Journal of Natural Gas Science and Engineering 9 (2012) 184e195

Contents lists available at SciVerse ScienceDirect

Journal of Natural Gas Science and Engineering

journal homepage: www.elsevier.com/locate/jngse

Unified Fracture Design for very low permeability reservoirs

S. Bhattacharya, M. Nikolaou, M.J. Economides*
Department of Chemical and Biological Engineering, University of Houston, TX, USA

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

Article history: Unified Fracture Design (UFD) is a methodology for the design of optimal hydraulic fracturing treatments
Received 31 May 2012 to maximize well performance. Many successful applications of UFD have been demonstrated for
Received in revised form reservoirs with low to high permeability. However, UFD has not been systematically applied to reservoirs
15 June 2012
with very low (nanodarcy) permeability, such as shales. For such reservoirs, massive hydraulic fracturing
Accepted 16 June 2012
combined with horizontal drilling has enabled the production of significant amounts of hydrocarbons in
Available online 2 August 2012
recent years. Yet a method for the design of optimal hydraulic fracture treatments for these reservoirs is
currently lacking, with present practice relying mainly on rules of thumb and trial and error. This paper
Keywords:
Hydraulic fracturing
fills this gap by developing a design methodology through extension of UFD that addresses two
Nanodarcy permeability reservoirs (shale) important elements: (a) nanodarcy permeability reservoirs, and (b) fractures with highly elongated
Elongated drainage areas drainage areas, that are necessary because of the extremely small reservoir permeability. Simple explicit
Unified Fracture Design functions are developed that can be used for efficient computation of optimal fracture dimensions and
the productivity index for large values of Proppant Number and highly elongated drainage area around
each fracture. A case study is presented to demonstrate how the developed UFD extension can be used in
practice and to illustrate the qualitative differences between standard UFD and the developed extension.
Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction hydraulic fracture treatments for these reservoirs is currently

lacking, with present practice relying mainly on rules of thumb and
Unified Fracture Design (UFD) (Economides et al., 2002) is trial and error. The objective of this paper is to fill this gap by
a methodology for the design of optimal hydraulic fracturing developing a design methodology through extension of UFD that
treatments. Given the quantity and properties of proppant to be addresses two important elements: (a) nanodarcy permeability
used, reservoir permeability, pay-zone thickness, and dimensions reservoirs, and (b) fractures with highly elongated drainage areas.
of the fracture drainage area, UFD efficiently determines optimal In the rest of the paper, we first provide a brief overview of UFD.
values of fracture dimensions (fracture half-length, xf, and fracture Next, we develop an extension of UFD to very low-permeability
width, wf) which maximize the productivity index, JD. An important reservoirs as discussed previously, and subsequently present
concept introduced by UFD is the Proppant Number, Np, which a case study to illustrate how the developed UFD extension can be
combines all of the given variables into a single dimensionless used in practice. Finally we draw conclusions and provide recom-
number that uniquely determines the optimal values of CfD and, by mendations for future work.
extension, xf, wf, and the maximum JD. Successful application of
UFD has been demonstrated for reservoirs with low to high
permeability (Daal and Economides, 2006; Demarchos et al., 2004, 2. UFD background and motivation
2005; Economides et al., 2010; Porcu et al., 2008, 2009). However,
UFD has not been applied to reservoirs with very low (nanodarcy) The UFD methodology is as follows. Given formation perme-
permeability, such as shales. For such reservoirs, massive hydraulic ability (k), pay-zone thickness (h), fracture height (hf), mass of
fracturing combined with horizontal drilling (Fig. 1) has enabled proppant to be injected in the pay zone (Mp), and fracture drainage
the production of significant amounts of hydrocarbons in recent area dimensions (xe,ye), the Proppant Number, Np, is first calculated
years. Yet a method for efficient systematic design of optimal as

2kf Vp
Np ¼ (1)
khxe ye
* Corresponding author.
E-mail addresses: [email protected], [email protected] where kf is fracture permeability, a function of the type of the
(M.J. Economides). proppant used; and Vp is the propped volume inside the pay zone

1875-5100/$ e see front matter Ó 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jngse.2012.06.005
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 185

where the aspect ratio yeD of the fracture drainage area is defined as
ye
yeD ¼ (5)
xe
For each value of CfD, the corresponding value of the dimen-
sionless productivity index, JD, is computed through a series of
calculations (Appendix A) culminating in Eq. (A.13), which provides
a direct boundary-element solution of the equations that govern
single-phase fluid flow through the porous matrix into the fracture
at pseudo-steady state. The largest of the resulting values of JD is the
optimum, JD,max. Associated with JD,max is the optimal CfD value,
CfD,opt, from which the optimal fracture dimensions can be calcu-
lated as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
kf Vf
xf ;opt ¼ (6)
CfD;opt kh
Fig. 1. Section of a horizontal well in a rectangular reservoir of constant height with
vertical rectangular fractures perpendicular to the well. and
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
CfD;opt kVf
wf ;opt ¼ (7)
kf h
which includes the two fracture wings and the void space between
the proppant grains, calculated as where


