APPLIED GEOPHYSICS, Vol.13, No.1 (March 2016), P. 80-92, 14 Figures.

DOI:10.1007/s11770-016-0531-7

The application study on the multi-scales integrated

prediction method to fractured reservoir description*
Chen Shuang-Quan♦1,2, Zeng Lian-Bo1, Huang Ping3, Sun Shao-Han4, Zhang Wan-Lu1,2, and Li Xiang-Yang1,2,5

Abstract: In this paper, we implement three scales of fracture integrated prediction study by
classifying it to macro- (> 1/4λ), meso- (> 1/100λ and < 1/4λ) and micro- (< 1/100λ) scales.
Based on the multi-scales rock physics modelling technique, the seismic azimuthal anisotropy
characteristic is analyzed for distinguishing the fractures of meso-scale. Furthermore, by
integrating geological core fracture description, image well-logging fracture interpretation,
seismic attributes macro-scale fracture prediction and core slice micro-scale fracture
characterization, an comprehensive multi-scale fracture prediction methodology and technique
workflow are proposed by using geology, well-logging and seismic multi-attributes. Firstly,
utilizing the geology core slice observation (Fractures description) and image well-logging
data interpretation results, the main governing factors of fracture development are obtained,
and then the control factors of the development of regional macro-scale fractures are carried
out via modelling of the tectonic stress field. For the meso-scale fracture description, the
poststack geometric attributes are used to describe the macro-scale fracture as well, the
prestack attenuation seismic attribute is used to predict the meso-scale fracture. Finally, by
combining lithological statistic inversion with superposed results of faults, the relationship
of the meso-scale fractures, lithology and faults can be reasonably interpreted and the cause
of meso-scale fractures can be verified. The micro-scale fracture description is mainly
implemented by using the electron microscope scanning of cores. Therefore, the development
of fractures in reservoirs is assessed by valuating three classes of fracture prediction
results. An integrated fracture prediction application to a real field in Sichuan basin, where
limestone reservoir fractures developed, is implemented. The application results in the study
area indicates that the proposed multi-scales integrated fracture prediction method and the
technique procedureare able to deal with the strong heterogeneity and multi-scales problems
in fracture prediction. Moreover, the multi-scale fracture prediction technique integrated
with geology, well-logging and seismic multi-information can help improve the reservoir
characterization and sweet-spots prediction for the fractured hydrocarbon reservoirs.
Keywords: Multi-scales; Fracture prediction; Heterogeneity; Reservoir characterization,
Sweet-spots prediction.

Manuscript received by the Editor February 9, 2015; revised manuscript received February 16, 2016.
*This research is supported by the national oil and gas major project (No. 2011ZX05019-008) and National Natural Science
Foundation of China (No. 41574108 and U1262208), presented at the Exploration Geophysics Symposium 2015 of the
EAGE Local Chapter China.
1. State Key Laboratory of Petroleum Resources and Prospecting, China University of Petroleum, Beijing 102249, China.
2. CNPC Key Laboratory of Geophysical Prospecting, China University of Petroleum, Beijing 102249, China.
3. Research Institute of CNPC Southwest Oil and Gas Field Branch, Chengdu 641500, China.
4. Geophysical Research Center, BGP, PetroChina, Zhuozhou 072751, China.
5. British Geological Survey, Edinburgh EH9 3AZ, U. K.
♦Corresponding author: Chen Shuang-Quan (Email: chensq@cup.edu.cn)
© 2016 The Editorial Department of APPLIED GEOPHYSICS. All rights reserved.

