6857_.book Page IX Mercredi, 16.

janvier 2013 3:00 15

> STDI FrameMaker Couleur

Introduction:
Goals and organisation
of this book

This study of earth subsurface modeling is intended for several categories of readers geologists, engineers, and managers involved in the study and evaluation of subsurface reservoirs
and hydrocarbon exploration. The authors cover progress in various fields of computer sciences such as solid modeling or knowledge engineering, which can provide useful techniques
for solving several crucial issues concerning the modeling of geological reservoirs. The book
will also be of interest to researchers in computer science. Problems of major importance
have recently made their appearance in earth modeling and are likely to provide opportunities for researchers to implement some of the many innovative techniques that are currently
being developed, on actual case studies. More broadly, earth modeling is a good example of
an engineering field that involves complex methodologies, a range of skills, and multiple
participants. This book will also be of interest to engineers, researchers, and students who
wish to apply advanced knowledge-based techniques to complex engineering problems.
Over the last two decades, earth modeling has become a major investigative tool for
evaluating the potential of hydrocarbon reservoirs. During that time, significant progress has
been made in interactive and user-friendly software that is commonly used by professionals
for oil and gas exploration. At present, petroleum companies are facing a number of challenges in exploration. Not only must they investigate newly identified resources but, given
existing hydrocarbon price fluctuations, they must also re-evaluate the potential of previously investigated reservoirs, examine new prospecting data, and consider divergent interpretations. However, significant difficulties are involved. Earth models used for oil and gas
exploration and production rely on various types of raw data (seismic data, well-log data,
thin-section observation, rock sample analyses). They are synthetic representations that
incorporate a variety of different interpretations made at successive steps of the modeling
process (seismic interpretation, stratigraphic correlations, petrological property identification) by professionals with different interests and abilities (geophysicists, geologists, petrologists, reservoir engineers). At present, the modeling procedures currently in use in the
industry provide no way to link a range of data and interpretations with a final earth model.
In the best cases, these links can only be retrieved by examining the written reports that
accompany the model itself. This situation is very unsatisfactory, and presents a serious
obstacle to evaluating the quality of a given model and improving it by modifying some of
the assumptions on which it rests.
In the early 2000s, a research group was assembled, involving researchers from
IFP Energies nouvelles (IFPEN), the cole des Mines de Paris (ENSMP), and other institu-

6857_.book Page X Mercredi, 16. janvier 2013 3:00 15

> STDI FrameMaker Couleur

Shared Earth Modeling

tions with the goal of finding practical means to overcome these difficulties. The work of
this group is based on two major assumptions. The first consists in treating geology more
specifically the geological objects present within earth models and the chronology of the
geological events with which those objects are associated as a form of critical knowledge
that would form the core of the earth modeling process. Accordingly, the engineering system corresponding to the reservoir modeling chain (i.e., the succession of procedures
enacted during the modeling process) should be designed so that a permanent link can be
established between the objects examined by various specialized fields (seismic reflectors,
well markers, geometrical surfaces) and the geological objects (geological units, geological
boundaries) with which they have been associated by geological interpretation.
The second assumption consists in shedding new light on the concept of Shared Earth
Modeling, which was defined some years ago by earth model makers [L. Cosentino,
2001] 1. The term Shared Earth Modeling expresses the idea that earth models should be
built in such a way that experts and end users can have access, at any time, to the data and
assumptions used to produce the model. In light of this goal, it is our belief that shared earth
modeling should not be data-driven but knowledge-driven. This implies that the geo-modeling process should not be directly dependent on data but on the interpretations of geoscientists [Rainaud et al., 2005; Mastella et al., 2007] 2.
Over the last decade, IFPEN/ENSMP research group has been studying practical solutions to enable geoscientists to produce Shared Earth Models driven by geological knowledge. The goal of this book is to present the key results that have been obtained and evaluate
them in terms of overall progress in the broader field of earth modeling. It is the result of the
work of the 30 contributors listed at the end of the book, all of whom shared in the writing of
this book. The complete text has been revised by the editors (MP, JFR).
The book is organized into five parts. The first provides an overview of geological models and representations, examines what they are and what they are used for. Chapter 1 lists
key issues associated with models and representations in general and earth models in particular. It describes how constructing earth models depends on the requirements of users and
the choices made by modelers. And it shows how interpretation is a key component of the
geological modeling process, and lists some major problems that should be addressed when
designing geological models.
Chapter 2 introduces the various geological objects to be modeled and the various types
of data involved. It considers certain features of these objects and data, and identifies key
issues related to the building of earth models. Chapter 3 describes the various types of earth
models (structural, stratigraphic, reservoir models) currently in use in hydrocarbon reservoir
studies, as well as the software and workflows used to construct them. Present-day issues
related to data integration are discussed along with various collaborative efforts (EPISEM,
1. Cosentino, L., 2001, Integrated Reservoir Studies, 336 p., Editions Technip, Paris.
2. Rainaud, J. F., Perrin, M., Bertrand, Y. Innovative Knowledge-Driven Approach for Shared Earth
Model Building; Joint SPE/EAGE Conference, Madrid, June 2005, Extended Abstract, SPE 94172 PP.
Mastella, L., Perrin, M., Abel, M., Rainaud, J.-F., Touari, W. Knowledge Management for Shared
Earth Modelling; in 69th EAGE Conference and Exhibition, incorporating SPE EUROPEC 2007,
London, 2007a. EarthDoc. URL: http://www.earthdoc.org/detail.php?paperid=D021&edition=27.

