SPE 120897

Predicting Water Influx From Common Shared Aquifers

Munenori Shimada, Norwest Corporation, and Turhan Yildiz, SPE

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, 8–11 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
In this paper, we present 3-dimensional analytical models to forecast water influx from a common shared aquifer to
several producing reservoirs. We consider finite aquifers and an arbitrary number and distribution of multiple reservoirs. The
new analytical models have been verified by comparing them against the existing analytical models for the special case of
symmetrical distribution of multiple identical reservoirs. The computation of the equations representing the analytical models
is straightforward.
The new analytical models proposed in this study may be used for aquifer representation in reservoir simulators and
material balance based reserve estimate studies. Using the analytical models, we show a sample calculation for irregularly-
shaped reservoir on an aquifer.

Introduction
In a bottomwaterdrive hydrocarbon reservoir, the aquifer is located beneath the reservoir. When hydrocarbons are produced
from the reservoir, the water in the aquifer encroaches into the reservoir due to the pressure difference created between the
reservoir and the aquifer. Since the reservoir performance is influenced by the aquifer, it is important to estimate the water
influx accurately.
In some hydrocarbon producing basins, multiple reservoirs are supported by a common shared aquifer. In this case, the
reservoirs are in hydrodynamic communication through the aquifer. Therefore, the individual reservoir performance is
influenced by not only its own production but also the production from the neighboring reservoirs. This field condition is
called common aquifer.
There are many examples of common aquifers all over the world. Well known examples of the multiple reservoirs
communicating through a common aquifer are historical Woodbine basin in East Texas and Leduc D3 pool in Alberta. The
other field examples of common aquifers include Central Basin Platform in West Texas, El-Sahara Fields in Libya, z-Field in
Kuwait, Bowen Basin in Queensland, Abkatun-Pol-Chuc system in offshore Mexico, and Algyo Field in Hungary.
The strength of the common aquifer support varies from weak to extensive. In this type of basins, the individual reservoir
performance is influenced not only by the aquifer support but also the interference between the multiple reservoirs. Only very
few studies have focused on the prediction of water influx from a common aquifer into multiple reservoirs. The available
models to predict water influx from a common aquifer are limited to 1-dimensional and 2-dimensional flow.
In the literature, there are several models to forecast the cumulative water influx from a bottomwaterdrive aquifer into a
single reservoir. However, for the case of multiple reservoirs, there is no appropriate model that can handle the vertical flow
component in the bottomwaterdrive aquifer. The basic motivation for this study is to construct a common bottomwaterdrive
aquifer model to simulate the water influx into multiple reservoirs.
Once the new aquifer model is completed, it may be useful for reserve estimation, reservoir simulation, and other
reservoir engineering applications. The objectives of this study are; to develop an analytical model predicting water influx
into multiple reservoirs from a common shared aquifer, to verify the analytical model for single and multiple reservoir cases,
and to demonstrate the application of the new model.

Literature Review
In this section, we review three groups of relevant literature. The first group is the study of hydrocarbon producing fields with
common aquifers. In the second review, we briefly examine available transient flow models for bottomwaterdrive aquifers.
The third review is on available models to evaluate the pressure interference effect in common aquifers.
2 SPE 120897

Hydrocarbon Producing Fields with Common Aquifers. In the literature, we found numerous field studies of common
aquifers. The examples of the common aquifer support are found in Woodbine Basin in East Texas (Bell and Shepherd
1951), Central Basin Platform in West Texas (Moore and Truby 1952), Leduc D3 pool in Alberta (Horsfield 1958), Bowen
Basin in Queensland (Tucker 1989), z-field in Kuwait (Sheshodia et al. 1989), Algyo field in Hungary (Werovsky et al.
1990), Abkatun-Pol-Chuc system in Mexico (Rodriguez et al. 1996), and El-Sharara Fields in Libya (Mohammed et al.
2005). In these studies, an unusual pressure behavior such as a pressure decline before production was typically reported.
For example, Hawkins field was discovered in 1940 in Woodbine Basin, East Texas. The Hawkins field consists of two
sections separated by a fault; the eastern and western Hawkins. During the pressure surveys, the original pressure in the
eastern part was observed as 1,710 psi while a pressure of 1,900 psi was obtained in the western part. Based on examination
of cores and production tests, the fluid contact levels were known to be different in the two parts. Comparisons with the other
Woodbine fields showed that the pressure in the western Hawkins is in a normal original pressure regime. Eventually, it was
concluded that the pressure in the eastern Hawkins had been reduced from the original pressure by the production in the
neighboring fields.

Bottomwaterdrive Aquifer Models. In the verification of the new model for common aquifers, we use the bottomwaterdrive
aquifer models existing in the literature. These bottomwaterdrive aquifer models are developed for a single cylindrical
reservoir concentrically located at the top of a cylindrical aquifer. There are three existing bottomwater aquifer models; the
Coats, the Allard-Chen, and the Yildiz-Khosravi models.

Coats Model. In 1962, Coats modeled the bottomwaterdrive aquifer-reservoir system and presented a mathematical
solution for the aquifer interface pressure. Coats considered an infinite aquifer. The aquifer is sealed at the top and bottom
boundaries except at the aquifer-reservoir interface. The aquifer is considered to have constant but unequal permeabilities in
the horizontal and vertical directions.
To forecast the cumulative water influx from a bottomwater aquifer into a reservoir, Coats solved the two-dimensional
diffusivity equation. The dimensionless groups for time, aquifer radius, and vertical distance were introduced. The diffusivity
equation was solved for the case of constant water flow rate. Coats used Hankel transformation and separation of variables to
obtain the solution. His solution includes the integrals of Bessel functions. Using the solution, Coats calculated the aquifer-
reservoir interface pressure as a function of time. The results are reported for different values of dimensionless aquifer
thickness and dimensionless time.

Allard-Chen Model. In 1988, Allard and Chen built a numerical model for finite bottomwaterdrive aquifers. The constant
terminal pressure drop at the aquifer-reservoir interface was imposed. Allard and Chen considered the same aquifer-reservoir
model as Coats, and started with the same diffusivity equation. However, instead of attempting to formulate an analytic
solution, Allard and Chen used a numerical simulator. The results are presented in the form of tables for dimensionless
cumulative water influx. The dimensionless cumulative water influx is tabulated as a function of dimensionless time,
dimensionless aquifer radius, and dimensionless aquifer thickness. Allard and Chen presented a sample calculation to
illustrate the use of their solution. In the sample calculation, they used the superposition principle to consider the variable
pressure at the aquifer-reservoir interface.

Yildiz-Khosravi Model. In 2007, Yildiz and Khosravi presented an analytical model to predict water influx from
bottomwaterdrive aquifers. Initially, a solution for constant interface pressure condition was constructed. They used Laplace
transformation and separation of variables to develop a solution in Laplace space solution. The Stehfest algorithm was
utilized to invert the constant terminal pressure solution back to real time domain. Yildiz and Khosravi used the superposition
principle to handle the time-dependent interface pressure cases.

