TPG4150 Reservoir Recovery Techniques

Please read as supplementary text to lectures. Please focus on

the discussions of physical behavior. Sections on reservoir
simulation are not relevant for this course

Dual Porosity, Dual Permeability Formulation for Fractured Reser-

voir Simulation
K. Uleberg and J. Kleppe, Norwegian University of Science and Technology (NTNU), Trondheim
RUTH Seminar, Stavanger 1996

Abstract Based on the theory of fluid flow in fractured po-

rous media developed in the 1960's by Barrenblatt
et al.1, Warren and Root2 introduced the concept
This study reviews key physical mechanisms and
of dual-porosity models into petroleum reservoir
calculation methods for modelling of fluid flow in
engineering. Their idealized model of a highly
North Sea fractured reservoirs. The main matrix-
interconnected set of fractures which is supplied
fracture fluid exchange mechanisms described are
by fluids from numerous small matrix blocks, is
gravity drainage, capillary imbibition and molecu-
shown in Fig. 1. Kazemi et al.3 were the first to
lar diffusion. Important issues such as capillary
incorporate the dual-porosity concept into a nu-
continuity between matrix blocks, reinfiltration of
merical model, with application to fluid flow on a
fluids from higher to lower blocks and effect of
large scale.
block shape on flow processes are also addressed.

Simulation studies of water-flooding in fractured

reservoirs are reported for the purpose of identify-
ing the effects of gravity and capillary forces on
oil recovery. Included are studies of effects of ca-
pillary continuity and degree of wetting. The re-
sults show that for intermediately wetted systems,
such as the Ekofisk reservoir, capillary continuity
has a major effect on oil recovery.

Laboratory processes involving high pressure gas

injection in fractured systems have been studied
Fig. 1 Idealization of a fractured system (from
by compositional simulation. The results show
Warren and Root2)
that changes in interfacial tension caused by diffu-
sion, may have dramatic effects on oil recovery.
Since that time, numerical modelling of naturally
fractured reservoirs using dual-porosity models
Computational aspects of fluid exchange process-
has been the subject of numerous investigations.
es are discussed, including conventional dual po-
In the dual-porosity and dual-porosity/dual-
rosity formulation, use of matrix-fracture transfer
permeability formulations most commonly used to
functions, and detailed numerical calculation. The
model fractured reservoirs, proper representation
only solution to more representative modelling of
of imbibition and gravity drainage is difficult. In
flow in fractured reservoirs is more detailed calcu-
some formulations, attempts have been made to
lations. A multiple grid concept is proposed which
represent correct behavior by employing a gravity
may drastically increase the detail of the simula-
term, and assuming a simplified fluid distribution
tion.
in the matrix.4-6 Several authors7-10 have made use
of capillary pressure pseudofunctions for the ma-
Introduction trix and/or the fracture that employ matrix fluid
distributions obtained through some type of histo-
ry matching with a fine-grid model of a single
 Combined Gas/Water Injection Subprogram

matrix block. Others6,11-12 have refined the matrix during a gas/oil drainage process has not been
blocks into multiple blocks. The more recent of extensively treated in the literature.
these publications on dual-porosity modelling deal
with enhancements of the representation of gravi- It is obvious that the current computational proce-
ty effects in the matrix-fracture transfer calcula- dures are insufficient to represent the physical
tion. flow phenomena taking place between fracture
and matrix in highly fractured reservoirs such as
In the North Sea, the Ekofisk Field has been sub- in the North Sea. Clearly, additional details in
ject to several simulation studies, such as Phillips'
study of water imbibition13 and Petrofina's study,14
which includes an evaluation of the effect of ca-
pillary continuity on recovery.

Current state-of-art in the area of fractured reser-

voir simulation, with emphasis on gas/oil gravity
drainage in terms of block-to-block processes, is
reviewed by Fung.15 He classifies the current
methods into four groups: 1) gravity-segregated,
2) subdomain, 3) pseudofunction, and 4) dual-
permeability models. Fung proposes a new ap-
proach for calculation of pseudo capillary pressure
pressures, either a priori if vertical equilibrium
can be assumed, or by fine grid simulation if not.

Although several publications have discussed re-

imbibition of oil from higher matrix blocks into
lower matrix blocks, all models reviewed by
Fung15 neglect capillary continuity (Fig. 2), except
the dual-permeability model, where the matrix is
assumed to be completely continuous.

Fig. 3 Reinfiltration of fluids from higher to lower

matrix blocks (from Fung15)

the description of the flow processes in the mod-

els are required.

