Edinburgh Napier University

  This paper presents details of the development and implementation of a stochastic-based performance-prediction model using the Markov chains method for road network pavement maintenance. As well as the difficulty in the process of developing the transition probability matrix (TPM) on the basis of historical data, the way of presenting the results and their interpretation become challenges for the use of stochastic-based performance prediction models. The analysis uses a database developed by the State of Victoria, Australia, consisting of 2197 road sections. Predicted distributions of network-level pavement conditions after maintenance treatments based on Markov chain principles are presented, and their value and utility are discussed. Methods of visual presentation and assessment of stochastic-based performance prediction of different maintenance actions on different pavement types are put forward. The analyses show that a level of maintenance strategy higher than routine maintenance is required in worst road states/conditions. A steady-state analysis of the embedded Markov chains is proved to be effective in assisting the decision process when applied, and has been tested with the selected actual sample to confirm the prediction results. This paper provides tools for road authorities to select optimal maintenance measures based on a more informed network-level performance prediction model. This approach has been shown to be proficient given the uncertainty of pavement behaviour