80
 Chen et al.

Introduction depending on their lengths: 1) macro-scale: stratigraphic

and lithological intervals or faults, with a range over
dozens of meters in length; 2) meso-scale: cross-
Fracture is the key factor to dominate the rock bedding and ripple laminations etc, with lengths on
physical parameters in oil and gas reservoirs. The precise the order of centimeters to meters, which are less than
prediction of fracture development plays a significant one-quarter seismic wavelength and greater than one
role for the hydrocarbon exploration and development, percent of seismic wavelength so that they are unable
especially for the sweet-point prediction of the to be observed in seismic profile; 3) micro-scale: grain-
unconventional reservoirs. Currently, the most widely scale crack, pores or pore throats in rock samples, with
used methods for seismic fracture prediction are to a range less than one percent of seismic wavelength,
utilize the azimuthal anisotropy characteristic displayed only can be observed by electron microscope and the
when the seismic wave propagates in the fractured seismic reflection characters of which are too small to be
medium, including reflected amplitude, velocity, travel observed by seismic data.
time, attenuation, etc. These seismic anisotropy variations Besides, it may obtain the same result in terms of
with azimuth are used to predict the fracture intensity fracture development at different scales to predict fractures
and direction. Recently, studies on the theory and using seismic anisotropy characteristics (Wang et al.,
application of azimuthal anisotropic characteristic in the 2015). These two fracture models have the same density
fractured media, including laboratory measurements and but different scales. Because they have the same fracture
real field data applications, have achieved big progresses. density, the conventional seismic response attributes
Among them, Rüger and Tsvankin (1997) established resulted from the fracture density, such as amplitude
the equations and analyzed the azimuthal anisotropy and velocity, can’t been used to effectively distinguish
characteristic of the prestack seismic reflection the fracture scale and accurately characterize the multi-
coefficients for the HTI (Horizontal Transverse Isotropy) scales fractured reservoir. Therefore, geophysicists are
media, which are the basic for fracture prediction by expecting to find a more effective method to distinguish
using the azimuthal anisotropy attributes. Based on and interpret the anisotropy response characteristic caused
the theory, fracture prediction method was proposed by fractures with different scales.
for the prestack seismic gather dataset by utilizing the The basic theories for analyzing seismic anisotropic
azimuthal anisotropy characteristic, and has been applied responses of a fractured medium are equivalent medium
in the real field data achieving very good results (Shen rock physics models, such as Hudson model (1981)
et al., 2002；Tod et al., 2007). For the VSP seismic and Schoenberg model (1983). Nevertheless, both of
data, the fractured reservoir can also be characterized the models only consider the effects of single fracture
by analyzing the azimuthal anisotropy attributes (Liu scale on the elastic moduli. Chapman (2003) proposed
et al., 2011；Behura et al., 2012). Besides, Yin et al. a meso-scale fracture model in which the effect of
(2012) implement the azimuthal anisotropy analysis and microcracks, meso-scale fractures and pores are taken
fracture prediction for the seismic dataset acquired from into considerations. And an analogous model by using a
the laboratory physical model that is made of artificial T-matrix approximation is proposed by Jakobsen (2004).
fractured media. As well, the study on seismic azimuthal In the T-matrix approximation, the effect of fractures
attenuation attributes in fractured media has been carried (pores, cracks) is introduced as a perturbation of the
out by previous researchers (e.g., Chichinina et al., 2006; solution for the elastic background medium. Agersborg
Clark et al., 2009; Behura et al., 2012). Chen et al. (2012) et al. (2007) show that the meso-scale fracture model
study the frequency-dependent attributes using a physical is a special case of the T-matrix approach. Both of the
model dataset, and achieve the comparative analysis of rock physics models are further extended to the case
the seismic response characteristics by saturating the of multiple sets of fractures by Jakobsen and Chapman
fractured media with different liquids. Later on, Wang (2009). In all of these multi-scales rock physics models,
et al. (2014) make a step in the application of fracture the fluid flow between fractures and equant pores is
prediction by using the prestack attenuation attributes to governed by the finite fluid-pressure difference between
a real field data. fracture and equant pore space, and the fluid flow derived
Actually, seismic anisotropy characteristics are by the two models above will cause energy diffusion
comprehensive response of fractures at different scales and frequency-dependent anisotropy in the seismic
within the reservoirs. According to MacBeth and Li frequency band. This phenomenon has been verified in
(1999), fractures can be classified into three groups real field seismic exploration. Liu et al. (2003) reveal the

81
 Fractured reservoir description

frequency-dependent anisotropy using multicomponent macro-scale fractures by means of conventional

VSP data from a fractured gas reservoir. Using this data migration imaging method such as Ray theory (e.g., Aki
set, Maultzsch et al. (2003) analyzed frequency-dependent and Richards, 1980). Mirco-scale fractures are difficult to
shear-wave splitting and found that the time delay be analyzed by seismic data since their scales are far less
between fast and slow splitting shear waves decreases as than the seismic wavelength. Such type of fractures can
frequency increases. They used the meso-scale fracture only be analyzed by rock parameter test and core slice
model, which quantifies frequency-dependent anisotropy observation. But for meso-scale fractures that are between
resulting from mesoscopic fractures, to invert for fracture micro-scale and macro-scale fractures, we can take
density and fracture radius. By analyzing multi-azimuthal advantage of rock physics model in equivalent medium
walkaway VSP data from a fractured hydrocarbon to help analyze their anisotropy response characteristics,
reservoir, Maultzsch et al. (2007) observed significant explore their sensitive anisotropy attributes, finally
azimuthal variations in P-wave attenuation at reservoir establish a technical procedure and scheme of multi-
depth. They found that the presence of attenuation scales fracture prediction by comparing those prediction
anisotropy was due to aligned meso-scale fractures, by methods of fractures in three types of scales.
simulating this effect using the meso-scale fracture model. We investigate the sensitive attributes to meso-scale
Liu et al. (2007) compared the velocity and attenuation fractures by analyzing seismic anisotropy response of
anisotropies caused by two fracture sets (one being the fracture models at different scales. In order to simulate
fluid conduit) embedded in a porous medium and found the seismic response characteristics of fractures at
that the symmetry axes in the azimuthal variations of different scales, the study applies forward modelling
velocity and attenuation were significantly different. method which is mainly based on Chapman’s multi-
In this paper, we aim to develop the method and scales rock physics model (Chapman, 2003). Chapman’
technical procedure for multi-scales fracture prediction s model considers the case of the model consisting of
which is effective in practical industrial application. It two aligned vertical fracture sets, then the frequency-
begins with multi-scales rock physics model, conducts dependent effective elastic tensor can be written as
forward modeling and attribute analyzing of fracture
models at different scales, and forms a method and Cijkl (Z ) Cijkl
0
 I Cijkl
p
(Z )  H1Cijkl
f1
(Z )  H 2Cijkl
f2
(Z ), (1)
technique for detecting meso-scale fractures by using
attenuation anisotropy attribute. Then, we obtain the where ω is angular frequency, φ is porosity, ε is
geological fracture prediction distributions by regional fracture density, C 0 is the elastic stiffness matrix of
geological structure analysis and stress field analysis and the background medium, and C p, C f1, C f2 denote the
predict macro-scale fractures integrated by combining contributions of pores, the first fracture set and the
with distinguishable poststack seismic attributes. Finally, second fracture set, respectively. The subscripts of i,
we observe core slices for micro-scale fractures, pores j, k, and l denote that the constant C is a tensor. The
and the filling shale. We can obtain the integrated first correction here is the isotropic elastic tensor for
prediction result of multi-scales fractures by combining the background medium, the second correction is the
the results of seismic, geology and well logging isotropic elastic tensor associated with the spherical
information for fractures under different conditions of pores, and the last two corrections correspond to the
three different scales. By applying the proposed method fractures. The fracture corrections are proportional to
and technical procedure to a real field limestone fractured the relevant fracture densities if we only consider the
reservoirs, a high coincidence between the fracture first order approximation of anisotropy intensity by
prediction result and imaging logging interpretation result ignoring the interactions between the cracks and pores
has been established, which demonstrates the feasibility (i.e., ignore the second order term). The tensors for the
of the integrated prediction method. fracture corrections are transversely isotropic, but they
will take the familiar simplified form of the TI tensor
when they are referred to a system of coordinates in
Methodology and prediction which one of the axes is parallel to the fracture strike; in
the material coordinate system, simple fracture elastic
procedure modulus should be taken account of and then we can
rotate the fracture corrections back into the material
Since the lengths of macro-scale fractures are greater coordinate system through coordinate transformation.
than the resolution of seismic waves, we can evaluate The modelling result is shown in Figure 1. Figure 1a

