Modeling the pothole generation process by composite statistical models for pavement management on expressway bridges

It is empirically known that the generation frequency of potholes on bridge surface pavement is influenced by deterioration state of bridge decks. On the basis of this concept, in this study, data analysis methods to estimate the frequency of generating potholes is proposed. Specifically, the authors develop a statistical model of a composite process including the generation of potholes and the deterioration of decks. The process of generating potholes applies Poisson process while the process of deteriorating decks applies Markov process. Moreover, posterior distribution of the parameters of the complex models is estimated by using Bayesian methods and Markov chain Monte Carlo methods, incorporating the inspection data both on pavement and on decks into the model. On the other hand, these two kind of inspection data are not always acquired at the same time due to the nature of actual inspection works. Therefore, the model estimation methods as correcting such a systematic sampling bias are developed. Lastly, the effectiveness of the methodology proposed in this study is discussed for inspection data on existent expressway bridges. As a result, the generation rate of potholes is estimated by using the deterioration rate of decks while the deterioration rate of decks is estimated by using the generation rate of potholes. The result in this study contributes to reduce inspection frequency and costs rationally, leading to efficient pavement management.