1

Simulation of high molecular weight fluids through

porous media



Mid-term report
BACHELOR OF TECHNOLOGY IN CHEMICAL ENGINEERING
Major Project Part-I (CHD 411)

by
Nishant Kumar
(2011CH10100)



Department of Chemical Engineering
Indian Institute of Technology Delhi



2

Contents

S. No. Title Page No.
1. Introduction 3
2. Literature Review 3
4. Mathematical Model 4
5. Boundary Conditions 7
6. Results and Discussion 8
7. Future Plan 10
8. References 11


3

Introduction

Dealing with porous media has always been a challenge for the engineers. Here the work presented
is trying to simulate the flow of fluids through a porous media. First part of the work only deals
with diffusion through a porous media avoiding the convection or the velocity term. The basic idea
was to look at the inlet and the outlet concentration and the variation of these concentrations with
time. All the simulations have been done using Lattice Boltzmann method. A large part of the
project time was spent in understanding this model and applying it to various models and situations
for better understanding.

Objectives of the work

The objective was to model flows of fluids or analyze fluid behavior when passing through a
porous media. The flow behavior changed as the parameters influencing the flow were altered.
The lattice Boltzmann model was understood for different situations. Different boundary
conditions were taken into account.

Literature Review

Numerous literatures have been published related to modelling flows through porous media.

A one dimensional dual porosity model was developed for the purpose of studying saturated
water flow and solute transport in structured soils and fractured rocks. The model involved two
overlaying continua at the macroscopic level: a macro pore or the fracture pore system and a less
permeable matrix pore system. Water in both pores was assumed to be mobile. Saturated water
flow in the matrix as well as in the fracture pore system was described with the Richards
equation and solute transport equation was described with convection-dispersion equation.
Transfer of water and solutes between two pore regions was simulated by means of first order
equation. The mass transfer components for solute transport included both convective and
diffusive terms. The proposed model was capable of simulating flow situations using parameters
which can be related to the physical and chemical properties of the system.

Analytical solutions were presented for movement of chemicals with lateral or intra-aggregate
diffusion. The liquid phase in the porous medium was divided into mobile and immobile regions.
Sorption process in both dynamic and stagnant regions of the medium were assumed to be
instantaneous and adsorption isotherm is assumed to be linear. The model described in this paper
illustrated the extensive tailing observed during the flow through an unsaturated medium and
also explained the often observed early breakthrough of chemical in the effluent.

Mass transfer between mobile and immobile zones is a consequence of simultaneous process. A
multirate model was developed by Haggerty R. and Gorelick S. M. that allowed modelling of
small scale variations in rates and types of mass transfer by using a series of mass transfer
equations to represent each mass transfer process. The multirate model is then used in the
dispersive equation.
4

Mathematical model

The mathematical model used for simulating fluid flow through porous media is Lattice Boltzmann
method. This method has been increasingly attracting the researchers in many areas relating to
turbulent fields to multi-phase simulations in porous media. This method is one of the methods
used for computational fluid dynamics. Instead of the continuous Navier Stokes equation, the
discrete Boltzmann equation is used because of several advantages over the conventional methods.

Introduction

The above method can be used for simulating any transport phenomenon. Basically there are two
main approaches for simulating transport equations (heat, mass and momentum), continuous and
discrete. In continuous approach, ordinary or partial differential equations are obtain by
conservation of mass, momentum and energy for an infinitesimal control volume. Since often
these equations take form into complex and unsolvable equations. Even if they can be solved, it
requires extreme calculations and very high computational power and resources. Often these are
limited. Thus various schemes such as finite difference, finite element, and finite volume were
used to convert these complex differential equation to simpler and solvable algebraic forms. The
procedure is as follows. The first step is to identify the governing equations i.e. the physics
behind the problem at hand. The next step is to discretize the domain into volume, grids or
elements depending on the method of the solution. This step is looked at as if the volume or the
node or the element contains a collection of particles described by a distributions function. This
is macroscopic scale. The velocity, temperature and pressure of each particle are averaged over a
finite volume. The logical properties such as viscosity, thermal conductivity, diffusivity
coefficient are all predetermined.
The other extreme that can be talked of here is the microscopic scale. The medium can be
considered made of small particles (atoms or molecules). At this scale we need to identify inter
particle forces and solve the equations related to newtons second law. At each time step,
location and velocities of each particle is identified to solve the trajectory of each particle. The
physical parameters such as temperature, pressure, viscosity, heat capacity etc. have no definition
at this level. For example, temperature and pressure are related to the kinetic energy of the
particle. This method is called molecular dynamics.
Lattice Boltzmann method (LBM) has advantages of both microscopic and macroscopic level
and is often known as a mesoscopic level.
LBM has many advantages. It is easy to apply for complex domains, easy to treat multi-phase
and multi-component flows without a need to trace interfaces between different phases. It can
also be naturally adapted to parallel processing computing. Moreover, there is no need to solve
Laplace equation at each time step to satisfy the continuity equation of incompressible, unsteady
flows as it is in solving Navier-Stokes equation. Also, this method can handle problems at micro
and macro scales with reliable accuracy.

