solution for case 2 - shodhgangashodhganga.inflibnet.ac.in/bitstream/10603/73163/12... · 3.2e+005...
TRANSCRIPT
CHAPTER 5
SOLUTION FOR CASE 2
5.1 Introduction
This chapter describes the numerical as well as experimental results of natural con
vection with surface radiation from planar heat generating element mounted freely in
a vertical channel formed between two parallel plates. To study the effect of surface
radiation, the surface emissivities of the heat source and the side adiabatic plates are
varied. As per the observation in chapter 4, it is proposed to perform three dimensional
numerical simulations using commercial software FLUENT 6.3.
5.2 Experiments conducted
Experimental methodology, procedure and data analysis for the experiments conducted
are described in chapter 3. Experiments are conducted by varying the surface emis
sivities of the heat source and the side adiabatic plates by coating these surfaces with
different paints. The emissivity of the paint coated surfaces are estimated based on
the procedure described in section 3.4. The spacing between the adiabatic side plates
and the stream-wise location of the heat source inside the channel has been varied to
study its effect on the heat transfer characteristics of the heat source. Thermocouples
are placed at different locations on the adiabatic side plate and the heat source so as to
measure their surface temperature. All the thermocouples are connected to a PC based
data acquisition system. The power input to the heat source is given by regulated DC
power supply system.
5.3 Discussion of Results from Experiments
The important variable geometry related parameters namely, channel aspect ratio has
been varied as 6.25, 5, 4.17, and 3.5, while the stream-wise location of the heat gen
erating element studied are the top, middle, and the bottom locations of the channel.
The surface emissivities of both the heat source (ch)and the side plate (C8) have been
varied in the range (0.05-0.85). The heat input is varied from 1-40W, as a result volu
metric heat generation (q*) and the modified Rayleigh number (Ra*) were changed in
the range 3.18-12.59 and 3.2 x 105- 1.6 X 107 respectively.
5.3.1 Effect ofstream-wise locations ofheat generating elementand
channel aspect ratio on heat transfer
The measured surface temperature of the heat generating element for three different
stream-wise locations of the heat generating element such as the bottom, middle and
top for a channel aspect ratio, (AR=6.25) shows that for a given modified Rayleigh
number, and for fixed value of emissivities of the heat source and channel walls, the
temperature of the heat generating element is independent of its stream-wise location.
It must be noted that this aspect ratio corresponds to the minimum channel spacing
possible with the experimental setup. The heat generating element, placed inside the
channel can be regarded as a heated short plate kept in an infinite fluid medium, hence
the physical situation remains identical for different stream-wise locations of the heat
generating element. Consequently, one would expect identical temperature distribution
for all stream-wise locations as observed experimentally. However, the effect of stream
wise location may be prominent for narrow channel spacings. Table 5.1 shows the tem
perature of the heat generating element for four different channel aspect ratios. The
results indicate that the temperature slightly increases with channel aspect ratio, but the
difference is not discernible. The maximum deviation is only 1.8f{ for Ra* = 8.1 x 106 ,
which suggest that the influence of channel aspect ratio on the temperature of the heat
generating element is negligible when compared to the average surface temperature of
65
Table 5.1: Measured surface temperature of the heat generating element for differentchannel aspect ratio (Heat source located at the middle of the channel, th =
0.85, t s = 0.05)
Modified Rayleighnumber (Ra*)
3.3 X 105
8.1 X 106
Surface temperature (K) of the heat sourcefor different channel aspect ratio (AR)
AR=6.25 AR=5 AR=4.17 AR=3.57310.9 311.1 311.4 310.7366.7 366.6 366.3 365.1
the heat generating element. This insignificant effect of aspect ratio on temperature is
attributed to the same reason discussed previously for the effect of stream-wise loca
tion. However, it is worth mentioning that the effect of aspect ratio on heat transfer
characteristics had been reported in previous studies conducted on geometries like par
allel plate with protruding heat source [9], parallel plates [22], enclosures [24] and side
vented cavities, [30] and [31]. From the preceding discussion, it can be concluded that
for the range of geometrical parameters considered, the effect of channel aspect ratio
and stream-wise location of the heat generating element on heat transfer performance
is negligible. Therefore, further experimental investigations have been performed by
placing the heat generating element at the geometrical center of the channel and for a
channel aspect ratio of AR = 6.25.
