Comparative analysis of different probability distributions of random parameters in the assessment of water distribution system reliability

During recent years, several methods based on the probabilistic approach have been proposed for the analysis of the performance of water distribution systems (WDSs). Uncertain elements are described by probabilistic laws chosen and parameterised on the basis of the network characteristics. However, the choice of the most suitable probabilistic distribution and of the statistical parameters can be difficult because of the lack of information about the WDSs. Among the stochastic parameters that affect the network performance, a fundamental role is played by the times to failure and repair of the system components. The impact of the chosen probability distributions of these fundamental variables on the evaluation of water distribution network reliability is analysed. The study is performed by using a technique capable of considering the mechanical failure of the network components, the spatial and temporal variations of the water demand and the uncertain distribution of the pipe roughness. This analysis allows quantification of the effect of any inaccuracy that may occur in the probabilistic characterisation of the random parameters.