Thesis Title: Dynamic Prescriptive Analytics for Logistics Service Providers

Date and Time: Friday, July 29th, 2022, at 10 am (EDT)

Meeting Location: https://gatech.zoom.us/j/6598438705

Committee:

Dr. Benoit Montreuil, School of Industrial and Systems Engineering, Georgia Tech (Advisor)

Dr. He Wang, School of Industrial and Systems Engineering, Georgia Tech (Co-Advisor)

Dr. Siva Theja Maguluri, School of Industrial and Systems Engineering, Georgia Tech

Dr. Mathieu Dahan, School of Industrial and Systems Engineering, Georgia Tech

Dr. Jennifer Pazour, School of Engineering, Rensselaer Polytechnic Institute

Abstract:

With the strong growth in e-commerce and distribution volume comes a demand for more efficient logistics infrastructure to support the development. Third-party logistics service providers (3PLs) are facing increased demands for their services. This thesis focuses on the application of analytics-based approaches to improve 3PL’s capacity management, and the efficiency of warehouse processes such as picking.

 In chapter 2, we introduce a decision-making framework for 3PLs to dynamically manage its assets. We address the three layers of analytics: descriptive, predictive, and prescriptive analytics and show how a 3PL can use these to transform into a proactive hyperconnected logistics player.

 In chapter 3, we introduce the demand and capacity management problem that cold-chain logistics player face when operating temperature-controlled warehouses. At each time step, the 3PL decides whether to change the temperature of their storing rooms (capacity management) and whether to accept or reject an incoming customer request (demand management) for temperature controlled space. We show that optimal solutions are not feasible and introduce data-driven rollout-based algorithms that outperform greedy heuristics.

In chapter 4, we discuss the development of a simulation model that allows a deeper understanding of the trade-off between order consolidation and timely order fulfillment in multi-order picking systems in warehouses. We develop a domain informed, regret-based threshold policy that addresses the trade-off and compare its performance to greedy heuristics.