You are on page 1of 0

40 Oileld Review

Resolving Carbonate Complexity


Assessing basic rock properties using traditional logging suitesusually a straight-
forward process in sandstone reservoirsmay be difcult or impossible in carbonate
reservoirs. Also, when dealing with carbonates, determining optimal locations for
new wells from petrophysical analysis often becomes little more than a statistical
exercise. However, new tools, techniques and interpretation methodologies are
helping petrophysicists unravel the complexities posed by carbonate reservoirs.
Equipped with this information, operators are able to drill and produce these reser-
voirs while better managing uncertainty.
Mariam Ibrahim Al-Marzouqi
Sultan Budebes
Emad Sultan
Abu Dhabi Marine Operating Company
Abu Dhabi, UAE
Iain Bush
Gatwick, England
Roger Grifths
Kais B.M. Gzara
Raghu Ramamoorthy
Abu Dhabi, UAE
Alexis Husser
Sugar Land, Texas, USA
Ziad Jeha
Juergen Roth
Ahmadi, Kuwait
Bernard Montaron
Beijing, China
Srinivasa Rao Narhari
Sunil Kumar Singh
Kuwait Oil Company
Ahmadi, Kuwait
Xavier Poirier-Coutansais
Mabruk Oil Company
Tripoli, Libya
Oileld Review Summer 2010: 22, no. 2.
Copyright 2010 Schlumberger.
For help in preparation of this article, thanks to Lisa
Stewart, Cambridge, Massachusetts, USA; and Joelle Fay,
Gatwick, England.
AIT, Carbonate Advisor, DeepLook-CS, EcoScope, ECS,
FCM, FMI, HRLA, Litho-Density, MD Sweep, Petrel, Q-Land,
Sonic Scanner and SpectroLith are marks of Schlumberger.
31424schD7R1.indd 40 8/10/10 10:19 PM
Summer 2010 41
Characterizing and evaluating carbonate reser-
voirs from conventional logging data can be
daunting. Traditional approaches that work per-
fectly well for determining basic petrophysical
properties in siliciclasticssuch as porosity,
saturation, permeability and rock mechanical
propertiesmay yield inaccurate results in car-
bonates. In addition to the difculties in evaluat-
ing rock properties, many carbonates have lateral
structural heterogeneities; rock properties vary
greatly across the eld. Drilling to maximize pro-
duction can thus become a statistical exercise:
Drill enough wells and some will be successful.
Experts estimate that 60% of the worlds oil
reserves, as well as vast quantities of natural gas,
lie in carbonate reservoirs. The rewards for deci-
phering these enigmatic formations are very
attractive. But to do so, petrophysicists and engi-
neers who evaluate and produce hydrocarbons
from carbonates have learned that they must use
methods that differ substantially from those used
for sandstones. Fortunately, new tools are avail-
able that increase analysts reservoir understand-
ing and decrease risks associated with eld
development and reservoir management.
This article describes several recently intro-
duced techniques, beginning at the drill bit and
extending to eldwide seismic studies that strive
to clarify carbonate complexity. Included are
advances in logging-while-drilling (LWD) technol-
ogy that help geologists overcome difculties they
encounter evaluating carbonates when using con-
ventional logging suites. We also review an inte-
grated software workow that addresses
characteristics unique to carbonates. In addition,
a seismic workow method is presented that, com-
bined with other data sources, identies high-
quality reservoir sections by detecting fracture
corridors. Case studies from the Middle East dem-
onstrate applications of these techniques.
The Problem with Carbonates
Carbonate sediments differ from siliciclastics in
nearly every aspect: origin, deposition, diagene-
sis, oil lling and evolution.
1
Because abundant
examples exist in the literature describing these
differences, it might seem that carbonates are so
well understood that new techniques would pro-
vide only incremental assistance in their evalua-
tion. However, the problems experienced by log
analysts evaluating carbonates still provide sig-
nicant opportunities for the development of
new technologies and interpretation methods.
The problem is not that carbonates are poorly
understood; geologists and petrophysicists have
been studying and describing them since the
dawn of the oil industry. They have developed
numerous classication systems that focus on
particular carbonate peculiarities, such as tex-
ture, pore size and internal rock structure
(above).
2
These efforts, however, do not equate to
understanding specic reservoir rock properties
in a given well or eld.
Difculties begin with quantifying basic in
situ mineral, uid and textural properties using
conventional logging tools. Petrophysicists use
these log data to characterize and identify quality
reservoir rocks and guide drillers to the best pro-
ducing zones. Because of the complexities of car-
bonate reservoirs, evaluation programs often rely
on conventional coring to decipher heterogene-
ities in rock properties. Coring provides lithology,
qualitative and quantitative estimation of porosity
1. For more on carbonates and carbonate evaluation:
Akbar M, Petricola M, Watfa M, Badri M, Charara M,
Boyd A, Cassell B, Nurmi R, Delhomme J-P, Grace M,
Kenyon B and Roestenburg J: Classic Interpretation
Problems: Evaluating Carbonates, Oileld Review 7,
no. 1 (January 1995): 3857.
Akbar M, Vissapragada B, Alghamdi AH, Allen D, Herron M,
Carnegie A, Dutta D, Olesen J-R, Chourasiya RD, Logan D,
Stief D, Netherwood R, Russell SD and Saxena K:
A Snapshot of Carbonate Reservoir Evaluation,
Oileld Review 12, no. 4 (Winter 2000/2001): 4260.
Ahr WM, Allen D, Boyd A, Bachman HN, Smithson T,
Clerke EA, Gzara KBM, Hassall JK, Murty CRK, Zubari H
and Ramamoorthy R: Confronting the Carbonate
Conundrum, Oileld Review 17, no. 1 (Spring 2005): 1829.
2. For more on carbonate classication systems:
Scholle PA and Ulmer-Scholle DS: Carbonate
Classication: Rocks and Sediments, in Scholle PA and
Ulmer-Scholle DS (eds): A Color Guide to the Petrography
of Carbonate Rocks: Grains, Textures, Porosity, Diagenesis.
Tulsa: American Association of Petroleum Geologists,
AAPG Memoir 77 (2003): 283292.
>
Carbonate classication systems. The Dunham classication system (top), devised in 1964, is based on
rock texture and grain size. (Adapted from Akbar et al, 2000/2001, reference 1.) The Ahr classication
system (bottom), published in 2005, maps pore geometry and attempts to relate stratigraphy to eld-level
permeability predictions. (Adapted from Ahr et al, reference 1.) Although these parameters are important
for characterizing carbonate rock properties, neither classication system adequately describes key
reservoir storage capacity or ow characteristics.
TSFigure 01
Mudstone Wackestone Packstone Grainstone Boundstone Crystalline
Less than
10% grains
More than
10% grains
Grain supported Lacks mud,
grain supported
Original
components
bound together
Depositional
texture not
recognizable
Mud supported
Contains mud, clay and fine silt-size carbonate
Original components not bound together during deposition
Depositional texture recognizable
Depositional
Hybrid 1
Hybrid 2
Hybrid 3
Fracture Diagenetic
Reduced
Compaction
Cementation
Replacement
Diagenesis influences
brittle behavior.
Depositional character
influences fractures.
Depositional
aspects dominate.
Enhanced
Dissolution
Replacement
Recrystallization
Porosity
Size and shape
Vugs separate
Vugs touching
Diagenetic
aspects
dominate.
Interparticle
Intraparticle
Fenestral
Shelter or keystone
Reef
Intraskeletal
Interskeletal
Stromatactis vugs
Constructed voids
Detrital infill
31424schD7R1.indd 41 8/10/10 10:19 PM
42 Oileld Review
and permeability, and invaluable fracture infor-
mation. Even when rock properties are quantied
for a particular well, measurement analogs
beyond the near wellbore may not be valid at res-
ervoir scales because of the inherent heterogene-
ity and diagenetic history of the carbonates within
the eld.
Petrophysicists must overcome a number of
difculties when evaluating carbonates. To begin
with, carbonates differ from sandstones in that
they often have some type of organic origin and are
more susceptible to chemical and mechanical
reactions. They usually consist of skeletons and
shells of animals that settled near where they
livedtypically in warm, shallow marine environ-
ments. Those biological structures were built from
the calcium carbonate the animals extracted from
seawater. The climatic conditions, the types of
organisms and the manner in which they existed
in their ecosystem all contribute to the reservoir
heterogeneity of carbonate structures.
By contrast, the particles that make up sand-
stone and mudstone deposits may travel thou-
sands of kilometers to reach their nal resting
place. Their size, shape and sorting have much to
do with the energy of the depositional environ-
ment. Because carbonate sediments usually are
not transported far from their source, these depo-
sitional characteristics are not nearly as impor-
tant. And, although most carbonate reservoirs
are biogenically sourced, deepwater carbonate
accumulations and precipitations that are not of
biological origin have also been discovered.
These can cover wide expanses and also act as
hydrocarbon traps.
When the skeletal remains of biogenic car-
bonates stay where the organism lived, such as
coral or algal reefs, geologists refer to these accu-
mulations as autochthonous.
3
Lacking the inter-
granular permeability of clastics, these structures
usually require additional internal connectivity
to be productive, most often in the form of natu-
ral fractures (above). In contrast, allochthonous
carbonate deposits are composed of transported
shells and skeletal remains or bioclastic frag-
ments eroded from reworked deposits.
Once the carbonate fragments come to rest,
they eventually become cemented together, gen-
erally with calcite, in a process of lithication.
Because these deposits can consist of ne-
grained particles or broken shell fragments, they
may have clastic characteristics similar to those
of sandstone. During lithication, the deposits
often undergo chemical and biological diagene-
sis, which produces metastable compounds that
are susceptible to change (see Diagenesis and
Reservoir Quality, page 14). After deposition,
these rocks can become radically altered through
diagenesis, which can enhance hydrocarbon stor-
age and production capacity (porogenesis) or
destroy it (poronecrosis).
The most abundant carbonate form is cal-
cium carbonate, or calcite [CaCO
3
]. A less stable
polymorph, aragonite, has the same chemical
composition. Calcite is one of the more common
minerals on Earth, accounting for 4% by weight
of the Earths crust. Its chemical instability
makes it susceptible to transformation into
other mineral types.
4
Siderite [FeCO
3
] can form
when calcite is exposed to iron. Various other
carbonate varieties exist, each having character-
istic physical properties that affect matrix den-
sity and texture. The two most common
carbonate reservoir rocks are limestone and
dolomite. Limestone refers to the sedimentary
rock form that contains calcite, although these
two terms are often used interchangeably.
Determining the correct lithologybe it lime-
stone, dolomite or a combination of mineralsis
an important step in carbonate reservoir evalua-
tion.
5
Lithology establishes the matrix density, or
grain density, used for computing porosity from
density tools. It is also an input for other porosity
measurements, such as those from thermal and
epithermal neutron measurements. An accurate
porosity value is a crucial input for calculating
water and hydrocarbon saturations, determining
total uid volumes and estimating reserves.
>
Complexity of carbonates. The carbonate matrix often tends to be complex and is composed of
varying concentrations of limestone, dolomite and other minerals. Vuggy facies may make up a
signicant portion of carbonate reserves. Wells with connectivity through vug-to-vug contact in
fracture networks generally are more prolic producers than wells with matrix permeability alone.
(Core slab photograph courtesy of the Whiting Petroleum Corporation, used with permission.)
TSFigure 02
>
Matrix effects on density-porosity measurements.
Density porosity is computed using a value for
matrix density. If the input is unknown or incor-
rect, the density-porosity measurement error can
be substantial. For example, a 10% porosity
limestone has a bulk density of 2.539 g/cm
3
. If the
rock is dolomite, the porosity is 17% with that
same bulk density measurement. This 70% error
could be the difference between a commercial
well and abandonment.
TSFigure 03A
C
a
l
c
u
l
a
t
e
d

