code_saturne User's Forum

Hi Yvan,

In the calculation of the wall distance, is the only outputted value the magnitude of the distance (distb) between the boundary face centre and the cell centre? If the distance is calculated at the start (or restart) of the simulation, doesn't it need to be recomputed again in computation when the transient turbomachinery module is activated?

There may be a bug in wall BC when the transient turbomachinery module is activated. With the transient turbomachinery module, it doesn't seem to recognize the icodcl value if itypfb(ifac) = iparoi. I had to explicitly state (or hardcode) the icodcl value along with the rcodcl values for the wall in order for the code to recognize that it was a smooth wall rotating with a velocity. This problem doesn't seem to occur for the frozen model for some reason. Please also note this problem is with my compressible algorithm and I don't know if this problem occurs with the release version (v4.0) but I wanted to make you aware that it may exist.

Regards,
Andrew

Hello,

Which version are you using ? 5.0 ?

Not sure why this should happen, or why the compressible module should have a different behaviour regarding icodcl (an I worked on part of the rotor-stator model, so I am familiar with parts of it), but the best solution is probably to check if we can reproduce this issue to better understand it.

If you have a (hopefully small) case which illustrates this, the best solution would be to file a bug (as the bug-tracker is a better medium for follow-ups, and "semi-old" bugs are less easily forgotten tha Forum posts which can be "masked" by newer posts (If your case is confidential but can be shared with EDF, you can send it to me or to the saturne-support e-mail).

Cheers,

Yvan

I am using v4.0.1.

I have no problems with using frozen rotor model and I have compared my results using the frozen rotor model to Fluent and they compare well. It's only when I use the transient model that I have troubles. A few months ago I noticed that the wall BC wasn't working correctly and I used icodcl(ifac,iu) = 1 for the wall BC, but looking at my results recently I noticed that the turbulent viscosity was being greatly overestimated and realized that there was no wall distance. Specifying icodcl(ifac,iu) = 5 for the BC activates the smooth wall BC and appears to be improving my results, but it will take a few days to know if it has fixed the problem. I looked through the source code many times and can't find a reason why specifying itypfb=iparoi doesn't automatically set icodcl(ifac,iu) = 5 for when transient turbomachinery model is being used. I have never used the transient model without the compressible model and I don't know if the error still occurs. I will update you when my simulation progresses further.

I will also file a bug report. I can give you a scale down version of my mesh (8 millions cells), but the compressor is transonic.

Thank you,
Andrew

Hi Yvan,

After spending more time on this problem, I believe the problem is unrelated with the compressible module and is a bug with the turbomachinery module. When the BC are updated in the navstv routine (line 762 and below), the boundary coefficients are recomputed if a wall function is being used. This recalculation may lead to unrealistic values. Using a coordinate transformation for coefficients yields better results for a rotating pipe. I will update the bug report and I will let you know if I receive better results for my compressor.

Regards,
Andrew