[M_Tidswell]

Dear All

I wondered if anyone had any thoughts regarding this:

Quantifying error/uncertainty for steady state RANS models is a relatively well published subject (Richardson extrapolation etc) and it appears a lot of those leanings can be applied to unsteady RANS models although I suspect careful selection of the subject variable would be important.

Quantification of error for LES models seems to be a relatively young subject with a number of ratio tests between the SGS element and the LES element and modified Richardson extrapolation.
Where I'm becoming unstuck is how to apply these learning's to hybrid models such as SAS-SST, the grid resolution element is fully transferable but I'm struggling with how to include the additional modelling error contributed from the SST element. There are a number of options that could be modified to try and assess the changing error, ideally I would modify one that would enhance the near wall resolution but nothing is jumping out at me of the available list in CFX.

Has anyone else attempted to do this?

Many thanks

Mark