How is TLEN working - Turbulence Model

Ben-Bourget · June 16, 2009, 11:26

Hello dear users,

here is my question,

I am currently making a free surface 3D simulation using the RNG turbulence model on v9.3.1.
I notice a great importance of the TLEN value for my results.
If I set a very low TLEN the simulation is running well and I reach a stationary state.
If I set a TLEN a little bit higher, I have a simulation that never really stabilizes, with a velocity and a pressure field changing from a time step to another.
If I set a TLEN that is high (as I firstly want to do) The simulation doesn't run because of the viscous stress condition.

Do you have any idea of how to run the simulation with a high TLEN?

I also have a second question :

If I do not specified a TLEN value, I read in Flow3D user manual that TLEN is chosen as 7% of a characteristic dimension of the flow.
What is this characteristic dimension ? How is it evaluated by Flow3D ?

Thanks for your help.

Ben

ck5285 · June 16, 2009, 20:40

Ben
I believe this is command problem in turbulent modeling by F3D. The results are much dependent on TLEN value that user specify (at least based on my experienced). So, hopefully the dependency on TLEN can be improve/remove in new version to ensure the solver is robust.

Btw, if not mistaken the 7% is based on smaller cell size in the computational domain.

Hope this help.

Ben-Bourget · June 17, 2009, 03:44

Thanks for your answer.

I hope that the default TLEN is not related to the smallest celle size...
In the turbulence theory, TLEN is supposed to be 7% of a characteristic dimension of the geometry, that is to say much bigger than 7% of the smallest cell size.
I am thinking that the TLEN could be related to the smallest water depth or to the smallest distance between two solids.

Does someone know precisely how the default TLEN is fixed?

stefmascio · June 17, 2009, 10:06

the default tlen is taken as the 7% of the smallest domain dimension, that is the smallest length of the three dimensions of your mesh block. So it doesn't depend on the cell size, and the smallest domain dimension should be about of the same magnitude order of your geometries.
You can also check which tlen value is using your simulation by watching at the "runtime options", while a simulation is running.

I think in the new version the dependency from tlen will be removed.

ck5285 · June 20, 2009, 08:31

Stefano
Really? the dependency of TLEN will be remove in new version? that's good news.

stefmascio · June 22, 2009, 02:40

however, as far as I know, not completely. I think there will still be tlen, but the sensitivity on the results from tlen value will be much more smaller.
FlowScience can correct me in case I'm wrong.

June 16, 2009, 11:26	How is TLEN working - Turbulence Model	#1
Ben-Bourget New Member Ben-Bourget Join Date: Jun 2009 Posts: 3 Rep Power: 16	Hello dear users, here is my question, I am currently making a free surface 3D simulation using the RNG turbulence model on v9.3.1. I notice a great importance of the TLEN value for my results. If I set a very low TLEN the simulation is running well and I reach a stationary state. If I set a TLEN a little bit higher, I have a simulation that never really stabilizes, with a velocity and a pressure field changing from a time step to another. If I set a TLEN that is high (as I firstly want to do) The simulation doesn't run because of the viscous stress condition. Do you have any idea of how to run the simulation with a high TLEN? I also have a second question : If I do not specified a TLEN value, I read in Flow3D user manual that TLEN is chosen as 7% of a characteristic dimension of the flow. What is this characteristic dimension ? How is it evaluated by Flow3D ? Thanks for your help. Ben

June 17, 2009, 03:44	Default TLEN	#3
Ben-Bourget New Member Ben-Bourget Join Date: Jun 2009 Posts: 3 Rep Power: 16	Thanks for your answer. I hope that the default TLEN is not related to the smallest celle size... In the turbulence theory, TLEN is supposed to be 7% of a characteristic dimension of the geometry, that is to say much bigger than 7% of the smallest cell size. I am thinking that the TLEN could be related to the smallest water depth or to the smallest distance between two solids. Does someone know precisely how the default TLEN is fixed?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Adding a Turbulence Model	doug	OpenFOAM Running, Solving & CFD	11	May 21, 2018 13:54
SimpleFoam case with SpalartAllmaras turbulence model implemented	nedved	OpenFOAM Running, Solving & CFD	2	November 30, 2014 22:43
KOmega Turbulence model from wwwopenFOAMWikinet	philippose	OpenFOAM Running, Solving & CFD	30	August 4, 2010 10:26
Discussion: Reason of Turbulence!!	Wen Long	Main CFD Forum	3	May 15, 2009 09:52
Turbulence Model	GG	Siemens	3	March 3, 2008 19:06

June 16, 2009, 20:40		#2
ck5285 Senior Member Join Date: Mar 2009 Posts: 129 Rep Power: 17	Ben I believe this is command problem in turbulent modeling by F3D. The results are much dependent on TLEN value that user specify (at least based on my experienced). So, hopefully the dependency on TLEN can be improve/remove in new version to ensure the solver is robust. Btw, if not mistaken the 7% is based on smaller cell size in the computational domain. Hope this help.

June 17, 2009, 10:06		#4
stefmascio Member Stefano Join Date: Mar 2009 Posts: 59 Rep Power: 17	the default tlen is taken as the 7% of the smallest domain dimension, that is the smallest length of the three dimensions of your mesh block. So it doesn't depend on the cell size, and the smallest domain dimension should be about of the same magnitude order of your geometries. You can also check which tlen value is using your simulation by watching at the "runtime options", while a simulation is running. I think in the new version the dependency from tlen will be removed.

June 20, 2009, 08:31		#5
ck5285 Senior Member Join Date: Mar 2009 Posts: 129 Rep Power: 17	Stefano Really? the dependency of TLEN will be remove in new version? that's good news.

June 22, 2009, 02:40		#6
stefmascio Member Stefano Join Date: Mar 2009 Posts: 59 Rep Power: 17	however, as far as I know, not completely. I think there will still be tlen, but the sensitivity on the results from tlen value will be much more smaller. FlowScience can correct me in case I'm wrong.