Risk Assessment for Concrete Buildings under Geothermal Induced Seismicity Using Gaussian Process Regression

  • Cebulj, Sonja (Technical University of Munich)
  • Taddei, Francesca (TU Wien)
  • Müller, Gerhard (Technical University of Munich)

Please login to view abstract download link

Geothermal energy offers considerable potential for a carbon-neutral, weather-independent energy supply. However, the associated risk of induced seismicity requires evaluation, given that geothermal power plants are typically established in proximity to populated areas ensuring energy provision close to consumers. An existing method of estimating the risk of induced seismicity is the adaptive traffic light system (ATLS), as presented by Mignan et al. [1]. These systems provide a seismic forecast and function as a decision-making tool for operating the geothermal power plant. In the context of the dynamic response of the buildings, the risk assessment can be expressed by the fragility and comfortability curves. This approach is employed in the study by Khansefid et al. [2], which focuses on masonry buildings subjected to induced seismicity. The objective of this study is to predict the dynamic response of concrete buildings to geothermal-induced seismicity and to incorporate this prediction within a risk assessment and decision-making framework like the ATLS. Therefore, it is essential that the run-times are minimal. Consequently, surrogate models are developed using Gaussian process regression. This facilitates a real-time calculation of building responses for a range of building parameters and induced seismic inputs. REFERENCES [1] A. Mignan, A. P. Rinaldi, F. Lanza, S. Wiemer, A multi-lasso model to forecast induced seismicity at enhanced geothermal systems, Geoenergy Sci. Eng., 236: 212746 , 2024. [2] A. Khansefid, S.M. Yadollahi, F. Taddei, G. Müller, Fragility and comfortability curves development and seismic risk assessment of a masonry building under earthquakes induced by geothermal power plants operation, Struct Safety, 103: 102343, 2023.