SPE 166279

Estimation of Effective Fracture Volume Using Water Flowback and

Production Data for Shale Gas Wells
Ahmad Alkouh, SPE, Texas A&M University, Steven McKetta, SPE, Southwestern Energy, Robert A.
Wattenbarger, SPE, Texas A&M University

Copyright 2013, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, 30 September–2 October 2013.

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
Gas production data may be the main tool used to analyze shale gas reservoirs. Water production is usually not included in
the analysis. In this paper, post-frac water flowback and long-term water production are added to the analysis. These water
data are usually available but are analyzed separately and not combined with long-term gas production data. In this paper,
procedures and examples are presented for including water flowback and water production data in the analysis of shale gas
wells using rate transient analysis.
A number of simulation cases were run. Various physical assumptions were used for the saturations and properties that
exist in the fracture/matrix system after hydraulic fracturing. Water flowback and long-term production periods were then
simulated. The results of these simulations were compared with data from actual wells using diagnostic and specialized plots.
These comparisons lead to certain conclusions which describe well/reservoir conditions after hydraulic fracturing and during
production.
This paper shows the benefits of a new method for combining water flowback and long-term water production data in
shale gas analysis. Water production analysis can provide effective fracture volume which was confirmed by the cumulative
produced water. This will help evaluating fracture-stimulation jobs. It also shows some pitfalls of ignoring flowback data. In
some cases the time shift on diagnostic plots changes the apparent flow regime identification of the early gas production data
as well as water production data. This leads to different models of the fracture/matrix system. The presented work shows the
importance of including water flowback data in the long-term production analysis.

Introduction
Water production is usually ignored when analyzing and forecasting shale gas well. The process involves pumping thousands
of barrels of water with proppant and additives into the rock at high pressure. Many operators and Alkouh and Wattenbarger
(2013) indicates that the percent of injected fluid recovered (Load Recovery) in shale gas wells ranges from 10-40%.
Flowback is the early data (water/gas rate & pressure) gathered after fracture stimulation of the well which might be followed
by a shut-in. The flow sequance of the usual shale well is presented in Fig. 1 where there is a shut-in period between
flowback and production due to delays in pipeline connection. Most operators ignore flowback data and do not combine it
with production data.
This paper shows the benefits of combining flowback data with production data. Also, a new method is presented to
calculated effective water volume using combined (flowback and production) water data. This method was developed using
the conventional practice of analyzing gas production in shale wells but applied to water production.
The paper has four main parts: (1) a review of the literatures related to flow regime identification and diffusivity equation.
(2) A verification of the new method using simulated model. (3) Examples of single wells and 5 wells pad illustrating the
new method. (4) Discussion and conclusions. The aim of this paper is to present a new method that can be applied on water
data from both flowback and production periods combined.
2 SPE 166279

Fig. 1— The flow sequence of a usual shale well starts with frac job. Then a flowback period has high water rates and gas would start after 2 to 5
days and some wells are shut-in due to connection delay. Then production starts.

Literature Review
There is a lack of literature dealing with quantitative analysis of flowback data (gas/water). Flowback usually deals with first
1 to 4 days of the data and often removed from the long-term production data especially if followed by a shut-in. In an
attempt to analyze flowback data, Crafton and Gunderson (2007) presented a method to calculate hydraulic fracture
conductivity using early flowback data. In a personal contact with Crafton, there is no work done to calculate fracture volume
using water data without the knowledge of other parameters like permeability.
Abbasi et al. (2012) presented an attempt to analyze water data from flowback period. The author divided the production
profile shown in Fig. 1 into three regions based on type of fluid flowing. First region water production dominates and Second
is water production decline and gas production increase. The third region is when gas production dominates. The authors
observed linear flow relationship between rate normalized pressure (RNP) and material balance time (tMB) for all regions. The
designed model was for the first region but the data used is from all the three regions. In the model gas compressibility was
not accounted for although it is the dominant compressibility in the second and third regions. Hydraulic fracture permeability
is needed to find fracture volume and vise versa which adds uncertainty to the calculation.
On the other hand, long-term production data (gas/water) has been the interest of a lot of research and publications. In
production data analysis (PDA), gas production (production period only) is analyzed using analytical solutions ignoring the
flowback period if there is a shut-in and ignoring water data. To the authors’ knowledge, there is now publication that
combines flowback data with production data if separated by a shut-in period.
Simulation is another tool to analyze production data and can help in verifying assumed models. Unfortunately, shale gas
researchers are not comparing simulation models with analytical solutions, Alkouh et al.(2012). In this work, simulation is
used to verify the new method of analyzing water data using PDA. This method is developed by looking at the condition of
gas dominating the diffusivity equation as in the next section.

