Laboratory Earthquakes Simulations—Typical Events, Fault Damage, and Gouge Production

We propose a numerical model of laboratory earthquake cycle inspired by a set of experiments performed on a triaxial apparatus on sawcut Carrara marble samples. The model couples two representations of rock matter: rock is essentially represented as an elastic continuum, except in the vicinity of the sliding interface, where a discrete representation is employed. This allows to simulate in a single framework the storage and release of strain energy in the bulk of the sample and in the loading system, the damage of rock due to sliding, and the progressive production of a granular gouge layer in the interface. After independent calibration, we find that the tribosystem spontaneously evolves toward a stick‐slip sliding regime, mimicking in a satisfactory way the behavior observed in the lab. The model offers insights on complex phenomena which are out of reach in experiments. This includes the variability in space and time of the fields of stress and effective friction along the fault, the progressive thickening of the damaged region of rock around the interface, and the build‐up of a granular layer of gouge accommodating shear. We present in detail several typical sliding events, we illustrate the fault heterogeneity, and we analyze quantitatively the damage rate in the numerical samples. Some limitations of the model are pointed out, as well as ideas of future improvements, and several research directions are proposed in order to further explore the large numerical data set produced by these simulations. Plain Language Summary Earthquakes are due to sudden sliding in faults several kilometers in the ground. A common laboratory practice is to reproduce such sliding events on dedicated lab devices. In this work, we present a novel numerical model aiming to reproduce such experiment in a computer simulation, in order to enhance our understanding of the phenomena at stake. This model is novel because it couples two different representations of the rock matter, namely a continuous and a discrete one. It therefore allows to reproduce in the same framework the bulk deformation of rock and the granular phenomena occurring at the sliding interface. The model is calibrated and leads to the spontaneous occurrence of unstable sliding, that is, of earthquakes of the same kind as those observed in the lab. We further explore into more detail some typical sliding events, and focus our attention of the interface damaging and wear during sliding. This work is likely to clar