Simulation of fractured reservoirs using the dual-

porosity (and dual-permeability) approach in-
volves discretization of the solution domain into
two continua called the matrix and the fracture.
The original idealized model of Warren and Root2
assumes that the matrix acts essentially as a
source or sink to the fracture, which is the primary
conduit for fluid flow. In multiphase flow situa-
tions, however, complex gravity and capillary
pressure driven fluid exchange between fractures
and matrix is occurring, and these processes are
not well understood.

Especially the gas-oil gravity-drainage process in

fractured reservoirs will require improved model-
Fig. 2 Effect of vertical capillary continuity on
ling in order to obtain efficient field-scale simula-
saturation distribution (from Fung15)
tion. The gravity segregation in a dual-porosity
medium is highly affected by capillary continuity
None of the models treats the reinfiltration phe-
between matrix blocks across fractures, and by the
nomenon properly (Fig. 3). Finally, the im-
process of oil reinfiltration from fractures to ma-
portance of the effect of gas-gas and gas-liquid
trix blocks. These mechanisms are not adequately
diffusion on recovery from fractured reservoirs
 Dual Porosity, Dual Permeability Formulation for Fractured Reservoir Simulation

represented in today's simulation models and sev- capillary forces and gravity, therefore the term
eral research groups are working on the prob- reinfiltration should be used instead of the much
lem.15-19 Luan20 presented a systematic study and a used word reimbibition, since the latter could im-
comprehensive discussion of the fundamental ply the capillary effect only.
physical involved in gravity drainage of fractured
reservoirs. The flow from one block to another (reinfiltration)
is either achieved by 1) film flow across contact
The computational side of fractured reservoir points or 2) by liquid bridges. This liquid trans-
modelling is particularly difficult due to the large missibility across the fracture is therefore an im-
scale differences. Important flow processes are portant parameter for calculating the rate of drain-
taking place at a scale much smaller than the grid age of a stack of matrix blocks. Fig. 3 illustrates
blocks normally employed in dual porosity mod- the contact points and the liquid bridges.
els. Uleberg21 has made a study of application of
local grid refining techniques in dual-porosity Experimental results18 have shown that the trans-
formulations for detailed computation of matrix- missibility across a fracture is very sensitive to the
fracture fluid exchange. fracture aperture, but not so sensitive to the num-
ber of contact points or contact area. The reinfil-
Physical Characteristics and tration mechanism is also time dependent, since
liquid bridging provides the main transmissibility
Fluid Flow Mechanisms in the initial stage of the gravity drainage process.
Later the oil saturation in the fractures will be
Capillary Continuity very low and the main liquid transmissibility from
The concept of capillary continuity between ma- block to block is due to film flow. This final peri-
trix blocks in fractured reservoirs is now widely od is of long duration and is very important for
accepted. A schematic comparison of capillary- the overall recovery. The reinfiltration process is
gravity dominated saturation distributions for a not adequately modelled in the reservoir simula-
discontinuous system and a system with capillary tors used to predict gravity-drainage oil recovery
contact between matrix blocks is shown in Fig. 2. in fractured reservoirs.
However, the discontinuous concept is still being
used in most commercial dual-porosity models to Da Silva and Meyer16 conducted a simulation
handle block to block interactions. study of reinfiltration, and concluded that this
may be an important mechanism for systems of
Festøy and van Golf-Racht22 presented simulation capillary continuity and large fracture angles.
results where the fracture system allowed for var-
ious degrees of matrix to matrix contact. Their Diffusion
results show dramatically higher recoveries for Oil may be recovered by diffusion during gravity
capillary continuous systems compared to discon- drainage in fractured reservoirs. Methods for es-
tinuous ones. They suggested that the matrix is timating the amount and rate of this recovery in
better described as tortuously continuous. such reservoir processes are in early stages of de-
velopment and poorly tested. It is very limited
Luan´s20 discussion on this subject concludes that published data against which theories and predic-
the end effects may be important in a drainage tion methods can be tested adequately. However,
process in fractured reservoirs. The saturation dis- according to Orr,25 the scaling efforts based on
tribution at the endface of the blocks is dependent fundamental physics must be continued, specially
on the wetting conditions and the properties of the in fractured reservoir problems, including work on
fractured medium. Experimental studies23,24 show diffusion/dispersion as well as gravity, viscous
that the end effects (caused by saturation disconti- and capillary forces. An interesting effect may be
nuity) can be reduced by increased contact areas due to interfacial tension gradients caused by dif-
between blocks (applying overburden pressure in fusion of gas into the oil. The interfacial tension
the laboratory). induced capillary pressure gradients may result in
unexpected saturation profiles.26-27 Hua et al.27
Reinfiltration conducted a series of simulations, and were able
An important aspect in gas-oil gravity drainage of to reproduce the experimental results. However,
fractured reservoirs is the process of reinfiltration. the effects of diffusion on overall recovery is
When drained oil from an upper matrix block en- probably very small and can for most systems be
ters into a matrix block underneath, the process is neglected for practical purposes.
called reinfiltration. Several publications17,18 have
shown that the reinfiltration is a function of both Matrix Block Shape and Size
 Combined Gas/Water Injection Subprogram

