Vous êtes sur la page 1sur 18

Chapter 5

Numerical Applications of TLLBM

Part I: Flows in 2D Lid-Driven Cavity*

5.1 Background

Driven cavity flows are industrially important because these flows and the

structures that they exhibit play an important role in industry. For example, Aidun et al.

(1991) pointed out the direct relevance of cavity flows to coaters, and in melt spinning

processes used to manufacture microcrystalline material. The eddy structures found in

driven cavity flows give insight into the behavior of such structures in applications as

diverse as drag-reducing riblets and mixing cavities used to synthesize fine polymeric

composites (Zumbrunnen et al. 1995). However, in our view the greatest importance of

these flows is in the basic study of fluid mechanics. In no other class of flows are the

boundary conditions so unambiguous. Consequently, driven cavity flows offer an ideal

framework in which meaningful and detailed comparisons can be made between results

*
The works in this chapter have been published by:
1. C. Shu, X. D. Niu and Y. T. Chew, “Taylor series expansion- and least square-
based lattice Boltzmann method: two-dimensional formulation and its applications”, Physical
Review E., vol. 65, 036708, p1-13, 2002, The American Physical Society.
2. Y. T. Chew, C. Shu and X. D. Niu, “Simulation of Unsteady Incompressible Flows by Using Taylor
Series Expansion- and Least Square-Based Lattice Boltzmann Method”, International Journal of
Modern Physics C, vol.13, No. 6, p719-738, 2002, World Scientific Publishing Company.
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 86

obtained from experiment, theory, and computation. In fact, as hundreds of papers

attest, the driven cavity problem is one of the standards used to test new computational

schemes. We here mention only Benjamin & Denny (1979), Ghia et al. (1982), Fuchs

& Tillmark (1985), Soh et al. (1988), Zang et al. (1993) and Ku et al. (1987). Another

great advantage of this class of flows is that the flow domain is unchanged when the

Reynolds number is increased. This greatly facilitates investigations over the whole

range of Reynolds numbers, 0<Re<∞. Thus the most comprehensive comparisons

between the experimental results obtained in a turbulent flow (Prasad & Koseff 1989)

and the corresponding direct numerical simulations (DNS) (Deshpande & Shankar

1994a, b; Verstappen & Veldman 1994) have been made for a driven cubical cavity.

Furthermore, driven cavity flows exhibit almost all phenomena that can possibly occur

in incompressible flows: eddies, secondary flows, complex 3D patterns, chaotic

particle motions, instabilities, transition, and turbulence. As a striking example, it was

in such flows that Bogatyrev & Gorin (1978) and Koseff & Street (1984b) showed,

contrary to intuition, that the flow was essentially 3D, even when the aspect ratio was

large. In this sense, cavity flows are almost canonical and will continue to be

extensively studied and used.

In this chapter, the proposed Taylor-series-expansion- and least-square-based

LBM (TLLBM) is applied to simulate the lid-driven cavity flows including 2D steady

polar and square cavity flows, and 2D unsteady square cavity flow. The numerical

work of Ghia et al. (1982), numerical and experimental work of Fuchs & Tillmark

(1985) and numerical work of Zang et al. (1994) are used as benchmarks to evaluate

the present results.


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 87

5.2 Flows in A Square Cavity

In this case, the top lid moves from left to right with a constant velocity U.

Non-uniform meshes of 49×49 for Re = 100 and 400, 97×97 for Re = 1000, 145×145

for Re = 5000 and 10000 were used respectively for the calculation. The Reynolds

number was defined as Re = UL/υ (based on the lid velocity and the length of the

square cavity). The use of non-uniform mesh is desirable, especially for the cases of

high Reynolds number. This is because the thin boundary layer is attached to the solid

boundaries. So, to capture the thin boundary layer, the mesh spacing near the wall

should be very small. Apart from the solid wall, relatively large mesh spacing was

used. In this way, we can correctly simulate the thin boundary layer, and at the same

time, we can save the computational effort.

Initially a constant density, ρ =1, was prescribed for the whole cavity field, and

the velocities in the interior of the cavity were set to zero. On the top, the x-component

velocity is U, which was set to 0.15, and the y-component velocity is zero. At the end

of each time step, the density distribution function fα at the top was set to the

equilibrium state. The whole halfway wall bounce back boundary conditions were used

on the other three solid walls. For the upper two corners between the stationary wall

and the moving wall, which are singular points, it was found that treatment with the

moving wall or the stationary wall points had little difference in our simulations.

