Dynamic resilience for power distribution and customers

Power distribution supplies energy directly to customers. However, power distribution is particularly vulnerable to severe weather, where large scale failures can interrupt electricity service to millions of customers for extended durations. Resilience metrics are needed for characterizing failures, recoveries and impacts to customers. While various resilience measures are studied previously including those from the IEEE standards, we find, using real data from Hurricane Sandy, that two such commonly-used metrics are inadequate for characterizing resilience. We develop a new metric that incorporates randomness and dynamics of failures and recoveries from severe weather events. The metric is formulated from bottom-up based on a model, i.e., a spatiotemporal non-stationary random process. Such a model integrates failures and recoveries, and importantly, impacts (costs) to customers. The resulting metric consists of a tuple as the infrastructure- and service-resilience. The dynamic metric thus characterizes resilience for the infrastructure (power distribution), services and customers. Large-scale real data from Hurricane Sandy is used to evaluate the resilience metric for the operational power grid and the cost on customer service interruption.