Pressure as a predictor of occurrence of pipe breaks in water distribution networks

Abstract

Sudden bursts in water distribution networks may lead to costly consequences. As pressure is one of the causes of such events, pressure management could reduce the probability of failure. In this work, a methodology is proposed to analyse the relationship between pipe breaks and water pressure by means of pressure-related indicators. The objective is to identify the most influential indicators for the probability of occurrence of pipe breaks. The methodology compares the cumulative distribution function (CDF) conditioned to breaks and 100 random sets of the same size sampled from the CDF of the indicator. The most influential indicators are related to the greatest number of rejected cases of the Kolmogorov-Smirnov test. Moment indicators and their calculation period emerge from sensitivity analyses. The methodology is applied to six sectors of Madrid (Spain) while two sectors are used for validation. Results show that the pressure range is the best indicator of breaks.