5.3.2 Heat transfer rates
In a combined mode of energy transport involVing surface radiation and natural convec
tion, the parameters such as emissivity and Rayleigh number playa vital role. Firstly,
the influence of surface emissivity on the heat transfer rate were examined. Figure
5.1 illustrates the influence of surface emissivity of the heat generating element on the
relative contribution of natural convection and surface radiation heat transfer to the to-
tal heat transfer rate. The curves are drawn for the cases with the lowest emissivity
of channel walls, t s = 0.05. As seen from this Figure. the convective heat transfer
66
.,................ ,....0" ~ 7....•.. Ra =1.14 >< 100.25
1.00 r-~~;o;-.-----------.-.
0.75 ~
0.50
•......0.00 !--.....:......I------'----'---'---'-"""'t
........................... ' .•.... Ra' =8.1 x 106
•.........0.00 I-_......JL.....I-~----'__--'-__'-_-'-......j
1.00 I-------,.---~-----
0.75 ~
ri 0.50
a0.25
1.00 I--=--~----0--0----%-Q-,--..-..-.-..-.-Q-,----l
~~ ~ ~---~
1.0
Ra' =6.58 x 10'
0.2 0.4 0.6 0.8Emissivity of heat source, (~)
...................,.',.',.'.. ',.'.. '.,'",•....0.00 L....:.......--l.._~--'-~_.L..-~--J.__--J
0.0
0.50
0.25
Figure 5,1: Variation of convective and radiative heat transfer with emissivity of heatsource for different modified Rayleigh number (Es = 0.05)
67
0.8o ell = 0.05
.-. 0.7 ... e h = 0.55
g: .6h =0.75
a • E h = 0.85-- 0.6...
'* •Ie 0.5... ••-m • •• • • • • • •• •J: 0.4 •c:: .A •••0 ...... • • • • • • • •
16 ......... ... ... ... .A ... ... ...=c 0.3~1llc::
0.20:0::0~LL
0.1
poooo 0 o 0 0 0 0 0 0
0.03.2E+005 3.3E+OO6 6.4E+OO6 9.6E+006 1.7E+OO7
Modified Rayleigh number (Ra*j
Figure 5.2: Variation of fractional radiation heat transfer with modified Rayleigh number (Es = 0.05)
decreases with the increase of surface emissivity of heat source, but at the same time
radiative heat transport from the heat source increases. This trend is expected, as the
largest temperature gradient between the heat source and the fluid occurs at lowest sur
face emissivity for which the natural convection is the dominant mode of heat transfer.
But, as surface emissivity increases the temperature of heat generating element drops,
thereby reducing the convection transport. This is because, the strength of convection
current is determined by the buoyancy force which depends on the fluid temperature
differential and isobaric volume expansivity. To provide further insight into the rela
tive strength of surface radiation and natural convection in the total heat exchange, the
fractional radiative heat transfer as a function of modified Rayleigh number is plotted
in Fig. 5.2. Evidently, for high surface emissivities of the heat generating element, the
contribution of surface radiation is found to be maximum at the lowest value of modi
fied Rayleigh number encountered in the respective cases. Stated differently, the surface
radiation is seen to playa significant role as compared to natural convection at very low
68
thermal loads. It can be seen from Fig. 5.2 that the maximum radiation accounts for
about 54% of the total heat removed for the highest emissivity considered in this study.
At low values of modified Rayleigh number, the buoyant force becomes weaker and
the resulting slow fluid circulation reduces the natural convection heat transfer coeffi
cient. This also brings out the fact that for high surface emissivities, radiation effects
are predominant even for a moderate temperature differential, typically encountered
during the operation of electronic components. Therefore, surface radiation effects has
to be taken into account for the accurate estimation of heat transfer rates even at low
thermal loads. This observation is also in accordance with the results reported in [22].
Figure 5.2 also reveals that the radiative heat transfer increases with increasing surface
emissivity for a fixed value of modified Rayleigh number. However, for a fixed surface
emissivity of the heat generating element the relative radiation contribution decreases
with increasing modified Rayleigh number, reaches a minimum for a Rayleigh number
of approximately Ra* = 2.5 x 106 , and thereafter both natural convection and surface
radiation are in competing mode, contributing almost equally to the total heat transfer.
It may be noted that for a heat generating element with low emissivity (Elt = 0.05),
natural convection is the dominant mode of heat transfer contributing to about 96% of
the total heat transferred. Figure 5.3 shows the influence of surface emissivity of heat
source and side plate on convective and radiative heat transfer. From Fig. 5.2 and Fig.
