Upload
amr-al-sayed
View
94
Download
0
Embed Size (px)
DESCRIPTION
This paper presents an analytical model for unconventional reservoirs with horizontal wells with multiple fractures. The model is an extension of the "trilinear flow" solution, but it subdivides the reservoir into five regions instead of three. This enables it to be used for more-complex reservoirs. Accordingly, the model can simulate a fracture that is surrounded by a stimulated region of limited extent (fracture branching), whereas the remaining reservoir is nonstimulated. In addition to modeling flow within the fracture and flow within the stimulated region, the model takes into account flow from the surrounding nonstimulated region, both parallel to and perpendicular to the fracture. The model can be used to simulate the flow in tight reservoirs with multifractured horizontal wells. In many cases, the fractures do not have a simple biwing shape, but are branched. This effectively creates regions of higher permeability around each fracture, which obviously affect the production performance significantly. However, in many tight reservoirs, in spite of their low permeability, the surrounding nonstimulated region can also be a significant contributor to long-term production. The five-region model accounts for this contribution. Thus, it is particularly valuable when generating production forecasts for reserves evaluation. The model was validated by comparing its results with numerical simulation. We found that analytical and numerical results are in good agreement only when the geometry of the system falls within certain limitations. However, these limitations are met in most cases of interest. Therefore, the model is useful for practical engineering purposes.
Citation preview
Analytical Model forUnconventional Multifractured
Composite SystemsEkaterina Stalgorova and Louis Mattar, IHS
Summary
This paper presents an analytical model for unconventional reser-voirs with horizontal wells with multiple fractures. The model isan extension of the trilinear flow solution, but it subdivides thereservoir into five regions instead of three. This enables it to beused for more-complex reservoirs. Accordingly, the model cansimulate a fracture that is surrounded by a stimulated region oflimited extent (fracture branching), whereas the remaining reser-voir is nonstimulated. In addition to modeling flow within thefracture and flow within the stimulated region, the model takesinto account flow from the surrounding nonstimulated region, bothparallel to and perpendicular to the fracture.
The model can be used to simulate the flow in tight reservoirswith multifractured horizontal wells. In many cases, the fracturesdo not have a simple biwing shape, but are branched. This effec-tively creates regions of higher permeability around each fracture,which obviously affect the production performance significantly.However, in many tight reservoirs, in spite of their low permeability,the surrounding nonstimulated region can also be a significant con-tributor to long-term production. The five-region model accounts forthis contribution. Thus, it is particularly valuable when generatingproduction forecasts for reserves evaluation.
The model was validated by comparing its results with numeri-cal simulation. We found that analytical and numerical results arein good agreement only when the geometry of the system fallswithin certain limitations. However, these limitations are met inmost cases of interest. Therefore, the model is useful for practicalengineering purposes.
Introduction
A common way to produce hydrocarbons from tight reservoirs isto use horizontal wells with multiple hydraulic fractures. It hasbeen postulated that, in many cases, the propagation of a fracturemay create a branch pattern (Daneshy 2003). Fracture branchingcreates a stimulated region around each hydraulic fracture, whichcan be modeled by introducing a region of higher permeability(Fig. 1).
Brown et al. (2009) suggested an analytical solution todescribe the pressure distribution for a system in which the en-hanced region occupies all the space between fractures (Fig. 2a).They considered that the model consists of three regions: fracture,high-permeability region, and low-permeability region. A 1D lin-ear flow was assumed within each region. Brohi et al. (2011) pro-vided another solution for the same configuration; the maindifference between these two solutions is the way in which thethree regions are coupled.
In our earlier study (Stalgorova and Mattar 2012), we sug-gested a solution for the case when the enhanced region occupiesonly part of the space between the fractures, but the regionbeyond the fractures was neglected (Fig. 2b) Similar to Brownet al. (2009), we considered three linear-flow regions (fracture,
high-permeability region, and low-permeability region), but thelocation of the regions and their interaction were different.
In this paper, we present a solution for the case shown inFig. 2c. To provide the pressure distribution for such a system, weconsider five regions of linear flow; therefore, the suggestedmodel is referred as a five-region model. This case is a generaliza-tion that encompasses both the trilinear-model solution suggestedby Brown et al. (2009) and the enhanced-fracture-region modelpresented by Stalgorova and Mattar (2012).
Five-Region Model
The mathematical formulation of the proposed model is anextension of the one presented in our earlier study (Stalgorovaand Mattar 2012). We present an outline of the solution in thissection, whereas the detailed derivation is given in AppendicesA and B.
To model a reservoir with a multifractured horizontal well, weassume that all fractures have the same length and conductivityand are spaced uniformly along the horizontal well. This assump-tion is realistic because it is common field practice to designequally spaced hydraulic fractures with similar properties. Weintroduce a region of higher permeability around each fracture torepresent fracture branching. Because of the symmetry of the sys-tem, we can perform calculations on one-quarter of the spacebetween the fractures. Flow in the model is treated as a combina-tion of five linear flows within contiguous regions, as shown inFig. 3. We formulate the 1D flow solutions for each region andthen couple them by imposing flux and pressure continuity acrossthe boundaries between regions.
A schematic for the suggested five-region model is presented inFig. 3. The model consists of Regions 1 through 4 and the FractureRegion. To keep the solution general, we allow different properties(permeability, porosity, and total compressibility) in each region.However, for practical purposes, properties for Regions 2, 3, and 4can be considered identical because they represent the original res-ervoir rock, whereas Region 1 represents the region of enhancedpermeability immediately near the fracture.
