Comparison of Solutions to a
Three-Dimensional Black-Oil
Reservoir Simulation Problem
Aziz s. Odeh, SPE, Mobil Research and Development Corp.
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 Amdahl
470/V5 and IBM 370/168. Intercomp used Cray-l
and Harris17. Mobil and SSC used CDC Cyber 175,
and Shell used Univac 111012C 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 single-
point upstream mobility weighting. Shell used two
points upstream. Constraints and data are given in
the text.
Statement of the Problem
Areal and cross-section views of the reservoir are
given in Figs. 1 and 2. The grid system is given in Fig.
0149·2136/81/0001·9723$00.25
Copyright 1981 Society of Petroleum Engineers of AIME
JANUARY 1981
1. Stratification and reservoir properties are given in
Fig. 2. The reservoir is initially undersaturated. A gas
injection well is located at Grid Point (1, 1), and a
producing well is located at Grid Point (10, 10).
Pertinent data and constraints are given in Table 1.
PVT properties and relative permeabilities are given
in Tables 2 and 3. The participants were asked to
make the runs and report the results described below.
Runs To Be Made
Case 1
Let the bubble-point (saturation) pressure be con-
stant with a value equal to the original value.
Case 2
Let the saturation pressure vary with gas
saturation - 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
The following results are to be reported.
I.Plots of:
a. Oil rate vs. time.
b. GOR vs. time.
2.Report annually and at abandonment:
a. The pressures of the cell where the injector
and producer are located. *
b. Gas saturation at Grid Points (1, 1, 1), (1, 1,
2), (1, 1,3), (10, 1, 1), (10, 1,2), (10, 1,3), (10,
10, 1), (10, 10,2), and (10, 10,3).
3.Report at the end of 8 years:
a. Tables of gas saturation.
b. Tables of cell pressures. *
c. Tables of saturation pressures for the variable
saturations-pressure case. *
'Report of all pressures referred to a depth of 8,325 It or at the center of the
respective blocks. If both are available, report both.
13
 GAS INJECTION OIL PRODUCTION
100 MM SCFI D WELL

.,. H. FT. KX
K
Ky (LINK) Sw So
8325 FT.
LAYER 1 .3 20 500 500 . 12 .88 8335 FT •
50

LAYER 2 .3 30 50 50 . 12 .88 8380 FT .

25

LAYER 3 .3 50 200 200 . 12 .88 8400 FT •

,.
Ci=±=ti=.t,=~~_~~~~ I : J: 1 2 3 4 5 8 7 8 II 10
8425 FT.
1000 FT.

Fig. 1 - Reservoir and grid system . Fig. 2 - Diagonal cross section.

TABLE 1- DATA AND CONSTRAINTS

Initial reservoir pressure, psia at 8,400 ft 4,800

Gas injection rate, MMscfID 100
Maximum oil production rate, STBID 20,000
Minimum oil rate, STBID 1,000
Minimum flowing bottom hole pressure, psi 1,000
Maximum saturat ion change during time step 0.05
Rock compressibility, 1/psi 3x10- 6
Porosity value of 0.3 was measured at a base pressure of 14.7 psi
Wellbore radius, ft 0.25
Skin o
Capillary pressure o
Reservoir temperature, & F 200
Gas specific gravity 0.792
Runs are terminated either at the end of 10 years or when
GOR5/s 20,000 scf/STB or when the oil production rate51 ,000
STB/D; whichever occurs first terminates the run.

