Comparison of Solutions to a
ThreeDimensional BlackOil
Reservoir Simulation Problem
Aziz s. Odeh, SPE, Mobil Research and Development Corp.
Summary
A comparison of solutions to a threedimensional
blackoil 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 blackoil 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 Crayl
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 singlepoint upstream mobility weighting. Shell used two
points upstream. Constraints and data are given in
the text.
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 bubblepoint (saturation) pressure be constant with a value equal to the original value.
Case 2
Let the saturation pressure vary with gas
saturation  i.e., this is a variable saturationpressure
case. The PVT lines at pressures above the calculated
saturation pressures are parallel to the original line.
Results To Be Reported
Areal and crosssection views of the reservoir are
given in Figs. 1 and 2. The grid system is given in Fig.
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
saturationspressure case. *
01492136/81/00019723$00.25
Copyright 1981 Society of Petroleum Engineers of AIME
'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.
Statement of the Problem
JANUARY 1981
13
OIL PRODUCTION
WELL
GAS INJECTION
100 MM SCFI D
LAYER 1
.,.
H. FT.
.3
20
K
Ky (LINK) Sw
KX
500 500
So
8325 FT.
. 12 .88
8335 FT
.88
8380 FT .
. 12 .88
8400 FT
50
.3
LAYER 2
30
. 12
50
50
25
LAYER 3
,.
Ci==ti=.t,=~~_~~~~
I : J: 1
1000 FT.
.3
50
200 200
8425 FT.
4
II
10
Fig. 2  Diagonal cross section.
Fig. 1  Reservoir and grid system .
TABLE 1 DATA AND CONSTRAINTS
4,800
100
20,000
1,000
1,000
0.05
Initial reservoir pressure, psia at 8,400 ft
Gas injection rate, MMscfID
Maximum oil production rate, STBID
Minimum oil rate, STBID
Minimum flowing bottom hole pressure, psi
Maximum saturat ion change during time step
Rock compressibility, 1/psi
Porosity value of 0.3 was measured at a base pressure of 14.7 psi
Wellbore radius, ft
Skin
Capillary pressure
Reservoir temperature, & F
Gas specific gravity
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.
3x10 6
0.25
o
o
200
0.792
TABLE 2  PVT PROPERTIES
Saturated Oil PVT Functions
Reservoir
Pressure
(psi a)
14.7
264.7
514.7
1014.7
2014.7
2514.7
3014.7
4014.7
5014.7
9014.7
FVF
(RB/STB)
Viscosity
(cp)
Density
(Ibm/cu ft)
1.0620
1.1500
1.2070
1.2950
1.4350
1.5000
1.5650
1.6950
1.8270
2.3570
1.0400
0.9750
0.9100
0.8300
0.6950
0.6410
0.5940
0.5100
0.4490
0.2030
46.244
43.544
42.287
41 .004
38.995
38.304
37.781
37.046
36.424
34.482
Saturated Water PVT Functions
Solution GOR
(scf/stb)
1.0
90.5
180.0
371 .0
636.0
775.0
930.0
1270.0
1618.0
2984.0
Reservoir
Pressure
(psi a)
FVF
(RB/bbl)
Viscosity
(cp)
Density
(Ibm/cu tt)
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
Gas/Water Ratio
(scf/bbl)
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
Undersaturated Oil PVT Functions
Reservoir
Pressure
(psi a)
4014.7
9014.7
Gas PVT Functions
FVF
(RB/STB)
1.6950
1.5790
Viscosity
(cp)
0.5100
0.7400
Density
(Ibm/cu ft)
37.046
39.768
Undersaturated Water PVT
Functions
Reservoir
Pressure
(psia)
FVF
(RB/bbl)
Viscosity
(cp)
4014.7
9014.7
1.0290
1.0130
0.3100
0.3100
14
Density
(Ibm/cu tt)
62.964
63.959
Reservoir
Pressure
(ps ia)
FVF
(RB/bbl)
Viscosity
(cp)
Density
(Ibm/cu tt)
14.7
264.7
514.7
1014.7
2014.7
2514.7
3014.7
4014.7
5014.7
9014.7
0.166666
0.012093
0.006274
0.003197
0.001614
0.001294
0.001080
0.000811
0.000649
0.000386
0.008000
0.009600
0.011200
0.014000
0.018900
0.020800
0.022800
0.026800
0.030900
0.047000
0.0647
0.8916
1.7185
3.3727
6.6806
8.3326
9.9837
13.2952
16.6139
27.9483
Pseudo Gas Potential
M(f)
(psia Icp)
O.
