wall treatment for EARSM model

fxzf · May 28, 2019, 06:49

Hi All,

I open a question here. When we develop EARSM turbulence model.

Does wall function like omegaWallFunction, nutUSpaldingWallFunction, nutkRoughWallFunction need to be adapted for EARSM with nonlinear term in?

In my EARSM model, I found the results is good when mesh Y+ = 1, but when my Y+ is not 1, results is very weird. I am wonder if wall function need to be adapted as well.

Anyone has any idea would appreciated!

Cheers,

mAlletto · May 29, 2019, 06:17

Yes you have to modify the existing wall functions to be consistent with the model since they are only valid for linear eddy viscosity models.

For standard wall functions what you assume is constant shear stress over the first cell and equilibrium between production and dissipation.

Since both Production and shear stress are different compared to linear eddy viscosity, the wall functions for nut have to be modified.

The omega wall function most probably too.

fxzf · May 30, 2019, 04:53

Quote:

Originally Posted by mAlletto

Yes you have to modify the existing wall functions to be consistent with the model since they are only valid for linear eddy viscosity models.

For standard wall functions what you assume is constant shear stress over the first cell and equilibrium between production and dissipation.

Since both Production and shear stress are different compared to linear eddy viscosity, the wall functions for nut have to be modified.

The omega wall function most probably too.

@mAlletto

Thanks very much for your reply. Finally, I can find someone to talk this.

I checked omegaWallFunction, for omega value near wall, it mainly based on y+ https://www.cfd-online.com/Wiki/Near...k-omega_models so seems not too much related to nonlinear terms. However, there is G0 in the code to update production term. I think this one need to add nonlinear stress term in.

After I applied this change. It changed results very little,

Therefore, I think the main focus should be in nutUSpaldingWallFunction, nutkRoughWallFunction (if I want to apply roughness).

The problem is it seems we need derive it from scratch. Do you know if there is any paper talk about nonlinear stress near wall?

Thanks very much.

Cheers,

mAlletto · May 30, 2019, 08:18

Unfortunately I'm not aware of any paper describing. But there is a good report describing wall functions in general and the assumptions made which lead to the formulation of wall functions: http://www.tfd.chalmers.se/~lada/pos..._report_WF.pdf

Regarding the EARM i think deriving an adequate nut wall function should not be to difficult.

Since you an assume the flow close to the wall as nearly one dimensional (if you assume equilibrium and neglect stream line curvature effect and also the influence of the pressure gradient) all higher order terms of the expansion made to derive the EARM can be neglected and you end up with the only term remaining $\nu_t S_{ij}$ . Since in the EARM you can express $\nu_t$ similarly to the base line model as $C_{\mu} k / (\omega \beta^*)$ . (see https://turbmodels.larc.nasa.gov/easmko.html)
I think you can just replace the constant cmu in the nut wall function by the variable of the EARM.

But if I wear you I would try to derive the relations by your own.

Best

Michael

fxzf · June 3, 2019, 03:44

Hi Michael,

Thanks so much for your reply and information. These are very useful.

First, I am actually using the following paper, which is K Omega SST with Nonlinear correction. https://www.researchgate.net/publica...rbulence_model . So in this paper, Cmu is a constant 0.09 as normal k omega sst model.

Second, I quickly went through the wall equations. Then, I end up the similar conclusion which you suggested me. If we assume the near wall is nearly 1D flow and high order term are very small. It just need to replace some constant like Cmu by variables in nut wall function. In my case, my Cmu still a constant, so it seems I don't need to do anything. However, as my result shown, my wall shear is messed when mesh y+ is big.

So I am trying to go through some details and derive this relations.

Keep in touch.

Cheers

mAlletto · June 3, 2019, 08:17

Can you just post some figures to better understand the difference between the wall resolved (y+ ~ 1) and the wall modeled simulation. Depending on the wall model it may be that the wall functions are only valid if the first cell center is in the log layer (y+ > 40).