Common Aquifer Models. In the literature, we found only a few models focusing on the prediction of water influx from a
common aquifer into multiple reservoirs. The examples of the common aquifer model are presented by Mortada (1955),
Hurst (1960), Sageev and Horne (1985), and Rodriguez et al. (1996). Among these four models, the first three models deal
with two radially-shaped reservoirs surrounded by an infinite aquifer. By setting the physical models in this manner, they
introduce limitations in number and shape of reservoir, and type of aquifer. On the other hand, Rodriguez et al. model has
more flexibility. Its physical model consists of any number of arbitrary-shaped reservoirs. The reservoirs are discritized into
sub-domains and total flux of the reservoir is split into the summation of fluxes for each active interface cell. The model
presented by Rodriguez et al. is 2-dimensional and it can handle finite aquifers.
In all the modeling examples mentioned above, the reservoirs are surrounded by the aquifer on their lateral surface. This
means that water influx only takes place in the horizontal direction. This assumption may be adequate where we have
edgewaterdrive mechanism. When there is bottomwaterdrive aquifer, these models can not be used because the vertical
component of the water flow is not modeled.
SPE 120897 3

Analytical Models for Common Aquifers

In the first part of this section, we described the details of mathematical model developed for simulating the water flow from
a common shared aquifer to multiple petroleum reservoirs. In the second part, we verify the new analytical model.

Model Development. This section describes the development of an analytical transient flow model to compute water influx
from a bottomwaterdrive aquifer to multiple reservoirs supported by the aquifer. The analytical models derived for several
different flow conditions at the reservoir-aquifer interface; constant water influx rate, constant pressure drop, and time-
dependent pressure drop. The analytical model considers finite aquifers. However, an infinite aquifer may be mimicked by
assigning numerically large values for the aquifer length and width. The physical model is shown in Fig. 1. Both aquifer and
reservoirs have rectangular shape. The reservoirs are located at the top of the common aquifer. Hence, a bottomwaterdrive
condition exists. Water flow from common aquifer to all the reservoirs is three-dimensional.
The aquifer has a length of xa, width of ya, and thickness of ha. IR number of reservoirs is considered. The jth reservoir has
length of xej and width of yej. The position of the jth reservoir is specified with the coordinates of the left-lower corner as xcj
and ycj. Water flow is allowed only through the aquifer-reservoir interfaces and all the other external boundaries of the aquifer
are sealed.
In the model development, we consider the assumptions listed below:

• Aquifer porosity and thickness are constant.

• Aquifer has a uniform but anisotropic permeability.
• Permeabilities in x-, y- and z-directions are kx, ky, and kz, respectively.
• Aquifer rock and water have constant compressibility.
• Aquifer rock is completely saturated with water.

The development of the analytical models is shown in Appendix A. The solutions are derived for constant flow rate,
constant pressure drop, and time-dependent pressure drop at the aquifer reservoir interfaces. The solutions are obtained by
means of Laplace and finite Fourier cosine transforms. The final solutions are in Laplace space. Numerical algorithms
(Stehfest, 1970 and Iseger, 2006) are used to invert the Laplace space solutions back to real time domain.

Verification of Analytical Models. In this section, the new model is verified step by step. Initially, a single reservoir is
considered and the new model is compared with the models in the literature. In the second part, the multiple reservoir
solution is verified by considering the symmetric distribution of multiple identical reservoirs and comparing them to the
single reservoir case.

Verification for Single Reservoir Case. The new solutions are developed for three different flow conditions at the
aquifer-reservoir interface; constant water influx rate, constant pressure drop, and variable pressure drop. The new model for
the single reservoir case is verified for all the three flow assumptions.
Among the models available in the literature, we choose the 2D finite radial models developed by Coats, Allard and
Chen, and Yildiz and Khosravi. These three models are developed for a radially-shaped aquifer-reservoir system whereas
rectangular shape and Cartesian coordinate system are considered in the new models. To compare the results from the new
models and the radial models in the literature, we construct equivalent aquifers and reservoirs for both the new Cartesian
models and the radial models. To ensure equivalency, we consider isotropic porous media. Additionally, we make sure that
aquifer volume, aquifer-reservoir interface area, and aquifer thickness are the same in both the Cartesian and cylindrical
coordinate systems. Furthermore, to replicate the concentric cylindrical aquifer and reservoir geometry of radial models, we
consider square reservoir and aquifer pair and enforce symmetry with respect to aquifer center point and external boundaries.
For the case of a single bottomwaterdrive reservoir, we conducted an extensive comparison study between the new
models and the models presented by Coats, Allard and Chen, and Yildiz and Khosravi. The comparisons were carried out for
finite and infinite aquifers. Although the comparison study covered three different interface boundary conditions, constant
flow at the interface, constant pressure drop at the interface, and time-dependent interface pressure drop, we only present one
single comparison case for the sake of conciseness.
The verification of the variable pressure drop at interface is performed by comparison with the Allard-Chen model. Allard
and Chen presented a sample calculation to illustrate the use of their model. In the sample calculation, they consider the
hypothetical reservoir-aquifer system with the properties shown in Table 1. In the Allard-Chen study, the system was
numerically simulated. They reported average reservoir pressure, average pressure at the aquifer-reservoir interface,
cumulative water influx, and cumulative oil production on a monthly basis. For additional details, the reader is referred to
Table 6 of the publication by Allard and Chen.
Using the new model, we predicted the cumulative water influx for the aquifer-reservoir system described in Table 1. The
calculated cumulative water influx values from both models are graphically shown in Fig. 2, a Cartesian plot of We versus t.
As illustrated in Fig. 2, the values of cumulative water influx from the new model and the Allard-Chen model agree very
well.
4 SPE 120897

Verification for Multiple Reservoir Case. For the special case of multiple reservoirs supported by a common aquifer, we
could not find any comparable model in the literature. Therefore, to verify our solution for the multiple reservoir case, we
consider the symmetric distribution of the identical multiple reservoirs in a common aquifer and compare them with the
equivalent single reservoir counterpart. To illustrate the methodology, we consider four identical reservoirs on top of a
common aquifer. Four reservoirs are symmetrically distributed with respect to aquifer boundaries, aquifer center point, and
each other. The locations of the identical reservoirs are chosen such that the aquifer-reservoirs system could be divided into
four symmetrical sub units. This is graphically shown in Fig. 3. We also enforce identical flow boundary conditions at all
four interfaces between the common aquifer and four reservoirs.
In Fig. 3, Reservoirs 1 through 4 are located at the center of the one-quarter of the square aquifer. The equivalent single
reservoir-aquifer system is also depicted on the right-hand-side of Fig. 3. Since one quarter of the aquifer-reservoir system is
equivalent to the other sub-systems and identical flow conditions are induced at all four aquifer-reservoir interfaces, there are
flow-induced “no-flow” boundaries along the dashed lines in Fig. 3.
The comparisons for the verification of multiple reservoir case were conducted for three different flow conditions at the
interface; constant flow rate, constant pressure drop, and time-dependent pressure drop. Both finite and infinite aquifer sizes
were tested. Here, we present an example comparison for the case of variable pressure drop at interface. We once again
utilize the data presented by Allard and Chen. We consider the distribution of the multiple reservoirs as depicted in Fig. 3.
The Alard-Chen simulation considers an infinite aquifer. A very large aquifer size is selected to mimic the response of an
infinite aquifer. The relevant reservoir and fluid data have already been presented in Table 1. We impose the same interface
pressure profile, as reported by Alard and Chen, for both the multiple and single reservoir cases. The calculated cumulative
water influxes for the multiple and single reservoir cases and Allard and Chen model are all graphically shown in Fig. 4. The
comparison displayed in Fig. 4 verifies that our common aquifer model and the code developed for the model are accurate.

