Вы находитесь на странице: 1из 8

SPE 121256

Application of the Monte Carlo Simulation in Calculating HC-Reserves


Zsolt Komlosi, Julia Komlosi, SPE, Partners for MOL

Copyright 2009, Society of Petroleum Engineers


This paper was prepared for presentation at the 2009 SPE EUROPEC/EAGE Annual Conference and Exhibition held in Amsterdam, The Netherlands, 811 June 2009.
This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect any position of the Society of Petroleum Engineers, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Society of Petroleum Engineers is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
One of the key features of E&P companies is their proved reserves in hydrocarbon deposits. Reserve estimation requires
knowledge of Initial Hydrocarbon in Place, technical reserves and economic conditions including annual cash flow estimation
in the forecast period. Since all parameters used in evaluation procedure are burdened by rather more than less certainty.
Therefore, in a sophisticated evaluation process, there should be determined not the expected values only (deterministic way),
but errors/uncertainty of estimation as well (stochastic way) applying Monte Carlo simulation.
The estimation procedure comprises three main stages (the third stage /economic modeling/ is not discussed in this paper). In
the first stage, key input data (e.g., area, thickness, porosity, and so on) are treated as statistical variables, and the result of the
simulation is probability distribution function of HCIIP. This is an input of next stage.
In the second stage technical reserves (recoverable resources) should be estimated. There could be several assumptions for
production procedure for a reservoir (as e.g., drive mechanism, hydrodynamic system, phase behavior of reservoir fluids, well
spacing, water injection, presence of pressure barriers etc). Each regime (i.e. scenario) can be modeled applying input
parameters as statistical variables. This method is named a multiscenario method in the literature. Simulation result for each
scenario is a probability distribution function (PDF). While, expected value of PDF reconstructs the deterministic result and
gives a basis for project evaluation, the width of PDF is proportional with uncertainty of the estimation. Estimating
probability of each scenario a combined technical reserve PDF can be derived. Its first percentile can yield proved reserve for
booking procedure after economic limit test.
Authors show some case histories how to apply method after a brief theoretical summary referring to SPE-PRMS accepted.
Introduction
The fundamental issue in reserve estimation is the volume of hydrocarbon that can be economically recovered from the
reservoir. This is a complex task. Experts of several disciplines should closely cooperate for reaching a good solution
moreover, we will always have only limited amount of information. This is the reason why we focus on the Monte Carlo
simulation (MCS) procedures in this paper (except the economic modeling). We expect that both experts of geo-sciences and
petroleum engineers will be interested.
As this paper will not cover profitability estimation, we can not speak about reserves according to international standards
but only about recoverable resources. But we wish to highlight that the subject of our analysis is closer to reserves than to
resources, therefore we will use the term of technical reserve as formerly used in the industrial practice: technical reserve is a
resource, which can be economically recovered using regular production technologies by expectation of technical experts, but
the profitability was not specifically analyzed.
Theoretical background
We will approach the problem in two directions in order that the goal set out in the title can be accomplished. First we will
describe the reserve assessment process, and, as a result, we will present some types of reserves and resources Secondly, we
will briefly describe the simulation method used for the stochastic modeling, and, within that, for the Monte Carlo simulation.
Finally, we will combine the said two directions through case studies.
We must admit that extremely wide and theoretically sound financial and banking processes were first considered when the
methods were developed, but in this process we had to realize that we had no chance for developing such a theoretical system.
We were striving for implementing a practical and user-friendly solution managing complexity of the problem and the limited
amount of information available. The authors have the view that any process description can only be developed through
permanent application and regular and sector-level analysis of results and lift it among international standards finally.

SPE 121256

Reserve estimation
There are several national and international standards for reserve estimation and qualification. Always the given company
decides which method it intends to apply depending regulation of the countries where company is operating. The authors have
the view that the Petroleum Resources Management System (see SPE (2007), Ross (2007)) can be deemed as the best method
to follow. The standard applies the following three key words or categories:
Reservoir [in-place volumes]; Project [production/cash flow]; Property [ownership/contract terms].
Reserve
Natural
conditions
HCIIP
Fluid
properties

Regional
environment

Reservoir rock

Rock
properties

Human
conditions
Technology

Depletion
technique

Geometry

Economics

Well pattern

Kuhn & Komlosi (1991) described the critical reserve factors


as it shown on Fig. 1. On the left side of the flowchart (blue cells)
one can see the (objective) parameters determined by nature,
while the right side (yellow cells) presents the parameters that
experts or decision-makers may select from with a certain
degree of freedom depending on natural conditions. So first we
have to estimate the volume of hydrocarbons assumed in the
reservoir, and then we should determine the projects we wish to
implement and the volume (i.e. the reserve) we expect to recover
as a result of these projects.

