IPTC 11489

Capturing Complex Dynamic Behaviour in a Material Balance Model

Jalal Mazloom and Mike Tosdevin, SPE, Sasol Petroleum International, and Dominique Frizzell, Bill Foley, and
Mike Sibley, SPE, Chevron

Copyright 2007, International Petroleum Technology Conference

This paper was prepared for presentation at the International Petroleum Technology
Conference held in Dubai, U.A.E., 46 December 2007.
This paper was selected for presentation by an IPTC Programme 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 International Petroleum Technology Conference
and are subject to correction by the author(s). The material, as presented, does not
necessarily reflect any position of the International Petroleum Technology Conference, its
officers, or members. Papers presented at IPTC are subject to publication review by Sponsor
Society Committees of IPTC. Electronic reproduction, distribution, or storage of any part of this
paper for commercial purposes without the written consent of the International Petroleum
Technology Conference 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, IPTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
Sometimes, a simple and quick material balance method is
preferred to using a numerical simulation model. This
preference can be justified when preparing the development
plan and production optimization for a collection of
hydrocarbon reservoirs (lean and rich gas condensate, oil rim
and gas cap), some connected to an aquifer, and the reservoirs
cannot be modelled separately. This situation can occur when
multiple gas reservoirs are needed to be developed in order to
provide enough gas for a particular project. A significant
drawback of this modelling approach is the simplification
introduced when a single tank model (Material balance
method) is being used instead of a fine grid simulation model.
The material balance method assumes every well contacts all
hydrocarbons and that geological heterogeneity is not a factor
in recovery. It is necessary to know how reliable are final gas
and condensate recovery factors and gas, condensate and
water production profiles predicted by a material balance
model. In this study, we address all these uncertainties. A
sensitivity analysis has been carried out on different aquifer
strengths, gas condensate richness, and reservoir heterogeneity
which are related to the real and field data set. Introduction of
a generic method for selecting the important input data to the
material balance model (relative permeabilities and well
productivities) in order to have reliable results is the target of
this study. The material balance results are compared to a fine
grid simulation. It is observed that using the introduced
method, the effect of reservoir heterogeneity and aquifer
influx on final gas recovery factor can be captured in a
material balance model.
Introduction
Predictions of oil and gas reservoirs behavior and hydrocarbon
production profiles from them are crucial steps for planning
fields development. Although it is believed numerical
simulation (3 dimensional models) gives more reliable results
than a material balance (zero dimensions) evaluation, a

material balance method can be utilized in an acceptable range

of uncertainties. Material balance has been used as a reliable
tool for calculating hydrocarbon volume initial in place and
reservoir drive mechanism and prediction production profile1,
2
.
Sometimes material balance can be used for narrowing down
uncertainties
around
in
place
volume
and
compartmentalization and presence of faults before
simulation3.
Recent years have witnessed efforts for
improving the material balance method4, 5.
Also, some
studies have shown that material balance can be utilized for
performance prediction of gas condensate reservoirs6.
However, it is still important to understand whether field
performance, as predicted by a tank model, is reliable enough
for making a financial investment decision. In this paper,
reservoir performance and production profiles predicted by
material balance and 3D simulation model are compared with
each other. It is explained how the tank model can capture
the effect of aquifer and condensate drop-out on reservoir
performance if the model is tuned properly.
Model construction and sensitivity analysis
1. Simulation model construction
A heterogeneous 3D geological model that had been
constructed in GOCAD was selected as the reservoir model.
General characteristics of the model can be found in table 1.
Table 1: Grid model characteristic
As the main cause of difference between outcomes from the
material balance and 3D simulation methods is the high level
of heterogeneity in geological properties (e.g. Porosity, NTG,
Permeability), the authors believe the selected geological
model is an appropriate example for purpose of this research.
Figure 1 shows the porosity distribution in the reservoir.