Discussion
In this section, we use the new models to examine the cumulative water influx into reservoirs in more detail. At first, we
investigate the general shape of WeD-tD curve and identify the specific flow regimes. Then, to illustrate the versatility of the
model, we investigate the influence of reservoir shape, and show one example for irregularly-shaped aquifer-reservoir
interface.
In the verification process, the calculations have been performed only for the square reservoir(s) located on the center of
common aquifer or its sub-systems. However, the use of the new model is not limited to that condition but can be applied for
any rectangular shape. In the analysis, the shape of reservoir is changed from square to rectangle to investigate the effect. To
specify the shape of the rectangles, we use the aspect ratio.
As described above, the new model can be used for any rectangle. In addition, the new model can be used for multiple
reservoirs on a common aquifer. These two characteristics enable us to calculate water influx for multiple reservoirs with
irregular shapes. We choose an actual reservoir with irregular shape from the literature and calculate water influx into the
reservoir. Several rectangles are fitted on the shape of the reservoir and simulate the reservoir as a composite of rectangles.

Flow Regimes. To identify the flow regimes, we compute WeD as a function of tD for both the infinite and finite aquifer cases
and compare them on a log-log plot. The WeD-tD relationships for both aquifer cases are displayed in Fig. 5. The basic data set
used to generate Fig. 5 is given in Table 2. Fig. 5 demonstrates that there are four distinct flow regimes in WeD-tD relationship
for a finite aquifer; linear flow period, infinite acting transient flow period, boundary-dominated flow period, and pot flow
period. Fig. 5 shows that, at the times earlier than tD = 0.01, WeD-tD relationship is characterized by a half-slope straight line
on the log-log plot. This observation indicates the existence of the linear flow period. The duration of the linear flow period
is controlled by the aquifer thickness and permeability anisotropy. For an aquifer thickness of 200 ft and isotropic
permeability conditions, the linear flow period for both the finite and infinite aquifer cases prevails until tD = 0.01. In the
linear flow period, the part of the aquifer immediately below the reservoir responds and the flow in the aquifer is mainly in
the vertical direction.
The WeD-tD relationship for both infinite and finite aquifer cases are identical until around tD = 50, including the linear
flow period. After that, the curve for finite aquifer case starts flattening out whereas the curve for infinite aquifer case
continues to rise with nearly the same trend. Basically, Fig. 5 demonstrates that, the lateral aquifer boundaries do not have
any influence on the aquifer performance until about tD = 50. For times earlier than tD = 50, the finite aquifer behaves as if it
is an infinite aquifer. Hence, the time interval from the end of the linear flow period to the start of lateral boundary influence
is called infinite acting flow period. At the end of the infinite acting flow period, the pressure disturbance created at the
aquifer-reservoir interface reaches the lateral aquifer boundary in the horizontal direction.
In the finite aquifer case, it is observed that the increment in WeD becomes smaller and smaller from tD = 50 to tD = 300.
The finite aquifer curve finally becomes completely flat after tD = 300. In the time interval 50 < tD < 300, water flow is
dominantly influenced by the impermeable aquifer boundaries. Hence, this flow period is called boundary-dominated flow
period.
At tD = 300, the water influx reaches to its maximum value. There is no more additional water influx after tD > 300. In
this flow period, water aquifer responds as if it’s a simple tank containing compressed water. The porous media nature of the
SPE 120897 5

aquifer becomes immaterial. Beyond tD > 300, cumulative water influx could be represented by formulating the
decompression of pressured water in a tank. This flow period is appropriately called pot flow period.

Effect of Reservoir Shape. In the radial models for the bottomwaterdrive aquifer, the aquifer performance is reported as a
function of the aquifer-reservoir size ratio and time. For example, in the Allard-Chen model, the dimensionless cumulative
water influx values are reported as a function of dimensionless aquifer radius, dimensionless aquifer thickness, and
dimensionless time. In the radial models, since the physical systems consist of concentric cylinders, there is no consideration
for different shapes of reservoirs. In the new model, it is possible to change the shapes to any rectangles. Rectangles with any
aspect ratios can be modeled by providing the length and width of the reservoir.
In this section, the aquifer performance is examined for different aspect ratios of rectangular reservoirs. We start with a
square reservoir at the center of a square aquifer and change the aspect ratio of the reservoir from 1:1 to 1:5, 1:10, and 1:20,
while keeping the aquifer-reservoir contact area constant. The same dimensionless pressures are applied at the interface for
all four cases.
At first, we investigate the infinite aquifer case so that we can simplify the discussion by eliminating the boundary-
dominated flow period. Then, we analyze the finite aquifer case to see the effect of the aquifer boundaries.

Infinite Aquifer Case. As discussed above, we start with a square reservoir at the center of the square aquifer. In this
case, the aspect ratio of the reservoir is 1:1. The side lengths of the reservoir are selected arbitrarily as 2,000 ft. The infinite
aquifer condition is implemented by assigning large values for the length and the width of the aquifer. The thickness of the
aquifer is also arbitrarily selected as 200 ft. The dimensionless pressure drop at the aquifer-reservoir interface is assigned as
pD = 1.
The dimensionless cumulative water influx, WeD, and the dimensionless water influx rate, qeD, are calculated for all the
above four cases. The results for WeD-tD and qeD-tD are plotted in Figs. 6 and 7.
In Fig. 6, it is observed that both the linear flow period and the infinite acting transient flow period exist for all the cases.
In the linear flow period, all the curves are overlapped and there is no difference in the aquifer performance. The linear flow
period also appears in Fig. 7 and it is confirmed that all the curves are identical.
Fig. 6 shows that, in the infinite acting transient flow period, the dimensionless cumulative water influx becomes higher
for larger aspect ratios. For example, at tD = 0.4, WeD = 0.4 for aspect ratio of 1:1 while WeD = 0.6 for aspect ratio of 1:10.
Fig.7 shows that, at tD = 0.4, qeD = 0.4 for aspect ratio of 1:1 and qeD = 0.7 for aspect ratio of 1:10. This implies that the water
influx in the infinite acting transient flow period is dependent upon the shape of the reservoir.
In Fig. 7, although the dimensionless water influx rate is higher for larger aspect ratio, this effect becomes smaller at large
times. At tD = 0.1, the difference in the dimensionless water influx rate is obvious but the differences in the rates decline as
the time increases. As a matter of fact, for tD > 0.2, the rate responses of 1:1 and 1:2 cases are nearly identical.
In order to explain the above phenomena, we consider the propagation of the pressure disturbance in the aquifer. In the
linear flow period, the pressure disturbance travels mainly in the vertical direction and mostly the section of the aquifer
immediately below the reservoir responds. The water in that part of the aquifer moves upward. Since the area of the aquifer-
reservoir interface is kept constant for all the cases, there is no difference in the cumulative water influx in the linear flow
period. Therefore, the aquifer performance in the linear flow period is considered as a function of the contact area at the
interface, not a function of the shape.
In the infinite acting transient flow period, the pressure disturbance travels not only in the vertical direction but also in the
horizontal direction. When the pressure disturbance propagates in the horizontal direction, the shape of interface should
influence the way the drainage area expands. To explain this, the shapes of pressure disturbances for square and rectangular
interfaces are illustrated in Fig. 8. The path of water flow is also indicated with streamlines. In Fig.8, the streamlines for the
square interface are nearly radial, corresponding to the symmetric shape of the interface. On the other hand, the pressure
disturbance for the rectangular interface is more elliptical. In both cases, the streamlines in the vicinity of the reservoir are
forced to converge into the interface. However, due to the circular shape of the pressure disturbance, the streamlines are more
congested for the square interface, and the water feels more resistance to flow. On the other hand, in the rectangular interface,
the streamlines are nearly linear in the areal plane around the interface and water should flow more easily.
Because the water flows easily in the rectangular interface, the water influx rate is higher and the cumulative water influx
becomes larger as observed in Figs. 6 and 7. The level of flow convergence is lower in rectangular interface; therefore, the
flow is greater in case of the rectangular reservoir with larger aspect ratio.
The difference in the flow convergence is more obvious when the approximate radius of influence created by pressure
disturbance is near the interfaces. However, far from the interface, the difference in flow convergence becomes smaller and
smaller. This is why the difference in aspect ratios appear to be small at large times in Fig. 7.

