Comput. & Indus. Ensng, Vol, l, pp. 139-144. Perllamon Press, 1977.

Printed in Great Britain

COMPUTER APPLICATIONS IN PETROLEUM

RESERVOIR RESEARCH

DWIGHT F. RYCHEL
Cities Service Oil Company, Tulsa, OK, U.S.A.

(Received 12 August 1976)

Abstract--The computer is an essential tool utilized by the research engineer in petroleum reservoir
evaluations. The magnitude and technical difficulty of the work usually precludes any other study technique.
The typical reservoir study is aimed at determining the optimally economic primary development plan for a newly
discovered field, or the feasibility of a variety of enhanced recovery techniques for a nearly-depleted field.
The reservoir study begins by defining the problem, identifying the desired output, and choosing the
appropriate approach for the degree of detail needed. The computer is utilized to retrieve, reduce and
statistically analyze the geological data, and production history from the company-wide data bank. The
properties of the oil are determined through a non-linear curve fit, matching the theoretical test results of the
computer modeled oil with lab test results. The resulting data is used to model the specific reservoir with a
complex simulator. Physical parameters of the simulated reservoir are varied to match historical production,
if any, validating the computer model.
When the engineer is satisfied the computer simulator adequately describes the performance of the
reservoir, alternate development plans can be forecasted and compared. This information provides a basis for
the economic analysis that determines the best plan of action. The computer time required for such a study
typically costs thousands of dollars. However, the cost of drilling a 5000 ft well can exceed $100,000 and
deeper wells have run over $I million, so an initial investment in computer and man time can make a
tremendous difference in final development costs and receipts.

INTRODUCTION
The computer is an essential tool for people involved in the exploration, production, and
distribution of energy. Few businesses rely as heavily on such a large amount of historical
information, or risk as much money on future predictions and estimates as the oil industry. Each
new move in the exploration or production of oil is based on the cumulative knowledge of all such
moves in the past, and adds more information for future moves.
The computer is utilized by geologists, geophysicists, engineers of all kinds, planners,
economic analysts, technical specialists, accountants and managers at all levels. The scope of this
paper is limited to the uses of the computer in Reservoir Research, by the research engineer. The
engineer, to conduct a reservoir study, must be familiar with the physical process of reservoir
mechanics as well as the systems concepts of modeling and simulation. The complexity and
scope of most projects make the computer the only alternative of problem solution. This paper
will describe most of the stages that might be included in any particular study. The.purpose of the
typical reservoir study is to determine the optimally economic primary development plan for a
newly discovered field, or the feasibility of a variety of enhanced recovery techniques for a
nearly depleted field.

PROBLEM DEFINITION
Like all engineering problems, care must be taken to properly define the nature and scope of
the problem before a solution is attempted. The most important question the engineer must ask is:
"Exactly what information do I need?". N o solution can begin untilitis known what answers are
needed and in what form. Another consideration in the determination of the solution technique is
the ever-present constraints on time and money. A thorough investigation resulting in certain
conclusions may be of no value if they are too late.A quick answer, using a simple technique or
model, may provide a timelier,if less accurate,,answer. A most important factor in determining
the solution technique is the degree of detail needed. Everyone may want as much detail as the
most sophisticated simulator is capable of giving,but too often the incremental information is not
justified by the increased resources required. Similarly, a question often overlooked by people
who use models, is what degree of problem solution is meaningful in view of the amount and
139
140 DWIGHT F. RYCHEL

accuracy of the input data. If information is scarce, and assuredly at a depth of 10,000 ft it is, then
assumptions are made that give much less detail to the reservoir description. No model or
simulation can give more detail than what is put in, even though it blindly and boldly prints
answers to eight significant figures. It is advantageous both to the engineer and the person
requiting the answer to use no more sophistication in the model than is required for the solution.
There is no virtue in keeping a computer computing for hours on a simulated problem that could
best be answered with a simpler solution method.