5.3, it is clear that increase of surface emissivity of the side plates results in greater
interaction of surface radiation.
5.3.3 Heat transfer correlations
The experimental design consists of a total of ninety eight experiments corresponding
to different values of heat input, surface emissivity of the heat generating element and
adiabatic side plates for a channel of AR=6.25. Since the height of the heat generating
element is small, no appreciable variation in the local temperature of the heat source
along the height (also the flow direction) is observed. The average convective Nusselt
number for the heat transfer process is evaluated using Eq. (3.12) and the total Nusselt
69
0.980t---------------,
0.736 ................. : .0.492 0-- A
0.248 eh=O.85,e.=0.85,relatlveNu,---0- eh = 0.85, e. = 0.85. relative Nu c
.......... eh = 0,05. e. = 0.05, relative Nu,~ eh = 0,05. e. = 0.05. relative Nu c
3.2E+006 6.1 E+006 9.0E+006Modified Rayleigh number (Ra~)
0.248 f
0.0051-k=A===""~==Cio:::==~==~-l
3.3E+005
c 0.005o~ 0.9801-----------------1.c~ 0.736 .o~0.492 0-::::l v---o----<>----o------<>------o6°,248 ........ "h = 0,75, e. = 0.75. relative Nu,... ~"h = 0.75, e. = 0.75, relative Nu c~ O,0051---~---'---~--..L------lEE0.9801------------------;-'i 0.736 .(fj
~ 0.492~ 0.248 D-u--O------{::::::J_----O-~-_{~J_--_I-n::Ji eh = 0.55, ",=0.55, relative NUi
-0- "h=0,55, ".=0.55. relative Nu c~ 0,0051--------"-----.....;".------'---\
0.980 r-*'··~..,.......,..........~.........., ..........~~........~.........., ........~~........___j
0.736 -
0.492 -
Figure 5.3: Variation of relative Nusselt number with modified Rayleigh number
70
number, due to the collective contribution of convective and radiative modes is evalu
ated using the Eq. (3.14). A correlation for the average convective Nusselt number for
the present conjugate heat transfer problem is developed as:
(5.l)
The agreement of the experimental data with the proposed correlation is repre
sented in the form of a parity plot shown in Fig. 5.4. The negative exponent in the term
consisting of emissivity of heat source indicates that emissivity of heat source weakens
the convective heat transport. This is clearly reflected in the experimental observation
reported in Table 5.2. However, it is interesting to note that the strength of convection is
increased with the emissivity of the adiabatic side plates. The reason for this observa
tion is clear from the numerically simulated velocity and temperature profiles shown in
Figs. 5.5 and 5.6 respectively. The Figures suggest that the increase of convective heat
transport with the emissivity of the adiabatic side plates is attributed to the modification
37
__ 33..::L.o.£ 29
u;:,Z
~ 25.cE;:,
.: 21Q)IJ)IJ)
i 17Q)
>'';::;
~ 13>cou 9
Correlation coefficien t =0.95 +1 O~.'O
Std error = =0.02
-10%
9 13 17 21 25 29 33 37
Convective Nusselt number, Nuc(Exptl.)
Figure 5.4: Parity plot of average convective Nusselt number
71
0.25
0.20
0.15-III- 0.10E-Ql....ctl 0.05C.Ql
"0 0.00III'tl 0.25r::ctlQl(,) 0.20~
::s0III.... 0.15ctlQl
.r::r:: 0,10QlQl;;:.... 0.05Ql
.Q
0.ctl 0.00O'lQl 0.25.r::....IIIIII 0.200~
(,)ctl>- 0.15....'u0
0.10Ql>>- 0.05
Ra" = 81 x 1013
/Ra·=6.58 x 105
-- ell = 0.85, e s = 0.858 11 = 0,0, l!s =0,0
._ •• , 1:11
= 0.85. 1:. = 0.0
o.00 1!..-_-'--_...1...-_-'-_--L._---'-_----L_----'_~
0.000 0.004 0,008 0.012 0.016Spacing between the heat source and one side plate(m)
Figure 5.5: Variation of fluid velocity across the gap between heat source and one sideplate measured at the mid-height of the heat source
72
0.0160,012
•
0.008
-
Ra'=114 x107
..···0·..·· Z:n-{).{}'~5=o.o
,•• ..:.- •• 8",=0.1l:5, 8 s ='O.{J• llh=0.85,8§=0.85
...,.0.. ,.. llh ~ 0.0, 8 1 = 0.0,•• 06 ••• !lM'" 0.85, l!s~' (LO
• l:h=0.85,lls=O.85
.....0..... eh = 0.0, 8: 5 =0.0••• "" ••• 8:h=0.85., 'a:s=O.O
• Bh = 0.85,1::. = 0,85
0.004
400
380
- 360~-<U.... 340(l:J
0-(l) 320~en"'0 300s::(l:J
(l) 400<J'-=0 380en-(l:J 360(l)
.t:s::(l) 340(l)
3:-(Ll 320.c0-(l:J 3000)
(l).t: 400-I/)en0 380'-<J(l:J
(l) 360'-=-(l:J 340'-(l)0-E 320(l)
I-
3000.000
Spacing between the heat source and one side plate(m)
Figure 5.6: Variation of fluid temperature across the gap between heat source and oneside plate measured at the mid-height of the heat source
73
Table 5.2: Variation of surface temperature, convective and total Nusselt number withemissivity of heat generating element corresponding to C8 = 0.05 and Ra* =
8.1 x 106
Emissivity of Surface temperatureheat Source of heat source
(ch) (K)0.05 374.70.55 371.70.75 367.00.85 366.7
Convective Nusseltnumber(Nuc)
23.718.618.016.6
Total Nusseltnumber(NUt)24.727.929.130.0
of the thermal and flow field under the influence of surface radiation. The fluid tem-
perature profiles indicate that for a given heat input, the temperature of the heat source
decreases with the increase of its surface emissivity. This reduction in temperature of
the heat source at high values of emissivity is primarily due to larger heat energy radi
ated to the cold surfaces as compared to that radiated from the heat source at low and
zero emissivity cases. Consequently, to satisfy the energy balance the radiative heat
flux leaving the heated surface needs to be absorbed by the cold surfaces. At zero and
very low values of emissivity of the adiabatic side plates, the incident radiation reflected
from the side plates enable it to maintain low temperatures. Under this situation, the
emitted radiation from the heated surface and the reflected radiation from the cold adia-
batic side plates are expected to be absorbed by the ambient; the third body considered
in the radiative heat transfer analysis. As emissivity of the side plate is increased, the
incident radiation will be absorbed and re-emitted. The process of absorption and emis
sion continues, and this eventually results in the increase of temperature of the side plate
when thermal equilibrium is established. Thus, as shown in Fig. 5.6, the temperature
gradient on the adiabatic side plate is strikingly different from the case with no surface
radiation and consequently heat transfer is affected by the surface radiation. Hence, it
appears that at high values of emissivity of the heat source and adiabatic side plates, a
radiation induced boundary layer is developed along the length of the side plates. In
other words, with high emissivity of both heated and unheated surfaces, the unheated
surface becomes an active surface and part of the input heat to the heat generating ele
ment is convected to the fluid from the unheated surface. This observation is reflected
74
by the positive exponent of emissivity of the side plates in Eg. (5.1).
In a conjugate heat transfer problem like the one considered here, an accurate predic
tion of total heat transfer rate is of great importance from an engineering view point.
So, to report the overall heat transfer characteristics of the heat generating element, the
experimental data of the total Nusselt number is expressed as a function of modified
Rayleigh number, emissivity of the heat generating element and adiabatic side plates.
A multiple regression analysis results in the folloWing correlation for the average total
Nusselt number.
(5.2)
Figure 5.7 illustrates the trend of predicted values of the total Nusselt number obtained
from Eg. (5.2) with the experimental values. It can be seen that the distribution of
data points about the parity line is unbiased and the data points are distributed within
42
38
--...034
(.)-..:::I:-:30(jj..cE~ 26::Gltil
gj 22:zS{: 18
14
Correlation coefficient = 0.97
Std error = ±0.04+10%
10 "'""'''''"-.L.-.......L-''''--...L..............L..._...I...-..I-.............................I
10 14 18 22 26 30 34 38 42
Total Nusselt number, NUt (Exptl.)