Finite Aquifer Case. In the previous subsection, we examined the effect of the reservoir shape for the infinite aquifer and
discussed the results from the viewpoint of pressure disturbance. In this subsection, we investigate the finite aquifer case and
examine the boundary-dominated flow period. We start with the 2,000 ft-2,000 ft square reservoir. Then, we change the
aspect ratio from 1:1 to 1:2, 1:5, and 1:10 while keeping the interface area constant. We arbitrarily set both the length and
width of the aquifer as 18,000 ft for all four cases.
6 SPE 120897

The effect of reservoir shape on the simulated WeD-tD and qeD-tD relationships for finite aquifers is depicted in Figs. 9 and
10. To see the difference between the infinite acting flow period and the boundary-dominated flow, the infinite and finite
aquifer cases are compared for WeD and qeD in Figs. 11 and 12, respectively. In Figs 11 and 12, only the cases for aspect ratio
of 1:1 and 1:10 are plotted.
In Figure 9, the linear flow period, the infinite acting transient flow period, and the pot flow period are clearly observed.
In Figures 11 and 12, the curves for infinite and finite aquifer cases are deviating at large times. Therefore, we can also see
the boundary-dominated flow period between the infinite acting flow period and pot flow period.
In Figure 9, it is observed that there is no distinct difference in the ending time of the boundary-dominated flow period for
all four curves. The pot flow period starts at the dimensionless time of tD = 300 for all the cases.
It is difficult to see the starting time of the boundary-dominated flow period in Figs. 9 through 12. In order to see the start
of the boundary-dominated flow period, we use the relationship of qeD-tD. The qeD-tD curves for infinite and finite aquifer
cases are compared in Figs. 13 through 16. In Figs. 13 through 16, it is observed that the start of the boundary-dominated
flow period ranges from tD = 10 to tD = 20. As the aspect ratio increases, the boundary-dominated flow period emerges
earlier.
As we discussed in the previous subsection, the level of flow convergence depends on the shape of interface at early
times. However, far from the interface, the difference in flow convergence becomes smaller and smaller. The pressure
disturbance far from the interface is illustrated in Fig. 17. In Fig. 17, the pressure disturbance is considered to be nearly
identical for the square and the rectangular interfaces when the pressure disturbance travels away from the interface.
From the analysis above, we conclude that the aquifer size in this case is large enough to minimize the difference in
starting and ending times of the boundary-dominated flow period.

Water Influx for Irregularly-Shaped Reservoir. The new model can be used to calculate the water influx into the
reservoir with an irregular shape. As an example, we consider the shape of the main pool of Leduc-D3 basin as shown in Fig.
18. In Fig. 18, five rectangles are patched on the map to fit with the original shape of the reservoir. We consider the original
interface as a composite of the five sections. The water influx into each sectional interface can be calculated and the sum of
the sectional water influxes represents the total cumulative water influx through the total interface.
For illustration purposes, the aquifer shape and size are arbitrarily selected as shown in Fig. 19. We specify the location
and size of each rectangle by measuring the coordinate of the left-lower corner and length and width of each rectangle. The
area of each sub-reservoir is reported in Table 3. The pressure drop for each aquifer-reservoir interface is set as pD = 1 for all
the interfaces and is kept constant.
The cumulative water influx is calculated with the assumption described above. The results are graphically shown in Fig.
20. In Fig. 20, it is observed that there is the linear flow period until tD = 0.1. After the infinite acting transient flow period
and the boundary-dominated flow period, the water influx is terminated at tD= 2,000. tD= 2,000 marks the beginning of the
pot flow period. The durations of infinite acting and boundary-dominated flow periods could not be identified on Fig. 20. To
determine the durations of these two flow regimes, the WeD-tD curves for both the infinite and finite aquifers should be
compared.
This example calculation is performed for illustration purposes and the calculation is not limited to one composite
reservoir case or the constant pressure drop case. We can also calculate the water influxes for multiple composite reservoir
cases and for variable pressure drop at each interface section.

Conclusions
The main objective of this study was to develop a semi-analytical model to predict water influx into multiple reservoirs on a
common shared aquifer. This objective has been achieved. The water influx solutions are obtained for the following three
conditions at the aquifer-reservoir interface; constant water influx rate, constant pressure drop, and variable pressure drop. In
the case for variable pressure drop at the aquifer-reservoir interface, the step-wise pressure profile is directly transformed into
Laplace space, instead of applying the superposition principle in real time domain.
The final solutions are in the Laplace space and are numerically inverted to the real time domain. Computer codes in
FORTRAN language are built to calculate the solutions efficiently.
The new model developed in this study has been verified successfully for the constant water influx rate, constant pressure
drop, and variable pressure drop cases. Both finite and infinite aquifer cases are considered in the verification. The single
reservoir solution is compared with the models in the literature and very good agreements were established. The multiple
reservoir solution is tested by considering the symmetric distributions of identical reservoirs and comparing the results with
the single reservoir solution.
The specific results and conclusions of this study are listed below.

• Four distinct flow periods are identified on the log-log plot for WeD–tD relationship for a finite aquifer; the linear flow
period, the infinite acting flow period, the boundary-dominated flow period, and the pot flow period.
• In the linear flow period, the part of the aquifer immediately under the reservoir responds and the flow is dominantly
in the vertical plane. Therefore, the duration of the linear flow period is a function of the aquifer thickness.
SPE 120897 7

• Once the linear flow period ends, the pressure disturbance starts extending in the lateral direction. When it hits the
nearest boundary in x- or y- directions, the infinite-acting flow period ends and the water influx rate becomes lower
compared with the infinite aquifer case.
• The water influx rate and cumulative water influx in middle times are functions of the reservoir shape. The higher the
aspect ratio of the rectangular interface is, the higher the water influx rate and cumulative water influx are. This
phenomenon is related to the flow convergence associated with the pressure disturbance propagation.
• At large times, the difference, observed in the water influx rate for different shapes of the reservoirs, diminishes. This
is attributed to the negligible difference in flow convergence far from the reservoir-aquifer.
• The new model can be used for the reservoirs with irregular shape. The sample calculation is presented using the
actual reservoir shape from Leduc D-3 basin, Alberta.

