Copyright 2001, Society of Petroleum Engineers Inc.

This paper was prepared for presentation at the 2001 SPE Annual Technical Conference and
Exhibition held in New Orleans, Louisiana, 30 September3 October 2001.

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, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes 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 where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
The estimation of oil and gas reserves is necessary for
assessing the value of oil and gas properties, and for meeting
the reporting and disclosure requirements imposed on oil and
gas companies by various governmental and regulatory
bodies. Reservoir simulation is a sophisticated technique of
forecasting future recoverable volumes and production rates,
and is an increasingly popular tool for reservoir management
and optimization. Reservoir simulation is also progressively
being used in the process of forecasting and estimating
reserves. However, as with any reservoir engineering
technique, certain precautions must be taken when relying on
reservoir simulation as the means for estimating reserves.
This paper identifies some of the unique issues encountered in
the reserves evaluation process when reservoir simulation is
utilized, and offers strategies for addressing these issues.

Introduction
The assessment of a companys reserves is an important
process, regardless of the companys size or corporate
structure. The level of reserves can have a direct impact on a
companys earnings and balance sheet, and can significantly
affect the cost and availability of capital necessary for its
growth.
Reserves are determined using a variety of geological and
engineering methods. Regardless of the evaluation methods
used, however, any estimate of future recovery, no matter how
reasonable, does not necessarily qualify as an estimate of
reserves. Specific criteria must be met to qualify estimated
recoverable volumes as reserves. These criteria are generally
defined in the form of Reserves Definitions.
For an exploration or production company, the value of the
reserves usually constitutes a major portion of the total
company value. Therefore, it is critical that reserves be
estimated as accurately as possible within the constraints
imposed by the relevant Reserves Definitions. As long as the
requirements of the definitions are upheld, reservoir
simulation can be a valuable tool for improving the accuracy
of the reserves estimate. However, due to the nature of
simulation models (their non-uniqueness and complexity) their
use as a tool for this purpose is not always straightforward, as
this paper will discuss.

Discussion
Reserves Definitions
Several sets of Reserves Definitions have been published by
various regulatory bodies and professional organizations
throughout the world. Most sets of Reserves Definitions
specify different grades of reserves depending on the level of
certainty associated with the estimated recovery of those
reserves. In many systems, the top grade of reserves is
designated as proved reserves. As would be expected,
proved reserves require a very high degree of confidence that
the hydrocarbons will actually be recovered. Lower grades of
reserves such as probable or possible require diminishing
standards of certainty.
Examples of two sets of definitions commonly
encountered by companies based or operating in the United
States are 1) those established by the United States Securities
and Exchange Commission (SEC), and 2) those established by
the Society of Petroleum Engineers and the World Petroleum
Congress (SPE/WPC). There are many similarities between
these two sets of definitions, but there are important
distinctions as well which are beyond the scope of this
discussion.
Publicly held oil and gas companies in the United States
must file specific forms with the SEC on a regular basis and
whenever stock is issued. Some of these forms require that
the company disclose volumes of reserves and the estimated
future income attributable to those reserves. SEC regulations
stipulate that the reserves estimated for such purposes must be
computed in accordance with the definition of proved reserves
contained in Part 210.4-10 (a) of Regulation S-X.
On the other hand, some lending institutions, financial
advisors, and corporate planners work with reserves estimated

SPE 71430
The Adaptation of Reservoir Simulation Models for Use in Reserves Certification
Under Regulatory Guidelines or Reserves Definitions
M.R. Palke and D.C. Rietz, Ryder Scott Company

