SPECIAL

R e s e rSECTION:
v o i r dR e sse c
r vroii pr tdie o
s cnr i p t i o n

Shale-gas reservoir-prediction study

in Daanzhai, Eastern Sichuan Basin
ZHONGPING LI, SINOPEC
FANGYAN HUANG, XINWEI HE, WANLONG ZHANG, and YUTING HE, CGG

Abstract (Daanzhai Formation), which is approximately 2200 m deep

In recent years, the economic viability of large-scale shale-
gas exploration and development projects has been impacted
directly by improvements in horizontal-well drilling and hy-
draulic-fracturing technologies. However, results from mul-
tiple shale-gas projects indicate that accurate reservoir descrip-
tions and resulting predictions of sweet spots are also crucial to
maximize production from those projects. Accurate reservoir
description results in the most appropriate placement of well
locations and trajectories, hydraulic-fracture design, and reser-
voir engineering. The successful experience in shale-gas explo-
ration and development outside China can aid in the search for
the most favorable geologic characteristics and rock-physical
properties and the examination of appropriate well-log inter-
pretation and seismic-reservoir-prediction technologies for a
shale-gas study in the eastern area of Sichuan Basin, China.
Favorable parameters in a shale-gas reservoir are mid- to high
kerogen content, low clay volumes, high brittleness, high effec-
tive porosity, and high permeability, along with well-developed
microfractures. Of those parameters, mid- to high total organic
carbon (TOC) is favorable for shale-gas accumulation, and
high brittleness is beneficial to hydraulic-fracture propagation.
Hence, the aim is to find sweet spots with high TOC and high
brittleness. Starting with petrophysical analyses that combine
laboratory measurements, core data, and well logs, a statistical
model is used to estimate relative volumes and compositions
of TOC and minerals and to establish a qualitative evalua-
tion criterion for shale-gas reservoirs in this area. Petrophysical
modeling and rock-property analyses establish the relation-
ship among petrophysical properties
(such as TOC, effective porosity, and
gas saturation) and elastic properties
(such as P-impedance, VP /VS ratio,
Poisson’s ratio, and Young’s modulus)
of the formation, which are related
to brittleness. Rock-property analy-
sis, prestack simultaneous inversion,
and facies-fluid probability-analysis
technologies help to predict favorable
development zones for the shale-gas
reservoir. Finally, the values of brittle-
ness, TOC, and shale-gas lithology
volume are combined to predict the
sweet spots of the gas shale and to
guide subsequent developments.
and 45 m thick. The Daanzhai Formation (Lower Jurassic
system) was formed in a shallow-semideep lacustrine envi-
ronment and consists of black–dark gray shale intercalated
with thinly bedded shell limestone deposition. Porosity of the
target shale reservoir is less than 5%, and average permeabil-
ity is 0.737 mD. The regional structure is gentle, high in the
northwest and low in the southeast, but microfractures are
well developed (Figure 1).
Core analysis for the target Da-2 member shows that the
overall content of total organic carbon (TOC) is about 1%. In
the Da-2 member, there are two layers with respect to TOC,
a layer with 0.5% to 1% TOC, comprising 58% of the total
interval, and a layer with 1% to 2% TOC, which is 40% of the
total interval. Organic-carbon maturity is primarily Type I–II
kerogen, and the vitrinite emission rate (RO) is 0.6% to 1.3%.
These values indicate favorable geologic conditions for shale gas.
Analysis of reservoir mineral compositions (Figure 2)
indicates that clay content is approximately 35% and brit-
tle-mineral content is approximately 65%, whereas average
quartz content is approximately 25%, indicating that this set
of shale has good brittle-rock components and is favorable for
fracturing operations.
Through the above analysis, the Daanzhai Da-2 member
is characterized by developed source rocks and microfrac-
tures, thick reservoir, and moderate depth of burial, whereby
it is quite favorable for economically viable shale-gas develop-
ment. The same target reservoir, the Da-2 member, in the
middle of the basin has achieved industrial gas-production

Introduction
The work area is in the eastern
part of the Sichuan Basin, and the
target formation is the Da-2 member Figure 1. Core microphotographs of the target layer.

