You are on page 1of 14

A New Method for Evaluating the

Productivity Index of Nonlinear Flows


Eugenio Aulisa and Akif Ibragimov, Texas Tech University, and Jay R. Walton, Texas A&M University

Summary Ibragimov et al. (2005) rigorously proved this fact. Traditional


This paper addresses the effects of nonlinearity of flows on the methods for PI evaluation use semianalytical solution of transient
value of the productivity index (PI) of the well. Experimental data problem and, as a rule, impose serious restrictions on the geometry
show that, during the dynamic process of hydrocarbon recovery, of the reservoir.
the PI stabilizes to some constant value, which, in general, is a Ibragimov et al. (2004, 2005) presented an alternative method
nonlinear function of both the pressure drawdown and the pro- for evaluating both the PSS and BD PIs on the basis of the solution
duction rate. of the corresponding time-independent, steady-state boundary value
Linear Darcy flow is well understood, and excellent approxi- problems (BVPs) (see Appendix A). From the reservoir-engineer-
mate formulas are available for the PI in various well/reservoir ing point of view, these new relations between the solutions of
geometries. To handle the more realistic nonlinear situation, the transient and steady-state problems enable direct use of powerful
current practice is to solve the nonlinear problem multiple times finite element/volume machinery. In addition, variational interpre-
for different values of production rate and then add ad-hoc correc- tation provides an effective tool for PI estimation without solving
tive parameters in the linear formulas to reproduce the nonlinear the corresponding problem (Aulisa et al. 2007). These methods
nature of the flow. become particularly useful in evaluating the PI for a well of arbitrary
In this paper, we propose a rigorous framework to measure the configuration in a general 3D flow case because the aforementioned
PI of a well for nonlinear Forchheimer flows. Our approach, based steady-state problems are solved in a bounded domain modeling the
on recent progress in the modeling of transient Forchheimer flows, actual reservoir shape. The results were presented at the SPE Annual
uses both analytical and numerical techniques. It provides, for a wide Technical Conference and Exhibition in Houston in 2004, where it
class of reservoir geometries, an accurate relation between the PI for was suggested to test the developed methods for non-Darcy flows.
nonlinear Forchheimer flows and the PI for linear Darcy flows. For these flows, a semianalytical technique based on the superposi-
The proposed method of building look-up tables and analyti- tion principle is not applicable. The current work reports the progress
cal formulas serves as an effective tool for fast PI evaluation in we have made in modeling one-phase nonlinear flows in porous
nonlinear cases. media and indicates directions for future research.
The purpose of this paper is to emphasize the effect of the
Introduction nonlinear nature of the flow filtration on the value of the PI. We
The PI is one of the more basic characteristics of well performance introduce a generalized nonlinear Darcy equation with permeability
not requiring assumptions about the equations of the flow motion tensor dependent on the gradient of the pressure, which enables
and the state of the fluid. The concept of PI expresses the following: modeling of “two-terms” and “power” Forchheimer flows within
Once the well production is, in some sense, stabilized, then the ratio one mathematical framework. This framework does not depend on
between the production rate and the pressure drawdown (difference the type of nonlinearity in the generalized Darcy equation; however,
between the reservoir average pressure and the well average pres- the computational results and analytical solutions are different for
sure) is practically independent from the production history or even these two cases. In this paper, numerical and analytical studies are
from the operating conditions (Muskat 1937; Dietz 1965; Dake presented primarily for the two-terms Forchheimer law. In addition,
1978; Raghavan 1993; Larsen 2001). The higher the value of the PI some analytical results for power law are reported in the Power
is, the better the performance of the well in the reservoir is. Forchheimer Law section. Possible extensions of the model of non-
We are particularly interested in the asymptotic (late-time) linear flows for other boundary conditions in the general case are
value of the PI. For an isolated reservoir with no-flow condition discussed in the Conditions on the Boundaries section.
on the outer boundary and constant production rate, stabilization of Background
the PI means that the pressure drawdown becomes time invariant;
this flow regime is called pseudosteady state (PSS). In the case of Darcy’s law describes the following common observation in reser-
constant wellbore pressure, both the production and reservoir pres- voir engineering: The higher the production rate Q is, the higher
sure change in time, but the PI asymptotically stabilizes to a con- the pressure drawdown P is. The observed data shows that the
stant value, leading to the flow regime called boundary-dominated ratio between these two quantities stabilizes in time. This value is,
(BD) (Dietz 1965; Dake 1978; Raghavan 1993; Larsen 2001). in general, a nonlinear function of both P and Q. Forchheimer
In case of Darcy flow in a constant-thickness reservoir, the introduced two different ways to model nonlinearity in observed
most popular way to evaluate the PSS PI is based on the Dietz data: two-terms law, AQ + BQ2 = P, and power law, CQn = P
formula (see Appendix A). To compute the BD PI, one can use (1 ≤ n ≤ 2) (Fig. 1).
shape factors as defined by Helmy and Wattenbarger (1998). In Today, it is commonly accepted that the inertial terms cause
practice, for simple polygonal domains with large enough dimen- the deviation from the linear Darcy law. It is important to men-
sionless radii, the same shape factors are used in the evaluation of tion that the nature of these inertial forces is not fully understood.
both PSS and BD PIs. However, the BD PI of a well, in general, As pointed out by Bear (1972), “most of the experiments indicate
is different from the PSS PI. Helmy and Wattenbarger (1998) that the actual turbulence occurs at a Reynolds number at least one
observed from numerical simulation of time-dependant problem order of magnitude higher than the Reynolds value at which the
that the PSS PI is always greater than or equal to the BD PI, and deviation from the Darcy law is observed.” Detailed discussion of
these types of nonlinear approximation for the inertial terms and
of the conditions when those approximations are applicable can be
found in work by Muskat (1937), Bear (1972), Whitaker (1996),
Copyright © 2009 Society of Petroleum Engineers
Li and Engler (2001), and Tavera et al. (2006).
This paper (SPE 108984) was accepted for presentation at the SPE Annual Technical In a number of publications on mathematical and numerical mod-
Conference and Exhibition, Anaheim, California, 11–14 November 2007, and revised
for publication. Original manuscript received for review 9 July 2007. Revised manuscript eling of the multidimensional nonlinear Forchheimer flow (Wheeler
received for review 14 July 2008. Paper peer approved 29 July 2008. and Peszyńska 2002; Ewing et al. 1999, Douglas et al. 1993;

December 2009 SPE Journal 693


⎛ k1,1 k1,2 k1,3 ⎞
1⎜
k2 ,1 k2 ,2 k2 ,3 ⎟ be a dimensionless, symmetrical, and
Q
Let K =
k0 ⎜⎜ ⎟
2 ⎝ k3,1 k3,2 , k3,3 ⎟⎠
positive definite permeability tensor. Here, k0 is a normalization
factor characterizing the pivotal direction of the filtration. The
parameter k0 can be taken equal to some average value of the per-
meability, for example, k0 = 3 k1,1k2 ,2 k3,3 or k0 = (k1,1 + k2,2 + k3,3)/3.
1 Coefficients of the matrix are not homogeneous and can depend
3 on spatial Ivariables.
I I
( )
Let Ab , b = ∑ ai , j bi b j be a bilinear form, where b = (b1 , b2 , b3 )
i, j
is a generic vector and A = (ai,j) is a matrix. Let  and  be the
viscosity and the density of the fluid,  = /k, and  =  , where
 is the Forchheimer factor (Dake 1978).
Different ways exist to account for the anisotropy of the non-
linear flow (Dmitriev et al. 2005; Whitaker 1996; Wheeler and
Peszyńska 2002). In this article, we introduce a convenient frame-
ΔP work by constituting the two following equations of motion (Aulisa
et al. 2007):
I
(K I I I
)
Fig. 1—Relation between pressure drawdown and production −1
rate: (1) two-terms Forchheimer, (2) power Forchheimer, and v + v , v v = − K ∇p . . . . . . . . . . . . . . . . . . . . . . . . (1)
(3) Darcy.
for the two-terms law and

cn (K −1 I I
v, v ) n −1 I
v = − K ∇p . . . . . . . . . . . . . . . . . . . . . . . . . . (2)
Balhoff et al. 2009), the continuity equation and Forchheimer/
for the power
I law.
Darcy momentum equations are treated as a coupled system of
Here, v is the velocity of filtration; p is the reservoir pressure,
partial differential equations (PDE). We suggest an alternative
1 ≤ n ≤ 2; and c is a constant depending on parameter n and is
approach, specifically, to constitute the specific nonlinear Darcy
equal to  when n = 1. ∇p = ( px1 , px2 , px3 ) is the pressure gradi-
equation to reduce the coupled system into one PDE for pressure
ent, where the subindices in (x1, x2, x3) are used to indicate the
function only (Aulisa et al. 2006, 2007, 2009; Douglas et al. 1993).
Euclidian coordinates.
This enables application of fundamental methods of parabolic and
Both equations share important features and can be replaced by
elliptic PDE for a rigorous study of the properties of the nonlinear
the generalized nonlinear Darcy equation. Let
flow and for numerical calculation of hydrodynamic parameters of
the recovery process. In a recently published paper (Auriault et al.
 (∇p) = ( K ∇p, ∇p ) =
1
2007), a similar idea is used for the implementation of homogeni- ∑ ki , j pxi px j . . . . . . . . . . . . . . . . . . . (3)
k i, j
zation technique.
For the so-called “1D” flows, the PDE equation reduces to an We have shown in Aulisa et al. (2006, 2007) that Eq. 1 is
ordinary differential equation (ODE), and, therefore, most of the equivalent to equation
analysis can be performed analytically. However, for 1D flows, I
a more effective approach exists for calculating hydrodynamic v = − M ( (∇p)) K ∇p , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (4)
characteristics on the basis of the method of material balance
(Bear 1972; Maksimov et al. 1993). In this case, the Forchheimer where
equation can be derived in a nondeformed streamtube with a vary-
2
ing cross section, and, then, hydrodynamic characteristics of the M( ) = . . . . . . . . . . . . . . . . . . . . . . . . . . . (5)
fluids filtrations can be obtained explicitly from specific formulas.  +  + 4  2