The shape and size of matrix blocks will strongly A major cause of the difficulties reproducing ex-
affect the matrix-fracture fluid exchange process. perimental results is probably that the capillary
Torsæter and Silseth28 conducted water imbibition pressure and relative permeability curves used in
experiments on chalk and sandstone cores of dif- the simulations were measured at flow conditions
ferent shapes and sizes, and typical results are different from those of the experiments. An im-
presented in Fig. 4. Obviously, within a gridblock portant issue to be addressed in fractured reservoir
of a dual-porosity simulation model, matrix simulations is the one of co-current vs. counter-
blocks of varying shape and size will exist.29 One current flow. As reported by Bourbiaux and
improvement that could be incorporated in such Kalaydjian,36 the experimental imbibition results
models, would be the inclusion of some form of are very much affected by the boundary condi-
distribution of size and shape, as indicated in Fig. tions imposed.
5. The computation of fluid exchange between
matrix and fracture in a matrix block would then Water Flooding Simulation
be the sum of computed exchange from a number
of different geometries. Studies
Matrix-fracture fluid exchange in a fractured res-
ervoir is controlled mainly by a combination of
capillary and gravity forces. The shape and size of
matrix blocks, and the inclination of fractures will
affect recovery of oil by water flooding. Several
simulation studies37-42 investigating the effects of
these factors have been conducted. In the follow-
ing, the most important results will be presented.

Most of the studies in the past dealing with oil

displacement by water in a fractured reservoir per-
tain to strong water-wet matrix where capillary
imbibition dominates the process of fluid transfer
Fig. 4 Effect of size and shape on imbibition oil
recovery (from Torsæter and Silseth28) between the matrix block and the fracture. Few
studies have been reported dealing with water
flooding in intermediate wet systems where both
capillary and gravity forces play key roles. In
certain situations, the role played by gravity forc-
es may be so dominant that the limited recovery
due to spontaneous imbibition in case of an in-
termediate wet rock is significantly improved.43

Block to block processes for a gas-oil system has

been intensively studied by many authors.15,22,44
Festøy and van Golf-Racht22 showed that for a
gas-oil system, matrix to matrix contact area in a
vertical stack of blocks significantly improves the
recovery. No such study for oil-water system has
been reported in the literature.
Fig. 5 Frequency distribution of representative
matrix blocks (from Torsæter et al.29) Pratap40 conducted fine grid simulations using a
similar geometry in order to understand in detail
Co-Current and Counter-Current Flow the block to block processes involved in oil dis-
Several authors have reported simulation studies placement by water in a fractured reservoir.
for the purpose of matching results of laboratory GeoQuest´s Eclipse 100 model was used for the
experiments.30-35 Such matching of experimental simulations. The study makes use of oil-water
results is not trivial, as reported by Kvalheim23 relative permeability and capillary pressures for
and by Beckner et al.24 The outcome of the simu- the Ekofisk Field, as reported by Thomas et al.45
lations is very sensitive to the shape and magni-
tude of the capillary pressure curve. Results of the study show that a significant frac-
tion of oil expelled from the down-dip matrix
block into the separating fracture reinfiltrates
 Dual Porosity, Dual Permeability Formulation for Fractured Reservoir Simulation

into the block above. This reentering of oil into

the matrix block above is a result of the change in
oil pressure in the separating fracture. In the early
phase of production, the oil pressure in the sepa-
rating fracture is less than that of the matrix grid
cells above as well as below. Thus, oil will flow
to the separating fracture from both the matrix
blocks. The oil discharged to the separating frac-
ture flows laterally to the vertical fracture. There
it joins the mainstream of oil produced by sponta-
neous imbibition of water from the lateral sides of
the blocks as they are surrounded by water in the
vertical fracture. After some time, the oil pressure
of the separating fracture exceeds the oil pressure
of the matrix grid cells of the up-dip block in con-
tact with the fracture. This results in significant Fig. 6 Effect of capillary continuity on oil recov-
fraction of oil in the separating fracture (produced ery for water wet (I) and mixed wet (II) systems
from lower block) to flow into the upper block. (from Pratap30)
This block to block process which starts at the
edge blocks marks the beginning of the late
phase. As time progresses, more and more inner
matrix grid blocks undergo the process of oil
redrainage. This late phase production continues
for a long time.