Diffusivity Equation. Another way to divide the well’s production profile is based on dominating phase into two regions.
Water dominating region is where the compressibility of water dominant and it is very short. Gas dominating region is the
main region and it start at early times. In production data analysis, it is assumed that the dominant phase is gas which makes
the analyst ignore the previous period since gas dominates the diffusivity equation. Eq. 1 shows the diffusivity equation for a
two phase system (water and gas).

∙ ∅ …………….…………………………………… (1)
Where the total mobility is defined as

…………….……………………… (2)

Where the total compressibility is defined as

…………….………………………………... (3)

When significant gas flows, two conditions are satisfied.

and
SPE 166279 3

If those conditions are true, then the real gas diffusivity equation can be used to analyze gas production data, using the real
gas pseudo-pressure.

2 ∅
…………….……………………………….............. (4)

The satisfaction of mobility condition depends on relative permeability of the gas which depends on water saturation. Based
on several simulation runs, any water saturation below (1-Sgirr) is dominated by gas. For example, if Sgirr = 0.2 then any
location with water saturation below 0.8, gas will dominate this location in the system. The satisfaction of compressibility
condition depends on water saturation also. Usually, the difference between volumetric gas compressibility (Sg cg) and total
compressibility is 3% at Sw = 0.7. In general, if the water saturation is below 0.7, then gas is dominating the diffusivity
equation (mobility and compressibility).
Due to the high flow rates in the flowback period and the size of the hydraulic fracture is small compare to the whole
system, gas can dominate the system at early times. With this in mind, gas compressibility cannot be ignored in the water
analysis. This concept is the main drive for this research. Even though the analyzed fluid is water, gas compressibility should
be used which was verified using a simulation model.

Flow Regimes. In the production data of shale wells, several slopes can be observed on log-log plots of rate time or material
balance time (tMB). Transient linear flow dominates the early part of the production profile and can last for a year in usual
cases. This linear flow can be from hydraulic fracture or a stimulated matrix (higher than original matrix permeability).
Transient linear flow would be shown as a half slope on the log-log plot (t & tMB) and a straight line on the square root of
time plot.
If two transient linear systems are flowing at the same time, a signature of transient bilinear flow will be observed. This
can be interpreted as linear flow in matrix and linear flow in natural fracture. A transient bilinear flow or a quarter-slope line
in the log-log plot (t & tMB). When the transient period ends, the boundary of the well is reached. This period is called
boundary dominated flow (BDF) which has a different signature on time and material balance time. BDF has an exponential
signature on log-log plot of t and unit slope line on a log-log plot of tMB.
Water production data usually does not show the transient period and in some cases BDF is present. This signature is due
to the fact that the hydraulic fracture and induced natural fractures are filled mainly with frac water. Alkouh and
Wattenbarger (2013) showed that natural fractures can trap 60% of the water and flowback the rest. The volume that was
occupied by the produced water is called effective fracture volume since it is displaced by the gas. This effective volume can
be partially from hydraulic fracture which might trap water and the rest from natural fracture which will trap water. It is
called “effective fracture volume” because it provided the volume for the gas to flow and it is in direct communication with
the horizontal well. This method does not work if there is an underlying aquifer which will force the well to produce more
water higher than the injected volume. The calculated effective fracture volume can be less than the produced water volume
due to water hits from nearby frac stimulation jobs.

Model Simulation
A 3-D gas-water black oil simulator is used to simulate a single fracture filled with water and gas in the matrix. The
simulation properties are shown in TABLE 1. The relative permeability in the fracture is gravity segregated as shown in Fig.
2. In Fig. 3, the two simulation cases are shown and the single phase flow has water flowing in fracture only and the two
phase has gas flowing into the fracture which is filled with water. Water volume in both cases in the fracture is constant. The
concept behind the cases is to build linear flow of fracture filled with water and compare it with the effect of gas flowing
from matrix which provides an extra compressibility to the water flow. In both cases, water flow is analyzed to calculate
different parameters.
4 SPE 166279

TABLE 1—SHALE GAS SIMULATION PROPERTIES

Initial pressure, pi (psi) 3000 Matrix porosity, φm (fraction) 0.06
Flowing BH pressure, pwf (psi) 500 Reservoir thickness, h (ft) 300
Specific gravity, SGg 0.65 Matrix permeability, km (md) 1.5 x 10-4
Reservoir temperature, T (°F) 160 Hydraulic Fracture Conductivity, CHF (md-ft) 4
Fracture porosity, φF (fraction) 1 Fracture spacing, LF (ft) 500
Water volume in the fracture, Vw (STB) 6,995 Fracture half length, xf (ft) 550
Water formation volume factor, Bw (Res bbl/STB) 1.01 Formation Compressibility, Cf (psi-1) 1 x 10-6
-1 -4
Gas Compressibility, Cg (psi ) 3.9 x 10 Water Compressibility, Cw (psi-1) 2.9 x 10-6

1
Relative Permeability

0.8

0.6

0.4 Krg

0.2 Krw

0
0 0.2 0.4 0.6 0.8 1
Water Saturation

Fig. 2— Gravity segregated elative permeability curve of the fracture.

Horizontal well

Fig. 3— Two simulated cases and in the single phase case water that filling the fracture is flowing only (Gas from matrix is not flowing); in the two
phase case, gas is flowing from matrix into fracture that is filled with water.

The result of the two cases is presented in Fig. 4 with water rates plotted versus time for a single fracture. The single
phase case represents linear flow in a fracture with the signature of ½ slope and then declines exponentially. This curve can
be reproduced using analytical solution. When the gas flow from matrix is included as in two phase flow, the linear flow is
departed from the single phase case coming to another linear due to flow of gas from matrix into fracture. The second linear
flow line is not informative since it depends on several unknown parameters and this type of flow is very early and is not
observed in field water data. Water rate declines in different manner if compared to the single phase case.
SPE 166279 5

In Fig. 5, both cases are plotted using material balance time to observe the boundary dominated flow regime as a unit
slope line. In both case water volume in the fracture is similar but the results indicate that those volumes are different since
both unit slope lines are not overlying each other. Since gas is flowing into the fracture, it will tend to increase the total
compressibility which will affect water volume calculations. A new method is implemented to rigorously account for the gas
present effect on water volume calculations.
Water rate normalized by pressure (RNPw) is plotted for both cases as in Fig. 6. RNPw is used to correct for variable
bottomhole pressure (pwf) as in the case of an actual well (the simulated cases are constant pwf). Eq. 5 is used to calculate
water volume in STB where mpss is the slope of the Cartesian plot in the period of unit slope line as in Fig. 6 where the line or
RNPw is over RNP’w (derivative) line. mpss can be also calculated using Eq. 6 with the data in the unit slope period of Fig. 6.

…………….……………………………………………. (5)

……………………………………………………………. (6)

1.E+4

1.E+3

1.E+2
Water Rate, bbl/D

1.E+1

1.E+0

1.E‐1
Single Phase
1.E‐2
Two Phase
1.E‐3
1.E‐4 1.E‐3 1.E‐2 1.E‐1
1.E+0 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5
Time, Days
Fig. 4— water flow rate of a single fracture with/without gas flow from matrix, linear flow (1/2 slope) in single phase and then declines
exponentially; higher flow rates and longer time in the case with gas flow from matrix into water filled fracture increasing compressibility.

1.E+4

1.E+3

1.E+2
Water Rate, bbl/D

1.E+1

1.E+0

1.E‐1

1.E‐2
Single Phase
1.E‐3
Two Phase
1.E‐4
1.E‐4 1.E‐3 1.E‐2 1.E‐1 1.E+0 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6
Water Matrial Balance Time, tmBw
Fig. 5— Water flow rate versus water material balance time showing two different unit slope lines indicating two different volumes when the same
volume is used for both cases, this is due to gas compressibility.
6 SPE 166279

1.E+8
RNP'w; Two Phase
1.E+7 RNPw; Two Phase
RNP'w; Single Phase
1.E+6 RNPw; Single Phase

RNPw and RNP'w, psi/bbl/D

1.E+5

1.E+4

1.E+3

1.E+2

1.E+1

1.E+0

1.E‐1
1.E‐4 1.E‐3 1.E‐2
1.E‐1 1.E+0 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6
Water Matrial Balance Time, tmBw
Fig. 6— Water flow rate versus water material balance time showing two different unit slope lines indicating two different volumes when the

Total compressibility in the single phase is water and formation compressibility but in the two phase case it is different.
The two phase case compressibility is first dominated by water then after gas saturation increase in the fracture, gas
compressibility is dominating. To get the correct water volume in the fracture, an approximation is used by equating total
compressibility to gas compressibility.
The results of both cases are presented in Table 2 where the inputted water volume is back calculated. In the single phase
case, the total compressibility (Water and Formation) is used. But, gas compressibility is used to calculated water volume in
the fracture. This method will be used in field data for further confirmation.

TABLE 2—CALCULATIONS RESULTS OF SIMULATION CASES

Parameters Single Phase Case Two Phase Case
mpss 37 0.37
ct, psi-1 3.9 x 10-6 3.9 x 10-4
Input Vw, STB 6,955 6,955
Calculated Vw, STB 6,999 6,930

Field Examples
The examples are from Fayetteville and Barnett formations with two single wells and two pads with multiple wells. Well FF-
1 was studied thoroughly to which shows the details of calculating effective water volume and the effect of flowback period
on flow regime identification. Well B-151 was an example of on how to apply the method if flowback data is not available.
Clark pad illustrated the use of effective fracture volume as a monitoring and tracking method for the effect of nearby frac
jobs. Finally, rowlett pad show how the frac sequence and placement of the well can affect the effective fracture volume and
its relationship with gas production.

