Ph.D. Thesis Proposal

By

Leah Josefine Ruckle

(Advisor: Prof. Dimitri N. Mavris)

9:30 AM, Wednesday, December 14, 2016

Weber Space Science and Technology Building (SST-II)

Collaborative Visualization Environment (CoVE)

A multidisciplinary approach to solving the express shipment service network design problem

ABSTRACT:

Express service package delivery is a multi-billion dollar industry in the United States and abroad. It has become an essential part of business operations and, with the rise of e-commerce, an expectation of the everyday consumer. Express package delivery is challenged by extremely tight service guarantees, the huge volume of material that must be moved in a multimodal network, and the large geographical area of guaranteed service. These challenges necessitate the use of large jet aircraft fleets, thus prompting the desire of package delivery companies to use these aircraft as efficiently as possible.

Express package delivery operations are mathematically modeled as a type of service network design problem called the express shipment service network design problem (ESSNDP). The resulting mixed integer linear program is too large to solve exactly within a reasonable runtime for realistic problem instances and so approaches such as branch-and-price, column generation, neighborhood search heuristics and Benders decomposition are used to produce high-quality solutions.

This work explores the effectiveness of multidisciplinary design optimization (MDO) to solve the ESSNDP, a previously untested approach. Multidisciplinary design optimization is a popular optimization technique, particularly within aerospace and mechanical engineering, which is used in the design of complex physical systems such as aircraft, spacecraft, automobiles, wind turbines and bridges. In MDO, large and complex optimization problems are solved by decomposing the problem into disciplines, solving those disciplines semi-independently, and iteratively reconciling the differences between the disciplines until a high-quality feasible solution emerges.

The ESSNDP decomposes into two well-structured subproblems. One is the package routing subproblem which decides how the packages will flow through the network from their origins to their destinations via at least one sorting hub. The other is the aircraft scheduling subproblem which decides how to assign the limited heterogeneous fleet to arcs in the network to provide capacity to move the packages. These two subproblems can then be used and solved as disciplines in an MDO architecture.

Multidisciplinary design optimization has proven itself to be useful in other, physical, design realms, but it has not been applied to logistics problems. I will explore its effectiveness in solving the ESSNDP, a large and complex logistics problem, and compare its performance to an undecomposed approach and a Benders decomposition approach.

Committee Members:

Professor Dimitri N. Mavris, GT AE

Professor Brian J. German, GT AE

Professor Alejandro Toriello, GT ISYE