SPE 81499
Numerical Simulation of Non-Darcy Flow Utilizing the New Forchheimer’s
Diffusivity Equation
H.A. Belhaj, K.R. Agha, A.M. Nouri, S.D. Butt, H.F. Vaziri*, SPE and M.R. Islam, SPE, Dalhousie University
*BP America
Copyright 2003, Society of Petroleum Engineers Inc.
This paper was prepared for presentation at the SPE 13th Middle East Oil Show & Conference
to be held in Bahrain 5-8 April 2003.
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
Recently, it became evident that non-Darcy flow occurs not
only in gas reservoirs, fractured reservoirs and multi-
permeability systems within oil reservoirs experienced non-
linearity due to non-Darcy flow behavior. Most reservoir
simulators currently used encountered false predictions due to
their dependency on the traditionally used diffusivity equation.
This paper introduces alternative diffusivity equation to
replace the one derived from Darcy’s law. The new equation
was derived from the commonly known as Forchheimer’s
equation which is basically Darcy’s equation plus an inertia
term to account for high velocity fluid flow in porous medium.
Mathematical derivation of the diffusivity equation based
on Forchheimer equation has been presented in a previous
paper by the authors.
The newly derived diffusivity equation has been
numerically simulated. Correlations used to estimate the non-
Darcy coefficient “ β ” have been comprehensively reviewed;
nine correlations found suitable for use in this study for
technical reasons. A new dimensionless number (Be) relating
β , velocity, density and viscosity has been introduced to
differentiate between Darcy and non-Darcy flow in porous
medium for any rock type and any flowing fluid. Evidences
show that this new dimensionless number cannot be
considered a declaration of turbulence flow in porous medium
rather the energy loss is contributed to the nature of both
flowing fluid and the porous medium. The point of deviation
from the Darcian behavior to the non-Darcian behavior has
been found at Be = 0, for practical use it has been determined
that Be = 0.0526 at 5% deviation from Darcy’s linear trend.
A range of permeability from 1 md to 1000 md with
porosity changing accordingly has been verified with the new
model, velocity as low as 0.0001 cm/sec and as high as 700
cc/sec has been tested as well. Both Darcy and non-Darcy
behaviors have been identified for the domain of testing, and
the numerical model has proven of good agreement in
all cases.
Intoduction
Traditionally, when mentioned, non-Darcy behavior means
gas flow in reservoirs and the attention goes to the
Forchheimer’s equation. On the other hand, despite that Darcy
equation has been around for about one and a half century and
essentially based on empirical experimental approach with no
mathematical foundation, until today, it is considered the
corner stone of fluid flow in porous medium.
The uncertainties usually associated with the predictions of
commercial reservoir simulators mostly attributed to a number
of factors such as poor quality and lack of in-put data,
unrealistic history matching, assumptions encountered, etc. It
is only lately that the basic governing equations of flow
behavior become questioned. In fact, transmissibility and so
saturations distribution within a reservoir at any time and
coordinates are direct functions of pressure predicted by the
diffusivity equation employed. As it controls the evaluation of
other parameters, if this pressure is in doubt then the ultimate
simulator predictions are consequently misleading.
Non-Darcy flow has been encountered in oil and gas
reservoirs produced by vertical and horizontal wells alike. The
severity of the non-Darcy effect is indeed affected by the
velocity of the flow and both physical properties of the
flowing fluid and the characteristics of the medium.
The productivity of a horizontal well has been affected by
the non-Darcy behavior around the well bore. To include this
high flow rate effect on the productivity model of a horizontal
well the flow in the near wellbore has been assumed normal to
the well trajectory and a radial flow region is supposed to take