The pressure distribution in each region is governed by the
flow equation r q klrq
@
@t/q (Dake 1978). For the case
of liquid flow, this equation can be rewritten as
r2p /lck
@p
@t1
Note: for the case of gas flow, the equation should be rewritten interms of pseudopressure, as discussed later in the Possible ModelModifications section.
To simplify the solution, we rewrite this equation for eachregion in dimensionless terms and convert it to the Laplace do-main. Definitions of dimensionless variables are given in Appen-dix A; derivations of equations and solutions are given inAppendix B.
Region 4
The diffusivity equation becomes
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
CopyrightVC 2013 Society of Petroleum Engineers
This paper (SPE 162516) was accepted for presentation at the SPE CanadianUnconventional Resources Conference, Calgary, 30 October1 November 2012, andrevised for publication. Original manuscript received for review 7 December 2012. Revisedmanuscript received for review 21 March 2013. Paper peer approved 30 May 2013.
246 August 2013 SPE Reservoir Evaluation & Engineering
@2p4D@y2D
sg4D
p4D 0 2 where p4D is the pressure in Region 4 in the Laplace domain. Boundary Condition 1: No-flow condition at the outer reser-voir boundary (y y2)
@p4D@yD
yDy2D
0 3
Boundary Condition 2: Pressure continuity between Regions4 and 2 (at y y1)
p4Dy1D p2Dy1D 4
Region 3
The diffusivity equation becomes
@2p3D@y2D
sg3D
p3D 0 5
Boundary Condition 1: No-flow condition at the outer reser-voir boundary (y y2)
@p3D@yD
yDy2D
0 6
Boundary Condition 2: Pressure continuity between Regions3 and 1 (at y y1)
p3Dy1D p1Dy1D 7
Region 2
The diffusivity equation becomes
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
a b c
k1 k1 k1
k2
k2k2
Fig. 2(a) Trilinear-flow model (Brown et al. 2009), (b) enhanced-fracture-region model (Stalgorova and Mattar 2012), and (c) five-region model (this study).
Stimulatedregion width (x1)y
Region 3: Region 4:
k3, 3, c3
Region 1:
k1, 1, c1
kf, f, cf
Region 2:
k2, 2, c2
k4, 4, c4
w/2
y 1 =
Fr
act
ure
half
leng
th ((
x f)y)
Fracture:
Fracture half width = w/2
Half distance between fractures (x2)
x
Res
ervo
ir ha
lf wi
dth
(y 2)
Fig. 3Schematic and dimensions for five-region model (one-quarter of a fracture).
a b
k1
k2
Fig. 1Horizontal well with multiple branch fractures (a) and its representation by a model (k1> k2) (b).
August 2013 SPE Reservoir Evaluation & Engineering 247
@2p2D@x2D
k4k2y1D
@p4D@yD
y1D
sg2D
p2D 0 8
Boundary Condition 1: No-flow boundary midway betweenthe fractures
@p2D@xD
x2D
0 9
Boundary Condition 2: Pressure continuity between Regions1 and 2
p2Dx1D p1Dx1D 10
Region 1
The diffusivity equation becomes
@2p1D@x2D
k3k1y1D
@p3D@yD
y1D
sg1D
p1D 0 11
Boundary Condition 1: Flux continuity between Regions 2and 1
k2l
@p2D@xD
x1D
k1l
@p1D@xD
x1D
12
Boundary Condition 2: Continuity between Region 1 andFracture Region
p1DwD=2 pFDwD=2 13
Fracture Region
The diffusivity equation becomes
@2pFD@y2D
2FCD
@p1D@xD
wD=2
sgFD
pFD 0 14
Boundary Condition 1: No flow through the fracture tip(yD y1D)
@pFD@yD
y1D
0 15
Boundary Condition 2: Obtained by applying Darcys law atthe wellbore (yD 0). (A correction to account for flowline con-vergence in the fracture, caused by the well being horizontalrather than vertical, is introduced later in Eq. 20). In dimension-less terms, in the Laplace domain, it becomes
@pFD@yD
0
pFCD s 16
Flow directions and boundary conditions for each region aresummarized in Fig. 4.
It can be shown (see detailed derivation in Appendix B) thatthe solution of the preceding equations with boundary conditions(Eqs. 2 through 16) is:
pFDyD pcoshyD 1
c6s
p FCD s
c6s
psinh c6sp 17
where c6(s) is calculated as shown in Appendix B. The solution atthe wellbore is given by
pwD pFD0 p
FCD s c6s
ptanh c6sp 18
This solution can be inverted from the Laplace to the time domainwith the numerical algorithm given by Stehfest (1970).
It is worth noting that, in the case when the stimulated region(Region 1) occupies all the space between fractures (x1 x2), thefive-region model reduces to the trilinear model of Brown et al. 2009(Fig. 2a). In the case when the contribution of Regions 3 and 4 isneglected (y1 y2), the five-region model reduces to the enhancedfracture region model of Stalgorova and Mattar (2012) (Fig. 2b).Therefore, the five-region model is a generalization that covers boththe trilinear model and the enhanced fracture region model.
Possible Model Modifications
This section follows Stalgorova and Mattar (2012).
Gas Flow
The solution was derived for a liquid system. To use the solutionfor gas flow, the dimensionless pressure should be expressed interms of real-gas pseudopressure (Eq. A-1).
