Cluster Analysis of the LTPP Data of NM State to Estimate MEPDG Traffic Input and Sensitivity Analysis

Abstract

Mechanistic-Empirical Pavement Design Guide (MEPDG) uses vehicle type, load spectra and number of axle to characterize traffic inputs for the pavement designs, as an alternate of the previously used equivalent single axle load (ESAL). Long term pavement performance (LTPP) study has 8 sites in New Mexico (NM) which collects the weigh-in-motion (WIM) data all over the year. These WIM collected data can be used as the traffic inputs for a reliable pavement design using the MEPDG. In this study the LTPP traffic data were extracted and then clustering technique was used to find out the similarities between different sites. Then the traffic data from different cluster was compared to the default values in the MEPDG software. The monthly adjustment factors (MAF), vehicle class distribution (VCD) and axle load distributions for different cluster deviated from the MEPDG’s default values. The effect of these differences on the pavement performance for New Mexico conditions was established through a sensitivity analysis. It was found that longitudinal cracking is mostly sensitive to MAF. Alligator cracking is found sensitive to axle load distribution. Whereas rutting was found sensitive to all the variables studied, however international roughness index (IRI) was not significantly affected by any of the variables in this study.