SWUTC Research Project Description
Title of Project: Quantifying Travel Time Variability in Transportation Networks
Project Number: 167275
Principal Investigator:
Travis Waller
(512) 471-4539
P.I. Affiliation: University of Texas at Austin
Project Monitor:
Arash Mirzaei
(817) 695-9261
Project Status: Active
Date Started: 9/1/07
Estimation Completion Date: 8/31/08
Estimated Cost - Current Fiscal: $40,000
Estimated Cost - Total Planned: $40,000
Project Summary:
Project Abstract:
Since transportation systems are inherently complex, interconnected with other systems, and driven by human behavior, it is impossible to specify future conditions with any degree of accuracy. In this study, we focus on short-term sources of uncertainty in travel times, and their impacts. Although much research has been invested in how these systems respond to uncertainty, considerably less attention has been given to how to quantify uncertainty in the first place. While ITS devices allow this variability to be directly measured, in most cases these are present on only a small minority of roadway segments in a given region. This project will develop a method for estimating the variation in travel time when no exact measurement is possible, allowing the application of stochastic transportation planning models while obviating the need to collect this data directly and the associated costs.
Project Objectives:
The three primary objectives of this research are as follows:
Task Descriptions:
Task 1. Review of Relevant Literature on Travel Time Uncertainty
A careful study will be made of existing research on travel time uncertainty, both of its causes and its effects. This will include review of safety and operations literature to identify factors that lead to disturbances in traffic flow. These will be of use in determining the most appropriate and useful scope for this project.
Task 2. Determination of Modeling Scope
In conjunction with Task 3 (data acquisition), and guided by Task 1, the scope of this project will be explicitly delineated based on data availability, previous research, and a determination of the most significant factors affecting travel time variability.
Task 3. Data Acquisition and Adaptation into Relevant Formats
In this task, the research team will collect data from a variety of sources, such as ITS data archives, metropolitan planning organizations, weather records, and incident databases. As these are obtained from different agencies and locations, a substantial amount of effort will be needed to geocode and synthesize these into the format most useful for this project.
Task 4. Generation of Variability Prediction Model
Statistical techniques will be applied to the data collected in Task 3 to isolate the impact of various uncertain factors on travel time. Such techniques may include, but are not limited to, regression (regular or nonparametric), time-series analysis, Bayesian networks, kriging, and hazard-based statistical models. Additionally, the context of particular links in a larger network will be included. It is likely that several techniques will be applied (either simultaneously or sequentially) to assess the impacts of a broad spectrum of factors.
Task 5. Validation of Variability Prediction Model
The generalizability of this model will be examined, including internal consistency, transferability to other geographic regions, and suitability for various applications, such as prediction of toll revenues, transportation planning, and mode choice modeling.
Task 6. Final Report
A comprehensive final report will be written, describing all research work performed.
Index Terms:
Network Modeling, Operational Uncertainty, Information Provision, Value of Reliability, Safety