Well FF-1
This well is in Fayetteville formation and has the properties shown in Table 3. The well follows the trend shown in Fig. 2
with 12 days of flowback and a shut-in of 30 days and then production of 4 years. The well was injected with 72,600 STB of
slick water. Fig. 7-A shows water and gas data for the flowback period and the shut-in is after the start of gas production due
to delays in pipe connection. Production period data is shown in Fig. 7-B which indicates a bilinear flow for 100 days then
followed by linear flow. This signature indicate the possibility of linear flow in natural fractures while the matrix is flowing
linearly too which imply that natural fractures have great domination in the system.
Gas rate normalized by pressure difference versus time for production period is presented in Fig. 8-A to remove the effect
of bottomhole pressure and the bilinear is present. Instead of plotting time, material balance time is plotted for the production
period as in Fig 8-B also with the bilinear-linear flow signature. All the presented diagnostic plots indicate that this well has a
long period of bilinear.
SPE 166279 7

TABLE 3—PROPERTIES OF WELL FF-1

Initial pressure, pi (psi) 1736 Matrix porosity, φ (fraction) 0.04
Specific gravity, SGg 0.58 Reservoir thickness, h (ft) 293
Reservoir temperature, T (°F) 118 Injector Water, Vinj (STB) 72,600
No. of Hydraulic Fractures, nF 24

1.E+5 1.E+4 1.E+3

qw ‐ Flowback qg ‐ Production
qg ‐ Flowback qw ‐ Production

Water Rate, bbl/Day

Water Rate, bbl/Day

Gas Rate, Mcf/D

Gas Rate, Mcf/D

1.E+4 1.E+3 1.E+2

1.E+3 1.E+2 1.E+1

A
1.E+2 B
1.E+0 1.E+1
1.E+0 1.E+1 1.E+2
1.E+2 1.E+3 1.E+4 1.E+0 1.E+1
Time (Flowback), Days Time (Production), Days
Fig. 7— A: flowback period with gas starting to flow before shut-in; B: production period with gas data showing bilinear flow up to 100 days
followed by linear flow. Water data does not have a clear signature.

1.E‐4 1.E‐4
PNRg ‐ Production
qg / [m(pi)‐m(pwf)],Mcf/D/psi2 /cp

PNRg ‐ Production
qg / [m(pi)‐m(pwf)],Mcf/D/psi2/cp

1.E‐5 1.E‐5

1.E‐6 1.E‐6

A B
1.E‐7 1.E‐7
1.E+1 1.E+2 1.E+0
1.E+3 1.E+4 1.E+1 1.E+2 1.E+3 1.E+4 1.E+0
Time (Production), Days Gas Material Balance Time (Production), Days
Fig. 8— A: rate normalized by pressure versus production time indicates the bilinear flow followed by linear flow; B: The same as plot A, but with
material balance time..

Although it is the practice in some companies, but there is no reason to remove flowback data from production if they are
separated by a shut-in period. In Fig 9, we combined flowback data with production and noticed a big difference in the gas
rate signature of the well. The bilinear flow disappeared and was replaced by a longer linear flow with a spike at the
beginning which is normal for the gas rate to start flowing at higher rates after a month of shut-in. Including flowback period
with the shut-in time shifted the production data in time to give the correct signature of longer linear. Taking the first point of
the production data to be at 1 day instead of 45 days (adding flowback and shut-in) would make the early linear flow look
like bilinear. Although it is not shown, gas material balance time can be affected if the flowback period is ignored due to the
high gas rates at this period and smaller OGIP would be calculated. Any previous gas flow should be included in the analysis
even if it is separated by a shut-in period.
8 SPE 166279

1.E+5 1.E+4

Gas Rate, Mcf/Day 1.E+4 1.E+3

Water Rate, bbl/Day

1.E+3 1.E+2

1.E+2 qg ‐ Production 1.E+1

qg ‐ Flowback
qw ‐ Flowback
qw ‐ Production
1.E+1 1.E+0
1.E+0 1.E+1 1.E+2 1.E+3 1.E+4
Time (Combined), Days

Fig. 9— Combining flowback and production data made the bilinear flow in gas disappear and replaced by a longer linear with a spike at the
beginning due to the shut-in of a month.

