A front-line and cost-effective model for the assessment of service life of network pipes

Abstract

In any water utility, a reliable assessment of the service life of the network pipes is a key piece within the big puzzle of assets management. This paper presents a new statistical model (basic pipes life assessment, BPLA) to assess the service life of pipes, to locate the pipes on the failures bath curve and to forecast the expected failures in future years. Its main novelties are the processing of pipe information (is that information what is adapted to the classical maintenance engineering and not the other way back) and the definition of two different time variables that can be analyzed in parallel. The first novelty makes the model less demanding in terms of data and software tools than others currently available, and the second one allows to get all the results after one single stage of calculation. To show its usability, the BPLA has been applied to a pipe network that supplies water to 500,000 citizens for which two years of failure records are available. Procedures and results have been compared to the well-known Weibull proportional hazard model (WPHM), with final relative errors lower than 10% and 15% on each particular result.