The study also examines the effect of matrix to

matrix contact on oil recovery from a vertical
stack of blocks. The result shows that in absence
of capillary continuity between the blocks, the oil
recovery is low. The presence of matrix to matrix
contact increases the ultimate recovery from the
stack dramatically. However, the oil recovery
rate is higher at early times if the contact area be-
tween the matrix blocks exceeds 5 %. With a ma-
trix to matrix contact of only 1 %, a long time is Fig 7 Effect of degree of capillary contact on oil
required to obtain the ultimate recovery. Some of recovery (from Pratap30)
these results are shown in Figs. 6-7. showed that with a contact of 25 % between the
matrix blocks, the time to obtain the ultimate re-
The position of the matrix to matrix contact with covery is close to that with 100 % contact, and
respect to the vertical axis of the cylindrical the present study for an oil-water system confirms
blocks also affects the recovery time. For exam- their findings. It also shows that 30% contact be-
ple, a 10 % contact at the edge of the matrix block tween matrix blocks is sufficiently effective to
gives faster ultimate recovery as compared to the drain the oil in a reasonable time.
same contact located at the center. The study on
a gas-oil system by Festøy and van Golf-Racht22 A comparison has been made with a case of
strong water wet matrix block (Fig. 7). The results
show that oil redrainage does not take place if the
rock is strongly water wet, as the oil phase pres-
sure of separating fracture never exceeds that of
the upper matrix block.

The process of oil redrainage from the lower ma-

trix block to the upper matrix block in case of oil-
water system has not previously been reported in
the literature. Neither is the finding that the posi-
tion of the contact influences the time to reach
ultimate recovery. The effect of vertical capillary
continuity and its effect on oil recovery for in-
 Combined Gas/Water Injection Subprogram

termediate wet rock has been extensively studied Physically, the mechanism for gas entering from
for the first time. the fracture to an undersaturated oil would require
an ultra-thin contact zone at the fracture/matrix
High Pressure Gas Injection interface. In this zone a small amount of equilibri-
um gas will exist, and diffusion transfer between
fracture and matrix can occur via this zone, as
Despite the efficiency of waterflooding in frac-
gas-gas diffusion between fracture gas and equi-
tured reservoirs, considerable oil will be left be-
librium gas in the two-phase zone, and as liquid-
hind due to relatively high residual saturations.
liquid diffusion between undersaturated matrix oil
This residual oil may be a target for high pressure
and saturated oil in contact zone.
gas injection.
Fig. 8 shows the system layout of an experiment
The recovery mechanisms involved in high pres-
performed at Reslab by Øyno and Whitson,46
sure gas injections in fractured reservoirs are
where methane was injected around cores filled
complex and not fully understood. They include
with a highly undersaturated oil. The cores were
viscous displacement, gas gravity drainage, diffu-
initially filled with Ekofisk oil, and methane gas
sion, swelling and vaporization/stripping of the
was then injected into the annulus system.
oil. Interfacial tension gradients caused by diffu-
sion may also play an important role on the over-
all recovery. Viscous displacement normally plays
a minor role, except perhaps in the near vicinity of
the wells where the pressure gradients are large.

Contrary to conventional reservoirs, diffusion

may play an important role in fractured reservoirs.
The injection gas has a tendency to flow in the
fractured system, resulting in relatively large
composition gradients between fracture gas and
matrix hydrocarbon fluids. Thus, there is a poten-
tial for transport by molecular diffusion. This is
especially the case in reservoirs with a high de-
gree of fracturing (small matrix block sizes). Dif-
fusion is difficult to model correctly in a simula-
tor. The diffusion flux for a given phase is often
modelled as

Jip = Td (φScp)Dip ΔCip,

where Td is the diffusion transmissibility, Dip is Fig. 8 Laboratory setup (from Øyno and Whit-
the effective diffusion coefficients, and the prod- son46)
uct φScp represents the fraction of the gridblock
interface where diffusion takes place. The satura- The experiment was simulated using a fully im-
tion term, Scp, is normally chosen as the minimum plicit compositional simulator (SSI's COMP4
saturation of the adjacent grid blocks. Using this model) that accounts for both molecular diffusion
formalism, a problem arises when gas is injected and dynamic IFT-scaled capillary pressure. The
in an undersaturated reservoir. The minimum con- effective matrix block height of the laboratory
tact saturation, Scp, between the fracture and ma- system was lower than the capillary entry pres-
trix grid block will be zero, and no diffusion be- sure, so without any modifications to the system
tween the two media will be calculated. With ma- set-up, no oil production could be observed in the
trix block heights lower than the capillary entry simulations. When a thin (1 mm) two phase zone
height, no mass transfer between fracture and ma- was introduced (Fig. 9), mass transfer by diffusion
trix systems will occur. In some simulators this could take place. Fig. 10 shows the experimental
problem is circumvented by choosing Scp as the results and the simulated best-fit using a thin two-
maximum saturation between neighbouring grid phase zone.
blocks or by introducing gas-liquid diffusion. This
does not give a physically correct description of a
diffusion process in general, and is therefore not
recommended.
 Dual Porosity, Dual Permeability Formulation for Fractured Reservoir Simulation

contact zone. During this stage there is some vis-

cous flow from the center of the core to the frac-
ture, due to swelling of the oil and interfacial ten-
sion gradients. The oil produced to the fractures is
vaporized by the injection gas, and no free oil is
observed.

