Fuel 128 (2014) 373–380

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Experimental study of hydraulic fracturing for shale by stimulated

reservoir volume
Tiankui Guo a,b,⇑, Shicheng Zhang b, Zhanqing Qu a, Tong Zhou b, Yongshun Xiao c, Jun Gao d
a
College of Petroleum Engineering, China University of Petroleum, Huadong 266580, China
b
College of Petroleum Engineering, China University of Petroleum, Beijing 102249, China
c
Department of Engineering Physics, Tsinghua University, Beijing 100083, China
d
International Well Service Company of Greatwall Drilling Company, PetroChina, China

h i g h l i g h t s

 Hydraulic fracturing simulation experiments of shale outcrops were first carried out.
 Fracture morphology was observed for the first time by high-energy CT scanning.
 The effects of multiple factors on fractures propagating in shale play were studied.
 CT scanning images were combined with internal fractures photographs for analysis.
 Hydraulic fracturing of horizontal well was simulated for shale specimens.

a r t i c l e i n f o a b s t r a c t

Article history: ‘‘Stimulated reservoir volume’’ (SRV) makes it possible for commercial production of shale gas by means
Received 28 November 2013 of multistage fracturing of horizontal wells. However, the formation mechanism of effective fracture net-
Received in revised form 11 March 2014 work has not been well understood. The mechanism of fracture propagation in shale with hydraulic frac-
Accepted 13 March 2014
turing needs to be further explored, in order to realize the control on morphology of fracture network
Available online 26 March 2014
with SRV and increase the single well production of shale gas. In this article, the true triaxial test system
was deployed for horizontal well hydraulic fracturing simulation experiments of shale outcrops for the
Keywords:
first time. The effects of multiple factors on propagating rules of fractures of horizontal well in shale with
Shale
Stimulated reservoir volume
SRV were studied, and the fracture morphology of post-fracturing rock cores was observed for the first
Fracture propagation time by high-energy CT scanning using the large-scale non-destructive testing system based on linear
Horizontal in situ stress difference accelerator. The results show that the influence of flow rate (for SRV) on fracture complexity differs when
Flow rate its value falls in different intervals. When the horizontal in situ stress difference is less than 9 MPa, the
hydraulic fracture easily propagates along the natural fractures, forming a fracture network. In this range,
when the stress difference is increased, the appearance of the main hydraulic fracture contributes to
interconnecting more natural fractures, forming a relatively more complex fracture system. Under the
condition of the same horizontal stress difference, if the coefficient of stress difference Kh > 0.25, there
is an obvious trend to form single main fracture. The effects of viscosity of fracturing fluid and flow rate
on the fracture propagation can be expressed by the parameter ql. When the order of magnitude of ql-
value is 109 N m, it is favorable to the generation of fracture network, but too small or too large values
are both harmful. The development and cementing strength of depositional beddings seriously affect the
propagating complexity of fracture network with hydraulic fracturing. Refracturing helps to produce
some new fractures which are different from the initial ones, and better fracturing effect is achieved.
The fracture initiation morphology in perforation fracturing is closely related to the distribution of nat-
ural fractures (depositional bedding) around the perforations. If the pressure curve fluctuates signifi-
cantly, it is indicated that the shale core contains a large number of opened natural fractures or
depositional beddings before fracturing, which causes the serious loss of fracturing fluid in the propaga-
tion process.
Ó 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +86 053286981701.

E-mail address: [email protected] (T. Guo).

http://dx.doi.org/10.1016/j.fuel.2014.03.029
0016-2361/Ó 2014 Elsevier Ltd. All rights reserved.
374 T. Guo et al. / Fuel 128 (2014) 373–380

