Top Factors Leading to Water Main Failures—An Analysis of 12 Canadian Cities

Water main deterioration is a global challenge which can jeopardize the ability of water systems to deliver clean water safely. A variety of factors can affect pipe breakage, yet previous research has focused on different subsets of data. The key objective of this study is to identify the most important factors in predicting water main failure across multiple Canadian cities. The study included data organizing, data cleaning, and dimensionality reduction analyses. Two approaches were applied to analyze the importance of factors affecting water main rate of failure, factor analysis of mixed data (FAMD) and random forest. Data from 12 water utilities across Canada were analyzed, containing information on pipe physical characteristics, historical information, protection activities, environmental factors, and operational condition. Available attributes were different for each city. Overall, the two methods rated failure month and protection as the most important factors. While random forest consistently found age to be the most important attribute, FAMD results indicated that either failure month or material was the most important factor. Random forest results were observed to be biased as they consistently rated numerical attributes as more important than categorical. Thus, FAMD results are expected to be more reliable. Future data collection should focus on pipe characteristics and protection activities. Next steps of the project will include conducting other dimension reduction approaches and running correlation analyses. Findings will support the development of a framework for pipe and break data collection. This will help water utilities develop cost-effective and accurate water main deterioration models, enabling more reliable renewal strategies for water distribution systems.