European Journal of Scientific Research

ISSN 1450-216X Vol.35 No.3 (2009), pp.347-354

© EuroJournals Publishing, Inc. 2009
http://www.eurojournals.com/ejsr.htm

Prediction of Natural Convection in a Square Cavity with

Partially Heated from Below and Symmetrical Cooling from
Sides by the Finite Difference Lattice Boltzmann Method

Nor Azwadi Che Sidik

Thermo-fluid Department, Universiti Teknologi Malaysia
UTM Skudai, Johor, Malaysia
E-mail: azwadi@fkm.utm.my
Tel: +60-7-5534718; Fax: +60-7-5566169

Abstract

In this paper, the lattice Boltzmann method, a numerical tool based on the particle
distribution function is used to simulate thermal fluid flow problem. A well-known finite
difference technique with third order accuracy in space is combined with the double
population thermal lattice Boltzmann model to solve two-dimensional natural convection in
a square cavity. The basic idea is to solve the velocity and the temperature field using two
different distribution functions. The simulation of localized heated bottom and symmetrical
cooling from the sides were carried out in order to validate this approach. The combination
of finite difference with double population thermal lattice Boltzmann model is found to be
an efficient and stable numerical approach for low and moderate Rayleigh number
calculations.

Keywords: Lattice Boltzmann, Double Population, Finite Difference, Natural Convection,

Localised Heating

1. Introduction
For more than a decade, lattice Boltzmann method (LBM) has been demonstrated to be a very effective
numerical tool for a broad variety of complex fluid flow phenomena that are problematic for
conventional method (Bernsdorf, Brenner and Durst, 2000; Nor Azwadi and Tanahashi, 2006).
Compared with traditional computational fluid dynamics, LBM algorithms are much easier to be
implemented especially in complex geometries and multicomponent flows (Michael, Orlandini, Orborn
and Yeomans, 1996).
Historically, LBM was derived from lattice gas (LG) automata (Frish, Hasslacher and Pomeau,
1986). It utilizes the particle distribution function to describe collective behaviors of fluid molecules.
Although promising, the current LBM still have a few shortcomings that limit its general application as
a practical computational fluid dynamics tool. One of these shortcomings, which are specifically
addressed in this paper, is low order accuracy of the current LBM model.
Generally, there are three types of thermal lattice Boltzmann models have been proposed;
multi-speed model (McNamara and Alder, 1993), passive scalar model (Shan, 1997) and double-
distribution function (DDF) model (He, Chen and Doolen, 1998). Among these models, the DDF
model is reported to be the most stable (Peng, Shu and Chew, 2003) and widely used in simulating
Prediction of Natural Convection in a Square Cavity with Partially Heated from Below and
Symmetrical Cooling from Sides by the Finite Difference Lattice Boltzmann Method 348

thermal fluid flow problems (Onishi, Chen and Ohahsi, 2001; Peng, Shu and Chew, 2002; Nor Azwadi
and Tanahashi, 2007).
In this paper, the DDF thermal LBM is combined with finite difference technique. The finite
difference technique is applied to solve the advection term in the governing equations of DDF thermal
LBM. This combination contributes in allowing us to increase the accuracy both in time and space. In
order to verify the proposed model, a natural convection of air, rectangular enclosure with localized
heating from below and symmetrical cooling from the sides is considered.

2. Double-Distribution Function Thermal LBM

The governing equation of the DDF thermal LBM are
∂f i ∂f
[ f i − f i eq ]+ F
1
+c i = (1)
∂t ∂x τ f
∂g i ∂g
[g i − g ieq ]
1
+c i = (2)
∂t ∂x τ g
where the density distribution function f = f(x,c,t) is used to simulate the density and velocity fields and
the internal energy density distribution function g = g(x,c,t) is used to simulate the macroscopic
temperature field. c in Eqs. (1) and (2) is the microscopic velocity and F is the external force. Noted
that the first term in the right hand side of Eqs. (1) and (2) is the collision term where the BGK
approximation (Bhatnagar, Gross and Krook, 1954) has been applied. In this approximation, τ is the
time to reach equilibrium condition during collision process and known as the time relaxation of
distribution function.
Macroscopic density ρ, velocity u and temperature T of the fluid are determined by the
following moments of the distribution functions
ρ = ∫ f eq dc , ρ u = ∫ cf eq dc , T = ∫ g eq dc (3)
eq eq
The equilibrium distribution functions, f and g are chosen such that the continuum
macroscopic equations approximated by the governing equation correctly describe the hydrodynamics
of the fluid. They are given as
⎡ 9 3 ⎤
f i eq = ρω i ⎢1 + 3c i ⋅ u + (c i ⋅ u ) − u 2 ⎥
2

⎣ 2 2 ⎦ (4)
g i = T ω i [1 + 3c i ⋅ u ]
eq
(5)
The lattice geometry for density and internal energy density distribution function are shown in
Fig. 1