526 The Leading Edge May 2014

 R e s e r v o i r d e s c r i p t i o n
Figure 2. Rock-mineral constituents and contents in the zone of
interest derived from core analysis and well logs.
flow. Geologic research indicates that the area has similar geo-
logic conditions and is useful as an analogue in predicting the
properties required for shale-gas production.
Multiwell formation-parameter evaluation
Shale-gas exploration indicates that mineral constituents
determine the quality of a shale-gas reservoir, influence gas
content, give insight into the maturity of the organic con-
tent, and provide valuable information for placement of well
locations, design of well-drilling programs, well completions,
and design of hydraulic-fracture jobs. For instance, experi-
ments indicate that shale with rich organic matter has more
micropore spaces than that with poor organic matter and can
adsorb more natural gas under the same temperature and
pressure. Hence, the critical factor that influences volume of
adsorbed gas is the content of organic carbon.
High content of quartz and other brittle minerals causes
more tendency to fracturing. Conversely, smectite and other
expansion clay minerals, because of filling or closing frac-
tures, are unfavorable for hydraulic fracturing. A carbonate
rock layer with high calcite content can be reactive to acids
and form dissolved pores, large volumes of pyrite can decrease
measured resistivity, and so forth. Therefore, determining
TOC and mineral composition in the zone of interest is a
critical first step in evaluation of unconventional formations.
A stochastic or statistical model is used to estimate mul-
tiwell volumes and distributions of TOC and of minerals
May 2014 The Leading Edge 527
 R e s e r v o i r d e s c r i p t i o n

Figure 3. TOC and mineral compositions calculated by statistical modeling (red lines), compared with core analysis (green dots). From left:
Column 1: photoelectric absorption index of the interface log (red), gamma-ray log without uranium (green), borehole-diameter log (gray).
Column 2: thorium log (green), potassium log (red), uranium log (black). Column 3: measure depth. Column 4: deep-investigation induction
log (red), shallow-investigation induction log (blue). Column 5: neutron-porosity log (blue), density log (red), P-sonic log (green). Column
6: mineral compositions from core data — clay (black), sand (yellow), calcite (light blue), and so forth. Column 7: mineral compositions
calculated from statistical modeling — clay (black), sand (yellow), calcite (light blue), and so forth. Columns 8 through 14 in sequence show
clay-volume log, sand-volume log, calcite-volume log, pyrite-volume log, TOC log, porosity, and water saturation. Red lines represent the
statistical-modeling result; green dots represent core analysis.

(Moses and Harrison, 1985; Mitchell and Nelson, 1988,
1991) by solving the system of linear equations:
fi = ∑ mj eijVj ,i = 1,···n, (1)
where fi represents log readings related linearly with each
mineral, Vj is the volume of mineral j, eij is the computed
value of parameter i in mineral j, and m is the number of
minerals being solved for.
The model calculates the parameters by minimizing the dif-
ference between the normalized acquired logs and the comput-
ed synthetic logs:
, (2)
where n is the number of variables in the equation, mi is the
normalized measured values of parameter i, fiis the comput-
ed values of parameter i, and bi is the variance of parameter i
for each component of the model. Although some of the frac-
tional volumes might be gas-filled porosity or liquid-filled
porosity rather than minerals, all volumes will be described
as “mineral” volumes for simplicity.
The parameters can be neutron-porosity log, volume-
density log, P-sonic log, gamma-ray spectrometry U-content
log, KTh calculated clay-content log, and so forth. In this
study, the calculated volumes are dry clay content, quartz
content, limestone content, pyrite content, organic-carbon
Shale gas Tight limestone Porous limestone Shale
VBRIT • VDCL ” VDCL ” Else
VKERO • VCAL • VCAL •
ĭH” ĭH 烍
Table 1. Evaluation standard of shale-gas formations.
content, water-filled pore volume, and hydrocarbon-filled
pore volume.
There are many ways to control the quality of estimated
results; for example, core data are the control mechanism to
validate the model. Passey methods (Passey et al., 1990) and
modified Passey methods (Sondergeld et al., 2010) can be
used as a quality-control check on TOC volume. In addi-
tion, uranium content, a strong indicator of TOC, can be
used in the analysis. Results obtained in the study area show
that rock-mineral content, TOC content, and porosity are in
good agreement with core analysis data (Figure 3), indicating
that this statistical model is suitable for shale-gas reservoir in-
terpretation.
Because shale-gas reservoirs display highly variable char-
acteristics and reservoir-forming mechanisms, criteria for
shale-gas reservoir evaluation vary from area to area. Table
1 summarizes the initial qualitative evaluation criteria used
in this study. VBRIT is the sum of brittle-minerals content
(quartz, calcite, pyrite, and so forth), VKERO is TOC content,

