Date of Award

2017

Document Type

Open Access Dissertation

Department

Civil and Environmental Engineering

Sub-Department

College of Engineering and Computing

First Advisor

Nathan Huynh

Abstract

Over the last 50 years, international trade has grown considerably, and this growth has strained the global supply chains and their underlying support infrastructures. Consequently, shippers and receivers have to look for more efficient ways to transport their goods. In recent years, intermodal transport is becoming an increasingly attractive alternative to shippers, and this trend is likely to continue as governmental agencies are considering policies to induce a freight modal shift from road to intermodal to alleviate highway congestion and emissions. Intermodal freight transport involves using more than one mode, and thus, it is a more complex transport process. The factors that affect the overall efficiency of intermodal transport include, but not limited to: 1) cost of each mode, 2) trip time of each mode, 3) transfer time to another mode, and 4) location of that transfer (intermodal terminal). One of the reasons for the inefficiencies in intermodal freight transportation is the lack of planning on where to locate intermodal facilities in the transportation network and which infrastructure to expand to accommodate growth. This dissertation focuses on the intermodal network design problem and it extends previous works in three aspects: 1) address competition among intermodal service providers, 2) incorporate uncertainty of demand and supply in the design, and 3) incorporate multi-period planning into investment decisions. The following provides an overview of the works that have been completed in this dissertation.

This work formulated robust optimization models for the problem of finding near-optimal locations for new intermodal terminals and their capacities for a railroad company, which operates an intermodal network in a competitive environment with uncertain demands. To solve the robust models, a Simulated Annealing (SA) algorithm was developed. Experimental results indicated that the SA solutions (i.e. objective function values) are comparable to those obtained using GAMS, but the SA algorithm can obtain solutions faster and can solve much larger problems. Also, the results verified that solutions obtained from the robust models are more effective in dealing with uncertain demand scenarios.

In a second study, a robust Mixed-Integer Linear Program (MILP) was developed to assist railroad operators with intermodal network expansion decisions. Specifically, the objective of the model was to identify critical rail links to retrofit, locations to establish new terminals, and existing terminals to expand, where the intermodal freight network is subject to demand and supply uncertainties. Addition considerations by the model included a finite overall budget for investment, limited capacities on network links and at intermodal terminals, and due dates for shipments. A hybrid genetic algorithm was developed to solve the proposed MILP. It utilized a column generation algorithm for freight flow assignment and a shortest path labeling algorithm for routing decisions. Experimental results indicated that the developed algorithm can produce optimal solutions efficiently for both small-sized and large-sized intermodal freight networks. The results also verified that the developed model outperformed the traditional network design model with no uncertainty in terms of total network cost.

The last study investigated the impact of multi-period approach in intermodal network expansion and routing decisions. A multi-period network design model was proposed to find when and where to locate new terminals, expand existing terminals and retrofit weaker links of the network over an extended planning period. Unlike the traditional static model, the planning horizon was divided into multiple periods in the multi-period model with different time scales for routing and design decisions. Expansion decisions were subject to budget constraints, demand uncertainty and network disruptions. A hybrid Simulated Annealing algorithm was developed to solve this NP-hard model. Model and algorithm’s application were investigated with two numerical case studies. The results verified the superiority of the multi-period model versus the single-period one in terms of total transportation cost and capacity utilization.

Rights

© 2017, Fateme Fotuhiardakani

Share

COinS