Book Description

Natural mineral waters are groundwaters subject to strict quality standards, limiting the possibilities of intervention in case of presence of natural undesirable elements of geogenic or anthropogenic origin, such as manganese, iron or ammonium. The targeted removal of manganese by physical-chemical oxidation is commonly used in the industry concerned. However, there is no regulatory process for the elimination of ammonium, which represents a health risk due to its potential for oxidation into nitrite. The removal of these compounds by biological sand filtration is successfully applied in several countries and described in the literature. However, the acquisition of experimental data for an industrial application under restricted conditions is necessary.This thesis project focused on the study of nitrification and manganese oxidation processes in RSFs. For this purpose, a pilot filter installation was set up, fed by groundwater from an industrial borehole. These filters were first operated without parameters modification, to observe the installation of oxidation activities. Obtained experimental data enabled the construction of a mathematical model to simulate these activities. However, these setup times, from 2 to 3 months, were considered too long for an industrial application, as water is lost during this period, resulting in significant costs and environmental footprint.An optimization strategy specific to this context has been developed based on preliminary data. It consists in accelerating the specific growth rate of active bacteria entering the system, by increasing the water temperature and the substrate concentration during the start-up phase. An average start-up time reduction of 66% was observed for nitrification, while maintaining the removal efficiency back to baseline conditions. On the other hand, the biological removal of manganese was not impacted by the modification of these parameters.Molecular biology analyses showed a logical selection of active populations in the depth and at the outlet of the biofilters, particularly nitrifying bacteria. Moreover, an evolutionary trend within the nitrifying populations could be observed, with mainly AOB of the Nitrosomonadaceae genus during the start-up phase, and mainly commamox of the Nitrospira genus afterwards, back to reference conditions. The expression of the mmcO gene, involved in the biological oxidation of manganese via the multi-copper oxidase enzyme, was highlighted by a metagenomic study. Finally, quantitative analyses showed that there was no significant increase in the amount of total planktonic bacteria between the inlet and outlet of the biofilter after the installation of the activities.During substrate saturation experiments, no inhibition of manganese oxidation by the presence of nitrite and ammonium was observed. On the other hand, a residual presence of activity and microbial populations could be deducted in the lower half of the biofilter. This improved the understanding and long-term representation of the attachment of active biomass on the filter media, in the model definition.A model of manganese oxidation, including both biological and chemical pathways, was proposed and successfully applied. This is an important added value of knowledge, as the distinction of these processes has been very little studied in the literature.Finally, industrial scenarios were tested via the model with a system of partial effluent water recirculation, to drastically increase the gains obtained during the experimental optimization strategy: up to 75% reduction of start-up time, and up to 95% savings of water and thermal energy.This thesis provided additional knowledge on an already proven technology, thanks to a strict industrial context. The sum of knowledge and data, thanks to the different levels of analysis carried out, allowed the construction of a model allowing to reproduce, then simulate oxidation activities and their distribution in the biofilters.