0.777916
0.267580
0.875262
0.270709
0.386910
0.516118
0.803963
0.112256
0.251845
E + 07
E + 08
E+ 08
E + 09
E + 09
E+09
E + 09
E+10
E + 10
JOURNAL OF PETROLEUM TECHNOLOGY
20
18
MOBIL
16
14
IXI
12
....
SHELL
AMOCO
INTERCOMP
EXXON
 SSC
(/)
(II)
10
0
....
W 8
!i
a:
...J
6
EXXON AGREES WITH MOBIL BETWEEN 37 YEARS
4
EXXON AGREES WITH INTERCOMP BETWEEN 810 YEARS
AMOCO AGREES WITH SHELL AFTER 7 YEARS
CMG AGREES WITH MOBIL THROUGH 7 YEARS AND
WITH INTERCOMP BETWEEN 810 YEARS
10 TIME, YEARS
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 Simulators
Amoco's Model
The IMPES method was used, with semiimplicit
adjustments in well rates. This method proved quite
satisfactory; additional computations for implicit
handling of inter block flow were not needed.
Maximum timestep size can vary with time and is
input. The model determines internal timestep sizes
to satisfy both the current maximum l1t and the
maximum saturation change for any grid block (5070
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 semiimplicit
production rates.
3. Iterative computation of gridblock pressure
JANUARY 1981
TABLE 3  RELATIVE PERMEABILITY DATA'
OilGas
0
0.001
0.02
0.05
0.12
0.2
0.25
0.3
0.4
0.45
0.5
0.6
0.7
0.85
1.0
~ ~
0.0
0.0
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
1.0
1.0
0.997
0.980
0.700
0.350
0.200
0.090
0.021
0.010
0.001
0.0001
0.000
0.000
0.000
'This is a twophase, gas/oil problem. Set the relative permeability to water
equal to zero for all values of water saturations.
15
20
,.
...,m
16
MOBIL
,.
  SHELL
AMOCO
12
INTERCQMP
'".. 10
  EXXON
<.>
 .. ~
'"
"r
"
sse
EXXON AGREES WITH MOBIL UP TO 3 YEARS
sse
AGREES WITH MOBIL BETWEEN 27 YEARS
AMOCO AGREES WITH INTERCOMP AFTER 6 YEARS
4
CMG AGREES WITH MOBIL THROUGH 7 YEARS
AND WITH INTERCOMP BETWEEN 710 YEARS
0
1
10 TIME , YEARS
Fig. 4  Case' GOR vs. time.
INTERCOMP,
5500
SHEll,
AMOCO

MOBIL
SHELL
AMOCO
INTERCOMP
_
EXXON
_._ . sse
AMOCO AGREES WITH SHEll UP TO 3 YEARS
.'"
4500
EXXON AGREES WITH AMOCO BETWEEN 610 YEARS
sse
AGREES WITH AMOCO BETWEEN 35 YEARS
ui
II:
::>
'"'"
.
W
II:
3500
MOBil,
INTERCOMI;P:::::::::::;::::~;~~.~;;.~_~.;;_;:;.Z
CMG AGREES WITH MOBIL BETWEEN 14, AND 9  10 YEARS,
AND WITH AMOCO BETWEEN 58 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 PETROlUM TECHNOLOG Y
changes using slice successive overrelaxation.
4. Noniterative computation of gridblock saturation changes and of semiimplicit adjustments in
production rates.
5. If any saturation change exceeds the maximum,
reduce !:J.t and go to Step 2.
6. Noniterative computation of gridblock bubblepoint pressure changes (for variable saturationpressure case only).
Production rates for each step were the sum of
rates at the start of the step plus semiimplicit adjustments 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).
point upstream, and centralized upstream
weightings. The time discretization is by backward
differences with a modified CrankNicholson
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 ordering, a bandwidthreducing 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 twopoint 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 solutiongas formulation. 2 The
pseudo solutiongas formulation allows both variable
bubblepoint problems and fixed bubblepoint
problems to be handled in a simple manner.
For this problem the simulator was run in threephase, threedimensional 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
CMG's BlackOil Model
CMG's blackoil simulator models threephase
water/oil/gas systems or twophase water/oil
systems. The model includes the effects of gravity
and capillary pressure. It can be run in the one, twoor threedimensional 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 singlepoint upstream, two
30
 M O BIL

SHELL, AMOCO
INTERCOMP
SSC
IZ
w 20
a:
INTERCOMP, SHELL, AMOCO
Q.
