Evaluation of Clustered Traffic Inputs for Mechanistic-Empirical Pavement Design: Case Study in New Jersey

Abstract

Truck traffic is one of the significant inputs in design and analysis of pavement structures. This paper focuses on comprehensive cluster analysis of truck traffic in New Jersey for implementation of mechanistic-empirical pavement design. Multiple year traffic data were collected from a large number of weigh-in-motion stations across New Jersey. Statistical analysis was first conducted to analyze directional and temporal (yearly) variations of traffic data. Hierarchical cluster analysis was conducted and three optimum clusters were found for axle load spectra (single, tandem, tridem), vehicle class distribution, and axle/truck ratio, respectively. Road functional classifications were employed to identify different clusters as no common geographic trend could be perceived. The results illustrate that the predicted performance using the site-specific traffic data is comparable with that using the traffic cluster for the selected 10 sites. Among four different traffic inputs, the cluster traffic inputs generated the closest predictions of pavement life as compared with those using site-specific traffic input and the default traffic inputs yielded the highest error. It is recommended to use traffic clusters in mechanistic-empirical pavement design when site-specific data is unavailable.