this is because the loadBalance() method of Dune::CpGrid is seriously
broken at the moment: it throws away elements even in sequential mode
for some more relevant grids. ("more relevant" == "Norne". Norne is is
buggy itself because it features two completely disjoint parts.)
but it is even worse than that: if the bug mentioned above bites,
Dune::CpGrid cannot even agree for itself how many elements it has
which causes the CartesianIndexMapper to segfault.
at least, they compile as far as eWoms is concerned. Some external
libraries (in particular everything which uses SuperLU) still have
issues.
Also, there seem to be issues with the precision that is achievable
by the Newton method when using float.
an interesting side effect is that the Newton convergence output is
only defined if the VTK output is enabled.
Also, this patch could be implemented more efficiently by retrieving
the EnableVtkOutput only once per output module (instead of once per
element).
i.e., there is now a base class for the EclGridManagers and one class
for each type of grid (Dune::ALUGrid, Dune::PolyhedralGrid and
Dune::CpGrid). Selecting the concrete grid type is now done by
deriving the EclProblem's type tag from the type tag of the respective
grid manager.
that's because its applicability for non-cornerpoint grids is _very_
limited. Also the class is renamed to 'AluCartesianIndexMapper'
(because its purpose is to be used in conjunction with dune-alugrid)
and the namespace is changed from 'Dune' to 'Ewoms'.
operator>=() is available for Opm::Evaluation, and its efficiency is
the same as explicitly comparing the value. Let's thus remove this to
reduce optical noise.
i.e., calculate the pressure which the exterior cell of a face would
exhibit at the depth of the interior one (instead of bringing both to
the depth of the face).
because the average fluid density is used, there should not be a
difference because of this.
for large problems with many wells the performance impact of this is
probably even measureable.
Also, this patch makes it possible to access the well manager outside
of the problem. In the normal case, this should be rarely needed, but
it can come in handy for debugging purposes.
the EQUIL init code from opm-core chokes hard if any Dune grid other
than Dune::CpGrid is passed to it. Until opm-core gets its act
together or EQUIL init is properly implemented within eWoms, let's
just disable EQUIL initialization in ebos (and print a warning) if
ALUGrid is used.
before, it could have been written multiple times if the first time
step of the simulation failed and the timestep was repeated with a
smaller step size.
re-ordering the attributes avoids some padding for EclPeacemanWell
objects and -- more importantly -- makes their ordering slightly more
logical.
initializing them avoids a valgrind complaint when writing the initial
condition to the ECL summary file.
obviously if the switching variable is interpreted as x_g^O, the gas
phase is present, because in this case it is the only phase. Also,
when the oil phase appears, the gas saturation is 1-Sw, not 1. (the
last issue only happened in the vaporized case because the switch
variable would never get the meaning of "oil component's mole fraction
in the gas phase".)