Mp h=hf Vp
Vp ¼
 (2) Vf ¼ (8)
2
r p 1  fp
is the fracture volume per wing (Economides et al., 2002).
with rp and 4p as the specific gravity and porosity of the proppant While the procedure outlined above is straightforward, the
pack. repetitive computations leading to Eq. (A.13) are cumbersome.
kf wf
Next, the dimensionless fracture conductivity, CfD ¼ ˆ is These computations can be avoided through the use of simple
kxf
computed as correlations (Daal and Economides, 2006), which efficiently
produce JD,max as a function of CfD,opt, depending on Np for a given
Np yeD yeD (see Appendix B for specifics). However, these correlations are
CfD ¼ (3)
Ix2 available only for yeD  0.1 and for Np  100. Yet, in many very low-
permeability reservoirs, yeD can reach values much lower than 0.1,
for several values of the penetration ratio
as a very large number of fractures can be spaced very close to each
2xf other, resulting in a highly elongated drainage area around each
Ix ¼ (4) fracture (Vincent, 2011). Similarly, for very low-permeability
xe

yeD=1.0
10
JD,max
Dimensionless porductivity index, JD

Ix=1 JD,max (plateau) = 1.9

Np=10000
Np=1000
1
Np=100
Np=10
Np=1

0.1
1 10 100 1,000 10,000 100,000 1,000,000 10,000,000
Dimensionless fracture conductivity, CfD

Fig. 2. Dimensionless productivity index analysis for yeD ¼ 1.0 with increasing values of Proppant Number. The coordinates of the squares represent (CfD,opt, JD,max) for a particular Np
and the dashed line corresponds to Ix ¼ 1.
186 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195

Fig. 3. Dimensionless productivity index analysis for yeD ¼ 1.0 and Np  0.1 from Daal and Economides (2006). The coordinates of the squares represent (CfD,opt, JD,max) for
a particular Np.

reservoirs, Np can easily reach values much higher than 100 as 10, 100, 1000, 10,000 produces the results shown in Fig. 2, which is
illustrated in the case study discussed later in this work. For the counterpart of the results produced by Daal and Economides
yeD << 0.1 and Np >> 100 no correlations are available for efficient (2006), reproduced here in Fig. 3. From comparison of these two
implementation of UFD in a fashion similar to that described in figures, the following important differences emerge:
Appendix B. In the next section, we develop such correlations by
analyzing the results of detailed calculations for high values of Np - While CfD,opt z 1.6 for all Np  0.1, as shown in Fig. 3, the
and small values of yeD. optimal value of CfD (shown by the squares in Fig. 3) increases
substantially as Np increases above 1. The reason for this
3. UFD extension behavior of CfD,opt is that when Np increases above 1, the cor-
responding optimal Ix reaches its maximum value of 1 (corre-
We now show how an extension of the analysis performed by sponding to 2xf ¼ xe (Fig. 1) for a fracture spanning the entire
Daal and Economides (2006) results in qualitatively distinct length of the fracture drainage area), hence, by Eq. (3), CfD,opt
outcomes for yeD << 0.1 and Np >> 100. keeps increasing as Np increases.
Application of the procedure outlined in Appendix A for yeD ¼ 1 - While JD,max is an increasing function of Np for Np < 0.1, the
(namely a square drainage area around a single fracture) and Np ¼ 1, corresponding JD,max for Np > 1 eventually reaches a plateau

yeD=0.5
10
JD,max
Dimensionless productivity index, JD

JD,max (plateau) = 3.8

Ix=1
Np=10000
Np=1000
Np=100
1 Np=10

Np=1

0.1
0 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000
Dimensionless fracture conductivity, CfD

Fig. 4. Dimensionless productivity index analysis for yeD ¼ 0.5 with increasing values of Proppant Number. The coordinates of the squares represent (CfD,opt, JD,max) for a particular Np
and the dashed line corresponds to Ix ¼ 1.
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 187

yeD=0.1
100
JD,max

Dimensionless productivity index, JD JD,max (plateau) = 19

10 Np=10000
Np=1000
Ix=1
Np=100

1 Np=10

Np=1

0.1
0 1 10 100 1,000 10,000 100,000 1,000,000 10,000,000
Dimensionless fracture conductivity, CfD

Fig. 5. Dimensionless productivity index analysis for yeD ¼ 0.1 with increasing values of Proppant Number. The coordinates of the squares represent (CfD,opt, JD,max) for a particular Np
and the dashed line corresponds to Ix ¼ 1.

value of 1.9 in Fig. 2. This is because for Np >> 1 the conduc- associated with Np  10,000. Similarly, Fig. 6 and Fig. 7 show results
tivity of the fracture reaches a practically infinite value for for even more elongated fracture drainage areas, namely for
a certain amount of proppant, and consequently any additional yeD ¼ 0.05 and yeD ¼ 0.01, respectively. In these figures, JD,max rea-
amount of proppant allocated into the fracture will not increase ches a plateau at much higher values of Np, which are not shown in
JD,max. the respective figures.
From inspection of Fig. 8, it is clear that JD,max is higher for
Similar patterns emerge for smaller values of yeD. Shown in Fig. 4 fracture drainage areas with more pronounced elongation (small-
are the results for yeD ¼ 0.5, namely for a rectangular fracture eryeD) when Np is high. The situation is reversed for low values of
drainage area. In this case, increase in Np results in increase in Np, for which JD,max is higher for fracture drainage areas with less
CfD,opt, which leads to JD,max (plateau) ¼ 3.8 for CfD,opt  1000, asso- pronounced elongation. The reason for this reversal is related to the
ciated with Np  1000. Shown in Fig. 5 is the pattern emerging for optimal fracture length in each case and the resulting pattern of
yeD ¼ 0.1, namely for a fairly elongated fracture drainage area. A drainage flow. As Fig. 9 indicates, for high Np the optimal fracture
similar value of JD,max (plateau) ¼ 19 appears for CfD,opt1000, length is such that the optimal fracture reaches the boundary of its

yeD=0.05
100

JD,max (plateau) = 39 JD,max

Dimensiobless productivity index, JD

Np=10000
10
Np=1000

Ix=1 Np=100
1
Np=10

Np=1
0.1

0.01
0 0 1 10 100 1,000 10,000 100,000 1,000,000
Dimensionless fracture conductivity, CfD

