CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Is it possible to ignore conduction in a fluid zone? (

bawfuls August 11, 2011 14:03

Is it possible to ignore conduction in a fluid zone?
I am using a UDF for this problem but my question isn't really about UDF functionality so I'm posting here.

I am modeling heat transfer and fluid flow in a high temperature participating medium, using a UDF and outside code for the radiation model. (The outside code computes radiation heat transfer and sends source terms to Fluent.)

For bench-marking purposes, I wish to model radiation heat transfer between parallel infinite plates with a participating medium between. However, the bench-mark should only include radiation heat transfer, and ignore conduction in the participating medium, i.e. the fluid. Since I am providing Fluent with a source term for the energy equation (via the UDF), is there a way to tell Fluent to ignore conduction completely? I tried setting thermal conductivity of the gas to 0, but that just gives an error (Divergence detected in the temperature solver.) If i set the thermal conductivity to anything non-zero, the steady state solution will eventually converge based on the conduction, and I won't have a good radiation bench-mark.

Why can't Fluent ignore conduction if I'm providing it a source term for the energy equation?

Amir August 11, 2011 15:37

I think it's not weird to face convergence issues by eliminating conduction term; as you know, this term has dissipative role and by eliminating of that, small instabilities can easily grow without any damping; this issue is more probable by introducing source terms, especially explicit ones.
Anyway, I think you may achieve convergence if you implement other dissipative schemes, such as second order upwind or reducing conductivity in some stages and also good initialization. Share your findings if you can solve this issue.


Micael August 11, 2011 16:50

Did you derivated the source term in respect to T? I mean, do you give any value for dS[eqn] ? That usually help stability.

bawfuls August 11, 2011 21:50

Thanks for the advice, I neglected to mention that I'm also specifying zero velocity for this benchmark, so that make the problem of dissipation worse.

However, I have since found a suitable benchmark that includes radiation and conduction, so for now I'm pursuing that method as it seems more appropriate for Fluent.

All times are GMT -4. The time now is 06:19.