Professional Documents
Culture Documents
ABSTRACT
Perhaps no industry is more vitally concerned with risk
than the oil and gas industry, and few professional men
other than petroleum engineers are required to recommend higher investments on the basis of such uncertain
and limited information. Tn recent }'ears, the number of
methods dealing with risk and uncertainty has grown
extensh-el}' so that the classical approach, using analytical procedures and single-yalued parameters, has undergone a significant transformation. The use of stochastic "ariables, such as those frequentl,r encountered in the
oil industry, is now economically feasible in the eyuluation of an increasing number of problems by the applica~
tion of Monte Carlo techniques.
This paper defines the Monte Carlo method as a subset
of simulation techniques and a combination of sampling
theory and numerical analysis_ Briefly, the basic technique of Monte Carlo simulation inyob.-es the representation of a situation in log-ical terms so that, when the
pertinent data are inserted, a mathematical solution becomes Jlossible. Using random numbers generated by an
"automatic penny-tossing machine" and a cumulatiye freQuency distribution, the beha'\'iour pattern of the particular case can be determined by a process of statistical experimentation_ In practical applications, the probabilistic
data expressed in one or several distributions may pertain
to geological exploration, discover).' processes, oil-in~place
e\'aluations or the productivity of heterogeneous reserYoirs. The great variety of probabilit).' models used to
date (e..[_, normal, log-normal, skewed log~normal, linear,
multi-modal, discontinuous, theoretical, experimental) confirms a broad range of experimental computations and a
g-enuine interest in realistic representations of random
impacts encountered in practice.
Emphasis in this paper is directed to the salient characteristics of the Monte Cado method, with particular reference to applications in areas related to the oil and ,gas
industry_ Attention is focused on reservoir engin{'erin~
models. Nevertheless. management facets of the oil and
gas business are considered alan,!! with other applications
in statistics, mathematics, physics and engineering. Sample size reducing techniques and the use of digital com
puters are also discussed.
INTRODUCTION
120
--'-
Concept of
BASIC TECHNIQUES.
,~
'.
1
;
SALIENT CHARACTERISTICS
Drawing from various publications (1, 2). the salient characteristics of the Monte Carlo method appear
to be as follows;
(I)-The Monte Carlo' method is associated with
probability theory_ However, whereas the relationships -of probability theory have been derived from
theoretical considerations of the phenomenon of
.chance, the Monte Carlo method uses probability to
find answers to physical problems that maJ~ or may
not be related to probability.
(2)-The application of the Monte Carlo method offers a penetrating insight into the behaviour of the
s~ystems studied. Frequently, problems become deceptively simple. In this sense, effective Monte Carlo
techniques are self-liquidating.
(3)-The results of Monte Carlo computations are
treated as estimates within certain confidence limits
rather than true or exact values. Actually, all meaningful physical measurements are expressed in this
way. In many cases where relationships in a model
cannot be evaluated at all because of either mathematical or practical considerations, Monte Carlo techniques can be used to obtain approximations.
(4)-As in any other method, there is a need for
adequate basic information; data for the implementation of the Monte Cado method, however, may be
obtained b~r standard rlata processing procedures.
(5)-The method is flexible to the extent that the
intricacies of a problem. as may be reflected by either
a great number of parameters or complicated geometry, do not alter its basic character; the penalty
paid for complexity is increased computing time and
costs.
(6)-A practical consideration is that the iterative
calculations necessary for attaining a certain level of
confidence can be distributed among several computers, working simultaneously in one or more places.
(7)-The l\ionte Carlo solutions are approximate,
however, they can be up-graded commensurately with
the time and money allocated to the problem.
(8)-The Monte Carlo method's purely numerical
character requires careful scrutiny of all results.
(9)-Solutions of the Monte Carlo method are
numerical and apply onbr to the particular case studied_
RANDOM NUMBERS
--
--,
..:
'
Ran~omness
,':
',.
(Eg. 1)
where:
R n is the nth random number;
random number;
Ie
Rn+l
is the (n+l)th
~'
Technology, July
121
~~:~ ...
Thb
i~
be
USE OF COMPUTERS
ApPLICATIONS IN MATHEMATICS
123
..
.....
~~
.:
~~;
.... ..
~
ApPLICATIONS TN STATISTICS
:Monte Carlo techniques are widely used in the statistical design of experiments, and mayor may not
be connected \....ith techniques referred to as sequential, multiplex and inverse sampling. Typical applications include; con-elation of )andom valiables; study
of interdependent events; branching processes fol' design of new equipment and analysis of failure patterns; quality control and scheduling of maintenance.