~a:
::::l
ti:
(f)
EXXON AGREES WITH SHELL, AMOCO THROUGH 6 YEARS,
AND WITH INTERCOMP BETWEEN 710 YEARS
10
(f)
<C
(!)
CMG AGREES WITH MOBIL THROUGH 3 YEARS, AND
WITH INTERCOMP BETWEEN 410 YEARS
SSC AGREES WITH SHELL BETWEEN 49 YEARS
o+~~~~~~~~~_+~
456
TIME, YEARS
10
Fig. 6  Case 1  gas saturation vs. time for producing wel l Cell 10, 10,3.
JANUARY 1981
17
7500
MOBIL

SHELL
INTERCOMP
EXXON
_......... CMG
6500
...0;
AMOCO AGREES WITH SHELL UP TO 6 YEARS,
AND WITH EXXON AFTER 6 YEARS
a:
SSC AGREES WITH SHELL BETWEEN 12
YEARS, AND 45 YEARS,
WITH MOBIL BETWEEN 34 YEARS, AND
WITH EXXON BETWEEN 610 YEARS
:>
~ 5500
a:
..
CMG AGREES WITH
MOBIL UP TO 4 YEARS
MOBILINTERCOMP~...
4500
MOBILINTERCOMP::;~~~::;;:
3900__~__~__~__~__~__~__~__~__~__~.
10 TIME , YEARS
Fig. 7 Case 1  pressurevs. time for injection well Cell 1, 1, 1.
4500
_.
..
..'"
_._. _.
4000
:!
._.
 ..
MOBIL
W
a:
:>
'"'"a:w
.
SHELL
AMOCO
INTERCOMP
EXXON
._._. SSC
3500
..
CMG AGREES WITH AMOCO
3000
1,1
2,2
3 ,3
I~M
4 ,4
5,5
6,6
7,7
GRID POINT LOCATION
8,8
9,9
10,10
Fig. 8 Case 1  pressure vs. gridpoinllocation, time= 8 years, top layer.
18
JOURNAL OF PETROL EUM TECHNOLOGY
S= M
. .1.... sf
1
50
S
A'I ' E= M
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
_ _ SSC
........... CMG
......t......" ...., ...+ ,... 1=I S
SSC = E
SSC , E=A
....
z
w
0
.
0:
W
E=I
CMG~ A
45
0:
:1
::>
E= M
t(
<II
<II
sse=
""
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. gridpoint location , lime =8 years, top layer.
50
I, E:A
CMG: M
40
MOBIL
SHELL
AMOCO
INTERCOMP
EXXON
__ SSC
.......... CMG
SSC = A
....
Z
0:
W 30
o
~
0:
'c::>
CMG=
GAS SAT. < .005 AT:
1) 4 ,4 SHELL, EXXON, AMOCO
20
2) 5 ,5 ALL PARTICIPANTS
3) 6 ,6 ALL PARTICIPANTS EXCEPT MOBIL,
CMG
4 ) 7 .7_SHELL, AMOCO
<II
<II
""
slc,
E= CMG
5} 8,8 SHELL
10
A, E, CMG= M
I, Ef A
CjG= M
E= CjG
r SSC, CMG: M
0 ~~~~'~~"'~
" ""~"~'~1~~
1,1
2,2
3 ,3
4,4
5,5
6,6
7,7
GRID POINT LOCATION
8,8
9 ,9
10,10
Fig. 10  Case 1 _ gas saturation vs. gridpoint location, time=8 years, middle layer.
JANU ARY 1981
19
Exxon's General Purpose Simulator
Exxon's 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
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 semiimplicitly, 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 semiimplicitly.)
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
threedimensional 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.
Intercomp's BlackOil Simulator
Intercomp's BETA II black~oil model is designed to
simulate numerically two or threephase 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 blackoil reservoir problems, ranging from
individual well behavior (coning simulations or well
test analysis) to large, multireservoir fields. To
complement the threephase simulation capabilities,
BET A II contains two distinct segments of code
20
EXXON,AMOCO
18
SHELL
= MOBIL
 M O BIL
16
SHELL
AMOCO
_._ . SSC
14
0
"
a:I
~
(JJ
12
(\')
... 10
0
~
II:
I
(5
6
INTERCOMP AGREES WITH MOBIL UP T O 7 YEARS,
AND WITH AMOCO BETWEEN 710 YEARS
EXXON AND AMOCO AGREE FOR 10 YEARS
2
0
CMG AGREES WITH AMOCO BETWEEN 58 YEARS,
AND WITH MOBIL BETWEEN 810 YEARS
10 TIME, YEARS
Fig. 11Case2oil ratevs. time.
