interFoam (OpenFOAM) or general: pressure boundary conditions and tank draining
 

I want to model 2-D incompressible fluid flow out of a tank, where two layers of fluid are initially in the tank. Someone solved this problem years ago for the situation I'm interested in. I'm trying to reproduce the results in OpenFOAM using the interFoam solver with pressure inlet and outlet boundary conditions.

So far I can't reproduce the results using OpenFOAM/interFoam. The results are very different from the original non-OpenFOAM results in terms of velocity. I can't figure out if I'm going wrong somewhere or if there's a bug in one of the codes. I can't find any published benchmarks for OpenFOAM/interFoam and pressure boundary conditions. Someone who has used interFoam told me that it's tough to deal with pressure boundary conditions, but I'm not sure if this means impossible!

I am looking for someone to help me figure out what the problem is. For example, maybe someone could:

1. Provide a result based on a different code that I can compare to the results I currently have. This should help me figure out which code has the problem. Or

2: Help me figure out how to deal with interFoam. I'm a beginner, so maybe there's some numerical setting I haven't set correctly or some other problem.

Any other ideas would be welcome.

I have received a response from Workshop Technologies; if anyone has any experience with using them, I'd love to know. Thank you.

is this what you're looking for?

https://www.youtube.com/watch?v=axcnw8nrNe8

https://www.youtube.com/watch?v=TePoU5lhGVs

Thanks very much for those links, cfdan. I'm looking for a person or for information to help in identifying where my problem is.

A benchmark case that has been run based on interFoam or some other code with pressure boundary conditions would be great. I need to know which code (OpenFOAM interFoam or the other one) has the problem.

Alternatively, someone who knows enough about the numerics to tell me if either OpenFOAM interFoam or the other code may have problems with pressure boundary conditions because of their numerical basis.

Thanks.

Hi SFr,

We can surely help you in this.
We are into CAD Design, CFD Analysis, Machine Design, Automation, Web-Design and Digital Marketing.

We have a proven track-record of handling Multi-Disciplinary projects and give results well within time.

You can see our profile at: http://workshopwale.com and mail us at: adarsh@workshopwale.com

Regards,
Workshop Technologies

Thanks to those who responded. I found someone to help and the problem is now solved.

Hi SFr,

Do you mind sharing the solution?

Thank you in advance!

Hi aiman:

Not at all. It turned out to be related to the mesh. I never would have figured it out or even realized that the mesh was the problem if I hadn't sought advice.

I wanted to deal with two dimensions, so set the front and back patches to "empty." When generating a mesh, snappyHexMesh deformed some of the mesh faces. Not enough for me to see, but sufficiently to generate an error about the mesh (I don't remember the specific error and can't access it right now, but can provide it later if anyone is interested). I couldn't figure out what on Earth the error meant and no matter how many times I remeshed, it wouldn't disappear. My simulations seemed to run just fine, so I ignored it.

The person who helped advised me to set my front and back patches as walls with slip. This immediately increased my velocities. The position of the interface between my two fluid layers is now much closer to the published results I'm using for my benchmarking.

Unfortunately, however, I just realized that I only match the other results for very simple cases--i.e., where the viscosity and density in my two fluid layers are identical and the inlet BC pressure is low. So there's still a problem somewhere. Right now I'm challenging my fairly basic math skills by trying to figure out if there's some fundamental difference between the volume of fluid method that interFoam is based on and standard Galerkin finite element methods (which is what the other code is based on) that mean I'll never get a perfect match.

Regardless, the tip about the mesh has helped a lot.