Optimization of a Transient Heat Transfer Analysis 1

[Alexander Koss]

Hello everyone,

I am currently working on a complex heat transfer model which deals with some self-heating objects exchanging heat with each other. The analysis takes a very long period of time (on a server, much more powerful than my pc), and usually results in a crash somewhere midway through it, because it can't achieve equilibrium, even with a 10^-8 minimum increment size.

Are there any ways of optimizing the analysis? Potentially, change some settings in the job or step, or potentially the heat flux equilibrium criterion can be manipulated as well.

 

[FEA way]

Try increasing max allowable temperature change per increment and reducing initial time increment. Avoid using very low minimum increment size. You can also try switching the solution technique to quasi-Newton.

If you take a closer look at the message file, it should point you to the reason of nonconvergence.

 

[Alexander Koss]

It seems that the reason is "Heat flux equilibrium not achieved within tolerance". It crashes midway after about half the simulation is complete due to too many attempts for an increment.
I'm going to try to solve with quasi-newton technique now, with an increased temperature change per increment.

In my solution sometimes self-heating occurs at around 250°C/s, what would you say would be an appropriate minimum increment size?

 

[Alexander Koss]

Also, do you think potentially switching to iterative solution method would help resolve the problem?

 

[FEA way]

The choice of initial time increment is much more important (there's even a simple formula that allows you to determine proper initial increment size for particular problem). For minimum time increment it should be enough to just reduce it 2-3 orders of magnitude from default value.

You can try with iterative solver but I doubt it helps.