20
JOURNAL OF PETROLEUM TECHNOLOGY
20
18
MOBIL
16
AMOCO
EXXON
12
,
U.
(,)
10
III
'"
ri
0
SHELL
_. _ . SSC
14
III
tIII
"
AMOCO AGREES WITH SHELL BETWEEN 8  10 YEARS
EXXON AGREES WITH SHELL UP TO 4 YEARS. AND
WITH MOBil BETWEEN 58 YEARS
,
,
INTERCOMP AGREES WITH MOBIL UP TO 7 YEARS,
AND WITH SHELL BETWEEN B1 0 YEARS
,,
4
2
CMG AGREES MOBIL THROUGH 9 YEARS
AT 10 YEAR GOR 20 ,000
0
1
10 TIME. YEARS
Fig. 12Case2GORvs.lime.
6000
MOBil
SHELL
AMOCO
EXXON
SHELL
AMOCO
UP TO
4 YEARS
_.  . SSC
........... CMG
~
5000
INTERCOMP AGREES WITH SHELL UP TO
4 YEARS, AND WITH MOBIL AFTER THAT
III
0.
CMG AGREES WITH MOBIL THROUGH
2 YEARS
rr
::>
III
III
W
rr
0.
4000
"'.
3000 ++~~~~~T~~~~~~
o
1
2
3
4
5
6
7
8
9
10
TIME, YEARS
Fig. 13Case 2  pressure vs. time for producing well Cell 10, 10,3.
JANUARY 1981
21
which are designed specifically for optimal solution
of twophase problems: one for water/oil (dissolvedgas content constant) problems and one for
gas/water (no oil phase) problems. BETA II has a
large variety of useroriented features such as input/ output options and well control options.
There is a oneequation implicit pressure formulation in which the equations are decoupled and
solved in this order: pressure, gas saturation, and
water saturation. For the oneequation formulation,
there are options to solve for both saturations explicitly (lMPES) and either or both saturations
implicitly (sequential). There is an alternative twoequation 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 bandwidth Gaussian elimination, (2) several forms of oneline, twoline, and planar successive overrelaxation
(SOR), or (3) SIP.
Mobil's All Purpose Reservoir Simulator (ALPURS)
ALPURS is a threedimensional, threephase,
multiwell, blackoil reservoir simulator which uses a
strongly coupled, fully implicit method to solve
simultaneously for all unknowns. 5 The nonlinear
intercell flow equations and wellconstraint
equations are linearized and iterated to converge
using NewtonRaphson iteration. Linear equations
are solved with block successive overrelaxation. A
typical block is an xz, yz, or xy 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 compressibility, gravity, gas dissolved in both the oil and
water phases, constant or variable bubblepoint
pressures, hysteresis in saturationdependent data,
tubing string pressure drop, and flash surface
separation. Modern concepts of well flow equations
are incorporated, including pseudo gaspotential
function, skin factor to account for damage or
improvement, nonDarcy flow effect, and flow
restriction due to limited entry such as partial
penetration.
SSCModel
SSC's blackoil 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
_ ._ . SSC
IZ
a:
20
Q.
fi
a:
::>
t(
UJ
UJ
EXXON AGREES WITH INTERCOMP BETWEEN
310 YEARS
10
<
CMG AGREES WITH INTERCOMP
CJ
O ++++~_r~r~~~
5
6
TIME, YEARS
10
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
_ . _ . SSC
.......... CMG
6500
...
'"
:i
MOBILSHELLEXXON'.
\
EXXON AGREES WITH AMOCO
BETWEEN 610 YEARS
SSC AGREES WITH MOBIL
AFTER 6 YEARS
5500
::>
'"'"a:w
..
CMG AGREES WITH EXXON
BETWEEN 610 YEARS
4500
3900' ~+~~~~~~~~:;
10 TIME, YEARS
Fig. 15 Case 2  pressure vs. time for injection well CellI, I , 1.
L
4500
...'"
4000
MOBIL

SHELL
ui
a:
AMOCO
::>
'"'"a:w
INTERCOMP
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
GRID POINT LOCATION
8,8
9,9
10,10
Fig. 16  Case 2  pressure vs. gridpoint 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 implicitness where and when needed and for only as
long as needed.
The simulator also provides a wide variety of useroriented features. For example, one can override
AIM and operate in a fully implicit, partially implicit, or an IMPES mode. Variable bubblepoint
problems, such as that in Case 2, are handled by
variable substitution. The simulator is a general
purpose package offering threedimensional capability 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, stocktank 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. Additionally, twopoint 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. 2124, 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 Conference and Exhibition, Dallas, Sept. 2124, 1980.
