Development of network-level pavement deterioration curves using the linear empirical Bayes approach

ABSTRACT

Modelling the pavement deterioration process is essential for a successful pavement management system (PMS). The pavement deterioration process is highly influenced by uncertainties related to data acquisition and condition assessment. This paper presents a novel approach for predicting a pavement deterioration index. The model builds on a negative binomial (NB) regression used to predict pavement deterioration as a function of the pavement age. Network-level pavement condition models were developed for interstate, primary, and secondary pavement road families and were compared with traditional non-linear regression models. The linear empirical Bayesian (LEB) approach was then used to improve the predictions by combining the deterioration estimated by the fitted model and the observed/measured condition recorded in the PMS. The proposed approach can improve the mean square error prediction of the next-year pavement condition by 33%, 36% and 41% for Interstate, Primary, and Secondary roads, respectively, compared with the measured pavement condition without further modelling of the pavement deterioration.

KEYWORDS:

• Pavement management system
• negative binomial
• linear empirical Bayes
• pavement deterioration models

Acknowledgements

The authors gratefully acknowledge the Virginia Department of Transportation (VDOT) for providing access to the data used in this research.

Disclosure statement

No potential conflict of interest was reported by the authors.