d
e
n
s
i
t
y

p
o
r
o
s
i
t
y
,

%
Limestone matrix
2.71 g/cm
3
Dolomite matrix
2.85 g/cm
3
20
15
10
5
0

density
= density porosity

matrix
= matrix density, or grain density

bulk
= bulk density measurement

fluid
= fluid density

matrix

density

bulk

matrix

fluid
=
70% error
31424schD7R1.indd 42 8/10/10 10:19 PM
Summer 2010 43
Measuring Basic Properties
Porosity is a basic petrophysical measurement,
usually obtained from well logs. It is commonly
computed from bulk density data. Density poros-
ity is sensitve to both the pore uids and the
matrix, especially the matrix. There are several
methods available for computing porosity, and
these often are affected by the uids in the rock
and the mineralogy. Depending on environmen-
tal conditions and operational constraints, inte-
grating these measurements plays a role in
decoupling the effects of the matrix on the
porosity value.
Examples of porosity measurements include
those from lithology-dependent thermal neu-
tron, lithology-independent neutron, acoustic,
thermal neutron capture spectroscopy and
nuclear magnetic resonance (NMR) tools.
Neutron and NMR porosity tools are blind to the
presence of gas, and NMR measurements are
also blind to porosity lled with tar, bitumen,
microporosity-bound water and hydrates.
In contrast to the NMR and neutron tools, bulk
density tools respond to both uid and lithology.
Density porosity (
density
) is computed using two
xed inputs, matrix density (
matrix
) and uid den-
sity (
uid
), and the bulk density measured by the
tool (previous page, bottom). The uid density
used in calculating porosity is that of the uid ll-
ing the pores of the formation, typically 1.0 g/cm
3
,
while the matrix density depends on the rock type.
The matrix density of limestone is 2.71 g/cm
3
, dolo-
mite is 2.85 g/cm
3
, siderite is 3.89 g/cm
3
and sand-
stone (quartz) is 2.65 g/cm
3
.
Uncertainty in lithology translates into large
errors in computed porosity. For instance, a 10%
porosity limestone formation has a measured bulk
density of 2.539 g/cm
3
. However, a dolomite matrix
could have the same measured bulk density but its
porosity would be 17%. If the rock type is not cor-
rectly identied, this signicant discrepancy in
the computed porositya 70% errormight be
the difference between commercial viability and
the decision to abandon a well.
The matrix may be a single mineral type but is
often a mixture. Small concentrations of minerals,
if unaccounted for, can introduce considerable
error in the computed porosity. A common noncar-
bonate mineral associated with limestone reser-
voirs, the evaporite anhydrite, has a bulk density of
2.98 g/cm
3
. Dispersed within the rock matrix, a
small percentage of anhydrite can signicantly
increase the measured bulk density. When the
anhydrite is found in the form of nodules, the mea-
sured porosity will be lower than the true value
because logging tools average the response from
both rock types (above). The formation may
appear to be of poor quality, although the carbon-
ate portion may, in fact, have good porosity and
permeability but be masked by the anhydrites
effects on the measurement.
6
Low-porosity carbonates with heavy minerals,
such as anhydrite, are emerging as major sources of
bypassed hydrocarbons. Understanding the manner
in which these minerals affect porosity measure-
ments and reservoir producibility is crucial for
geologists who study carbonates. Core analysis
often becomes a major factor in determining
commerciality of a eld. Logging data lack the
ne resolution of core analysis, but they provide a
continuous record of petrophysical properties
such as porosity and lithology.
Complexity, Texture and Relative Permeability
Perhaps the most common lithology-determina-
tion method from logging data uses the photo-
electric effect (PEF) measurement, which
responds primarily to the minerals in the forma-
tion. This measurement is routinely acquired
using formation density devices, such as the
Litho-Density and LWD density tools.
7
Although
useful in differentiating pairs of minerals among
sandstone, limestone, dolomite and anhydrite,
additional measurements are required when
more than two minerals are present. Also, the
measurement is affected by barite in drilling-mud
systems, and borehole conditions such as thick
mudcake and hole rugosity may render it useless.
A better method for solving complex litholo-
gies and determining mineralogical concentra-
tions, which may vary widely across a eld
depending upon the diagenetic history and uids
percolating through the reservoir, is an elemental
thermal neutron capture spectroscopy measure-
ment. For example, the ECS elemental capture
spectroscopy and the LWD EcoScope tools offer
this type of measurement.
8
These tools measure
the concentrations of specic elements that cor-
respond to mineralogy. Various matrix properties
3. Vernon RH: A Practical Guide to Rock Microstructure.
Cambridge, England: Cambridge University Press
(2004): 3437.
4. There is disagreement on how dolomite forms in nature;
some scientists suggest that biogenic origins are the
primary source. For more on dolomite: Al-Awadi M,
Clark WJ, Moore WR, Herron M, Zhang T, Zhao W,
Hurley N, Kho D, Montaron B and Sadooni F: Dolomite:
Perspectives on a Perplexing Mineral, Oileld Review 21,
no. 3 (Autumn 2009): 3245.
5. For more on difculties with carbonate reservoir
evaluation: Ramamoorthy R, Boyd B, Neville TJ,
Seleznev N, Sun H, Flaum C and Ma J: A New
Workow for Petrophysical and Textural Evaluation
of Carbonate Reservoirs, Petrophysics 51, no. 1
(February 2010): 1731.
>
Mineralogical effects. Anhydrite is just one of many minerals found within
carbonate reservoir rocks. The manner in which this mineral is dispersed may
affect uid ow in the reservoir. It may also impact the porosity measurement.
In the case of anhydrite nodules, the porosity of the reservoir rocks tends to
be underestimated and uid ow is not greatly affected (core photograph,
right). If the anhydrite is dispersed within the pore structure (micrograph, left),
the porosity measurement will be reduced, as will uid ow. (Adapted from
Ramamoorthy et al, reference 5.)
TSFigure 04
Pore-filling anhydrite
Anhydrite nodule
6. Ramamoorthy et al, reference 5.
7. The PEF is a log of photoelectric absorption (P
e
) properties
of the rock matrix that is acquired along with formation
density measurements. Common minerals encountered in
oil and gas wells have specic P
e
values: sandstone (1.9),
dolomite (3.1), limestone (5.1) and anhydrite (5.0).
8. Japan Oil, Gas and Metals National Corporation (JOGMEC),
formerly Japan National Oil Corporation (JNOC), and
Schlumberger collaborated on a research project to
develop LWD technology that reduces the need for
traditional chemical sources. Designed around the
pulsed neutron generator (PNG), EcoScope service uses
technology that resulted from this collaboration. The
PNG and the comprehensive suite of measurements in
a single collar are key components of the EcoScope
service that deliver game-changing LWD technology.
31424schD7R1.indd 43 8/10/10 10:19 PM
44 Oileld Review
can also be computed from the yields, including
grain density.
9
Grain density represents an effec-
tive matrix density and varies according to the
elements present in the formation. It yields more-
accurate density porosity than when computed
using a xed-value matrix density.
Texture and pore geometry are also important
properties for identifying reservoir-quality rock
because knowledge of correct mineralogy and
porosity measurement alone is not sufcient to
infer ow characteristics in carbonate reservoirs.
In fact, some experts believe that characteriza-
tion of pore geometry is the most important com-
ponent in carbonate evaluation.
10
Complex pore
shapes and sizes often result from reservoir depo-
sition and the ensuing processes of dissolution,
precipitation and recrystallization. Although
time-consuming, core analysis can reliably iden-
tify and quantify pore geometry. The standard
resistivity and porosity measurements of a triple-
combo logging suite often do not respond to
changes in pore size and texture. NMR data, how-
ever, have been shown to identify changes in pore
size distribution not detectable by these conven-
tional logs (left).
To better evaluate reservoir rock quality using
logging data, experts developed a technique for
characterizing carbonate pore geometry by parti-
tioning the total porosity measurement into three
classes of pore spaces based on sizemicro-
(less than 0.5 microns), meso- (0.5 to 5 microns)
and macroporosity (larger than 5 microns). From
these partitions, reservoir quality and uid-ow
properties are inferred.
11
Partitioning of forma-
tion porosity by pore size uses specic ranges of
transverse relaxation times, T
2
, from NMR data.
12