TABLE 2 - PVT PROPERTIES

Saturated Oil PVT Functions Saturated Water PVT Functions

Reservoir Reservoir
Pressure FVF Viscosity Density Solution GOR Pressure FVF Viscosity Density Gas/Water Ratio
(psi a) (RB/STB) (cp) (Ibm/cu ft) (scf/stb) (psi a) (RB/bbl) (cp) (Ibm/cu tt) (scf/bbl)
14.7 1.0620 1.0400 46.244 1.0 14.7 1.0410 0.3100 62.238 0.0
264.7 1.1500 0.9750 43.544 90.5 264.7 1.0403 0.3100 62.283 0.0
514.7 1.2070 0.9100 42.287 180.0 514.7 1.0395 0.3100 62.328 0.0
1014.7 1.2950 0.8300 41 .004 371 .0 1014.7 1.0380 0.3100 62.418 0.0
2014.7 1.4350 0.6950 38.995 636.0 2014.7 1.0350 0.3100 62.599 0.0
2514.7 1.5000 0.6410 38.304 775.0 2514.7 1.0335 0.3100 62.690 0.0
3014.7 1.5650 0.5940 37.781 930.0 3014.7 1.0320 0.3100 62.781 0.0
4014.7 1.6950 0.5100 37.046 1270.0 4014.7 1.0290 0.3100 62.964 0.0
5014.7 1.8270 0.4490 36.424 1618.0 5014.7 1.0258 0.3100 63.160 0.0
9014.7 2.3570 0.2030 34.482 2984.0 9014.7 1.0130 0.3100 63.959 0.0

Undersaturated Oil PVT Functions

Gas PVT Functions
Reservoir
Pressure FVF Viscosity Density Reservoir Pseudo Gas Potential
(psi a) (RB/STB) (cp) (Ibm/cu ft) Pressure FVF Viscosity Density M(f)
37.046 (ps ia) (RB/bbl) (cp) (Ibm/cu tt) (psia Icp)
4014.7 1.6950 0.5100
9014.7 1.5790 0.7400 39.768 14.7 0.166666 0.008000 0.0647 O.
264.7 0.012093 0.009600 0.8916 0.777916 E + 07
514.7 0.006274 0.011200 1.7185 0.267580 E + 08
Undersaturated Water PVT
1014.7 0.003197 0.014000 3.3727 0.875262 E-+ 08
Functions
2014.7 0.001614 0.018900 6.6806 0.270709 E + 09
Reservoir 2514.7 0.001294 0.020800 8.3326 0.386910 E + 09
Pressure FVF Viscosity Density 3014.7 0.001080 0.022800 9.9837 0.516118 E+09
(psia) (RB/bbl) (cp) (Ibm/cu tt) 4014.7 0.000811 0.026800 13.2952 0.803963 E + 09
4014.7 1.0290 0.3100 62.964 5014.7 0.000649 0.030900 16.6139 0.112256 E+10
9014.7 1.0130 0.3100 63.959 9014.7 0.000386 0.047000 27.9483 0.251845 E + 10

14 JOURNAL OF PETROLEUM TECHNOLOGY

 20

18
-MOBIL
16 - SHELL

- AMOCO
14
- INTERCOMP

~ 12 - EXXON
IXI
....
(/) ----- SSC
(II) 10
0
....
W- 8
!i
a:
...J
0 6
EXXON AGREES WITH MOBIL BETWEEN 3-7 YEARS
4
EXXON AGREES WITH INTERCOMP BETWEEN 8-10 YEARS
AMOCO AGREES WITH SHELL AFTER 7 YEARS
2
CMG AGREES WITH MOBIL THROUGH 7 YEARS AND
WITH INTERCOMP BETWEEN 8-10 YEARS
0
1 2 3 4 5 6 7 8 9 10 TIME, YEARS

Fig_ 3 - Case 1 - oil rate VS. time.

Results TABLE 3 - RELATIVE PERMEABILITY DATA'