Detailed discussion of this approach for radial flow of gases, sub-


jected to the two-terms law with different boundary conditions, is We have also shown that Eq. 2 can be reduced to
presented by Civan (2000). I
In this paper, we primarily apply the PDE approach to model v = − N ( (∇p)) K ∇p , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
nonlinear flow in a reservoir with arbitrary geometry. To imple-
ment the PDE framework, we introduce the generalized nonlinear where
Darcy equation to model the power and the two-terms laws by
( )
( 1− n ) / n

assuming the coefficients of the permeability matrix to be func- N ( ) = c −1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)


tions also of the pressure gradient. Using this model, we obtained
a compact and easy-to-apply formula for the evaluation of the PI Clearly, Eq. 4 converges to the classical, anisotropic Darcy
similar to the one in the linear case. equation as →0. The same conclusion can be obtained by setting
n = 1 in Eq. 7.
The Generalized Darcy Law These arguments lead to the introduction of the generalized
Different approaches exist for deriving the momentum equation nonlinear Darcy equation:
for the non-Darcy phenomena (Forchheimer 1901; Bear 1972; Li I
and Engler 2001). The equation of motion can be obtained from v = − G [ (∇p) ] K ∇p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (8)
the more general hydrodynamic equation (Payne and Straughan
1999) or by using capillary models to describe the fluid filtration Here, the nonlinear term G depends on the gradient of pressure
through the porous media (Whitaker 1996; Dmitriev et al. 2005; and is defined as G[ (∇p)] = M [ (∇p)] in case of two-terms law,
Li and Engler 2001). In this paper, we implement a more intuitive and G[ (∇p)] = N [ (∇p)] in case of power law.
approach based on the constitutive equation; specifically, we build
two different vector equations for the power and the two-terms Mathematical Model for the PI of the Well
Forchheimer laws. In the theoretical part of the paper, we consider We assume the fluid to be slightly compressible, (p) = (p),
the 3D case in its generality. where the constant characterizes the compressibility of the fluid.

694 December 2009 SPE Journal


Using standard arguments about slightly compressible liquids in wQ ,G
= 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13)
the equation of continuity and accounting for the generalized w

Darcy equation of motion, we arrive at the following nonlinear


diffusive parabolic equation for pressure: ∂wQ ,G
I = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (14)
∂n
e
∂p
= div [ G ( (∇p) K ∇p ] . . . . . . . . . . . . . . . . . . . . . . . . . . (9)
∂t Here, V is the volume of the drainage area. Hereafter, we will
refer to Eqs. 12 through 14 as the auxiliary BVP. Note that
Note that Eq. 9 depends on the type of nonlinearity in G[ (∇p)], Q
but the forthcoming construction is formally the same for both p( x , t ) = B − t + wQ ,G ( x ) is the solution of Eq. 9, where the constant
types of Forchheimer flows. V
B is uniquely defined by the value of the initial reserves.
The auxiliary BVP (Eqs. 12 through 14) is time independent
Definitions. Eq. 9 characterizes the fluid (oil or gas) flow in the (steady-state); therefore, the calculation of the PSS PI in Eq. 11
reservoir generated by different regimes of well exploitation. Sev- has the advantage of reduced computer time if compared to the
eral boundary conditions (see Raghavan 1993; Kumar 1977; Civan original transient problem.
2000; Ibragimov et al. 2004, 2005) can be used to simulate the Eq. 11 gives the value of the PSS PI (i.e., it is explicitly assumed
corresponding regimes of production. In our intended application, that the PSS is reached from the Ibeginning of production at time
the exterior boundary of the reservoir is considered impermeable. t = 0). Note that the value JG ( p, v , t ) is dimensional.
The boundary condition on the well, the nonflux condition on the
exterior boundary, and the initial pressure distribution form the Uniqueness. If the pressure is uniform along the wellbore, the
initial BVP (IBVP) for Eq. 9. At this time, we will not specify any value of the PSS PI JQ,G is unique and Eq. 11 is its analytical
particular conditions on the well. representation. The rigorous proof of this property can be found
Let U be the drainage region bounded by the well
w and by the in Ibragimov et al. (2004) for linear Darcy case and in Aulisa
exterior boundary
e. Let p(x,0) = p0(x) be the given initial pressure et al. (2007) for two-terms Forchheimer case.
distribution in the reservoir U. Let PU(t) be the average pressure in In many applications, it is not reasonable to expect that the PSS
the reservoir U and Pw be the average pressure on the well
w. assumption holds that, initially, at time t = 0, the reservoir pressure
Definition of
I the PI. Let the pressure function p(x,t) and vector is distributed as wQ,G(x). In this case, PI is time dependent, and it is
velocity field v be subjected to the generalized Darcy equation, important from both theoretical and application points of view to
Eq. 8. Then, the PI of the well in the isolated drainage area U is compare it with PSS PI. Numerical experiments and field data sug-
defined as the ratio gest that the value of the PSS PI, as provided by Eq. 11, is the only
attainable value of the PI. We were able to prove that, if pressure
I Q
JG ( p, v , t ) = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (10) is uniformly distributed along the well, then time-dependent PI
PU − Pw converges exponentially to PSS PI for linear Darcy and two-terms
Forchheimer laws (Ibragimov et al. 2004; Aulisa et al. 2008).
Here, Q is total flux over the well boundary
w (i.e., Q is the
production rate). The Two-Terms Forchheimer Law for
Definition of the PSS Regime. Let the production rate Q be
Homogeneous Isotropic Reservoir
time independent; the regime of production is PSS if the corre-
sponding pressure drawdown [the difference between the pressure For a homogeneous isotropic reservoir, the generalized Darcy equa-
average in the reservoir PU(t) and the pressure average on the well tion takes a simpler form. To distinguish this case, we denote
PW(t)] is constant.
2
Another important regime of production, the BD regime, and G (∇p) = f ( ∇p ) = . . . . . . . . . . . . . . . . (15)
the associated PI for the linear Darcy case were studied in detail  +  2 + 4 ∇p
by Ibragimov et al. (2005, 2004). In the case of BD regime, the
bottomhole pressure is fixed and the rate of production is vanish-
ing in time. Because the focus of this study is the time-invariant Here, ∇p = px1 2 + px2 2 + px3 2 . Then, the generalized Darcy equa-
(late-time) PI for the nonlinear Forchheimer flows, the effects of tion for two-terms law takes the following form:
the nonlinearity become negligible over time. Therefore, in this I
article, we study the PSS regime. v = − f ( ∇p ) ∇p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (16)
A brief review of some other regimes of filtration and the associ-
ated boundary conditions on the external boundary and on the well Correspondingly, the auxiliary BVP becomes
surface are given in the Conditions on the Boundaries section.

Formula for PSS PI. For short vertical and slanted wells, the
(
div f (∇wQ , )∇wQ , = − ) Q
V
, . . . . . . . . . . . . . . . . . . . . . . . (17)
wellbore pressure can be assumed to be uniformly distributed on
the wellbore at any time. If a horizontal well has infinite conduc- wQ , = 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (18)

w
tivity, then the pressure is uniformly distributed on the wellbore
as well. Note that this constraint fails for very long slanted wells. ∂wQ ,
Although the method is applicable in case of nonconstant pressure I = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (19)
on the well, in this paper, we will assume that at each moment the ∂n
e

pressure on
w is uniformly distributed.
If the well produces with constant rate Q, the value of the PSS On the basis of Eq. 15, the PSS PI formula, Eq. 11, for convenience
PI is given by will be rewritten in the form

I I QV
JG ( p, v , t ) = JQ ,G =
QV
. . . . . . . . . . . . . . . . . . . . . . . . (11) JQ , f ( p, v , t ) = JQ , = . . . . . . . . . . . . . . . . . . . . . . (20)
∫w Q ,G dV
U
∫ wQ , dV
U

Once again, to compute the PSS PI, the auxiliary BVP (Eqs.
Here, wQ,G is the solution of the following steady-state BVP:
17 through 19) must be solved first. For most of the cases, it can
be done only numerically, but, for the axial symmetric case, the PI
(
div G (∇wQ ,G ) K ∇wq ,G = − ) Q
V
, . . . . . . . . . . . . . . . . . . . . . . (12) can be evaluated explicitly using an analytical formula.

