SPE 68165

Single Medium Simulation of Reservoirs with Conductive Faults and Fractures

Paul van Lingen (Institut Franais du Ptrole), Mustafa Sengul (Saudi Aramco), Jean-Marc Daniel (Institut Franais du
Ptrole), Luca Cosentino (Beicip-Franlab)
Copyright 2001, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the 2001 SPE Middle East Oil Show held in
Bahrain, 1720 March 2001.
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300
words; illustrations may not be copied. The abstract must contain conspicuous
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract
This paper describes a technique to incorporate conductive
faults and -fractures in a reservoir simulation model. The
technique is appropriate for reservoirs with significant primary
matrix productivity, and a large spacing between fractures (in
the order of tens of meters). The single porosity (one grid)
formulation is applied. Any fracture geometry can be
represented accurately without grid modifications. The key
elements of the approach are pseudo relative permeability
curves for grid bocks containing fractures. These curves are
determined through an analytical procedure, based on the local
fracture and matrix properties. The method is applied
successfully to a Middle East carbonate reservoir.
Introduction
Conductive faults and -fractures are common features in
reservoirs worldwide. Several methods have been proposed to
model fluid flow in fractured reservoirs. Well known is the
Warren and Root [1] approach, which applies homogenization
of the fracture properties and a dual porosity formulation
(separate matrix and fracture grids). The homogenization of
the fracture properties is possible when the Representative
Elementary Volume (REV) [2] of the fracture network is
smaller than the reservoir simulation grid.
This paper focuses on fracture patterns that can not be
homogenized, because the fracture REV is several times larger
that the grid block size. Figure 1 provides an illustration. This
figure represents a 10 by 22 kilometer sector of a much larger
carbonate reservoir. The displayed fracture pattern is obtained
from fracture analysis [3]. It is superimposed on a 250 by 250
meter simulation grid. In this example the conductive fractures
are identified as the key contributor to unstable movement of

the water injection front. Consequently, the fracture pattern

must be incorporated when a simulation model is to be built.
Several methods are available in the literature to integrate
conductive fractures in a reservoir simulation model. Some of
these methods are described below, with their strengths and
weaknesses summarized in Table 1.
Phelps et al [4] apply local grid refinement (LGR) to
model conductive fractures explicitly. This approach is
thorough in the handling of the physical elements of the
displacement process. However, numerical difficulties may be
experienced due to a large flow rate in grid blocks with a small
pore volume. This is especially the case when a rapidly
advancing water tongues develop due to gravity segregation in
the fractures. Henn et al [5] extend the LGR technique by
applying vertical lumping of the fractured grid blocks, in
combination with an analytical treatment of capillary/gravity
equilibrium in the fracture. Their method strongly improves
numerical performance of the simulation model. The general
disadvantage of these LGR approaches is that they are
impractical for complex fracture patterns like the one shown in
Figure 1. Cosentino et al [6] apply a dual media (dual
permeability dual porosity) formulation to model conductive
fractures explicitly. Their approach facilitates the capture of
any complex fracture geometry without the need for
modifications to the grid geometry. The method is thorough in
its treatment of the processes that may take place during water
flooding. The numerical challenges, with respect to high flow
rates in grid cells with a small pore volume, are comparable to
those experienced in the conventional LGR approach.
Subsequently, relatively small time steps are required to
adhere to stability criteria.
The present paper describes another alternative to include
conductive fractures in a fluid flow simulator. Strong points of
this alternative are a single medium formulation (an
appropriate dual media formulation is not always available in
commercial software) and the possibility of dealing with any
fracture geometry on a regular grid. The approach is based on
a fixed order of oil displacement from grid cells containing
fractures: the fracture volume within a grid block is assumed
to fill with water before the matrix volume of this grid block is
flooded. This displacement sequence occurs when capillary
driven fluid exchange between the matrix and the fracture
system is inferior to the so-called viscous and

