Objective

Fluctuations in travel demand and irregular incidents give rise to recurrent and non-recurrent congestion delay on highway segments and thus make the system performance unstable and unreliable. The research attempts to deliver a proof of concept that optimal planning and management strategies can be formulated through applying robust optimization methodology such that limited resources could be allocated more rationally, and reliability of a highway network improved more efficiently.

Background

The potential sources of disruption to highway traffic operations are numerous, ranging from irregular and random incidents, like earthquakes, terrorist attacks, floods, adverse weathers, traffic accidents, breakdowns, signal failures, roadwork etc, to regular fluctuations of travel demand in times of day, days of the week, and seasons of the year. The scales, impacts, frequencies and predictability of these disruptive events will of course vary enormously, with natural or man-made disasters at one extreme, and routine events that happen every now and then at the other.

While little can be done about their scales, frequencies or predictability, particularly where natural disasters and accidents are concerned, it remains possible to design and manage highway networks so as to minimize the disruption such events can cause. It is not realistic to expect the performance of a highway network under catastrophic disasters is the same as that under minor incidents. Therefore, the required functionalities and ways to ensure the functionalities for those two extreme types of disruptive events would differ significantly. The research focuses on improving performance reliability of a highway network at its routine operations, addressing everyday incidents.

Various policies and activities can be adopted to improve highway network reliability. In order to facilitate delivering a proof of concept, we build our modeling framework to consider road improvement and incident management. Other policies, such as providing information through advanced traveler information systems (ATIS) can be accommodated by the proposed modeling framework without much difficulty.

With a limited budget available to highway network management, engineers and planners sometimes need to decide on which links improvement works, such as maintenance and rehabilitation or road expansion, should be implemented in order to maintain or improve the effectiveness of a highway system. It is quite often that the decisions are made without considering the impacts of disruptive events on the system performance. As a result, the system performance may deteriorate significantly upon the onset of those disruptive events. In some rare cases, in order to deal with the impacts, sensitivity analyses have been performed to evaluate sensitivities of the decisions to the uncertainties (onsets of disruptive events). However, such practices are intrinsically posteriori or reactive, and provide no direct mechanism for controlling the sensitivities.

On the other hand, traffic incident management is emerging as a proven solution to ensure highway reliability. It is a planned and coordinated process to detect, respond to, and remove traffic incidents and restore traffic capacity as safely and quickly as possible. As one component of incident management, incident response teams and freeway service patrols (FSP) facilitate the quick removal of incidents through fast response and clearance times. Given a limited budget, setting up FSP beats that tow trucks patrol on and assigning FSP resources to beats so as to maintain sufficient service intensity are very important to success of the system. However, these decisions are often made in a heuristic manner. For example, criteria for funding allocations in the FSP program in California have been based on population, urban freeway lane miles, and vehicle hours of congestion delay. A wiser allocation of available funding could be made from a systematic perspective such that the effectiveness of the FSP system can be maximized.

Methodology

In general, the decision problem that we tackle in the research is, with a given budget, choosing links from a network to implement road improvement works and determining the corresponding investment magnitudes, and/or choosing links to implement FSP program and determining the corresponding service intensity so as to improve performance reliability of the network at most. We apply robust optimization to determine a robust improvement scheme for a highway network. Robust optimization is a modeling methodology to solve optimization problems in which the data are uncertain and only known to belong to some uncertainty set. The approach is to seek optimal (or near optimal) solutions that are not overly sensitive to any realization of uncertainty. The approach is not to propose some reliability index and then maximize it to obtain the optimal improvement scheme. In contrary, we attempt to gracefully trade off effectiveness vs. guaranteed robustness and reliability. The objective will still be maximization of effectiveness, but the fundamental idea is to seek a robust optimal solution that tolerates changes of travel demand and network supply caused by disruptive events, up to a given bound known a priori. The solution is therefore neither careless (without considering uncertainty at all) nor overly conservative. Because the improvement scheme is robust, we expect that the resulting highway network will remain ÒcloseÓ to its designated performance under any realization of the uncertainties (onsets of disruptive events). In this sense, the network will be more reliable and predictable.

Project Completed Final Report: UCB-ITS-PRR-2005-35