DATA RETRIEVAL AND REDUCTION

The most tedious, but yet most important part of a reservoir study is in obtaining and
correctly reducing raw data into a form that the simulator can use to accurately describe the
reservoir. As more work is done describing the reservoir, more reliable answers are obtained,
and less work is required in the validation of the model.
Much of the information required is stored in a company-wide data base. This is true of all
production histories of all company-owned wells, including the monthly production of oil, natural
gas and water, required for the validation, or history matching. Another data base contains the
geological information obtained by lab analysis of cores taken from wells. This includes
information on rock type, the porosity of the reservoir rock, the ease of flow of oil, gas and water
in the rock, and the location of the sample. To retrieve the data on any specific well requires a
retrieval system. The purpose of the retrieval system is four-fold: (1) selectively extract data
elements from existing pre-defined computer files; (2) perform computations on specific items; (3)
arrange the selected items by sequencing and grouping, such as the aggregation of core samples in
groups representing distinct layers in the reservoir; (4) tabulate totals, cumulatives, counts,
averages, maximums, minimums, and other summary-type information; and (5) output page
numbers, date, title lines, and other format features. Other more specific information such as
special well test results and instrumentation surveys of wells must be obtained from the
responsible production engineer.
Prior to the initiation of a reservoir study, a fluid sample is taken in the field. It undergoes
several tests in the laboratory, depending on the type of study. If the sample is not volatile, its
component composition is relatively constant with pressure and the results of the lab tests can be
used directly in the simulator to describe the physical properties of the fluid. More volatile fluids
must be modeled as multi-component fluids, since the composition of the fluid can be expected to
change over the simulation period. To model the fluid, the hypothetical component physical
properties are adjusted with a non-linear regression program so the results of simulating the lab
tests on the hypothetical components matches the actual fluid behavior in the lab tests.

SIMULATOR VALIDATION--HISTORY MATCHING

The validity of the simulator is tested by matching actual preduction history. Although this is
an important step and necessary to have full confidence in the simulation, it is not possible in a
new field where there is no history to match. In these cases, experience becomes important to
judge the accuracy of the output.
More often than not, there is some production (gas, water, oil) to match and often reservoir
pressures to match as well. With the simpler simulators, the matching may be accomplished
automatically by a non-linear curve fit routine that will vary specified reservoir parameters until
the simulated performance matches the actual performance. The more complex simulators, with
more parameters, requires the research engineer to make a run, compare the simulated
performance to actual performance by tabular or graphical comparison, adjust a parameter or
parameters, such as the production rate or reservoir size, and re-run the problem. This iterative
man-machine interaction continues until the engineer feels there is a good match and the model
reservoir describes the actual reservoir. Obviously, the data gathering step, if done properly, can
reduce the work required in history-matching considerably.
t
PREDICTION
Most of the work is complete before the engineer can use a model to predict reservoir
performance. Typically, the variables of interest through time are gas, oil and water production
rates, the physical state of the reservoir, the composition of produced and reservoir fluid and the
 Computerapplicationsin petroleumreservoirresearch 141
saturation pattern of any fluids injected into the reservoir. The ultimate criterion for alternate
simulations is economic, so all distinct alternatives that could be the most economical must be
simulated. Some of the variables that would change for specific situations would include the
number of producing and injecting wells, the geometric pattern of these wells, the best ground
processing equipment operating conditions, different injection and production rates, and different
injection fluids. Each change makes a change in the streams of revenues and costs. The predictive
mode of the simulator will ultimately provide the best solution for the original problem, if a
solution exists.

COMPUTER GRAPHICS
There are three common methods of presenting graphical information. These are with the line
printer, the pen plotter, and the CRT device. All of them are used for the same reason: to convey
a substantial amount of information in a concise manner to be quickly and accurately evaluated
and compared to similar information. The line printer uses alphanumeric characters and is limited
to the width.of the letter for accuracy. It is mostly used to plot graphs of a single computed
variable, such as reservoir pressure vs time, and compared with real data, read in and plotted on
the same axis. It may also be used to show contours of constant measurement of such variables
as pressure or oil saturation at any instant of time in the reservoir. Figures 1 and 2 are examples
of the graphics from a line printer.
The pen plotter is much more versatile than the line printer. With the appropriate software
support, discrete data can be fit and plotted in virtually any form. In most reservoir studies, the
simulated production information over time is of most interest to the engineer. This is typically
oil, gas, and water production, and the ratios of water/oil and gas/oil. Also of interest is the
injection rate, for those reservoirs under injection. All of these variables take on an added
dimension when they can be plotted on the same time axis. One figure can summarize the entire
predicted reservoir performance. Figure 3 is an example of the pen plotter and production
history. The plotter can also be used in a more specialized study of a well, or a pattern of wells, to

--Z~L -
P~N1 SYMBOCS ¢0 ~ DAT& O z O~SE~VEO A = C&LCULATE~ * • COI~CIDE~r

30.u * . . . . . . . . . . .
x ~. . . . !
- ! . . . . . .

! . . . . . . . . . . . . . . . .

! I
It~O,O * *
. . . . ~ . . . . . . . . . . . . . . . . . . . . . . . T

. . . . . . . . . . . . . . . . . . . . . . r

! !

330.U * x

..T]~':,. . . . . . . . . . . . . . !

,tt)*~), I - - -

i . . . . T
480,0 * o x

l 0 x
. . . . . . . . . . . t- ]
I • x !