PAUL VAN LINGEN, MUSTAFA SENGUL, JEAN-MARC DANIEL, LUCA COSENTINO

gravitational displacement processes. The method is thus

appropriate for reservoirs with significant primary matrix
productivity and a large spacing between fractures (in the
order of tens of meters). The method can be used regardless of
the fracture spacing when the matrix is oil wet. Under this
condition capillary driven flow of oil from the matrix into the
fracture system does not occur. The key element of the
implementation of the approach is the use of pseudo relative
permeability curves. Expected numerical difficulties are
comparable to those of the conventional LGR technique. We
apply the method successfully during a field study, in which
we compare it with the dual media approach mentioned earlier
[6].
Detailing of the Approach
It is assumed that a fracture map and a simulation grid of the
matrix are available. The objective is to incorporate the map of
conductive fractures in the simulation grid. Figure 1 illustrates
the modelling problem. The picture shows a 250 by 250 meter
simulation grid with superimposed a fracture pattern. In this
figure, each of the bold lines represent a fracture swarm. A
fracture swarm may consist of many parallel fractures, but is
conceptually viewed as a void (porosity = 1) with an aperture
df and a permeability of kf. The grid blocks affected by the
fractures are colored grey. The porosity, permeability, and
relative permeability curves of these grey grid blocks will be
modified to reflect the presence of fractures. Additional
modifications are applied to the transmissibility of the
boundary between the fractured and unfractured grid blocks.
Finally, a special treatment is required for fractured grid
blocks in which a well is located.
The following procedures assume that fracture swarms add
to the porosity and permeability of a grid block without
reducing the matrix contribution (i.e. fracture swarms are
present without taking the place of the matrix). This
assumption simplifies the equations without significantly
affecting their accuracy.
Conductivity. Formally, a tensor notation is required to
incorporate the fracture conductivity in the matrix grid.
Formally, a tensor notation is required to incorporate the
fracture conductivity in the matrix grid. This option is not
implemented in todays commercial simulation software, and
is not required for practical purposes. We apply a two step
procedure. In the first step, the following permeability kb is
calculated for the grid blocks containing fractures:

kb = k m +

k f nf d f
db

................................................. (1)

Where subscripts b, m, and f stand for Grid block, Matrix

and Fracture respectively. km is the matrix permeability, kf
is the effective permeability of the fracture swarm, df is the
cumulative fracture swarm aperture, db is the grid block
spacing (assumed equal in x and y directions), and nf gives the
number of fracture swarms present in the fractured block. In
the second step, we reset the transmisivity of the perimeter of

SPE 68165

fractured grid blocks to its original (matrix) value using a

transmisivity multiplier MT,ij (this multiplier thus applies to
boundaries between fractured and non fractured grid blocks):

M T ,ij =

Tij , m
Tij ,b

............................................................. (2)

where Tij,m is the transmisivity between grid blocks i and j

considering only matrix, and Tij,b is the transmissibility
between block i and j after incorporation of the fracture
permeability using Eq. 1.
The transmissibility Tij between grid blocks i and j is given
by the following harmonic average:

Tij =

2k i k j

d b (k i + k j )

,..................................................... (3)

Note that the conductivity modifications described above

are absolutely correct only when the orientation of all fracture
swarms in a given grid block is the same. In the presented
field case, this is generally the true. If not so, the permeability
of some fractured cells is overestimated, which may not have a
dramatic effect on the results.
Vertical Permeability. In an automated implementation of
the procedure, it may be useful to allow for the modification of
the KV/KH ratio of fracture grid blocks, as this ratio is
possibly higher than in the unfractured parts of the field. Note
that in our application, the use of a unit KV/KH ratio required
unpractical small timesteps. A high vertical permeability
causes a rapidly advancing water tong in the fracture system
due to gravity segregation. A small time step is required to
maintain a numerically stable solution.
Porosity. The average porosity b of grid blocks with
fractures is calculated using the following arithmetic average:

b = m +

flf df
d b2

................................................... (4)

where lf is the cumulative length of the fracture swarms in

the given grid block, f is the fracture porosity (taken as 1),
and m is the matrix porosity. Note again in Eq. 4 that the
small reduction in matrix volume due to the presence of
fractures is ignored.
Pseudo Relative Permeability Curves
The key to obtaining the rapid water advance due to the
presence of conductive fractures is the use of pseudo relative
permeability curves. These curves are calculated analytically,
based on the assumption that the fracture volume of a grid
block is filled with water prior to imbibition of water into the
matrix.
Pseudo End-Points. The effective residual saturations and
end-point relative permeabilities of grid blocks are changed by
the presence of fracture swarms.

SPE 68165

SINGLE MEDIUM SIMULATION OF RESERVOIRS WITH CONDUCTIVE FAULTS AND FRACTURES

Residual Saturations. The effective residual oil saturation

of the fractured gird blocks Sor,b is calculated using the
following arithmetic average:

S or ,b =

S or , m m d + S or , f f l f d f
2
b

m d b2 + f l f d f

S wr ,b =

m d b2 + f l f d f

k oe ,b =

km d b + k f n f d f

........................ (6)

k we , m k m d b + k we , f k f n f d f
km d b + k f n f d f

S wc, f )l f d f f
+
S wc, f )l f d f f

........................ (8)

where kwe,m is the matrix end point relative permeability to

water (at the residual oil saturation) in the matrix, and kwe,f is
the end point relative permeability to water in the fracture.
Relative permeability curves for fractured cells. As
mentioned earlier, matrix and fracture relative permeability
curves are combined under the assumption that, within a grid
block, the fractures are invaded by water before the matrix.
The procedure of generating the pseudo curves essentially
consists of the combining of the relative permeability curves
of the two media fracture and matrix- in one single pseudo
curve. This process is illustrated in Figure 2. First, the original
matrix and fracture curves are normalized between 0 and 1,
both with respect to the saturation and with respect to the endpoint relative permeability. The parameters f, f,w and m,o
(Figure 2) required to combine the matrix and the fracture
relative permeability are defined below.
f, the contribution of the fracture volume to the total
mobile porosity in a grid block, is given by:

S wc , f )l f d f f

or ,m

.................................... (9)

S wc,m )d b2m

f,w, the contribution of fracture to the maximum grid block

relative permeability to water, is defined as:
f ,w =

k f k rwe, f n f d f
k f k rwe, f n f d f + k m k rwe ,m db

............................... (10)

m,o, the contribution of the matrix to the maximum grid

block relative permeability to oil, is defined as:
m ,o =

k m k roe,m d b
................................. (11)
k f k roe, f n f d f + k m k roe ,m d b

The points of the original matrix curves are now modified

using the following transformations:
S wn,b = S wn ,m (1 f ) + f

k rwn,b = k rwn,m (1 f ,w ) + f ,w .................................... (12)

k
ron,b = f ,o k ron,m

.......................... (7)

where koe,m is the matrix end point relative permeability to

oil (at the residual water saturation) in the matrix, and koe,f is
the end point relative permeability to oil in the fracture.
Similarly, the effective endpoint relative permeability kwe,b
to water in grid blocks containing fractures is calculated as:

k we ,b =

or , f
or , f

......................... (5)

where Swr,m is the residual water saturation in the matrix

and Swr,f is the residual water saturation in the fracture.
End-Point Relative Permeabilities. The effective endpoint
relative permeability koe,b to oil in grid blocks containing
fractures is calculated as:

k oe , m k m d b + k oe , f k f n f d f

(1 S
(1 S
(1 S
+
(1 S

or , f

where Sor,m is the residual oil saturation in the matrix and

Sor,f is the residual oil saturation in the fracture.
Similarly, the effective connate water saturation of the
fractured gird block (Swr,b) is calculated as:

S wr , m m d b2 + S wr , f f l f d f

f =

In addition, the following two points are added to generate

the grid block pseudo relative permeability curves*:
k rwn,b ( S wn,b = 0) = 0
k ron,b (S wn,b = 0) = 1

................................................ (13)

Treatment of wells. We implemented the approach for

vertical wells. When a vertical well is located in a grid column
containing a fracture, it is as if this fracture intersects the well
over the complete well length. Generally, this will not be the
case. The probability of a vertical well intersecting a vertical
fracture is small (such occurrence would give a clear signature
in the field data). In general, there will either be some distance
between the well and the fracture, or the fracture will cut the
well locally (rather than over the complete reservoir height).
To reflect that the fracture does not intersect the whole well,
two modifications are applied, which are detailed below.
The first modification concerns a multiplier for the
productivity index** (PI) in perforated grid blocks that contain
fractures:
k
M PI ,i ,n = Cn m .................................................... (14)
k b i ,n
Here, MPI,i,n is a local productivity (or injectivity) index
multiplier for perforated grid block i of well n, km is the matrix
permeability of grid block i, and kb is the permeability of
*

Note that any shape of fracture relative permeability curve

can be applied by slightly extending the procedure.
**
or injectivity index for injection wells.

PAUL VAN LINGEN, MUSTAFA SENGUL, JEAN-MARC DANIEL, LUCA COSENTINO

matrix plus fracture in grid block i. Cn represents a global

multiplyer for all perforated grid blocks of well n. This
multiplier can be increased above 1, if required, to reflect a
global PI increase due to the proximity of a fracture. The idea
behind this PI modification is the following. Due to its high
vertical permeability we may consider a fracture as a constant
pressure boundary. Equally, the vertical well may be
considered as a constant pressure boundary. If the well and
the fracture are parallel, the flux between these constant
pressure boundaries is proportional to the permeability inbetween. Thus the shape of a flowmeter profile is not affected
by a fault running parallel to a well, as long as the fault does
not cut the well. For a fracture cutting the well at a certain
depth, it is appropriate not to readjust the PI in that location.
The second modification to perforated grid blocks that
contain fractures is to reset the relative permeability curves to
their original (matrix) shapes. This modification again reflects
that the fracture does not actually intersect the well trajectory.
In the field application, the modifications to fractured grid
blocks containing a well were clearly justified through
comparison of the simulation results with field data, in
particular flow meter logs.
Horizontal wells. The approach was not tested for
horizontal wells. However, a similar treatment as to vertical
wells is expected to be appropriate. This means application of
the two modifications described above for grid blocks where
no intersection between the well and the fracture is expected,
and no modifications in grid blocks where the well and the
fracture intersect (these intersections are probable for
horizontal wells).
Automated Implementation
Application of the proposed method requires an automated
procedure. For this purpose, an interactive Visual Basic
application was written. An important aspect of the
implementation is the clustering of the calculated pseudo
relative permeability curves. Due to the spatial variations in
parameters required computing the pseudos, a unique set of
water- and oil curves is obtained for each fractured grid block.
This may easily lead to thousands of pseudos for a medium
sized simulation model, thus exceeding the capacity of the
simulation software. The appropriate algorithm to cluster the
calculated pseudo curves in a manageable number of groups
may vary depending on the reservoir characteristics. In the
present case, the grouping is performed firstly on a layer-bylayer basis, secondly on the number of fracture swarms present
(nf), and thirdly on the grid block matrix permeability (km).
Table 2 shows a fictive example in which 18 different groups
of pseudos are defined. Within each group, average values are
determined for the parameters required to calculate the pseudo
curves.
After performing all steps described above, the application
exports the results in the form of a simulator input deck.