1. Introduction the research factors were relatively comprehensive. The effects of

horizontal stress, horizontal stress difference coefficient, flow rate,
Shale matrix has very low porosity and permeability, and shale viscosity of fracturing fluid and perforation on fracture propagating
gas has different existential states [1,2], which brings great difficul- morphology were explored for the fracturing of shale horizontal
ties for the exploitation of shale. Hydraulic Fracturing by Stimu- well. Thirdly, the large-scale ‘‘IPT4106D’’ non-destructive testing
lated Reservoir Volume (SRV) [3] is a major technology to system based on linear accelerator was used for the first time to
achieve commercial development of shale gas. The in-depth under- carry out high-energy CT scanning for observing the structure of
standing of propagation mechanism of fractures in shale with SRV complex shale fracture network after treatment. So, the analysis
can contribute to guide the design of fracturing operation, control- about fractures combined with internal fractures photographs
ling the morphology of fracture network of shale with SRV, thus was more accurate. The research findings would provide theoreti-
improving the single well production of shale gas. The key factors cal basis and technical support for the efficient development of
affecting the morphology of post-fracturing network include hori- shale gas reservoirs.
zontal in situ stress difference, rock brittleness and natural fracture
system (sedimentary bedding) [4–6]. In addition, the morphology 2. Experiments
of fracture network can also be affected by the fracturing operation
factors (volume of fracturing fluid, flow rate, the spacing between 2.1. Test apparatus and specimen processing
fracturing segments [7,8]) and fracturing techniques (horizontal
well multistage fracturing, synchronous fracturing, zipper fractur- Experiment apparatuses included the large-size true triaxial
ing and refracturing [9]). hydraulic fracturing simulation system [16] and the large-scale
The hydraulic fracturing simulation experiment is an important ‘‘IPT4106D’’ non-destructive testing system (6 MeV) based on lin-
and accurate tool to study the mechanism of fracture propagation. ear accelerator [27]. The latter provided high-energy computed
So far, many fracturing simulation experiments of coals [10,11], ce- tomography (CT) scanning and digital radiography (DR) scanning
ment blocks [11,13] and sandstones [14,15] have already been car- for large-size work pieces. CT scanning and DR scanning can pro-
ried out globally. The propagation mechanisms of hydraulic vide tomographic images and integral perspective views of speci-
fractures in relevant rocks under certain conditions have been re- mens, respectively.
vealed, and some studies have also been presented to analyze the Specimens were collected from Sichuan shale outcrops in China,
effect of natural fractures on hydraulic fracture propagation belonging from the No. 5 layer of the Triassic Xujiahe Formation in
[14,16–18]. However, lots of experimental studies under triaxial Sichuan Basin. During sampling, the weathered rock layer was re-
loading show that shale rock has its own unique physical and moved. The well-preserved fresh shale outcrops were obtained,
mechanical properties [19–21]. The propagation mechanism of and then they were processed into 30 cm  30 cm  30 cm cubes.
fractures in shale with SRV is not yet clear [22,23], and the reports By using a special chemical glue, the steel liquid injection tube
of true triaxial fracturing simulation experiments on shale cores are with the length of 12.5 cm was fixed to eyehole to simulate the
very rare [24], Zhang et al. [25] carried out the hydraulic fracturing wellbore, and a 5 cm open hole section was formed. To simulate
simulation experiment on shale, but only a few factors had been perforation fracturing, the steel tube was designed with a sealed
investigated. Moreover, this experiment was not able to achieve bottom, a length of 17.5 cm and four perforations.
the precise positioning of fractures propagation morphology. It is The shale mineral analysis showed that the average contents of
very difficult to obtain accurate structural information of complex carbonate, quartz and clay were 13.1%, 41.3% and 39.9%, respec-
fracture network of shale only by the conventional method of tively. As shown by the brittleness index calculated by Young’s
observing the fracture indicated by the tracer through rock break- modulus–Poisson’s ratio method [6] and the ratio of uniaxial com-
ing after hydraulic fracturing [16,17]. Due to the reasons of the large pressive strength to tensile strength [28], the brittleness of the
size, high clay content and abundant natural fractures in shale spec- shale core is high.
imens, the acoustic emission energy is pretty weak during fracture
propagation. Therefore, it is also difficult to locate complicated frac- 2.2. Experiment design
tures by acoustic emission monitoring system [18,26].
In this article, the uniqueness of experiment was showed as fol- In order to investigate the effects of horizontal stress, horizontal
lows. Firstly, the true triaxial horizontal well hydraulic fracturing stress difference coefficient, flow rate, viscosity of fracturing fluid
simulation experiments were conducted for the large-size natural and perforation on the fracture propagating rules in the fracturing
shale cores. Therefore, the experimental results could make more of shale horizontal well, the experiment program was designed as
instructive value to the actual shale gas development. Secondly, shown in Table 1.

Table 1
Experiment program for hydraulic fracturing simulation of shale.

