With increasing concerns about non-renewable resources, geothermal energy is becoming increasingly important for heat and power generation. Germany has set a goal to obtain a significant share of its energy from renewable sources by 2050. Deep geothermal plants are expected to make a significant contribution to meeting heat demand. However, these plants cause micro-seismic events that affect the serviceability of buildings and human comfort. Until now, the measurement of seismic activity in the Bavaria region relied on the free-field vibration sensor network. In a novel approach, additional sensors were recently installed on different levels of a low-rise building on the site of a geothermal plant in the south of Munich. This strategic sensor network aims to collect and analyze unique vibration data. The project focuses on improving the understanding and prediction of building response to micro-earthquakes through data-driven model updating.
The first phase will involve processing and preparing existing sensor data using classical and modern machine learning approaches, such as neural networks, for automatic data processing and feature extraction. The second phase will involve developing surrogate models to quantify uncertainty and replace detailed nonlinear analyses. These models will use experimental data to improve the prediction accuracy of structural behavior under induced seismicity. By improving existing building models using experimental data, this project aims to develop a warning system and a shaking prediction model to mitigate the effects of induced seismicity in geothermal plants. By monitoring building integrity under induced seismicity, the project will contribute to the safety and efficiency of geothermal operations in Munich and potentially worldwide.
Reference: MSDI page
Person in charge: Wei-Teng, Kao ([email protected])
Supervisor: Dr.-Ing. Francesca Taddei
Please check Task description for following information.