I'm modelling flow in a cylinder. The problem of interest is axisymmetric, therefore can be done in 2-D, but STAR doesn't solve in 2-D, therefore I have modeled a 10 degree wedge of a cylinder, with length L, circular cross section, Wall boundary condition (no-slip) on the exterior and Cyclic boundary condtion along the sides, using cylindrical coordinates.

I cannot seem to get the correct flow field however... The maximum velocity should be directly along the centreline of the cylinder, I however get a flow field that is not correct, namely the max velocity is shifted upwards from the line of symmetry. I think this is due to the fact that the first cells, i.e. where R=0, are not 4-sided, but 3-sided.

Does anyone know how to deal with this? I've searched for a similar topic and nothing seem to came up...

thanks in advance, Emmanuel

This is a picture of a simple solution. note that at the line of symmetry, the flow is not at max velocity.

http://img219.imageshack.us/img219/6...eenshoted4.png

It seems the problem comes from that your cells at the axis are actually "4-sided", as you've called it, when they have to be actually "3-sided". I.e. now your cells next to the axis are hexas with one face collapsed. This face, by default, has a wall boundary. So you're modelling a cylinder with a thin wire at the axis. These cells should be prisms for correct solutions. You have to merge vertices in your model (see VMERGE command), run a check for double vertices (with Check Tool) and then redefine the cells it will find as prisms (see CFLIP command).

And, BTW, you can use only one layer of cells to get axisymmetric solution.

Thank you so much!

Hello Anton, I don't think I follow you after all... My cells at the axis are not "4-sided", but "3-sided"... that is how pro-STAR creates the cells touching the axis. I followed your suggestions and merged vertices (that step removes double vertices), but it was already done. I still have the problem of 3 sided cells along the axis, and STAR for some reason interprets that as a wall boundary condition along that edge... any help out there?