Hello,
A quick question - I have been staying clear of the k-omega SST model in code_saturne because I was told that it was 'broken'. Looking into it further the reference for this information is this forum post, which has no replies:
http://code-saturne.org/forum/viewtopic ... wall#p4703
Was this a genuine bug or a case of not setting the model up correctly (eg wrong value on icdpar)? If it was a bug, is it now fixed/gone in version 3? Original post is from June 2011.
Thanks in advance and happy new year!
k-omega SST query
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
Yvan Fournier
- Posts: 4220
- Joined: Mon Feb 20, 2012 3:25 pm
Re: k-omega SST query
Hello,
I am not sure this was really a bug, in the sense that the behavior switching from smooth to rough wall was non continuous, but changing the model to improve it would require more tests, to make sure we do not degrade other cases. I am not sure who uses k-omega with rough walls, so this has not not been a priority for improvement, but If anyone looks into this closely enough, we will try to follow suit.
A fix relative to very low y+ values (for smooth walls) was implemented in version 3.1, when "scalable wall functions" are not used (not in 3.0 if I am correct).
Otherwise, nothing was changed relative to this model. For finer analysis, the code for the smooth wall is in src/base/clptur.f90, and that for the rough wall in src/base/clptrg.f90.
The parent post seems you link to refer to tests done using the "scalable wall functions" (ideuch=2).
Regards,
Yvan
I am not sure this was really a bug, in the sense that the behavior switching from smooth to rough wall was non continuous, but changing the model to improve it would require more tests, to make sure we do not degrade other cases. I am not sure who uses k-omega with rough walls, so this has not not been a priority for improvement, but If anyone looks into this closely enough, we will try to follow suit.
A fix relative to very low y+ values (for smooth walls) was implemented in version 3.1, when "scalable wall functions" are not used (not in 3.0 if I am correct).
Otherwise, nothing was changed relative to this model. For finer analysis, the code for the smooth wall is in src/base/clptur.f90, and that for the rough wall in src/base/clptrg.f90.
The parent post seems you link to refer to tests done using the "scalable wall functions" (ideuch=2).
Regards,
Yvan
-
Brian Angel
Re: k-omega SST query
Hello Susan,
We have recently used the k-omega SST model in Code_Saturne and have compared predicted data using this turbulence model with DNS results for a bouyancy driven and a mixed bouyancy-forced convection flow.
For the bouyancy driven flow, the low Reynolds number version of this model correctly predicted the wall heat transfer when compared to the DNS data.
For a mixed natural-forced convection flow, both the high and low Reynolds number versions gave acceptable results when compared to DNS data.
Regards,
Brian Angel.
We have recently used the k-omega SST model in Code_Saturne and have compared predicted data using this turbulence model with DNS results for a bouyancy driven and a mixed bouyancy-forced convection flow.
For the bouyancy driven flow, the low Reynolds number version of this model correctly predicted the wall heat transfer when compared to the DNS data.
For a mixed natural-forced convection flow, both the high and low Reynolds number versions gave acceptable results when compared to DNS data.
Regards,
Brian Angel.
Re: k-omega SST query
Thank you both for your replies and the information.
Apologies for the very late response also!
Best regards
Susan
Apologies for the very late response also!
Best regards
Susan