Global sensitivity analysis of a phenomenological wastewater treatment plant influent generator. 8th IWA Symposium on Systems Analysis and Integrated Assessment

The objective of this paper is to present the results of a global sensitivity analysis (GSA) of a phenomenological model that generates wastewater treatment plant (WWTP) dynamic influent disturbance scenarios. This influent model is part of the Benchmark Simulation Model no 2 (BSM2) and creates realistic dry/wet weather files describing diurnal, weekend and seasonal variations through the combination of different generic models blocks, i.e. households, industry, infiltration, rainfall and transport. The GSA is carried out by combining Monte Carlo simulations and standard regression coefficients (SRC), followed by classification of the influence of model parameters on the model output into strong, medium and weak. The results show that the method is able to decompose the variance of the model predictions (R2 > 0.9) satisfactorily for several flow rate descriptors calculated at different time resolutions. Catchment size (PE) and the usage of wastewater per person equivalent (QperPE) are two parameters that strongly influence the yearly average dry weather flow rate and its variability. Wet weather conditions are mainly affected by three parameters: 1) the probability of occurrence of a rain event (Llrain); (2) the catchment size, incorporated in the model as a parameter representing the conversion from mm day-1 to m3 day-1 (Qpermm); and, (3) the quantity of rain falling on permeable areas (aH). Very importantly, the case study shows that the SRC parameter ranking changes when the time resolution is changed, both for dry and wet weather conditions. The paper ends with a discussion on the interpretation of GSA results and of the advantages of using synthetic flow rate data for WWTP simulation studies. The discussion section also includes suggestions on how to use the influent model to adapt the generated time series to each modeller’s demands.