustible gases like CO, CH4, and H2. Extinguishing high-
temperature coal fires by means of water injection alone may enhance reaction rates and cause serious
explosions. Field tests have proven that a combination of fire fighting measures, i.e. excavation of high-
temperature coal followed by outcrop sealing and injection of composite gel and slurry is most effective
for extinction.
摘要
煤田火灾中燃烧中心温度通常非常高。本文提出了一个描述煤田火区化学反应和传质过程的模
型：（1）火区高温煤体表面发生氧化反应和气化反应并产生CO和CO 2，并且使煤炭转化成焦
炭；（2）产生的CO向煤层表面移动，并且与从地表下沿裂缝和裂隙移动的O2混合，可引起燃烧
而产生火焰；（3）高温煤体容易与水和CO2反应产生像CO、CH4和H2等可燃气体。通过注水或
二氧化碳灭火时，有时反而会提高燃烧速度甚至引起灾难性的爆炸。野外实验表明，综合灭火技
术，如剥挖高温煤体，接着对露头煤层进行封堵，再注入复合胶体和泥浆等，是效果最好效率最
高的灭火措施。
544
1 Introduction
Coal fires are a severe problem in northern China. In some areas, they have been known for
more than 100,000 years; the deepest among these fires reach down to almost 400 m below
surface (Guan 1998). Coal fires presently occur in Xinjiang, Ningxia, and Inner Mongolia
Autonomous Regions, as well as Shaanxi, Shanxi, Qinghai, and Gansu Provinces. 35
fire sites are reported from both sides of Mount Tian in Xinjiang. Out of the c. 720 km2
currently affected by coal fires in northern China, 17 to 20 km2 are actually burning fire
zones (Guan 1998). These fires eat away about 13 million t of coal annually – a serious
economic loss and source of environmental pollution for the country (Liu 2004).
Rosema et al. (2001) studied the development of coal fires in Rujigou Coalfield of
Ningxia Autonomous Region in numerical simulations; Xu (2002) and Deng and Xu
(2005) had the same geographical focus, but explicitly researched coal self-ignition
at low temperature ranges. The coal burning in underground coal seam fires is always
covered by overlying rock. Cracks within this rock allow air to penetrate from the
surface. Temperatures in the combustion center of a fire are generally very high and
liken those of a furnace. The highest temperatures measured in Shangyi Fire Zone of
Rujigou Coalfield were 3,000 °C and above. Reaction mechanisms, kinetic parameters,
and development trends of coal fires at high temperatures are quite different from those
at lower ranges. There exist only a few studies on coal fire development and combustion
mechanisms at high temperatures.
2 Reaction Mechanisms of High-Temperature Coal Fires
Coal seam outcrops have been burning for years at many locations in China. The
combustion area sometimes extends over more than 1,000 m2 and may advance into
the ground to a depth of more than 100 m. Cracks and fissures riddle the rock above
burning seams and allow air to enter the ground and reach to the combustion center,
thus providing the oxygen (O2) needed for combustion. New cracks form in the wake
of land subsidence and cave-ins of burned coal bodies, or as a result of inhomogeneous
temperature distribution in the fire zone. In the semi-closed environment of a burning
seam, the heat generated during coal oxidation can accumulate, and ultimately raises
the temperature inside the seam to degrees as high as 3,000 °C. Excavations proved that
temperatures at about 2 m above burning coal seams were high enough to produce raging
flames.
545
flames.

2.1 Reactions in burning coal at different temperatures
2.1 The
Reactions inof
temperature burning coal
a coal fire at different
varies temperatures
remarkably from over 3,000 °C in the combustion
center to ambient temperature at the far outer edge of the fire zone. With rising tempera-
The temperature of a coal fire varies remarkably from over 3,000 °C in the combustion
ture, the production rate of volatiles rises, too. Volatiles are combustibles and may burn
center to ambient temperature at the far outer edge of the fire zone. With rising
in flames.theTheir
temperature, production
production rate
rate of increases
volatiles rises,with
too.increasing combustion
Volatiles are time,
combustibles andalong
may the
burnfollowing
in flames.equation:
Their production rate increases with increasing combustion time, along the
following equation:
dV § E ·
(1) k 0 exp¨ ¸Vmax V .
(1) dW © RT ¹
where τ is the time of combustion, k0 the apparent pre-exponential factor, E the activation
energy,
whereR the universal
W is the timegas
of constant, Vmax kthethe
combustion, maximum
apparentproduction rate of factor,
pre-exponential coal volatiles,
E the activa-
0
and V the production rate of volatiles with time.
tion energy, R the universal gas constant, Vmax the maximum production rate of coal
volatiles,
While and V
the overall the production
production rate ofrate of volatiles
volatiles withthe
increases, time.
rate of increment slows down
with increasing combustion time. After a certain time of burning, almost all coal has turned
into carbon (C). According to the combustion theory of Xu and Xu (1990), the reaction of
C with O2 under different temperature regimes is as follows:
1) At the edge of the fire zone, where temperatures

4 are below 700 °C, the chemical
reactions that take place are:
(2) 3C + 2O2 → C3O4
(3) 2C3O4 + O2 → 2CO + 4CO2
These processes are characterized by the diffusion of O2 through the bedrock to the coal
and subsequent oxidation, i.e. production of carbon monoxide (CO) and dioxide (CO2)
from the reaction of coal with O2. There is no further combustion of CO and the gaseous
reaction products diffuse into the bedrock to leave at the surface.
2) As temperatures rise to 800 to 1200 °C in direction of the combustion center, the

reactions between coal and O2 are as follows:
(4) 3C + 2O2 → C3O4
(5) C3O4 → 2CO + CO2
(6) CO + O2 → CO2
546
O2 diffuses to the coal and reacts with carbon; CO and CO2, the products of this oxidation
reaction, diffuse back into the bedrock. CO at these temperatures may also combust, forming
open flames even at a distance to the coal seam. The CO2 produced in this latter reaction
diffuses into the bedrock, where it no longer forms part of the combustion process.
3) At temperatures above 1,300 °C, the main reaction at the coal surface is gasification; it
can be described as:
(7) C + CO2 → 2CO
The CO produced in this gasification reaction diffuses into areas at a distance from
the coal (with lower temperatures), where it oxidizes to CO2 in flaming combustion.
Reactions at the flaming front follow equation (6). The CO2 generated may partly
diffuse back to the coal surface and react again with C in the same reaction.
2.2 Effects of water injection on high-temperature burning coal
Water (H2O) is reactive with high-temperature coal and produces partly explosive gases
upon contact, leading to disastrous explosions upon injection of great volumes of water or
rainfall. In 2001, such explosions took place in Shangyi Fire Zone of Rujigou Coalfield
when water was injected into a high-temperature zone. The injection was followed by a
geyser-like eruption of water and water vapor through one of the boreholes (Figure 1).
Figure 1: Water eruption after injection of water in Shangyi Fire Zone of Rujigou
Coalfield, 2001
547
Water evaporates in proximity to high-temperature coal, where the vapor reacts with the
coal to produce explosive gases. The reaction between C and H2O at high temperatures is
as follows:
(8) H2O + C → H2+ CO
When hydrogen (H2) and CO are mixed with O2 at high temperatures, explosions may
result as in the following two reactions:
(9) H2 + O2 →H2O
(10) CO + O2 →CO2
Reactions (9) and (10) are controlled by diffusion. The diffusivity of H 2 is largest,
followed by H2O and CO; CO2 has the smallest diffusivity among the four gases. Thus,
the rate of reaction (10) at high temperatures is larger than that of (9). As a result,
injecting only small amounts of water into a larger body of high-temperature coal may
in fact enhance combustion. However, as both reactions (9) and (10) are endothermic,
they have the potential to lower temperatures in the coal and thus reduce their own
reaction rates given there is no additional heat source that counteracts this effect. The
injection of an amount of water large enough to affect the whole coal body may hence
effectively reduce temperatures and ultimately extinguish a coal fire.
3 Coal Fire Physical Model
In Shangyi Fire Zone of Rujigou Coalfield, coal fires are often extinguished by excavating
burning coal. Most of these excavations come to reveal abandoned mine shafts or
workings, through which the fire has propagated underground. A lot of cracks develop in a
coal seam and its overlying bedrock during a fire. Temperatures even several meters above
the seam are relatively high, and flames can reach as far as to the surface, where they are
visible especially at night. A schematic cross-section of Shangyi Fire Zone is shown in
Figure 2.
548
Figure 2: Schematic cross-section and gas flow of Shangyi Fire Zone in Rujigou
Coalfield
The high-temperature zone in Shangyi is situated around an abandoned laneway. The

flaming zone in the rock lies about 2 m above the burning coal seam; it measures 1 to 2 m
in height, as opposed to the 10 m-thick coal seam. Temperatures in the combustion center
can reach up to 3,000 °C and above; at the flaming zone they are at 1,000 to 1,500 °C.
Heat flow from the burning coal to the bedrock takes place in the form of conduction and
convection.
Reactions (9) H2 + O2 → H2 O and (10) CO + O2 → CO2 are endothermic reactions. If they

are the only reactions to take place on the surface of the burning coal seam, temperatures
will decrease and the fire will ultimately die out. Long-term monitoring of Shangyi Fire
Zone, however, showed that the temperatures in the combustion center had hovered at
3,000°C for years. This proves that in addition to reactions (9) and (10), other, exothermic
oxidation reactions must take place underground; they keep the fire zone in a dynamic heat
equilibrium. Based on these insights, the following model of governing coal fire processes
in Shangyi was developed:
Air penetrates through cracks and abandoned mine workings to the coal body, where
O2 reacts with coal to produce CO and CO2 as in reactions (2), (5), and (6). The CO2
549
generated may directly react with high-temperature coal to produce CO following reaction
mechanism (7). Buoyancy caused by inhomogeneous temperature distribution (temperature
gradients) propels gas circulation from the combustion center to the surface and vice versa,
establishing a constant counter-directional flow of flue gases and fresh air through cracks
and fissures in the coal and bedrock. Air (O2) from two sides of the high-temperature zone
flows to the flaming zone and reacts with CO and other volatiles to form CO2 and H2O
in flaming combustion. Finally, CO2 and H2O, as well as inert nitrogen gas (N2) stream
upward to the surface and diffuse into the air (Figure 2).
4 Analysis of Fire Fighting Measures
In view of the specific reaction mechanisms of high-temperature coal combustion and the
geological/reaction characteristics of Shangyi Fire Zone, the most elementary measures to
be taken in fire fighting are to:
1) Lower the temperature of the combustion zone by injection of water and/or slurry; and
2) Reduce or best halt the oxygen supply by sealing the entire fire zone with loess; or
3) Remove, i.e. excavate the burning coal.
Excavation can extinguish a fire, but in most cases the velocity of fire propagation into
the coal seam is greater than that of excavation. In Shangyi Fire Zone, excavation is
only used in those sections where the overburden is not too thick. Sealing with loess,
on the other hand, can quench a fire or at least inhibit its progress by reducing O 2
concentration around the coal; but it lowers temperatures underground at only a very
slow pace. New cracks can form in the loess layer and the fire may rekindle after a
phase of dormancy.
All reaction processes involved in coal combustion can be restrained by reducing air
leakage and lowering coal and rock temperatures. Xu et al. (2003) and Zhang et al. (2004)
proposed a composite gel for fire extinction – a very economical and effective method
to fight spontaneous coal seam fires. Much less composite gel is needed for filling of
cracks than for example loess for surface sealing. Another advantage of gel is the speed
at which it reduces temperatures underground, faster for example than sealing or water
injection. Since the evaporation rate of water in gel is much slower than that of pure water,
explosions are fairly unlikely. When water or slurry are being used for extinction, a large
portion of the water pumped into the rock may leak away through fractures and cracks; gel
550
in contrast is highly viscous and can remain for a much longer time in the high-temperature
zone.
Fire-fighting based on composite gel injection proceeds in the following three steps: First,
a suspension slurry is prepared using local materials such as rock powder, sand, loess, or
fly ash, mixed with water and a few other additives. In a second step, the slurry is pumped
into the ground near the fire site, after which a smaller amount of gelling agent is added.
The mixture is then injected into the cracks above the coal fire, where it will gel and thus
hinder further air leakage. This, in turn, will lower temperatures in the coal and bedrock
and eventually extinguish the fire.
5 Combining Different Fire Fighting Measures for Successful

