General question about SST turbulence model

Richard Renaud · December 4, 2015, 14:49

I have a general question about when it is beneficial to use an SST turbulence model.

I've been reading up on turbulence lately. Prior to this recent study, I had a habit of using the SST turbulence model without checking the y-plus of my grid. For example, I used SST when modelling a cylindrical feed well with a tangential inlet pipe that contains baffles. Since there are baffles perpendicular to the flow, separation is likely to occur and I thought that SST would be a good choice to model this. However, I checked my mesh and my y-plus is on the order of 100!

I am wondering, as a general question, does SST offer any advantages over a standard k-ε model when the boundary layer is not well refined. If it is computationally infeasible for me to refine the boundary layer with a y+ of 1, would you recommend that I stick with the SST model anyways? Or does the SST model offer no benefits over a k-ε model when the mesh refinement in the boundary layer is insufficient?

Thanks,
Richard Renaud

hazari.aaquib · December 19, 2015, 05:13

Dear Richard,

If you're planning to use a grid with a large yplus of the order of 100 I suggest that you should use k-e model.
Only if your yplus is less than 2 (it can extend till 4) you should use the SST or k-w model.

At a wall gradients are steep and basically the SST and k-w model directly solve the equations at the grid cells. While k-e uses wall functions to solve the gradients hence less computationally expensive.

Furthermore, SST is a combination of k-w and k-e.

Good luck!

December 4, 2015, 14:49	General question about SST turbulence model	#1
Richard Renaud New Member Richard Renaud Join Date: Mar 2015 Posts: 6 Rep Power: 11	I have a general question about when it is beneficial to use an SST turbulence model. I've been reading up on turbulence lately. Prior to this recent study, I had a habit of using the SST turbulence model without checking the y-plus of my grid. For example, I used SST when modelling a cylindrical feed well with a tangential inlet pipe that contains baffles. Since there are baffles perpendicular to the flow, separation is likely to occur and I thought that SST would be a good choice to model this. However, I checked my mesh and my y-plus is on the order of 100! I am wondering, as a general question, does SST offer any advantages over a standard k-ε model when the boundary layer is not well refined. If it is computationally infeasible for me to refine the boundary layer with a y+ of 1, would you recommend that I stick with the SST model anyways? Or does the SST model offer no benefits over a k-ε model when the mesh refinement in the boundary layer is insufficient? Thanks, Richard Renaud

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Hoz to fix Y+ dependence on inlet Mach number? SST turbulence model	LuReym	CFX	5	January 20, 2015 17:28
boundary and initial condition for SST k-omga turbulence model	mb.pejvak	Main CFD Forum	3	May 1, 2012 21:06
General question: turbulence and laminar models	fluentmonkey	Main CFD Forum	12	March 17, 2011 15:47
Low Reynolds k-epsilon model	YJZ	ANSYS	1	August 20, 2010 14:57
Advanced Turbulence Modeling in Fluent, Realizable k-epsilon Model	Jonas Larsson	FLUENT	5	March 13, 2000 04:27

December 19, 2015, 05:13		#2
hazari.aaquib New Member Syed Aaquib Hazari Join Date: Apr 2015 Location: Delft Posts: 16 Rep Power: 11	Dear Richard, If you're planning to use a grid with a large yplus of the order of 100 I suggest that you should use k-e model. Only if your yplus is less than 2 (it can extend till 4) you should use the SST or k-w model. At a wall gradients are steep and basically the SST and k-w model directly solve the equations at the grid cells. While k-e uses wall functions to solve the gradients hence less computationally expensive. Furthermore, SST is a combination of k-w and k-e. Good luck!