5.12 introduced lots of new options and dependencies, could
not get it to work...

Docker in Docker is not possible with our current images.
Reverted back to previous setup.

Needed by the createIsoparametricFiniteElement() function.

Things to be built on master only are now executed in a stages
post() part. This allows for more parallelism of the sub-jobs.

More changes:

- Removed setting of option OGS_CPU_ARCHITECTURE on eve frontend
  builds
- Failed header compilation in stage 'Check headers' marks the
  build as unstable
- Tests-Large now runs on singularity1 or envinf1

Random fixes

The classes aren't used anywhere in the code aside the self-tests.

The length of the passed initializer list is expected
to be short, well below of 100 entries or so.

HM mass formulation

after HM implementation tolerance for pressure
was sligthly too tight in the test hm2_1Dcolumn1
for testing on Frontend2 (serial)

LIE/HM fracture permeability model: ConstantPermeability

HM mass formulation caused in this test
(square_1e2_unconfined_compression_early)
minimal differences in the pressure solution

due to mass formulation of HM the hydraulic flow
is now rho_fr times larger than before

in mass formulation p_neumann has to be
rho_fr times larger than before
rho_fr = 1000
1e-5 * rho_fr = 1e-2

influence of g was neglected with zero density,
now density (solid and fluid) is 1, gravity is 0,
result is the same

    added Kpu, multiplied rho_fr where necessary

 Added nonequilibrium_stress to ThermoMechanics process

# Conflicts:
#	ProcessLib/LIE/HydroMechanics/CreateHydroMechanicsProcess.cpp

Zhang Ning authored 5 years ago and Dmitri Naumov committed 5 years ago

Tested with salt cavern model.

Florian Zill authored 5 years ago and Tom Fischer committed 5 years ago

error_dx and global_component are swapped

LIE/HM fracture permeability models; CubicLaw

Remove specific file name in \file macro to avoid matching
it against other/new Permeability.h.