* reference solution needs to be updated because test uses time-depended Dirichlet BC
* old behaviour was actually wrong

* only local element needs to be bound

Melanie Lipp authored 6 years ago and Kilian Weishaupt committed 6 years ago

[freeflow][outflow BC] Introduce that there is no gradient normal to the outflow boundary in cases where the outflow boundary is the normal face with respect to the assembled scvf.

[test][navierstokesnc] Fix test test_msfreeflow. Make binaryDiffusionCoefficient have the same argument as in fluidsystem base class.

Bernd Flemisch authored 6 years ago and Timo Koch committed 6 years ago

Also fix the test itself in the following sense. By default, in the
box method, material law parameters are evaluated at scv centers,
while primary variables are evaluated at vertices. If no interface
condition is used, the secondary variables handed to the Vtk
writer might now differ depending on the order in which the scvs
are evaluated for each vertex. In order to saty consistent, this
test evaluates the material law parameters at the vertex position.

Also replace all calls to isLiquid by equivalent calls to isGas

move the calculation of enthalpy and viscosity to the volumevariables
and adapt the interphaces of the solve-methods

* internally forward to isOnWallAtPos
* needed for coupling

Kilian Weishaupt authored 6 years ago and Thomas Fetzer committed 6 years ago

* remove now obsolete NavierStokesNCIndices

* test was passing but vtk #39 was compared although there are 40 steps

* get rid of fluidSystemPhaseIdx

We can use one-phase adapter for reusing m-phase fluid systems now.
Reverts 7bb9a857 which introduced a bug for the diffusion coefficients