Optimal reliability centered vegetation maintenance scheduling in electric power distribution systems

One maintenance activity performed by many electric utilities is vegetation maintenance. Vegetation maintenance involves trimming or removing trees near the rights-of-way to reduce the number of tree related outages and maintain system reliability. Since vegetation maintenance programs are very expensive, it is beneficial for a utility and its customers to find low cost yet highly reliable maintenance schedules.; This dissertation devises a new reliability centered vegetation maintenance scheduling algorithm for electric power distribution systems. To implement this algorithm, a reliability model is developed that quantifies the impact of vegetation maintenance on system reliability. Incorporated into the maintenance optimization algorithm is a new visual inspection scheduling algorithm. Inspection allows the results given by the reliability model to be updated, yielding potentially more accurate vegetation maintenance schedules.; To model the time-varying vegetation failure rate of overhead lines, a modified artificial neural network model is developed. This model uses climate data and tree density to predict the overhead line failure rates as a function of time since the bug maintenance activity was performed.; With a time-varying model for the vegetation failure rate created, a combinatorial optimization algorithm is formulated that determines the vegetation maintenance schedule for each circuit in the system. The algorithm considers different objective functions and constraints including crew availability, cost, and reliability. Three solution techniques are reviewed including genetic algorithms, simulated annealing, and tabu search. A hill climbing technique is tested as a post-processing algorithm. Genetic algorithms combined with a hill climbing technique find the highest quality maintenance schedules.; A new Markov inspection model is then developed that establishes the optimal feeder inspection frequency based on the results from the maintenance scheduling algorithm. By performing inspection, the actual state of a feeder can be determined and a corresponding condition rating assigned. The condition rating is incorporated into the failure rate model and new vegetation failure rates are predicted. The new failure rates are given to the maintenance scheduling algorithm, allowing potentially more accurate vegetation maintenance schedules to be devised.