[dasssj]

Hi,
I'm using SU2(3.2.7) for CFD research. In my case(RAE2822, C grid, Ma = 0.729, Re = 6.5E6, AoA = 2.31),the solver can not converge in 5000 steps. It takes so much time so I think maybe there are some options in the configuration file have not been properly set.
My cfg file and the mesh file are attached bellow. Can anybody tell me what should I do so that the solver will converge a little faster?
Thank you!

[Santiago Padron]

After looking at your config file, I would suggest you turn on multigrid to greatly reduce the number of iterations.
You can change it from MGLEVEL= 0 to MGLEVEL= 2 or 3.

[dasssj]

Hi, Santiago, thank you for your reply.
I do as you said, but this time it divergences when I set MGLEVEL= 2 or 3. Is that any other option I have to change?

[fpalacios]

Hi,

I agree with Santiago, if the MG works then it is one of the best methods. You are running a C grid and the MG should help. Could you please try with

MGLEVEL= 2
MGCYCLE= V_CYCLE
MG_PRE_SMOOTH= ( 1, 1, 1, 1 )
MG_POST_SMOOTH= ( 0, 0, 0, 0 )
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 0.75
MG_DAMP_PROLONGATION= 0.75

On the other hand, If you are going to run RANS I would use
CONV_NUM_METHOD_FLOW= ROE-2ND_ORDER

In fact, looking at your config file, I have noticed that you are running an old version of SU2. Could you please try with the v4.0.1? In fact, in that version you will have a new option called adaptive CFL. And you can try something like:

CFL_ADAPT= YES
CFL_ADAPT_PARAM= ( 1.5, 0.5, 1.0, 25.0 )

Best,
Francisco

[dasssj]

Thank you for your reply, Francisco.
I falied to compile v4.0.1 on windows 7,so I still use v3.2.7. I change the options you suggested except CFL_ADAPT and CFL_ADAPT_PARAM, but it seems that the sover still converge slowly.
I'm not really understand what do these options mean:
MG_PRE_SMOOTH= ( 1, 1, 1, 1 )
MG_POST_SMOOTH= ( 0, 0, 0, 0 )
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 0.75
MG_DAMP_PROLONGATION= 0.75
Is there any document about these options?
Thank you very much！

Jason

[dasssj]

Originally Posted by fpalacios

Hi,

I agree with Santiago, if the MG works then it is one of the best methods. You are running a C grid and the MG should help. Could you please try with

MGLEVEL= 2
MGCYCLE= V_CYCLE
MG_PRE_SMOOTH= ( 1, 1, 1, 1 )
MG_POST_SMOOTH= ( 0, 0, 0, 0 )
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 0.75
MG_DAMP_PROLONGATION= 0.75

On the other hand, If you are going to run RANS I would use
CONV_NUM_METHOD_FLOW= ROE-2ND_ORDER

In fact, looking at your config file, I have noticed that you are running an old version of SU2. Could you please try with the v4.0.1? In fact, in that version you will have a new option called adaptive CFL. And you can try something like:

CFL_ADAPT= YES
CFL_ADAPT_PARAM= ( 1.5, 0.5, 1.0, 25.0 )

Best,
Francisco

Thank you for your reply, Francisco.
I falied to compile v4.0.1 on windows 7,so I still use v3.2.7. I change the options you suggest except CFL_ADAPT and CFL_ADAPT_PARAM, but it seems that the sover still converge slowly.
I'm not really understand what do these options mean:
MG_PRE_SMOOTH= ( 1, 1, 1, 1 )
MG_POST_SMOOTH= ( 0, 0, 0, 0 )
MG_CORRECTION_SMOOTH= ( 0, 0, 0, 0 )
MG_DAMP_RESTRICTION= 0.75
MG_DAMP_PROLONGATION= 0.75
Is there any document about these options?
Thank you very much！

Jason