Date of this Version
Today’s electric utilities are confronted with a myriad of challenges that include aging infrastructure, enhanced expectation of reliability, reduced cost, and coping effectively with uncertainties and changing regulation requirements. Utilities rely on Asset Management programs to manage inspections and maintenance activities in order to control equipment conditions. However, development of strategies to make sound decisions in order to effectively improve equipment and system reliability while meeting constraints such as a maintenance budget is a challenge.
The primary objective of this dissertation is to develop models and algorithms to study the impact of maintenance toward equipment/system reliability and economic cost, and to optimize maintenance schedules in a substation to improve the overall substation reliability while decreasing the cost.
Firstly, stochastic-based equipment-level reliability and economic models are developed depending on maintenance types. Semi-Markov processes are deployed to represent deteriorations, failures, inspection, maintenance and replacement states for reliability modeling; semi-Markov decision processes are implemented for economic cost evaluations considering capital investment, operations and maintenance cost, and outage cost.
Secondly, substation level reliability and economic cost models are established based on equipment level models. Sensitivity studies for analyzing the impact of equipment maintenance toward system level reliability and overall system cost are conducted.
Finally, maintenance optimization scenarios and solutions are developed, to determine optimal equipment maintenance rates that maximize substation reliability or minimize overall cost, while meeting operational and economic cost constraints, based on Particle Swarm Optimization techniques.
Moreover, fuzzy Markov and Markov decision processes are designed to calculate fuzzy reliability indices and economic cost; a parallel Monte-Carlo simulation method is also proposed to perform reliability evaluations through simulation method, in which the accuracy and computation speed are testified.
The algorithms developed in this dissertation are valuable for system reliability evaluation, maintenance planning, maintenance prioritizations, and maintenance policy. The programs developed can assist asset managers in making maintenance-related decisions, to effectively balance the system level reliability and associated maintenance cost.