It is, however, important to keep in mind that, for the case of
gas, the diffusivity term g k/lc
is not constant but changes with
pressure, so the diffusivity equation becomes nonlinear. One ofthe ways to deal with this problem is to use pseudotime (Andersonand Mattar 2005).
Dual Porosity
If desired, a dual-porosity modification can be introduced to themodel: The inner region can be treated as a dual continuum. Theessence of the modification consists of replacing s with u sf(s)in Eq. 11, in which f(s) is a function defined by the dual-porositycharacteristics of the system. A detailed description of the dual-porosity modification is given in Brown et al. (2009).
Skin Because of Convergence (Choking Skin). We assumed 1Dlinear flow within the hydraulic fracture. This would be appropri-ate if the well intersecting the fracture were vertical. However, inour present scenario, the well is horizontal, and the flowlines inthe fracture itself are not linear all the way, but as they approachthe horizontal well, they converge toward the wellbore (Fig. 5).This convergence results in an additional pressure drop, which isreferred to as a skin because of convergence (choking skin).Mukherjee and Economides (1991) provided a formulation (sc) toaccount for this additional pressure drop. Eq. 18 is modified as
. . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . .
. . . . . .
No-flow
No-flow
No-
flow
No-flow
Region 3 Region 4
Region 1 Region 2
p1 = p3
p F =
p 1
p 1 =
p 2
p2 = p4y1 = (xf)y
y2
y
w/2 x1 x2pFy
qF
x
k 1p
1y
p2
x=
k 2
Fig. 4Five-region model: flow directions and boundaryconditions.
248 August 2013 SPE Reservoir Evaluation & Engineering
pwD p
FCD s c6s
ptanh c6sp
scs
19
where sc is the skin because of convergence, which can be cal-culated from
sc k1hkf w
lnh
2rw
p
2
20
Wellbore Storage
The wellbore-storage effect can also be incorporated as suggestedby Brown et al. (2009). To account for wellbore storage, modifythe wellbore pressure given in Eq. 19 to
pwD;storage pwD
1 CDs2pwD21
where CD is the dimensionless wellbore-storage coefficient.
Model Validation
We compared the results of the five-region model with the resultsof numerical simulation. We used a single-phase liquid for thisexercise. For numerical simulation, we used a 2D finite-differencemodel. To ensure the accuracy of the model, we defined a finegrid throughout the reservoir, and enforced additional refinementnear the fracture and near the boundaries of the regions. One-quarter of a fracture stage was modeled with 33 91 grid cells.Overall, the cell size was approximately 2 3 ft, and the cellsnear the fracture and near the region boundaries were 0.6 0.6 ft.Other model parameters are presented in Table 1. For these pa-rameters, the five-region-model results showed excellent agree-ment with the numerical simulation, as can be seen in Fig. 6.
Alternative Model (for Reservoirs Elongated inx-Direction)
In the original five-region model, it was assumed that flow can beapproximated by a combination of linear flows in each of the fiveregions, as was shown in Fig. 4. This assumption is valid for some
cases. However, there are other cases when such an approxima-tion produces inaccurate results. For example, if x2 is much largerthan y2, we can expect the flowlines in Region 4 to be aligned inthe direction of the pressure sink (the fracture). To account forsuch reservoir geometries, we developed an alternative five-regionmodel that directs the flow in Region 4 toward Region 3 ratherthan Region 2. This is depicted in Fig. 7. The modifications to therelevant equations and boundary conditions were straightforward.
Figs. 8 and 9 compare the original five-region model with thealternative five-region model for two different geometries. For thecase in which y2>>x2 (Fig. 8a), the original five-region model is inclose agreement with the numerical model, whereas the alternativemodel produces inaccurate results (Fig. 8b). On the other hand, forthe cases in which x2>>y2, the alternative five-region model givesmore accurate results than the original one (Figs. 9a and 9b).
For a typical field case of a horizontal well with multiple frac-tures, x2 100 ft (0.5 times the horizontal-well length/number offractures) and y2 500 ft (half the distance between two neigh-boring horizontal wells). Therefore, the use of the original five-region model (shown in Fig. 4) is more appropriate for most casesof interest.
It was shown previously that, under certain reservoir configura-tions, the original five-region model is applicable, whereas otherconfigurations require the use of the alternative five-region model.Unfortunately, there are cases when both the original and the alter-native models are not accurate enough. One such case is whenboth x2 and y2 are much larger than the fracture half-length (xf)y(e.g., a single fracture in an infinite reservoir). Fig. 10 shows thatfor such a configuration, the numerical solution is significantly dif-ferent from that of either model. The explanation for this inaccur-acy is that, at distances far away from the fracture, the flowlinesshould be radial not linear; however, the five-region models do notallow for this. The five-region model was developed to deal withmultiple-fracture horizontal wells. The geometry of these systemsis not likely to lead to radial flow. Therefore, rather than resolvingthis problem, we have focused instead on defining the conditionsof the applicability of the original five-region model.
Applicability of the original five-region model
To evaluate when the Five-Region Model is applicable, we per-formed comparisons with numerical models for different reservoirconfigurations. Agreement between the five-region model and thenumerical model depends on many parameters such as x1, x2,(xf)y, y2, k1, k2, and qF. It may seem difficult to test all possible
. . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
TABLE 1MODEL PARAMETERS
Initial pressure pi 10,000 psi
Viscosity l 1 cpFormation volume factor B 1 bbl/STB
Compressibility c 1106 psi1Horizontal-well length Le 2,000 ft
Reservoir width Ye 1,000 ft
Fracture half-length
(in y-direction)
(xf)y 100 ft
Dimensionless fracture
conductivity
FCD 20.8
Distance from well to
permeability boundary
x1 25 ft
Number of fractures nf 20
Net pay h 100 ft
Permeability of
stimulated region
k1 1 md
Permeability of
nonstimulated regions
k2 k3k4 0.01 md
Porosity / 0.1
Saturation Sw 1
Production rate q 400 STB/D
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
Pwf,
psi
0 200 400 600 800 1,000time, days
Pwf (Five-Region Model)Pwf (numerical simulation)
Fig. 6Comparison of five-region-model results with numeri-cal-model results.
Fracture
Horizontal wellbore
Fig. 5Convergence of flowlines toward the wellbore withinthe fracture.
August 2013 SPE Reservoir Evaluation & Engineering 249
combinations of parameters. However, looking at the solution,one can notice that the number of parameters that actually affectthe result can be minimized, if we compare pD vs. tD plots (dimen-sionless) instead of comparing p vs. t plots. Fig. 11 illustrates thecomparison of the dimensionless type curves obtained from theoriginal five-region model and the numerical simulator for onereservoir geometry.
We ran a series of such comparisons with different combina-tions of dimensionless parameters, and we observed that the nu-merical and five-region-model type curves show reasonableagreement if all the following conditions are met:
x1 0.1 x2.In other words, the stimulated zone occupies at least 10% of
the space between fractures. (xf)y 0.1 y2.This condition ensures that the fracture length is not too small
compared with the reservoir size. If the fracture length is verysmall, radial flow is more likely to occur, and this is not handledby the five-region model. The (xf)y is usually at least 100 ft; y2 is
the half-distance between neighboring horizontal wells, whichcan be estimated at approximately 1,000 ft; therefore, this condi-tion is usually met.
y2 2 x2. The area drained by each fracture is elongated inthe y-direction (if the geometry were more elongated in the x-direction, the alternative model would be more appropriate). Thex2 is the half-distance between adjacent fractures, so it can beexpressed as x2 Le/(2nf), in which Le is the horizontal-welllength (approximately 4,000 ft) and nf is the number of fractures(at least 10). Therefore, we can safely assume that x2 is less than400; thus, y2 2 x2, and the condition y2 2 x2 is typically met.
Conclusions
We have presented an analytical model to simulate flow through ahorizontal well with branch fractures surrounded by a nonstimu-lated zone. The following conclusions can be made: The proposed five-region model is a generalization for both the tri-
linear model (Brown et al. 2009) and the enhanced-fracture-region
y2
y No-flow
Region 3
Region 1
Region 4
Region 2
No-
flow
No-
flow
No-flow
y1 = (xf)y
pFy
qF
w/2 x1 x2x
p1 = p3p F
=
p 1
p 3 =
p 4
p 1 = p 2
k1p1y
p2x
= k2
Fig. 7Alternative five-region model: flow directions and boundary conditions.
Original
Alternative
a b
x1 = 25 ft
k2 = 0.1 md
x2 = 250 ft
(xf) y
= 10
0 ft
y 2 =
10
00 ft
k1 = 1 md
10000
9000
8000
7000
6000
5000
4000
3000
2000
1000
00 2,000 4,000 6,000 8,000 10,000
time, days
pwf,
psi
Original Five-Region Model
Alternative Five-Region Model
Numerical Simulation
Fig. 8y2x2: Model schematic (a) and comparison of the original and alternative five-region models with numerical simulation(b).
250 August 2013 SPE Reservoir Evaluation & Engineering
model (Stalgorova and Mattar 2012). Equations describing themodel have been solved analytically in the Laplace domain.
The flow through a horizontal well with multiple branch frac-tures can be represented by five regions of linear flow.
We compared the results obtained from the five-region modelwith the results of numerical simulation. The results showedclose agreement for typical well and reservoir configurations.
Depending on the reservoir geometry, the five-region model isnot always applicable. We suggested the alternative five-regionmodel for cases when x2y2. However, this condition does notapply for most of the practical cases of multiple-fractured hori-zontal wells.
We found that the five-region model gives accurate (close tonumerical) results when the stimulated zone is not too smallcompared with the whole reservoir (in the x- and y-direction),and the model is elongated in the y-direction. Formally, it canbe expressed as x1 0.1 x2, (xf)y 0.1 y2, and y2 2 x2. It wasshown that these conditions are met in most practical cases.
Nomenclature
B liquid formation volume factor, RB/STBc compressibility, 1/psi
CD dimensionless wellbore storagect total compressibility, 1/psi
FCD dimensionless fracture conductivityh net pay, ftk permeability, md
OriginalAlternative
k2 = 0.1 md
k1 = 1 md
x1 = 50 ft
(xf) y
= 10
0 ft
y 2 =
10
00 ft
x2 = 1000 ft
5000
4500
4000
3500
3000
2500
2000
1500
1000
500
0
pwf,
psi
b
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000time, days
aOriginal Five-Region Model
Alternative Five-Region Model
Numerical Simulation
Fig. 10x2 and y2(xf)y: Model schematic (a) and comparison of the original and alternative five-region models with numericalsimulation (b).
