code_saturne User's Forum

Hi, I have met a problem when coupling the simulation. First I independently run the fluid and solid simulation on CS and Syrthes, both using 4 processors, the simulation can run successfully. After that, I set 1 processors for the solid domain and 1 for the fluid domain in the coupling_parameter file, and execute runcase in terminal, it can start the calculation. After that, I set 2 processors for the solid domain and 2 for the fluid domain, and there appears an error, which is shown on the screenshot, I'd like to know what is the problem? Thank you! (Previously I've reinstall the MPI library, and reinstall Code Saturne).

Hello,

Did you check the log files indicated in the message? what do they say ?

Regards,

Yvan

Hi, I've read the Syrthes.log and listing file but I can't understand what they are talking about. Thank you.

Hi:

Previously I set 4 processors for CS and 4 for Syrthes, and it failed like this. This time I set 4 processors for CS and 2 processors for Syrthes, and it can run the coupling simulation, but the calculation finish very fast. I click into the result folder of the monitoring point, only one iteration has been run. Thanks for the time.

Regards
Jundi He

Hello,

What partitioning method are you using for Syrthes ? For this very small case, using Metis may lead to a rank with an empty mesh, which crashes Syrthes.

Try Scotch on the Syrthes side (though I'm not sure the Code_Saturne 4.0 handles this part of the Syrthes script; it does in version 5.0, with a first beta which will be released this week), or a larger mesh

Regards,

Yvan

Hi, I've solve the problem, by setting only one processors for Syrthes, if I set 2 processors for Syrthes, it will crash like the first picture, which tells you some MPI problems. Like I have 8 processors in my computer, I can only use 1 for Syrthes, the other 7 for Code_Saturne. In this setting, I can run the coupling simulation with fluid and solid domain. Thank you.

Regards!
Jundi He