Fig. 1: Reserve critical factors flow chart

Term of reserves (and their categories) is a dynamic one. It is changing time by time as a function of project status and
information earned in the field development and production procedures. First we meet the initial status. We will gradually
obtain more and more information during the project development and production procedures, due to the logic of the system,
and it will enable us to determine the initial and actual status with much higher level of accuracy. This is clearly an iteration
process (first we can assume it is convergent). We will reach the ideal accuracy as we get closer to the finish of the production
process, in accordance with the rules of formal logic and as described in the professional literature (e.g. Garb 1988).
However, when Demirmen (2005) analyzed the relevant statistics he came to the conclusion that the said decreasing
uncertainty was a myth. The authors also share this view pursuant to their practical experiences. The most presumable reasons
behind is that reservoirs are so complex (and some reservoirs are much more complex) that if we really want to understand
them (and to prepare the 100% relevant exploitation model) the amount of the necessary information would need so much
expenses that makes entire production project uneconomic. Therefore experts should look for optimum solution between
information maximization and expenses minimization and dream about 100% accurate production model. Even at the end of
exploitation we cannot be assured in theoretical reserve figure since we dont know how much HC would have been produced
if we had selected another production regime or technology at the beginning.
Managing uncertainty of reserves there are defined some categories for reserves which can be applied for technical
reserves too as analogy. The most important category is the best estimate (2P). It is the basis for planning field development
and project profitability estimation. One part of the best estimate is the low estimate (1P) derived very conservatively. It has a
high probability (90%) occurrence and serves as an objective for company financial reporting (Etherington (2006)). Ratio of
low and best estimate shows some certainty index (CI) for reserve assessment. Being it is near to 100% that proves certainty of
best estimate reserve or, its a pity, weakness of simulation methodology applied.
The third category of reserves are the high estimate (3P) showing all hopes of experts with low probability occurrence
(10%).
Hydrocarbon initially in place (HCIIP)
Basically two kinds of approaches can be seen in practice. One of them strives for perfection; this is the isovol mapping
method, where the volumes of hydrocarbons over unit of area (=isovol) are calculated at a well on the basis of information and
measurement data obtained from the well or seismic processing. The process is as follows: (1) Isovols are determined well by
well (2) Isovols are interpolated to the area in between the wells and an isovol map is compiled with software or manually and
(3) Integral of the isovol map yields HCIIP.
Another solution is a simplified one. It can be used in lack of sufficient amount of data and/or time. It is a calculation of a
general-cylinder (having the same area as the reservoir) where the characteristic properties are represented by the average data
of the reservoir investigated:
HCIIP = A * h * POR * (1 SW ) / BHCi

Of course, the first solution provides a more reliable estimate on the HCIIP, yet at the same time, by showing the spatial
distribution of the storage properties of the formations, it provides key data for the production technology study as well.
Advantage of the simplified method is the flexibility and operability; furthermore it allows running Monte Carlo simulation.

SPE 121256

A third method could be the one when HCIIP is extrapolated from production history data. Result of these extrapolations is
uncertain too since measurement of volumes, temperatures, pressures are uncertain as well. Although, this is a good control for
data derived volumetrically.
Technical reserve (recoverable resource)
We can exploit only a part of the HCIIP (defined in the previous section). How big this part is, well, this is determined by
the relevant geo-technical circumstances and the selected technology. Several technologies can be applied under the given
conditions and each has its typical ultimate recovery ratio:
HCTR = HCIIP * UR

Such technology options, called scenarios have a development project has with own CAPEX needs. When we select the
method and/or the technology, we should take not only the produced volume into account but also expenses and profitability.
Any decision selecting a technology that offers higher reserve should be made on business basis. One must investigate whether
it that reserve secures higher profit as well.
Stochastic modeling, Monte Carlo simulation (MCS)
The data used for assessment are burdened by rather more than less certainty, as they (such as porosity, water saturation,
and area in case of HCIIP estimation) can only be determined with a level on inaccuracy. Therefore, the need has arisen to
assess the accuracy of the properties determined.
One of the solutions used in practice when in addition to the most probable version an optimist and pessimist version is
also calculated. However, is it sufficient? Let us take two parameters, porosity and thickness. Porosity may tend to follow the
best case, the worst case scenario or the realistic terms (these are 3 cases). Thickness may have the same three options (which
are further 3 cases). As porosity is not dependent on thickness, this yields 9 different options. When you take into account that
you work with a lot more than two parameters, quite a great number of versions required calculation.
It is not the calculation, which is difficult, as modern computers can cope with the task easily; more problematic is the
evaluation of the results. No decision maker can review and evaluate the several hundred versions calculated from the dozen
parameters.
The possible solution is the estimation of reserves using the stochastic modeling tool, Monte Carlo simulation that can also
be considered as a wide ranging combined sensitivity test as well. In the simulation process, key input data (area, thickness,
porosity, and so on) are treated as statistical variables, and the evaluation process uses not their most probable or expected
value, but with values sampled in accordance with their probability distribution. The foundation of MCS is the deterministic
calculation model.
During the preparatory phase, the analysts determine the probability distribution function (PDF) of the input parameters in
the light of the known geological and technical relationships. The probability theory describes several PDFs, but truly abstract
distributions that cannot be followed by the evaluator are better not to apply as they may distort the modeling efforts. It is
advisable to apply triangular and the similar PERT distribution in addition to the uniform distribution, normal distribution and
lognormal distribution. In the PERT case the occurrence probability of minimum and maximum parameter values equals zero,
and there is a maximum probability occurrence value in between the two. There is a curve resembling to the normal or
lognormal distribution obtained between the three respective points. A benefit of this approach is that the tailings outside the
minimum - maximum interval will not appear. (Fig. 2 shows the distribution function of lognormal and Pert distributions).
Not all of the parameters are independent from each other. For instance, in most formation larger porosity will entails
lower water saturation level under similar circumstances. The software
Pert
used by the authors (@RISK) allows for the definition of a correlation
Lognorm al
coefficient between any pair of parameters and sampling is being made
Min-Mod-Max
with due attention paid to this coefficient during the iteration process.
During the simulation procedures, the software will calculated several
hundred or several thousand times as programmed the HCIIP taking
samples in accordance with the predefined input probability functions and
it will gather the calculated results as well.
Finally after the run of all the cycles a frequency function is complied
using the calculated results, which can be used to derive characteristic
properties such as expected value, standard deviation, any percentile, and
so on.
0

100

200

300

400

Fig. 2: Comparison Pert and lognormal probability


distribution functions

In other words, the outcome of the calculations will not be one figure, the volume of the HCIIP but its PDF and any
properties derived from it.

SPE 121256

Useful service of MCS is sensitivity analysis which is a multivariable stepwise regression analysis showing how strong
is impact of an input on an output parameter. When regression co-efficient is near zero then no effect when it is near +1 or -1
then effect is strong by the given input parameter to the result. Coefficients are displayed in form of tornado charts (see later).
The complete procedure described above can be repeated for technical reserve estimation (and further for reserve
estimation or even for profitability modeling) running with relevant evaluation input parameters as HCIIP probability
distribution function, ultimate recoveries and production forecasts for various scenarios which can be defined by production
technology selected. The only speciality is at (technical) reserve assessment modeling, that some input variables are not
continuous. E.g. whether is there water drive or not, which can be assumes as two scenarios as mentioned by Ross (2007).
Analyst should develop two models: one for water drive and one for closed reservoir, and the results will be two PDFs.
Estimating probability occurrences of scenarios a final PDF valid for entire field can be derived.
Case histories
In the past years, estimation of HCIIP was carried out in a number of fields. The deterministic HCIIP was estimated by
mapping isovol parameters in accordance with the practice of MOL.
The simulation models were based on deterministic models in each case using the average values, standard deviations or
various percentiles derived from data (as porosity, net pay, water saturation, BHCi) applied for isovol mapped.
Authors assume these cases arent finished yet. Results are to be discussed by experts of the field and of various disciplines
and, finally, the model and/or input values should be justify. Each case is supplied with some comments (i.e. critics) showing
weakness of MCS modeling.
Supporting this analysis, there are some assumptions which can be applied as rules of thumb e.g.: (1) Volume/area is the
strongest uncertainty drive; (2) Water saturations error is bigger than porositys error in clastic reservoirs; (3) Water saturation
correlates with porosity at various confidential levels; (4) Porosity error is significant in low porosity reservoirs.
Study one of the cases was extended to direction of technical reserve assessment. Various technologies were assumed as a
more or less reliable solution to exploit reservoirs. Each technology with its special conditions was named a scenario during
discussion.
HCIIP cases
General assumptions applied at each field: (1) No overestimation in uncertainty. Authors assumes in this stage of
methodology development we should rather call attention of relevant experts and decision makers than frighten away them
with horrible figures; (2) Uncertainty of BHCi derived from laboratory measurement statistics if any; (3) Input of porosity,
water saturation, and net pay are described as a truncated normal distribution derived from petrophysical data applied at
evaluation of each reservoir.
Key properties of the case study fields are contained in the Table 1.
Field #1
Mean values of porosity, net pay and water saturation levels and associated standard deviation for the purposes of
simulation were calculated from the values determined for the individual formation sections. When determining the area error,
we started from a simplified model stating that the error in gas-water contact will result in some area error, in other word a
deeper laying GWC will result in larger reservoir area. Calculating from the inclination of the strata, 1 meter of inaccuracy in
GWC will result 6% inaccuracy in area calculations (see Fig. 3).
Table 1
Feature
OWC/GWC
Formation

