[Artur]

Hi Foamers!

I have recently switched from using U-RANS to LES simulating unsteady flow around mariner propellers. I am interested in having long time-traces of data and so restarting my simulations multiple times is a must, given maximum walltime of 60 hours per job on the cluster I have access to.

I tend to run my simulations at maxCo of about 2.5, starting from a mapped RANS solution. Everything works just fine in the first run. However, when I restart there is a short period of time (5-10 time steps, sometimes more) where forces values become completely unphysical, solvers take far more iterations to converge and adaptive time step falls down by about a factor of 10. After this period things seem to recover and everything works just as well as it did during the first run. I tried various scheme/solver/subgrid model configurations with no effect. The only thing that seems to have an effect is manually editing the deltaT value in the last time directory and setting it to a small value, like the one I used at the beginning of the first run (~1e-7 s).

Question is: If I record time traces of forces on the blades, pressures at various locations, etc., how will I know what part of them to discard due to these perturbations exactly and how may I be sure that these will not affect the flow in some weird way (create/destroy vortices, low/high pressure regions, you get the point)?

Many thanks,

Artur