Figure 5.7: Parity plot of total Nusselt number
75
Table 5.3: Variation of surface temperature, convective and total Nusselt number withemissivity of adiabatic side plate corresponding to Ch = 0.85 and Ra* =8.1 x 106
Emissivity of Average surfaceadiabatic side plate temperature of
(1:8) heat source (K)0.05 366.70.55 359.5
0.75 359.20.85 359.1
Convective Nusselt
number(Nue)
16.619.3
19.619.7
Total Nusselt Relative contribution of
number convective Nusselt number(NUt) (%)30.1 44.832.6 40.7
32.9 40.433.0 40.3
an error band of ±10%. It is apparent that the total Nusselt number is an increasing
function of both modified Rayleigh number and emissivity of the heat source. Also,
the exponents of the governing parameters of buoyancy driven convection as well as
surface radiation indicate the strength of two distinct heat transfer modes in the com
bined process of thermal energy transport. The dependence of total heat transfer on
emissivity of the side plate is only minimal. In a conjugate heat transfer problem in
volving surface radiation and natural convection, the influence of radiative heat transfer
is controlled by the surface emissivity and as a consequence of which the total Nusselt
number is an increasing function of the emissivity. The exponent of modified Rayleigh
number is larger in Eq. (5.1) compared to Eq. (5.2). This larger value of exponent
in the convective Nusselt number correlation is an indication of strong dependence of
natural convection on Rayleigh number even in the presence of other modes of heat
transfer. However, in the conjugate problem considered here, the extent to which the
natural convection is affected by the surface radiation is found to be a strong function of
the emissivity of the surface of heat source. For this reason, the exponent of Rayleigh
number in the total Nusselt number correlation is expected to have a lower value as
compared to the exponent of Rayleigh number in the convective Nusselt number cor
relation. Experimental data shown in Table 5.3 gives a quantitative idea of effect of
emissivity of the side plate on heat transfer characteristics for a given value of emis
sivity of the heat source and heat input. It is clear that the convective Nusselt number
slightly increases with emissivity of the side plate, but the relative contribution to the
total heat transfer decreases due to increase of the radiative Nusselt number. Since the
76
o
Correlation coefficient = 0.996
Std error = ± 0.04
0.9
-:- 0.8...oU~ 0.7
e:::l10 0.6~c.E 0.5CllI-III~ 0.4
§.~ 0.3Cll
Eis 0.2
0.1
0.0 ~~'--,--------,'--'-------.J~'-------.J'-----'---J~-l~--l~-----'-~-----'-
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Dimensionless Temperature,e (Exptl.)
Figure 5.8: Parity plot of dimensionless temperature of the heat generating element
total Nusselt number also increases, the effect of increase of emissivity of the side plate
is to decrease the temperature of the heat source. In fact, the proposed Nusselt number
correlations are expected to mirror these experimental observations in a realistic way.
In many practical design problems, particularly in cooling of electronic application de
vices, the primary requirement is the accurate prediction of the operating temperature
corresponding to a given thermal load. With this in view, a correlation for dimension
less temperature of the heat generating element in terms of dimensionless volumetric
heat generation, and emissivities of the heat generating element and the adiabatic side
plates has been developed as
e (5.3)
The parity plot shown in Fig. 5.8 suggests that the predicted values agree well with
the experimental data. Equation (5.3) demonstrates that the dimensionless temperature
77
decreases with both fh and f s , while the effect of fh is strikingly visible. As temperature
is a direct function of volumetric heat generation (q*), a stronger exponent of q* is jus
tifiable. Furthermore, a value close to unity signifies that (j is monotonically increasing
with q*.
5.4 Numerical Analysis
The numerical analysis of the problem has been performed using FLUENT 6.3. The
physical geometry is three dimensional consisting of a fluid zone between the parallel
plates, extended regions at the inlet, exit and sides of the channel and a solid zone rep
resenting the heat generating element. The fluid medium (air) is assumed to be incom
pressible with constant thermophysical properties. However, the body force arising due
to density variation imparted by the temperature gradient is accounted by an additional
source term in the vertical direction of the momentum equation and this source term is
modeled using Boussinesq approximation [64]. In-order to model surface radiation the
fluid medium is treated as transparent for radiation exchange.
5.4.1 Governing Equations
The governing equations presented in section 4.2.1 are for the two dimensional problem.