Combining flowback and production data affected material balance plot of water normalized rate. In Fig. 10-A, pressure
normalized rate is plotted for the production period (dark blue) and a unit slope line (1) is indicated at the end of the data.
This data is missing around 10 points of high rates from the flowback period which will shift production data since the
material balance time value will be higher due to higher cumulative. In the same plot, flowback period (gray) is plotted with
production period (aqua). It is noticed that the production period fall on the same unit slope line (2) that started in the
flowback period. The unit slope (1) from production period would give smaller effective water volume compared to the
combined data unit slope (2). Ignoring water flowback period might indicate a smaller effective water volume if the flowback
period is long with high rates.
Fig. 10-B is showing combined data with the derivative to find the slope mpss which is the slope of the Cartesian plot in
the period of unit slope line where the line or RNPw is over RNP’w (derivative) line. The solutions are shown in Table 4 with
the calculated effective water volume as 14% of the injected water volume. Around 96% of the calculated effective water
volume is produce which is in agreement with the low water rates 2 STB/D.

1.E+1 1.E+4
PNRw vs tMBw (Combined)

1.E+0 2 PNRw vs tMBw (Production)

1.E+3
PNRw vs tMBW (Flowback)
qw / [pi ‐ pwf], STB/D/psi

[pi ‐ pwf] / qw , psi/STB/D

1.E‐1 1.E+2

1
1.E‐2 1.E+1

1.E‐3 1.E+0 RNPw ‐ Flowback & Production

A B RNP'w ‐ Flowback & Production

1.E‐4 1.E‐1
1.E+1 1.E+0
1.E+2 1.E+3 1.E+4 1.E+5 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+0
Water Material Balance Time, days Water Material Balance Time (Combined), days
Fig. 10— A: The lower points (dark) are for the production rates and times. The upper curve (light) is created by adding flowback rates and times.
The time shift results in a larger effective water volume ; B: unit slope line is used to calculate slope of BDF period to estimate effective water
volume.

TABLE 4—EFFECTIVE WATER VOLUME CALCULATIONS FOR WELL FF-1

Slope of BDF period, mpss 0.158 Water FVF, Bw (res bbl/STB) 1
Gas compressibility, cg (psi-1) 3.9 x 10-6 Cumulative produced water, (STB) 10,177
Calculated Water Volume, Vw (STB) 10,550
SPE 166279 9

Well B - 151
This well is in Barnett formation and was analyzed in several papers. This well was chosen to show how to include flowback
period even if the data is not available and its effect on time plot. As in Fig. 11-A, the well has bilinear flow for 150 days
then followed by linear flow. From the completion data, there is a period of 15 days between the fracing of the well and the
time of production. If this delay period is included in the production data, gas rate will be as in Fig. 11-B which is a longer
linear flow. If we compared water decline in plot A and B, we notice that in A the data is stretched with a slope of 1 or less
but in B it follows the normal decline of slopes higher than 1. Even if the flowback data is not available, the delay time
should be included to avoid stretching of data and giving the wrong reservoir signature.

1.E+4 1.E+4

Gas & Water Rate, Mcf/D & bbl/D

Gas & Water Rate, Mcf/D & bbl/D

1.E+3 1.E+3

1.E+2 1.E+2

Gas Rate Gas Rate ‐ Shifted

Water Rate A Water Rate ‐ Shifted B
1.E+1 1.E+1
1.E+0
1.E+2 1.E+1
1.E+3 1.E+4 1.E+1 1.E+0
1.E+2 1.E+3 1.E+4
Time, D Shifted Time, D
Fig. 11— A: bilinear flow for 150 days followed by linear flow in gas rate and water has a low slope of around 1; B: shifting the data 15 days due to
delay cause the bilinear flow to disappear and a long linear flow is created instead and water has the normal high slope of higher than 1.

Well FF-18 in Clark Pad

This well is oldest well compared to the other 3 wells drilled next to it. The flow sequence of the well is studied over three
periods each one is separated by newly drilled wells as in Fig. 12. The effective volume will increase due to newly drilled
wells and water and gas production will increase accordingly.
The injected water volume of 132,586 STB is shown in period 1 of Table 5. After 515 days of production, the cumulative
is 6% of the injected volume and water production is around 5 STB/D as in Fig. 13. Using Fig. 14, water effective volume
for this well can be calculated which is reported in Table 5 as 8% of the injected volume and 77% of this effective water
volume is already produced.
Well FF-13 was fraced which increased water production but the effective water volume did not change significantly. This
can be observed in Fig. 14 where period 1 and 2 are almost on the same unit slope line. This is not the case in period 3 where
the slope is shifted to the right indicating an increase in effective water volume due to fracing wells FF-19 and FF-20. This
increase in water volume affected gas rates which increased too. The new effective water volume is 10% of the injected
volume and 95% of this volume was produced water as in Table 5. This method can be a mean of tracking the interference
effect of newly drilled wells on effective water volume.
10 SPE 166279

Period 1 Period 2 Period 3

515 Days 1216 Days 1466 Days

Fig. 12— A map of all wells in Clark pad. The production of the well FF- 18 is divided into 3 periods where period 1 is before well FF-13 is fraced
and period 2 is before wells FF-19 and FF-20 are fraced and period 3 is until the end of the data.

