Hi all,

I have found this site very very useful for my study, and I think I can get a very good reply for my question.

I am trying to simulate Channel Flow using LES, can some one tell me how to go about calculating iteratively the Frictional Velocity (U_tau) and Yplus, in order to apply the log law along the wall of the channel, I found this method metioned by Piomelli and Balaras, in their paper " WAll-Layer models for LES", but I am stuck about in calculating the the U_tau and Yplus.

Ur help will be highly appreciated,

Thanks in advance,

Harry.

Use Newton-Raphson. It converges very fast and typically in 1 iteration starting from the conditions at the previous time step.

However, an instantaneous log-law is a fairly poor near wall model for LES. Various approaches for LES wall functions were studied in the 80s (various Stanford reports come to mind) and the early LES researchers like Deardorf and Schumann used such models so reading their papers should prove useful.

HI Andy,

Thanks for the reply, as directed by you, I got hold of the papers by Deardoff,Schumann and also Piomelli and Ferzier.

In Piomelli and Ferziers model, they propose something called the shifted and ejection model, they also define a term called ∆s and they define it to be, y*tan(8) for yplus 30-60 and y*tan(13) for larger yplus values.

So where exactly is this ∆s added, is it to the streamwise and spanwise velocities..while calculating the <up> and <wp>...

p-> is the first point over the wall <.> is the plane averaged value immediately above

the wall

Thanks,

Harry.

It would probably be unwise for me to guess without looking at the paper and the the details of the model.

In my experience with near wall LES models the challenge does not lie with making minor improvements to the prediction of healthy boundary layers with the addition of terms of debatable physical significance. The main challenge lies in creating a model that behaves reasonably in the presence of recirculations (i.e. large eddies which the LES model is good at predicting unlike RANS models). Around such regions empirical information based oh healthy boundary layers is not applicable.