Flow analysis and leak detection with the CFPD method in the paris drinking water distribution system
Author: Peter van Thienen, Frank Montiel
Full Paper




In principle, DMA flow data can provide valuable information about new leaks, unregistered changes in valve status, etc. However, distinguishing one from the others is often difficult. The CFPD (Comparison of Flow Pattern Distributions) method is a flow time series data transformation which facilitates the identification, quantification, and interpretation of changes in the amounts of water supplied. In this way, it helps to distinguish e.g. new leaks from operational signals and demand changes. In the past years, it has been successfully applied at several Dutch drinking water companies. In this paper, we illustrate the application of the CFPD method by presenting selected results from CFPD analyses of flow data for 16 of 30 DMAs in the city of Paris. The findings are compared to a list of registered leaks.