Multistate Travel Time Reliability Models with Skewed Component Distributions

Abstract

The multistate travel time reliability model has demonstrated superior performance, a close relationship with underlying traffic conditions, and ease of interpretation for travel time reliability reporting. This study advances the multistate model by using skewed component distributions—for example, the gamma and lognormal distributions—to accommodate nonsymmetrically distributed travel times, which are commonly observed in congested states. Six alternative models, the single-state normal, gamma, and lognormal distributions and their multistate versions, were fitted to field-collected data. The performance of the models was compared with Akaike's information criterion. The results indicate that the multistate lognormal model consistently out performs alternative models, especially during peak hours. The improved fitting of the lognormal model is mainly reflected in the mode and tail portion of the data distribution. During off-peak hours, the single-state model could provide a compatible but parsimonious alternative to multistate models. The impact of using the multistate lognormal model is demonstrated on travel time reliability reporting by using the travel time buffer index. The study concludes that the multistate lognormal model is the optimal alternative model for modeling travel time under moderate to heavy traffic conditions along freeways.