Fig. 5.1 showed the streamlines for different Reynolds numbers. The effects of

the Reynolds number on the flow pattern and the structure of the steady recirculating
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 88

vortices were clearly seen in this figure. Initially, the center of the primary vortex,

which is located 0.24 below the lid in the mid-plane, moved a little lower and to the

right when Re = 100 (Fig. 5.1(a)). But it was found that, for Re = 400 (Fig. 5.1 (b)), the

center of the primary vortex moves lower and back towards the center plane; and, as

Fig. 5.1 showed, as Re increases further, there is uniform tendency for the vortex

center to move towards the geometric center of the cavity.

To facilitate the discussion of the secondary vortices, we designated them

bottom right, bottom left, and top left; they were simplified as BR1, BR2, ..., BL1, BL2,

..., TL, where the subscripts indicated, except for TL, the member in a presumably

infinite sequence. As Re increases, although both BR1 and BL1 grew in size, BR1's

growth is greater, as are their strength (as can be seen from Fig. 5.1). The trajectory of

the vortice centers is complex, with the distance above the cavity bottom of the center

of BL1 being actually greater than that of BR1 for Re = 5000. Fig. 5.1(e) also showed

the growth of BR2 and BL2, which are small and weak at Re = 5000.

The emergence of the upper upstream vortex (TL1) represents a genuine change

in flow topology. Hints of its imminent appearance can be seen in the streamline

patterns at Re = 1000 (Fig. 5.1 (c)). Having emerged at a Reynolds number of around

1200 (Benjamin & Denny 1979), it grew in size and strength at least until Re = 10,000

(Fig. 5.1(e)). One must note that this secondary vortex, attached to a plane wall, is

quite different in character from the lower-corner vortices.


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 89

It should be clear from the above that even the two-dimensional flow in a cavity

of simple geometry can be complex and it would be very difficult to qualitatively

predict the changes that are likely to take place as the Reynolds number increases. The

vorticity contours of Fig. 5.2 provide insight into some general features of the flow

field as the Reynolds number increases. As Re→∞, one would expect thin boundary

layers to develop along the solid walls, with the central core in almost inviscid motion.

This was indeed seen in the figure. As Re increases, there is a clearly visible tendency

for the core fluid to move as a solid body with uniform vorticity. Apparently, the flow

structures agree very well with those of Ghia et al. (1982).

Table 5.1 gave the detailed comparison for locations of the vortex center

obtained by the present method and Ghia et al. (1982). The relative errors between the

two solutions were less than 4%. In this table, we also gave the CPU time (seconds)

spent in the present computation on the personal computer with Pentium III 866 and

364M memory. Although other information about this aspect is lacking, we still

believe, from our numerical simulations, that the computational efficiency of present

method was as good as that of the traditional CFD tools. The U and V velocities along

their respective central line were displayed in Fig. 5.3 for different Reynolds numbers.

Obviously, our simulation results are in good agreement with those of Ghia et al.

(1982).
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 90

5.3 Flows in A Polar Cavity

The polar cavity case was used to show the capability of the present method in

treating the flow problem with complex geometry. The geometry with a non-uniform

mesh was given in Fig. 5.4. The Reynolds number (based on the lid velocity and the

radius of the inner circle) was 350. Initially a clockwise velocity of U = 0.15 was set

on the inner lid and other conditions were the same as those in the square cavity case.

Fig. 5.5 showed the steady state azimuthal and radial velocity profiles along the line of

θ = 0. Results, which were obtained on 49×49, 65×65 and 81×81 non-uniform meshes,

together with the experimental and numerical results of Fuchs and Tillmark (1985) are

included in the figure for comparison. The results obtained by the present method

agreed well with those of Fuchs and Tillmark’s numerical simulation. The present

solutions also compared well with the experimental data and the discrepancy between

them may be attributed to the three-dimensional effect in the experiments in which

three-dimensional flow structures are observed. In Fig. 5.6, the streamlines obtained by

the present method on the mesh size of 81×81 were compared with those obtained

from the solution of Navier-Stokes equations (Zang et al. 1994). Again, good

agreement was achieved in the size of the vortices and location of the separation and

reattachment points.

