Summer Scholarship Programme Project Proposal

Fractures occur in an elastic material when the local strength is exceeded by the stress. During deformation fractures develop in stages: by nucleation, growth and coalescence. The combination of spatial heterogeneity in material strength, the complexity of the resulting stress field, and the non-linear physics of fracturing, makes the prediction of their evolution very difficult using analytical or vector computational methods.

Recently, Geology and Geophysics have developed a domain decomposition technique to model the evolution of fractures in the form of tensile joints that form near the earth's surface. Joints provide pathways for fluid flow and contaminant transport in the subsurface, so their properties are also of practical interest. The code hass already been parallelised using a message passing interface and the AZTEC parallel library but the fracture part still hast to be added.

The new work will include adding failed elements (fratctures, as in an existing serial code) to the parallel code - the remaining elastic strain field is then calculated automatically - aslo modifying boundary and loading conditions to represent joint formation more realistically. This involves using distributed internal stress relaxation rather than simply external stresses applied at a boundary, and including physico- chemical weakening processses that degrade rock strength at crack tips by an amount which depends on reaction time and local stress.

The results will show how realistic fracture patterns might develop in space and time in natural conditions. Several runs will show the statistical properties of the resulting geometry (alignment, degree of clustering, size distribution etc.) and the underlying statistical mechanics (strain energy distribution). If time remains it might be worhtwhile trying to visualise/animate these results using AVS.

The original author of the code, Javier Sabadell, no longer works in G&G but he would be available for consultation should the need arise which should facilitate the code development.

week 1: Courses.
week 2: Look at the code.
week 3: Begin Modifying the code.
week 4:
week 5: Modifications should be completed by this stage.
week 6: Begin performing simulations and attempt to visualise results.
week 7: If we get this far further tasks would be gained from G&G.
week 8:
week 9:
week 10: Report Writing.

Knowledge of Fortran and MPI.

MPI (at least)

Access time on lomond.

Webpage maintained by www@epcc.ed.ac.uk