SPE Paper by A. S. Odeh For Exercise 3

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Comparison of Solutions to a

Three-Dimensional Black-Oil
Reservoir Simulation Problem
Aziz S. Odeh, SPE,

Mobil Resemch 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 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.

1. Stratification and reservoir properties are given in


Fig. 2. The reservoir is initiall~ 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 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

Areal and cross-section 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.


1.Plots Ot
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), (lo, 1, 1), (lo, 1, 2), (lo, 1, 3), (lo,
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.*

0149.2136/81 /0001 .9722S00.25


Copyright 1981 Society of Petroleum Engineers of Al ME

Report of all pressures referred to a depth of 8,325 ft or at the center of the


raspectlva blocks. If both are available, report both.

Statement of the Problem

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

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
3X1 O-6

Initial reservoir pressure, psia at 8,400 ft


Gas injection rate, MMscf/D
Maximum oil production rate, STWD
Minimum oil rate, STB/D
Minimum flowing bottomhole pressure, PSI
Maximum saturation change during time step
Rock compressibility, I/psi
Porosity value of 0.3 was measured at a base pressure of 14.7 PSI
Wellbore radius, ft
Skin
Capillary pressure
Fleservoir

temperature,

0.25
0
0
200
0.792

&F

Gas specific gravity


Runs are terminated either at the end of 10 years or when
GOR%20,000 scf/STB or when the oil production ratesl ,000
STB/Q whichever occurs first terminates the run.

TABLE 2- PVT PROPERTIES


Saturated Oil PVT Functions
ReSeNOlr
Pressure
(psla)
14.7

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

Saturated Water PVT Functions

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

Undersaturated Oil PVT Functions


Gas PVT Functions

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

Undersaturated Water PVT


Functions
Reservoir
Pressure
(psla)
4014.7
9014.7
14

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

Pseudo Gas Potential


M(P)
(psia2/cp)

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

.IOURNAL OF PETROLEUM TECHNOLOGY

! 6

14

MOBIL

SHELL

AMOCO

INTERCOMP

EXXON

---

SSc

EXXON AGREES WITH MOBIL BETWEEN 3-7 YEARS

EXXON AGREES WITH INTERCOMP BETWEEN 8-10


AMOCO AGREES WITH SHELL AFTER 7YEARS

2
0

YEARS

CMG AGREES WITH MoBILTHRolJGH


WITH iNTERCOMP BETWEEN 8-10

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

TABLE 3- RELATIVE PERMEABILITY DATA*


Oil-Gas

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

This Is a two.phaae, gas/oil problem. Set the relative permeability to water


equal to zero for all values of water saturations.

MOBIL

SHELL

AMOCO

INTERCOMP

EXXON

---

Ssc

EXXON AGREES WITH MOBIL UP TO 3 YEARS


SSC AGREES WITH MOBIL BETWEEN 2-7 YEARS
AMOCO AGREES WITH INTERCOMP

AFTER 6 YEARS

CMG AGREES WITH MOBIL THROUGH 7 YEARS


AND WITH INTERCOMP BETWEEN 7-10 YEARS

t
2345676

i
10 TIME, YEARS

9
Fig, 4 - Case 1- GOR vs. time.

INTERCOMP.

5500
MOBIL

INTERCOMP

SHELL

EXXON

AMOCO

-o-

Ssc

AMOCO AGREES WITH SHELL UP TO 3 YEARS


4500

EXXON AGREES WITH AMOCO BETWEEN 6-10


\~

YEARS

SSC AGREES WITH AMOCO BETWEEN 3-5 YEARS

3500
MOBIL, lNTERCOMP~

CMGAGREES WiTH MOBIL BETWEEN I-4, AND$-10


AND WITH AMOCO BETWEEN 5-8 YEARS
2500

34

YEARS,

10

TIME,5YEARS6
Fig. 5 - Case 1- pressure vs. time for producing well Cell 10,10,3.
JOURNAL OF PETROLEUM TECHNOLOGY

,.

changes using slicesuccessiveoverrelaxation.


