Date and Time: November 15, 2022, 3pm EST

Teams link: Join conversation (microsoft.com)

Advisors:

Dr. Pinar Keskinocak, Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Benoit Montreuil, Industrial and Systems Engineering, Georgia Institute of Technology

Thesis Committee: 

Dr. Mathieu Dahan, Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Alan Erera, Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Ozlem Ergun, Mechanical and Industrial Engineering, Northeastern University

Dr. Chelsea (Chip) White, Industrial and Systems Engineering, Georgia Institute of Technology

Abstract:

This dissertation uses optimization techniques to tackle and solve real-world problems, focusing on time sensitive restoration and operations of supply chain networks.

In network restoration, we consider The Network Connectivity Restoration Problem and aim to re-establish physical connectivity between nodes in the network by restoring the edges. In some cases, the progress of restoration activities is endogenous, i.e., depending not only on the structure of the network but also on previous decisions and actions. We model this problem as a Mixed Integer Program (MIP) and present a portfolio of optimal-solution-structure based heuristics. This problem has multiple applications, including in the humanitarian aspect of post-disaster debris clearance.

In supply chain network operations, we consider time sensitive supply networks in both for-profit and humanitarian supply chains. For The Post Disaster Kit Routing Problem, we aim to deliver kits from a regional warehouse to mobile storage units where the kits are distributed to the population affected by a disaster. Those kits are comprised of items required for the survival of the affected population and need to be delivered in a timely manner. We model this problem as a multi-objective MIP while optimizing various aspects of the post-disaster times such as effectiveness, fairness, autonomy, and cost.

For the for-profit supply chains, we consider The Logistic Network Design for Fresh Flowers - a temperature-controlled network for a yearly supply of several types of fresh cut flowers from multiple supply locations. We aim to satisfy demand by different customer types, located in various demand locations. We propose an efficient and effective three-fold heuristic approach, aiming to replace current reliance on air transportation.

In order to illustrate the models' and algorithms' capabilities, each problem is associated with a comprehensive case study comparing results to other known algorithms or current practice solutions.