advice needed on which solver to use

Numerical methods and mathematical models of Elmer
Post Reply
bptsj
Posts: 28
Joined: 22 Sep 2012, 21:16
Antispam: Yes

advice needed on which solver to use

Post by bptsj »

Hello All,

I'm using ElmerGUI to solve a steady-state heat transfer problem of a semiconductor with a heat sink exposed to forced convective air cooling. I have been using the heat equation along with a diffusion defined for a boundary condition on all the heat sink surfaces. I'm setting a heat transfer coefficient and external temperature and want to know if this correctly defines my problem. Even with what I think is a large heat transfer coefficient I don't see the cooling effect I expect.

I have convection set to none in my one heat equation. How is this different then defining a diffusion boundary equation at the heat sink surface? If I select convection as constant in the heat equation and specify velocities does that apply to all bodies selected?

Also, I read a tutorial on the Advection Diffusion solver. How is this different from using convection in the heat solver.

Thank you for your advice and patience :)

Brian
bptsj
Posts: 28
Joined: 22 Sep 2012, 21:16
Antispam: Yes

Re: advice needed on which solver to use

Post by bptsj »

:( I've had no feedback yet.
I am modeling a semiconductor heat sink using Elmer GUI. Are there any examples showing how convection is set up? I see a convection tab in the Heat Solver that allows setting velocities. Does this apply to all bodies selected for that 1 equation? I also see the ability to set an external temperature in the Boundary condition but it is associated with a diffusion flux. Is this appropriate for a convection dominated boundary?
raback
Site Admin
Posts: 4812
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

Re: advice needed on which solver to use

Post by raback »

Hi

In order to account for convection you need to include a flow solver also. There is some tutorial on the Raleigh-Bernard instability that has a coupled heat transfer - cfd problem. Now the downside of this is that

1) You must also model the surroundings for the CFD problem
2) You will introduce a small time scale and probably also size scale resulting to explosion of resources required

Now to overcome problems of the scales people usually use RANS turbulence models with boundary layer models. This is not the strongest point of Elmer so you might not be able to control everything by the GUI. Even without the GUI challenges the natural convection & RANS has not been thoroughly explored in Elmer.

-Peter
bptsj
Posts: 28
Joined: 22 Sep 2012, 21:16
Antispam: Yes

Re: advice needed on which solver to use

Post by bptsj »

Thanks Peter,

I'll dig into the tutorial you suggested and also become more familiar with running the non-GUI solver,

Much appreciated,
Brian
Post Reply