Sample number Triaxial stress (MPa) rm/rH/rh Completion method Displacement (ml/min) Fracturing fluid viscosity ( mPa s)
1# 25/15/12 Open hole 10 2.5
2# 25/15/12 Open hole 50 2.5
3# 25/15/12 Open hole 100 2.5
4# 25/18/12 Open hole 100 2.5
5# 25/18/12 Open hole 50 2.5
6# 25/18/6 Open hole 50 2.5
7# 25/18/9 Open hole 50 2.5
8# 25/18/9 Open hole 100 2.5
9# 25/9/6 Open hole 50 2.5
10# 25/12/6 Open hole 50 2.5
11# 9/12/6 Open hole 50 2.5
12# 25/15/12 Open hole 50 65
13# 25/14/8 Open hole 50 2.5
14# 25/15/12 Perforation 50 2.5
 T. Guo et al. / Fuel 128 (2014) 373–380 375

σH
σh

σν

Fig. 1. Fracture morphology after hydraulic fracturing of specimen 7#: (a) and (b) DR perspective views; (1–4) CT scanning images.

2.3. Experiment procedure

In order to simulate horizontal well fracturing, the direction of σH

the wellbores was always parallel to the sedimentary bedding and
σν σh
along the direction of minimum horizontal stress. The overburden
pressure was first applied in the horizontal direction perpendicular
to bedding planes, and then the maximum horizontal stress was
applied in the horizontal direction parallel to bedding planes. The
red ink was added into the fracturing fluid as tracer. Finally, the
minimum horizontal stress was applied in the perpendicular direc-
tion. The ‘‘IPT4106D’’ testing system was used to observe the mor-
phology of shale fracture network. The fracture structure as
indicated by tracer was directly observed by breaking the rock
specimens.

3. Experiment results and analysis

The high-energy CT scanning results and the statistical results of

the red fractures zone (tracer) were combined to analyze the frac-
ture characteristics. It is found that the CT scanning pictures of shale
after hydraulic fracturing mainly contain four types of fractures.

a. Opened structural fracture before hydraulic fracturing, such

as specimen 7#, as shown in Fig. 1 (The scanning positions of
CT scanning images ‘‘Fig. 1(1–4)’’ are located in (1)–(4) of DR
perspective views ‘‘Fig. 1(a and b)’’, respectively). There is a
clear dividing line between fractures and the matrix. This
type of fracture is presented as a pure black trace with a rel- Fig. 2. Fracture morphology after hydraulic fracturing of specimen 1#: (a) and (b)
DR perspective views; (1) and (2) CT scanning images.
atively large width of about 0.6–2 mm.
b. Opened bedding fractures after hydraulic fracturing, such as
specimen 1# (Fig. 2) and specimen 5# (Fig. 3). This type of d. Shear fractures after hydraulic fracturing, such as specimen
fracture is usually partially or penetratively opened. The 5# (Fig. 3) and specimen 7# (Fig. 1). The fracture faces will
fracture interfaces are usually very clear, which are pre- not engage in situ, and the fracture interface is relatively
sented as one or more relatively straight traces (Fig. 1). fuzzy. The fracture width varies considerably and is about
c. Open type fracture after hydraulic fracturing, namely, trans- 0.4–1 mm, slightly larger than the open type fracture. The
verse main fracture, such as specimen 5# (Fig. 3). In the fracture strike is very complex.
absence of proppant pack, this fracture will close in situ to e. Fracture with the width of about 1 mm caused by linear gum
form ‘‘aligned fracture’’ after pressure relief. The fracture fracturing fluid (65 MPa s), such as specimen 12# (Fig. 4). As
width is very narrow, about 0.1–0.5 mm and the fracture a result of the large opening of fracture, the fracture inter-
interfaces are not clear. faces are very obvious.
376 T. Guo et al. / Fuel 128 (2014) 373–380

σν
σH σh

σH
σν σh

Fig. 3. Fracture morphology after hydraulic fracturing of specimen 5#: (a) and (b) DR perspective views; (1–6) CT scanning images.