Lattice Boltzmann equation

Historically, the lattice Boltzmann approach developed from lattice gases, although it can also be
derived directly from the simplified Boltzmann BGK equation. In lattice gases, particles live on
5

the nodes of a discrete lattice. The particles jump from one lattice node to the next, according to
their (discrete) velocity. This is called the propagation phase. Then, the particles collide and get
a new velocity. This is the collision phase. Hence the simulation proceeds in an alternation
between particle propagations and collisions.



Figure 1 The simulation of particles between particles propagation and collision
It can be shown that lattice gases solve the Navier-Stokes equations of fluid flow. The major
disadvantage of lattice gases for common fluid dynamics applications is the occurrence of noise.
If the main interest is a smooth flow field one needs to average out over a very large lattice and
over a long time. The lattice Boltzmann method solves this problem by pre-averaging the lattice
gas. It considers particle distributions that live on the lattice nodes, rather than the individual
particles.


Figure 2 Illustration of laatice node of D2Q9 model

6




The general form of the lattice Boltzmann equation is


Where fi is the concentration of particles that travels with velocity ci. With the discrete velocity ci
the particle distributions travel to the next lattice node in one time step .

Figure 3 Illustration of streaming process of a lattice node
The collision operator i differs for the many lattice Boltzmann methods. In the lattice
Boltzmann Bhatnagher-Gross-Krook method (LBGK) that we use, the particle distribution after
propagation is relaxed towards the equilibrium distribution

(, ), as


The relaxation parameter determines the kinematic viscosity of the simulated fluid,
according to,


The equilibrium distribution

(, ) is a function of the local density and the local velocity .

These are the first and second order moments of the particle distribution as,


And


The equilibrium density

(, ) is calculated as


7

In which cs is the speed of sound, the index =

and tp is the corresponding equilibrium

density for . For the three-dimensional, nineteen velocity lattice (D3Q19) that we have used
in our simulations,

= 1/3 (rest particle),

1
= 1/18 (particles streaming to the face-connected
neighbors) and
3
= 1/36 (particles streaming to the edge-connected neighbors).



Boundary Conditions

Boundary conditions (BCs) are central to the stability and the accuracy of any numerical
solution.
In this project, we explore two of the most widely used BCs: Bounce-back BCs and Zou-He
velocity and pressure (density) BCs.

Bounce-back BCs

Bounce-back BCs are typically used to implement no-slip conditions on the boundary. By the so-
called bounce-back we mean that when a particle (discrete distribution function) reaches a
boundary node, the particle will scatter back to the fluid along with its incoming direction.
Bounce-back BCs come in a few variants and we focus on two types of implementations: the on-
grid and the mid-grid bounce-back.
The idea of the on-grid bounce-back is particularly simple and preserves a decent numerical
accuracy.

Figure 4 Illustration of on grid bounce back
In this configuration, the boundary of the fluid domain is aligned with the lattice points. One can
use a Boolean mask for the boundary and the interior nodes. The incoming directions of the
distribution functions are reversed when encountering a boundary node. This implementation
does not distinguish the orientation of the boundaries and is ideal for simulating fluid flows in
complex geometries, such as the porous media flow.
8

The configuration of the mid-grid bounce-back introduces fictitious nodes and places the
boundary wall centered between fictitious nodes and boundary nodes of the fluid. At a given
time step t, the distribution functions with directions towards the boundary wall would leave the
domain. Collision process is then applied and directions of these distribution functions are
reversed and they bounce back to the boundary nodes. We point out that the distribution
functions at the end of bounce-back in this configuration is the post-collision distribution
functions.

