Physical and Social Systems Resilience Assessment and Optimization

Resilience has been measured using qualitative and quantitative metrics in engineering,economics, psychology, business, ecology, among others. This dissertation proposes a resilience metric that explicitly incorporates the intensity of the disruptive event to provide a more accurate estimation of system resilience. A comparative analysis between the proposed metric and average performance resilience metrics for linear and nonlinear loss and recovery functions suggests that the new metric enables a more objective assessment of resilience for disruptions with different intensities. Moreover, the proposed metric is independent of a control time parameter. This provides a more consistent resilience estimation for a given system and when comparing different systems. The metric is evaluated in the study of community resilience during a pandemic influenza outbreak and the analysis of supply chain resilience. As a result, the model quantifies constant, increasing and decreasing resilience, enables a better understanding of system response capabilities in contrast with traditional average performance resilience metrics that always capture decreasing resilience levels when the disruptive events magnitude increases. In addition, resilience drivers are identified to enhance resilience against disruptive events. Once resilience drivers have been found, then a multi-objective resource allocation model is proposed to improve resilience levels. Previous resilience optimization models have been developed mainly based on a single resilience metric. The existing bi-objective models typically maximize resilience while the recovery cost is minimized. Although the single metric approach improves system resilience some of their limitations are that the solution is highly dependent on the selected resilience index and generally few optimal points are found. To overcome the rigidity of a unique metric a bi-objective model is proposed to maximize two key resilience dimensions, the absorptive and restorative capacities. This approach has the potential to offer multiple non-dominated solutions increasing decision makers alternatives where the single metric solutions are included