[Hixon]

Hello everyone,
I have got a problem that different solutions were generated for a single case on different PCs.
I am trying to do some simulation about the laser cladding process. My
case mainly contains a CLSVOF model and a udf energy source at the liquid-gas interface.
And I have two dell workstations. The Old One is T5500 with dual CPUs, 12 cores 24 threads, win 7 OS. The New One is T7910 with dual CPUs,20 cores, 40 threads, win 10 OS. Fluent 18.0 installed on Both.
The solution on the Old workstation shows that the highest temperature in melting pool is about 1950K, and the profile of the melting pool matches with my experiment results.
The solution on the New workstation shows that the highest temperature is below 1800K, and the volume of the melting pool is much smaller than the previous one. So it seems there is a energy leak?
I've tried using same threads (20t), using a newer version (19.1). But the difference remained.
Anyone has any idea about this ? Can anyone help me?

[Leonardo.flores]

I would run it in serial for some hours to discard any parallel problem. And would check if the domain is splited in the same way in both pcs when running in parallel.

[Hixon]

Originally Posted by Leonardo.flores

I would run it in serial for some hours to discard any parallel problem. And would check if the domain is splited in the same way in both pcs when running in parallel.

Thanks for your suggestion.
However, my calculation takes 3 days using 20t, so I may not run it in serial.
I've tried to re-partition the mesh using different methods, but the results show that partition may have nothing to do with my problem.
Thanks again!