Research

Project Number:
03-06

Research Project:
Robust Investment Decisions for Highway Capacity Expansions

P.I. Name & Address:
Fernando Ordonez
Department of Industrial and Systems Engineering
University of Southern California
Los Angeles, CA 90089-0193
Tel:(213) 821-2413
Fax:(213) 740-1120
Email: fordon@usc.edu

Project Objective:
According to Caltrans' 2001 HOV Annual Report, projects to add new HOV lanes to the Los Angeles highway system for an estimated $2.9 billion are currently under study.The order in which these future projects are executed will influence significantly the driving conditions in the Los Angeles highway system for the next decade.

In this project we consider the problem of deciding which project to execute, i.e. where to add new HOV lanes in the Los Angeles highway system, taking into account the existing uncertainty in the data that is present in the problem.The problem is modeled assuming there is a fixed budget and trying to minimize the total travel time or average travel time throughout the system.The solution to this problem depends on the assumed origin-destination flows, which are estimates of the true unknown origin-destination flows.In addition this information evolves over time, thus creating the need to find a solution that is not significantly affected by changes in this data.

We propose to study and develop a technique to obtain a robust solution to this problem.A robust solution is feasible and close to optimal for all plausible data instances of the problem.For problems where the optimal solution exhibits a strong dependence on the data, a robust solution is likely to outperform the optimal solution on the true unknown data instance, thus providing a solution that is in practice a more successful policy than what is obtained with an optimal solution.Additionally, the difference between the robust solution and an optimal solution can provide insights into the tradeoffs between the optimality of a solution and its robustness for this application.

We will develop this technique by applying a newly developed mathematical framework that obtains robust solutions for problems with data uncertainty which holds promise of being applicable to this particular transportation problem.

Task Descriptions:
1. Study of robust optimization and its applicability to the NHOV problem. (3 months)
2. Modeling of NHOV problem and formulation of a robust NHOV (RNHOV).(5 months)
3. Exploration of efficient computational strategies to solve the RNHOV problem.(5 months)
4. Study of robust solution.(2 months)
5. Preparation of final report.(2 mos.)

Milestones, Dates:
June 1, 2003 - May 31, 2004; final draft report 5/31/04

Total Budget:
$35,000

Student Involvement:
One Student @ 25% time for 12 months

Relationship to Other Research Projects:
Project is part of infrastructure focus area

Technology Transfer Activities:
Project report will be posted on the website

Potential Benefits of the Project:
Better tools for capital investment decisions

TRB Keywords:
HOV lanes, decision analysis

Primary Subject:
Transportation planning, economics, and institutional issues

Enabling Research:
Tools for modeling and design

Modal Orientation:
Highway