ApPLICATIONS IN OPERATIONS RESI::ARCH
The .Monte Carlo method introduces a new dimension of flexibility in many OR analJfses. Single-valued
input variables, such as mode, median or mean. are
replaced by distribution:) expressing the situation in
a more meaningful manner and providing room for
"optimistic" and "pessimistic" limits, Monte Carlo
techniques extend the procedures for solving the following classes of problems:
(a), Linear programming: mixing problem using
random variables.
(b). Resource allocation: proration to markets; allocation of work to pl"oces~ing machines; ~tocl( distribution.
(c). Optimization: warehouse location; shortest
route problem and k-th best route through a networiL
(d). Inventory: decision rules; inventory control
and procurement.
(e L Transportation and merging: simulation and
control of traffic and highway merging and congestion; flow through networks.
(0. Queuing or waiting line theory: terminal operation of trucks, buses and ~hips; determination of
peak demands.
(g). Analysis of type PERT (Program Evaluation
and Re"ie\.... Technique).
(hl. Dynamic programming: maximum time - rate
of return; multif'tage and Markoff renewal processe~.
ApPLICATlONS IN PHYSICS
ApPLICATIONS IN THE OIL AND GAS INDUSTRY
of 'Monte
.J. "V. Gibbs (1876) introduced the stochastic method in statistical mechanic~ through the hypothesis
that the average random walk over the time is similal to the average over the phase space. On this basi!i.
Ihermodynamic quantities were determined by integrating over thp ensemble of phase ~pace (e,g., ga:;
c.ompres~ibility factor),
T24
~:
'.
,
"
Nature) using well the deck of cards in her possession, responds to the search for oil and gas b~r signals
that are partially random. Exploration for oil a.nd
gas therefol'e involves many uncertainties. Each elemental assumption in the exploratory search involves
its own degree of llncertainty_ Together. these assumptions p:1rramid to a total uncertainty of critical
proportions. The uncertainty is not really related to
the determination of the success ratio of a company
with infinite resomces, which can conduct exploration on such a large scale that it can rely on 'longrange" average results; nor is the problem simply to
decide whether to drill a well or not. The problem
seems instead to be that of the operator with limited
. funds ''''ho wishes to explore a particular basin and
seeks a decision from an individual and specific point
of view. This is so because it is well knmvn that operators proceeding in a prudent manner have drilled
as many as twentJ' unsuccessful prospects although
the industry-wide success ratio in the area was as
high as one discovery in five trials.
Monte Carlo tl2!chniques can be used profitably in
simulations of exploratory programs to focus the attention on critical "reas (5): Although not a substitute for the undertaking of ,,,'orthwhile risks, the
Monte Carlo method assists in the following:
(I)-Reduction of the number of possible exploration programs to a manageable size.
(2)-Application of statistical and personal probabilities to change a decision from one of uncertainty
to one of assumed risk.
Typical applications in exploration are:
(I)-Tracing the history of an individual oil compan]! by branching processes and sequential sampling
from appropriate probability functions. Individual experiments can be terminated following attainment of
a fixed number of ,'eutures, a certain level of economic
- stabilit]r, an amount of profit, a merger or bankruptcy. This would include individual "random walks."
(2)-Bidding models on the basis of the particular
company's policy regarding returns and utilit,}, probability of discovery, likely competitors and their bidding patterns. The model would be useful in the determination of the calculated risk toward the acceptance of a particular bid.
(3)-Preparation of stochastic models of the discover}' process, including:
(a). Calculation of the number of prospects to drill
in a particular basin so that there will be a reasonable statistical probability of obtaining production
in at least one prospect.
(b). Dete-rmination of type of hydrocarbon upon
discovery (I.e., gas or oil)(c). Calculation of the size of oil or gas discovery,
using appropriate distributions_
Cd). Calculation of "extension" and secondarJ' recovery reserves by drawing relative appreciation
factors from distributions Ustratified" by "year
since discover}'."
TcchnologYI July - Sep~ember, 1965, Montreal
'-.
,
,
'
'
..
k~: .. :
~;:...'
.~.
Ih
(Eq. 2)
This is considered to be the best method of determining the specific surface (6),
Warren and Skiba (7) studied an idealized miscible displacement process in a three-dimensional heterogeneous medium by means of experimental computation based on a Monte Carlo model. In their study,
displacement processes were described in terms of the
probability of "residence time" of "mathematical"
particles representing input fluid.
