A risk-based multi-criteria maintenance planning of distribution feeders

This paper proposes a risk-based multi-criteria approach for maintenance planning of distribution feeders that uses Outage Management System (OMS) historical data to obtain a minimum-risk maintenance plan. It considers simultaneously system-oriented and local-oriented reliability indices per interruption causes to ensure the necessary granularity of the analysis and the consistency (fairness) of reliability across the system. Uncertainty in reliability indices per interruption causes is described by intervals obtained using the bootstrap method and the historical OMS data. Using the interval extension of the TOPSIS technique and the interval-valued reliability indices per interruption causes, the ranking of interruption causes by the reliability indices is performed according to a decision maker's risk preferences and preferences regarding consistency (fairness) of reliability across the system. The ranking list is further assessed using the Pareto principle to obtain a list of interruption causes, i.e., pairs (feeder, interruption cause), responsible for most power interruptions. Then, maintenance actions are defined that reduce the influence of those causes on power interruptions. For a maintenance action, intervals of cost and maintenance effectiveness factors are defined to account for the uncertainty about cost and benefit estimation. Next, using the interval extension of the TOPSIS technique, maintenance actions, i.e., pairs (feeder, maintenance action), are ranked according to a decision maker's risk preference, reliability balancing preference, and the trade-off preference between benefit and cost. The ranking list is then iteratively used within the proposed procedure to obtain an initial minimum-risk maintenance plan used as the initial solution in the proposed multi-criteria simulated annealing (MCSA) algorithm. It generates a set of Pareto-optimal maintenance plans ranked using the interval extension of TOPSIS to obtain a minimum-risk maintenance plan according to the adopted decision maker's risk preference, reliability balancing preference, and the trade-off preference between benefit and cost. The effectiveness of the proposed approach is illustrated through case studies of the real-life power distribution company that supplies about 1 million customers through around 3000 medium-voltage feeders.