Sensitivity Analysis of Yield Estimate of Urban Water Supply Systems

Abstract

Sensitivity analysis (SA) theory and techniques were used in this study to estimate the sensitivity of input variables on the yield estimate of an urban water supply system. The SA techniques considered were Morris method and Fourier amplitude sensitivity test (FAST), including the related extended FAST. A case study on a simple urban water supply system was conducted to assess the applicability and to study the limitations of these techniques and the SA framework adopted. Findings showed that the streamflow dominated all experiments, with the supply reliability threshold, the upper restriction rule curve and the consecutive months in restrictions threshold of subsequent importance. In a screening pass, importance ranking of the 26 considered variables from the Morris method were verified with FAST and extended FAST. Once minor errors were overcome by increasing the number of model simulations, a high resolution pass quantified the importance of the top 10 ranked variables. The case study also highlighted the need to improve the adopted SA framework by considering a different methodology considering different climate scenarios and alternative input variable handling strategies.