Bayesian model averaging for the prediction of water main failure for small to large Canadian municipalities

Water utilities often rely on water main failure prediction models for developing preventive or proactive repair and replacement action programs. Due to inherent uncertainties in modeling, it is challenging to understand the water main failure processes and to predict the failure effectively. In this study, Bayesian model averaging (BMA) method is presented to identify the influential covariates and to predict the failure rates of water mains considering model uncertainties. To accredit the proposed model, it is implemented to predict the failure of pipes of the water distribution network of the City of Kelowna, BC and Greater Vernon Water, BC, Canada. Results indicate that the proposed BMA approach captures the effect of the potential explanatory variables more effectively through the posterior probabilities in contrast to that of the p-value given by the classical regression analysis. Moreover, BMA approach performs better compared to classical regression analysis when limited pipe failure data are available.