Manual Partioning
 

Hello,

I have set up SU2 for parallel running, and have obtained faster convergence with the QuickStart NACA model supplied. However when I run my own case it is slower, requires more iterations and returns worse residuals (though still reasonable).

When I look at the flow contours they seem physical, with no distortions or jumps. However when I look at the Coefficient of Pressure and Friction over one of my surfaces there is a point of missing data where I assume the partion cuts through. I have pressure outlet and inlet BCs and I'm wondering whether there is a conflict with them.

My current simulation is implicit, I have tried to run Explicitly though I get a core dump error. I am using the supplied Metis. Is there anyway with SU2 to view the partions? and then control them manually such as specifying a plane to cut through?

Cheers,
Ross

Hi Ross,

Thanks for using SU2, could you please share the mesh and config file that you are using?

Please be sure that you are using the latest SU2 version... we have improved the stability of the code and now it is using ParMetis instead of Metis.

Best Regards,
Francisco
SU2 lead developer

Francisco,

I have used v3.2.9 from Git with the supplied parMetis and it now works perfectly.

Thank you for your help and best of luck with development, it's a great project.

Kind regards,
Ross.