Professional Documents
Culture Documents
Copyright © 2010 Society of Petroleum Engineers Rate Equation for a Partially Penetrating
This paper (SPE 137051) was revised for publication from paper SPE 118845, first Vertical Well
presented at the SPE Reservoir Simulation Symposium, The Woodlands, Texas, USA,
2–4 February 2009. Original manuscript received for review 14 August 2009. Revised We begin by deriving a simple formula for the well index that is
manuscript received for review 13 January 2010. Paper peer approved 28 January 2010. suitable for partially perforated vertical wells, as shown in Fig 2.
q
Pk ro
Δy kz
x
hp
kr
H
∆x
Δx
q = 2 KH
( e − wf ). . . . . . . . . . . . . . . . . . . . . . . . . . . . . (6)
parameters in Eq. 8 and the value of skin factor obtained from the
well test or by formulas (Odeh 1980; Brons and Marting 1961).
⎛r ⎞
ln ⎜ e ⎟ + Sc Implementation in a Reservoir Simulator. For a gridblock k,
⎝ rw ⎠
which contains a perforation of a partially penetrating vertical
In Eq. 6, Sc is called completion (perforation) skin factor. Comple- well, we can write the rate equation using the form suggested in
tion skin factor can be calculated from well tests or formulas. Sev- Eq. 7 rather than Eq. 1:
eral authors have suggested different formulas for the completion
skin. We will use skin factors suggested by Odeh (1980) and Brons
qk = 2 K z k f
( k − wf ). . . . . . . . . . . . . . . . . . . . . . . . (10)
and Marting (1961). The expression for flow from restriction can pk ⎛r ⎞
have a more complicated form than Eqs. 4 and 6; however, our ln ⎜ o ,3 D ,k ⎟
objective is to provide a reasonable approximation that can correct ⎝ rw ⎠
the radial-flow model presented by Peaceman (Eq. 1).
Comparing the flow-rate equation proposed by Kozeny (1953) In Eq. 10, zk is the thickness of the gridblock containing the
and Craft and Hawkins (1959) (Eq. 4) with Eq. 6, we see that Eq. 6 open perforation and fpk is the partial-penetration factor for the same
includes the effect of vertical permeability Kz through the skin fac- gridblock, which can be calculated by Eq. 8 or Eq. 9. The term k in
tor expressed by Odeh (1980) and Brons and Marting (1961) . Eq. 10, fluid potential for gridblock k, is calculated by the simulator,
In this paper, for convenience, we will use the form suggested and ro, 3D, k is the unknown equivalent wellblock radius on which k
by Eq. 4 by introducing a new parameter called fp, the partial- resides. The flow rate qk, calculated by Eq. 10, assumes that comple-
penetration factor. With the new parameter, Eq. 4 becomes tion k, in addition to horizontal flow, receives vertical flow from
above and below. Therefore, the term zk fpk is the effective gridblock
q = 2 Kh p f p
( e − wf ). . . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
thickness that accounts for both horizontal and vertical flow; hence,
zk fpk > zk for partially penetrating well with fpk > 1.
⎛r ⎞ Rearranging Eq. 10, the new well index or perforation index
ln ⎜ e ⎟
⎝ rw ⎠ for a vertical well becomes
x k-1
Δy Φk
k qk
Δx
Δz
k+1
Kz
ΦB ΦB
KX
Δx
k=N-1
Fig. 3—Constant potential boundary conditions in areal direction.
k=N
Δx z
The first term in Eq. 12 represents the total horizontal flow rate
∆x
qh,k. The remaining terms in Eq. 12 describe the total vertical-flow
rate qv,k into the gridblock, and qk is the total withdrawal rate form
the perforation.
Transmissibilities Tx and Tz are defined in the Nomenclature. Δx
Eq. 12 can be abbreviated as
Fig. 5—Definition of wellblock radius for the vertical-flow
qh ,k + qv ,k = qk, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (13a) effects, areal view.
