Issue running Streamwise periodic flow tutorial

[Cristopher-Morales]

Hi!!

I hope you are doing well,

I tried to run the tutorial case Stream periodic flow and it seems that it does not converge. I am using the current SU2 develop. Also the solution seems to have some unphysical behaviour

the residuals:

additionally, I observed that the solutions files were updated 3 months ago, when I run the adjoint solver using SU2_CFD_AD DA_configMaster.cfg the simulation diverges:

Could someone have a look about this tutorial case and guide me how to make it work again?

Thank you so much in advance!!!

[bigfooted]

There seems to be a problem with SST

1. with turbulence off (Lower Reynolds), everything converges to machine precision and results look correct. Temperature converges very slowly but this might be the periodic implementation.
2. Switching to SA, everything looks fine, convergence to machine precision. Temperature converges slowly.

[Cristopher-Morales]

Hi,

Thank you so much for your check!!

Yes, I run it with SA and the results look fine and convergence is also fine.

[bigfooted]

Problem occurred after tag v8.2.0 If you lower cfl to 10 and wait for a long time, the flow starts to converge but turbulence does not. Maybe the SST modifications of @Bot-Enigma-0 ?

[bigfooted]

Problem started to occur after #2533

commit 0b6e0ad10d5c874ac61b09254e2b583089a4ff32 (HEAD) Author: Rutvik Khedkar [email protected] Date: Tue Jul 29 21:15:44 2025 +0100

Updating the Cross-Diffusion Term Discretisation in k-w SST Model (#2533)

[Bot-Enigma-0]

Can you try it with a lower under relaxation factor? Maybe: MAX_UPDATE_SST= 0.7

[bigfooted]

The underrelaxations don't help, or switching to accurate jacobians.

[pcarruscag]

We also had the bug fix in the eddy viscosity

[bigfooted]

Before commit https://github.com/su2code/SU2/commit/0b6e0ad10d5c874ac61b09254e2b583089a4ff32 after commit: