In the case of specific problem, such as natural convective flows, probably a modification of the standard wall functions could improve the predictions. Does anyone know if it is possible to modify the wall functions giving, for example, non linear relationships between the constant of the velocity profile and the variables such as y+ or wall shear stress ?

Thanks you in advance

Hi,

The wall function approach has been extensively validated and fiddling with the values of constants is not a good idea. If you need to fiddle with wall function constants then you should abandon wall functions. If the default wall functions are not proving accurate I recommend not using wall functions but integrating to the wall (ie y+=1).

Glenn Horrocks

I don't completely agree with you in the sence that the standard wall functions have been formulated in equilibrium forced convective boundary layers. In recirculating flows or when the natural convection plays a significative role, the departure from the standard wall functions is significative. In these case some authors provided new wall functions and their predictions are in well agreement with the corresponding experimental data.

You won't make standard wall functions accurate in recirculating or natural convection flows by fiddling with the model constants! The entire approach is not applicable so that is why you need to abandon wall functions and integrate to the wall.

Glenn Horrocks

Glenn is absolutely right , if u at all want to use wall function then forget about resolving boundary layer or the thermal boundary layer as in your case (assuming u r using CFX solver ) . All I could suggest is that first of all change the turbulence model to SST and use automatic wall function only if you mesh have a yplus betweeb 0 to 10 , though I would reccomend to you to have yplus between 0 to 1 but it depends on many factors of your simulation goal. Ya modification of wall function is not easy just by some number for sure .