Sei sulla pagina 1di 29

SPE-174023-MS

Volumetric Analysis of Two-Phase Flowback Data for Fracture


Characterization
Y. Xu, O.A. Adefidipe, and H. Dehghanpour, University of Alberta; C.J. Virues, Nexen Energy ULC

Copyright 2015, Society of Petroleum Engineers


This paper was prepared for presentation at the SPE Western Regional Meeting held in Garden Grove, California, USA, 2730 April 2015.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
Analysis of early-time production data obtained during the flowback period presents the earliest
opportunity to characterize a stimulated reservoir volume (SRV). Previous studies have developed
analytical models/methods to analyze two-phase flowback data for fracture characterization. However, the
mechanisms responsible for the early-time gas production in shales are poorly understood. The objective
of this paper is to understand the mechanisms responsible for early gas production and develop a
mathematical model to estimate effective fracture volume.
The study incorporates a comprehensive field data analysis from 8 wells of a single well pad completed
in the Horn River Basin. Firstly, we develop several diagnostic plots [production rates, cumulative gas
production (Gp) vs. cumulative water production (Wp), GWR vs. Gp] to identify the early time trends/
signatures. The Gas-Water-Ratio (GWR) plots from the wells considered indicate a V-shape trend,
dividing the flowback data into two regions: 1) Early Gas Production-EGP and 2) Late Gas ProductionLGP. Water rate increases during EGP, resulting in a decreasing GWR curve. Production then rolls over
to the gas dominant phase with a positive GWR slope.
Secondly, we introduce a method to estimate effective fracture volume by assuming a simplistic
two-phase tank model for the fracture system. Conventional p/Z analysis shows that the fracture network
can be approximated by a tank model during the EGP phase and that the dominant main mechanisms
during EGP include the expansion of initial free gas in fractures (IGIF), fracture closure and the expansion
of residual frac fluids (water). Effective fracture volume is calculated using a modified material balance
approach for the two-phase system. The material balance approach enables the estimation of effective
fracture volume regardless of the fracture geometry. Finally, the proposed model is validated by numerical
simulation and is applied to estimate effective fracture volume using field production data.

Introduction
Shale gas reservoirs have gradually become an important source of hydrocarbon due to the dwindling
supply of hydrocarbon from conventional reservoirs coupled with rapidly increasing energy demands. The
United States and Canada are reputed to have technically recoverable shale gas reserves of up to 1161 Tcf
and 573 Tcf, respectively (2013). A combination of horizontal well technology and hydraulic fracturing
has made exploitation of these hydrocarbon resources possible. After drilling horizontal wells, high

SPE-174023-MS

pressure fracturing fluids are injected into the wells to create fractures, which largely increase permeability of shale plays. The fracturing fluidsare recovered during a post stimulation period known as
flowback. The flowback data, although usually neglected in traditional production data analysis, present
the earliest opportunity for characterizing the hydraulic fractures and providing useful insights into
fracture design and long term production forecasts.
Abbasi et al. (2012, 2014) presented a series of diagnostic plots that separate flowback data of tight
oil/gas reservoirs into three distinct regions. The first region, which occurs during the very early time,
indicates single-phase water flow. They developed a rate transient model that adequately describes the
first region, estimating fracture permeability and storage coefficient by history matching early-time
flowback data. Some other studies (Crafton and Gunderson, 2006; Crafton and Gunderson, 2007) have
also demonstrated the use of high frequency single-phase flowback data for estimating fracture permeability and conductivity. However, the flowback data from the shales studied in this paper do not follow
this trend. The single-phase region does not occur; instead showing a trend of instant two-phase
production (gas and water) once the wells are opened. Other authors have also observed a unique V-shape
trend in GWR plots when analyzing flowback data of similar well pads in the Horn River Basin (Ghanbari
et al., 2013; Abbasi et al., 2012). These GWR plots are based on the diagnostic plots first presented by
Ilk et al. (2010) to analyze tight reservoirs. Using similar diagnostic plots, Adefidipe et al. (2014) divided
shale gas flowback data into two regions: (1) Early Gas Production-EGP and (2) Late Gas ProductionLGP. The EGP region is characterized by a negative slope on the GWR diagnostic curve, indicating free
gas depletion from fracture network. The LGP region occurs once significant gas saturation builds up in
the secondary fractures connecting to the primary hydraulic fractures.
Flowback is essentially a two-phase process. As a result, existing single-phase dual/triple porosity
models have to be modified to account for multiphase flow during the fracture flowback. This is especially
the case in shale gas reservoirs where instant multiphase flow is observed once the wells are opened
(Adefidipe et al., 2014). Recently, several methods have been presented to analyze transient multiphase
flowback data. Ezulike et al. (2014) modified existing dual porosity models by incorporating an explicitly
determined dynamic relative permeability (DRP) function which accounts for the two-phase flow of gas
and water. In addition, several techniques using stochastic history matching and multi-phase type-curve
matching have been proposed recently to analyze flowback data (Clarkson and Williams-Kovacs, 2013;
Williams-Kovacs and Clarkson, 2013). Li et al. (2013) carried out simulation studies by varying several
fracture parameters to develop a correlation between early gas production and the key fracture parameters
in a shale reservoir. They presented diagnostic plots which allow comparison of fracture parameters from
different wells and/or reservoirs but are not able to characterize the hydraulic fractures quantitatively.
Alkouh et al. (2014) combined two-phase flowback data with long term gas production data to analyze
shale gas reservoirs. They presented a new method to estimate effective fracture volume from flowback
water production data. By conducting several simulation runs, they concluded that gas is the dominant
phase in the system and that water production is driven by gas expansion. However, the contribution of
fracture closure as a drive mechanism was ignored in their analysis.
Diagnostic plots of water gas ratio (WGR) by Zhang and Ehlig-Economides (2014) separate flowback
data into two regions. By analyzing data from 32 wells from the Horn River and Barnett Shale formations,
they identified two apparent trends in the slopes of the WGR plots. The chage in slopes is thought to
signify two separate fracturing fluid production mechanisms; the negative half slope signifying drainage
followed by a negative unit slop signifying vaporization. Another study by Jones et al. (2014) also
suggests the existence of different production mechanisms during the flowback process. They developed
a method to estimate the initial reservoir pressure by calculating well flowing pressure (pwf). In the
analysis, calculated pwf is initially above the reservoir pressure due to frac charge and falls gradually
until it levels out near the time of the first hydrocarbon production. This minimum point is the estimated