The first stage ended when some of the oil within

the core first became saturated. This marks the
beginning of the second stage, where a free gas
saturation advances toward the center of the core.
As the gas front advances, the gas-gas diffusion
will play a more dominant role on the recovery
process. During this stage, mainly light-
intermediate, and the intermediate components of
the oil are vaporised.

When the light-intermediate and most of the in-

termediate components have been recovered, the
third stage begins. During the last stage, mostly
heavy-intermediate and heavy components are
vaporized. This stage is slow compared to the first
two stages, but a large additional recovery may be
achieved.
Fig. 9 Grid system used for modelling of labora-
tory experiment (from Øyno et al.49)
It is worth noticing that there was no significant
IFT reduction during production and therefore a

Fig. 10 Experimental and modelled oil recovery

vs. time (from Øyno et al.49)

The simulated results indicate that the recovery Fig. 11 IFT profile at core center as function of
can roughly be divided into three production stag- time - with diffusion (from Øyno et al.49)
es negligible production by gravity drainage. The
1. primary swelling of the oil IFT changes due to diffusion may in other cases
2. secondary swelling and vaporization have a drastic effect on the recovery.47-48 Fig. 11
3. final vaporization of the oil. shows the interfacial tension profile in a core after
The initial stage is dominated by swelling of the high pressure gas injection around a core filled
oil inside the core, due to liquid-liquid diffusion with recombined reservoir oil.49
between the undersaturated oil inside the core and
the saturated oil at the outer surface of the core. In this experiment the frontal IFT is reduced dras-
The light components of the oil (mostly methane), tically as gas advances down the core. Composi-
diffuse into the core, while the intermediate oil tional changes due to gas diffusion cause the IFT
components from the core diffuse to the outer behind the gas front to increase again. The result-
 Combined Gas/Water Injection Subprogram

ing IFT gradient is so strong that oil is sucked most 100 % recovery by gravity drainage. We feel
upward against gravity, resulting in the somewhat that the dynamic composition variations, and the
strange saturation profile given in Fig. 12. influence on IFT's, are important issues that need
to be addressed in more detail.
If diffusion and IFT scaling of capillary pressure
are not included in the simulator, these IFT effects
will not be accounted for. Fig. 13 shows the satu-
ration profile for the same experiment when diffu-

Fig. 14 IFT profile at core center as function of

time - no diffusion (from Øyno et al.49)

Fig. 12 Saturation profile at core center as func- Mathematical Model

tion of time - with diffusion (from Øyno et al.49)
Development
Current state of the art has been reviewed by
Fung.15 Current models are insufficient for proper
modelling of most fluid exchange processes be-
tween matrix blocks and fractures. In North Sea
fractured reservoirs, matrix blocks sizes are typi-
cally much less than 1 m3. Thus, a normal grid
block employed in reservoir simulations would
contain several tens of thousands of matrix
blocks. Obviously, some form of detailed descrip-
tion of the flow processes must be included in the
models.

The dual-porosity and the dual-porosity/dual-

permeability formulations are the most common
approaches used to represent a large number of
individual matrix blocks in larger computational
blocks. In these models, the processes of water/oil
Fig. 13 Saturation profile at core center as func-
imbibition and gas/oil drainage have caused par-
tion of time - no diffusion (from Øyno et al.49)
ticular difficulties. Attempts to represent correct
sion is not included. The IFT gradients are not so
behavior by modifications of the gravity term or
pronounced as in the diffusion case, and therefore
by use of capillary pressure pseudofunctions have
not strong enough for the oil to imbibe upwards.
generally not been successful. Others have refined
This results in a smaller oil saturation behind the
the matrix blocks into multiple blocks.
advancing gas-oil front, see Fig. 14.
Fig. 15 presents the multiple grid concept for im-
It was observed that the IFT at the gas front for
proving the fractured reservoir simulation. The
some cases (mainly depending upon initial gas-oil
upper grid system represents the coarse grid nor-
ratio), were almost vanishing. This resulted in al-
mally used in dual-porosity simulators. Inside
 Dual Porosity, Dual Permeability Formulation for Fractured Reservoir Simulation