2 M.R. PALKE, D.C. RIETZ SPE 71430
in accordance with the SPE/WPC Reserves Definitions.
Because these definitions allow for probable and possible in
addition to proved reserves, it is feasible to express the
reserves in terms of a range of possible outcomes, as implied
by the amount of reserves allocated to each classification.
Different grades of reserves are used for different
purposes. In general, proved reserves are used for financial
reporting and lending, where the requirement for certainty is
the greatest. However in order to make intelligent business
decisions in activities such as prioritization of capital spending
and property acquisitions, it is also important to recognize and
quantify the amount of probable and possible reserves.
Companies buying or selling assets regularly consider
upside potential, which is usually represented as probable
and possible reserves. The degree to which entities value
these non-proved categories varies dramatically due to the
reduced certainty in recovery of the volumes.
The SPE/WPC definition contains a general requirement
that proved reserves have a reasonable certainty of being
recovered. Other, more specific, criteria must also be met for
reserves to be classified as proved. The definitions for
probable reserves are less stringent, requiring that a general
test of more likely than not be satisfied
1
.
This paper focuses on deterministic reserves estimation
incorporating reservoir simulation modeling results. Many of
the issues apply also to the probabalistic reserves setting, but
many other issues, not discussed herein, are unique to
probabalistic reserves estimation
2
.

Prevalence of Reservoir Simulation
One of the advantages of reservoir simulation is that it enables
an engineer to simultaneously and rigorously consider almost
all of the geological and engineering data pertinent to the
production behavior of the reservoir
3
. In essence, reservoir
simulation is the construction of a numerical model that is
expected to behave like a particular oil or gas reservoir.
Certain properties of the reservoir (e.g. porosity, permeability,
structure, thickness, etc.) and certain properties of the fluids
contained in that reservoir (viscosity, density, etc.) are
described in numerical terms suitable for input into a
simulation package. Once constructed, a reservoir model is
typically history matched. After the history match is achieved,
the model can be run to predict future performance under a
variety of future development and operating scenarios.
Simulation has become increasingly practical due to
advances in computer software, hardware, and simulation
expertise. There are a number of stable and efficient
commercial programs available, each with its own strengths
and weaknesses. Simulation involves a great deal of
numerical computation; however, as computer power has
continued to increase, it has become feasible to run
increasingly detailed models on relatively inexpensive
hardware. Familiarity with simulation has become more

common as engineers have acquired simulation training and
experience earlier in their education and careers than their
counterparts of previous generations.
We have observed that reservoir simulation has been
increasingly promoted as a means to estimate reserves. It
appears that the use of simulation for reserves estimation is a
logical step in the continued evolution of reserves estimation.

Capabilities and Limitations of Simulation
As with any analysis tool, it is important to recognize the
limitations as well as the capabilities of reservoir simulation.
Only then can model results be appropriately interpreted for
estimation of reserves or for any other purpose. As discussed
below, the accuracy of most models is impacted by the
presence of complexities within the reservoir that are not
incorporated into the model.
A model is composed of discrete cells or grid blocks, each
of which represents a specific volume of the reservoir. Each
grid block is assigned certain rock properties that either 1)
dictate the amount and types of fluids originally contained in
that portion of the reservoir (e.g. pay thickness, porosity,
connate water saturation, etc.), or 2) control the movement of
fluids into and out of that portion of the reservoir (e.g.
permeability, net-to-gross ratio, relative permeability, etc.). In
a model, these parameters are uniform within any given grid
block, whereas in nature there may be considerable variation
within the volume of the reservoir represented by the grid
block. At best, the model block will be assigned values
approximating some type of average for the reservoir over the
volume of the block.
However, it is not always possible to determine the
average block properties with a high degree of reliability
because of the difficulty in directly measuring many of the
reservoir parameters, and the inherent sparseness of the data.
In some instances, undetected reservoir boundaries or
structural features will not be represented in the model.
Although geostatistical methods help to improve estimation of
the properties in areas of the reservoir lacking measured data,
it is generally not possible to overcome the limitations of the
data set with a high degree of certainty.
Because of the above factors, a model is generally less
heterogeneous than the reservoir it is intended to represent.
Unfortunately, heterogeneity in a reservoir leads to uneven
sweep and incomplete drainage, both of which reduce the
recovery of hydrocarbons. Consequently, since a model
generally understates the degree of heterogeneity present in
the reservoir, the model will tend to overstate the recovery of
hydrocarbons unless compensatory adjustments are made to
the simulation data. For example, homogeneity may be
overestimated, but may be held in check with a lower than
actual overall permeability value or using adverse pseudo
relative permeability curves. Such adjustments are usually
made to achieve a history match. The history matching
process is discussed in greater detail later in this paper.

