Prototypes of software tools existed for pipe failure prediction and water supply service reliability at a system level prior to the CARE-W project. These methods were extended to single pipelines and single users, respectively, i.e. pipes and nodes including their individual characteristics.
The description showed within a unified framework, both, theoretical and operational aspects, including data requirements of each model. Data for pipe failure prediction and water supply service reliability were generated with the existing models and through theoretical considerations made by consortium members. This task was based on a preliminary state-of-the-art report (paper published in a scientific conference) that has been developed as a preparation for the CARE-W project.
Several models were compared in practical tests. The test results were used for improvement of the models. Generic data for failure forecasting on a network level will be developed, based on expert judgement and statistics from case cities. Methods for forecasting failures of single pipes based on statistics were developed, and compared to a physical based (deterministic) method, which currently is under development. Potential and limitations of the various models will be described as a function of data availability and other relevant criteria. The impact and importance of each pipe to the reliability of all networks and water supply of consumers were set up by the reliability model.
Since CARE-W includes models that partially overlap with respect to their objectives and results, a procedure for the selection of models in practical situations were established. The objectives were to ensure that the end-users apply the most efficient models for their particular needs
Published January 14, 2008