Remove mesh velocity from VOF scaling #1328
Conversation
|
@mbkuhn I have to rebuild my spack environment to test this due to some recent changes to dependencies in spack. The changes here should remove mesh_velocity from the thickness of the interface in Nalu-Wind. I am in the process of fixing my spack and will test it out on a spinning patch |
|
I'll try to test this today! |
|
do you expect this to be equivalent in the static case? Because I'm not seeing that |
|
The two are similar after 500 steps of the flat interface problem, but I was expecting them to be closer. |
|
PR vs master (velocity_magnitude)
|
|
They will not be completely the same for static cases since the volume flow rate includes pressure gradient terms in Nalu-Wind. This is due to Nalu-Wind being collocated and requiring pressure terms in the volume flow rate. Using straight velocity, as is done here, removes all pressure and mesh velocity influences |
|
Sounds good, let's go ahead with it, then. In this static case, the interface is thicker in the center because there are smaller spurious currents there, but that is just the nature of the case. In the moving case, it made a clear improvement to the consistent thickness of the interface. |
This PR uses a computed face velocity rather than the volume flow rate to calculate the velocity scaling of all VOF sharpening and diffusion at liquid/gas interfaces. This removes erroneous dependence of such interfaces on the mesh_velocity, which appears in the volume flow rate.