Is OpenFoam Slower Than Starccm?
 

Dear All,

Greetings!

I am running, 2D simulation with standard kepsilon model, I have around 100k grid cells.

I run the 10 sec, simulation in Starccm V6. it takes around 4 days in single processor with all second order dicretization.

but the same simulation if I run in OpenFOAM by parallel computing (2 processors) to complete 10 sec simulation is takes 6.5 days. Single processor takes more.


My fvSchemes and fvSolution file are attached here. How can I speed up my case?

Please, help. I will be ever grateful.

Attachment 19930

Attachment 19931

In the fvSolution file, Try GAMG (multigrid method) to solve p and pFinal.

And I think that you should avoid under relaxation when using the PISO method. Under relaxation should be used when using the PIMPLE method and employing more than one outer corrector.

Greetings to all!

Another possibility is the Operating System used. Since the initial attachments are of type "docx", that leads me to suspect that both Star-CCM+ and OpenFOAM were executed on Windows. It's best that you use both on Linux, for the best performance!

Yet another note: 2D in Star-CCM+ is probably optimized; while OpenFOAM always does things in 3D, although it can disable the 3rd dimension and run things as if it were 2D.

Best regards,
Bruno

Dear fsaltara and Bruno,

Greetings!

Thank you very much for your valuable suggestions. I run Starccm in windows and OpenFOAM in Ubuntu.

First, I run Kepsilon model without under relaxation than I found the simulation was used to explode after 10 to 15 iterations. Then I used UR to make the simulation stable. I will remove and try again as u guys have suggested.

Whatever, Thank you very much for our kind advice.