Extinction
Coal seam outcrops in Shangyi Fire Zone have been burning for decades. The fire area
covers several square kilometers and combustion reaches far into the ground, threatening
the safety of nearby mine workings. Measures such as water and composite gel injection,
sealing, and excavation have been combined to extinguish coal fires here. Figure 3
illustrates their application.
Figure 3: Combination of coal fire fighting measures in Shangyi Fire Zone (schematic
cross-sectional view)
551
In Shangyi, high-temperature coal at a coal seam outcrop was excavated and the area
exposed covered with loess for isolation. Boreholes were drilled from a platform above
the burning coal seam to inject composite gel and slurry. Some 9,000 m3 of composite gel
were pumped into an area of roughly 2,000 m2. In response, temperatures at the surface
quickly approached ambient temperature, and smoke ceased to rise from the fire zone. The
combination of different fire fighting measures, evidently, was successful.
6 Conclusion
Coal at high temperatures is reactive with H2O and CO2, and fire fighting by water injection
alone may cause disastrous explosions under these conditions. Stopping air flow to the fire
zone can inhibit all reactions involved in coal combustion. Filling cracks and fissures in
coal and bedrock, i.e. pathways of O2, with composite gel is a more effective means of air
blockage and temperature reduction than surface sealing with loess or injections of water
and slurry. Field tests have proven that a combination of measures, i.e. excavation of high-
temperature coal followed by outcrop sealing, injection of composite gel and slurry, is
more successful in extinguishing coal seam fires than any single measure alone.
Acknowledgements
The research behind this paper was supported by the Natural Science Fund of Shaanxi
Province, P.R. China (2005D02). Some of the ideas propounded were sparked by Prof.
Guan Haiyan’s presentation at the International Conference on Coal Fire Research.
References
Deng, J., Xu, J.C., and Wen, H., 1998. Experimental studies and numerical simulations of short-term coal
self-ignition. Journal of Xiangtan Mining Institute 8(1): 11-16. (in Chinese)
Liu H.P., 2004. Underground fire swallows 13 million tons of coal annually. Liaowang News Weekly 11:
30-31. (in Chinese)
Rosema, A., Guan, H.Y., and Veldc, H., 2001. Simulation of spontaneous combustion to study the causes
of coal fires in the Rujigou Basin. Fuel. 80(1): 7-16.
552
Xu, J.C., Zhang, X.H., Wen, H., and Deng, J., 2003. Theory and technology of extinguishing spontaneous
coal fires with gel. Beijing: Coal Industry Publishing House. (in Chinese)
Xu, J.Y. and Xu, T.M., 1990. Combustion. Beijing: Machinery Industry Publishing House. (in Chinese)
Zhang, X.H., Deng, J., Xu, J.C., and Wen, H., 2004. Technique of extinguishing and preventing coalfield
spontaneous combustion. In Wang, Y.J., Huang, P., and Li, S.C. (Eds.), Progress in safety
science and technology. Proceedings of the 2004 International Symposium on Safety Science
and Technology (ISSST), Shanghai, P.R. China, 25-28 October 2004. Vol. II. Beijing/New
York: Science Press, 299-301.
553
Control and Prevention of Coal Fires
in Sealed-Off Mine Areas
Using Different Means of Inertization
D.D. Tripathi
Mine Fire Laboratory, Central Institute of Mining and Fuel Research (CIMFR),
Dhanbad, Jharkhand, India
Abstract
Fire hazards have always been associated with underground coal mining in India. Owing to widely
varying geo-mining conditions, even at one and the same site, there is no distinct technology to date that
could deal with all types of underground mine fires comprehensively. Several factors are involved in
selecting suitable techniques: the state and extent of a fire, site-specific conditions, resource availability,
and cost-efficiency. Some techniques are more effective with extensive sealed-off fires, while others
are more effective with newly developed and thus limited fires, either in goafs, crushed coal pillars, or
abandoned mine workings. The purpose of this paper is to explicate the application of inert gases, with or
without endothermic chemicals, and different means of isolation for control and prevention of coal fires in
sealed-off mine areas, i.e.: (i) N2; (ii) CO2; (iii) N2 and CO2 (60:40); (iv) water fog, chemical fog, nitrogen
fog, fog gel, foam, nitrogen foam; and (v) silica gel, sealant, or other coating materials. By means of three
case studies, this paper aims to highlight the selection criteria for each technique, their modus operandi,
scope of application, as well as successes and limitations in practice.
摘要
印度的煤火灾害常常与地下采矿活动密切相关。由于地质开采条件变化较大，即使在同一个地方
也没有一种独特新技术能够综合解决各种类型的煤矿井下火灾。当选择适用的技术时要涉及到几
方面因素，其中包括了煤火的状态和范围、火区现场的具体条件、可用的资源和经济性。有些技
术用于大范围密闭火时比较有效，还有一些其他技术对在废弃矿、破裂的煤柱、废弃的矿井巷道
等新发现的局部火灾还是有效的。本文目的是阐明在密闭采矿区中利用惰性气体或添加吸热化合
物的惰性气体，使用不同的隔离方法控制和预防煤火，即：（1）N2，（2）CO2，（3）N2和CO2
（比例：60:40），（4）水雾、化学雾、氮雾、雾凝胶、泡沫、氮气泡沫等，（5）硅胶、密封
胶，或其他密封材料。本文通过三个案例研究，旨在说明在实践中各种技术的选择标准、操作方
式、适用范围以及可行性和局限性。
554
1 Introduction
The first reported case of coal fires in Indian coal mines occurred in 1865; only a
few years after large-scale coal mining had begun in the country. In this early period,
traditional methods such as isolation, trench cutting, and water injection were used in
fire fighting. Systematic research on detection, control, and prevention of coal fires was
instigated in the early 1960s by the Central Mining Research Institute (CMRI; now
joined with the Central Fuel Research Institute (CFRI) to the Central Institute of Mining
and Fuel Research (CIMFR)) in Dhanbad, Jharkhand. The gist of their work amounts
to the following: the earlier a fire is attended to, the easier it can be controlled and
extinguished.
Board and pillar mining with a generally low level of recovery (60 to 80 %) is predominant
in India (> 80 %). During both development and depillaring (i.e. the second phase of
mining where the coal pillars needed for stabilization are extracted by caving or stowing),
20 to 40 % of coal remains in the depillared panel as stooks, crushed coal on the mine
floor, or hard coal in the mine roof and floor – it is these remaining deposits that represent
a most serious fire hazard. The rise in temperature in abandoned panels, in pillars left
between adjacent panels, or in gobs may be due either to shortcomings in the mechanized
mining method applied or to the properties of the coal, which may be particularly prone to
spontaneous heating and combustion.
(1) Coal + O2 = δH (93.7 kcal mol-1 O2 ) + Combustible products
The heat generated during oxidation of coal varies between 0.3 and 25 J mol -1 O 2,
depending on the state of the coal, i.e. whether it is weathered or has only recently
been exposed to air (Banerjee 1985). Under favorable geo-mining conditions, δH
tilts the balance of heat generation and dissipation and ultimately leads to self-
ignition in vulnerable sites. In order to design optimal methods of fire fighting and
control, it is important to assess the fire risk potential in a mine panel, delineate the
fire area, and determine its intensity and extent. Temperature gauges such as infrared
scanners or sensors are often unable to trace the thermal profile inside vast gobs, and
it is similarly difficult to drill an adequate number of boreholes for sampling into
the goafs of multi-seam workings. Examining fire indices based on compositional
analysis of indicator gases is the only means of determining coal temperature in
non- or barely accessible locations. However, there exists to date no single index
that could be universally employed to make accurate predictions on the status of a
555
fire. Instead, combinations of fire indices are generally used for interpretation and
for selection of appropriate fire control and prevention techniques. It is commonly
held among mine managers and researchers that this selection is very much site-
and condition-specific. The techniques used today are aimed at minimizing heat
generation or increasing its dissipation either by treatment or protection. These
techniques are:
• Use of water with or without additives in the form of fog or mist,

• Foam (high expansion or solid),
• Infusion of silica gel into coal pillars,
• Fire-protective sealant coating,
• Isolation by stoppings and inert gas infusion (carbon dioxide (CO2) and nitrogen (N2)) to
create an extinctive environment behind the seal.
Their merits and demerits of these techniques as from experience with fire fighting
campaigns in Indian coal mines are given in Table 1.
Two research projects were launched by the Indian Ministry of Coal to examine the scope
of inert gases for use in control and prevention of coal fires. Their results have been tested
and standardized in several underground coal mines, including Bagdigi (1999), Tata Sijua
(2000), and Haripur (2005) Collieries.
This paper describes the first-time application of fog, chemical fog, fog gel, and nitrogen
fog gel to decrease coal fire intensity in the above coal mines. Optimal compositions for
stable gels lasting in a goaf for at least 12 to 15 months were determined in laboratory
tests and the gels brought to field testing in the GDK-9 Incline Mine of Singareni Colliery
Company Limited. Nitrogen fog gel was applied to form an impervious coating on the coal
surface in unapproachable areas (goafs) where heating was expected to occur. To delay the
vaporization of water from the gel, highly hygroscopic salt and hydrated calcium chloride
were added to the solution.
556
Table 1: Selection of fire fighting techniques and their merits and demerits as from
experience in Indian coal mines
Methodology Selection Merits Demerits Remarks
• Colliery, Year Criteria
CO2 (l/g) Intensive fire in Minimizes air leakage Absorbed by CO2 is 1.5 times
Solid ice limited area from seam bottom. surrounding bedrock. heavier than O2
and thus reduces
• Sudamdih Seat of fire in mine Minimizes explosion risk. Poisonous air leakage
Colliery, 1977 floor
• Bulan Barari Endothermic reduction Once infused, difficult
Colliery, 1978 High air leakage to flush out.
• Haripur Colliery, Distance from
2005 working panel
N2 (l/g) Superficial/newly Mother foam N2 does not flow easily Graham’s Ratio is
Nitrogen foam developed fire in encapsulates coal and to fire source. not much useful
Hard Foam extensive area halts oxidation. in assessing the
Very poor heat status of a mine
• New-Kenda, 1993 High air leakage Isotropic expansion extraction fire during N2
• Moonidih infusion.
Colliery, 1994 Easy distribution
• Bagdigi Colliery, throughout goaf
1999 N2 gas can be easily
infused from single
points.
Creates positive pressure
inside the goaf, thereby
reducing air leakage from
the outside.
Non-poisonous and cost-
effective
Gel infusion Fractured pillar Prevents air leakage. Bitumen emulsion Both gel infusion
Bitumen coating to be used on cool and bitumen
Surface blanketing Cracks in mine roof Halts fire progression. surfaces only, sealing were
and floor otherwise adverse developed by
• Jitpur Colliery, Seals fractured pillars.
Surface subsidence effects. CMRI and used
1992 on a number of
• Bankola Colliery, High-pressure mine fires.
1995 difference in sealed-
• Madhushudanpur, off area
1996
CO2 + N2 infusion Fire in intervening Halts air leakage from Possibilities of CO Pressure
Pressure balancing strata seam bottom. formation balancing is
Bitumen coating difficult in highly
Water jetting Fractured parting Cools down rapidly. fractured areas.
• Jitpur Colliery, Pressure difference Seals fractures in mine
1992 across stoppings roof and floor.
• Tata Sijua Superficial fire in
Colliery, 1998 extensive area
Chemical fog Fire in working panel Chemical fog and fog gel Installation of PSA Used first by
Fog gel retain moisture. (Pressure Swing CMRI to combat
Nitrogen fog Fire in adjacent Adsorption) for coal fires /increase
Nitrogen fog gel sealed-off goaf Rapid cooling nitrogen fog generation incubation period
• GDK-9 Incline Fire in overlying Fog gel inertizes goaf and at colliery level is in GDK-9.
worked-out seam cools surrounding area. difficult.
Mine, 1999-2001
557
2 Designing Optimal Fire Fighting Measures

Once a fire ignites in an underground coal mine, huge resources are required for control
and extinction. Unnecessarily high costs may be incurred, if adequate fire fighting
techniques fail to be chosen. Appropriate selection depends on good analysis and
interpretation of the following parameters (Tripathi & Sen 1996a,b):
1) Assessment of fire intensity and extent, i.e.:

• Superficial fire in limited area,
• Superficial fire in extensive area,
• Intensive fire in limited area,
• Intensive fire in extensive area.
2) Identification of leakage pathways by thermo-compositional analysis or sulfur
hexafluoride gas (SF6).
3) Assessment of direction of fire progress.
4) Thermal infrared surveys, surface borehole measurements, or thermo-compositional
analysis of gases in sealed-off mine areas.
5) Visual inspection of:
• Cleats, micro- and macro-cracks and fissures;
• Side spilling, weathering, coal left in mine roof and floor;
• Surface subsidence.
6) Ventilation surveys of:
• Pressure difference of gas in- and outflow,
• Pressure difference in- and outside sealed-off area.
7) Open fires in developed pillars/galleries:
• Sealed-off fire in isolated goaf,
• Opencast mining of underground developed pillars in multi-seams workings.
3 Oxygen Concentration and Fire Status
The most important function of inert gas infusion in sealed-off fire areas is O2 depletion
and heat extraction. Experiences show that (Chakrabarty 1995):
1) Flaming combustion ceases when O2 concentration falls below 12 %.

