Anda di halaman 1dari 16

Computers and Chemical Engineering 27 (2003) 385 /400

Monte Carlo simulation of flow of fluids through porous media

Semant Jain, Madhav Acharya 1, Sandeep Gupta, Ashok N. Bhaskarwar *
Department of Chemical Engineering, Indian Institute of Technology, Hauz Khas, New Delhi 110016, India

Received 30 July 1999; received in revised form 26 August 2002; accepted 17 September 2002


This simulation employs Monte Carlo technique for studying fluid flow through a porous medium in the capillary regime. The
medium has been modelled as a 2 or 3-dimensional network of elements, some of which are randomly closed to the fluid flow.
Dijkstra’s algorithm has been employed to identify the least-resistance pathway, which is instrumental in determining the minimum
pressure required to achieve break-through across the network. At higher pressures, network resistance has been calculated by
determining the manner in which the cluster forms and by accounting for the nature of flowpaths. The simulation yields a linear
relationship between the pressure applied across the network and flowrate showing similarity to Darcy’s law. Polynominal fitting of
the data on the fraction of openable pores open as dependent on pressure applied across the network has been carried out and the
coefficients determined.
# 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Monte Carlo; Simulation; Porous media; Fluid flow; Darcy’s law; Scaling

1. Introduction The term ‘porous media’ encompasses a wide variety

of contacting devices such as packed towers, sand beds
Monte Carlo technique is a method of computer and substances like limestone rock, filter paper and
simulation of a system with many degrees of freedom. It catalytic particles. It is desirable to classify the porous
makes use of random numbers to numerically generate media according to the types of pore spaces they
probability distributions, which might otherwise not be contain. A proposed classification was by dividing the
explicitly known since the considered systems are so pore spaces into voids, capillaries and force spaces
complex (Binder, 1979). Monte Carlo simulation pro- (Manegold, 1937). Void spaces are characterized by
vides a good comparison between data from experi- the fact that walls have little or no effect on hydro-
ments on real systems to those from the model. It is used dynamic properties in the interior; in capillaries, the
in areas like simulation of thermodynamic properties of walls do affect the hydrodynamics but do not bring the
fluids, crystal growth, combustion of coal particles etc. molecular structure of the fluid into evidence; and in
Its classical application includes evaluation of multiple force spaces, the molecular structure of the fluid is of
integrals in statistical mechanics. considerable importance. This work concentrates on
pores of the size of capillaries.
Flow through a porous medium requires a description
of both the medium and the flow. A porous medium can
Abbreviations: APO, actual number of pores that have opened up;
be represented as an extremely complicated network of
FOPO, fraction of openable pores open.
* Corresponding author. Tel.: /91-11-6591028; fax: /91-11-
channels, including those containing obstructions and
6581120. dead ends too (Bernsdorf, Brenner & Durst, 2000). The
E-mail addresses: (S. Jain), distribution of channels is obtained from assumed (M. Acharya), statistical descriptions. The pores in the network can (A.N. Bhaskarwar).
Current address: Catalyst Technology Laboratory, ExxonMobil
be interconnected or non-connected, depending on
Refining and Supply Company, Process Research Laboratories, 1545 whether they are a part of a continuous network of
Route 22 East, Annadale, NJ 08801, USA. pores that exists within the medium or not. Put another
0098-1354/03/$ - see front matter # 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 9 8 - 1 3 5 4 ( 0 2 ) 0 0 2 1 1 - 9
386 S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Small letters
fb fraction of pores that have been blocked
n number of flowpaths present
r capillary radius
sg size of the grid along one dimension or side
Capital letters
Pa applied pressure (on inlet face of network)
Pc capillary pressure
Pcr break-through pressure
Pmax maximum pressure
Q total flowrate
Qi flowrate of ith flowpath
Qmax maximum flowrate through the network
Rc radius of curvature of meniscus
Ri resistance of i th flowpath
Greek symbols
s surface tension
f contact angle

may also govern the process, as in the case of water

way, any channel classified as ‘interconnected’ will
flowing through limestone under gravity.
ultimately be filled by fluid flowing through the
According to the nature of the problem, the random
medium, while ‘non-interconnected’ elements will re-
mechanism can be attributed to the fluid or the medium.
main devoid of any fluid flow. The fraction of inter-
The former falls under the category of diffusion
connected channels gives an indication of the accessible
processes. A typical example is the motion of one
porosity of the medium.
To describe the flow through a porous medium, we molecule in a gas as it undergoes collisions with other
molecules. In case of a dilute gas, each collision event is
also need to specify two parameters */applied pressure
totally random as it is not influenced by other collisions
across the network and the flowrate (i.e. the net amount
that have occurred in the past. In other words, the
of fluid passing through the network per unit time).
medium has no ‘memory’ of its past history. Also, the
We have divided the paper in six main sections. The
medium (which is essentially the molecules), is continu-
theory section has an overview of percolation theory,
ously varying after each collision and so is not invariant
capillary pressure and pore structure models. The
in time.
section following that describes the approach we used
to develop the simulation code. It covers the algorithm The other, relatively less common, is known as
percolation. In percolation processes (such as a fluid
used to identify the least resistant pathway, how the
soaking into a porous medium), there is a distinction
status of a pore changes from ‘closed’ initially to ‘open’
between the fluid particles and the scattering medium.
to finally ‘part of a flowpath’, the manner in which
This medium, although it varies in random fashion from
subsequent flowpaths are identified and finally the
point to point, is invariant in time. Thus ‘memory’
equations used to compute the flowrate through the
effects cannot be neglected as in diffusion, and the
medium. After briefly mentioning our constraints, we
random scattering of the particles of the fluid must be
describe our results and finally present the conclusions.
treated as being an inherent property of the medium.
This difference between percolation and diffusion can
be mathematically understood through the 1-dimen-
2. Theoretical background sional Polya walk. The medium is described as a set of
points placed at equal intervals along a straight line and
the particles of fluid can move in steps of unit length in
2.1. Percolation theory either direction with equal probability. In the case of
diffusion, the points constitute the fluid as well as the
There are many physical phenomena in which a fluid medium and so they can move in random fashion
spreads randomly through a medium, e.g. it may be a without any constraints. In the corresponding percola-
solute diffusing through a solvent, molecules penetrating tion process, the points of the medium are assigned a
a porous solid, or electrons migrating over an atomic direction to the left or right with equal probability. A
lattice. Besides the random mechanism, external forces particle entering the medium moves in accordance with
S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400 387

Fig. 1. Particles ‘a’ and ‘b’ are trapped in the medium due to
orientation of arrows.

