Both approaches are valid. Here are boundary conditions.

High Re form wall boundary conditions are:

nut  --> nutUSpaldingWallfunction (or nutKWallfunction or Equivalent)

k  -->   kqRWallFunction

omega -> omegaWallFunction

lowReForm boundary conditions are:

nut  > calculated

k  > fixed value (1e-12 to 1e-40) (low value basically)

omega > omegaWallFunction or fixedValue ( with calculation given by menter omega wall boundary condition.)

Y+ for High_Re form should between (30 - 300) (log region)

Y+ for Low_Re should be between ( 0 to 5) ( 1 and below gives great results though).

I will just add something from personal view:

I prefer to use kOmegaSST in low Re form. Because even though wall functions work quite well there are some circumstances when they don't represent the flow in the proper physics that it should. ( Example : Circumstances of large flow separation).

@Mohamed Thanks a lot for your comprehensive answer. I was doing the same at high Re but I had some doubts about that for low Re. Apparently, in OF version I am using (2.3), there are plenty of turbulence fields BC, those include nutLowReWallFn. I was wondering about the use of this one, specially when I looked at the source code and found that it does nothing except setting nut to zero and allowing the calculation of y+. Do you know when to use it? Maybe it is made for other low Re models?

I think Ashok is talking about ANSYS Fluent, As long as I am aware OpenFoam does not have an interface up to date. Abdel, NutlowReWallfunction is appropriate for transitional models such as KKL omega ( 3 eqn) model. I am not sure what other turbulence models this boundary condition works for.  

Regards and best of Luck

and in case of transitional modelling using RANS, should I actually also keep my y+>30 (in the log region)? I have some doubts about that because in Low Re in general, we need a very coarse mesh to keep it in the log region when we use the wall function, which affects the quality of the resolved fields. I hope that the answer is no and that we are actually allowed to resolve the first point in the laminar region.

KKl omega in OpenFOAM is a low reynolds transitional model. It needs Y+ to be

@Mohamed That's the answer that makes me satisfied. Thanks a lot taking the time to reply.

Have a nice weekend

 @Mohamed I had some doubts about the zeroGradient boundary condition about nut(nuSgs).The  nusgs file in incompressible/pisoFoam/les/pitzDaily/0(OpenFoam 2.4.0) sets all boundary conditions to zeroGradient, and it seems that there is no wall function used.I wonder what is the difference between calculated and zeroGradient.

Thanks.

Calculated is calculated from other patch fields.

ZeroGradient means that the normal is zero. Look at the source codes of patchFields for these two boundary conditions if you want more details.

Regards

 @Mohamed I am wondering why use omegawallfunction in BC with Low Re modelling.

Thanks in advance

OmegaWallFunction just sets the correct omega at wall for each first cell. Alternatively you can also use fixedValue boundary condition, calculate omega at wall and use this number. But as the first cell height varies over the geometry it would be easier to use the automatic omega values calculated from OmegaWallfunction. Dont you think