Framework for optimizing survivability in complex systems

Abstract

Increasing high probability low frequency events such as extreme weather incidents in combination with aging infrastructure in the United States puts the nation's critical infrastructure such as hydroelectric dams' survivability at risk. Maximizing resiliency in complex systems can be viewed as a multi-objective optimization that includes system performance, survivability, economic and social factors. Systems requiring high survivability: a hydroelectric dam, typically require one or more redundant (standby) subsystems, which increases system cost. To optimize the tradeoffs between system survivability and cost, this research introduces an approach for obtaining the Pareto-optimal set of design candidates ("resilience frontier"). The method combines Monte Carlo (MC) sampling to estimate total survivability and a genetic algorithm (GA), referred to as the MCGA, to obtain the resilience frontier. The MCGA is applied to a hydroelectric dam to maximize overall system survivability. The MCGA is demonstrated through several numerical case studies. The results of the case studies indicate that the MCGA approach shows promise as a tool for evaluating survivability versus cost tradeoffs and also as a potential design tool for choosing system configuration and components to maximize overall system resiliency.