3. Spillette, A.G., Hillestad, J.G., and Stone, H.L.: "A HighStability Sequential Solution Approach to Reservoir
Simulation," paper SPE 4542 presented at the SPE 48th
Annual Meeting, Las Vegas, Sept. 300ct. 3, 1973.
SSC:I
.............
A:~
55
1~
E: S ....
S7 E
1
MOBIL
CMG:A
I
_....... ......
Q.
~.
Z 50
SHELL
AMOCO
INTERCOMP
EXXON
_ _SSC
IX:
CMG
A:S
fi
IX:
::J
t:c
f/)
<
_.. .,
CMG.L S
_ . .
f/)
45
CMG:M
I
SSC, CMG: A
. .
A, CMG : M
1, 1
2,2
3,3
4 ,4
5 ,5
6 ,6
ssc: E
I
'f
CMG~ A
40
SSC : E
7 ,7
.............
8 ,8
............
9 ,9
10,1 0
GRID POINT LOCATION
Fig. 17  Case 2  gas saturation vs. grldpoint location, time = 8 years, top layer.
24
JOURNAL OF PETROLEUM TECHNOLOGY
..
1= M
50
CMG= A
'.
MOBIL
CMG= E
... 40
zw
o
a:
SHELL
AMOCO
INTERCOMP
E,CMG=M
.~.J
EXXON
_._. SSC
Q.
........... CMG
30
A=I
a:
::)
GAS SAT. < .005 AT:
ti
1) 4,4 AND 5,5SHELL, AMOCO,
SSC,EXXON
2) 4,4CMG
rJ)
20
rJ)
ct
3) 6,6SHELL,
10
ssc
S, I, E, CMG= M
SSC,S =E
rSSC,CMG=M
o1~+~=~~l~~i.~. .~.~. ~. .~..~~~. I~~~...~.~..~.. ~'.~~~~~_~I~~~:S~'A~'~E~=I
4,4
5,5
6,6
7,7
8,8
9,9
10,10
GRID POINT LOCATION
Fig. 18  Case 2  gas saturation vs. gridpoint location, time == 8 years, middle layer.
1,1
2,2
3,3
4. Watts, J.W.: "A Conjugate GradientTruncated Direct
Method for the Iterative Solution of the Reservoir Simulation
Pressure Equation, " paper SPE 8252 presented at the SPE 54th
Annual Technical Conference and Exhibition, Las Vegas, Sept.
2326, 1979.
5. Bansal, P.P., Harper, J.L., McDonald, A.E., Moreland, E.E.,
Odeh, AS., and Trimble, R.H.: "A Strongly Coupled, Fully
Implicit, ThreeDimensional, ThreePhase Reservoir
Simulator," paper SPE 8329 presented at the SPE 54th Annual
Technical Conference and Exhibition, Las Vegas, Sept. 2326,
1979.
Nomenclature
h
I
J
k
k rg
k ro
kx
ky
M(p)
=
=
=
=
thickness
number of grid points in the x direction
number of grid points in the y direction
permeability
relative permeability to gas
relative permeability to oil
permeability in the x direction
permeability in the y direction
pseudo gas potential
JANUARY 1981
Sg
So
Sw
ill
<I>
gas saturation
oil saturation
water saturation
time step
porosity
SI Metric Conversion Factors
bbl
cp
cu ft
of
ft
Ibm
psi, psi a
scf
x 1.589 873 E01
x 1.0*
E03
x 2.831 685 E02
CF  32)/1.8
x 3.048*
E01
x 4.535 924 E01
x 6.894 757 E + 00
x 2.863 640 E  02
Conversion factor is exact.
m3
Pas
m3
C
m
kg
kPa
stdm 3
JPT
Original manuscript recieved in Society of Petroleum Engineers office Nov.
18,1980. Paper (SPE 9723) accepted for publication Nov. 26,1980.
25
Discussion of Comparison of Solutions
to a ThreeDimensional BlackOil
Reservoir Simulation Problem
(SPE 9741)
William Hurst, SPE, consultant
I read with interest the article by Aziz S. Odeh in the
Jan. 1981 issue of JPT (Pages 1325). Although the
article is excellently illustrated with colorcoded
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
conformal mapping to simulate reservoir performance. 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 42 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.
'
Original manuscript (SPE 9741) received in the Society of Petroleum
Engineers office Feb. 5,1981.
01492136/81/00039741$00.25
Copyright 1981 Society of Petroleum Engineers of AIME
552
JPT
JOURNAL OF PETROLEUM TECHNOLOGY