Figure 1: (a) Nine-velocity Model, ω0 = 4/9, ω1~4= 1/9 and ω5~8=1/36 and (b) Four-velocity Model, ω1~4=1/4

(a) (b)
349 Nor Azwadi Che Sidik

The macroscopic continuity, momentum and energy equations can be obtained from Eqs. (1)
and (2) by using multi-scaling expansion procedure (Hou, Zou, Chen and Doolen, 1995). As a result,
the viscosity and thermal diffusivity in these models are related to the time relaxations as below
τf
υ=
3 (6)
χ = τg (7)

3. Finite Difference DDF Thermal LBM

In the finite difference formulation of LBM, Eqs. (1) and (2) are once again discretized in phase space
and time and solved using one of the several available finite difference numerical schemes. In this
paper, the temporal discretization is obtained using second order Rungge-Kutta (modified) Euler
method. The time evolution of particle distributions is then derived by

fi
n+
1
2
= fi + n Δt ⎡
⎢− c i ⋅ ∇ f i +
n ( f i eq , n − f i n )⎤
⎥
2 ⎣⎢ τf ⎦⎥
(8)
⎡ ⎛ eq , n + 1 n + ⎞⎤
1
⎢ ⎜ fi 2
− fi 2 ⎟⎥
⎝ ⎠⎥
1
f i n +1 = f i n + Δt ⎢−c i ⋅ ∇f i 2 +
n+

⎢ τf ⎥ (9)
⎢ ⎥
⎣ ⎦
The third order upwind scheme was applied to calculate the advection term in the evolution of
density and temperature equations. From this combination, the accuracy of the finite difference DDF
thermal LBM is upgraded to second order in time and third order in space.

4. Numerical Results
In this section, the proposed model is applied to simulate the phenomenon of natural convection in a
square cavity with localized heating from below and symmetrical cooling from the sides. Symmetrical
cooling from the sides is expected to be an efficient cooling option, while partial heating at the lower
surface simulates the electronic components such as chips.
Fig. 2 shows a schematic diagram of the setup in the simulation. The lower wall has a various
centrally located heat source, l = H/5, 2H/5, 3H/5 and 4H/5, which is assumed to be isothermally
heated at a constant temperature, TH.
The Boussinesq approximation is applied to the buoyancy force term. So the external force in
Eq. (1) is
F = 3G (c − u ) f i eq (10)
The dynamical similarity depends on two dimensionless parameters: the Prandtl number Pr and
the Rayleigh number Ra,
υ
Pr =
χ (11)
g β ΔTH 3
Ra =
υχ (12)
In all simulations, Pr is set to be 0.71 in order to simulate air cooling of electronic components.
Prediction of Natural Convection in a Square Cavity with Partially Heated from Below and
Symmetrical Cooling from Sides by the Finite Difference Lattice Boltzmann Method 350

Figure 2: Geometry and Boundary Conditions of the Problem

Figure 3: Velocity Vectors for Ra = 103 and l = H/5

Fig. 3 shows velocity vectors for Ra = 103 and l = H/5. They describe that the hot fluid rises
above the source until it reaches the top wall, then moves upwards along the horizontal wall before
moving downwards along the sidewalls under the effect of cooling. The same phenomenon is observed
for every simulated condition.
The main characteristics of natural convection flow are shown in terms of streamlines and
isotherm in Figs. 4 to 7. For each case, 30 equally spaced contours have been used to define the
corresponding structure.
351 Nor Azwadi Che Sidik
Figure 4: Streamlines for (a) Ra = 103, (b) Ra = 104 and (c) Ra = 105

(a) (b) (c)

Figure 5: Isotherms plots for Ra = 103 (a) l = H/5 (b) l = 2H/5 (c) l = 3H/5 and (d) l = 4H/5

(a) (b)

(c) (d)

The flow pattern is characterized by primary flows that rotate clockwise in the right half and
counter-clockwise in the left half of the cavity as shown by streamline plots in Fig. 4. Owing to the
symmetry, the flow in the left and right halves of the enclosure is identical except for the sense of
rotation. Owing to the symmetrical boundary condition, the flow reaches an asymptotic steady state
exhibiting a symmetric motion about the vertical centerline of the cavity for all values of Ra and l
considered. Due to the same pattern shown by all the simulated conditions, only results obtained for l =
H/5 are brought as references.
Figs. 5 to 7 show the isotherm plots for each case considered.
Prediction of Natural Convection in a Square Cavity with Partially Heated from Below and
Symmetrical Cooling from Sides by the Finite Difference Lattice Boltzmann Method 352

Figure 6: Isotherms plots for Ra = 104 (a) l = H/ (b) l = 2H/5 (c) l = 3H/5 and (d) l = 4H/5

(a) (b)

(c) (d)