100000
10000
1000
100
10
1
0.1
p D
0 1 10 100 1000 10000 100000 1000000tD
pD (Five-Region Model)pD (Numerical Simulation)
x1D = 1 x2D = 5 y2D = 20 k2 = 0.01
Fig. 11Comparison of dimensionless type curves obtainedfrom the five-region model and numerical simulation.
OriginalAlternative
k2 = 0.1 md
x2 = 3000 ft
(xf) y
= 10
0 ft
k1 = 1 md y 2 =
30
0 ft
x1 = 50 ft
5000
4500
4000
3500
3000
2500
2000
1500
1000
500
0
pwf,
psi
Original Five-Region Model
Alternative Five-Region Model
Numerical Simulation
0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000time, days
a
b
Fig. 9x2y2: Model schematic (a) and comparison of the original and alternative five-region models with numerical simulation(b).
August 2013 SPE Reservoir Evaluation & Engineering 251
kf fracture permeability, mdki permeability of the ith region, mdLe horizontal-well length, ftnf number of fracturesp pressure, psi
pD dimensionless pressurepFD dimensionless pressure in the fracturepi initial reservoir pressure, psi
piD dimensionless pressure in the ith regionpp pseudopressure, psi2/cppwf wellbore flowing pressure, psiq well-production rate, SDB/D, Mscf/DqF production rate through one fracture, SDB/D, Mscf/Drw wellbore radius, fts Laplace-transform parametersc skin because of convergenceSw water saturation, fractiont time, hoursT reservoir temperature, RtD dimensionless timew fracture width, ft
wD dimensionless fracture widthx coordinate along the horizontal well, ftx1 distance from fracture to permeability boundary, ftx2 half-distance between fractures, ftxD dimensionless coordinate along the horizontal wellx1D dimensionless distance from fracture to permeability
boundaryx2D dimensionless half-distance between fractures
(xf)y half fracture length, fty coordinate perpendicular to the horizontal well, fty1 fracture half-length, fty2 half-distance between horizontal wells, ftyD dimensionless coordinate perpendicular to the horizontal
wellYe reservoir size in y-direction (distance between horizontal
wells), ftz gas-compressibility factorg diffusivity, ft2/hr
gFD dimensionless diffusivity in the fracturegi diffusivity in the ith region, ft2/hrgiD dimensionless diffusivity in the ith regionl viscosity, cp/ porosity, fraction/i porosity of the ith region, fraction/f fracture porosity, fraction
Acknowledgments
The authors would like to thank Karel Zaoral for his advice, valu-able and stimulating discussions, and contribution to the softwareimplementation.
References
Anderson, D.M. and Mattar, L. 2005. An Improved Pseudo-Time for Gas
Reservoirs With Significant Transient Flow. Paper 2005-114 presented
at the Canadian International Petroleum Conference, Calgary, Alberta,
79 June. http:/dx.doi.org/10.2118/2005-114.
Brohi, I., Pooladi-Darvish, M., and Aguilera, R. 2011. Modeling Fractured
Horizontal Wells as Dual Porosity Composite ReservoirsApplica-
tion to Tight Gas, Shale Gas and Tight Oil Cases. Paper SPE 144057
presented at the SPE Western North American Region Meeting, An-
chorage, Alaska, 711 May. http:/dx.doi.org/10.2118/144057-MS.
Brown, M., Ozkan, E., Raghavan, R. et al. 2009. Practical Solutions for
Pressure Transient Responses of Fractured Horizontal Wells in Uncon-
ventional Reservoirs. Paper SPE 125043 presented at the SPE Annual
Technical Conference and Exhibition, New Orleans, Louisiana, 47
October. http:/dx.doi.org/10.2118/125043-MS.
Dake, L.P. 1978. Fundamentals of Reservoir Engineering. Amsterdam:Elsevier.
Daneshy, A.A. 2003. Off-Balance Growth: A New Concept in Hydraulic
Fracturing. J. Pet Tech 55 (4): 7885. http:/dx.doi.org/10.2118/80992-MS.
Mukherjee, H. and Economides, M. 1991. A Parametric Comparison of
Horizontal and Vertical Well Performance. SPE Form Eval 6 (2):
209216. http:/dx.doi.org/10.2118/18303-PA.
Stalgorova, E. and Mattar, L. 2012. Practical Analytical Model to Simu-
late Production of Horizontal Wells With Branch Fractures. Paper SPE
162515 presented at the SPE Canadian Unconventional Resources
Conference, Calgary, Alberta, 30 October1 November. http:/dx.doi.
org/10.2118/162515-MS.
Stehfest, H. 1970. Numerical Inversion of Laplace Transforms. Communi-cations of the ACM 13 (1): 4749. http:/dx.doi.org/10.1145/361953.
361969.
