Accuracy of LRR and SSG in turbulent channel flow

dradenkovic · August 16, 2016, 15:25

Hello.

I am simulating steady, incompressible flow in channel. Dimensions of channel are:
length 6 m, width 350 mm, and height 35 mm.

I did simulations with two RSMs - LRR and SSG. When I compare results of simulations with experimental and DNS data, I see that accuracy of simulations can be improved. Measuring section is at 5.8 m from the inlet.

Figures are on the following links (from some reason, I couldn't attach pictures on this page)

LRR comparison

Here is comparison for SSG simulation:
SSG comparison

I don't have exact DNS data for this case, but I used data for channel flow where Re number is around 40 000 so I believe it can be used for verification, when data are nondimensionalised.

Information about setup are here:

CheckMesh

fvSchemes

fvSolution

Any suggestion to improve accuracy of these simulations?

Best regards,
Darko

dradenkovic · August 23, 2016, 14:03

Here are velocity profiles that I have as a result of a several simulations.

https://www.dropbox.com/s/eyd4tk3rid...Phase.eps?dl=0

I have error around 5% in velocity profile. What can I do to reduce it? I did simulations on various meshes, but the result is the same.

I have trouble to verify RMS values (from above post). Maybe that difference in RMS values (which is causing me troubles) is due to difference in velocity profiles. That is why I need to correct velocity profile.

Here is complete case of steady, incompressible, turbulent channel flow, k - epsilon model.

https://www.dropbox.com/s/z1w7bu9mv1...imple.zip?dl=0

Any suggestion is appreciated.

Best regards,
Darko

August 16, 2016, 15:25	Accuracy of LRR and SSG in turbulent channel flow	#1
dradenkovic Member Darko Radenkovic Join Date: Oct 2015 Posts: 38 Rep Power: 11	Hello. I am simulating steady, incompressible flow in channel. Dimensions of channel are: length 6 m, width 350 mm, and height 35 mm. I did simulations with two RSMs - LRR and SSG. When I compare results of simulations with experimental and DNS data, I see that accuracy of simulations can be improved. Measuring section is at 5.8 m from the inlet. Figures are on the following links (from some reason, I couldn't attach pictures on this page) LRR comparison Here is comparison for SSG simulation: SSG comparison I don't have exact DNS data for this case, but I used data for channel flow where Re number is around 40 000 so I believe it can be used for verification, when data are nondimensionalised. Information about setup are here: CheckMesh fvSchemes fvSolution Any suggestion to improve accuracy of these simulations? Best regards, Darko

August 23, 2016, 14:03		#2
dradenkovic Member Darko Radenkovic Join Date: Oct 2015 Posts: 38 Rep Power: 11	Here are velocity profiles that I have as a result of a several simulations. https://www.dropbox.com/s/eyd4tk3rid...Phase.eps?dl=0 I have error around 5% in velocity profile. What can I do to reduce it? I did simulations on various meshes, but the result is the same. I have trouble to verify RMS values (from above post). Maybe that difference in RMS values (which is causing me troubles) is due to difference in velocity profiles. That is why I need to correct velocity profile. Here is complete case of steady, incompressible, turbulent channel flow, k - epsilon model. https://www.dropbox.com/s/z1w7bu9mv1...imple.zip?dl=0 Any suggestion is appreciated. Best regards, Darko