each grid block, a large number of individual ma-

trix blocks exist. It is, of course, not possible to do
individual computation on each matrix block.
However, assuming that the large scale grid
blocks are chosen so that all the matrix blocks
inside exhibit similar behavior (for practical pur-
poses), we may do individual computation on one
representative matrix block inside each large grid
block, and multiply the results with the number of
grid blocks present. In principle, as previously
suggested, a distribution of different matrix block
geometries could be included in the matrix block
descriptions, so that individual computations are
conducted on a number of block groups. The ma-
trix block would be locally gridded in 1, 2 or 3
dimensions, depending on the situation, and the
behavior simulated using local time steps for fixed
boundary conditions during each large time step
of the coarse model. This system would be well
suited for parallel computing using massive paral-
lel computers.

Fig. 16 presents another improvement suggested

for improvement of the fluid flow representation Fig. 16 Improved Warren and Root model
in dual-porosity models. It is proposed that the
conventional Warren and Root model with discon- with adjacent blocks. This is of particular im-
tinuous matrix blocks is replaced by a model portance in the vertical direction.
where all matrix blocks have capillary contact
The method has so far been developed and tested
for single and two phase flow50-52 and is currently
being extended to a three phase model. The fine
grid calculations together with the iterations be-
tween the two grid systems require substantially
more computing time. However, the method is
well suited for parallelization, since each inner
block solutions can be computed separately once
the boundary conditions from the coarse grid are
given. So far, the computational speed has been
improved by a factor of 10 by vectorization and
parallelization (Fig. 17).

Fig. 15 Multiple grid concept for fractured reser-

voir simulation

Fig. 17 Computational speed improvements by

vectorization and parallelization.
 Combined Gas/Water Injection Subprogram

7. Thomas, L.K., Dixon, T.N., and Pierson, R.G.:

"Fractured Reservoir Simulation," SPEJ (Feb.
Conclusions 1983) 42-54.
8. Dean, R.H. and Lo, L.L.: "Simulations of Nat-
urally Fractured Reservoirs," SPERE (May
1. Based on a systematic study of physical mech-
1988) 638-48.
anisms and parameters affecting flow in frac-
9. Rossen, R.H. and Shen, E.I.: "Simulation of
tured reservoirs, it is concluded that all major
Gas/Oil Drainage and Water/Oil Imbibition in
flow mechanisms and flow processes must be
Naturally Fractured Reservoirs," SPERE (Nov.
incorporated in a model for fractured reser-
1989) 464-70; Trans., AIME, 287.
voirs. These include gravity, capillary forces,
10. Coats, K.H.: "Implicit Compositional Simula-
gravity drainage, diffusion, capillary continui-
tion of Single-Porosity and Dual-Porosity Res-
ty, and reinfiltration.
ervoirs," paper SPE 18427 presented at the
2. The key to improved modelling of fractured
1989 SPE Reservoir Simulation Symposium,
reservoirs is to include sufficient detail in the
Houston, Feb. 6-8.
calculation so that all these flow mechanism
11. Gilman, J.R.: "An efficient Finite-Difference
are represented. A multiple grid concept for
Method for Simulating Phase Segregation in
this purpose is recommended.
the Matrix Blocks in Double Porosity Reser-
3. Simulation studies of water flooding of frac-
voirs," SPERE (July 1986) 403-13.
tured reservoirs show that capillary continuity
12. Saidi, A.M.: "Simulation of Naturally Frac-
between matrix blocks is of high importance
tured Reservoirs," paper SPE 12270 presented
for mixed wetted systems.
at the 1983 SPE Reservoir Simulation Sympo-
4. High pressure gas injection may yield high oil
sium, San Francisco, Nov. 16-18.
recoveries due to reduction in interfacial ten-
13. Dangerfield, J.A. and Brown, D.A.: "The Eko-
sion caused by diffusion.
fisk Field," North Sea Oil and Gas Reservoirs
5. Literature review shows that current models
Seminar I, J. Kleppe et al. (eds.), Graham and
based on the dual-porosity, dual-permeability
Trotman, London (1986).
concept using simple fluid transfer terms be-
14. daSilva, F.: "Primary and Enhanced Recover-
tween fractures and matrix blocks, are not suf-
ies of the Ekofisk Field - A Single and Dual
ficiently accurate or representative in most
Porosity Numerical Simulation Study," paper
cases.
SPE 19840 presented at the 1989 SPE Annual
Technical Conference and Exhibition, San An-
References tonio, Oct. 8-11.
15. Fung, L.S.-K.: "Simulation Of Block-to-Block
1. Barrenblatt, G.D. et al.: "Basic Concepts in the Processes in Naturally Fractured Reservoirs,"
Theory of Homogeneous Liquids in Fissured SPERE (Nov. 1991) 477-484
Rocks," J. of Applied Math. (1960) 24, no. 5, 16. da Silva, F.V. and Meyer, B.: "Improved For-
1286-1303 (USSR). mulation for Gravity Segregation in Naturally
2. Warren, J.E. and Root, P.J.: "The Behavior of Fractured Reservoirs," paper presented at the
Naturally Fractured Reservoirs," SPEJ (Sept. 6th European IOR-Symposium in Stavanger,
1963) 245-255. May 21-23, 1991.
3. Kazemi, H., Merrill, L.S., Porterfield, K.L., 17. Firoozabadi, A. and Markeset, T.: "An Exper-
and Zeman, P.R.: "Numerical Simulation of imental Study of the Gas-Liquid Transmissibil-
Water-Oil Flow in Naturally Fractured Reser- ity in Fractured Porous Media," paper SPE
voirs," SPEJ (Dec. 1976) 317-326. 24919 presented at the 1992 SPE Annual
4. Sonier, F., Souillard, P., and Blaskovich, F. T.: Technical Conference and Exhibition, Wash-
"Numerical Simulation of Naturally Fractured ington DC, Oct. 4-7.
Reservoirs," SPERE (Nov. 1988) 1114-22. 18. Barkve, T., Firoozabadi, A.: "Analysis of Re-
5. Litvak, B. L.: "Simulation and Characteriza- infiltration in Fractured Porous Media," paper
tion of Naturally Fractured Reservoirs," Proc., SPE 24900 presented at the 1992 SPE Annual
Reservoir Characterization Technical Confer- Technical Conference and Exhibition, Wash-
ence, Dallas, Academic Press, New York, ington DC, Oct. 4-7.
(1985) 561-83. 19. Kazemi, H., Gilman, J.R. and Elsharkaway,
6. Gilman, J.R. and Kazemi, H.: "Improved Cal- A.M.: "Analytical and Numerical Solution of
culations for Viscous and Gravity Displace- Oil Recovery From Fractured Reservoirs With
ment in Matrix Blocks in Dual-Porosity Simu- Empirical Transfer Function," SPERE (May
lators," JPT (Jan. 1988) 60-70; Trans., AIME, 1992) 219-27.
285.
 Dual Porosity, Dual Permeability Formulation for Fractured Reservoir Simulation