can leak into the natural fractures and beddings to reduce the
cohesion of fracture planes as a ‘‘lubricant’’, and reduce or even
eliminate the shear strength of fractures. At the same time, the
fluid pressure could further open natural fractures and beddings,
thus creating conditions for creating a new fracture. The CT scan-
ning image of specimen 2# shows an ‘‘X’’ type fracture for cross-
ing and propagating at a low approaching angle (<30°) and a ‘‘T’’
σH type fracture with a high approaching angle (>60°). At the flow
rate of 100 ml/min, the hydraulic fracture begins to propagate
σν σh from the original opened natural fracture in specimen 3#, and
continues to propagate along the face of this natural fracture. In
this process, the hydraulic fracture mostly crosses the closed nat-
ural fractures and propagates along the opened natural fractures.
The complexity (SRV) of the final fracture network is lower than
that at 50 ml/min.
When the horizontal stress difference increases to 6 MPa, the
obvious transverse fractures are formed, as shown in Fig. 3. The
fracture density of specimen 4# is less than that of specimen 5#.
In the specimen 4#, the interference of natural fractures is not sig-
nificant, and the hydraulic fracture continues to propagate along
the direction perpendicular to the minimum horizontal stress.
There is also an ‘‘X’’ type fracture at a low approaching angle in
CT scanning image of specimen 4#. A complex fracture network
structure is formed around the transverse main fracture of speci-
men 5#.
Fig. 4. Fracture morphology after hydraulic fracturing of specimen 12#: (a) and (b) When the horizontal stress difference increases to 9 MPa, the
DR perspective views; (1) and (2) CT scanning images.
high-density fracture network occurs in the CT scanning image of
specimen 7#, but no obvious transverse fracture is presented. This
specimen is special and the experiment result is closely related to
3.1. Effects of flow rate the highly dense opened natural fractures in the core inside before
hydraulic fracturing, as shown in Fig. 1. A large amount of opened
Three groups of experiments were carried out. Group I: 10 ml/ natural fractures around the wellbore can cause hydraulic fracture
min (specimen 1#), 50 ml/min (specimen 2#), 100 ml/min (speci- to propagate along the natural fractures (Fig. 5(7#)). The hydraulic
men 3#). Group II: 50 ml/min (specimen 5#), 100 ml/min (speci- fracture propagates along the weakly cemented natural fractures
men 4#). Group III: 50 ml/min (specimen 7#), 100 ml/min (Fig. 1(3) and (4)) and the bedding planes (Fig. 1(1) and (2)), form-
(specimen 8#). ing a ‘‘dendritic fracture network’’. When the flow rate increases to
For the horizontal stress difference of 3 MPa, no obvious trans- 100 ml/min, the specimen 8# shows a large number of bedding
verse fractures are traced in the three specimens. At the flow rate fractures (Fig. 5(8#)), which is closely related to many partially
of 10 ml/min, the specimen 1# has obvious characteristics of opened bedding fractures before hydraulic fracturing. Although
propagating along bedding fractures and pre-existing natural the test was performed under the flow rate of 100 ml/min and
fractures, as shown in Fig. 2. The fracture network is the most the horizontal stress difference of 9 MPa, the obvious transverse
complex when the flow rate is 50 ml/min. The fracturing fluid main fracture is still formed.
 T. Guo et al. / Fuel 128 (2014) 373–380 377

Fig. 5. Fractures distribution on the surfaces of different specimens after hydraulic fracturing.

3.2. Effects of in situ stress maximum horizontal stress to cross more natural fractures. Other-
wise, the high-density fracture zone is formed only near wellbore
3.2.1. Influence of horizontal in situ stress difference area, and the corresponding stimulation effect will be greatly
Two groups of experiments were designed. Group I: 3 MPa reduced.
(specimen 2#), 6 MPa (specimen 5#), 9 MPa (specimen 7#),
12 MPa (specimen 6#), the flow rate of 50 ml/min. Group II: 3.2.2. Influence of horizontal stress difference coefficient
3 MPa (specimen 3#), 6 MPa (specimen 4#), 9 MPa (specimen The horizontal stress difference coefficient is defined by the fol-
8#), the flow rate of 100 ml/min. lowing formula:
The experimental results show that when the horizontal stress
K h ¼ ðrH  rh Þ=rh ð1Þ
difference is less than 9 MPa, the hydraulic fracture tends to devi-
ate from the natural fractures to form fracture network at the flow where rH is the maximum horizontal in situ stress; rh is the mini-
rate of 50 ml/min, as shown in Fig. 5(5#) and (7#). When the hor- mum horizontal in situ stress.
izontal stress difference increases to 12 MPa, a single transverse The greater the Kh, the more obvious the horizontal stress dif-
fracture is easily formed, as shown in Fig. 5(6#). For the second ference for the two directions will be. Two groups of experiments
group, the complex fracture network can still be formed in speci- were designed. Group I: 0.25 (specimen 2#), 0.5 (specimen 9#).
men 8# with horizontal stress difference of 9 MPa, but it includes Group II: 0.5 (specimen 5#), 0.75 (specimen 13#), 1 (specimen
a large number of bedding fractures as shown in Fig. 5(8#). 10#).
In addition, we found that the complexity of the fractures does Based on the experiment results, the fracture density of speci-
not increase with the decrease of horizontal stress difference under men 9# is less than that of specimen 2#, and the obvious trans-
the condition of a low level of stress difference, such as at 3 MPa verse fracture is formed. Specimen 10# presents an obvious
and 6 MPa. The reason is that with the increasing of horizontal transverse fracture (Fig. 5) and its fracture density is slightly less
stress difference, the forming of transverse main fracture is benefi- than that of specimen 13#, but both are less than that of specimen
cial to communicate more natural fractures to form a more com- 5#. The number of formed fractures in the interior of specimen 10#
plex fracture morphology. But for a lower stress difference, it is is slightly higher than that of specimen 8# with the same Kh, be-
very easy to deflect, so the hydraulic fracturing fluid inclines to cause the stress difference of specimen 8# is greater.
infiltrate and propagate only along the weakly natural fractures. It can be seen that with the same horizontal stress difference, Kh
Thus, the ideal SRV requires not only the easy deflection of hydrau- controls the propagation direction of hydraulic fracture. The great-
lic fracture, but also to extend farther along the direction of the er the value, the more obvious the propagation trend along the
378 T. Guo et al. / Fuel 128 (2014) 373–380

