This paper first discusses the limitations of mathematical models for studying car following behaviour in an urban network. It then describes an urban network "event scanning" simulation package in which each driver decision is taken into account as an "event", and the results of the different events are forecast at the time the event itself happens. This kind of simulation is particularly suitable for conflict analysis, or when it is necessary to know which precise event caused a particular problem. The paper also describes a car following experiment using an instrumented Renault 12 car over 8 journeys in and around Paris. The average intervehicle spacing, speed, braking period and driver reaction time were measured. The average reaction time and intervehicle spacing distance were used to validate the car following rules used in the simulation. (A) This paper is also published in Seminar E, (Volume P307) p27- 39 (see IRRD 842406).
Samenvatting