Application of cluster analysis and Markov chain model for network-level highway infrastructure management

Khyber Pakhtunkhwa province, Pakistan’s third-largest provincial economy, is totally dependent on road transport. To upkeep the highway infrastructure in good condition, an effective, efficient, economical, and sustainable management technique needs to be devised. The pavement deterioration prediction is a vital element for an efficient pavement management system that can help the road authorities in determining the appropriate maintenance technique. The research aimed to develop the pavement deterioration prediction model at the networklevel using Markov Chain. However, at the network-level, the road characteristics are not uniform due to which it is difficult to classify the road sections into families for applying Markov Chain. Thus, first Agglomerative Cluster Analysis was applied to define pavement families with similar characteristics. Then, the deterioration prediction models were developed for pavement families using Markov Chain. The combination of AHC with Markov Chain found was found an effective technique for pavement deterioration models at the network-level.