Early in the project the researchers set out a framework for incorporating reliability into planning and operation models that distinguishes between the demand and supply side. Travel demand may be static, as in typical planning models; dynamic for planning and operational models; or activity-based. Supply–in other words, the capacity of each part of the network–may be fixed, stochastic, or systematically varying. The SHRP 2 Reliability focus area identified seven sources of nonrecurring congestion: incidents, weather, work zones, special events, traffic control devices not working properly, unusual fluctuations in demand, and bottlenecks that can exacerbate these sources of unreliability. These nonrecurring sources of congestion can affect supply, demand, or both; for example, work zones affect supply; special events, demand; and incidents and weather, both. These supply and demand factors influence the travel time for origin—destination (O-D) pairs across the network and, in turn, the distribution of travel time from which various reliability measures can be derived. To explain how to address reliability when using micro- and meso-simulation models, the framework was extended to distinguish between sources of nonrecurring congestion external (exogenous) to a simulation model and internal (endogenous) to it. Exogenous factors include incidents, weather, and work zones, whereas endogenous factors include heterogeneity of driver behaviour and vehicle type on the demand side and breakdown of flow, traffic control, and differences in car-following behaviour on the supply side. Micro-simulation models are widely used in the transportation field to understand how vehicles behave in detailed settings, such as a series of traffic signals along an arterial street, freeway on-ramps, or a small network of roads. Meso-simulation models are suitable for higher-resolution analysis and can be applied to networks of varying sizes, including an entire region. Both micro- and meso-simulation models are based on some form of traffic physics, in contrast to a standard four-step demand model. This project focused considerable attention on how micro- and meso-simulation models could address travel time reliability. The essence of the approach is to sandwich a simulation model between a pre- and post-processor such that together, all three components can portray travel time reliability on a network or part of it. The researchers developed two software prototypes that were tested with both a widely used meso-simulation model and a widely used micro-simulation model. The first software prototype, the Scenario Manager, consisted of the pre-processor for either type of simulation model. The Scenario Manager produces random scenarios involving various sources of nonrecurring congestion such as traffic incidents, weather, and work zones. It can also address scenarios based on historical data or scenarios previously constructed for planning purposes. The other software prototype is the Trajectory Processor. This post-processor determines the distribution of travel time for every O-D pair on a network. Nearly all the travel time reliability metrics, including standard deviation and the Planning Time Index, can be derived from the travel time distribution. For information about how to use the two prototypes, see their user guides. This report provides more information about the Scenario Manager and the Trajectory Processor, as well as the research. The research also produced SHRP 2 Report S2-L04-RR-1: Incorporating Reliability Performance Measures into Operations and Planning Modelling Tools: Application Guidelines, about a micro- or meso-simulation model with pre- and post-processors. Private sector software vendors may wish to closely examine the prototype software to determine the merits of incorporating similar capability into the products they have on the market. The application guidelines and user guides should help private vendors make informed decisions. It is worth noting that a similar scenario manager and procedures for compiling the distribution of travel time were also developed and applied in the SHRP 2 L02 project, Incorporation of Travel Time Reliability into the Highway Capacity Manual. The Transportation Research Board Committee on Highway Capacity and Quality of Service approved a motion to incorporate this new approach into the Highway Capacity Manual. The SHRP 2 L04 project also drew on earlier work performed in the SHRP 2 Capacity focus area under a project titled Improving our Understanding of How Highway Congestion and Pricing Affect Travel Demand (SHRP 2 C02). Reliability was introduced into successively richer utility functions, beginning with the traditional variables of out-of-pocket costs and travel time, and progressively adding other variables including travel time reliability. The researchers describe how to place a value on travel time reliability given other relevant terms in the utility function and emphasize that the value of reliability is not a constant; rather, it varies with such factors as vehicle occupancy and household income. This project on incorporating reliability into planning and operation models absorbed important aspects of the earlier research performed within the SHRP 2 Capacity focus area. Finally, a substantial effort was undertaken within this project to provide guidance on how to integrate reliability into a modelling system that uses activity-based models on the demand side and a fine-grained, time-sensitive model on the supply side (e.g., a meso-simulation model). This guidance appears in the project’s reference material report (SHRP 2 Report S2-L04-RR-1: Incorporating Reliability Performance Measures into Operations and Planning Modelling Tools: Reference Material). (Author/publisher)
Samenvatting