A comparison of the results is given in Figs. 3 Oil·Gas

through 18. No comparison of saturation pressures is
given because the values reported by the seven ~ ~ ~
0 0.0 1.0
companies were within 20 psi of each other. 0.001 0.0 1.0
0.02 0.0 0.997
Description of the Simulators 0.05 0.005 0.980
0.12 0.025 0.700
Amoco's Model 0.2 0.075 0.350
The IMPES method was used, with semi-implicit 0.25 0.125 0.200
0.3 0.190 0.090
adjustments in well rates. This method proved quite 0.4 0.410 0.021
satisfactory; additional computations for implicit 0.45 0.60 0.010
handling of inter block flow were not needed. 0.5 0.72 0.001
Maximum time-step size can vary with time and is 0.6 0.87 0.0001
0.7 0.94 0.000
input. The model determines internal time-step sizes 0.85 0.98 0.000
to satisfy both the current maximum l1t and the 1.0 1.0 0.000
maximum saturation change for any grid block (5070 'This is a two·phase, gas/oil problem. Set the relative permeability to water
equal to zero for all values of water saturations.
PV). A sequence of runs using maximum l1t 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 l1t of
1.0 month.
For each internal time step the computation
sequence was as follows.
1. Well rates.
2. Coefficients including terms for semi-implicit
production rates.
3. Iterative computation of grid-block pressure
JANUARY 1981 15
 20

,.
16 -MOBIL

,. - - SHELL

_ AMOCO

...,m 12 - INTERCQMP

'".. 10 - - EXXON
<.> - .-. ~ sse
'"
"r£
0
•
" EXXON AGREES WITH MOBIL UP TO 3 YEARS
6 sse AGREES WITH MOBIL BETWEEN 2-7 YEARS

AMOCO AGREES WITH INTERCOMP AFTER 6 YEARS

4
CMG AGREES WITH MOBIL THROUGH 7 YEARS
AND WITH INTERCOMP BETWEEN 7-10 YEARS
2

0
1 2 3
• 5 6 7
Fig. 4 - Case' -GOR vs. time.
• 9 10 TIME , YEARS

INTERCOMP,
SHEll,
5500 AMOCO

- MOBIL - INTERCOMP
- SHELL _ EXXON
- AMOCO _._ . sse

AMOCO AGREES WITH SHEll UP TO 3 YEARS

4500 EXXON AGREES WITH AMOCO BETWEEN 6-10 YEARS

.'"
~

ui
sse AGREES WITH AMOCO BETWEEN 3-5 YEARS

II:
::>

'"'"
.
W
II:
3500
MOBil, INTERCOMI;P:::::::::::;::::~;·~-~.~-;;.~_~.;;_;:;.Z

CMG AGREES WITH MOBIL BETWEEN 1-4, AND 9 - 10 YEARS,

AND WITH AMOCO BETWEEN 5-8 YEARS

2500 +-_-+_ _~_~_~_ _+-_-+_ _~_~_-4_~

o 1 2 3 4 5 6 7 8 9 10
TIME, YEARS
Fig. 5 -Case 1 - pressure vs. time lor producing well Cell 10, 10, 3.

16 JOURNAL OF PETROl£UM TECHNOLOG Y

changes using slice successive overrelaxation.
4. Noniterative computation of grid-block sat-
uration changes and of semi-implicit adjustments in
production rates.
• 5. If any saturation change exceeds the maximum,
reduce !:J.t and go to Step 2.
6. Noniterative computation of grid-block bubble-
point pressure changes (for variable saturation-
pressure case only).
Production rates for each step were the sum of
rates at the start of the step plus semi-implicit ad-
justments as saturations changed in the well block.
An important exception is that the oil rate was held
constant if, at the start of the step, the well had
excess computed productivity (Le., if the computed
bottomhole pressure exceeded the minimum value of
1,000 psi).
CMG's Black-Oil Model
CMG's black-oil simulator models three-phase
water/oil/gas systems or two-phase water/oil
systems. The model includes the effects of gravity
and capillary pressure. It can be run in the one-, two-
or three-dimensional mode. Variable grid spacing
can be used. The nonlinear equations are solved by
Newtonian iteration with the derivatives of the
Jacobian matrix evaluated numerically. The model
contains several possible options for the weighting of
mobilities. These include single-point upstream, two-
point upstream, and centralized upstream
weightings. 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. MuItiblock 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 or-
dering, 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 weightings are used or when muitiblock
completion wells are 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 three-
phase, 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

30

- M O BIL
- SHELL, AMOCO
- INTERCOMP
-----SSC
I-
Z
w 20
~
o
a:
w
Q. INTERCOMP, SHELL, AMOCO
Z
o
~a:
::::l
ti:
(f) 10
EXXON AGREES WITH SHELL, AMOCO THROUGH 6 YEARS,
(f) AND WITH INTERCOMP BETWEEN 7-10 YEARS
<C
(!) CMG AGREES WITH MOBIL THROUGH 3 YEARS, AND
WITH INTERCOMP BETWEEN 4-10 YEARS
SSC AGREES WITH SHELL BETWEEN 4-9 YEARS

o+-----~----~----~----~--~~--~----~----~----_+----~
o 1 2 3 456 7 8 9 10
TIME, YEARS
Fig. 6 - Case 1 - gas saturation vs. time for producing wel l Cell 10, 10,3.

JANUARY 1981 17
 7500

-MOBIL
- SHELL
- INTERCOMP
- EXXON
6500 _......... CMG

...0;
.
w
AMOCO AGREES WITH SHELL UP TO 6 YEARS,
AND WITH EXXON AFTER 6 YEARS

a: SSC AGREES WITH SHELL BETWEEN 1-2

:>
YEARS, AND 4-5 YEARS,
~ 5500

..
w
a: CMG AGREES WITH
MOBIL UP TO 4 YEARS
WITH MOBIL BETWEEN 3-4 YEARS, AND
WITH EXXON BETWEEN 6-10 YEARS

MOBIL-INTERCOMP~-...

4500

MOBIL-INTERCOMP::;~~~::;;:
3900±-__~__~__~__~__~__~__~__~__~__~.
o 1 2 3 4 5 6 7 8 9 10 TIME , YEARS
Fig. 7 -Case 1 - pressurevs. time for injection well Cell 1, 1, 1.

4500
_.
.-.
-. .
. _._. _.
.
4000 ._.
..'"
:!

W
-MOBIL
- SHELL
- .-.-
a: - AMOCO
:>

'"'"a:w
- INTERCOMP

. 3500
- EXXON
._._. SSC

.-.
CMG AGREES WITH AMOCO

I~M
3000
1,1 2,2 3 ,3 4 ,4 5,5 6,6 7,7 8,8 9,9 10,10
GRID POINT LOCATION
Fig. 8 -Case 1 - pressure vs. grid-poinllocation, time= 8 years, top layer.

18 JOURNAL OF PETROL EUM TECHNOLOGY

 S= M

. .1.... sf
1
+ M
S
A'I ' E= M -MOBIL

50 SSC = E
......t......" ...., .•..+ ,... 1=I S -
-
SHELL
AMOCO
SSC , E=A - INTERCOMP
- EXXON
....
z ·_ · _ · SSC
w ........... CMG
0

.
0:
W

Z E=I
0
~
0:
45 CMG~ A ·············:1
::> E= M
t(
<II
<II sse= A

"" CMG ~ 1 '1' _._._.

CMG= A
40 ~~-+~~~~~~~-r~=-~~~-=~+-~~~~~~~
1,1 2,2 3 ,3 4,4 5,5 6,6 7,7 8 ,8 9,9 10,10
GRID POINT LOCATION
Fig. 9 - Case 1 - gas saturation vs. grid-point location , lime =8 years, top layer.

50 I, E:A

CMG: M
40 -MOBIL
SSC = A - SHELL
- AMOCO
....
Z - INTERCOMP
W
o - EXXON

.
0:
W 30

Z
···T····
E= CMG
· _·_ · SSC
..........• CMG slc,
CMG=
o
~ GAS SAT. < .005 AT:
0: 1) 4 ,4- SHELL, EXXON, AMOCO
'c::> 20 2) 5 ,5 -ALL PARTICIPANTS
<II 3) 6 ,6- ALL PARTICIPANTS EXCEPT MOBIL,
<II CMG
4 ) 7 .7_SHELL, AMOCO

"" 10
5} 8,8- SHELL

A, E, CMG= M
I, Ef A
CjG= M E= CjG I r SSC, CMG: M

0 ~~~~'~~"'~
1,1
" ""~"~'~1~~
2,2 3 ,3 4,4 5,5 6,6 7,7 8,8 9 ,9 10,10
GRID POINT LOCATION
Fig. 10 - Case 1 _ gas saturation vs. grid-point location, time=8 years, middle layer.

JANU ARY 1981 19

Exxon's General Purpose Simulator
Exxon's general purpose reservoir simulator (GP-
SIM) 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
boundary between grid blocks, and these total
velocities are 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 method 4 was used
to solve for pressures, and strongly implicit
procedure (SIP) was used to solve for saturations.
(The full saturation solution was needed because
mobilities 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 vapo-
rization 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 two-
dimensional ones. If the central memory made
available is sufficiently large, the program
automatically will eliminate the temporary data
storage, keeping all data within core.
Intercomp's Black-Oil Simulator
Intercomp's BETA II black~oil model is designed to
simulate numerically two- or three-phase com-
pressible 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 rec-
tangular or cylindrical coordinates. Alternate solu-
tion 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,
BET A II contains two distinct segments of code

20
EXXON,AMOCO
18
SHELL = MOBIL - M O BIL
- SHELL
16
- AMOCO
_._ . SSC
14
0
"-
a:I
~
(JJ 12
(\')

... 10
0

w
~
II:
8
-I
(5
6
INTERCOMP AGREES WITH MOBIL UP T O 7 YEARS,
4 AND WITH AMOCO BETWEEN 7-10 YEARS

EXXON AND AMOCO AGREE FOR 10 YEARS

2
CMG AGREES WITH AMOCO BETWEEN 5-8 YEARS,
AND WITH MOBIL BETWEEN 8-10 YEARS
0
1 2 3 4 5 6 7 8 9 10 TIME, YEARS
Fig. 11-Case2-oil ratevs. time.

20 JOURNAL OF PETROLEUM TECHNOLOGY

 20

18
-MOBIL
- SHELL
16 - AMOCO
- EXXON
14 _. _ . SSC

III
t- 12
,
III
U.
(,)
III 10
'"
ri
0 8 AMOCO AGREES WITH SHELL BETWEEN 8 - 10 YEARS
" EXXON AGREES WITH SHELL UP TO 4 YEARS. AND
6
, WITH MOBil BETWEEN 5-8 YEARS

, INTERCOMP AGREES WITH MOBIL UP TO 7 YEARS,

,,
4 AND WITH SHELL BETWEEN B-1 0 YEARS

CMG AGREES MOBIL THROUGH 9 YEARS

2 AT 10 YEAR GOR 20 ,000=
0
1 2 3 4 5 6 7 8 9 10 TIME. YEARS
Fig. 12-Case2-GORvs.lime.

6000

SHELL- -MOBil
AMOCO - SHELL
UP TO - AMOCO
4 YEARS
- EXXON
_. - . SSC
........... CMG
5000
~ INTERCOMP AGREES WITH SHELL UP TO
III
0. 4 YEARS, AND WITH MOBIL AFTER THAT
W CMG AGREES WITH MOBIL THROUGH
rr
::> 2 YEARS
III
III
W
rr
0.
4000

"-'-.
3000 +----+--~~--~--~----~---T--~~--~--~~~
o 1 2 3 4 5 6 7 8 9 10
TIME, YEARS
Fig. 13-Case 2 - pressure vs. time for producing well Cell 10, 10,3.

JANUARY 1981 21
which are designed specifically for optimal solution
of two-phase problems: one for water/oil (dissolved-
gas content constant) problems and one for
gas/water (no oil phase) problems. BETA II has a
large variety of user-oriented features such as in-
put/ output options and well control options.
There is a one-equation implicit pressure for-
mulation in which the equations are decoupled and
solved in this order: pressure, gas saturation, and
water saturation. For the one-equation formulation,
there are options to solve for both saturations ex-
plicitly (lMPES) and either or both saturations
implicitly (sequential). There is an alternative two-
equation formulation in which the program solves
implicitly for pressure and gas saturation and then
solves for water saturation. Similar to the previous
formulation, water saturation can be treated either
implicitly or explicitly. In any of the formulations,
multiple outer iterations may be taken to account for
the nonlinearity of the basic flow equations.
However, if no vaporization or resolution of gas is
occurring, only one or two iterations are required to
converge the nonlinearities adequately.
The large systems of linear algebraic equations
may be solved by a variety of methods, anyone of
which may offer significant speed advantages on a
given problem: (1) direct solution by reduced band-
width Gaussian elimination, (2) several forms of one-
line, two-line, and planar successive overrelaxation
(SOR), or (3) SIP.
Mobil's All Purpose Reservoir Simulator (ALPURS)
ALPURS is a three-dimensional, three-phase,
multiwell, black-oil reservoir simulator which uses a
strongly coupled, fully implicit method to solve
simultaneously for all unknowns. 5 The nonlinear
intercell flow equations and well-constraint
equations are linearized and iterated to converge
using Newton-Raphson iteration. Linear equations
are solved with block successive overrelaxation. A
typical block is an x-z, y-z, or x-y reservoir slice,
which is solved by sparse elimination. The relaxation
parameter is computed automatically using the
power method and Rayleigh quotients. ALPURS
accounts for reservoir heterogeneity, rock com-
pressibility, gravity, gas dissolved in both the oil and
water phases, constant or variable bubble-point
pressures, hysteresis in saturation-dependent data,
tubing string pressure drop, and flash surface
separation. Modern concepts of well flow equations
are incorporated, including pseudo gas-potential
function, skin factor to account for damage or
improvement, non-Darcy flow effect, and flow
restriction due to limited entry such as partial
penetration.
SSCModel
SSC's black-oil model employs an Adaptive Implicit
Method (AIM). This technique, which recently was
developed at SSC, seeks to achieve an optimum with

30

-MOBIL
- SHELL, AMOCO
- INTERCOMP
I-
_ ._ . SSC
Z
w
o 20
a:
w
Q.

Z
o
fi
a:
::>
t( EXXON AGREES WITH INTERCOMP BETWEEN
UJ 10 3-10 YEARS
UJ
< CMG AGREES WITH INTERCOMP
CJ

O +----+----+----+----~--_r----~---r----~--~--~
o 1 2 3 4 5 6 7 8 9 10
TIME, YEARS
Fig. 14 - Case 2 - gas saturation vs. ti me for produci ng well Ceil 10, 10, 3.

22 JOURNAL OF PETROLEUM TECHNOLOGY

 7500

-MOBIL
- SHELL
- AMOCO
- INTERCOMP
- EXXON
6500 _ . _ . SSC
.......... CMG

... MOBIL-SHELL-EXXON'.
\
EXXON AGREES WITH AMOCO
'" BETWEEN 6-10 YEARS

:i
::>
5500 SSC AGREES WITH MOBIL
AFTER 6 YEARS
'"'"a:w
.. CMG AGREES WITH EXXON
BETWEEN 6-10 YEARS

4500

3900' ~--+---~--~--~--~--~--~--~--~---:;
o 1 2 3 4 5 6 7 8 9 10 TIME, YEARS
Fig. 15 -Case 2 - pressure vs. time for injection well CellI, I , 1.

L
4500

...'" 4000
-MOBIL
- SHELL
ui - AMOCO
a:
::> - INTERCOMP

'"'"a:w
. - EXXON
_._" SSC
3500 .......... CMG

SSC AGREES WITH MOBIL AT 4,4 THROUGH 10, 10

CMG AGREES WITH EXXON AT 1,1 THROUGH 6 ,6

3000~~-+~~+-~~~~~~~~~-r~~~~-+~~~~~
1,1 2 ,2 3 ,3 4,4 5,5 6,6 7,7 8,8 9,9 10,10
GRID POINT LOCATION
Fig. 16 - Case 2 - pressure vs. grid-point location, time=8years, top layer.