* L
b30,G x
1
x I

1!
!
ao 1
I
]
780°U

Fig. 1. Observedandcalculateddatadisplay.
142 DWIGHTF. RYCHEL

303n00o
OOOUOOOt~vO0,'
. . . . . . . . . . . O0 Oil O~FOC~O O 0 0 0 ~ O 0 0 0 0 0 0 QO C C O 0 . . . . .
O O O O u OOOiau~J OOO O 0 0 u O i l o : l U U V o O 0 0 0 u
O000uOOOOOuOOOOOOOOOOa~OOOOOOOOO~n~
bOI~OOOOOOuuOuOOOOOOJOJOOOOFIOOOUO~O;JJ
O000000OOOuOOoOOOOOOUdOOOOOOOOOO,':ubO00
l) O000OO rio u O00000UI)O C(]OC 00 U0 UO UU0 U~ , JO U
. . . . . . . . . . . . . . .
. . . . . O. 00 O00 OO-O00000 ,) 000 (I~'0 OO0 OFj00~730 ~0 O,3"0O-frO
O()O f~O000001)00000 OUOV~)OOOOuO u O,)uO u ~ UUU u u UO ,C~UOhOU
OOOO0~O0000000000000000oOOOOOoO~oOOOOuOOOOu.nuOOhO0
O000000uuOUUV~IUOUVOUUUVOUUOCUUOuuOO~;uuvuuU~, ,uOucOu~,
• OOOOOOoOOOOOOOOOOOoOOOOoOOOOOOOonOO~uOOoOuoojno~oono~1 •
00000(~O00UO O0 OO00OL) G0 U00000000q)U ) UI) UCL)UUUu O0 V r} U~ UUOIlO ri
. . . . . . . . . . 00(l (iO 0 0 0 0 0 IlO 0 O 0 0 6 O 0 0 O 0 0 0 0 0 n 00 u 0 0 0 q~3 fin 000"OOOrl;l,TO 0000tlO~-) . . . . .
0000000000flU0000000000 b000O 0O I I i l ~ I 0 ~ i J U O0 O 0 0 0 0 O 0 0 u O UUU~I~ I1(~ UU
(]O0000000000¢IO000000000UOOOOOOOoOOOOOOOOOOOOO[%O,JOOOOOOUOO~
• O00000OOOOOOUU~OUOOOOOUUOOOOOOVOOOO~IUOOOOOUOO.VPUOOOO~UUO )u •
O 0 0 0 0 0 0 0 0 0 0 0 h O 0 0 0 0 0 0 OhOOO~)OOOflO 0 0 0 0 0 0 q 00(~0 C)O000OIl~;(~O(JuO 00 f)O 0
O000000000UOOUU OOO~U uOQOUUOUU O O Q U O O U O ~ I t J O u O O000¢JO N ' I U U O Q U O 0 0 U U
. . . . . . . . . . . . . . . "0 (~0fl 000 I) OXTITOQ/)~TO O'O'O'ITO u u 0 u O'ITOITO'O O00-O'OXTDITr~O 0 ~ u ,) ,~u n ~
O0~O00oOOt)OOOOOUUO00000000ov O00noouo O000tJt~OOOOlFt, Oul) OUO Onrju
00000000O000000000000000000 0 d O 0 0 0 0 0 0 0 0 0 0 0 , ) u O 0 0 0 UO0()O O
• O000OOOUOOOOUOOOOUUOUOOO00 OOUO htiOOOOO UUOuO U~ 0 0 0 0 { ~ 1 •
000000000000 i 00O0000000000000000n0~00
O0000OO0000 I!II IiI II IiI II ~Jun(~ouOoO~Ouc~oOoooOhOU
......................... 0 ~ o o o o o ~0"~--T~lll I~I-I-I . . . . . T l l . . . . . . . n0~J'0 ~ u ut)u u u u u u o o u ~ u n(,
OUOOOOOOOOOOU II 11 ~¢~ ~I 'U O(IU 000 OUd UC00 UU~)OIIflCk
O00OO00000000 ii 2Z 2222 22 II ,lO 0 0 0 0 0 0 0 0 n r h ~ 0 0 0 0 0 0 ~
• Ot}OOO000OO000 II ~?~e'?2d ~2
0000000000000 II P-? 33 ~,'~ 3 ? I nooooooooooooooono000
0000U000000'~}00 II ?P 3333 t+~+ b 5 ~ 3 ZI 0000UU0000000uVOuU00U0