r
Age
Porosity
Type of porosity
Water saturation
dr
dLKH
Type of trap
Area
Fig. 3: Dependence between GWC and area
estimation (Angle is taken from a seismic cross Effective thickness
section)
Number of layers
Number of wells
In the simulation Pert distribution was used Formation content
for each parameter. The calculated mean B ; B
oi
gi
values of porosity, net pay and water HCIIP (best estimate)
saturation levels were used as the medium

Field:
Unit
m

%
%
km2
m

MMm3

1900
2265
2250
Shaly Sandstone
Basalt
Turbidities
Middle Pliocene
Miocene
Miocene
22.4
17.7
12.3
Primary
Double
Primary
22
85
46
Anticline
Lithological Lithological
3.5
1.4
5.6
27
63
43
6
1
8
7
9
6
Gas
Oil
Wet Gas
6.3
1.27
200
2670
1.74
880

SPE 121256

value of the function, while the minimum and maximum values were determined as the 30% and 70% percentiles,
corresponding to an interval in standard deviation of a normal distribution.
Best estimate was 2670 MMm3, while the low estimation was 2465 MMm3 (CI=92%). Probability distribution of the
HCIIP can be seen on Fig. 4 and the sensitivity tornado chart on Fig. 5.
Critics: (1) Net pay impact is overestimated, but if we assume that net pay regulates area as well it can be accepted. (2)
Relation of impacts porosity and saturation is questionable. Opposite habit were expected (uncertainty of SW estimation is
higher than uncertainty of porosity estimation).
X <= 2465
10%

X <= 2901
90%

0.87

h, m
0.35

A,km2

0.25

UR

0.20

POR
SW -0.01
2200

2400

2600

2800

3000

-0.2

3200

Sensitivity
0

0.2

0.4

0.6

0.8

Fig. 4-5: GIIP PDF for Field #1 and sensitivity coefficients

Field #2
There were no available PDF of porosity, water saturation and net pay, so an artificial method was applied. Several
versions were prepared from the deterministic estimate of OIIP using various porosity cut offs. This way, reserve calculation
variables (area, net pay, porosity) changes as a function of the porosity cut offs, which was described by linear regression.
A minimum error (+0.6%) was used for porosity determination. Due to the lithological character of the deposit, the water
saturation estimate was considered to be the most uncertain factor. Based on the processing of petrophysical data, the most
probable saturation level was 85%, but experts held the view that 78% or 92% saturation levels were just as much possible.
Every parameter was described with Pert PDF.
Best estimate was 1.91 MMm3, while the low estimation was 1.39 MMm3 (CI=72%). Probability distribution of the HCIIP
can be seen on Fig. 6 and the sensitivity tornado chart on Fig. 7.
Critics: (1) Water saturation impact is overestimated; (2) Impact of area and net pay is missing. Correction of MCS model
is required, but PDF of OIIP seems to be reliable.
X <= 1.39
10%

X <= 2.23
90%

-0.98

SW

-0.14

PORcut

-0.11

Boi

POR
1

1.5

2.5

0.07

Sensitivity
-1

-0.8 -0.6 -0.4 -0.2

0.2

Fig. 6-7: OIIP PDF for Field #2 and sensitivity coefficients

Field #3
For reasons of too few wells available and the poor information given by the 2D seismic interpretation materials two
deterministic model versions were prepared, one optimistic and one realistic (percentiles 10%, 90% and mean value of area are
shown on Fig. 8 for each reservoir). For the purposes of simulation, average porosity, net pay and water saturation levels were
calculated in every well with the associated standard deviation, using the well log interpretation results sampled by 20
centimeters.