place around the wellbore. The impact of high flow rate region
is predicted to occur near the wellbore and as it gets away
from the wellbore in the flow field the effect vanishes. It has
been found that non-Darcy effect, if occurred, in a horizontal
well under radial or pseudo radial flow cases become
significant in a range of 10 to 20 feet around the wellbore,
beyond that the effect become insignificant and practically
fades out 1.
The relationship between pressure drop and flow rates can
be complicated in producing gas condensate reservoirs more
predominantly near the wellbore. The main reason of this
2 SPE 81499
relationship complexity is the very high gas velocity caused by
the inertial effect “non-Darcy flow” and the increase in
mobility at high capillary number referred to as viscous
stripping or positive coupling 2.
Simulators experienced problems when trying to history
matching. A satisfactory matching has not been achieved
without the inclusion of high-velocity effect (non-Darcy
effect) in the model used.
This paper discusses a numerical model for simulating
non-Darcy flow behavior based on a diffusivity equation
derived from the Forchheimer’s3 formula. Detailed derivation
of the new diffusivity equation has been presented in a
previous publication by the authors4. The derived model
includes a velocity term that expresses the non- linearity
caused by deviation from the non-Darcian behavior. The
Darcian model does not include any velocity terms.
The Numerical Model
The non-Darcy diffusivity equation in two dimensions that has
been derived4 from Forchheimer’s equation is used for
numerical simulation of both Darcy and non-Darcy flow in
porous medium:
∂ 2P ∂ 2P µ  ∂ P  ∂ P ∂ P   (1)
+ = c  + 2 βρ v   φ + v +  
∂x 2
∂y 2
K  ∂ t  ∂x ∂y  
Where: K is the permeability, φ is the porosity and β is the
non-Darcy coefficient of the porous medium. ρ is the density
and µ is the viscosity of the flowing fluid. c is the
isothermal compressibility factor and v is the velocity.
This partial differential equation would represent the non-
Darcy flow at high fluid velocity in porous medium and if the
velocity of the fluid is low enough the velocity terms become
negligible and the equation is reduced to the traditional
Darcian diffusivity equation:
∂ 2P ∂ 2P φµc ∂P (2)
+ =
∂x 2 ∂y 2 K ∂t
The original Forchheimer equation includes both Darcy’s
term and the inertial term introduced by Forchheimer.
Partial derivatives in eq. 1 and 2 have been transferred into
finite differences in order to be modeled numerically. Crank-
Necholson 5,6 and Barakat-Clark 7 methods have been used for
numerical modeling processing. Both methods have shown
converging and stability of solution at a very acceptable
accuracy4. Crank-Nicholson solution was adopted for further
parametric analysis.
Results and Discussions
Recently, it has become evident that turbulence does not play
any significant role in fluid flow behavior in porous media
rather the deviation from Darcy’s law is attributed to2: one the
convective acceleration and deceleration (to move at lower
speed) of fluid particles as they travel through the pores and
two the displacement of fluid particles from a straight line due
to the tortuous path in the pores system. Bear2 excluded
turbulence as the cause for non-Darcy effect. The main
reasons for that is in turbulent flow through pipes, the linear
term (Darcy’s term) in Forchheimer equation does not exist,
the transition from laminar to turbulent flow in pipes is not
gradual but rather a sharp change and the critical Reynolds
number at which the transition begins is much higher than that
at which the non-Darcy effect begins.
Numerous correlations for estimating the non-Darcy
coefficient β are available in the literature8. Only the single
phase flow correlations and particularly those include either
permeability or both permeability and porosity have been
selected in this study. The reason for that is in this study only
single phase flow is considered and the correlations involve
the tortuosity been excluded since determining tortuosity with
