The effects of traffic mix (the percentage of cars, trucks, buses and so on) are of particular interest in the speed-volume relationship in urban signalized arterials under various geometric and control characteristics. The paper presents some empirical observations on the relation between travel speed, traffic volume and traffic composition in urban signalized arterials. A methodology based on emerging self-organizing structures of neural networks to identify regions in the speed-volume relationship with respect to traffic composition and Bayesian networks to evaluate the effect of different types of motorized vehicles on prevailing traffic conditions is proposed. Results based on data from a large urban network indicate that the variability in traffic conditions can be described by eight regions in speed-volume relationship with respect to traffic composition. Further evaluation of the effect of motorized vehicles in each region separately indicates that the effect of traffic composition decreases with the onset of congestion. Moreover, taxis and motorcycles are the primary affecting parameter of the form of the speed-volume relationship in urban arterials. (Author/publisher).
Abstract