THE ADAPTATION OF RESERVOIR SIMULATION MODELS FOR USE IN RESERVES CERTIFICATION
SPE 71430 UNDER REGULATORY GUIDELINES OR RESERVES DEFINITIONS 3
Applying Simulation Results to
Estimate Proved Reserves
Reservoir simulation studies are rarely performed with the
objective of estimating proved reserves. Usually, the primary
objective of reservoir simulation is to improve the
understanding of the reservoir, so that the reservoir may be
optimally developed and managed. Frequently, a development
plan based on proved reserves alone (as compared to the most
likely volumes) would under-develop the field, and in fact
could result in reduced overall production (over the depletion
history of the field). Thus it makes sense to consider the best
estimate of total potential when making development and
management decisions.
As a result, the presumed most likely scenario is most
commonly modeled with the reservoir simulator. Most
likely is a level of recoverable volumes that is more
consistent with proved + probable reserves, rather than proved
alone. This is mainly due to the specific definitions of proved
reserves. One can strongly believe (most likely) that volumes
are present in the subsurface, but based on the proved reserves
definitions, they may not be recognized for proved reserves
estimation. Therefore, it is very common that results from a
simulation model cannot be directly applied to the proved
reserves category, even if they are passed through a cashflow
analysis to prove economic viability.
It is not just original hydrocarbon in place that may not fit
the definition of proved reserves. Models may include
pressure support from aquifers or rock compressibility that are
not proved. Numerous other parameters would also fall into
this category. The key is to search for sources of reservoir
drive energy that are not proven to exist. As will be discussed
later, sensitivity studies can be used to root out such
parameters from a history matched model.
Even though reservoir simulation is usually undertaken to
forecast recovery and future production rates under the most
likely geological scenario (which may not conform to the
proved reserves definition), simulation can still be utilized in
the estimation of proved reserves if appropriate steps are
taken. There are two approaches for applying the simulation
technique in such a way as to ensure that the definition of
proved reserves is honored in all respects.
First, the model can be modified such that the reservoir
configuration described by simulation data is in strict
compliance with the clear mandates of the proved reserves
definitions. Consider, for example, the case of a reservoir for
which the level of the hydrocarbon-water contact has not been
established from the geological and engineering data. In this
situation, the hydrocarbon-water contact in the model should
be set at the lowest observed occurrence of hydrocarbons
(lowest known oil/gas, or oil-down-to), as specified in the
definition of proved reserves. As long as the other
components of the definition are also honored, the results
generated from this model could be utilized in the estimation
of proved reserves.
The act of modifying a proved + probable model into a
proved model or vice versa can be more difficult than it
sounds. Such a modification to a proved + probable model is
not merely a question of modifying the reservoir description.
The planned wells and facilities (manifested through field rate
constraints) must also be adjusted to fit the proved reserves
categories. For instance, if a proved + probable model is
modified to be consistent with proved hydrocarbon in place,
the recovery factor might tend to be too high if all of the
probable wells are left within the model.
In addition to the question of constraints, substantial
modifications to the original grid/description could also be
required. For instance, models derived from seismic data
often feature thickening between wells based on reasonable
interpretations of the data. This thickening may or may not be
permitted under the reserves definitions.
We do not recommend, however, that models be originally
constructed to comply with proved reserves definitions
except under special circumstances (litigation, contentious
reserves estimation situations, etc.). In general, constructing a
model in compliance with proved definitions might regrettably
eliminate a great deal of utility in terms of planning and
optimization. It seems likely that it is better to build the larger
model (proved + probable + possible), and then in turn modify
the model, rather than construct a proved model, and have to
add on probable reserves later.
Alternatively, less desirable but perhaps more practical, a
model that is not in compliance with the proved reserves
definition may be used in the estimation of reserves with
appropriate modifications to the simulator results. In this
situation, the reserves evaluator must perform calculations
external to the model to estimate the recovery that reasonably
approximates what would have been calculated had the model
actually been constructed in accordance with the proved
reserves definition. An abundance of simulation output data is
necessary in order to perform the required adjustments.
Obviously, some of the rigorous nature of the simulation is
lost in the manual process of manipulating the output data.
An example of this second approach occurred, when the
authors were requested to opine on the use of a simulation
model of a field for proved reserves estimation. The field
consisted of stacked reservoirs which were not in vertical
communication except through pipe. The problem arose
because the volumes in the various sands were associated with
different reserves categories based on whether the fault block
they were in had been production tested. Furthermore, the
model volumes were not identical to the volumes certified by
the authors geological associates (due to differences in
interpretation of contact depths and contouring). The model
was calibrated so that the production tests could be reliably
replicated by the model.
The solution used by the authors was to separate the
production streams from the various sands. Production
streams from sands that did not qualify as proved were
eliminated. The remaining proved streams were scaled so that
their initial rates and ultimate recovery factors were preserved,
but the ultimate recovery factor was appropriate for the
certified OOIP rather than the OOIP present in the model.

