I am using flotherm for simulation of power supplies of various power range. For natural convection(without fan) power supplies, Flotherm doesnt predict temperature accurately. I performed some testing also. We measured the actual power dissipation for main components( Transformer, choke, Fet, Capacitor). After getting the actual power dissipation, i performed Simulation again and compre with actual temperature measurement. I got the difference of -10% to +18%. for me it is huge. It would be great if someone share their ideas.
My two cents:
Turbulence Model = Laminar (as long as you are sure it's laminar)?
Choke Modeling is IMHO often not so easy (material settings, losses in the core etc ...
Gridding ok (if in doubt raise an SR)?
Measurement "setup" in experiment and in simulation identical? I've seen lots of cases where the position of thermocouples and monitor points were not identical ... IR is also critical as in the experiment you have to open the chassis and in the FT model is very often closed ...
Thanks Dirk for your reply!
of course , we need to take care of radiation in this case.
I have put turbulent flow in the simulation. as far as concerned with testing setup and thermocouple location, i monitor it closely and i believe it is correct.
For Boundary condition, i have put 2x above, x below, x/2 right and left where x is dimension of power supply.
In some case i am getting low level stable and solution not converge. I am not sure low level stable predict correct temperature or not.
Low level stable is usually ok. Do you have some high conducting materials (Cu or matierial with k>k(Cu)) inside?
In [Solve/Overall Control] switch on the Field Error for Temperature and solve one more iteration. Visualize the Field Error: Where is it worst? For high conducting materials you often see some random (or even periodic) pattern in the solids in a very small intervall. That's what is causing the low level stability. But you should check ...