Fig. 3. Intermediate stage of percolation in 2-dimensional network.

Fig. 2. Percolation pathway found in the network. Pores part of
pathway have been shown connected with a dashed line. Black filled 2.2. Theory of capillary pressure
circles are inter-connected pores. Open circles are blocked pores.
Arrows indicate direction of motion of a particle at a point in the
medium. Consider the hydrostatics of two immiscible fluids or
phases that exist simultaneously in a porous medium
(Greenkorn, 1983; De Weist, 1969). In general, one
phase will wet the solid. The entrance of one fluid into a
small pore against the other fluid is opposed by surface
tension forces between the wetting fluid and the pore
the arrows at each point and so the medium plays the walls (Scheidegger, 1963; Muskat, 1982). The result is
active role. As can be seen from Fig. 1, a particle can get that a certain pressure differential in the displacing
trapped and be forced to oscillate indefinitely between phase versus the displaced phase will have to be
two points (whereas this does not occur in the case of produced to maintain equilibrium. This pressure is
diffusion). In such a case, no percolation path can be called the capillary pressure. In a single capillary, the
struck between two ends of a line. curvature Rc of the interface gives rise to the pressure
This simple 1-dimensional illustration can be ex- differential equal to
tended further to 2 and 3-dimensional networks of Pc 2s=Rc (1)
elements, where some elements may also be blocked
off to fluid flow. Consider a 2-dimensional matrix The radius of curvature of the meniscus is equal to
through which a path has to be found (Fig. 2). If all Rc r=cos f (2)
elements are open to fluid flow, then they form part of a
So that for a single circular capillary
single ‘percolation path’ for the entire medium. As
elements are blocked off, the size of the percolation Pc 2s cos f=r (3)
path is reduced, but it still connects both ends of the As can be seen from the above expression, the surface
network. At a certain critical fraction of closed elements, tension force is inversely related to capillary radius.
the percolation path ceases to exist. It has been observed Hence, capillary pressure can be regarded as the
that this fraction is 0.41 for 2-dimensional networks and resistance offered by a capillary to the flow of fluid
0.69 for 3-dimensional ones (Efros, 1982) (Fig. 3). through it */the larger the capillary radius, the lower the
388 S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

resistance. In this simulation, the resistance of an 3. Simulation

element is dimensionless and so surface tension values
are irrelevant, i.e. the analysis is system non-specific. 3.1. Initialization

2.3. Pore structure models The simulation requires a random generation of

resistances of the elements that mimic the porous
This simulation uses a simplified model of porous medium. The parameters of relevance are the size of
medium. The model consists of a network of elements the network or lattice and the desired phase fraction of
that represent cylindrical capillary tubes of different blocked elements (Monteagudo, Rajagopal & Lage,
diameters and equal lengths. A single pore, then, is a 2002). The resistances are generated over a range of
series of elements placed one after another and so values as would exist in a porous medium, but the actual
incorporates the effect of varying diameter along its distributions as reported in literature have not been
length. Our model has all pores of same length. The used.
effect of ‘arrows’ (as in Polya walk) is obtained by the At a given instant, each pore is either ‘closed’, ‘open’
random generation and assignment of resistances (i.e. or ‘part of a flowpath’. To start with, all nodes are
through assigning radii values) to elements of the marked ‘closed’. It should be noted that there might be
medium. The concept of least resistance is used to pore(s) having finite resistance but are surrounded
determine the percolation path of the fluid. completely by ‘blocked-off’ pores. As these pores cannot
If pressure is applied to a fluid-filled porous medium, be reached from the entry face, their status would
or to the fluid at the entrance to a capillary system, the remain unchanged from ‘closed’ throughout the dura-
fluid will penetrate those capillaries whose capillary tion of the simulation. As the pressure is increased and
pressure is lower than the applied pressure. In other the fluid begins to percolate in the medium, pores that
words, the largest-diameter capillaries would be filled are filled are marked ‘open’. For a pore’s status to
first and at increasing pressures, the smaller capillaries become ‘part of a flowpath’, it must become a part of
would get filled. This is referred to as the concept of either a dependent or an independent flowpath. It is
‘least resistance’. possible for pores to remain ‘open’ and yet not ‘part of a
In actual porous media, the pores can be fully filled, flowpath’ because these pores could be dead ends or a
partially filled or be completely empty. Although the sequence of pores that have as yet not succeeded in
simulation assumes a pore under consideration as being forming a flowpath.
either fully filled or completely empty, it is possible to
model a partially filled pore as a combination of two 3.2. Least resistant pathway
adjacent pores */one being fully-filled and the other
completely empty. The simulation employs Dijkstra’s algorithm to
In this simulation, the pressure applied across the determine the least resistant pathway. Since both
network is incremented by a very small amount at every pressure and resistance are dimensionless, the resistance
iteration. This approach is validated by experimental value of the least resistant pathway can be equated to
observations where the effect of hysteresis was dimin- give the ‘break-through’ or the minimum pressure
ished or eliminated by carrying out the experiment required to cause the first flowpath to appear.
sufficiently slowly (Dullien, 1992). Varying pore geome- The simulation has an entry face for the fluid to enter
try in a porous medium essentially implies a pore of a into the network which can be regarded as a ‘single-
greater or smaller ‘capillary pressure’ than a correspond- source’ for the algorithm. By definition, the resistances
ing pore of uniform diameter. This implies that this pore of all pores are positive implying positive edge weights
requires a greater or smaller pressure for break-through. for the graph. The concept of edge weight is equivalent
Such effects have been accounted for by selecting the to the resistance of a pore and is stored in the pore itself.
minimum and maximum pore radii at the beginning of The simulation maintains a priority queue that
the simulation. The random selection of a radius value contains all the pores whose source distance is yet to
between the above limits for a pore incorporates all be finalized. All pores present in the queue are sorted in
media effects that can affect the resistance of a pore ascending order of their source distance. To begin with,
(Dullien). all pores are assigned a source distance value as infinite.
Thus, though this simplified model does not account Then the source distance of the pores on the entry face is
for the presence of films, ‘permanent hysteresis’ (Dul- equated to the resistance of the pore. This is followed by
lien, 1992) or different connectivities of the pore repeatedly selecting a pore with the least source distance
elements with their neighbours (Cordero, Rojas & in the priority queue and relaxing all pores that are
Riccardo, 2001) in the system, it incorporates most connected to it. The process of relaxation of the source
features of porous media that would have a strong distance of a pore involves updating the current value of
bearing on the flowrate. the source distance with the sum of the resistance of the
S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400 389

