Establishment of Network-Level Pavement Deterioration Models of Asphalt Overlays Using Markov Chains

This paper presents the establishment of Markov models to analyze the deterioration of asphalt pavement overlays. The regular Markov chains were first developed, and the differences between homogeneous and nonhomogeneous models, as well as between two-state and three-state transitions, were investigated. An optimized Markov chain was also developed using arithmetic and geometric assumptions for model simplification, and two methods were proposed to predict alligator cracking. The research revealed that the optimized Markov chain is effective and provides good prediction capabilities. It also proved effective in incorporating site-specific properties such as overlay thickness, average daily traffic (ADT), and truck percentage. In terms of application, the method was capable of predicting other indices and was employed to reasonably adjust the trigger values for pavement overlays. This study demonstrates that the Markov chain has strong potential for pavement distress prediction and supporting pavement maintenance.