* 2p1cni model to simulate steam injection

Holger Class authored 8 years ago and Bernd Flemisch committed 8 years ago

lecture. Unlike there, we use here a non-zero capillary pressure to use
rel-perms and capillary pressures from dumux stable

The installexternal.sh script can be used to install opm
and ert and to apply patches to opm-common and opm-parser.

With Opm release 2016.04, the module dune-cornerpoint has been
renamed to opm-grid. Interfaces changed without deprecation. There
doesn't seem to be a preprocessor macro to check for the employed
Opm version. Therefore, drop support for older versions and only
support 2016.04.

lecture. Unlike there, we use here a non-zero capillary pressure to use
rel-perms and capillary pressures from dumux stable

Everything in our .hh files is defined within the namespace `Dumux`.
There were numerous occurrences of `Dumux::` qualifiers that were
unnecessary. Remove those occurrences. Necessary qualifications
remain, such as `typedef Dumux::H2O<Scalar> H2O;`.

* Class can only be used for mpnc

this means
* the first vtk (normally the initial solution) is *-00000.vtu
* if all timesteps are writen the time step number and vtk number correspond

Reviewed by heck

Instead of calculating it in the local residual, the effective permeability
is now calculated within a new file permeabilityrutqvisttsang.hh in the
folder dumux/material/fluidmatrixinteractions.
This is just one possible effective permeability model. Several others can be
added and set via the property EffectivePermeabilityModel.

Up to now, the permeabilities were averaged twice, first by calculating
the mean of the mean of the intrinsic permeabilities in the flux variables
and later by averaging the effective permeabilities in the local residual.
This is redundant. Now, only the mean of the effective permeability is
calculated in the local residual.
This change required to change the reference files for the sequential and
the parallel test.

Move all testing related scripts to subfolder testing

The reference solutions have been changed because of the previous commit (change pointers to shared pointers)

Add a FieldType enum that controls the field type we generate,
e.g. log10, scalar. Could be extended by e.g. tensor in the future.

* Remove EntityPointer deprecated in dune2.4
* Use range-based for loops over gridview
* Use std::vector instead of blockvector

Add test cases where an intersection
* is on an edge
* goes through a corner
* one-d element is split by face

Add timer for intersection computation.

* Constructs bounding box tree
* Test intersection with points
* If dune-foamgrid available tests for intersection with a network
* Test all above tests for invariance w.r.t. grid coord scaling

Things that could be added in the future:
* Test with unstructured grids
* Test with different grid managers

The current implementation does not work with tensor permeabilities.
Not outputting the permeability as standard is suboptimal so we should
think about a better solution in the future.

* Remove permeability from the reference vtu files

We still need to patch Dune-multidomaingrid. Since we anyway need
to patch, the issue with GCC 6 can also be solved by the patch.

Thanks to Thomas for noticing.

The installexternal script is updated to fetch and build the correct Opm
modules.

No patch is required anymore for dune-multidomaingrid. In order to keep that
situation, add `#include <numeric>` to the problem files of the multidomain
tests. This is required if the tests are built with GCC 6.

Adapt README files.

As before, use UGGrid, at least until we have fixed things for
employing dune-alugrid master.

Use UGGrid, at least until the local vertex numbering when using
ALUGrid is fixed.

Sequential MPFA relies on the local vertex numberings of the individual
grid managers. This numberings have changed for the 2d-ALUGrid. Until this
is fixed, change the corresponding tests to use UGGrid instead. Alternatively,
master-old of dune-alugrid can be used.

Also revert changes to the input file that never were reflected in the
corresponding spatial parameters file.

*All ZeroEq test use Yasp now

Bernd Flemisch authored 9 years ago and Kilian Weishaupt committed 9 years ago

UGGrid is not needed here. Increases the consistency with the
reference results.

Kilian Weishaupt authored 9 years ago and Timo Koch committed 9 years ago

* Use input file for structured grids instead of dgf files for
  fluidmatrixinteractions tests

Kilian Weishaupt authored 9 years ago and Timo Koch committed 9 years ago

*remove #include <dune/grid/io/file/dgfparser/dgfyasp.hh>
*remove #include <dune/grid/io/file/dgfparser.hh>
*ect

Kilian Weishaupt authored 9 years ago and Timo Koch committed 9 years ago

* Use input file for structured grids instead of dgf files for
  zeroeq2c tests

Kilian Weishaupt authored 9 years ago and Timo Koch committed 9 years ago

* Use input file for structured grids instead of dgf files for
  zeroeq tests