Big data-driven prediction of watermain failures in semi-tropical regions: Case study of Hong Kong's distribution network

Watermain failures universally challenge urban infrastructure, causing water loss, service disruptions, and maintenance challenges. There is a lack of research attention towards semi-tropical regions, which have unique environmental conditions affecting the deterioration of watermains differently. This paper develops a big data-driven prediction model for watermain failure types (no leak, leak, no burst, burst) in semi-tropical regions. While using Hong Kong's water distribution network as a case study, eight key factors are analyzed through correlation coefficients. Ordinal logistic regression models are developed using a dataset of over 1 million assets. The models achieved prediction accuracies of 72.8 % for leak/no-leak and 75.7 % for burst/no-burst conditions. Pipe material and soil corrosivity emerged as the most significant predictors. These findings provide water utilities in semi-tropical regions with a practical tool for proactive maintenance planning. Future research can incorporate additional environmental variables and expand the model's application to other regions for enhanced generalizability.