SPE-174023-MS

initial reservoir pressure and usually corresponds to the time when the first drop of hydrocarbon is
produced.
Overall, the mechanisms responsible for instant gas production in some shale gas reservoirs are poorly
understood. This early-time behavior is crucial to fracture/flowback design because a significant percentage of recoverable fracture fluid is produced in this period. The EGP data is often masked by wellbore
effects, resulting in inaccuracy when analyzed by existing dual/triple porosity models. Furthermore,
uncertainties in the input parameters for the current flowback models results in the non-uniqueness of the
results. This paper seeks to understand the mechanisms responsible for the early-time production, and to
develop a method to estimate effective fracture volume using a material balance approach. The key
advantage is that it enables the estimation of the connected fracture volume from flowback production data
regardless of the fracture geometry. Fewer input parameters reduce the uncertainties associated with the
current flowback analysis models which try to model the complex fracture networks.
The first portion of this paper qualitatively interprets actual field data collected from 8 wells from a
single pad drilled and stimulated in the Horn River Basin. Various diagnostic plots are developed based
on the work of Ilk et al. (2010). These plots describe the flowback behavior in the shale gas wells
considered. The production signatures indicate that the fracture system behaves like a closed system
during the first few hours of flowback. The initial production is from the gas and water initially in place
within the fractures with negligible influx from the matrix system. The second portion of this paper
conducts a volumetric analysis on water and gas production data to estimate the effective fracture volume
by a simple material balance technique. The fracture system is analyzed as a two-phase tank model by
assuming negligible gas influx from the matrix. Finally, the proposed material balance approach is
verified by analyzing production data from 1) a simulated fractured well and 2) a multi-well pad
completed in the Horn River Basin.

Section 1: Reservoir and Well Pad Description


The Horn River Basin is located in the northeastern part of British Columbia and extends into the
northwest territories of Canada (Johnson et al., 2011). Mineralogical studies of the Horn River shales
(Chalmers et al., 2012) show that they are characterized by a high quartz content; thus they are excessively
brittle and have a high amount of pre-existing natural fractures. Fig.1 shows its stratigraphic section: it is
of Devonian age and belongs to Western Canada Sedimentary Basin (WCSB). The Horn River Basin can
be divided into three shale formations from top to bottom: the Muskwa (MU), the Otter Park (OP) and
the Evie (EV) shale formations. Fig. 2 shows the layouts of the well pad X studied. It includes four wells
completed in each of the Muskwa and Otter Park formations. Table 1 summarizes completion/fracture
designs, including number of fracture stages, number of perforation clusters per stage, horizontal well
lengths, flowback time, total injected volume (TIV) and total load recovery (TLR) for the eight wells
studied.

SPE-174023-MS

Figure 1Stratigraphic section of Devonian-Mississippian strata (Gal and Jones, 2003)

Figure 2Layout of well pad X in the Horn River Basin


Table 1Completion/Fracture design parameters for well pad X

Well ID
A
B
C
D
E
F
G
H

Formation
Name

Fracture
Stages

Perforation
Clusters

Horizontal
Well Length
(m)

TIV (m3)

Flowback
Time (hrs)

TLR (m3)

Otter Park
Muskwa
Otter Park
Muskwa
Otter Park
Muskwa
Otter Park
Muskwa

15
17
20
15
20
15
20
17

4
4
1
1
4
4
1
1

1500
1700
1900
1400
1970
1500
1900
1600

60590.0
66246.0
75504.0
69673.0
54217.0
43927.0
58678.0
54217.0

2463
2024
948
1910
1996
2459
1372
789

12185
5030
5020
7882
9147
12186
6914
4453

Section 2: Diagnostic Plots


Ilk et al. (2010) suggested a workflow to analyze flowback data in tight/shale reservoirs. The workflow
includes generating diagnostic plots of production rates (gas and water) and GWR from flowback data.
In order to understand flowback behavior in shales, we apply the diagnostic plots on well pad X to identify
production signatures during the flowback.

SPE-174023-MS

Fig. 3 shows gas and water production rates in the first 200 hours for the 8 wells studied. A surprising
trend of immediate gas breakthrough is observed in the production rate plots. This instantaneous
two-phase production is observed in all the wells analyzed. This is different from tight gas/oil reservoirs
where a single-phase (water) region is observed during the first few hours of flowback (Abbasi et al.,
2014). As a result, existing single-phase flowback models derived for tight reservoirs (Abbasi et al., 2014;
Crafton and Gunderson, 2006, 2007) are not applicable in shales. In addition, frequent changes in choke
sizes at the beginning of flowback results in noisy rate plots; therefore, we are unable to identify
flowback regions accurately using the production rate plots. Instead, a method to divide shale gas
flowback data into two regions using the producing gas-water-ratio (GWR) will be presented here.

SPE-174023-MS

Figure 3Early-time flowback production rates (gas and water) for Pad X. Instantaneous two-phase flow is observed after opening the
wells. The data are very noisy due to frequent choke size changes and wellbore storage effects.

