In light of the increasing and pressing need to efficiently manage scarce water resources, there has been renewed interest by water distribution network owners to develop and implement water management strategies and tools that would assist in the integrated and automated management of those networks. Such asset management strategies should assist the network owners to evaluate the condition of the water distribution network, assess historical incident data (leakage or breakage) and risk of failure, visualise areas of high risk, propose “repair or replace” strategies and prioritise the work based on the inherent risk and cost of action. The methodology and support system outlined in this paper can form an integral part of a leakage management strategy and provide a useful decision-making tool. The work presented outlines an integrated methodology and a decision support system for arriving at such “repair-or-replace” decisions, as part of a long-term pipeline asset management program that could be undertaken by a water utility to improve on the reliability of the water distribution networks.

Sustainable management of urban water distribution networks should include not only new methods for monitoring, repairing or replacing aging infrastructure, but also (and more importantly) expanded methods for modelling deteriorating infrastructure, for pro-actively assessing the risk of failure and for devising replace or repair strategies. The study presented herein describes a framework for proactive risk-based integrity monitoring of urban water distribution networks and the results obtained from a case-study based on a 5-year data sample. A combination of artificial neural network and statistical modelling techniques stemming from parametric and nonparametric survival analysis (Kaplan-Meier survival curves with Epanechnikov's kernel) are utilized in the investigation of identified risk factors and for estimation of the forecasted time to failure metric. The data is stratified for different pipe groups for a more targeted analysis.