Publication Type

PhD Dissertation

Version

publishedVersion

Publication Date

5-2025

Abstract

This thesis investigates order fulfillment operations for online retailing. Order fulfillment encompasses the entire process of receiving, processing, and delivering customer orders. The rapid growth of e-commerce has heightened the demand for efficient fulfillment operations. This thesis explores strategies for optimizing order fulfillment in three scenarios. The first scenario studies cellular bucket brigade, a method for coordinating workers in an assembly line or order-picking environment to improve workflow efficiency. The second scenario investigates zone picking, serial bucket brigades and cellular bucket brigades with hand-off times. The third scenario focuses on autonomous robotic systems within package sortation centers.

Chapter 2 investigates cellular bucket brigades, a way to effectively coordinate workers by folding the serial bucket brigade line in the middle such that workers can work on both sides. We study the performance of cellular bucket brigades on discrete stations with non-preemptible work contents and exponential processing times. We analytically derive the average throughput by tracing the system status immediately before each job completion. We determine the analytical performance gap between serial bucket brigades and cellular bucket brigades in terms of each worker’s average work speed and expected idle time. This gap is due to the additional wasted capacity by each worker in one system compared to the other. We extensively compare their performance in manufacturing and warehouse order-picking based on synthetic data as well as real data from a hardware distributor and an e-commerce platform. Our results suggest that the cellular bucket brigade is generally more productive than its serial counterpart, but for order-picking, their throughput gap becomes smaller as the pick intensity increases. We find that strategies that make faster workers cover stations with larger work contents are more productive. Furthermore, a high speed variation combined with a highly-skewed work-content distribution can further improve the performance of the most productive strategies. We provide practical guidelines for managers to optimize serial and cellular bucket brigades under different situations.

Chapter 3 focuses on hand-off in order fulfillment strategies, specifically speaking, zone picking, serial bucket brigades, and cellular bucket brigades. Order picking is a critical yet labor-intensive operation in warehouses, often accounting for a significant portion of total operating costs. With the growing demand for rapid and accurate order fulfillment, efficient order-picking strategies have become essential. Among these, zone picking and bucket brigades (both serial and cellular) are widely adopted to improve workflow coordination and productivity. However, frequent task hand-offs inherent in these systems introduce inefficiencies, consuming time and disrupting operational flow. To incorporate hand-off times into the model, we assume that both processing times and hand-off times between consecutive workers follow exponential distributions. We explicitly track workers engaged in hand-offs, reanalyze the system state, and analytically derive the average throughput of the three systems by tracing system status immediately before each job completion. Through numerical studies, we evaluate the performance of zone picking, serial buckets, and cellular bucket brigades with hand-off times under both identical and heterogeneous hand-off rates. The results suggest that the bucket brigades generally remain more productive than zone picking.

Chapter 4 studies an autonomous robotic system, operating in a package sortation center where performs packages stowage and retrieval over a multi-period horizon. The sortation center includes a storage area with multiple storage pods and multiple work stations surrounding it. Autonomous robots lift and transport storage pods to workstations, where human operators stow or retrieve packages. At the start of each period, we select pods from the storage area to stow arriving packages at each station and each package is associated with a specific delivery route. At the end of the period, when certain delivery routes are called for dispatch, we retrieve packages from the pods to fulfill the dispatch. Our objective is to minimize the total travel cost of the pods. We first formulate a mixed-integer programming model under a deterministic setting, assuming the number of arriving packages is known. To address computational challenges, we propose a two-phase heuristic approach that decomposes the integrated problem into two subproblems. Next, we extend our base model to account for uncertainty in package arrivals. We consider using Sample Average Approximation (SAA) and Restricted Linear Rule (RLR) to make the model with uncertainty tractable. In our numerical studies, we first compare our two-phase heuristic approach with two existing policies: class-based storage assignment rule and correlated storage assignment rule. We then compare the performance of SAA and RLR by varying the models parameters.

Degree Awarded

PhD in Business (Ops Mgmt)

Discipline

E-Commerce | Entrepreneurial and Small Business Operations | Strategic Management Policy

Supervisor(s)

LIM, Yun Fong

First Page

1

Last Page

150

Publisher

Singapore Management University

City or Country

Singapore

Copyright Owner and License

Author

Download

Available for download on Sunday, May 31, 2026

Share

COinS