INTRODUCTION TO RESERVOIR PERFORMANCE PREDICTIONS

BY SIMULATION APPROACH
Prof. T. Kumar
Dept. of Petroleum Engg.
IIT(ISM) Dhanbad 826004

1. Introduction

A reservoir study concerns the factors governing the behaviour of a petroleum reservoir, the
end result of which is a recommended operating program which will yield the maximum net
income from the property.
Complete reservoir studies usually require months, or even years to complete,
depending on the complexity of the reservoir and the thoroughness of the study. It should be
emphasized that a thorough reservoir study cannot be completed in a few days or a few
weeks.
So-called "thumb-nail" reservoir studies, requiring only a few days or a few weeks to
complete, are often made, but these studies are of value only to indicate trends in reservoir
behaviour, or to determine the need for a full-scale study.

2. Parts of reservoir study

Stated broadly, a reservoir study is comprised of three basic parts:

(1) Matching past behaviour,
(2) Predicting future behaviour under various possible operating programs (various fluid
withdrawal rates, water injection, gas injection), and
(3) Recommending the best overall operating plan for the reservoir.

3. History matching

The first step in any reservoir study after the data have been assembled is the matching of
past behaviour. Referring to Figure 1, the solid lines represent the actual pressure-production
and gas-oil ratio history of the reservoir.
 If a reliable reservoir study is to be made the amount of past production from the
reservoir must be substantial. The exact amount of past history which has to be accumulated
before a reservoir study can be considered reliable is difficult to establish due to the many
variable factors involved.
However, a rule of thumb which many practicing reservoir engineers use is that it is
safe to predict ahead the same number of years for which past production history is available.
In other words, if a reservoir has been producing for 10 years then it should be safe to predict
ahead for 10 additional years. As more and more data are accumulated the accuracy of
reservoir predictions will normally increase.

4. Predicting future performance

Starting at a time when N, = 0, and using the data which have been accumulated, a series of
reservoir performance predictions are made. No attention at all is paid to the actual reservoir
history. The reservoir performance predictions are made to cover the time from original
production to the current data.
These data, the "predicted" performance over the past producing history of the
reservoir are then plotted on Figure 1, which also shows the actual production history.
A typical plot is shown by the dashed lines in Figure 1. In the illustration, the calculated
reservoir performance does not match the actual reservoir performance. This is often the
result of the first attempt.
It can be seen from Figure 1 that if future predictions of reservoir behaviour had begun
immediately, in all probability the predicted future behaviour would not have matched the
actual future behaviour. If the predicted past behaviour does not match the actual past
behaviour, there is no basis for believing that the predicted future behaviour will match the
future actual behaviour.
5. Reasons for mismatch of predicted and actual behaviour of the reservoir

It is now necessary to determine why the predicted and actual past production histories do
not match. There are two possible sources of error: (1) either incorrect equations, or (2)
incorrect data, are being used. In most cases, proper care has been taken in selecting the
equations, and it must be assumed that these equations are correct.
Occasionally equations used for predicting behaviour of water drive reservoirs may
need some change. The only remaining area of potential misinformation lies in the numbers
used in the equations. The data which are used in the equations may be subdivided into four
parts: (1) reservoir pressure data, (2) reservoir fluid data obtained from either subsurface or
recombined samples (3) oil, water, and gas production data, and (4) gross reservoir data,
such as size of original oil zone, gas cap size, and water influx.
If the equations used to predict the past behaviour are correct, then if an incorrect
answer is obtained it is obvious that incorrect data have been used in the equations.
Therefore, the next step in the reservoir study is a re-evaluation of all the reservoir
data in order to correct the discrepancies. This may be a tedious, time- consuming job, but
nevertheless it is necessary if a sound reservoir study is to be made.
A close appraisal of all the reservoir data may indicate possible errors in some of the
data, in which case a tentative correction in the data is made and past reservoir performance
is again "predicted." If this correction fails to yield a perfect match between predicted past
behaviour, then further study of the data must be made.
In cases where a study of all reservoir data fails to yield a potential source of error, it
may be simpler to examine the calculated results to determine what factors in the equations
would have to be changed in order to yield the desired match.
For example, it might be determined that a small change in original reservoir pressure
would result in a match between predicted and actual behaviour. Then the original reservoir
pressure measurement could be critically examined to determine whether or not there would
be any justification for the change.
Likewise, it might be determined that a small change in the reported gas production
would yield the desired match, in which case the gas production data could be examined very
closely to locate possible errors in gas measurement.

6. Relook at the source data for any possible input error

Before finally changing any basic data to provide a match between predicted and actual past
behaviour there should be some justification for the change. Arbitrarily changing original
reservoir pressure just because this would achieve the desired match is not good practice.
This may temporarily solve the problem, but if the real problem is the omission of some of the
gas production, then in future prediction work this may result in a compounding of errors.

7. Importance of matching past performance with future prediction

A reservoir study is of value only if the future performance of the reservoir can be accurately
predicted. Matching predicted past performance with actual past performance is a major step
in the attainment of this goal.

8. Optimization of operating parameters for maximizing net income

The heart of a reservoir study lies in the information derived in predicting future behaviour of
the reservoir. The object of the reservoir study is to determine the operating methods which
will yield the greatest net income for the property.
Therefore, it will usually be necessary to study the effects of: (1) different reservoir
withdrawal rates, (2) varying amounts of gas injection, (3) varying amounts of water injection,
and (4) other fluid injection or secondary recovery methods.
All or part of these various parameters may be studied for anyone reservoir. It
becomes immediately obvious that the number of calculations becomes enormous if several
combinations of varying factors are studied.
For example, consider the number of complete calculations required if the effects of
three different rates of oil production, three different rates of gas production, and three
different rates of water production are studied.
There are 27 different combinations of withdrawal rates. If to this, three different
amounts of gas injection are added, this results in 81 sets of complete calculations, which
would require several man-months of time if a standard desk calculator were used.
Fortunately, computer facilities are available which reduce the calculation time to a
matter of minutes, rather than months. Most routine methods of reservoir studies have been
programmed, and therefore little preliminary work is required for solution of the problem on
computers.
The application of the computer to reservoir engineering studies has eliminated the
laboratories and unrewarding aspects of reservoir studies and permits the engineer to more
effectively utilize his time in the development of good data, and in the interpretation of the
results. It should be emphasized that, even with the aid of computer facilities, a thorough
reservoir study still requires months for completion.
9. Method of presentation of predicted performance

The results of the predicted future behaviour under various operating plans is usually
shown in graphical form. The dashed lines to the right of the vertical line in Figure 1 show
typical results for three different gas injection rates with one set of fluid withdrawal rates.
Because of the uncertain future actions of the various regulatory agencies in
controlling reservoir fluid withdrawal rates, it is usually desirable to determine the effect of
at least three different withdrawal rates for oil.
It may be necessary due to other uncertainties to also consider the effects of three
different gas producing rates for each oil producing rate, and three water producing rates
for each of these production rates.

10. Selection of the best method of operation

After the effects of various operating methods on ultimate oil recovery have been predicted,
it is necessary to select the best overall method of operation for recommendation to
management.
The method to be recommended will normally be the one which will yield the
greatest ultimate net income, as there often will be factors present which will render
uneconomic the recovery of a maximum amount of oil.
For example, the return of 95% of the produced gas to the reservoir might result in
the recovery of more oil than if only 50% of the produced gas were returned, but if the cost
of additional gas and compressor facilities was more than the value of the additional oil
recovered, there would be no incentive to construct the additional facilities.