December 2009 SPE Journal 695


From Eq. 25 follows a useful comparison formula between
productivity indices for two-terms Forchheimer and Darcy cases:

JQ ,0 ε I2
= 1+ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (26)
JQ ,  I1

rw Let dimensionless parameter sk be

ε I2
sk = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (27)
re  I1
Parameter sk can be interpreted as skin factor, which indicates
deviation of the Forchheimer PI from the Darcy PI. Assuming rw <<
re, we obtain the following approximate formula for skin factor:

ε re 2
sk ≈ . . . . . . . . . . . . . . . . . . . . . . . . . (28)
 rw [ ln(re / rw ) − 3 / 4 ]
Combining Eqs. 25 and 28 results in the approximate equation
for skin factor:

Q 1
sk ≈ . . . . . . . . . . . . . . . . . . . (29)
Fig. 2—2D scheme of fully penetrated vertical well in a cylindri-  2 rw H [ ln(re / rw ) − 3 / 4 ]
cal reservoir.
Note that Eq. 27 is exact. Table 5 presents a comparison
between two values of dimensionless PIs: the first by solving the
PSS PI for Circular Drainage Area and Skin Factor. This sec- auxiliary problem (Eqs. 17 through 19) and applying Eq. 20 and
tion is dedicated to the classical reservoir-engineering problem the second by using the value of the dimensionless Darcy PI and
of fluid filtration in a cylindrical reservoir with a fully penetrated applying the formula
vertical well positioned in its center (Fig. 2).
To find the analytical formula for PSS PI in case of axial-sym- JQ , = J Darcy / (1 + sk ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (30)
metric two-terms Forchheimer flow, we first derive an alternative
formula for the PI in terms of the radial velocity v,q(r) generated Here, JQ, , and J Darcy are dimensionless productivity indices for
by PSS pressure distribution p(r,t). two-terms Forchheimer and Darcy laws respectively.
Henceforth, Eq. 30 will be referred to as the PI with skin. The
QV Q2 skin factor in Eq. 29 has a clear hydrodynamic interpretation,
JQ , = = . . . . . . . (21)
∫w
re
dV which will be discussed in the Skin Factor—General Case subsec-
U
Q ,
2 H ∫ ( ( vQ , )2 +  ( vQ , )3 )rdr tion dedicated to comparison between two productivity indices.
rw In Table 1, the comparison between the analytical and the
numerical results for the radial two-terms Forchheimer law for
Next, the auxiliary BVP (Eqs. 17 through 19) is rewritten in
some particular values of the parameters is presented.
polar coordinates for the radial component of the velocity vector
The geometrical parameters are
• The radius of the reservoir re = 100 m.
r −1
d
dr
(
rvQ , = −) Q
, r < r < re ,
H (rw 2 − re 2 ) w
• The reservoir thickness H = 10 m.
• The well radius rw = 0.3 m.
vQ , (re ) = 0 . . . . . . . . . . . . (22) The fixed hydrodynamic parameters are
•  = 1.01 × 1010 (Pa·s)/m2.
• = 0.658 × 10−9 Pa−1.
The function
The varying hydrodynamic parameters are
Q(re 2 − r 2 ) •  = 2.4318 × 1011, 2.4318 × 1012, and 2.4318 × 1013 Pa·s2/m3.
vQ , (r ) = . . . . . . . . . . . . . . . . . . . . . . . . . . (23) • Q/H = 10−4, 10−3, and 10−2 m2/s.
2 Hr (re 2 − rw 2 ) Table 1 reports the numerical values of the well PI vs. the
varying hydrodynamic parameters  and Q. Column I contains the
solves the BVP (Eq. 22). This expression can be used to derive the
values of the PI obtained using the numerical solution of the BVP
PI in terms of velocity. Specifically, the radial velocity is plugged
(Eqs. 17 and 18) in Eq. 20. The column titled “Analytic” contains
into the alternative formula for JQ, in Eq. 21, resulting in
the PI values obtained with the analytical Eq. 24. For all  and Q
2 H (re 2 − rw 2 )2 values, the difference is negligible.
JQ , = . . . . . . . . . . . . . . . . . . . . . . . . . . . . (24) In Fig. 3, parameters re, , H, and are the same,  =
 I1 + ε I 2 2.4318 × 1011 Pa·s2/m3, and Q/H = 10−2 m2/s. The graph in Fig. 3
emphasizes the effect of the value of the well radius on the ratio
Here,
z = JQ , / J Darcy ; as the radius of the well decreases, the effect of
Q nonlinearity increases. This reflects the nature of Forchheimer
ε= , flows: The higher the gradient of the pressure, the bigger the
2 H (re 2 − rw 2 ) deviation from Darcy.
re
Comparison between two productivities indices for other values
of the parameters are presented in Table 5.
I1 = ∫ (re 2 − r 2 )2 r −1dr ,
In the radial case, the velocity fields for the Darcy and the two-
rw
terms Forchheimer flows are the same and independent from the
parameter . Therefore, if  and Q are known, then the values of
and
the PI for the two flows can be obtained from each other simply
re using the algebraic relation. In fact, this was used to derive Eq.
I 2 = ∫ (re 2 − r 2 )3 r −2 dr . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (25) 30. On the other hand, the pressure distribution depends on the
rw

696 December 2009 SPE Journal


0.98
z
0.97
0.96
0.95
0.94
0.93
0.92
0.91
0.9
rw
0.89
0.05 0.075 0.1 0.125 0.15 0.175 0.2 0.225 0.25 0.275 0.3

Fig. 3—Dependence of the ratio z = JQ , / JDarcy on rw (in meters).

parameters  and Q. For the PSS regime, the steady-state part


of the pressure distribution is the solution of the BVP (Eqs. 17 ( )
div′ f ( ∇′w ′ ) ∇′w ′ = −1 , on U ' , . . . . . . . . . . . . . . . . . . (32)
through 19):
w ′
′ = 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (33)
w
Q(r 2 − rw 2 − 2re 2 ln(r / rw )2 )
wQ , (r ) = −
H 4 (rw 2 − re 2 ) ∂w␦′
I = 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (34)
Q (r rw + r rw + rrw − 6rrwre + 3re )
2 3 2 2 3 2 4 ∂n
e′
+ (r − rw ) .
H 2 12 2rrw (rw 2 − re 2 )2
Here, div, ∇′ , U,
′w and
′e denote the divergence, the gradient,
Difference in the pressure distributions can be very big for quite a the drainage volume, the well boundary, and the exterior boundary
large range of parameters, as shown in Fig. 4. in dimensionless coordinates. In Eq. 32,
The explicit representation for pressure function for two-terms
2 Q
law is not new and is similar to the corresponding formula for f ( ∇ ′w ′ ) = , with = . . . . . . . . . . (35)
pseudopressure for gases in Civan (2000) (see also Chapter 2, 1 + 1 + 4 ∇′w′  L2
Section 4.2 in Maksimov et al. 1993).
Then, the PSS PI can be rewritten as
Properties of the PSS PI. The results in Table 1 clearly indicate
that, for all  and Q with the constant value of the product ·Q, L
the PSS PI is the same. It appears to be a common property for JQ , = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (36)
the two-terms Forchheimer flows. Assuming the most general 3D  ∫ w ′ dV ′
U′
flow in the reservoir and considering the following dimensionless
variables and parameters 
For a given reservoir and given parameter  = , Eq. 36 states that
k
 the PSS PI depends only on the parameter c0 = Q.
x = Lx ′, wQ , ( x ) = w ′( x ′), with L3 = V , . . . . . . . . . . . . . (31) Property 1. For a given constant value c0 = Q, the PSS PI depends
L
only on  = /k, and on the geometry of the drainage area.
the auxiliary problem (Eqs. 17 through 19) can be rewritten in its This result was originally obtained through numerical simula-
dimensionless form as tion and has a clear physical interpretation: In the Forchheimer
equation, the coefficient  and the rate of production Q have
the same effect on the value of the well PSS PI. It can also be
B applied for the solution of the inverse problem—estimating the
hydrodynamic characteristics of the reservoir through analysis of
2500
the production data.
Pressure Radial Distribution

To highlight another important feature of the PSS PI for two-


terms Forchheimer law, consider the PSS PI formula for the cor-
2000 responding Darcy case (Ibragimov et al. 2005):

VQ
1500 J Darcy = , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (37)
∫ wQ ,0 dV
U

1000 A
where wQ,0 is the solution of the corresponding linear auxiliary
BVP (Eqs. A-7 through A-9). Note that, in terms of  and Q, the
500 Darcy PSS PI is a particular case of the Forchheimer PI, specifi-
cally JDarcy = JQ,0 = J1,0.
Property 1 does not remove the existing gap between the for-
2000 4000 6000 8000 10000 mula for the evaluation of the PSS PI for the Darcy Eq. 37 and the
Radius two-terms Forchheimer (Eq. 20) cases. The PI for the Darcy case
does not depend on the value of the production rate Q and can be
Fig. 4—Pressure distribution in the radial reservoir: (A) Darcy computed by solving the auxiliary problem (Eqs. A-7 through A-9)
case and (B) two-terms Forchheimer case. only once. In the nonlinear case, Eq. 20 requires the solution of the

December 2009 SPE Journal 697


auxiliary problem for each given production rate Q. This is unfor- This relation enables us to rewrite the formula for the PSS PI in
tunate, but it reflects the nonlinear nature of Forchheimer flows. terms of the velocity vector field:
In the next section, it will be shown that under certain con-
straints on the flow, the PSS PI for the two-terms law can be QV Q2
JQ , = = I 2 I 3 . . . . . . . . . . . . (40)
computed by solving only one auxiliary problem.
∫w
U
Q , dV ∫ ( vQ , +  vQ , ) dV
U
PSS PI for Uniform Flows. The aim of this section is to obtain I
a more convenient approximate formula for the PI not requiring Assume that the velocity vector field vQ, ( x , t ) uniformly
multipleIsolving of the auxiliary problem. depends on the rate of the production in the sense
Let vQ, ( x ,) be the velocity vector field generated by the PSS
Q I Q I
pressure distribution p( x , t ) = B − t + wQ , ( x ) and defined by vQ , ( x , t ) = vQ , ( x , t ) . . . . . . . . . . . . . . . . . . . . . . . . . . . (41)
U Q1 1
the equation
The above condition obviously holds for 1D flows. Under the con-
I I I straint (Eq. 41), the following approximate formula can be derived
−∇wQ , =  vQ , +  vQ , vQ , . . . . . . . . . . . . . . . . . . . . . . . . (38) for PSS PI in terms of the velocity field for the Darcy case:

From Eq. 38 follows 1


J ,Q = I 2 I 3 . . . . . . . . . . . . . . . . . . . . . (42)
I I 2 I 3
−∇wQ , ⋅ vQ , =  vQ , +  vQ , . . . . . . . . . . . . . . . . . . . . (39) ∫ ( v1,0 + Q v1,0 ) dV
U

The left-hand side in Eq. 39 is the scalar product between two vectors. Now, to calculate the PI for all possible values of Q, only one
Recalling Eq. 16 and that wQ, solves the auxiliary problem (Eqs. 17 auxiliary problem (Eqs. A-7 through A-9) for linear PDE needs
Q I 2 I 3 to be solved. In Tables 1 through 4, the comparison between the
through 19), it can be proved that ∫ wQ , dV = ∫ ( vQ , + vQ , ) dV . true value (without constraint Eq. 41) of the PSS PI for a given
VU U

TABLE1—PRODUCTIVITY INDEX FOR THE CYLINDRICAL RESERVOIR


AND FULLY PENETRATED VERTICAL WELL

Q/H = 10–4 Q/H = 10–3 Q/H = 10–2

Analytic I Analytic I Analytic I


11
β = 2.4318×10 0.1976 0.1976 0.1971 0.1971 0.1928 0.1928
β = 2.4318×1012 0.1971 0.1971 0.1928 0.1928 0.1582 0.1582
β = 2.4318×1013 0.1928 0.1928 0.1582 0.1582 0.05659 0.05659

TABLE 2—PRODUCTIVITY INDEX FOR THE RECTANGULAR RESERVOIR AND FULLY PENETRATED VERTICAL WELL

Q/H = 10–4 Q/H = 10–3 Q/H = 10–2

T I II T I II T I II
11
β = 2.4318×10 0.1480 0.1483 0.1483 0.1477 0.1480 0.1480 0.1450 0.1452 0.1452
β = 2.4318×1012 0.1477 0.1480 0.1480 0.1450 0.1452 0.1452 0.1224 0.1225 0.1224
β = 2.4318×1013 0.1450 0.1452 0.1452 0.1224 0.1225 0.1224 0.0481 0.0480 0.0476

TABLE 3—PRODUCTIVITY INDEX FOR THE RECTANGULAR RESERVOIR


AND FULLY PENETRATED FRACTURE

Q/H = 10–4 Q/H = 10–3 Q/H = 10–2

θ = 0° I II I II I II
11
β = 2.4318×10 0.9537 0.9537 0.9535 0.9535 0.9516 0.9516
β = 2.4318×1012 0.9535 0.9535 0.9516 0.9516 0.9327 0.9327
β = 2.4318×1013 0.9516 0.9516 0.9327 0.9327 0.7779 0.7778
θ = 45° I II I II I II
11
β = 2.4318×10 1.2106 1.2106 1.2104 1.2104 1.2078 1.2078
β = 2.4318×1012 1.2104 1.2104 1.2078 1.2078 1.1826 1.1824
β = 2.4318×1013 1.2078 1.2078 1.1826 1.1824 0.9856 0.9768
θ = 90° I II I II I II
11
β = 2.4318×10 1.4940 1.4940 1.4937 1.4937 1.4900 1.4900
β = 2.4318×1012 1.4937 1.4937 1.4900 1.4900 1.4550 1.4542
β = 2.4318×1013 1.4900 1.4900 1.4550 1.4542 1.1970 1.1723

698 December 2009 SPE Journal


TABLE 4—PRODUCTIVITY INDEX FOR THE CYLINDRICAL RESERVOIR
WITH VERTICAL DEVIATED WELL

θ = 0° θ = 15° θ = 30°

βQ I II Analytic I II I II

0 0.1988 0.1988 0.1976 0.2015 0.2015 0.2017 0.2017


10
8×2.4318×10 0.1941 0.1941 0.1928 0.1967 0.1967 0.1969 0.1969
11
8×2.4318×10 0.0593 0.0585 0.0600 0.0590 0.0697 0.0581
θ = 45° θ = 60° θ = 75°

βQ I II I II I II

0 0.1974 0.1974 0.1878 0.1878 0.1693 0.1693


10
8×2.4318×10 0.1927 0.1926 0.1834 0.1833 0.1655 0.1654
11
8×2.4318×10 0.0590 0.0571 0.0573 0.0550 0.0531 0.0506

rate Q and the approximate value from Eq. 42 is presented and heimer coefficient, and permeability increase and the viscosity of
proves to be very small. the fluid decreases, the deviation from Darcy becomes proportion-
ally larger.
Skin Factor—General Case. From both applied and theoretical Geometry of the reservoir/well system contributes to the value
points of view, it is important to estimate the deviation of the PI for of the by means of the maximum value of the modulus of the
nonlinear flows from the Darcy PI. In the current paper, we primar- velocity on the well, which is inversely proportional to the radius
ily concentrate on theoretical analysis; numerical experiments were and length of the well. For example, in the radial case, this value
performed for a number of particular geometries. is equal to 1/2 rwH—the inverse of the surface area of the well.
It can be shown (see proof in Appendix B) that Unfortunately, the parameter does not measure deviation
of the two productivity indices accurately. Obtaining a precise
I
QV  max v1,0 formula for skin in a general case is a challenging problem that
J Darcy − JQ , ≤ U
J Darcy . . . . . . . . . . . . . . . . . . (43) we plan to explore in future research. At this time, as a practical
 solution, for fast PI evaluation, we suggest a heuristic approximate
engineering formula with skin:
In a homogeneous reservoir, the maximum of the modulus of
the velocity reaches its value on the well surface
w; therefore,
Jskin (Q,  ) ≈ J Darcy / (1 + sk ) . . . . . . . . . . . . . . . . . . . . . . . . . (45)
from Eq. 43, one can conclude
I Here,
QV  max v1,0

w
J Darcy − JQ , ≤ J Darcy = J Darcy . . . . . . . . . . . . (44) Q
 sk = J , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (46)

w Darcy
Parameter is nondimensional and can be regarded as the measure
of the deviation of the Forchheimer PI from the Darcy PI. This
w is the surface area of the well and J Darcy is the nondimensional
parameter depends on the hydrodynamic characteristics of the flow Darcy PI.
and the geometry of the reservoir. The effect of the hydrodynamic Tables 5 through 9 present comparisons between dimension-
parameters on the skin is explicit: As the rate of production, Forch- less PIs and are organized as follows:

TABLE 5—COMPARISON BETWEEN PIs: ACTUAL VS. FORMULA WITH SKIN FACTOR
FOR FULLY PENETRATED VERTICAL WELL IN CYLINDRICAL RESERVOIR

Q1 ,β1 Q2 ,β1 Q3 ,β1 Q1 ,β2 Q2 ,β2 Q3 ,β2 Q1,β3 Q2 ,β3 Q3 ,β3

A1 0.19766 0.19766 0.19766 0.19766 0.19766 0.19766 0.19766 0.19766 0.19766


A2 0.00025 0.00252 0.02525 0.00252 0.02525 0.25248 0.02525 0.25248 2.52481
A3 0.19761 0.19716 0.19279 0.19716 0.19279 0.15782 0.19279 0.15782 0.05608
A4 0.19760 0.19710 0.19280 0.19710 0.19280 0.15820 0.19280 0.15820 0.05659
A5 0.00006 0.00033 0.00003 0.00033 0.00003 0.00243 0.00003 0.00243 0.00906

TABLE 6—COMPARISON BETWEEN PIs: ACTUAL VS. FORMULA WITH SKIN FACTOR
FOR FULLY PENETRATED VERTICAL WELL IN THE RECTANGULAR RESERVOIR

Q1 ,β1 Q2 ,β1 Q3 ,β1 Q1 ,β2 Q2 ,β2 Q3 ,β2 Q1,β3 Q2 ,β3 Q3 ,β3

A1 0.14835 0.14835 0.14835 0.14835 0.14835 0.14835 0.14835 0.14835 0.14835


A2 0.00019 0.00189 0.01895 0.00189 0.01895 0.18950 0.01895 0.18950 1.89497
A3 0.14832 0.14807 0.14559 0.14807 0.14559 0.12472 0.14559 0.12472 0.05125
A4 0.14830 0.14800 0.14520 0.14800 0.14520 0.12250 0.14520 0.12250 0.04810
A5 0.00017 0.00049 0.00271 0.00049 0.00271 0.01812 0.00271 0.01812 0.06539

December 2009 SPE Journal 699


TABLE 7—COMPARISON BETWEEN PIs: ACTUAL VS. FORMULA WITH SKIN FACTOR
FOR FULLY PENETRATED FRACTURE IN THE RECTANGULAR RESERVOIR
β1 Q1,0° Q2,0°1 Q3,0° Q1,45° Q2,45° Q3,45° Q1 ,90° Q2,90° Q3,90°

A1 0.96119 0.96119 0.96119 1.23444 1.23444 1.23444 1.54197 1.54197 1.54197


A2 0.00003 0.00030 0.00300 3.8E-05 0.00038 0.00385 4.8E-05 0.00048 0.00481
A3 0.96116 0.96090 0.95831 1.23439 1.23396 1.22971 1.54190 1.54123 1.53459
A4 0.95370 0.95350 0.95160 1.21060 1.21040 1.20780 1.49400 1.49370 1.49000
A5 0.00776 0.00770 0.00701 0.01927 0.01910 0.01781 0.03107 0.03084 0.02906

