Numerical Characterization of Gross Vehicle Weight Distributions from Weigh-in-Motion Data

Abstract

This paper presents a method for quantifying gross vehicle weight (GVW) distributions of commercial vehicles using weigh-in-motion (WIM) data. Finite mixture models are used to fit a combination of normal distributions to the overall GVW distribution to identify peak parameters more precisely. The GVW distribution is commonly bimodal or trimodal with prominent peaks occurring in the loaded and unloaded weight ranges. The GVW characteristics of FHWA Class 9 vehicles are commonly used for assessing WIM accuracy by visual interpretation of frequency histograms. Temporal changes in the GVW distribution are difficult to detect using common visualization techniques. Mixture models enable the statistical identification of the modal peaks and the proportion of traffic belonging to those peaks for ongoing monitoring purposes. Mixture models are applied to a WIM site in Indiana to illustrate the analysis method and benefits. Numerical monitoring of the GVW distribution is shown to have some advantages over the widely accepted metric based on the Class 9 steer axle weight. The proposed metric should not be used singularly, rather as an additional tool to complement existing metrics.