SPE-174023-MS

Fig. 4 shows GWR plots for the wells considered. Generally, a striking V-shape behavior is observed
in the GWR plots: it is characterized by a negative sloping GWR curve at first, immediately followed by
an increasing/positive sloping GWR trend. In the pad analyzed, five wells (Well A to E) show remarkable
V-shape trend; Well G shows a delayed V-shape response; Well F and H show discontinuous V-shape
trends: the GWR first follows a V-shape trend, then reaches a higher value which does not follow the
V-shape. This V-shape trend has also been reported in other Horn River wells (Abbasi, 2013; Ghanbari
et al., 2013). A similar trend is also observed in GWR-time plots (Adefidipe et al., 2014). On this basis,
flowback data can be divided into two distinct regions: Early Gas Production (EGP) and Late Gas
Production (LGP). Water rate increases in EGP, resulting in the decreasing GWR. Production then rolls
over into LGP region where gas production dominates, resulting in an increasing GWR. The LGP region
has been discussed in several papers (Ilk et al., 2010; Ezulike et al., 2014) and is generally understood to
represent bi-linear gas flow in the fracture and matrix system. In conventional reservoir analysis, LGP
region is usually the first identifiable flow regime once the wellbore storage effects become negligible. In
tight gas reservoirs, this LGP region occurs once enough gas saturation builds up in the secondary fracture
network. This gas build up causes the flow of gas from the matrix through the secondary fractures into
the primary fractures.

SPE-174023-MS

Figure 4 Diagnostic GWR plots for pad X. Five wells (Well A to E) show remarkable V-shape trend; Well G shows a delayed V-shape
response; Well F and H show discontinuous V-shape trend.

SPE-174023-MS

There is usually a single-phase (water) region in tight gas reservoirs before significant gas saturation
builds up in the secondary fractures. However, the shales considered in this study do not show this
single-phase region. They instead show a surprising trend of instantaneous gas production once the wells
are opened. This suggests the presence of initial free gas saturation in the primary fracture network and
within pre-existing natural fractures. This hypothesis is backed by imbibition experiments conducted on
samples from cores in the same reservoir (Dehghanpour et al., 2012, 2013; Makhanov et al., 2014). Gas
is released into the fractures during the shut-in period by forced and spontaneous water imbibition and also
by the redistribution of free gas initially in the natural fractures. An extended shut-in period following the
flowback operation gives enough time for the imbibition and redistribution processes to take place.
Possible reason for the negative slope of the GWR curve during the EGP region is water and free gas
depletion from a relatively closed system with negligible influx of gas from the matrix system. Production
during this EGP region is primarily from the expansion of free gas in the primary fractures. If the roll
over time from the EGP to the LGP region can be estimated, early time data during the EGP region can
be analyzed using gas material balance techniques to estimate initial fracture volume and/or free gas
volume initially in the fractures. In Appendix A, we present a methodology to model mathematically the
GWR data to estimate the roll over time from the EGP to LGP. We believe that the roll over time is a
key fracture performance indicator and is crucial in the analysis of flowback data. The V-shape behavior
and roll over time can be correlated to the fracture/completion parameters and can be used as a guide
in future development plans.

Section 3: Two-phase p/Z analysis during early-time fracture flowback


The decreasing GWR trend suggests the possibility of modeling the fracture network as a two-phase tank
system during the EGP phase. To test this hypothesis, we conduct a gas mole balance analysis on flowback
production data using p/Z plots. Eq. 1 is the gas mole balance equation derived for a volumetric (closed
tank) gas reservoir where the volume of gas in the reservoir always remains constant because gas always
expands to fill the available reservoir pore volume. It indicates that a plot of p/Z vs. Gp will yield a straight
line for a volumetric/closed reservoir where production is driven by the expansion of free gas only
(Ahmed, 2006). The slope of the straight line is directly related to the volume of gas initially in place at
reservoir conditions while the x-axis intercept reflects the volume of gas initially in place at standard
conditions.
(1)

Several studies (Singh, 2013; Rahman et al., 2006; Moghadam et al., 2011) have considered improved
gas mole balance equations for unconventional gas reservoirs that take into account several reservoir drive
mechanisms. Fig. 5a graphically highlights possible deviations from this straight line relationship for
non-volumetric gas reservoirs such as over-pressured reservoirs, aquifer supports and CBM reservoirs
(Moghadam et al., 2011). In these reservoirs, the tank volume continuously changes that leads to the
deviation of the straight line relationship (Eq. 1). Futher, Ramagost et al., (1981), King (1993) and
Rahman (2006) provided mamthematical fomualtions for these deviations.

10

SPE-174023-MS

Figure 5(a) p/Z vs. Gp plots for different reservoirs (Moghadam et al., 2011) (b) p/z vs. Gp plots for a naturally fractured reservoir
(Kuchuk et al., 2014)

Kuchuk et al. (2014) conducted a similar p/Z analysis for naturally fractured reservoirs. The authors
identified three different depletion patterns by observing the changes in the slope of the p/Z plots as shown
in Fig. 5b. The first flow regime observed is a fracture dominated pseudo steady state flow. During this
phase, fluid is depleted from the primary fracture network, with negligible support from the matrix. The
x-axis intercept of the straight line represents the initial free gas in the fracture network. This first region
is typically observed in wells with a finite fracture volume where the formation matrix is very tight. In
cases where there is a significant secondary fracture volume, a second flow regime indicated by a change
in the slope is observed. This slope change corresponds to depletion from the secondary fracture network.
As the matrix contribution becomes more significant, there is another deviation, which signifies the
beginning of matrix depletion.
Fig. 6 shows the flowback p/Z analysis for the 8 wells analyzed in this study. Generally, we observe
a linear p/Z trend with a constant slope during the first few hours of flowback. The linear behavior at the
beginning supports the hypothesis that the fractures can be viewed as a closed tank at early times and
that gas expansion is the predominant drive mechanism during EGP. The x-axis intercept of the
extrapolated line should give an estimate of the initial free gas volume in the fracture network at standard
conditions. As the flowback process continues into the LGP region, the closed tank system assumption no
longer applies due to gas influx into the fractures from the matrix system. The change in slope signifies
the onset of matrix depletion. This behavior is especially seen in wells B and C shown in Fig. 6. Some
wells (such as Well E and G) show a slightly delayed response before the linear trend is observed. We
believe this response is caused by the effects of fracture closure and significant water production during
the early times. The two effects can violate the assumption of constant gas chamber volume which is
required for observing a linear behavior. Other possible explanation could be the effect of wellbore storage
on masking of early-time fracture behavior.

