SWUTC Research Project Description
Title of Project: Multimodal Network Models for Robust Transportation Systems
Project Number: 167867
Principal Investigator:
S. Travis Waller
(512) 471-4539
P.I. Affiliation: University of Texas at Austin
Project Monitor:
Andrew Griffith,
Research and Technology Implementation Office
(512) 465-7908
Project Status: Active
Date Started: 9/1/05
Estimation Completion Date: 8/31/06
Estimated Cost - Current Fiscal: $31,000
Estimated Cost - Total Planned: $31,000
Project Summary:
Project Abstract:
It is well known that travel demand, which is often forecast decades into the future, is not known with complete certainty. Previous research has shown that a transportation network designed to be optimal for one future scenario, may perform poorly if future demand differs from predictions. Instead of planning for one projected scenario, better solutions will emerge if planning considers a range of feasible future scenarios. The purpose of this project is to examine the impact of considering transport demand as uncertain on the performance of dynamic multimodal networks. This project expands on previous work by the authors considering uncertainty in dynamic traffic networks by incorporating the transit mode. Multiple user classes will be used to differentiate between different types of travelers. The assignment model will be iterated with a mode choice model until convergence is achieved. Measures of robustness will be used to assess the effectiveness of each mode considered in handling unexpected future demand scenarios.
Project Objectives:
The four study objectives are stated in sequential order as follows:
Task Descriptions:
Task 1. Review Relevant Existing Literature
In this task, papers and research reports on multimodal network evaluation in dynamic contexts will be collected and reviewed critically, especially those works dealing with demand uncertainty. It will also be necessary to review literature on the integration of dynamic network models for routing and logit models for demand. To determine appropriate user classes, it will be necessary to review literature on multiclass transit routing.
Task 2. Algorithm Development
Models developed in the previous year’s SWUTC project for robustness in dynamic traffic equilibrium will be used as a starting point for modeling robustness in multimodal dynamic user equilibrium. A mathematical formulation will be devised to represent and analyze the problem of multiclass multimodal dynamic user equilibrium.
Task 3. Solution Method Development
Based on the formulation developed in Task 2, appropriate solution methods will be explored to find statistically valid estimates of robust system performance. Potential approaches may include demand alterations (such as Chance Constraint, planning demand inflations or other data modifications), Monte Carlo evaluation, and other sophisticated sampling procedures.
Task 4. Software Development
The proposed solution methodologies will be implemented with C++ and interfaced with an existing transportation toolset with DTA capability (VISTA: Ziliaskopoulos and Waller, 2000). VISTA will need to be altered to allow for integration with a mode choice model and transit assignment. Currently VISTA has the capability to model the flow of buses, but much effort will be required to enforce rules that guide bus behavior based on the demand for transit.
Task 5. Modification of DTA Test Cases for Multimodal Analysis
Various test cases will be assembled by leveraging on other ongoing research by the PI. These cases will include network topologies, supply measures, and estimates of time-dependent demand (with statistical measures describing uncertainty). To evaluate the impact of transit on future system performance, a variety of transit configurations must be tested. These configurations could be developed based on intuition and input from transit practitioners, or produced from a random network generator.
Task 6. Evaluation of Methodology
The developed strategies and computer code will be evaluated in this task. The evaluation will examine the statistical confidence, computational time, and data requirements for each approach. Further, opportunities for method improvement will also be explored in this task. The results from each type of transit configuration and distribution of OD demand will be compared.
Task 7. Development of Final Procedures for Large-Scale Dynamic Networks
This task will address the requirements for large-scale network implementations of the proposed strategies. First, bounded relationships will be explored so that if optimality must be sacrificed then some confidence can still be attained. Second, decomposition approaches will be explored that exploit the inherent network structure. Finally, heuristic and sample-based techniques will be explored for deployable approaches to large-scale networks such as Tabu Search and Genetic Algorithms.
Task 8. Final Report
A comprehensive final report will be developed. Algorithms and software developed and results will be fully documented.
Index Terms:
Travel Demand; Multimodal Transportation; Transportation Modes; Traffic Equilibrium; Traffic Forecasting; Traffic Estimation; Dynamic Traffic Assignment; Mode Choice; Traffic Models