Problem with pressure on processor patches with buoyantBoussinesq solvers
 

Hello,


I have been using buoyantBoussinesqPimpleFoam to simulate a fluid with two layers of different density in a box with the moving lid. The solution is good if I run it in serial, but in parallel there is a strange behaviour of velocities between processor patches. I have found the same behaviour in OpenFOAM 3 and 5 and in buoyantBoussinesqPisoFoam of FOAM-extend 4.



This behaviour appeared in both, laminar and turbulent cases, but in laminar it does not seem to affect the solution, where in turbulent, both LES and RAS, it becomes dominant part of the solution.



I checked velocities before solving for pressure on the first time step and this behaviour is absent. It is also absent if I switch off gravity term. This leads me to conclusion that the problem is in linear solver for pressure, but it appeared with all linear solvers that I have tested.



I am attaching a test-code which can reproduce this and an images of this behaviour in laminar and with LES.


Any help is welcome!:)


Best,
Sara


EDIT:

Problem solved on https://develop.openfoam.com/Develop...us/issues/1047 .