Significance of uncertainties derived from settling tank model structure and parameters on predicting WWTP performance - A global sensitivity analysis study

Uncertainty derived from one of the process models – such as one-dimensional secondary settling tank (SST) models – can impact the output of the other process models, e.g., biokinetic (ASM1), as well as the integrated wastewater treatment plant (WWTP) models. The model structure and parameter uncertainty of settler models can therefore propagate, and add to the uncertainties in prediction of any plant performance criteria. Here we present an assessment of the relative significance of secondary settling model performance in WWTP simulations. We perform a global sensitivity analysis (GSA) based on Monte Carlo analysis in combination with multi-linear regression using four different simulation scenarios. We use the Benchmark Simulation Model Nr. 1 (BSM1), in which, we consider two operation scenarios for wastage of activated sludge (WAS), i.e. from the recycle of activated sludge (RAS) stream and from the last aerobic bioreactor upstream to the SST (Garrett/hydraulic method). For model structure uncertainty, two one-dimensional secondary settling tank (1-D SST) models are assessed, including a first-order model (the widely used Takács-model), in which the feasibility of using measured parameters for calibration is limited. The other SST model is a state-of-the-art, second-order, convection-dispersion tool (Plósz et al., 2007). The sensitivity results obtained from the four scenarios consistently indicate that the settler models and their parameters are among the most significant sources of uncertainty contributing to the predicted plant effluent data as well as sludge production and aeration demand, followed by influent wastewater fractions and biokinetic parameters, especially for autotrophs. Additionally, our results show that the first-order model systematically underrepresents uncertainty. The outcome of this study contributes to a better understanding of uncertainty in WWTPs, and explicitly demonstrates the significance of secondary settling processes that are crucial elements of model prediction under dry and wet-weather loading conditions.