ParaGeo - Multi-Field Geomechanical Simulation Tool

 

 

ParaGeo is an adaptive coupled simulation tool that can accommodate  large topological changes via geometry evolution and adaptive remeshing strategies.  The key elements and assumptions in the design of ParaGeo are listed here. For a more detailed overview of the key features and ParaGeo applications see the pages within the current folder.

 

 

Massively parallel design

 

All operations have been designed to allow parallel computation on multi-node architecture (including re-generation of the mesh).

 

 

 

Solution Strategy

 

Specifically targeted at geomechanical applications with  single-field, coupled geomechanical/porous flow and full thermo-hydro-mechanical solution algorithms. Currently the porous flow field is limited to single-phase flow.

 

 

 

Geometry Driven

 

ParaGeo works directly from the geometry defined in terms of points, lines, surfaces and volumes rather than from a pre-defined finite element mesh.  This facilitates:

•Updating the geometry e.g. due to sedimentation of new material, erosion of existing material, or due to evolving geometrical changes; e.g. diapirism, pinchout of stratigraphy, etc.

•Definition of boundary conditions which is made on geometry entity basis rather than node-by-node basis.

 

 

 

Integration with Open Source Libraries and Software

 

ParaGeo uses several open libraries including:

•PETSc (Portable, Extensible Toolkit for Scientific computation).  This is a large suite of parallel linear, nonlinear equation solvers with interfaces for C, C++, Fortran and Python.  PETSc supports parallel matrix and vector assembly routines that provides efficient overlap of communication and computation.

•MUMPS (MUltifrontal Massively Parallel sparse direct Solver) which is a distributed multifrontal solver (F90, MPI based) which uses dynamic distributed scheduling to accommodate both numerical fill-in and multi-user environment

•MPI (Message Passing Interface) Mpich2 standard functions

•HDF5 (Hierarchical Data Format) library which HDF  is a library of functions for creation/manipulation of very large datasets.  HDF files are self-describing (Similar to XML documents) and support complex data relationships and dependencies.

•TetGen A Quality Tetrahedral Mesh Generator and Three-Dimensional Delaunay Triangulator