| Title | Detection and diagnosis of abnormal conditions in the feed pumps of a geothermal binary power plant using feature-based time-series analytics |
|---|---|
| Authors | P.M.B. Abrasaldo, S.J. Zarrouk, A.W. Kempa-Liehr, A. Mudie, J. Cen, C. Siega |
| Year | 2023 |
| Conference | New Zealand Geothermal Workshop |
| Keywords | Geothermal binary plant, artificial intelligence, machine learning, feature engineering, time-series analysis, fault detection, condition monitoring |
| Abstract | Industries worldwide have seen innovative growth fueled by volumes of data generated by inexpensive sensors deployed across critical points of various systems. These large datasets have led to artificial intelligence and machine learning applications in designing and operating energy systems. Furthermore, analytics-driven decision-making in industrial operations is a field of research that has benefitted from the confluence of developments in data storage, capture, and processing technologies. This study investigates the application of time-series analytics in detecting and diagnosing abnormal operating conditions in circulating feed pumps operating in a geothermal binary power plant. Pump power and speed data have shown that the pumps experience episodes of cavitation that have led to reduced performance and reliability in the past. Operators have documented five cavitation events within one year, but operators believe there may have been more events that were not recorded. Systematic time-series feature engineering, supervised, and other machine learning methodologies were applied to detect and diagnose the drivers of the pump cavitation events. A fully-labelled dataset was used to develop predictive models to forecast the occurrence of the target events and diagnose the primary drivers that may have triggered such events in the past. The workflow deployed in this study can be used as a foundation for developing fit-for-purpose tools that can automatically detect abnormal operating conditions for different components and systems across other geothermal operations. |