JANUARY 1981 2l
respect to stability, truncation errors, and computer
costs. Typically, only a small fraction of the total
number of grid blocks during simulation experience
sufficiently large surges in pressure and/or saturation
to justify implicit treatment. When it is needed,
implicit treatment may not be required in all phases
or for long periods of time. Moreover, those cells
requiring implicit treatment will change as the
simulation proceeds. Consequently, a model offering
a fixed degree of implicitness to all cells for all time
steps is not always the most desirable. For example, a
fully implicit model, while ensuring stable answers,
amounts to overkill in most of the cells most of the
time, while an IMPES model can cause under kill.
With AIM there is no problem of over- or under kill.
Various degrees of implicitness are invoked
regionally or individually cell by cell- i.e; the
solution is advanced with adjacent cells having
different degrees of implicitness. As the calculations
proceed, the degrees of implicitness locally and
dynamically shift as needed - all automatically. The
whole idea is to apply the right amount of im-
plicitness where and when needed and for only as
long as needed.
The simulator also provides a wide variety of user-
oriented features. For example, one can override
AIM and operate in a fully implicit, partially im-
plicit, or an IMPES mode. Variable bubble-point
problems, such as that in Case 2, are handled by
variable substitution. The simulator is a general
purpose package offering three-dimensional capabil-
ity in both Cartesian and cylindrical coordinates.
Shell Development Model
The Shell reservoir simulation system operates an
IMPES mode or an implicit mode. There are three
pseudocomponents: water, stock-tank oil, and
separator gas. There are three phases: aqueous,
hydrocarbon liquid, and hydrocarbon vapor. The
aqueous phase contains water. The hydrocarbon
liquid and vapor phases can contain both oil and gas.
A fourth component is also available for modeling
polymer or carbon dioxide. There are several indirect
and direct solution methods as a user option. Ad-
ditionally, two-point upstream weighting is used to
calculate phase mobilities.
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 Vinsome, P.K.W.:
"Techniques for Fully Implicit Reservoir Simulation," paper
SPE 9302 presented at the SPE 55th Annual Technical Con-
ference and Exhibition, Dallas, Sept. 21-24, 1980.
3. Spillette, A.G., Hillestad, J.G., and Stone, H.L.: "A High-
Stability Sequential Solution Approach to Reservoir
Simulation," paper SPE 4542 presented at the SPE 48th
Annual Meeting, Las Vegas, Sept. 30-0ct. 3, 1973.

SSC:I
t
.............
A:~ 1~ S7 E

55
E: S .... ·1·- -MOBIL
CMG:A - SHELL
- AMOCO
I-
Z - INTERCOMP
W
0 - EXXON
IX:
W
Q.

Z 50
_....... ......
~.-
· _· _·SSC
··········· CMG
0
A:S
fi
IX:
::J
t:c
f/)
f/)
<
_ . .- _.. .,
CMG.L S

0 45 CMG:M
SSC : E
I . .- ssc: E
SSC, CMG: A 'f
1 I
CMG~ A + .............
. !
............
A, CMG : M
40
1, 1 2,2 3,3 4 ,4 5 ,5 6 ,6 7 ,7 8 ,8 9 ,9 10,1 0
GRID POINT LOCATION
Fig. 17 - Case 2 - gas saturation vs. grld·point location, time = 8 years, top layer.