Nomenclature
B = matrix coefficient
cea= effective aquifer compressibility, psi-1
ha = aquifer thickness, ft
haD = dimensionless aquifer thickness
H= unit step functions
IR = number of reservoir
kV = vertical permeability, darcies
kx = permeability in x-direction, md
ky = permeability in y-direction, md
kz = permeability in z-direction, md
kH = horizontal permeability, darcies
l= integer parameter for finite Fourier cosine transform
Lh = characteristic length, ft
Lp = characteristic length, ft
Lt = characteristic length, ft
Lv = characteristic length, ft
m = summation index, integer parameter for finite Fourier cosine transform
Nt = number of pressure pulse
p= aquifer pressure, psia
pD = dimensionless aquifer pressure
q= water influx rate, rbbl/d
q0 = water influx rate, rbbl/d
qeD = dimensionless water influx rate
ra = aquifer radius, ft
raD = dimensionless aquifer radius
rR = reservoir radius, ft
s= Laplace space variable corresponding to dimensionless time
t= time, days
tD = dimensionless time
We = cumulative water influx, STB
WeD = dimensionless water influx
xa = aquifer length, ft
xc = x coordinate for left-lower corner of reservoir, ft
xe = reservoir length, ft
ya = aquifer width, ft
yc = y coordinate for left-lower corner of reservoir, ft
ye = reservoir width, ft
αt = conversion factor
a = aquifer porosity, fraction
μw = water viscosity, cP

Subscripts:
0= reference
a= aquifer
D = dimensionless
i= initial
j= jth reservoir
8 SPE 120897

×= aquifer-reservoir interface
w= water

References
Al-Ajmi, N., Ahmadi, M., Ozkan, E. and Kazemi, H. 2008. Numerical Inversion of Laplace Transform in the Solution of Transient Flow
Problems with Discontinuities. Paper SPE 116255 presented at the SPE Annual Technical Conference and Exhibition, Denver,
Colorado, USA, 21-24 September
Allard, D.R. and Chen, S.M. 1988. Calculation of Water Influx for Bottomwater Drive Reservoirs. SPERE 3 (2): 369-379. SPE-13170-PA.
Bell, J.S. and Shepherd, J.M. 1951. Pressure Behavior in the Woodbine Sand. Trans., AIME, 192: 19-28 SPE-951019-G.
Coats, K.H. 1962. A Mathematical Model Water Movement about Bottom-Water-Drive Reservoirs. SPEJ 2 (1): 44-52. SPE-160-PA.
Delauretis, E.F., Yarranton, H.W. and Baker, R.O. 2008. Application of Material Balance and Volumetrics to Determine Reservoir Fluid
Saturations and Fluid Contact Levels. Journal of Canadian Petroleum Technology, 47 (3): 39-48.
Horsfield, R. 1958. Performance of the Leduc D-3 Reservoir. JPT 10 (2): 21-26. SPE-855-G.
Hurst, W. 1960. Interference Between Oil Fields. Trans., AIME, 219, 175-192. SPE-1335-G.
Iseger, P.D. 2006. Numerical Transform Inversion Using Gaussian Quadrature. Probability in the Engineering and Informational Sciences,
20: 1-44.
Mohammed, K., Elbekshi, A., Ghnia, S., Al-lottai, O., 2005. Reservoir Management for Fields Sharing Common Aquifer: A Case Study
from North African Sandstone Reservoirs El-Sharara Fields A, H and M NC-115, Libya. Paper SPE 94084 presented at the SPE
Europec/EAGE Annual Conference, Madrid, Spain, 13-16 June.
Moore, W.D. and Truby, L.G. 1952. The Pressure Performance of Five Fields Completed in a Common Aquifer. Trans., AIME, 195: 297-
302. SPE-238-G.
Mortada, M. 1955. A Practical Method for Treating Oilfield Interference in Water-Drive Reservoirs. Trans., AIME, 204: 217-226. SPE-
513-G.
NETLIB. AT&T Bell Laboratories, http://www.netlib.org/linpack/. Downloaded 5 March 5 2008.
Rodriguez, F., Samaniego-V, F., and Cinco-Ley, H. 1996. A Model for the Production Interference of Multiple Reservoirs Sharing a
Common Aquifer. SPE Advanced Technology Series, 4 (1): 38-43. SPE-26975-PA.
Sageev, A. and Horne, R. 1985. Interference Between Constant-Rate and Constant-Pressure Reservoirs Sharing Common Aquifer. SPEJ 25
(3): 419-426. SPE-12711-PA.
Sheshodia, T.S, Mubarak, M.S., and Rahman, B.A. 1989. History Match and Future Production Strategy of a Multi-Reservoir Field. Paper
SPE 17957 presented at the SPE Middle East Oil Technical Conference and Exhibition, Manama, Bahrain , 11-14 March.
Stehfest, H. 1970. Numerical Inversion of Laplace Transforms. Communications of the ACM 13 (1): 47-49
Tucker, C.C. 1989. Modeling Gas Deliverability From a Group of Fields That Communicate Via an Aquifer. Paper SPE 19492 presented at
the SPE Asia-Pacific Conference, Sydney, Australia, 13-15 September.
Van Everdingen, A.F. and Hurst, W. 1949. Application of the Laplace Transformation to Flow Problems in Reservoirs. Trans., AIME,
186:305-324. SPE-949305-G.
Werovsky, V., Tromboczky, S., Miklos. T. and Kristof, M. 1990. Case History of Algyo Field, Hungary. Paper SPE 20995 presented at
Europec 90, Hague, Netherlands, 22-24 October.
Yildiz, T. and Khosravi, A. 2007. An Analytical Bottomwaterdrive Aquifer Model for Material Balance Analysis. SPEREE 10 (6): 618-
628. SPE-103283-PA.

Appendix A – Bottomwaterdrive Common Aquifer Model

The three-dimensional (3D) diffusivity equation governs the single phase flow of water from a common aquifer into multiple
reservoirs. Initially, we assume that flow rate at all the reservoir-aquifer interfaces are known and time-dependent. A uniform
but time-dependent flux distribution is considered at interfaces.
The model is described in terms of dimensionless parameters. We introduce the following dimensionless parameters.

kL p
pD = ( pi − p ) ............................................................................................................................................... (A-1)
887.3q 0 μ w

k x
xD = ................................................................................................................................................................. (A-2)
k x Lh

k y
yD = ................................................................................................................................................................. (A-3)
k y Lh

k z
zD = ................................................................................................................................................................. (A-4)
k z Lv
SPE 120897 9

k ha
haD = ............................................................................................................................................................... (A-5)
k z Lv

αt k
tD = t .......................................................................................................................................................... (A-6)
φ a μ w c ea L2t

q ei (t )
qeiD (t D ) = .............................................................................................................................................................. (A-7)
q0

k = 3 k x k y k z .................................................................................................................................................................. (A-8)

In the equation above, Lp, Lh, Lv, and Lt are the characteristic lengths and q0 is a reference constant water influx rate.
The 3D dimensionless diffusivity equation governing the single phase flow of aquifer water is

2 2
∂ 2 pD ∂ 2 pD ⎛L ⎞ ∂ 2 p D ⎛ Lh ⎞ ∂p D
+ + ⎜⎜ h ⎟ ⎜ ⎟
⎟ ∂z 2 = ⎜ L ⎟ ∂t ............................................................................................................... (A-9)
∂x D2 ∂y D2 ⎝ Lv ⎠ D ⎝ t ⎠ D

The initial and boundary conditions are

tD = 0 pD (0, xD , yD , z D ) = 0 ............................................................................................................................... (A-10)

∂ pD
1. x D = 0 = 0 ................................................................................................................................................. (A-11)
∂ xD

∂ pD
2. x D = x aD = 0 ............................................................................................................................................. (A-12)
∂ xD

∂ pD
3. y D = 0 = 0 ............................................................................................................................................. (A-13)
∂ yD

∂ pD
4. y D = y aD = 0 ............................................................................................................................................. (A-14)
∂ yD

∂ pD
5. z D = 0 = 0 ............................................................................................................................................. (A-15)
∂ zD

∂ pD qeiD (t D )
IR
L L
6. z D = haD
∂ zD
= p2 v
Lh
∑i =1 xeiD yeiD
U xiDU yiD ...................................................................................................... (A-16)

where

U xiD = H x (x D − x ciD ) − H x (x D − x ciD − x eiD ) ................................................................................................................ (A-17)

U yiD = H y ( y D − y ciD ) − H y ( y D − y ciD − y eiD ) ............................................................................................................... (A-18)

In the expressions above, IR is the number of reservoirs supported by the common aquifer and H(ξ-a) is the Heaviside unit
step function.