TABLE 8—COMPARISON BETWEEN PIs: ACTUAL VS. FORMULA WITH SKIN FACTOR
FOR FULLY PENETRATED FRACTURE IN THE RECTANGULAR RESERVOIR
β2 Q1,0° Q2,0°1 Q3,0° Q1,45° Q2,45° Q3,45° Q1,90° Q2,90° Q3,90°

A1 0.96119 0.96119 0.96119 1.23444 1.23444 1.23444 1.54197 1.54197 1.54197


A2 0.00030 0.00300 0.02998 0.00038 0.00385 0.03850 0.00048 0.00481 0.04809
A3 0.96090 0.95831 0.93321 1.23396 1.22971 1.18868 1.54123 1.53459 1.47122
A4 0.95350 0.95160 0.93270 1.21040 1.20780 1.18260 1.49370 1.49000 1.45500
A5 0.00770 0.00701 0.00055 0.01910 0.01781 0.00511 0.03084 0.02906 0.01103

TABLE 9—COMPARISON BETWEEN PIs: ACTUAL VS. FORMULA WITH SKIN FACTOR
FOR FULLY PENETRATED FRACTURE IN THE RECTANGULAR RESERVOIR
β3 Q1,0° Q2,0°1 Q3,0° Q1,45° Q2,45° Q3,45° Q1,90° Q2,90° Q3,90°

A1 0.96119 0.96119 0.96119 1.23444 1.23444 1.23444 1.54197 1.54197 1.54197


A2 0.00300 0.02998 0.29978 0.00385 0.03850 0.38500 0.00481 0.04809 0.48091
A3 0.95831 0.93321 0.73950 1.22971 1.18868 0.89129 1.53459 1.47122 1.04123
A4 0.95150 0.93270 0.77790 1.20780 1.18260 0.98560 1.49000 1.45500 1.19700
A5 0.00711 0.00055 0.05190 0.01781 0.00511 0.10580 0.02906 0.01103 0.14960

• Darcy PI, A1 = J Darcy based on Eq. A-5 and the solution of the rectangular reservoir for extremely high values of both parameters
auxiliary problem (Eqs. A-6 through A-8). Q and . The graph in Fig. 5 emphasizes the effect of the orien-
Q tation of the fracture in a rectangular reservoir on the two ratios:
• Skin factor A2= sk = J . Jskin (Q,  ) / J Darcy (Curve a), and JQ, / J Darcy (Curve b) The model of

w Darcy the fracture and corresponding parameters are given in the 2D Fracture
• Two-terms Forchheimer PI with skin factor A3 = Jskin (Q,  ) = in a Rectangular Reservoir subsection. For this case,
w is defined as
J Darcy / (1+ sk ) . the area of fracture surface 2H(Lf + df). In Fig. 5, skin factor sk depends
• Darcy PI, A4 = JQ, based on Eq. 20 and the solution of the on the angle  between fracture and vertical axis, and it increases with
auxiliary problem (Eqs. 17 through 19). angle sk (0°) = 0.029998 , sk (45°) = 0.0385 , sk (90°) = 0.04809 .
• Relative error A5 = JQ , − Jskin (Q,  ) / JQ , . In Fig. 5,  = 2.43178 × 1013 Pa·s2/m3 and Q/H = 10−1 m2/s.
In most cases, the relative error does not exceed 3%. The only All other parameters, , H, , Lf, and df, are the same as in the 2D
high deviation of 14% was observed in the PI of the fracture in a Fracture in a Rectangular Reservoir subsection.

0.975
z1(z2)

0.97

0.965
a

0.96

0.955
b
0.95

0.945

0.94
0.002998 0.0385 0.04809 sk(θ)

Fig. 5—Dependence of the ratios z1 = Jskin ( Q,  ) / JDarcy and z2 = JQ , / JDarcy on sk ( ) .

700 December 2009 SPE Journal


The Power Forchheimer Law Q(re 2 − r 2 )
In the case of a homogeneous, isotropic reservoir, the generalized vQn (r ) = . . . . . . . . . . . . . . . . . . . . . . . . . . . (56)
2 Hr (re 2 − rw 2 )
Darcy equation takes a simpler form. To distinguish this case, we
denote
Substituting Eq. 56 into Eq. 55, we get
( ∇p )
( 1− n ) / n

G (∇p) = gn ( ∇p ) = c −1
. . . . . . . . . . . . . . . . . . . . . (47)
QV (2 )n H nQ − ( n −1) (re 2 − rw 2 )n +1
JQn = = . . . . . . . . . . . (57)
∫ w dV
re
Here, ∇p = px 2 + py 2 + pz 2 . Then, the generalized Darcy equa- n

∫ (c ((r )
− r 2 ))n +1 r − n dr
Q n 2
tion for power law takes the following form: U e
rw
I
v = − gn ( ∇p ) ∇p . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (48)
Note that JQ1 = J Darcy does not depend on the production rate Q.
Correspondingly, the auxiliary BVP becomes At the same time, if n > 1, then the PI is inversely proportional
to Q(n−1), which is similar to what we observed in the PSS PI for
Circular Drainage Area and Skin Factor subsection for two-terms
(
div gn (∇wQn )∇wQn = − ) Q
V
, . . . . . . . . . . . . . . . . . . . . . . . . . (49) law. In contrast to two-terms law, obtaining the skin for this case is
a more difficult task because the integral in the denominator cannot
wQn
= 0, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (50) be reduced explicitly to elementary functions that depend on rw ,
w re , and n. However, a simple nonexplicit formula can be obtained
from the following analysis. It is easy to see that
∂wQn
I = 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (51)
∂n re
n +1

e
∫ ⎡⎣(r − r 2 ) ⎤⎦ r − n dr
2
( n −1) e
J Darcy c Q (re − rw )
n 2 2 2
=
rw
Hereafter, the symbol (⋅)Qn indicates the case of the power law. .
 ( 2 ) H n −1 (re 2 − rw 2 )n +1
n −1 re
JQn 2
∫ ⎡⎣(r
−1
Eq. 11 for the PSS PI takes the form
e
2
− r ) ⎤⎦ r dr
2

rw
I QV
JQn ( p, v , t ) = JQn = . . . . . . . . . . . . . . . . . . . . . . . . . . (52) . . . . . . . . . . . . . . . . . . . . . . . (58)
∫ wQ dV
n

U If rw << re , then Eq. 58 reduces to:


Once again, to compute the PSS PI, one should first solve the
auxiliary BVP (Eqs. 49 through 51). For the majority of geom- re
n +1
∫ ⎡⎣(r
r − n dr − r 2 ) ⎤⎦
2
etries, it can only be done numerically; but, for the 1D case, the n ( n −1) e
PI can be evaluated explicitly using formulation of the problem J Darcy cQ
=
rw
.
 ( 2 ) H n −1 (re 2 − rw 2 )n +1 ln(re / rw ) − 3 / 4
n −1
in terms of ODE along the streamline with respect to the velocity JQn
field inside the so-called nondeformed streamtube with a varying . . . . . . . . . . . . . . . . . . . . . . . (59)
cross section. This ODE can be solved analytically. The obtained
representation for velocity can be used directly to calculate the PI The right-hand side in Eq. 59 is an exact formula for skin, which
of the well. Using this approach, we obtain an analytical formula is similar to the right-hand side of Eq. 26 in the case of two-terms
for the PSS PI of axially symmetrical flow subjected to power law.
Forchheimer equation. Denote
Let the flow in a cylindrical reservoir be generated by fully
penetrated vertical well in its center (Fig. 2). To find the analytical re

∫ ((r − r 2 ))n +1 r − n dr
2
PSS PI formula, we firstI derive an alternative formula for the PI e
in terms of the velocity v. I (rw , re ) =
rw
. . . . . . . . . . . . . . . . . . . . . . (60)
From Eq. 49, boundary conditions Eqs. 50 and 51, it follows (re 2 − rw 2 )n +1
Q
that ∫ gn (∇wQn ) ∇ wQn ∇ wQn dV = ∫ wQn dV .
U
VU Let sk be the factor that measures deviation of the Darcy PI from
Recollecting formulas linking the velocity of the filtration the power Forchheimer PI. From Eq. 59, it follows that if rw >> re
I n −1 I then sk can be approximated by
field with the gradient of the pressure c n vQn vQn = −∇wQn and
In
vQ = − gn (∇wQn )∇wQn , we get the identity c nQ ( n −1) I (rw , re )
sk = . . . . . . . . . . . . (61)
 (n − 1) ( 2 ) H [ e / rw ) − 3 / 4 ]
n −1 n −1
ln(r
I n +1 Q
∫c vQn dV = ∫ wQn dV . . . . . . . . . . . . . . . . . . . . . . . . . . (53)
n

U
VU Therefore, for fast evaluation of the PI for power Forchheimer flow,
one can use the formula
Substituting Eq. 53 into Eq. 52, we obtain an alternative for-
mula for the PI of the well in case of power Forchheimer flow: JQn = J Darcy / sk . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (62)

I Q2 Comparing Eqs. 62 and 30, it is not difficult to conclude that the


JQn ( p, v , t ) = JQn = I n +1 . . . . . . . . . . . . . . . . . . . . . . (54) dependence of the power and two-terms PIs on parameters rw , re , H,

n
c vQn dV and Q is similar qualitatively. In particular, if rw is small and Q/H is
U
large, then, for some constants C1 and C2 depending on , , c, and n
n −1
In axially symmetrical case, Eq. 54 reduces to r H ⎛ H⎞
(n > 1), JQ , ≈ C1 w , JQn ≈ C2rwn −1 ⎜ ⎟ , and J Darcy ≈ ln −1 re / rw .
Q ⎝ Q⎠
QV Q2
JQn = = , . . . . . . . . . . . . . (55) The asymptote for the power law follows from Eq. C-9.
∫ w dV
re
I
( )
n
Q
2 H ∫ c n ( vQn (r ))n +1 rdr The same extensive analysis as performed for the two-terms law
U
rw
can be conducted for the power law in reservoirs with complicated
geometries. This study is beyond the scope of the current article
and the velocity field can be written explicitly as and will be addressed in the ongoing research.