200
H=100 ft , kx =kz =100 md
Dx=820 ft, Dz=20 ft, b=3,000 psi
f p ,k= 1
2 π K Δz fp(Pk –Pw)/q
0
f p , k = 0.3
Fully Penetrating Well
-100
Partially Penetrating Well
-200
ro,2D ro,3D
-300
0.0001 0.001 0.01 0.1 0.2 0.6 1.0
rw /Δ x
r /Δ x re / Δx
ro/Δx
4 New Method
5
Peaceman's Method
6
10
11
q k-1
qk
Δz
z
q k+1
k=N-1 q Nc
k=N
Δx z
fL =
sinh ( az ) cosh ⎡⎣ a ( H − z )⎤⎦ . . . . . . . . . . . . . . . (18b)
s s
⎧sinh ( az ) cosh ⎡ a ( H − z ) ⎤ ⎫
⎪
⎨
s⎣ ⎦ ⎪
⎬
s Nc
t ( z ) = ∑ z ≤ z j ( )⎧
q j sinh ( )
az j cosh ⎡⎣ a H − z j ( )⎤⎦
⎪⎩+ cosh ( az ) sinh ⎡⎣ a ( H − z ) ⎤⎦ ⎪⎭
s s j =1
⎪sinh ( ) (
az j cosh ⎡⎣ a H − z j ⎤⎦ ⎫
⎪ )
⎨ ⎬
and ⎪⎩+ cosh ( )
⎡ ⎤(
az j sinh ⎣ a H − z j ⎦ ⎪
⎭ )
cosh ( )
az s cosh ⎡⎣ a ( H − z s ) ⎤⎦ ×
( az )
cosh
fR = . . . . . . . . . . . . . . (18c) ⎡ aK x ysinh ( )
az j ⎤⎦
⎪ ( ⎣ )
⎧sinh az s cosh ⎡ a ( H − z s ) ⎤ ⎫
⎦ ⎪
⎣ z
⎨
( )
⎬
⎪⎩+ cosh az s sinh ⎡⎣ a ( H − z s ) ⎤⎦ ⎪⎭ + z > z j( )⎧
(
q j sinh ⎡⎣ a H − z j ⎤⎦ cosh az j) ( )
( ) ⎡
⎪ sinh az j cossh ⎣ a H − z j ⎦ ⎪ ( ⎤⎫ )
⎨ ⎬
Partial-penetration factor fp,k can be determined by the methods
described earlier in this paper. Our suggestion is to use Odeh’s ⎩⎪ ( )
+ cosh az j sinh ⎡⎣ a H − z j ⎤⎦ ⎪ ( ⎭ )
method (Odeh 1980) for perforations not located at the top or cosh ⎡⎣ a ( H − z ) ⎤⎦
bottom because it yields results closer to the fine-grid numerical ×
{ )}
.
solution. To use the Odeh (1980) or the Brons and Marting (1961) aK z x ysinh ⎡⎣ a H − z j ⎤⎦(
method, we set re = x in Eq. 9.
. . . . . . . . . . . . . . . . . . . . . . . (21)
⎛ x ⎞
ln ⎜ ⎟ The equivalent wellblock radius ro,3D,k is obtained from Eq. A-24
H ⎝ rw ⎠ q
fp = . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (19) by evaluating the term v ,k as defined by Eqs. A-27 and A-28 in
z ⎛ x ⎞ qk
ln ⎜ ⎟ + Sc
⎝ rw ⎠ Appendix A. The in Eq. 21 is the Dirac delta function, which
assumes the value of unity if the argument is true, otherwise zero.
Sc in Eq. 19 is defined by formulas (Craft and Hawkins 1959;
Odeh 1980; Brons and Marting 1961). The perforation index WIk Numerical Experiments and Verification of the New Equivalent
is calculated by Eq. 11. Wellblock Radius. A 3×3×11 test model (Fig. 9) was used to test
For a general case, the measured rate (well test) can be used the new method and compare with Peaceman’s method. A vertical
to select the best correlation for this well by matching the calcu- well was placed in the center of the central gridblock. Gridblocks
lated rate by this method with the measured rate by changing the surrounding the central well were assumed to have constant poten-
correlation. tial, 4,780 psi. In the areal plane, square grids were used with 820-ft
sides. Vertical layer thickness was assumed to be constant for all
Multiple Perforations. The analytical solution for the potential of the 11 layers (9.09 ft).
equation can be obtained by superposing the analytical solution for Two completion scenarios were considered—(1) completion
each perforation (Fig. 8). Let t(z) represent the total potential interval at the top of the formation, Layers 1 and 2, and (2) comple-
drop at location z from multiple sources located at the well qj, j = tion interval at the middle of the formation, Layers 5 and 6.