TABLE 5—Water Data for Well FF-18

Water Volume, STB Period 1 Period 3
Injected (Frac Job) 132,586 132,586
Cumulative Produced 8,799 12,992
Calculated 11,351 13,695

1.E+4 1.E+3
qg qw
FF‐13
FF‐19 & FF‐20

1.E+3 1.E+2

Water Rate, bbl/Day

Gas Rate, Mcf/Day

1.E+2 1.E+1

1.E+1 1.E+0
0 300 600 900 1200 1500
Time (Combined), Days
Fig. 13— Water and gas production of the combined period (Flowback and Production) for well FF-18 with the other wells effect as in FF-19 & FF-
20 which increased gas and water production.

1.E+4
Period 1 1.E+3
Period 2 Period 1
1.E+3
Period 3 Period 2
[pi ‐ pwf] / qw, psi/STB/D

Period 3
1.E+2
1.E+2

1.E+1

1.E+0
1.E+1
1.E‐1 1.E+2 1.E+3 1.E+4
1.E+1 1.E+0
1.E+2 1.E+3 1.E+4 1.E+5
Water Material Balance Time (Combined)
Fig. 14— Water RNP showing the three periods with a unit slope indicating boundary dominated flow and in period 3 the increase in water
effective volume (unit slope moves to the right) related to fracing of wells FF-19 and FF-20.
SPE 166279 11

Rowlett Pad
This pad consists of 5 wells as seen in Fig. 15 that were analyzed by Harpel et al. (2012) in order to find the optimum frac
job. Three wells have the same well’s properties and frac job treatment (FF-21, FF-22, FF-24). One well had the same well’s
properties but the frac job injected volume was decreased to 60% (FF-23). One well had a larger well length and the same
frac job as the three wells (FF-25). This pad offered a great opportunity to test the presented methodology since the properties
the same and only frac job injected volume is different.

Fig. 15— Map of the 5 Rowlett wells with the well FF-23 in the middle with the frac volume reduced to 60% of the originally injected volume in the
other wells.

Table 6 shows the well’s data and it is important to notice the sequence of fracing which might have an effect over the
created water volume. Since FF-25 has a longer well length, it was removed from the comparison analysis. Cumulative water
produced and calculated water volumes are shown in Table 7. It is shown in Fig. 16 that the water values vary between the
wells which suggest that not all the injected volume contribute to the created volume.
The result shows that FF-21 has the highest cumulative water and calculated volume. Although FF-22 used the same frac
fluid volume, it produced less water and the calculated water volume is less if compared to FF-21. The difference between
FF-21 and FF-22 might be due to the sequence of fracturing since FF-21 was fraced before FF-22. We notice the same trend
between FF-24 and FF-21 and FF-22 which might be due to sequence and location of the well.
On the other hand, well FF-23 used 60% of the originally injected volume and produced 50% of cumulative water if
compare to an average of the other wells’ cumulative. The calculated water volume using the propose method gives almost
the same percentage. Well FF-23 produced more than the calculated water volume which might be due to communication
with the other wells. The total cumulative produced water is 91% of the total calculated water volume which might suggest
that the wells are in communication.

TABLE 6—Well’s Data for Rowlett Pad

Well Well Length No. of Frac Injected Fluid, bbl Frac Sequence
FF-21 4,542 60 93,221 1
FF-22 4,722 60 91,085 2
FF-23 4,722 60 62,052 5
FF-24 4,722 60 91,215 3
FF-25 5,137 66 97,659 4

TABLE 7—Water Data for Rowlett Pad

Well Cum. Water, bbl % Rec. Injected Water Volume Calc., bbl % Rec. Calc. Water Vol.
FF-21 11,000 12 12,900 85
FF-22 9,300 10 10,500 88
FF-23 4,500 7 4,200 107
FF-24 6,400 7 6,700 95
FF-25 7,300 7 7,600 80
Total 38,500 41,900 91
12 SPE 166279

1.E+0
1.E+0
FF‐21 1 FF‐21
FF‐22
FF‐23
FF‐24
1.E‐1 1.E‐1 2
FF‐23
qw / [pi ‐ pwf] , STB/D/psi

qw / [pi ‐ pwf] , STB/D/psi

1.E‐2 1.E‐2

1.E‐3 1.E‐3

A B
1.E‐4 1.E‐4
1.E+1 1.E+0
1.E+2 1.E+3 1.E+4 1.E+1 1.E+2 1.E+3 1.E+4 1.E+0
Water Material Balance Time (Combined) Water Material Balance Time (Combined)
Fig. 16— A: Water RNP showing a unit slope indicating BDF in all wells with the smallest volume in FF-23 and largest in FF-21; B: the start of
BDF is fast in FF-23 compare to FF-21 which is delayed and the bigger water volume, the start is delayed as seen in arrow 1 and 2.