December 2009 SPE Journal 701


Conditions on the Boundaries pressure outside the drainage domain is known and equal to Pe
This section presents a brief overview of a possible extension and the pressure inside the bottomhole of the well is equal to Pw.
of the proposed framework for different boundary conditions in Then, the process of filtration can effectively be modeled as the
I following mixed IBVP:
the drainage domain and arbitrary initial data. Hereafter, vn is the
normal component of the velocity inside the boundaries of the
drainage area; the boundaries consist of the well surface
w and ∂p
= div ( G ( (∇p) K ∇p ) , . . . . . . . . . . . . . . . . . . . . . . . . (71)
∂t
I exterior boundary of the reservoir
e (see formal definition of
the
vn in Appendix C). I
The mathematical model of this process for two-terms and vn + be ( p − Pe )
= 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (72)
e
power Forchheimer flows is derived in terms of IBVP for pressure
function only. I
vn + bw ( p − Pw )
= 0, . . . . . . . . . . . . . . . . . . . . . . . . . . . . (73)
w

∂p
= div ( G ( (∇p) K ∇p ) , . . . . . . . . . . . . . . . . . . . . . . . . (63)
∂t p( x , 0) = p0 ( x ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (74)
I
∫ vn ds = Q , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (64) Positive coefficients be, bw are dimensional; be characterizes

w conductivity of the external boundary of the drainage domain, and
bw takes into consideration the presence of the so-called “thin-skin”
I
vn
= 0, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (65) zone around the well. Numerical experiments regarding thin-skin
e
computations using mixed boundary condition on the well were
presented by Khalmanova (2004) for linear Darcy law.
and Obviously, in case of nonlinear flow, the only time invari-
ant solution of Eqs. 71 through 74 is steady-state, and it can be
p( x , 0) = p0 ( x ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (66) obtained as the solution of the BVP:
Let pressure be uniformly distributed along the well. (The physics
of this assumption means that the conductivity of the well is infinite
( )
div G ( (∇w Q ,G ) K ∇w Q ,G = 0 , . . . . . . . . . . . . . . . . . . . . . . (75)
compared to conductivity of the porous media.) In many applica-
tions, it is not reasonable to expect that, at reference time t = 0, the I
vn + be ( w Q ,G − Pe )
= 0, . . . . . . . . . . . . . . . . . . . . . . . . . . . (76)
reservoir pressure is equivalent to wQ,G(x). For initial pressures not e

equal to wQ,G(x), the PI does depend on time. Engineers know that, I


when the production rate is constant, the PI converges to PSS PI. vn + bw ( w Q ,G − Pw )
= 0. . . . . . . . . . . . . . . . . . . . . . . . . . . (77)
w
In the linear case, the rate of convergence can be estimated using
the standard method of separation of variables and depends on Correspondingly, the steady-state PI will be equal to
eigenvalues of the BVP (Eqs. A-1 through A-4) (Ibragimov et al.
I
2004). In the nonlinear case, the method of separation of variables
is not applicable. Nevertheless, using modern methods of energy
∫ v ds
n

JG =

w
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (78)
estimates for the solution of the quasilinear parabolic equations, Pe − Pw
we proved in Aulisa et al. (2008) the following statement:
For any initial reservoir pressure p0(x), solution of the IBVP
Numerical Analyses and Simulation
(Eqs. 63 through 66) will converge to PSS solution. The PI in this
case stabilizes to the value of the PSS PI exponentially. In this section, we evaluate and compare the PI in standard reser-
voir/well geometries for the following cases:
I QV • The transient PI where Eq. 10 is computed solving the original
JG ( p, v , t ) → J q ,G = , as t → ∞ . . . . . . . . . . . . . . (67) Forchheimer system of equations.
U
∫w q ,G dV • The PSS PI evaluation on the basis of Eq. 20 and computed
by solving the steady-state auxiliary BVP (Eqs. 17 through 19).
The rate of convergence in Eq. 67 depends on the coefficients in • The PSS PI evaluation on the basis of Eq. 42 and computed
G and geometrical parameters of the domain. by solving the steady-state auxiliary BVP (Eqs. A-6 through A-8)
Exterior boundary condition can also be generalized as follows: for the Darcy flow.
Let the flux (in or out) per unit surface on the external boundary • The PSS PI evaluation on the basis of formulas with skin
of the drainage domain be time independent and equal qext(x). Eqs. 30 and 45 by solving the steady-state auxiliary BVP (Eqs.
Then, the construction introduced in the Mathematical Model A-7 through A-9) for the Darcy flow.
for the PI of the Well section and the corresponding formula for All the simulations have been performed using COMSOL Multi-
PSS PI is valid. In this case, the PSS solution can be obtained as physics (COMSOL; Stockholm; 2008). Timestep and grid size have
Q − Qext been refined until convergence in time and space is reached. Per-
p1 ( x , t ) = − t + B + w1Q ,qext ,G ( x ) , and w1Q ,qext ,G ( x ) is solving formed numerical simulations estimate the effect of the three leading
V
parameters of interest—the rate of production Q, the Forchheimer
the auxiliary BVP.
coefficient , and the geometry—on the value of the PI.
Note that, in all computations, the PI is dimensionless. To con-
( ( )
div G ∇wQ ,Qext ,G K ∇wQ ,Qext ,G = − ) Q − Qext
V
, . . . . . . . . . . . . (68) vert the dimensional PI into a dimensionless one, we multiply the
PI by the factor /(2 k) for 2D cases and by the factor /(2 kH)
for 3D cases (Dake 1978).
wQ ,Qext ,G = 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (69) The transient PI obtained by solving the original Forchheimer

w
system converges and stabilizes to the value of the PI obtained by
I using the PSS PI formula for all cases and for a wide variety of
vn
= qext ( x ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (70) initial data.
e

Here, Qext is a total flux (in or out) on the exterior boundary (see Rectangular Reservoir and Fully Penetrated Vertical Well
Appendix C). (2D Case). Henceforth, the original Forchheimer system will be
Next, consider the steady-state boundary condition for the fol- referred to as system of Eq. 1 coupled with a continuity equation
lowing engineering problem. Assume that the reference average and equation of state for slightly compressible fluid.

702 December 2009 SPE Journal


Lx2
Lx2 = 4000 cm D = 500
θ

Lf

Lx1

Lx1 = 8000 cm

Fig. 6—2D scheme of the fully penetrated vertical well in a Fig. 7—2D scheme of fully penetrated fracture in a rectangular
rectangular reservoir. reservoir.

In this section, we compare the results obtained by solving the The varying geometric parameter is
original transient Forchheimer system and those obtained with both • The angle  = 0, 45, and 90° between the fracture and the
the steady-state auxiliary BVP (Eqs. 17 through 19) and the steady- x2 axis.
state auxiliary BVP (Eqs. A-7 through A-9) for the Darcy case. The fixed hydrodynamic parameters are
The reservoir domain is modeled as a 3D rectangular box. We •  = 1.01 × 1010 (Pa·s)/m2.
consider a fully penetrated vertical well. Because of the bound- • = 0.658 × 10−9 Pa−1.
ary conditions on the well and the exterior boundary, the problem The varying hydrodynamic parameters are
reduces to the 2D geometry sketched in Fig. 6. •  = 2.4318 × 1011, 2.4318 × 1012, and 2.4318 × 1013 Pa·s2/m3.
In numerical simulations, we study the effect of the hydrody- • Q/H = 10−4, 10−3, and 10−2 m2/s.
namic parameters on the value of the PI. In Table 2, Column T Results of computation of the well PI for different values of
contains the PI computed using the original transient equation after the angle , the Forchheimer coefficient , and production rate Q
PSS is reached. In Column I, the PI is obtained using the numerical are reported in Table 3. The PI values in Column I are obtained
solution of the BVP (Eqs. 17 through 19) and Eq. 20. In column II using Eq. 20, while the values in Column II are obtained using the
the PI is obtained using the numerical solution of the BVP (Eqs. approximate Eq. 42.
A-6 through A-8) and the approximate Eq. 42. Results in Columns I and II are very close for the small values
The fixed geometric parameters are of production rate Q and the Forchheimer coefficient . As Q and 
• The midpoint of the well. increase, the difference increases as well. The maximum difference
• The radius of the well rw = 0.3 m. (approximately 5%) can be observed for the horizontal well when
• The height of the box H = 10 m.  = 90°. It reflects the fact that, in this configuration, the uniform
• Lx1 = 80m, Lx2 = 40m, and D = 5 m. flow constraint is violated to the highest extent.
The fixed hydrodynamic parameters are
•  = 1.01 × 1010 (Pa·s)/m2. Vertically Deviated Well in a Cylindrical Reservoir (3D Case).
• = 0.658 × 10−9 Pa−1. In this section, we present numerical results for a 3D cylindrical
The varying hydrodynamic parameters are reservoir with a well deviated from the vertical axis (Fig. 8).
•  = 2.4318 × 1011, 2.4318 × 1012, and 2.4318 × 1013 Pa·s2/m3. The fixed geometric parameters are
• Q/H = 10−4, 10−3, and 10−2 m2/s. • The radius of the reservoir re = 100 m.
The values of the PI (Column T) for the solution of the original • The reservoir thickness H = 10 m.
transient Forchheimer system (Eq. 1) after stabilization become • The radius of the well rw = 0.3 m.
almost identical to values of the PSS PI (Column I). The numerical • The length of the well Lw = 8 m.
computation of the Forchheimer transient values takes a long time
and requires a fine grid and adjustment of the initial data. This
justifies the use of the generalized Darcy equation.
Deviations of the PI value between Columns I and II are very
small for given values of the production rate and Forchheimer
coefficient . Once more, this highlights the fact that, for simple
flow geometry around a vertical well, the effect of Forchheimer
phenomena, at least at the level of the PI, can be handled alge-
braically from the solution of the corresponding problem for the
linear Darcy case.

