* However, based on the velocity vectors in the direct neighbours of the volume face, it is not possible to compute all components of the stress tensor.
* Only the gradients of u and v in the direction of the vector between the cell centres is available.
* To avoid the reconstruction of the full velocity gradient tensor based on a larger stencil,
* the horizontal contribution to the eddy viscosity (in the mixing-length model) is computed using only the velocity gradients normal to the face:
It should be noted that this simplified approach is formally inconsistent and will result in a turbulent viscosity that is dependent on the grid (orientation).