Dealing with uncertainty in modelling of wastewater disinfection by peracetic acid

Disinfection is fundamental in wastewater treatment plants (WWTPs) for matching the microbiological standards on treated wastewater, especially in case of water reuse. In recent years, the modelling of disinfection is receiving increasing interest due to the need for supporting tools for the operation of reactors. In particular, when dealing with chemical disinfectants, the main goal is the optimization of dosage for providing the requested dose for microbial inactivation while minimizing the cost of reagents, the generation of disinfection by-products (DBPs) and the discharged residual disinfectant, having ecotoxicological implications on the aquatic ecosystem. Peracetic acid (PAA) is an emerging disinfectant whose dosage in disinfection reactors must account for PAA strong decay in wastewater. Moreover, even if PAA is not quoted for the generation of DBPs, the discharge in surface water of residual components of commercial solutions (PAA, hydrogen peroxide) can result in the occurrence of ecotoxicity phenomena. Several attempts were reported in the past about the modelling of PAA disinfection, based on deterministic and non-deterministic approaches, also aimed at developing a control system for optimal dosing of the disinfectant. However, no research work included the modelling of uncertainty related to the process so far. Uncertainty is a key aspect of many engineering processes, possibly resulting in significant effects on process performance. In the present work, a modelling framework accounting for uncertainty is defined and discussed, based on stochastical techniques to estimate the uncertainty related to each part of process and to propagate it. The behaviour of a full-scale disinfection reactor was evaluated as a function of operating conditions, identifying the potential non-compliance and proposing operational practices for maximizing performance while satisfying different process goals, while relevant indications for the optimization of a full-scale disinfection reactors were determined. The proposed approach can be easily adapted to other WWTPs and disinfectants.