a reasonable accuracy is not an easy task. It is advised though,
that the non-Darcy coefficient is to be determined for each
case depending on the relevant available data for that
particular case. Once permeability and porosity determined
from laboratory or field testing, pressure gradient and flow
rate (velocity) data should be utilized to determine β
corresponding to that permeability and porosity values using
Forchheimer equation.
Table (1) summarizes the correlations used to estimate β
for single phase flow. The applications, limitations and
sources of these correlations are also mentioned in Table (1).
The proposed numerical model has been verified using these
nine correlations.
The behavior of pressure drop with velocity (or flow rate)
change as predicted by the numerical model using different
non-Darcy coefficient correlations is described in Figs. 1-3. In
general, it can be seen that at low velocity both Darcy’s and
Forchheimer’s models seem to predict identical flow behavior;
the trend of velocity versus pressure gradient in both cases is
almost the same at low velocity region. As velocity increases
the Forchheimer’s model predicts higher pressure gradient
than Darcy’s model and higher velocity would make the gap
between the two trends even larger. Darcy’s pressure gradient
trend increases linearly (straight line) with velocity increase,
this is understandable since inertia term does not exist in
Darcy’s model. The inertia term in Forchheimer’s model
(includes v 2 ) has a very small value “negligible” at low
velocity, but at high velocity the term become significant
because of “ v 2 ” effect.
It is evident that the non-Darcy coefficient is a very critical
factor in the Forchheimer’s model, this is obvious from the
plots in Figs. 1-3 since each of the nine correlations used to
estimate β gives completely different pressure gradient
predictions with velocity variation. Accurate determination of
β is recommended and if β has to be estimated from
correlations, a careful selection of the right correlation is
essential. The nine correlations used to estimate β mentioned
in Table (1) has been divided to three groups according to
their predictions when used in Forchheimer’s model, β1 takes
distinctive trend, β2 to β 7 take another different group-trend
and β8 and β9 take a third distinctive behavior as been
demonstrated in Figs. 1-3.
SPE 81499
The point at which the pressure gradient deviated from the
Darcian behavior as predicted by the proposed model is
considered the start of the non-Darcy behavior. This point is of
great importance since it declares the beginning of the non-
Darcy regime; this point has been given careful investigation
in this study. It has been found that the point of deviation from
the Darcian trend is function of fluid properties, porous
medium properties and flow characteristics. Specifically, a
dimensionless group “Be” consists of viscosity, density,
permeability, non-Darcy coefficient and velocity terms has
been identified and found to be a decisive element to
distinguish the Non-Darcian behavior. This dimensionless
group “Be” takes the following form:
ρβ Kv (3)
Be =
µ This paper takes a numerical approach based on a
The quantity “Be” is a unique and always equal constant value
at the point of deviation from the Darcian flow to the non-
Darcian flow no matter what fluid or porous media type is
used. Theoretically, it has been considered that “Be” equals
zero at the point of deviation. “Be” found to be constant at this
point. Fig. 4 indicates a relationship between “Be” and the
ratio of pressure drop predicted by Darcian model and
pressure drop predicted by Forchheimer’s model. It is clear
that there is a unique value of “Be” at each pressure drop ratio.
For practical reasons, “Be” has been taken at 5% deviation
from the Darcian trend (0.95 pressure drop ratio) and found to
be equal 0.0526.
Appendix “A” describes a mathematical proof shows that
“Be” is a constant value at the point of deviation from
Darcian behavior.
Parametric Study. To verify the proposed numerical model
three type fluids of different properties have been used. Water