2D Fracture in a Rectangular Reservoir (2D Case). In this


section, we consider a horizontal well in a 3D box modeled as a
vertical fully penetrated fracture. The vertical flow in direction x3 is
ignored. This leads to the classical geometry presented in Fig. 7.
θ
Here, the midpoint of the well coincides with the midpoint of the
reservoir. In the numerical computation, we compare the effect
of the geometry and the hydrodynamic parameters on the value
of the PI.
The fixed geometric parameters are
• The midpoint of the well.
• The thickness the fracture df = 0.6 m.
• The length of the fracture Lf = 38 m.
• The height of the box H = 10 m. Fig. 8—3D Scheme of the cylindrical reservoir with a vertically
• Lx1 = 80 m and Lx2 = 40 m. deviated well.

December 2009 SPE Journal 703


The varying geometric parameter is x = space variable

• The angle  = 0, 15, 30, 45, 60, and 75° between the well = = coefficient, Pa·s/m2
and the vertical axis. k
The fixed hydrodynamic parameters are.  = coefficient, Pa·s2/m4
•  = 1.01 × 1010 (Pa·s)/m2. = coefficient, Pa−1
• = 0.658 × 10−9 Pa−1.
e = boundary of the reservoir
The varying hydrodynamic parameters are
w = surface of the well
s = Q, with = dimensionless parameter
•  = 0, 2.4318 × 1012, 2.4318 × 1013 Pa·s2/m3. P = drawdown, Pa
• Q = 8 × 10−2 m3/s.  = viscosity, Pa·s
As one can see from Table 4, the results are very close for  = density
small values of c0; as c0 increases, the difference also increases
slightly. The difference between the two columns also increases Acknowledgments
as the angle  increases for a given value of c0. The maximum This research was particularly supported by National Science
difference (approximately 5%) is observed for  = 75°. It reflects Foundation grant DMS-0813825, by Advanced Research Program
that, in this configuration, the uniform flow constraint is violated grant 0212-44-C399, and by award number KUS-C1-016-04,
to the highest extent. made by King Abdullah University of Science and Technology.
The authors want thank P. Valko for stimulating discussions and
Conclusions valuable consultations. We would like to extend our thanks to the
In this paper, a new framework to evaluate well performance in reviewers whose comments exceeded editorial responsibilities and
case of nonlinear Forchheimer flows is proposed. The developed led to new results and significant improvement of the paper.
technique for evaluation of the PI combines generalized nonlinear
Darcy equation and easy-to-apply numerical and analytical meth- References
ods. The major benefits of the approach include Aulisa, E., Cakmak, A., Ibragimov, A., and Solynin, A. 2007. Variational
• It offers effective and accurate methods for building tables and Principle and Steady State Invariants for Non-Linear Hydrodynamic
evaluating the PI for vertical, horizontal, and deviated wells in Interactions in Porous Media. Dynamics of Continuous, Discrete and
reservoirs with complicated geometries. Impulsive Systems, Series A: Mathematical Analysis 14 (S2): 148–155.
• The derived analytical formula for PI with skin provides a fast Aulisa, E., Ibragimov, A., Valko, P., and Walton, J. 2006. Mathematical
and explicit tool to measure the effect of hydrodynamic and geo- Framework for the Productivity Index of Wells with Fast Forchheimer
metrical parameters of the reservoir/well system on the deviation (non-Darcy) Flow in Porous Media. Proc., COMSOL Users Confer-
of the Forchheimer PI from the Darcy PI. ence, Boston, Massachusetts, USA, 22–24 October.
• For a wide class of reservoir/well geometries and a wide range Aulisa, E., Ibragimov, A., Valko, P., and Walton, J. 2009. Mathematical
of hydrodynamic parameters, the derived algebraic relation Framework for Productivity Index of the Well for Fast Forchheimer
between the two-terms Forchheimer and Darcy PIs are proved to (non-Darcy) Flow in Porous Media. Mathematical Models and Methods
be accurate for feasible constraints on the flows in the reservoir; in Applied Science.
this approximation becomes less realistic only for extremely Auriault, J.-L., Geindreau, C., and Orgéas, L. 2007. Upscaling Forchheimer
large values of the product c0 = Q. law. Transport in Porous Media 70 (2): 213–229. DOI: 10.1007/
Mathematical and numerical analyses reveal several important quali- s11242-006-9096-x.
tative properties of the PIs for two-terms Forchheimer flow, which have Balhoff, M.T. and Wheeler, M.F. 2007. Predictive Pore-Scale Model for
clear engineering interpretations. Below are some of them: Non-Darcy Flow in Anisotropic Media. Paper SPE 110838 presented
• If the product between the Forchheimer coefficient  and the at the SPE Annual Technical Conference and Exhibition, Anaheim,
production rate Q is constant, then the PSS PI is invariant with California, 11–14 November. DOI: 10.2118/110838-MS.
respect to any variation of the production rate and the Forch- Balhoff, M.T., Mikelic, M., and Wheeler M.F. 2009. Polynomial Filtra-
heimer coefficient. tion Laws for Low Reynolds Number Flows Through Porous Media,
• The Forchheimer PI increases linearly as the product c0 = Q Transport Porous Med, April (electronic version).
decreases and converges to the Darcy PI. Bear, J. 1972. Dynamics of Fluids in Porous Media. Mineola, New York:
• Even for small values of the parameter c0, the effect of nonlin- Dover Publications.
earity is significant if the value of the pressure gradient is large Civan, F. 2000. Leaky-tank reservoir model including the non-Darcy effect.
(e.g., in a highly nonhomogeneous reservoir). Journal of Petroleum Science and Engineering 28 (3): 87–93. DOI:
10.1016/S0920-4105(00)00073-5.
Nomenclature Dake, L.P. 1978. Fundamentals of Reservoir Engineering. Amsterdam:
A = matrix of the bilinear form ( n −1)
Elsevier Science BV.
c = coefficient, [ Pa ⋅ s / m 3 ]1/ n ([s] / [m]) n Dietz, D.N. 1965. Determination of Average Reservoir Pressure From
G, M, N = nonlinear functions of gradient of pressure Build-Up Surveys. J. Pet Tech 17 (8): 955–959; Trans., AIME, 234.
SPE-1156-PA. DOI: 10.2118/1156-PA.
JQ,G = PSS PI for generalized Darcy law
Douglas, J. Jr., Paes-Leme, P.J., and Giorgi, T. 1993. Generalized Forch-
JQ, = PSS PI for two-terms Forchheimer law heimer Flow in Porous Media. In the book Boundary Value Problems
JDarcy = PSS PI for Darcy law for Partial Differential Equations and Applications, 99–111, RMA Res.
JQn = PSS PI for power Forchheimer law Notes Appl. Math, 29 Masson, Paris.
JG = steady-state PI for generalized Darcy law Dmitriev, M.N., Dmitriev N.M., and Maksimov, V.M. 2005. Representation
J = dimensionless PI of the Functions of the Relative Phase Permeabilities for Anisotropic
k = permeability, m2 Porous Media. Fluid Dynamics 40 (3).
L = distance Ewing, R.E., Lazarov, R.D., Lyons, S.L., Papavassiliou, D.V., Pasciak, J.,
Q = production rate, m3/s and Quin, G. 1999. Numerical well model for non-Darcy flow through
re = external radius of cylindrical reservoir, m isotropic porous media. Computational Geosciences 3 (3–4): 185–204.
DOI: 10.1023/A:1011543412675.
rw = radius of the well, m
Forchheimer, P.F. 1901. Wasserbewegung durch Boden. Zeitschrift des
sk = skin factor Vereines deutscher Ingenieure 45 (5): 1781–1788.
T = time Helmy, M.W. and Wattenbarger, R.A. 1998. New Shape Factors for
U = reservoir domain Wells Produced at Constant Pressure. Paper SPE 39970 presented at
V = drainage volume, m3 the SPE Gas Technology Symposium, Calgary, 15–18 March. DOI:
w = solution of the auxiliary problem 10.2118/39970-MS.

