Further Aspects of Traffic Arrival Patterns on a Multilane Highway

The model presented in this study describes the arrival pattern of vehicles on a multilane unidirectional highway. It is based on the assumption that in discrete time intervals, the vehicle arrival patterns can be described as a Markov chain process. The Markov properties conveniently enable the expression of interdependence of vehicle arrivals within and among lanes. From the empirical data collected, it was found that in the fast lane, the arrival process is overdispersed and the tendency for bunching is greater than under Poisson distribution conditions. The arrival process in the slow lane is underdispersed and the vehicles tend to distribute themselves deterministically along the lane. The cross-correlation function between the arrival processes in two adjacent lanes indicates a clear dependence between the two processes. This dependence might limit the number of potential overtakings, increases the friction between lanes and increases the probability for a side collision.