1, Nc, where Nc is the total number of completions. By superposing The flowing bottomhole pressure was set to 4,424 psi. Reservoir
the potential drops, oil viscosity was 3 cp. Reservoir permeability was set to 500 md in x
and y directions. The vertical permeability was 250 md. A simulator
Nc (Dogru et al. 2002) was run to predict the oil rate for three cases:
( )
t ( z ) = ∑ z , z j , . . . . . . . . . . . . . . . . . . . . . . . . . . . (20) • Base case, coarse grid using Peaceman’s well index (Fig. 9)
j =1 • Locally refined grids for the central cell where the well is
located (81×81×11 grid with x = 10.12 ft) using Peaceman’s well
where index (Figs. 10 and 11)
• New method, coarse grid with new well index (Fig. 9) Nomenclature
For the top-completion-interval scenario (1), model runs showed fp = partial-penetration factor
that oil-production rate predicted by the coarse grid, Case 1, was H = total formation thickness, cm
22% less than the oil rate calculated by locally refined grids, Case 2.
k = gridblock index
Coarse grid with new well index, Case 3, yielded rates very close
to the locally refined case, within 0.6% difference. For this scenario, Kx = permeability in x direction, darcies
the new well index was 1.31 times the Peaceman’s well index. We Ky = permeability in y direction, darcies
have used Muskat’s method to calculate the skin factor. Kz = permeability in z direction, darcies
For the middle-completion-interval scenario (2), Layers 5 and N = number of gridblocks
6, the completed interval receives flow from the top and the bottom P = pressure, atm
of the completion interval, and, hence, the effect of vertical flow q = production rate, cm3/s
is more pronounced. re = drainage radius, cm
Model runs showed that the difference between Case 1 and Case 2 ro = wellblock radius, cm
was approximately 30% (29.5%). Case 3 with the new well index rw = well radius, cm
resulted in oil rate very close to that of Case 2, with a difference of
T = transmissibility
0.7%. The new method well index was 1.43 times the Peaceman’s
well index. We have used the Odeh (1980) method to calculate the Tx = Kxyz/x
completion skin factor. Ty = Kyxz/y
Tz = Kzyx/z
Conclusions WI = well index
• A new practical formula and procedure have been developed for = potential, atm
the equivalent wellblock radius and perforation index for par-
tially penetrating vertical wells for square areal grids. The new Subscripts
formula is easy to implement in a numerical reservoir simula- B = boundary
tor. Analytical solutions developed in this paper can be used for c = completion
homogeneous and anisotropic reservoirs with uniform grids. e = external (drainage)
• The theory presented here is valid only if the potential around f = flowing or fraction
the wellblocks remains constant and approximately steady-state j = completion number
flow conditions exist. Well indices calculated by this method k = cell number (index)
can be repeated at every timestep during the simulation or once
L = left
at the beginning of the simulation, depending on the conditions.
Constant-potential approach for the neighboring gridblocks is p = perforated or partial
generally justifiable for large, high-permeability, thick reservoirs t = total
for full-field-simulation models. w = well
• New formulas for the equivalent wellblock radius and well indi- wf = flowing well
ces depend only on the grid properties and not on the well bound- x = areal x direction
ary condition (specified flow rate or bottomhole pressure). y = areal y direction
• Numerical experiments have shown that Peaceman’s method z = vertical direction
(Peaceman 1978, 1983) underpredicts the well index, while the
new well index can correct the error in flow rates. For a model Acknowledgments
problem, the new method reduced the error from 30% to less than
1%. An alternative to the new method is to use locally refined The author would like to thank Hussein Kazemi and Dean Oliver
grids around the wells. Depending on the degree of refinement, for valuable discussions and suggestions. The author would also
this approach can be very costly for full-field-simulation models like to thank Jorge Pita and Larry Fung for reviewing the manu-
with many wells. script and Tom Dreiman for model building.