Since the analysis presented in this chapter is three dimensional, the relevant equations
are presented in the divergence free form in the following
Conservation ofmass
V.u = 0
Conservation ofmomentum
1(u.\7)u = --\7P + v\72u + B
p
78
(5.4)
(5.5)
where B is the body force vector defined by [0, -g{3(T - Too), O]T
Conservation ofenergy
(a) Fluid domain
(b) Solid domain
1/1
(lis\72Ts + gc = 0. p p
5.4.2 Boundary conditions
(5.6)
(5.7)
No slip for velocity components and zero normal pressure gradient are the boundary
conditions used on all solid walls. The velocity components at the inlet and exit of the
extended domain not known a priori, are obtained using appropriate boundary condi
tion for pressure. The stagnation pressure at the inlet of the domain is equated to the
atmospheric pressure, whereas static pressure at the exit of the domain is taken as at
mospheric pressure. The emissivity of the surfaces, used for the numerical simulations,
are same as those obtained in section 3.4. In general the mathematical representation of
the boundary conditions can be described as follows
On the side walls
Inlet of extended domain
Pressure Po = Patml Temperature T = Too
Exit of extended domain
79
Pressure Pst = Patm, aT = 0al1
5.4.3 Computational domain and grid
Initially, a 3D computational domain without an extended domain has been created
for the numerical solutions. The computational domain is discretized using structured
non-uniform hexahedral cells. Fine grids have been employed over all the solid-fluid
interfaces to resolve the gradient of field variables. Numerical simulations have been
carried out for the same heat input, used in experimental investigation. The heat input
has been modeled as a volumetric heat generation term, specified in Eg. 3.17. The
results show that the simulated temperature over-predicts the measured temperature of
the heat source. This observation has become the motivation to adopt an extended
domain approach for the numerical simulations. In view of this, numerical simulations
have been carried out by extending the domains of the channel by 25%, 50%, 100% of
the channel height and the results of the studies are summarized in Table 5.4.
It can be seen that the numerical simulations performed with extended domains
50% and 100% of channel lengths provide close results. From this observation, we
have chosen 50% extended domain for numerical studies. With 50% extended domain,
a grid sensitivity study has been conducted to arrive at an optimum grid for numerical
simulations. A summary of the results of grid sensitivity study is given in Table 5.5. The
results are found to be insensitive to grid beyond the one with 192750 cells, consisting
of 594295 faces and 208964 nodes. The final computational domain along with the
Table 5.4: Summary of domain independence study (Eh=0.85,E8 =0.05, AR=6.25)
Modilled Rayleigh Measured surface Numerically predicted surface temperature of heat source (K)number temperature of for different percentage of extended size of the domains
heat source
(Ra·) (K) 0% 25% 50% 100%.
3.3 x 105 310.9 312.8 312.2 311.6 311.68.1 x lOG 366.7 373.3 371.1 369.5 369.4
80
Table 5.5: Summary of grid sensitivity study (Eh = 0.85, Es = 0.05, AR=6.25, Ra* =8.1 x 106)
Number of Measured surface Numerically predicted Temperature at different y-Iocationscells temperature of the surface temperature on the adiabatic plate
heat source of the heat source (y is in mm)
(K) (K) Y = 50 y = 100 Y = 125 Y = 150 Y = 200
36640 376.7 303.9 304.6 305.7 306.5 307.994200 370.3 304.0 304.9 305.9 306.6 308.1132000 369.7 304.1 305.4 306.2 306.8 308.1192750 366.7 368.4 304.2 305.1 306.2 306.8 308.2241680 368.3 304.2 305.2 306.2 306.8 308.2
I~Y
~-L..:X
_--- Pressure outlet
Extended domainat channel exit
---Adiabatic wall
Heatgeneratingelement
Extended domainat channel inlet
-~--_ Pressure inlet
Figure 5.9: Computational domain along with the grids used for the 3D numerical simulations
81
a.One fourth height of the heat source
c.Three fourth height of the heat source
b.Half height of the heat ource
d.Top edge of the heat source
Figure 5.10: Temperature field plotted along the dimensionless horizontal distance andchannel span at different levels along the height of the heat source.(Eh =0.85, lOs = 0.05, Ra* = 8.1 x 106
)
82
grid used is shown in Fig. 5.9. Variations of the predicted temperature cOlTesponding
to different locations along the height of the heat source are shown in Fig. 5.10. The
plots undoubtedly indicate that the temperature variation along the z-direction (span
of the heat generating element) is almost uniform except close to the supports. Stated
differently, the temperature field is nearly two dimensional in the x-y plane.