2) Combustion ceases when O2 concentration falls below 5 %.
3) Combustion is completely arrested when O2 concentration falls below 3 %.
558
4 Oxygen Dilution by Inert Gas Infusion
The following formula can predict O2 concentration during inertization in a sealed-off fire
area after a time t:
(2) t = V / q ln C0 / Ct,
where V is the sealed-off volume in m3, q the inert gas flow rate in this area in m3 min-1;
C0 the initial O2 concentration at t = 0, and Ct the O2 concentration after the time t.
Combustion can be stopped by lowering the O2 concentration down to about 5 to 6 %,
i.e. Q/V = 1. In a perfectly sealed-off fire area (no air leakage), the quantity of inert gas Q
necessary to reduce O2 concentration from the atmospheric 21 to 2 % is Q/V = 2.3. The
ratio must be kept considerably higher than 2.3 in real conditions as the fire will continue
even at as low an O2 concentration as 1.31 %.
In case of an extensive fire, large quantities of inert gas are required to penetrate thermal
barriers and reach the fire source. Under such circumstances, we generally prefer nitrogen
foam, fog, or gel.
5 Mitigation of Underground Coal Fires – Case Studies
5.1 Bagdigi Colliery (Bharat Coking Coal Limited, BCCL), 1999
Fire problems in shaft pillars at 10 pit in XI/XII seam of Bagdigi Colliery BCCL (Figure 1)
were reported to CMRI in 1999. The fires had to be brought under immediate control so
that safe extraction of coal below XI/XII could continue (CMRI 1999).
559
Figure 1: Plan of XI/XII seam and cross-section of 10 pit of Bagdigi Colliery BCCL
In our preliminary investigations, we found that:
• Carbon monoxide (CO) varied from 0.344 to 0.7 % at inset stoppings nos. 1 and 2
(thermo-compositional analysis).
• The stoppings were warm or hot (c. 50 to 60 °C), and smoke was being emitted through
cracks and fissures; the temperature 10 m inside the stoppings was above 140 °C.
• The temperature inside the bypass gallery was above 200 °C.
• The overlying coal seams XIIIB, XIIIA, and XIII (including XI/XII at 10 pit) were
exhausted and engulfed in fire up to the surface.
After data interpretation, the following measures were taken to control the fire and save the
shaft pillars:
1) The surface above the fire area was blanketed with clay/soil and/or soil/bentonite (1:5),
and leveled with a bulldozer to suppress air entry.
2) Existing stoppings in XI/XII seam were repaired and new stopping were constructed at
both sides.
560
3) Silica gel (with rheological properties) was infused into the shaft pillars to reduce air
entry through micro- and macro-cracks. A layer of fire-protective coating was sprayed
onto the vulnerable areas of the shaft.
4) The bypass gallery in XI/XII seam (around the shaft) was filled with bentonite slurry
admixed with fire-retardant chemicals, fly ash, and lime. Nitrogen foam was flushed
through boreholes at the surface.
5) Nitrogen foam (70,000 l liquid N2 and 250 l foam solution) was flushed into the sealed-
off fire area through sampling points in stoppings nos. 1 and 2.
6) Thermo-compositional monitoring was carried out for six consecutive months to
improve work safety below XI/XII seam.
Gel infusion into shaft pillars

To control air leakage through micro- and macro-cracks, silica gel – a medium with
rheological properties and controllable setting time – was infused into the coal pillars, the
shaft, and into roof and floor cavities. The gel blocked air circulation permanently.
The application required in a first step that holes of 40 mm in diameter be drilled at
different angles into the pillars and the mine roof and floor, up to a depth of 0.5 to 2 m
and at 3 to 4 m distance. After drilling, the silica gel was infused (using a CMRI-designed
infusion gun) at a pressure of 3 to 4 kg cm-2. The pressure mainly depended on fractures
in surrounding strata and was adjusted if needed. In order for the gel not to harden during
operation or ooze out after infusion, its setting time was maintained at 45 minutes, through
a constant pH of 9. The silica gel consisted of a 10 % solution of sodium silicate, admixed
with 25 % of a 10 % diammonium phosphate (DAP) solution. DAP is a fire-retardant.
Nitrogen foam infusion

The large number of cracks, fractures, and interconnections to the surface justified infusion
of nitrogen foam into the bypass gallery. The breaking foam bubbles encapsulated the
burning coal and reduced the risk of further heating.
Results and discussion

Through the above operations, the fire in the shaft pillars at 10 pit in XI/XII seam was contained
successfully. The temperature inside the bypass gallery dropped from 200 to 85 °C within
15 days; inside the stoppings, it fell from 140 to 40 °C. CO and CO2 concentrations came
down to 0.01 and 2.5 in 10 days. However, the continued presence of O2 (13.31 %) indicated
substantial air leakage from the surface. Improvements in surface blanketing/sealing were
therefore suggested to the mine management.
561
5.2 Haripur Colliery (Eastern Coalfields Limited, ECL), 2003
A spontaneous fire in stowed goaf panels D and F at the intake side between inclines no. 1
and 2 of Haripur Colliery ECL was reported to CMRI in 2003. CMRI was asked to ascertain
the state and extent of the fire and give scientific advice on its handling (Figure 2).
Figure 2: Plan of coal seam R-V in panels D and F of Haripur Colliery ECL
Preliminary investigations revealed that the coal seam R-V in panels D and F was 6.25 m
in thickness and depillared along the bottom, leaving a 1 m roof layer of coal. First
sampling was carried out on 9 January 2003 at stoppings nos. D-5, D-12, and F-5. The gas
concentrations measured are given in Table 2 (CMRI 2005).
Table 2: Gas concentrations as sampled on 9 January 2003 from panels D and F in coal
seam R-V of Haripur Colliery ECL
Stopping Sampling O2 CO2 CO CH4 N2 CO/O2
Time (%) (%) (%) (%) (%)
D-5 1.05 p.m. 12.60 5.65 0.0000 0.00 81.73 0.000
D-12 1.00 p.m. 10.20 7.83 0.0322 0.12 81.93 0.279
F-5 12.45 p.m. 17.50 1.32 0.0000 0.00 81.18 0.000
Thermo-compositional analysis confirmed an active fire in panel D and enormous air

leakage in panel F. With the existing sampling points, it was not possible to locate either
fire source or leakage pathways. Therefore, the supplementary sampling points D-7, D-14,
and D-18, as well as F-13 and F-22 were drilled into existing stoppings.
Gas samples were collected from all stoppings. Laboratory analysis confirmed that the
fire source must be located somewhere near stoppings D-12 and D-14. O2 concentration
562
in panel F was found to be at nearly 19 %. In immediate reaction, all stoppings, and
mine roof and floor areas were strengthened and plastered. O 2 concentration in panel
F, in consequence, dropped considerably. Since the active fire in panel D persisted in a
limited area on the floor of the stowed goaf, infusions of CO2 were administered at certain
intervals. Almost 7,000 kg of liquid CO2 (solid ice) were infused through stoppings D-14
and D-18 during a period from August 2003 to June 2004.
Results and discussion

Through the measures described above, CMRI was able to bring the fire in goaf panels D
and F of Haripur Colliery ECL under control. CO concentration was reduced to 0.001 %;
O2 concentration dropped to 4 to 5 % in panel D and 12 to 14 % in panel F. For further
reduction of O2 concentration in panel F, plastering of roof and floor spaces and stoppings
was continued. As panels exist at the mouth of intakes nos.1 and 2, and the possibility of
leakage cannot be ruled out completely, continuous thermo-compositional monitoring at
these spots is imperative.
5.3 GDK-9 Incline Mine (Singareni Colliery Company Limited, SCCL),

2001
During extraction of developed pillars in seam IV of GDK-9 Incline Mine SCCL, incidents
of spontaneous heating were frequent. Seams III and IV in GDK-9 are 10.5 m and 2.6 to
4 m in thickness and occur contiguously with a stone parting of 3 to 7 m. Seam IV was
developed on board and pillars, below a caved gob of seam III. During a main collapse,
the parting between the two seams caved in and weathered coal from seam III fell into
seam IV, causing spontaneous combustion there. In panel 6F(ii)a, this roof fall occurred on
11 December 1998, and signs of heating were noticed as early as 12 January 1999; a full-
blown fire developed within a month. CMRI was contacted to design the guidelines and
methodologies necessary for both suppression and prevention of spontaneous heating in
the seams.
After preliminary investigations, the following measures were taken (CMRI 2000):
1) The roof fall of weathered coal from seam III into seam IV, a prime cause of heating,
was delayed by leaving small ribs in coal pillars at the mouth of the mine panel. The
size, design, and strength of these ribs depended on site conditions.
563
2) Depillaring at shallow depth was halted and supplanted by pressure balancing across
the isolation stoppings of the sealed-off area; overburden was dumped onto the panels’
depillared areas to prevent air leakage through cracks.
3) Water fog, chemical fog, nitrogen fog, and fog gel were applied to decrease fire
intensity.
4) Regular thermo-compositional monitoring was instituted for early detection of
spontaneous heating.
Application of water fog, chemical fog, nitrogen fog, and fog gel
After the roof fall in panel 6F(ii)a, a surge in CO concentration was detected on 12 January
2000 (3 % within two days). In response, the panel was shielded and atomized silica gel
(nitrogen fog gel) was infused through boreholes with a fog fighter at high pressure (3
to 7 kg m-2). The silica gel consisted of a 10 % solution of sodium silicate, admixed with
25 % of a 10 % DAP solution (N2 21 %; phosphate 53 %; (NH4)2 HPO4). This combination
was chosen with regard to its setting time of 45 minutes (at a pH of 9), which would keep
the gel from hardening during infusion. Following application, CO concentration dropped
from 2.6 to 0.0017 %, and conditions were back to normal within 2 to 3 days (Figure 3).
Figure 3: Status of fire as per fire indices in panel 6F(ii)a of GDK-9 Incline Mine SCCL
before and after fire fighting operations
JTR = John Tricket Ratio
564
5.4 Extraction of panels adjoining 6F(ii)a
In panels 6F(ii)b, as well as 6F(iii)a and b, chemical fog and fog gel were sprayed in
goaved out and remote areas to increase coal moisture content and delay self-heating.
Sodium chloride is a highly hygroscopic salt, which adsorbs water vapor from air and thus
binds moisture. Among the chemicals for spraying, dilute solutions of DAP (0.01 % w/v)
and ammonium chloride (NH4Cl; 0.025 % w/v) were found most suitable. The temperature
was monitored in thermal infrared surveys; it remained below ambient temperature (24 °C)
during extraction.
A variety of aspects were considered in the selection and application of chemicals for
fire fighting in the three case studies of Bagdigi, Haripur, and Singareni Collieries. They
included: (i) properties under various geo-mining conditions, i.e. coal left in mine roof
and floor areas, or in developed or depillared panels; (ii) instances of corrosion, abrasion,
or toxicity; (iii) cost effectiveness; as well as (iv) mixing, pumping, and storing qualities.
Silica gel proved to be a most effective fire-retardant. Its advantages are as follows:
1) Due to very low viscosity and controllable setting time, silicate fog gel (sodium
silicate + DAP) can be easily sprayed onto surfaces and into cavities, i.e. overlying
goafs, voids between two fallen stone pieces inside goafs, and inaccessible areas that
can only be attended to from distances of 25 to 30 m.
2) Silicate fog gel forms an impervious coating on coal surface/broken coal that lasts for at
least 12 to 15 months. Hence, a substantial increase in incubation periods.
3) Silicate fog gel can hold large quantities of water and has proven an effective inhibitor
of coal oxidation at temperatures of up to 150 °C.
4) In laboratory experiments, CO concentration was reduced from 0.33 to 0.16 % when
mixed with a sodium silicate solution (2 % w/v).
5) Loose coal fallen from the seam roof or overlying seams after collapse should be treated
simultaneously with endothermic chemicals to avoid heating. These chemicals must be
cost-effective, easily available, and miner-friendly.
6) Silicate fog gel and nitrogen fog gel can be applied to vulnerable spots from all
directions and at safe distance.
565
Acknowledgements
The author is thankful to Dr. A. Sinha, Acting Director of CIMFR, for his kind permission
and partial financial support to present this paper at the International Conference on Coal
Fire Research.
References
Banerjee, S.C., 1985. Spontaneous combustion of coal and mine fires. New Delhi, Oxford, Rotterdam:
IBH Publishing.
CMRI, 2005. Scientific investigation and advice on dealing with fire in goaved out panels D and F lying
between nos. 1 and 2 incline of Haripur Colliery of Kenda area, ECL. Project report GC/
MS/102/2002-2003. Central Mining Research Institute (CMRI), Dhanbad, Jharkhand, India.
CMRI, 2000. Design of economic methods for fire-safe extraction of coal pillars at GDK-9 incline mine of
SCCL and advice thereof. Project report MS/GC/68/98-99. Central Mining Research Institute
CMRI, 1999. Dealing with shaft pillar fire of 10 pit in XI/XII seam at Bagdigi Colliery BCCL and advice
thereof. Project report GC/97/97. Central Mining Research Institute (CRMI), Dhanbad,
Jharkhand, India.
Tripathi, D.D. and Sen, S.K., 1996a. Monitoring – First criteria for preventing and controlling of mine
fires. Journal of Mine Metals and Fuels 231: 199-202.
Tripathi, D.D. and Sen, S.K., 1996b. Prediction of fire status – A new approach to an old problem. Coal
International 244(3): 118-120.
566
Control of Shallow-Depth Underground
Coal Fires in Jharia Coalfield
D.D. Tripathi
Mine Fire Laboratory, Central Institute of Mining and Fuel Research (CIMFR),
Dhanbad, Jharkhand, India
Abstract
With nine large opencast coal blocks, Jharia Coalfield is the only source of prime coking coal in India.
Most of the reserves harbor extensive coal fires. The fires are generally combated in a haphazard manner
with high-pressure water injections, lacking due consideration of fire status, necessary period of cooling,
or method of watering (i.e. whether fog, mist, or spray; or with or without endothermic chemicals).
Undoubtedly, water is the most widely used means of fire fighting in India and throughout the world.
Small droplets of water, though, either in the form of fog or mist, do not have a fire fighting impact as
such – they need to be applied at or reach to the seat of the fire. In most cases, flame- and heat-induced
upward flow resists this motion. Experiences show that more than 80 % of the water applied in fire
fighting does not reach its target and is hence useless. Realistic and convenient techniques for effective
fire fighting in coalmines have been formulated by the Central Mining Research Institute (CMRI) with
a view to: (i) control and containment of intensive or blazing coal fires, and (ii) rapid cooling of heated
overlying rock strata and overburden debris.
A variety of fire-retardant and endothermic chemicals were used to improve the extinguishing capability
of water mist. The addition of wetting agents reduces surface tension and facilitates the formation of
small drops necessary for rapid heat absorption. Highly hygroscopic salt and hydrated calcium chloride
were added to delay vaporization. Two successful case studies are discussed in this paper, one at Shatabdi
Opencast Mine and the other at South Tisra Colliery.
摘要
切里亚(Jharia)煤田是印度唯一生产主焦煤的矿区，该煤田有九个大型露天开采的煤柱。遗憾的是
大部分露天开采资源中都广泛发生煤火。通常都采用高压注水方法随意灭火，而没有考虑煤火的
状态、必要的冷却周期和注水的方法，就是说：是雾状、薄雾状还是喷雾状注水、有无吸热化学
材料等。
567
无疑，水是在印度及世界各地最常用的灭火手段，但是雾状或薄雾状的小水珠同样都没有足够的
灭火能力，它们只有达到发生煤火的位置才能起作用，但在大多数情况下，煤火焰和产生的向上
热流阻止了水流向火源流动。经验表明：超过80%的灭火用水都没有达到预期的位置，因此都是
没有效果的。所以，本文基（1）抑制火焰、熄灭强燃烧火和（2）快速冷却受热的顶板岩层和用
碎屑物覆盖的观点，详细叙述了煤矿中实用和便捷的有效灭火技术。
过去常用各种阻燃及吸热化学材料来提高水雾的灭火能力。将润湿剂添加到水中降低了水表面张
力，有利于形成可迅速吸热的小水珠。而将高吸湿盐和含水氯酸钙加入到水中可以延缓蒸发。本
文为描述灭火技术的应用讨论了两个成功案例，一个是在Shatabdi露天煤矿的案例，另一个是在
Tisra煤矿的案例。
1 Introduction
Coal fires have been burning in Jharia Coalfield (JCF), India for the last 90 years. The
earliest fires started in shallow-depth coal mines but have meanwhile spread to cover
9.6 km2 in altogether 41 mines, destroying every year 3 Mt of superior grade coal (Tripathi
& Jayanthu 1996). Efforts to control and contain these fires have been underway ever
since their emergence. In 1988, an “Integrated Fire Control Plan of Jharia Coalfield” was
formulated with the following two objectives:
1) Assessment of the state and extent of the fires to stop further propagation.
2) Control and containment for resource conservation and environmental protection.
One of the main technologies suggested for surface fire control was sealing with proper
compaction, which prevents oxygen access and thus limits fire progress. After sealing,
coal fire-heated overlying rock strata need considerable time to cool down unless cooling
is accelerated artificially. Most fires occur in outcrops at shallow depth, baring access to
areas that are most suitable for opencast mining. Unless steps are taken for rapid cooling,
production targets cannot be met. In view of these obvious economic and environmental
imperatives, experiments for: (i) control and containment of intensive or blazing coal
fires, and (ii) rapid cooling of coal fire-heated overlying rock strata and overburden debris
were carried out in Shatabdi Opencast Mine and South Tisra Colliery, Bharat Coking Coal
Limited (BCCL).
2 Geo-Mining Characterization of Jharia Coalfield
JCF lies within 230°39’to 48’ N latitude and 860°11’to 27’ E longitude; and belongs to
the Raniganj (58 km2) and Barakar (210 km2) Coal Measures. Its total coal reserves are
568
estimated at 5,300 Mt. JCF has 18 regular coal seams (XVIII to I from top to bottom) and
46 coal horizons. Many of these consist of discontinuous splits, namely top and bottom,
upper and lower, as well as local seams (0, A, B, -- --; etc.), with a varying thickness
of 0.5 to 55 m, and a gradient (slope angle) of 5 to 40° up to a depth of 1200 m. JCF is
marked by geological disruptions such as dykes, sills, igneous intrusions, folds, and more
than 250 faults. The thermal gradient in the area varies from 35 to 65 K km-1, which has
contributed to the conversion of large quantities of prime coking coal to Jawa. The upper
coal seams XVIII to X contain prime coking coal; the lower seams coal of inferior quality.
The overburden is sandy and porous.
3 Fire Control Measures in Jharia Coalfield
After nationalization of the Indian coal industry, five shallow-depth underground coal
fires were extinguished in Jharia’s Nudkhurki (seam X), Kooridih (seam X), East Katras
(seam XIV), Jogidih (seam X), and West Modidih (seam IX/X) Collieries, saving about
95 Mt of prime coking coal for extraction. These first successes spawned 22 national fire
projects with a total budget of INR 114.57 billion, targeting another 59 fires in JCF. The
technologies employed in these projects were:
• Trench cutting and filling with incombustible materials;

