We developed a hybrid model combining elements from statistics following a Gutenberg-Richter frequency magnitude scaling for fracture size and reservoir physics via Coulomb frictional stress variations to provide a better estimate of the induced seismic hazard from reservoir operations. Our model is a generalization of previous approaches and provide estimates that are in line with field observations without being limited to the actual physics at play. As such, it enables to extend predictive capabilities about expected maximum induced moment magnitudes by also providing essential constraints on novel operation scenarios (i.e. cylclic fatigue).
By additionally, combining the novel theory with Physics preserving Machine learning surrogate models, we are also able to perform extensive sensitivity analysis via ensemble modelling and uncertainty quantifications.
Cacace, M., Hofmann, H., Shapiro, S. (2021): Projecting seismicity induced by complex alterations of underground stresses with applications to geothermal systems. - Scientific Reports, 11, 23560.
Further information on the programme: