[Ship Designer]

Hello all,

I'm running ship resistance cases with interFoam in parallel. At the far field and centre plane boundaries, p_rgh anomalies occur in way of the processor boundaries. High and low pressure form next to each other in areas of uniform flow, far from the hull geometry and then propagate downstream.








Typical unaffected p_rgh values at the far field are < 100 pa, at hte anomalies they reach the thousands. The solution converges in spite of this but gets peaks and ripple in the calculated forces which still bothers me:




The anomalies come and go throughout the computations, in some they are more insistent than in others. After noticing that they occur at the processor boundaries, I stopped interFoam, executed reconstructPar for a time where the anomalies were prominent and continued the solution from there in single-core mode. The anomalies disappeared immediately at their point of origin and within 100 time steps also all their traces downstream. I therefore conclude that this is not caused by the boundary conditions (symmetryPlane and wall with slip) or any fvSchemes or fvSolution settings but is tied to the parallel decomposition somehow.

The grid with 2.6 mio. cells in constant is error free, however if I run checkMesh within the processorN folders, some wrongFaces are reported. It turns out, that they are not near the processor boundaries where the phenomenon occurs, yet I find it still odd that decomposePar does this. For it I used:

numberOfSubdomains  4;

            method  scotch;

Does anyone know what might cause and how to avoid this? Anyone experienced this? I'm grateful for any hints!


Have a nice weekend, Claudio