Domain-informed Wear and Fatigue Modeling for Hydropower Plants

Working in collaboration with Argonne National Lab, other research institutions, and industry partners, the goal of this research project is to use data from operating hydroelectric plants to more accurately predict when parts will fail, allowing plant operators to minimize downtime from unexpected repairs and maintenance while getting the most use out of each component. By analyzing operational data from the Hydropower Research Institute (HRI) database, our team at Idaho National Lab is working to develop mapping between failure-and-degradation models and legacy hydropower models. My contribution to the project was to outline a method for selecting data from the HRI database and create a correlation matrix for sensors in different systems across
the unit, allowing us to analyze how each component responds a particular type of failure event and identify trends among different systems