SPE 68165

Field Application
The described procedure was developed and implemented in
the context of an integrated study of a Middle East carbonate
reservoir. The reservoir is formed by an extensive anticline,
and is stratigraphically positioned in the ARAB D formation.
Within this reservoir, the study site covers an area of 10 by 22
km (Figure 1).
The study site has been on mild depletion from a few wells
for almost 30 years. Real development started in 1995 with the
implementation of a peripheral water injection scheme.
Shortly after the start of injection, several parts of the study
site experienced early water breakthrough, corresponding to a
rapid advance of the waterfront with up to 40 meters per day.
The first phase of the study concerned the geological
characterization of the study site [7]. This characterization
focussed on the Matrix, the Fractures, and the so-called
statiform Super-K intervals*. The fracture map that was
produced showed a clear match with the pattern of early water
breakthrough seen in the study site, thus pointing towards
fracturing as the primary cause for unstable displacement.
The following simulation phase focussed on the
construction of the simulation model, history matching, and
forecasting. In fact, the primary focus of the simulation work
was the construction of a dual media simulation model [6].
This model was history matched with remarkable ease, thus
validating the geological characterization and the dual media
simulation approach.
On the request of the field operator, the history-matched
data were transferred into a single media model, using the
procedure described above. To this end, the stratiform SuperK and Matrix were merged using straightforward arithmetic
averaging of permeability and porosity. Subsequently, the
approach described in this paper was applied to incorporate
the fracture pattern. Clustering led to 61 sets of pseudos,
which are plotted in Figure 3. This figure illustrates the visual
difference between the produced pseudos.
Results of the single medium model, over the history
match period were remarkable, and comparable to those of the
dual media model. This indicates that the applied single media
approach is a viable option for modeling water flooding in
reservoirs with conductive fractures.
As both the single medium and dual media models were
available for the study, the dual medium formulation was
chosen as the preferred option for the continued modeling
work on the field. This choice is justified by a more complete
treatment of all elements of the displacement process, in
particular capillary exchange between the fracture and the
matrix. This mechanism may become more important as the
field reaches maturity. With resect to the CPU time, the single
media model was slightly faster, but not enough to count as a
persuasive argument in favor of the approach.

Thin, discontinuous layers, with a productivity of more than

500 STB/ft/Day and a typical permeability of several Darcy.

SPE 68165

SINGLE MEDIUM SIMULATION OF RESERVOIRS WITH CONDUCTIVE FAULTS AND FRACTURES

Conclusions
A procedure is developed to include an irregular pattern
of conductive fractures in a single medium simulation
model.
The method accurately captures any fracture geometry
without modification of the grid geometry.
The method is appropriate for reservoirs with significant
matrix productivity and a large spacing between fractures
(in the order of hundreds of meters).
The method was successfully applied during an integrated
study on a giant Middle East carbonate reservoir.

References
[1] Warren, J.E. and Root, P.J.: The Behavior of Naturally
fractured reservoirs, SPE Journal (Sept. 1963), 245-255.
[2] Bear, J. and Buchlin, J-M: Modelling and Application of
Transport Phenomena in Porous Media, (Chapter 1),
Kluwer Academic Publishers, Dordrecht, (1991).
[3] Sabathier, J., Bourbiaux, B., Cacas, M. and Sarda, S.: A
New Approach of Fractured Reservoirs, in Proceedings
SPE Int. Pet. Conf. & Exh., Villahermosa, Mexico, (March
1998), pp. 49-59, (SPE 39825).
[4] Phelps, R., Pham, T., and Shahri, A.: Rigorous Inclusion
of Faults and Fractures in 3-D Simulation, Saudi Aramco
Journal of Technology, WINTER 1999/2000, P. 43-52
[5] Henn, N., Bourbiaux B., Quintard, M., and Sakthikumar
S., Modelling conductive faults with a multiscale
approach involving segregation concepts, Paper 2022,
EAGE 10th European Symposium on Improved Oil
Recovery, Brighton, U.K., august 18-20, 1999.
[6] Cosentino L., Coury, Y., Daniel. J., Manceau, E. Ravenne,
C., van Lingen, P., Cole, J. and Sengul, M. Integrated
Study of a Fractured Middle East Reservoir with
Stratiform Super-K Intervals Part 2: Upscaling and Dual
Media Simulation, in Proceedings of the Middle East Oil
Show, Bahrain, 2001 (SPE 68184).
[7] Ravenne, C., Daniel J.M., Lecomte J.C., Camus D.,
Chautru J.M., Cosentino L., Coury Y., Cole J., Sengul M.:
Integrated Study of a Fractured Middle East Reservoir with
Stratiform Super-K Intervals Part 1 : Geological Model,
in Proceedings of the Middle East Oil Show, Bahrain, 2001
(SPE 68183).