5.4.4 Numerical solution procedure
FLUENT solver uses control volume approach to solve the governing differential
equations offluid flow and heat transfer [65]. The segregated steady state solver option
has been selected for the present computation. A second order upwind scheme is used
to interpolate the unknown cell interface values required for modeling the convection
tellliS. The coupling between velocity and pressure is resolved by selecting the SIMPLE
(Semi-Implicit method for pressure linked equations) option and radiation heat transfer
by the discrete ordinate model. In using discrete ordinate model the optical thickness
is set equal to zero, thereby the fluid medium is assumed as non-participating. Non
linearity of the equations requires the solution to be progressed in a controlled manner
with the use ofrelaxation factors. The under relaxation factors used in the present study
are 0.3 for pressure, 0.7 for momentum, 0.9 for energy and 0.9 for intensity of radiation.
Convergence ofthe solution is checked by examining the residues ofdiscretized conser
vation equations of mass, momentum, energy and intensity of radiation. The iteration
is terminated only when the maximum of all the residues reaches less than 1 x 10-6 to
assure conservation of quantities.
5.4.5 Discussion of results from Numerical analysis
Numerical simulations have been performed for all the values of variable parameters,
used in the experimental investigations with the heat generating element placed at the
geometrical center of the channel and for a channel aspect ratio of AR = 6.25. Figure
5.1 0 shows the temperature field plotted across the gap between the heat generating
83
40 .---------------------,
35 ...'"'"
:i '" ~
Z '" l8! l8!
..: 30 '" l8!Q)
'"l8!
.c'"
l8!E ~
:::J l8!c • NUl experimental... 25 ~
"ai l8! NUl numerical(/)(/) t:::JZ i~
20~
~
15
10 Ll----''-----'-_.........._.l...-........_.....L..._"''''--------J_---'------I
3.2E+005 3.7E+006 7.0E+006 1.0E+007 1.4E+007 1.7E+007
Modified Rayleigh number, Ra*
Figure 5.11: Comparison between the experimental and numerical values of total Nusselt number (ch = 0.85, Cs = 0.05)
element andone side wall corresponding to four span-wise locations. A close examina
tion of the sub-figures reveals that the heat generating element appears as an isothermal
body, which is in full agreement with the experimental observation reported earlier.
Furthelmore, these figures communicate that the vmiation of thennal field is predom
inant in the x-y plane, which means that the heat generating element can be regarded
as a planar heat source. Figure 5.11 presents a comparison of experimentally and nu
merically estimated values of total Nusselt number as a function of modified Rayleigh
number for the case Ch = 0.85 and Cs = 0.05. As can be seen, the experimental and
numerical results compare well for low values ofRayleigh numbers, while the deviation
between the experimental and numerical values progresses with Rayleigh number, the
maximum deviation is found to be 16.6% at Ra* = 1.6x 107. The reason for this trend
84
Table 5.6: Mass flow rate through the channel for the case with Ra*AR=6.25
Emissivity ofheat generating element (Eh)
0.050.85
% increase of rrif
Es = 00.002410.00236
-2.07
Mass flow rate, kglsfor different Es
Es = 0.05 Es = 0.550.00256 0.002760.00303 0.00398
18.35 44.20
Es = 0.850.002770.00403
45.48
can be traced to the fact that the unaccountable heat losses, which is inherent to any
experimental setup, increases with increase of temperature. Revisiting the Figs. 5.5 and
5.6, it can be seen that in the absence of radiation (Eh = 0.0 and Es = 0.0), the boundary
layer formed adjacent to the heat source is similar to natural convection boundary layer
over a vertical isothennal plate kept in an infinite fluid medium. With increasing sur
face emissivity of the walls, the fluid velocity profiles in Fig. 5.5 tend to shift towards
the channel side wall which is indicative of flow field redistribution effected by surface
radiation. Clearly, the flow field redistribution under the influence of surface radiation
is seen to be affecting the mass flow rate through the channel. Table 5.6 gives the mass
flow rate through the channel computed for different values of the surface emissivity of
the heat generating element and channel walls. For a given value of emissivity of the
heat generating element, mass flow rate increase with increase of emissivity of the side
walls. It may be noted that for low emissive surface of the heat generating element,
(Eh = 0.05) the increase ofmass flow rate with an increase in the emissivity of side wall
