SPE Paper by A. S. Odeh For Exercise 3
SPE Paper by A. S. Odeh For Exercise 3
SPE Paper by A. S. Odeh For Exercise 3
Three-Dimensional Black-Oil
Reservoir Simulation Problem
Aziz S. Odeh, SPE,
Summary
A comparison of solutions to a three-dimensional
black-oil reservoir simulation problem is presented.
The test of the problem and a brief description of the
seven simulators used in the study are given.
Introduction
Seven companies participated in a reservoir
simulation project to compare the results obtained by
different black-oil simulators. The companies were
chosen to give a good cross section of the solution
methods used in the industry. The participants were
Amoco Production Co., Computer Modelling Group
of Calgary (CMG), Exxon Production Research Co.,
Intercomp Resource Development and Engineering
Inc., Mobil Research and Development Corp., Shell
Development Co., and Scientific Software Corp.
(SSC). The paper presents the text of the problem, a
comparison of results in graphical form, and a brief
description of each model. The descriptions were
supplied by the participants.
A variety of computers was used. Amoco used
IBM 3033, IBM 370/168, and Amdahl V/6. CMG
used Honeywell 6000 DPS, and Exxon used Amclahl
470/V5 and IBM 370/168. Intercomp used Cray-1
and Harris/7. Mobil and SSC used CDC Cyber 175,
and Shell used Univac 1110/2C Level 36. The
number of time steps and the central processor times
varied considerably. Those interested in the actual
values should contact the individual companies.
Except for Shell, all the participants used singlepoint upstream mobility weighting. Shell used two
points upstream. Constraints and data are given in
the text.
Runs To Be Made
case 1
Let the bubble-point (saturation) pressure be constant with a value equal to the original value.
Case 2
Let the saturation pressure vary with gas
satiation i.e., this is a variable saturation-pressure
case. The PVT lines at pressures above the calculated
saturation pressures are parallel to the original line.
Results To Be Reported
JANUARY 1981
13
x
QA5I
OIL PRODUCTION
WELL
QAS INJECTION
100 MM SCFID
k
LAYER I
H, FT.
.3
20
KX
Ky (LI~K) SW
600 600
GO
.12 .88
8326 FT.
( I 8336 FT.
.12 .aa
#~S360 FT.
.12 .88
FT.
50
.3
LAYER2
60
30
50
28
LAYER3
.3
200 200
60
1
WELL
1: J=123450
tooo FT.
FT.
78
temperature,
0.25
0
0
200
0.792
&F
264.7
514.7
1014.7
2014.7
2514.7
3014.7
4014.7
5014.7
9014.7
IWF
(RBISTB)
Density
(Ibm/cu ft)
viscosity
(Cp)
1.0620
1.0400
1.1500
1.2070
1.2950
1.4350
1.5000
1.5650
1.6950
1.8270
2.3570
0,9750
0.9100
0.8300
0.6950
0.6410
0.5940
0.5100
0.4490
0.2030
Solution GOR
(scflstb)
46.2444il.544
42.287
41.004
38.995
38.304
37.781
37.046
36.424
34.482
1.0
90:5
180.0
371.0
636.0
775.0
930.0
1270.0
1618.0
2984.0
Reservoir
Pressure
(psia)
FVF
(RBlbbl)
Viscosltv
(Cp) -
Densitv
(ibm/cu it)
Gas/Water Rat io
(scflbbl)
14.7
264.7
514.7
1014.7
2014.7
2514.7
3014.7
4014.7
5014.7
9014.7
1.0410
1.0403
1.0395
1.0380
1.0350
1.0335
1.0320
1.0290
1.0258
1.0130
0.3100
0.3100
0.3100
0.3100
0.3100
0.3100
0.3100
0.3100
0.3100
0.3100
62.238
62.283
62.328
62.418
62.599
62.690
62.781
62.964
63.160
63.959
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Rf3St3NOlr
Pressure
(psia)
4014.7
9014.7
FVF
(RBISTB)
1.6950
1,5790
Viscosity
(Cp)
Density
(ibm/cu ft)
0.5100
0.7400
37.046
39.768
FVF
(RBlbbl)
1.0290
1.0130
Viscosity
(Cp)
Density
(ibm/cu ft)
0.3100
0.3100
62.964
63.959
Ff(3St3NOlr
Pressure
(psia)
14.7
264.7
514.7
1014.7
2014.7
2514.7
3014.7
4014.7
5014.7
9014.7
FVF
(RBlbbl)
0.166666
0.012093
0.006274
0.003197
0.001614
0.001294
0.001080
0,000811
0.000649
0.000386
viscosity
(Cp)
0.008000
0,009600
0.011200
0.014000
0.018900
0.020800
0.022800
0.026800
0.030900
0.047000
Density
(ibm/cu ft)
0.0647
0.8916
1.7185
3.3727
6.8806
8,3326
9.9837
13.2952
16.6139
27.9483
0.
