An approach to predict the failure of water mains under climatic variations

Abstract

Urban water distribution systems are critical infrastructures, and their failure can lead to significant economic, environmental, and social losses including flood streets and loss of treated drinking water. Identifying the failure patterns of water mains over time under various conditions is an inexpensive approach for estimating the structural deterioration of water distribution systems. It is also an alternative method for direct inspection that requires intensive efforts and budget. Time-dependent factors such as temperature and precipitation variations can lead to changes in frost depths and ground movements, resulting in stresses that exceed design values and increasing the potential of water main failures. A few studies have addressed the impact of climatic variations on the failure prediction of water mains. To fill this gap, a temporal approach for the failure prediction of water mains under climatic variations is presented. The proposed approach can predict the failure of water mains at selected locations (not only one location) and allow to not only predict the failure by a one time-step ahead but also obtain accurate failure predictions up to 9 months ahead. Another purpose of the proposed model is to accommodate additional variables to predict the failure of water mains at selected locations. To achieve this objective, a vector autoregression model with exogenous variables that incorporates the impact of climatic variations was developed. Spatiotemporal data of water mains failure events and climate data are collected for this study from Quebec and Ontario, Canada. Monte Carlo method was applied to validate the reliability of the predictive model. In other words, the failure prediction of water mains uncertainties was generated using Monte Carlo simulation. Results show that climatic variations can provide valuable information for the failure prediction of water mains. Results also prove that the proposed model can accurately predict the temporal failure patterns of water mains at two water distribution systems simultaneously.