528 The Leading Edge May 2014

 R e s e r v o i r d e s c r i p t i o n
qe is effective porosity, VDCL is clay content, and VCAL is calcite
content. We define the shale-gas reservoir as where brittleness
is more than 50% and TOC content is more than 1%.
Petrophysical modeling and rock-property analysis
The existing seismic-amplitude data associated with key
parameters of the shale-gas reservoir allow extension of the
analysis beyond well control to the entire field to identify the
most attractive facies. Hence, petrophysical modeling and
rock-property analysis play the role of bridge and link.
Based on various mineral contents, porosity, and water
saturation calculated with the statistical model, the petrophys-
ical-modeling method uses the self-consistent approach (Bu-
diansky, 1965; Hill, 1965; Wu, 1966) to characterize the field
in such a way that at any wellbore, data can be reproduced by
seismic-amplitude data. Therefore, reliable nonmeasured logs
such as shear-wave velocity can be synthesized.
High TOC is favorable for shale-gas accumulation,
whereas high brittleness is beneficial to fracturing. Hence,
the two parameters are critical to reservoir prediction and
subsequent production fracturing. The aim is to seek a rela-
tionship between petrophysical properties (such as TOC) and
elastic properties (such as P-impedance, VP /VS ratio, Poisson’s
ratio, and Young’s modulus) through rock-physical analysis
technology. Elastic parameters, such as Young’s modulus or
shear modulus, indicate which rocks are brittle. Rickman et
al. (2008) derive equations to estimate a brittleness index,
B, from well logs. Equation 3 is a generalized form of those
equations. The same equations can be applied to inversion
results as follows:
BE ≈ (E – Emin)/(Emax – Emin), (3a)
BV ≈ (V – Vmin )/(Vmax – Vmin ), (3b)
and
B ≈ (BE + BV )/2 , (3c)
where E is Young’s modulus, B E is brittleness estimated from
Young’s modulus, V is Poisson’s ratio, B V is brittleness esti-
mated from Poisson’s ratio, and subscripts min and max denote
minimum and maximum values of Young’s modulus and
Poisson’s ratio, respectively
From core analysis, we know that brittle components are
mostly quartz, calcite, and pyrite, whereby in this area, we
consider the sum of those as the brittle volume. Through the
crossplot of Young’s modulus (vertical axis) and Poisson’s ratio
(horizontal axis) (Figure 4a), we observe that high Young’s
modulus and low Poisson’s ratio are related to a higher vol-
ume of brittle minerals (brittle sum of quartz, calcite, py-
rite content, and so forth, here defined as brittle volume).
Hence we used the multiple-regression method to establish
a relationship among brittle volume, Young’s modulus, and
Poisson’s ratio, as described by the following equation:
Brittle volume = 0.452 – 0.614 =PR + 0.010 =YM , (4)
where PR is Poisson’s ratio and YM is Young’s modulus.
Low Young’s modulus and low Poisson’s ratio also can in-
dicate higher TOC. The crossplot of TOC, brittleness, and
May 2014 The Leading Edge 529
 R e s e r v o i r d e s c r i p t i o n

Figure 4. (a) Crossplot of brittle volume, Young’s modulus, and Poisson’s ratio. The color scale describes brittle volume. (b) Crossplot of TOC,
P-impedance, and brittleness index. The color scale describes the brittleness index.

Figure 5. (a) Crossplot of brittle volume and brittle contents. (b) Crossplot of TOC estimated by P-impedance and brittleness and TOC
estimated from statistical modeling. The color scale describes the brittleness index estimated from equation 3 (Rickman et al., 2008).

P-impedance (Figure 4b) indicates that TOC increases, followed Time Minoffset Stack 1 Stack 2 Stack 3
by the increase of brittleness and P-impedance to some extent. (ms) (m) (m) (m) (m)
We also used the multiple-regression method to establish
0 60 100 100 100
an empirical relationship among TOC, brittleness index, and
P-impedance, given by 500 60 420 500 580
1000 60 900 1300 1700
TOC = 0.038 + 0.06 × B – 4.580e –006 × Pimp , (5) 1300 60 1140 2020 2900
where B is brittleness calculated with equation 3, and Pimp is 1600 60 1220 2100 2980
P-impedance of wells. Table 2. Partial-stack scheme.
For comparison with well-log parameter evaluation,
the crossplot in Figure 5a describes estimated brittle-vol- Prestack simultaneous inversion
ume results (horizontal axis) and brittle contents (brittle Prestack simultaneous inversion (Contreras et al., 2006),
sum of quartz, calcite, pyrite content, and so forth) (verti- based on the constrained sparse-spike inversion algorithm, in-
cal axis). The crossplot in Figure 5b describes the estimated verts partial-stack seismic-amplitude data into elastic properties
TOC (horizontal axis) and TOC estimated from statistic such as P-impedance, S-impedance, VP /VS ratio, density, and so
modeling (vertical axis), whereas the colored points iden- forth. Poisson’s ratio can be calculated from the VP /VS ratio, and
tify the brittleness index (estimated with equation 3). We Young’s modulus can be calculated from P-wave velocity, S-wave
observe that equations 4 and 5 match core-analysis and velocity, and density. Therefore, seismic-derived TOC and brit-
well-log interpretation results very well. tleness volume can be calculated with equations 3, 4, and 5.

530 The Leading Edge May 2014

 R e s e r v o i r d e s c r i p t i o n

In this area, seismic-amplitude measurements were sam- Facies-fluid probability analysis

pled at 2 ms and have a frequency band of 10 to 90 Hz, with The facies-fluid probability-analysis method (FFP) is
a central frequency of 50 Hz. In the target reservoir, the maxi- based on prestack simultaneous inversion to generate vol-
mum offsets which are useful to seismic-amplitude data are umes of lithology probability that incorporate estimated
2900 m (the 3D survey spaced at 20-m intervals in inline and uncertainties resulting, for example, from property overlap
crossline) We generate three partial stacks (Table 2). Minoff- of lithologies, resolution, and seismic noise. FFP calculates
set in Table 2 represents the minimum
offset used in the study. In the target
reservoir (the time is approximate, from
1300 to 1600 ms), the offset ranges of
Stack 1, Stack 2, and Stack 3 are (60–
1140), (1140–2020), and (2020–2900).
In addition, small time-shift corrections
are applied to the partial stacks to reduce
event misalignment caused by process-
ing errors such as residual NMO correc-
tions.
The seismic-amplitude data lack
low-frequency information (zero to
10 Hz in our case), so low-frequency
volumes of P-impedance, VP /VS ratio,
and density are required for prestack si-
multaneous inversion. In this study, we
have four well logs. We use two of them
(Wells 2 and 4) to interpolate these vol-
umes by using the geologic model as a
guiding framework based on the ho-
rizon interpretation of the tops of the
main geologic formation. Three well
locations (Wells 1, 2, and 4) are used
Figure 6. Section view of P-impedance, VP /VS ratio, S-impedance, density, TOC, and brittle
to estimate each partial-stack multiwell volume. The black arrow identifies the target reservoir.
wavelet.
The deterministic inversion makes
use of partial-stack data, the previously
estimated multiwell wavelet of each
stack, and the low-frequency volumes
of elastic properties to produce the elas-
tic volumes. Then equations 3, 4, and 5
are used to estimate the brittle volume
and TOC.
Figure 6 shows the inversion result
and the estimated result in section view.
Wells 1 and 3 are blind wells and were
not used in the interpolation of low-
frequency volumes. We observe that the
inversion result (P-impedance, S-imped-
ance, VP /VS ratio, and density) matches
the measured logs very well. The TOC
and brittle volumes also match the well’s
TOC and brittle volume calculated with
the statistical model.
An integrated workflow using pet-
rophysical modeling, rock-property ana-
lysis, and prestack simultaneous inver-
sion can yield a reliable brittle and TOC
volume, laying the foundation for subse- Figure 7. Probability density function (PDF). Green, yellow, and pink data points and ellipses
quent field development. identify shale, shale gas, and porous limestone, respectively.

