Enhanced Prediction of Urban Road Pavement Performance under Climate Change with Machine Learning

In light of climate change, increasing traffic demands, and aging infrastructures, flexible pavements face escalating challenges in terms of resilience and longevity. This paper highlights the potential of Machine Learning (ML) to integrate with Mechanistic-Empirical pavement design, aiming to facilitate proactive maintenance and rehabilitation and ultimately enhanced resilience of urban road pavements. A comprehensive analysis comprising 4800 case studies across 10 major Canadian cities was conducted, encompassing various scenarios reflecting climate change pathways, pavement structures, and traffic levels. The findings indicate an increased risk of failure, particularly rutting, under projected future climate conditions. The study demonstrates that developed artificial neural network models exhibit high accuracy in predicting fatigue cracking (R2: 0.96) and rutting (R2: 0.98). Furthermore, it emphasizes the potential of ML techniques in conducting impact assessments and devising strategies for climate change adaptation, considering the evolving landscape of urban complexities.