maximum horizontal stress direction will be. For the measured distribution of natural fractures and sedimentary beddings. The
samples, when Kh > 0.25, there is an obvious trend to form a single opened natural fractures will seriously affect the propagation path
main fracture. of hydraulic fractures. The density, orientation and cementing
strength of closed pre-existing natural fractures are three factors
3.2.3. rH > rm > rh dominating the formation of fracture network. Although the
Under the condition of the tectonic stress for strike-slip fault, hydraulic fracture is generally considered as likely propagate along
namely, rm:rH:rh = 9:12:6 MPa, the fracture propagation law of natural fractures at the low approaching angle (<30°) and cross the
specimen 11# was tested. As can be seen in Fig. 5(11#), due to natural fractures at high approaching angle (>60°), many speci-
the reason that the maximum stress is applied parallel to the direc- mens in this paper also show that the hydraulic fracture intersects
tion of shale bedding, and the stress difference is only 3 MPa be- the natural fracture to propagate at the low approaching angle
tween the overburden pressure and the minimum horizontal (<30°) to form a ‘‘X’’ fracture or cause the opening of natural frac-
stress, a large number of bedding fractures are very easy to form ture at a high approaching angle.
during hydraulic fracturing. In addition, the effects of beddings on fracture propagation can-
not be ignored. A series of experimental results show that if the
3.2.4. Dimensionless net pressure, pnet,D sedimentary bedding of shale is developed many weakly cemented
Beugelsdijk et al. [12] considered that with other conditions beddings will be split considerably along with the appearance of
being the same, the fracture propagation morphology was con- transverse fracture. In the meantime, the ‘‘barrier-type’’ fracture
trolled by both flow rate and horizontal stress difference. A dimen- structure will be generated due to the perpendicular intersection
sionless net pressure can be deployed to judge whether the of both, such as in specimens 5#, 8#, 9# and 12#, thus greatly
hydraulic fractures deviate from the preferred fracture plane, see increasing the SRV.
Eq. (2).
pnet 3.5. Fracturing curve
p0net;D ¼ ; ð2Þ
rH  rh
The pressure curve (fracturing curve) is closely related to frac-
where p0 net,D is the dimensionless net pressure and pnet is the net ture propagation in hydraulic fracturing process for shale. There
pressure. is no obvious correlation between the degree of fluctuation of pres-
For a large dimensionless net pressure, the fracture can more sure curve and the propagation complexity of fractures. The severe
easily deviate from the preferred fracture plane. The smaller the fluctuation of pressure curve indicates that a large number of the
p0 net,D, the easier it is for the transverse fracture to form. However, original opened natural fractures (specimen 2#) or sedimentary
based on the above results, we believe that the fracture propaga- beddings (specimen 6#, 9# and 12#) are present in the specimens
tion morphology is decided by the flow rate, the horizontal stress of shale core, causing considerable fracturing fluid loss in propaga-
difference and the horizontal stress difference coefficient at the tion. In contrary, the propagation of hydraulic fracture in the closed
same time. Thus, the above equation could be corrected as Eq. (3): natural fractures or bedding does not cause large fluctuation of the
pnet pnet rh curve.
pnet;D ¼ ¼ ð3Þ Under the condition of rm > rH > rh, the obvious breakdown
ðrH  rh ÞK h ðrH  rh Þ2
pressure usually indicates that an obvious transverse main fracture
There is a certain limit for the net pressure to influence on the has been formed (sample 4#, 5# and 7#). When the flow rate of
deviation tendency of fracture. When the net pressure is too high, fracturing fluid is great, such as at 100 ml/min, the decreasing
it is very hard for the fracture to deviate from the preferred frac- amplitude of extended pressure will be very small compared with
ture plane. the breakdown pressure in pressure curve. In this condition, the
pressure curve is smooth (sample 3#, 4# and 8#) and does not
3.3. Effects of fracturing fluid viscosity present severe fluctuation.

To understand the effect of viscosity of fracturing fluid on the 3.6. Perforation fracturing
fracture morphology, two groups of experiments based on
65 mPa s linear gel and water were carried out. This type of fractur- In this paper, the perforation fracturing experiment was only
ing fluid is often used in shale reservoirs with high clay content conducted for specimen 14#. We did not perforate on the borehole
and weak brittleness [4,6]. As can be seen in Fig. 4, the fracturing wall, and only made four shallow cavities in the wall facing the
fluid with such a viscosity can easily cross weakly cemented bed- perforations, which would contribute to fracture initiation. Refrac-
dings and natural fractures, leading to the formation of a single turing was continuously carried out twice for specimen 14#. Since
transverse fracture with a greater fracture width. the specimen was not perforated, fracture initiation becomes very
The influence of the flow rate and viscosity of fracturing fluid on difficult, resulting in extremely high breakdown pressure for the
the morphology of hydraulic fracture can be expressed as the prod- first hydraulic fracturing. The shale reservoir has an ultra-low per-
uct of these parameters, ql. Its unit is N m. It is found that the meability, and very little fracturing fluid can leak into the shale
morphologies of hydraulic fractures created are similar for the matrix. So after the initial hydraulic fracturing, the fracture system
same ql-value. For specimen 1# (4.2  1010 N m), specimen 2# will always remain in a high pressure state for a long time. When
(2.1  109 N m) and specimen 12# (5.4  108 N m), the above the subsequent refracturing is carried out, this high-pressure frac-
experiment results show that when the order of magnitude of ture system could cause the hydraulic fractures produced in the
ql-value is 109 N m, it is favorable to the creation of fracture net- subsequent fracturing to deviate from the original fracture path.
work, but too low and too high ql-value are both harmful. Therefore, refracturing can make the hydraulic fracture more easily
to get into contact with new shale area, reducing the breakdown
3.4. Effects of natural fractures and sedimentary bedding pressure of the second and third fracturing. The three fracturing
curves show different peak points, indicating that refracturing does
Based on the experiment results, we found that one of the most produce fractures different from those by the initial fracturing. The
important factors affecting fracture morphology of hydraulic ultra-low permeability of shale matrix is a key factor for the suc-
fracturing for shale specimens is the development degree and the cessful implementation of refracturing in the Barnett shale.
 T. Guo et al. / Fuel 128 (2014) 373–380 379

Although the three hydraulic fracturings were carried out con- Acknowledgements
tinuously for specimen 14#, a single tortuous transverse fracture
was formed. This is closely related to the encounter of only one This work was sponsored by the ‘‘National Science and Technol-
of four perforations and the natural fracture. For the casing com- ogy Major Project’’(Grant No. 2011ZX05048-10HZ). We also thank
pletion of horizontal well, the perforation is designed to ensure Beijing Guhong Technology Co. Ltd. for providing X-ray computed
that hydraulic fracture can communicate more nature fractures tomography technique using large-scale ‘‘IPT4106D’’ non-destruc-
to reduce the initiation pressure and obtain an ideal SRV. tive testing system based on linear accelerator.