IPTC 11489

Gas, Condensate & Water Relative Permeability

Relative Permeability

1
0.8
Krw

0.6

Kro
0.4

Krg

0.2
0
0

0.2

0.4
0.6
Phase Saturation

0.8

Figure 2: Phases relative permeability curves

Figure 1: Porosity distribution in the selected reservoir

The reservoir permeability maps were calculated using the
correlation between porosity and permeability. This equation
has been generated based on the available log and core data.
It is our belief that the question about the capability of the
material balance method to capture the effect of water
encroachment and condensate drop-out around the well-bore
on reservoir performance and the final recovery has not been
answered yet. In order to investigate this issue, sensitivity
analyses have been carried out on the aquifer strength and gas
condensate richness. The aquifer permeability has been
selected as the only representative of strength, other
parameters such as areal extent of the aquifer were not
considered in this study. Aquifer permeabilities of 1, 50, 500
md indicate very weak, moderate and strong aquifer
respectively. In order to simulate fluid phase behavior in the
reservoir, black oil tables were generated for lean and rich gas
samples. Table 2 shows the condensate gas ratio and the dew
point for two fluid samples that were used as the lean and rich
gases in this study.
Table 2: Fluid properties of Rich and Lean samples
Lean Sample
Rich Sample

CGR(Stb/MMscf) Dew Point(Psi)

15
3300
99
3818

The gas water contact of 7757ft and the initial pressure of

4200 psi were considered in model. The calculated initial gas
and the vaporized condensate-in-place using two different
fluid samples can be observed in table 3.
Table 3: Gas and vaporized condensate in places
Lean samle
Rich Sample

Gas in place(Bscf)
419.2
419.5

Vaporized condensate in place(MMstb)

6.2
41.3

The relative permeability curves in figure 2 were used in

model.

The reservoir produces gas through 9 wells. To prevent water

coning, wells have been perforated in the top 21 layers. The
well-head pressure of 700 psi is the only constraint in the
model. A vertical well trajectory was used to generate the
vertical lift curves and it has been assumed that the tubing
extends from top of the reservoir to the surface. The software
PROSPERTM (Petroleum Experts product) was utilized for
generating VLP curves. According to the different fluid
models and aquifer strength, 6 simulation models were
generated at the end (Table 4).
Table 4: Simulation model cases
Simulation model no.
Description
Case1
Lean sample, strong aquifer
Case2
Lean sample, moderate aquifer
Case3
Lean sample, very weak aquifer
Case4
Rich sample, strong aquifer
Case5
Rich sample, moderate aquifer
Case6
Rich sample, very weak aquifer
2.

Material balance model construction

MBALTM (Petroleum Experts product) was utilized to

generate and run zero dimensional models. The gas-in-place,
initial volume average pressure, fluid properties in tank and
simulation models are identical. Same as for the 3D models,
Carter-Tracy aquifers are connected to the tanks and aquifer
permeability is considered as strength criteria. Number of
wells and VLP tables that have been used in 3D models were
duplicated in the material balance model. Two sets of well
productivities and Darcy factors have been calculated for
wells in the material balance model. In the first method,
similar productivity was calculated for all wells using average
permeability and thickness of whole reservoir. In the second
method, well productivity for each well was estimated based
on the average reservoir permeability and thickness around
each well.
Therefore 12 tank models were generated in
MBAL (Table-5). The results of the material balance models
are compared with results of the corresponding simulation
models in the result section.

IPTC 11489

Table 5:Tank model cases

Predicted Condensate Recovery Factors Using Single Tank Method
60

Final Recovery factor(%)

Condensate Recovery Factor(%)

37.8

40

36.8

37.4

36.8

37.0

36.4
31.3

31.5

31.4

31.4

30.8

30.9

20

1- Date cutoff: production profiles up to 6.5 years were

used to calculate recovery factors
2- Rate cutoff: 20 MMSCF/d was considered as the
minimum acceptable gas production rate from field.
Because the reservoir heterogeneities have been captured at a
fine scale and because the reservoir fluid properties and
relative permeabilities in the model initialization are coming
from field data, results from the simulation model are
considered to be representative of real field performance. Here
we compare results from the material balance model with
results from simulation to find out under what circumstances a
single tank model can predict the field performance. Answers
are given for predicting gas and condensate recovery and the
impact of aquifer influx in separate sections.
1- Is a single tank model able to capture the effect
of reservoir heterogeneity and condensate dropout on field performance?

