Roughness Model Accounting for Heterogeneity Based on In-Service Pavement Performance Data

Abstract

Deterioration modeling is a key component in a transportation infrastructure management system. It serves to capture and predict the performance of a facility. A sound deterioration model should incorporate: (1) relevant variables that affect deterioration process; (2) physical principle that reflects deterioration mechanism; and (3) rigorous statistical approach to estimating the model. This paper addresses these critical aspects with particular focus on highway pavements. Data collected from in-service pavement sections as part of the Minnesota Road Test project are used to capture the real-world pavement deterioration process. A number of variables involved in the system are thoroughly investigated, which include pavement design, materials, traffic, environment, and maintenance factors. Performance uncertainty due to unobserved heterogeneity is incorporated in the proposed model. The unobserved heterogeneity stems primarily from variability in the materials and the construction process. The model is estimated through maximum simulated likelihood estimation. It is demonstrated that the results are consistent with observations and engineering judgment in the context of pavement design and performance. In addition to population-level parameters representing the general deterioration mechanism, section-specific or individual-level parameters are obtained through Bayesian approach. The two levels of parameters can be used to accommodate network- and project-level analysis in pavement management.