Fig. 6. Dimensionless productivity index analysis for yeD ¼ 0.05 with increasing values of Proppant Number. The coordinates of the squares represent (CfD,opt, JD,max) for a particular
Np and the dashed line corresponds to Ix ¼ 1.
188 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195

yeD=0.01
100
JD,max (plateau) = 204 JD,max
Dimesnionless productivity index, JD

10 Np=10000

Np=1000

1
Ix=1 N =100
p

0.1 Np=10

Np=1

0.01
0 0 1 10 100 1,000 10,000 100,000 1,000,000
Dimensionless fracture conductivity, CfD

Fig. 7. Dimensionless productivity index analysis for yeD ¼ 0.01 with increasing values of Proppant Number. The coordinates of the squares represent (CfD,opt, JD,max) for a particular
Np and the dashed line corresponds to Ix ¼ 1.

drainage area (Ix,opt ¼ 1) regardless of yeD. In this case, the more 4. Generalized correlations for high Proppant Numbers
elongated the drainage area, the more efficiently it can be drained
by fractures penetrating the drainage area, as shown in Fig. 10, After analyzing all the plots in the previous section, correlations
right. However, for Np < 5000, the optimal fracture length is which are applicable for low permeability reservoirs are developed
a function of yeD, and the optimal fracture does not reach the in this section.
boundary of its drainage area (Ix,opt<1) as shown in Fig. 9. In fact, for For all cases of yeD, JD,max increases with increase in Np and then
low Np, the more elongated the drainage area, the less efficiently it reaches a plateau value. This implies that once JD,max reaches
can be drained, since the corresponding fracture length is small, a plateau value JD,max (plateau) for a certain amount of proppant, use
thus making it difficult to drain the domain near the boundary of of a larger amount of proppant would not result in higher
the drainage area, as shown in Fig. 10, left. Thus, many fractures, in productivity; rather, increase in Np (resulting from increase in Mp)
e.g., shales, also mean almost fully penetrating fractures for effi- would only increase expenses. Moreover, the value of Np at which
cient drainage of the reservoir volume. the plateau value is attained for JD,max increases with decrease in

1000

yeD=0.01
Increasing
Maximun dimensionless productivity index, JD,max

100 elongation in
drainage area
yeD=0.05

yeD=0.1
10

yeD=0.5

yeD=1.0
1

0.1

0.01
1 10 100 1,000 10,000 100,000 1,000,000 10,000,000 100,000,000
Proppant Number, Np

Fig. 8. Maximum dimensionless productivity index analysis for different aspect ratios with increasing Proppant Number.
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 189

1.1

Optimum penetration Ratio, Ix, opt 0.9

yeD=0.05
yeD=0.01

yeD=0.1
yeD=0.5
0.8
yeD=1.0

0.7

0.6

0.5

0.4
1 10 100 1,000 10,000
Proppant Number, Np

Fig. 9. Optimum penetration ratio for different aspect ratios with increasing Proppant Number.

yeD. In fact, plotting JD,max (plateau) against yeD in a logelog plot An empirical equation for CfD,opt as a function of yeD and Np was
(Fig. 11), reveals a simple dependence of JD,max (plateau) on yeD as obtained as

2 ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
v


JD;maxðplateauÞ ¼ (9) u
yeD u 1 þ 0:03N 2 1 þ 1:3N2 y1:2 
u p p eD
CfD;opt ¼ 1:6t   (10)
To find the dependence of CfD,opt and JD,max on yeD and Np in 1 þ 0:1Np2 y3:2 eD
terms of empirical equations, function approximation by quasi-
polynomial ratios was considered.
Comparison between actual and empirical values of CfD,opt (ob-
tained from Eq. (10)) is shown in Fig. 12.
Similarly, an empirical equation for JD,max as a function of yeD and
Fracture Np was obtained as
Undrained
region
0:6y0:7 1:6 2:7 2
eD þ 14yeD Np þ 28yeD Np
JD;max ¼ (11)
1 þ 35y1:75
eD Np þ 14yeD Np
3:7 2

Low yeD
Comparison between actual and empirical values of JD,max (ob-
tained from Eq. (11)) is shown in Fig. 13. For very high values of
Drained CfD,opt, the value of JD,max computed by Eq. (11) becomes equal to the
region
value of JD,max (plateau) suggested by Eq. (9).

1000

Actual
Jd_actual
Empirical
JD_emp

100
Plateau JD,max

High yeD

10

1
Low N p High N p 0.01 0.1 1
Aspect Ratio, yeD

Fig. 10. Optimal fracture lengths and drainage patterns corresponding to low Np (left) Fig. 11. Plateau value of maximum productivity index obtained at different aspect
and high Np (right) for drainage areas of different aspect ratios yeD. ratios.
190 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195

100000000
yeD=0.1

Optimum dimensionless fracture conductivity, CfD, opt

10000000

1000000 Actual yeD=0.5

— Empirical
100000
yeD=0.01
10000
yeD=0.05
1000

100
yeD=0.01
10

0.1

0.01
1 10 100 1000 10000 100000 1000000 10000000 100000000

Proppant Number, Np

Fig. 12. Comparison between actual values (dots) and empirical values (continuous lines) of optimum dimensionless fracture conductivity.

1000
Maximum dimensionless productivity index, JD,max

yeD=0.01
Actual
100
— Empirical
yeD=0.05
yeD=0.1
10
yeD=0.5
yeD=1.0
1

0.1

0.01
1 10 100 1000 10000 100000 1000000 10000000 100000000

Proppant Number, Np

Fig. 13. Comparison between actual values (dots) and empirical values (continuous lines) of maximum dimensionless productivity index.