82
 Chen et al.

shows the characteristic of P-wave velocity varying with the velocity dispersion will move toward low frequency
frequency without meso-scale fractures. We can find that with the increase of fracture scale. When the meso-scale
the P-wave velocity doesn’t change with frequency fractures in underground medium are longer than 0.1 m,
(< 1000 Hz) in the seismic and well logging frequency frequency dispersion effect will be present in the seismic
band when there aren’t meso-scale fractures, whilst frequency band (< 100 Hz). The comparison of both
frequency dispersion occurs when the frequency is up to figures demonstrates that meso-scale fractures are the
ultrasonic frequency range. Figure 1b shows the variation significant cause of frequency dispersion and attenuation
with frequency of P-wave velocity parallel to fracture in the seismic frequency band. For this reason, meso-
direction with the presence of meso-scale fractures (i.e., scale fractures can be studied based on that feature.
all the fractures are at meso-scale). It can be found that

(a) (b)
2.9 2.6
2.85
2.8
2.75 2.55
Vp (Km/s)

Vp (Km/s)
2.7
2.65
1m 0.1 m 0.01 m
2.6 2.5
2.55
2.5 Mirco

2.45 2.45
0 2 4 6 0 1 2 3
Log frequency Log frequency
Fig.1 The frequency-dependent characteristics of P-wave velocity. (a) model without meso-scale
fractures (b) model with meso-scale fractures.

Let’s consider the difference of azimuthal anisotropy Hz. The modelling results demonstrate that maximum
attributes of meso-scale fracture models in the seismic velocity and minimum attenuation correspond to the
frequency band. Figure 2 shows the modelling results of direction parallel to fracture, which also confirms that
fracture models with simple fracture set, the azimuthal both the anisotropy characteristics of P-wave velocity
variation of P-wave velocity (Vp) and attenuation (1/ (Li et al., 2003) and attenuation anisotropy can be used
Q), where the incident angle is 30 degrees, the strike to help predict fracture orientation. Besides, in Figure
direction of fractures is 50 degree and frequency is 10 2, velocity and attenuation anisotropy characteristics

(a) 2.6 (b) 0.025

0.001m
2.55 0.01m
0.1m
1m 0.02
2.5
Attenuation (1/Q)

2.45 0.015
Vp (Km/s)

2.4

2.35 0.01
0.001m
2.3 0.01m
0.005 0.1m
1m
2.25

2.2 0
0 50 100 150 200 250 300 350 0 50 100 150 200 250 300 350
Azimuth angle (degree) Azimuth angle (degree)

Fig.2 The azimuthal variation of P-wave velocity (a) and attenuation (b) in a medium with meso-
scale fractures. For the parameters in the modelling, the incident angle is 30 degree, the fracture
strike direction is 50 degree, and frequency is 10 Hz.

83
 Fractured reservoir description

also show a trend of cosine variation with azimuth scale fractures (0.001m) and the variation with azimuth
(as an ellipse in the plane) so that we can analyze of attenuation enhances apparently when the lengths of
the characteristics of anisotropy intensity based on fractures are longer than 0.01 m (meso-scale fractures);
ellipse fitting in the plane. Remarkably, the attenuation the velocity variation with azimuth doesn’t have
anisotropy enhances gradually with the increase of apparent difference under different conditions of fracture
fracture scale (see Figure 4b) when the scale of the scale. All the evidences demonstrate that attenuation
simulated fracture model increases; attenuation does anisotropy is sensitive to fracture scale while velocity
hardly vary with azimuth under the condition of macro- anisotropy is insensitive to scale information.

(a) 3.9 (b) 0.024

0.022
3.8
0.02
3.7 0.018

Attenuation (1/Q)
100 0.016
Vp (Km/s)

3.6
0.014
3.5 0.012
3.4 0.01
80
0.008
3.3
0.006
3.2 0.004
100 200 300 100 200 300
Azimuth angle (degree) Azimuth angle (degree)
Fig.3 The modelling P-wave velocity (a) and attenuation (b) anisotropy of models with two fracture
sets under the condition of the same fracture density. One set of fractures is in mesoscale (1 m)
and the fracture strike direction is 80 degrees, and the other is in microscale (0.001 m) and the
fracture strike direction is 120 degrees. The anisotropy intensities of the both two fracture sets
are three percent and the fractures and pores are filled with water.

