quenstion in transient turbulence modeling

cfdlover · August 6, 2009, 17:23

hello,
i have found that the RNG k-e model provides best data fit for branching flow under steady state conditions. Is it expected that this model will also give the best fit under transient conditions?
Please advice.
Thanks

NewtonKF · August 9, 2009, 10:16

Hi cfdlover... I may be wrong but k-e models are not the best way to analyze transient flows... You have to set the correct constants and to do it, you have to use some PIV or LDA data.... I think the best way is to use LES or DNS. This last for academic pourposes of course....

Good luck....

vishyaroon · August 9, 2009, 19:06

I agree with Newton. LES (if not DNS) is the best for transient turbulent flows. Make sure you have the computational capability for LES. DES can be an alternative which can bring down the cell count, but if you know that you'd encounter massively separation. Also be careful in meshing for DES, as the solution is very sensitive to the mesh you generate.

As for RANS models, I'd not trust the results as much.

cfdlover · August 12, 2009, 17:04

thank you.
what is "DES"?
Also, i have around 10 million nodes in my geometry and the process is very rapid (5-20 seconds). Do you have any idea about the computation time limitation and time step in CFD codes?
Thanks
cfdlover

vishyaroon · August 12, 2009, 17:26

DES is Detached Eddy Simulation. In simple words it is a hybrid RANS-LES method which considers the regions closer to the boundary as RANS and regions away from teh wall as LES.

Your time step will again depend on the smallest grid point you have in your domain. It's difficult to say from teh number of grid points as to what your minimum time step will be.

cfdlover · August 12, 2009, 18:16

thanks alot for the useful information.
i have the smallest grids in the y direction where there are 215 nodes in 4 cm length. Thus the smallest grid is around 0.19 mm.
cfdlover

August 6, 2009, 17:23	quenstion in transient turbulence modeling	#1
cfdlover New Member afm Join Date: Jul 2009 Posts: 7 Rep Power: 16	hello, i have found that the RNG k-e model provides best data fit for branching flow under steady state conditions. Is it expected that this model will also give the best fit under transient conditions? Please advice. Thanks

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Discussion: Reason of Turbulence!!	Wen Long	Main CFD Forum	3	May 15, 2009 09:52
May Focus Area: Turbulence Modeling	Jonas Larsson	CFD-Wiki	0	May 2, 2006 11:48
Turbulence modeling for fan/compressor blading	Jonas Larsson	FLUENT	1	April 13, 2005 14:08
Turbulence model for Transient Pipe Flow Modeling	Brian	FLUENT	0	June 7, 2002 12:15

August 9, 2009, 10:16		#2
NewtonKF Member Newton KF Join Date: Mar 2009 Posts: 36 Rep Power: 17	Hi cfdlover... I may be wrong but k-e models are not the best way to analyze transient flows... You have to set the correct constants and to do it, you have to use some PIV or LDA data.... I think the best way is to use LES or DNS. This last for academic pourposes of course.... Good luck....

August 9, 2009, 19:06		#3
vishyaroon Senior Member Aroon Join Date: Apr 2009 Location: Racine WI Posts: 148 Rep Power: 17	I agree with Newton. LES (if not DNS) is the best for transient turbulent flows. Make sure you have the computational capability for LES. DES can be an alternative which can bring down the cell count, but if you know that you'd encounter massively separation. Also be careful in meshing for DES, as the solution is very sensitive to the mesh you generate. As for RANS models, I'd not trust the results as much.

August 12, 2009, 17:04		#4
cfdlover New Member afm Join Date: Jul 2009 Posts: 7 Rep Power: 16	thank you. what is "DES"? Also, i have around 10 million nodes in my geometry and the process is very rapid (5-20 seconds). Do you have any idea about the computation time limitation and time step in CFD codes? Thanks cfdlover

August 12, 2009, 17:26		#5
vishyaroon Senior Member Aroon Join Date: Apr 2009 Location: Racine WI Posts: 148 Rep Power: 17	DES is Detached Eddy Simulation. In simple words it is a hybrid RANS-LES method which considers the regions closer to the boundary as RANS and regions away from teh wall as LES. Your time step will again depend on the smallest grid point you have in your domain. It's difficult to say from teh number of grid points as to what your minimum time step will be.

August 12, 2009, 18:16		#6
cfdlover New Member afm Join Date: Jul 2009 Posts: 7 Rep Power: 16	thanks alot for the useful information. i have the smallest grids in the y direction where there are 215 nodes in 4 cm length. Thus the smallest grid is around 0.19 mm. cfdlover