Modeling the recovery process: A key dimension of resilience

The recovery process is a key determinant of system resilience because it describes the capability of a system to restore its performance after a disruption. In this work, we construct recovery functions that satisfy the necessary conditions to model the performance of critical infrastructure systems over time, during periods of loss and restoration following a disruptive event. To characterize the responses by various systems, we estimate the parameters of those functions by using empirical data about major infrastructure disruptions, including their recovery processes. The validity of our procedure is then illustrated by applying data from the three major utility companies that served New Jersey in the aftermath of Hurricane Sandy. Models are selected via standard methods to compare those recovery processes. Our results indicate that the recovery functions proposed here can capture the sensitivity of a system response to key parameters, thereby supporting the design of a more resilient system.