pore and the source distance of the neighbouring pore ‘opened-up’ and the ‘flow resistance’ of the pore already
(that has just been popped from the queue), if the value a part of some existing flowpath. ‘Flow resistance’ of a
of the latter is smaller (Cormen, Leiserson & Rivest, pore is defined as the cumulative resistance of all the
2000). pores starting from the pore under consideration to a
After all the pores have been assigned the minimum pore on the exit face. Other than the above mentioned
possible source distance value, the source distance of all case, there is a distinct possibility of the existence of
pores on the exit face is compared and the one having multiple series of pores that have opened up from a pore
the least value is selected. This is the value of resistance inside the network and joined existing flowpaths. In
of the ‘least resistant pathway’ or the ‘break-through’ order to take into account the branching of flow from
pressure required across the network. such pores, the resistance of each branch is computed
For a pathway to exist between the entry and exit face separately by linear addition of the resistances of the
of the network, the value of ‘break-through’ pressure pores that have ‘opened-up’ (starting at the pore from
must be finite. If a pathway has been found, the which the branching begins till a pore which is an
elements are stored in order of their appearance in the immediate neighbour of a pore whose status is ‘part of a
percolation path and consequently, the order in which flowpath’) with the ‘flow-resistance’ of the pore which is
they will be filled up by the fluid. a ‘part of a flowpath’. Thus, the ‘flow resistance’ of the
pore from which the branching begins is the parallel
3.3. Opening of a node summation of the resistances of all the branches from
that pore onwards. The effective resistance of the
The next step is to carry out a depth-first search ‘dependent flowpath’ is thus the linear addition of the
across all pores having finite resistance which have resistances of all pores from the entry face to the pore
opened up. A pore is deemed to have ‘opened-up’ when from which branching begins and the ‘flow resistance’ of
its capillary pressure is smaller than the effective that pore.
pressure available at that pore. Effective pressure at a
pore is the difference of the cumulative resistance of all 3.5. Flowrate
pores that precede the current pore from the applied
pressure across the network. The recursion for a pore In the case of an actual porous medium, the physical
ends when the capillary resistance is greater than the quantity ‘flowrate’ is defined only when the fluid
effective pressure or all adjacent pores having finite actually exits from the pores at the end opposite to the
resistance have been explored. one at which it entered (assuming 1-dimensional perco-
The pressure is incremented in steps until all the pores lation). The simulated network, in reality, has several
that have finite resistance and are reachable have exit points, each of which has its own flowrate. The
‘opened-up’. Although the pressure is incremented flowrate through the network is defined as the cumula-
slowly, at a given increment there can be multiple pores tive flowrate from all the pores on the exit face.
that ‘open-up’. This is similar to the morphological In this simulation, mass and volume conservation
approach used to study fractal dimensions (Hilpert & have been assumed to hold. The pores do not rupture in
Miller, 2001). the pressure range being studied. Since fluid enters from
one face and leaves from the opposite face, the
3.4. Additional flowpath determination cumulative flowrate could be alternatively defined as
the sum of the flowrates entering the network through
The process of repeatedly increasing applied pressure the pores on the entry face.
would eventually lead to other flowpaths opening up. A When the first flowpath is obtained (using Dijkstra’s
new flowpath comes into existence when an independent algorithm), the net flowrate through it (and in this case,
or dependent flowpath comes into existence. An ‘in- through the network) is zero. As the inlet pressure is
dependent flowpath’ consists of a series of pores that increased, a new sequence of pores starts getting filled
have ‘opened-up’ from the entry to the exit face, while a by the fluid, which may result in another flowpath
‘dependent flowpath’ is a sequence of pores from the joining one of the existing flowpath or flowpaths
entry face to a pore already part of an existing flowpath. emerging from the exit face. All flowpaths have the
The resistance offered by an ‘independent flowpath’ is same inlet and outlet pressures at any given time and
the sum of the resistances of all pores that form the vary only in their individual resistances. Thus, the net
flowpath from the entry to the exit face. pressure driving force across a particular path, rather
For a ‘dependent flowpath’, the simplest case would than the inlet pressure is taken for calculation of the
involve just one sequence of pores from the entry face to flowrate. The flowrate through a path is then computed
a pore already part of an existing flowpath. Here, the by dividing the driving force by the path resistance.
resistance of the newly opened flowpath equals the
cumulative resistance of all pores that have just recently Pi Pa Ri (4)
390 S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Fig. 4. Fraction of openable pores open vs. applied pressure for a 150 /150 grid at 20 % blockage. Continuous line represents the least-squares
averaged trend line.

Fig. 5. Flowrate vs. applied pressure for a 150/150 grid at 20% blockage. Continuous line represents the least squares averaged trend line.

Qi Pi =Ri (5) attained, the matrix is regenerated and the process
repeated a few dozen times over.
The total flowrate is then the sum of individual
flowrates and can be expressed as
3.6. Constraints
Q Qi (6) The random nature of the trials causes variation in the
1 values of ‘break-through’ and ‘maximum’ pressures.
Both these pressures vary with matrix size and fraction
The flowrate is calculated each time a pressure of pores blocked-off.
increment is made and also a new pore ‘opens up’. The choice of matrix size and blocked fraction is
The pressure after which no more pores open up is dictated by two factors */computer memory size and
called ‘maximum pressure’. Once ‘maximum pressure’ is critical phase fraction values. The memory factor limits
S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400 391

Fig. 6. Fraction of openable pores open vs. Applied pressure for 80/80 grid at 30% blockage. Continuous line represents the least-square averaged
trend line.

Fig. 7. Flowrate vs. applied pressure for 80/80 grid at 30 % blockage. Continuous line represents the least-squares averaged trend line.

the maximum network size that can be simulated but 4. Results

there is also a certain minimum value below which the
simulation is unable to generate statistically significant In this work, we have a pore throat diameter
data to mimic the flow properly. At these ‘finite’ sizes, distribution (‘pore size distribution’) which is generated
the randomness associated with Monte Carlo simulation by a pseudorandom number generator according to a
cannot be used effectively and errors result. uniform distribution between the minimum and max-
392 S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Fig. 8. Fraction of openable pores open vs. applied pressure for 70/70 grid at 40% blockage. Continuous line represents the least-squares averaged
trend line.

