K omega SST vs Realizable K epsilon

AhmadHij · August 7, 2022, 12:26

Greetings,
I am working on a 2D transient simulation of vertical axis wind turbine. here are some of the Dimensions:
Diameter =2 m
Chord=0.265 m
Number of blades=2
pitch angle=6 degrees
airfoil: NACA 0021
Boundary conditions :
inlet velocity=8 m/s
turbulence intensity=0.5%
Reynolds number is around 10^5.
I tried first the K omega SST turbulence model, and it gave very low results in comparison with the experimental data (by comparing the average Cp value), so I then tried the Realizable K epsilon model, and It gave much better results.
My question is can I use the Realizable K epsilon model for vertical axis wind turbines where there are a lot of flow separation and why it is giving much better results than K omega SST model ?

jola · August 7, 2022, 15:22

The SST k-omega model often predict separation on convex surfaces too early. If your operating point has significant suction-side separations then the SST k-omega model will probably predict too early separations and thereby incorrect Cp values.

Your Reynolds number is quite low and you will probably have a fairly large laminar part of your suction-side boundary layer. This will make the separation problem even worse.

Using a Relizable k-epsilon model can improve the results. If you have experimental data for a similar case you should use that to decide which turbulence model to use.

Another way to improve the results is to use a transition model, but since you have so low turbulence levels this is a bit tricky since you might have natural transition and not by-pass transition caused by the free-stream turbulence.

August 7, 2022, 12:26	K omega SST vs Realizable K epsilon	#1
AhmadHij New Member Ahmad Hijazi Join Date: Jul 2022 Posts: 7 Rep Power: 4	Greetings, I am working on a 2D transient simulation of vertical axis wind turbine. here are some of the Dimensions: Diameter =2 m Chord=0.265 m Number of blades=2 pitch angle=6 degrees airfoil: NACA 0021 Boundary conditions : inlet velocity=8 m/s turbulence intensity=0.5% Reynolds number is around 10^5. I tried first the K omega SST turbulence model, and it gave very low results in comparison with the experimental data (by comparing the average Cp value), so I then tried the Realizable K epsilon model, and It gave much better results. My question is can I use the Realizable K epsilon model for vertical axis wind turbines where there are a lot of flow separation and why it is giving much better results than K omega SST model ?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Setting change from k omega to k omega SST	bingo641	OpenFOAM Programming & Development	0	May 22, 2018 06:15
At high Y+ values does the K Omega SST model just behave like the K Epsilon model?	JuPa	CFX	0	December 22, 2015 07:44
how to calculate the omega at inlet boundary in k omega sst	Scabbard	OpenFOAM Running, Solving & CFD	2	September 30, 2014 14:06
problem from k epsilon to k omega	NJG	OpenFOAM Running, Solving & CFD	0	April 15, 2013 09:24
Boundary Conditions for k omega SST	dancfd	OpenFOAM Pre-Processing	0	June 10, 2011 00:25

August 7, 2022, 15:22		#2
jola Administrator Jonas Larsson Join Date: Jan 2009 Location: Gothenburg, Sweden Posts: 838 Rep Power: 10	The SST k-omega model often predict separation on convex surfaces too early. If your operating point has significant suction-side separations then the SST k-omega model will probably predict too early separations and thereby incorrect Cp values. Your Reynolds number is quite low and you will probably have a fairly large laminar part of your suction-side boundary layer. This will make the separation problem even worse. Using a Relizable k-epsilon model can improve the results. If you have experimental data for a similar case you should use that to decide which turbulence model to use. Another way to improve the results is to use a transition model, but since you have so low turbulence levels this is a bit tricky since you might have natural transition and not by-pass transition caused by the free-stream turbulence.