C
as
e6
b

C
as
e6
a

C
as
e5
a

C
as
e5
b

C
as
e4
b

C
as
e4
a

C
as
e3
b

C
as
e3
a

C
as
e2
b

C
as
e2
a

C
as
e1
b

Figure 4: Condensate recovery factors predicted by tank

model

Figures 5 and 6 indicate the final gas and condensate recovery

factors estimated using a 3D simulation model.

Predicted Gas Recovery Factors by Simulation Model

100

Gas Recovery Factor(%)

80

Final Recovery factor(%)

Results and recommendations

1. Results and discussion
In order to calculate the final gas and condensate recovery
factors, two cut-offs were implemented on the gas and
condensate production profiles.

C
as
e1
a

68.1

60

56.1

54.7

50.5
37.8

40

42.2

20

0
Case1

Case2

Case3

Case4

Case5

Case6

Figure 5: Final recovery factors predicted by simulation

model
Predicted Condensate Recovery Factors by Simulation Model

Predicted Gas Recovery Factors Using Single Tank Method

Final Recovery factor(%)

100
80

Gas Recovery Factor(%)

74.8

73.8

74.8

73.8

74.6

73.8

70.5

70

70.5

70

70.0

70

60
40
20

C
as
e1
a
C
as
e1
b
C
as
e2
a
C
as
e2
b
C
as
e3
a
C
as
e3
b
C
as
e4
a
C
as
e4
b
C
as
e5
a
C
as
e5
b
C
as
e6
a
C
as
e6
b

Figure 3: Gas recovery factors predicted by tank model

60

Condensate Recovery Factor(%)

50
44.4
Final Recovery factor(%)

Figures 3&4 demonstrate the predicted final gas and

condensate recoveries using the material balance method in
the different cases that have been defined in table 5. As can
be observed, two different methods of well productivity
estimation (well productivities calculated based on the global
average K&H and local average K&H around each well)
generate very similar results. For example, recovery factors of
cases 1a&b or cases 2a&b are close to each other. Therefore,
for comparing the results of tank and simulation model
outcomes, only one of a or b cases will be selected.

39.7
40

36.9

36.1

35.6

33.3

30

20

10

0
Case1

Case2

Case3

Case4

Case5

Case6

Figure 6: Condensate recovery factors predicted by tank

model
By comparing the results in figure 5 and figure 3, it can be
concluded that the material balance model, significantly
overestimates the gas recovery factor. This is due to the lack
of capability for capturing the reservoir heterogeneity in the
single tank model. In the single tank model, it is assumed that
the entire area in the reservoir is accessible by all the wells
and this is the main reason for predicting a high gas recovery
factor by this method.
Also, as can be observed in figure 5, the final gas recovery
factors for the lean gas fluid are bigger than same parameter in
the rich fluid sample by a maximum of 120%. The well

IPTC 11489

productivity impairment due to the liquid drop-out in area very

close to the well-bore is the main reason for this fact. But in
the single tank model (Figure 3), the maximum difference
between the final gas recovery factors of lean and rich fluid
sample is only 4% and therefore material balance method
doesnt capture this effect.
Contrary to the predicted gas recovery factors, the cumulative
condensate production has been underestimated slightly in the
material balance model. This is again due to using only one
grid block (Zero dimensional method) in the tank model for
the whole volume of reservoir. In reality, as it has been
captured in 3D simulation, the condensate drop-out happens
only in a small volume around well-bore as long as the
average reservoir pressure is above dew point. But in the tank
model, liquid-dropout and condensate loss happens in the
entire reservoir as soon as the average reservoir pressure drops
below the dew point. Therefore, the answer to the question at
the beginning of this section question is:
-

Material Balance overestimates the gas recovery

factor and predicts slightly less condensate
production than 3D simulation model.