20. Luan, Z.: "Oil Recovery Mechanisms in Natu- 34. Luan, Z.: "Discussion of Analytical and Nu-
rally Fractured Reservoirs," RUTH Project merical Solution of Waterflooding in Fractured
Report, NTH, Nov. 1992. Reservoirs With Emperical Transfer Func-
21. Uleberg, K.: "Improved Methods for Simula- tions," SPERE (July, 1992).
tion of Fractured Reservoirs," Sivilingeniør 35. Uleberg, K., Torsæter, O. and Kleppe, J.: "A
Thesis, NTH, May 1993. Review of Fluid Flow Mechanisms for Model-
22. Festøy, S. and van Golf-Racht, T.: "Gas Gravi- ling of North Sea Fractured Reservoirs," Proc.,
ty Drainage in Fractured Reservoirs Through a 1993 RUTH Seminar, Stavanger, Oct. 13-14.
New Dual-Continuum Approach," SPERE 36. Bourbiaux, B.J. and Kalaydjian, F.J.: "Experi-
(Aug. 1989) 271-278. mental Study of Cocurrent and Countercurrent
23. Suffridge, F.E. and Renner, T.A.: "Diffusion Flows in Natural Porous Media," SPERE (Aug.
and Gravity Drainage Tests to Support the De- 1990) 361-368.
velopment of a Dual Porosity Simulator," pa- 37. Kløv, T.: "Et studium av fluidstrøm mellom
per presented at the 6th European IOR- matriseblokk og sprekk i et oppsprukket reser-
Symposium in Stavanger, May 21-23, 1991. voar," Project Report, NTH, June 1994
24. Catalan, L. and Dullien, F.A.L.:"Applications 38. Skaar, I.: "Toporøsitetssimulering," Project
of Mixed-Wet Pastes in Gravity Drainage Ex- Report, NTH, June 1994.
periments," JCPT (May 1992) 28-30. 39. Kløv, T.: "Et studium av fysikalske prosesser i
25. Orr, F.M. Jr.: "Use of Gas for Improved Oil naturlig oppsprukkete reservoarer ved bruk av
Recovery," presentation at the 1992 RUTH- simuleringsmodeller," Sivilingeniør Thesis,
seminar, Stavanger, Dec. 7-8. NTH, Dec. 1994.
26. Schechter, D.S., Dengen, Z., and Orr, F.M. Jr.: 40. Pratap, M.: "A study of the effect of vertical
"Capillary Imbibition and Gravity Segregation capillary contact between matrix blocks in a
in Low-IFT Systems" paper SPE 22594 pre- fractured reservoir under water injection,"
sented at 1991 SPE Annual Technical Confer- [Link]. Thesis, NTH, Dec. 1994.
ence and Exhibition, Dallas, Oct. 6-9. 41. Ese, A. M.: "Effect of gravity and capillary
27. Hu, H., Whitson, C.H. and Qi, Y.: "A Study of pressure on recovery of oil in intermediate wet
Recovery Mechanisms in a Nitrogen Diffusion reservoirs," Project Report, NTH, June 1995.
Experiment," paper presented at the 6th Euro- 42. Fougner, C.: Detailed simulations of matrix-
pean Symposium on Improved Oil Recovery, fracture fluid exchange in fractured reservoirs
Stavanger, May 21-23, 1991. under various processes," Project Report,
28. Torsæter, O. and Silseth, J.: "The Effects of NTH, June 1995.
Sample Shape and Boundary Conditions on 43. Kyte, J. R.: "A Centrifuge Method to Predict
