[nakonakata]

Hi everyone, I have encountered a discussion about SU2's old version (from 2019) in this link: https://github.com/su2code/SU2/issue...ment-460811699 regarding solver accuracy & speed of SU2 at the version of that time. This stirred up some concerns in my head about the issues about numerics and linear algebra in compressible CFD solvers.

What else can you say about the issues regarding not able to go up to high CFL numbers for an implicit, coupled density-based solver (not just SU2) in high Reynolds number RANS simulations? Are the reasons listed below possible causes of this?

1. Using 1st order approximations when forming the flux Jacobian of spatial terms' residuals

2. Not having good enough convergence at inner iterations for linear solver (i.e. GMRES with some kind of preconditioner)

3. Not having good enough convergence at outer iterations

4. Using aggresive limiters for gradients of flow variables

5. Not coupling the turbulence variables to the implicit system (i.e. solving them in a segregated manner after the coupled system of continuity, momentum and energy equations)

6. Mesh quality issues near the no-slip wall boundaries

I'd be glad if you can refer to some relevant discussion/article you have encountered before similar to the link above about these matters if you'd like, thank you!

Best,
Atakan