NLH with SA (Extended Wall Functions)

Cappytbg · March 1, 2018, 06:57

Hello!

I have a question to NLH method. Particularity, I saw in Numeca Fine/Open User's Guide restrictions for use of NLH method, it says that the only possibility for this kind of analysis is the use of SA turbulence model, but... can you tell me, if Numeca Fine/Open can run simulation with SA (with Extended Wall Functions) model?

I am running simulation of stator and rotor and want to catch their interference.

Thank you for your help!

Cappytbg · March 1, 2018, 15:29

I mean - simulation is running with this set, but my question is if Numeca does not changes automaticaly from this model.

DarylMusashi · March 2, 2018, 06:23

Dear Cappytbg,

May I ask which kind of mesh you are using? If you created a "low Reynolds" mesh you resolve the boundary layer with a typical first cell height of 1E-5 [m] for turbomachinery applications and result in y+ values < 10. In this case a wall function approach does not make sense.

The second scenario would be a "high Reynolds" mesh with only one cell in the boundary layer, the recommended y+ values are between 30 and 100. In this case an (extended) wall function approach is necessary.

For turbomachinery applications I recommend the first. Is there a particular reason why you want to use the SA (extended wall function) model instead of the standard SA model?

Kind regards,
Holger

Cappytbg · March 2, 2018, 06:40

Dear DarylMusashi,

Thank you for your answer! I am using unstructured mesh. And Re is high.
You see, I have conducted a series of analyses so far, but I am making investigation.
I have done analysis with y+<1, but now I want to check the wall functions approach.

About y+ values, I have read Numeca's User Guide, and it says that to be in WF range y+ should be 20-50, what I obtained. And here I used SA (with Extended Wall functions).

But, I want also to run NHL simulation and I do not know if Numeca (by automatic) does not change SA (w EWF) to pure SA.

Do you happen to know this?

DarylMusashi · March 2, 2018, 07:01

Dear Cappytbg,

I don't think that it is reset to SA model without wall function approach. What you can do is to analyse the convergence histories of two computations with the standard SA model and the SA model with extended wall functions. You can easily load multiple computations in the monitor to compare them. The residuals and convergence of the flow quantities should slightly differ.

I have attached the button in the attached figure.

Kind regards,
Holger

Cappytbg · March 2, 2018, 08:27

Thank you for reply and your advice!
I will try to do as you said. Thanks!

Kind regards!

Cappytbg · March 15, 2018, 06:26

To finish this thread: I have compared those convergence histories and they are different, thus SA with EWF works with NLH method.

I have also found out (seen somewhere on Numeca's site) that from 6.1 (?) also two equation Turbulence Models are available, but I have not tried it, thus its not confirmed).
However in User's Guide 6.2 still is written that only SA is available.

March 1, 2018, 06:57	NLH with SA (Extended Wall Functions)	#1
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Hello! I have a question to NLH method. Particularity, I saw in Numeca Fine/Open User's Guide restrictions for use of NLH method, it says that the only possibility for this kind of analysis is the use of SA turbulence model, but... can you tell me, if Numeca Fine/Open can run simulation with SA (with Extended Wall Functions) model? I am running simulation of stator and rotor and want to catch their interference. Thank you for your help!

March 2, 2018, 06:23		#3
DarylMusashi Senior Member Holger Dietrich Join Date: Apr 2011 Location: Germany Posts: 174 Rep Power: 15	Dear Cappytbg, May I ask which kind of mesh you are using? If you created a "low Reynolds" mesh you resolve the boundary layer with a typical first cell height of 1E-5 [m] for turbomachinery applications and result in y+ values < 10. In this case a wall function approach does not make sense. The second scenario would be a "high Reynolds" mesh with only one cell in the boundary layer, the recommended y+ values are between 30 and 100. In this case an (extended) wall function approach is necessary. For turbomachinery applications I recommend the first. Is there a particular reason why you want to use the SA (extended wall function) model instead of the standard SA model? Kind regards, Holger eng.abdul likes this.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Wall functions - questions about implementation	AlmostSurelyRob	OpenFOAM Programming & Development	44	May 6, 2020 11:11
Wall functions vs. boundary layers	kandelabr	OpenFOAM Running, Solving & CFD	3	June 12, 2017 08:05
Boundary Field and Wall Functions, time 0 & time >0	NJG	OpenFOAM Running, Solving & CFD	1	April 5, 2013 11:48
wall functions implementation	besh	OpenFOAM Programming & Development	0	January 1, 2013 10:14
[ICEM] Export ICEM mesh to Gambit / Fluent	romekr	ANSYS Meshing & Geometry	1	November 26, 2011 13:11

March 1, 2018, 15:29		#2
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	I mean - simulation is running with this set, but my question is if Numeca does not changes automaticaly from this model.

March 2, 2018, 06:40		#4
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Dear DarylMusashi, Thank you for your answer! I am using unstructured mesh. And Re is high. You see, I have conducted a series of analyses so far, but I am making investigation. I have done analysis with y+<1, but now I want to check the wall functions approach. About y+ values, I have read Numeca's User Guide, and it says that to be in WF range y+ should be 20-50, what I obtained. And here I used SA (with Extended Wall functions). But, I want also to run NHL simulation and I do not know if Numeca (by automatic) does not change SA (w EWF) to pure SA. Do you happen to know this?

March 2, 2018, 08:27		#6
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	Thank you for reply and your advice! I will try to do as you said. Thanks! Kind regards!

March 15, 2018, 06:26		#7
Cappytbg New Member Kacper Palkus Join Date: Mar 2018 Location: Rzeszow Posts: 9 Rep Power: 8	To finish this thread: I have compared those convergence histories and they are different, thus SA with EWF works with NLH method. I have also found out (seen somewhere on Numeca's site) that from 6.1 (?) also two equation Turbulence Models are available, but I have not tried it, thus its not confirmed). However in User's Guide 6.2 still is written that only SA is available.