6857_.book Page XI Mercredi, 16. janvier 2013 3:00 15

> STDI FrameMaker Couleur

Introduction: Goals and organisation of this book

XI

POSC). Finally, we show how knowledge integration has become an additional challenge
for Shared Earth Modeling, a theme that will be discussed throughout the book.
Part 2 examines the major issues that must be addressed at various stages of the earth
modeling building process and suggests innovative solutions for resolving them. Chapter 4
considers present-day issues associated with seismic interpretation. It then proposes a new
approach based on cognitive vision 1. The proposed method relies on establishing links to
the geological objects that must be identified, assigning visual attributes to seismic images,
and using specific procedures to deal with the high volume of data that must be processed.
The method has already provided encouraging preliminary results.
Chapter 5 considers the various constraints and requirements associated with the representation of individual geological surfaces, which are the basic elements of structural models. We show that there are many advantages in choosing a representation based on triangulated surfaces and a model that operates on a local scale. Particular solutions to issues such
as data simplification and hole filling are then proposed as a means to improve triangulated
surfaces based on seismic interpretation and to model horizons and faults in accordance with
geological requirements.
Chapter 6 examines the conditions that must be respected for producing geologically
consistent surface assemblies within a 3D structural model. Using geological maps as a template, it shows how the topology of a structural model is strictly correlated with the geological and chronological relationships that exist among the various surfaces. It further shows
that it is possible to determine the topology of a given structural model from a formalized
geological interpretation of the set of geological surfaces that compose it. Finally we show
how this property can be used to define a methodology to automate the construction of structural models, and we examine the data models needed to do so.
Chapter 7 describes existing methods of 3D meshing and how such meshing can be used as
a support for the attribution of local properties inside geological volumes. The chapter identifies the constraints imposed by geology and simulation protocols, and examines practical solutions for accommodating them. Special attention is given to solving specific problems, such as
fault insertion and horizon flattening. Chapter 8 addresses the problem of data completion and
extension by considering geostatistical methods for completing property patterns within earth
models and discusses the quality of the results in light of the expectations of geologists.
Part 3 is dedicated to geological knowledge, its formalization, and its use in achieving the
goal of knowledge-driven shared earth modeling. Chapter 9 characterizes ontology as the
best available tool for formalizing the scientific and technical knowledge used to construct
earth models. We discuss the types of ontologies that should be used and the conditions that
must be met for them to be robust and well-constructed. Chapter 10 examines ontologies for
structural model building. We present the GeoSciML model, currently being used to model
knowledge associated with geological maps, and discuss its advantages and limitations. We
also show how complementary ontologies can assist in describing shared chronological and
topological relationships among the geological objects used to create structural models. We
1. Maillot, N., Thonnat, M., 2008, Ontology Based Complex Object Recognition, Image and Vision
Computing 26,1 (2008), pp. 102-113.

6857_.book Page XII Mercredi, 16. janvier 2013 3:00 15

XII

> STDI FrameMaker Couleur

Shared Earth Modeling

then show how the ontologies that we have defined are used in practice. Two examples are
given, one involving a multi-agent method for structural model validation, and the other a
correlation of time scales and stratigraphic successions. Chapter 11 deals with the issue of
translating natural language representations so that their semantic content can be used by a
software system. Two methods for building ontologies from and for textual material exploitation are presented, one description-oriented and the other document-oriented.
Chapter 12 provides an in-depth examination of an ontology-based rock description and
interpretation. Two software tools from UFRGS (Brazil), Petroledge and Strataledge, are
presented, which make use of microscopic and core data, respectively. The software offers
several advantages: it integrates the practical knowledge of petrologists, defined verbally
and visually, and it can be used to formulate inferences automatically. Additionally, the use
of a knowledge-based approach provides users with high levels of interactivity. For these
reasons, Petroledge and Strataledge are exemplary tools for demonstrating the value of
ontology-based approaches.
Part 4 discusses practical solutions for managing ontology-based systems. Chapter 13
identifies ontology-based databases as a key tool for the management of semantic data. In
the case of earth model building, difficulties arise from the fact that many different forms of
technical knowledge are required, each described by one or more ontologies, which must
then be integrated or, at least, made compatible. Specific solutions are suggested for solving
these problems by using semantic annotation and ontology integration. Other issues are
associated with the need to formally describe and manage not only static technical knowledge but also the various activities at work in current earth modeling chains. Solutions are
proposed for solving this issue and for providing model makers with automated tools to
facilitate workflow management tasks. Chapter 14 further examines software tools for managing semantic data and suggests a tentative architecture for a platform of Web services
based on Semantic Web technologies. Finally, Chapter 15 provides a full-scale example of a
knowledge-based system for structural model building.
In Part 5, the editors draw general conclusions and evaluate the progress that has been
made over the past 12 years in knowledge-oriented earth modeling. Concrete solutions are
already available for solving a number of issues associated with this ambitious goal but they
must still be integrated in functional workflows that are accepted and implemented by industry professionals. Trends for future research are also described.
A multidisciplinary subject such as earth modeling requires the examination of issues
that are relevant to diverse scientific fields. We realize that readers may confront a number
of issues with which they are unfamiliar. We have tried to address this difficulty by avoiding
unnecessary technical detail whenever possible. At the end of each chapter, we also provide
references for readers who would like a more in-depth approach to specific issues or simply
want to familiarize themselves with new areas of study. A glossary of all the technical terms
used in this book is available at the end of the volume. Finally, we have tried to make this
book accessible to anyone with a good scientific background and a minimum of curiosity.
The Editors
Michel Perrin, Jean-Franois Rainaud