A key requirement for de-risking deep geothermal projects is the ability to robustly and reliably forecasting the reservoir and seismicity evolution during stimulation and operation in near real time. This requires more sophisticated computational modelling and data assimilation approaches than available today. Building on the approaches developed by ETH, EOST and LBNL, and using existing datasets - and the ones collected in real-time in WP4 - we will here evaluate pseudo-prospective test for a next generation of forecasting, risk assessment and mitigation models.
Lead: ETH
Participants: EOST, LBNL, IEG, GES
Contact: Antonio Rinaldi and Luigi Passarelli
Task 3.1: An automatized testbench for induced seismicity forecast evaluation
Task 3.2: History matching with seismicity forecast and risk models
Task 3.3: Development of Machine Learning approaches for induced seismicity forecasting and
systematic comparison with other forecasting models
Task 3.4: Constraining models with laboratory measurements and investigations for the
characterisation of representative reservoir rocks
D 3.1: Extensive comparison of existing forecasting induced seismicity models with existing datasets
D 3.2: Development of Machine Learning based induced seismicity forecasting model
D 3.3: Release of new hybrid model for induced seismicity forecasting
A list of all deliverables for DEEP is available here.