References
4. Conclusion
[1] Ross DJK, Bustin RM. Impact of mass balance calculations on adsorption
The true triaxial horizontal well hydraulic fracturing simula- capacities in microporous shale gas reservoirs. Fuel 2007;86:2696–706.
[2] Clarkson CR, Solano N, Bustin RM, Bustin AMM, Chalmers GRL, He L, et al. Pore
tion experiments were carried out for the large-size shale
structure characterization of north American shale gas reservoirs using
outcrops for the first time. The effects of horizontal stress, coeffi- USANS/SANS, gas adsorption, and mercury intrusion. Fuel 2013;103:606–16.
cient of horizontal stress difference, flow rate, viscosity of fractur- [3] Cipolla CL, Warpinski NR, Mayerhofer MJ, Lolon EP, Vincent MC. The
ing fluid and perforation on the fracture propagating rules were relationship between fracture complexity, reservoir properties, and fracture-
treatment design. In: SPE 115769, presented at the SPE Annual Technical
explored. The non-destructive testing system based on linear Conference and Exhibition, Denver, Colorado; September 21–24, 2010.
accelerator was first used to carry out CT scanning to observe [4] King GE. Thirty years of gas shale fracturing: what have we learned? In: SPE
the structure of complex shale fracture network after hydraulic 133456, presented at the SPE Annual Technical Conference and Exhibition,
Florence, Italy; September 19–22, 2010.
fracturing. The propagation rules of fractures in shale are obtained [5] Britt LK, Schoeffler J. The geomechanics of a shale play: what makes a shale
as follows: prospective! In: SPE 125525, presented at the 2009 SPE Eastem Regional
Meeting, Charleston, West Virginia; September 23–25, 2009.
[6] Rickman R, Mullen M, Petre E, Grieser B, Kundert D. A practical use of shale
1. Flow rate has different impacts on fracture complexity in petrophysics for stimulation design optimization: all shale plays are not clones
hydraulic fracturing when its value falls in different intervals. of the barnett shale. In: SPE 115258, presented at the 2008 SPE Annual
For a low flow rate (10 ml/min), the fracturing fluid flows usu- Technical Conference and Exhibition, Denver, Colorado; September 21–24,
2008.
ally along the natural fractures and beddings. When the flow [7] Soliman MY, East L, Augustine J. Fracturing design aimed at enhancing fracture
rate increases to a certain extent (50 ml/min), the free deviation complexity. In: SPE 130043, presented at the SPE EUROPEC/EAGE Annual
from the maximum horizontal stress direction becomes very Conference and Exhibition, Barcelona, Spain; June 14–17, 2010.
[8] Chen Z, Xue CJ, Jiang TX, Qing YM. Proposals for the application of fracturing by
easy for hydraulic fractures. The hydraulic fracture can commu-
stimulated reservoir volume (SRV) in shale gas wells in China. Nat Gas Ind
nicate natural fractures and greatly increase the SRV. If the flow 2010;30(10):30–2.
rate continues to increase to a certain stage (100 ml/min), the [9] Mayerhofer MJ, Lolon EP, Warpinski NR, Cipolla CL, Walser D, Rightmire CM.
excessive net pressure will cause the hydraulic fracture to cross What is stimulated rock volume (SRV)? In: SPE 119890, presented at the 2008
SPE Shale Gas Production Conference, Fort Worth, Texas; November 16–18,
easily the closed natural fractures and make hydraulic fracture 2008.
more likely to propagate along the optimal fracture plane, [10] Du CZ. Study on theoretics of hydraulic fracturing in coal bed and its
reducing the fracture complexity. applications. Doctoral thesis, China University of Mining and Technology;
2008.
2. When the horizontal stress difference is less than 9 MPa, the [11] Huang BX. Research on theory and application of hydraulic fracture weakening
hydraulic fracture tends to deviate from natural fractures to for coal–rock mass. Doctoral thesis, China University of Mining and
form fracture network. The complexity of the generated frac- Technology; 2008.
[12] Beugelsdijk LJL, Pater CJD, Sato K. Experimental hydraulic fracture propagation
tures does not increase with the decreasing horizontal stress in a multi fractured medium. In: SPE 59419, presented at the 2000 SPE Asia
difference under the condition of a low level of stress difference, Pacific Conference, Yokohama, Japan; April 25–26, 2000.
such as at 3 MPa and 6 MPa. [13] Zhou J, Chen M, Jin Y, Zhang GQ. Experiment of propagation mechanism of
hydraulic fracture in multi-fracture reservoir. J China Univ Petrol (Ed Nat Sci)
3. For the same horizontal stress difference, the greater the Kh- 2008;32(4):51–4.
value, the more obvious the propagation trend along the maxi- [14] Casas LA, Miskimins JL, Black AD, Green SJ. Laboratory hydraulic fracturing test