In recent years, the mathematics and application of
statistical and Monte Carlo techniques have received
increased attention in mineral sciences (8, 9).
According to Collins (6), if a mathematical theory
of flow through porous media is possible at all, it
must take the form of a statistical theory describing
the macroscopic features of the flow, in the sense of
the "ergodic hypothesis" of Gibbs. Several investIgators (l0, 6) have heated this problem in terms of
random walks of a "drunkard '''ithout or with 'some
memory;" Le., with or without autocorrelation. The
randomness of a porous medium, however, has yet to
be successfully represented.
125
",.
l'"
..
".
Warren and Price (11) studied the effect of the disposition of heterogeneous permeabilities on single-
TABLE
phase flow for a known permeability distribution function. They compared pl"obabilistic permeability solutions fOi' radial flow between concentric cylinders with
permeabilities calculated on the basis of arillnuetic,
geometric and harmonic means, as well as of the
median and the mode.
F(~j
~O
57
6.9
78
85
Oil-i?/-Place Calculations
0.1l0
0.239
0.306
0.389
0.500
0.581
0.697
0.807
1.000
96
10.3
TABLE
The porosity. {P, was introduced as a cumulative frequency function and jH shown in Table I and Figure
1_ The formation water saturation, 8 . . . , was expressed
by five distributions, The effect of structural position on the \\'ater saturation wa::; not considered in
this ca~e. Formation water saturation distribution~
therefore account only for the porosity \'ariation and
are shown in Tables II to VI. They are the result of
an "educated interpretation" of the "scattering" of
data in a l/J-S .. plot. An alternative would have been
to use simple dala proces!iing procedures to alTi\'e
fit statistical distributions. Stratified sampling according to structura! position and type of porosity
was considered. but. in the intere;;;t of simplicity, was
not implemented, A :iequential sampling technique,
howevel', was employed. First, a pseudo-random number was generated and the porosity drawn from its
distribution. A second random number was entered
in the appl'opriate fOI-mation water distribution. depending upon the porosity value sampled by the first
random number. Corresponding porosity and formation water values were used to calculate the oil-inplace using the exprelision:
~
13.3
14 l
h ([eeU = 70
(SPE Standard Symbol's)
A (acres) = 27.000
Bot (fnction) = 1.<110
llii
lEg.
,?j
""
:::O~
Ftl1ll' 0
s;:
G
"
./
~55
385
,11.5
44.5
47.0
.~
.. Ell
,,/
~~
w
>
,J /
>
po. Ci>l11
"
---
~c
:5
;:/
/'
..----~
/,,0/ /
,
POR051TY
126
"
~,
0.006
0.021
0.086
0.2\9
0.501
0.754
0.913
0.974
1.000
"
20.5
Z3.5
26.5
29.5
32.5
,
~
[1
Vi
0'
"
PE"~EA8111TT
mil
"
'00
'/
. ~~(:~<;
~c~:.'''::
TABLE III
TABLE
F,(S.)
13.5
15.9
18.3
20.7
23.1
25.5
27.9
30.3
32.7
35.5
0
0.006
0.Ql8
0.041
0.083
0.284
0585
0890
0949
1000
TABLE
~:;;:.~
,.>
S. (%)
;;;:._:~:':_'
~.
S.. (%)
7.0
8.9
10.8
12.7
14.6
16.5
18.4
20.3
22.2
24.0
IV
0.017
0.044
0.189
0.439
0.669
0.806
0.898
0.956
1000
TABLE
VI
,..
