Hi,

I haven´t found StarCD's minimum requirements for grid quality. Do you know that? Aspect ratio? min. Angle? Warpage?

Thanks, Andreas

Hello,

You could find that in the first chapter of the User Guide.

Ok,

I've seen these parameters. But I think these are not the minimum requirements. Especially the internal angle with 45° seems to me very conservative.

Well, if you can't improve your mesh quality any more (esp. tet quality <0.01), try running your case in double-precision and with the SIMPLISO algorithm. Should help lessen the effect of numerical errors from those badly-formed cells.

I have the same feeling. Updating version is make computation more and more robust. Warpage up to even 75 degree does not incur trouble. However what happens to outcome? I did not make any comparison. Anyone has some idea?

I once observed the difference in hex qualities: One model had near-perfect hexas and converged within 60 iterations while another (same geometry, differently meshed by a beginner) had an entire block (that I was able to spot) with *reflex* (>180'; probably around 190') angled cells and took 500+ iterations to converge. That model had 8 blocks of hexahedral cells.

Exept for the time required to converge, were there any differences between the two calculations, and how big were them? Cheers Roadracer

No other differences in set-up between the models. All block edge cell counts are fixed for this simple exercise. <10k cells as I remember.

Sorry, I haven´t been clear, I meant if there were differences in the results!! Cheers Roadracer

No idea. Never bothered to scrutinize it. Grid quality aside, the mesh was just too coarse (and hex dimensions will differ as positioning block edges is not an exact science) to achieve grid-independent results...