This paper describes DRACULA, a new kind of urban traffic assignment model under EPSRC-funded development since mid-1993. The major difference from conventional approaches is that DRACULA models the day to day evolution of traffic conditions, rather than some abstract long-term equilibrium state. Drivers respond to congestion through a learning sub-model, and daily select routes and departure times using a behavioural choice sub-model. A demand sub-model represents day-to-day variability in the network capacities (e.g. due to accidents). Finally, a traffic sub-model is used to move vehicles through the network on each day. Various forms and levels of detail of these sub-models may be selected by the user. At its most detailed level, DRACULA represents (i) the choices and experiences of individual drivers, and (ii) the movement of individual vehicles through the network ("microsimulation"), as these evolve in real-time and from day-to-day. The linking, for the first time, of such a detailed traffic model with an assignment capability has particularly important applications in the modelling of responsive signals (e.g. SCOOT) and environmental outputs, such as exhaust emissions. The application of DRACULA in an EC-funded study of driver information systems is presented as an example. (A)
Samenvatting