Performance Model Development for Flexible Pavements via Neural Networks

One of the major goals of pavement management and design is to increase pavement life considering the effects of materials, environment, traffic, and rehabilitation actions. Billions of dollars are required every year for the maintenance and rehabilitation (M&R) of road networks. However, transportation agencies are required to prioritize their M&R actions due to the rate of pavement deterioration and limited budget allocation. Therefore, there is a need for reliable and accurate pavement performance models that can estimate future pavement conditions, identify rehabilitation needs, and analyze rehabilitation impacts. However, the literature review showed that M&R actions are challenging to be incorporated into modeling. This study developed performance models for asphalt pavements considering traffic and climate loads, pavement age, initial roughness condition, and M&R interventions. The data was retrieved from the Long-Term Performance Pavement ( LTPP) program database and an artificial neural networks (ANNs) technique was used for the model development. The developed models efficiently characterized the deterioration behavior of asphalt pavements over time, and effectively capture the effect of M&R interventions. The predicted international roughness index (IRI) values were in good agreement with observed values, and the developed models were found to be reasonably accurate. Therefore, the models developed in this study are suitable for transportation agencies to assess pavement conditions to schedule and prioritize M&R actions for the critical asphalt pavement sections.