5..(%)
F,CS,J
S.(%)
9.7
118
13.9
16.0
18.1
20.2
22.3
24.4
26.5
28.4
0
0.01I
0.031
0.060
0.145
0.371
0.684
0.912
0.974
1000
5.1
6.8
8.6
10.3
12.1
13.8
15.6
F,(S,,)
0
0.027
0.086
0.214
0.457
0.629
0.810
0.888
0.936
1.000
17.3
19.1
20.8
. -::
.,:~
.~
"
j
TABLE
VII
Expected
Experiment
n
Porosity
Fonnation
Water
Oil-in-Place
Oil-in-Place
~(%)
S, (%)
N(MMSTB)
:ENjn(MMSTB)
10
20
30
40
50
60
70
80
90
100
13.4
8.8
14.2
14.4
5.6
5.7
14.2
10.3
8.0
5.8
12.9
19.3
11.3
13.7
19.3
19.7
15.4
22.0
23.4
1217.651
737.2,0
1310.395
1293085
425.127
479.882
1I88.813
905.860
645.354
460.447
823.032
774.251
799.850
808.604
790.657
812.068
799.383
811.405
800.999
7911I7
275.212
301.351
304.773
325.986
315.583
316.727
328.290
326.595
322.565
326.405
200
300
400
500
600
700
800
900
10.4
11.2
13.4
143
10.4
5.7
13.6
7.5
17.6
16.8
113
119
13.7
27.6
10.3
26.4
892.440
968.747
1232.252
1312.445
935.m4
127.020
1264.661
577.540
833.400
838.692
819.106
826.402
819.129
824.285
823.709
82I.I74
325.827
314.459
318.590
317.283
317.503
319.679
319.795
318.735
22.0
679.361
778.989
626.699
1260.481
1027.008
887.212
641785
004.052
228.109
648.883
823.625
823.465
823.171
821.I56
820.745
820.275
820.026
818.568
818.324
819.763
317.379
316.755
316.889
317.612
316.893
316.996
317.319
316.962
317.353
316.792
1000
1010
1020
1030
1040
1050
1060
1070
1080
I090
Technology~
27.6
8.4
8.8
7.7
14.3
12.4
10.1
7.9
9.5
3.4
7.6
15.0
22.2
15.5
20.2
15.4
22.1
16.5
35.3
18.3
Standard
Deviation
S(MMSTB)
f:
127
Productivity
TABLE
the well,
~iodel
=
2:
r
Ri-l R j + i
r.. -
((/. 61
r,,
P,
B"
~o
h avg.
k avg.
2,~A7
psig
~ 0.289
.r
(using k ""16.1
r,.
r.
~Vdl:
[,
12-16-10-55
11.6 feet
0,066
1,490 rl~cl
- 0.263 feet
~
8.06 Old
!;-",
('1. /l
log (!;-l + l~
lClg (kJrl + U
An individual experiment is c.onducted by drawing r
pseudo-random numbers and obtaining permeability
values from appropriate distributions. The cumulative
frequency distribution for the pool, Leafland Viking,
and '...ell, 12-16-40-55, are shown in Table V[II and
Figure 2. Ring boundaries are calculated from Equa-
TABLE
VIII
126
:::: kill
Well k (md)
1.450
0.4 ep.
7.5 feet
6.28 md
where
C is a proportionality factor determined from:
C
IX
Pool k (md)
Di~rrilmlioll
F(k)
0.10
0.10
Oll
O.I~
O.ll
0.12
0.18
0.22
0.16
0.26
0.3:1
0.23
0.3B
0.4.
0.37
O.5B
0.55
0.76
O.q.
0.66
2..10
1.95
0.77
1LOa
7.50
O.BB
100.00
100.00
LaO
(Eq. S)
equal~:
J (STB/D/psil =
~
0.00307 kIll h_
B II (LII log (r,,/r.1
(Eq.9)
As expected from an inspection of FiguJ'e 2, the calculated permeabilit)r and produeti"it~.. indices are higher \"'hen using the permeability distribution for the
pool (i.e., Model 2). The e.ffect of geometry Hppenr~
to be negligible. Differences between core Hnalysi::l
and actual productivit)r te~t~ cannot be reconciled by
serial flow models used in the Monte Carlo calculation and must be explained by well stimulation 01' lacle
of stabilization during tests.
CONCLUSIONS
"
"
'
TABLE
ft-Icdel2
1l1cdell
"Georneir}' a"
Experiment
n
k" xr
(md)
1. .... " ..
2 .........
3 ........
4 .......
5 .......
6 ...... , ..
7 .........
8 .........
9 ...... .. .
10 .........
j; (bbl(
"GeomeiTY a"
"Geomefry b"
(bbl(
day(p,i)
.j;
k,,:o:p
(md)
j; (bbl(
day/psi)
"Geometry b"
Irexi
j; (bbl(
day(psi)
day/~si)
k""I'
(md)
0.233
0.322
0.448
0,462
0.442
0.435
0.391
0,465
0.435
0,491
0.0025
0.0034
0.0047
0.0049
0.0047
0.0046
0.0041
0.0049
0.0046
0.0052
0.232
0.320
0.459
O,4eO
0.455
0.440
0.396
0.469
0.439
0.508
0.0025
0.0034
0.0049
00051
0.0048
0,0047
0.0042
0.0050
0.0046
0.0054
LI50
0.783
0,603
0.590
0.513
0,486
0,452
0,455
0,482
0,448
0.0122
0.0083
0,0064
0.0062
0.0054
0.0051
0.0048
0,0048
0.0051
0.0047
0,357
1.240
0.872
0,769
0.698
0.604
0,564
0.523
0,482
0,472
0.0038
0.0131
0.0092
0.0081
0.0074
0.0064
0,0060
0.0055
0.0051
0.0050
20 .........
