Hello preCICE community,
I am coupling a steady-state CFD solver with a transient material response solver using preCICE.
The material solver advances in physical time, while the CFD solver is used to obtain a steady-state solution for the current wall boundary conditions, such as wall temperature and blowing rate.
I am considering a serial implicit coupling scheme. From the documentation, I understand that participants should write a checkpoint at the beginning of a time window and reload it if the coupling iteration does not converge.
My question is:
Should the steady-state CFD solver also perform checkpoint/reload in implicit coupling?
My concern is that if the CFD solver is rolled back after every non-converged coupling iteration, its residual is reset each time. Then the CFD residual cannot decrease cumulatively across coupling iterations unless the CFD solver is sufficiently converged within each individual coupling iteration.
So, for a steady CFD solver, is the recommended approach to fully or sufficiently converge the CFD solution inside each coupling iteration before writing data to preCICE?
Or is it acceptable to continue the CFD solution from the previous coupling iteration without rollback?
Best regards