Transient Heat tTransfer - optimal solution settings 1

[Alexander Koss]

Dear Forum,

I am currently working on a very large heat transfer analysis, which deals with factors like radiation and self-heating elements. Due to its complexity, it sometimes takes a very long amount of time to produce a solution. Therefore, I would love to hear some opinions from you guys, since I am not very experienced

1. Direct vs. Iterative solution method - would it make a large difference?
2. Quasi-Newton seems to not work for some reason - any idea why that could be the case?
3. Should I extrapolate the steps linearly, or rather in a parabolic way?
4. Matrix Storage: should a certain way be specified, or should it go by the default value?
5. Any other suggestions on speeding up a transient heat transfer solution?

Thank you very much in advance!

 

[FEA way]

It all depends on particular case. But, generally speaking:
- iterative solver should be faster but may result in convergence issues
- quasi-Newton method is a way to achieve convergence in problematic analyses and reduce computational cost
- extrapolation choice is a matter of tests for specific case
- unsymmetric scheme may improve convergence when radiation is used in the model

Other ways of speeding up such analyses are related to modeling approach. For example you can try reducing mesh density (if it doesn’t affect accuracy too much) or introducing some simplifications.