Empirical modeling of pavement overlay crack progression with field data

We present the development of an empirical pavement overlay crack progression model using condition survey data for the highway system in the state of Washington. The crack progression model uses random-effect panel data regression techniques, with correction for incidental truncation, endogeneity bias, and unobserved heterogeneity. The parameter estimation results show that existing cracking prior to the overlay, traffic loading, overlay thickness and materials, and some environmental factors play important roles in explaining crack progression. The model developed in this paper is used jointly with a probabilistic model of overlay crack initiation as inputs to a Monte Carlo simulation of overlay cracking trends over time. These trends are realistic for overlays in the estimation database, but caution should be used when applying the model to other states. Specifically, it should only be used in situations where the maintenance policy used is similar to that used by the Washington DOT.