(a) 3.9 (b) 0.025

3.8
0.02
3.7
Attenuation (1/Q)

0.015
Vp (Km/s)

3.6 86

3.5 0.01
3.4 80
0.005
3.3

3.2 0
100 200 300 100 200 300
Azimuth angle (degree) Azimuth angle (degree)

Fig.4 The simulated P-wave velocity (a) and attenuation (b) anisotropy of models with two
fracture sets under different conditions of fracture density. One set of fractures is in mesoscale
(1m) and the fracture strike direction is 80 degree, and the other is in microscale (0.001 m) and
the fracture strike direction is 120 degrees.

The anisotropy response characteristics under the is at micro-scale with an azimuth of 120 degrees and a
condition of two sets of fractures at different scales are length of 0.001 m. The anisotropy intensities for the both
considered to further verify the sensitivity to meso- of the fracture sets are three percent and the fractures
scale fracture of attenuation anisotropy. The first model and pores are filled with water. Figure 5 shows the
designed here consists of two sets of fractures. One modelling results for the azimuthal variation of velocity
set of fractures is at meso-scale with an azimuth of 80 and attenuation. We can find that both velocity and
degrees and a length of 1m; the other set of fractures attenuation show anisotropy characteristics due to the

84
 Chen et al.

presence of fractures. However, the maximum velocity attributes related to seismic attenuation to directly
anisotropy corresponds to the average of the two fracture predict meso-scale fractures.
strike directions (100 degrees), and the minimum Based on the theoretical analysis and rock physics
attenuation corresponds to the azimuth of meso-scale numerical modelling in previous section, the results
fractures (80 degrees). demonstrate that meso-scale fractures can be well
Then two fracture sets were designed to study their characterized by seismic attenuation attribute, thereby
anisotropy characteristics under different fracture meso-scale fractures can be estimated by using azimuthal
densities conditions. The model consists of two sets of attenuation from prestack seismic data. Combining with
fractures. One set of fractures is at mesoscale with an the prediction methods of macro-scale and meso-scale
azimuth of 80 degree, a length of 1m and a density of fractures described previously, a technical procedure
three percent; the other set of fractures is at microscale of fracture prediction which combined geology, well
with an azimuth of 120 degree, a length of 0.001m logging and seismic multi-attributes comprehensively
and a density of four percent. Figure 4 shows the is proposed (see in Figure 5). The technique can be
modelling results of the azimuthal variation of velocity summarized as: 1) For macro-scale fractures, they
and attenuation. We can find that both velocity and are described by integrating geology with analysis of
attenuation show anisotropy characteristics due to the poststack distinguishable multiple attributes; 2) For
presence of fractures. However, the maximum velocity meso-scale fractures, we depict the fractures based
anisotropy corresponds to the weighted average of on the attribute parameter inversion result associated
the two fracture strike directions (86 degrees), and the with prestack attenuation; 3) For micro-scale fractures,
minimum attenuation still corresponds to the azimuth the description is mainly based on imaging logging
of meso-scale fractures (80 degrees). The modelling interpretation results, integrating rock physics parameter
results of the both models demonstrate that attenuation test with core slice observation and analysis, and
anisotropy is sensitive to meso-scale fractures but interpreting the reservoir fractures.
insensitive to micro-scale fractures and velocity The method can be summarized as the workflow chart
anisotropy is the weighted average of the responses of in Figure 5:
multiple fracture sets. So we can make use of anisotropy

Imaging logging Core Poststack seismic Prestack gathers Rock sample

data data
Fracture
description of Fault Prestack azimuthal Parameter test
intensity and interpre- Attributes such as frequency gradient
tation curvature, attribute inversion
orientation coherence and
variance Core micro-scale
fracture
Fracture Stress field Attribute optimization Azimuthal characteristic observation
interpretation simulation fitting
Fracture prediction Lithological
based on poststack inversion
attributes Micro-scale
Regional macro-scale Meso-scale fracture
fracture prediction prediction fractures

Multi-attributes and multi-scales fracture

prediction and interpretation

Fig.5 The technical procedure of fracture prediction which combines geology, well logging and seismic comprehensively.

For the real field application of the integrated 2) Describing the fracture intensity and orientation for
technology, some basic data, including imaging logging all core data in the study area, combining the fault data
data, rock cores, poststack seismic data set, prestack interpreted by poststack seismic data with the imaging
seismic gathers and rock samples, are required. The logging interpretation result, simulating the stress field
main technical processes or workflows include: then interpret it, and then the fracture development
1) Utilizing core and imaging logging data to interpret pattern in the region can be obtained;
the orientations and densities of fractures; 3) Calculating the attributes from poststack seismic
85
 Fractured reservoir description

