[Bruno Machado]

Guys, I built an entire fuel cell model using UDF. This steady state model has RP variables, 4 UDSI and many UDMI.

The problem is that, once I close fluent, I can not use the previous result as an initial condition. It simply diverges all the variables at first iteration.

Did anyone have this kind of problem before? Is there any way I should save it or variables that should be stored to start the simulation using the previous results?

Thank you in advance.

[brunoc]

I think Fluent discards all UDS gradients, so if you're using those to assemble the equations at the start it is probably just using garbage, which would crash the solver. Is this your case?

I remember seeing something like this in a UDF from ANSYS. I don't quite remember the workaround, but what I think they did (I don't really trust my memory) was have the solver re-calculate gradients on the fly.

[Bruno Machado]

Originally Posted by brunoc

I think Fluent discards all UDS gradients, so if you're using those to assemble the equations at the start it is probably just using garbage, which would crash the solver. Is this your case?

I remember seeing something like this in a UDF from ANSYS. I don't quite remember the workaround, but what I think they did (I don't really trust my memory) was have the solver re-calculate gradients on the fly.

Hi Bruno,

Do you mean I should compute the gradient from all the UDS before starting it again? In some kind of DEFINE_ON_DEMAND, for example?

Obrigado pela ajuda haha.

Cheers

[brunoc]

Something like that, yes.

Estamos aqui pra ajudar