Figure 5 Illustration of mid-grid bounce back
Although the on-grid bounce-back is easy to implement, it has been verified that it is only first-
order accurate due to its one-sided treatment on streaming at the boundary. However the centered
nature of the mid-grid bounce-back leads to a second order of accuracy at the price of a modest
complication.

Zou-He Velocity and Pressure BCs

In many physical situations, we would like to model flows with prescribed velocity or pressure
(density) at the boundary. This particular velocity/pressure BC we discuss here was originally
developed by Zou and He in. For illustration, we consider that the velocity ul = (u, v) is given on
the left boundary.
After streaming, f0, f2, f3, f4, f6 and f7 are known. What's left undetermined are f1, f5, f8 and .
The idea of Zou-He BCs is to formulate a linear system of f1, f5, f8 and .
Results and Discussion

Since the objective was to use lattice Boltzmann method for the rest of the project, a good
understanding of the model was necessary. Thus codes were written on various platform to test
the speeds and efficiency of different programming languages.

A simple one dimensional model was started with. This was coded in C++. The model was
D1Q3. The collision term and streaming functions were employed here. The code was very
simple.

The next thing that was done was a two dimensional model D2Q9 for flow around a cylinder.
This was done in python test the visualization libraries. Python has a very easy syntax and good
visualization libraries. The code is short and the output is very fast.


9


Figure 6 Simulation of flow around a cylinder

Then a diffusion problem was taken where a slab at uniform temperature is subjected to a
constant temperature at one of the walls. The temperature distribution was determined in the slab
after a time of 400 units. The computer codes can be found in the appendix.

After that the diffusion through a porous media is modeled where the concentration was
determined at various time steps with a two dimensional structure. The following graphs have
been plotted using the data generated. The following are generated with iterations of 400 and
1500 respectively.


Figure 7 Concentration distribution through a porous media with 400 iterations
10



Figure 8 Concentration distribution through a porous media with 1500 iterations


Future Plan

The future plan includes firstly adding convection term to the diffusion equation. A pressure
gradient would be applied across the ends of the porous media for the fluid to flow. Further,
particles sticking to the surface of the porous media would be modeled in equilibrium with the
particles being transported. Then flow in reservoirs would be looked at to include some practical
analysis into the model.














11

References

1. Mohamad, A. A., Lattice Boltzmann method, Fundamentals and Engineering
Applications with Computer Codes, Springer 2011
2. Nabovati, A., and Sousa, A. C. M., Fluid Flow Simulation in random porous media at
pore level using the Lattice Boltzmann method, Journal of Science and Technology,
Vol.2 No. 3 (2007)
3. Bao, Y. B., and Meskas, J., Lattice Boltzmann method for Fluid Simulations, 2011
4. Gerke, H. H., and MT van Genuchten. "A dualporosity model for simulating the
preferential movement of water and solutes in structured porous media." Water
Resources Research 29.2 (1993): 305-319.
5. Van Genuchten, M. T., & Wierenga, P. J. (1976). Mass transfer studies in sorbing porous
media I. Analytical solutions. Soil Science Society of America Journal, 40(4), 473-480.
6. Haggerty, R., & Gorelick, S. M. (1995). Multiplerate mass transfer for modeling
diffusion and surface reactions in media with porescale heterogeneity. Water Resources
Research, 31(10), 2383-2400.
7. Hornung, U., & Showalter, R. E. (1990). Diffusion models for fractured media. Journal
of Mathematical Analysis and Applications, 147(1), 69-80.
8. Khaled, A. R., & Vafai, K. (2003). The role of porous media in modeling flow and heat
transfer in biological tissues. International Journal of Heat and Mass Transfer, 46(26),
4989-5003.
9. Lenormand, R., Touboul, E., & Zarcone, C. (1988). Numerical models and experiments
on immiscible displacements in porous media. Journal of fluid mechanics, 189, 165-187.
10. Deans, H. A. (1963). A mathematical model for dispersion in the direction of flow in
porous media. Society of Petroleum Engineers Journal, 3(01), 49-52.