Comparisons of Deterministic and Stochastic Traffic Loading Models

Abstract

Stochastic route choice models have been proposed as a more realistic alternative to the simpler deterministic route choice representations in models for loading traffic onto a network. However, studies on stochastic route choice models have thus far been limited to small experimental networks. Because stochastic models are in general considered to be more realistic in their assumptions than deterministic models, it becomes pertinent to examine their implementation and the corresponding traffic loading on a network of realistic size. Such an implementation for two stochastic route choice models is examined. The results indicate that the differences among the models, at the network level, for predicted link flows and average travel times between origin and destinations are not major, and therefore, deterministic models may suffice for long-range network planning. However, stochastic models predict the usage of more links in the network, and the difference in the flows at the link level may play an important role in travel advisory and route guidance systems. Some suggestions for future research are presented.