Deep lEarning foRecasting of Induced Seismicity for risK management operations

Descripción del proyecto

Una explotación mundial más segura de la energía geotérmica

Para combatir el cambio climático, es necesario producir energía limpia y sostenible. Los sistemas geotérmicos mejorados (EGS) pueden aprovechar el calor de la Tierra en muchas regiones. Sin embargo, existe un obstáculo importante a la tecnología EGS que podría dificultar el funcionamiento de dichas plantas: la sismicidad inducida. En los últimos años se han producido terremotos destructivos asociados a los proyectos de EGS, lo que ha provocado su cierre y suscitado inquietud en la sociedad. Con el apoyo del programa Marie Skłodowska-Curie Actions, el proyecto DERISK tiene como objetivo optimizar la vigilancia de la sismicidad inducida. Esta innovadora iniciativa combina tecnologías punteras de adquisición de datos con técnicas novedosas de aprendizaje profundo y explota todo el potencial del análisis de big data para reducir el peligro sísmico de los EGS. El objetivo es desarrollar un marco de análisis capaz de detectar, localizar y caracterizar la sismicidad inducida.

Objetivo

Climate change mitigation requires a fast and efficient transition to clean and sustainable energy production.Enhanced Geothermal Systems (EGS) play a key Climate change mitigation requires a fast and efficient transition to clean and sustainable energy production. In this context, Enhanced Geothermal Systems (EGS) can play a key role in facing this challenge since, with this technology, clean energy production from the Earth's heat is no longer confined to volcanic or hydrothermal regions. Despite this potential, EGS presents society and economy-related problems that need to be solved to ensure operational safety, continuity, and public acceptance of such industrial projects. Induced seismicity is the major obstacle to the development and social acceptance of EGS projects. In the last years, several damaging earthquakes have been associated with EGS, leading to the definitive closure of the involved projects and raising social concerns against this form of energy production. With DERISK we aim to develop new paradigms for induced seismicity analysis, combining the latest available data acquisition technologies, such as distributed acoustic sensing (DAS), with innovative deep-learning techniques. To characterize induced seismicity with unprecedented resolution and accuracy, we want to develop a next-generation data analysis framework combining deep-learning and waveform-based seismic imaging techniques. Our final goal is to produce deep-learning-based enhanced microseismicity catalogs that will be used to test the performance of new induced seismicity forecasting models exploiting the recent research advances in the field of physics-informed machine learning. The techniques developed within DERISK will be tested and validated with high-quality induced seismicity datasets collected at different EGS sites. If successful, DERISK will open the way to a safer and more widespread development of EGS projects, contributing to the transition to sustainable energy production.