Probabilistic Service Life Model of Pavement Marking by Degradation Data

Abstract

Pavement markings are important features for improving road safety. As these markings must possess adequate retroreflectivity, road agencies are concerned about retroreflectivity maintenance. Consequently there is an increased focus on modeling the durability of pavement markings. Such models use time-to-failure data from life test experiments. The current study proposes a new method that consists of three steps: (i) designing an accelerated experiment for data collection; (ii) fitting a pavement marking retroreflectivity model using a generalized linear mixed model and the degradation data from the planned experiment, confirming the goodness of fit of the fitted model through the residual analysis; and (iii) conducting Monte Carlo simulation (with at least 10,000 runs of simulated vectors of the fitted model’s coefficients) to obtain the pavement marking service life distribution for a failure retroreflectivity threshold level. The simulation of service life of pavement marking provides relevant information, such as service life expectation and empirical quantities of interest (as median). The proposed method is applied to data on retroreflectivity degradation obtained from an experimental test site, a Brazilian highway with a high traffic volume, with two fixed factors (waterborne paints and glass bead application rates, both with three levels). The model indicates that retroreflectivity decreases at an average rate of 2% per day, and how materials differ from each other. The proposed model is easily implementable and can help management teams to adequately plan the maintenance time for pavement markings.