Risk Based Maintenance Optimization using Probabilistic Maintenance Quantification Models of Circuit Breaker
MetadataShow full item record
New maintenance techniques for circuit breakers are studied in this dissertation by proposing a probabilistic maintenance model and a new methodology to assess circuit breaker condition utilizing its control circuit data. A risk-based decision approach is proposed at system level making use of the proposed new methodology, for optimizing the maintenance schedules and allocation of resources. This dissertation is focused on developing optimal maintenance strategies for circuit breakers, both at component and system level. A probabilistic maintenance model is proposed using similar approach recently introduced for power transformers. Probabilistic models give better insight into the interplay among monitoring techniques, failure modes and maintenance techniques of the component. The model is based on the concept of representing the component life time by several deterioration stages. Inspection and maintenance is introduced at each stage and model parameters are defined. A sensitivity analysis is carried to understand the importance of model parameters in obtaining optimal maintenance strategies. The analysis covers the effect of inspection rate calculated for each stage and its impact on failure probability, inspection cost, maintenance cost and failure cost. This maintenance model is best suited for long-term maintenance planning. All simulations are carried in MATLAB and how the analysis results may be used to achieve optimal maintenance schedules is discussed. A new methodology is proposed to convert data from the control circuit of a breaker into condition of the breaker by defining several performance indices for breaker assemblies. Control circuit signal timings are extracted and a probability distribution is fitted to each timing parameter. Performance indices for various assemblies such as, trip coil, close coil, auxiliary contacts etc. are defined based on the probability distributions. These indices are updated using Bayesian approach as the new data arrives. This process can be made practical by approximating the Bayesian approach calculating the indices on-line. The quantification of maintenance is achieved by computing the indices after a maintenance action and comparing with those of previously estimated ones. A risk-based decision approach to maintenance planning is proposed based on the new methodology developed for maintenance quantification. A list of events is identified for the test system under consideration, and event probability, event consequence, and hence the risk associated with each event is computed. Optimal maintenance decisions are taken based on the computed risk levels for each event. Two case studies are presented to evaluate the performance of the proposed new methodology for maintenance quantification. The risk-based decision approach is tested on IEEE Reliability Test System. All simulations are carried in MATLAB and the discussions of results are provided.