Beruflich Dokumente
Kultur Dokumente
To cite this article: A. Amiri , K. Vafai & T. M. Kuzay (1995) EFFECTS OF BOUNDARY CONDITIONS ON NON-DARCIAN HEAT
TRANSFER THROUGH POROUS MEDIA AND EXPERIMENTAL COMPARISONS, Numerical Heat Transfer, Part A: Applications: An
International Journal of Computation and Methodology, 27:6, 651-664, DOI: 10.1080/10407789508913724
To link to this article: http://dx.doi.org/10.1080/10407789508913724
T. M. Kumy
Argonne National Laboratory, Argonne, lllinois 60439, USA
The present work centers around the numerical simuhtion offorced conuectioe incompressible flow thmugh porous beds. Inertial as weU as uiscous effects are considered i n the
momenhrm equation. The mathemafr'cal model for energy transport was based on the
two-phase equation model, which does not employ local thermal equilibrium assumption
between the Jluid and the solid phases. The transport processes for hvo diflerent types of
boundnry conditions are studied. 7'he analysis was pegarmed i n terms of nondimensional
parameters that successfuUy cast together OU the pertinent influencing effects. Comparisons
were made between our numerical findings and experimertlal resub. OueraU, the
comparisons that were madefor the constant waU h e d m boundary condition display good
agreement.
INTRODUCTION
Forced convection heat transfer in porous media has been extensively investigated. The phenomenal enhancement in heat transfer level produced by using
porous materials as transport media has stimulated significant interest for researchers to develop rigorous models to simulate thermal behavior in porous
media. Typical examples of applications that utilize porous media include absorption and adsorption processes, packed filters, pebble-type heat exchangers, and
energy storage units. In addition, utilization of porous insertions in high heat flux
applications as a technique to augment heat transfer has proved to be a promising
research field, as evidenced by the experimental investigations of metallic porous
media [I, 21. These experimental results have shown phenomenal increases in the
level of heat transfer.
There exists a significant amount of published literature concerned, in the
main, with the problem of forced convective flow through bounded porous media.
Received 28 April 1994; accepted 10 August 1994.
Address correspondence to Dr. Kambiz Vafai, Department of Mechanical Engineering, The
Ohio State University. Columbus, OH 43210-1107, USA.
A. AMlRI ET AL.
652
NOMENCLATURE
"
.t
C~
d~
dp/&
Da
Dh
F
hSf
H
i
I
J
k
K
L
LTE
Nu
P
Pr
9,
Re
Re,
I
T
U
'
'
'
u
x, Y
a
A,
M
P
I=
~~r"~F(~,H/pr)l
Subscripts
d
ed
eff
f
m
s
w
x
Y
0
dispersion
inlet
effective property
fluid
mean
solid
wall
x component
y component
reference
Superscripts
f
s
fluid
solid
Symbol
The complicated and interesting phenomena associated with porous beds provide a
wide area for fruitful discussion. In the work of Vafai and Tien [3], the inertial and
impermeable boundary effects on momentum and energy transport were discussed
in great detail. The investigation provided a simple characterization scheme for
interpreting the applicability of the familiar Darcy's law for various flow conditions
and bed configurations.
The fluid-to-solid heat transfer coefficient has received considerable attention in terms of experimental investigations to formulate its quantity for porous
beds of different geometries and configurations. Studies by Barker [4] and Wakao
et al. [5,6] can be regarded as examples of investigations to formulate the
fluid-to-solid heat transfer coefficient. Moreover, the flow through complicated
structures such as porous beds introduces flow disturbance, mixing, and recirculation of local fluid streams as the fluid passes through the bed. This phenomenon is
653
ANALYSIS
A. AMIN El'AL.
Governing Equations
V(u)
Momentum
where u is the velocity vector, p, the fluid density, the average porosity, pf the
dynamic viscosity of the fluid, K the permeability of the porous bed, F the
geometric function, J = u,/lupl the unit vector oriented along the pore velocity
vector up,( P)' the average pressure read off a pressure gage, c,, the fluid-specific
heat, (T,)' the intrinsic phase average of the fluid temperature k,, the fluid
effective thermal conductivity, h , the fluid-to-solid heat transfer coefficient, a,,
the specific surface area of the porous bed, p, the solid density, c,, the solid
specific heat, (T,)" the intrinsic phase average of the solid temperature, and k,,,,
the solid effective thermal conductivity.
