Disaster Preparedness and Response through Optimization
By Violet Chen
Monday started with a diverse line up of talks, including the session “Disaster Preparedness and Response through Optimization,” chaired by Dr. Joseph Szmerekovsky, Professor of Transportation and Logistics at North Dakota State University. Dr. Shaolong Hu from Texas State University opened up this session with his joint work with Dr. Zhijie Dong on “A Multi-stage Stochastic Programming Model for Relief Distribution Considering the State of Road Network.” Their approach captures various factors which have potential impacts on relief distribution. Examples are multiple types of vehicles, fluctuation of rental, and the state of road network. Specifically, the road network state is characterized using uncertain and dynamic road capacity, and they use a scenario tree to represent possible states. They propose a progressive hedging algorithm (PHA) to solve the model in large scale, and provide insights from numerical analysis on data from the Ya’an earthquake in China. One managerial insight is relief distribution operations are more efficient when there are more initial supplies reserved for areas prone to earthquake impacts.
Tugce Canbilen, PhD student in Industrial Engineering from Middle East Technical University, presented “Management of Interdependent Infrastructure Networks under Disaster-related Uncertainties.” In this joint project with Dr. Sakine Batun and Dr. Melih Celik, they propose a two-stage stochastic programming approach for deciding restoration of interdependent critical infrastructure networks subject to disaster-related uncertainties. Their model aims to recover infrastructure’s service as quick as possible. To assist decision making, the solution gives which subset of network arcs should be reinforced before the disaster, and which subset of arcs should be repaired after the disaster. Uncertainties are included by defining scenarios with different probabilities of arcs breaking during a disaster. Through computational experiments on the Sioux Falls network, they observe solution structure, the most important arcs and most inaccessible nodes in the network.
The third talk was given by Bahar Shahverdi, PhD student in Civil, Environmental, and Infrastructure Engineering from George Mason University. She talked about “Post-disaster Inspection and Restoration Prioritization in Roadway Networks,” a joint project with Dr. Elise Miller-Hooks and Dr. Hsi-Hsien Wei. A main goal of this research is to fill a gap in existing literature about post-disaster operations, that is, how to jointly schedule restoration and inspection activities for traffic and other lifelines post disaster. They adopt a stochastic programming framework to include different types of uncertainties, including variations in repair time, inspection time, and level of damage, into the formulation. Using scenario trees, they study both the exogenous case where time of new information arrival is known, and the endogenous case where timing of new information realization is unknown. Their ongoing work looks for solution techniques to apply their models to real-world questions with large scale.
In the fourth talk “Managing Humanitarian Logistics through Public-Private Partnership: A Collaborative Decision-making Approach to the Pre-positioning of Hurricane Relief Kits,” Dr. Yue Ge from the University of Central Florida shared a somewhat different perspective on disaster relief management. In this joint project with Sara Iman and Dr. Joseph Szmerekovsky, they apply operations management techniques to address the social science question of utilizing public-private partnership for emergency management. Previous social science research has established that some private firms have the potential to build expertise in addressing disasters, yet it is not well understood how to take advantage of public-private partnership. Through conducting web surveys and personal interviews of different stakeholders, they aim to learn knowledge to guide the design of better modeling and simulation approaches.
Last but not least, Vahid Eghbal Akhlaghi from the University of Iowa presented “Fuel Distribution Planning for Islands: Models and Case Study for Puerto Rico.” Vahid is a PhD student in Management Sciences, and he worked with Dr. Ann Melissa Campbell and Dr. Renato E. de Matta on this research. They focus on disaster recovery on islands, an important but often neglected topic. In this study, they propose an optimization approach for the fuel storage and distribution problem on islands under possible disaster scenarios. Their mixed integer program-based model consists of two tiers: the first tier ensures fairness in allocating fuel to affected areas with an equity-seeking objective, the second tier optimizes an efficiency objective among equitable candidate allocation schemes. Their case study on Puerto Rico recovers insights about fuel reserve prepositioning and fuel distribution.