Hello all,

I'm new on heat sink analysis and I would appreciate it if anyone here can give me some ideas on the topic as follows.

The heat sink I'm simulating is not fin-based. It is a small dome-shaped aluminum heat sink (maximum extent=35mm) with a few specially designed through holes which increase surface area. The heat transfer mechanism is natural convection. I've got measurements of heat sink surface temperature and room temperature. The goal of the simulation is to find out heat sink temperature for other scaled heat sinks of the same shape. Now my question is how to determine heat transfer coefficient on curved surfaces.

Thank you, Ken

Use

HTC = q" / ( Tsurf - Tfluid)

For any surface ( in Fluent) you get wall heat flux and then calculate the HTC using surface temperature and average film/fluid temperature.

Thank you very much for answering my question. But I have another confusion. In heatsink analysis, is it necessary to model both heatsink (solid zone) and surrounding air (fluid zone)? In my case, the focus is on heatsink surface temperature due to natural convection. So I modelled heatsink only. Now I'm matching the measured heatsink temperature (in a specific location) by adjusting HTC and for smplicity I'm using only one HTC value for all the surfaces exposed to air. Later, I will apply this HTC to other scaled heatsinks of the same kind. Is it correct to do so? By the way, is the HTC calculated by q"/(Tsurf-Tfluid) an AVERAGE HTC over a surface?

Thanks,

Ken

If you could calculate HTC and apply that could provide you first order solution. Later you could modify the HTC which varies with respect to location etc... Biggest problem is getting accurate prediction of HTC... if you get it right you have a accurate solution...

You could check the results with varying dT or heat load and see how closely you match the results...

Best of luck, AJ

Hello AJ,

Like you said, HTC is the culprit. Although I'm matching the heatsink temperature by adjusting HTC, I feel uncomfortable doing this way because I think it is like making a guess. Just for curiosity, is there any serious way of doing this kind of analysis?

Regards,

Ken