0.777916 E+ 07
0.267580 E+ 08
0.875262 E+ 08
0.270709 E+ 09
0.386910 E+ 09
0.518118 E+09
0.803963 E+ 09
0.112256 E+ 10
0.251845 E+ 10
! 6
14
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
---
SSc
2
0
YEARS
7 YEARSAND
YEARS
~~~
234567
10 TIME,YEARS
89
Fig. 3 - Case 1- oil rate vs. time.
Results
A comparison of the results is given in Figs. 3
through 18. No comparison of saturation pressures is
given because the values reported by the seven
companies were within 20 psi of each other.
Description of the Shuulators
Amocos Model
The IMPES method was used, with semi-implicit
adjustments in well rates. This method proved quite
satisfactory; additional computations for implicit
handling of interlock flow were not needed.
Maximum time-step size can vary with time and is
input. The model determines internal time-step sizes
to satisfy both the current maximum At and the
maximum saturation change for any grid block (5070
PV). A sequence of runs using maximum At of 0.25,
0.5, 1, and 2 months yielded virtually identical
results, confirming the applicability of the IMPES
method. The final results are for a maximum At of
1.0 month.
For each internal time step the computation
sequence was as follows.
1. Wellrates.
2. Coefficients including terms for semi-implicit
production rates.
3. Iterative computation of grid-block pressure
JANUARY 1981
3o
0.001
0.02
0.05
0.12
0.2
0.25
::
0.45
l::
0.7
0.65
1.0
[9
k
1.00
;:;
0.0
0.005
0.025
0.075
0.125
0.190
0.410
0.60
0.72
0.87
0.94
0.98
1.0
:::97
0.980
0.700
0.350
0.200
0.090
0.021
0.010
0.001
0.0001
0.000
0,000
0.000
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
---
Ssc
AFTER 6 YEARS
t
2345676
i
10 TIME, YEARS
9
Fig, 4 - Case 1- GOR vs. time.
INTERCOMP.
5500
MOBIL
INTERCOMP
SHELL
EXXON
AMOCO
-o-
Ssc
YEARS
3500
MOBIL, lNTERCOMP~
34
YEARS,
10
TIME,5YEARS6
Fig. 5 - Case 1- pressure vs. time for producing well Cell 10,10,3.
JOURNAL OF PETROLEUM TECHNOLOGY
,.
MOBIL
SHELL, AMOCO
lNTERC(WIP
--
SSC
point upstream,
and centralized upstream
weighings. The time discretization is by backward
differences with a modified Crank-Nicholson
method included as an option. The well model
permits the placing of wells at various positions in a
grid block. Multiblock completion wells are included
and are modeled in a manner which does not increase
the matrix bandwidth. Finally, an efficient solution
routine is included in the model. This routine
provides Gaussian elimination with block D4 ordering, a bandwidth-reducing option, and two
different iterative solutions methods: AB and
COMBINATIVE.1
The model is fully implicit in its basic formulation.
It becomes highly implicit, not fully implicit, when
the options for two-point upstream or centralized
upstream weighings are used or when multiblock
completion wellsare modeled.
Disappearance of the gas phase is not handled by
the conventional variable substitution technique but
by a novel pseudo solution-gas formulation.2 The
pseudo solution-gas formulation allows both variable
bubble-point problems and fixed bubble-point
problems to be handled in a simple manner.
For this problem the simulator was run in threephase, three-dimensional mode. The basic fully
implicit formulation
was used. The time
discretization was backward differences. The matrix
problem was solved by the AB iterative routine. 1
34
10
TIME, !EARS
Fig. $ - Case 1- gas
JANUARY 1981
saturation
vs.
