Risk analysis and traffic signal design

Lam, Wa-Kwai (1989) Risk analysis and traffic signal design. University of Southampton, Doctoral Thesis.

Abstract

In current traffic signal design philosophy, the signal settings of isolated intersections are usually determined from using single estimates of the key traffic parameters. There are no considerations of the possible impact that may have been caused by the traffic variabilities and uncertainties surrounding the estimates. These pre-timed signal settings may be used in a range of circumstances subject to stochastic fluctuations of traffic. Whilst the design objective is usually to optimise the operational effectiveness, these signal settings fail to fulfil such an objective when the traffic variables are subject to substantial variabilities, in spite of the uncertainties associated with the estimated values. In conventional design methods, the variabilities and estimation uncertainties inherent in the traffic signal parameters are usually ignored. This research has examined in detail the impact of risks and uncertainties in traffic signal designs. A new approach using Risk Analysis Techinque was proposed to tackle such uncertainties arising in practice. The Risk Analysis Delay Model (RADM) developed in this study utilizes probabilistic simulation to assess the performance of a signalised intersection subject to risk and uncertainty. RADM has been applied to explore the impact of risks and uncertainties on an example and real-life signalised intersections. The intensive studies have affirmed the importance of acknowledging traffic variabilities and estimation uncertainties in traffic signal calculations. Notably, the greatest effect on the efficiency of a signal was mainly governed by the inherent variabilities of traffic demand and saturation flow. The extensive assessments on the real-life intersection have also examined the likely impact caused by some possible areas of risk and uncertainty which might arise in practice.

This record has no associated files available for download.

Contributors

Download statistics