SPE-174023-MS

11

Figure 6 p/Z analysis for flowback production data for Pad X.

12

SPE-174023-MS

Fig. 6 also shows another trend in the p/Z plots, which may be peculiar to flowback production data.
The p/Z plots of some wells (A, D, F and H) show a rightward shift of the initial straight line (shown
by arrows in Fig. 6). This unexpected rightward shift could be explained in two ways. First, the right
shifted line has the same slope as the original one, but gives a larger x-intercept. This suggests that the
gas volume at reservoir conditions remains the same but increases at standard conditions. This suggests
a sudden gas amount increase in the same space underground. One possible explanation is shale gas
desorption from matrix-fracture interface. During flowback, pressure drops to a point where gas may
desorb from the fracture-matrix interface, resulting in this rightward shift. Second, the rightward shifted
line indicates that there is some gas produced without significant p/Z depletion. This may be due to
substantial communication between the fracture networks of adjacent wells within the same well pad. The
inter-wellbore communication has the effects as a pressure source, causing the rightward shift line on
the p/Z plots. This suggests that not all fractured horizontal wells may be representative candidates for the
simplified mole balance approach for analyzing flowback data. Wells that show a significant distortion in
the p/Z plots (due to gas desorption effects, inter-connectivity of fractures, wellbore storage effects, etc)
may require more robust analytical methods.

Section 4: Material Balance Equation for Fracture System


The p/Z analysis above shows that fractures can be approximated by a two-phase tank model in EGP
phase. On this basis, we develop a material balance equation applicable to early-time two-phase
production in shales. Fig. 7 explains schematically the material balance concept. The control volume is
the effective fracture system that includes all active/connected fractures (including primary hydraulic
fractures, secondary fractures and natural factures) in the same pressure system. The effective fracture
system is initially saturated with initial free gas and water. After opening the wells, gas and water
production are driven by the expansion of initial free gas in fractures (IGIF), expansion of water and
fracture closure. Several studies (Singh, 2013; Rahman et al., 2006; Moghadam et al., 2011) have
considered advanced gas material balance equations for unconventional gas reservoirs that take into
account several reservoir drive mechanisms.

Figure 7Schematic illustration of the material balance concept in EGP. Gas and water are produced from three combined mechanisms: gas expansion, water expansion and fracture closure.

(2)

SPE-174023-MS

13

Fracture closure is a function of the fracture stiffness, defined by the reciprocal of fracture compliance
(Craig, 2006). Assuming that the fracture stiffness remains constant during the EGP, fracture closure term
can be defined as
(3)
Where Pf is the fracture pressure and Pc is the minimum pressure required to keep the effective
fractures open. Substituting Eq. 3 into Eq. 2 gives
(4)
Assuming water compressibility remains constant during EGP,
becomes:

, Eq. 4
(5)
(6)

Notice that GfiBgi and WfiBwi are volumes of initial gas in effective fractures (IGIF) and initial water
in effective fractures (IWIF) at initial conditions before flowback. They are both related to initial effective
fracture volume (Vfi) by
(7)
Here, Swi and Sgi are the average gas and water saturation in the fractures before flowback. Thus, Eq.
6 can be simplified using the volume of initial free gas in effective fractures (IGIF):
(8)
Eq. 8 can be simplified by defining the following variables:
Underground withdrawal,
(9)
Expansion coefficient,
(10)
Thus,
(11)
Eq. 11 is the final form of the material balance equation used in our study. It predicts a linear
relationship between the Underground withdrawal (F) from the effective fracture system and the total
Expansion coefficient (E). The expansion coefficient E is a dimensionless term that accounts for all three
drive mechanisms considered. The first term,
, represents gas expansion effect. The second term,
, combines the effects of water expansion and fracture closure. Eq. 11 shows that
plot of Underground withdrawal (F) vs. Expansion coefficient (E) yields a straight line which passes
through the origin. Volume of IGIF can be obtained from the slope of the straight line. Furthermore,
effective fracture volume (Vfi) can be estimated by

14

SPE-174023-MS

(12)

Section 5: Verification of the Proposed Material Balance Model by


Simulation
In this section, the proposed material balance model is used to analyze simulated production data. Rather
than modeling the actual field production data, the objective here is to verify the proposed model by
estimating the effective fracture volume of a synthetic fractured well by using the simulated production
data. The numerical model is developed following the approach presented by Cheng (2012a) in simulation
studies on fractured reservoirs. The model considers two-phase flow of gas and water toward a fractured
horizontal well in a reservoir. Basic reservoir, fracture and well properties are listed in Table 2. The model
takes advantage of the symmetrical configuration of the system by modeling a quarter of reservoir volume
around two hydraulic fractures (two stages). Fig. 8 shows the 3D view of the simulation model built: 61m
in x-direction, 110m in y-direction and 17 m in z-direction. Two hydraulic fractures run perpendicular to
the horizontal wellbore with a fracture spacing of 30.5 m and half-length of 110 m. The actual effective
fracture volume (33 m3) is calculated from the designed fracture geometry.
Table 2Fracture and reservoir properties for simulation case
Parameters

Values

Matrix Permeability
Fracture Permeability
Matrix Porosity
Fracture Porosity
Fracture Thickness
Flowing Wellbore Pressure
Initial Reservoir Pressure

0.0001 md
2000 md
20 %
60 %
0.91cm
6800 kPa
20680 kPa

Figure 8 3D view of the numerical model. Two hydraulic fractures are symmetric and run perpendicular to the production wells.

