I am afraid I need to do some user coding to implement a partial cyclic boundary condition. In my case, velocities are periodic at the two boundaries, but pressure and temperature are not. What's more, the Pro-STAR built-in partial cyclic BC is not suitable for my situation.

So, the first big problem coming out is how I can exteact out the relations between a pair of corespendont facets on the boundaries. This is not very hard for a structured mesh, but how about an unstructured mesh? and how about if the small facets on both boundaries are not in one-to-one relationship?

The second problem is: which one should I deal with in order to implemet cyclic BC, the variables on the boundaries (in STAR-CD termonology, the shell cells) or the variables on the centroids of cells (i.e. the normal cell data)? STAR-CD methodology does not say much about this. The things unclear are which part of job needs to be done by the user subroutine, and which part is already done internally by STAR.

Some minor problems include how I can print out all index numbers of shell cells on a region of boundary, how to print out all neighering normal cells to the boundary, etc.

Welocme all suggestions or hints. Thanks a lot.

Jiaying

Anybody got idea about these?

I think the second problem I have to contact STAR-CD support to get technical details of STAR. But how about the first problem? That is, how can I map both boundaries if the small faces on them don't have one-to-one coorespondence? This is the biggest problem during user coding.

Thanks,

Jiaying

How about using the coordinate as the guidance in user 's subroutine? I mean, for example, for rotating machine, at the same radius, the velocity magnitude is the same. So, you can set a constant, and at the accordingly coordinate the velocity can be set to fulfill your demand.