par Bouillot, P.;Servais, Pierre 
;Billen, Gilles ;Levi, Y.
Référence Revue des sciences de l'eau, 5, page (33-49)
Publication Publié, 1992
;Billen, Gilles ;Levi, Y.Référence Revue des sciences de l'eau, 5, page (33-49)
Publication Publié, 1992
Article révisé par les pairs
| Résumé : | En production d'eau potable, la nature bactérienne de l'abattement du carbone organique dissous biodégradable (COOB) observé dans les filtres à charbon actif en grains (CAG) a été démontrée. les performances de fonctionnement de ce type de contacteur biologique ont été principalement étudiées sur pilotes. Dans la présente étude, elles sont vérifiées et transposées en condition d'exploitation sur une usine de production d'eau potable de la banlieue parisienne. la colonisation bactérienne du CAG a été suivie et montre que l'équilibre biologique est atteint après filtration d'environ 12500m3 d'eau/m3 de CAG. Durant cette phase de colonisation, la biodégradation se substitue progressivement à l'adsorption pour abattre le COD. Après colonisation, l'efficacité des filtres biologiques, exprimée en terme d'abattement de COOB, est fonction du temps de contact quelle que soit la vitesse de filtration (dans la gamme de 2 à 18 m/h). les résultats de suivis de deux filtres sur deux ans montrent que l'efficacité a été globalement meilleure en 1989 qu'en 1990, celte différence s'explique par les fluctuations plus importantes de CODB dans l'influent en 1989. Un modèle mathématique, établi à partir des équations cinétiques des processus bactériens dans les filtres à CAG (modèle CHABROL), développé sur base d'observations antérieures, permet de simuler correcte ment les observations faites au cours de la présente étude. Avec la mesure du CODB, le modèle CHABROL constitue un outil très bien adapté pour contrôler les performances des contacteurs biologiques. Ils permettent, entre autre, de définir le temps de contact optimal de l'eau dans le filtre en fonction d'une température et d'une qualité d'eau donnée dans l'influent et d'une qualité d'eau souhaitée dans l'effluent. _____________________________________________________________________________________________________________________________________ In drinking water production, filtration on granular activated carbon (GAC) is generally used in order to remove by adsorption the dissolved organic matter. Nevertheless, the adsorption capacity of GAC is rapidly saturated and it is so necessary to regenerate the GAC. An interesting alternate has been applied in sorne treatment plants. It consists to use GAC filtration without regeneration taking benefit of the activity of the microbial community which colonize the GAC particles (RITTMANN and HUCK, 1989). In fact, this biological filtration offers the advantage to specially remove the biodegradable fraction of the dissolved organic carbon (BDOC), which is responsible for the problem of bacterial growth into the distribution networks. The bacterial nature of the BDOC removal achieved by the biological filtration on GAC has been now clearly demonstrated (SERVAIS et al., 1991) and some important results of the functioning of the se filters has been obtained in studies conducted on pilots filters (BOUILLOT et al., 1990 ; SERVAIS et al., 1992). These studies have for example shown that only a very small part of the bacterial biomass produced in the filter is exported with the outflow. ln the present study, biological filtration has been investigated in a full scale treatment line at Choisy-le-Roi in the Parisian suburbs and the results compared with those gained on pilot filters. The working conditions of the three GAC filters studied are presented in table 1 and compared with those of pilot filters used in a previous study conducted at Neuilly-sur-Marne (table 2). The microbial colonization has been followed in two of the filters. It lasted roughly 3 months to reach biological equilibration, it corresponds to a water volume filtrated of 12 500 m3 per m3 of GAC. Efficiency of the removal during this period is presented in figure 2. Progressively, biological processes take turn with adsorption (fig. 1). As already demonstrated by SERVAIS et al. (1992), the efficiency of biological filtration, calculated in percentage of BOOC removal, increases with increasing contact time whatever the filtration velocity cou Id be in the range 2 mlh to 18 m/h (fig. 3). However, the percentage of BOOC, at similar temperature, is higher in the GAC filters at Choisy-le-Roi than at Neuilly-sur-Marne. The fixed bacterial biomass is also higher at Choisy-le-Roi (average 7.5 j.lgC/cm3) than at Neuilly-sur-Marne (average 2 j.lC/cm3). Following during two years the functioning of the n° 56 and 38 filters (tables 3, 4 and fig. 5, 7), it seems that the global efficiency of filtration is better in 1990 than in 1989. This can be linked to the greater fluctuations in BOOC in the influent water in 1989 than in 1990, as shown on figure 8. Fluctuations in the quality of the influent water requires a period to reach the equilibrium during which the effluent is charchacterized by a lower quality (fig. 8). This period is longer at low temperature. The mathematical model based on the kinetics of the basic microbiological processes involved in biological filtration (the CHABROL model) has been previously developed (BILLEN et al., 1992) in order to simulate the performances of the filtration. It can be used to simulate the vertical profiles of BOOC and bacterial biomass in the filters of the Choisy-le-Roi treatment plant, with modifying only one parameter in the mOdel, the average bacterial mortality "kd" (fig. 4). BOOC decreases versus empty bed contact time (EBCT) calcula¬ted by the model are presented on figure 6 for the Choisy-le-Roi and Neuilly¬sur-Marne treatment plants and for Iwo temperatures. From a management point of view, the minimum BOOC is reached for contact time between 15 and 20 minutes at Neuilly-sur-Marne, while at Choisy-le-Roi it is rather between 10 and 15 minutes. ln conclusion, BDOC measurements and CHABROL model constitute powerful tools for management and design of biological GAC filters. |
