Mesh size with y+ > 30 criterion too rough
 

Hello everybody,

I'm creating a workflow for shape optimization of a cylindrical nozzle using pisoFoam. I've finished as a first step a laminar simulation, know I'm trying to create an adequate mesh for a turbulent simulation. I will use the k - epsilon model, therefore I have to ensure that the height of my fist cell near to the wall is higher than y+ = 30 to meet the log-law-region, because I want to use the openFoam standard wall functions. My inflow-velocity of interest causes a Re-number of 10000 at the inlet with a diameter of 1m. So I can assume a fully developed turbulent flow.
My Problem is now, that when I estimate the height of the first cell height using the y+ calculater e.g. http://www.pointwise.com/yplus/ I get a size of 0.05 m for y+ = 30. That is much too rough, so I can't adequate dissolve the flow. So what I can do? Is my Re-number too low for wall functions or can I use the wall functions with finer sizes without causing trouble?
Of course I can use y+ < 1 and solve the viscous sublayer without using wall functions, but later I will simulate flows with higher Re numbers and I want prevent the additional numerical cost, since the boundary layer are not important for my optimization.

Thank you very much for your help!

Best regards

Gerrit

[You can use the low Reynolds formulation, it is appropriate for y+ around 1 (it should be in the viscous sublayer, which would be y+<5).
So use kLowReWallFunction, epsilonLowReWallFunction and nutLowReWallFunction.]

Sorry, I think what I said up here is incorrect as pointed by Fabian in the post below. The kLowReWallFunction seems to be only applicable to the v2f model.

Thanks, that's working fine!

Moreover, you should consider using a lowRe turbulence model. As far as I know, kEpsilon is only highRe. But this strongly depending on your case. Try LaunderSharmaKE.
Besides, kLowReWallFunction wallFunction is only for v2f model.

Have a look into this thread http://www.cfd-online.com/Forums/ope...cing-step.html

Cheers

Fabian

Thanks for this hint! To be on the safe side I will simply use k = 0 instead of
the kLowReWallFunction in combination with LaunderSharma. Unfortunately, I have a y+ value around 4, I hope this bc is still accurate. With y+ = 1 my mesh was much too big,
while using y+ around 30 and using highRe models my mesh was too rough with respect to my application. It's hard to find the balance.

Best regards

Gerrit

Precisely, Fabian. Although, I haven't been able to actually figure if the kLowReWallFunction and epsilonLowReWallFunction are actually only for the v2f, im taking a look at the code.

Hi,

I'd didn't say that epsilonLowReWallFunction is only suitable for v2f model. Only kLowReWallFunction is for v2f. epsilonLowReWallFunction can be used with other kEpsilon lowRe models such as LaunderSharma etc.

Cheers

Fabian

Hai Fabian and others,

could anyone help me? I went through the derivation part of wall function for kPlus and epsilonPlus and also the codes kLowReWallFunction, epsilonLowReWallFunction in OF. As far as I know, derivation for kPlus and epsilonPlus do not depend on V2 or f term. So why can't these be used for kEpsilon Turbulence model?

can anyone explain why kLowReWallFunction cannot be used for kEpsilon model? My point here is, even if kEpsilon model equations were used to derive for kPlus and epsilonPlus, the algebraic equations are not going to change, since the k and epsilon equations of both kEpsilon and V2F models are the same. Please correct me if I am wrong.

Thanks,
Karthick