30 ..... ".
40 ....... '
50 .........
60 .........
70 ...... .. .
80 .... ... ..
90 ..... .. ".
100. , .......
1I0., .... '"
120 ........ ,
0.380
0.342
0.353
0.345
0.338
0.348
0.349
0.350
0.336
0.329
0.320
0,0040
0.0036
0.0037
0.0037
0.0036
00.037
0.0037
0.0037
0.0036
0.0035
0.0034
0.391
0.356
0.359
0.349
0.346
0.357
0.357
0.353
0.340
0.332
0.323
0,0041
0,0038
00038
0.0037
0.0037
0.0038
0.0038
0.0037
0.0036
0.0035
0.0034
0,488
0,471
0.501
0.516
0.544
0.525
0.515
0.505
0.534
0.540
0.561
0.0052
0.0050
0.0053
0.0055
0.0058
0.0056
0.0055
0.0053
0.0056
00057
0.0059
0,462
0,401
0.418
0.501
0.476
0,466
0.495
0,475
0,466
O,4N
0.48<1
0.0049
0.0042
0.0044
0.0053
0.0050
0.0049
0.0052
0.0050
0.0049
0.0050
0.0051
126 .........
127 ........ ,
128 ....... '
129 ......
130 ...... ..
131. ...... ..
132 ... , , ....
133 ........
134 ... " ....
135 ... ._-.- -
0.316
0.318
0.317
0.315
0.314
0.316
0.315
0.315
0313 '
0.312
0.0033
0.003i
0.0034
0,0033
0.0033
0.0034
0.0033
0.0033
0.0033
0.0033
0.319
0.320
0.319
0.317
0.316
0.319
0.318
0.318
0.316
0.315
0.0034
0.0034
0.0034
0.0034
0.0033
0.0034
0.0034
0.0034
0.0034
0.0033
0.562
0.559
0,559
0.563
0.565
0.564
0.563
0,565
0.5f.5
0.5('2
0.0059
0.0059
0.0059
0.0060
0.0060
0.0060
O.OOED
O.OOEO
O.OOED
0.0059
0,482
0.481
0,480
0.482
0.482
0,482
0,480
0.487
0,486
0.483
0.0051
0.0051
0.0051
0.0051
0.0051
0.0051
0.0051
0.0051
0.0051
0.0051
(md)
.' .
~.-
~<.
--
.0 : ' , . '
.~
';
.;
ACKNOWLEDGMENTS
,
(I
.,
,1
:,
i
several Monte Carlo programs, - N _ Collins for selection of specific applications and E. J _ Morin for reviewing the manuscript. The opinions expressed in
the paper, hOlrl'ever, al'e entire!)' those of the author.
REFERENCES
-I September,
1965. Montreal
(6) Collins, R. E., "Flow of Fluids Through Porous :i\rlaterials," Reinhold Publishing Corporatioll, New York,
1961, 270 p.
(7) JVal"J"en, J. E., and Skiba, F. F.J ui\faeroscopic Dispersion,'J Society of Petroleum Engineers Journal,
September, 1964, pp. 215 - 230.
(8) Hazen, S. Hr' J /7-., "Statistical Analysis of Sample
Data for Estimating Ore," Bureau of i\Iines, Vrashington, 1961, 27 p.
(9) Hewlett, R. F_, rlSimulating Mineral Deposits Using
Monte Carlo Techniques and Mathematical Models,"
Bureau of Mines, Washington, 1964, 27 p.
(10) Scheidegger, .A. E., "The Physics of Flow Through
Porous Media," University oj To-ronto Press, 1963,
313 p,
(11) Wan'en, J. E., and Pl-ice, H. 8., "Flow in Heterogeneous Porous Media," Society oj Pet"oleu7n Enginec?,s JOU?7u,Ll, September. 1961, pp. 153 - 169.
129
' ..
";:~."." .
.:::
: ....
.. ;