[Tommy Chen]

Hi Guys. I have ran a wing-body case at mach 0.85, the flow converged well. However when I solved the adjoint, the res always diverged, I used the same settings as the steady M6 RANS optimization case. So anyone has experience on tuning the adjoint sovler parameters to handle this ?

[economon]

Hi Tommy,

Have you tried to adjust the 4th order dissipation coefficient for the modified JST scheme when solving the adjoint? You can try to increase the dissipation to see if you get convergence. Possibly switch to the Roe scheme and activate the limiter.

The typical strategies for converging the flow problem also apply... check that the CFL isn't too high and try turning off the multigrid too.

Hope this helps,
Tom

[Tommy Chen]

Originally Posted by economon

Hi Tommy,

Have you tried to adjust the 4th order dissipation coefficient for the modified JST scheme when solving the adjoint? You can try to increase the dissipation to see if you get convergence. Possibly switch to the Roe scheme and activate the limiter.

The typical strategies for converging the flow problem also apply... check that the CFL isn't too high and try turning off the multigrid too.

Hope this helps,
Tom

Hi Tom:

Is that OK if I solve the flow using JST but solve the adjoint using Roe ?