The specific surface area of the bed is formulated as [13,15]
where d p is the pore diameter. The fluid-to-solid heat transfer coefficient was
655
where k, and k , are the fluid and solid thermal conductivities, respectively. It is
worth noting that Pr is assumed constant due to the small temperature range
considered in this study.
The local heat flux is cast in a dimensionless form, namely, the Nusselt
number. The definition of Nu is based on the hydraulic diameter:
In the above equation, q, is the wall heat flux, Dh is the hydraulic diameter, Tw is
the wall temperature, and T,, is the mixed mean fluid temperature defined in a
manner similar to that for classical channel flows. It is important to point out that
the total heat flux generated for the case of a constant wall temperature is
computed by summing the heat flux generated from both the fluid phase and the
solid phase.
The particle Reynolds number Re, and the Darcy number Da were utilized
to discuss the energy transport in porous beds. In addition, different permeability
values and different materials were considered for the solid phase. The numerical
computations were carried out for Da =
and 5 x
The main criterion in
selecting the solid phase materials was to obtain a wide range of values for the
solid-to-fluid thermal diffusivity ratio a,/af since the thermophysical properties of
the porous material may differ widely from one application to another. Table 1
presents the thermophysical properties of the materials considered in the present
investigation.
For completeness of the mathematical model, one needs to specify the
porosity of the bed and the pore diameter. For practical reasons, the latter is taken
to be 1 mm, whereas the former is taken to be 0.95 and the geometric function is
A. AMIRl ET AL.
656
Water
Solid phase
Copper
Bronze
Sandstone
Density,
kg m - 3
Specific heat,
J ~ ~ K- - II
997
4179
8933
8800
2200
385
420
710
Thermal
conductivity,
w m-l
K-I
0.613
Dynamic
viscosity,
x104 kgm-ls-l
8.55
401
52
1.83
taken as 0.08. On the other hand, the permeability and the geometric function used
for comparison with the experimental results were obtained directly from the
experimental investigations of Kuzay et al. [2] in a manner similar to that reported
by Vafai and Tien [16].
Boundary Conditions
657
equation energy mode has been reported. The wall heat flux boundary condition
may be viewed in two different ways. The first is to assume that each representative
elementary volume (which contains both fluid and solid phases) at the wall surface
receives a prescribed heat flux that is equal to the wall heat flux q,. As a result,
the q, will be divided between the two phases on the basis of the physical values of
their effective conductivities and their corresponding temperature gradients. The
second approach is to assume that each of the individual phases at the wall surface
will receive an equal amount of heat flux q,. Due to the non-slip boundary
condition at the wall, the convective mode between the two phases tends to vanish
at the wall surface. Consequently, the LTE assumption prevails close the wall, and
as a result, it is reasonable to assume that each of the individual phases at the wall
surface will receive an equal amount of heat flux q,. The numerical experimentation in this study was found to validate this argument, and it was therefore
implemented in this work.
SOLUTION METHODOLOGY
Equations (1)-(4) were solved by using the finite difference method. The
momentum and the energy equations were solved separately, since they are not
coupled. Central differencing was used for the spatial derivatives except for the
convective term in Eq. (31, for which the upwind differencing scheme was employed. The momentum equation was first solved by a tridiagonal matrix algorithm
after linearizing the nonlinear term. Next, the energy equations were solved after
specifying the exit boundary condition. A three-point backward differencing was
employed for the spatial x derivatives at the grid points on the right boundary. The
strong parabolic nature of the forced convective flows validates this procedure.
Furthermore, this procedure was validated by a systematic increase in the computational length of the bed, until the numerical results within the physical domain
were no longer affected by further increase in the computational length. The
steady state solutions of these equations were then obtained. The source terms that
appear in Eqs. (3) and (4) were updated after each iteration due to their temperature dependency.
Variable grid size was implemented in the y direction, while constant grid
size was used in the x direction. A fine, equally spaced grid size was positioned
within 16% of the total height from each solid boundary, while a relatively coarser
equally spaced grid size was used for the core region. It was ensured that the
results were grid independent. This was done through a systematic decrease in grid
size until a grid-independent result was obtained. A 101 x 101 grid configuration
was found to satisfy this criterion. The results were assumed to have converged
when the temperature values for the fluid and the solid phases for two consecutive
iterations were below the convergence criterion of 10-lo.
RESULTS AND DISCUSSION
In order to examine the accuracy of our numerical results, the numerical
results were compared against the most closely related analytical solutions. Limiting cases of analytical solutions were used for this purpose. The necessary adjust-
A. AMlRI
658
ET AL.
ments were made to our numerical code to reduce it to a system equivalent to the
simplified available solutions. O u r numerical finding for the velocity field was first
compared with the analytical solution of Vafai and Kim [17]. The comparison is
depicted in Figure 2 a . Next, the temperature distribution was compared with the
analytical results of Vafai and Kim [17] for the case of constant wall heat flux, and
with the analytical solution of Vafai and Thiyagaraja 1181 for the case of constant
wall temperature. Figures 2b and 2c demonstrate such comparisons. It is worth
noting that all of the above comparisons were made in terms of the dimensionless
ra5
04
03
- rrw-
02
01
--m
aW.lo
02
04
"tu.
oe
aa
Figure 2 0 . Cornpanson of velocity profile of the present work w ~ t hthe analytical solution of Vafa~and Kim 1171.
Figure Zb. Comparison of the fully developed temperature profile of the present
work with the analytical solution of Vafai
and Kim 1171 for q, = const.
Figure Zc. Comparison of the fully developed temperature profile of the present
work the analytical solution of Vafai and
Thiyagaraja 1181 for T, = const.
659
variables that appear in the above-mentioned studies. As may be seen from Figure
2, all comparisons display excellent agreement.
The results of this study focus on the thermal response of the porous bed to a
number of influencing parameters such as the particle Reynolds number, the Darcy
number, and the solid-to-fluid thermal diffusivity ratio. The results are presented
in terms of Nusselt number for boundary conditions of constant wall temperature
and constant wall heat flux. However, since the experimental results are based on a
constant wall heat flux, the constant wall heat flux boundary condition is considered for the experimental comparisons. In addition, the wall temperature and the
average friction factor will also be highlighted when comparisons are made with
experimental results. The average friction factor is defined in a manner similar to
that for classical channel flows:
In the above equation, u , is the mixed mean fluid velocity defined in a manner
similar to that for classical channel flows.
Effect of t h e Particle Reynolds Number
In order to determine the effect of the particle Reynolds number Re, on the
Nusselt number Nu, cases with different Darcy number Da were compared. These
comparisons were performed using bronze as the solid phase. Figures 3a and 36
show the effect of the Re, on Nu for the two different boundary conditions under
consideration. As expected, it is observed that as Re, increases, the thermal
boundary layer becomes thinner, resulting in more energy being carried away by
convection than by conduction, and as a result, larger values of Nu are encountered. The Nu values are found to be much larger for the constant wall temperature than for the constant wall heat flux. This is primarily due to the way that the
Nu is calculated from the two-equation model. The contribution of the solid phase
to the total heat flux generated, which is incorporated in the Nu calculation for the
constant wall temperature, has significantly increased the magnitude of Nu.
