Search

To improve the viability of nuclear power plants, there is a need to reduce their operational costs. Operational costs account for a significant portion of a nuclear plant's yearly budget. This is due to their reactive and uninformed maintenance approach. In order to reduce these costs, proactive condition monitoring methods are required that can estimate the state of a machine in real-time and aid operators in optimizing maintenance schedules. In this research, we use Bayesian networks to develop a condition monitoring platform that can diagnose pump faults, infer their root cause, and estimate it's remaining-useful-life. This solution provides an informed probabilistic viewpoint of the pumping system for the purpose of preventative maintenance. In this thesis, we analyze a centrifugal pump to estimate its current state. We combine domain expertise with physical laws to create a cause-and-effect relationship between the pump components to estimate the root cause of two major pump faults. We incorporate survival analysis with Bayesian statistics to forecast the health of the pump conditional to its current state. We complete our condition monitoring analysis by successfully implementing the final Bayesian network in case studies for three modes of pump operation.