2- Does a single tank model have the capability of

predicting the effect of aquifer influx on the gas
condensate reservoir performance?
Water encroachment reduces the gas recovery factor by
leaving gas behind a water front propagated at a higher
pressure than abandonment pressure under volumetric
behaviour7,8,9. The residual gas saturation controls the volume
of gas trapped in that section of the reservoir that water
encroachment has happened. The volume and location of the
residual gas are controlled by the distribution of the
petrophysical properties. More heterogeneity causes more
residual gas in reservoir because of bypassing of lower
permeability rock. The results of simulation models in figure
5 are consistent with this observation. As observed in figure 5,
the gas recovery factors for the rich and lean gas condensate
fluid are reduced by an increase in aquifer strength. But this
fact is not seen in the material balance results. Figure5 shows
that all the constructed tank models do not reflect the effect of
the aquifer strength and the water encroachment on the
reservoir performance.
Not including reservoir
heterogeneity/geometry in the tank model is the main source
of this deficiency. It was mentioned that the volume of the
residual gas is controlled by the petrophysical distribution and
because in the single tank model there is no distribution of the
reservoir properties, this control is missing.
Another interesting observation that we came across during
this research was the effect of pressure support by the aquifer
on condensate recovery. As was discussed, the aquifer has a
negative effect on gas recovery. But as it can be seen in figure
6, the condensate recovery increased slightly due to the
provided pressure support by water encroachment. This
situation causes less liquid drop-out and condensate loss in the
reservoir and more condensate production at the surface. So,
two main conclusions for this section are:

2.

Material balance method does not reflect the

effect of an aquifer on the gas recovery factors.
In the simulation model, due to the pressure
support from the aquifer and less condensate
drop-out inside reservoir, the condensate
production increases with a stronger aquifer.
Multiple Tanks Material Balance

As was mentioned earlier, the lack of capability in the single

tank model for capturing the reservoir heterogeneity/geometry
is the main reason for the difference between the material
balance and simulation models results. In the material balance
model, it is assumed that the entire gas in-place in the model is
accessible by all wells according to their well productivities.
Also for the same reason (no heterogeneity in single tank
model), it is not possible to predict an increase in the residual
gas saturation and the reduction in the gas recovery factor due
to the water influx in the single tank model. In order to have a
better prediction by material balance method a multiple tanks
model is suggested. The procedures of constructing of this
model are as below:
1- The number of tanks in the multiple tanks model
should be equal to the number of wells in 3D
simulation model. The well productivity for each
well has been calculated according to the average
reservoir thickness and permeability around well. As
was mentioned earlier, in the single tank model it is
assumed that the entire gas in reservoir is accessible
by all wells. But in reality, only gas in the drainage
area of each well can be produced according to the
well productivity. In order to capture this fact in the
multiple tanks model, the gas-in-place for each tank
should be calculated based on distribution of gas
volume vs. well productivity or K*h. So, the gas-inplace in each tank is defined using:
a.

b.

c.

In the geological model, all the grid blocks

should be divided to the different districts
according to their permeability and
formation
thickness.
The
average
permeability and thickness at each district is
equal to the average K&H that were used for
well productivity calculation in the tank
model.
In the geological model, the amount of gas
in place that is located at each range of
K&H should be defined. In this step we
identify the distribution of gas-in-place vs.
K*H.
These numbers should be used as gas-inplace numbers for each tank in the multiple
tanks model.

2- The residual gas saturation at each tank should be

defined according to the average reservoir porosity
and the equation below (introduced equation at
reference 9).

IPTC 11489

(1)
Sgrm = 0.969 * + 0.5473
This equation has been developed to be used in 3D simulation
models, but since by using the multiple tanks we are trying to
approach a heterogeneous simulation model from a single tank
model, this assumption is valid to some extent. The
implementation of this recommended method is given in an
example below.
Table 6 shows the average reservoir permeability and
thickness that have been used for calculation of well
productivities in the tank model.

is connected to a different tank and in-place in each tank has

been calculated by multiplying percentage number in table 7
and total gas-in-place. The residual gas saturation in each tank
has been estimated using equation 1.

Table6: The average reservoir permeability and thickness

around each well
Well No.
1&2
3
4
5
6
7
8
9

K*H(md.m)
34
30
28
25
21
17
13
8

The percentage of the total gas in place which is located at the

different range of permeability and thickness can be observed
in table 7.

Figure 8: Multiple tank model with different gas in place

in each tank
Figure 9 indicates the final gas recovery factors for cases that
were defined in table 6. These numbers have been estimated
using proposed method in this paper.

Table7: Percentage of the total gas in place which are

located at different ranges of kh
Percentage of gas in place
50.0%
12.8%
4.3%
5.3%
7.4%
4.3%
3.2%
4.5%
8.3%

Figure 7 show the configuration for the single tank model and
figure 8 demonstrates the same model after implementing the
suggested procedures.

Figure 7: Single tank model

In figure 7, all wells are connected to only one tank with gas
in place of 419 Bscf. The residual gas saturation in the single
tank model is coming from the relative permeability curves
data that were showed in section 2. But in figure 8, each well

Final Recovery factor(%)

Range of KH(md.m)
>34
34>Kh>30
30>Kh>28
28>Kh>25
25>Kh>21
21>Kh>17
17>Kh>13
13>Kh>8
8>Kh

Predicted Gas Recovery Factors Using Proposed Method

100

Gas Recovery Factor(%)

80
60.0

62.0

65.0

60

55.0

57.0

Case4

Case5

60.0

40

20

0
Case1

Case2

Case3

Case6

Figure 9: The final recovery factors calculated using

proposed method
As can be observed, the final recovery factors in figure 9 are
relatively close to the numbers in figure 3 that have been
calculated using 3D simulation model. Particularly, in the very
weak aquifer cases (Case 3&6), there is a good consistency
between outcomes of the 3D simulation and the proposed
method. Also note that in Case3 the material balance
approach actually under predicts recovery factor by 3%.
Therefore, it can be concluded that applying the proposed
method, the reservoir heterogeneity effect has been captured
better than single tank model approach. Also by increasing
aquifer strength, the final recovery factor is reduced; this trend
was not captured in the single tank method but an
improvement achieved using the multiple tanks approach.
Additional sources of error not accounted for by this multi
tank model approach include well interference due to the
actual level of communications in the reservoir and any
inaccessible pore volume thats not drained by the planned
wells.

IPTC 11489

Conclusions
The effect of reservoir heterogeneity, water encroachment and
condensate drop-out around well-bore on the gas condensate
reservoir performance was studied by using a fine scale
heterogeneous reservoir model. Using the geological model,
the recovery factors of gas and condensate for lean and rich
gas condensate reservoir connected to a strong or very weak
aquifer were calculated. Single tank models identical to the
simulation models were constructed and results of the material
balance model were compared with 3D simulation model
outcomes. Main results of this investigation are:
o

The
single
tank
method
overestimates
considerably the gas recovery factor. This is due
to the inability for capturing the reservoir
heterogeneity in the single tank model.

The cumulative condensate production is

underestimated slightly in the material balance
model. This is due to using only one grid block
(Zero dimensional method) in the tank model for
the whole volume of reservoir. In reality, as it has
been captured in 3D simulation, the condensate
drop-out happens only in a small volume around
well-bore as long as the average reservoir pressure
is above dew point. But in the tank model, liquiddropout and condensate loss happens in the entire
reservoir as soon as average reservoir pressure
drops below the dew point.

