Predicting the fate of micropollutants during wastewater treatment: Calibration and sensitivity analysis

- A MP fate model was implemented and calibrated with full-scale measurements.
- Considering biodegradation in the secondary clarifier was important for modelling some MPs.
- The most influential parameters of the model depended on the dominant fate process
- Sensitivity analysis suggests suitable measurement conditions for improved calibration.

The presence of micropollutants in the environment and their toxic impacts on the aquatic environment have raised concern about their inefficient removal in wastewater treatment plants. In this study, the fate of micropollutants of four different classes was simulated in a conventional activated sludge plant using a bioreactor micropollutant fate model coupled to a settler model. The latter was based on the Bürger-Diehl model extended for the first time to include micropollutant fate processes. Calibration of model parameters was completed by matching modelling results with full-scale measurements (i.e. including aqueous and particulate phase concentrations of micropollutants) obtained from a 4-day sampling campaign. Modelling results showed that further biodegradation takes place in the sludge blanket of the settler for the highly biodegradable caffeine, underlining the need for a reactive settler model. The adopted Monte Carlo based calibration approach also provided an overview of the model's global sensitivity to the parameters. This analysis showed that for each micropollutant and according to the dominant fate process, a different set of one or more parameters had a significant impact on the model fit, justifying the selection of parameter subsets for model calibration. A dynamic local sensitivity analysis was also performed with the calibrated parameters. This analysis supported the conclusions from the global sensitivity and provided guidance for future sampling campaigns. This study expands the understanding of micropollutant fate models when applied to different micropollutants, in terms of global and local sensitivity to model parameters, as well as the identifiability of the parameters.

200