Fig. 9. Flowrate vs. applied pressure for a 70/70 grid at 40% blockage. Continuous line represents the least-squares averaged trend line.

imum values. These values were assumed to be 1 and 10, 7, 9 and 11). In these plots, the values of the y -intercept
respectively. The pseudo-random numbers are used for must be negative. It is indicative of the fact that only at a
assigning values to the required throat diameters, so that certain finite positive pressure does the flow through the
those are uncorrelated, i.e. the size of one throat is porous medium begin to take place. The values of
independent of the size of any other throat. Porous parameters that have been estimated using linear least-
media may have polymodal and/or spatially correlated squares technique, show a consistency in the predicted
pore size distributions, that can also be handled by the values regardless of the matrix size for a specified
model, but these are not considered in this work. blockage fraction (Tables 1/5). Due to simulation
The flowrate vs. applied pressure relationships ob- time constraints we could not carry out runs at grid
tained from the model for different sets of parameters sizes greater than 200 for 2-dimensional networks.
show a linear dependence similar to Darcy’s law (Figs. 5, However looking at the graphs of the dominant para-
S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400 393

Fig. 10. Fraction of openable pores open vs. applied pressure for 18/18/18 grid at 40% blockage. Continuous line represents the least-squares
averaged trend line.

Fig. 11. Flowrate vs. applied pressure for a 18/18/18 grid at 40% blockage. Continuous line represents the least-squares averaged trend line.

meter vs. grid size (Fig. 12) and cumulative flowrate vs. are not ‘blocked-off’.
grid size at a specific applied pressure (Fig. 13), we For 2-dimensional systems;
believe that an infinite medium can be simulated (7)
satisfactorily in the 200/300 grid size range for 2- Openable Pores  s2g (1fb )
dimensional networks. For 3-dimensional systems;
The flowrate is dependent on the number of flowpaths (8)
Openable Pores  s3g (1fb )
and that in turn on the accessible porosity of the
FOPO APO=Openable Pores (9)
medium. In our view, fraction of openable pores open
(FOPO) is a good measure of the accessible porosity. To On plotting FOPO vs. applied pressure for different
calculate FOPO, we noted the actual number of pores configurations, i.e. grid sizes and blockage fractions, we
that have ‘opened-up’ (APO) while incrementing the note an initial linear rising trend, attaining an asympto-
pressure and then divided it by the number of pores that tic value subsequently (Figs. 4, 6, 8 and 10). This trend is
Table 1
Values of coefficients for fraction of openable pores open and flowrate for 2 and 3-dimensional grids at blockage of 20%

Dimension Size Fraction of openable pores open Flowrate

R2 R2

S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Coefficients Coefficients Q

P6 P5 P4 P3 P2 P1 P0 P1 P0

2 40 /3.00E/16 1.00E/12 /1.00E/09 7.00E/07 /2.00E/04 3.28E/02 /9.49E/02 0.9089 4.89 /29.30 4863.60 0.8195
50 /8.00E/17 3.00E/13 /4.00E/10 3.00E/07 /1.00E/04 2.52E/02 /8.47E/01 0.9306 5.98 /40.76 5935.94 0.7727
60 /2.00E/17 1.00E/13 /2.00E/10 2.00E/07 /9.00E/05 2.08E/02 /8.66E/01 0.9441 7.81 /47.56 7760.15 0.8829
70 /9.00E/18 5.00E/14 /1.00E/10 1.00E/07 /6.00E/05 1.72E/02 /8.27E/01 0.9539 9.28 /41.79 9235.21 0.9014
80 /4.00E/18 2.00E/14 /6.00E/11 7.00E/08 /5.00E/05 1.46E/02 /8.03E/01 0.9606 10.51 /48.02 10456.98 0.9080
90 /2.00E/18 1.00E/14 /3.00E/11 5.00E/08 /3.00E/05 1.25E/02 /7.73E/01 0.9641 11.65 /81.98 11567.02 0.8840
100 /8.00E/19 6.00E/15 /2.00E/11 3.00E/08 /3.00E/05 1.11E/02 /7.65E/01 0.9679 12.88 /77.17 12797.83 0.9263
120 /2.00E/19 2.00E/15 /8.00E/12 2.00E/08 /2.00E/05 8.60E/02 /7.08E/01 0.9746 16.79 /44.28 16744.72 0.9195
150 /4.00E/20 5.00E/16 /3.00E/12 7.00E/09 /9.00E/06 6.40E/02 /6.58E/01 0.9780 21.10 /353.29 20744.71 0.9570
175 /5.00E/21 1.00E/16 /8.00E/13 3.00E/09 /5.00E/06 4.60E/02 /5.15E/01 0.9777 23.58 /259.32 23324.68 0.9112
190 1.00E/20 /2.00E/16 6.00E/13 /6.00E/10 /1.00E/06 2.50E/02 /2.16E/01 0.9727 26.98 /590.17 26384.83 0.9728
200 1.00E/20 /1.00E/16 5.00E/13 /6.00E/10 /8.00E/07 2.20E/02 /1.85E/01 0.9748 28.20 /120.83 28082.17 0.9568
3 10 /2.00E/13 2.00E/10 /8.00E/08 2.00E/05 /1.50E/03 7.16E/02 /2.30E/01 0.9821 11.45 /99.17 11349.83 0.9570
11 /3.00E/14 5.00E/11 /2.00E/08 6.00E/06 /8.00E/04 4.88E/02 /1.03E/01 0.9496 13.97 /77.61 13892.40 0.9681
12 /2.00E/14 3.00E/11 /2.00E/08 4.00E/06 /6.00E/04 4.28E/02 /8.75E/02 0.9644 17.06 /96.06 16959.95 0.9729
13 /1.00E/15 2.00E/11 /1.00E/08 3.00E/06 /5.00E/04 3.75E/02 /8.07E/02 0.9680 20.25 /110.19 20136.81 0.9679
14 /5.00E/15 8.00E/12 /6.00E/09 2.00E/06 /4.00E/04 3.14E/02 /2.83E/02 0.9735 23.28 /85.03 23297.97 0.9680
15 /2.00E/15 5.00E/12 /4.00E/09 1.00E/06 /3.00E/04 2.89E/02 /4.22E/02 0.9752 27.24 /143.68 27097.32 0.9765
16 /1.00E/15 3.00E/12 /2.00E/09 1.00E/06 /2.00E/04 2.46E/02 /2.17E/02 0.9799 30.44 /164.09 30276.91 0.9786
17 /4.00E/16 1.00E/12 /1.00E/09 6.00E/07 /2.00E/04 2.10E/02 1.70E/03 0.9805 35.76 /187.90 35573.10 0.9842
18 /3.00E/17 3.00E/13 /5.00E/10 4.00E/07 /1.00E/04 1.77E/02 3.88E/02 0.9831 39.38 /212.43 39162.57 0.9805
19 1.00E/16 1.00E/13 /9.00E/11 2.00E/07 /7.00E/05 1.45E/02 /4.50E/03 0.9637 44.23 /252.98 43973.02 0.9725
20 8.00E/17 /8.00E/14 /6.00E/11 1.00E/07 /6.00E/05 1.31E/02 8.50E/03 0.9670 50.05 /283.51 49761.49 0.9684

R2 represents the regression coefficient values. Q represents the value of flowrate at an applied pressure of 1000 units. FOPO (Pa) /P6P6a/P5P5a/P4P4a/P3P3a/P2P2a/P1Pa/P0, Flowrate (Pa) /
P1Pa/P0, Q/P1/1000/P0.
Table 2
Values of coefficients for fraction of openable pores open and flowrate for 2- and 3-dimensional grids at blockage of 25 %

Dimension Size Fraction of openable pores open Flowrate

S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Coefficients R Coefficients Q R2

P6 P5 P4 P3 P2 P1 P0 P1 P0

2 40 /3.00E/16 8.00E/13 /1.00E/09 6.00E/07 /2.00E/04 3.05E/02 /8.67E/01 0.8789 3.37 /23.16 3344.14 0.7737
50 /7.00E/17 3.00E/13 /4.00E/10 3.00E/07 /1.00E/04 2.46E/02 /9.09E/01 0.9162 4.60 /23.86 4575.55 0.7250
60 /2.00E/17 1.00E/13 /2.00E/10 2.00E/07 /8.00E/05 2.01E/02 /8.97E/01 0.9325 5.33 /34.30 5298.41 0.8331
70 /8.00E/18 4.00E/14 /9.00E/11 1.00E/07 /6.00E/05 1.70E/02 /8.84E/01 0.9444 6.22 /49.19 6174.31 0.7869
80 /4.00E/18 2.00E/14 /5.00E/11 7.00E/08 /4.00E/05 1.49E/02 /8.95E/01 0.9540 7.58 /38.44 7540.87 0.8264
90 /2.00E/18 1.00E/14 /3.00E/11 4.00E/08 /3.00E/05 1.28E/02 /8.39E/01 0.9618 8.93 /123.30 8811.10 0.8281
100 /8.00E/19 6.00E/15 /2.00E/11 3.00E/08 /3.00E/05 1.11E/02 /8.45E/01 0.9626 9.06 /54.21 9007.69 0.8748
120 /2.00E/19 2.00E/15 /8.00E/12 2.00E/08 /2.00E/05 8.60E/03 /7.74E/01 0.9710 12.04 /141.63 11902.57 0.9463
150 /4.00E/20 5.00E/16 /3.00E/12 7.00E/09 /9.00E/06 6.40E/03 /7.25E/01 0.9761 14.72 /67.14 14647.86 0.8777
175 /9.00E/21 /6.00E/17 /4.00E/14 1.00E/09 /3.00E/06 3.80E/03 /4.23E/01 0.9708 16.81 /1.86 16810.14 0.9140
200 9.00E/21 /1.00E/16 4.00E/13 /2.00E/10 /1.00E/06 2.50E/03 /2.67E/01 0.9715 21.83 /103.21 21725.79 0.9471
3 10 /2.00E/13 2.00E/10 /7.00E/08 1.00E/05 /1.50E/03 7.05E/02 /2.24E/01 0.9696 10.30 /46.54 10253.47 0.9596
11 /4.00E/14 5.00E/11 /2.00E/08 6.00E/06 /8.00E/04 5.07E/02 /1.47E/01 0.9550 11.34 /61.61 11282.40 0.9627
12 /2.00E/14 3.00E/11 /2.00E/08 4.00E/06 /6.00E/04 4.34E/02 /1.12E/01 0.9602 13.95 /79.27 13873.73 0.9470
13 /1.00E/14 2.00E/11 /1.00E/08 3.00E/06 /5.00E/04 3.83E/02 /1.06E/01 0.9673 16.49 /94.49 16392.51 0.9704
14 /5.00E/15 9.00E/12 /6.00E/09 2.00E/06 /4.00E/04 3.43E/02 /1.03E/01 0.9694 19.91 /111.98 19796.02 0.9684
15 /3.00E/15 5.00E/12 /4.00E/09 1.00E/06 /3.00E/04 2.91E/02 /6.88E/02 0.9726 21.35 /127.20 21222.80 0.9694
16 /1.00E/15 3.00E/12 /2.00E/09 1.00E/06 /2.00E/04 2.60E/02 /5.82E/02 0.9774 25.06 /146.22 24916.78 0.9800
17 /6.00E/16 1.00E/12 /1.00E/09 1.00E/07 /2.00E/04 2.23E/02 /2.82E/02 0.9813 27.20 /160.49 27038.51 0.9834
18 /2.00E/16 7.00E/13 /8.00E/10 5.00E/07 /1.00E/04 1.97E/02 /1.02E/02 0.9816 30.77 /183.64 30585.36 0.9806
19 1.00E/16 /6.00E/14 /2.00E/10 2.00E/07 /8.00E/05 1.56E/02 /3.82E/02 0.9596 34.90 /216.07 34680.93 0.9673
20 2.00E/16 /2.00E/13 2.00E/11 1.00E/07 /6.00E/05 1.38E/02 /1.75E/02 0.9642 38.75 /244.84 38504.16 0.9660

R2 represents the regression coefficient values. Q represents the value of flowrate at an applied pressure of 1000 units. POPO (Pa) /P6P6a/P5P5a/P4P4a/P3P3a/P2P2a/P1Pa/P0, Flowrate (Pa) /
P1Pa/P0, Q/P1/1000/P0.

Table 3
Values of coefficients for fraction of openable pores open and flowrate for 2- and 3-dimensional grids at blockage of 30 %

Dimension Size Fraction of openable pores open Flowrate

R2 R2

S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Coefficients Coefficients Q

