The solvers based on automatic differentiation have received a general overhaul to documentation and examples to make it easier to get started with the corresponding modules. We have also created a new module, ad-eor, which contains extended versions of the polymer solvers previously found in ad-blackoil, as well as solvers for surfactant.
The functionality in the ad-fi module has been replaced and/or superseded by ad-core and ad-blackoil. The module will not be further maintained and will most likely not be included in future releases of MRST.
The ad-eor module has been added! It contains extended versions of the polymer solvers previously found in ad-blackoil as well as brand new solvers for problems with surfactant. The polymer solvers have also been extended with velocity dependent viscosity (shear thinning and thickening)
A new module has been added that implements two and three-phase black-oil solvers based on a sequential formulation. These solvers use separate models for pressure and transport and allows a great deal of flexibility for choices in nonlinear and linear solvers. The current implementation uses a fractional-flow formulation based on total velocity and can be significantly faster than the fully implicit solvers for larger problems.
The HFM module implements routines for creating problems with embedded fractures, as well as extensions to the msrsb module to enable the solution of embedded fracture models with the multiscale restriction-smoothing basis (MsRSB) solver. The module can also be used to generate transmissibility and topology for the AD-solvers, making it straightforward to solve problems with embedded fractures. The module was created by Swej Shah from the TU Delft DARSim group in collaboration with researchers from SINTEF. Swej Shah won the SPE Regional Student Paper Contest in Europe for this work. A paper documenting the work is published in J. Comput. Phys,DOI: 10.1016/j.jcp.2016.05.001.
Added a new module for the solution of linear elasticity and poro-elastic problems based on the virtual element method (VEM). This module is not entirely well documented and tested to the standard of other modules in MRST, but we nevertheless release it as a preview of future features after several user requests.
A new module for steady-state upscaling of saturation-dependent fluid properties. This includes relative permeability and polymer property upscaling. This module is the result of the work done by Sindre T. Hilden in his PhD work at NTNU/SINTEF, see https://brage.bibsys.no/xmlui/handle/11250/2388331
The module has been renamed to a more descriptive name, and several new examples and features have been added. This includes unstructured construction of support/interaction regions and C-accelerated routines for computing the MsRSB basis functions on arbitrary grids.
The 'deckformat' module gained initial support for constructing fundamental static and dynamic MRST objects from ECLIPSE output files. In particular, the new functions 'eclOut2mrst', 'convertRestartToStates', and 'convertSummaryToWellSols' construct, respectively, the grid and rock structures and associate transmissibilities on the ECLIPSE flow graph, a collection of dynamic restart data (e.g., temporally varying pressure and saturation fields), and a collection of well summary data for visualisation through module ad-core's 'plotWellSols' function. We refer to the new examples 'readAndConvertEclipseOutput', 'readAndVisualizeEclipseOutput', and 'diagnosticsFromEclipseOutput' for demonstrations of how to use these tools.
Tutorials have been moved in from the core module and reworked to present a better introduction to the various incompressible solvers in MRST.
Tutorial examples have been renamed and reworked, and new several new ones have been imported from the MRST User Guide by K.-A. Lie.
Examples have been reworked and renamed to give a better introduction to the features of the module. New examples have been imported from the MRST User Guide by K.-A. Lie.
The module has been reworked so that it now is compatible with the new object-oriented automatic differentiation library (ad-core, ad-blackoil, etc.). As part of this work, a large number of the examples and underlying computational routines have been tidied up and improved.
New functionality includes
Published December 14, 2015