General Aquifer Solution. For the initial-boundary value problem described by Eqs. A-9 through A-18, we apply Laplace
transformation on time variable. In the Laplace space, we use the finite Fourier cosine transforms on x and y directions. This
yields a second order homogeneous ordinary differential equation in z-direction. The solution to the second order ordinary
10 SPE 120897

differential equation is well known and given in terms of simple exponential function. We apply inverse finite Fourier cosine
transforms on the solution for the ordinary differential equation. The final general aquifer solution in the Laplace space is

( )+ x ∑ A (e )cos⎛⎜⎜ myπy ⎞
∞
1 2
p D (s, x D , y D , z D ) =
~ A0 e λ z + e − λ z 0 D 0 D
m
λm z D
+ e −λ m zD D
⎟
⎟
x aD y aD aD y aD m =1 ⎝ aD ⎠
⎛ lπx D ⎞
∑ A (e )
∞

)cos⎛⎜⎜ lπx x ⎞ ⎛ mπy D ⎞

2
∑ ∑ A (e
∞ ∞
+ λl z D
+ e −λ z cos⎜⎜l D
⎟+ 4
⎟ x y
λlm z D
+ e −λ lm z D D
⎟ cos⎜ ⎟ ............................. (A-19)
x aD y aD
l ⎟ ⎜ y ⎟
⎝ x aD
lm
l =1 ⎠ aD aD l =1 m =1 ⎝ aD ⎠ ⎝ aD ⎠

where

L p Lv IR

∑ q~ (s ) ..................................................................................................................... (A-20)
1
A0 = λ0 haD eiD
L2h λ 0 (e − e −λ0 haD ) i =1

2
L ⎛ Lh ⎞
λ0 = v ⎜ ⎟ s ....................................................................................................................................................... (A-21)
Lh ⎜L ⎟
⎝ t ⎠

q~eiD (s )
IR
L p Lv
∑
1
Am = λ m haD
Tymi ............................................................................................................. (A-22)
L 2
h λm (e − e−λ m h aD
) i =1 yeDi

2
Lv m 2 π 2 ⎛ Lh ⎞
λm = + ⎜⎜ ⎟⎟ s ......................................................................................................................................... (A-23)
Lh y aD
2
⎝ Lt ⎠

L p Lv I
q~eiD (s )
∑
R
1
Al = λl haD −λ h
Txli ................................................................................................................ (A-24)
L 2
h λ l (e −e ) i =1 x eiD
l aD

2
L l 2 π 2 ⎛ Lh ⎞
λl = v + ⎜⎜ ⎟ s ............................................................................................................................................ (A-25)
⎟
Lh x aD
2
⎝ Lt ⎠

q~eiD (s )
IR
L p Lv
∑x
1
Alm = λlm haD −λlm haD
TxliT ymi ............................................................................................... (A-26)
L2h λlm (e −e ) i =1 eiD y eiD

x aD ⎧⎪ ⎡ lπ (xciD + xeiD ) ⎤ ⎡ lπxciD ⎤ ⎫⎪

Txli = ⎨sin ⎢ ⎥ − sin ⎢ ⎥ ⎬ ....................................................................................................... (A-27)
lπ ⎪⎩ ⎣ x aD ⎦ ⎣ x aD ⎦ ⎪⎭

y aD ⎧⎪ ⎡ mπ ( y ciD + y eiD ) ⎤ ⎡ mπy ciD ⎤ ⎫⎪

T ymi = ⎨sin ⎢ ⎥ − sin ⎢ ⎥ ⎬ ................................................................................................. (A-28)
mπ ⎪⎩ ⎣ y aD ⎦ ⎣ y aD ⎦ ⎪⎭

2
Lv l 2π 2 m 2π 2 ⎛L ⎞
λlm = + + ⎜⎜ h ⎟ s .......................................................................................................................... (A-29)
⎟
Lh 2
x aD 2
y aD ⎝ Lt ⎠

In Eqs. A-19 through A-29, s is Laplace space variable corresponding to real time and l and m are integer parameters for
finite Fourier cosine transforms.

Interface Solutions. The solution given in Eqs. A-19 through A-29 allows us to predict the pressure at any point in the
aquifer. Now, we evaluate the solution at the aquifer-reservoir interface by simply specifying zD=haD in Eq. A-19.
SPE 120897 11

As stated previously, in our model, we initially consider uniform but different flux at all reservoir-aquifer interfaces. This
yields a variable pressure distribution at the interface location. To have a uniform dimensionless pressure at the interface, we
employ the pressure-averaging technique. Hence, the pressure at the interface between the jth reservoir and the aquifer is

p jD (s, x D , y D , haD ) = ~
~ p D (s, x D , y D , z D = haD ) ....................................................................................................................... (A-30)

xcjD + xejD ycjD + yejD

∫ ∫
1
p× jD (s ) =
~ p jD (s, x D , y D , haD ) dy D dx D ......................................................................................... (A-31)
~
x ejD y ejD xcjD ycjD

where the × sign in the subscript on the left-hand-side represents the interface. The × sign does not have anything to do
with x-direction or multiplication.
Evaluation of the integral in Eq. A-31 yields the following solution.

IR
p× jD (s ) = ∑ B ji (s ) q~eiD (s ) ...................................................................................................................................... (A-32)
~
i =1

B ji (s ) = β1 + β 2 + β 3 + β 4 ............................................................................................................................................ (A-33)

1 L p Lv coth (λ0 haD )

β1 = ................................................................................................................................. (A-34)
x aD y aD L2h λ0

L p Lv ∞
⎡ coth (λ l haD ) ⎤
∑ ⎢⎣T
2 1
β2 = xli T xlj ⎥ .................................................................................................... (A-35)
x aD y aD L2h x ejD x eiD l =1
λl ⎦

coth (λm haD ) ⎤ ........................................................................................................ (A-36)

∞
L p Lv ⎡
∑ ⎢⎣T
2 1
β3 = ymiTymj ⎥
xaD yaD L2h yejD yeiD m =1
λm ⎦

L p Lv ∞ ∞
⎡ coth (λ lm haD ) ⎤
∑ ∑ ⎢⎣T
4 1
β4 = T xli T xlj ymi T ymj ⎥ ......................................................... (A-37)
x aD y aD Lh x ejD y ejD x eiD y eiD
2
l =1 m =1
λlm ⎦

The expression presented in Eq. A-32 is the general interface solution. In Eq. A-32, neither interface pressure nor flow
rate is constrained. In the following section, using Eq. A-32, we present three specific solutions.

• Solution for constant water influx rate at the aquifer-reservoir interface

• Solution for constant pressure drop at the aquifer-reservoir interface
• Solution for variable pressure drop at the aquifer-reservoir interface

Solution for Constant Water Influx Rate. In the solution described by Eq. A-32, we specify constant water flow rate as

q eiD = Constant ............................................................................................................................................................. (A-38)

Laplace transform of Eq. A-38 is

q
q~eiD (s ) = eiD ............................................................................................................................................................... (A-39)
s

Substituting Eq. A-39 into Eq. A-32, we obtain

IR

∑B
1
p× jD (s ) =
~
ji ( s ) q eiD ........................................................................................................................................... (A-40)
s i =1
12 SPE 120897

The dimensionless interface pressure solution in Laplace space, Eq. A-40, may be inverted to the real time domain using
numerical inversion algorithms (Stehfest, 1970 and Iseger, 2006).

