[AlmostSurelyRob]

Dear Foam Users and Developers,

I am really interested in understanding wall functions well. The main reason for this interest is that I would like to know precisely which RAS model is compatible with each of those functions.

Therefore, I embarked on the quest to understand the implantation in OF. To paraphrase one sketch, I would like to emerge one day from this journey, probably a little bruised, but hopefully a little wiser. So here I am after a few days of literature review and perusing the code with a set of compiled questions - not yet wiser, but not too bruised as well. I am hoping to get both with this post.

I am looking at OF 2.0

Q1)
nutUWallFunction stands for Standard Wall function. yplus is estimated with at most 10 Newton iterations under given U and y and then turbulent viscosity is modified.

According to the literature e.g. Wilcox having u_\tau allows to adjust k and epsilon/omega. But in OF I don't see any equivalent wall boundary conditions for k and omega. Instead there is an implementation which changes omega/epsilon and adjusts the production term of k. In fact k and omega/epsilon have only one wall function each. Is nutUWallFunction compatible with epsilon and kqR wall functions?


Q2)
G is k production i.e. \bar{uv} dU/dy. Please look at the comments, the code is taken from omega wall function.

G[faceCellI] =
//This is an estimate of shear stress \bar uv
            (nutw[faceI] + nuw[faceI])
           *magGradUw[faceI]
//This is an estimate of dU/dy from the logarithmic velocity profile
           *Cmu25*sqrt(k[faceCellI])
           /(kappa_*y[faceI]);

Are these comments right and if they are what is the point of having laminar viscosity in shear stress calculation and then calculating the gradient from log-law anyway?

Q2') When is this actually invoked? k equation is always solved second in RAS. omega/epsilon is always solved before and it has this suspiciously looking method boundaryManipulate.

    omegaEqn().boundaryManipulate(omega_.boundaryField());

Q3)
What is db object? It appears from for example here:

    volScalarField& G =
        const_cast<volScalarField&>(db().lookupObject<volScalarField>(GName_));

Q4) Am I right that k wall function in OF is just zeroGradient?

Q5) Finally, what is the correct boundary condition for nut in k-epsilon low-Re models? Should I put calculated or should I put nutLowReWallFunction or... is there no difference? The latter seems to have only constructors and cloning functions.


I will appreciate any comments or hints on the above questions. I find OF implementation very neat but as I outlined above I have still some doubts and gaps in my knowledge.