Internal Fan simulation doesn't produce temp changes.

[JohnTomasik6493]

I set this up based on the tutorial at the link below.  There doesn't seem to be much heat conduction to the heat sink.  I've changed the LED material type (from PCBA) and there was no change in temp.  I also removed the 0 pressure boundary conditions of the air surface at the bottom of the LED's, with no change in the temp.  I've even increased the power from 150W each to 900W and the temp remained the same.  

Any suggestion?....besides doing something else for a living....lol.

[srhusain]

Looking at your model, it appears that the part with ids 4, 10, 15, 20, 25, and 30 are assigned steel material- the fans attached to these will not work correctly. You need to assign these as fluid parts.

[JohnTomasik6493]

Ok, I'll change it, but curious why the fans can't be attached to housings, even if it's just a very small edge?  There's an internal air part on either side of the fan's flow direction.  It seems to flow correctly.

[JohnTomasik6493]

It didn't work.  It still has the same thermal result.  Next suggestion, please.  Oh, and now it does not flow correctly...the air is blown  all over the housing instead of directed over the components to be cooled.

[srhusain]

Try using a slip factor less than 1

[JohnTomasik6493]

Didn't change a thing when it comes to temp.

[srhusain]

[Royce_adsk]

Hi John,

Lets see if we can get this moving forward for you.

The internal fan material is designed to have a full face on the inlet and outlet side. Make the fan a smaller diameter so that it is the same size as the ID or the housing that it is blowing into.

Why do you have the small air blocks below each unit with a zero pressure?  That configuration seem a bit odd, was that your intention?

Your outlet extension may need to be a bit longer. It needs to be long enough to have fairly developed flow, if you still see recirculation near the outlet then you need to make it longer.  

The inlet extension should be a little longer as well so that the inlet temperature boundary condition isn't co-planar with the housing which can create non-physical conduction into the housing.  I would accomplish all of this by creating Solid Hockey puck looking caps in CAD so that the model will automatically void fill when launched into CFD instead of leveraging the void fill tool in CFD.  Creating the caps in CAD is considered the more superior workflow.

Could you make those changes and address the question I had?  Once you setup the model again could you attach the new support file?

Thanks,

[JohnTomasik6493]

Royce,

Do the housings that the fans are connected to need to be air as requested prior?  

The blocks of air was my attempt to conduct/convect heat away from the LED boards on the outside face of the board.  I don't know what the software assumes on the outside surface of the LED board.  Will it behave as though the entire unit is surrounded by air?

I'll do the other changes.

John

[Royce_adsk · Accepted solution]

Hi John,

The housing does not need to be air, but it was a workaround to allow the fan to operate more as intended.

Also, if I run your model out it does produce a temperature change.  Maybe it is more expectation on when you should see the temperature change.  Since you are running the auto forced convection option you will not see any temperature change until the flow has finished solving which should be no more than 300 iterations (as you have defined).  It might stop sooner based on convergence criteria.

The picture posted earlier is of the finished result with no changes.

In terms of the LED board, I assume the round circles would be where the LEDs would be located, the pressure boundary condition isn't really the correct way to capture this.  A more appropriate method would be remove the 0 pressure and just apply a film coefficient like 5W/m2/k with the ambient temperature as the reference. It would be shortcut way to model this interaction. Otherwise, we would want to consider modeling the external domain to capture the heat convecting away from the LED and that face of the board.

Since making the assumption that if we neglect this possible means of heat to escape would be conservative in regards to operating temperature, I wouldn't model it and just assume that all the heat is driven into the heat sink when examining the performance of your design.  In this case I would remove the pressure boundary condition and just leave it as stagnent air with no boundary conditions applied to that face.

[Royce_adsk]

Hi John,

How are the changes coming along?

[JohnTomasik6493]

Royce, 

Yeah, it seems to be working.  Thanks for your help.  

John