data such as coherence, variance and curvature, applying forms, large variations in axial, and intertwined multi-
attribute fusion and optimization and the fracture directional structural forms so that the structure is
prediction based on poststack attributes are obtained, stressed unevenly and complicatedly which provides
including fracture intensity and orientation; favorable conditions for the fractures developing. The
4) Using the azimuthal gathers to invert azimuthal main target formation is limestone fractured reservoir.
frequency gradient attribute and applying azimuthal Fractures are main flow channel and important
characteristic fitting, and then obtaining the prediction of reservoir space in limestone tight oil reservoirs, which
meso-scale fracture intensity and orientation; controls the distribution and production of oil and
5) Based on the results of rock physics parameter (e.g., gas. It is significantly important to guide the reservoir
P-wave and S-wave velocity) measurements, reservoir exploration and development in the area by studying the
lithology inversion combined with prestack gathers are development characteristics and control factors of the
implemented and the relationship between lithology fractures within reservoirs, evaluating the qualitative and
characteristic and reservoir fractures are evaluated; quantitative parameters of fractures quantitatively and
6) Slicing the rock samples and observing the growth predicting the distribution pattern of fractures.
of micro-scale fractures, and obtaining their development
pattern; 7) Verifying the reliability of macro-scale and Core analysis and stress field modelling
meso-scale fracture prediction by comparing with the Some research have indicated that the formation and
imaging logging and stress field simulation results, development of fractures are influenced and dominated
and thereby an integrated interpretation result of the by many geological factors such as tectonic stress,
developments of macro-scale, meso-scale and micro- lithology characteristic, layer thickness and structure.
scale fractures is established comprehensively. However, geological factors such as lithology, layer
thickness, geological structure and sedimentation,
are internal factors of dominating the formation and
Application to a real field data distribution of structural fractures, while stress field
is the external factor. Therefore, we must analyze the
internal factors of fracture formation by core data of
In this section, an application to a real field data,
the work area at first, then make certain of the external
in which fractures are developed in the reservoir, is
factor by stress field modelling, finally form integrated
implemented by using the proposed method and technical
analysis of the controlling factors of fractures. Figure
procedure. There are many sets of oil-bearing series
6 shows the interpretation results of part of cores and
in the study area L. There lies two major fault zones,
fractures in the study area, where fracture development
and the covering suffers from little folds in the area.
of four types of lithology, including high dip angle
However, it has many types of structures, complicated

Argillaceous limestone, high dip angle structural fractures

mestone, low dip angle shear fractures Marlstone, high dip angle tension fractures Mudstone, nearly horizontal diagenesis cracks
Fig.6 The observation results of part of cores in the study area.

86
 Chen et al.

structural fractures of argillaceous limestone, low dip conduct directional interpretation. By analyzing the
angle shear fractures of limestone, high dip angle tension fracture orientation interpretation using the imaging
fractures of marlstone and nearly horizontal diagenesis logging data in the study area (Figure 7a), we find that
cracks, are displayed. At the same time, we collect the main directions of reservoir fractures are north-east
statistics on fracture development of cores in the study (NE), north-west-west (NWW) and north-west (NW)
area and the results demonstrate that high-dip-angle (60 directions (in Figures 7b-7c).
degrees to 90 degrees) fractures, oblique-dip-angle (30 By combining core observation data with imaging
degrees to 60 degrees) fractures and low-dip-angle (0 logging interpretation results, we numerically simulate
degree to 30 degrees) fractures grow well, where the the stress field. Figure 8 shows the development in plane
distribution frequency of low-dip-angle fractures is up and distribution pattern of fractures within limestone
to 40% while high-dip-angle fractures accounts for 30%. reservoirs predicted by regional stress field modelling,
By analyzing the fracture development with different where the color scale denotes fracture intensity and the
lithologic characteristic, a classified statistic on fracture gray lines denote fault zones. As shown in the figure,
development for the lithology in all zones indicates fractures of limestone reservoirs in the work area are
that the development degree of fractures in the area is distributed along the fault zones and it can be noted that
significantly controlled by lithology, where fractures regional fractures are dominated by fault zones, which
within limestone are most developed gradually, followed provides evidences for analysis of seismic fracture
by sandstone and thirdly mudstone. As suggested prediction results.
above, the study points out that we should focus on the
fracture development in the limestone reservoir and the Fracture prediction using seismic attributes
development angle is low-dip-angle fractures. Although stress field modelling is able to help
The fracture groups and orientations are the basic analyze the origin of fractures, it cannot be applied to
parameters and foundation for well pattern arrangement precisely predict the fracture development. However,
in the low-permeability reservoir hydrocarbon the distribution of underground fractures can be depicted
development. Therefore, the determination of fracture by seismic data. In this section, 3D wide-azimuth
groups and orientations is the basic premise to help seismic data are used to help predict the distribution
describe fracture parameters quantitatively. Generally, of underground fractures. Currently, seismic fracture
the favorable method to determine the extension prediction can be classified into two basic types, i.e.,
direction of fractures is orientational coring. If there prestack and poststack. Prestack prediction mainly makes
aren’t any orientational coring data available, it can use of the azimuthal anisotropy response characteristics
be implemented by using the imaging logging data to of fractures to predict meso-scale fractures by means of

(a) Static electronic imaging Dynamic electronic imaging Imaging result Quantitative calculation

2770

2780

2790

87
 Fractured reservoir description

(b) 0 (c) 0
330 30 330 30

300 60 300 60

270 90 270 90

240 120 240 120

210 160 210 160

180 180

Fig.7 The imaging logging result of zones of interest and statistical charts of fracture orientations.
(a) The first two columns are static and dynamic images. The third column is the result of imaging
interpretation where the red small tadpole symbol denotes fracture development and azimuth. The
fourth column is the result of quantitative calculation which shows the fracture intensity of zones
of interest. (b - c) The diagrams of fracture orientations got by imaging logging interpretation.

