[tafaugl]

Hello everyone,
This is my first forum post so hopefully I don't break any unsaid forum rules.
I am working on modelling the heat transfer in a cyclotron target. The target is aluminum with a rectangular void filled with water and held to a constant pressure. A proton beam heats the water. There are also 6 coolant channels that run through the target. The heat is transferred from the target and water to the coolant, which has a set mass flow rate/temperature. All exterior boundaries of the target are adiabatic since the target is inside a vacuum. The "open" boundaries for the target chamber are treated as adiabatic walls. In reality these are clear windows used for videoing the fluid motion/activity within the target chamber.

I am attempting to solve this heat transfer problem in two ways. The first is to suppress the coolant domain and apply a calculated average heat transfer coefficient to the channels. I calculated the heat transfer using the known mass flow rates for each of the channels using the Dittus-Boelter correlation for a heated fluid flowing through a cylindrical channel.
The second is to apply the inlet and outlet boundary conditions for the coolant and set up the domain interface between the liquid (water) coolant and the aluminum target. I specified this interface as a general connection with conservative interface flux.

I ran a model involving only the fluid flow through the coolant channels to verify my calculations of the heat transfer coefficient and these were within 10% error from the heat transfer coefficients I calculated externally.

The first model met the default convergence criteria, but the results did not align with what is seen physically (maximum temperatures above saturation temperature), so I think that I am modeling this incorrectly.

The second model did not meet the default convergence criteria for heat transfer after 500 iterations. I checked the results file anyways, even though I had little confidence in the results. The results seemed to be much more reasonable compared to our experimental data, but that fact that the convergence criteria has not been met gives me no confidence in the accuracy of this solution.

My questions:
1.) Am I incorrect to expect these models to produce similar results?
2.) Is there a better setting for heat transfer from a fluid to a solid through a boundary?
3.) Does anyone have any insight as far as my lack of convergence being a physics problem or potentially a meshing problem? (I know this topic is discussed in depth in a different area of the forum).
Thanks so much
-Tim

[ghorrocks]

The forum "rules" are here - http://www.cfd-online.com/Wiki/Ansys...ible_answer.3F

Don't worry, your question is a good one

Heat transfer can still occur in a vacuum - by radiation. Have you checked whether radiation is significant?

Your questions:
1) Do you mean final results or convergence history? There is no reason either of them will be similar.
2) I do not know the specific correlation you used, but the approach is usually valid.
3) This FAQ will help: http://www.cfd-online.com/Wiki/Ansys...gence_criteria

Also in conjugate heat transfer simulations like this the solid time scale is usually far slower than the fluid time scale. So using a big solid time scale factor (100 or 1000) is often very good for speeding convergence.