I came across a web article somewhere about a wall function method that actually involved moving the solid surface slightly so as to achieve the required y+ constraint. The thoery being that the results are far less sensitive to moving a wall very slightly compared to trying to get the required grid distribution.

I simply can't find the article any longer. Can anyone point me in the right direction? Thank you.

It seems interesting but what should you do then for a stagnation point where u+ tends to zero ?

Sylvain

Good point. Don't know. As with all these things, it is a shame to not proceed with an idea just becasue of a 'but' or 'what if'. I'm sure it was a commercial CFD vendor who published this info. Can't for the life of me remember who.

It looks like a CHAM (Phenix) or CD (Star CD) idea. I did look at such a think, but I gave up - question of time and of stagnation point ;-).

(1). Still talking about a variation of the black art? (2). In my opinion, most implementations of the wall function were incorrect. (or severely constrainted)

...which is why there's so much scope for novel methods, new approaches etc. Just because it ain't there yet is no reason not to push on.

(1). I have discussed the right approach here many times already. And I have used it since early 70's. (2). Since the turbulent flow is very common in real world application, incorrect implementation had already made the published results un-reliable, for the last 30 years I guess. (3). Once you become familiar with the low Reynolds number model implementation and its actual usage in an application, it is not difficult to understand the problem with the wall function implementations for the attached flows.

Try the CFX site. They've implemented it into TASCflow.