[Kellis]

Hello all,

I am interested in a two-phase flow problem of counterflow between a gas and liquid film which runs down a flat surface. This can be simplified into two dimensions: the direction of the gas flow, and the direction normal to the wall which the film runs down. In this case, the film domain contains only a single dimension, aligned with the gas and film flow.

As a preliminary exercise I am testing the performance of the film module released as part of OpenFOAM v11 in a single dimension. The test domain is a 1.0 m x 0.05 m x 0.001 m box, with 100 cells in the long direction and 1 cell each in the remaining directions. Film runs down the wall from the top and out the bottom. I have the boundary conditions set up following the rivulet panel case, using the filmWall patch type on the wall, empty on the film surface, and patch types for the inlet at the top of the domain, and the outlet at the bottom. For the side surfaces, I have tried several boundary condition types: symmetry, cyclic, empty, and zeroGradient. However, regardless of the boundary condition type I choose, there is a sharp jump in the velocity field in the first cell below the inlet, which propagates down the remainder of the domain. The inlet value is specified as 0.0412 m/s, but the velocity magnitude throughout most of the domain is ~0.14 m/s, over 3x higher than the prescribed inlet value. The film thickness decreases slightly, from the inlet value of 0.173 mm to a value of 0.12 mm within the domain, but this does not compensate for the increased velocity.

Obviously, this leads to a significant conservation error. I have tried a variety of solutions: increasing the number of cells, various boundary conditions, more correctors, tighter solution tolerances, etc. However, none of them seem to fix what should be a very simple problem. Am I missing something critical in my case? Or is the film formulation just not compatible with 1D simulation? Any insight would be appreciated.

I have attached a simple case which replicates this error for those interested in experimenting.

Thanks,
Kellis