Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 102 (2015) 1520 – 1525

The 7th World Congress on Particle Technology (WCPT7)

A Coupling Model of DEM and LBM for Fluid Flow through

Porous Media
Liu-Chao Qiu*
College of Water Resources and Civil Engineering, China Agriculture University, Beijing, 100083, China

Abstract

A two-dimensional coupling model of the discrete element method (DEM) and the lattice Boltzmann method (LBM) was used to study the fluid
flow through porous media. The fluid flow was modeled at a mesoscale level by using the LBM. The particle motion was modeled at microscale
level by means of the DEM. The interfacial interaction between the fluid and the solids is resolved by an implicit immersed boundary method
(IBM). The proposed method has validated by comparing the numerical results with theoretic solutions in the literature. It demonstrates the
ability of the proposed method to simulate fluid flows through complex porous media.
© 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
© 2014 The Authors. Published by Elsevier Ltd. Selection and peer-review under responsibility of Chinese Society of Particuology, Institute of
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Process Engineering,
Selection Chinese
and peer-review underAcademy of Sciences
responsibility (CAS).
of Chinese Society of Particuology, Institute of Process Engineering, Chinese Academy
of Sciences (CAS)
Keywords: Discrete Element Method; CFD; Lattice Boltzmann Method; Porous Media Flow; Immersed Boundary Method

1. Introduction

Fluid flows through porous media are found in many industrial processes ranging from filtration to petroleum
exploration and recovery and have been subjected to considerable scientific investigation. Substantial efforts have
been made in the past to develop numerical techniques for modelling fluid flows in porous media. The techniques
applied require generally two essential steps including the numerical construction of the packed particle sample
(solid skeleton) and the modelling of the fluid flow in the porous spaces. Due to the fluid–solid interaction in porous
media flow occurs at the pore scale, it seems more natural to simulate the process from a microscopic or mesoscopic
perspective. To this end, we develop a LBM–DEM coupling scheme to simulate the porous media flow. In this

* Corresponding author. Liu-Chao Qiu, Tel.: +86-010-62736040.

E-mail address: qiuliuchao@cau.edu.cn

1877-7058 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Selection and peer-review under responsibility of Chinese Society of Particuology, Institute of Process Engineering, Chinese Academy of Sciences (CAS)
doi:10.1016/j.proeng.2015.01.286
 Liu-Chao Qiu / Procedia Engineering 102 (2015) 1520 – 1525 1521

model, fluid flow in the pore structure is explicitly modeled using the LBM [1]; the geometrical representation and
the mechanical behavior of the solid skeleton are simulated by the particulate DEM [2]; the interfacial interaction
between the fluid and the solids is resolved by an IBM [3]. The key issue in the IBM is the computation of restoring
force. The widely used way is the penalty method [3, 4]. This method introduces a user-defined spring parameter
which may have a significant effect on the computational efficiency and accuracy. To avoid using the user-defined
parameter, Feng and Michaelides [5] propose the direct forcing method. As direct forcing method needs to solve N-
S equations to compute the force at the boundary point, it may spoil the merits of LBM when IBM is combined with
it. Recently, a simple method for computing the restoring force was proposed by Niu et al [6], in which the
momentum exchange at the boundary is used to compute the force. Within the conventional IBM, the forcing term is
computed explicitly in advance. However, the no-slip condition at the walls cannot be guaranteed, and streamline
penetration may be observed. In order to overcome this drawback, Wu and Shu [7] introduced an implicit IBM in
which the forcing term is treated as an unknown and determined by enforcing the non-slip boundary condition. In
this study, the DEM and implicit IBM are coupled to simulate fluid flows in porous media.

