Thesis Title: Hyperconnected Fulfillment Optimization: A Holistic Dynamic Approach

Thesis Committee:

Dr. Benoit Montreuil (advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Santanu S. Dey (co-advisor), H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Dr. Matthieu Lauras, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology 

Dr. Walid Klibi, Supply Chain Center of Excellence, KEDGE Business School

Dr. Mathieu Dahan, H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology

Date and Time: Monday, November 17th, 9:00am - 11:00am (EST)

Meeting Link: Jiachen Shi Thesis Defense | Meeting-Join | Microsoft Teams

Abstract:

Modern supply chains are complex networks with various actors such as manufacturers, distributors, retailers, and customers that are tightly interconnected. In today’s turbulent high-velocity environment, supply chain networks face continuously evolving challenges, which are further amplified as e-commerce continues to expand. Focusing on solving the challenges from the perspective of order fulfillment and inventory management, we aim to find a breakthrough from the introduction of the Physical Internet (PI). In light of PI, logistics infrastructure is shifting from private networks to an open web of interconnected networks, each involving multiple organizations collaborating to share resources, optimize operations, and enhance service responsiveness. The Physical Internet enabled hyperconnected fulfillment, where fulfillment operations are no longer isolated within a retailer’s private network; instead, decision makers can access a much wider range of open-source fulfillment facilities that are interconnected, openly available, and their service capacity available for contract on demand. This thesis addresses algorithmic, modeling, and application challenges faced by fulfillment centers of e-commerce retail supply chain through a novel hyperconnected approach in collaboration with a real-world e-commerce retailer.

In Chapter 2, we propose a unified framework that studies inventory replenishment provisioning and e-commerce fulfillment jointly.

As the digital capabilities of supply chains rapidly increase, it is now much easier technologically to interconnect supply chain nodes, share data, automate processes, and obtain sensor-based signals. The proposed framework takes advantage of the augmenting visibility and interoperability potential of a sentient supply chain. We formulate the replenishment provisioning problem as a mixed integer program that accounts for the interplay between inventory management and e-commerce fulfillment, as well as lateral transshipment and product substitution. The framework is applied to a real word retailer test case.

A wide range of decision-making problems in the supply chain and logistics fields includes a time component, where optimization occurs across multiple time periods. In Chapter 3, we propose a systematic modeling framework to stabilize solutions through frequent resolving in such sequential decision making, by leveraging decisions of the optimal solutions that are traditionally not implemented. The application of the proposed framework is illustrated through the dynamic decision-making problem studied in Chapter 2. Various formulations are incorporated into the replenishment provisioning MILP model to address the instability inherent in frequent resolving of a rolling horizon framework.

Chapter 4 tackles the Hyperconnected Fulfillment Facility & Service Contracting problem that consists of determining contracts for open-source facilities and services for e-commerce demand. Conventional fulfillment networks operate under a closed paradigm, the null hypothesis that retailers own and manage all facilities and services. Our proposed framework integrates PI principles and facility location modeling and is modeled as a Mixed Integer Linear Program (MILP). We leverage Bender's decomposition that’s efficient for solving large-scale instances of the proposed model. Experimental results from the e-commerce retailer case study demonstrate the advantages of leveraging open fulfillment centers through dynamic contracts over traditional industry practice.