Capillary Imbibition" paper presented at the Matrix-Block Recovery in Fractured Reser-
1985 Chalk Symposium, Stavanger, May 21- voirs," SPEJ (June 1970) 164-170
22. 44. Saidi, A. M., Tehrani, D. H., and Wit, K.:
29. Torsæter, O., Kleppe, J. and van Golf-Racht, "Mathematical Simulation of Fractured Reser-
T.: "Multi-Phase Flow in Fractured Reser- voir Performance Based on Physical Model
voirs", in Fundamentals of Transport Pheno- Experiments," paper PD10(3), Proc., 10th
mena in Porous Media, J. Bear and M.Y. World Petroleum Congress, Bucharest (1979)
Corapcioglu (eds.), NATO ASI Series, Nijhoff 45. Thomas, L. K., Thomas, N. D., and Pierson, R.
Publishers, Dordrecht (1986) G.: "Fractured Reservoir Simulation," SPEJ
30. Kleppe, J. and Morse, R.A.: "Oil Production (Feb. 1983) 42-54
From fractured Reservoirs by Water Dis- 46. Øyno L. and Whitson C. H.: "Experimental
placement" paper SPE 5084 presented at 1974 Study of Gas Injection for Secondary and Ter-
SPE Annual Technical Conference and Exhibi- tiary Recovery in Fractured Chalk Reservoirs,"
tion, Houston, Oct. 6-9. Proc., 1993 RUTH Seminar, Stavanger, Oct.
31. Kazemi, H. and Merill, L.S.: "Numerical Sim- 13-14.
ulation of Water Imbibition in Fractured 47. Morel, D.D., Bourbiaux, B., Latil, M., and
Cores" SPEJ (June 1979) 175-182. Thiebot, B.: "Diffusion Effects in Gas Flooded
32. Kvalheim, B.: "Kapillær oppsuging i kalkstein: Light Oil Fractured Reservoirs," paper SPE
Numerisk simulering av laboratorieforsøk," 20516 presented at the 1990 SPE Annual
Sivilingeniør Thesis, NTH, Dec. 1984. Technical Conference and Exhibition, New
33. Beckner, B.J. et al.: "Imbibition-Dominated Orleans, Sept. 23-26.
Matrix-Fracture Fluid Transfer in Dual Porosi- 48. Hu, H., Whitson, C.H., and Yuanchang, Q.: "A
ty Simulators" paper SPE 16981 presented at Study of Recovery Mechanisms in a Nitrogen
1987 SPE Annual Technical Conference and Diffusion Experiment," Proc., Sixth European
Exhibition, Dallas, Sept. 27-30.
 Combined Gas/Water Injection Subprogram

Symposium on Improved Oil Recovery, Sta-

vanger, May 21-23 (1991) 543-553.
49. Øyno L., Uleberg K and Whitson C.H.: "Dry
Gas Injection in Fractured Chalk Reservoirs -
An Experimental Approach," paper presented
at the 1995 International Symposium of the
Society of Core Analysts, San Fransisco, Sept.
12-14.
50. Gimmestad, O.: "Beregning av strømning i to-
porøsitets reservoarsystemer," Project Report,
NTH, June 1994.
51. Gimmestad, O. and Uleberg, K.: "The Pro-
gramming of Fracsim - A Fast Fractured Res-
ervoir Simulator," Project Report, NTH, Aug.
1994.
52. Gimmestad, O.: "Utvikling av effektive rutiner
for lokal beregning av strømning i
oppsprukkete reservoarer," Sivilingeniør The-
sis, NTH, Dec. 1994.