Models of traffic flow and traffic safety have traditionally concentrated on either a macro-level of analysis where vehicles are seen as particles and no account is taken of individual driver reactions, or at a micro-level where only the driver's reactions to the immediate surroundings are considered. Models at a meso-level which link these two aspects have not previously been developed. Meso-level models focus on interactions between road users, with consideration for both traffic stream and driver characteristics. These models are necessary if we want to incorporate safety criteria in traffic control systems. The paper describes the work done to develop such a model and to derive criteria which predict the point at which traffic flows become particularly sensitive to disturbances which can lead to accidents. The purpose of this model is to identify points at which interventions aimed at drivers can still be made to avert accidents. (Author/publisher)
Abstract