Fig. 9a shows the gas formation volume factor (FVF) and viscosity used in the simulation model. The
numerical model is initialized by injecting 100 m3 of water into the existing fractures to increase the
fracture pressure to approximately 48263 kPa which is close to the fracturing pressure. This mimics the

SPE-174023-MS

15

reservoir conditions immediately after fracturing. Then the wells are shut-in for 54 days before production. The extended shut-in period allows for sufficient time for the imbibition process. Water imbibes into
the matrix through both forced and spontaneous imbibition while gas is expelled from the matrix into the
fracture network. This causes a gradual build-up of free gas saturation in the fracture network. At the end
of shut-in time, the average gas saturation (Sgi 0.35) is calculated by averaging the gas saturations in
all the fracture grid blocks. The wells are then produced at a constant drawdown of 6800 kPa. Fig. 9b
shows the simulated production rates which indicate instant gas production once the wells are opened, in
agreement with observations from field production data. Fig. 9c shows that the simulated production data
follow the same V-shape GWR trend as field production data. Finally, the proposed mathematical model
will be validated by comparing the effective fracture volume obtained from analyzing production data and
the actual (calculated) fracture volume from the fracture geometry.

Figure 9 (a) Gas formation volume factor (FVF) and viscosity used in the simulation model (b) Production rates plot of simulation
model (c) V-shape GWR and (d) Material balance equation of the simulation model

Eq. 11 is used to analyze the production data from the numerical model. Fig. 9d shows a straight line
relationship as predicted by the material balance equation applicable during the EGP phase of flowback.

16

SPE-174023-MS

Please note that there is a positive intercept on y-axis which is expected by Eq. 11. This intercept may be
due to removal of a few data points at the beginning affected by wellbore storage effects. The slope of
the line indicates that the volume of IGIF is 12.7 m3. From Eq.12, the effective fracture is estimated at
36 m3, which is a reasonable match to the actual (calculated) fracture volume of 33 m3. Therefore, our
proposed material balance approach should give reasonable results when analyzing early-time flowback
data.

Section 6 Field Application of the Proposed Material Balance Approach


The results from analyzing simulated flowback production data demonstrate the feasibility of the proposed
model in estimating effective fracture volume. In the final step, we demonstrate the field application of
the proposed material balance model. Eq. 11 suggests a linear relationship between two-phase Underground Withdrawal (F) and total Expansion Coefficient (E), where the slope represents IGIF. However,
analysis of flowback data in EGP phase using Eq. 11 requires the average initial free gas saturation in
effective fractures (Sgi) which is unknown in field cases. Therefore, the field application brings up an
interesting challenge in that either the initial free gas saturation or IGIF must be determined by
independent methods. Here we demonstrate the application of the material balance approach to field data
under three different scenarios: (1) Effective fracture volume is estimated by assuming reasonable average
initial free gas saturation; (2) A sensitivity analysis is done to observe the effects of initial gas saturation
on the effective fracture volume. (3) The material balance model is constrained by the IGIF determined
from the p/Z plots to estimate the initial free gas saturation and the corresponding effective fracture
volume.
Although gas builds up in the fractures by water imbibition during the shut-in period, the large volume
of water injected ensures that a significant portion of the fractures is saturated with water. Fig. 10 shows
the material balance analysis for the 8 wells assuming an initial free gas saturation (Sgi 15 %). It should
be noted that the data correspond to the production during the EGP phase only. A linear relationship is
observed between the Underground Withdrawal (F) and the Expansion coefficient (E) as predicted by Eq.
11. The key output is the slope of the straight line that can be interpreted to estimate IGIF. However, some
wells (such as well B, C, E and H) show a small y-axis intercept which is not expected from the derived
material balance equation (Eq. 11). These tiny intercepts can be viewed as a skin effect arising from the
data points ignored at the very beginning of flowback. Well G shows a relatively large intercept on y-axis.
Recall that Well G shows a delayed V-shape GWR response. This suggests that the tank model
assumption is not applicable for Well G at the very early times. Thus, we neglected the early-time data
points, resulting in this non-zero y-intercept.

SPE-174023-MS

17

Figure 10 Material balance analysis for well pad X (Assuming Sgi 15 %). Plots of F vs. E show straight lines across the origin where
the slopes incicate IGIF. Well G shows a non-negligible y-intercept due to removal of early-time data points.

18

SPE-174023-MS

Fig. 11 shows the typical response of effective fracture volume and IGIF to the changes in the initial
gas saturation observed in all 8 wells. In all the cases, IGIF becomes less sensitive as the gas saturation
increases. The effective fracture volume, on the other hand remains highly sensitive to the gas saturation.
The general understanding in fractured horizontal wells is that the total effective fracture volume cannot
exceed the TIV during the fracturing operation (Ezulike et al., 2014). Thus, from the observations in the
sensitivity analysis, there exists a minimum gas saturation that ensures that the effective fracture volume
is always less than the TIV. This gas saturation is summarized in Table 3 as Sgi-min for all cases and
represents the minimum limit of the initial free gas saturation in the fractures before starting the flowback.
The material balance analysis of flowback production data shows that a significant portion of the created
fracture volume must be saturated with gas. The minimum initial gas saturation for this well pad ranges
between 5 and 20 % of the effective fracture volume.

Table 3Summary of MBE analysis for Pad X


Well ID
A
B
C
D
E
F
G
H

Fracture
Stages
15
17
20
15
20
15
20
17

Perforation
Clusters
4
4
1
1
4
4
1
1

Flowback
Sequence

TIV (m3)

Gp* (std.
Mm3)

IGIF
(res. m3)

Sgi-min

Sgi

Vfi (m3)