4. Noniterative computation of grid-block saturation changes and of semi-implicit adjustments in
production rates.
5. If any saturation change exceeds the maximum,
reduce At and go to Step 2.
6. Noniterative computation of grid-block 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 semi-implicit adjustments as saturations changed in the well block.
h important exception is that the oil rate was held
constant if, at the start of the step, the well had
excess computed productivity (i.e., if the computed
bottomhole pressure exceeded the minimum value of
1,000psi).
CMGSBlack-Oil Model

CMGS 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-, twoor 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
nobilities. These include single-point upstream, two-

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

INTERCOMP, SHELL, AMOCO

EXXON AGREES WITH SHELL, AMOCO THROUGH 6 YEARS,


AND WITH INTERCOMP BETWEEN 7-10 ?EARS
CMG AGREES WITH MOBIL THROUGH 3 YEARS, AND
WITH INTERCOMP BETWEEN 4-10 YEARS
SSC AGREES WITH SHELL BETWEEN 4-9 YEARS

34

10

TIME, !EARS
Fig. $ - Case 1- gas
JANUARY 1981

saturation

vs.

time for producing well Cell 10,10,3.


17

7500T

6500

MOBIL

SHELL

INTERCOMP

--

EXXON

*o*oDw~=
CMG
AMOCO AGREES WITH SHELL UPT06
AND WITH EXXON AFTER 6 YEARS

YEARS,

SSC AGREES WITH SHELL BETWEEN 1-2


YEARS, AND4-5
YEARS,
WITH MOBIL BETWEEN 3-4YEARS,
AND
WITH EXXON BETWEEN 6-10 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

CMG AGREES WITH AMOCO

I=M

3000

1,1

2,2

3,3

4,4

5,5

6,6

T,7

8,8

qg

10,10

GRID POINT LOCATION


Fig. 8 - Case 1- pressure vs. grid-point location, time= 8 years, top layer.
18

JOURNAL OF PETROLEUM TECHNOLOGY

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

GAS SAT. <.005

AT:

1) 4,4- SHELL, EXXON, AMOCO


2) 5,5 -ALL

PARTICIPANTS

3) 6,6- ALL PARTICIPANTS EXCEPT MOBIL,


CMG
4) 7,7-SHELL,
AMOCO
5) 8,8- SHELL

CMG : M

I
,

1,1

2,2

10,10

3,3

GRID POINT LOCATION


Fig.
JANUARY 1981

10-

Case 1- gas saturation

vs. grid.point

location,

time=

8 years,

middle

layer.

19

Exxons General Purpose Simulator

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

INTERCOMP AGREES WITH MOBIL UP TO 7 YEARS,


AND WITH AMOCO BETWEEN 7-10 YEARS
EXXON AND AMOCO AGREE FOR 10 YEARS

,
,
2345678

CM(3 AGREES WITH AMOCO BETWEEN 5-8 YEARS,


AND WITH MOBIL BETWEEN 8-10 YEARS
,
I
,
,
I
9
10 TIME, YEARS

*
Fig.

11-

Case 2- oil rate vs. time,


JOURNAL OF PETROLEUM TECHNOLOGY

MOBIL

SHELL

hMOCO
EXXON

--

Ssc

/*

AMOCO AGREES WITH SHELL BETWEEN 8-10

YEARS

~
!

EXXON AGREES WITH SHELL UP TO 4 YEARS, AND


WITH MOBIL BETWEEN 5-8 YEARS

INTERCOMP AGREES WITH MOBIL UP TO 7 YEARS,


AND WITH SHELL BETWEEN 8-10 YEARS

I
~

CM(3 AGREES MOBIL THROUGH 9 YEARS


AT 10 YEAR GOR= 2O,OOO

,,

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

well Cell 10,10,3.

