Assessing the Impact of Pipe Rehabilitation on Decreasing Watermain Break Rates Using Random Survival Forest Models

Abstract

North America is facing an aging watermain infrastructure crisis. Pipes are surpassing their expected service life, and surveys continue to suggest that the rate of pipe breaks is increasing dramatically. This paper investigates two utilities in Ontario, Canada, that have managed to halt, and even decrease, break rates substantially over the last two decades. Specifically, the impact of adopting early rehabilitation for both cities, including hot-spot cathodic protection, retrofit cathodic protection, lining, early adoption of PVC, and proactive pipe replacement are examined. A random survival forest model is developed for major pipe cohorts for both cities, and using this machine-learning survival analysis model, the numbers of breaks avoided due to pipe rehabilitation are examined. The results indicate pipe rehabilitation techniques avoided 14,572 breaks between 1986 and 2015 for City A and 764 breaks between 1990 and 2020 for City B, and, in the absence of these undertakings, incurred break rates would have been more than 50% greater for both cities. Therefore, pipe rehabilitation is a major factor that has resulted in the declining break rates identified in both cities, suggesting that other cities struggling with increasing break rates should consider similar rehabilitation strategies.