as it has unique properties; methanol that is less viscous than
water and mineral oil that is more viscous than water have
been used to predict velocity using the numerical model. Fig.
5 shows velocity predictions of the model using water,
methanol and mineral oil at 5% deviation from the Darcian
trend. Velocity distribution takes the same trend described
earlier, when using the correlation β1 the predicted velocity
become too small to be noticed on the plot in Fig. 5, when
using correlations β2- β7 velocities are still small but it can be
seen on the plot of Fig.5. Velocity is slightly increasing as we
go from β2 to β7. When correlations β8 and β9 are applied
another significantly higher velocity predictions is identified.
Figs. 6-11 show the effect of both viscosity and density on
velocity of the flowing fluid. The scheme of dividing
predicted behaviors according to the β correlation mentioned
before is still valid. Water has been taken as a reference with a
density of 1 gm/cm3 and viscosity of 1 cp (gm/cm-sec), the
domain of both density and viscosity changes is from 50% to
200% of the parameter as shown in Figs. 6-11. It can be
inferred that density and viscosity are in opposite effect on the
predicted velocity, velocity would be reduced with increasing
density and/or increasing viscosity. Although, their behavior
has been categorized to three groups, density and viscosity
3
using β2- β3 correlations have a very similar behavior, β5- β6
correlations have another similar behavior and β8- β9
correlations take a third similar trend.
It should be noted that changing permeability and porosity
accordingly has been tested to check the validity of the model,
permeability values as low as 1 md and as high as 1000 md
has been tested. Porosity has been evaluated from permeability
using existed correlations9 for different oil fields. Although,
using different correlations for β estimate would give different
values of β, that by itself can be considered
permeability/porosity validation since the coefficient β
includes both permeability and porosity.
Conclusions
mathematical sound derivation of a new diffusivity model
from Forchheimer’s equation and using empirical correlations
inferred from both experimental and field dada. The following
conclusions can be summarized:
1. A new two dimensional numerical simulation model based
on the diffusivity equation derived from Forchheimer
equation has been introduced in this paper. This model
may describe both Darcian and non-Darcian behaviors. It
includes the viscous term of Darcy’s equation and the
inertial term of Forchheimer’s equation, at low velocity the
inertia term drops out and the model reduces to Darcian
behavior and as velocity increases this term become
significant.
2. A new dimensionless group term “Be” consists of
viscosity, density, velocity, permeability and the non-
Darcy coefficient has been introduced to verify the non-
Darcy behavior. This dimensionless term is equals to zero
at the point of deviation from Darcian flow but for
practical aspects it can be considered as 0.0526 at which
the deviation is 5% from the Darcian behavior.
3. A parametric study has been carried out to verify the
model. The effect of density, viscosity, permeability and
the non-Darcy coefficient estimated from available
empirical correlations have been studied and the model
found valid and gave realistic results.
4. It is recommended that the traditional diffusivity equation
derived from Darcy’s law that is currently used for
pressure predictions in reservoir simulators should be
replaced by the new diffusivity equation presented in this
paper. This equation has been numerically modeled and
validated. It gives accurate results and accounts for the
extra pressure drop caused by high flow rate encountered
usually around the wellbore or when fractured reservoir
is involved.
Acknowledgment
The authors are thankful to the Killam trustees, the Natural
Science and Engineering Research Council of Canada and
Atlantic Innovation Fund for the financial support of this
research program.
4 SPE 81499

