[Vivio]

Hi everyone,


I am doing some validation of compressible solvers using a simple shock tube case. Initially I was using foam-extend but have had to switch to using ESI for use of its compressible overset solver overRhoPimpleDyMFoam as there is no compressible overset solver in foam-extend.



To this end I decided to compare the single grid solution for rhoPimpleFoam from ESI and foam-extend to make sure there is no difference due to the fork of OF used. I am using the Preconditioned Bi-Conjugate Gradient (PBiCGStab) solver for the velocity field, whereas foam-extend has the BiCGStab solver, which, despite the name, is also pre-conditioned. As the two linear solvers use the same technique, I would expect the results to be the same.


However, the version with foam-extend provides better results for every field except temperature, where it is notably off from where the exact solution is supposed to be.


Is there a reason for this? Is it more likely to be due to the linear solver or the thermodynamic model? Also, is there a consensus as to which gives better perfomance between PBiCGStab and foam-extend's BiCGStab?


The attached results were computed on a grid of 1000 cells, same conditions, I've zoomed in on the compression/expansion shock location for the pressure/velocity/density plots to show the difference in the results.