4 M.R. PALKE, D.C. RIETZ SPE 71430
While this approach is not completely rigorous, it at least
relies upon rate forecasts and recovery factors predicted by a
well constructed model. This approach treats the simulated
reservoirs as analogies to the actual reservoirs in terms of
initial rate and recovery factor. It is felt that this approach
does meet all of the requirements of proved reserves
estimation, which the original model, no matter how well
constructed, did not.

Immature Reservoirs
The status of a reservoir represented by a simulation model
can range from immature to mature depending on the
stage of depletion at the time the model is to be used for the
estimation of reserves. For the remainder of this paper, we
shall discuss the considerations for using models at these
extreme ends of the maturity spectrum, although many
reservoirs will actually fall between the two extremes. As will
be evident from this discussion, the reserves estimator will
require an understanding of the concepts involved in reservoir
simulation as well as experience in the application of the
Reserves Definitions in order to utilize the model results
properly for this purpose.
Immature reservoirs have limited or no actual production
history. In some cases, the history is limited to formation tests
of various wells.
When building models of such reservoirs, it is necessary to
rely primarily on geophysical and geological data to set the
reservoir size and characteristics. A history match of the
model to the reservoir is easy to obtain since there are few if
any performance points to be matched. Because it is so easy
to obtain, however, the match is not very meaningful in terms
of calibrating and improving the reliability of the model.
It is unlikely that the most likely estimate of
hydrocarbons in place will adhere to the pertinent guidelines
for proved reserves. The proved reserves definitions contain
specific rules that limit the reservoir area to be considered in
the reserves assessment. These rules place restrictions on
assumptions regarding the position of the water-hydrocarbon
contact, the extent of the reservoir defined by logged and
tested wells, and so forth. Furthermore, the model is probably
being constructed for purposes such as optimization, or
facilities design, rather than proved reserves estimation alone.
As a result, most models of immature reservoirs are unlikely
to be acceptable for the purposes of proved reserves
estimation.
However, even models not built in accordance with the
Reserves Definitions are helpful for estimating the
hydrocarbon recovery efficiency to be used in the volumetric
calculation of reserves. Important to this process is the use of
the model to study the sensitivity of recovery to certain
reservoir characteristics not explicitly governed by the
reserves definition (e.g. relative permeability, drive
mechanism, pore volume compressibility, etc.). The use of
simulation in this fashion is referred to as performing
sensitivity studies. The understanding gained by sensitivity
studies will aid in the determination of the hydrocarbon
recovery efficiency that is reasonably certain to be attained.
Unless contradicted by analogy data (or experience) it would
be legitimate to use a value of recovery efficiency established
by such means in the estimation of proved reserves,
particularly if one treats this information in their analysis as
another analogous reservoir.
Furthermore, in many cases it is simple enough to modify
the non-proved-volumes model that was used for planning
purposes, to contain a proved volume in place for the purposes
of a proved reserves forecast.

