[aerocfdD]

Hey,

I could not use the wall function option in su2 version 7.1.0 . I am trying to solve basic external flow problem with farfield (farfield BC) and a blunted body (heat flux BC).

The error I encountered : Tau_wall evaluation has not converged in CNSSolver.cpp

Could you please help me about using the wall function?

Thanks.


% Viscous wall markers for which wall functions must be applied. (NONE = no marker)
% Format: ( marker name, wall function type -NO_WALL_FUNCTION, STANDARD_WALL_FUNCTION,
% ADAPTIVE_WALL_FUNCTION, SCALABLE_WALL_FUNCTION, EQUILIBRIUM_WALL_MODEL,
% NONEQUILIBRIUM_WALL_MODEL-, ... )
MARKER_WALL_FUNCTIONS= ( body, STANDARD_WALL_FUNCTION )

[CSMDakota]

Hi aerocfdD,

I have seen this issue as well, at least in v7.0.0.

At one time I looked farther into it and concluded the iterative solver for Tau_wall is not always robust enough to converge. If I recall correctly, it runs a maximum of 10 iterations with a 25% correction of the error between the initial value and updated value for Tau_wall. In some cases (e.g. a handful of cells in a domain) I found that the algorithm will reach a condition where it toggles between two values for Tau_wall (well apart from the convergence condition) and thus it never converges. I was able to replicate the scenario in a spreadsheet, and even when given thousands of iterations to converge it never stabilizes in these cases. So for now I've concluded that the wall function scheme isn't yet reliable for some problems, so when I need accurate near-wall modeling it's best to adjust my grids to get yPlus ≈ 1 and not use the SU2 wall models.

Regards,
Brandon

[pcg]

Hello,
This has been "work in progress" for a long time but now actual progress is being made https://github.com/su2code/SU2/pull/1204
The person working on it improved that iterative solver.
Stay tuned or go over to github and use the branch directly (with caution) attention from the "outside world" is always good motivation, and pretty much the only reward we developers get.
Cheers,
Pedro

[bigfootedrockmidget]

Hi, "the person working on it" here. I would appreciate it if you could try your case on the Work-In-Progress feature branch. I'm mainly working on incompressible cases, so I've not done any testing yet with compressible, and I'm assuming that your case uses the compressible solver? The original implementation also did not have walls with nonzero heatflux implemented. Do you have a heated wall or is your heat flux zero?

[CSMDakota]

Hi bigfoot,


That's awesome you're working on the wall functions, I think this could be a big improvement to the code! I'll get my hands on your WIP code and try it on a grid here soon. Yes, my cases in the past were compressible with a mix of isothermal and adiabatic wall conditions.

[aerocfdD]

Originally Posted by bigfootedrockmidget

Hi, "the person working on it" here. I would appreciate it if you could try your case on the Work-In-Progress feature branch. I'm mainly working on incompressible cases, so I've not done any testing yet with compressible, and I'm assuming that your case uses the compressible solver? The original implementation also did not have walls with nonzero heatflux implemented. Do you have a heated wall or is your heat flux zero?

Hello,

Thanks for all your helps and care. Sorry I could not write on the form earlier. Yes ı am trying to solve the compressible flow case with zero heat flux wall. However, I am also interested with different problems such as internal flow problems with heated walls . Are the wall function code going to be fully usable for all type of problems ? I will try Work-In-Progress feature branch version of the code as soon as possible for the case ı asked and other cases. Thanks again for this wonderfull open-source code and your supports .

[bigfootedrockmidget]

I made an update over the weekend and it runs smoothly now for my incompressible diffuser case. At the moment it does not handle (ignores) the effect of isothermal boundary conditions on the boundary layer.