Assessing Induced Seismicity Models for Use in Deep Geothermal Energy Projects

Abstract:

The decision to phase out nuclear power in Switzerland by 2034 accelerated research on deep geothermal energy, which has the ability to contribute to long-term energy resources. Induced seismicty is a nessesary tool to create an enhanced geothermal system; however, potential seismic hazard poses a major challange to the widespread implementation of this technology. Monitoring and controlling induced seismicity with warning systems requires models that are updated as new data arrive and that are cast in probabilistic terms. Our main question is: is it possible to forecast the seismic response of the geothermal site during and after stimulation with models based on observed seismicity and hydraulic data? To answer the question, we explore the predictive performance of various stochastic and hybrid models. The goal is to find the most suitable model or model combination for forecasting induced microseismicity and unexpected events in geothermal reservoirs.

In this study, we consider the Basel 2006 dataset and generate forecasts of the number and spatial distribution of seismicity in the next six hours. We explore two models: (1) a hydro-geomechanical stochastic seed model based on pore pressure diffusion with irreversible permeability enhancement; and (2) four variants of a 3D “Shapiro” model which combine estimates of seismogenic index with a spatial forecast based on kernel-smoothed seismicity and temporal weighting. For both models, hydraulic and seismic parameters are calibrated against data from a learning period (starting at the beginning of injection) every six hours. We assess the models using metrics developed by the Collaboratory for the Study of Earthquake Predictability: we check the overall consistency of forecasts with the observations by comparing the number, magnitude and spatial distribution of forecast events with the observed induced earthquakes. We also compare the models with each other in terms of information gain, allowing pairwise ranking.