7500T
6500
MOBIL
SHELL
INTERCOMP
--
EXXON
*o*oDw~=
CMG
AMOCO AGREES WITH SHELL UPT06
AND WITH EXXON AFTER 6 YEARS
YEARS,
MOBIL-INTERCOMP
4500 MOBIL-INTERCOMP
39004
t
6709
012345
Fig. 7 -Casel-
pressurevs,timeforinjection
i
10 TIME, YEARS
wellCelll,
l,l,
4500
----
-1
)----
.
------
----
1-----
.-.....
,-----
-.
. ...-
.-*-
.-. -..
4-s-.
-1
1
I
. .. .. .... 1
MOBIL
SHELL
Nhloco
INTERCOMP
EXXON
--
Ssc
%
I
---.QI
I=M
3000
1,1
2,2
3,3
4,4
5,5
6,6
T,7
8,8
qg
10,10
SyM
v ......-
A, S,E=M
=
--
1=s
t:J&ti
sSC, E= A
m
1
- 1=s
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
SSC
-=~=~m=
CMG
A=M
S,Af
T
CMG = A ,
M
1
t
----..............
E=l
+
E=M
E=M
v
40411:22:33:44:
s
~5~
66:77:88:9,9
GRID P&NT
L06ATION
10,10
Fig. 9 - Case 1- gas saturation vs. grid-point location, time= 8 years, top layer..
I,Ez
T
G=M
-.
CMG = M
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
--
SSC
=s=CMG
SSC = A
&
.........*.
1
E = CMG
1
Ssc,
CMG = I
AT:
PARTICIPANTS
CMG : M
I
,
1,1
2,2
10,10
3,3
10-
vs. grid.point
location,
time=
8 years,
middle
layer.
19
Exxons general purpose reservoir simulator (GPSIM) uses a sequential implicit solution procedure.3
The first step in this approach is the solution of a set
of pressure equations, This set consists of a single
equation for each grid block, and solving it yields a
complete new pressure distribution at the end of a
time step.
This pressure distribution then is used to calculate
the sum of the velocities of all phases at each
brxm!ary between grid blocks, and these total
vclncities we used in a set of saturation equations. If
either capillary pressure or relative permeability is
being treated semi-implicitly, this set consists of two
coupled equations per grid block and is solved
simultaneously to yield saturation distributions at the
new time, Otherwise, the equations are uncoupled
and can be solved point by point explicitly, in the
normal IMPES fashion.
Several options are available for solving the
matrices involved. In the problem discussed here, a
preconditioned conjugate gradient method4 was used
to solve for pressures, and strongly implicit
procedure (SIP) was used to solve for saturations.
(The full saturation solution was needed because
nobilities were treated semi-implicitly.)
As is common in modern reservoir simulators,
GPSIM can account for reservoir heterogeneity, rock
compressibility, and solution of gas in both oil and
water. Less common features it can model are vaporization of oil into the gas phase and hysteresis in the
capillary pressure and relative permeability data.
GPSIM has only minor restrictions on the number
of grid blocks that it can use; large problems can be
run using only relatively modest amounts of central
memory. This desirable feature is accomplished by
using disks to store data temporarily by planes for
three-dimensional problems or by rows for twodimensional ones. If the central memory made
available is sufficiently large, the program
automatically will eliminate the temporary data
storage, keeping all data within core.
Intercomps Black-Oil Simulator
lntercomps BETA 11black:oil model is designed to
simulate numerically two- or three-phase compressible flow in heterogeneous hydrocarbon
reservoirs. Gas is assumed to be soluble in oil but not
in water; neither oil nor water can exist in any phase
other than its own. Solutions are obtained in one,
two, or three spatial dimensions using either rectangular or cylindrical coordinates. Alternate solution procedures provide for efficient modeling of all
classes of black-oil reservoir problems, ranging from
individual well behavior (coning simulations or well
test analysis) to large, multireservoir fields. To
complement the three-phase simulation capabilities,
BETA 11 contains two distinct segments of code
-nzlb-w...
INTERCOMP :CM
SHELL = MOBIL
MOBIL
SHELL
AMOCO
----
Ssc
,
,
2345678
*
Fig.
11-
MOBIL
SHELL
hMOCO
EXXON
--
Ssc
/*
YEARS
~
!
I
~
,,
2345678
10 TIME, YEARS
9
Fig. 12- Case 2-GOR vs. time.
MOBIL
SHELL
AMOCO
-
EXXON
--0
Ssc
*.~~**~o==
CMG
INTERCOMP AGREES WITH SHELL UP TO
4 YEARS, AND WITH MOBIL AFTER THAT
CMCI AGREES WITH MOBIL THROUGH
2 YEARS
Fig.
JANUARY 1981
13-
3
Case 2-pressure
,
,
6
5
TIME, YEARS
vs. time for producing
10
MOBiL
SHELL, AMOCO
iNTERCOMP
-=-
Ssc
1
1
I
2
I
34
,
6
,
7
,
8
#
9
1
10
TiME, tEAFiS
..
7600
MOBIL
SHELL
_
AMOCO
INTERCOMP
EXXON
--
Ssc
9==o====M
CMG
6500
MOBIL-SHELL-EXXON\.
EXXON AGREES WITH AMOCO
BETWEEN 6-10 YEARS
SSC AGREES WITH MOBIL
AFTER 6 YEARS
CMG AGREES WITH EXXON
BETWEEN 6-10 YEARS
t
*
10 TIME, YEARS
2
3
4
5
6
7
8
9
Fig. 15-Case 2-pressure vs. tlmefor lnJectlon well Celll,l,l.
4s00
..-.
--
AMOCO
INTERCOMP
1--z ---EXXON
350a
MOBIL
M-u-SHELL
wU
a
m
a
w
a
a
SSC
====-== CMG
1,1
2,2
3,3
4,4
5,5 6,6
7,7
GRID POINT LOCATION
vs. gridpoint
8,8
9,9
1 O,1O
References
1. Behie, A. and Vinsome, P.K. W.: Block Iterative Methods for
Fully Implicit Reservoir Simulation,
paper SPE 9303
presented at the SPE 55th Annual Technical Conference and
Exhibition, Dallas, Sept. 21-24,1980,
2. Au, A. D. K., Behie, A,, Rubin, B., and V]nsome, P. K,W,:
Techniques for Fully Implicit Reservoir Simulation, paper
SPE 9302 presented at the SPE 55th Annual Technical Conference and Exhibition, Dallas, Sept. 21-24, 1980.
3. Spillette, A.G., Hillestad, J.G., and Stone, H. L.: A HQhStability Sequential Solution
Approach
to Reservoir
Simulation,paper SPE 4542 presented at the SPE 48th
Annual Meeting, Las Vegas, Sept. 30-Ott. 3,1973.
* S= E
:M
~~s.O-
55
CMG k A
...*. ......
---R
AiS
50
45
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
--
SSC
=o*-8nm9m*
CMG
-.R
CMG = S
SSC, CMG = A
CMG = A
4a
1,1
1
2,2
3,3
4,4
............
5,5
6,6
7,7
8,8
9,9
10,10
IVOBIL
SHELL
AMOCO
INTERCOMP
EXXON
S=l
I=L
=
E, CM(3 ~ M
-.
---SSC
0=0===0==~
CMG
A=l
Ssc
S, 1, E, CM(3 = M
SSC, S = E
E=A
r- Ss
L1
1,1
Fig.
2,2
* 3,3
It) - Case 2-
gas saturat
Nomenclature
h = thickness
I = number of grid points in the ~ direction
= number of grid points in they direction
L=relative
permeability
permeability togas
permeability to oil
~==reIative
permeability in the x direction
=permeability in they direction
kx
ky
M(p)
=
JANUARY 1981
.CMG= M
Iocatlon,
8,8
time =8 years,
middle
layer.
od saturation
water saturation
time step
porosity
x 1.589873 E-01
E-03
x 1.O*
x 2.831 685 E-02
(eF-32)/l.8
E-01
X 3.048
x 4.535924 E-01
x 6.894757 E+ 00
x 2.863 640 E 02
=
=
=
=
=
=
=
=
m3
Pans
m3
c
m
kg
kPa
std m3
Conversionfaxrorisexact.
25