Discussion Forum

Posted: 1 decade ago 9 déc. 2009, 01:41 UTC−5

Your question (part a) is a little ambiguous. I assume you are talking about the convective heat transfer coefficient, h. This coefficient is not associated with a domain but with a fluid/solid interface. Thus, you would want to impose this (technically called a Neumann boundary condition) on the interface touching the air (I think it is the air/glass interface in your problem). You will need to choose a reasonable value for h; say, h=10 in SI units for natural convection, or use some common engineering correlations to estimate its value. I suggest that you do a little bit of a sensitivity study wrt h within a reasonable range to see if your results strongly depends on it. If that turns out to be the case maybe you need to consider another more definitive method to approach your problem such as solving the actual air flow on the outside as a coupled fluid/thermal problem. Further, you may need to consider whether radiation heat transfer is important as well. As for part b - there are various ways of exporting the results. I would consult the COMSOL documentation for that. Hope this helps, Ozgur [QUOTE] Hi guys, This is more of conceptual question. I am modeling a transient heat transfer of two layer mirror system in COMSOL. A thick substrate layer (5cm) made of glass and the top thin coating layer (100nm) of gold. This gold coated mirror is initially at higher temperature (50 deg C) and I want to model the cooling time required for the mirror to reach the ambient/air temperature (25 deg C). The mirror cools by dissipating excessive heat to surrounding air from the coating layer.All the boundaries are insulated except the top gold layer. My basic doubt is related to the specification of heat transfer coefficient in Physics->Boundary Setting in COMSOL. a) Which of the three heat transfer coefficients (air or Au or the glass) could one use to correctly model the cooling rates of the mirror? b) How do I export/save the 2D temperature data in COMSOL for the reflecting layer of the mirror after certain time lapse? This final temperature distribution is required to model the behavior of the mirror in different application outside COMSOL. I thank the COMSOL user community in advance and looking forward to hear your expert views and comments. -Sandra [/QUOTE]

Posted: 1 decade ago 9 déc. 2009, 02:26 UTC−5

Thanks a lot for that clarification Ozgur. If I understand it correctly what you say is that the heat transfer coefficient h would depends on the nature of fluid/solid interface. So its value is going to be different for different cases, e.g. glass/air, metal/air, glass/water etc. That sounds reasonable. Another point is that the metal coating layer is extremely thin compared to the substrate thickness. So would that have any bearing on the choice of h? -Sandra

Posted: 1 decade ago 9 déc. 2009, 03:09 UTC−5

In fact, h depends only on the fluid properties, nature of flow and the geometry. It is completely independent of the material solid phase is made of. [QUOTE] Thanks a lot for that clarification Ozgur. If I understand it correctly what you say is that the heat transfer coefficient h would depends on the nature of fluid/solid interface. So its value is going to be different for different cases, e.g. glass/air, metal/air, glass/water etc. That sounds reasonable. Another point is that the metal coating layer is extremely thin compared to the substrate thickness. So would that have any bearing on the choice of h? -Sandra [/QUOTE]

Posted: 1 decade ago 21 déc. 2009, 18:02 UTC−5

In addition, you can model the convective heat transfer processes by coupling your model with weakly compressible navier stokes equation.In that case, a buoyancy driven convective flux can be solved. See a sample model in Model Library>Heat transfer>Tutorial>thermos laminarflow & thermos lamina hcoeff. This models give clearer picture of the problem

Posted: 1 decade ago 22 déc. 2009, 08:33 UTC−5

Hello I can add the "h" figures I commonly use as "engineering" values, taken from the litterature, and to be understood as "as is": solid air intrface low convection exchange : 2[W/K/m^2] typically for optical domain when you want to minimise turbulences for more natural environment, i.e. electric circuit boards, but no forced convection: 5[W/K/m^2] for forced convection you can go up to 4-6 times higher Good references are "handbook of heat transfer" Mc Grah Hill, or "Convection heat transfer" by Bejan, Wiley, or "Convective Heat and Mas transfer", by Kays, Mc Graw-Hill In your case with only 50°C temperature difference radiative exchange can be most probably neglected, its slightly more tricky to use as you need realistic emissivity values to and to know your surroundings further litterature "Thermal Radiative Heat Transfer" by R. Siegel Taylor and Francis Good luck Ivar