[kandelabr]

Hi.

I noticed the recent addition of nutUBlendedWallFunction to OF1712 but I'm not sure it's what I think it is:

Description
     This boundary condition provides a turbulent kinematic viscosity condition
     when using wall functions, based on a blending of laminar sub-layer and
     log region contributions.

That means it handles cases where boundary cell size in mesh is neither y+<1 for LowRe nor y+>30 for wall functions?

I changed nutkWallFunction to nutUWallFunction in simpleFoam/pitzDaily tutorial and RAS model to kOmega but it produces a floating point exception right away. Should I change something else as well? What's the difference between nutk* stuff and nutU* stuff?

Thanks!

[Bazinga]

Have not used it, but AFAIK blended wall functions can be used for any y+. They work in the laminar, transition and log-layer. nutk are wall functions which use k for approximation and nutU use the velocity at the wall.

In Doxygen this paper is referenced:

Menter, F., Carregal Ferreira, J., Esch, T., Konno, B. (2003).
The SST Turbulence Model with Improved Wall Treatment
for Heat Transfer Predictions in Gas Turbines.
Proceedings of the International Gas Turbine Congress 2003 Tokyo

There is also a note saying:

The full 'automatic wall treatment' description also requires use of the Foam:megaWallFunction with the blended flag set to 'on'

This might be the reason for the simulation blowing up.

[kandelabr]

This is all very confusing. I found this 'blended' wall function in ESI-CFD version but now I'm looking in OpenFOAM Foundation omegaWallFunction and it also provides some kind of blending.

What's more, the source also says

Recent tests have shown that the standard switching method 
provides more accurate results for 10 < y+ < 30 when used with high 
Reynolds number wall-functions and both methods provide accurate results 
when used with continuous wall-functions. Based on this the standard 
switching method is used by default.

And there's also kLowReWallFunction which, according to source:

Description
    This boundary condition provides a turbulence kinetic energy wall function
    condition for low- and high-Reynolds number turbulent flow cases.
    The model operates in two modes, based on the computed laminar-to-turbulent
    switch-over y+ value derived from kappa and E.

So if I can provide y+>10 I can 'safely' use switched k and omega functions and there's no mention of switching in any nut wall function anyway. I have even seen all 'nut' fields on wall set to 'calculated' so why use blended wall functions or why even use 'nut' functions if it can be calculated from k and omega?

Thank you for your effort, I appreciate your help. I'll also read the paper though I'm afraid it'll take some time to grasp all the gory details

[Bazinga]

This is what I remember from my research some time ago. Someone might be able to correct my if I am wrong.

The wall functions for omega and u_tau as described in the above-mentioned paper (see Eq. 16 and 18) are the blended versions. According to the paper, they can be used independently of the mesh size.

This new wall boundary condition has been implemented in
combination with the SST model (Grotjans and Menter, 1998). It
exploits the simple and robust near wall formulation of the
underlying k-ω model and switches automatically from a
low-Reynolds number formulation to a wall function treatment
based on the grid density. The advantage is that the user can make
optimal use of the advanced performance of the turbulence model,
for a given grid. The automatic wall treatment avoids the
deterioration of the results typically seen if low-Re models are
applied on under-resolved grids.

In OpenFOAM 17.06 the Equations 15 and the Equations 17 are used for omega and u_tau and therefore nut (overall 4 equations) by default, a switch of calculation takes place dependent on the y+ value. However, you can use Eq. 16 and 18 using "blended yes". The kqRwallfunction is a zeroGradient condition. The kLowReWallFunction is explained in the following work, but I have not used this WF or checked the source thoroughly:
http://www.tfd.chalmers.se/~hani/kur...nfoamFinal.pdf

nut wall functions are used for better estimation at the wall. There is switching for nut in the nutUBlendedWallFunction as explained above.

Till now I either used meshes with y+> 30 using wall functions for omega, k and nut or y+<5 where I did not use wallfunctions or only the omegawallfunction. There are some threads here in the forum giving some advice.

If I had to make a guess for meshes which vary y+ from 1 to above 30 I would use omegaWallFunction with blended yes, nutUBlendedWallFunction and kqRwallfunction.

[kandelabr]

Finally an answer that makes sense. Thank you, this is what I needed.

[snak]

There was a bug in nutUBlendedWallFunction of v1712. Using nutUBlendedWallFunction in simpleFoam/pitzDaily tutorial case will reproduce this error. Zero gradient at the initial state leads to yPlus of zero.
https://develop.openfoam.com/Develop...lus/issues/696


It is fixed in development-version. You can get it from OpenFOAM-plus gitlab.

You have to have an account there.
https://develop.openfoam.com/Development/OpenFOAM-plus

[kandelabr]

Hehe, I never thought I'd find one.
Thanks for the information.

I'm currently stuck on Windows with OF4Win from Turbomachinery CFD and I'll also miss the cyclicRepeatAMI BC. Not sure which one I'd miss more.