Mature Reservoirs & History Matching
A mature reservoir is one that has produced long enough to
develop well-established production and pressure trends. A
history match of a model of a mature reservoir is usually more
difficult to obtain than for an immature reservoir, but is more
meaningful in terms of enhancing model reliability.
There are two basic components of a successful history
match. First, the model properties should be set such that
simulated reservoir pressure levels are reasonably close to
measured field pressures at the proper time and location within
the reservoir. Second, model properties should be set such
that simulated produced fluid ratios (water cuts and gas-oil
ratios) and contact movements are reasonably close to the
observed fluid ratios and contact movements overall, and in as
many individual wells as is possible.
A match to pressure alone is necessary but may not be
sufficient to definitively establish the size of the reservoir. For
example, a small reservoir connected to a gas cap or a large
aquifer can exhibit pressure behavior similar to a large
reservoir connected to a small aquifer. Furthermore, the
reservoir pressure behavior is sensitive to fluid composition
and phase changes. This means that any error in the fluid
characterization can cause a significant error in the size of the
reservoir required to achieve the pressure match. Thus, unless
the presence and size of a gas cap and aquifer are known from
independent geological information and the fluid
characteristics are known with reasonable certainty, the
pressure match alone does not guarantee a unique and correct
solution for the reservoir size.
On the other hand, a model that matches both pressure
behavior and produced fluid ratios has a much better chance of
providing a correct representation of the actual reservoir. A
match to observed pressures and water cuts for example, will
require a reservoir/aquifer combination of particular size and
relative proportions, thus reducing the range of possible
solutions.
The quality of a history match is also an important
consideration for gaining confidence in a models ability to
predict future reservoir behavior. For example, a model with a
good field history match, but a poor well-by-well match,
would generally be less trustworthy than a model where both
the field data and the individual well data were matched.
Further, if a well-by-well match was obtained by adjusting the
model properties in the vicinity of existing wells, the model
would not be expected to give reliable predictions for
THE ADAPTATION OF RESERVOIR SIMULATION MODELS FOR USE IN RESERVES CERTIFICATION
SPE 71430 UNDER REGULATORY GUIDELINES OR RESERVES DEFINITIONS 5
conditions significantly different from those already
encountered (e.g. the drilling of infill wells). Whenever
possible, localized changes used in the history match should
be propagated in a reasonable fashion to new well locations.
The history match technique should be focused on making
logical adjustments to model parameters that are consistent
with geological and engineering evidence.
We should recollect however, that history matching is
generally a somewhat subjective process, and it is unlikely
that any two engineers would arrive at the exact same solution.
Furthermore, it is normal that certain parameters that have a
limited impact upon the history match would have a dramatic
impact upon the predictions from the same model. Aquifer
dimensions are perhaps the most obvious of such parameters.
It may be surprising, but original hydrocarbon in place is
frequently another such parameter!
Thus the authors strongly recommend that any parameters
suspected of falling into this category be tested through the
use of sensitivity studies so that the engineer who estimates
reserves can study the possible range of future recovery
forecasts.
Many times the OOIP modeled for a history matched
mature reservoir does not comply with the Reserves
Definitions. For situations where a model has been history-
matched to field pressures and fluid ratios, and where there is
virtually no other way to match observed reservoir behavior,
the model would generally be a reliable tool for estimating
proved reserves. Sensitivity studies are a reliable tool for this
analysis. In all other cases, the size of the reservoir in the
model should be scaled back as necessary to prevent the
modeled reservoir from exceeding the size implied by the
criteria of the Reserves Definitions.
It would be wise to remember that reserves from very
mature reservoirs are often based solely on the application of
empirical decline curve methods, and are not necessarily tied
to geological-volumetric analysis of original hydrocarbons in
place. In such a case, it is not critical that the original
hydrocarbons within the model comply with the Reserves
Definitions. Thus, in cases where there is compelling
evidence that the model is representative (reasonable
predictions, and a high quality history match) the model
results should be used in a manner consistent with the use of
traditional production trend analysis, even though the original
hydrocarbon in-place cannot be fully reconciled with the
reserves definitions. Meaning that the model should not be
discounted because every detail cannot be explained.
However, this is only appropriate for cases where reserves
could otherwise be assigned without respect to volumetric
analysis. As usual, the model results should be considered in
the context of all available data.

Model Appropriateness
When using a model to estimate reserves, it is imperative that
reasonable assumptions be made with regard to future
development and operations of the reservoir. As with any
determination of proved reserves, environmental, regulatory,
contractual, market, and other types of practical factors must
be taken into account when estimating the likelihood and
timing of future development and operational changes.
Assumptions related to future wellhead pressures, the drilling
of future wells, etc. should be made with these practical
factors in mind.
It is also important to recognize situations where the
physical processes governing reservoir behavior are expected
to be different in the future than they have been in the past,
and to adjust expectations for the model accordingly. For
example, if a model has been matched to the history of a
reservoir that has produced primarily under solution gas drive,
there will be less confidence in that models ability to predict
reservoir performance under waterflood than if the reservoir
were to continue to produce under solution gas drive. The
match under solution gas drive provided assurance that the
factors important under that depletion regime (such as gas-oil
relative permeability functions) were reasonably correct in the
model. If there were little or no water movement or
production to match in the past, however, the model would not
have been tested for the adequacy of factors such as the water-
oil relative permeability functions, which would impact the
performance under waterflood. Observations from analog or
nearby fields or laboratory test data could be incorporated into
the model to improve the confidence when forecasting under
different depletion mechanisms.
As a final check, the evaluator should verify that the
transition from historical to predicted production is smooth if
the model is run as a status quo, or do nothing case. An
abrupt change at the end of history is indicative of an
inappropriate model, even if the history match appears to be
reasonable in all other respects.

Mixed Reserves Categories
Thus far, we have focussed on down-grading a proved +
probable model for use in proved reserves estimation. There
are many other permutations that are interesting. In general, it
is best that an acceptable proved + probable or proved +
probable + possible model be developed, history matched, and
forecast. This model would then be modified for the higher
category runs. The probable reserves would be the increment
between the proved and proved + probable reserves (as
determined through cashflow analysis of the simulation
results), and the possible reserves, if desired, would be the
final increment between proved + probable + possible
reserved and the proved + probable reserves. As a warning
though, removing hydrocarbon volumes from a history
matched models can have unforeseen consequences that must
be investigated. The authors have seen cases where rate
forecasts from proved reserves reach higher rates than the
reserves forecasts from the history matched proved + probable
model, despite featuring less OOIP and fewer wells. This
could be caused by mobility contrasts between the cases, and
enhanced contact with the aquifer when bands of hydrocarbon
are removed from a history matched model

6 M.R. PALKE, D.C. RIETZ SPE 71430
There are other means of modeling the different categories.
One is to create separate models. One that contains only
proved category hydrocarbons, a second that includes only
probable hydrocarbons, and a third (if so desired) that contains
only possible category reserves. The models would be run
separately, and reserves estimated based on three separate
streams. This approach is less desirable however, because it
would tend to underestimate interference between categories
through the reservoir and through facilities and wellbores.

