Transition probability matrices for pavement deterioration modelling with variable duty cycle times

Abstract

Probabilistic pavement models, with Markov chains as the most widely used type, are considered to capture an accurate representation of the in situ pavement performance. Homogeneous Markov chain models present the same transition probability matrix (TPM) for all the transitions of the period and require data from multiple duty cycles of one or two years. The aim of this paper is to explore the feasibility of developing homogeneous Markov chain models with variations of the duty cycle (in increments of either one or two years). Without considering maintenance and rehabilitation works, this research found that TPMs for a one-year duty cycle can be calculated from the two-year duty cycle, without a noticeable effect on accuracy using International Roughness Index (IRI) values from the Spanish State Road Network. However, for developing coherent TPMs, two primary assumptions were made: (1) heavy vehicle traffic volumes determine the traffic category (TC), and (2) only roads from the same climatic region were modelled. The satisfactory results verified the validity of the methodology and overcame the disadvantages of homogeneous Markov models. Furthermore, the results suggest that pavement sections are adequately designed in Spain for each TC because of the similar deterioration patterns.