5. Case study
Table 1
In this section, we demonstrate how the UFD extension Reservoir and proppant parameters.
developed above can be applied to horizontal wells with multiple
Porosity of proppant pack, 4p 0.35
transverse fractures in very low-permeability reservoirs. The Specific gravity of proppant, rp 2.65
objective is to compare two different wells in two different Pay-zone thickness, h (ft) 100
reservoirs: Well #1 is a horizontal well with a single transverse Fracture height, hf (ft) 100
fracture in a tight reservoir of permeability k ¼ 0.1 md. Well #2 Dimensions of square reservoir 2000  2000
drainage area, xR, yR (ft)
includes instances of a horizontal well with multiple transverse
Horizontal well length, L (ft) 2000
fractures in a reservoir of permeability k ¼ 1, 10, or 100 nd. To Pressure difference, Dp (psi) 3000
compare the two wells, the number of fractures nfe in Well #2 Formation volume factor, B 1.1
will be calculated that results in the same production as the (res bbl/STB)
production of Well #1. The parameters for this study are shown Oil viscosity, m (cp) 2
Proppant permeability, kf (md) 200,000
in Table 1. Note that in comparing Well #1 and Well # 2, two Mass of proppant 100,000;
cases are distinguished for the mass of proppant, Mp, used: In (per fracture or well), Mp lbs) 500,000;
Case 1, the same mass per fracture is allocated for both Well #1 1,000,000
and Well #2. In Case 2, the same total mass is allocated for both Tight oil permeability, k (md) 0.1
Shale oil permeability, k (nd) 1, 10, 100
Well #1 and Well #2.
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 191

Table 2
Comparison between tight and shale reservoirs for 100,000 lbs of proppant per fracture (Case-1).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 9.304 1 0.95 10.4 1.66 3.45E-02 1.571 151.7
Shale oil
k ¼ 1 nd Eq. (13) 281 2.61E þ 08 3.56E-03 0.87 1.23E þ 06 561.7 3.17E-07 561.6 152.4
k ¼ 10 nd Eq. (13) 89 8.28E þ 06 1.12E-02 0.89 1.16E þ 05 177.7 3.26E-06 177.6 152.6
k ¼ 100 nd Eq. (13) 29 2.70E þ 05 3.45E-02 0.92 1.10E þ 04 57.62 3.35E-05 57.51 161.1
k ¼ 100 nd Eq. (14) 90 8.37E þ 05 1.11E-02 0.89 1.16E þ 04 176.6 2.31E-05 175.9 1529

Table 3
Comparison between tight and shale reservoirs for 500,000 lbs of proppant per fracture (Case-1).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 46.52 1 1.00 46.74 1.918 7.27E-03 1.892 182.7
Shale oil
k ¼ 1 nd Eq. (13) 308 1.43E þ 09 3.25E-03 0.87 6.19E þ 06 615.9 6.32E-08 615.9 183.2
k ¼ 10 nd Eq. (13) 98 4.56E þ 07 1.02E-02 0.89 5.85E þ 05 195.9 6.50E-07 195.9 185.4
k ¼ 100 nd Eq. (13) 31 1.44E þ 06 3.23E-02 0.92 55,190 61.91 6.69E-06 61.89 185.3
k ¼ 100 nd Eq. (14) 98 4.56E þ 06 1.02E-02 0.89 58,460 195.2 4.60E-06 195 1846

Table 4
Comparison between tight and shale reservoirs for 1,000,000 lbs of proppant per fracture (Case-1).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 93.04 1 1.00 93.09 1.958 3.65E-03 1.944 187.8
Shale oil
k ¼ 1 nd Eq. (13) 312 2.90E þ 09 3.21E-03 0.87 1.24E þ 07 624 3.16E-08 623.9 188
k ¼ 10 nd Eq. (13) 99 9.21E þ 07 1.01E-02 0.89 1.17E þ 06 198 3.25E-07 197.9 189.3
k ¼ 100 nd Eq. (13) 32 2.98E þ 06 3.13E-02 0.92 110,600 63.95 3.34E-06 63.94 197.6
k ¼ 100 nd Eq. (14) 99 9.21E þ 06 1.01E-02 0.89 117,000 197.6 2.30E-06 197.5 1888

To compute nfe, given the parameter values in Table 1, proceed as e Compute JDTH from Eq. (C.2).
follows: f Compute the production rate for Well #1,q1,pss, using Eq. (C.1).
2. For each of the three permeabilities considered for Well #2 and
1. For Well #1 and for nf ¼ 1 (single fracture in the reservoir for multiple values of nf ¼ 1, 2, 3,. (multiple fractures in the
drainage area): reservoir drainage area):
a Compute JD,opt from Eq. (11) for a value of Np calculated (for a Compute JD,opt from Eq. (11) for a value of Np calculated using
given Mp) using Eq. (1) and yeD ¼ 1. Eq. (2) and for the corresponding yeD, where
b Compute CfD,opt using Eq. (10) for the above values of Np and
yeD. ye yR =nf
yeD ¼ ¼ (12)
c Compute the optimum fracture dimensions xf,opt and wf,opt xe xR
from Eq. (6) and Eq. (7), respectively.
d Compute the choke skin sc from Eq. (C.3).
b Compute CfD,opt using Eq. (10) for the above values of Np and
yeD.
c Compute the optimum fracture dimensions xf,opt and wf,opt
350
from Eq. (6) and Eq. (7), respectively.
300 k=1 nd
d Compute the choke skin sc from Eq. (C.3).
e Compute JDTH from Eq. (C.2).
No. of equivalent fractures, nfe

250 f Compute the production rate for Well #2, q2,pss, using Eq. (C.1).
3. Determine the number of equivalent fractures nfe for Well #2 as
200 the smallest value of nf in the above step (2) for which