1g167
pc1

1q108
1q101
x
1q3 1g167
1g2
1g7
1g001-12 1g10 1q101x 1q104x
1g001-3 1g2 1g10
1g9 1g001-6 1q105x
1g9 1g001-12
1g18
1g18 1g001-6
1g1 1g171 1g001-3 1g001-7
1g001-10 1g001-1
1q102x 1g001-10 1g1
m 1g19 1g001-20
1.8
1g161 1g16
1g16 1g172
1.4
1g161
Industrial 1g21
oil & gas well 1.0
1g21 0 3000 6000 9000
0.6

Fig.8 The development in plane and distribution pattern Fig.9 The plane distribution of reservoir fractures
of fractures predicted by tectonic stress field numerical predicted by using poststack coherency attributes.
modelling.

ellipse fitting of different azimuthal anisotropy attributes. gathers. Firstly, using prestack azimuth gathers to
Poststack prediction mainly takes use of the discontinuity calculate azimuth attributes, and then applying ellipse
of fractures for prediction so that it is mainly used for fitting, finally obtaining fracture orientation and intensity
macro-scale fracture prediction. The poststack attributes (Qian et al., 2007). The seismic attributes sensitive to
associated with fractures contain geometric attributes fractures are reflection amplitude, interval velocity,
such as coherency, variance, azimuth angle, dip angle travel time difference, azimuth AVO gradient, azimuth
and curvature. In this study, coherency attribute is used interval frequency, frequency attenuation, frequency
as the result of poststack fracture prediction. Figure 9 gradient, stack amplitude, amplitude azimuth error
shows the plane distribution of reservoir fractures by and so on. The conventional amplitude attributes are
using poststack coherency attributes. In the figure, the normally influenced by offset and become unstable, and
dark black stripes denote the macro-scale fractures of thereby leading to poor prediction results. Although AVO
fracture zones associated with faults. gradient attribute can be used to predict fractures well,
It mainly takes use of azimuthal anisotropy attribute it can’t be used to distinguish fracture scales due to its
to predict fractures when using prestack wide-azimuth insensitivity to fracture scales. According to the above

88
 Chen et al.

analysis, we introduce attenuation-dependent frequency the intensities of fractures and their orientations as
gradient attribute to help predict meso-scale fractures. well. Figure 11 shows the fracture orientation predicted
Figure 10 shows a laminated chart of the intensity by using prestack frequency gradient (colorful rose
result of meso-scale fracture prediction and the poststack diagram). The predicted orientations are mainly NW
macro-scale fracture prediction result, where the and NE directions as a whole and there are also a few
multicolor denotes the meso-scale fracture intensities. fractures lying in nearly WE direction, which reasonably
In the study area, the result of meso-scale fracture agrees with the imaging logging statistics. In Figure
prediction is mainly divided into east and west regions 12, the imaging logging interpretation result of fracture
(divided by red dashed line in the figure). In the figure, orientation with the prediction result of orientation are
the fractures in west region develop better than those in compared based on two prestack attenuation anisotropy
east region, and two strong-development regions, A and characteristics, and it can be found that the former is in
B areas, can be observed in the study area. Based on the favorable agreement with the latter.
attributes of prestack azimuth gathers, we can predict

A
1q108 1q108

1q3 1g167 1q3 1g167

1g7 1g7
1q101x 1q104x 1q101x 1q104x
1q105x 1g2 1g10 1q105x 1g2 1g10
1g9 1g001-12 1g001-12
1g9
1g18 1g001-6 1g18 1g001-6
B 1g171 1g001-3 1g001-7
1g001-1 1g171 1g001-3 1g001-7
1g001-1
1q102x 1g001-10 1g1 2.4 1q102x 1g001-10 1g1
1g19 1g001-20 1g19 1g001-20
2
1g161 1g16 1g161 1g16
1g172 1.6 1g172

1g21 1.2 1g21

0.8
0 3000 6000 9000 0 3000 6000 9000

Fig.10 A laminated chart of the prediction result using Fig.11 The result of fracture orientation predicted
prestack frequency gradient and the prediction result by using prestack frequency gradient, where the
using poststack coherence, where the multicolor colorful rose diagram shows the statistics on the
denotes the meso-scale fracture intensities. predicted orientations.

(a) 0 (b) (c)

330 30

300 60

270 90

240 120

210 160
180
Fig.12 (a) Fracture orientation interpreted by imaging logging. (b, c) The comparison of fracture orientation predicted by
using two prestack attenuation anisotropy attributes.

Lithologic inversion and interpretation distribution of lithological characters can be calculated

by means of prestack seismic lithologic inversion.
The core analysis above demonstrates that the
Based on the prestack seismic data in the study
development of meos-scale fractures is mainly
area, we can carry out logging-constrained reservoir
influenced by lithological characters. Therefore, the
89
 Fractured reservoir description

prediction and make clear the distribution of reservoir analysis of Mavko et al. (2009), the P-wave velocities
development. We apply lithology interpretation of pure limestone are generally over 5000 meters per
method to conduct interpretation and prediction of the second while the S-wave velocities are over 3000
distribution and thickness of limestone for the study area. meters per second. However, in the study area, the
Figure 13 shows the inversion result: the upper figure is P-wave velocities are about 4500 to 5000 meters per
the probability distribution of limestone by geostatistical second while the S-wave velocities are about 2500 to
inversion, which has a high vertical resolution and a 2800 meters per second. Studies show that the main
good continuity in horizon direction; the lower figure factors leading to the change of limestone velocities
shows the ultimate inverted thickness of limestone, in includes two aspects: 1) Shale content. Whether the
which the different colors represent different thickness sample contains shale or not and the variation of shale
and the thickest is about 18 meters while the thinnestis is content will lead to great changes of P-wave and S-wave
about 3 meters. In the figure, the red dashed line makes velocities. With the increasing of shale content within the
it clear that the limestone in the west of the study area rock, the wave velocity of rock shows a greater degree
develops well, which is in line with the distribution result of decrease; 2) The development of microcracks and
of fracture prediction using prestack seismic attribute. pores. Both microcracks and pores are able to affect the
Thereore, we can verify that meso-scale fractures are wave velocity of limestone and the effects of them are
dominated by lithological factors. the same as that of shale content on wave velocity. The
presence of microcracks and pores can make the P-wave
lq103 lq101x lg2 lg001-3 lg001-6 lg001-7 lg1
1926 2067 2181 2295 2409 2522 2635 2748 2861 2971
and S-wave velocities of sample fall down apparently.
1185 1163 1134 1105 1076 1048 1032 1045 1070 1070 Therefore, analyzing of the core slice is required to
1350 make sure whether there are both factors above or not in
1400
the study area. Figure 14 shows the electron microscopic
Time (ms)

