[tyskies]

Hello everyone

I'm currently computing the turbulent flow over a flate plate in order to prove that wall functions are useful to reproduce the theoretical laws of the wall.
My simulation settings are quickly summed up below:
- solver: simpleFoam.
- boundary layer mesh spec: around 120 cells in 0.1m with refinement near the wall.
- turbulence model: k-epsilon.
- wall functions: kqRWallFunction, epsilonWallFunction, nutkWallFunction.

Here is how I evaluate y+ and u+:
1) First, I get the value of the wallShearStress t_w (simpleFoam -postProcessing -func wallShearStress):

t_w = sqrt(t_w(x)² + t_w(y)² + t_w(z)²)

2) Then, I evaluate the friction velocity value:

u_tau = sqrt(t_w)

NB: Units of wallShearStress values in openFoam are m²/s² instead of kg.m^-1/s². So we don't need do divide by rho (density).

3) Third, I evaluate y+ and u+ for each y:

y+ = y * u_tau / nu, where nu is the kinematic viscosity.

u+ = u(y) / u_tau, where u(y) is the x-component of the velocity field

Is there anything wrong in my methodology ?
Someone know how yPlus is calculated by openFoam with the utility: simpleFoam -postProcessing -func yPlus ?

I hope someone will be able to help me. Thanks in advance!

[HPE]

The line of computations is OK to my knowledge. Yet didn't quite understand the main question.

Two points:
- For the zero-gradient flat plate case, consider t_w(x) only - and ignore the other two components.
- Make sure using mag(t_w(x)) to avoid the negative sign - otherwise sqrt will throw a domain error.

[tyskies]

Thanks for your quick reply and advice !

I just dont understand why my y+ and u+ dont match with theoretical values (Spalding law). Is that because my mesh is too refined near the wall (openfoam gives me a yPlus of 0.32) for a kEpsilon model ?

Otherwise, I noticed that the pressure isnt really converged (residual=10^-3).

NB: My case geometry is based on the following one: https://www.openfoam.com/documentati...plate-zpg.html

[HPE]

`kEpsilon` is tricky for y+<30 meshes since the derivation assumption of this turbulence closure model is y+>30 (i.e. it cannot model below the inertial sublayer). Having said that you are using wall functions - which should allow you to model below y+ 30, so I presume the issue is relevant to the settings used for `kEpsilon` or `k`-`epsilon` fields.

Note that in the link you have shared, the turbulence model being used is `kOmegaSST`.

See if the ZPG flat plate tutorial helps you: https://develop.openfoam.com/Develop...ulentFlatPlate where you can find some settings for `kEpsilon`. This tutorial is more or less the replica of the ZPG case available in the NASA turbulence website.

Hope these help.

[tyskies]

Thanks, you help a lot !

I computed the case from the link you shared. It seems that with kOmegaSST it's easier to reproduce the log-law and I'm not surprised with that results because this model is way better for modelling the boundary layer. However, for the kEpsilon model it still not good... It looks like wall functions aren't working well in my case.