Hello, STAR users,

I am encountering a case in which inflow occurs at oulet boundary, i.e., a small portion of fluid flows into the domain at the boundary. This inflow is not negibile! Therefore, the default outlet boundary condition built in the STAR is not suitable for the boundary. Also, a prescribed pressure distribution is not available, which ctually is what I want.

So, how should I treat this boundary? I'll appreciate any suggestion/idea.

Thanks,

Jiaying

In addition, I can't change the location of the boundary to somewhere no flow recirculation.

For your imagination, the flow is traveling through a geometrically periodic structure. To save the computation cost, I limit my computation domain in a periodic unit of the structure. The flow recirculation seems not avoidable.

Jiaying

You can describe a prescribed pressure at the outlet, using a 'pressure' boundary region.

What you are describing sounds more like a partial cyclic boundary. You can prescribe either are pressure drop or a mass flow rate for this.

Yes, you are right. The boundary I am dealling with is partially cyclic to another.

But the pressure drop is unknown, actually it is a variable we need to calculate. Giving the mass flow rate is an flexible option. However, STAR-CD manuals tell that the partial cyclic b.c. can only be implemented where the boundary faces on the boundary region pairs have one-to-one corespondent -- this is exactly not case for me. The mesh generator I used (ABAQUS/CAE) can not produce such kind of mesh.

So either, I implement the b.c. with my own coding, which is the way I am going; or, I change to a better mesh generating tool, which I haven't got a clue. The alternative tool we can have a try is ICEM/CFD package, but anyone experienced can tell me if it is capable? I never use the package.

On the other way round, the prescribed pressure boubdary is not suitable, too. Because the pressure distribution on the boundary is not known (and again, it is the variable we need to calculate out). Do you have any further idea?

Thanks for your reply.

Jiaying

"However, STAR-CD manuals tell that the partial cyclic b.c. can only be implemented where the boundary faces on the boundary region pairs have one-to-one corespondent"

I think you are wrong. See the User Guide, arbitrary cyclic boundaries, i.e. no one-to-one corespondance is needed.

Yes, User Guide says one can use arbitrary match option to link the boundary pair, but that is only allowed when the two boundaries are fully cyclic.

The Methodology manual tells us that arbitrary cyclic matching involving pairs whose boundary faces are not in one-to-one correspondence with each other is not allowed for partial cyclic conditions (Chapter 5 Section 8).

I met some Star-CD people a couple of days ago, and was told this might be no longer a problem with v3.15. But I just checked manuals v3.15. It is clearly stated there. I am using STAR-CD v3.15 for IRIX.

Thanks.

Jiaying

Instead of worrying about what the manuals do or don't say, why don't you just try it?