Lead University: Carnegie Mellon University
PI: Mario Berges, Civil and Environmental Engineering
Co-PIs: Barnabas Poczos, Machine Learning
Embankment dams, like most other civil infrastructure systems, are exposed to unpredictable environments. However, their design specifications and as-is properties are not generally known due to among other things, their age and the difficulties associated with assessing their internal structure. Hence, evaluating measurements collected from instruments used to monitor their behavior requires sound engineering judgment and analysis, as well as robust statistical analysis techniques to prevent misinterpretation. In the U.S., the current practice of analyzing the structural integrity of embankment dams relies primarily on manual a posteriori analysis of instrument data by engineers, leaving much room for improvement through the application of automated data analysis techniques. Thus, we propose to test if applications of robust statistical anomaly detection techniques to piezometer data from embankment dams detect anomalies that are indicative of internal erosion, which is the most common failure mode in embankment dams, more accurately and earlier than qualitative examinations performed by engineers. In this research, we plan to investigate different categories of anomaly detection techniques that have been widely validated in various domains. In addition, we will simulate different degrees of anomalous severities using a physics-based engine for seepage flow in order to closely replicate more realistic anomalous scenarios.
Through collaboration with the US Army Corps of Engineers (USACE) we will incorporate expert feedback and real-life dense instrumentation data into the approach. The USACE has been putting tremendous efforts to ensure dam safety. In addition, Geosyntec, a specialized engineering firm that has broad experience in solving engineering problems including safety and risk evaluation of geotechnical infrastructures, will also support us by providing critical reviews on our results. By leveraging the USACE engineers’ past experience and the support from Geosyntec engineers, there is opportunity to improve rational data analysis tools for the existing instruments, thus benefit the dam safety domain.