150 q2;pss  q1;pss (13)

k=10 nd
100 Results for Case-1 corresponding to Mp of 100,000 lb per frac-
ture are summarized in Table 2, which shows that 281, 89, or 29
50 k=100 nd
fractures in Well #2 for corresponding permeability 1, 10, or 100 nd,
0
respectively, would produce the same as a single fracture in Well
0 500,000 1,000,000 #1, namely q1,pss ¼ 152 STB/d. Similar numbers corresponding to Mp
Mass of proppant per fracture of 500,000 and 1,000,000 lb per fracture are shown in Table 3 and 4
respectively. All of the results in Table 2, Table 3 and Table 4 are
Fig. 14. Number of equivalent fractures for different permeability reservoirs that
produce same as single fracture in a tight (0.1 md) reservoir with different mass of summarized in Fig. 14, which shows that increasing the proppant
proppant per fracture (Case-1). mass increases the number of equivalent fractures, nfe from 100,000
192 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195

Table 5
Comparison between tight and shale reservoirs for 100,000 lbs of proppant per well (Case-2).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 9.3 1 0.95 10.4 1.66 0.03 1.57 151.67
Shale oil
k ¼ 1 nd Eq. (13) 315 9.30E þ 05 3.18E-03 0.87 3935 525.1 9.95E-05 499 151.8
k ¼ 10 nd Eq. (13) 101 9.30E þ 04 9.90E-03 0.89 1159 166 3.28E-04 157.4 153.6
k ¼ 100 nd Eq. (13) 32 9304 3.13E-02 0.92 345.5 52.09 1.07E-03 49.34 152.5
k ¼ 100 nd Eq. (14) q2,pss  10  q1,pss not achievable

Table 6
Comparison between tight and shale reservoirs for 500,000 lbs of proppant per well (Case-2).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 46.52 1 1.00 46.74 1.918 7.27E-03 1.892 182.7
Shale oil
k ¼ 1 nd Eq (13) 316 4.65E þ 06 3.17E-03 0.87 1.96E þ 04 607.6 2.00E-05 600.3 183.2
k ¼ 10 nd Eq (13) 100 4.65E þ 05 1.00E-02 0.89 5852 191.8 6.50E-05 189.5 183
k ¼ 100 nd Eq (13) 32 4.65E þ 04 3.13E-02 0.92 1727 61.2 2.14E-04 60.41 186.7
k ¼ 100 nd Eq. (14) 136 4.65E þ 04 7.35E-03 0.88 436.9 153.2 6.20E-04 139.9 1837

Table 7
Comparison between tight and shale reservoirs for 1,000,000 lbs of proppant per well (Case-2).

Criterion nfe Np yeD Ix,opt CfD,opt JD,max sc JDTH qpss (STB/d)

Tight reservoir
k ¼ 0.1 md 1 93.04 1 1.00 93.09 1.958 3.65E-03 1.944 187.8
Shale oil
k ¼ 1 nd Eq (13) 316 9.30E þ 06 3.17E-03 0.87 3.92E þ 04 619.6 9.98E-06 615.7 187.9
k ¼ 10 nd Eq (13) 100 9.30E þ 05 1.00E-02 0.89 1.17E þ 04 195.8 3.25E-05 194.6 187.9
k ¼ 100 nd Eq (13) 32 9.30E þ 04 3.13E-02 0.92 3455 62.57 1.07E-04 62.15 192.1
k ¼ 100 nd Eq. (14) 114 9.30E þ 04 8.77E-03 0.89 1033 178.9 2.61E-04 170.9 1881

to 500,000 but does not affect nfe much after further increase. The In this scenario it is found that nfe ¼ 90, 98, and 99 for Well #2
same tables also indicate that for the calculated value of nfe, the having permeability of k ¼ 100 nd corresponding to mass of
resulting value of Np for Well #2 is very high (more than 106). proppant of 100,000; 500,00; and 1,000,000 lb per fracture
Moreover, the choke skin is negligible for Well #2, making JDTH respectively.
effectively equal to JD. However, Case 1 looks impractical because Results for Case-2 corresponding to Mp of 100,000 lb per well,
the total amount of proppant for Well #1 would be i.e. mass of proppant per fracture equal to Mp/nf, are summarized in
281  1,000,000 ¼ 281,000,000 lb, which is unrealistic used in the Table 5. This table shows that 315, 101, or 32 fractures in Well #2 for
comparative calculation. To find the optimal amount of proppant corresponding permeability 1, 10, or 100 nd, respectively, would
per well an economic optimization would be needed, which will be produce the same as a single fracture in Well #1. Similar numbers
part of future work. corresponding to Mp of 500,000 and 1,000,000 lb per well are
Finally, one more scenario is evaluated for Case 1, in which Eq. shown in Tables 6 and 7, respectively. All of the results in Table 5,
(13) is replaced by Table 6 and Table 7 are summarized in Fig. 15, which shows that the
number of equivalent fractures nfe remains the same with
q2;pss  10  q1;pss (14) increasing mass of proppant. For the scenario corresponding to Eq.
(14), production q2,pss  10  q1,pss is not achievable by Well #2
with 100,000 lb of proppant mass per well, because very low
350
amount proppant is available per fracture as nf increases. This leads
k=1 nd to insufficient fracture conductivity and fracture half-length. For
300
proppant mass of 500,000 lb and 1,000,000, nfe is found to be 136
No. of equivalent fractures, nfe

250 and 114 respectively, in 100 nd permeability reservoir.

