Influence of NOM concentration on parameter sensitivity of a mechanistic ozone decomposition model

Aqueous ozone decomposition proceeds through a complex chain mechanism of radical reactions. When natural organic matter (NOM) is present, the system becomes much more complex and often (semi-)empirical modelling approaches are used to describe ozonation of water and wastewater systems. Mechanistic models, however, can be of great value to gain knowledge in the chemical pathways of ozonation and advanced oxidation processes in view of engineering applications. However, the numerous model parameters and model complexity often restrict their applicability. Model simplification is then an option to cure these drawbacks. In this study, sensitivity analyses (SAs) were used to determine the most important elementary reactions from the complex kinetic model. Additionally, SAs were used to understand the reaction mechanism. It was demonstrated that only seven of the twenty-eight first and second order rate constants showed to impact ozone and HO• concentrations. Processes involving HO• scavenging by inorganic carbon were of minor importance. Mass-transfer related parameters kLa and [O3*] were of major importance in all cases. Hence, it is of extreme importance that these parameters are determined with high accuracy. It was shown that the aqueous ozone concentration is extremely sensitive to parameters involving NOM at very low scavenger concentrations implying that impurities should always be considered in models, even in ultrapure water systems. Uncertainty analysis showed that especially the HO• concentration is susceptible to variations in influent composition. The uncertainty regarding this species significantly reduced with increasing levels of scavengers and especially NOM. It was demonstrated that simplification of the elementary radical scheme should be considered. On the other hand, a model extension with regard to reactions involving NOM should be performed in order to improve the applicability of future wastewater ozonation models.

http://ift.tt/2pBqpF3