Acknowledgements
The authors thank Saudi Aramco management for the
permission to publish this paper.
Nomenclature
d
aperture or width, L
k
permeability, L2
l
length, L
M
multiplier
n
number of
S
saturation
T
transmisivity, L

parameter used for the calculation of pseudos

parameter used for the calculation of pseudos

porosity
Subscripts
b
grid block (matrix + fracture)
e
end-point
f
fracture
i
grid block indicator
m
matrix
n
normalized, or well indicator
o
oil
w
water

Table 2 Fictive example of clustering into 18 pseudo

groups, based on the grid layer, the number of
fractures, and the matrix permeability

Table 1 - Strengths and weaknesses of approaches to modeling conductive

fractures
NUMERICAL
PERFORMANCE FLEXIBILITY

AVAILABILITY

PHYSICS

CONVENTIONAL LGR

++

++

LGR + VERTICAL LUMPING

--

++

DUAL MEDIA
(EXPLICIT FRACTURES)

++

++

PSEUDO'S

++

++

Pseudo
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

layer
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8
1-4
5-8

nf
1
1
2
2
3+
3+
1
1
2
2
3+
3+
1
1
2
2
3+
3+

k m (mD)
0.1 - 10
0.1 - 10
0.1 - 10
0.1 - 10
0.1 - 10
0.1 - 10
10 - 50
10 - 50
10 - 50
10 - 50
10 - 50
10 - 50
50 - 100
50 - 100
50 - 100
50 - 100
50 - 100
50 - 100

PAUL VAN LINGEN, MUSTAFA SENGUL, JEAN-MARC DANIEL, LUCA COSENTINO

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

SPE 68165

9 10 1

40

40

39
38

39
38

37
36
35

37
36
35

34
33
32

34
33
32

31
30
29

31
30
29

28

28

27
26

27
26

25
24
23

25
24
23

22
21
20

22
21
20

19
18
17

19
18
17

16
15

16
15

14
13

14
13

12
11
10

12
11
10

9
8
7

9
8
7

6
5
4

6
5
4

3
2
1

3
2
1
1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

9 10 1

Figure 1- Top view of an Irregular pattern of conductive fractures swarms superimposed on a 250x250 meter simulation grid

wa
t er

oil

normalized water saturation

r
te
wa

il

normalized water saturation

1
PSEUDO CURVES

l
oi

m ,o

Normalized Kr

MATRIX CURVES

Normalized Kr

Normalized Kr
0

FRACTURE CURVES

wa

te

f ,w
0

normalized water saturation

Figure 2- Generation of pseudos by combining the fracture and the matrix relative permeability curves

SPE 68165

SINGLE MEDIUM SIMULATION OF RESERVOIRS WITH CONDUCTIVE FAULTS AND FRACTURES

1.0

0.8

water relative permeability

oil relative permeability

1.0

0.6

0.4

0.2

0.0

0.8

0.6

0.4

0.2

0.0
0.0

0.2

0.4
0.6
water saturation

0.8

0.0

0.2

0.4
0.6
water saturation

Figure 3- 61 pseudo relative permeability curves obtained for the field case. Oil (left) and water (right)

0.8