VARIABILITY OF TRAVEL TIMES CAUSED BY DELAYS AND DISTURBANCES ON LINKS AND NODES AS A QUANTITATIVE DESIGN CRITERION FOR RELIABLE TRAFFIC NETWORKS

[Pieter Vansteenwegen]

for Chris Tampère:

1. Abstract of the planned research
Road traffic networks are characterized by increasing levels of congestion, resulting in unreliability of travel times. The objective of this research is to gain fundamental insight into contributors to variability of travel times and to use this insight for the improved design of traffic networks. The focus is on amplifications of exogeneously imposed fluctuations in demand and/or capacity due to the non-linear dynamic behaviour of multi-commodity traffic networks. In order to find vulnerable network elements, we gain in-depth knowledge about the causes of non-linear dynamics: network topology, origin-destination pattern, and (flexible) route choice behaviour. Apart from these causes, special attention is paid to traffic operations at intersections and their (non-linear) impact on travel time variability. In this respect, one crucial aspect is how traffic flow operations on intersections – both signal controlled and priority controlled (incl. roundabouts) – are affected by queues spilling back from downstream links (‘multi-regime’ traffic flow operations). The research proceeds along two tracks:
• quantitative assessment of travel time reliability and analysis of the influence of the above mentioned aspects; this quantification lays the foundation of a design methodology for 'reliable' traffic networks;
• empirical research and modelling of multi-regime traffic operations on intersections, as sources of (irregular) delays in traffic networks.

2. Problem statement and scope
This project proposal focuses on the identification of network elements (especially nodes) having large influence on travel time variability, and ways to design networks in order to minimize travel time variability.
Travel time variability is the result of disturbances in demand or supply (due to the stochastic character of ‘regular’ traffic or due to incidents, e.g. capacity reductions due to weather, accidents). Interestingly, the impact on travel time variability is not independent of the location, time or type of the disturbance. Moreover, very often the impact of disturbances on travel times is non-linear: under certain conditions, doubling the severity (duration, capacity reduction,…) of an incident might increase total travel times by a factor much larger than two. This amplification effect is due to the strongly non-linear dynamic behavior of flows in heavily loaded traffic networks. At least two important characteristics of traffic flows in networks are responsible for this non-linear behavior: congestion spillback and interactions at nodes (junctions, intersections). An irregular surge in demand or reduction of capacity might cause jams to spill back onto upstream junctions and links, hence influencing the travel times of traffic that has no direct relation to the original disturbance. Especially at nodes that are close to saturation, where delays are known to be strongly non-linear with demand, such influences might further amplify the impact of the original disturbance.
The generation of disturbances, their impact, as well as ways to control them is a separate area of research, including for instance: incident management, adequate traffic control suppressing the impact of disturbances, adequate planning of road works or manifestations causing demand surges. However, disturbances are considered exogenous inputs in this proposal. The central question then is: how and under which conditions can non-linear traffic dynamics cause these disturbances to have disproportional impact on travel time variability? To answer this question, the impact must be studied of network topology, demand structure (OD-table), route choice (both prior to and during the trip), behavior of multi-commodity traffic flow on links, and the lay out of and behavior of multi-commodity traffic at junctions and intersections. Finally, the ambition of this project is to apply the new insights into this complex matter for the design of traffic networks that are less susceptible to disturbances and hence provide more reliable travel times.