P6 P5 P4 P3 P2 P1 P0 P1 P0

2 40 /2.00E/16 6.00E/13 /7.00E/10 5.00E/07 /2.00E/04 2.64E/02 /7.83E/01 0.8487 2.39 /22.77 2363.33 0.6176
50 /5.00E/17 2.00E/13 /3.00E/10 2.00E/07 /1.00E/04 2.14E/02 /8.51E/01 0.8784 3.12 /19.96 3104.14 0.6818
60 /9.00E/18 5.00E/14 /9.00E/11 1.00E/07 /5.00E/05 1.54E/02 /7.23E/01 0.8944 3.43 /28.82 3396.98 0.7054
70 /5.00E/18 3.00E/14 /6.00E/11 7.00E/08 /4.00E/05 1.44E/02 /8.11E/01 0.9227 4.04 /34.01 4004.39 0.8041
80 /2.00E/18 1.00E/14 /4.00E/11 5.00E/08 /3.00E/05 1.26E/02 /7.98E/01 0.9203 4.72 /20.15 4698.95 0.7591
90 /9.00E/19 7.00E/15 /2.00E/11 3.00E/08 /3.00E/05 1.09E/02 /8.05E/01 0.9383 5.43 /40.26 5392.14 0.8179
100 /4.00E/19 3.00E/15 /1.00E/11 2.00E/08 /2.00E/05 9.60E/03 /7.90E/01 0.9534 6.90 /48.15 6855.05 0.8026
120 /1.00E/19 1.00E/15 /6.00E/12 1.00E/08 /1.00E/05 8.00E/03 /7.87E/01 0.9572 7.84 /39.59 7804.61 0.8617
150 /2.00E/20 2.00E/16 /2.00E/12 5.00E/09 /8.00E/06 6.00E/03 /7.42E/01 0.9679 10.44 /126.99 10316.01 0.8970
175 /1.00E/21 6.00E/17 /6.00E/13 2.00E/09 /5.00E/06 4.40E/03 /6.10E/01 0.9728 13.02 /34.94 12980.06 0.9100
190 1.00E/20 /1.00E/16 5.00E/13 /5.00E/10 /1.00E/06 2.40E/03 /2.68E/02 0.9657 12.42 /71.08 12352.92 0.8820
200 1.00E/20 /1.00E/16 6.00E/13 /9.00E/10 /2.00E/07 1.80E/03 /1.83E/01 0.9643 13.69 /144.91 13544.09 0.8988
3 10 /5.00E/14 6.00E/11 /3.00E/08 7.00E/06 /8.00E/04 4.90E/02 /7.78E/02 0.9693 7.63 /68.05 7558.55 0.9357
11 /3.00E/14 4.00E/11 /2.00E/08 6.00E/06 /8.00E/04 5.16E/02 /1.93E/01 0.9671 9.69 /78.84 9615.77 0.9626
12 /2.00E/14 3.00E/11 /2.00E/08 4.00E/06 /6.00E/04 4.44E/02 /1.51E/01 0.9558 10.75 /72.81 10680.19 0.9530
13 /1.00E/14 2.00E/11 /1.00E/08 3.00E/06 /5.00E/04 4.01E/02 /1.52E/01 0.9599 12.98 /86.76 12896.24 0.9738
14 /5.00E/15 9.00E/12 /6.00E/09 2.00E/06 /4.00E/04 3.43E/02 /1.22E/01 0.9885 14.71 /98.58 14614.42 0.9621
15 /3.00E/15 5.00E/12 /4.00E/09 2.00E/06 /3.00E/04 3.06E/02 /1.12E/01 0.9732 16.80 /114.86 16684.14 0.9601
16 /2.00E/15 4.00E/12 /3.00E/09 1.00E/06 /3.00E/03 2.80E/02 /1.13E/01 0.9723 18.70 /123.03 18576.97 0.9706
17 /3.00E/12 6.00E/09 /6.00E/06 /2.80E/03 /6.85E/01 8.28E/01 /2.70E/02 0.9777 21.56 /141.11 21415.89 0.9740
18 /3.00E/16 9.00E/13 /1.00E/09 5.00E/07 /2.00E/04 2.11E/02 /5.63E/01 0.9628 24.47 /137.24 24334.76 0.9765
19 8.00E/18 2.00E/13 /4.00E/10 3.00E/07 /1.00E/04 1.77E/02 /9.66E/02 0.9581 27.45 /191.75 27262.25 0.9839
20 1.00E/16 /6.00E/14 /1.00E/10 2.00E/07 /8.00E/05 1.55E/02 /6.76E/02 0.9613 30.88 /213.58 30666.42 0.9707

R2 represents the regression coefficient values. Q represents the value of flowrate at an applied pressure of 1000 units. FOPO (Pa) /P6P6a/P5P5a/P4P4a/P3P3a/P2P2a/P1Pa/P0, Flowrate (Pa) /
P1Pa/P0, Q /P1/1000/P0
Table 4
Values of coefficients for Fraction of Openable Pores Open and Flowrate for 2- and 3-dimensional grids at blockage of 35 %

Dimension Size Fraction of openable pores open Flowrate

S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Coefficients R2 Coefficients Q R2

P6 P5 P4 P3 P2 P1 P0 P1 P0

2 40 /2.00E/16 5.00E/13 /6.00E/10 3.00E/07 /1.00E/04 1.95E/02 /5.43E/01 0.6144 1.30 /21.16 1275.94 0.4851
50 /9.00E/18 5.00E/14 /9.00E/11 9.00E/08 /5.00E/05 1.31E/02 /4.75E/01 0.7336 1.80 /16.92 1784.38 0.5058
60 /4.00E/17 5.00E/14 1.00E/10 1.00E/07 /5.00E/05 1.97E/02 7.44E/02 0.7235 2.05 /41.64 2010.56 0.5494
70 /7.00E/19 7.00E/15 /2.00E/11 3.00E/08 /2.00E/05 9.70E/03 /5.72E/01 0.8182 2.32 /20.71 2300.49 0.5519
80 /1.00E/18 9.00E/15 /2.00E/11 3.00E/08 /3.00E/05 1.00E/02 /6.96E/01 0.8293 3.11 /22.08 3088.02 0.7216
90 2.00E/19 /5.00E/16 /1.00E/12 6.00E/09 /9.00E/06 5.60E/03 /3.58E/01 0.8645 4.22 /19.90 4196.30 0.7665
100 /4.00E/19 3.00E/15 /9.00E/12 1.00E/08 /1.00E/05 6.70E/03 /5.38E/01 0.8276 4.38 /108.20 4268.90 0.6773
120 1.00E/19 /7.00E/16 1.00E/12 3.00E/10 /4.00E/06 4.00E/03 /3.82E/01 0.8983 4.97 /129.83 4842.57 0.7234
150 6.00E/20 /5.00E/16 2.00E/12 /2.00E/09 /2.00E/07 2.20E/03 /2.43E/01 0.9486 5.95 /83.84 5861.36 0.7084
175 2.00E/20 /3.00E/16 1.00E/12 /2.00E/09 1.00E/06 9.00E/04 /5.20E/03 0.8808 7.86 /483.51 7372.69 0.8720
200 1.00E/20 /2.00E/16 8.00E/13 /2.00E/09 1.00E/06 9.00E/04 /6.94E/02 0.9425 /425.30 8163.10 8163.1 0.8574