SPE 121256

X <= 766
10%

10%

#Reservoir

X <= 996
90%

Mean
7

90%
6

Area, km2

Fig. 8: Area scenarios for reservoirs


of Field #3

600

800

1000

1200

Fig. 9: GIIP PDF for Field #3

Normal distribution was assumed with regard to porosity and water saturation for the purposes of the simulation, but only
values within the interval of the standard deviation were allowed to be processed (the out-of-range values were truncated).
When dealing with net pay, it was assumed that the extent of supposed error would change between 0.6 and 1.5 meter, as a
function of the net pay.
Using the two types (i.e. optimistic and realistic) of deterministic isovol values a mean and standard deviation were
calculated.
Input parameters were formulated for each and every sampling operation during the iteration process and isovol values
derived from them, which then were compared to deterministic isovol parameters. The area necessary for the reserves was
assumed to have a uniform probability distribution between the optimistic and realistic estimates.
Best estimate was 880 MMm3, while the low estimation was 766 MMm3 (CI=87%). Probability distribution of the HCIIP
can be seen on Fig. 9.
Critics: Missing sensitivities due to accumulation data of 8 reservoirs. Reorganization of MCS model can solve the
problem.
Technical reserve case
Solution before production or at the beginning
Now we will have a close look on case in Field #3. During the initial production phase we could give fairly high
probability to a recovery ratio between 53-69% and 60% as the most probable figure. In this case we get the probability
distribution as presented in Fig. 10, and, as a consequence, the best estimate (analog with 2P reserves) will be 531 MMm3 the
low estimate technical reserve (analog with proved, 1P reserve) will be 460 MMm3 (CI=86%). The technical reserve
sensitivity for two input parameters are displayed on Fig. 11.
X <= 460
10%

X <= 606
90%

GIIP,
MMm3

0.89

0.46

UR

Sensitivity
0

300

400

500

600

700

0.2

0.4

0.6

0.8

800

Fig. 10-11: Technical reserve PDF for Field #3 and sensitivity coefficients

Solution with some production history


A study was prepared at recovery of 20-30% in Field #3, where various production scenarios were analyzed using the
material balance method. Deciding questions in connection with scenarios are as follows: (1) If all wells are producing the
same reservoir or each well produces a reservoir separated from the others? (2) Are reservoirs closed or are there any water
drive? (3) Is it planned to drill a new producing well or not? Table 2 presents the parameters of the various scenarios. The
probability distribution of the technical reserves is defined as product of the GIIP probability distribution and relevant recovery
ratio.

SPE 121256

Lets see first the scenario 4. It is the only one which doesnt need additional investment (i.e. the cheapest solution). Result
PDF is shown on Fig. 12, the best estimate will be 624 MMm3 the low estimate technical reserve will be 551 MMm3
(CI=88%).
Table 2
X <=551
10%

X <=704
90%

Scenario

WaterDrive

500

700

4
Several

Closed

Open

New Well

300

One

Reservoir No

Ultimate recovery, %

84

59

55

55

T echn. Reserve,MMm 3
Probability of occurence, %

486

674

736

624

38

13

44

900

Fig. 12: Technical reserve PDF for Scenario 4 in Field #3

The probability distribution of the technical reserves for scenario 13 can be seen on Fig. 13. Dashed lines represent PDF
of each scenario separately, while the thick line represent their integrated PDF derived with probability of occurrence (see the
bottom row in Table 2), as these can be used for modeling the behavior of the technical reserve in the all scenarios.
When the material balance was prepared, the technical reserve was estimated with the expected line of pressure and
cumulative production data series. Since data points are not fitted on a line exactly, petroleum engineers derived a (wrapping)
line for minimum pressures and one line for maximum pressures in case of scenario 1 and 2. These lines yielded minimum and
maximum ultimate recoveries for both scenarios. Generating PERT distribution functions from those data-triads and
performing a simulation result PDF was yielded (see Fig. 14). One can see that in case of scenario 1 there is a good
coincidence between two PDFs, but dispersion of technical reserves derived from material balance is less in scenario 2 than
dispersion of PDF derived volumetrically.
X <=533
10%

X <=779
90%

TR(1)@GIIP
TR(2)@GIIP

Scen-1-3
Scen-2

TR(1)@MatBal.
TR(2)@MatBal.