704 December 2009 SPE Journal


Ibragimov, A., Khalmanova, D., Valko, P.P., and Walton, J. 2004. Analyti-
wQ ,0
= 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-3)
cal Method of Evaluating Productivity Index for Constant Production w

Rate or Constant Wellbore Pressure. Paper SPE 89935 presented at


the SPE Annual Technical Conference and Exhibition, Houston, 26–29 ∂wQ ,0
I = 0. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-4)
September. DOI: 10.2118/89935-MS. ∂n
e
Ibragimov, A.I., Khalmanova, D., Valko, P.P., and Walton, J.R. 2005.
On a Mathematical Model of the Productivity Index of a Well From
The dimensionless PI for PSS regime is equal to
Reservoir Engineering. SIAM J. Appl. Math. 65 (6): 1952–1980. DOI:
10.1137/040607654. VQ
Khalmanova, D. 2004. A Mathematical Model of the Productivity Index. J PSS = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-5)
PhD dissertation, Texas A&M University, College Station, Texas. ∫U
wQ ,0 dV
Kumar, A. 1977. Strength of Water Drive or Fluid Injection From Transient
Well Test Data. J. Pet Tech 29 (11): 1497–1508; Trans., AIME, 263. In Eq. A-5, wQ,0 is the solution of the auxiliary steady-state
SPE-5054-PA. DOI: 10.2118/5054-PA. problem:
Larsen, L. 2001. General Productivity Models for Wells in Homogeneous
and Layered Reservoirs. Paper SPE 71613 presented at the SPE Annual
Technical Conference and Exhibition, New Orleans, 30 September–3
(
div  −1K ∇wQ,0 = − ) Q
V
, . . . . . . . . . . . . . . . . . . . . . . . . . . (A-6)
October. DOI: 10.2118/71613-MS.
Li, D. and Engler, T.W. 2001. Literature Review on Correlations of the
wQ ,0
= 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-7)
Non-Darcy Coefficient. Paper SPE 70015 presented at the SPE Permian w

Basin Oil and Gas Recovery Conference, Midland, Texas, 15–17 May.
DOI: 10.2118/70015-MS. ∂wQ ,0
Maksimov, A.M, Kochina, I.N., and Basniev, K.S. 1993. Underground I = 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-8)
Hydrodynamics (Подземная гидродинамика). Moscow: Nedra. ∂n
e

Muskat, M. 1937. The Flow of Homogeneous Fluids Through Porous


Media. New York: McGraw-Hill Book Company. Because of the linearity, PI for Darcy law does not depend on the
Payne, L.E. and Straughan, B. 1999. Convergence and Continuous Depen- rate of production.
dence for the Brinkman- Forchheimer Equations. Studies in Applied
Mathematics 102 (4): 419–439. DOI: 10.1111/1467-9590.00116. Appendix B—Estimation of the Difference
Raghavan, R. 1993. Well Test Analysis. Englewood Cliffs, New Jersey: Between Two-Terms Forchheimer and
Prentice Hall Petroleum Engineering, Prentice-Hall. Darcy PIs
Tavera, C.A.P., Kazemi, H., and Ozkan, E. 2006. Combine Effect of From Eqs. 20 and 37, it follows that
Non-Darcy Flow and Formation Damage on Gas Well Performance of
Dual-Porosity and Dual Permeability Reservoirs. SPE Res Eval & Eng
9 (5): 543–552. SPE-90623-PA. DOI: 10.2118/90623-PA. VQ ∫ wQ , dV − ∫ wQ ,0 dV
Wheeler, M.F. and Peszyńska, M. 2002 Computational engineering and
J Darcy − J q , = U U
. . . . . . . . . . . . . . (B-1)
science methodologies for modeling and simulation of subsurface
applications. Advances in Water Resources 25 (8–12): 1147–1173. U
∫ wQ ,0 dV ∫ wQ , dV
U
DOI: 10.1016/S0309-1708(02)00105-7.
Whitaker, S. 1996. The Forchheimer equation: A theoretical develop- To evaluate the numerator in Eq. B-1, we multiply Eq. 17 times
ment. Transport in Porous Media 25 (1): 27–61. DOI: 10.1007/ wQ,0 and Eq. 38 times wQ, and then apply the Green formula. It
BF00141261. then follows that

Appendix A—PI for Linear Darcy Case


U
∫ (Qw Q , − QwQ ,0 )dV =

( )
In case of linear Darcy flow for slightly compressible fluid, the
V ∫ ⎡ −1 − f ∇wQ , ⎤ ∇wQ , ∇wQ ,0 dV . . . . . . . . . . . . . . . (B-2)
most popular way to evaluate the PI for constant thickness reservoir ⎣ ⎦
is based on the representation of the PI as the product, PI = F0 J . U

2 kh
Here, F0 = depends only on the fluid and on the porous-media From the definition of the function f, it can be obtained that
Bf 
properties together with the reservoir thickness. J is the dimen-  2
sionless PI, which depends on the well/reservoir geometry and  −1 − f ∇wQ , = ( ) ( )
f ∇wQ , ∇wQ , . . . . . . . . . . . . . (B-3)
 
on the type of the flow regime. The PSS and the BD productivity
indices are traditionally estimated by the equation Substituting Eq. B-3 into Eq. B-2 and applying the Cauchy-
Schwartz inequality produces
1
J Dietz = , . . . . . . . . . . . . . . . . . . . . . . . . (A-1)
1 ⎛ 4A ⎞
ln ⎜ + s
2 ⎝ e C Arw2 ⎟⎠ ∫ (Qw
U
Q , − QwQ ,0 )dV


where is the Euler’s constant, A is the drainage area, rw is the
wellbore radius, and s is the skin factor. The shape factor CA for
≤V
 U∫ 
( 2
f 2 ∇wQ , ∇wQ , ∇wQ ,0 dV . )
computing the PSS PI can be obtained from Dietz (1965).
In Ibragimov et al. (2004, 2005), a method for evaluating both After some analytical work and taking into account inequality
the PSS and BD productivity indices of a well are presented. This f 2 ≤  −1 and the fact that wQ, is a solution of BVP (Eqs. 17
method is based on the solution of the time-independent (steady- through 19), we get
state) BVPs. It was shown that, in a reservoir of the volume V,
the dimensionless PI for BD regime is equal to J BD = V . Here,
λ is the first eigenvalue of the mixed BVP problem for elliptic ∫ (Qw Q , − QwQ ,0 )dV
differential equations: U

 Q
≤V max ∇wQ ,0 ∫ wQ , dV .
div ( K ∇u ) + u = 0 , . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-2) 2 U VU

December 2009 SPE Journal 705


Substituting the last inequality into the right-hand side of Eq. B-1, I I
vn = G ⎡⎣ (∇w Q ,G ) ⎤⎦ K ∇w Q ,G , n . . . . . . . . . . . . . . . . . . . . . . (C-6)
we get

Derivative with respect to normal direction on boundary


of
the domain U is defined as
 Q
V 2 max ∇wQ ,0 ∫ wQ , dV
 U VU ∂w I
J Darcy − JQ , = I = (∇w, n )
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-7)
∫ wQ ,0 dV ∫ wQ , dV
U U
∂n

 In the Conditions on the Boundaries section, total flux on the


QV 2 max ∇wQ ,0 QV  max vI1,0
defined as
≤  U
= U
J Darcy . . . . . . . . . . . (B-4)
∫ Q ,0
w dV 
I
U
∫ v ds = Q

n ext . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-8)

re
Appendix C—MathematicalI Notations n +1
∫ ⎡⎣(r − r 2 ) ⎤⎦ r − n dr
2
I e
For two vectors a = (a1 , a2 , a3 ) and b = (b1 , b2 , b3 ) , the dot product
I (rw , re ) = =
rw
is defined as (re 2 − rw 2 )n +1
I I
re

(a , b ) = ∑ ai bi . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-1)
n
2(n + 1) ∫ ⎡⎣(re 2 − r 2 ) ⎤⎦ r 2− n dr
n
r
+
w rw
. . . . . . . . . . . . . . . . . (C-9)
I 1− n (11 − n)(re 2 − rw 2 )n +1
Correspondingly, modulus of the vector a is equal to
I2 I I
a = ( a , a ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (C-2)
Akif Ibragimov is a professor of mathematics at Texas Tech
Notation Y(x)|
for any variable of interest Y(x) means the value University. Ibragimov’s research interest includes mathematical
I on the boundary
.
of the Y(x) analyses and modeling of flows in porous media and
Let n be an outside normal to the boundary
of the domain dynamical systems. He holds a PhD degree from Lomonosov,
U, then normal component of the velocity field of the generalized Moscow State University, and a doctor of science degree
Darcy law on the boundary is defined as a dot product. from Steklov Mathematical Institute in Moscow. Eugenio Aulisa
is assistant professor in the department of Mathematics and
I I
vn
= (G ( (∇p) ) K ∇p, n )
. . . . . . . . . . . . . . . . . . . . . . (C-3)
Statistics at Texas Tech University. He holds a PhD degree in
energetic, nuclear, and environmental control engineering
from the University of Bologna, Italy. His primary research
For the equation on the exterior boundary
e in the auxiliary
I interests are computational fluid mechanics, finite element
steady-state problems (Eqs. 12, 13 14, 72, 73, 74, and 78), vn methods, multigrid solvers, nonlinear flows in porous media, and
reduces to fluid/structure interaction problems. Jay R. Walton is professor
of mathematics and aerospace engineering at Texas A&M
I I
vn = G ⎡⎣ (∇wQ ,G ) ⎤⎦ K ∇wQ ,G , n , . . . . . . . . . . . . . . . . . . . . . (C-4) University. He holds a PhD degree in mathematics from Indiana
University. Walton’s research interests include mathematical
analysis and modeling in solid mechanics, especially fractures,
I I time-dependent material behavior, nonlinear theories, flow
vn = G ⎡⎣ (∇wQ ,Qext ,G ) ⎤⎦ K ∇wQ ,Qext ,G , n , . . . . . . . . . . . . . . . . (C-5) through porous material, and biological materials.

706 December 2009 SPE Journal

You might also like