Core data are often used to rene T
2
measure-
ment ranges (left).
Another partitioning method maps relative
pore geometry into eight rock classes (next page,
bottom left).
13
The resulting ternary diagram was
rst developed through systematic analysis of
texture-sensitive borehole logs, which included
NMR data, borehole images, full-waveform acous-
tic logs and dielectric data.
14
A similar ternary
diagram has been derived from mercury injection
capillary pressure (MICP) tests on core.
For macroporosity evaluation, geophysicists
have recently begun to use acoustic data, such as
those from the Sonic Scanner tool, to estimate
the fraction of vuggy porosity. One application of
these data is to ne-tune the cementation expo-
nent, m, in Archies water saturation equation.
Vugs tend to increase the cementation exponent,
while large intergranular pores do not. Use of
macroporosity fractions from NMR data alone
>
Pore size and geometry. Measurements from NMR logging tools are more
sensitive to pore size and geometry than are resistivity and other porosity
measurements. The gamma ray log (Track 1), resistivity logs (Track 2) and
porosity measurements (Track 3) are consistent throughout the interval shown.
The NMR data (Track 4) indicate a large increase in pore size above X,040 ft
that is not seen in the other measurements. (Adapted from Ramamoorthy et al,
reference 5.)
T
2
Distributions
Depth,
ft
X,050
X,000
0 100 gAPI
Gamma Ray
6 16 in.
Caliper
6 16 in.
Bit Size
0.1 1,000 ohm.m
Array 1
Array 2
Array 3
Array 4
Array 5
R
xo
Resistivity
45 15 %
Neutron Porosity
45 15 %
Array Porosity
3 13
PEF
1.95 2.95 g/cm
3
Bulk Density
0.3 6,000 ms
T
2
Log Mean
>
NMR porosity partitioning. When NMR logging tools were introduced to
the oil industry, the T
2
distributions were scaled as pore sizes. For a number
of reasons, this practice was abandoned. However, the concept works fairly
well for carbonates. Pore sizes are determined according to a range of T
2

distributions, and then the porosity is partitioned into macro-, meso- and
microporosity based on these measurements. The longest T
2
distributions
correspond to macroporosity, large pores and vugs. The shortest T
2

distributions respond to microporosity. Oil migrating into water-lled rock
displaces water in macro- and mesopores rst. Micropores generally remain
water lled.
TSFigure 07
Total porosity
Oil in place
0.5
microns
5
microns
Mesoporosity Macroporosity Microporosity
Porosity
below short
T
2
cutoff
NMR T
2
response
Porosity
above long
T
2
cutoff
31424schD7R1.indd 44 8/10/10 10:19 PM
Summer 2010 45
can result in elevated estimates of m because the
measurement is based on pore size, not shape.
Combining vuggy porosity estimates from full-
waveform acoustic data improves log-derived
estimations of the m exponent.
NMR data are also used to compute permea-
bility. The technique evolved from empirically
derived relationships, which work well in sand-
stones but are not always relevant in carbonates
because the pores may not be connected.
Relative permeabilities and fractional ow in
hydrocarbon zones may, however, be derived
from array resistivity log data when the well is
drilled with water-base mud.
15
The invading mud
ltrate acts as an uncontrolled two-phase ow
experiment that can be analyzed in a manner
similar to relative permeability measurements
conducted on core.
This mud-ltrate invasion method not only
provides information about in situ fractional ow
and relative permeabilities, it also improves the
accuracy of formation resistivity measurements
and water saturation estimates. The processing
involves forward modeling based on relative
permeability parameterization, radial invasion
models, petrophysical models and tool response
to specic conditions. The inputs required for
computing water saturations using Archies equa-
tionformation water and bulk formation resis-
tivitiesare more accurate when obtained using
this method, as are the ultimate computed uid
volumes. Even so, log analysts have discovered
that Archies equation may not be as reliable for
characterizing uids in carbonate reservoirs as it
is in sandstones.
Whats Wrong with Archie?
In 1942 Gus Archie laid the foundation for mod-
ern log interpretation by introducing a relation-
ship linking water resistivity, formation porosity
and formation resistivity to uid saturation
(right). Variables in the equationa, m and n
are empirically t based on reservoir characteris-
tics. In the absence of specic data they are
generally assumed to equal 1, 2 and 2, respec-
tively.
16
Assumptions in the formulamorphol-
ogy of the pore space, connectivity of the pores
and wettability of the rockare best suited to
>
A ternary diagram based on pore size. Carbonate pore geometry and size are
inputs to this ternary diagram, which indicates reservoir quality. On the lower left
side of the triangle, permeability is a function of grain size. For the upper section,
permeability is controlled by the volume of macropores. On the lower right, the
permeability is a function of both grain and pore size.
TSFigure 06
k = 0.35
2
( T2LM

)
2
Carbonate rocks with intergranular
T2LM is the
logarithmic mean of
the T
2
measurement.
porosity (no macroporosity)
Permeability, k, is controlled by
porosity and the average pore
(grain) size. k = 1.0 V
macro
/(V
meso
+ V
micro
) ] [
2
Carbonate rocks with abundant
macroporosity
We
pore throats)
ll-connected pores (large
Permeability is controlled by
porosity and the volume of
macroporosity (V
macro
).
100%
microporosity
100%
mesoporosity
Carbonate Pore System Classes and Permeability
100%
macroporosity

2
>
Archies water saturation equation (bottom).
Porosity and R
t
are log-derived measurements.
R
w
is either derived from water salinity or
measured from produced water and converted to
downhole temperature. Variables a, m and n are
empirically t based on reservoir characteristics.
They are assumed equal to 1, 2 and 2, respectively,
in the absence of specic data. A sensitivity
analysis (top) demonstrates the effects of varying
m and n on computed water saturation. First, n
is set to 2 and m is varied from 2.3 to 1.7 (Track
1). Next, m is xed and n is varied from 2.5 to 1.0
(Track 2). The baseline water saturation curve
using default inputs for m = n = 2 is presented in
both tracks (red curve). (Adapted from Grifths et
al, reference 17.)
TSFigure 06A
S
w
= Archies water saturation
R
w
= resistivity of formation water
R
t
= true formation resistivity
a = formation-factor multiplier
= porosity
m = cementation exponent
n = saturation exponent
%
n = 2,
m = 2.3 to 1.7
100 0 %
m = 2,
n = 2.5 to 1.0
100 0
Water Saturation Water Saturation
S

R
R a
m
=
t
w
w
n
9. For a thorough review of neutron capture spectroscopy:
Barson D, Christensen R, Decoster E, Grau J, Herron M,
Herron S, Guru UK, Jordn M, Maher TM, Rylander E
and White J: Spectroscopy: The Key to Rapid, Reliable
Petrophysical Answers, Oileld Review 17, no. 2
(Summer 2005): 1433.
10. Archie GE: Classication of Carbonate Reservoir Rocks
and Petrophysical Considerations, AAPG Bulletin 36,
no. 2 (1952): 278298.
11. Hassall JK, Ferraris P, Al-Raisi M, Hurley JF, Boyd A and
Allen DF: Comparison of Permeability Predictors from
NMR, Formation Image and Other Logs in a Carbonate
Reservoir, paper SPE 88683, presented at the Abu Dhabi
International Petroleum Conference and Exhibition,
Abu Dhabi, UAE, October 1013, 2004.
12. In NMR logging, transverse relaxation time, T
2
, results
from interactions of hydrogen atoms with their
surroundings, including effects of bulk uids, pore
surfaces and diffusion in magnetic eld gradients.
Short T
2
times correspond to small pores, and longer
T
2
times correspond to larger pores.
13. Hassall et al, reference 11.
14. Ramamoorthy et al, reference 5.
15. For more on this technique: Ramakrishnan TS,
Al-Khalifa J, Al-Waheed HH and Cao Minh C:
Producibility Estimation from Array-Induction Logs
and Comparison with MeasurementsA Case Study,
Transactions of the SPWLA 38th Annual Logging
Symposium, Houston, June 1518, 1997, paper X.
16. The a constant, a tortuosity or consolidation factor,
was not in Archies original equation but was added
later as a means of correcting for saturation in known
water-lled reservoir rocks. For more on this subject:
Archie GE: The Electrical Resistivity Log as an Aid in
Determining Some Reservoir Characteristics,
Petroleum Transactions of AIME 146 (1942): 5462.
Winsauer WO, Shearin HM, Masson PH and Williams M:
Resistivity of Brine Saturated Sands in Relation to Pore
Geometry, AAPG Bulletin 36, no. 2 (1952): 253277.
31424schD7R1.indd 45 8/10/10 10:19 PM
46 Oileld Review
siliciclastic rocks.
17
Although most water satura-
tion methods utilize some form of Archies equa-
tion, it is generally recognized that there are
problems with this approach when applied to
carbonates. Even Gus Archie stated that he
doubted the applicability of his equation in car-
bonate evaluation.
18
In addition, the complex nature of carbonates
makes determination of the a, m and n variables
difcult, and these values may change rapidly
throughout the reservoir.
19
Other problems with
using Archies saturation equation in carbonates
include matrix complexity, pore size heterogene-
ity, pore shape and distribution, variability in for-
mation water salinity and uncertainty in the true
formation resistivity measurement.
The process of lling the reservoir creates
some of the difculties encountered when using
Archies water saturation equation: Water lls
the pores initially and then hydrocarbons enter,
charging the complex carbonate structure. The
macropores ll rst, because they have the low-
est capillary entry pressure. A proportion of the
mesopores ll next and, because of capillary
pressure, micropores may remain water lled. As
a result of the basic nature of carbonate grain
surfaces, there is an afnity for crude oil, which
typically contains acidic components. Hence, the
pores that ll with oil may become oil wet, while
micropores that never ll with oil remain water
wet. This results in a mixed-wettability rock.
Moved by natural or injected water sweeping
through producing elds or by ltrate during
drilling, reservoir uids are displaced in the larg-
est pores rst. Because of the altered wettability
in the rock, these pores present the least resis-
tance to the ingress of the uids. Fluid capillary
effects and differences between the original
charging pressure and reservoir pressure during
production may result in some of the mesopores
remaining oil lled even as the macro- and micro-
pores are water lled. This creates a complex
uid distribution inside the pore network. Thus,
Archie parameters are different for the invaded
rock of the near-wellbore area than for the unin-
vaded zones of the same rock (above).
The complex wettability of carbonates makes
use of Archies saturation equation problematic
as well. Unlike sandstone reservoirs that are usu-
ally strongly water wet, most carbonate reservoir
rocks have some degree of moderate oil-wet char-
acter. Preferentially oil-wet surfaces, located on
the walls of meso- and macropores, have been in
contact with oil. This reduces the connectivity of
the water phase in the porous rock and contrib-
utes to an increase in the resistivity compared
with the value predicted by Archies equation.
On the other hand, micritic grainstightly
packed micron-size calcite crystals with sub-
micron poresare fully water saturated and
water wet and dramatically enhance the connec-
tivity of water in the medium. The effect of
micrite counteracts the effect of oil-wetness on
the rocks electrical properties. Carbonate rocks
with a large volume fraction of micrite may have
a resistivity similar to that of shaly sandstone
rocks. Carbonate rocks with little or no micritic
content, such as dolomite, may have a pro-
nounced opposite response typical of oil-wet
rocks. These resistivity behaviors can be modeled
by the connectivity equation.
20
In Archies saturation equation, the term for
formation water, R
w
, assumes a simple uid distri-
bution with a single value of formation water resis-
tivity. Complex uid distributions, such as mixed
ltrate or injection waters, are a departure from
>
Carbonate reservoir lling and resistivity measurements. Water (blue) originally lls the pore spaces of carbonate
reservoirs (left). As oil (green) migrates into the rock, large pores ll rst. If there is no connectivity, some pores may remain
water lled (center). Because resistivity tools measure through a path of least resistance (red line), the current may bypass
oil-lled pores (right), which will increase the measured resistivity. Thus the resistivity values may be substantially lower
than expected and not be representative of the true bulk resistivity.
TSFigure 08

density
, is the density porosity

matrix
, is the matrix density, or grain density

bulk
, is the tool measurement

fluid
, is the fluid density

matrix

density

bulk

matrix

fluid
=
Micropores
Mesopores Macropores Water-filled vug Path of least resistance
>
Sigma equation for water saturation. Standard values for the matrix sigma,
grain
, are shown (top),
although the measurement can be rened with spectroscopy data. Values for
water
can be calculated
using uid salinity, computed from log responses or directly measured from produced water samples.
This equation (bottom) provides a water saturation value that is not based on resistivity measurements.
TSFigure 09
Lithology
Sandstone = 4.3
Dolomite = 4.7
Calcite = 7.1
Anhydrite = 12
, cu 0 5 10 15 20 25 30 35
Clays
40 45 50
Fluid Gas Oil Fresh water Increasing salinity

density
, is the density porosity

matrix
, is the matrix density, or grain density

bulk
, is the tool measurement

fluid
, is the fluid density
S
w
= formation water saturation
= formation porosity

bulk
= measured formation capture cross section

water
= capture cross section of the water

grain
= formation grain capture cross section

HC
= hydrocarbon capture cross section

matrix

density

bulk

matrix

fluid
=

bulk

grain
w =
( )
grain

HC
( )

water

HC
( )

31424schD7R1.indd 46 8/10/10 10:19 PM


Summer 2010 47
the model. The true resistivity in the reservoir, R
t
,
can be difcult to measure as well. Water-lled
microporosity and mesoporosity provide paths of
lower resistance for the sensor current. Thus the
average bulk resistivity measured in the formation
is signicantly lower than that in rocks with identi-
cal porosity and uid saturations but with uni-
modal pore-throat or pore-body size distributions.
These occurrences, referred to as low-resistivity
pay formations, have led to underestimated hydro-
carbon reserves and bypassed hydrocarbons.
For these and other reasons, Archies satura-
tion equation is unlikely to be accurate for car-
bonates without making empirical adjustments
to the input variables. An alternative to Archies
equation derives saturation from the macro-
scopic thermal neutron capture cross section
measurement, or sigma (, measured in capture
units, cu), which has been used for cased hole
evaluation for many years. A pulsed-neutron gen-
erator (PNG) emits high-energy neutrons that
interact with the nuclei of the elements present
in the surrounding formation. Of the elements
generally found in the reservoir, chloride ions
[Cl

], primarily found in salt water, have the


greatest neutron capture capacity, also referred
to as capture cross section. The rate of neutron
capture is predominately a function of chloride
concentration, which can be related to the vol-
ume and salinity of the formation water.
Hydrocarbons have a low capture capacity, and as
long as there is sufcient salinity in the forma-
tion water to produce a usable sigma contrast
between hydrocarbons and water, sigma can be
used to compute water saturation.
Inputs for computing water saturation using
sigma are porosity and macroscopic capture cross
section for formation matrix (
grain
), formation
water (
water
), expected hydrocarbons in place
(
HC
) and the sigma measured by the tool (
bulk
)
(previous page, bottom). If the lithology is known,
matrix sigma can be input as a constant, or it can
be derived from the elemental thermal neutron
capture spectroscopy measurement in a manner
similar to determining grain density for porosity
calculations. The value of
water
can be measured
directly, estimated from downhole measurements
or calculated from the salinity of produced sam-
ples. Finally,
HC
, a constant used in the satura-
tion equation for the hydrocarbon type, is derived
from expected uid properties at downhole tem-
perature and pressure.
The depth of investigation of the sigma
measurement is quite shallow compared with
that of resistivity measurements. Thus the ability
to characterize the uninvaded portion of the res-
ervoir may be signicantly hindered because
mud ltrate invades the near-wellbore zone dur-
ing the drilling process. The sigma measurement
may respond primarily to the ltrate. As a conse-
quence, wireline sigma measurements acquired
in open hole have not proved useful for evaluat-
ing water saturation in the virgin zone. One
exception to this occurs when the invaded and
uninvaded zones remain similar, such as when
drilling in oil-bearing formations at irreducible
water saturation with oil-base mud. In this case
the time of the measurement does not matter, but
the assertion that the formation is at irreducible
water saturation must be validated.
Cased hole sigma logs have proved more
benecial than openhole logs because they are
acquired after the ltrate has dissipated. Even
so, the measurement may be degraded by the
effects of casing, cement and residual uids. This
has led to differences between saturations mea-
sured with cased hole tools and those derived
from openhole logs.
An alternative to openhole and cased hole
sigma measurements from wireline tools is sigma
measured using an LWD tool. Depending on such
factors as drilling rate of penetration (ROP), for-
mation porosity, formation permeability, mud
properties, mud pressure overbalance and the
elapsed time between the rst drilling in the for-
mation and the time of acquiring the sigma mea-
surement, the invaded zone may not extend into
the region of the measurements depth of investi-
gation. Acquiring data close behind the drill bit
and prior to invasion overcomes many of the limi-
tations of sigma acquisition using wireline meth-
ods. This capability has been available for several
years with the EcoScope tool, a multifunction
LWD service that combines resistivity sensors
with a PNG for sigma and sourceless thermal neu-
tron porosity logging (above). The EcoScope tool
17. Grifths R, Carnegie A, Gyllensten A, Ribeiro MT,
Prasodjo A and Sallam Y: Evaluation of Low Resistivity
Pay in CarbonatesA Breakthrough, Transactions of
the SPWLA 47th Annual Logging Symposium, Veracruz,
Mexico, June 47, 2006, paper E.
18. Grifths et al, reference 17.
19. Grifths et al, reference 17.
20. For more on wettability and carbonates, especially
modeling of resistivity: Montaron B: Connectivity
TheoryA New Approach to Modeling Non-Archie
Rocks, Transactions of the SPWLA 49th Annual Logging
Symposium, Edinburgh, Scotland, May 2528, 2008,
paper GGGG.
>
EcoScope LWD tool. The EcoScope tool incorporates resistivity, neutron
porosity, sigma and neutron capture spectroscopy sensors into a single
compact device. Wireline and LWD tools generally use chemical sources for
neutron porosity and neutron capture spectroscopy measurements. The
EcoScope tool generates neutrons with a pulsed-neutron generator that
operates only when mud is being pumped through the tool.
Phase resistivity
TSFigure 10
Attenuation resistivity
Sigma, sourceless neutron
porosity, spectroscopy and
neutron-gamma density
Azimuthal density
and PEF
31424schD7R1.indd 47 8/10/10 10:20 PM
48 Oileld Review
is considered sourceless because once power
which is generated from mud owing through the
toolis no longer applied to the PNG, it ceases to
emit neutrons. Conversely, chemical sources are
always on.
The neutron output from the PNG also makes
thermal neutron capture spectroscopy measure-
ments possible. Similar to the measurements
from the wireline ECS tool, the EcoScope spec-
trometry service delivers elemental yields of sili-
con [Si], calcium [Ca], iron [Fe], sulfur [S],
titanium [Ti], gadolinium [Gd], potassium [K],
hydrogen [H] and chlorine [Cl]. Although the
EcoScope tool was not able to differentiate lime-
stone from dolomite in the past, the tool response
was recently recharacterized to include a magne-
sium [Mg] measurement (below). The ability to
measure Mg is fundamental for distinguishing
dolomite from limestone. In barite-weighted mud
systems, this becomes a crucial measurement for
determining formation lithology because the PEF
measurement from a Litho-Density tool is ren-
dered unusable by the effects of the barite. In
complex mineralogy the spectroscopy measure-
ment helps identify mineral constituents and pro-
vides an effective matrix density, or grain density,
for more-accurate density-porosity computations.
Complex Middle East Carbonate
Recently the EcoScope tool was run in an offshore
Abu Dhabi carbonate eld.
21
Production from this
eld began in 1968 from Lower Cretaceous, Upper
Jurassic, Upper Permian and Lower Triassic for-
mations. In 2006 Total decided to drill and develop
the Late Triassic (Gulailah) and Lower Jurassic
(Hamlah) Formations, which had not been previ-
ously produced.
The Hamlah reservoir is 50 m [164 ft] thick
and comprises two intervals separated by shale.
The lower interval is a micro- to very ne-grained
crystalline dolomite interbedded with limestone
streaks. The upper interval grades between lime-
stone, wackestone to packstone, with some grain-
stone and dolomite. Porosity ranges from 6% to
8%, and permeability ranges from very low to low.
The Gulailah reservoir is 250 m [820 ft] thick,
with alternating dolomitic and anhydritic beds.
The dolomites are sucrosic to nely crystalline,
anhydritic and occasionally argillaceous. Porosity
ranges from 8% to 13% and permeability is low to
very low.
Deviated wells were drilled using 1.35-g/cm
3
[11.3-lbm/galUS] barite-weighted mud systems.
This barite signicantly degraded the PEF mea-
surement. The EcoScope tools spectroscopy
measurement was able to accurately distinguish
calcite from dolomite and provide the matrix
grain density.
Another common complication encountered in
evaluating deviated wellsespecially in carbon-
atesis resistivity anomalies caused by shoulder-
bed effects. These arise when the measurement
volume includes regions with large conductivity
contrasts. Electromagnetic averaging and charge
buildup along the interface between layers result
in polarization horns, seen as anomalous spikes in
the resistivity data (next page).
22
Although shoulder-bed effects are generally
small in vertical wells, for deviated and horizon-
tal wells these effects may be prominent in long
intervals as wells approach, intersect and depart
from layer boundaries. Resistivities affected by
shoulder beds can produce misleadingly high
hydrocarbon saturations when calculated using
Archies saturation equation.
>
Rening lithology determination. Standard SpectroLith processing (left)
cannot distinguish calcite from dolomite in the absence of a PEF or
magnesium measurement and assumes that all calcium is associated with
calcite. When lithology is computed using the PEF measurement from a
Litho-Density tool, the software is able to distinguish dolomite from calcite
(center), but the PEF measurement can be affected by barite in the drilling
uids and by hole conditions. The excessive anhydrite shown in the center
track is attributed to these effects. If more than two minerals are present,
the PEF measurement is less accurate. Spectroscopy that includes a
magnesium measurement (right) distinguishes dolomite from calcite and is
not affected by hole conditions and uid properties. Other minerals can be
accurately quantied as well.
Carbonate
Pyrite
Anhydrite-Gypsum
Clay
Quartz-Feldspar-Mica
Illite
Bound Water
Quartz
Anhydrite
Calcite
Dolomite
Illite
Bound Water
Quartz
Anhydrite
Calcite
Dolomite
Standard SpectroLith Calcite-Dolomite
from PEF Processing
Calcite-Dolomite from
Enhanced Spectroscopy
Oilfield Review
Autumn 10
CleanPhase Fig. 11
ORAUT10-CLNPSE Fig. 11
31424schD7R1.indd 48 8/10/10 10:20 PM
Summer 2010 49
The superiority of sigma-based saturation
measurements over conventional methods is
compromised in the presence of signicant mud-
ltrate invasion. Resistivity-response modeling
has shown that invasion less than 5 cm [2 in.] has
negligible effects on the sigma measurement.
Generally, because the measurement is taken so
close to the bit, the formation does not have time
to become signicantly invaded before the
EcoScope tool acquires data. The tools resistivity
sensor array, collocated with the sigma measure-
ment, can determine the degree of invasion in
the area sampled.
21. Grifths R and Poirier-Coutansais X: Complex
Carbonate Reservoir EvaluationA Logging While
Drilling Field Example, paper AA, presented at the
SPWLA Regional Symposium, Abu Dhabi, UAE,
April 1618, 2007.
22. Grifths and Poirier-Coutansais, reference 21.
>
Shoulder-bed effects on LWD resistivity measurements. Averaging of resistivity measurements affects the output at
bed boundaries. In wells drilled nearly perpendicular to the layering (top left), these effects tend to be localized as
the tool crosses a resistivity interface. Horizontal wells may cross multiple zones with large resistivity contrasts (top
right). In this situation, charges accumulate at the interface and induce a polarization horn, or spikeswhich are
dependent on the depth of investigationthat are not representative of the actual resistivity (middle). If not
accounted for during interpretation, the elevated resistivities produce misleadingly high hydrocarbon saturations
using Archies saturation equation. The sigma measurement (bottom) does not suffer from the polarization effect,
permitting a more accurate evaluation of the hydrocarbon saturation in high-angle wells.
TSFigure 12
1 ohm.m
50 ohm.m
R
e
s
i
s
t
i
v
i
t
y
,

o
h
m
.
m
5,000 5,010 5,020
Distance from boundary, ft
5,030 5,040
1,000
100
10
1
1 ohm.m 50 ohm.m
S
i
g
m
a
,

c
u
5,000 5,010 5,020
Distance from boundary, ft
5,030 5,040
1,000
100
10
1
1 ohm.m 50 ohm.m