At low Rayleigh number, Ra = 103, the isotherms deviate slightly from a diagonally symmetric
structure indicates that the convection intensity is very weak and the main heat transfer mechanism is
by conduction. As the Rayliegh number increased, Ra = 104, the intensity of the recirculation inside the
cavity increase, and the cores of the cells move upwards. The effect of convection can be seen from
isotherms lines. At Ra = 105, the formation of thermal boundary layers can be observed, because of
increased recirculation intensity. For all values of Ra considered, increasing l for a fixed Ra results in
an increase of average Nusselt number (not shown). All of these observations are in good agreement
with the results reported in the previous studies (Tsutahara, 1999; Nor Azwadi and Rosdzimin, 2008;
Corvaro and Paroncini, 2008).
353 Nor Azwadi Che Sidik
Figure 7: Isotherms plots for Ra = 105 (a) l = H/5 (b) l = 2H/5 (c) l = 3H/5 and (d) l = 4H/5

(a) (b)

(c) (d)

5. Summary and Concluding Remarks

In this paper, the natural convection in a square cavity with localized heating from below and
symmetrical cooling from the sides has been studied using finite difference double-distribution
function thermal lattice Boltzmann model. The evolution of lattice Boltzmann equation has been
disrcetised using the third order accuracy finite difference upwind scheme. The flow pattern including
vortices and thermal boundary layer can be seen clearly. The results obtained demonstrate that this
approach is very efficient procedure to study flow and heat transfer in a differentially heated cavity
flow. The extension to 3D computations and high Rayleigh numbers is the subject of further
investigations within ongoing research.

Acknowledgment
The author would like to thank Universiti Teknologi Malaysia and Malaysia government for
supporting this research activity.
Prediction of Natural Convection in a Square Cavity with Partially Heated from Below and
Symmetrical Cooling from Sides by the Finite Difference Lattice Boltzmann Method 354

References
[1] Bernsdorf, J., G., Brenner, and F., Durst, 2000. “Numerical Analysis of the Pressure Drop in
Porous Media Flow with Lattice Boltzmann (BGK) Automata”, Computer Physics
Communications 129, pp. 247-255.
[2] Bhatnagar, P. L., E. P., Gross, and M., Krook, 1954. “A Model for Collision Processes in
Gasses. 1. Small Amplitude Processes in Charged and Neutral One-Component System”,
Physics Review 94, pp. 511–525.
[3] Corvaro, F., and M., Paroncini, 2008. “A Numerical and Experimental Analysis on the Neutral
Convective Heat Transfer of a Small Heating Strip Located on the Floor of a Square Cavity”,
Applied Thermal Engineering 28, pp. 25-35.
[4] Frish, U., B., Hasslacher, and Y., Pomeau, 1986. “Lattice Gas Automata for the Navier-Stokes
Equation”, Physics Review Letter 56, pp. 1505-1508.
[5] He, X., S., Chen, and G., Doolen, 1998. “A Novel Thermal Model for the Lattice Boltzmann
Method in Incompressible Limit”, Journal of Computational Physics 146, pp. 282-300.
[6] Hou, S., Q., Zou, S., Chen and G., Doolen, 1995. “Simulation of Cavity Flow by the lattice
Boltzmann Method”, Journal of Computational Physics 118, pp. 329-347.
[7] McNamara, G., and B., Alder, 1993. “Analysis of the Lattice Boltzmann Treatment of
Hydrodynamics”, Physica A 194, pp. 218-228.
[8] Michael, R. S., S. F., Orlandini, W.R., Osborn and J. M., Yeomans, 1996. “Lattice Boltzmann
Simulations of liquid-Gas and Binary Fluid Systems”, Physical Review E 54, pp. 5041-5052.
[9] Nor Azwadi, C.S., and T., Tanahashi, 2006. “Simplified Thermal Lattice Boltzmann in
Incompressible Limit”, International Journal of Modern Physics B 20, pp. 2437–2449.
[10] Nor Azwadi, C.S., and T., Tanahashi, 2007. “Three-Dimensional Thermal Lattice Boltzmann
Simulation of Natural Convection in a Cubic Cavity”, International Journal of Modern Physics
B 21, pp. 87–96.
[11] Nor Azwadi, C.S., and A. R. M., Rosdzimin, 2008. “Simulation of Natural Convection Heat
Transfer in an Enclosure using Lattice Boltzmann Method”, Jurnal Mekanikal 27, pp. 42–50.
[12] Onishi, J., Y., Chen and H., Ohashi, 2001. “Lattice Boltzmann Simulation of Natural
Convection in a Square Cavity”, JSME International Journal Series B 44, pp. 53-62.
[13] Peng, Y., C., Shu and Y. T., Chew, 2003. “Simplified Thermal lattice Model for incompressible
Flows”, Physical Review E 68, pp. 026701-026708.
[14] Peng, Y., C., Shu and Y. T. Chew, 2003. “A 3D Incompressible Thermal Lattice Boltzmann
Model and its Application to Simulate Natural Convection in a Cubic Cavity, Journal of
Computational Physics 193, pp. 260-274.
[15] Tsutahara, M, 1999. “Lattice Gas Method. Lattice Boltzmann Method (In Japanese)”, Corona
Publishing, Tokyo.