k-epsilon wall boundary conditions

Iain Barton

I am currently implementing a standard k-epsilon model.

I have noticed that interestingly there appear to be two (main) schools of thought how to implement the wall boundary conditions...

option 1 (supposedly the standard approach)

calculate y_plus, find tau_wall, replace source_k terms using special production and dissipation terms, then solve k

and for the epsilon value (adjacent to the wall) calculate using an expression based on k and height.

this seems to be an interesting modelling approach, since diffusion and convection of k is being modelled.

option 2 (seems to appear in more recent papers)

calculate u_friction using low of the wall replace the k and epsilon values (adjacent to the wall) both with prescribed values based on the frictional velocity, distance, etc.

this seems to be a fairly brute force approach.

==================================================

There are many variations one could come up with of course. Has anyone have any comments?

Why isn't there an option "0" where the source terms for both k and epsilon are modified for the "cells" adjacent to the wall, in terms of implementation this would be the easiest for me to do.

(I would suppose people would argue yes there is they are, in effect, the low-Reynolds wall dampening models).