is marginal, the increase is estimated as 14.9% for an increase of Es from 0.0 to 0.85. On
the other hand, with highly emissive surface ofthe heat generating element (Eh = 0.85)
the effect of increasing the emissivity of channel walls on the mass flow rate is strikingly
noticeable, the maximum increase in the mass flow rate is 70.8%. This considerable en
hancement in the mass flow rate suggests that the average convective Nusselt number
increases due to surface radiation interaction. However, it is important to note that with
the increase of emissivity of the heat source from Eh = 0.05 - 0.85, the mass flow rate
is found to decrease by 2.07%. This is because the convection could be suppressed by a
drop in temperature of the heat source with highly emissive surface. Figure 5.12 shows
85
0.00 '-_........._--1.__""'---_---'-_---'__....1...-_---'-_-'
0.000 0.004 0.008 0.012 0.016
Spacing between the heat source and one side plate(m)
._..._.... Top
-Middle----·Bottom
Ra'=8.1 X 106
...-~Ear 0.30 ....------------'---------------,-.!!!c.Q)
~ 0.25"0s::ccQ)
f:::J 0.20otJ)
1UQ)
.c:s:: 0.15Q)
iQ).cgo 0.100)
Q).c:-VI~ 0.05I(Jcc~'uo~>-
Figure 5.12: Variation of fluid velocity, across the gap between the heat source and oneside wall, drawn at three different heights of the heat source obtained from3D numerical simulations (th = t s = 0.85, Ra* = 8.1 x 106
)
the variation of fluid velocity across the gap between the heat generating element and
one side wall of the channel, plotted at three different heights of the heat generating
element, viz. bottom, middle and top. The velocity distribution is nearly uniform in
the gap c1oseto the leading edge of the heat source with almost symmetric boundary
layers on either side. From middle height on-wards, the fluid accelerates relatively more
over the surface of the heat generating element. Consequently the velocity distribution
becomes skewed with increase of height. However the boundary layer thickness over
the surface of the heat generating element is found to be nearly same. As seen in many
numerical studies reported earlier [7], [9] and [20], 2D numerical analysis of the prob
lem was started with a view to reduce computational efforts. It is interesting to note that
the temperature field predicted by the 2D numerical simulation shows close agreement
with 3D numerical prediction. However, the velocity distribution shown in Fig. 5.12
86
.-...- .... Top-Middle----. Bottom
-.!!J.EQ) 0.30 r-----------------------,10 . ,-........."Q. ( ).:-:.....,....(1) I ../' ",~•....;;.~~---_
"C 'I .. ..."iii 0.25 t i ~ "',CI> ,'.f . '""' " .....c: , J , , .•O· I: ".
"C If ,e'i .co I, ~.Q) 0.20 I i "\e ! ~~ I ~o : . "-f/I .f \. \.... \, \m0.15 ! \.c ( '.~ \Q) i \i 0.10 ! \Co ico '.~ \~ 0.05 Ra':: 8.1 x 106 \f/Ie i(JCO~ 0.00 '--_-'-__'--_----'-__'---_-'-_---J__-'--_~
g 0.000 0.004 0.008 0.012 0.016
~ Spacing between the heat source and one side plate(m)>-
Figure 5.13: Variation of fluid velocity at three different stream-wise locations on theheat source obtained from 2D numerical simulations (Ch = Cs = 0.85,Ra* = 8.1 x 106)
and 5.13 plotted across the gap between the heat source and side wall, obtained from
3D and 2D simulations are distinctly different. It is clear from Fig. 5.12 that the average
velocity increases with increase of height of the heat source, which is due to the possi
ble effect of flow entrainment into the rising plume from the end faces of the channel.
On the other hand for 2D numerical simulation (Fig. 5.13), velocity magnitude remains
same at different heights of the heat source. This observation shows that 2D numerical
simulation for the present problem is inadequate for the accurate simulation of the flow
field.
87
5.5 Summary
The details of experiments and numerical simulations conducted to investigate coupled
laminar natural convection with surface radiation heat transfer from a short planar heat
generating element mounted freely between two thermally insulated vertical parallel
plates along with the results obtained are explained. Heat transfer correlations devel
oped from experimental data are presented and discussed. The results of numerical
simulations have been compared with experiments. The flow field modifications due to
surface radiation is also presented.
A logical extension to the present research is to investigate the heat dissipation char
acteristics of multiple heat sources, since multiple heat sources are encountered in a .
number of electronic equipment that are undergoing miniaturization in their develop
ments. So the problem of natural convection from free standing planar heat sources has
been selected to investigate next.
88