A random lattice solid model for simulation of fault zone dynamics and fracture processes

Abstract

The particle-based Lattice Solid Model (LSM) was developed to provide a basis to study the physics of rocks and the non-linear dynamics of earthquakes (Mora and Place, 1993; Place and Mora, 1999). In previous work, intact material was modelled as particles bonded together by elastic-brittle links in a regular 2D triangular lattice and the model was applied to the study of fault zone evolution. In numerical experiments involving a transform fault model in which a weak and heterogeneous fault zone was sheared, a fault gouge layer formed with features similar to those observed in laboratory experiments with simulated fault gouge (Beeler et al., 1996). As in the laboratory experiments, gouge layer strength correlated with gouge layer thickness in the numerical experiments and the gouge layer effective friction initially oscillated around a value of about 0.6. However, after a large displacement, a re-organisation of the model fault gouge was observed with slip becoming highly localised in a narrow and anomalously weak basal shear zone. The long time required for the self-organisation process is a possible reason why the weak gouge layers observed in the numerical experiments, and which could explain the Heat Flow Paradox (Mora and Place, 1999), have not yet been observed in the laboratory. A new modular and flexible LSM approach has been developed that allows different micro-physics to be easily added and removed at the grain scale to enable the effect of different microphysics on macroscopic behaviour to be studied. The model is extended to allow three-dimensional simulations to be performed and particles of different sizes to be specified. We demonstrate the new approach by simulating bi-axial fracture experiments in a 2D model with random particle sizes. The new model provides a basis to investigate nucleation, rupture and slip pulse propagation in complex fault zones without the previous model limitation of a regular low-level surface geometry.