Got something peculiar when I switched from k-e to SST turbulence model in Pre:

The (normally empty) physics errors window got all red and accusing, telling me I've got Wall roughness as a parameter on all my wall boundaries, but its an invalid parameter. Fixing was as easy as opening the wall boundary specification window and saving it again, but it got me thinking why is wall roughness not an option when running the SST model. Not that I've used it before, (usually specified smooth wall) but I can't remember it NOT being there in 5.5 when I've used SST...

I was under the impression that wall roughness was not possible in CFX 5 when using omega based turbulence models. According to Wilcox the implementation is quite straigtforward but it hasn't been done in CFX.

Hi Bart, how does the roughness work with SST? Does the roughness height still have to be smaller than the first cell height ? Would that not then make the roughness height very small as for the SST model you are going down to the levels of Y+ =1-2 ?? Just curious as we have just come across this issue ourselves. Cheers Bob

According to the manual of CFX 5.6 SST is k-w based (near the wall at least). In that case it is not possible in CFX (see page 76 of solver theory) to work with wall roughness (not for free slip, laminair and omega base turbulence models). If you want to use a wall roughness you have to use epsilon based turbulence models. The fact that your first grid height should be higher than 50% of the wall roughness still applies. You will get warnings during the solver run about this. For high wall roughness this results in very high y+ values. Apperantly you will have to accept this. I have now idea what this does to your solution (wall shear stress, pressure loss, etc). Maybe you can help me there, what the general implications are of such a high y+. It is still strange that CFX does not have k-w models with wall roughness as it is easier (according to Wilcox) to implement them (anyone from CFX to comment on this?)

I indeed thought that there's no roughness in k-Omega based model since one integrates to the wall, ie roughness has to be part of the described geometry.

[venkatesh92]

Hi,


Looks like quite an old thread! I believe that wall roughness can be incorporated in omega-based models by including the roughness as part of the geometry. There are a few situations where the performance of omega models are much more superior than epsilon based models. Please correct me if you think this is not the way to approach this problem. Also suggest alternatives to implement wall roughness with omega models.

Thanks

[Opaque]

Not sure if using an old thread is any good for starting a discussion for the treatment of wall roughness when using omega based models.

Reading the documentation of most recent releases, you will find out how wall roughness is incorporated into omega based models within ANSYS CFX or ANSYS Fluent.

Using the latest implementation as starting point will save discussion time, and contributions will help improve (if needed) the current understanding.

[venkatesh92]

Thanks for your reply!

The latest documentation from Ansys remains silent on implementation of roughness in SST k-w for instance. Is there an UDF for this purpose? I would be grateful if someone can share it!

Cheers