Rock-deformation laboratory experiments and numerical modelling provide essential complementary physical constraints (e.g. friction rupture laws) to our understanding of earthquake mechanics and thermo-hydro-mechanical-chemical coupling. They allow for a systematic investigation of representative rock samples under controlled environmental conditions to investigate the sensitivity of fault stability, poro-elastic and frictional behaviour, and enable upscaling and transfer to other tectonic regimes.
A key feature of FEAR is the opportunity to test scaling relationships and robustness of the results on multiple rock-deformation apparatuses, in order to simulate in the laboratory the larger-scale experiments conducted in the BedrettoLab. The novelty of the proposed approach is to control the on-fault state of stress to understand the role of barriers, pre-stress, loading rates and fluid injection on the generation, propagation and temporal occurrence of frictional instabilities in a controlled environment. In particular, finding a relation between the on-fault state of stress and the resulting deformation prior to and after the onset of frictional instabilities constitutes a breakthrough research goal to validate across scales.
Experimental simulations will also contribute to the calibration and interpretation of precursory signals recorded during WP1. Experiments will be performed on three rock-deformation apparatuses (SHIVA at INGV, LabQuake(X) at ETH, and HIGHSTEPS co-owned by EPFL and ETH), simulating as closely as possible the in-situ reservoir conditions of the BedrettoLab. The three apparatuses are constructed with different experimental setups. Their combined use will enable a realistic integration and verification of the in-situ BedrettoLab experiments. FEAR will effectively build the first integrated network of experimental rock deformation machines in Europe. Experiments carried out on different rock types will also enable the application of the results obtained in the BedrettoLab to other geological conditions.
Numerical modelling is used as a powerful tool for up- and downscaling, process understanding, real-time forecasting and hypothesis testing and, finally, real-time risk mitigation. Detailed modelling and validation of the behaviour of even a single well-characterized fault segment remains a cutting-edge challenge for current numerical methods. Improving the capability to solve the strongly coupled, non-linear and time-varying thermo-hydro-mechanical-chemical processes on realistic geometric representations of a fault network is an important FEAR objective.
The modelling component of FEAR will integrate and evolve current numerical models on coupling processes, such as contact/friction/slip on rough fracture surfaces, poro- and thermo-elasticity, elastic rock deformation, fracture initiation and propagation, and fluid pressure evolution. Model development, calibration and validation will be closely integrated with laboratory and in-situ experiments, and will be performed with experts and hardware provided by the Swiss National Supercomputing Center (CSCS), giving us access to Europe’s currently fastest High Performance Computer.