Material balance method does not reflect the

effect of the aquifer on the gas recovery factors.

In the simulation model, due to the pressure

support from the aquifer and less condensate
drop-out inside the reservoir, condensate
production is increased by a stronger aquifer.

Using the proposed multiple tanks method, the

effect of heterogeneity/geometry is captured in a
reasonable range

In the multiple tanks method, by increasing

aquifer strength final recovery factor decreases.
This trend was not captured in the single tank
method. But it seems this method still doesn't
reflect the aquifer effect on gas recovery factor
perfectly.

3D simulation models are useful to help verify the

accuracy of material balance models and can be
used to help fine tune then in order to account for
the effects of reservoir heterogeneity.

Nomenclatures
X : Average grid block size in X direction
Y : Average grid block size in Y direction
Z : Average grid block size in Z direction
nX: Number of grid blocks in X direction

nY: Number of grid blocks in Y direction

nZ: Number of layers in the grid model
CGR: Condensate gas ratio
Ave : Average porosity
Sgrm : Residual gas saturation
AveK : Average permeability
h : Formation thickness
VFP: Vertical flow performance
Acknowledgments
The authors wish to thank the management of Sasol Chevron,
Chevron, and Sasol Petroleum International for permission to
write this paper. Some of the ideas explained in this research
came from helpful discussion with Duke Snyder. The authors
wish to thank their colleagues particularly Anand Rao, at
Sasol Petroleum International; Rachel Preece, Shah Kabir and
Andrew Harding at Chevron for their technical support to
conduct this research.
References
1. Farough, Ali, S.M., Nielsen, R.F., 1970. The Material
Balance Approach vs reservoir Simulation as an aid to
Understanding Reservoir Mechanics. Paper SPE 3080,
presented at the 45th annual fall meeting of SPE,
Houston, Tex, October 47.
2. Merchant, A.R., Arnold, M.D., 1973. A Technique for
Improving Material balance Accuracy in Reservoir
simulation Model. Paper SPE 4548, presented at the 48th
annual fall meeting of SPE, Las Vegas, Nev., October.
3. Eugene, E., Dresda, S., Monico, C., 2004. Use of Material
Balance to enhance 3D Reservoir Simulation: A Case
Study. Paper SPE 90362, presented at the SPE Annual
Technical Conference, Houston, Tex., 26-29 September.
4. Pletcher, J.L., 2000. Improvements to Reservoir MaterialBalance Methods. Paper SPE 75354, presented at the
SPE Annual Technical Conference, Dallas, Tex., 1-4
October.
5. Yildiz, T., Khosravi, A., 2006. An Analytical Bottom Water
Drive Aquifer Model for Material Balance analysis. Paper
SPE 103283, presented at the SPE Annual Technical
Conference, Sanantonio, Tex., 24-27 September.
6. Zeidouni, M., Movazi, M., Pourghasam, B.,2006.
Performance Prediction of a Rich Gas Condensate
Reservoir through Material balance and PVT Behavior
Case Study. Paper SPE 99830, presented at the SPE
Gas Technology Symposium, Calgary, Alberta, 15-17
May.
7. Hower, T.L., Jones, R.E., 1991. Prediction Recovery of Gas
Reservoirs under Water Drive Conditions. Paper SPE
22937, presented at the SPE 66th annual technical
conference, Dallas, Tex., October.
8. Cronquiest, C.,1973. Effect of Permeability Variation and
Production Rate on Recovery from Partial Water Drive
Gas Reservoirs. Paper SPE 4635, presented at the SPE
48th annual fall meeting, Las Vegas, Nev., October.
9. Holtz, M.H., 2002. Residual Gas Saturation to Aquifer Influx:
A Calculation Method for 3-D Reservoir Model
construction. Paper SPE 75502, presented at the SPE
Gas Technology Symposium, Calgary, Alberta, 15-17
May.