O00OO0(~O000000O 11 ?? J ~ 5 b6 /85,+ O00000Ot)OOOOOOOuOOOO000

000000000000000 i 2~33 t, 5 6 7 8 000~000d0000O00000~000ff
• 0000000000000000 iX ~?-~ ~ 5 ~t7 ~ ~ ~3210000UOOOOOOOOOOuOOUU~)O0 •
- 0000000000000000 ~ 77 II 73 ~ ~8 O00000000000OO00000OO0
O0000UOUOOOUUO000 t*5 ( } 7 7 7 7 7 7 7 7 5~3~ O0000OOdOOOOOOOu iX 2a
O'0W'0u0OOuu0uu0"0~0 '" ~,~ ~"----/'7"--6~5"-~;"3~Z'T-~0"uuuuuvuvuuuuuu ~
00000000000000000 iii 6? I ,i 5 3 2 I 0000O0000000u000 3
.... 0-tlO-0-0O000"00WO0q'JO~O " 1"~ ~ 8 ? e , St+ ~ I O0000"O00OOOlrO0-O'O ~r
0U0000UO000000000U Ill1 3~'5 ? 6 b~' 3 ~ 2 11 0 n 0 0 0 0 0 0 0 0 0 0 0 u 0
. . . . . . . 0-000"000(T000ff00Ouuu ~F12 -t+- ~ ~-3~-~1- ~oouu000oou0 . . . . . . .
oOOOOOOOOO00000000 2 ~, $ ~, 33 2 II 00000O00o0
uvuuuuuuuuuuuuuuuuu I ~ 3 ~ .~.~ ~ t~ uuuuuuvuuuu
OOOO00000OO00000000000 ~ 2 ~ 3 ;~2 I O0000000OOOV
. . . . ~0000~0000oo0O0ff00-0-- IT -- lq 0~0oO~r00~n0~o " T
O000000000000000000000 I i I Ill 00000000000000
. . . . . . . . . . . UUUUOOOOOOOOWO~OUUU~O - I'ITI-i~ uOOuuuOOuffOffn~ . . . . . . . .
OOOO000000000000000000 ! OOO()O000000000
uuuuuvOuuuuuuuuuOuuuuu uuuuuuuuuuuuuuu
00000000000000000tJ000000000OO0000000000
• . . . . . . . ~ O0"000~ O ~ 0 0 0 O if00"if00 u u u 00 0"00qT00000 ~T- . . . .
OOO0000000000000OO0000000000
. . . . . . . . . . . . . . . --~rOO~WO0~O'O~ . . . . . . . . . . . . . . . . . . . .

Fig. 2. Lineprintercontourmaps.

track fluid components. Figure 4 shows the streamlines of injected fluids from the injectors to the
producers.
The CRT can, in general, display anything the line printer or plotter can. It has the advantage
of speed, and allows the engineer to interact and control the display.

CONCLUSIONS
The computer time required for a reservoir study may be from tens to hundreds of CPU
hours. For an average project, where the reservoir is divided into 200 grid blocks, it requires on
the order of one minute of CPU time for one time step. The time step may be one-half day or 30
days, depending on how fast parameters are changing in the simulation. A full time study would
simulate up to 30 yr or more. Different alternatives would require repeating all or part of the
simulation. (Most simulators are written to stop and restart at discrete points under the user's
control.) The cost for the computer time, as well as the man time, is substantial. It is small,
however, in comparison with the costs of drilling a well and small in comparison with the
difference in receipts if the reservoir is not developed in the optimum manner.
The computer plays an integral and necessary part in the analyses involved in petroleum
reservoir research. It represents a substantial investment, but the petroleum industry is
increasingly required to make substantial investments to squeeze a reluctant world for its
remaining oil.
 Computer applications in petroleum reservoir research 143

ACI IH ~ITU 81TUHEN RECOVER~

r,-' OEVELeP£O 5-SP~T
WELL SPACIWC = 500. r l
m' MIOPOINT TEMPi 1~6.0 r
:=,. tO X 6 - - OLD VERSION

["'" J
I "'~o,, ....... ;,

o
~== ==

/,,,'7 i/, ,,
!7 if" i
I I /"" \
I \ / \
I \X // \ X
I \~ f "'X 1~
I
~ "2: !
/
r-'
W'

r,l'

50
TIRE X 10"= (OATS)

Fig. 3. Production plot.

CRSE 2. ISOLRTED 5-5POT, ERCH I N J E [ T O R ' 5

RRTE I5 - 0 . 2 5 TIMES THE PRODUCER'S RRTE

Fig. 4. Fluid streamline plot.

144 DWIGHT F. RYCHEL

REFERENCES
1. K. H. Coats, Use and misuse of reservoir simulation models, J. Petroleum Technology, 1391 (November 1969).
2. K. H. Coats, R. L. Nielson, M. H. Terhune & A. G. Weber, Simulation of three-dimensional, two-phase flow in oil and gas
reservoirs, Society of Petroleum Engineers J. 377 (December 1969).
3. B. C. Craft & M. F. Hawkins, Applied Petroleum ReservoirEngineering,Prentice-Hall, Englewood Cliffs, N.J. (1959).
4. W. T. Ford, Elements of Simulation of Fluid Flow in Porous Media, Texas Tech. University Mathematics Series No. 8
(1971).
5. M. A. Nobles, Using the Computer to Solve Petroleum Engineering Problems, Gulf Publ. (1974).