• Accuracy of equivalent wellblock radius and perforation indices
calculated by the new method depends on the accuracy of the References
rate formula. If the perforation skin is used in the rate formula, it Aavatsmark, I. and Klausen, R.A. 2003. Well Index in Reservoir Simula-
needs to be calculated accurately. It can be estimated from a well tion for Slanted and Slightly Curved Wells in 3D Grids. SPE J. 8 (1):
test or from correlations. In either case, it is advisable to calibrate 41–48. SPE-75275-PA. doi: 10.2118/75275-PA.
the new rate equation used to existing historical production data. Aziz, K. and Settari, A. 1979. Petroleum Reservoir Simulation. Essex, UK:
Alternatively, published formulas for the skin factor can be used. Elsevier Applied Science Publishers.
• The new method can be expanded to horizontal wells or multilat- Babu, D.K. and Odeh, A.S. 1989. Productivity of a Horizontal Well. SPE
eral wells by following the same methodology of this paper. Res Eng 4 (4): 417–421. SPE-18298-PA. doi: 10.2118/18298-PA.
∂ ( z )
∂z
= C1 a exp ( az ) − C 2 ( )
a exp − az . . . . . . . . . (A-5b)
⎛ ∂ ⎞
q = K z x y ⎜ at z s = 0. . . . . . . . . . . . . . . . . . . . . (A-6)
Fig. A-1—Grid system. ⎝ ∂z ⎟⎠ z = 0
s
Close examination of the ratio b/a reveals that this term repre- qL =
sinh( az ) cosh ⎡⎣ a ( H − z )⎤⎦ q . . . . . . . . . . . . (A-19)
s s
⎧sinh ( az ) cosh ⎡ a ( H − z ) ⎤ ⎫
k
sents the average potential around the perforation cells:
⎪ s ⎣ ⎦ ⎪ s
⎨ ⎬
b
B = . ⎪⎩+ cosh ( az ) sinh ⎡⎣ a ( H − z ) ⎤⎦ ⎪⎭
s s
a
and
Specified Bottomhole Pressure. In the case of specified poten-
tial at the wellbore wf , the boundary conditions for a single perfo-
ration at the top of the formation will be qR =
sinh ⎡⎣ a ( H − z s ) ⎤⎦ cosh ( az s ) qk . . . . . . . . . . . (A-20)
⎪ ( ⎣ )
⎧sinh az s cosh ⎡ a ( H − z s ) ⎤ ⎫
⎦ ⎪
⎨ ⎬
⎛ ∂ ⎞
K z x 2 ⎜
⎝ ∂z ⎟⎠ z = 0
( )
= WI k k − wf . . . . . . . . . . . . . . . . . . . (A-9) ( )
⎪⎩+ cosh az s sinh ⎡⎣ a ( H − z s ) ⎤⎦ ⎪⎭
Eqs. A-15 and A-16 together with Eqs. A-19 and A-20 com-
Substituting Eqs. A-5a and A-5b into Eq. A-9 and using Eq. pletely define the analytical solution for a single perforation located
A-7, we obtain between the top and bottom of the formation. For convenience, we
⎛ ⎞ will use fL and fR as the fractions of the total rate to simplify Eqs.
⎜ ⎛b ⎞ ⎟ A-19 and A-20. By using this definition, Eq. A-19 and A-20 can
WI k ⎜ + wf ⎟ be shortened because qL = fLqk and qR = fRqk. Using the definition
b ⎜ ⎝a ⎠ ⎟
(z) = − − ⎜ ⎟ b
a ⎜ ⎧ K z x asinh asinh ( H − z s ) ⎫ ⎟
2 of B = − , Eqs. A-15 and A-16 can be written as
⎪ ⎪ a
⎜⎨ ⎬⎟
⎜⎝ ⎪+WI cosh ⎡⎣ asinh ( H − z s ) ⎤⎦ ⎪ ⎟⎠
⎩ k ⎭
L (z ) = B −
( az )
qL cosh
⎣ (
× cosh ⎡⎢ a H − z ⎤⎥ .