Solution for Constant Pressure Drop. The constant pressure drop at aquifer-reservoir interface is specified as follows.

p× jD = Constant ............................................................................................................................................................ (A-41)

Laplace transform of Eq. A-41 is

p xjD
p× jD (s ) =
~ ............................................................................................................................................................. (A-42)
s

In case of constant terminal pressure at the interface, the dimensionless pressure is defined differently. This definition is
provided below.

pi − p(t , x, y, z )
p D (t D ) = ............................................................................................................................................ (A-43)
pi − p 0

where p0 is a reference pressure. The dimensionless water influx is defined as below.

887.3 qi (t )μ w Lv
qeiD (t D ) = ...................................................................................................................... (A-44)
k z xej yej k k z ( pi − p0 )

Combining Eqs. A-42 and A-32, we obtain

IR

∑B
i =1
~
ji qeiD (s ) = p× jD s ................................................................................................................................................ (A-45)

To solve for rates at the interfaces, we re-write Eq. A-45 for all the reservoirs and recast the resulting system of linear
equations in the form of a matrix.

IR

∑B
i =1
~
1i q eiD (s ) = p×1D s ................................................................................................................................................ (A-46)

IR

∑B
i =1
~
2i qeiD (s ) = p× 2 D s ................................................................................................................................................ (A-47)

IR

∑B
i =1
~
3i qeiD (s ) = p×3D s ................................................................................................................................................ (A-48)

IR

∑ B(
i =1
~
I R −1)i q eiD (s ) = p× (I R −1 )D s ................................................................................................................................... (A-49)

IR

∑B
i =1
~
I Ri qeiD (s ) = p× I RD
s .............................................................................................................................................. (A-50)

Set of simultaneous linear equations may be put into a matrix as shown below.
SPE 120897 13

⎡ B11 B12 B13 " B1(I R −1) B1I R ⎤ ⎡ q~e1D (s ) ⎤ ⎡ p×1D s ⎤

⎢ B B 22 B23 " B2(I R −1) B 2 I R ⎥⎥ ⎢ q~ (s ) ⎥ ⎢ p ⎥
⎢ 21 ⎢ e2D ⎥ ⎢ ×2 D s ⎥
⎢ B31 B32 B33 " B3(I R −1) B3 I R ⎥ ⎢ q~e3 D (s ) ⎥ ⎢ p×3 D s ⎥
⎢ ⎥ ⎢ ⎥ =⎢ ⎥ ........................................ (A-51)
⎢ # ⎥ ⎢ # ⎥ ⎢ # ⎥
⎢ B(I −1)1 B(I R −1)2 B(I R −1)3 " B(I R −1)(I R −1) B(I R −1)I R ⎥ ⎢q~e(I −1)D (s )⎥ ⎢ p× (I −1)D s ⎥
⎢ R ⎥ ⎢ ~R ⎥ ⎢ R
⎥
⎢⎣ B I R 1 BI R 2 BI R 3 " B I R (I R −1) B I R I R ⎥⎦ ⎢⎣ q eI R D (s ) ⎥⎦ ⎢⎣ p× I R D s ⎥⎦

The solution of the matrix defined in Eq. A-51 yields the flow rates at all the interfaces. Once the rates are obtained as a
function of time then cumulative water influx could be estimated by simply integrating the rates over time.

tD
WejD (t D ) = ∫ 0
qejD (τ ) dτ ............................................................................................................................................. (A-52)

The dimensionless cumulative water influx in Eq. A-52 is defined as expressed below.

Wej (t )
WejD (t D ) = .................................................................................................... (A-53)
L2t x ej y ej
0.1781φ a c ea k z k ( pi − p0 )
Lv

Applying Laplace transformation on Eq. A-53, we obtain

~ 1
WejD (s ) = q~ejD (s ) ........................................................................................................................................................ (A-54)
s

The dimensionless water influx rate and the dimensionless cumulative water influx in Laplace space are inverted to the
real time domain using numerical inversion algorithms.

Solution for Variable Pressure Drop. The assumption of constant pressure drop at the interface is not realistic. During
the course of hydrocarbon production from subsurface reservoirs, the interface pressure varies with time. In the conventional
methods for computing cumulative water influx, a staircase interface pressure profile is generated from the measured average
reservoir pressure values. Here, we also assume step-wise pressure profiles at all the aquifer-reservoir interfaces.
The step-wise pressure profile for the jth reservoir-aquifer interface may mathematically be expressed in terms of
Heaviside unit step functions.

N tj
p× jD (t D ) = ∑ p {U (t ) ( )}............................................................................................(A-55)
kj
× jD D − t (k j −1)D − U t D − t k j D
k j =1

where Ntj is the number of average reservoir pressure measurements and t0D=0. Applying Laplace transformation to Eq. A-55,
we obtain

N tj

∑p
1 k j ⎛ − st (k j −1) − st k j ⎞ ...................................................................................................................... (A-56)
p× jD (s ) =
~
× jD ⎜ e −e ⎟
sk ⎝ ⎠
j =1

Rewriting Eq. A-56 for all reservoirs, and combining with Eq. A-32, we end up with a matrix similar to one presented in
Eq. A-51. Only the column on the right-hand-side of Eq. A-51 does change.
14 SPE 120897

⎡
∑ ( ) ⎤
Nt 1
− st − st
⎢ p×k11 D e (k1 −1) − e k1 ⎥
⎢ k1 =1 ⎥
⎢ ⎥
∑ ( )
Nt 2
⎢ − st
p×k22 D e ( k2 −1) − e k2
− st ⎥
⎡ B11 B12 B13 " B1( I −1) B1I ⎤ ⎡ q~e1D (s ) ⎤ ⎢ ⎥
⎢ B ⎢ q~ (s ) ⎥
B2 I ⎥⎥
R R

B22 B23 " B2 ( I −1) ⎢ k2 =1 ⎥

⎢ 21 ⎢ e2 D ⎥ ⎢
( ) ⎥
R R Nt 3
⎢ q~e3 D (s ) ⎥ ∑
− st − st
⎢ B31 B32 B33 " B3( I −1) B3 I ⎥ 1 ⎢ p×k33 D e (k3 −1) − e k3 ⎥
⎢ R R
⎥ ⎢ ⎥ = ⎢ k3 =1 ⎥ .. (A-57)
⎢ # ⎥ ⎢ # ⎥ s⎢
# ⎥
⎢ B( I −1)1 B( I −1)2 B( I −1)3 " B( I −1)( I −1) B( I −1)I ⎥ ⎢q~e(I −1)D (s )⎥ ⎢ Nt ( I R −1) ⎥
⎢ ⎥ ⎢ ~R ⎥ ⎢ ⎞⎥
∑
st(k −1)
p× ((IIR −−1)1)D ⎛⎜ e ( I R −1) − e ( I R −1)
R R R R R R R
k − − stk
⎣⎢ BI 1 BI 2 BI 3 " BI ( I −1) BI I ⎦⎥ ⎢⎣ q eI R D (s ) ⎥⎦ ⎢ ⎟⎥
R R R R R R R R ⎝ ⎠
⎢k( I R −1) =1 ⎥
⎢ NtI R ⎥
∑
− st(k −1)
p× IIR D ⎛⎜ e I R − e I R ⎞⎟
− stk
⎢ k
⎥
⎢ R ⎝ ⎠ ⎥
⎣ k I R =1 ⎦

