Hi,

I have a problem with the simulation of „Cooling fins", rather the Wall Heat Transfer. My simulation deals with a horizontal, cylindrical cask containing an energy-source. There are "Cooling fins" on the surface of the cask to increase the Heat Dissipation. Due to the quite huge dimensions it is not possible to build such a fine mesh to simulate the cooling fins in real. So I have to model them in another way. I have some ideas, but they are either physically not trustworthy, or it seems they cannot be implemented in CFX. Hence I have to model the wall of the cask for temperature-distribution on the one hand, but on the other hand the fluid outside for some fluid mechanical issues, it is not possible to reduce one of the components to "Wall Boundary Condition" and therefore I need a domain interfaces (fluid-solid) between the two domains.

But how do I increase the Heat Flux (rather the Wall Heat Transfer Coefficient) from the solid domain to the Fluid domain in order to take account to the cooling fins? I know there are "User Defined Wall Functions" to achieve that on walls, but they are not applicable for Domain-Interfaces? Or does anyone have an idea how to do that: Ways I tried so far are:

1. Between Solid and Fluid I defined a "Porous Medium" for the cooling fins. But this increases only the velocity loss, but without an increase of the Heat Transfer

2. Additionally I defined energy sources on the domain interface to subtract energy on the solid side (by a factor of round about 2 regarding the normal Heat Transfer, because of the surface increase by the cooling fins), and to add it again in the whole domain of the porous medium. Here I have the problem, that I am only able to calculate an average heat flux over the surface, which is contrary to the fact, that the temperatures (and therefore the heat flux) are locally different. Also this method seems to be quite dominant to the heat transfer, so I fear I will have a huge deviation from reality.

3. I tried to increase the "wall roughness", but once again the influence on my Heat Flux is too small

The Hotline does not have any better ideas. They mentioned in CFX12 (??) will be the option to manipulate the "Wall Heat Transfer Coefficient" on Domain Interfaces, but I am working with CFX 11. Are there other ways to increase the Heat Flux between the domain interfaces? Are there other ways to simulate cooling fins without modelling them? Or is it simply impossible? As mentioned, it should make physically sense.

I am looking forward to answers.

With best regards

B. Simon

(Sorry for the long text... But I feared to neglect important things)

Hi,

A few points:

1) Accurate heat transfer modelling with coarse grids is difficult. You may be forced to use a mesh resolution around y+=1.

2) You can play with the wall function prandtl number to increase the heat transfer but this is not recommended. The default value is widely accepted so you are just covering up an error elsewhere in the analysis by doing this.

3) Could radiation be a source of significant heat transfer?

4) I would not use a porous medium to model the cooling fins. It has not been designed to do that.

Glenn Horrocks

Thanks you for your reply,

1.) Yes, my final mesh will be around y+=1. But at the moment I am testing my ideas with a quite coarse mesh.

2.)This I think I will check, even if it is not recommended. Thanks.

3.)I included a Radiation-Modell in my simulation; "Monte Carlo" due to the Domain-Interfaces. Yet I know there will be an error because of my smaller radiation surface (without fins)

4.) Thats a pity. Then I have to look for another backround to this problem

Thanks for your advice,

B.Simon

Hi I was just wondering where you're up to with your problem, because I was trying to model something very similar. I ended up leaving it but I thought I worked out a strategy. My problem involves a cooling coil with an extraordinary amount of fins (Solidworks has a hard time creating the geometry) and air goes through it. I thought a would model in great details a smaller area to calculate accurately the pressure loss for a range of velocities and the heat transfer from the fins. then I would apply this data to a porous media. That's the only way I worked out but maybe my problem set up is different from yours?