2. Methodology

2.1. Implicit IB-LBM

The implicit IB-LBM developed in the work of Wu and Shu [7] is based on the scheme proposed by Guo et al. [8]
which can be written as
 eq
fD x  eD Gtt  Gt  fD xt  fD xt  fD xt  FD Gt , (1)
W

 e u e ˜u
FD  wD D   D  ˜ eD ˜ f , (2)
W cs cs


Uu ¦
D
eD f D 

fGt , (3)

Here f is the force density at the Eulerian point; W is the single relaxation parameter; fD is the distribution function; eD
is the particle velocity; wa are the coefficients in the equilibrium distribution function; cs =c/  is the sound speed
of the model, where c = Gx/Gt, Gx and Gt are the lattice spacing and time step. For the case of Gx=Gt, c is taken as 1.
For the D2Q9 model used in this work, the corresponding equilibrium distribution function is
 
eq eD ˜ u eD ˜ u  cs _ u _
fD UwD>   @, (4)
cs cs

The physical velocity at a fluid node can be written as

u u  Gu , (5)
 
where u
U ¦
D
eD fD is intermediate velocity and Gu
U
fGt is the velocity correction.

In the IBM, the boundary of rigid body is represented by a set of Lagrangian points X lB , l = 1, 2, …., m (see Fig.

1). An unknown velocity correction vector GUlB is set at every Lagrangian point. The velocity correction Gu at the
1522 Liu-Chao Qiu / Procedia Engineering 102 (2015) 1520 – 1525

Eulerian point can be obtained by Gu ¦ GU

l
l
B Dij x ij  X lB 'Sl  l m , where ƸSl is the arc length

of the boundary element, Dij is the continuous kernel distribution proposed by Peskin [3].

Fig. 1. Immersed boundary illustration.

To satisfy the non-slip boundary condition, the following equation system must be satisfied

U lB ¦ u D 'x'y  ¦ ¦ GU
ij
ij
ij l
l
B Dij 'Sl Dij 'x'y, (6)

which can be further re-written as the following matrix form

AX B, (7)

the elements of matrix A are only related to the Lagrangian boundary points and their neighboring Eulerian points.
This matrix can be inverted in order to solve for the unknown velocities Gu . When these are known, the forces in the
Eulerian frame can be obtained. The detailed solution procedure of present approach can be found in reference [7].

2.2. DEM

In fluid-particle systems, the translational and rotational motions of a particle can be described individually
according to Newton’s second law:
dU
M FC  Ffp  Mg (8)
dt

dΩ
I TC  T fp (9)
dt

in which t is the timeˈM , I, U, : are the mass, momentum of inertia, linear velocity, and angular velocity,

respectively, of the solid particle. g is the body force (gravity) per unit mass, Fc, Tc are the contact force and torque

due to particle-particle interaction respectively. Ffp, Tfp are respectively the force and torque due to fluid-particle

interaction. In the contact force calculation, the Hertz theory [9] is used in modelling the normal force–
 Liu-Chao Qiu / Procedia Engineering 102 (2015) 1520 – 1525 1523

displacement relationship while the Mindlin and Deresiewicz theory [10] is employed for the tangential force–

displacement calculations. The fluid–particle interaction force on a single solid particle given by F fp ³
 fds and
*

Tfp  ³
*
X B  X C u f ds , where XC is centre of the circle particle, *is the boundary of solid particle.

3. Numerical examples

3.1. Flows over a circular cylinder

To verify the developed LBM-DEM code for fluid-particle interaction, we modeled a flow past a cylinder of
radius=0.5 as that shown in Fig. 2. The computational domain is taken as 50 × 40 and the corresponding LBM grid
is 500 × 400. The cylinder is located at (20, 20). The cylinder surface is represented by 120 Lagrangian points with
uniform distribution. The fluid density is taken as U=1.0 and the free stream velocity is UĞ=0.1, The computation
starts with the given free stream velocity. At the far field boundaries the equilibrium distribution functions are used
to implement the boundary conditions. The Reynolds number Re=UĞD/v used in this case is 100 .
In the present work, the force can be directly computed from the velocity correction. Fig. 3 shows the time
evolution of the drag coefficient on the surface of cylinder. We can see that the present time-averaged drag
coefficient cd=1.365 agrees with the second-order result of Lai and Peskin [11]. Fig. 4 shows the streamlines. Fig. 5
shows the instantaneous vorticity contours of vortex shedding, where the dotted and solid lines denote the negative
and positive levels of vorticity respectively. The Karman vortex street of the flow over a circular cylinder can be
clearly seen in these figures.
1.37
Cd

1.36

150 160 170 180 190 200

Dimensionless time

Fig. 2. Flow over cylinder setup. Fig. 3. Evolution of drag coefficients.

Fig. 4. Streamlines. Fig. 5. Vorticity contours.

3.2. Pore-scale fluid flow through idealized porous media

In this section, we simulated the pore-scale fluid flow in idealized porous media using the proposed method. A
channel of height H and width 4H is filled with cylindrical objects of diameter H/8 and distance H/8 as that shown
in Fig. 6. The left inlet, top and bottom walls are subject to Dirichlet condition by imposing a velocity U0=1. The
1524 Liu-Chao Qiu / Procedia Engineering 102 (2015) 1520 – 1525

right outlet has a prescribed reference pressure P 0=0. The fluid density is taken as U=1000. The computational
domain is discretized by 256 x 64 LBM grids. Each cylinder surface is represented by 24 Lagrangian points with
uniform distribution. Fig. 7 shows the pressure contours with streamlines. Fig. 8 shows two pressure plots along two
different horizontal lines. The results have a well agreement with that of Geller et al. [12].

Fig. 6. Definition of porous flow setup. Fig. 7. Pressure contours with streamlines.

Fig. 8. Pressure plots along horizontal lines : y =0.5H (left) and y =0.625H (right).

4. Conclusions

In this paper, we proposed an approach to investigate fluid flow through porous media using the discrete element
method for modeling particles dynamics and the lattice Boltzmann method for simulating the fluid motion. The
interfacial interaction between the fluid and the solids is resolved using an implicit immersed boundary method.
Some preliminary results of simulation of porous flows using the proposed LBM-DEM model are presented. It is
demonstrated that the developed lattice Boltzmann–discrete element coupling system is a powerful fundamental
research tool for investigating porous media flow. More research in this direction is required. Our future work will
address the issue of hardware acceleration using GPU technology.

Acknowledgements

The financial support of National Science Foundation of China (Grant No. 11172321) and Chinese Universities
Scientific Fund (Grant No. 2011JS047) is highly appreciated.

References

[1] Y. Qian, D. d’Humieres, P. Lallemand. Lattice BGK models for Navier–Stokes equation. Europhysics Letters, 1992, 17(6):479–484.
[2] P.A. Cundall, O.D.L Strack. A discrete numerical model for granular assemblies. Geotechnique, 1979, 29(1):47–65.
[3] C.S. Peskin, Numerical analysis of blood flow in the heart, J. Comp. Phys. 25 (1977) 220–252.
[4] Z. Feng, E. Michaelides, The immersed boundary-lattice Boltzmann method for solving fluid-particles interaction problems, J. Comp. Phys.
195 (2004) 602–628.
[5] Z. Feng, E. Michaelides, Proteus: a direct forcing method in the simulations of particulate flows, J. Comp. Phys. 202 (2005) 20–51.
[6] X.D. Niu, C. Shu, Y.T. Chew, Y. Peng, A momentum exchange-based immersed boundary-lattice Boltzmann method for simulating
incompressible viscous flows, Phys. Lett. A 354 (2006) 173–182.
 Liu-Chao Qiu / Procedia Engineering 102 (2015) 1520 – 1525 1525

[7] J. Wu and C. Shu. Implicit velocity correction-based immersed boundary-lattice Boltzmannmethod and its applications. J. Comput. Phys.,
228(6):1963-1979, 2009.
[8] Z. Guo, C. Zheng, B. Shi, Discrete lattice effects on the forcing term in the lattice Boltzmann method, Phys. Rev. E 65 (2002) 046308.
[9] K.L. Johnson. Contact Mechanics. Cambridge: Cambridge University Press; 1985.
[10] R.D. Mindlin, H. Deresiewicz. Elastic spheres in contact under varying oblique forces. ASME J Appl Mech. 1953;20:327–344.
[11] M.C. Lai, C.S. Peskin, An immersed boundary method with formal second-order accuracy and reduced numerical viscosity, J. Comp. Phys.
160 (2000) 705–719.
[12] S. Geller, M. Krafczyk, J. Toelke, S. Turek, and J. Hron. Benchmark computations based on lattice-Boltzmann, finite element and finite
volume methods for laminar flows. Computers & Fluids, 35 (2006) 888–897.