Fuzzy approach in the uncertainty analysis of the water distribution network of Becej

Abstract

One of the most important steps in the development of water distribution system simulation models is providing the dependence of results of simulation on the uncertainty of input values. Information about propagation of uncertainty through the simulation model provides a basis for general model improvement in an efficient and economic (low cost) way. Besides statistically based uncertainty and sensitivity analyses, the fuzzy set theory provides an alternative that does not require crisp statistical measures of input parameter distributions. Uncertainty is represented by fuzzy set parameters, which do not have direct correspondence with statistical background, although indications of uncertainty can be easily recognized. In the analysis presented in this article, applied to the model water distribution system of the Becej community, roughness of old pipes has been selected as the uncertain parameter, as the correct values and their statistical distribution are unknown. As fuzzy uncertainty analyses for non-linear models and non-monotonic results require efficient optimization methods, a method based on evolutionary genetic algorithm has been used.