*********************************
There is now a CONTENT FREEZE for Mercury while we switch to a new platform. It began on Friday, March 10 at 6pm and will end on Wednesday, March 15 at noon. No new content can be created during this time, but all material in the system as of the beginning of the freeze will be migrated to the new platform, including users and groups. Functionally the new site is identical to the old one. webteam@gatech.edu
*********************************
Thesis Title: Service Network Design for Parcel Trucking
Advisors:
Dr. Alan Erera, School of Industrial and Systems Engineering, Georgia Tech
Dr. Alejandro Toriello, School of Industrial and Systems Engineering, Georgia Tech
Committee members:
Dr. Martin Savelsbergh, School of Industrial and Systems Engineering, Georgia Tech
Dr. Benoit Montreuil, School of Industrial and Systems Engineering, Georgia Tech
Dr. German Riano, Amazon
Date and Time: 7 pm ET, Wednesday January 13, 2021
Meeting URL: https://bluejeans.com/100142235
Meeting ID:
100 142 235
Abstract:
Last-mile logistics has become an essential part of the economy to make possible the transportation of goods from producers to end-consumers. As a result of the explosive growth of e-commerce in the past decade, from 2014 to 2019, during which time e-commerce sales ratios nearly tripled globally, demand for last-mile delivery is soaring and is expected to grow by 78% globally by 2030. Thus, one of the main objectives of last-mile delivery logistics is to deliver packages as affordably, quickly and efficiently as possible.
Logistics and service network design models are the most used tools for developing novel enabling technology and enhancing current practices in parcel and freight logistics. In these planning processes, the enabling technology provides the planner with the choice of paths for shipments and the services or resources necessary to execute them. The selection of hubs for cross-docking activities is an important tactical decision to take advantage of consolidation opportunities. Building feasible consolidation plans that allow to operate in a seamless manner offer useful advantages in practice such as conforming in-tree plans. In this thesis, we address the challenges discussed above. We develop a large-scale package express service network design in which we build flat network models that aim to capture relevant timing constraints. The first part focuses on a detailed intracity scheduling service network design problem for megacities whereas the last two works focus on linehaul consolidation planning.
In Chapter 2, we present a service network design problem in the context of intra-city courier service in megacities. The problem consists of designing a cost-effective service network to move packages from their origins to their destinations taking into account the committed service levels. In this study, we focus on shuttle activities and develop optimization technology for the design shuttle services using novel rate-based models to determine package as well as vehicles routes. To ensure operational simplicity, we build repeatable vehicle routes in the form of cycles, which are executed throughout the day, taking into account capacity and timing requirements. A computational study using data from a large Chinese package company shows that our technology produces a cost-effective service network design for shuttle schedules with excellent on-time performance.
In Chapter 3, we present a strategic hub selection problem within the context of service network design for a package courier system operating fast time-definite services in large urban areas. In this work, we address the selection of intermediate hubs for cross-docking, incorporating another layer of complexity on top of service network design. The aim is to select the hubs (either gateway hubs or local hubs) that are the most cost-effective to perform cross-docking activities. We develop a cost-effective greedy heuristic approach that employs smaller tractable integer programming based (IP-based) models to add one intermediate hub on each iteration. We develop three IP-based greedy heuristic variants: (1) a Greedy-Hub, (2) a Greedy-Hub-Full which solves larger IPs on each iteration, and (3) a GRASP-Hub that randomizes the selection of intermediate hubs. Computational studies show that the greedy approach selects well geographically distributed cost-effective hubs for cross docking, and moreover, the heuristic outperforms the full optimization model by a 20% gap difference for the relevant cases.
Finally, in Chapter 4, we build an operationally feasible plan in which the hubs allowed for cross docking are fixed. We generalize the in-tree concept by introducing the generalized in-tree, referred to as GIT, which has useful operational benefits and encompasses a combination of the destination, time requirement and remaining service time of commodities. We develop three approaches: (1) a GIT-based approach, (2) a path-based approach and (3) an optimized GIT-based formulation. We demonstrate, via a computational study, that the solutions of the approaches return high percentage of a GIT structure, they show more than 90% of a 2 and 4-hour discretized GIT structure for the large instances. We also demonstrate that imposing different GIT structures bears a penalty cost between 2% and 4%. Furthermore, the flexibility to employ longer paths, having more than one intermediate stop, employing the same set of intermediate hubs selected in Chapter 3, allows to improve solution costs by 3%.
