Using a break prediction model for drinking water networks asset management: From research to practice

Abstract

Break prediction models can help water utility decision-makers to build pipe rehabilitation programs. For many years, using them has been a specialist matter. After more than 15 years of research into the ageing of water pipes, Irstea (formerly Cemagref) has developed the Linear Extension of the Yule Process (LEYP) model based on counting process theory, which relies not only on a pipe's characteristics and environment but also on its age and previous breaks. It was then decided to develop a break prediction tool usable by water utilities: the ‘Casses’ freeware. To make this possible, it was necessary to deal with several constraints. To cope with the diversity of available data for various water utilities, flexible input data formats were designed as well as an importation module which checks the conformity and coherence of data. Tools for data management and an advice module dedicated to model calibration were conceived for non-statistician users. The break prediction results can be used directly to compare break evolution with different rehabilitation strategies and they can also feed multicriteria decision tools. In this case, the ‘Casses’ freeware can work as a ‘slave’ of the integrated application.