10

which are designed specifically for optimai solution


of two-phase problems: one for water/oil (dissolvedgas content constant) problems and one for
gas/water (no oil phase) problems. BET,.4 II has a
large variety of user-oriented features mch as input/output options and well control options.
There is a one-equation implicit pressure formulation in which the equations are decoupled and
solved in this order: pressure, gas saturation, and
water saturation. For the one-equation formuiaticn,
there are options to solve for both saturations explicitly (IMPES) 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,
muitiple outer iterations maybe taken to account for
the mmiinearity of the basic flow equations.
However, if no vaporization or resolution of gas is
occurring, oniy one or two iterations are required to
converge the noniinearities adequately.
The large systems of linear algebraic equations
may be solved by a variety of methods, any one of
which may offer significant speed advantages on a
given problem: (1) direct solution by reduced bandwidth Gaussian elimination, (2) several forms of oneline, two-line, and planar successive overrelaxation
(SOR), or (3) SIP.

MOBiL

SHELL, AMOCO

iNTERCOMP

-=-

Ssc

1
1

I
2

I
34

Fig. 14- Case 2-gas

Mobils Ail Purpose Reservoir Shnuiator (ALPURS)

ALPURS is a three-dimensionai, three-phase,


multiwell, black-oil reservoir simulator which uses a
strongly coupled, fully implicit method to solve
simultaneously for ail unknowns. 5 The noniinear
intercell flow equations and weii-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
y, rock compressibility, gravity, gas dissolved in both the oii 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-potentiai
function, skin factor to account for damage or
improvement, non-Darcy flow effect, and flow
restriction due to limited entry such as partial
penetration.
SSC Model
SSCSblack-oil model employs an Adaptive Implicit
Method (AIM). This technique, which recently was
developed at SSC, seeks to achieve an optimum with

EXXON AGREES WiTH iNTERCOMP BETWEEN


3-10 YEARS
CMG AGREES WiTH INTERCOMP

,
6

,
7

,
8

#
9

1
10

TiME, tEAFiS

saturation vs. time for producing

well Cell 10,10,3.

JOURNAL OF PETROLEUM TECHNOLOGY

..

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

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


CMQ AGREES WITH EXXON AT 1,1 THROUGH 6,6
3ooa

1,1

2,2

3,3

4,4
5,5 6,6
7,7
GRID POINT LOCATION

Fig. 16- Case 2-pressure


JANUARY 1981

vs. gridpoint

8,8

9,9

1 O,1O

location, time =8 years, top layer.


23

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 afl 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 underkill.
With AIM there is no problem of over-or underkill.
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 bubble-point
problems, such as that in Case 2, are handled by

variable substitution. The simulator is a general


purpose package offering three-dimensional 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, 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. Additionally, two-point upstream weighting is used to
calculate phase nobilities.

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

GRID POINT LOCATION


Fig. 17- Case 2-gas
24

saturation vs. grid-point location, time =8 years, top layer.


JOURNAL OF PETROLEUM TECHNOLOGY

IVOBIL

SHELL

AMOCO

INTERCOMP

EXXON

S=l

I=L
=

E, CM(3 ~ M

-.

---SSC

0=0===0==~
CMG

A=l

GAS SAT. C .005 AT:


}) 4,4 AND 5,5-SHELL, AMOCO,
Ssc, EXXON
2) 4,4-CMG
3) 6,6-SHELL,

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

4,4 * !5,6 6,6 7,7


GRID POINT LOCATION
ion vs. grid.point

4. Watts, J.W.: A Conjugate Gradient-Truncated


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.
23-26,1979.
5. Bansal, P, P., Harper, J. L., McDonald, A. E., Moreland, E. E,,
Odeh, AS., and Trirnble, R. H.: A Strongly coupled,Fully
Implicit,
Three-Phase
Reservoir
Three-Dimensional,
Simulator, paper SPE 8329 presented at the SPE 54th Annual
Technieal Conference and Exhibition, Las Vegas, Sept. 23-26,
1979,

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)
=

pseudo gas potential

JANUARY 1981

.CMG= M

Iocatlon,

8,8

time =8 years,

middle

layer.

Sg= g$s saturation


=
:: =
At =
@=

od saturation
water saturation
time step
porosity

S1 Metric Conversion Factors


bbl
Cp
tuft
F
ft
lbm
psi, psia
scf

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.

orlglrral rr?8rw8cript recleved In SOC18tyof Petroleum Engineers office Nov.


18,1980. Paper (SPE 9723) acceptedtor publication Nov. 26,1980.

25

You might also like