5.4 Oscillating Flow in Lid Driven Cavity at Re = 400

The study of oscillating flow in a lid driven cavity poses a challenge to

researchers in fluid mechanics. In this case, a periodic velocity waveform was imposed

on the cavity lid and the time evolution observed in the flow was compared with the
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 91

numerical results of Soh et al. (1988) using the N-S equations. The periodic lid

velocity was given by a sinusoidal waveform:

u(t)=Ucos(ωt) (5.1)

where U is the maximum lid velocity during the cycle, ω is the frequency of the

oscillation and t is the time. The period of oscillation, T, is related to the frequency by

T=2π/ω. (5.2)

The simulation was performed with Re = 400 (Re = UL/ν, where L is the

characteristic length set as the length of top wall), frequency of ω = 1 and U = 0.15. A

non-uniform grid of 97×97 with denser distribution near the boundaries was used, and

the time step was set as, in the unit of L/U, 4.5×10-4. The oscillatory flow was assumed

to reach the periodic steady state when the differences of the velocity components u

and v of each point in the domain at two subsequent flow cycles are within a small

tolerance of ε=10-5.

The flow reached steady state after 7 cycles. Fig. 5.7 showed the time evolution

of the viscous drag on the lid, which was estimated using the same formula as given by

Soh et al. (1988). It can be seen that the drag settles down to be periodic very quickly,

more quickly than the entire flow field. The maximum drag occurs approximately at

phase of 15.96° before the lid velocity reached its peak.


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 92

The instantaneous streamlines obtained by present method and those of Soh et

al. at t = χT where χ = 0.2, 0.3, 0.35, 0.4, 0.45, 0.5, 0.7, 0.8, 0.85, 0.9, 0.95 and 1.0

were shown in Fig. 5.8-1 and Fig. 5.8-2, respectively. As time advances, the direction

of the lid movement and the center of the vortex changes. The lid velocity passes

through zero at t = T/4, after which it reverses direction. As a result, at t = 0.3T a

counter rotating vorticity is formed in the flow field at the left corner of the cavity. As

the magnitude of the velocity increases in the negative x direction, the size of the

second vortex created in the upper left corner of the cavity also increases. At the same

time, the primary vortex continuously shrinks until t = T/2. At this point, the velocity

reaches its maximum in the negative x direction and the second vortex, which has

formed in the left corner of the cavity, attains its maximum size and occupies the entire

domain.

After this point, the streamlines at each time step are the mirror images of the

streamlines at time from t = 0 to T/2. This conclusion can be made by comparing Fig.

5.8-1 and Fig. 5.8-2. The results are in good agreement with those of Soh et al. (1988).

5.5 Concluding Remarks

The 2D steady flows in a lid-driven square cavity at different Reynolds

numbers were first investigated by the new TLLBM. The present numerical

computation confirmed the flow features in the cavity domain obtained in numerical

and experimental studies by other researchers.


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 93

Simulations of 2D steady flows in a lid-driven polar cavity showed that our

TLLBM can be applied to the problems with complicated geometrical boundary which

cannot be done by the standard LBE.

Since the LBE can be recovered to the unsteady Navier-Stokes equations, it

should be able to simulate the unsteady problems and this has been confirmed by

applying our TLLBM to the oscillatory flow in a lid-driven square cavity.


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 94

Table 5.1 Comparison for locations of primary vortex center of the lid-driven
square cavity at different Reynolds numbers

Reynolds number 100 400 1000 5000 10000

Ghia
(0.61,0.73) (0.56,0.61) (0.54,0.56) (0.52,0.54) (0.51,0.51)
Vortex et al.
Center Present
(0.61,0.73) (0.56,0.60) (0.54,0.56) (0.53,0.56) (0.51,0.52)
method
CPU (seconds) by
present method 195.521 600.3833 3567.650 20443.85 64401.57
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 95

(a) Re = 100, 49×49 (b) Re = 400, 49×49 (c) R e= 1000, 97×97

(d) Re = 5000, 145×145 (e) Re = 10000, 145×145

Figure 5.1 Streamlines of the lid-driven square cavity flows at different Reynolds
numbers
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 96