Effect of t h e Darcy Number
The effect of Da was examined in a manner similar to that carried out in
determining the effect of Re,. Figure 4 depicts the effect of Da for the case of
bronze as the solid phase. The thickness of the momentum boundary layer for a
low-permeability porous medium is a hnction of Da only, while it becomes a
function of Da and the inertia parameter A , as the permeability increases [17]. In
addition, the flow inside the channel becomes more of a slug flow as Da decreases,
causing the magnitude of Nu to increase. Thus, the magnitude of Nu reaches its
asymptotic maximum value as Da decreases. It is important to point out that the
maximum asymptotic value, as illustrated in Figure 4, is different for different Re,.
v-.mm=-
?rn
(b)
,I
Figure 3. Variation of the Nusselt number for bronze with Da = lo-' and 5 X
lo-' for (a) constant wall temperature
and ( b ) constant wall heat flux.
The differences between the asymptotic maximum values are primarily due to the
dependence of the dispersion parameter on the velocity vector, which as a result,
affect the magnitude of the thermal resistance by a considerable margin, depending on the magnitude of the velocity.
Figure 4. Variation of the Nusselt number for bronze with Rep = 40, 100, and
200.
Figure 5. Variation of the Nusselt number for copper, bronze, and sandstone for
Da = 5 x
Table 2 lists the numerical values of the geometric functions and the corresponding
permeabilities (presented here in terms of Da). For the sake of quantitative
comparisons, the hydraulic diameter was set equal to the diameter of the square
channel.
A. AMIRI ET AL.
Table 2. Characteristics for the different mesh types
Mesh
Formed
Foam
Porosity, %
Darcy number
Geometric
function
85
95
2.2 x l o 4
5.1 X lo-'
0.01 1
0.010
Figure 6. Comparison of the friction factor predictions of the present work with
the experimental data reported by Kuzay
et al. [21.
Nusselt number predictions of the present work with the experimental data
reported by Kuzay et al. [2].
CONCLUSIONS
T h e problem of forced convection in a channel filled with a porous material
is analyzed. T h e two-equation model is utilized to perform the analysis. Both the
constant wall temperature and the constant wall heat flux were analyzed and it is
shown that the constant wall temperature has a n edge. T h e analysis was performed
in terms of three nondimensional parameters that successfully cast together all the
pertinent influencing effects. T h e investigation was completed by comparing the
numerical analysis with a recent experimental study. T h e comparisons display good
agreement for the Nusselt number predictions and the wall temperature variation.
REFERENCES
1. F. E. Megerlin, R. W. Murphy, and A. E. Bergles, Augmentation of Heat Transfer in
Tubes by Use of Mesh and Brush Inserts, ASME J. Hear Transfer, pp. 145-151, 1974.
2. T. M. Kuzay, J. T. Collins, A. M. Khounsary, and G. Morales, Enhanced Heat Transfer
with Metal-Wool-Filled Tubes, Proc. ASME / JSME Thermal Eng. Cont, pp. 145-151,
1991.
3. K. Vafai and C. L. Tien, Boundary and Inertia Effects on Flow and Heat Transfer in
Porous Media, Int. J. Heat Mass Transfer, vol. 108, pp. 195-203, 1981.
4. J. J. Barker, Heat Transfer in Packed Beds, Ind. Eng. Chem., vol. 57, no. 4, pp. 43-51,
1965.
5. N. Wakao, S. Kaguei, and T. Funazkri, Effect of Fluid Dispersion Coefficients on
Particle-to-Fluid Heat Transfer Coefficients in Packed Beds, Chem. Eng. Sci., vol. 34, pp.
325-336, 1979.
6. N. Wakao and S. Kaguei, Heat and Mass Transfer in Packed Beds, Gordon and Breach,
New York, 1982.
7. S. Yagi and D. Kunii, Studies on Effective Thermal Conductivities in Packed Beds,
AIChE J., vol. 3, no. 3, pp. 373-381, 1957.
8. S. Yagi and D. Kunii, Studies on Heat Transfer near Wall Surface in Packed Beds,
AIChE J., vol. 61, no. 1, pp. 97-104, 1960.
664
A. AMlRl ET AL.