1450
observations on core slice. We can observe that the
1500
microcracks and pores in the rocks of study area develop
1560
1600
well and clay fillers also exist. It verifies further that
there are not only macro-scale fractures dominated by
fault zones but also meso-scale and micro-scale fractures
controlled by diagenesis in the work area. The three
1q108
types of fractures constitute the main reservoir spaces of
limestone reservoirs in the work area.
1q3 1g167
1g7
The application result of limestone fractured reservoirs
1q101x 1q104x in the actual work area demonstrates that the development
1g2 1g10
1q105x
1g9 1g001-12 of meso-scale fractures within the area is mainly
1g18 1g001-6
1g171 1g001-3 1g001-7
1g001-1
influenced by lithological characters. Hence, combining
1g19
1q102x 1g001-10 1g1
18
with the result of rock physics parameter test, we obtain
1g001-20
1g161 1g16 13.5
the thickness of limestone by conventional prestack
1g172
9 inversion and lithological statistical inversion. The
1g21
4.5 lithologic inversion result verifies that the development
0 of lithological characters is in an reasonable agreement
0 3000 6000 9000
with the result of meso-scale fracture prediction. Due
to the influence of limestone thickness, we divide the
Fig.13 The well tie profile of probability distribution of limestone
by geostatistical inversion (upper) and the distribution result meso-scale fracture development areas in the study area
of thickness in horizon direction (lower), in which the different into East and West areas. Meanwhile, comparing the
colors represent different thickness and the thickest is about 18 thickness chart in Figure 13 with A and B areas in Figure
meters while the thinnest is around 3 meters. 10 where significant fractures develop, we can find that
meso-scale fractures mainly lie in the area which has a
Finally, we analyze the development of micro-scale thickness of about 13 meters, which means that meso-
fractures by rock physics parameter test. The core test scale fractures don’t lie in the thickest limestone area.
demonstrates that the P-wave and S-wave velocities Due to the influence of tectonic effect, the fractures
of limestone in the study area are mostly lower than development in the thickest limestone area are macro-
those of conventional limestone. According to the scale fractures associated with fault zones.

90
 Chen et al.

Based on the meso-scale fracture distribution of scale fractures (see Figure 10), the development of meso-
prestack azimuthal frequency attenuation attribute scale fractures are dominated by lithologies as well as
prediction, the main orientations of meso-scale fractures faults. All the evidences demonstrate that the development
are NW and NE directions (see Figure 12) and there of fractures in the study area is mainly dominated by
are also a few fractures lying in nearly WE direction. tectonic origin, followed by lithological origin.
Compared with the prediction result of faults and macro-

Shale

Microcracks
Pores

Fig.14 Electron microscopic observations on core slice. The micro pores and microcracks can be observed
clearly as well as argilliferous filling materials.

Conclusions hence help verity the prediction result. Therefore, the

proposed method and technical procedure of multi-scales
fracture prediction, which combine multiple attributes
The paper begins with rock physics model of multi- synthetically, can provide a good potential reference for
scales fracture and modelling research on anisotropy unconventional fractured reservoir prediction and sweet
characteristic of meso-scale fractures, and the calculation point prediction.
results demonstrate that attenuation attribute can be
used toassess meso-scale fractures well. For this reason,
we propose to predict meso-scale fractures by using Acknowledgements
attenuation-dependent frequency gradient attribute. In
terms of prediction of the development for macro-scale
Thanks for the joint funding of National oil and gas
fractures, based on having mastered the development
major project (No.2011ZX05019-008) and National
angle of fractures and the lithology influence factors,
Natural Science Foundation of China (No.41574108,
we mainly begin with core fracture observation and
No.U1262208) for our studies on methods and
statistical analysis, and then tectonic stress field modelling
techniques. At the same time, thanks for the approval of
is applied to aid analyze the distribution of macro-scale
Research Institute of CNPC Southwest Oil and Gas Field
fractures. The result from tectonic stress field modelling
Branch for the publish of the field work data application
demonstrates that fault zones have apparent controlling
result in the article.
effect on macro-scale fractures. The macro-scale fractures
mainly lie in the vicinity of faults and can be depicted by
geometric attributes such as poststack coherence attribute.
In this paper, we propose a technical procedure of
References
multi-scales fracture prediction by integrating geology,
well logging and seismic information comprehensively, Agersborg, R., Jakobsen, M., Ruud, B. O., and Johansen,
apply multi-scales fracture integrated prediction to the T. A., 2007, Effects of pore fluid pressure on the seismic
objective tight oil reservoirs, and thereby determine the response of a fractured carbonate reservoir: Stud
spatial distribution characteristic of fractures within the Geophys Geod, 51, 89−118.
area. The fracture prediction result reasonably agrees Aki, K., and Richards, P. G., 1980, Quantitative
with the result from imaging logging interpretation, and Seismology, Theory and Methods: W. H. Freeman and

