Developing a model to predict water main failures

Many water authorities own water mains that date to the beginning of the last century. As these assets begin to reach the end of their useful lives, failures are becoming increasingly more common. With only limited financial resources available, water authorities are faced with the challenge of whether to repair or replace damaged water mains. A useful decision making tool in this area is a water main failure prediction model. ActewAGL is a water utility operating and maintaining the water supply network in the ACT for the asset owner Actew Corporation. The network consists of in excess of 3000 kilometres of water mains and services approximately 360,000 customers. Although not currently considered a problem, the water main failure rate in the ACT appears to be on the rise. Therefore, Actew recognizes the need to better understand the causes of water main failures and to be proactive in developing policies to most efficiently deal with the effects of failure. The objective of this study was to investigate the methods, parameters and theory used in existing water main failure models and to use these findings to develop a customised water main failure prediction model taking into account limitations with the type and quality of water main failure data available.As a result of this study a number of recommendations were made for further work in order to improve Actew’s ability to monitor system performance and predict water main failures. These are as follows. • Compile historical failure records into a central water main failure database and add links to the GIS and asset management systems. • Incorporate physical parameters into water main failure data collection processes. • Conduct further testing, validation and improvement of the multivariate statistical models developed in this study. • Conduct preliminary investigations into the development of a physically based water main failure prediction mode