ABSTRACT
The paper presents a Markov chain–based procedure for modeling roughness progression in Mexican federal roads at the network level. The IRI has been collected in Mexico for more than 15 years, however, these data have not been used for deterioration modeling. The proposed procedure comprises three steps. First, road sections are classified into families, and IRI time series are prepared for each family. Next, the sign test is used to identify years with likely road works or measurement errors; those years are assumed to delimit deterioration cycles. Finally, Markov chains are generated using data from the identified deterioration cycles. The relevance of the models is evaluated by comparing their predictions with the original time series, and by exploring their sensitivity to deterioration explanatory variables. The obtained results show that the proposed method constitutes a feasible alternative for modeling pavement deterioration in Mexico at the network level.
