Search results

A bi-level programming approach has been used to tackle an area traffic control optimisation problem subject to user equilibrium traffic flows. In the upper level problem, the signal timing plan for coordinated fixed time control has been defined. In the lower level problem, user equilibrium traffic assignment obeying Wardrop's first principle has been formulated as a variational inequality problem. Mathematical expressions for various components of the performance index in the upper level problem and the average delay in the lower level problem have been derived and reported (Chiou 1997a). A mixed search procedure has been proposed as the solution method to the bi-level problem and a range of numerical tests have been carried out (Chiou 1997b, 1998a,b). In this paper, further numerical tests are performed on Allsop and Charlesworth's (1977) road network in which various traffic loads are taken into account. Effectiveness in terms of the robustness and reliability of the mixed search procedure in congested and uncongested road networks is thus investigated further. Comparisons of the performance index resulting from the mixed search procedure and that of mutually consistent TRANSYT-optimal signal settings and traffic flows are made for the congested road network.

This paper investigates the temporal inflow profile that minimises the total cost of travel for a single route. The problem is formulated to consider the case in which the total demand to be serviced is fixed. The approach used here is a direct calculation of the first order variation of total system cost with respect to variations in the inflow profile. Two traffic models are considered; the bottleneck with deterministic queue and the kinematic wave model. For the bottleneck model a known solution is recovered. The wave model proves more difficult and after eliminating the possibility of a smooth inflow profile the restricted case of constant inflow is solved. As the space of possible profiles is finite dimensional in this case, the standard techniques of calculus apply. We establish a pair of equations that are satisfied simultaneously by the optimal inflow and time of first departure.

In this paper we consider two traffic control strategies relying on user response to information and/or flow restriction. Ultimately, the control strategies are designed to function in real time, hence provide command values based on actual conditions and requiring little computational effort. The proposed control strategies are based on the idea that the network load, as measured by instantaneous travel times for instance, should be shared as equally as possible between paths. In order to achieve such an aim, the commands are designed to make the system state converge towards a state in which instantaneous travel times of paths relative to any given OD tend to be equal.

The estimation of queue length and delays in queues that are oversaturated for some part of a study period is of substantial importance in a range of traffic engineering applications. Whiting’s co-ordinate transformation has provided the basis for several approaches to this. We analyse this approach and present an explicit form for the derivative of queue length with respect to time, which we then use to establish various properties. We also report the results of numerical comparisons with exact formulae for certain special cases and show that these offer little or no advantage over the co-ordinate transformation approximations and can be computationally impractical in study periods of moderate duration.

Purpose – Departure time choice not only depends on the desire to carry out activities at certain times and places; it is a complex decision making process influenced by travel conditions, congestion levels, activity schedules, and external trip factors. To estimate departure time choice models capturing the factors influencing it in appropriate form, a complex data collection procedure allowing to obtain detailed input data from different sources and at different time periods is required.

The main aim of this chapter is to describe and discuss the survey methodology we used in a time-of-day choice project, involving the collection of revealed preference (RP) and stated preference (SP) data to estimate hybrid discrete departure time choice models incorporating latent variables. Preliminary model results are also presented as an example.

Methodology/approach – Data was obtained from 405 workers at different private and public institutions located in the centre of Santiago, Chile. The survey process had three different stages and used various collection methods (e-mail, web-page, and personal interviews at the workplace) in order to satisfy efficiency, reliability and cost criteria.

The RP component survey design was based on the last origin-destination survey implemented in Santiago (i.e. a travel diary filled under an activity recall framework). Relevant level-of-service measures at different time periods were obtained from GPS data measured from instrumented vehicles in the public and private transport networks. A SP-off-RP optimal design considering dependence among attribute levels was also developed. Finally, several 1–7 Likert scale questions were included to incorporate the latent variables.

Findings – The survey methodology described in this chapter represents a successful experience in terms of collecting high quality data, from different sources, with the aim of estimating appropriate time-of-day choice models. The data collection process was carried out in different stages, by means of web pages, email, and personal interviews. The data was further enriched with level-of-service attributes measured at different times of the day with unusual precision. Preliminary results reported in this chapter show that data obtained through this methodology are appropriate to model time-of-day choices.

Originality/value of chapter – The novelty of the survey methodology described in this chapter is the collection of data of a different nature for time-of-day choice modelling through the integration of different collection techniques.

Acquisition of very precise information about preferred departure/arrival times, level of service at different times of the day, detailed information about flexibility in schedules, employment information and attitudes towards departure times, should allow practitioners to estimate hybrid time-of-day choice models incorporating latent variables.