k ⎦ ) . . . . . . . . . . . . . . . . . . .(A-10) aK x sinh ( az )
z
2
s
( )
the volume-balance equation for the cell k in Eq. A-22a (Donnez
b qL cosh az 2007), we obtain
L (z ) = − − ,
a aK z x 2sinh az s ( )
⎛ x ⎞ f ⎛ q ⎞
ln ⎜ ⎟ = p ⎜ 1 − v ,k ⎟ . . . . . . . . . . . . . . . . . . . . . . (A-23)
for 0 < z < zs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-15) ⎝ ro ,3 D ,k ⎠ 2⎝ qk ⎠
⎡ ⎛ ⎧cosh ⎡ a ( H − z 2 ) ⎤ ⎫ ⎞ ⎤ ×
q j sinh ( )
a z j cosh ⎡⎣ a H − z j ( )⎤⎦
⎢
⎢
⎜ ⎪
⎨
⎣ ⎦ ⎪⎟ ⎥
⎬ ⎥
⎧sinh
⎪ ( ) (
az j cosh ⎡⎣ a H − z j ⎤⎦ ⎫
⎪ )
⎜ ⎪⎩− cosh ⎣ a ( H − z1 ) ⎤⎦ ⎪⎭ ⎟ ⎥
⎡ ⎨ ⎬
ro ,3 D ,k ⎢
= x exp ⎢ −
f p ,k
⎜ 1− ⎟ .
⎩⎪
+ cosh ( ) (
az j sinh ⎡⎣ a H − z j ⎤⎦ ⎪ )
2 ⎜ a zsinh aH ( ⎟⎥ ) ⎭
⎢
⎢
⎜
⎜
⎟⎥
⎟⎥ ×
( az )
cosh
+ z > zj ( )
⎢⎣ ⎝ ⎠ ⎥⎦ ⎡
⎣ aK x ysinh (
z az j ⎤⎦ )
. . . . . . . . . . . . . . . . . . . . . (A-26)
( k)
where
sinh az cosh ⎡⎣ a ( H − zk ) ⎤⎦
fL = k −1 − k = t ( z k −1 ) − t ( z k )
⎪ ( ⎣ )
⎧sinh az k cosh ⎡ a ( H − z k ) ⎤ ⎫
⎦ ⎪
⎨ ⎬
⎪⎩ + cosh az k(sinh ⎡
⎣ a ( H)− z k )⎦ ⎭
⎤⎪ and
k +1 − k = t ( z k +1 ) − t ( z k ). . . . . . . . . . . . . . . . . . (A-34)
and
The analytical solution is obtained by substituting Eq. A-31 into
sinh ⎡⎣ a ( H − z k ) ⎤⎦ cosh( az ) Eq. A-34 to obtain analytical expressions for k−1 − k and k+1 −
fR = k .
k. The resulting expression is substituted into Eq. A-33 and Eq.
⎪ ( ⎣ )
⎧sinh az k cosh ⎡ a ( H − z k ) ⎤ ⎫
⎦ ⎪ 16 to obtain the equivalent wellblock radius.
⎨ ⎬
( )
⎪⎩+ cosh az k sinh ⎡⎣ a ( H − z k ) ⎤⎦ ⎭⎪ Ali H. Dogru is the chief technologist of computational mod-
eling at Saudi Aramco’s Advanced Research Center. His pri-
The terms fL and fR are the fractions of the total flow rate qk. mary areas of interest are parallel reservoir simulation, reservoir
engineering, large-scale parameter estimation, and high-per-
qL = f L qk , qR = f R qk , f L + f R = 1. . . . . . . . . . . . . . . . . . . . (A-29) formance computing. He holds a PhD degree in petroleum
engineering with a minor in applied mathematics from The
Substituting Eq. A-28 into Eq. A-24, we obtain the equivalent University of Texas at Austin. His industrial experience is with
well radius. Core Labs and Mobil R&D in Dallas, Texas, and Saudi Aramco.
He was chairman of the SPE JPT Special Series Committee
2004–08. Dogru worked with various other SPE committees,
Multiple Perforations. Let the well have NC number of completions including Editorial Review and SPE Fluid Mechanics. Currently,
for the perforations j = 1, NC. We need to note that all perforations he is a director of the SPE R&D Technical Section Committee
do not have to be continuous. We may have a set of perforations and a member of JPT Special Series Committe. Dogru received
followed by a nonperforated interval and again followed by a new the 2008 SPE Reservoir Description and Dynamics award.