A risk-averse stochastic optimization model for community resilience planning

Community resilience planning is challenging as it involves several large-scale systems with interdependency, populations with diverse socio-economic characteristics, and numerous stakeholders. This study introduces a new optimization model to decrease a community's burden in developing viable alternative sets of decisions while considering costs and risks associated with uncertain hazard events. The model captures the essential features of a community, and its scope extends beyond infrastructure and buildings to include social goals. Structural engineering and social science approaches are adapted and incorporated into the model formulation to facilitate the identification of engineering decisions meeting the social goals of minimizing population dislocation and time for recovery. A risk-averse approach frames the optimization problem as a two-stage mean-risk stochastic programming model, which enables effective planning for low-probability, high-consequence hazard events. A case study simulating flood hazards in Lumberton, North Carolina, is developed, and the model is run with the generated data set to showcase the model's capability in developing risk-informed mitigation and recovery plans to achieve resilience goals. The insights drawn from the numerical experiments show the effect of changing risk preference on community resilience metrics.