Stress-Based and Convolutional Forecasting of Injection-Induced Seismicity: Application to the Helsinki Geothermal Reservoir Stimulation

Jean-Philippe Avouac, & Taeho Kim

Submitted September 11, 2022, SCEC Contribution #12346, 2022 SCEC Annual Meeting Poster #206

Induced seismicity observed during Enhanced Geothermal Stimulation (EGS) near Helsinki, Finland is modelled. A physical model is built upon stress evolution due to pore pressure diffusion and earthquake nucleation assumed either instantaneous or governed by rate-and-state friction. Assuming instantaneous failure and a uniform failure strength yield a poor fit. Including rate-and-state effects significantly improves the fit due to the nonlinear dependence on the stress history and the tendency of the direct effect to act as a threshold triggering pressure. The model suggests that the Omori law decay during injection shut-ins results mainly from stress relaxation by pore pressure diffusion. Our study shows that the hydraulic diffusivity of the medium inferred from the migration of the seismicity front is biased without accounting for the effect of the direct effec on earthquake nucleation. We suggest a heuristic method to account for this rate-and-state effect that is independent of the earthquake magnitude detection threshold. A statistical model performs convolution of the injection history with a kernel based on the Omori law. The convolution model is extended to space by learning the spatial distribution of seismicity from the best fitting physical models. The statistical method is computationally efficient and could be used to design traffic light systems or control\& optimization schemes. Whether the convolution model is valid is found to depend strongly on the degree of the Kaiser-effect. Both the physical and statistical models indicate that the Kaiser-effect was not strong in Otaniemi, likely due to the variation of injection locations between stimulation stages.

Key Words
Induced Seismicity, Geothermal Well Stimulation, Seismicity Rate Forecasting

Citation
Avouac, J., & Kim, T. (2022, 09). Stress-Based and Convolutional Forecasting of Injection-Induced Seismicity: Application to the Helsinki Geothermal Reservoir Stimulation. Poster Presentation at 2022 SCEC Annual Meeting.

Related Projects & Working Groups
Earthquake Forecasting and Predictability (EFP)