A numerical insight into rock sliding: Reproducing lab earthquakes with a coupled discrete-continuum model

Abstract:

Lab earthquakes have become a major tool for the understanding of seismic and aseismic sliding in natural faults. Outstanding progress has been made in the instrumentation of such experiments, allowing to get more and more local information on the phenomena at stake. For practical reasons, it remains however difficult for experimentalists to acquire the full mechanical and kinematic data on a segment of fault, at high sampling rates in space and time. This is where numerical simulations can help.

In this communication, we present a numerical model which aims at reproducing existing triaxial experiments on saw-cut marble samples. This model is in 2D and at a reduced scale when compared to its real-world counterpart, but contains a large part of the relevant physics. The two marble blocs are represented using a meshfree continuum approach, apart from the areas located in the neighborhood (within a few hundreds of micrometers) of their contacting surfaces. In these areas, polygonal grains bonded by a cohesive-zone model are implemented in a DEM framework. Realistic boundary conditions (in terms of the elasticity of the loading system, of the absorption of the elastic waves and of the fluid pressure applied on the lateral boundaries) are introduced. Constitutive and cohesive laws are calibrated based on experimental results found in the literature.

Upon loading, this model provides information on the system behavior that nicely complement the experimental data, such as (i) the progressive damaging of the contacting surfaces, leading to the emission of granular matter in the interface, to the formation of a gouge layer, and to a modification of the interface rheology, (ii) the space and time distribution and statistics and the detailed kinematics of the slip events related to the interface evolution, and (iii) the acoustic wave emission and propagation in the medium associated with such events.

The model shows that, depending on the experimental conditions (confining pressure, surface roughness, etc.), and without relying to any prior choice of slip- or rate-dependent friction laws, a large number of sliding regimes can emerge from this system. This includes large stress drops, regular stick-slip, or stable sliding. This model thus provides an unprecedented view of both local and global phenomena at stake during lab earthquakes.