Heat sink analysis
 

Hi,

I am simulating a electronic enclosure with a huge heat sink (overall dimensions around 30cmX4cmX4cm) on the top of 10W heat source, one inlet and one outlet and 3 fans of 5CFM. I am running laminar analysis due to low Re.
My BC are: Inlet at 55 degrees Celsius and pressure 0; outlet: pressure 0.
Interestingly, if I take the heat sink out of my system the temperature on the heat source decreases!

Any suggestion/idea of what is going on???

Thank you,

it would be helpful to see an image of your study...

possibly the heat sink does not have the correct material properties so it is insulating?

CFDesign v10 and 2011
 

Hi,

Thanks for your repply Mithrandir, but unfortunately I cannot post a screenshot of my model here. It is confidential.

However I have an interesting fact that I could not find an answer. I run my simulation (heat transfer) on CFDesign V10 and on CFDesign 2011 and the results were totally different, how come? The results from CFDesign 2011 seems much more reliable and are close to what I have been measuring on the lab.

hmm... interesting.... is the flow solution the same? or just the thermal?

The flow solution is the same. It is just the thermal solution, any ideas about what is going on?

I dunno....

Prolly someone with a good knowledge of the "Math" involved could answer this better....

but hopefully.. the more current version of the software gives a better solution.....

If you brought the model forward from a previous version, double check the model. The math doesn't change, so the problem lies with either the model definition or the way the pre-processor is handling the model.

If you are modeling the fins in the HS, be sure that the inter-fin flow field is being handled correctly. A transient analysis may also yield some insight into problem area(s).

Good luck!

Having supported CFdesign through all of these versions, the two versions you list have notably different approaches to the thermal solution.
Through the evolution of V10 diagnosing and getting thermal models to run would vary in the method slightly between the 3builds of v10 that existed.

From v2010 moving forward (and in to the current 2014 Simulation CFD) the strategy for thermal has been consistent and much more robust than what was previously done.

Considering you reference 2011 as being more accurate, I would suggest sticking with that and moving forward, not to mention that v10 was not parallelized and is much slower than 2011