Conclusions
In general, simulation results should be treated as if they are
actual results from an analog field. If the simulation model is
very detailed, properly constructed, and well history matched,
then the model can be treated as a nearly perfect analog. If the
model or its history match are less impressive, then the results
can be treated as a less directly comparative analog. Thus it is
our conclusion that when incorporating simulation modeling
results into reserves estimation, the model should be treated as
additional data, rather than the sole source of data.
Additionally we reach several more particular conclusions
related to the incorporation of simulation modeling results into
reserves estimation.
1. Although reservoir simulation is a sophisticated technique,
it does not always produce reliable or applicable results,
especially in situations where data are sparse or of poor
quality.
2. Reservoir simulation should be used to improve the
understanding of a reservoir, but should not be used to
circumvent the terms of the Reserves Definitions. Reservoir
simulation is increasingly being used as a tool for reserves
estimation, but reservoir simulation results do not necessarily
constitute reserves estimates.
3. Simulation models are typically designed to capture the
most likely reservoir description.
4. Because most likely is a level of confidence generally
associated with proved + probable reserves, models are
generally not designed to estimate proved reserves.
5. Aside from OOIP/OGIP, other types of reservoir energy
may be present in a model that do not qualify as proved, or
even probable in nature.
6. Models that are not in compliance with proved reserves
definitions can be modified to comply with the definitions, but
this process may be difficult.
7. Such modifications (# 6 above) may require substantial
alteration of the simulation grid/description, and require
attention to the development plan (wells and constraints)
applied to the model.
8. Results from models that are not in compliance with proved
reserves definitions can also be used through the alteration of
the simulation output itself. This requires a great deal of
simulation output and may provide less rigorous solutions.
9. For immature reservoirs, simulation is useful primarily for
estimation of the hydrocarbon recovery efficiency, and to test
the limits in terms of uncertain parameters (permeability,
aquifer support, OOIP, OGIP).
10. Some parameters will be uncertain, even in a history
matched model. These parameters may strongly influence the
prediction mode results. The impact of uncertain parameters
should be studied through the use of sensitivity runs.
11. Models of mature reservoirs should feature reasonable
history matches before they are accepted for reserves
purposes. The uniqueness and the quality of the history match
affect the confidence to be placed in a models ability to
predict future performance, and thus dictate the models
appropriate usage in the process of estimating reserves.
12. If the original hydrocarbon in place violates any reserves
definitions, sensitivity studies should be used to assure that the
original hydrocarbon in place is strictly necessary for history
matching before a model is accepted for reserves estimation.
Under certain conditions, history-matched models can be
useful for confirming and resolving the in-place volume and
the recoverable reserves for mature reservoirs.
13. When using a model to estimate reserves, it is imperative
that reasonable assumptions be made with regard to future
development and operations of the reservoir.
14. Care must be taken in estimating reserves when a model is
used to assess the impact on recovered volumes caused by the
introduction of a process not included in the history match of
the reservoir.
15. A do nothing or status quo predictive run is required to
test a history matched model. If a history matched model does
not smoothly transition between the historical data to the
predictive mode, it may need to be modified before it can be
used to estimate reserves.

Acknowledgments
We extend our thanks to Kent Williamson, Don Roesle, and
Ron Harrell for valuable assistance granted writing this paper.
We also thank Ryder Scott Company for supporting the
publication of this paper.

References
1. Petroleum Reserves Definitions, Published by the SPE and
WPC, Richardson TX (1997).
2. Acuna, G.H., and D.R. Harrell: Adapting Probabalistic
Methods to Conform to Regulatory Guidelines, paper 63202
presented at the 2000 Annual Technical Conference and
Exhibition, Dallas, Texas, October 1-4.
3. Mattax, C.C. and Dalton, R.L.: Reservoir Simulation,
Monograph Series, SPE, Richardson, TX (1990).