Barbara W.Y. Siu

Equilibrium Trip Scheduling in Congested Traffic under Uncertainty

Whilst traditional modeling and research efforts put emphasis on the effect of congestion on travel choices, recently more and more studies examine the notion of travel time uncertainty on travel decision. In particular, travel time reliability forms an important component in modeling route choice and trip scheduling. This paper considers equilibrium trip scheduling under random travel delay in a single bottleneck. Travelers form a heterogeneous population with distinctive requirements on the probability of punctual arrival and each with a different valuation toward earliness and lateness. The analysis shows that random delay plays a significant role in travel costs and introduces substantial differences in the queuing pattern, departure and arrival times, as compared with the traditional deterministic model.

Punctuality-based departure time scheduling under stochastic bottleneck capacity: formulation and equilibrium

This paper proposes a departure time scheduling model for a single bottleneck that is subjected to degradable capacity. Instead of considering travel cost due to uncertainty (either in form of schedule delay or travel time variance) as an additional cost term to be traded with congestion delay and other out-of-pocket costs, it is supposed that travellers deal with uncertainty by adjusting their departure times in order to achieve a certain desirable level of punctuality in accordance to their behavioural preference and trip characteristic. This paper develops the formulation in solving for the scheduling equilibrium, with a detailed exposition on how the expected travel time and schedule delays can be computed. Moreover, this paper discusses the existence of equilibrium of the proposed scheduling equilibrium. In the deterministic bottleneck model and subsequent works that extend to the case of stochastic bottleneck, it is often assumed that the start time of congestion (first appearance of queue) is fixed; in this paper, it is demonstrated that the congestion start time is in fact dependent on the random bottleneck capacity. Lastly, some interesting properties regarding the equilibrium departure profile are discussed and compared with the deterministic bottleneck results.

Punctuality-based route and departure time choice

Predominantly, existing dynamic traffic assignment studies presume that travel time is deterministic, merely subjected to congestion due to capacity limitations. In view of the inevitability of travel-time uncertainty, a lot of effort has been spent to investigate how uncertainty affects travel choices. Extending from the bottleneck scheduling model, this article establishes the connection between trip scheduling and punctuality reliability by considering that travellers value earliness and lateness differently according to their different degrees of punctuality reliability. Punctuality reliability refers to the probability of being not late for a scheduled activity, which is heterogeneous among travellers and depends on their degrees of risk aversion. By incorporating the notion of punctuality reliability, we can produce the sensible result that risk-averse travellers (those with higher punctuality reliability) choose to depart from home at earlier times, while such a mapping or feature is absent in the original model by Small [1982. The scheduling of consumer activities: work trips. American Economic Review, 72, 467–479]. The proposition is confirmed by our empirical study. The modelling framework is then demonstrated numerically first to the scheduling problem in a single bottleneck then to parallel bottlenecks that offer route choices.

Time-Dependent Transport Network Design and Urban Development

In the long run, there will be changes in the population, socio-economic factors, land-use, etc. within a region. To cope with these changes, it will be necessary to improve the transportation network and alter the tolls, if any, accordingly over time. In this study, we extend the current technique in network design as a static, one-time event to an approach that explicitly captures the time dimension in the network design problem. Thus, we can make optimal decisions on project initiation time, scaling, and phasing in accordance to the design objectives. Through a small network example, we demonstrated that a well-planned, sequential upgrade plan performs better than any one-time network improvements.