532 The Leading Edge May 2014

 R e s e r v o i r d e s c r i p t i o n

lithology probabilities using joint probability density func-
tions (PDF). A joint or multivariate PDF is a parametric
model of the statistical relationship between multiple prop-
erties (usually between P-impedance and S-impedance or
any other elastic property yielded by deterministic inver-
sion). It can be used to transform elastic information into
geologic information.
Under the guidance of rock-property analysis, a lithol-
ogy PDF template (Figure 7) is established based on the
difference in distribution range of various lithologies on
the crossplot of P-impedance and VP /VS ratio. In Figure 7,
the horizontal and vertical axes are P-impedance and VP /
VS ratio, respectively. Data points and probability ellipses,
with each lithology type indicated by a different color, cor-
respond to the probability-distribution range of different
lithologies, i.e., green, yellow, and pink data points and el-
lipses indicate shale, shale gas, and porous limestone, re-
spectively.
The probability-density-function template is then applied
to the elastic data volumes obtained from prestack simultane-
ous inversion to estimate the probability distribution of shale
gas and a volume of the most likely lithology at each point
in the 3D volume (Figure 8). It can be observed from Figure
8 that prediction results match the qualitative evaluation of
lithology in Well 3, a blind well.
Conclusions
Accurate reservoir descriptions and resulting prediction
of sweet spots are recognized as crucial for maximizing pro-
duction in shale-gas projects. In this study, we have success-
fully applied a workflow to describe the shale-gas reservoir of
the Daanzhai Formation in the Sichuan Basin by combining
core analysis, geologic knowledge, petrophysical modeling,
rock-property analysis, prestack simultaneous inversion, and
facies-fluid probability analysis. In particular, the seismic-
derived TOC, brittle volumes, and shale-gas likelihood li-
thology distribution can predict the shale-gas sweet spots of
the area and can guide the
subsequent development pro-
gram to optimize produc-
tion.
reservoirs in the central Gulf of Mexico: Geophysics, 71, no. 4,
E41–E48, http://dx.doi.org/10.1190/1.2212276.
Hill, R., 1965, A self-consistent mechanics of composite materials:
Journal of the Mechanics and Physics of Solids, 13, no. 4, 213–
222, http://dx.doi.org/10.1016/0022-5096(65)90010-4.
Mitchell, W. K., and R. J. Nelson, 1988, A practical approach to
statistical log analysis: Transactions of the Society of Profes-
sional Well Log Analysts, 29th Annual Logging Symposium,
Paper S.
Mitchell, W. K., and R. J. Nelson, 1991, Statistical log analysis made
practical: World Oil, 212, no. 6, 115–119.
Moses, B., and R. Harrison, 1985, Statistically valid log analysis meth-
od improves reservoir description: Presented at the SPE Offshore
Europe Conference, SPE 13981.
Passey, Q. R., S. Creaney, J. B. Kulla, F. J. Moretti, and J. D. Stroud,
1990, A practical model for organic richness from porosity and
resistivity logs: AAPG Bulletin, 74, no. 12, 1777–1794.
Rickman, R., M. J. Mullen, J. E. Petre, W. V. Grieser, and D. Kundert,
2008, A practical use of shale petrophysics for stimulation design
optimization: All shale plays are not clones of the Barnett Shale:
Annual Technical Conference and Exhibition, SPE, SPE 115258-
MS, http://dx.doi.org/10.2118/115258-MS.
Sondergeld, C. H., K. E. Newsham, J. T. Comisky, M. C. Rice, and C. S.
Rai, 2010, Petrophysical considerations in evaluating and producing
shale gas resources: Presented at the SPE Unconventional Gas Confer-
ence, SPE 131768-MS, http://dx.doi.org/10.2118/131768-MS.
Wu, T. T., 1966, The effect of inclusion shape on the elastic mod-
uli of a two-phase material: International Journal of Solids
and Structures, 2, no. 1, 1–8, http://dx.doi.org/10.1016/0020-
7683(66)90002-3.
Acknowledgments: We thank the Geophysical Exploration Research
Institute of the Jianghan oil field for permission to publish this ar-
ticle. The study has been performed using Jason and PowerLog
software. We are grateful to Carlos Torres-Verdín, Ted Holden, and
Lucia Levato for constructive suggestions. Our gratitude goes to
Sara Pink-Zerling for her assistance with the article and to Jiapeng
Ye for help with preparing materials.
Corresponding author: Fangyan.Huang@CGG.com

References
Budiansky, B., 1965, On the
elastic moduli of some
het erogeneous materials:
Journal of the Mechan-
ics and Physics of Solids,
13, no. 4, 223–227, http://
dx.doi.org/10.1016/0022-
5096(65)90011-6.
Contreras, A., C. Torres-Verdín,
and T. Fasnacht, 2006, AVA si-
multaneous inversion of par-
tially stacked seismic ampli- Figure 8. Section view showing shale-gas probability volume and maximum-likelihood lithology volume.
tudedataforthespatialdelineation The color bar on left shows lithology distribution of inversion in the blind-well location. The color bar on
of lithology and fluid units right shows lithology interpretation of the well log. Green, yellow, light blue, and pink identify shale, shale
of deepwater hydrocarbon gas, tight limestone, and porous limestone, respectively

534 The Leading Edge May 2014