Abstract

This thesis presents novel models and optimization strategies for dynamic resilient network flow problems under consecutive disruptions. The study addresses the need for resilient network design by focusing on the integration of stochastic elements, partial information, and flexible recovery
strategies within supply chains and other critical infrastructure networks. Three mathematical frameworks are developed, each targeting specific aspects of network resilience, including dynamic flow optimization, collaborative network design, and flexible recovery under uncertain conditions.
The first part of the thesis introduces a generalized structure for interconnected networks, enabling collaboration among different entities to enhance resilience. This model leverages both proactive and reactive strategies, using shared resources and interdependencies to mitigate the impact of disruptions more effectively. Computational experiments demonstrate that the proposed intertwined network structure significantly improves network stability and reduces disruption costs.
The second part focuses on dynamic network flow models that optimize rerouting strategies under consecutive disruptions with partial information. By employing two-stage stochastic programming and rolling-horizon optimization, the models dynamically adjust decisions based on real-time updates about disruption severity and timing. Results show that the dynamic models maintain network functionality more effectively than traditional static approaches, particularly in scenarios with evolving disruptions.
The final part extends the dynamic flow models to include overlapping disruptions and flexible recovery strategies. Different recovery options, characterized by varying costs and timeframes, are integrated
into the optimization framework, allowing decision-makers to select recovery actions based on disruption conditions. Simulation results reveal that flexible recovery strategies substantially reduce total disruption costs and recovery times, proving critical in managing complex cascading disruptions.
Overall, this thesis advances the field of network resilience by providing comprehensive models that address dynamic disruptions, optimize recovery decisions, and offer practical solutions for real-world applications in supply chains, logistics, and transportation networks.