Markov chain pavement deterioration prediction models for local street networks

Purpose: Pavement deterioration prediction models play a crucial role in determining maintenance strategies and future funding needs. While deterioration prediction models have been studied extensively in the past, applications of these models to local street networks have been limited. This study aims to address this gap by sharing the results of network level deterioration prediction models developed at a local level. Design/methodology/approach: Network level pavement deterioration prediction models are developed using Markov chains for the local street network in Syracuse, New York, based on pavement condition rating data collected over a 15-year time period. Transition probability matrices are generated by calculating the percentage of street sections that transition from one state to another within one duty cycle. Bootstrap sampling with replacement is used to numerically generate 95% confidence intervals around the transition probability values. Findings: The overall local street network is divided into three cohorts based on street type (i.e. avenues, streets and roads) and two cohorts based on pavement type. All cohorts demonstrated very similar deterioration trends, indicating the existence of a fast-paced deterioration mechanism for the local street network of Syracuse. Originality/value: This study contributes to the body of knowledge in deterioration modeling of local street networks, especially in the absence of key predictor variables. Furthermore, this study introduces the use of bootstrap sampling with replacement method in generating confidence intervals for transition probability values.