Appendix ADimensionless Variables
Dimensionless pressure:
pD k1h
141:2qFBlpi p
k1h
1422qFTpppi ppp
for liquid
for gas
8>>>: A-1
where qF q/(number of fractures) is the flow rate through onefracture and pp(p) is the real-gas pseudopressure defined by
ppp 2ppref
p
lpzp dp A-2
Dimensionless time:
tD g1txf 2yA-3
Dimensionless distances:
xD x=xf yx1D x1=xf yx2D x2=xf ywD w=xf yyD y=xf yy1D y1=xf y xf y=xf y 1y2D y2=xf y
A-4
Dimensionless diffusivities for five regions:
g1D g1=g1 1g2D g2=g1g3D g3=g1g4D g4=g1gFD gF=g1
A-5
where gF 2:637 104kF
/ctFland gi 2:637 104
ki/ctil
(i 1,2, 3, or 4 indicates region).Dimensionless fracture conductivity:
FCD kf wk1xf y
kf wDk1
A-6
Appendix BProblem Derivation and Solution
The suggested solution is based on the assumption that the flowcan be represented by a combination of flows within contiguousregions, and that the flow within each region is 1D, as shown inFig. B-1. Formally speaking, this assumption is inconsistent. Forexample, assuming 1D linear flow in Region 4 means that pres-sures at A and B in Fig. B-1 are equal. At the same time, pressure
. . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
252 August 2013 SPE Reservoir Evaluation & Engineering
at C can be considered equal to pressure at B, as well as pressureat D to pressure at A (a result of pressure continuity); therefore,pressure at C should be equal to pressure at Dbut, at the sametime, we assume flow from C to D in Region 3.
Though this simplified formulation has some inconsistenciesfrom a formal point of view, it gives a solution that is close to theexact solution for the cases of interest (as was demonstrated in theApplicability of the Original Five-Region Model section); there-fore, the suggested solution can be used for practical purposes.
In the general case, for liquid pressure, distribution in eachregion is described by the diffusivity equation
r2p g @p@t
B-1
where g k/lc
; therefore, g is different for different regions.
In dimensionless terms defined in Appendix A, Eq. B-1 can berewritten as
r2piD 1giD@piD@tD
0 B-2
where index i indicates one of the regions1, 2, 3, 4, or fracture(F).
After conversion into the Laplace domain, Eq. B-2 becomes
r2piD s
giDpiD 0 B-3
Here, the bar above the pressure symbol indicates the Laplace do-main, and s is the Laplace-transform parameter with respect to tD.
Equation and Boundary Conditions for Region 4
We assume that flow in Region 4 is a 1D linear flow and p4D does
not depend on xD. In this case, r2p4D @2p4D@y2D
; so, Eq. B-3
becomes
@2p4D@y2D
sg4D
p4D 0 B-4
Boundary Condition 1: No-flow condition at the outer reser-voir boundary (y y2)
@p4D@yD
yDy2D
0 B-5
Boundary Condition 2: Pressure continuity between Regions4 and 2 (at y y1)
p4Dy1D p2Dy1D B-6
The general form of the solution for Eq. B-4 can be given as
p4DyD A4cosh yD y2Ds
g4D
r
B4sinh yD y2Ds
g4D
r B-7
By applying the boundary condition given by Eq. B-5, weobtain B4 0. By applying the boundary condition given by Eq.B-6, we obtain A4 p2Dy1D
cosh y1D y2Ds
g4D
r" #
Therefore, the pressure for Region 4 can be rewritten in termsof the pressure for Region 2 at the boundary between Regions 2and 4:
p4DyD p2Dy1D cosh yD y2D
s
g4D
r" #
cosh y1D y2Ds
g4D
r" # B-8
Flux between Regions 2 and 4 is proportional to
@p4D@yD
y1D
p2Dy1Ds
g4D
r tanh y1D y2D
s
g4D
r
B-9
Equation and Boundary Conditions for Region 3
We assume that flow in Region 3 is a 1D linear flow and p3D does
not depend on xD. In this case, r2p3D @2p3D@y2D
; so, Eq. B-3
becomes
@2p3D@y2D
sg3D
p3D 0 B-10
Boundary Condition 1: No-flow condition at the outer-reser-voir boundary (y y2)
@p3D@yD
yDy2D
0 B-11
Boundary Condition 2: Pressure continuity between Regions3 and 1 (at y y1)
p3Dy1D p1Dy1D B-12
The general form of the solution for Eq. B-10 can be given as
p3DyD A3cosh yD y2Ds
g3D
r
B3sinh yD y2Ds
g3D
r B-13
By applying the boundary condition given by Eq. B-11, weobtain B3 0. By applying the boundary condition given byEq. B-12, we obtain A3 p1Dy1D
cosh y1D y2Ds
g3D
r" #
. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . .
. . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
Region 3 Region 4
Region 1 Region 2
A B
C D
Fracture x
y
Fig. B-1Flow represented by a combination of 1D linear flows.
August 2013 SPE Reservoir Evaluation & Engineering 253
Therefore, the pressure for Region 3 can be rewritten in termsof the pressure for Region 1 at the boundary between Regions 1and 3:
p3DyD p1Dy1D cosh yD y2D
s
g3D
r" #
cosh y1D y2Ds
g3D
r" # B-14
Flux between Regions 1 and 3 is proportional to
@p3D@yD
y1D
p1Dy1Ds
g3D
r tanh y1D y2D
s
g3D
r
B-15
Equation and Boundary Conditions for Region 2
Eq. B-3 for Region 2 is
@2p2D@x2D
@2p2D@y2D
sg2D
p2D 0 B-16
To convert Eq. B-16 into a 1D form, we integrate each of itsterms from 0 to y1D with respect to yD.
Assuming that pressure in the second region does not dependon yD, integrating the first term of Eq. B-16 givesy1D
0
@2p2D@x2D
dyD y1D @2p2D@x2D
B-17
Integrating the second term of Eq. B-16 gives
y1D0
@2p2D@y2D
dyD @p2D@yD
y1D
@p2D@yD
0
B-18
There is a no-flow boundary along the horizontal well; there-
fore,@p2D@yD
0
0:Flux across the boundary between Regions 2 and 4 is continu-
ous; therefore, k2@p2D@yD
y1D
k4 @p4D@yD
y1D
.