3 10 /1.00E/13 1.00E/10 /6.00E/08 1.00E/05 /1.30E/03 6.77E/02 /2.74E/01 0.9703 5.62 /49.53 5572.47 0.9240
11 /2.00E/14 3.00E/11 /2.00E/08 5.00E/06 /7.00E/04 4.52E/02 /1.61E/01 0.9570 6.63 /41.61 6591.19 0.9507
12 /2.00E/14 3.00E/11 /1.00E/08 4.00E/06 /6.00E/04 4.47E/02 /1.83E/01 0.9499 8.18 /60.93 8120.87 0.9574
13 /1.00E/14 2.00E/11 /1.00E/08 3.00E/06 /5.00E/04 4.11E/02 /1.94E/01 0.9511 9.54 /72.61 9470.10 0.9595
14 /6.00E/15 1.00E/11 /7.00E/09 2.00E/06 /4.00E/04 3.63E/02 /1.78E/01 0.9645 10.97 /82.84 10887.16 0.9451
15 /3.00E/15 6.00E/12 /4.00E/09 2.00E/06 /3.00E/04 3.22E/02 /1.59E/01 0.9680 12.78 /92.88 12691.12 0.9546
16 /2.00E/15 4.00E/12 /3.00E/09 1.00E/06 /3.00E/04 2.90E/02 /1.47E/01 0.9690 15.49 /116.15 15377.85 0.9662
17 /1.00E/15 2.00E/12 /2.00E/09 9.00E/07 /2.00E/04 2.57E/02 /1.22E/01 0.9729 16.74 /134.24 16608.76 0.9692
18 /6.00E/16 1.00E/12 /1.00E/09 7.00E/07 /2.00E/04 2.43E/02 /1.64E/01 0.9755 14.01 /114.95 13890.05 0.9836
19 /2.00E/16 6.00E/13 /7.00E/10 4.00E/07 /1.00E/04 2.03E/02 /1.71E/01 0.9558 20.71 /162.10 20543.90 0.9772
20 8.00E/17 4.00E/15 /2.00E/10 2.00E/07 /9.00E/05 1.68E/02 /1.11E/01 0.9565 22.93 /182.16 22747.84 0.9680

R2 represents the regression coefficient values. Q represents the value of flowrate at an applied pressure of 1000 units. POPO (Pa) /P6P6a/P5P5a/P4P4a/P3P3a/P2P2a/P1Pa/P0, Flowrate (Pa) /
P1Pa/P0, Q /P1/1000/P0.

Table 5
Values of coefficients for Fraction of Openable Pores Open and Flowrate for 2- and 3-dimensional grids at blockage of 40 %

Dimension Size Fraction of openable pores open Flowrate

S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

Coefficients R2 Coefficients Q R2

P6 P5 P4 P3 P2 P1 P0 P1 P0

2 40 /4.00E/17 1.00E/13 /1.00E/10 1.00E/07 /4.00E/05 8.40E/03 /7.93E/02 0.4145 0.53 /4.57 525.53 0.1420
50 /4.00E/18 6.00E/15 9.00E/13 /5.00E/09 5.00E/07 1.50E/03 3.15E/01 0.3538 1.11 /48.56 1060.84 0.4811
60 /2.00E/17 7.00E/14 /1.00E/10 1.00E/07 6.00E/05 1.48E/02 /8.78E/01 0.4960 0.93 /39.81 894.19 0.3329
70 /6.00E/19 4.00E/15 /1.00E/11 1.00E/08 1.00E/05 4.80E/03 /1.93E/01 0.5226 0.96 /83.53 880.97 0.2412
80 /4.00E/18 /2.00E/14 4.00E/11 /4.00E/08 2.00E/05 /2.00E/03 3.63E/01 0.5690 1.35 /90.05 1262.75 0.3091
90 8.00E/19 /5.00E/15 1.00E/11 /1.00E/08 5.00E/06 1.00E/04 1.93E/01 0.6605 2.42 /97.95 2317.45 0.3922
100 /7.00E/19 5.00E/15 /1.00E/11 2.00E/08 /1.00E/05 4.80E/03 /3.47E/01 0.5954 1.78 /65.86 1709.44 0.3082
120 3.00E/19 /2.00E/15 5.00E/12 /5.00E/09 3.00E/07 2.40E/03 /3.62E/01 0.6712 3.07 /871.17 2197.03 0.3891
150 2.00E/19 /2.00E/15 6.00E/12 /1.00E/08 1.00E/05 /6.60E/03 1.59E/00 0.6274 3.85 /153.03 3697.67 0.5664
175 1.00E/20 /1.00E/16 7.00E/13 /2.00E/09 3.00E/06 /7.20E/03 5.66E/01 0.6346 5.01 /983.12 4029.18 0.5165
3 10 /5.00E/14 6.00E/11 /3.00E/08 7.00E/06 /9.00E/04 5.27E/02 /1.81E/01 0.9378 3.90 /24.87 3870.13 0.9471
11 /3.00E/14 5.00E/11 /2.00E/08 6.00E/06 /8.00E/04 5.28E/02 /2.68E/01 0.9569 5.20 /46.21 5154.49 0.9255
12 /2.00E/14 3.00E/11 /2.00E/08 4.00E/06 /7.00E/04 4.64E/02 /2.34E/01 0.9445 5.83 /49.80 5782.70 0.9363
13 /1.00E/14 2.00E/11 /1.00E/08 3.00E/06 /5.00E/04 4.19E/02 /2.45E/01 0.9559 7.19 /59.08 7134.32 0.9373
14 /6.00E/15 1.00E/11 /7.00E/09 2.00E/06 /4.00E/04 3.78E/02 /2.35E/01 0.9605 8.34 /70.91 8266.79 0.9439
15 /3.00E/15 6.00E/12 /5.00E/09 2.00E/06 /4.00E/04 3.40E/02 /2.27E/01 0.9638 9.60 /80.99 9519.11 0.9582
16 /2.00E/15 4.00E/12 /3.00E/09 1.00E/06 /3.00E/03 3.00E/03 /1.96E/01 0.9664 11.03 /91.04 10934.97 0.9712
17 /1.00E/15 3.00E/12 /2.00E/09 1.00E/06 /2.00E/04 2.76E/02 /1.94E/01 0.9715 12.25 /101.36 12148.64 0.9659
18 /6.00E/16 1.00E/12 /1.00E/09 7.00E/07 /2.00E/04 2.43E/02 /1.64E/01 0.9738 14.01 /114.95 13890.05 0.9703
19 /3.00E/16 8.00E/13 /9.00E/10 5.00E/07 /1.00E/04 2.17E/02 /2.19E/01 0.9470 16.06 /136.46 15924.54 0.9639
20 /7.00E/17 3.00E/13 /5.00E/10 3.00E/07 /1.00E/04 1.91E/02 /1.85E/01 0.9520 17.28 /150.29 17129.71 0.9755

R2 represents the regression coefficient values. Q represents the value of flowrate at an applied pressure of 1000 units. POPO (Pa) /P6P6a/P5P5a/P4P4a/P3P3a/P2P2a/P1Pa/P0, Flowrate (Pa)/
P1Pa/P0, Q /P1/1000/P0.
S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400 399

Fig. 12. Slope of flowrate vs. grid size at different blockage fractions. Simulations with grid size 40 and above rate for 2-dimensional grids, while
stimulations smaller than this size were for 3-dimensional grids.

