A probabilistic model for assessing uncertainty and sensitivity in the prediction of monochloramine loss in French river waters.

Abstract Monochloramine (NH2Cl) is increasingly used as alternative disinfectant to free chlorine in industrial plants. After use in cooling systems, the waters are released to the environment and residual NH2Cl may be discharged into the receiving waters. As NH2Cl is suspected to exhibit toxicity towards aquatic organisms, a proper risk assessment of its occurrence in environmental waters is needed to prevent adverse effects on wildlife. For this purpose, a comprehensive model simulating monochloramine loss in natural riverine waters was developed. This model incorporates the following processes: (i) autodecomposition; (ii) reaction with nitrite and bromide; (iii) oxidation with Dissolved Organic Carbon (DOC); (iv) oxidation with organic fraction of Suspended Particulate Matter (SPM); (v) reactions in bottom sediments and (vi) volatilization. The model was also designed to conduct uncertainty and sensitivity analysis. It was tested on several French rivers submitted to discharges of monochloraminated effluents and on several seasonal conditions. Uncertainty analysis allowed evaluation of confidence intervals related to NH2Cl half-lives in natural waters. It was shown that simulation intervals are in good agreement with experimental data obtained on the same rivers. Sensitivity analysis using an EFAST variance decomposition approach allowed identification of the most influential parameters on half-life determination. It was shown that the kinetic rate describing rapid reaction of NH2Cl with DOC is by far the most sensitive parameter, demonstrating the predominance of such reactions in the loss process. Variables or parameters involved in temperature dependence (temperature and activation energy) can also significantly influence model results. To a lesser extent, wind velocity is the most sensitive parameter explaining uncertainty in the prediction of volatilization, with a high level of interactions with other parameters, showing that loss through volatilization can be essential in some specific conditions only. This study then identified the most important research priorities for improving the prediction of NH2Cl half-lives in natural rivers.