# Peltier cooling simulation in CFD

#1

Hey, I’m trying to simulate a cooling system in a Peltier sensor, like the one below:

However, I’m facing a few issues to model the problem using conjugate heat-transfer analysis. Besides the cooler, heatsink and peltier in my project, I’m using a bigger block to simulate the air domain.
Basically, the heatsink starts at 25 ºC and the peltier has a constant heat flow of 55 W additionally, the cooler has a mass flow of 4 kg/s. The boundary conditions in the simulation are Velocity Inlet/Pressure outlet for the cooler, contact resistance between the heatsink and peltier and the heatsink has the temperature model as turbulent heat flux. The link for the project: https://www.simscale.com/projects/iaguiar/projeto_transcal/

#2

Hi @iaguiar!

Can you first make sure to set the project to public? Regarding the setup, is the cooler considered a solid or will the cylinder contain a fluid additionally dissipating heat from the heatsink?

We definitely have to make sure that the peltier sensor touches the surface of the heatsink and the surface of the heatsink touches the cooler in order to establish a proper CHT meshing process. I think we can work around the inlet-outlet condition for the cooler and let the component cool actively or passively. @CFD-SQUAD, feel free to add your thoughts here.

Cheers!

Jousef

#3

Hey @jousefm just set the project to public. The cooler is supposed to be solid, and the only fluid it is using is the air inside the domain. The peltier surface is in contact with the heatsink, as well as the cooler.

#4

Hey there. I think the boundary conditions on the cooler should not be there if it is supposed to be a solid. Try setting some velocity inlet boundary condition for your air domain. Also set a pressure inlet/outlet boundary condition somewhere for your air domain. These two boundary conditions should make sure you would properly simulate the flow of air inside your air domain.

If I understand correctly, you have 3 solids touching (so there are 2 interfaces) and exchanging heat with each other? I do not know about whether CHT is able to do more than one contact interface in SimScale, maybe some of the other guys can tell you more about that.

#5

@tcakir @jousefm I got confused in translating the terms, sorry about that, the cooler is supposed to be a fan, but the only way I thought was to do a solid part like that because I don’t know the geometry of the blade, just the mass flow and the respective rotation (rpm). I’ve chosen to do a separate domain of air because I was not sure if only the fan would be enough to “pump” air into the heatsink instead of using it as an air domain.
Thank you for helping me!

#6

I suggest you to take a look at @Retsam https://www.simscale.com/projects/Retsam/coanda_effect_drone_study/ project. Here he defines a momentum source if I remember correctly. It allows you to emulate a fan effect.

The air domain is needed since your computational domain has got to be bounded somewhere right? The air volume is the physical volume in which we perform the simulation, so you need that. Your project however also reminds me of @anirudh2821998 https://www.simscale.com/projects/anirudh2821998/gtx_fan_cooling_effect/ project Perhaps take a look at those project, if not we can try to guide you further.

#7

Alright @tcakir I’ll take a look at it. Thank you!

#8

Well, this is a special cooler with air flow of 415 m/s? Please tell me first how you can do it, with a cooler which has 10 cm of diameter. Well, diameter has nothing to do with that speed, as you go well over Mach 1. In your case it is possibly 4 kg/hour.

Also, in my opinion, this should be simulated as ‘external CFD’ and you will define your domain as BMB (with appropriate size and walls being ‘open air)’, so zero gradient walls. Your BMB does not need to be so ‘fine’, start with coarse grid (or manually using like 40 x 40 x 40 grid first. This could allow quicker cycling over successive meshes / simulation setups.

@tcakir provides already a couple of pointers, so please really take a look at them.

Cheers,

Retsam

#9

Yeah, I’ve made a mistake, the actual value is 0.058 kg/s. Thank you for the advice about the external mesh, i’ll try in this way.