Turbulent length scale setting

realanony87 · June 1, 2009, 09:31

Hello,
How would setting the values for k and the turbulence length scale for the SST turbulence model effect the convergence of a problem ? As I understand the length scale is used to calculate epsilon at the inlet. However if the problem to be solved is a flow over a wing , and the wing is far several chord lengths away from the inlet, I find in contour plots that the kinetic turbulent energy decreases rapidly after the inlet. So what difference does it actually make if I set the turbulence length scale to be the thickness of the wing or the thickness of the boundary layer ?
Thanks

ghorrocks · June 1, 2009, 18:40

Hi,

The k or turbulence intensity comes from the turbulence levels. Often for airfoil experiments that is just defined as a turbulence percentage. It is a bit trickier for the second turbulence variable as there is no simple measurement which can define it. So often people just guess a length scale and use the e value which comes from that.

This can lead to an over-estimate of e. As e is the turbulence dissipation this will then eat up the turbulence level at the inlet until it establishes a new equilibrium between turbulence creation (through shear and other mechanisms) and destruction (through e, turbulence dissipation). That is why you see a rapid reduction in turbulence levels near the inlet.

If you see this and you know your turbulence intensity is correct at the inlet, this means the turbulence intensity after the e has established equilibrium is wrong. You had too much e and should reduce it until the turbulence value a distance away from the inlet (when equilibrium between turbulence creation and destruction is established) matches the turbulence intensity you need.

Glenn Horrocks

realanony87 · June 1, 2009, 20:31

Hey Glenn, Thanks.
I am a bit puzzled about this statement:
"If you see this and you know your turbulence intensity is correct at the inlet, this means the turbulence intensity after the e has established equilibrium is wrong. You had too much e and should reduce it until the turbulence value a distance away from the inlet (when equilibrium between turbulence creation and destruction is established) matches the turbulence intensity you need."

I am supposed to set the turbulence dissipation e so that I get the turbulence level I want at the wing. However this means that to decrease e, I have to increase the turbulence length scale, which doesn't make sense for me. Shouldn't I be worried about modelling the dissipation in the region of flow over a body itself ?

From the simple relation e= C k^(3/2)/l , decreasing e means increasing the length scale. This may mean that the length scale many times larger than the wing itself if I have a large inlet region and want to have a 1% turbulence level at the wing. Is this a limitation of the turbulence modell that I have to deal with ?

ghorrocks · June 2, 2009, 00:58

Hi,

What do you mean "modelling the dissipation in the region of flow over a body itself"?

I don't see the significance of the turbulence length scale to your airfoil. I see no reason why the incoming turbulence could not have a length scale larger than the wing. That is why aircraft go bump when they fly through turbulence.

But your final comment is correct as well - The simple 2-equation turbulence models may not be appropriate. That is up to you to determine.

Glenn Horrocks

Bollonga · October 14, 2012, 13:29

Quote:

Originally Posted by ghorrocks

This can lead to an over-estimate of e. As e is the turbulence dissipation this will then eat up the turbulence level at the inlet until it establishes a new equilibrium between turbulence creation (through shear and other mechanisms) and destruction (through e, turbulence dissipation). That is why you see a rapid reduction in turbulence levels near the inlet.

If you see this and you know your turbulence intensity is correct at the inlet, this means the turbulence intensity after the e has established equilibrium is wrong. You had too much e and should reduce it until the turbulence value a distance away from the inlet (when equilibrium between turbulence creation and destruction is established) matches the turbulence intensity you need.

Glenn Horrocks

Hi Glenn,

In my case I have a 2D inclined flat plate at high angle of attack, 70º. I have 10 m/s and 10% Iu at the inlet. I am using 0.05 m length scale, and I am getting Iu reduced to 0.1% at some distance away from the inlet. I am also facing viscosity ratio limitation to 1e5 in some cells in the wake of the plate.

Should I use a bigger length scale so ephsilon will be lower and the Iu won't be reduced that much? Wouldn't that make my viscosity ratio even bigger so I'll get it limitated in more cells?

Thanks in advance.

ghorrocks · October 14, 2012, 18:03

There is huge benefits in you working this out for yourself. Why not do an experiment and just find out what happens yourself? And combining it with some reading to find out what the textbooks say should happen is of even greater value.

In setting your inlet boundary you need to set both the incoming turbulence intensity and the dissipation to an appropriate amount. It is easy to set the turbulence intensity is a standard parameter often measured, but the dissipation needs to be adjusted so it does not wipe out all the turbulence straight away.

k-e turbulence models have known issues with the pressure in stagnation points and in separations.

June 1, 2009, 09:31	Turbulent length scale setting	#1
realanony87 New Member Join Date: May 2009 Posts: 21 Rep Power: 16	Hello, How would setting the values for k and the turbulence length scale for the SST turbulence model effect the convergence of a problem ? As I understand the length scale is used to calculate epsilon at the inlet. However if the problem to be solved is a flow over a wing , and the wing is far several chord lengths away from the inlet, I find in contour plots that the kinetic turbulent energy decreases rapidly after the inlet. So what difference does it actually make if I set the turbulence length scale to be the thickness of the wing or the thickness of the boundary layer ? Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
how to find turbulent length scale ??	kapil	FLUENT	0	January 20, 2009 12:10
Warning 097-	AB	Siemens	6	November 15, 2004 04:41
turbulent length scale	Cathy	FLUENT	1	July 9, 2003 07:13
Turbulent Length Scale	James Date	Main CFD Forum	6	May 3, 2000 08:10

June 1, 2009, 18:40		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Hi, The k or turbulence intensity comes from the turbulence levels. Often for airfoil experiments that is just defined as a turbulence percentage. It is a bit trickier for the second turbulence variable as there is no simple measurement which can define it. So often people just guess a length scale and use the e value which comes from that. This can lead to an over-estimate of e. As e is the turbulence dissipation this will then eat up the turbulence level at the inlet until it establishes a new equilibrium between turbulence creation (through shear and other mechanisms) and destruction (through e, turbulence dissipation). That is why you see a rapid reduction in turbulence levels near the inlet. If you see this and you know your turbulence intensity is correct at the inlet, this means the turbulence intensity after the e has established equilibrium is wrong. You had too much e and should reduce it until the turbulence value a distance away from the inlet (when equilibrium between turbulence creation and destruction is established) matches the turbulence intensity you need. Glenn Horrocks

June 1, 2009, 20:31		#3
realanony87 New Member Join Date: May 2009 Posts: 21 Rep Power: 16	Hey Glenn, Thanks. I am a bit puzzled about this statement: "If you see this and you know your turbulence intensity is correct at the inlet, this means the turbulence intensity after the e has established equilibrium is wrong. You had too much e and should reduce it until the turbulence value a distance away from the inlet (when equilibrium between turbulence creation and destruction is established) matches the turbulence intensity you need." I am supposed to set the turbulence dissipation e so that I get the turbulence level I want at the wing. However this means that to decrease e, I have to increase the turbulence length scale, which doesn't make sense for me. Shouldn't I be worried about modelling the dissipation in the region of flow over a body itself ? From the simple relation e= C k^(3/2)/l , decreasing e means increasing the length scale. This may mean that the length scale many times larger than the wing itself if I have a large inlet region and want to have a 1% turbulence level at the wing. Is this a limitation of the turbulence modell that I have to deal with ?

June 2, 2009, 00:58		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	Hi, What do you mean "modelling the dissipation in the region of flow over a body itself"? I don't see the significance of the turbulence length scale to your airfoil. I see no reason why the incoming turbulence could not have a length scale larger than the wing. That is why aircraft go bump when they fly through turbulence. But your final comment is correct as well - The simple 2-equation turbulence models may not be appropriate. That is up to you to determine. Glenn Horrocks

October 14, 2012, 18:03		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,703 Rep Power: 143	There is huge benefits in you working this out for yourself. Why not do an experiment and just find out what happens yourself? And combining it with some reading to find out what the textbooks say should happen is of even greater value. In setting your inlet boundary you need to set both the incoming turbulence intensity and the dissipation to an appropriate amount. It is easy to set the turbulence intensity is a standard parameter often measured, but the dissipation needs to be adjusted so it does not wipe out all the turbulence straight away. k-e turbulence models have known issues with the pressure in stagnation points and in separations.