mum horizontal stress direction. For the measured specimens, on a rock with artificial discontinuities. In: SPE 103617, presented at the SPE
annual technical conference and exhibition, San Antonio, Texas, USA;
when Kh > 0.25, there is an obvious trend to form a single main
September 24–27, 2006.
fracture (transverse fracture). [15] Damani A, Sharma A, Sondergeld CH, Rai CS. Mapping of hydraulic fractures
4. The influence of flow rate and viscosity of fracturing fluid on the under triaxial stress conditions in laboratory experiments using acoustic
morphology of hydraulic fracture can be expressed as the prod- emissions. In: SPE 159604, presented at the SPE annual technical conference
and exhibition, San Antonio, Texas, USA; October 8–10, 2012.
uct of these parameters, ql. The experiment results show that [16] Zhou J, Jin Y, Chen M. Experimental investigation of hydraulic fracturing in
when the order of magnitude of ql-value is 109 N m, it is random natural fractured blocks. Int J Rock Mech Min Sci 2010;47:1193–9.
favorable to the creation of fracture network, but too low and [17] Zhou J, Chen M, Jin Y, Zhang GQ. Analysis of fracture propagation behavior and
fracture geometry using a tri-axial fracturing system in naturally fractured
too high ql-value are both harmful. reservoirs. J Rock Mech Min Sci 2008;45:1143–52.
5. The development degree and cementing strength of sedimen- [18] Liu JZ, Gao LS, Zhang X. Observations on the acoustic emission in an in-house
tary bedding of shale will seriously affect SRV. hydrofracturing simulation experiment. Acta Petrolei Sinica 1990;11(2):73–8.
[19] Sarout J, Molez L, Guéguen Y, Hoteit N. Shale dynamic properties and
6. Refracturing does produce fractures different those by the ini- anisotropy under triaxial loading: experimental and theoretical
tial fracturing in shale, increasing SRV. For the casing comple- investigations. Phys Chem Earth Parts A/B/C 2007;32:896–906.
tion of horizontal well, the perforation is designed to ensure [20] Muniz ES, Fontoura SAB, Soares AC. CU triaxial tests on north sea shale. In:
ARMA-05-814, presented at the 40th US. Symposium on Rock Mechanics
that hydraulic fracture can communicate more nature fractures (USRMS), Anchorage, Alaska, USA; June 25–29, 2005.
(beddings) to reduce the initiation pressure and to obtain an [21] Ewy RT, Bovberg CA, Stankovich RJ. Shale triaxial strength alteration due to
ideal SRV. brine exposure. In: ARMA-08-304, presented at the 42th U.S. Symposium on
Rock Mechanics (USRMS), Francisco, California, USA; June 29–July 2, 2008.
7. There is no obvious correlation between the degree of fluctua-
[22] Jeffrey RG, Bunger A, Lecampion B, Zhang X, Chen ZR, As AV, Allison DP, Beer
tion of pressure curve and the propagation complexity of WD, Dudley JW, Siebrits E, Thiercelin MJ, Mainguy M. Measuring hydraulic
fractures. The severe fluctuation of pressure curve indicates that fracture growth in naturally fractured rock. In: SPE 124919, presented at the
a large number of the originally opened natural fractures or SPE annual technical conference and exhibition, New Orleans, Louisiana, USA;
October 4–7, 2009.
sedimentary beddings are present in the specimens of shale [23] Weng X, Kresse O, Cohen C, Wu R, Gu H. Modeling of hydraulic fracture
core. network propagation in a naturally fractured formation. In: SPE 140253,
380 T. Guo et al. / Fuel 128 (2014) 373–380

presented at the SPE hydraulic fracturing technology conference and annual technical conference and exhibition, San Antonio, Texas; September
exhibition, the woodlands, Texas; January 24–26 2011. 29-October 2, 2002.
[24] Wand Y, Miskimins J L. Experimental Investigations of Hydraulic Fracture [27] Xiao YS, Hu HF, Chen ZQ, Zhang L, Yang G, Yong T. Application of large
Growth Complexity in Slickwater Fracturing Treatments. In: SPE137515, industrial computed tomography in nondestructive testing of key components
presented at the SPE tight gas completions conference, San Antonio, Texas; of railway vehicles. Comput Tomogr Appl 2009;18(3):72–8.
November 2–3, 2010. [28] Goktan RM, Yilmaz NG. A new methodology for the analysis of the relationship
[25] Zhang X, Jiang TX, Jia CG, Zhang BP, Zhou J. Physical simulation of hydraulic between rock brittleness index and drag pick cutting. J South Afr Inst Min
fracturing of shale gas reservoir. Petrol Drill Tech 2013;41(2):70–4. Metall 2005;105:727–33.
[26] Maxwell SC, Steinsberger N, Zinno R. Microseismic imaging of hydraulic
fracture complexity in the barnett shale. In: SPE77440, presented at the SPE