Scen-3
Scen-1

300

500

700

900

Fig. 13: Technical reserve for scenarios 1, 2, 3 (dashed lines)


separately and for all (thick line) in Field #3

300

500

700

900

Fig. 14: Comparison PDF of technical reserves derived


from GIIP and material balance evaluation for scenario 1
and 2 in Field #3

One can ask authors which technical reserve (see Fig. 15) is valid for
the Field #3? There is no one answer. There are answers depending on
circumstances (remember right side of Fig. 1) as follows:
(1) No decision-making yet scenario 14;
(2) Decided to drill a well scenario 13;
(3) Decided not to drill any well scenario 4.
Best estimate of the ultimate recovery in scenario 13 is higher by 53
MMm3 and in scenario 3 is higher by 112 MMm3 than best estimate of
technical reserves in scenario 4 (see Fig.15), but the latter one doesnt
requires additional investment (CAPEX). So profitability estimation helps
to make decision.

900

Tech.Reserve,
3
MMm

High
Best
Low

800

700

600

500

14

13

Scenario

Fig. 15: Technical reserve versions in Field #3

SPE 121256

Conclusions
Monte Carlo simulation is a good tool for (technical) reserve assessment. It generates reserve categories (1P, 2P, 3P)
objective on the basis of simulation model developed by (subjective) experts.
Introduction of this system should be done gradually since experts and decision makers have to get used it. Experiences
of introduction should be collected an analyzed in order to development good methodological regulation.
Certainty indices are fairly high (72-92%) due to underestimated uncertainty factors. Assuming reserve assessment
control statistics realistic approaches had to yield lower CI values.
PDF of volumes in place is a continuous function, while PDFs of (technical) reserves are multiplied by various scenarios,
but they can be combined (with simulation) into one PDF.
Dont forget basic rule: Monte Carlo simulations accuracy is determined by the reliability of the geo-technical model,
parameters and conditions forecasted (garbage in, garbage out).
Nomenclature
A
=
area, km2
=
formation volume factor (FVF) of HC
BHCi
CAPEX =
Capital expenditure, $m
dLKH
=
deviation of lowest known hydrocarbon, m
dr
=
deviation of radius of area, m
h
=
net pay, m =H*N/G
H
=
Gross reservoir thickness, m
HCIIP
=
Hydrocarbon initially in place, Mm3
HCTR
=
Hydrocarbon technical reserve, Mm3
MM
=
Million (prefix)
N/G
=
Net / gross thickness rate
PDF
=
probability density function
POR
=
porosity
SW
=
water saturation
UR
=
ultimate recovery
Acknowledgments
The author would like to thank Hungarian Oil and Gas Plc. (MOL) for permission to publish this paper.
References
Demirmen, F. (2005): Reliability and Uncertainty in Reserves: How the Industry Fails, and a Vision for Improvement SPE
HEES, Dallas 3-5 April (SPE 94680)
Etherington, J. (2006): Integrating Reserves and Resource Assessment with Portfolio Management A Hybrid Approach
SPE Distinguished Lecturer Series
Garb, F.A. (1988): Assessing Risk in Estimating Hydrocarbon Reserves and in Evaluating Hydrocarbon-Producing
Properties JPT jun. p.765-778. (SPE 15921)
Kuhn T. - Komlosi Zs. (1991): How to Estimate the Reliability of Reserves? Quick and Practical Solution 3rd International
Reservoir Characterization Technical Conference, Tulsa Nov. 3-5. PennWell Books, Tulsa, p.769-780.
Ross, J. (2007): Understanding and Applying SPE-PRMS Training Course Material prepared for MOL Group Exploration
& Production, Budapest, Hungary
SPE (2007): Petroleum Resources Management System Sponsored by SPE, AAPG, WPC and SPEE. March

Вам также может понравиться