DeepNL-2
A digital twin for modelling and forecasting induced seismicity
We propose a numerical model capable to predict observed seismicity (size, time and location of events) based on a known fault geometry and forcing. The model will be applied to Groningen where the forcing is due to gas production. Each induced earthquake produces ground acceleration felt by the population. We will use ground accelerations from our numerical model to develop a machine learning tool able to forecast timing and size of the next event. This can be used to assess seismic hazard due to gas production, tune production to safe levels and estimate seismicity after production stop.
Our project is part of a larger NWO programme. The aim of this DeepNL research programme is to improve the fundamental understanding of the dynamics of the deep subsurface under the influence of human interventions.
With DeepNL, NWO is providing a concrete response to the advice of the Dutch Safety Board: ensure that there is a structural and long-term research programme into the gas-extraction related problems in Groningen. This NWO research programme has been made possible in part by a financial contribution from NAM. DeepNL will be realised in accordance with the usual NWO procedures and quality standards and is entirely independent. NAM is not involved in the decision-making or the specific management of the programme. DeepNL is part of NWO’s contribution to the Top Sector Energy.