(a) Re = 100, 49×49 (b) Re = 400, 49×49 (c) Re = 1000, 97×97

(d) R e= 5000, 145×145 (e) Re = 10000, 145×145

Figure 5.2 Vorticity contours of the lid-driven square cavity flows at different
Reynolds numbers
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 97

1 0.6
Ghia's data
0.8 0.4
Present
0.2 result
0.6
0
0.4 Ghia's data
-0.2
Present
0.2 result -0.4

0 -0.6
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
U-Y X-V

(a) Re = 100, 49×49

1 0.6
Ghia's data
0.8 0.4
Present
0.2 result
0.6
0
0.4 Ghia's data
-0.2
Present
0.2
result
-0.4
0
-0.6
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1
0 0.2 0.4 0.6 0.8 1
U-Y X-V

(b) Re = 400, 49×49

1 0.6
Ghia's data
0.4
0.8 Present
0.2 result
0.6
0
0.4 Ghia's data
-0.2
Present
0.2 -0.4
result

0 -0.6
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
X-V
U-Y

(c) Re = 1000, 97×97


Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 98

1 0.6
Ghia's data
0.4
0.8 Present
0.2 result
0.6
0
0.4 Ghia's data
-0.2
Present
0.2 -0.4
result
-0.6
0
0 0.2 0.4 0.6 0.8 1
-0.5 -0.25 0 0.25 0.5 0.75 1
X-V
U-Y

(d) Re = 5000, 145×145

1 0.6
Ghia's result
0.8 0.4
Present Result
0.2
0.6 Ghia's
result 0
0.4 Present
-0.2
result
0.2
-0.4

0 -0.6
-0.5 0 0.5 1 0 0.2 0.4 0.6 0.8 1
U-Y X-V

(e) Re = 10000, 145×145

Figure 5.3 U (left) and V (right) velocity profiles along vertical and horizontal
central lines of the square cavity at different Reynolds numbers
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 99

Figure 5.4 Geometry and a typical non-uniform mesh for the flow in a polar lid-
driven cavity

0.75

ur / U 0.5

uθ / U
0.25

-0.25

-0.5
0 0.2 0.4 0.6 0.8 1

( r − r0 ) / r0

Figure 5.5 Comparison of radial ( ur ) and azimuthal ( uθ ) velocity profiles along


the line of θ = 0 0 in the polar cavity at Re = 350 („ numerical data by Fuchs &
Tillmark; experimental data by Fuchs & Tillmark; — Present result of 49×49; – – –
Present result of 65×65; — – — Present result of 81×81)
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 100

(a) Present result (81×81) (b) N-S solution of Zang et al. (1994)

Figure 5.6 Comparison of streamlines in the polar cavity between the present
method and the N-S Solver

40 40

30 30

20 20

10 10
Drag

Drag

0 0

-10 -10

-20 -20

-30 -30

-40 -40
0 0.5 1 t/T 1.5 2 5 5.5 6 t/T 6.5 7

(a) Initial evolution (b) Final stage of evolution

Figure 5.7 Time evolution of drag on the cavity lid for oscillatory flow
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 101

(a) t=0.2T (b) t=0.3T (c) t=0.35T

(a’) t=0.2T (b’) t=0.3T (c’) t=0.35T

(d) t=0.4T (e) t=0.45T (f) t=0.5T

(d’) t=0.4T (e’) t=0.45T (f’) t=0.5T

Figure 5.8-1 Instantaneous streamlines in the first half period for the oscillatory lid-
driven cavity flow (1st and 3rd rows) and Soh et al. (2nd and 4th rows)
Chapter 5 Numerical Applications of TLLBM Part I: Flow in 2D Lid-Driven Cavity 102

(a) t=0.7T (b) t=0.8T (c) t=0.85T

(a’) t=0.7T (b’) t=0.8T (c’) t=0.85T

(d) t=0.9T (e) t=0.95T (f) t=T

(d’) t=0.9T (e’) t=0.95T (f’) t=T

Figure 5.8-2 Instantaneous streamlines in the second half period for the oscillatory
lid-driven cavity flow (1st and 3rd rows) and Soh et al. (2nd and 4th rows)

Vous aimerez peut-être aussi