Seminar by
Melih Çelik
Georgia Institute of Technology.
Desautels Faculty of Management
McGill University
One of the devastating consequences of natural and man-made disasters is the resulting debris, which can be catastrophic for the economy, environment, and the health of affected communities. Despite its importance, there is little existing work on quantitative decision support for debris management. In this talk, we discuss the post-disaster debris clearance problem, in which the objective is to establish connectivity of supply and demand for relief commodities in a timely manner by determining a sequence for the clearance of debris-blocked roads under limited resource capacity.
In many real-life applications, accurate estimation of debris amounts is not possible due to lack of information or estimation methods. As clearance proceeds, the estimations are updated based on actual observations of the debris amounts, resulting in a trade-off between learning the network structure and greedily aiming to satisfy the demand. To solve this problem optimally, we develop a partially observable Markov decision process model. Due to intensive computational requirements of this approach, we also propose novel heuristics that enhance the efficiency of the search tree. Through computational experiments, we compare the proposed approaches to those that ignore the uncertainty in the debris amounts as well as those that mimic current practice. This research is joint work with Özlem Ergun and Pınar Keskinocak from Georgia Institute of Technology. Biographical Sketch
Melih Çelik is a Ph.D. candidate in the H. Milton Stewart School of Industrial and Systems Engineering and works at the Center for Health and Humanitarian Logistics at Georgia Institute of Technology. He received his B.S. and M.S. degrees in Industrial Engineering from Middle East Technical University in 2007 and 2009, respectively. He also holds an M.S. degree in Operations Research from Georgia Institute of Technology, which he received in 2012. His research interests include stochastic and online network design, humanitarian applications of operations research, warehouse logistics, and facility location and layout.
by
Melih Celik
