Improving Traffic Assignment Model Using Intersection Delay Function

In this paper a method is proposed to enhance traffic assignment process in traditional travel demand forecasting models. In this method link delay is calculated as a function of flow on that link as well as flow on adjacent links, using intersection delay calculations. This method employs a combination of Frank-Wolf and the method of successive averages to model multi-path vehicle assignment in a reasonable amount of computational time, for small and medium size transportation networks. Inclusion of volume-based intersection delays in regional planning models is not yet widespread in practice. The objective of this paper is to consider the impact of volumebased (dynamic) intersection delays on multi-path traffic assignments in medium size transportation networks. The approach selected is designed to provide more realistic results than those generated by models represents only link delays, due to considering flows on other links. Convergence has been an issue in assignment models that use intersection delay. Nevertheless, convergence has been observed in a reasonable amount of time in our model. The results of this paper suggest that intersection delay, as especially at unsignalized intersections where delays at minor legs are highly dependent on volumes, result in significantly different assignments. It has been also demonstrated that the model achieves a higher levels of calibration and it is more sensitive to intersection level policies.