200
6. Conclusions
150

k=10 nd
In this work, we have extended the UFD approach to the case of
100 elongated fracture drainage areas and very high Proppant Numbers,
common in very low-permeability reservoirs. The developed
50 k=100 nd
correlations would be useful for finding optimum fracture dimen-
0 sions, particularly, when the aspect ratio per fracture is very low.
0 500,000 1,000,000 The following conclusions can be drawn from the work presented.
Mass of proppant per well

Fig. 15. Number of equivalent fractures for different permeability reservoirs that
1. In contrast to the case Np  0.1, for which CfD,opt ¼ 1.6 regardless
produce same as single fracture in a tight (0.1 md) reservoir with different mass of of the value of Np, the value of CfD,opt increases and reaches
proppant per well (Case-2). values much higher than 1.6 when Np > 1. The reason behind
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 193

this behavior is that for very high Np, the optimal fracture Pseudosteady-state drawdown at any observation point i due to
length spans the entire length of the fracture drainage area, nw wells in terms of influence function is given by:
leading to increase in CfD,opt with increase in Np.
2. For low values of yeD, the optimal productivity JD,max reaches
a plateau value JD,max(plateau) with increasing values of Np, i.e. a1 mB Xnw h i
Dpo;i ¼ p  po;i ¼ qi a xoD;i ; yoD;i ; xwD;j; ywD;j ; yeD
after certain Proppant Number, Np,plateau, there is no increase in 2pkh
j¼1
JD,max for increase in Np. This arises from the fact that for
Np >> 1 the conductivity of the fracture reaches a practically a1 mB X
nw
¼ q a½o ; w  (A.1)
infinite value. 2pkh j ¼ 1 i 1 1
3. The value of JD,max (plateau) is found to be inversely proportional
to yeD yielding a simple correlation. Therefore, the pressure drop between two observation points
4. For high values of Np (greater than 1000), JD,max is higher for due to flow from reservoir to fracture is obtained by subtracting
fracture drainage areas with more pronounced elongation (A.15) for two different points. For example, for the first two
(smaller yeD). The reverse trend is observed for low values of Np observation points, we get
(lower than 100), for which JD,max is higher for fracture drainage
areas with less pronounced elongation. The reason for this a1 mB
reversal is related to the optimal fracture length and the DpR;1/2 ¼ ½q ða½o1 ; w1   a½o2 ; w1 Þ þ . þ qnw ða½o1 ; wnw 
2pkh 1
resulting pattern of drainage flow.
 a½o2 ; wnw Þ ðA:2Þ
5. The optimal fracture half-length xf,opt is found to be a function
of yeD for Np < 5000. For higher values of Np, xf,opt is indepen-
Darcy’s law inside the fracture states that
dent of yeD and the fracture always reaches the boundary of the
fracture drainage area. kf Af vp
6. General correlations are obtained for maximum dimensionless q ¼  (A.3)
productivity index, JD,max and optimal dimensionless fracture
m vx
conductivity, CfD,opt as functions of aspect ratio, yeD and Prop- Thus, the pressure drop at two observation points inside frac-
pant Number, Np which are applicable for Np > 1. ture is given by
7. The number of equivalent fractures nfe in a shale reservoir that
would produce equally to a single fracture in a tight reservoir
2a1 mB
depends highly on the permeability of the reservoir. The lower Dpf ;1/2 ¼ ½q ðx  x01 Þ þ . þ qnw ðxo2  x01 Þ (A.4)
kf hw 2 o2
the permeability, the more fractures are required. Also, prop-
pant mass, Mp has a significant effect in determining nfe.
Eqs. (A.16) and (A.18) represent the pressure drop at the first two
observation points. Therefore, by equating them we get
In future work, the methodology developed above for physical
optimization of hydraulic fracturing designs would be combined
with economic concepts to develop economically optimal frac- q1 ða½o1 ; w1   a½o2 ; w1 Þ þ . þ qnw ða½o1 ; wnw   a½o2 ; wnw Þ
turing designs for nanodarcy permeability formations. 4pk
 ½q ðx  x01 Þ þ . þ qnw ðxo2  x01 Þ ¼ 0 ðA:5Þ
kf w 2 o2
Appendix A. Direct boundary element method & influence
function All the terms in the above equation are made dimensionless by
using the notation
The description below is obtained from the previous applica-
tions (Daal and Economides, 2006; Romero et al., 2002; Valkó and qBm
qDi ¼
 (A.6)
Economides, 1998; Valkó et al., 2000). Calculations are done on 2pkh p  pwf
a quarter of the fracture and drainage area zone due to the
symmetric nature of the fracture and reservoir system. The origin is
xoi
at the center of a fracture. All lengths of this section are made xDoi ¼ (A.7)
xe =2
dimensionless with respect to the larger rectangular dimension, xe.
The quarter fracture is divided into nw equal sections length wise,
xf
and one observation point and source point is placed inside each Ix ¼ (A.8)
section as shown in Fig. A1. The flow rate in each section is obtained xe =2
by solving a system of nw linear equations. nw  1 equations are for i ¼ 1,.,nw yielding
obtained by equating pressure drop between two observation
points in the fracture.
qD1 ða½o1 ;w1 a½o2 ;w1 Þ þ.þqDnw ða½o1 ;wnw a½o2 ;wnw Þ
nw segments
!
4pkxf xe =2
 ½qD2 ðxDo2 xD01 Þþ.þqDnw ðxDo2 xD01 Þ ¼ 0
kf w xf

xe / 2
qD1 ða½o1 ;w1 a½o2 ;w1 Þ þ.þqDnw ða½o1 ;wnw a½o2 ;wnw Þ
xf
4p
 ½q ðx xD01 Þþ.þqDnw ðxDo2 xD01 Þ ¼ 0 (A.9)
CfD Ix D2 Do2
(0,0) ye / 2
Similarly, nw  2 equations can be obtained by equating the
Fig. A1. Quarter section of the fracture and drainage area. pressure drop inside the reservoir and fracture between the 1st
194 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195

observation point and nw  2 other observation points. This gives ST ¼ S1 þ S2 þ S3

a total of nw  1 equations. The last equation is obtained by setting
P
N t
m
the drawdown in Eq. (A.15) to be unity. In matrix notation S1 ¼ 2 cos½mpxD cos½mpxwD 
m¼1 m
! t n o
! S2 ¼  N ln ½1  cosðpðxD þ xwD ÞÞ2 þ½sinðpðxD þ xwD ÞÞ2
Aq ¼ d (A.10)
2
where tN n o
 ln ½1  cosðpðxD  xwD ÞÞ2 þ½sinðpðxD  xwD ÞÞ2
2 3 2 3 2
a11 a12 . a1 nw qD1 XN
1
6 a21 a22 / a2 7 6 qD2 7 S3 ¼ 2tN cos½mpxD cos½mpxwD  (A.17)
6
A ¼ 4 nw 7; q ¼ 6 7
« « « « 5 4 « 5; m¼1
m
anw 1 anw 2 / anw nw qDnw where N is chosen such that
2 3 jaðf ðNÞÞ  aðf ðN  1ÞÞjhtolerance (A.18)
1
607
d ¼ 6
4«5
7 (A.11)
Appendix B. Existing UFD correlations
0
with elements of the matrix A are given as Two sets of correlations are defined (Daal and Economides,
2006) that can be used to find JD,max and CfD,opt for fixed values of
  aspect ratios yeD shown in Table B1.
a1j ðj ¼ 1;.;nwÞ ¼ a o1 ; wj
Case-1: Np  0.1
ai1 jði ¼ 2;.;nwÞ ¼ a½o1 ; w1   a½oi ; w1 
1
JD;max ¼  
   4p

0:99  0:5ln Np CA
(B.1)
aij ði;j ¼ 1;.;nwÞ ¼ a o1 ; wj  a oi ; wj  xDoðminði;jÞÞ  xDo1
CfD Ix
where CA is the Dietz shape factor (Dietz, 1965) dependent on
(A.12) aspect ratio yeD and
After solving the above system of linear equations for q, the
dimensionless productivity index JD for the whole fracture is CfD; opt ¼ 1:6 (B.2)
obtained as four times the summation of dimensionless flow Case-2: Np > 0.1
rate from each of nw divisions of the quarter section of the
fracture. 1
JD;max ¼   (B.3)
0:63  0:5ln Np þ Fopt
X
nw
JD ¼ 4 qDi (A.13) The F-function is given by
i¼1

A þ Buopt þ Cu2opt þ Du3opt

Fopt ¼ (B.4)
Influence function a0 þ b0 uopt þ c0 u2opt

Influence functions (Ozkan, 1988) used in Eq. (A.15) are given by: Table B1
Coefficients for Fopt

a xD ; yD ; xwD; ywD ; yeD ¼ a1 ½maxðxD ; xwD Þ; maxðyD ; ywD Þ; yeD 1 0.7 0.5 0.25 0.2 0.1
minðxD ; xwD Þ; minðyD ; ywD Þ; yeD  A 17.2 17.4 21.4 38.3 35 30.6
8 0 B 65.5 55.5 54.3 46 59 89.6
 < a xD ; yD
; xwD; ywD ; yeD if ðyeD
 < 1Þ C 52.5 53.3 56.3 71.1 70 70.2
1
a xD ; yD ; xwD; ywD ; yeD ¼ 1 D 16.9 16.9 16.9 15.84 16.3 17.8
: a0 xD ; yD ; xwD; ywD ;
yeD For any yeD
(A.14) a0 10
b0 36
! c0 33
 1 yD y2 þ y2
a0 xD ; yD ; xwD; ywD ; yeD ¼ 2pyeD  þ D 2 wD þ ST
3 yeD 2yeD where uopt ¼ ln(CfD,opt) and coefficients are defined in Table. The
optimum CfD is given by
X
N
tm
ST ¼ 2 cosðmpxD ÞcosðmpxwD Þ (A.15)
m¼1
m

where 100yeD  CfD;0:1  

CfD; opt ¼  Np  0:1 þ CfD;0:1 (B.5)
100
cos h½mpðyeD  jyD  ywD jÞ þ cos hðmp½yeD  ðyD  ywD ÞÞ where
tm ¼
sin hðmpyeD Þ
(A.16)

The infinite summation ST is replaced by finite series (Valkó 4:5yeD þ 0:25 if 0:1  yeD  0:25
CfD;0:1 ¼ (B.6)
et al., 2000; Ozkan, 1988) as 1:6 if 0:25 < yeD  1
 S. Bhattacharya et al. / Journal of Natural Gas Science and Engineering 9 (2012) 184e195 195

Appendix C. Pseudosteady state production rate in horizontal Valkó, P.P., Economides, M.J., 1998. Heavy Crude Production from Shallow Forma-
tions: Long Horizontal Wells Versus Horizontal Fractures. SPE 50421 in 1998
well with multiple transverse fractures
SPE International Conference on Horizontal Well Technology in Calgary,
Alberta, Canada.
The production rate in the pseudo-state regime from nf fractures Valkó, P.P., Doublet, L.E., Blasingame, T.A., 2000. Development and application of the
in a horizontal well (Porcu et al., 2009) is multi-well productivity index (MPI). SPE Journal 5 (1), 21e31.
Vincent, M.C., 2011. Optimizing transverse fractures in liquid-rich formations. In:
SPE Annual Technical Conference and Exhibition. Society of Petroleum Engi-
khDp neers, Denver, Colorado, USA.
qpss ¼ nf J (C.1)
141:2Bm DTH Wei, Y., Economides, M.J., 2005. Transverse Hydraulic Fractures from a Hori-
zontal Well. SPE 94671 in 2005 SPE Annual Technical Conference and
In the above equation, JDTH is the dimensionless productivity Exhibition.
index of a transverse fracture intersecting a horizontal well (Wei
and Economides, 2005) given by Nomenclature