TLR/Vfi

4
3
2
1
6
5
7
8

60590
66246
75504
69673
54217
43927
58678
54217

3700
2700
1050
3950
2350
2000
1650
1950

12310
8985
3495
13140
7820
6655
5490
6488

0.19
0.13
0.05
0.18
0.14
0.14
0.09
0.12

0.226
0.185
0.058
0.207
0.184
0.184
0.167
0.163

54469
48570
60258
63475
42500
36170
32875
39815

0.22
0.10
0.08
0.12
0.22
0.37
0.21
0.11

SPE-174023-MS

19

Figure 11Sensitivity analysis on initial average gas saturation for Pad X

20

SPE-174023-MS

In the final step, the material balance analysis is constrained by the IGIF obtained from the intercept
of the p/Z plots. Notice that the x-intercept of p/Z plots also gives the volume of initial gas in fractures,
but in standard conditions (Reported as Gp* in Table 3). Thus, there exists an initial gas saturation Sgi that
best matches the IGIF obtained from slope of [F vs. E] and from p/Z intercept. Table 3 lists this best match
average gas saturation as Sgi and the corresponding effective fracture volume (Vfi). The gas saturations
obtained are the average gas saturation in the fracture network, including the secondary and natural
fractures while the effective fracture volume is the total fracture volume contributing to the flow. The
effective fracture volume includes not only the primary/hydraulic fracture network but also all active
secondary and natural fractures. Although the primary fractures are mostly saturated with water because
of the large volumes injected, gas builds up in the secondary/natural fractures during the shut-in period.
This gas is released once the wells are opened causing the instant gas production observed in many shale
gas wells.
There is a significant difference between TIV and the estimated effective fracture volumes (Vfi) for the
wells analyzed in Table 3. A significant portion of the injected volume is lost into pre-existing natural
fractures or imbibed into the shale matrix. On average, the effective fracture volume is about 75 % of the
TIV for this well pad. However, 75 % still represents a large portion of the fracturing fluid and does not
explain the poor load recovery in fractured shale wells. If indeed a large percentage of the injected fluid
is effective in creating hydraulic fractures, the load recovery should be much larger than 10-15 %
commonly observed during flowback operations. We believe that the total load recovered (TLR) during
the flowback is from the large, well connected hydraulic fractures. The water imbibed into the existing
natural fractures, trapped in small and poorly connected secondary fractures, and retained in the bottom
of vertical fractures (Parmar et al., 2014) can hardly be recovered during flowback. Table 3 compares the
total load recovery (TLR) with the effective fracture volume (Vfi). The results show that TLR/Vfi ranges
between 10 and 40 % for this well pad. Since the total load recovery (TLR) is predominantly from large,
well connected fractures, we may conclude that most of the effective fracture volume actually comes from
the existing natural fractures or the secondary fracture network. This highlights the importance of
considering the effects of secondary/natural fractures in modeling fluid flow in shale reservoirs.

Section 7 Correlating Effective Fracture Volume to Fracture Design and


FLowback Sequence
In this section, we attempt to investigate possible correlations between the effective fracture volume and
common fracture design/operational parameters, such as TIV, number of stages, number of clusters and
the flowback sequence of wells within a well pad. Fig. 12a shows the relationship between TIV and Vfi.
Expectedly, the wells with higher TIV show a higher Vfi. On the other hand, the estimated effective
fracture volume has no correlation with the number of perforation clusters (Fig. 12b) or the number of
fracture stages (Fig. 12c). This is in agreement with findings by Cheng (2012b) stating that increasing the
number of perforation clusters does not necessarily have a positive impact on production rate or ultimate
recovery because of the increased number of ineffective fractures. Fig. 12d we shows the effects of the
flowback sequence of wells in the well pad on the effectiveness of the fracturing operation. The
effectiveness of the fracturing operation is measured by Vfi/TIV (in Table 3), which means the fracture
volume created per unit volume of fracturing fluid injected into the formation. Fig. 12d shows that the
wells flowed earlier in the sequence have a higher Vfi/TIV ratio compared with the wells opened later in
the sequence. This is contradictory to our expectation that wells flowed earlier should have lower
fracturing effectiveness due to insufficient shut-in time. One possible explanation is that the wells flowed
earlier have more significant wellbore storage effects. In other words, the production data we collected
acturally partly comes from the vertical and horizontal wellbore sections, not only from the effective
fracture network. Thus, the effective fracture volume calculated using early-time data is indeed the sum

SPE-174023-MS

21

fracture network downhole and the wellbore. This is especially the case with the wells flowed earlier. This
suggests that wellbore storage effects influence the early flowback data. Purifying the early-time
flowback data, especially eliminating the wellbore storage effects is of great importance and should be
considered in future studies. Another possible explanation is that the wells flowed earlier are producing
fluids beyond their drainage area and this is reflected as a higher fracture volume calculated from material
balance analysis. This means that the wells flowed earlier in the sequence are not producing only from
fractures within their own drainage area, but also from interconnected secondary/natural fractures from
adjacent wells, which are still closed. This may suggest that a single well approach may not be ideal in
analyzing flowback data, especially in well pads that have severe inter-well communication. Analyzing
fractured horizontal wells as flow groups and estimating the effective fracture volume contributing to the
group by combining production data from all the wells remain the subject of future studies.

Figure 12Key correlations for Well Pad X. (a) Effective fracture volume vs. TIV (b) Effective fracture volume vs. clusters (c) Effective
fracture volume vs. stages (d) Vfi/TIV vs. flowback sequence