+ +
1 ohm.m
50 ohm.m
31424schD7R1.indd 49 8/11/10 9:55 PM
50 Oileld Review
In the Total well, the preinvasion sigma from
the EcoScope tool provided a valid water satura-
tion measurement independent of formation
resistivity. As an added benet, petrophysicists
were able to determine appropriate inputs to
Archies water saturation equation to match the
sigma-based measurement. Because carbonate
reservoirs often have unknown R
w
values, simul-
taneously solving for water salinity provided a
realistic R
w
and
water
output that satised both
equations (above).
Sum Greater than Parts
The EcoScope approach provides answers about
uid saturations in carbonates, but a preinvasion
sigma measurement is often unavailable.
Recognizing the challenges in carbonate
evaluation, Schlumberger scientists devised a
workow for petrophysical and textural evalua-
tion that integrates standard wireline logging
suites with recently introduced measurements.
Several independent research efforts focusing on
discrete aspects of carbonate evaluation are com-
bined using this systematic methodology. The
workow evolved into the Carbonate Advisor soft-
ware program (next page, top left). Each step in
the workow provides a piece of the puzzle and
facilitates subsequent steps.
Petrophysicists applied this methodology to a
Cretaceous Middle East carbonate well that had
a comprehensive suite of wireline logs. The log-
ging program included array resistivity (both
induction and laterolog), gamma ray, density,
thermal and epithermal neutron, NMR, full-wave-
form acoustic, neutron capture spectroscopy and
microresistivity imaging tools.
The analysis hierarchy began with lithology
and mineralogy determinations from uid- and
matrix-sensitive data, including NMR informa-
tion, density and neutron porosity logs, PEF logs
and neutron capture spectroscopy data. The pet-
rophysicist can emphasize the importance of a
particular measurement based on its relevance
and the borehole environment to obtain a simul-
taneous solution that includes input from all
measurements.
23
In this case the mineralogy con-
sists predominantly of calcite with small amounts
of dolomite. Siliciclastic material and anhydrite
were also observed (next page, top right).
Elemental thermal neutron capture spectros-
copy data quantied the dolomite, anhydrite,
>
Improved Archies equation and sigma saturation measurements. Apparent formation salinity is computed assuming the
formation is 100% water saturated (Tracks 3 and 5, green curves). Apparent salinity from the spectroscopy chlorine/hydrogen
(Cl/H) ratio measurement (Tracks 3 and 5, blue curve) is presented for comparison. Archie saturation is calculated using n and m
exponents set to 2 and an R
w
based on the assumed salinity corrected for downhole conditions (Tracks 4 and 6, blue curve).
Sigma-based saturations (red curve) are computed using two different water salinities: 250 and 150 parts per thousand (ppt).
The red lines in Tracks 3 and 5 indicate the salinity input used for each analysis. The analysis using 250-ppt salinity water
(Tracks 3 and 4), which was the original assumption, exhibits a large separation between the two saturation solutions. Also, the
SpectroLith apparent salinity (blue curve) does not match the salinity used in the analysis (red line). For the 150-ppt salinity
analysis (Tracks 5 and 6), the SpectroLith apparent-salinity curve (blue) tracks the salinity value used in the analysis (red line),
and both saturation methods are in much closer agreement (Track 6). This simultaneous solution yields a more reliable saturation
measurement and a more reasonable choice for formation-uid salinity. Note the lack of separation between deep and shallow
resistivities (Track 1) indicating shallow invasion and acceptable sigma measurement. Neutron and density porosities, adjusted
for matrix lithology from spectroscopy data, are also presented (Track 2). (Adapted from Grifths and Poirier-Coutansais,
reference 21.)
Oilfield Review
Autumn 10
CleanPhase Fig. 13
ORAUT10-CLNPSE Fig. 13
Resistivity Matrix-Adjusted Porosity
Neutron Porosity
Density Porosity
Total Porosity
0.2 2,000 ohm.m 50 0 % 400 ppt 4
SpectroLith Apparent Salinity
Sigma Apparent Salinity
250-ppt Salinity
a = 1, m = n = 2
100 % 0
Water Saturation
(Sigma)
Water Saturation
(Archie)
400 ppt 4
SpectroLith Apparent Salinity
Sigma Apparent Salinity
150-ppt Salinity
a = 1, m = n = 2
100 % 0
Water Saturation
(Sigma)
Water Saturation
(Archie)
50 0 %
50 0 %
400 ppt 4
100 % 0 100 % 0
Free Water
Irreducible Water
Clay-Bound Water
Free Water
Irreducible Water
40-in. Blended LWD Tool
40-in. 2-MHz Phase Shift
28-in. 2-MHz Phase Shift
16-in. 2-MHz Phase Shift
400 ppt 4
Clay-Bound Water
31424schD7R1.indd 50 8/10/10 10:20 PM
Summer 2010 51
quartz and clay (illite) volumes to generate an
effective grain density, allowing an accurate
porosity to be obtained.
The lithology-corrected porosity was next par-
titioned into pore geometry components based
on NMR data, which were ne-tuned with bore-
hole image and full-waveform acoustic data. In
contrast to the lithology and mineralogy, the pore
geometry was highly variable, with zones contain-
ing signicant amounts of macroporosity inter-
spersed with zones dominated by mesoporosity
and lesser amounts of microporosity (left).
>
Integrated carbonate solution. This owchart shows the workow sequence
for analyzing carbonate reservoirs using Carbonate Advisor software.
TSFigure 14
Density, PEF, neutron,
NMR, spectroscopy
NMR, borehole images,
acoustic data
Formation testers
NMR pore size
transforms
Resistivity, sigma,
dielectrics, 3D NMR data
Array resistivities,
formation tester data
Lithology, porosity,
fluid type
Input Data Outputs
Porosity partitioning
Permeability
Petrophysical
rock types
I
n
t
e
g
r
a
t
e
d