A: Drainage area, L2, ft2

1
JDTH ¼ (C.2) B: Formation volume factor, dimensionless, bbl/STB
1 CA: Dietz shape factor, dimensionless
þ sc CfD: Dimensionless fracture conductivity, dimensionless
JD
CfD,opt: Optimum dimensionless fracture conductivity, dimensionless
where sc is the choke skin (Mukherjee and Economides, 1991) f: Pseudo-skin factor, dimensionless
Fopt: Optimum F-factor
calculated as h: Formation thickness, L, ft

  Ix: Penetration ratio, dimensionless
kh h p Ix,opt: Optimum penetration ratio, dimensionless
sc ¼ ln  (C.3) JD: Dimensionless productivity index, dimensionless
wf kf 2rw 2
JDTH: Transversely fractured horizontal well productivity index, dimensionless
JD,max: Maximum dimensionless productivity index, dimensionless
with propped fracture width wf and JD is the dimensionless JD,max (plateau): Plateau value of maximum dimensionless productivity index,
productivity index for a single fracture. dimensionless
k: Formation permeability, L2, md
kf: Fracture permeability, L2, md
References L: Horizontal well length, L, ft
Mp: Mass of proppant, M, lb
Daal, J.A., Economides, M.J., 2006. Optimization of hydraulically fractured wells in nf: Number of fractures in a horizontal well, dimensionless
irregularly shaped drainage areas. In: SPE 98047 Presented at 2006 SPE Inter- nfe: Number of equivalent fractures, dimensionless
national Symposium and Exhibition on Formation Damage Control. Lafayette, LA. nw: Number of divisions of fracture for DBEM, dimensionless
Demarchos, A.S., Chomatas, A.S., Economides, M.J., Mach, J.M., Wolcott, D.S., 2004. Np: Proppant number, dimensionless
Pushing the limits in hydraulic fracture design. In: SPE International Sympo- Np,sat: Proppant Number when dimensionless productivity index achieves plateau,
sium and Exhibition on Formation Damage Control. Society of Petroleum dimensionless
Engineers, Lafayette, Louisiana. p: Average reservoir pressure, M/LT2, psi
Demarchos, A.S., Economides, M.J., Diyashev, I.R., Svaykin, V.A., 2005. Analysis of the pwf: Pressure at wellbore, M/LT2, psi
performance of hydraulic fracturing treatments and quantum design q: Flow rate, L3 T1, STB/d
improvements. In: SPE European Formation Damage Conference. Society of qpss: Flow rate at pseudosteady state, L3 T1, STB/d
Petroleum Engineers, Sheveningen, The Netherlands. re: Radius of drainage area, L, ft
Dietz, D.N.,1965. Determination of Average Reservoir Pressure from Build up Surveys. JPT. rw: Wellbore radius, L, ft
0 : Effective wellbore radius, L, ft
rw
Economides, M.J., Oligney, R.E., Valko, P., 2002. Unified Fracture Design: Bridging
the Gap between Theory and Practice. Orsa Press, Alvin TX. sc: Choke skin factor, dimensionless
Economides, M.J., Porcu, M.M., Yang, M., Martin, A.N., 2010. Fracturing horizontal sf: Fracture skin, dimensionless
transverse, horizontal longitudinal and vertical wells: criteria for decision. In: Vf: Fracture volume per wing, L3, ft3
SPE 137328 in Canadian Unconventional Resources and International Petroleum Vp: Proppant volume in the pay zone per fracture, L3, ft3
Conference. Society of Petroleum Engineers, Calgary, Alberta, Canada. wf: Fracture width, L, inch (ft)
Mukherjee, H., Economides, M.J., 1991. A parametric comparison of horizontal and wf,opt: Optimum fracture width, L, inch (ft)
vertical well performance. SPE Formation Evaluation. xe: Length of single fracture drainage area, L, ft
Ozkan, E., 1988. Performance of Horizontal Wells. PhD Dissertation, University of xf: Fracture half length, L, ft
Houston, Oklahama. xf,opt: Optimum fracture half length, L, ft
Porcu, M.M., Economides, M.J., Holditch, S.A., 2008. Economic and physical opti- xR: Length of horizontal well drainage area, L, ft
mization of hydraulic fracturing. In: SPE International Symposium and Exhi- xw: x-coordinate of any division of a fracture, L, ft
bition on Formation Damage Control. Society of Petroleum Engineers, Lafayette, ye: Breadth of single fracture drainage area, L, ft
Louisiana, USA. yR: Breadth of horizontal well drainage area, L, ft
Porcu, M.M., Wang, X., Economides, M.J., 2009. Delineation of application: physical yw: y-coordinate of any division of a fracture, L, ft
and economic optimization of fractured gas wells. In: SPE Production and Oper- yeD: Aspect ratio, dimensionless
ations Symposium. Society of Petroleum Engineers, Oklahoma City, Oklahoma. g: Euler’s constant, 0.577215664
Romero, D.J., Valkó, P.P., Economides, M.J., 2002. The Optimization of the Produc- m: Fluid viscosity, M/LT, cp
tivity Index and the Fracture Geometry of a Stimulated Well with Fracture Face rp: Specific gravity of proppant, dimensionless
and Choke Skins. Paper SPE 73758. 4p: Porosity of proppant, dimensionless