BUG: nut{k,U}WallFunction: incorrect y+ estimations for viscous sublayer

Summary

In a nutWallFunction-simulation, when yPlus FO is called, the FO calls yPlus() of the nutWallFunction (unless the new commit 3e41c100 is in place).

yPlus() contains the y+ definition specific to the chosen nutWallFunction rather than the formal definition of y+ in order to ensure various consistencies.

yPlus() of some of the nutWallFunctions produces incorrect results for y+<10 (i.e. approximately viscous sublayer).

The affected items are:

• yPlus FO
• compressible::alphatJayatillekeWallFunction#L239
• alphatJayatillekeWallFunction#L242

Minimal working example

See !381 (closed) for a MWE.

Problem

Although each nutWallFunction correctly computes and blends nut predictions for viscous and inertial sublayers according to their definitions, yPlus() computes only the inertial sublayer y+.

Therefore, predictions for viscous sublayer y+ from yPlus() are incorrect.

Environment

• OpenFOAM version : 7be30924/all maintained versions
• Operating system : opensuse
• Compiler : gcc/clang

Possible fixes

• Compute viscous and inertial sublayer y+ nominal (theoretical?) predictions in yPlus(), and blend them according to the chosen blending method. The missing information = how can we compute the nominal y+ for the sub-inertial region from dependent variables, e.g. k.
• Extend the analytical expression of the inertial sublayer up to the boundary by covering the viscous sublayer by using Spalding's law of the wall.