Nomenclature 13. Li, D, Svec, R.K., Engler, T.W. and Grigg, R.B.:
“Modeling and Simulation of the Wafer Non-Darcy Flow
Be Dimensionless group Experiments,” paper SPE 68822, Proc., the 2001 SPE
c Isothermal compressibility factor Western Regional Meeting, Bakersfield, CA, March 26-30.
φ Porosity
SI Metric Conversion Factors
k Permeability of the porous system To Convert from To Multiply by
P Pressure Ft m 0.3048
x Space Coordinate in the flow direction md m2 9.869E-16
y Space Coordinate perpendicular to the flow direction psi kPa 6.894757
β Non-Darcy coefficient lbm/ft3 kg/m3 16.01845
t Time lbm/ft-sec kg/m-sec 1.488163
v Volume-averaged fluid velocity
µ Viscosity of flowing fluid
ρ Appendix A: Mathematical Formulation of the
Fluid density
Dimensionless Group “Be”
References
1. Tang, Y. and Kelkar, M.: “Effects of Formation Damage Pressure gradient according to Darcy’s equation:
and High-Velocity Flow on the Productivity of Perforated
Horizontal Wells,” paper SPE 77534 presented at the SPE ∂P µ
= u
Annual Technical Conference and Exhibition, San ∂x K
Antonio, Texas, 29 Septemebr-2 October 2002
2. Bear, J.: Dynamics of fluids in porous media, Dover and the pressure gradient according to Fochheimer’s equation:
Publications, Inc., New York (1974).
3. Forchheimer, P.: “Wasserbewewegung Durch Boden,”
∂P µ
ZVDI (1901), 45, 1781. = u + ρβ u 2
4. Belhaj, H.A., Agha, K.R, Nouri, A.M., Butt, S.D., Islam, ∂x K
M.R. and Vaziri, H.F.: “Numerical Modeling of
Forchheimer’s Equation to Describe Darcy and Non-Darcy then the dimensioless group were defined as:
Flow in Porous Medium System,” paper SPE 80440 proc.,
the SPE Asia Pacific Oil and Gas Conference and Forcheimer Term
Exhibition, Jakarta, Indonesia, April 15-17, 2003. Be =
5. Patankar, S.V.: Numerical Heat Transfer and Fluid Flow, Darcy Term
McGraw-Hill, (1980).
6. Kobayashi, M.H. and Pereira, J.C.F.: “A Comparison of
Second Order Convection Discretization Schemes for ( ρβ u 2 )
Be =
Incompressible Fluid Flow,” Communication in Numerical µ
Methods, 12, (7), 395-411, July 1996.
( u)
K
7. Barakat, H.Z and Clark, J.A.: “On the Solution of the
Diffusion Equation by Numerical Methods,” Journal of
ρβ Ku
Heat Transfer, (November 1966), 421-427. Be =
8. Li, D and Engler, T.W.: “Literature review on correlations µ
of the non-Darcy coefficient,” paper SPE 70015 presented
at the SPE Permian Basin Oil and Gas Recovery The relationship between the dimensionless group “Be” and
Conference, Midland, Texas, 15-16 May 2001. the pressure drop ratio (pressure drop caused by the viscous
9. Timur, A.: “An Investigation of Permeability, Porosity, Darcy term to that caused by Fochheimer’s inertial term), is as
and Residual Water Saturation Relationships,” Proc., follows;
SPWLA Ninth Annual Logging Symposium, New Orleans ∂P 
(June 23-26, 1986) Paper K. 
10. Pascal, H., Quillian, R.G. and Kingston, J.: “Analysis of ∂ x  Darcy 's equation
Pressure Drop Ratio =
Vertical Fracture Length and Non-Darcy Flow Coefficient ∂P 
Using Variable Rate test,” paper SPE 9438 presented at the 
1980 SPE Annual Technical Conference and Exhibition, ∂ x  Forcheimer 's equation
Dallas, September 21-24. µ
11. Coles, M.E. and Hartman, K.J.: “Non-Darcy u
Measurements in Dry Core and the Effect of Immobile = K
Liquid” paper SPE 39977 presented at the 1998 SPE Gas µ
u + ρβ u 2
Technology Symposium, Calgary, Alberta, Canada, K
March 15-18.
12. Jones, S.C.: “Using the Inertial Coefficient, β , to 1
Pressure Drop Ratio =
Characterize Heterogeneity in Reservoir Rock,” 1 + Be
paper SPE 16949 presented at the SPE Annual
Technical Conference and Exhibition, New Orleans, And therefore “Be” will be a constant value for a particular
October 5-8. pressure drop ratio.
SPE 81499 5