So, integrating the second term of Eq. B-16 gives
y1D0
@2p2D@y2D
dyD k4k2
@p4D@yD
y1D
B-19
Assuming that the pressure in Region 2 does not depend onyD, integrating the third term of Eq. B-16 givesy1D
0
s
g2Dp2DdyD y1D
s
g2Dp2D B-20
Therefore, after integration, Eq. B-16 becomes
y1D@2p2D@x2D
k4k2
@p4D@yD
y1D
y1D sg2Dp2D 0 B-21
Dividing Eq. B-21 by y1D, we obtain
@2p2D@x2D
k4k2y1D
@p4D@yD
y1D
sg2D
p2D 0 B-22
Boundary Condition 1: No-flow boundary midway betweenthe fractures
@p2D@xD
x2D
0 B-23
Boundary Condition 2: Pressure continuity between Regions1 and 2
p2Dx1D p1Dx1D B-24
Using Eq. B-9 and assuming that p2D does not depend on yD,Eq. B-22 can be rewritten as
@2p2D@x2D
c1sp2D 0 B-25
where
c1s sg2D k4
k2y1D
s
g4D
rtanh y1D y2D
s
g4D
r B-26
The general form of the solution for Eq. B-25 can be given as
p2DxD A2coshxD x2Dc1s
p
B2sinhxD x2Dc1s
p B-27
By applying the boundary condition given by Eq. B-23, weobtain B2 0. By applying the boundary condition given byEq. B-24, we obtain A2 p1Dx1D
coshx1D x2Dc1s
p Therefore, pressure for Region 2 (Eq. B-27) can be rewritten
in terms of pressure for Region 1 at the boundary betweenRegions 1 and 2:
p2DxD p1Dx1D coshxD x2D
c1s
p coshx1D x2D
c1s
p B-28Flux between Regions 1 and 2 is proportional to
@p2D@xD
jx1D p1Dx1Dc1s
p tanhx1D x2D
c1s
p
B-29
Equation and Boundary Conditions for Region 1
Eq. B-3 for Region 1 is
@2p1D@x2D
@2p1D@y2D
sg1D
p1D 0 B-30
To convert Eq. B-30 to a 1D form, we integrate each of itsterms from 0 to y1D with respect to yD. In the same way that itwas performed for Region 2, we can obtain that after integrationEq. B-30 becomes
@2p1D@x2D
k3k1y1D
@p3D@yD
y1D
sg1D
p1D 0 B-31
Boundary Condition 1: Flux continuity between Regions 1and 2
k2l
@p2D@xD
x1D
k1l
@p1D@xD
x1D
B-32
Boundary Condition 2: Continuity between Region 1 andFracture Region
p1DwD=2 pFDwD=2 B-33
Using Eq. B-15 and assuming that p1D does not depend on yD,we can rewrite Eq. B-31 as
@2p1D@x2D
c2sp1D 0 B-34
. . .
. . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . .
. . . . . . . . .
. . . .
. . . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
254 August 2013 SPE Reservoir Evaluation & Engineering
where
c2s sg1D k3
k1y1D
s
g3D
rtanh y1D y2D
s
g3D
r B-35
The general form of the solution for Eq. B-34 can be given as
p1DxD A1coshxD x1Dc2s
p
B1sinhxD x1Dc2s
p B-36
By substituting xD x1D, we obtain that A1 p1D(x1D). If wesubstitute Eq. B-29 in the left side of the boundary conditiongiven by Eq. B-32 and use Eq. B-36 to calculate the right side ofthe boundary condition given by Eq. B-32, we can express B1 as
B1 p1Dx1D k2k1
c1sc2s
stanhx1D x2D
c1s
p p1Dx1D c3s B-37
where
c3s k2k1
c1sc2s
stanhx1D x2D
c1s
pB-38
Therefore, Eq. B-36 can be rewritten as
p1DxD p1DxD1 coshxD x1Dc2s
p
n o c3ssinhxD x1D
c2s
p
n o B-39
Applying the boundary condition given by Eq. B-33, weobtain
pFDwD=2 p1DxD1 coshwD=2 x1Dc2s
p
n o c3ssinhwD=2 x1D
c2s
p
n o B-40
Therefore,
p1DxD1 pFDwD=2
coshwD=2 x1Dc2s
p c3ssinhwD=2 x1D c2sp pFDwD=2
c4s B-41
where
c4s coshwD=2 x1Dc2s
p
c3ssinhwD=2 x1Dc2s
p B-42
So, the expression for the pressure in Region 1 (Eq. B-39) can berewritten in terms of the pressure in the fracture at the boundarybetween Region 1 and fracture:
p1DxD pFDwD=2
coshxD x1Dc2s
p c3ssinhxD x1D c2sp c4s
B-43Flux from Region 1 to the fracture is proportional to
@p1D@xD
wD=2
pFDwD=2c2s
p
sinhwD=2 x1Dc2s
p c3scoshwD=2 x1D c2sp c4s
pFDwD=2 c5s B-44
where
c5s c2s
p sinhwD=2 x1D
c2s
p c3scoshwD=2 x1D c2sp c4s
B-45
Equation and Boundary Conditions for FractureRegion (F)
Eq. B-3 for the Fracture Region is
@2pFD@x2D
@2pFD@y2D
sgFD
pFD 0 B-46
To convert Eq. B-46 to a 1D form, we integrate each of itsterms from 0 to wD/2 with respect to xD.