May 28, 2019, 06:49	wall treatment for EARSM model	#1
fxzf Member ff Join Date: Feb 2010 Posts: 81 Rep Power: 16	Hi All, I open a question here. When we develop EARSM turbulence model. Does wall function like omegaWallFunction, nutUSpaldingWallFunction, nutkRoughWallFunction need to be adapted for EARSM with nonlinear term in? In my EARSM model, I found the results is good when mesh Y+ = 1, but when my Y+ is not 1, results is very weird. I am wonder if wall function need to be adapted as well. Anyone has any idea would appreciated! Cheers,

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Natural convection in a closed domain STILL NEEDING help!	Yr0gErG	FLUENT	4	December 2, 2019 00:04
Radiation in semi-transparent media with surface-to-surface model?	mpeppels	CFX	11	August 22, 2019 07:30
Does realizable k-e model with enhanced wall treatment work for surface roughness?	zyy95900	FLUENT	5	November 15, 2017 16:13
Enhanced wall treatment, low re k-e model, k-w model with low Re correction	junz	FLUENT	0	April 21, 2015 15:36
Wall treatment - RSM turbulence model	djack	STAR-CCM+	1	September 12, 2012 10:51

May 29, 2019, 06:17		#2
mAlletto Senior Member Michael Alletto Join Date: Jun 2018 Location: Bremen Posts: 615 Rep Power: 15	Yes you have to modify the existing wall functions to be consistent with the model since they are only valid for linear eddy viscosity models. For standard wall functions what you assume is constant shear stress over the first cell and equilibrium between production and dissipation. Since both Production and shear stress are different compared to linear eddy viscosity, the wall functions for nut have to be modified. The omega wall function most probably too.

May 30, 2019, 08:18		#4
mAlletto Senior Member Michael Alletto Join Date: Jun 2018 Location: Bremen Posts: 615 Rep Power: 15	Unfortunately I'm not aware of any paper describing. But there is a good report describing wall functions in general and the assumptions made which lead to the formulation of wall functions: http://www.tfd.chalmers.se/~lada/pos..._report_WF.pdf Regarding the EARM i think deriving an adequate nut wall function should not be to difficult. Since you an assume the flow close to the wall as nearly one dimensional (if you assume equilibrium and neglect stream line curvature effect and also the influence of the pressure gradient) all higher order terms of the expansion made to derive the EARM can be neglected and you end up with the only term remaining $\nu_t S_{ij}$ . Since in the EARM you can express $\nu_t$ similarly to the base line model as $C_{\mu} k / (\omega \beta^*)$ . (see https://turbmodels.larc.nasa.gov/easmko.html) I think you can just replace the constant cmu in the nut wall function by the variable of the EARM. But if I wear you I would try to derive the relations by your own. Best Michael

June 3, 2019, 03:44		#5
fxzf Member ff Join Date: Feb 2010 Posts: 81 Rep Power: 16	Hi Michael, Thanks so much for your reply and information. These are very useful. First, I am actually using the following paper, which is K Omega SST with Nonlinear correction. https://www.researchgate.net/publica...rbulence_model . So in this paper, Cmu is a constant 0.09 as normal k omega sst model. Second, I quickly went through the wall equations. Then, I end up the similar conclusion which you suggested me. If we assume the near wall is nearly 1D flow and high order term are very small. It just need to replace some constant like Cmu by variables in nut wall function. In my case, my Cmu still a constant, so it seems I don't need to do anything. However, as my result shown, my wall shear is messed when mesh y+ is big. So I am trying to go through some details and derive this relations. Keep in touch. Cheers

June 3, 2019, 08:17		#6
mAlletto Senior Member Michael Alletto Join Date: Jun 2018 Location: Bremen Posts: 615 Rep Power: 15	Can you just post some figures to better understand the difference between the wall resolved (y+ ~ 1) and the wall modeled simulation. Depending on the wall model it may be that the wall functions are only valid if the first cell center is in the log layer (y+ > 40).