Professional Documents
Culture Documents
net/publication/257241313
CITATIONS READS
18 4,322
5 authors, including:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Habib Zughbi on 05 December 2017.
PII S0735-1933(98)00023-2
A S I M P L I F I E D N U M E R I C A L M O D E L FOR
M E L T I N G OF ICE W I T H N A T U R A L C O N V E C T I O N
M. A. Hastaoglu
Department of Energy Systems, Gebze Institute of High Technology, Gebze, Turkey
N. Sobh
Saudi ARAMCO, Dhahran, Saudi Arabia
ABSTRACT
Ice in a rectangular enclosure is melted by heating from the top, while maintained at its
melting point at the bottom. The other surfaces are insulated. In the enclosure near the hot
region, liquid phase starts forming as temperatures reach values higher than the melting
point of ice. This phenomenon is first modeled by ignoring the effect of natural convection
in the liquid phase. The resulting equations of conservation of energy are solved in each
phase. The motion of melting front is governed by an energy balance at the interface. This
conduction model is verified by applying it on a system for which an analytical solution is
available. The model is then extended to include convective heat transfer in such a way that
the liquid phase is assumed to be a mixed body subjected to natural convection from the top
surface and the liquid-solid interface. The flux at the interface is obtained by finding a heat
transfer coefficient for natural convection with a cold plate facing upward. Comparison of
the results of the numerical work with experiments performed on water/ice system shows a
strong effect of natural convection on melting of ice. The model involving natural
convection in the liquid phase agrees well with the experimental work.
© 1998 Elsevier ScienceLtd
Introduction
Melting/solidification problems belong to a class of heat transfer where there exists a phase
change and its location is not known a priori. Phase change problems are encountered extensively in
359
360 R. Kahraman et al. Vol. 25, No. 3
nature and in a variety of technologically important processes. Such processes include melting of ice,
freezing/thawing of moist soil, crystal growth, latent heat-of-fusion, thermal energy storage, purification
and casting of metals, welding and plastics manufacturing.
Phase change problems have been the subject of intensive research over recent years. A number
of studies dealing with analytical or numerical aspects of particular melting or freezing heat transfer
problems have appeared in the literature [1-9]. It has been observed that most studies are for the case of
cooling or heating from vertical walls and most models lack quantitative comparison with experiments.
Moreover, many of the earlier models for phase change problems are based on conduction type of heat
transfer, both in solid and liquid. However, the actual physical processes show that convection type of
heat transfer may be present in the liquid and often plays an important role [10-13]. For the case of
melting of ice, the melting process might be considerably affected by complex natural convection in the
melt owing to density inversion of water at 4°C. The objective of this study is to simulate the melting of
ice in a rectangular enclosure heated from above taking into account natural convection in the liquid
phase by assuming the liquid to be a mixed body subjected to natural convection from the heated surface
above and the phase front below. This is a short cut method for inclusion of natural convection in the
phase change analysis which reduces the computational time substantially.
Mathematical Formulation
In the following subsections, the governing equations for melting of ice in a rectangular
enclosure are presented. Ice at an initial temperature of To =-30°C is subjected to a temperature of
Thot = 70°C at an upper boundary (z = 0). Temperature of the solid at the bottom (z = 20 cm) is
maintained at T m = 0°C.
3T O2T
for O<z<s(t), t>0 (1)
3 t - ctl 3 z 2
and
OT C2T
- a, for 0.20 m >z > s(t), t>0 (2)
Ot 322
where T is temperature, t is time, a t is liquid thermal diffusivity, ~. is solid thermal diffusivity and s is
Os = k dT~ _klOTt
P'LT[ " 0-2 oz (3)
where/9 is density, L is latent heat of melting, k is thermal conductivity and subscripts l and s refer to
liquid and solid, respectively.
Liquid water is assumed to be a mixed body subjected to natural convection from the top heated
surface and the cold ice surface. In this case, the liquid (bulk) temperature, Tt, plays a vital role in
evaluating the fluxes used in determining the interface location. Once liquid starts forming at the top
where hot surface contacts the solid, a heat balance can be written on the liquid as
heat transfer coefficients are calculated from the following correlation involving Rayleigh number, Ra,
(for hot surface facing down or cold surface facing up) [15]
pl L aOt=lfs-~zz-hmeltITm-Z
s . o~ ,
*) (9)
Finite differencing with explicit technique was utilized. A fixed grid structure was devised. At
grid points near the boundaries, string-intersected approximations to derivatives were used [16-19].
Central differencing was performed to approximate the spatial derivatives at all interior grid points. If a
grid point was close to the melt front (closer than a pre-assigned tolerance, for example 10-5 m), that grid
362 R. Kahraman et al. Vol. 25, No. 3
point was assumed to be at the melting temperature to prevent the singularities that would arise
otherwise,
The physical properties used in the solutions are: liquid density, P/= 993 kg/m3; liquid thermal
conductivity, k/ = 0.624 W/m °C; solid thermal conductivity, ks = 2.367 W/m °C; liquid thermal
diffusivity, a / = 1.505x10 -7 m2/s; solid thermal diffusivity, a s = 1.3164x10 -6 m2/s; liquid specific heat
capacity, Cp/= 4174 J/kg °C and latent heat of melting, L = 333,790 J/kg.
Experimental
Water in a copper container (0.2×0.2x0.2 m 3) was frozen to -30°C with several thermocouples
imbedded within. The ice was placed in an ice/water bath. The top surface of the ice was contacted with
a copper box with hot water at 70°C circulating from a reservoir. The thermocouple readings
and ice front location were recorded during the experiment. There were 10 thermocouples
inserted along the center line of the box at various distances from its bottom. The melt thickness
was determined by observing the temperatures recorded by the thermocouples during an experiment.
Moreover, the experiment was repeated and the melt thickness was also occasionally measured
by inserting a dip stick in the melting box from the top after removing the heating surface temporarily
to support the melt thickness determination through thermocouple readings. The results were not
significantly different.
The problem was solved numerically with and without natural convection. The model ignoring
natural convection was tested by applying it on I-D melting of a solid which is at the melting
temperature and subjected to a higher temperature at a boundary, for which an analytical solution is
available if the solid phase stays at the melting temperature throughout and convection type of heat
transfer in the liquid phase is ignored [14].
Figures 1 and 2 compare the results of the numerical model with those obtained by the analytical
solution. Good agreement with the analytical solution verifies the numerical model for the case of
conduction type of heat transfer. The grid and time step sensitivity of the model was also checked, as
shown in Figure 3. The model was not stable for spatial to time step size ratios (Az/At) smaller than
about 0.001. On the other hand, while it was stable its prediction accuracy suffered for large step sizes
(for Az _>0.1 m o r A t _ > l s).
Vol. 25, No. 3 MELTING OF ICE WITH NATURAL CONVECTION 363
0.035
0.030
E 0.025
0.020
Analytical
0.015
0.010
Solution
Numerical Solution
0.005
0.000 I I I I
FIG. 1
Melt thickness vs time plot obtained from the conduction model
(Ax = 0.001 m, At =0.1 s) and the analytical solution for melting of
ice if the solid phase were maintained at the melting temperature.
80
70
Analytical
60
- - Numerical
50
40
30
20
10
-10
0.000 0.005 0.010 0.015 0.020
Z, m
FIG. 2
Temperature profile at time = 500 s obtained from the conduction model
(Ax = 0.001 m, At = 0.1 s) and the analytical solution for melting of
ice if the solid phase were maintained at the melting temperature.
Advance of melting front obtained from the conduction model for the problem of the present study
is shown in Figure 4. Temperature distribution is also shown from top towards bottom in Figure 5.
364 R. Kahraman et al. Vol. 25, No. 3
0.05
dz (m), dt (s)
0 1,1
" 0.01 1 I
0.04 o oo~, 1 i
-- 0.001,0.1 /
0 0.001.0.001 I~
E 0.01,0.001 ^ ,j.
0.1, 0.001 t~ ~ p
0.03 100000,0.001 ~ /
f:
I-- 0.02
A a a
0.01
0.00 I - - ~ T I
FIG. 3
Melt thickness vs time plot obtained from the conduction model (for
melting of ice if the solid phase were maintained at the melting temperature)
for different spatial step size (Ax) and time step size (At) combinations.
0.20
0.15
E
e-
0.10
0.05
0.00 i i- i i i
0 10 2o 30 40 50 60
Time, h
FIG. 4
Melt thickness vs time plot obtained from the conduction model.
The lower curve is for the case that the solid thickness is at its initial value. The top curve, on the
other hand, is for the case that no solid is left in the box. As seen from the figures, the time required
for the ice in the box to melt completely is 53 hours if natural convection in the liquid is ignored.
Vol. 25, No. 3 MELTING OF ICE WITH NATURAL CONVECTION 365
8O
4O
°~ 20 t=5 h
-20 t=O ~ _ ~
-40 , , ,
0.00 0.05 0.10 0.15 0.20
Z, m
FIG. 5
Temperature distribution from top to bottom of the melting ice
(obtained from the conduction model).
Figure 6 compares the advance of melting front obtained from the numerical solution involving
convection heat transfer with that obtained experimentally for the water/ice system. Since ice lifts after a
certain time, limited experimental values are reported. The results of the simplified model considering
natural convection in the liquid shows good agreement with the experimental work. The time of melting
estimated to be about 10 hours by interpolation of the experimental data is much lower than that
determined by the conduction model, indicating a strong effect of natural convection on melting of ice.
0.10
-- Numerical
• Experimental
E 0.08 e e ee ee ~
° j.-
c~ 0.06
0.04
0.02
0.00 , , ,
1 2 3
Time, h
FIG. 6
Melt thickness vs time plot obtained experimentally and from the
numerical model considering convection.
366 R. Kahraman et al. Vol. 25, No. 3
The model was also verified by comparing the experimentally measured temperatures to their
numerical predictions. The temperature profiles obtained experimentally and numerically (as shown in
Figure 7) were not significantly different.
7o?
60
o Experimental
50 z~ Numerical
40
30
9 20
10
-10
0 £ 0 ~ O ~'
8
-20
-30
0.00 0.05 0.10 0.15 0.20
z,m
FIG. 7
Temperature profile at a time of 3,000 s obtained experimentally and numerically.
Conclusions
Melting of ice in a rectangular enclosure heated from above is modeled, first, by ignoring the
effect of natural convection in the liquid phase. This numerical study for the case of conduction type of
heat transfer is verified by application on a system for which an analytical solution is available. The
model is then extended to include convective heat transfer in the liquid phase by assuming the liquid to
be a mixed body subjected to natural convection from the heating surface above and the ice surface
below. Comparison of the numerical and the experimental results show a strong effect of natural
convection on melting of ice. The model taking into account natural convection in the liquid phase in a
simplified manner agrees well with the experimental work. Future work is planned to develop a more
rigorous and general model considering the temperature dependency of water density and using all
components of the equations of motion and continuity.
Acknowledgments
This project has been funded by King Fahd University of Petroleum & Minerals under Project #
ME/NUMERICAL/184.
Vol. 25, No. 3 MELTING OF ICE WITH NATURAL CONVECTION 367
Nomenclature
References
1. M.C. Liang, and C.W. Lan, J. Comput. Phys. 127, 330 (1996).
2. H. Zhang, V. Prasad, and M.K. Moallemi, Numer. Heat Transfer B 29, 399 (1996).
3. C. Beckermann, and R. Viskanta, Trans. ASME, J. Heat Transfer 111, 416 (1989).
6. W.-J. Chang, and D.-F. Yang, Int. J. Heat Mass Transfer 39, 2333 (1996).
7. G.H. Yeoh, M. Behnia, G. De Vahl Davis, and E. Leonardi, Int. J. Numer. Meth. Engng. 30, 899
(1990).
9. Y.K. Wu, and M. Lacroix, Transactions of the Canadian Society for Mechanical Engineering 17,
281 (1993).
10. R.A. Brewster, and B. Gebhart, lnt. J. HeatMass Transfer 31,331 (1988).
11. D. Gobin, and C. Benard, Trans. ASME, J. Heat Transfer 114, 521 (1992).
13 S. Chellaiah, and R. Viskanta, Trans. ASME, J. Heat Transfer 111, 425 (1989).
368 R. Kahraman et al. Vol. 25, No. 3
15. J.P. Holman, Heat Transfer, 2nd ed., McGraw-Hill, New York (1990).
16. A. Negiz, M.A. Hastaoglu, and R.A. Heidemann, Chem. Eng. Sci. 48, 3507 (1993).
17. A. Negiz, M.A. Hastaoglu, and R.A. Heidemann, Numer. Heat TransferA 28, 175 (1995).
18. M.A. Hastaoglu, A. Negiz, and R.A. Heidemann, Chem. Eng. Sci. 50, 2545 (1995).
19. V. Vemuri, and W.J. Karplus, Digital Computer Treatment of Partial Differential Equations,
Prentice Hall, New York (1981 ).