22

SPE-174023-MS

Conclusions
Flowback production data from gas shales, with extended shut-in period, show instantaneous gas
production, which suggests the presence of initial free gas within the fracture network. Diagnostic plots
of Gas-Water-Ratio (GWR) divide the flowback data into two distinct regions. The first region is the Early
Gas Production (EGP) characterized by a decreasing GWR, indicating free gas depletion from fractures.
As flowback process continues, production rolls over into the Late Gas Production (LGP) characterized
by an increasing GWR.
Conventional gas material balance analysis using p/Z plots indicate that the fracture system may be
approximated by a closed/volumetric system during EGP. We observe a linear relationship in these plots
during the early time, with a gradual upward deviation as the flowback operation continues. This suggests
that there is negligible influx from the matrix during the EGP phase and early-time production data can
be analyzed using the conventional gas material balance techniques.
We demonstrated the possibility of analyzing flowback production data using the material balance
approach by analyzing simulated production data as well as real production data from 8 MFHWs
completed in the Horn River Basin. The proposed approach enables the estimation of the effective fracture
volume contributing to flow regardless of the fracture geometry. Results show that a significant percentage of the effective fracture volume comes from the secondary/natural fracture networks. Although the
primary fractures are mostly saturated with water because of the large volumes injected, gas builds up in
the secondary/natural fractures during the shut-in period. This gas is released once the wells are opened,
causing the instant gas production observed in many shale has wells.
The results of the analysis also suggest that the flowback sequence of wells within a well pad is an
important factor and must be considered in planning future flowback operations. It appears that wells
flowed earlier in the sequence are not only producing from fractures within their drainage area but also
from interconnected fractures from adjacent wells. We do not observe any direct correlation between the
number of clusters/stages and the estimated effective fracture volume for this well pad. Increasing the
number of perforation clusters or stages does not necessarily have a positive impact on the well
performance. This is possibly due to the increased number of ineffective fractures created during the
process. This suggests that although the importance of fracture design cannot be argued, operational
parameters (such as the flowback sequence) must also be considered when planning future fracturing
operations.
Nomenclatures
Af
B
C
Gfi
Gp
P
Pc
Pwf
S
t
t*
V
Wfi
Wp
Z

Fracture Area (m*)


Formation Volume Factor (Std. Volume/Res.Volume)
Compressibility Factor (l/Pa)
Initial Free gas Volume at standard condition (m3)
Cummulative Gas Production (Std.m3)
Pressure (Pa)
Critical Pressure for Fractures (Pa)
Well Flowing Pressure (Pa)
Saturation
Time (hrs)
Roll over time (hrs)
Volume (m3)
Initial Water Volume at standard condition (Std.m3)
Cummulative Water Production (m3)
Gas Compressiblility Factor

SPE-174023-MS

f
f

23

Fracture Stiffness (Pa/m)


Average Fracture Aperture (m)
Difference

Subscript
f
g
i
min
w
SC

fracture
gas
initial
minimum
water (fracturing fluid)
Standard Canditain

Acknowledgments
This research was funded by Natural Sciences and Engineering Research Council of Canada (NSERC),
Nexen Energy ULC, FMC Technologies, and Trican Well Service. We thank Ebrahim Ghanbari for his
help in developing the numerical simulation.

References
1. Abbasi MA. A comparative study of flowback rate and pressure transient behavior in multi
fractured horizontal wells.; 2012 (Msc Thesis). University of Alberta.
2. Abbasi, M., Ezulike D., Dehghanpour H, Hawkes R, (2014) A comparative study of flowback rate
and pressure transient behavior in multi fractured horizontal wells completed in tight gas and oil
reservoirs, Journal of Natural Gas Science and Engineering 17, 8293.
3. Adefidipe, O. A., Dehghanpour, H., &Virues, C. J. (2014). Immediate Gas Production from Shale
Gas Wells: A Two-Phase Flowback Model. Society of Petroleum Engineers. doi: 10.2118/
168982-MS
4. Adefidipe, O. A., Xu, Y., Dehghanpour, H., & Virues, C. J. J. (2014, September). Estimating
Effective Fracture Volume from Early-Time Production Data: A Material Balance Approach. In
SPE/CSUR Unconventional Resources Conference-Canada. Society of Petroleum Enginees.
5. Ahmed, T. (2006). Reservoir engineering handbook. Golf Professional Publishing.
6. Alkouh, A., McKetta, S., & Wattenbarger, R. A. (2014). Estimation of Effective-Fracture Volume
Using Water-Flowback and Production Data for Shale-Gas Wells. Journal of Canadian Petroleum Technology 53 (05), 290 303.
7. Chalmers, G. R., Ross, D. J., &Bustin, R. M. (2012). Geological controls on matrix permeability
of Devonian Gas Shales in the Horn River and Liard basins, northeastern British Columbia,
Canada. International Journal of Coal Geology, 103, 120 131.
8. Cheng, Y. (2012). Impacts of the Number of Perforation Clusters and Cluster Spacing on
Production Performance of Horizontal Shale-Gas Wells. SPE Reservoir Evaluation & Engineering, 15 (01), 3140. doi: 10.2118/138843-PA
9. Cheng, Y. (2012). Impact of Water Dynamics in Fractures on the Performance of Hydraulically
Fractured Wells in Gas-Shale Reservoirs. Journal of Canadian Petroleum Technology, 51 (02),
143151. doi: 10.2118/127863-PA
10. Clarkson, C. R., & Williams-Kovacs, J. (2013). Modeling Two-Phase Flowback of Multifractured
Horizontal Wells Completed in Shale. SPE Journal, 18 (04), 795812. doi: 10.2118/162593-PA
11. Crafton, J. W., & Gunderson, D. W. (2006). Use of Extremely High Time-Resolution Production
Data to Characterize Hydraulic Fracture Properties. Society of Petroleum Engineers. doi: 10.2118/
103591-MS