c
a
r
b
o
n
a
t
e
e
v
a
l
u
a
t
i
o
n
Capillary pressures
Fluid saturations
Fractional flow
>
Lithology dened by the ECS tool. The
measurement principle for neutron capture
spectroscopy is the same for both the ECS and
the EcoScope tools; the difference is the neutron
source. The ECS sonde has a chemical source
and the EcoScope tool uses a pulsed-neutron
generator with a higher neutron output.
Traditional methods for determining lithology use
PEF data from a Litho-Density tool (left). This
method is best suited for two-mineral models. By
adding elemental yield data from the ECS tool
(right), the lithology can be rened, providing a
more accurate density-porosity measurement
because the grain density reects the true
mineralogy. The porosity difference between
using a xed limestone matrix density value and
an effective grain density computed from ECS
mineralogy is presented (Track 2, orange
shading). (Adapted with permission of the
SPWLA from Ramamoorthy et al, reference 5.)
Anhydrite
Calcite
Dolomite
Illite
Dolomite
Calcite
Anhydrite
Quartz
Bound Water
Porosity Correction
Oilfield Review
Autumn 10
CleanPhase Fig. 15
ORAUT10-CLNPSE Fig. 15
23. Ramamoorthy et al, reference 5.
>
Porosity partitioning of NMR data. The distribution of T
2
transverse relaxation time data (Track 1) from
the NMR tool is partitioned based on cutoffs that can be rened from core analysis. In this example
volumes computed from distributions to the left of the red line (Track 1) represent microporosity, which
correspond to the blue shaded volume in Track 2. Microporosity measurements from core are plotted
along with the microporosity volume for conrmation. The area between the red and blue lines in Track 1
is mesoporosity, corresponding to the green shading in Track 2. The macroporosity (red shading) is
associated with remaining porosity (Track 1, right of the blue line). Permeability from core data is
plotted with permeability computed from NMR data (Track 3). The free-uid volume computed from
NMR data can be similarly partitioned (Track 4). Fluid volume to the right of the cutoff (blue line) is
associated with mesoporosity, and the volume to the left is macroporosity. Core data points agree with
computed data. (Adapted from Ramamoorthy et al, reference 5.)
Depth,
ft
0.5 50,000 ms
50 % 0
Total Porosity
50 % 0
Core Microporosity
0.5 50,000 ms
X,500
X,600
0.1 10,000 mD
Core Permeability
0.1 10,000 mD
Computed Permeability
30 % 0
Core Macroporosity
30 % 0
Macroporosity Cutoff
30 % 0
Free Fluid, NMR
Microporosity
Mesoporosity
Macroporosity
T
2
Distributions
T
2
Cutoff Short
T
2
Cutoff Long
Oilfield Review
Autumn 10
CleanPhase Fig. 16
ORAUT10-CLNPSE Fig. 16
31424schD7R1.indd 51 8/10/10 10:20 PM
52 Oileld Review
The partitioned porosity from NMR data had
good correlation with data from MICP test
results. Analysts next used the partitioned poros-
ity to estimate permeability. These log-derived
values compare well with minipermeameter
probe measurements made on core plugs.
Relative permeability and uid saturations
were computed using both array induction and
array laterolog resistivity measurements. Because
of the high salinity of the borehole uid, the induc-
tion measurement was unreliable at high resistivi-
ties in the main hydrocarbon section. The laterolog
data are preferred in these zones.
Drainage capillary pressures were also com-
puted based on NMR data transforms.
24
Because
the NMR data provide pore size from T
2
distribu-
tions, assuming bulk and diffusion effects are
minimal, by integrating the T
2
distribution, a cap-
illary pressure versus saturation relationship can
be developed. To convert T
2
data to capillary pres-
sure, a small calibration constant is required.
This constant is obtained by comparing the NMR
data with MICP measurements taken from simi-
lar core samples. Using the Carbonate Advisor
program, the analyst manually determines the
constant by comparing MICP entry pressures
with those computed from NMR log data.
The integrated approach of the Carbonate
Advisor software provides comprehensive evalua-
tion of key properties that describe reservoir
storage capacity and ow characteristics (above).
The software follows a set workow, but through-
out the process the petrophysicist has interactive
control over how data are input, a particularly
useful feature when measurement conditions
may be less than optimal.
>
Integrated output. Shown is the nal product from the Carbonate Advisor
program. These outputs provide an integrated and comprehensive
evaluation of the key properties that describe a reservoirs storage and ow
capacity. The petrophysicist may weight the data from specic tools and
choose between tools (Depth track, AIT array induction imager tool, green;
and HRLA high-resolution laterolog array, gold). Complex lithology and uid
volumes (Track 1) are shown along with a moved-hydrocarbon analysis
(orange) from microresistivity data. Fluid-ow models are constructed from
resistivity data (Track 2). Porosity from NMR data (Track 3) are partitioned
and the results graphically displayed (Track 4). A full ternary analysis (Track 5)
is useful for identifying better quality reservoir rock. Drainage capillary
pressures are computed from NMR pore geometry data, adjusted to match
MICP data when available, and then plotted with water saturation (Track 6).
The dark-blue shading indicates the pore space that can become oil lled at
low capillary pressure. The shading transitions from blue to red,
corresponding to successively higher capillary pressures required to ll
additional pore volumes. Thus the layer around X,600, with more dark-blue
shading than the mostly red and yellow layer around X,500, represents
better quality rock. (Adapted from Ramamoorthy et al, reference 5.)
AIT Tool
Moved Hydrocarbon
Hydrocarbon
Water
Depth,
ft
Pyrite
Quartz
Anhydrite
Calcite
Dolomite
HRLA Tool
Siderite
Kaolinite
Chlorite
Illite (dry)
Montmorillonite
Lithology
Contributing Flow
0 % 100
T
2
Distributions
50 0 %
Core Porosity
Total Porosity
50 0 %
Microporosity
Macroporosity
Mesoporosity
NMR Porosity Partition
Computed Permeability
0.1 10,000 mD
Core Permeability
0.1 10,000 mD
Microporosity
Micromesoporosity
Micromacroporosity
Mesomicroporosity
Macromicroporosity
Mesoporosity
Macromesoporosity
Macroporosity
Ternary Porosity Partition
X,400
X,500
X,600
Capillary Pressure
Min Max
100 % 0
Water Saturation
Oilfield Review
Autumn 10
CleanPhase Fig. 17
ORAUT10-CLNPSE Fig. 17
31424schD7R1.indd 52 8/10/10 10:20 PM
Summer 2010 53
Searching Above Ground
Approaches discussed so far apply to data acquired
downhole. Because of the heterogeneity of carbon-
ate reservoirs, the shallow depth of investigation
of most logging tools may limit their use for opti-
mizing well positioning. For instance, fracture ori-
entation obtained from imaging tools can be
inuenced by local effects and may not reect the
predominant trend in the reservoir. However, new
developments in seismic technology are providing
operators with assistance in detecting fracture
swarms within a reservoir and this knowledge can
be used to optimize well locations.
Three-dimensional surface seismic surveys
offer an expanded view of reservoir heterogene-
ity, extending over the entire eld. Variations in
the reservoir properties such as porosity, clay
content and water saturation can all be charac-
terized using seismic measurements, although
their resolution and detection level are limited
by the seismic wavelengths used, survey design
and other factors such as near-surfacegener-
ated noise. Recent developments in seismic
acquisition tools and processing techniques have
increased the usable bandwidth and signal-to-
noise ratio such that higher resolution data with
enhanced signal delity are now obtainable.
Consequently, geoscientists are able to charac-
terize in ner detail the heterogeneous porosity
and lithology variations and the multiscale frac-
ture networks present in carbonate reservoirs.
25
Most carbonate reservoirs are naturally frac-
turedfrom microscale diffuse fractures (less
than 1 m [3 ft]) to macroscale faults (greater
than 100 m [330 ft]). At the intermediate meso-
scale (10 to 100 m) subseismic faults and frac-
ture swarms, or corridors, may prevail (above). A
typical fracture corridor can consist of thousands
of parallel fractures of variable dimensions
densely packed together, forming a volume that is
typically a few meters wide, a few tens of meters
high and several hundred meters long.
Permeabilities in these corridors can range well
above 10 darcies. These corridors often act as
major conduits for uid owing within the reser-
voir and may be responsible for early water
breakthrough from natural drive or waterood-
ing. Therefore, to manage eld production effec-
tively and maximize total recovery, it is crucial
that the locations of fracture corridors are accu-
rately known and modeled.
24. For more on the computation of capillary pressure:
Ouzzane J, Okuyiga M, Gomaa R, Ramamoorthy R,
Rose D, Boyd A and Allen DF: Application of NMR T2
Relaxation to Drainage Capillary Pressure in Vuggy
Carbonate Reservoirs, paper SPE 101897, presented at
the SPE Annual Technical Conference and Exhibition,
San Antonio, Texas, September 2427, 2006.
25. Singh SK, Abu-Habbiel H, Khan B, Akbar M, Etchecopar A
and Montaron B: Mapping Fracture Corridors in
Naturally Fractured Reservoirs: An Example from
Middle East Carbonates, First Break 26, no. 5
(May 2008): 109113.
>
Multiscale seismically constrained fracture characterization. Fractures
may exist over a wide range of scales from very small cracks to very large
faults. Understanding their distribution and properties at these different
scales is essential to characterize naturally fractured reservoirs. The scales
can be divided into three ranges: micro- (less than 1 m), meso- (10 to 100 m)
and macro- (greater than 100 m). Microscale fractures include layer-bound
diffuse fractures that can pervade across a geologic layer and are
frequently observed in image logs such as those from the FMI fullbore
formation microimager. Typically, these fracture types are the primary
controls used to build geologic models containing fractures, such as implicit
fracture models or discrete fracture networks (DFN). Although these diffuse
fractures are smaller than surface seismic wavelengths, a large population
density of such fractures can be detected with seismic measurements by
analyzing the seismic anisotropy. Mesoscale fracture corridors and
subseismic faults are the most difcult scale of fractures to characterize;
they are at the lower end of surface seismic resolution and few wells may
intersect them. These narrow features cross layer boundaries and, with
suitable 3D seismic data and careful analysis such as with the fracture
cluster mapping workow, they can be detected as subtle discontinuities in
the data. Because mesoscale fracture corridors can have very high
permeabilities and have major inuence over reservoir dynamics, they
should be incorporated into geologic models as individual fracture patch
sets. In contrast to micro- and mesoscale fractures, macroscale faults are
comparatively easy to detect with 3D seismic data and form the basis for
structural modeling. Computer interpretation methods for fault detection,
such as the ant tracking algorithm used in the Petrel seismic-to-simulation
software, are available to automate the process and may be able to
overcome analyst bias. Detailed analysis of the seismically derived rock
properties around these faults may help in assessing fault transmissivity.
TSFigure 22
M
a
c
r
o
s
c
a
l
e
Faults Dislocated horizons Ant tracking, fault transmissivity Structural faults
M
e
s
o
s
c
a
l
e
Fracture corridors Subtle discontinuities and scattering Fracture cluster mapping Fracture patch sets
M
i
c
r
o
s
c
a
l
e
Geologic Features Seismic Observations Data Analysis Model Representations
Diffuse fractures Seismic anisotropy Anisotropy analysis and inversion Implicit fracture models or DFN
31424schD7R1.indd 53 8/10/10 10:20 PM
54 Oileld Review
One method for identifying these corridors
using seismic data is the FCM fracture cluster
mapping technique. Geoscientists have devel-
oped the FCM workow to identify discontinui-
ties in the 3D surface seismic data associated
with subseismic faults and fracture corridors.
Two key factors contributing to the success of
this technique are the suitability of the seismic
acquisition and processing. The workow
assumes that large clusters of natural fractures,
which constitute a fracture corridor, produce
coherent structural discontinuities that are
detectable with 3D seismic data. The complete
FCM workow integrates expert interpretation
of high-quality seismic data and borehole mea-
surements with geologic modeling and dynamic
simulation, which enables a detailed character-
ization of naturally fractured reservoirs.
The discontinuity extraction software identi-
es subtle inconsistencies that appear as linea-
ments in the seismic data. Generally, the raw
lineaments that are extracted are associated
with either geologic discontinuities in the reser-
voir or nongeologic residual features in the data
such as acquisition footprints or near-surface
noise contamination.
26
To focus on detecting frac-
ture clusters, the process is constrained and cali-
brated with a priori knowledge that includes
regional and local structural geology, tectonic
history, reservoir geomechanics, core analysis,
borehole images, sonic logs, vertical seismic pro-
le data, well tests and production history.
Results are strongly dependent on the seismic
acquisition geometry and data quality and will be
less reliable with poor imaging, poor spatial and
temporal bandwidth, low signal-to-noise ratio
and acquisition footprints. Thus, there are strin-
gent requirements on the 3D seismic data quality
to provide a meaningful input for detecting frac-
ture clusters. Custom design of processing and
data acquisition, especially when using single-
sensor data such as those provided by the Q-Land
seismic system, may be necessary.
27
The FCM technique offers a radically different
technology for characterizing fractured reservoirs.
Historically, only the properties of diffuse fractures
have been characterized through the interpreta-
tion of a variety of seismic attributes, such as azi-
muthal anisotropy observations. However, with the
fully integrated FCM workow, the location of indi-
vidual fracture corridors can be detected and
embedded into a multiscale 3D reservoir model
containing faults and diffuse fractures. Dynamic
simulation of the uid ow through these multi-
scale models and calibration with production logs
verify the major ow pathways. Operators can use
this information to locate injector and producer
wells to maximize reservoir sweep efciency and
minimize water breakthrough.
Locating the Well
The FCM workow was used to model ve
Jurassic carbonate reservoirs in Kuwait. One of
these elds, the Sabriyah eld, was selected as
the key area for study because of its challenging
structural setting and a drilling schedule that
included four new wells (above left). An abun-
dance of lineaments across the reservoir were
identied after initial analysis of the seismic
data. Further analysis of these lineaments
revealed a predominant population oriented
NNE-SSW along the main axis of the anticline
structure and a secondary population consisting
of orthogonal lineaments (next page). In con-
trast, borehole image data showed a dominant
ENE-WSW fracture orientation.
This analysis suggested that the dominant
NNE-SSW trend in the lineaments is probably asso-
ciated with longitudinal fold-related fractures and
that the secondary set of orthogonal lineaments
correlate with the fractures identied from the
borehole image data and are possibly Riedel
26. Acquisition footprints, seen on 3D seismic time slices,
are patterns that correlate to surface-acquisition
geometry and distort amplitude and phase of reections.
This form of noise can obscure true subsurface
reections and should be removed prior to
interpretation, if possible. Although the FCM workow
might detect them, an experienced interpreter should
be able to identify them as noise rather than fractures.
27. The Q-Land system is a point-receiver acquisition and
processing system capable of acquiring 30,000 channels
of data in real time. Point-receiver data are recorded
with variable densities and processed with
>
Surface relief map of Sabriyah eld in northern
Kuwait. This eld, the rst of ve to be analyzed,
was considered a key area in the study.
Geoscientists used the FCM workow to evaluate
existing seismic data. Wells X-5 and X-6 were to
be drilled based on study results. Borehole
images and core from these wells validated the
fracture clusters predicted by the FCM model.
TSFigure 19
X-6
X-5
X-1
X-4
X-3
X-2
2 km
1 mi
>
Crosswell seismic imaging. At the absolute best,
3D surface seismic data (left) can resolve features
down to tens of meters. Crosswell imaging, such
as the DeepLook-CS seismic imaging service,
acquires data from downhole sources and
receivers placed in separate wells. Using higher
frequencies extending to kilohertz provides
ultrahigh-resolution images between wells and
can resolve features as small as 1.5 m [5 ft]. Seen
in the crosswell data (right) is a subseismic fault
(magenta line) and the detailed multilayered
reservoir structure. Fracture corridors, interpreted
from discontinuities detected in a 3D seismic
volume, can also be veried from this type of
crosswell seismic imaging.
TSFigure 23
X,950
Depth,
ft
Y,000
Y,050
Y,100
Y,150
Y,200
complementary digital group forming (DGF) techniques.
DGF processed raw sensor measurements provide a
clean group-formed trace with improved resolution
and low noise.
28. Riedel shears produce a geometric fracture pattern
commonly associated with strike-slip fault systems.
They may form echelon patterns inclined 10 to 30 to
the direction of motion.
29. Refae AT, Khalil S, Vincent B, Ball M, Francis M,
Barkwith D and Leathard M: Increasing Bandwidth for
Reservoir Characterization with Single-Sensor Seismic
Data, Petroleum Africa (July 2008): 4144.
30. The nominal fold is dened as the number of different
source-receiver locations that illuminate a particular
subsurface sampling point or bin. Each of the many
source-receiver pairs, corresponding to a given bin
location, will record reections along different raypaths
and can be characterized by its nominal azimuth and
offset. A broad and uniform distribution of source-
receiver offsets and azimuths within each bin provides
more information for seismic reservoir characterization.
31. Singh et al, reference 25.
31424schD7R1.indd 54 8/10/10 10:20 PM
Summer 2010 55
shears.
28
While this limited study indicated the
presence of numerous structural discontinuities
across the eld that could be related to subseismic
faults or fracture corridors, such interpretations
can be validated only through further integration
of other data sources and ultimately through drill-
ing. An example of validation from other sources is
the use of ultrahigh-resolution crosswell seismic
imaging (previous page, top right).
To obtain more-detailed information about the
fractures in the carbonate reservoirs of Kuwait,
Kuwait Oil Company (KOC) acquired a state-of-
the-art 3D seismic pilot survey over 100 km
2

