[1pnc] remove too strict newton criterion as this leads to different timesteps with different compilers.
additionally shorten test times to a time where velocities are higher

until better solution is found

Timo Koch authored 5 years ago and Katharina Heck committed 5 years ago

The implementation due to commit a78e605e
was incorrect for cc-models. For the upcoming release we throw a
not-implemeneted compiler error until a correct implementation is provided.

Beatrix Becker authored 5 years ago and Katharina Heck committed 5 years ago

DUNE_THROW if error to reference solution gets too high

Beatrix Becker authored 5 years ago and Katharina Heck committed 5 years ago

[test compositionalflash] add tests against sequential routines to determine reference state in one- and two-phase state

Katharina Heck authored 5 years ago and Kilian Weishaupt committed 5 years ago

Positive diffusive fluxes mean that the inside enthalpy should be used
as it is an outflux

Also remove unnecessary inclusions of the header.

Katharina Heck authored 5 years ago and Kilian Weishaupt committed 5 years ago

mass reference system

Katharina Heck authored 5 years ago and Kilian Weishaupt committed 5 years ago

same molar mass. Otherwise velocity will arise due to diffusion which is
not intended in the tracer model and does not make sense physically

Katharina Heck authored 5 years ago and Kilian Weishaupt committed 5 years ago

molar masses of the components.

changes only affect documentation

Simon Emmert authored 5 years ago and Katharina Heck committed 5 years ago

H2OHeavyOil and 3pImmiscible with minor fixes or restructuring of
binary/diffusionCoefficient according to @bernd 's wishes.

Katharina Heck authored 5 years ago and Timo Koch committed 5 years ago

and correct dimensionless numbers in the references

fde55173 introduced the previous solution only for the reactionAssembler 
and not the twoPAssembler in the 2ptracer-test

Bernd Flemisch authored 5 years ago and Kilian Weishaupt committed 5 years ago

Test for the minimal interface of gasDiffCoeff and liquidDiffCoeff.
Allow throwing a NotImplemented.

This was previously not possible because of the bug in the
explicit/numdiff assembler. The large epsilon masked the error
in the storage derivative. Now it also works with small epsilon.

This reverts changes introduced in
aa5951f5 and the following commits.
It turned out that doing the flux computation on the basis of a support
point inside the cell and then a tpfa flux towards the facet elements
performed very bad for anisotropic tensors.