[altair279]

Hi,

I have a simulation of flow past a turbine rotor. I am using pimpleDyMFoam to run the unsteady case. I run it in parallel on 64 processors by doing a decomposePar. After the case has finished running, I did reconstructPar and deleted all the processor* folders. Now when I want to run the case from the lastest time step, I do decomposePar on the last time step and when I ran the case from there, the solution diverges after a few iterations. I am not sure what's happening. If anyone has experienced this or have any idea about this, please let me know.

[Traction]

Hey,

do you use adaptive timestepping ? Have you compared your initial deltaT (defined in controlDict) to the timeStep of the last iteration of your successful first run ? Sounds like there could be a difference between them which could be the reason for your sudden divergence.

[altair279]

Hi,

Thanks for the response. I am using adaptive timestepping and have a maxCo set. But as you said the initial deltaT(in controlDict) is much lower than the last iteration timestep. So when I run it again from the latestTime, should I set the time step to much lower value and let the solution stabilize? I thought because the solution is already converged, I wouldn't have to do this.

[Traction]

Unfortunately I thought that your initial timestep is larger than the timestep of your last iteration. Looks like this approach won't solve your problem. Anyway it could be worth a try to increase your initial timestep to the value of your last iteration.