Comparative Review of Water Main Failure Prediction Models: Physical and Data-Driven Approaches

Abstract

Predicting water main failures is a critical challenge for managing aging water infrastructure worldwide. A failure is herein understood as any type of break, such as circumferential or longitudinal cracks or holes, recorded in the utilities’ historical break records. Water pipe break prediction models, whether categorized as physical or data-driven are essential tools in the strategic planning of rehabilitation efforts. While physical models simulate real-world conditions such as soil and traffic loads and material degradation, data-driven models utilize various factors, and historical failure records to predict future water main failures by identifying patterns and correlations within the data. Recent review studies on water pipe failure prediction have predominantly concentrated on data-driven models, with limited emphasis on physical models, particularly despite significant advancements in the past decade. Moreover, a comprehensive comparative analysis between physical and data-driven approaches, including the factors they incorporate, remains unexplored. This study aims to provide a comprehensive overview of the progression of physical models for predicting water main breaks, compare them with data-driven approaches, and identify opportunities for enhancing data-driven models by integrating insights from this comparison. This paper highlights the need to bridge the existing gaps by incorporating new variables into the data-driven models such as detailed soil and pipe material properties, climate-related variables, water quality parameters, and transient/surge pressures, and detailed consideration of load effects into predictive models. By identifying new variables to be incorporated in data-driven models this study finds opportunities to improve the accuracy of failure prediction models and support and support more effective water infrastructure management and decision-making.

Khashei et al. 2025

Journal of Water Resources Planning and Management https://doi.org/10.1061/JWRMD5.WRENG-6866