Table. 1 “K-φ” Non-Darcy Coefficient Correlations* available in the literature.

No. β Correlation. Application Reference

-Field testing (variable rate test)
4.8x1010
1 β= -Low permeability medium 10
K 1.176 -Hydraulically fractured medium

3.51 x 1010 φ 0.449 -Collective data

2 β= 11
1.88 -Sandstone and Limestone
K
1.82 x10 8 8
3 β= 5 3 -Natural porous medium
K 4φ 4

2.018 x 10 9 -Experimental data of Sandstone

4 β= 1.55 and Limestone
12
K
8.17 x 10 9 φ 0.537 -Collective data
5 β= -Sandstone and Limestone
11
K 1.79
1.15 × 10 7 -N2 gas
6 β= -Berea Sandstone
13
Kφ
1.59 × 10 3 -Consolidated and
7 β= unconsolidated sandstone, 8
K 0.5 φ 5.5 limestone and dolomite

1 245 *10 8 -Different mediums 8

8 β=
φ 12 K φ -Gases (CO2, N2, Methane, H2)

1 1.8 *10 9 -Different mediums 8

9 β=
φ K φ -Gases (CO2, N2, Methane, H2)

* K in md and β in 1/cm
6 SPE 81499

2.00E+03

1.80E+03
β1
1.60E+03

1.40E+03
Forchheimer
1.20E+03
∆P, Mpa/m

1.00E+03

8.00E+02

6.00E+02

4.00E+02

2.00E+02 Darcy

0.00E+00
0 0.2 0.4 0.6 0.8 1
Velocity, cm/sec

Fig. 1 Pressure Gradient Predicted vs. Flow Velocity for the case of β1 using both models

1.00E+04

β8 β7

7.50E+03 β6

β5
∆P, Mpa/m

5.00E+03 Forchheimer

β4
β3
2.50E+03
Darcy

0.00E+00
0 5 10 15 20
Velocity, cm/sec

Fig. 2 Pressure Gradient Predicted vs. Flow Velocity for the cases of β2 to β7 using both models
SPE 81499 7

3.00E+05

2.50E+05 β9

β8
2.00E+05
∆P, Mpa/m

Forchheimer
1.50E+05

1.00E+05
Darcy

5.00E+04

0.00E+00
0 100 200 300 400 500 600 700

Velocity, cm/sec

Fig. 3 Pressure Gradient Predicted vs. Flow Velocity for the cases of β8 and β9 using both models

1 16
Mineral Oil Water Methanol
14
0.8
Pressure Drop Ratio

12
Velocity, cm/sec

0.6
10
0.4 8

0.2 6

4
0
0 2 4 6 8 10 12 14 16 18 20 2
Be
Rek
0
Fig. 4 Pressure drop ratio (Darcy to Forchheimer prediction)
as a function of dimensionless number “Be”. β1 β2 β3 β4 β5 β6 β7 β8 β9
Fig. 5 Predicted Flow Velocity at 0.95 Pressure Drop
Ratio using different β correlations.
8 SPE 81499

0.006

0.005

0.004
V, cm/sec

0.003

0.002

0.001

0
0.5 0.75 1 1.25 1.5 1.75 2
3
ρ, gr/cm

Fig. 6 Effect of Fluid Density on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the case of β1)

0.006

0.005

0.004
V, cm/sec

0.003

0.002

0.001

0
0.5 0.75 1 1.25 1.5 1.75 2

2
µ, gr/cm-sec * 10

Fig. 7 Effect of Fluid Viscosity on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the case of β1)
SPE 81499 9

2.5

β2
2

β3

1.5
V, cm/sec

β4
1

β5

0.5 β6

β7

0
0.5 0.75 1 1.25 1.5 1.75 2
3
ρ, gr/cm

Fig. 8 Effect of Fluid Density on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the
cases of β2 to β7).

2.5
β2
β3

1.5
V, cm/sec

β4
1

β5
β6
0.5
β7

0
0.5 0.75 1 1.25 1.5 1.75 2

2
µ, gr/cm-sec * 10

Fig. 9 Effect of Fluid Viscosity on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the
cases of β2 to β7)
10 SPE 81499

30

β8
25

β9

20
V, cm/sec

15

10

0
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
3
ρ, gr/cm

Fig. 10 Effect of Fluid Density on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the
cases of β8 and β9)

30

β8

25
β9

20
V, cm/sec

15

10

0
0.5 0.7 0.9 1.1 1.3 1.5 1.7 1.9
2
µ, gr/cm-sec * 10

Fig. 11 Effect of Fluid Viscosity on Flow Velocity at the 0.95 Pressure Drop Ratio Point of Deviation (for the
cases of β8 and β9)