Water distribution networks (WDNs) and road networks are crucial systems in modern cities, exhibiting a coupling effect. This study investigates how traditional pipeline attributes and traffic-related factors influence pipeline failure rates. Using a comprehensive dataset from a district in China, including GIS data and failure records, we assess the impact of factors such as pipeline diameter, depth, material, age, road classification, and intersection presence. Employing a LightGBM model and SHAP values, we find that both traditional and traffic-related factors significantly contribute to pipeline failures, with intersections, road presence, and pipeline-road angle being particularly influential. These insights highlight the need for integrated infrastructure management strategies considering both pipeline attributes and traffic conditions to enhance urban water supply system resilience and reliability. Future research should validate these findings in different contexts and explore specific stress dynamics at intersections to refine predictive models and mitigation strategies.