PROBABILISTIC BEHAVIOR OF PAVEMENTS

Since the AASHO road test of 1962, tremendous efforts have been devote d to improve the methodologies of pavement performance prediction. The successful implementation of the Network Optimization System (NOS) in the Arizona Department of Transportation (ADOT), in the 1980s, repres ented an advancement in the prediction methodology by using Markov-pro cess-based transition-probability matrices (TPMs) to define the transi tion process of pavement conditions. This paper addresses some of the inadequacies of the original NOS prediction model. Two approaches were used to evaluate the transition probability matrices. First, the curr ent pavement performance data base was used to develop new TPMs. Secon d, the Chapman-Kolmogorov method was used to examine the logical exten sion of the transition probability matrices from a single step to long -term pavement behavior. As a result, the concept of pavement probabil istic behavior curves (PBC) was established. The newly generated TPMs were modified with accessibility rules to improve the prediction of pa vement performance. More importantly, it was demonstrated that Markovi an prediction satisfactorily models actual pavement behavior. The deve lopments presented in this paper improve the reliability of the microc omputer-based NOS.