[38 mi
2
] of the Northwest Raudhatain eld using
the WesternGeco Q-Land technology. This system
employs maximum displacement vibroseis sweep
and single-sensor receivers (see Land Seismic
Techniques for High-Quality Data, page 28). The
MD Sweep technique enhances low-frequency
content by optimally designing the drive force and
variable sweep rate of the vibroseis units.
29
Single-
sensor deployment enables dense sampling of the
waveeld for removal of source-generated noise.
The advanced acquisition design consisted of
a wide-azimuth square patch, resulting in a very
high nominal fold of 990 for 12.5-m by 12.5-m
[41-ft by 41-ft] bin size with uniform offset-
azimuth distribution up to 6 km [3.7 mi].
30
This
design is ideal for seismic fracture characteriza-
tion using P-wave data. The Northwest Raudhatain
eld presents an additional challenge because
the seismic reections are contaminated by a
series of multiple-reected seismic waves that
interfere with the primary reections over the
reservoir. Advanced data processing is currently
being applied to suppress these multiples and
maximize the extraction of information from the
3D seismic data for an extensive seismically
guided fracture characterization.
In the past, engineers have proposed that
fracture corridors result in early water break-
through but did not have effective tools to detect
their presence. Historically, fracture clusters
detected in wellbores were incorporated in sto-
chastic 3D models to explain their effects on pro-
duction. The ability to identify fracture clusters
away from the wellbore using the FCM workow
and to visualize their orientation with 3D maps
will help optimize eld development and avoid
unexpected water breakthrough.
31
Hydrocarbons from Carbonates
Much of the worlds remaining hydrocarbon
reserves are thought to lie in carbonate rocks
whose complexity has often confounded petro-
leum engineers, geophysicists and geologists
working to extract their riches. Step-change
improvements in a wide variety of interpretation
techniques and sensor technologies are making it
possible for these professionals to more effectively
evaluate, drill and produce carbonate reservoirs.
By integrating techniques and technology, the sta-
tistical odds inherent in drilling and maximizing
recovery from carbonates are being shifted in
favor of todays petroleum technologists. TS
>
Rening and dening fracture clusters. Existing seismic data were processed using discontinuity extraction software (DES) models without lters (left),
and the orientation of the fractures is overwhelmingly in line with the axis of the anticlinal structure (NNE-SSW). Logging data from Wells X-3 and X-4
indicated ENE-WSW orientation (insets). This is attributed to Riedel shears caused by NNE-SSW strike-slip faults. Azimuth lters applied to the seismic
data detected fracture clusters with different orientations (right). The orientation of these clusters is masked in the original processing. (Adapted from
Singh et al, reference 25.)
TSFigure 20
X-5
X-1
X-2
X-5
X-1
X-3
X-4
X-2
Filters:
Search azimuth: All 360
Dip angle: Features dip > 70
Filters:
Search azimuth: 45 to 135 and 225 to 315
Dip angle: Features dip > 70 in-line
in-line
45
315 225
135
x-line x-line
in-line
in-line
45
315 225
135
x-line x-line
X-3
X-3 Dipmeter Data
X-4
X-4 Dipmeter Data
270
90
45
315 225
135
0 180
270
90
45
315 225
135
0 180
31424schD7R1.indd 55 8/10/10 10:20 PM

You might also like