Robust Emergency Relief Supply Planning for Foreseen Disasters Under Evacuation-Side Uncertainty

For foreseen natural disasters (e.g., hurricanes or floods), the uncertainties faced in relief logistics primarily stem from evacuation activities. We present a strategic planning problem to supply relief items by considering uncertainties in disaster location, intensity, duration, and evacuee compliance. To ensure time- and cost-effectiveness in relief distribution, we develop a robust optimization model to determine centralized supply locations, and supply quantities for different transportation modes in a five-tier network. In doing so, we consider the interaction between evacuation and supply-side activities and capture the inherent uncertainties using a combination of event and box uncertainty representations. Our model provides a decision maker with the flexibility of including or excluding the time dependency of evacuation-related uncertainties. Accordingly, it suggests a threshold time window for relief distribution, beyond which either the system cost increases or the benefits of early distribution diminish. Although the model primarily aids a policymaker in strategic preparedness, its tactical variant can aid the efficient distribution. We devise an enhanced Benders decomposition-based efficient solution method to solve realistic-size problems. In a case study using geographic information system data, we highlight the complex dynamics among various system components and discuss the resulting time-cost trade-offs that also influence the network structure.