Thesis Title: Planning the Operations of Parcel Express Shipment Service

Advisors: 

Dr. Martin Savelsbergh, School of Industrial and Systems Engineering, Georgia Tech

Dr. Anton Kleywegt, School of Industrial and Systems Engineering, Georgia Tech

Committee members:

Dr. Alejandro Toriello, School of Industrial and Systems Engineering, Georgia Tech

Dr. He Wang, School of Industrial and Systems Engineering, Georgia Tech

Dr. Yixiao Huang, SF Express (Group) Co., Ltd.

Date and Time: 9:00am EST, Thursday, June 16, 2022

Meeting URL (for Zoom):  https://gatech.zoom.us/j/97554923920

Meeting ID (for Zoom): 

975 5492 3920

Abstract:

Package delivery represents a significant part of the transportation industry. A critical aspect of package delivery is timely service, which is driven, in part, by the growth of e-commerce, which relies heavily on fast delivery. To provide an economically viable delivery service, express service providers need to carefully allocate and utilize their resources during operations. The primary challenge is to identify and exploit consolidation opportunities (so as to keep the costs down) while satisfying the service guarantees offered to customers (so as to maintain or increase market share).

The size of real-world instances and complicating restrictions faced by express service providers in practice often impose challenges to problem tractability and difficulties in directly applying standard optimization algorithms. Therefore, computationally efficient heuristics that can produce high quality solutions within a reasonably short period of time are often pursued. This dissertation aims to develop heuristic algorithms that can provide effective decision support for planning the operations of parcel express shipment service in practice. Background information and challenges in planning the operations of a parcel express shipment service are provided and discussed in Chapter 1.

In Chapter 2,  we study a novel service network design problem encountered in a real-world urban same-day delivery system, in which the number of vehicles that can simultaneously load or unload at a hub is limited. The problem is modeled on a time-expanded network and formulated as an integer program. To produce high-quality solutions in a reasonable amount of time, we propose an integer programming based heuristic as well as a hybrid matheuristic that takes advantage of the strengths of the proposed integer programming based heuristic and of a local search based metaheuristic. The efficacy of the heuristics is demonstrated on a number of real-world instances.

In Chapter 3, we introduce a challenging routing and scheduling problem arising in the city operations of a large package express carrier. Vehicles perform multiple trips during a planning horizon spanning multiple driver shifts, where a trip can involve deliveries only, pickups only, or deliveries followed by pickups. Complicating factors include split deliveries and pickups, cross-trip consistency requirements, and limited unloading capacity at the main hubs. We develop an optimization-based multi-phase heuristic solution approach seeking to minimize the number of vehicles used. An extensive computational study is conducted to provide insight into current operations, its bottlenecks, and potential adjustments to improve efficiency.

In Chapter 4, we explore an operational planning problem for an express shipment service network where flight routes and schedules are fixed, and the air container movements as well as the crossdocking plans need to be determined and fixed before the realization of uncertain demand. Using a scenario-based approach for capturing the uncertainty of demand, we formulate a two-stage stochastic programming model where cross-docking container movements are decided in the first-stage and package flows are determined after demand is revealed in the second stage. We propose a Benders decomposition based heuristic approach to reduce computational time for producing high-quality solutions. An extensive computational study using real-world instances demonstrates the effectiveness of the approach.