6.1.0

Overview of changes since 6.0.8 release.

The changes since the prerelease 6.1.0-rc1 contain few bug fixes and improvements of the CI.

Features:

New processes:

• ComponentTransport #1758
• PhaseField #1813 and #1895
• RichardsComponentTransport #1929
• ThermoMechanics #1800
• TwoPhaseFlow p-rho #1613

Other process' changes:

• New equation assembly approach for the staggered scheme. With that, the coupling assembly computations are changed from performing assembly across the different Process classes for a coupling to a single Process class for coupled processes, which was assumed only for the monolithic before. With the changes, the original Process class that for monolithic scheme originally can now handle both of the monolithic and staggered schemes. So far, HT classes get the staggered scheme based on this new framework of assembly. #1970
• Heterogeneous liquid flow properties (#1979, #2017)
• New boundary conditions added: Nonuniform Dirichlet (#1952) and Neumann (#1891), and normal traction (#1896)
• Framework for time stepping and a first application of adaptive time stepping, EvolutionaryPIDcontroller, and automatic time step control (#1803).
• Nodal source terms (#1977)
• Fix deformation processes to work correctly with PETSc parallelization (#1838).
• Fix the access to local data of PETScVector (#1797).
• Add damping factor to global Newton. #2004
• Extend extrapolator to vectorial quantities and replace the component-wise output of Darcy velocity and the stress/strain in mechanical processes with single vectorial output.

Material model changes

• Separate FractureModels in LIE #1971
• Add material forces as published in http://www.sciencedirect.com/science/article/pii/S0093641317303865 #1936

Testing and documentation:

• Insitu visualization with ParaView Catalyst. See presentation. #1744, #1732. As a consequence VTK 7.1 is now required.
• Benchmark docs are now part of the code (in web/content) and can contain
  interactive 3D visualizations via vtk.js. #1706, #1714, #1723, #1729.
• Migrated handling of test data files from git-submodule to git-lfs, see docs. #1964, #1982, #1984, #2010, #2012.
  Now git-lfs is required. Check the installation instructions.

Infrastructure:

• Fully moved to Conan for automatic third-party library handling. Can be enabled with OGS_USE_CONAN=ON, see docs. #1907
• Conan version 1.0 is now required.
• Dropped travis CI environment and added few new tests on Jenkins because of simpler maintenance.

CMake options changes:

• OGS_EIGEN_DYNAMIC_SHAPE_MATRICES defaults to OFF on Release
  config, ON otherwise. Can be overridden by explicitly setting the option. #1673
• New OGS_EIGEN_INITIALIZE_MATRICES_BY_NAN defaults to ON for easier spotting of non-initialized matrices. When OFF, the Eigen's default initialization to 0 is skipped resulting in slightly faster execution. #1917
• Set default Eigen's cmake flag disabling vectorization since this lead to several problems in different environments. #1919 and the issue linked there https://github.com/ufz/ogs/issues/1881

Other

• PETSc config is tested on Jenkins (envinf1)
• OGS binaries are provided as eve / envinf1 modules. See docs for details. #1753
• Migrated Data Explorer to Qt5. #1622, #1625
• Windows builds are tested on MSVC 2017 on own hardware and on MSVC 2015 on AppVeyor. #1802
• Minimum GCC version is 4.9 (tested on Travis and Docker builds), 6.2 is tested on UFZ cluster system (envinf1). #1810
• Removed Windows 32-bit builds from testing --> no Windows XP binaries are provided anymore. #1988
• Migrated Jenkins Pipeline scripts to new syntax called Declarative Pipeline. #1976
• Tests to check the gradient of shape function by comparing the given and the computed element volumes (#1873).