91
 Fractured reservoir description

Co., San Francisco. fluid-saturated rock with aligned fractures: implication of

Behura, J., Tsvankin, I., Jenner, E., and Calvert, A., 2012. fracture size estimation from anisotropic measurements:
Estimation of interval velocity and attenuation anisotropy Geophys Prospect, 51, 381−392.
from reflection data at Coronation Field: The Leading Maultzsch, S., Chapman M., Liu, E., and Li, X., 2007,
Edge, 31, 580−587. Modelling and analysis of attenuation anisotropy in
Chapman, M., 2003, Frequency-dependent anisotropy multi-azimuth VSP data from the Clair field: Geophys
due to meso-scale fractures in the presence of equant Prospect, 55, 627−642.
porosity: Geophys Prospect, 51, 369−379. Mavko, G., Mukerji, T., and Dvorkin, J., 2009, The rock
Chapman, M., 2009, Modeling the effect of multiple sets physics handbook: tools for seismic analysis of porous
of mesoscale fractures in porous rock on frequency- media, 2nd ed.: Cambridge Press.
dependent anisotropy: Geophysics, 74(6), D97–D103. Qian, Z., Li, X. Y., and Chapman, M., 2007, Azimuthal
Chen, S. Q., Li, X. -Y., and Wang, S. X., 2012, The variations of PP- and PS-wave attributes: A synthetic
analysis of frequency-dependent characteristics for study: 77th Ann. Internat. Mtg., SEG, Expanded
fluid detection: a physical model experiment. Applied Abstracts, 184−188.
Geophysics, 9(2), 195−206. Rüger, A., and Tsvankin, I., 1997, Using AVO for fracture
Chichinina, T., Sabinin, V., and Ronquillo-Jarillo, G., 2006, detection: Analytic basis and practical solutions: The
QVOA analysis: P-wave attenuation anisotropy for Leading Edge, 16, 1429−1434.
fracture characterization: Geophysics, 71, C37−48. Schoenberg, M., 1983, Reflection of elastic waves from
Clark, R. A., Benson, P. M., Carter, A. J., and Moreno, C. periodically stratified media with interfacial slip:
A. G., 2009, Anisotropic P-wave attenuation measured Geophys Prospect, 31, 265−292.
from a multi-azimuth surface seismic reflection survey: Shen, F., Sierra, J., Burns, D. R., and Toksoz, M. N., 2002,
Geophys Prospect, 57, 835−845. Azimuthal offset-dependent attributes applied to fracture
Hudson, J. A., 1981, Wave speeds and attenuation of elastic detection in a carbonate reservoir: Geophysics, 67, 355−364.
waves in material containing cracks:Geophysical Journal Tod, S., Taylor, B., Johnston, R., and Allen, T., 2007.
of the Royal Astronomical Society, 64, 133−150. Fracture prediction from wide-azimuth land seismic data
Jakobsen, M., 2004, The interacting inclusion model of in SE Algeria: The Leading Edge, 26, 1154−1160.
wave-induced fluid flow: Geophys J. Int, 158, 1168−1176. Wang, Y., Chen, S. Q., and Li, X. Y., 2015, Anisotropic
Jakobsen, M., and Chapman, M., 2009, Unified theory of characteristics of mesoscale fractures and applications
global flow and squirt flow in cracked porous media: to wide azimuth 3D P-wave seismic data: Journal of
Geophysics, 74(2), WA65−WA76. Geophysics and Engineering, 12(3), 448−464.
Li, X. Y., Liu, Y. J., Liu, E., Shen, F., Li, Q., and Qu, S., Wang, Y., Zheng, D., Li, X. -Y., Cui, Y., Sun, S., Wan, X.
2003, Fracture detection using land 3D seismic data from and Yang, P., 2014, The fracture-cavern system prediction
the Yellow River Delta, China: The Leading Edge, 22, method and its application in carbonate fractured-vuggy
680−683. reservoirs: Geophysical Prospecting for Petroleum (in
Liu, E., Queen, J. H., Li, X. Y., Chapman, M., Maultzsch, S., Chinese), 53(6), 727−736.
Lynn, H. B., and Chesnokov, E. M., 2003, Observation Yin, Z. H., Li, X. -Y., Wei, J. X., Di, B. R., Zhang, S.H.
and analysis of frequency-dependent anisotropy from a and Wu, M.S., 2012, A physical modeling study on the
multicomponent VSP at Bluebell-Altamont field, Utah: J. 3D P-wave azimuthal anisotropy in HTI media: Chinese
Appl.Geophys, 54, 319−333. Journal Geophysics. (in Chinese), 55(11), 3805−3812.
Liu, E., Chapman, M., Varela, I., Li, X., Queen, J. H., and
Lynn, H., 2007, Velocity and attenuation anisotropy: Chen Shuang-Quan, Ph.D, Associate Professor. He
Implication of seismic fracture characterizations: The graduated from Daqing Petroleum
Leading Edge, 26, 1170−1174. Institute with a Bachelor Degree
Liu, E., Zelewski, G., Lu, C., Reilly, J. M., and Shevchek, in geophysics in 1999, from China
Z. J., 2011, Mitigation of overburden effects in fracture University of Petroleum (Beijing) with a
prediction using azimuthal AVO analysis: An example Doctor Degree in Geo-resources and Geo-
from a Middle East carbonate field: The Leading Edge, engineering in 2008, and is undertaking
30, 750−756. teaching and research works now in
MacBeth, C., and Li, X., 1999, AVD: an emerging new the Department of Exploration Geophysics, College
marine technology for reservoir characterization; of Geophysics and Information Engineering, China
acquisition and application: Geophysics, 64, 1153−1159. University of Petroleum (Beijing). His research interesting
Maultzsch, S., Chapman, M., Liu, E., and Li, X. Y., 2003, includes multi-wave seismic exploration, seismic
Modelling frequency-dependent seismic anisotropy in anisotropy, fracture prediction and seismic inversion.

92