Integrating the first term of Eq. B-46 gives
wD=20
@2pFD@x2D
dxD @pFD@xD
wD=2
@pFD@xD
0
B-47
There is a no-flow boundary in the middle of the fracture (in
the y-direction); therefore,@pFD@xD
0
0:Flux across the boundary between Region 1 and the fracture is
continuous; therefore, kf@pFD@xD
wD=2
k1 @p1D@xD
wD=2
.
So, integrating the first term of Eq. B-46 gives
wD=20
@2pFD@x2D
dxD k1kf
@p1D@xD
jwD=2 B-48
Assuming that pressure in the fracture does not depend on xD,integration of the second and third terms of Eq. B-46 gives
wD=20
@2pFD@y2D
sgFD
pFD
dxD wD
2
@2pFD@y2D
sgFD
pFD
B-49Therefore, after integration, Eq. B-46 becomes
k1kf
@p1D@xD
wD=2
wD2
@2pFD@y2D
sgFD
pFD
0 B-50
Dividing Eq. B-50 by wD/2 and using the definition of FCD(Eq. A-6), we obtain
@2pFD@y2D
2FCD
@p1D@xD
wD=2
sgFD
pFD 0 B-51
Boundary Condition 1: No flow through the fracture tip(yD y1D)
@pFD@yD
y1D
0 B-52
Boundary Condition 2: Obtained by applying Darcys Lawat the wellbore (yD 0), as shown on Fig. B-2. In Field unitsDarcys Law through one half of the fracture is given by:
qF2 0:001127 kf h w
l B@p
@yB-53a
. . . . . . . .
. . . . . . . . . . . . . . . . . . . . . .
. . . . . . .
. . . . . . . .
. . . . . . . . . . . . . . .
. . . . . . . . . .
. . . . . . . . . . . . . . .
. . . . . .
. . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . .
August 2013 SPE Reservoir Evaluation & Engineering 255
If we convert@p
@yto
@pD@yD
by use of the chain rule and definitions
given by Eqs. A-1 and A-4 and then simplify the obtained equa-tion by use of Eq. A-6 and convert to the Laplace domain, theboundary condition becomes
@pFD@yD
0
pFCD s B-53b
Because this formulation ignores the radial convergence to-ward the horizontal well within the fracture, it will need to be cor-rected later on by incorporating a convergence skinsee thePossible Model Modifications section.
Using Eq. B-44 and assuming that pFD does not depend on xD,we can rewrite Eq. B-51 as
@2pFD@y2D
c6spFD 0 B-54
where
c6s sgFD 2
FCD c5s B-55
The general solution for Eq. B-54 is
pFDyD AFcoshyD y1Dc6s
p
BFsinhyD y1Dc6s
p B-56
Using the boundary condition given by Eq. B-52, we obtainBF 0, and using the boundary condition given by Eq. B-53b, weobtain
AF pFCD s
c6s
psinh c6sp y1D B-57
Therefore, the solution for the Fracture Region is
pFDyD pcoshyD y1D
c6s
p FCD s
c6s
psinh c6sp y1D B-58
Using that y1D 1 (Eq. A-4), we can calculate the solution atthe wellbore:
pwD pFD0 p
FCD s c6s
ptanh c6sp B-59
Calculation Summary
To summarize, wellbore pressure can be obtained with the follow-ing sequence of calculations:
c1s sg2D k4
k2y1D
s
g4D
rtanh y1D y2D
s
g4D
r
c2s sg1D k3
k1y1D
s
g3D
rtanh y1D y2D
s
g3D
r
c3s k2k1
c1sc2s
stanhx1D x2D
c1s
pc4s coshwD=2 x1D
c2s
p
c3ssinhwD=2 x1Dc2s
p
c5s c2s
p sinhwD=2 x1D
c2s
p c3scoshwD=2 x1D c2sp c4s
c6s sgFD 2
FCD c5s
pwD p
FCD s c6s
ptanh c6sp
As discussed in the body of the paper (the Possible ModelModifications section), modifications to this calculation procedurecan readily be made to account for the gas case, dual-porosityeffects, skin because of convergence of flow lines within the frac-ture, and wellbore storage.
Louis Mattar is with Fekete Associates, recently acquired byIHS. He specializes in well testing and production-data analy-sis, and has authored 70 technical publications. In 2003, hewas an SPE Distinguished Lecturer in well testing. In 2006, Mat-tar received the SPE International Reservoir Description andDynamics Award. In 2012 he was a Distinguished Member ofSPE. He holds MSc and PEng degrees.
Ekaterina Stalgorova is with Fekete Associates, recentlyacquired by IHS. She holds a degree in in mathematics fromMoscowState University andanMScdegree in petroleumengi-neering from the University of Alberta. Stalgorova specializes inanalytical modeling for well testing and production-data anal-ysis. Her research interests also include classical and nonclassi-cal numerical simulation.
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . .
. . . . . . . . .
. . . . . . . . . .
. . . . . . .
. . . . .qF/2 qF/2
w
h
Fig. B-2Flow within the fracture toward the wellbore.
256 August 2013 SPE Reservoir Evaluation & Engineering