Abstract Manager

Abstract

Lifecycle cost analysis (LCCA) is widely used in pavement engineering due to its potency in evaluating the total economic viability of various investment options across the entire service life of road pavements. Although current approaches often rely on fixed deterioration models and predetermined maintenance schedules, their accuracy in determining actual pavement performance is often compromised because pavement degradation is highly variable, complex, and influenced by unpredictable real-world factors. The emergence of machine learning offers a promising prospect to address these limitations by enabling the development of more advanced pavement performance prediction models that simulate these scenarios well. Scholarly engagement in this field amidst its emergence remains limited. This study intends to fill this gap by exploring a machine learning approach to applying LCCA (ML-LCCA). By integrating machine learning models with condition-based maintenance strategies for hot mix asphalt (HMA) pavements under Ontario’s freeze-related climate conditions, this research explores how data-driven predictions can replace traditional fixed-schedule approaches to improve both accuracy and cost-effectiveness in pavement management. The study does this by using data from the Long-Term Pavement Performance (LTPP) database, where three machine learning models, Random Forest, Extreme Gradient Boosting (XGBoost), and Artificial Neural Networks, are developed and evaluated for pavement condition prediction. Model performance is then compared using standard statistical accuracy measures to identify the most suitable approach. The best-performing model is subsequently integrated into a condition-based LCCA framework, enabling maintenance decisions to be triggered based on predicted pavement conditions rather than fixed intervention schedules. The study’s findings promise to advance the theoretical and practical understanding of pavement deterioration, highlighting how machine learning-driven condition prediction can influence maintenance timing and life cycle cost outcomes.

Keywords: Machine Learning, Life Cycle Cost Analysis (LCCA) Pavement Deterioration; Hot Mix Asphalt (HMA), Pavement Condition Index (PCI), Freeze–Thaw Climate, Condition-Based Maintenance, Ontario Pavements; Transportation Infrastructure.