Use of Pressure Management to Reduce the Probability of Pipe Breaks: A Bayesian Approach

Abstract

As pipe breaks in water distribution networks produce serious consequences, water authorities strive to minimize the frequency of their occurrence. Pressure management is an essential tool to reduce the frequency of breaks and it is closely linked to the proper analysis of a maximum pressure indicator. A methodology that compares the unconditional cumulative distribution function (CDF) and the parametric break-conditioned CDF of the maximum pressure indicator is proposed in this paper. The relationship between the CDFs compared is established by means of the Bayes’ theorem, which allows determining a probability ratio. The objective is to identify the range of operation of maximum pressure that is most likely to reduce pipe breaks. The methodology is applied to four sectors of the water distribution network in Madrid (Spain). In three of those sectors, the maximum pressure indicator is a good predictor of the probability of pipe breaks, confirming that the probability of breaks increases for high maximum pressure ranges. The methodology is validated in one sector, and results provide good agreement between predicted and observed failure rates.