Distinction between k-w sst and SA turbulence model in SU2

ari003 · December 3, 2021, 03:46

Hello everyone,
Note:- Shifted this thread from news and announcement section to general question section.
Hope you all are doing well. I came across a specific doubt that might be very simple for others and I would really appreciate if they can share their understanding with me. The description is as follows:-

I am running an airfoil simulation at M=0.8 with k-w SST and S-A turbulence model. My aim was to analyze how the y+ varies along the upper surface of the airfoil and the plot is attached beneath. Also, there is a significant difference in Cl value for the both, 0.26 for k-w SST while 0.025 for S-A. I am wondering what causing this huge difference in Cl values?
If I take a dig in y+ formulation it seems like the eddy viscosity calculation in both the case is making the difference.

So my question is, whether the script for the SA case (attached) is in correct order or not? Also, if anyone is aware of SA-LM or SA-BC does either of the one will give better result compared to standard S-A model in case of airfoil case?

bigfootedrockmidget · December 3, 2021, 06:27

Hi,

Not sure about the difference in CL, but for transition you should use the BC model, the LM model has been deactivated and needs to be reimplemented.

Regarding the difference in CL, what is available in the literature? There are probably studies out there where the SA and SST model were used in the same study for this or a similar airfoil. A quick google search shows some images of the Cl(alpha) curve for SA and k-omega-SST with significant differences. Doesn't mean it cannot be a problem in the SU2 code, but this would be the first to check.

Jiba · January 5, 2022, 10:40

Hi,

looking a the Y+, the first thing I spot is that separation seems to be predicted at different locations. This can be easily confirmed comparing the cf. If this is the case, surely the comparison of cp will show differences and explain the different cl, even though the difference is too high.

From the cfg file, the simulation is run at AoA=0 deg, but the header (maybe not updated) refers to a symmetrical airfoil.? If this is the case, cl must be equal to zero.

An additional comment is that Y+ should be a bit smaller.

December 3, 2021, 06:27		#2
bigfootedrockmidget Senior Member bigfoot Join Date: Dec 2011 Location: Netherlands Posts: 676 Rep Power: 21	Hi, Not sure about the difference in CL, but for transition you should use the BC model, the LM model has been deactivated and needs to be reimplemented. Regarding the difference in CL, what is available in the literature? There are probably studies out there where the SA and SST model were used in the same study for this or a similar airfoil. A quick google search shows some images of the Cl(alpha) curve for SA and k-omega-SST with significant differences. Doesn't mean it cannot be a problem in the SU2 code, but this would be the first to check.

January 5, 2022, 10:40		#3
Jiba New Member Join Date: Dec 2013 Location: Italy Posts: 29 Rep Power: 13	Hi, looking a the Y+, the first thing I spot is that separation seems to be predicted at different locations. This can be easily confirmed comparing the cf. If this is the case, surely the comparison of cp will show differences and explain the different cl, even though the difference is too high. From the cfg file, the simulation is run at AoA=0 deg, but the header (maybe not updated) refers to a symmetrical airfoil.? If this is the case, cl must be equal to zero. An additional comment is that Y+ should be a bit smaller.