• Surface sealing with soil/mitti (soil with high water-retaining capacity);
• Creation of water pools on the surface, water circulation under pressure, flooding of fire areas;
• Inert gas and/or foam infusion;
• Cementing and grouting, flushing with bentonite and/or fly ash;
• Filling of developed galleries with sand and/or incombustible materials; and
• Excavation.
Surface sealing and compaction

One of the main technologies suggested for containment of surface fires in JCF is sealing
and compaction. Surface sealing carried out with sand or soil, however, has often led to
disappointing results, owing mainly to new cracks in the surface layer that allow escape
of hot gases and percolation of rainwater. Proper maintenance, e.g. regular leveling and
compacting with bulldozers, requires a lot of work and investment, and still does not
guarantee satisfactory results, particularly at the edges and on undulating land. As a frame
of reference for evaluation of surface sealing and compaction, air permeability at different
degrees of compaction and porosity is given in Table 1.
569
Table 1: Air permeability (superficial velocity) through topsoil at different degrees of

compaction and porosity
Degree of Compaction Porosity Permeability / Superficial Velocity
(dimensionless) (m sec-1)
Loosely compacted 0.3 10-4-10-3
Slightly compacted 0.2 10-5-10-4
Densely compacted 0.1 10-6-10-5
Highly compacted 0.05 10-7-∞
Grain size distribution and water content in soil are two important factors in reducing air
permeation. In field tests (Banerjee 1990), it was found that a 10 cm layer of bentonite can
reduce air permeation by up to 90 %.
4 Control of Blazing Fires in Shatabdi Opencast Mine, 2005
Blazing fires in developed galleries and concomitant hot overburden in Shatabdi Opencast
Mine were reported to the Central Mining Research Institute (CMRI; now joined with
the Central Fuel Research Institute (CFRI) to the Central Institute of Mining and Fuel
Research (CIMFR)) in 2004 for immediate control and advice on production improvements
in the fire area.
Nearly, 87 % of the coal in Shatabdi Opencast Mine is held by the composite seam V/VI/VII,
with a varying thickness of 14.89 to 22.06 m. Approximately 65 % of the seam’s mineable
parts has been developed on pillars; the reaming 35 % is virgin in deep side (Tripathi et al.
2005).
During extraction of underground pillars in opencast mining, a blazing fire was discovered
at the gallery mouth. The fire had been kindling before in loose coal and only flared up
as oxygen was made available through the opening of the gallery. The following work
program was devised by CMRI to control the fire.
Preliminary measurements and experiments

To evaluate the intensity and extent of the fire, surface temperature measurements at
different locations were performed with an infrared thermometer (Table 2).
570
Table 2: Temperatures at different locations of seam V/VI/VII, Shatabdi Opencast Mine

Location Max. Temperature (ºC)
Burning loose overburden 410
Overburden just above gallery mouth 450
Flames at gallery mouth 540
Borehole at interface between coal and overburden at 6 m depth 145
Borehole in virgin (not burning) coal at 7 to 8 m depth 91
Coal at 2 to 3 m from gallery mouth 141
Loose overburden near top bench 143
Four sets of experiments were carried out to determine which chemicals ought to be used
for fire quenching at the gallery mouth and extraction of heat from the overburden debris.
Water pipelines were laid and sprayers installed so as to generate the required pressure
(> 7 kg cm-2) for spraying of hot surfaces. These were regularly monitored with an infrared
heat spy.
Selection of chemicals for fire fighting

A variety of aspects were involved in the (lab-based) selection of chemicals for effective
and economical fire fighting, namely: (i) properties under various geo-mining conditions;
(ii) instances of corrosion, abrasion, or toxicity; (iii) cost effectiveness; as well as (iv)
mixing, pumping, and storing qualities.
Some chemicals were found more effective in flame retarding, while others proved useful
for rapid cooling of heated rock strata. The chemicals ultimately used in Shatabdi Opencast
Mine were phosphoric acid, boric acid, sodium chloride, sodium silicate, diammonium
phosphate (DAP), and silica gel in varying proportions.
Fire fighting – Application technologies

For suppression of flaming combustion at the mouth of the opencast gallery, a water admix
with phosphoric acid (0.05 to 0.10 %) was sprayed onto the flames. They died down within
5 to 7 min.
In a next step, DAP (0.03 to 0.1 %) was sprayed with a fog fighter. After spraying,
temperatures in the debris dropped from 450 to 60 °C and below within 10 to 12 h; in the
overlying rock strata, they fell from 250 to 80 °C within fifteen to twenty days. Silica gel, a
mixture of sodium silicate and DAP, was applied afterwards to form an impervious coating
571
on the coal surface and thus reduce the risk of re-ignition. Temperatures in the entire fire
area were constantly below 100ºC so that water gas formation could be ruled out during
fire fighting operations.
5 Rapid cooling of strata in South Tisra Colliery BCCL, 1997
A fire that had originally started in X seam quarry of North Tisra Colliery propagated to
IX/X seam quarry of South Tisra Colliery in 1973. IX/X seam contains several goafs.
Its underlying seams VIII A, VIII, VII, and V/VI, are also goaved out and isolated. At a
certain point, the entire goaf section from IX/X down to V/VI was on fire, covering a total
area of 237.000 m2. For an earliest possible extraction of the burning coal, a project for
rapid cooling of coal fire-heated strata was launched on trial basis. The following three
experimental schemes were proposed for cooling (Tripathi et al. 1997):
• Cooling by water pools on the surface;

• Cooling by flushing with water plus additives, with inert gas or foam;
• Cooling by re-circulation of chilled goaf gases.
Preliminary works at experimental sites

In preparation of the cooling experiments, surface thermal profiles of the fire area were
acquired using an infrared heat spy. The temperature of the overlying rock strata varied
from 80 to 350 °C, and flames were visible at the quarry edges. In a second step, the
surface area was covered in three layers of topsoil, each 0.5 m in thickness, and dozed with
a heavy compactor. Finally, 43 boreholes of varying depth were drilled for:
• Thermal data logging,

• Thermo-compositional monitoring,
• Formation of two water ponds of different dimensions over the fire area and their filling
with water.
• Filling of voids and cracks in the heated overlying rock strata and flushing of cooling
materials.
Cooling of hot strata by water percolation

Two ponds of 84 × 66 × 5 m and 54 × 70 × 5 m were dug into the dozed ground above
the fire area and filled with water up to a height of 3 to 4 m. A suitable gauge had been
572
installed in both ponds prior to filling to measure the water level at regular intervals.
Boreholes C1 to C5 were nearest to the water pools. Temperatures in these boreholes were
recorded every week for a total of two years. Another six boreholes, C6 to C11, were drilled
for oxygen logging at the opposite ends of the water pools – boreholes C6 and C7 down to
IX/X seam, and C10 and C11 down to V/VI seam. Logging data for different time intervals
are given in Figures 1 and 2.
Figure 1: Temperature in boreholes C1 to C5 after sealing and under water percolation

from pools on the surface
Figure 2: Oxygen concentration in boreholes C6 to C11 after sealing and under water
percolation from pools on the surface
573
Permeability of soil and rapid cooling of strata
The rate of water percolation depends on soil permeability, i.e. the number and size of
Permeability of soil and rapid cooling of strata
Theconnected voidspercolation
rate of water that can give passageontosoil
depends water. The rate ofi.e.
permeability, water
thepercolation
number andin size
micro-
scopically disordered
of connected porous
voids that can givemedia in one-dimensional
passage steady-state
to water. The rate of water conditions
percolationcan
in be
microscopically disordered
calculated using Darcy’sporous
law: media in one-dimensional steady-state conditions can
be calculated using Darcy’s law:
dH
(1)
(1) q K ,
dx
where q is the flux density of water passing through the soil, K is the saturated hydraulic
where q isHthe
conductivity, fluxtotal
is the density of water
potential, andpassing
x is the through
position the soil, K is the saturated hydraulic
coordinate.
conductivity, H is the total potential, and x is the position coordinate.
Sometimes, water may percolate through cracks and fissures; under such conditions
percolation velocity becomes much higher than in discharge through a steady medium.
Sometimes, water may percolate through cracks and fissures; under such conditions
Laboratory tests showed that gravel (coarse to fine) is most permeable and exhibits
percolation
turbulent velocity clay
flow; whereas becomes much
and silt higher
always thanlaminar
exhibit in discharge through
flow. For a steadyofmedium
rapid cooling coal
only. Laboratory tests showed that gravel (coarse
fire-heated overlying rock strata it is thus advisable to: to fine) is most permeable and exhib-
its turbulent flow; whereas clay and silt always exhibit laminar flow. For rapid cooling
• Use gravel
of coal instead of
fire-heated clay for water
overlying pool beddings.
rock strata In a coarse
it is thus advisable to:gravel bed, water flow is
directly proportional to the square root of the mean pore size.
• Lower the percentage of fine particles so that more water can percolate down toward the
x Use gravel instead of clay for water pool beddings. In a coarse gravel bed, water
heated strata.
flow is directly proportional to the square root of the mean pore size.
• Expect cooling effects beyond the pool’s boundaries – the coefficient of soil permeability
Lower
x many
is thehigher
times percentage
when of finepercolates
water particles in
so parallel
that more water
to soil can percolate
stratification, not down toward
vertically.
Thusthe
cooling
heatedis strata.
more pronounced around the pool than directly below its bed.
• Perform thermal
x Expect monitoring
cooling effects through
beyond boreholes
the pool’satboundaries
a certain distance from the pool’s
– the coefficient edge.
of soil perme-
ability is many times higher when water percolates in parallel to soil stratification,
Cooling by flushing of water, foam, and chemicals
not vertically. Thus cooling is more pronounced around the pool than directly below
Flushing in coal fire fighting mostly refers to borehole injection of water admixed with
its bed.
endothermic chemicals or inert materials such as sand, clay, bentonite, or grout. Flushing
x Perform
of inert materialsthermal monitoring
seals the throughcoal
voids in burning boreholes at a certain
and bedrock. At thedistance from
same time, thethe pool’s
water
mix extinguishes
edge. the fire and facilitates cooling. The following materials were injected
through the 16 boreholes in the fire area of South Tisra Colliery BCCL:
• Water at a pressure of 2 to 3 kg cm-2,

