[Far]

Dear ghorrocks

I wanted to know which model is better for turbo machinery flows. BSL Reynolds Stress model or SSG Reynolds stress model.

I also wanted to know that the BSL RSM (uses the LRR in inner and outer region) is better than the SSG/LRR-w (uses the SSG in outer core and LRR based on omega in the wall region).

Any help will be highly appreciated.

Regards
Far

[ghorrocks]

Depends on your application, that is why they are both in there. Try them both in your case and compare against experimental data.

[Far]

I am simulating NASA rotor 37 at 17188.7 RPM and running from choking to stall.

Any help is highly appreciated.

Regards
Far

[Far]

Moreover since the SSG is high reynolds number model and BSL RSM is low reynodls number. Should I use mesh with y plus values between 30 and 60 or below 1 for two models.

Regards

[ghorrocks]

The different models have different requirements for meshes. Use the appropriate mesh for each one.

[Far]

Actually I am using the BSL Reynolds stress model with automatic wall treatment. I have fixed the y plus approx. 30. My concern is that "Am I only solving the outer part of flow where the BSL converts to K epsilon model in terms of omega variable?

[ghorrocks]

I think what you meant to say is that you are using wall functions rather than integrating to the wall. Wall functions can be more accurate than integrating to the wall so I would not dismiss them without checking first - and given that they run on a far coarser mesh they are MUCH easier to use.

But they only way to know for sure is to run both methods and find out what runs best in your case.

[Far]

I am attaching my results for five turbulence model for rotor 37. Here you can see that the RSM BSL is not good as compared to k epsilon model

[Far]

size of mesh is 1.0 million and wall function approach is enforced. Only one blade is modeled

[Far]

any one else is also welcome who have some questions or knowledge about the topic.

I am also ready to share my data for this case.

[ghorrocks]

Is your comparison so far based on a mesh with y+=30? If possible, try to do a y+=1 approx mesh and try the turb models which can integrate to the wall.

[Far]

Yes. all results are on based on y plus 30

[lauda]

Hi Far,

did you compare radial profiles of the efficiency and pressure ratio with exp? I did a relatively long study on R37 to validate our code. It was interesting to observe how the pretty good matching for the efficiency curve was due to small error cancellation between total temperature and pressure profiles.

[Far]

@lauda: Yes we did compare radial profiles. But I think for better matching, we should use LES for this case... What do you think?

[lauda]

I'm not a LES expert but we know it's more accurate generally speaking..although more expensive from a computational point of view. However there is a paper on LES for R37 probably you are aware of it.. "Large Eddy Simulation of Transonic Flow Field in NASA Rotor 37" by C. Hah from NASA. It's available online.

http://ntrs.nasa.gov/archive/nasa/ca...0090038700.pdf

The LES results are quite different when compared to most of RANS approaches found in the literature..still the 3D flow features don't match experimental data in a much better way...so I'm not sure if it's worth.

[swm]

Originally Posted by Far

any one else is also welcome who have some questions or knowledge about the topic.

I am also ready to share my data for this case.

Hi Far,
i found when "stage"is used in mixing model, downstream\velocity constraint option can cause different flux result("Stage Average Velocity" is a little bit smaller than "Constant Total Pressure" ). i'd like to ask what your "standard" option is and the reason.
thanks for your time.