24 JOURNAL OF PETROLEUM TECHNOLOGY

 .. 1= M

50
CMG= A '.
CMG= E -MOBIL
- SHELL
... 40 E,CMG=M
zw - AMOCO
o - INTERCOMP
-.~.J
a:
w
Q.
- EXXON
_._.- SSC
Z
o 30 ........... CMG A=I
t
~
a:
::)
GAS SAT. < .005 AT:
ti
rJ) 1) 4,4 AND 5,5-SHELL, AMOCO,
rJ) 20 SSC,EXXON
ct
~ 2) 4,4-CMG
3) 6,6-SHELL, ssc
10 S, I, E, CMG= M
SSC,S =E rSSC,CMG=M

o1-~-+~=-~~-l~~i.~. .~.~. ~. .~..~~~. I~~~...~.~..~.. ~'.·§···~·~·~~~·_~I~~~·:S~'A~'~E~=I

1,1 2,2 3,3 4,4 5,5 6,6 7,7 8,8 9,9 10,10
GRID POINT LOCATION
Fig. 18 - Case 2 - gas saturation vs. grid-point location, time == 8 years, middle layer.

4. Watts, J.W.: "A Conjugate Gradient-Truncated Direct Sg gas saturation

Method for the Iterative Solution of the Reservoir Simulation So oil saturation
Pressure Equation, " paper SPE 8252 presented at the SPE 54th
Annual Technical Conference and Exhibition, Las Vegas, Sept. Sw water saturation
23-26, 1979. ill time step
5. Bansal, P.P., Harper, J.L., McDonald, A.E., Moreland, E.E., <I> porosity
Odeh, AS., and Trimble, R.H.: "A Strongly Coupled, Fully
Implicit, Three-Dimensional, Three-Phase Reservoir
Simulator," paper SPE 8329 presented at the SPE 54th Annual SI Metric Conversion Factors
Technical Conference and Exhibition, Las Vegas, Sept. 23-26, bbl x 1.589 873 E-01 m3
1979.
cp x 1.0* E-03 Pa·s
Nomenclature cu ft x 2.831 685 E-02 m3
h = thickness of CF - 32)/1.8 °C
I = number of grid points in the x direction ft x 3.048* E-01 m
J = number of grid points in the y direction Ibm x 4.535 924 E-01 kg
k = permeability psi, psi a x 6.894 757 E + 00 kPa
k rg relative permeability to gas scf x 2.863 640 E - 02 stdm 3
k ro relative permeability to oil ·Conversion factor is exact.
JPT
kx permeability in the x direction
ky permeability in the y direction Original manuscript recieved in Society of Petroleum Engineers office Nov.
M(p) pseudo gas potential 18,1980. Paper (SPE 9723) accepted for publication Nov. 26,1980.

JANUARY 1981 25
Discussion of Comparison of Solutions
to a Three-Dimensional Black-Oil
Reservoir Simulation Problem
William Hurst, SPE, consultant
I read with interest the article by Aziz S. Odeh in the
Jan. 1981 issue of JPT (Pages 13-25). Although the
article is excellently illustrated with color-coded
curves for each of the contributors, I question what
this article accomplishes.
The same physical data are given to seven
programming groups to see if they could reproduce
the same results for the model under consideration.
This applied even for their mesh grid arrangements
shown in the paper in Figs. 1 and 2.
I recommend to Odeh and other authors that they
reproduce in their programming procedure the
analytical solutions of Morris Muskat and mine that
apply to reservoir performance. I have done this in
my text, Reservoir Engineering and Conformal
Mapping of Oil and Gas Fields in treating with
Original manuscript (SPE 9741) received in the Society of Petroleum
Engineers office Feb. 5,1981.
552
(SPE 9741)
conformal mapping to simulate reservoir per-
formance. This was the first thing that occurred to
me to test the validity of the method.
The results have been excellent for the transient
treatment of fluid flow for a well in an enclosed
circle, either at the center of the circle or offset in the
enclosure. In Table 4-2 of the text, I start with the
conformal rectangle, whereby I can reproduce
Muskat's solution that applies to the offset well in a
circle.
No one has made these comparisons, which should
be made available if we are to place validity on the
known methods of reservoir simulation published in
the literature. '
JPT
0149·2136/81/0003·9741$00.25
Copyright 1981 Society of Petroleum Engineers of AIME
JOURNAL OF PETROLEUM TECHNOLOGY