• Water infused with carbon dioxide at a pressure of 7 kg cm-2,
• Water and bentonite mixed in a ratio of 1:3,
• 2 % DAP solution , and
• Foam with or without nitrogen.
37
574
After infusion of either pure water or water admixed with chemicals, temperatures were
measured in 10 nearby boreholes at intervals of 7 days. They dropped considerably from
350 to 78 °C in borehole B1 to B5 at a distance of 5 to 10 m from the point of injection in
45 to 135 days. The technology of re-circulation of chilled goaf gases is yet to be applied
(Tripathi et al. 1997).
Acknowledgements
The author is thankful to Dr. A. Sinha, Acting Director of CIMFR for his kind permission
and partial financial support to present this paper at the International Conference on Coal
Fire Research.
References
Banerjee, S.C., 1990. Assessment of status and control of underground mine fires. Project report
GAP-V/11. Central Mining Research Institute (CMRI), Dhanbad, Jharkhand, India.
Tripathi, D.D. and Jayanthu, S., 1996. Numerical techniques to improve fire control measures during
thick-seam mining. Journal of Mine Metals and Fuels 44(5): 163-166.
Tripathi, D.D. et al. 2005. Studying the probable causes for avoiding the occurrence of explosion during
extraction of developed galleries operating under fire by opencast at Shatabadi OCP, BCCL
and advice thereof. Project report GC/MS/102/2004. Central Mining Research Institute
Tripathi, D.D. et al. 1997. Rapid cooling of strata heated by fires at South Tisra Fire Area of BCCL.
Project report GAP/V38. Central Mining Research Institute (CMRI), Dhanbad, Jharkhand,
India.
575
Sustainable Control of Coal Fires
Through Systematic Decision Processes
and Cyclic Management
Uli Barth1, Birte Grashorn1, and Friedrich-Karl Bandelow2
1
Department of Safety Engineering Methods, University of Wuppertal, Wuppertal, Germany
2
Abstract
This paper presents a multi-layer model of systematic decision-making for sustainable control of coal
fires. The model is based in principle on a method for tackling large damage situations, which has been
used successfully for many years in German emergency organizations.
摘要
本论文介绍了用于煤火可持续管理系统决策的多层模型，该模型主要是基于一种经德国突发事件
处理机构多年来成功使用的方法，立足于解决大量煤火危害问题。
1 Introduction
Spontaneous coal seam fires have been common in northern China for several thousand
years (Xu et al. 2003). Today, about 700 fire zones are reported from seven autonomous
regions and provinces in China.
Coal fires are a multi-dimensional problem with a disastrous impact on the environment,
causing among other things:
• Loss of coal resources,

• Environmental hazards (greenhouse gas (GHG) emissions, soil and water pollution),
576
• Impaired health of local residents, and

• Safety risks in mining.
This list alone justifies that immediate action be taken to bring coal fires, in China and
worldwide, under sustainable control (Pennig 2000). However, as capital, human resources,
and equipment are limited, systematic procedures are required to realize this control in a
most effective and efficient way. Pertinent advice and experience, for example, may be
gleaned from the management levels of municipal and industrial fire fighting teams that,
under similar conditions of limited resources, have to control and mitigate fire disasters. In
contrast to what these teams are generally faced with, spontaneous coal seam fires tend to
burn for years. The life cycle of an uncontrolled coal seam fire spans from:
• A phase of heat accumulation inside the coal seam, to

• A point of self-ignition, to
• A phase of fire propagation of up to many years, and finally to
• A phase of exhaustion and self-extinction when all oxidizable material has been burned
or gas flow is insufficient.
The impact of coal fires can be reduced significantly with a suitable and efficient national
extinguishing strategy, a strategy that also encompasses local tactics.
2 A Macroscopic or National Perspective on Coal Fires
On a macroscopic or national level, coal fires (in general) are characterized by:
• Coal stratification;
• Coal fire status
(e.g. suspected – identified – quantified – controlled – extinguished – monitored);
• Affected settlements, industries, recreation areas;
• Climatic/weather conditions;
• Infrastructure (e.g. electricity, road and railway systems); and
• Water, soil, and other resources as needed for fire fighting.
577
3 A Microscopic or Local Perspective on Coal Fires

On a microscopic or local level, coal fires (i.e. one particular coal fire) are/is
characterized by:
• Coal stratification;
• Age and extent (e.g. temperature profile);
• Accessibility for exploration and fire fighting;
• Availability of water, soil, infrastructure in immediate vicinity (e.g. electricity, roads,
mine workings, settlements); and
• Climatic/weather conditions.
4 Sustainable Control Needs a Fundamental Strategy
The strategy that is chosen for an extinction campaign is generally determined by urgency
and importance (Ferch & Melioumis 2005). It may depend on:
• National objectives and concepts of coal conservation;

• National objectives and concepts of environmental protection (e.g. commitment to GHG
reductions)
• National funding and resources for fire fighting (e.g. personnel, infrastructure, vehicles,
water pipelines); and
• Computer-aided models and simulations of coal fire development from self-ignition to
extinction.
In order to devise an optimal fire fighting strategy, “aims of protection” have to be defined
in a first step. These may be:
• Directly affected coal seams;

• Neighboring seams and coalfields;
• GHG emissions into the atmosphere;
• Returns on investments (costs of fire fighting as compared to value of saved coal).
In a second step, it is imperative to decide whether the budget shall define the results, or
the results define the budget; i.e. whether there are fixed objectives for the mission that
will dictate the overall budget, or whether the resources available shall set the frame within
which to achieve best possible results.
578
In this context, the relation between the “damage (in terms of costs)” and the “effort to
minimize the damage (in terms of costs)” is of great importance. A comparison of the “loss
of coal“, the “control of coal fires,” and the “degree of control (DOC)” in a qualitative
ERSEC Ecological Book Series – 4
manner may shed light on this relation. The value “DOC” is hypothetical and can be
calculated as follows:
Real degree of control (oberservable)

(1)
(1) DOC u 100 (%) (comp. Figure 1)
Maximum degree of control (theoretical)
“Loss of coal” and “control of coal fires” range between two borderline cases, where the
“Loss
“degree of of coal” (DOC)
control and “control of coal
is either 0 %,fires”
i.e. range betweenoftwo
a maximum borderline
danger existscases,
due towhere the
a minimum
spent“degree of control
on safety; (DOC)
or 100 %, is aeither
i.e. 0 %, i.e.ofa maximum
maximum of dangerousness
safety is achieved exists dueoftocosts.
by a maximum a
minimum
The total costs spent
graphon(Tsafety;
; TLCor= 100
CL +%,CCi.e.
) isathe
maximum
sum of of safety
“loss is achieved
of coal by a“control
(CL)” and maxi- of
LC
mum(C
coal fires ofC)”.
costs. Theatotal
It has costs graph
minimum (TLC; point
at T. This TLC =indicates
CL + CC)the
is the
most sum of “loss ofapproach
economical coal
with (C
anL)”
optimal degreeofofcoal
and “control safety
fires(DOC
(CC)”.opt). Theapoint
It has T (TLCmin
minimum at T.) This
divides theindicates
point cost field
the into
an uneconomical
most economical unsafe rangewith
approach (DOC < DOC
an optimal opt ; left)ofand
degree an uneconomical
safety safe range
(DOCopt). The point T
(DOC > DOC ; right)
(TLCmin) divides
opt the (Hagenkötter
cost field into 1973).
an uneconomical unsafe range (DOC < DOCopt; left)
and an uneconomical safe range (DOC > DOCopt; right) (Hagenkötter 1973).
TLC
Control of coal fires CC
Expected cost due to
Total costs TLC

Loss of coal CL
P1 P2
CL CC
T
economical DOC safety

DOC1
approach
" opt approach
"DOC2
Figure 1:1:Degree
Figure Degree of
of control (DOC)model
control (DOC) model
Data
Data source:Compes
source: Compes(in
(in Hagenkötter
Hagenkötter 1973)
1973)
With this approach, it is possible to quantify and evaluate the expense and benefits of an
extinction campaign – an important factor for financiers.
579
With this approach, it is possible to quantify and evaluate the expense and benefits of an
extinction campaign – an important factor for financiers.
5 Effective and Efficient Fire Fighting Needs Optimal Tactics
There exist three fundamental tactics to control fire hazards:
• Defense (trenches or barriers to isolate a fire from unaffected coal),

• Rescue (expedient extraction of unaffected but threatened coal),
• Attack (active extinction or cooling of a fire).
Additionally, when a fire exceeds the organizational and/or technical scope of a mission:
• Retreat (monitoring until self-extinction)
may be used as a tactic too (Figure 2).
Figure 2: Fire fighting tactics
580
In other words, “to control a coal fire” does not necessarily and in all situations mean “to
extinguish a coal fire.” Best-practice tactics or a combination of them can be selected for
control on a case-to-case basis (Rodewald 1998; Zhang 2003). Results from computer-
aided fire simulations (as those developed in Phase A of the Sino-German Coal Fire
of Coal Fires in North China” (DLR et al. 2007)) are helpful and needed in this respect.
Also, it is conceivable that “extinction models” applicable to certain “fire zone models” be
developed (Samland 2004).
6 Systematic Decision Process
All conditions, i.e. all parameters that determine both the occurrence of a fire and its
management, are important for the systematic decision process. These can be for instance
geology, orography, time, and climate.
Figure 3 shows a typical fire process, divided into five phases. After ignition, a phase
of smoldering and glowing sets in, which leads to full combustion after a brief flash-
over. The fire abates in the fifth phase and dies once all combustible material has been
exhausted.
Figure 3: Typical fire process
581
The typical fire process can be decelerated if the right decision is made at the right time.
Therefore, a fire management plan is needed. The first step in the draw-up of such a plan is
to collect information on possible areas and objects of operation (e.g. by remote sensing,
as in the Sino-German Coal Fire Research Initiative (DLR et al. 2007)). Detailed data
(e.g. a database inventory) on the status of assigned and available personnel (reserves) and
technical equipment (e.g. water supply, extinguishing agents, and additives) must be put at
the disposal of decision-makers. Such a database inventory needs continued maintenance.
7 Cyclic Management Process
Based on an analysis of the coal fire situation (parameters) and the protection aims defined
beforehand, the most effective and efficient decisions concerning fire management tactics
can be derived again and again from the pool of options, i.e. defense, attack, rescue,
and retreat. They have been illustrated in Figure 2 based on methods that have proven
successful in practice, as well as new innovative methods (Hermülheim 2001; Xu et al.
2003).
Executives have to reflect, decide, and act systematically in periodical and closed
procedures in order to:
• Assess the situation,

• Plan the tactics, and
• Command the mission.
Figure 4 illustrates this cyclic management process.
Figure 4: Cyclic management process

Data source: Hagenkötter (1973); Schläfer (1998)
582
8 Conclusion
This paper aimed to highlight the following:
• Coal fires are a very complex challenge to environmental protection, resource

conservation, and fire fighting.
• Efficient control of coal fires needs fundamental strategies that are adjusted dynamically
to the general (macroscopic) situation.
• Effective extinction of coal fires needs adequate tactics applied to particular
(microscopic) situations.
• Phase A of the Sino-German Coal Fire Research Initiative “Innovative Technologies
for Exploration, Extinction, and Monitoring of Coal Fires in North China” (DLR et al.
2007) has provided pertinent information and data to be integrated into an algorithm for
support of a systematic decision process in fire fighting missions.
• In Phase B of the Sino-German Coal Fire Research Initiative, a systematic and
transparent method of coal fire control should be devised. This control process needs to
be reviewed periodically for continuous improvement.
Sustainable control of coal fires in the future could be arranged as follows:
Figure 5: Sustainable control of coal fires
583
References
DLR et al., 2007. Coal Fire Research: A Sino-German Initiative, 2003-2006. Final Report of Phase A
of the Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China.” German Aerospace Center (DLR),
Wessling, Germany.
Ferch, H. and Melioumis, M., 2005. Führungsstrategie – Großschadenlagen beherrschen. Stuttgart:

Kohlhammer.
Hagenkötter, M., 1973 (Ed.). Literaturexpertise über theoretische Grundlagen des Arbeitsschutzes. Project
report no. 111. Bundesanstalt für Arbeitsschutz und Unfallforschung (BAU), Wilhelmshaven,
Germany.
Hermülheim, W., 2001. Zur Bekämpfung von Kohlebränden in oberflächennah anstehender Kohle oder
in aufgegebenen Bergwerken – Stand der Technik sowie Übertragbarkeit von Verfahren des
Grubenrettungswesens. Hauptstelle für das Grubenrettungswesen, Herne, Germany.
Xu, J.C., Deng, J., Zhang, X.H., and Wen, H., 2003. Study on the mechanism and technique of controlling
spontaneous combustion in coalfields. Xi’an. (in Chinese)
Pennig, L., 2000. Globale Dimension der Kohlebrandproblematik. Research Project of the German
Aerospace Center (DLR), Wessling, Germany. Jena, Germany.
Rodewald, G., 1998. Brandlehre. 5th edition. Stuttgart: Kohlhammer.
Samland, S. 2004: State-of-the-art-Analyse der chinesischen Kohlebrandproblematik und entsprechender

Schadensminderungsmaßnahmen. Study paper. Department of Safety Engineering Methods,
University of Wuppertal, Wuppertal, Germany.
Schläfer, H. 1998. Das Taktikschema. 4th edition. Stuttgart: Kohlhammer.
Zhang, X., 2003. The practice of fire fighting in Rujigou Coalfield. Project report, as a contribution to
the Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
Extinction, and Monitoring of Coal Fires in North China.”
584
CER-Trading as a Means of Funding
Coal Fire Fighting in China
Friedrich-Karl Bandelow, Hartwig Gielisch, and Jörg Schulz
Abstract
Coal fires are a source of serious environmental damage and threaten the health of people living and
working in their vicinity. Common throughout the world, the phenomenon is especially severe in China.
The volume of CO2 emissions from uncontrolled Chinese coal fires equals that of entire industrialized
nations such as Finland or Sweden. Hence, an urgent need for research and development in the field of fire
mitigation and extinction.
The Clean Development Mechanism (CDM) as part of the Kyoto Protocol opens the possibility to
generate Certified Emission Reductions (CER) through coal fire extinction. The objective of the Kyoto
Protocol is to limit the worldwide output of greenhouse gases (GHGs). CDM provides an incentive for
investments in projects that promote sustainable development and reduce GHG emissions in developing
countries. Successful fire fighting can create earnings sufficient to offset its own expenses. The cost per
ton of CO2 reduction was computed in a DMT case study in Xinjiang Autonomous Region, P.R. China; it
proved that fire extinction can very well be a viable investment through generation of CERs. This paper
presents the specific process of setting up and successfully registering a CDM project on coal fire fighting
in China. Suggestions on how to finance such a project are given too.
摘要
地下煤火引起了巨大的环境破坏并严重威胁到火区及附近地区的居民健康生活和工作。煤火在世
界上是常见的现象，但在中国尤为严重。来自中国未控制火区的CO2排放量等于芬兰或瑞典这样
的工业化国家的排放总量。因此,迫切需要研究和开发煤火的减灾与灭火技术。
京都议定书中有关清洁发展机制（CDM）展现了通过灭火工程产生核证减排量（CER）的可能
性。京都议定书的目标是限制全世界温室气体的排放量。清洁发展机制（CDM）为发展中国家提
供了促进可持续发展和减少温室气体排放量的项目投资激励措施,而成功的灭火可以获得经济收入
并足以弥补灭火费用。DMT在新疆地区煤火案例研究中进行了减少CO2的吨成本计算，证明灭火
是一个利用CER可行的投资项目。本文介绍了在中国确立CDM项目和将灭火项目注册为清洁发展
机制项目的具体过程，并提出了如何资助灭火工程项目的建议。
585
1 Introduction
China’s national economy is growing by approximately 9 % annually. Its demand for
energy is rising more or less proportionally. Coal is the most important source of energy in
China, with an output of almost 2 billion t in 2005, covering 67 % of the country’s energy
supply. It is expected to keep this prominent role in the decades to come.
China is the second largest consumer of primary energy worldwide (after the U.S.A.) and
the second largest emitter of carbon dioxide (CO2). Its potential for cutting CO2 emissions
is considerable. Chinese power plant efficiency, technical standards, and renewable energy
sources, then, are areas of foremost (international) concern and should be the focus of
international collaborative projects.
This paper presents how to generate Certified Emission Reductions (CERs) according to the
Kyoto Protocol’s Clean Development Mechanism (CDM) by extinguishing coal fires in China.
2 Coal Fires and Their Effects
Uncontrolled coal seam fires presently occur at about 750 locations in China. It is assumed
that 20 million t or about 1 % of the country’s annual coal production is burned directly
in these fires. What is more, about 200 million t are lost for mining each year, rendered
unexploitable by mechanical and thermal destruction.
Uncontrolled coal seam fires have dire environmental consequences. They threaten the
health of people living in affected areas; pollute the soil, ground- and surface water with
CH-condensates and hydrogen sulfide (H2S); and destroy natural vegetation and farmland.
The issue of CERs depends on the emission reductions generated by a CDM project, more
precisely, on the reduction of CO2 equivalents. Coal combustion, independent of its nature –
be it industrial or natural – produces greenhouse gases (GHGs). These are noxious gases
such as CO2, methane (CH4), sulfur dioxide (SO2), and nitrogen oxides (NOx). In complete
combustion of coal, e.g. in modern power plants, carbon is converted almost completely
into CO2. The amount of CO2 generated by complete combustion of 20 Mt of coal with an
average carbon content of 750 kg t-1 is about 56 Mt.
In different settings, however, for example in coal fires, complete combustion cannot
always occur due to limited oxygen supply. Under these conditions, incomplete
586
combustion results, which produces a portion of CH4 in addition to CO2. CH4 has a
Global Warming Potential 23 times higher than CO2 and thus makes for a substantially
higher emission of CO2 equivalent. In such a scenario, the same 20 Mt of coal would
produce about 100 Mt of CO 2 equivalent. For the following CDM project model
calculations, complete combustion was assumed to allow for conservative scientific
statements and predictions.
Cutting CO 2 emissions by coal fire fighting is of global interest, since emissions

affect the entire atmosphere, not simply that above China. To illustrate the impact of
Chinese coal fires, their CO 2 emissions are plotted against those of 15 EU countries
in Figure 1.
Figure 1: CO2 emissions from Chinese coal fires (VRC; red) and 15 EU countries in
comparison. Marked in yellow are Finland and Sweden with emissions of
roughly the same size.
Data source: Eurostat (2006)
Evidently, coal fires in China generate CO2 emissions the size of the annual emissions
of highly developed industrial nations such as Austria, Finland, Denmark, or Sweden.
This highlights the urgent need for research and development in coal fire mitigation and
extinction.
587
3 The Kyoto Protocol and the Clean Development Mechanism
The Kyoto Protocol came into force on 16 February 2005, following its ratification by
Russia, which raised the level of global GHG emissions covered by the signatories to the
required minimum of 55 % (Article 25 of the Kyoto Protocol). The ultimate objective of
the Protocol is the “stabilization of greenhouse gas concentrations in the atmosphere at a
level that would prevent dangerous anthropogenic interference with the climate system”
(Article 2, United Nations Framework Convention on Climate Change (UNFCCC)). In
order to achieve this, mandatory reduction targets were set for the industrialized world
(Annex I countries according to the UNFCCC; Annex B countries according to the Kyoto
Protocol). Developing countries like China are exempted from these targets; however, they
have the option to participate in the GHG credit trading mechanisms foreseen under the
Protocol.
The Kyoto Protocol provides three mechanisms to help Annex I countries comply with
their reduction commitments by offering financially attractive alternatives to domestic
emission cuts; these are:
• Joint Implementation (JI) – produces Emission Reduction Units (ERUs),

• Clean Development Mechanism (CDM) – produces Certified Emission Reductions
(CERs), and
• Emission Trading (ET) – produces Assigned Amount Units (AAUs).
By promoting investments abroad, CDM effectively assists Non-Annex I countries in

achieving sustainable development and yet still reduces the volume of global GHG
emissions. The Kyoto Protocol and CDM hold a key to generating income in coal fire
fighting-related projects, not only in China but also in other developing countries. The
reduction of 1 t of CO2 in an accredited CDM project is worth one CER unit. CERs are
issued directly to the project participants and can be sold to partners in Annex I countries
which do not wish or are not in the position to realize their emission reductions by
measures within their national boundaries alone. Successful fire fighting thus can create
earnings sufficient to offset the expenses incurred during project implementation.
The Chinese government is aware of the environmental and economic disaster unfolding in
the northern Chinese coalfields and has launched a number of programs over the last years
to tackle the problem. In the beginning, these programs were motivated rather by resource
protection and mining safety, but environmental concerns such as global warming have
come to be a prominent driving force, particularly after China’s ratification of the Kyoto
Protocol at the World Summit on Sustainable Development in Johannesburg 2002.
588
4 Feasibility of Coal Fire Fighting in Xinjiang
The Deutsche Montan Technologie GmbH (DMT) in cooperation with the Xinjiang Coal
Bureau carried out a case study in an uncontrolled coal fire area of Xinjiang Autonomous
Region to estimate the viability of a CDM project to be implemented there. It was shown that
fire fighting can indeed be a lucrative investment through generation of CERs (Görlich 2004).
In the case study, the annual loss of coal was estimated at about 150,000 t for a given
fire location. The resulting CO2 emissions, based on the complete combustion model
(cf. Chapter 2), were calculated at 420,000 t CO2 per year. As geological parameters
and empirical estimations suggested that the fire would continue to burn with the same
intensity for at least 10 years, emissions of more than 4 Mt of CO2 were to be expected
over its lifespan.
The costs for fire fighting and extinction were calculated based on comparable operations
in previous years by the Xinjiang Coal Bureau. The resulting figure amounted to
EUR 4 million. Consequently, the cost for preventing the emission of 1 t of CO2 would be
EUR 0.95 in this scenario. However, additional costs for CDM accreditation would have to
be accounted for too.
The following diagram illustrates the price trend for CERs from April to November 2005.
Figure 2: Price development of CO2 certificates

Data source: Junqueira (2005)
589
The yellow graph in this diagram refers to emissions allocated to European countries in the
unrelated-to-Kyoto EU Emissions Trading Scheme (EU ETS). The blue graph shows the
price development of CDM credits as applicable to this case study. A slight positive trend
can be recognized, leading to a CER trading price on the carbon market of c. EUR 7.00 to
8.00 at the end of October 2005.
Given the estimated costs of c. EUR 1.00 per CER in the Xinjiang case study, it is
reasonable to expect profits from investments in coal fire fighting projects (Görlich 2004).
Even under different conditions – economically as well as geologically – the presented
scheme would be of considerable economic interest.
It should be added at this point that the Chinese government has made the international
average price their norm and tries to prevent the sale of certificates at dumping prices
(Abele 2005). Also, the government has levied so called “royalty fees,” which have to be
paid for every CER generated.
5 The CDM Process – Accrediting a Project
5.1 Preconditions in China
The legal basis for CDM accreditation in China is the document “Measures for operation
and management of Clean Development Mechanism projects in China” (China Climate
Change Info-Net 2005). The National Development and Reform Commission (NDRC),
according to this scheme, has the function of Designated National Authority (DNA)
in China, i.e. it decides which national projects can apply for registration at the CDM
Executive Board in Bonn (CDM EB). The NDRC receives advice from the National
CDM Board, a counseling body chaired by both NDRC and the Ministry of Science and
Technology of the People’s Republic of China (MoST) (Abele 2005).
In the CDM investment climate index for Asia, China ranks second behind India with 74
out of 100 points. Its pros are favorable macro-economic conditions and a large consumer
market with more than 1.3 billion buyers. China has been a WTO member since 2001.
On the cons, a government-controlled market, accompanied by inevitable red tape and
legal insecurity, figure large. There is an imminent risk of economic overheating, and
uncertainties regarding the Chinese energy market and its reform remain (Abele 2005).
590
So far, the number of nationally approved and CDM EB-registered projects in China is
relatively low; the NDRC, for example, had accepted only 18 applications by the end of
November 2005. The number of submitted Project Design Documents (PDDs) is estimated
at a few hundred. One reason for this reserve on the part of the applicants may be the late
entry into force of the Kyoto Protocol (2005) – they simply did not have the time yet to file
their projects. A strong increase is projected for the near future.
5.2 Processing of a CDM project
The aim of a CDM project is to generate CERs. But before these can be issued, a strict
process of preparation and validation has to be passed successfully. The general steps after
approval by the DNA are presented in Figure 3.
• Formulation of a Project Design Document (PDD), including:

o Project description,
o Baseline methodology,
o Estimation of emission reductions,
o Monitoring plan, and
o Environmental assessment;
• Validation of PDD by Designated Operational Entity (DOE);
• Registration by CDM EB; and
• Implementation by project developer and monitoring by DOE.
Figure 3: CDM project workflow
A central and equally controversial step in this process is the calculation and quantification
of emission reductions. CDM projects must “result in real, measurable, and long-term
benefits related to mitigation of climate change,” and emission reductions have to be
“additional to any that would occur in the absence of the certified project activity” (Article
12 of the Kyoto Protocol). In order to fulfill these criteria, it is essential to use a baseline,
i.e. a theoretical volume of emissions that would occur if the project was not implemented.
With this reference scenario it is possible to calculate the resulting emission reductions by
subtracting the project emissions from the baseline (Butzengeiger & Michaelowa 2002).
The Kyoto Protocol also stipulates that the anthropogenic origin of the emissions be
proven. In the case of coal fires, this was unanimously agreed upon by the parties of the
Sino-German Coal Fire Research Initiative “Innovative Technologies for Exploration,
591
Extinction, and Monitoring of Coal Fires in North China” and by the participants of the
International Conference on Coal Fire Research.
6 CDM Project Financing and Risks

The amount of CERs generated by a CDM project is influenced by many parameters. Most
of these are intertwined. Personal judgments, unavoidable in most cases, further complicate
matters.
CDM application and accreditation are strictly regulated; they involve a large amount of
time and effort. And yet, rejection by either DNA, DOE, or CDM EB is looming at every
step of the process, threatening to nullify the work done. Reasons for rejection, especially
after national approval, may be inaccuracy of applied methods for baseline verification and
emission reductions, or undefined costs for accreditation of the first PDD for a project.
Once a CDM project has been approved and registered, there are different options for the
host country to capitalize on the CERs issued:
• Direct sale of CERs to the EU ETS;

• Involvement of project partners from Annex I countries:
o Project financing and re-payment of loans by CERs,
o Equity investment in a joint CER company, or
o Technology transfer in exchange for CERs.
7 Suggestions
Compared to the projects that have already been registered with the CDM EB and are
now being implemented, securing CDM accreditation of a coal fire project appears
to be a formidable task. Estimations of emission reductions here only rely on broad
approximations of the extent and lifespan of a fire. The same is true for CO2 baselines and
the amount of coal saved after extinction. Also, coal reserves that generate CERs have to
be evaluated regarding their mineability and their utilization as fuel in the future since later
exploitation would annihilate the projected emission reductions.
Nevertheless, the sheer amount of GHGs emitted from uncontrolled coal seam fires
justifies that further efforts be made to probe the possibility of a coal fire CDM project.
592
To date (2005), there is no PDD for such a project and, given the complexity of the
accreditation process, its formulation should not be shouldered by one party alone. Instead,
it seems advisable that international development agencies support and sponsor the first
CDM project on coal fires in China, so as to level the ground for private initiatives. A
Chinese precedent, moreover, would make it possible for other developing countries to
attract foreign investors and thus eradicate coal fires beyond the Chinese context.
References
Abele, C., 2005. CDM-Markt Kompakt, VR China. Bundesagentur für Außenwirtschaft (bfai), Köln,
Germany and Deutsche Investitions- und Entwicklungsgesellschaft mbH, Köln, Germany.
9 January 2006 http://www.bfai.de/nsc_true/DE/Content/__SharedDocs/Anlagen/PDF/CDM/
cdm-markt-china,property=publicationFile.pdf
Butzengeiger, S. and Michaelowa, A., 2002. Baselines. KEEN background paper. Hamburg Institute of
International Economics (HWWA), Hamburg, Germany. 6 January 2006 http://www.hwwa.
de/Projects/Res_Programmes/RP/Klimapolitik/Baselines1.pdf
China Climate Change Info-Net, 2005. Measures for operation and management of Clean Development
Mechanism projects in China. Office of National Coordination Committee on Climate
Change, Beijing, P.R. China. 6 January 2006 http://www.ccchina.gov.cn/english/source/ca/
ca2005112901.htm
Eurostat, 2006. Structural indicators for energy. Environment and energy. Statistical Office of the
European Communities (Eurostat), Luxemburg. 8 January 2006 http://epp.eurostat.cec.eu.int/
portal/page?_pageid=0,1136239,0_45571450&_dad=portal&_schema=PORTAL
Görlich, B., 2004. Untersuchung der Brandzone Shui Xi Gou (Westliches Wassertal) in der Autonomen
Region Xinjiang (China). Internal project report for RAG Group, Essen. Deutsche Montan
Technologie GmbH (DMT), Essen, Germany.
Jourdain, C., 2005. The Clean Development Mechanism. Ecoal – The quarterly Newsletter of the World
Coal Institute 52: 4-5.
Junqueira, M., 2005. Carbon Market Newsletter 11/05. Econergy, Sao Paolo, Brazil. 6 January 2006
http://www.carbontrade.com/Asp/DisplayPDF.ASP?NUMBER=90
UNFCCC, 2005. United Nations Framework Convention on Climate Change. United Nations Framework
Convention on Climate Change (UNFCCC), Bonn, Germany. 6 January 2006 http://unfccc.
int/2860.php
593
Part VIII
Summary
and Conclusions
Toward a Better Understanding of Coal Fires
for Sustainable Fire Control and Prevention:
Conclusions and Recommendations
Outcome of the “International Conference on Coal Fire Research –

ERSEC Conference on Understanding, Extinction, and Prevention of
Spontaneous Coal Seam Fires”
29 November - 1 December 2005, Beijing, P.R. China
1 Introduction
This paper summarizes the findings of the “International Conference on Coal Fire Research –
ERSEC Conference on Understanding, Extinction, and Prevention of Spontaneous Coal
Seam Fires,” held in Beijing, P.R. China from 29 November to 1 December 2005. The
conference took place within the context of the ongoing Sino-German Coal Fire Research
Fires in North China,” a project aiming to contribute with extensive research in a bilateral
dialogue to a better understanding of spontaneous coal seam fires, their extinction and
prevention in China and worldwide. Delegates represented a broad spectrum, ranging from
science and research to industry and administration. While most participants had their
base in either China or Germany – the two project partners–, scientists from France, India,
Kazakhstan, Mongolia, New Zealand, South Africa, and the U.S.A. also attended and
contributed with their expertise.
Spontaneous coal seam fires have been known in different parts of the world for millions
of years; they are indeed a global phenomenon. While modern mining technology has
meanwhile reached a very high technological standard and, in commercial mines at least,
is put to use efficiently where new shafts are built, the main problem today rests with
coal seams that started to burn long ago, and new fires in the wake of so-called “wild”
596
mining, i.e. illegal, mostly private, unsystematic coal mining over large areas. Self-ignition
may also occur in small mines that have not been sealed properly, or unattended goafs of
large-scale coal mines. In northern China, a virtual coal fire belt extends over more than
5,000 km from east to west, eating up each year coal reserves of an estimated 20 million t,
and barring another 200 million t from future extraction. Coal fires moreover produce large
quantities of greenhouse gases. Against the backdrop of this economic and environmental
disaster, the main and most urgent task for all parties concerned is to extinguish and
prevent coal fires in a most sustainable way. With a view to achieving this, researches
and officials in the countries most affected by the phenomenon – Australia, China, India,
Indonesia, South Africa, and the U.S.A. in particular, but also a number of East and South
European countries – are now focusing their efforts on gaining more in-depth knowledge
about:
• Catalytic compounds and conditions of spontaneous combustion;

• The process of self-ignition;
• Coal fire propagation;
• Heat transfer and gas flux;
• Extinction methods, strategies, technologies, and materials; as well as
• Re-ignition of coal fires and its reasons.
With global energy demand rising dramatically (especially so in countries experiencing

huge industrial and demographic growth such as China), it is only a logic imperative
to combine forces and capital in an internationally concerted move to mitigate the
tremendous impact of coal fires – this will protect local environment and public health,
save national resources, and contribute to the reduction of toxic and greenhouse gases on
a global scale.
597
2 Conference Summary and Statements
The following section is structured according to questions that were raised in the context
of the conference and answers provided by the speakers, organizers, and audience.
What is the overall problem?

• Coal fires cause a plethora of problems and hazards worldwide:
o Toxic and greenhouse gas emissions;
o Air and groundwater pollution;
o Soil degradation;
o Loss of coal = non-renewable resource;
o Destruction/degradation of ecosystems, land subsidence, collapse, and erosion;
o Health hazards (respiratory diseases); and
o Impaired mine safety.
Do we have institutional support for solving this problem?

• China and other affected countries have recognized the immense problem of coal fires.
• However, national institutions which could effectively guide and control coal fire
management are still missing.
• The Sino-German Coal Fire Research Initiative is an important step toward solving
these issues in the Chinese context, and will certainly be an example for other countries
faced with the same problem.
• The Clean Development Mechanism (CDM) of the Kyoto Protocol is a possible tool to
finance fire extinction, but implementation has not taken off yet.
What are the overall goals of coal fire research?

• Understand and explain (scientifically) how and why coal fires ignite; how they can be
detected and monitored; and, ultimately, how they can be extinguished on a permanent
basis.
• Identify and quantify local, regional, and global effects of coal fires.
• Develop, publicize, and apply mining methods that raise mine safety and have no or
less potential to cause spontaneous combustion.
598
What are the goals/purposes of this conference?

• Enable exchange of research results and networking between the Sino-German
project partners and other scientists, industry representatives, and (non-)governmental
institutions.
• Broaden the field of coal fire research to include other affected countries.
• Lay the foundation of an international coal fire research network (forum).
• Identify and integrate key procedures and policies for coal fire mitigation.
What has impeded progress toward these goals?

• There exists no international and interdisciplinary research community/platform so far.
• Researchers and fire-fighters think predominantly in local contexts.
• Spontaneous coal seam fires, in most cases, are underground phenomena – they are not
directly visible from the surface and difficult to access/research.
• Coal fire fighting is intricate and very costly.
Who should contribute to achieving these goals?

• Political bodies (governments) at all levels – local, regional, national, and international.
• International (non-governmental) institutions with a focus on coal fire research and
mitigation.
• The academia at large: experts, researchers, managers, and social scientists.
What progress has been made?

• The Sino-German Coal Fire Research Initiative has substantially deepened the
knowledge of coal fire-related processes in its first phase, by:
o Determining (natural and anthropogenic) boundary conditions for spontaneous
combustion;
o Developing mathematical models of coal fire development, convection processes,
gas emissions, rock mechanics, and temperature and burning processes;
o Performing in-situ measurements to determine reaction types;
o Establishing automated and semi-automated remote sensing methods as a valuable
tool in coal fire research; and
o Setting up a coal fire data warehouse for all parties to use at their discretion.
599
What is our time horizon to achieve success?

• Coal fire research is a continuous task. Significant progress, however, is possible within less
than five years. This time frame should be sufficient to develop techniques for exploration,
extinction, and monitoring of coal fires and to test them under field conditions.
What are the constraints on coal fire fighting?

• Technical solutions need an interdisciplinary and integrated approach.
• Coal production is thought by many to be more important than fire prevention.
• Small-scale “wild” mining is still not under control.
• Funds are generally lacking.
• Administrative and political responsibilities are either unclear or completely missing.
What techniques will lead to successful coal fire fighting?

• Optimization of existing techniques.
• More accuracy in fire detection.
• Prevention based on coal fire-sensitive mining and “early warning systems.”
• Site-specific combination and integration of models and methods into one consistent
workflow applied rigorously in all cases of spontaneous combustion.
• Efficient mechanisms to control mine air supply and ventilation.
What research is needed?

• Some basic coal fire processes and their preconditions still remain unknown – they
need to be researched in greater detail.
• Enhanced fire exploration and adequate description of geometry.
• Development and evaluation of baselines for CDM project implementation.
• “Non-technical” political and socio-economic investigations into coal fire causes.
• Evaluation/adaptation of models and extinction methods in/to different mining sites.
How can we prevent new coal fires?

• Completely stop or firmly control “wild” coal mining.
• Employ appropriate coal fire-sensitive mining methods.
• Set up coal fire “early warning systems” (risk analysis).
• Properly seal and/or refill abandoned mines.
600
• Train and educate staff on the risks of spontaneous combustion.

• … and, of course, save energy!
What monitoring methods should be encouraged?

• High-resolution remote sensing-based monitoring for selected areas.
• Borehole logging with temperature and emission gas surveys for extinguished fire sites.
• Combination of remote sensing and in-situ measurements in a GIS-based information
and simulation system.
• Real-time monitoring of mine ventilation.
What management structure is needed?

• Responsibilities must not be delegated – all parties involved in coal mining (companies,
communities, and provinces) should contribute to/participate in fire fighting and
prevention.
• A Chinese national bureau for coal fire management should be established.
• Current mine management structures should be investigated and realigned with a view
to efficiency in communication and information processing.
What community interaction is needed?

• No purchase of coal from illegal mines.
• Regular information of local people on environmental and health hazards.
• Improved stakeholder management and communication.
• Integrated management system according to ISO norms.
We are convinced that the conference and these proceedings give an adequate overview
of state-of-the-art coal fire research, outlining the basic processes of coal fire ignition and
propagation, and presenting all relevant extinction and monitoring methods known to date.
In this respect, the International Conference on Coal Fire Research in Beijing provided for
a new basic understanding of this complex phenomenon from the perspective of different
scientific fields and approaches. With its emphasis on the regional and global consequences
of coal fires, it can, and certainly will be, a starting point for a renewed, more sustainable
and environmentally friendly effort at coal fire control and prevention – in China and
worldwide.
601
进一步提高面向可持续发展的
煤火控制与预防的认识：
结论与建议
2005年11月29日-12月1日在中国北京召开的
“国际煤火研讨会——暨中国环境可持续发展生态研究（ERSEC）框架
下的煤层自燃探测、灭火及预防研讨会”总结
1 引言
本文总结了“国际煤火研讨会——暨中国环境可持续发展生态研究（ERSEC）框架下
的煤层自燃探测、灭火及预防研讨会”的成果。该会议是于2005年11月29日到12月1
日在中国北京召开的，是基于正在实施的中德煤火科技合作项目——“中国北方煤火
探测、灭火与监测新技术”的项目成果召开的。该项目旨在通过双边合作，深入研究
地下煤火的基本规律，探索中国和世界范围内防灭火的新技术。会议代表来自科研、
企业和管理部门，大多数与会人员来自中国和德国合作双方。来自法国、印度、哈萨
克斯坦、蒙古、新西兰、南非和美国的科学家也在会上介绍了各自的经验。
众所周知，几百万年以来煤层自燃一直是在世界各地普遍发生，的确已经成为全球
性现象。虽然现代采矿技术已经达到很高的水平，至少在已建立新通风系统的大规
模生产矿井中都得到有效的应用，但目前的主要问题在于煤层在很久前就已经开始
燃烧，并且在较大范围存在着许多不合法的、个体的、无序开采的煤矿引发的新煤
火问题。自燃也可能在未适当封闭的小煤矿和未被注意的已废弃大型煤矿中发生。
中国地下煤火带主要分布在北方地区的东西长约5 000公里的区域中，估计每年烧
失煤量约2 000万吨，还有近2亿吨的煤炭资源受到煤火的影响，同时还产生了大量
的温室气体。针对煤火带来的经济和环境问题，对于关注煤火问题的各方来说，当
前最主要和紧迫的任务是寻求适于可持续发展道路的灭火与煤火预防工作方式。为
此，来自受煤火影响国家的科学家和政府管理者——特别是来自澳大利亚、中国、
印度、印度尼西亚、南非、美国，也包括来自欧洲东部和南部一些国家的科学家和
政府管理者，都将研究重点放在：
602
• 煤自燃反应与形成的环境条件，
• 煤自燃过程，
• 煤火的发展，
• 煤火的热扩散及气体变化，
• 灭火方法、策略、技术和材料，
• 煤火复燃及其原因，
……
等方面，以便获得进一步认识。
随着全球能源需求的急剧增加（尤其是正在经历高度工业化和人口增长过程的国
家，如中国），迫切需要通过国际社会的共同努力，投入各种物力与财力，致力于
减轻煤火造成的各种影响与保护区域环境和公众健康，保护国家煤炭资源，并对减
少全球有毒气体和温室气体的排放有所贡献。
2 大会纪要
本部分内容是依据大会演讲者、组织方和与会人员所提出的有关问题的相关解释。
什么是煤火的基本问题？
• 煤火引发了一系列的全球范围的问题和灾害：
o 排放了有害气体和温室气体，
o 污染大气和地下水，
o 造成局部土地退化，
o 损失不可再生的煤炭资源，
o 导致生态系统退化、滑坡、坍塌和侵蚀，
o 引起健康问题（如呼吸系统的疾病），
o 威胁煤矿安全生产。
是否有机构支持来解决这个问题呢？
• 中国和其他受煤火影响的国家已经认识到煤火问题的严重性。
• 可以有效的指导和控制煤火管理的政府机构尚未健全。
• 中德煤火合作研究项目是解决中国煤火问题的重要努力，该项目一定将成为其他
面临煤火问题国家的一个典型案例。
• “京都议定书”下的清洁发展机制（CDM）是一种有可能解决灭火经费的途径，
但目前尚未启动。
603
什么是煤火研究总目标？
• 从科学的角度了解和解释煤火自燃的成因和途径，如何探测与监测，如何能够将
煤火永久性熄灭的灭火方法。
• 定性与定量分析煤火对局部、区域和全球的影响。
• 开发、普及和应用能够提高矿山安全水平，减少和消除煤火发生的采矿方法。
本次会议的主要目标和意图是什么？
• 使中德煤火项目合作伙伴和来自各国的科学家、产业界代表以及非政府机构共同
交流研究结果。
• 拓宽煤火研究的领域到其他受煤火影响的国家。
• 为建立煤火的国际研究体系奠定基础。
• 为煤火的减灾工作提出明确的范畴并集成主要的方法和有关政策。
实现这些目标的主要制约因素是什么？
• 至今尚未形成国际的、多学科交叉的煤火研究机构或者研究平台。
• 研究人员和灭火工作者认为煤火主要是地方性问题。
• 煤自燃一般情况下都发生在地下，在地表不易直接观测到，因此接近或研究是很
困难的。
• 灭火是非常复杂的问题，而投入的费用又很高。
谁应为实现这些目标有所贡献？
• 各级政府机构，包括地方的、区域、国家的和国际机构等。
• 关注煤火研究和减灾的国际非政府组织。
• 规模较大的学术机构：专家、研究人员、管理者和社会科学家。
已获得的进展是什么？
• 中德煤火项目第一阶段研究进一步加深了对煤火发展过程的认识：
o 确定了自然和人为原因下煤火发生的边界条件；
o 开发了煤层自燃发展、热对流过程、气体排放，岩石应力作用、温度变化和煤
燃烧过程的数学模型；
o 进行了实地测量，从而确定了煤火的各种类型；
o 建立了煤火信息的自动和半自动遥感提取方法；
o 建立了可供煤火合作研究人员使用的煤火数据仓库。
604
我们要达到成功的时间范围是？
• 煤火研究是一个长期的任务。但是，在近五年内将有可能取得显著的进展。五年时
间内足以开发各种用于煤火探测、灭火和监测的技术，并在现场条件下进行验证。
什么制约了灭火工作？
• 灭火技术方案需要有一个跨学科和综合性方法。
• 只注重煤炭生产，而忽视煤火预防。
• 小煤窑开采尚未得到有效控制。
• 一般都缺乏资金。
• 政府管理者的责任尚不明确或者缺位。
取得灭火工作成功的主要技术是什么？
• 基于现有灭火方法的优化技术。
• 进一步提高煤火勘查精度的探测技术。
• 基于易燃煤层防火型开采方法和早期预警系统的预防技术。
• 将各种特定条件下煤火灭火模型方法集成为可严格应用到各种条件下的统一的工
作流程。
• 控制矿井通风的有效方法。
仍需研究的内容是社么？
• 进一步深入研究一些煤火的发展过程和煤火发生的前提条件。
• 加强煤火勘查并可准确地确定煤火空间范围。
• 清洁发展机制（CDM）项目实施中基准的开发与评价方法。
• 研究煤火发生的有关社会政策和社会经济等非技术方面原因。
• 研究不同采矿条件下煤火评价与适用的模型和灭火方法。
如何才能防止新的煤火发生?
• 全面禁止和严格控制小煤窑。
• 开发适用于煤层易自燃条件下的采矿方法。
• 建立煤火的“早期预警系统”（风险分析）。
• 妥善密封和再回填废弃煤矿。
• 开展煤火风险方面的人员教育和培训。
• 当然，还有节约能源等方面！
605
应该鼓励的监测方法是什么？
• 在选定区域进行高分辨率遥感监测方法。
• 熄灭区进行钻孔温度和气体调查的监测方法。
• 基于GIS信息和模拟系统的遥感和现场观测结果综合分析方法。
• 矿井通风的实时监测方法。
需要怎样的管理结构？
• 所有参与煤炭开采的各方（公司、社团和各级地方政府）都应参与灭火和预防，
决不能转移责任；
• 中国政府应确立专门的部门研究煤火与管理灭火工作；
• 开展现有煤矿管理体系的调查，在信息处理和沟通方面加强协作，以便提高效
率。
需要不同机构的合作是什么？
• 禁止与非法煤矿的交易。
• 定期发布当地环境灾害和健康方面的信息。
• 加强利益相关者的管理和沟通。
• 根据国际标准化组织的规范（ISO标准）建立综合的管理体系。
我们深信：本次会议及其论文集概括了煤火研究方面的进展，提出了煤自燃和发展
基本过程，展示了至今已有的煤火监测与灭火方法。在此方面，本次会议为从不同
的科学领域和不同的途径进一步认识复杂的煤火问题奠定了新的基础。随着有关煤
火的全球和区域性影响认识的进一步加深，本次会议可能并一定会成为我们朝着在
中国乃至世界范围的煤火控制和预防中更新的、可持续的、注重环境友好方向努力
的新起点。
606
Spontaneous Coal Seam Fires: Mitigating a Global Disaster/ UNESCO Office Beijing
© 2008 by UNESCO Office Beijing
All rights reserved. No parts of this book may be reproduced, in any form or by any means,
without permission in writing from UNESCO Office Beijing.
Tsinghua University Press

100084 Beijing, P. R. China
http://www.tup.com.cn
版权所有，翻印必究。举报电话：010-62782989 13501256678 13801310933
图书在版编目(CIP)数据
煤火自燃的控制与可持续管理 / 联合国教科文组织北京办事处编.
—北京：清华大学出版社，2008.3
ISBN 978-7-302-17146-1
Ⅰ. 煤… Ⅱ. 联… Ⅲ. 煤—自燃—控制—文集— Ⅳ.TQ534.9-53
中国版本图书馆CIP数据核字(2008)第032952号
出版发行：清华大学出版社地址：北京清华大学学研大厦 A座
http://www.tup.com.cn 邮编：100084
c-service@tup.tsinghua.edu.cn
社总机：010-62770175 邮购热线：010-62786544
投稿咨询：010-62772015 客户服务：010-62776969
责任编辑：徐晓飞 (xuxiaofei@tsinghua.edu.cn)
整体设计：徐晓飞
封面设计：傅瑞学
责任校对：王淑云
责任印制：孟凡玉
印装者：北京雅昌彩色印刷有限公司
开本：250×180 印张：39.5 字数：1000千字
版次：2008年3月第1 版印次：2008年3月第1次印刷
印数：1~500
书号：ISBN 978-7-302-17146-1
本书如存在文字不清、漏印、缺页、倒页、脱页等印装质量问题，请与清华大学出版社出版部联系调换。
联系电话：010-62770177转3103 产品编号：029161－01

159538M PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

159538M PDF

Uploaded by

Copyright:

Available Formats

UNESCO Publication as part of “Ecological Research for

Sustaining the Environment in China (ERSEC)” under funding

Disclaimer ERSEC Project Secretariat

The “International Conference on Coal Fire Research – ERSEC Conference on

It is part of ERSEC’s agenda to publish the conference recommendations in a book

Prof. Dr. Walter R. Erdelen

1) Missing awareness of the ecological impact of coal fires,

Sufficient funding, appropriate mining and fire fighting techniques and

The “International Conference on Coal Fire Research – ERSEC Conference on

The central outcome of the conference, without doubt, is a common understanding

Dr. Stefan Voigt

Many thanks to the following people:

 Chinese Translations (non-Chinese authors)

Project Partners – China

 Shenhua Group Corporation Limited

 China Geological Survey (CGS)

 China National Administration of Coal Geology (CNACG)

 Aerophotogrammetry and Remote Sensing Bureau of China Coal (ARSC)

 Beijing Normal University (BNU)

 Beijing Ouhualian Science and Technologie Ltd. (BOST)

 Xi’an Branch of China Coal Research Institute (CCRI)

 Wuda Coal Mining Corporation, Shenhua Group

 Ningxia Coal Corporation, Shenhua Group

 State Coal Industry Bureau (SCIB), Inner Mongolia (IMBIC)

Project Partners – Germany

 German Aerospace Center (DLR)

 Leibniz Institute for Applied Geosciences (GGA)

 Freiberg University of Mining and Technology

Project Partners – Associated Members

 International Institute for Geo-Information Science and Earth Observation (ITC)

 Deutsche Montan Technologie GmbH (DMT)

 Chinese National Commission for UNESCO

 Ministry of Education of the People’s Republic of China (MoE)

Mr. Zhang Guocheng / 张国成先生················································· 28 / 30

Dr. Renate Loskill / 罗瑞娜博士······················································ 32 / 35

Dr. Yasuyuki Aoshima / 青岛泰之博士· ············································ 37 / 39

A. Sinha and V.K. Singh

Stefan Voigt and Horst Rüter

M. Naze-Nancy Masalehdani, Jean-Luc Potdevin, Fabrice Cazier, and Dominique Courcot

Ümüt Halik, Friedrich-Karl Bandelow, and Jörg Schulz

Christian Lohrer, Martin Schmidt, and Ulrich Krause

Jolanta Kus, Werner Hiltmann, and Aenne Balke

Yu Minggao, Jia Hailin, Zuo Qiuling, and Xu Jun

Manfred P. Hochstein and Christopher J. Bromley

Stefan Schlömer, Manfred Teschner, Jürgen Poggenburg, and Christian Seeger

Gerlinde Schaumann, Bernhard Siemon, and Yu Changchun

Zhang Jianmin, Xiang Yaling, Wang Mei, and Huan Zhongdan

Zhang Jianmin, Huan Zhongdan, Wang Mei, and Xiang Yaling

Li Guangliang, Wang Mei, and Xiao Haihong

Huan Zhongdan, Tian Yuan, Sun Yujing, and Zhang Jianmin

Winfried Kessels, Manfred W. Wuttke, Stefan Wessling, and Li Xuan

Ulrich Krause, Martin Schmidt, and Christian Lohrer

Zhang Xinhai, Xu Jianzhong, Xi Guang, Deng Jun, and Wen Hu

Uli Barth, Birte Grashorn, and Friedrich-Karl Bandelow

Friedrich-Karl Bandelow, Hartwig Gielisch, and Jörg Schulz

Distinguished guests, ladies and gentleman,

It is my extraordinary pleasure to be attending today the “International Conference on

Ladies and gentlemen,

Ladies and gentlemen,

I am delighted to welcome you on behalf of the German Federal Ministry of Education

This conference is a milestone in Sino-German cooperation in research and education. Its

Thirdly, this conference is meant to contribute to capacity-building in China and stimulate

Ladies and gentlemen,