Gas production data for the pad is presented in Fig. 17. Well FF-23 was the best producer in terms of gas rates (also
cumulative) and in terms of the product . The product is calculated from the slope of Fig. 17 – B where is
effective permeability and is cross sectional area. Although the frac job volume was reduced to 60%, the well had the
best performance which might be due to the placement in the middle of the 5 wells with the last one to be fraced. It is worth
mentioning that this well produced the highest gas cumulative and the lowest water cumulative which is in agreement with
reported results in Alkouh and Wattenbarger (2013).
On the other hand, well FF-21 was the lowest gas producer (also cumulative) with the highest water cumulative. This
might be due to the fact that it was fraced first with the left side of the well not stimulated since there is no well on this side.
The result of this pad suggests that the wells in the middle of the pad if fraced last, will produce more gas and less water if
compared to the rest of the wells in the pad.

1.E+4
200,000
FF‐21 FF‐22
[m(pi) ‐ m(pwf)] / qg, psi2 /cp/Mcf/D

FF‐22
150,000 FF‐24
FF‐23
FF‐22
Gas Rate, Mcf/D

1.E+3 FF‐23
100,000
FF‐24 FF‐22
FF‐21
50,000

A B
1.E+2 0
1.E+0
1.E+1 1.E+2 1.E+3 0 5 10 15 20 25
Time (Combined), Days Time (Combined)0.5 , Days0.5
Fig. 17— A: gas rate showing linear flow in all wells with well FF-23 has the highest gas rates although it was fraced with 60% of other wells ; B:
straight line indicating linear flow in all wells with well FF-23 has the highest and well FF-21 has the lowest although it was
fraced with the highest volume.
SPE 166279 13

Discussion
It is a general practice to ignore flowback data in production data analysis especially if the flowback and production periods
are separated by a shut-in as in Fig. 1. Flowback period is analyzed as a separate data in most literature and production data
analysis ignores flowback period. This paper showed the pitfalls of ignoring flowback data in misidentifying the flow regime.
Also, developed a new method to calculate effective fracture volume based on combined (Flowback and Production) data.
This method is based on modifying the total compressibility to take into account the gas compressibility which is dominating
the diffusivity equation due to domination of mobility and compressibility. In general, gas phase will dominate the diffusivity
equation if the water saturation is below 0.7 at any location in the fracture.
A simulation model was developed to test this theory and back calculate water volume in the fracture. Using the propose
correction, water volume was calculated. This method was tested on several field examples. In well FF-1, the cumulative
produced water is 96% of the calculated effective fracture volume and water rates are around 2 STB/D. The effective fracture
volume is 14% of the injected frac volume. Adding flowback period in this well made it easy to identify the boundary
dominated flow regime and the indicated fracture volume is larger than the fracture volume without the flowback period.
Including the flowback period gave the correct fracture volume calculation.
In well FF-1, gas data from the production period alone (excluding flowback period) indicated that the well has a bilinear
signature followed by linear flow as in Fig. 8. This meant that the reservoir has natural fractures that have a major linear flow
plus the matrix linear for 100 days. This bilinear was explored on several diagnostic plots of time and material balance time.
After adding the flowback period data, the flow regime signature changed to longer linear which had the expected rate spike
after a shut-in as in Fig. 9. Including flowback period would help in avoiding misinterpretation of the flow regime.
In some case the flowback period is not available; Well B-151 was missing flowback period data. Using diagnostic plots,
production data indicated that the well has a bilinear flow regime. The days of the gap between start of flowback and start of
production can be tracked using well’s official data. Including those days in the production data would shift the production
data to a longer linear flow on time plot as in Fig. 11. Material balance time plots should be used with care since the well is
missing data from flowback period that might affect gas and water results.
This method also helps in assessing the effect of fracing new wells on an existing well as in Clark pad. Well FF-18 was
the oldest well in the Clark pad and three wells were fraced in two periods. The location and the direction of the newly fraced
wells determined whether the calculated fracture volume was changed. In first period, the effective fracture volume did not
change. In the second period, the effective fracture volume increased which had a increase on gas production. This method is
a tool to know if the well was affected by the fracing of new wells.
Our final example is the Rowlett pad which was helpful in showing the effect of fracing sequence and the location of the
well in a pad. Four of the wells have the same well’s properties and one of them had a 60% reduction in frac load. In the case
of Well FF-24, the well produced more gas and less water since it was in between FF-21 and FF-22 and fraced after them.
Well FF-23 had a frac load reduction of 60% and was fraced last with a middle placement. Although frac load reduced, the
location and the frac sequence helped the well to produce more gas and less water if compared to the rest. In this pad, the
wells with the highest effective fracture volume had the lowest cumulative gas.
Including flowback data has an effect on both time and material balance time diagnostic plots. In time, it will shift the data
giving possibly a different signature than the original signature. In material balance time, the data is shifted due to adding
extra high rates in the flowback period which will give the correct BDF in water data. If this well was analyzed without
flowback data and compared with production the difference might be big. In some wells, including or excluding flowback
data would not make a difference in the production data analysis. To avoid misinterpretation, flowback data should be
included in the analysis.