For the staircase interface pressure profile, the dimensionless cumulative water influx is computed as described in Eqs. A-
52 through A-54.
The numerical inversion of Eq. A-57 requires special treatment. The column vector in Eq. A-57 contains exponential
terms. The Stehfest algorithm (1970), generally used to invert the solution from Laplace space to real time domain, cannot
handle the exponential terms properly. Hence, we use the Iseger method (2006) for the numerical inversion of the solution
given in Eq. A-57. The use of the Iseger algorithm for modeling the single phase flow problems in porous media has been
previously illustrated by Al-Ajmi et al. (2008).
SPE 120897 15

TABLE 1 – PROPERTIES OF RESERVOIR-AQUIFER SYSTEM IN THE

SAMPLE CALCULATION BY ALLARD AND CHEN
Parameter Value
rR, ft 2,000
ra, ft Infinity
ha, ft 200
kH, darcies 0.050
kV, darcies 0.002
φ 0.10
μ, cp 0.395
-1 -6
cea, psi 8 x 10
pi, psia 3,000
Bw, bbl/STB 1.0

TABLE 2 – SYSTEM PARAMETERS FOR A FINITE AQUIFER

Parameter Value
xa, ft 18,000
ya, ft 18,000
ha, ft 200
xe, ft 2,000
ye, ft 2,000
xc, ft 8,000
yc, ft 8,000
kx, md 1.0
ky, md 1.0
kz, md 1.0
pD 1.0

TABLE 3 – AREAS FOR SUB-RESERVOIRS

Sub-reservoir Area (acre)
1 4,320
2 9,600
3 640
4 2,560
5 640
Total 17,760

xa

(xcj+xej ycj+yej)
Common Aquifer

Reservoir j Reservoir j yej

(xcj, ycj) ya
Reservoir 1 xej
Reservoir IR Reservoir 1
Reservoir IR
y

Common Aquifer
x

Reservoir 1 Reservoir j Reservoir IR

z
ha
qej

Fig. 1 – Aquifer-reservoir model considered in this study.

16 SPE 120897

3000
This study
All ard -Chen model
Cumulative. water influx, MSTB

2500 1E+03
infinite aquifer case
2000 finite aquifer case
1E+02

1500
1E+01

W eD
1000
1E+00
500
1E-01
0
0 30 60 90 120 150 180 210 240 1E-02
Time, days
1E-03 1E -02 1E-01 1E+00 1E+01 1E+02 1E+03 1E+04

tD
Fig. 2 – We vs. t for a single bottomwaterdrive reservoir case.

Fig.5 – Flow regimes for bottomwaterdrive aquifers.

1E+01
Aquifer Aspect ratio = 1:1
Aspect ratio = 1:2
Aspect ratio = 1:5
Aquifer 1E+00 Aspect ratio = 1:10
Reservoir 1 Reservoir 2
WeD

Reservoir 1
1E-01

Reservoir 3 Reservoir 4

1E-02
1E-03 1E-02 1E-01 1E+00 1E+01
tD
Fig. 3 – Distributions of multiple reservoirs and reference
single reservoir case.
Figure 6 – Effect of reservoir shape on WeD vs. tD relatiotionship
in infinite aquifer case.
3000
Single reservoir
Cumula tive water influx, MSTB

Multi reservoir (Total)

2500
Allard-Chen model 1E+02
2000 Aspect ratio = 1:1
Aspect ratio = 1:2
1500 Aspect ratio = 1:5
1E+01 Aspect ratio = 1:10
1000
qD

500
1E+00
0
0 30 60 90 120 150 180 210 240
Time, days
1E-01
1E-03 1E-02 1E-01 1E+00 1E+01 1E+02
Fig. 4 – We vs. t for multiple reservoir case, model verification. tD

Figure 7 – Effect of reservoir shape on qeD vs. tD in infinite

aquifer case.
SPE 120897 17

1E+01
Aspect ratio = 1:1
Aspect ratio = 1:2
Aspect ratio = 1:5
Aspect ratio = 1:10
1E+00
Rectangular interface Square interface

qD
The front of pressure
1E-01

1E-02
1E-02 1E-01 1E+00 1E+01 1E+02
tD

Figure 10 – Effect of reservoir shape on qeD vs. tD in finite

aquifers.
Streamlines are less Streamlines are more
1E+02
Aspect ratio = 1:1 (infinite aquifer)
Aspect ratio = 1:1 (finite aquifer)
Figure 8 – Flow convergence around rectangular and square Aspect ratio = 1:10 (infinite aquifer)
Aspect ratio = 1:10 (finite aquifer)
interfaces at early times. W eD

1E+02 1E+01
Aspect ratio = 1:1
Aspect ratio = 1:2
Aspect ratio = 1:5
Aspect ratio = 1:10
1E+01
WeD

1E+00
1E+00 1E+01 1E +02 1E+03
tD
1E+00

Figure 11 – Influence of reservoir shape and aquifer boundary

on WeD vs. tD.

1E-01
1E-02 1E-01 1E+00 1E+01 1E +02 1E+03
tD 1E+00 Aspect ratio = 1:1 (i nfinite aquifer)
Aspect ratio = 1:1 (finite aquifer)
Aspect ratio = 1:10 (infinite aquifer)
Figure 9 – Effect of reservoir shape on WeD vs. tD relationship in Aspect ratio = 1:10 (finite aquifer)
finite aquifers.
qD

1E-01

1E-02
1E+00 1E+01 tD 1E+02 1E+03

Figure 12 - Impact of reservoir shape and aquifer boundary on

qeD vs. tD.
18 SPE 120897

1E+00
1E+00

1E -01
qD

1E-01

qD
Infinite aquifer case
Finite aquifer case Infinite aquifer case
Finite aqu ifer case
1E -02
1E-02
1E+00 1E+01 1E+02
tD 1E+00 1E+01 1E+02
tD

Figure 13 – Effect of aquifer boundary on qeD vs. tD for aspect

ratio of 1:1. Figure 16 – Boundary effect on qeD vs. tD for aspect ratio of
1:10.

1E+00
Flat interface Square interface

The front of pressure

disturbance
qD

1E-01

Infi nite aquifer case

Finite aquifer case

1E-02
1E+00 1E+01 1E+02
tD

Figure 14 – Influence of aquifer boundary on qeD vs. tD for Figure 17 - Drainage area influenced by pressure disturbance
aspect ratio of 1:2. at large times.

1E+00
qD

1E-01

Infinite aquifer case

Finite aqui fer case

1E-02
1E+00 1E+01 1E+02
tD

Figure 15 – Boundary effect on qeD vs. tD for aspect ratio of 1:5.

 SPE 120897 19

Figure 18 - Reservoir with irregular shape, Leduc-D3 basin,

main pool (Horsfield, 1958).

1.5 mile
5.8 miles
4.5 miles

1 mile

3 1 mile
4 miles
15 miles

2
2 miles
5 miles 5 miles
5 1 mile
2 miles 4
6 miles 1 mile 7 miles

2 miles
5 miles

ha = 500 ft
kx = ky = kz = 1.0

15 miles

Figure 19 – Discritized aquifer-reservoir interface, Leduc 3 Reservoir.

20 SPE 120897

1E+03

1E+02
W eD

1E+01

1E+00
1E-03 1E-02 1E-01 1E+00 1E+01 1E +02 1E+03 1E+04
tD

Figure 20 - WeD vs tD for a reservoir with irregular shape.