Fig. 13. Flowrate vs. grid size at different blockage fractions at an applied pressure of 1000 units. Simulations with grid size 40 and above were for 2-
dimensional grids while simulations smaller than this size were for 3-dimensional grids.

expected and holds irrespective of the grid size or the 5). This observation is in line with the fact that an ideal
fraction of pores blocked. This is similar to the observed simulation would have all grid dimensions as infinite.
trend when calculated cumulative mercury intrusion At low blocked fractions, there are few chances of
volume was plotted against applied pressure (Bryntes- pores with finite resistance being surrounded by
son, 2002). When the least-squares analysis is conducted ‘blocked-off’ pores. Thus all pores having finite resis-
and seven parameters of the polynomial estimated, it is tance can be considered ‘openable’. The asymptotic
observed that the parameter values gradually attain value of FOPO attained in all such cases is near 1 (Figs.
asymptotic values on increasing the grid size (Tables 1/ 4 and 6). As the blockage fraction increases, the
400 S. Jain et al. / Computers and Chemical Engineering 27 (2003) 385 /400

probability of finite- resistance pores being surrounded because theoretically an infinite number of runs are
by ‘blocked-off’ pores is higher and thus, a greater required before presenting any trend.
variation is seen in the asymptotic values of FOPO. This Further work in this field would include extension of
observation is most stark at blockage values within 5% the simulation to the region of Hagen /Poiseulle flow,
of the critical value (Fig. 8). This observation can be where the resistance is inversely related to the fourth
explained by the fact that near critical values there is a power of the radius of the element. The resistance
sharp decrease in the probability of finding a flowpath, beyond the laminar flow region decreases slightly and
the implication being that there is a greater variation in then reaches a constant value at very high flowrates in
the probability of finding a flowpath for a given the turbulent zone. This can be modelled as two separate
configuration. For this to be true, it would be natural flow regions, with elements having different (but con-
to expect a greater variation in the asymptotic value of stant) resistances in each region.
the openable pores. These observations have been
validated by the regression analysis while computing
the parameter values. The R2 trend values indicate that Acknowledgements
there is greater deviation from 1 at blockage fractions
near critical values, particularly at low grid sizes (Tables We are grateful to Dr M.N. Gupta, Senior Manager,
1 /5). As grid size increases the R2 values improve Computer Services Centre, IIT, Delhi, for the generous
indicating that the simulation is closer to that of an allocation of computational time. We would also like to
infinite medium. thank all staff members of CSC, IIT, Delhi, for their
cooperation throughout this project. In particular, we
would like to mention Mr. Gopal Krishen, the system
programmer, and Mr. Gulshan Naveriya, the hardware
5. Conclusions
engineer, for their invaluable assistance.
This simulation is a simplified but innovative way of
dealing with flow through porous media. Using the
Monte Carlo technique and concepts of percolation
theory, simulations have been carried out for 2 and 3- Bernsdorf, J., Brenner, G., & Durst, F. (2000). Numerical analysis of
dimensional models of porous media in capillary flow the pressure drop in porous media flow with lattice Boltzmann
regime. The flowrate vs. applied pressure results have (BGK) automata. Computer Physics Communications 129 , 247 /
been found to be in agreement with Darcy’s law. 255.
Binder, K. (Ed.), Monte Carlo methods in statistical physics (1979).
Parameter estimation of fraction of openable pores
that have ‘opened up’ vs. applied pressure has also Bryntesson, L. M. (2002). Pore network modelling of the behaviour of
been shown to be consistent. More accurate results a solute in chromatography media: transient and steady-state
could be obtained by using actual resistance distribu- diffusion properties. Journal of Chromatography A 945 , 103 /115.
tions of porous media. It might then be possible to Cordero, S., Rojas, F., & Riccardo, J. L. (2001). Simulation of three-
predict flowrate vs. applied pressure behaviour for dimensional porous networks. Colloids and Surfaces A: Physico-
chemical and Engineering Aspects 187 /188 , 425 /438.
systems with different physical properties (e.g. surface Cormen, T. H., Leiserson, C. E., & Rivest, R. L. (2000). Introduction to
tension and contact angle). algorithms . EEE: Prentice Hall of India.
Actual experimental data for flow in the capillary De Weist, R. J. M. (Ed.), Flow through porous media (1969). Academic
regime have not been reported in literature, where the Press.
major focus has been on statistical description of porous Dullien, F. A. L. (1992). Porous media */fluid transport and pore
structure . Academic Press.
media. Also, experiments that reproduce the exact Efros, A. L. (1982). Physics and geometry of disorder */percolation
conditions of the simulation would be difficult to theory . Mir Publishers.
perform. Therefore, the results mentioned herein should Greenkorn, R. A. (1983). Flow phenomenon in porous media . Marcel
be treated as being of a qualitative nature. Dekker.
The parameter values have been computed after Hilpert, M., & Miller, C. T. (2001). Pore-morphology-based simula-
tion of drainage in totally wetting porous media. Advances in Water
considering several thousands of points for each config- Resources 24 , 243 /255.
uration (matrix size and blocked fraction). Runs were Manegold, V. E. (1937). Uber Kapillarsysteme. Kolloid-Z 80 , 253.
concluded once the curve predicting FOPO vs. applied Monteagudo, J. E. P., Rajagopal, K., & Lage, P. L. C. (2002).
pressure exhibited an asymptotic behaviour. It has been Simulating oil flow in porous media under asphaltene deposition.
our observation that until this asymptotic behaviour is Chemical Engineering Science 57 , 323 /337.
Muskat, M. (1982). Flow of homogenous fluids through porous media .
observed the runs should not be concluded, otherwise IHRDC.
the parameter values would be misleading. This is also in Scheidegger, A. E. (1963). The physics of flow through porous media .
accordance with the random nature of the simulation University of Toronto Press.