24

SPE-174023-MS

12. Crafton, J. W., & Gunderson, D. W. (2007). Stimulation Flowback Management--Keeping a Good
Completion Good. Society of Petroleum Engineers. doi: 10.2118/110851-MS.
13. Craig, D.P., (2006). Analytical Modeling of a Fracture-injection/fall off Sequence and the
Development of a Refracture Candidate Diagnostic Test (Ph.D thesis). TexasA& M University,
Texas
14. Dehghanpour, H., Zubair, H.A., Chhabra, A., Ullah, A., (2012). Liquid intake of organic shales.
Energy Fuels 26(9), 5750 5758. http://dx.doi.org/10.1021/ef300979447.
15. Dehghanpour, H., Lan, Q., Saeed, Y., Fei, H., Qi, Z., 2013. Spontaneous imbibition of brine and
oil in gas shales: effect of water adsorption and resulting micro fractures. Energy Fuels 2013 (27),
3039e3049. http://dx.doi.org/10.1021/ef4002814.
16. EIA, U. (2013). Technically Recoverable Shale Oil and Shale Gas Resources: An Assessment of
137 Shale Formations in 41 Countries outside the United States. US Department of Energy/EIA,
Washington, DC.
17. Ghanbari, E., Bearinger, D., Abbasi, M. A., &Dehghanpour, H. (2013). Flowback Volumetric and
Chemical Analysis for Evaluating Load Recovery and Its Impact on Early-Time Production.
Society of Petroleum Engineers. doi: 10.2118/167165-MS
18. Ilk, D., Currie, S. M., Symmons, D., Rushing, J. A., Broussard, N. J., &Blasingame, T. A. (2010).
A Comprehensive Workflow for Early Analysis and Interpretation of Flowback Data from Wells
in Tight Gas/Shale Reservoir Systems. Society of Petroleum Engineers. doi: 10.2118/135607-MS
19. Johnson, M. F., Walsh, W., Budgell, P. A., & Davidson, J. A. (2011, January). The Ultimate
Potential for Unconventional Gas in the Horn River Basin: Integrating Geological Mapping with
Monte Carlo Simulations. In Canadian Unconventional Resources Conference. Society of Petroleum Engineers. doi: 10.2118/135607-MS
20. Jones, R. S., Pownall, B., & Franke, J. (2014, August 28). Estimating Reservoir Pressure form
Early Flowback Data. Society of Petroleum Engineers. doi: 10.2118/135607-MS.
21. Kuchuk, F. J., Biryukov, D., & Fitzpatrick, T. (2014). Rate Transient and Decline Curve Analyses
for Continuously (Dual-Porosity) and Discretely Naturally Fractured Reservoirs. Society of
Petroleum Engineers. doi: 10.2118/170698-MS
22. Li, C., Lafollette, R., Sookprasong, A., & Wang, S. (2013). Characterization of Hydraulic Fracture
Geometry in Shale Gas Reservoirs Using Early Production Data. International Petroleum Technology Conference. doi: 10.2523/16896-MS
23. Makhanov, K., Kuru, E., &Dehghanpour, H. (2014). Measuring Liquid Uptake of Organic Shales:
A Workflow to Estimate Water Loss During Shut-in Periods. Journal of Unconventional Oil and
Gas Resources7, 2232.
24. Moghadam, S., Jeje, O., &Mattar, L. (2011). Advanced Gas Material Balance in Simplified
Format. Journal of Canadian Petroleum Technology, 50 (01), 90 98 doi: 10.2118/139428-PA
25. Obinna, E. D., & Hassan, D. (2014). A model for simultaneous matrix depletion into natural and
hydraulic fracture networks. Journal of Natural Gas Science and Engineering. http://dx.doi.org/
10.1016/j.jngse.2013.11.004.
26. Parmar, J., Dehghanpour, & Kuru, E., H. (2014). Displacement of water by gas in propped
fractures: Combined effects of gravity, surface tension, and wettability. Journal of Unconventional Oil and Gas Resources5, 10 21.
27. Rahman, N. M. A., Mattar, L., & Anderson, D. M. (2006). New, Rigorous Material Balance
Equation for Gas Flow in a Compressible Formation with Residual Fluid Saturation. Society of
Petroleum Engineers. doi: 10.2118/100563-MS
28. Ross, D. J., & Bustin, R. M. (2008). Characterizing the shale gas resource potential of Devonianississippian strata in the Wesern Canada sedimentary basin: Application of an intergrated
formation evaluation. AAPG bulletin, 92(1), 87125. doi: 10/1306/09040707048

SPE-174023-MS

25

29. Singh, V. K. (2013). Overview of Material Balance Equation (MBE) in Shale Gas & NonConventional Reservoir. Society of Petroleum Engineers. doi: 10.2118/164427-MS
30. Williams-Kovacs, J. D., & Clarkson, C. R. (2013). Stochastic Modeling of Multi-Phase Flowback
From Multi-Fractured Horizontal Tight Oil Wells. Society of Petroleum Engineers. doi: 10.2118/
167232-MS
31. Zhang. Y., & Ehlig-Economides, C. (2014, August 28). Accouting for Remaining Injected
Fracturing Fluid in Shale Gas Wells. Society of Petroleum Engineeris. doi: 10.15530/urtec-20141892994

Acknowledgements
The authors are grateful to (1) Nexen Energy ULC and INPEX Gas British Columbia Ltd for providing
the flowback data and (2) FMC Technologies, Trican Well Service and National Sciences and Engineering
Research Council of Canada (NSERC) for funding this study.

26

SPE-174023-MS

Appendix
In order to estimate the roll over time of the GWR curve, we establish a relationship between the cumulative gas production
(Gp) and the cumulative water production (Wp) from flowback data. Fig. A1 shows that all 8 wells studied in Pad X suggest
that cumulative gas produced (Gp) can be approximated by a cubic function of the cumulative water production (Wp), with
R-square values of approximately 1. The Gp-Wp relationship can be expressed by Eq. A-1, where a, b, c, d are constants
obtained from early time flowback data. The instantaneous slope of the curve, obtained by differentiating Eq. A-1, gives the
GWR at any time. Eq. A-2 is the approximation equation to calculate the GWR from cumulative water and gas production data.
Fig. A2 compares the GWR obtained from field data and calculated from Eq. A-2. The cumulative water production
corresponding to the roll over time, t* is calculated from Eq. A-4.

SPE-174023-MS

27

Figure A1Cumulative Gas Production (Gp) vs. Cumulative Water Production (Wp) for Pad X. Cumulative gas production (Gp) can be
modeled as a cubic function of cumulative water production (Wp)

28

SPE-174023-MS

Figure A2GWR obtained from field data and mathematical model for Pad X.

SPE-174023-MS

29

(A-1)
(A-2)

(A-3)

At the minimum point,


Therefore, cumulative water production at the end of EGP is given by
(A-4)
where Wp* is the cumulative water production corresponding to roll over time, t*.

Potrebbero piacerti anche