Conclusions
Following conclusions can be drawn from the presented work:
 Gas dominates the diffusivity equation in water/gas system if mobility and compressibility conditions are met which
is usually around Sw < 0.7 in the fracture.
 A new method was developed to calculate effective fracture volume (water volume) by modifying the total
compressibility calculation to include gas compressibility.
 In production data analysis, ignoring flowback data can lead to misinterpretation in flow regime identification and
any early gas/water data should be included in the analysis even if it is separated by a shut-in period.
 Even if the flowback data is not available, the delay time should be included to avoid stretching of data and giving
the wrong reservoir signature.
 Including flowback data has an effect on both time and material balance time diagnostic plots.
 Water rate normalized pressure plot can be used as a tool to evaluate frac jobs based on effective fracture volume.
 Ignoring flowback period with high water rates might indicate a smaller effective water volume.
 This method can be a mean of tracking the interference effect of newly drilled wells on effective water volume of
the existing wells.
 Some wells suggested that less frac water is needed for the wells in the middle of a pad which will produce more gas
and less water if compared to the rest of the wells in the pad.
14 SPE 166279

Acknowledgement
The authors would like to thank the sponsors of the Computer Modeling Consortium at Texas A&M University for their
support and CMG for the use of their simulation software. Southwestern Energy data contribution made the completion of
this research possible.

Nomenclature
Acm = total matrix surface area draining into fracture system, ft2
BDF = boundary dominated flow
Bw = water formation volume factor, rbbl/STB
CHF = hydraulic fracture conductivity, md-ft
ct = total compressibility, psi-1
cg = gas compressibility, psi-1
cf = formation compressibility, psi-1
cw = water compressibility, psi-1
h = reservoir thickness, ft
k = permeability, md
keff = effective permeability, md
km = matrix permeability, md
kr = relative permeability, fraction
LF = hydraulic fracture spacing, ft
m(p) = real gas pseudo-pressure (gas) 2 , psi2/cp
mpss = slope of BDF period in a cartesian plot
nF = number of hydraulic fracture
OGIP = original gas in place
PDA = production data analysis
PNR = pressure normalized rate, Mcf/D/psis or
pi = initial reservoir pressure, psi
pwf = wellbore flowing pressure, psi
RNP = rate normalized pressure, sdfds
Sg = gas saturation, fraction
Sgirr = irreducible gas saturation, fraction
Sw = water saturation, fraction
SG = specific gravity, fraction
T = temperature, oF
t = time, days
tMB = material balance time, days
Vw = effective water volume, bbl
Vinj = injected frac volume, bbl
xf = hydraulic fracture half-length, ft

Greek symbols
  viscosity, cp
φ = porosity
λ = mobility

Subscript
eff = effective
F = hydraulic fracture
i = initial
m = matrix
w = water
SPE 166279 15

References

Abbasi, M., Dehghanpour, H., and Hawkes, R.V. 2012. Flowback Analysis for Fracture Characterization. Paper presented at
the SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada. Society of Petroleum
Engineers SPE-162661-MS. DOI: 10.2118/162661-ms.
Alkouh, A. and Wattenbarger, R.A. 2013. New Advances in Shale Reservoir Analysis Using Flowback Data. Paper presented
at the SPE Eastern Regional Meeting, Pittsburgh, Pennsylvania, USA. Society of Petroleum Engineers SPE-
165721-MS. DOI: 10.2118/165721-ms.
Alkouh, A.B., Patel, K., Schechter, D. et al. 2012. Practical Use of Simulators for Characterization of Shale Reservoirs. Paper
presented at the SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada. Society of
Petroleum Engineers SPE-162645-MS. DOI: 10.2118/162645-ms.
Crafton, J.W. and Gunderson, D.W. 2007. Stimulation Flowback Management--Keeping a Good Completion Good. Paper
presented at the SPE Annual Technical Conference and Exhibition, Anaheim, California, U.S.A. Society of
Petroleum Engineers SPE-110851-MS. DOI: 10.2118/110851-ms.
Harpel, J.M., Barker, L.B., Fontenot, J.M. et al. 2012. Case History of the Fayetteville Shale Completions. Paper presented at
the SPE Hydraulic Fracturing Technology Conference, The Woodlands, Texas, USA. Society of Petroleum
Engineers SPE-152621-MS. DOI: 10.2118/152621-ms.