FR :

L'efficacité de l'extraction sur résine XAD-4 a été étudiée à partir de neuf composés hydrophiles non ionisables (alcools, aldéhydes, cétones, esters). Le facteur de capacité (k') est déterminé expérimentalement par élution frontale de différentes fractions volumiques de méthanol. Les rendements de récupération sont évalués pour quatre solvants. L'efficacité maximale et le facteur de concentration optimal sont définis en fonction du volume de percolation et du volume d'élution. Le rendement moyen est de 80 % après élution par l'acétone.

L'entraînement dynamique à la vapeur est par ailleurs développé pour concentrer les composés dans 200 ml de distillat à partir de 4 litres d'échantillon. Le rendement moyen est de 77 %. Ces deux techniques sont associées afin d'accroître la spécificité de la méthodologie analytique vis-à-vis des composés hydrophiles non ionisables. Après un entraînement de l'échantillon d'eau, le condensat est ensuite percolé sur une colonne de résine XAD-4 (7 cm x 1,5 cm), l'élution s'effectuant par 14 ml d'acétone. L'éluat est alors concentré, à température ambiante, par un courant d'azote jusqu'à 4 ml, puis analysé par un chromatographe en phase gazeuse. Cette méthodologie a été appliquée à un eau naturelle artificiellement polluée par 15 composés hydrophiles non ionisables. Les rendements de récupération sont compris entre 42 % et 83 % (rendement moyen de 63 %).

EN :

Non-ionic hydrophilic compounds are very often used as solvents in industrial activities or for domestic use. These products are generally alcohols, aldehydes, ketones and esters. Although they are not very dangerous for the environment at low concentrations, for hazardous pollution conditions there is a need to be able to identify and quantify such compounds. Chlorination of humic substances during drinking water treatment can also generate chlorinated by-products (chlorinated ketones).

Because of their solubility in water, it is necessary to concentrate and extract these compounds before gas chromatography analysis. In this study, we describe a method where liquid/gas (steam distillation) and liquid/solid (XAD-4 resin) extractions were combined. The aim of this work was to define analytical conditions to obtain the best efficiency for the overall method.

In the first phase, we developed the liquid/solid extraction on XAD-4 resin by studying capacity factors, elution volumes and percolated water volumes. The sorption of nine non- ionic hydrophilic compounds (ethanol, butanol-1, pentanol-1, ethyl acetate, n-butyl acetate, methylisobutylketone, cyclohexanone, diisobutylketone, butanal) from water onto microparticulate XAD-4 resin, and their subsequent recovery into water-methanol mixtures, was used to estimate the capacity factor (k'water) for each analyte. According to the capacity factor values, we could define three classes of compounds:

1. highly hydrophilic compounds such as ethanol, which were not very well retained by the resin (k'water=11);

2. moderately hydrophilic compounds such as ethyl acetate, butanol-1, butanal, pentanol-1 for which capacity factors were less than 100;

3. weakly hydrophilic compounds that were well retained by the resin, such as diisobutylketone with a capacity factor equal to 1770.

Subsequently, mobilization/recovery by frontal elution into each of four different solvents (dichloromethane, acetone, acetonitrile, methanol) was studied in order to define for each of the analytes an optimal recovery procedure. Recoveries were dependent on the volume of the water sample passing through the column (100 mL, 200 mL, 400 mL) and on volume of solvent used for the elution. The best recoveries were observed when 200 mL of water were passed through the column and when the elution was performed by 14 mL of acetone. Mean recovery was 80%. Methylisobutylketone, pentanol-1 and ethyl acetate were extracted with yields higher than 80%, whereas butanal and 2,2-dichloroethanol showed a poor recovery of around 60%.

In the second phase of the study, steam distillation was also assessed as a technique for the concentration of analytes (200 mL condensate from 4 litres aqueous sample). Mean recovery was 77%. Higher recoveries were observed for ethyl acetate, butanol-1, pentanol-1 and 1,1- dichloroacetone. Dichloroethanol was not extracted with a suitable recovery (yield around 33%).

The two extraction methods were combined to increase the extraction specificity for non- ionic hydrophilic compounds. In a preliminary concentration step, the analytes were recovered by steam distillation from 4 litre water samples. Analytes were sorbed from the condensate onto a 7 cm x 1.5 cm column of XAD-4 resin, then eluted with 14 mL acetone. The column eluate was concentrated to 4 mL, at room temperature in a gentle stream of nitrogen, and analyzed by gas chromatography.

The extraction procedure was applied to 18 compounds (butanol-1, pentanol-1, hexanol-1, octanol-1, decanol, 2,2-dichloroethanol, ethyl acetate, n-butyl acetate, methyl hexanoate, methyl octanoate, methylisobutylketone, diisobutylketone, 1,1-dichloroacetone, octanone-2, butanal, hexanal, octanal, benzaldehyde) in spiked ultrapure water and also to 15 compounds (same compounds except for benzaldehyde and 2,2-dichloroethanol) in spiked natural water. Mean recovery percentages for the analytes in ultrapure water and in natural water were respectively around 60% and 63%. These results demonstrate that the ionic composition of the water medium did not interfere with the efficiency of the complete method. In the case of spiked natural water, recovery yields ranged between 42% (butanol- 1) and 83% (octanol-1).

Keywords

.

FR :

Cette étude présente un suivi de l'élimination des acides aminés totaux par les différentes étapes de traitement de deux usines canadiennes de Ville de Laval (Québec, Canada). L'influence des saisons sur l'efficacité de l'abattement ainsi que le devenir des acides aminés dans les réseaux de distribution de ces usines ont aussi été étudiés.

D'une manière générale, l'étape de coagulation-floculation-décantation permet une élimination important d'acides aminés totaux (34-72%), l'effet de la filtration sur sable et anthracite est peu prononcé et varie avec les saisons, l'ozonation provoque une augmentation (20-100%) et la filtration sur charbon actif biologique (CAB) permet un abattement en eau tiède (25-34%), mais semble relarguer en eaux froides (35%). Finalement, la post-désinfection au chlore permet une élimination supplémentaire (47%) à cause de la réactivité de cet oxydant envers les acides aminés. La post-désinfection au bioxyde de chlore permet elle aussi d'abattre les teneurs en acides aminés (12-39%) et ce, malgré le fait que cet oxydant soit moins réactif envers les acides aminés. L'élimination des acides aminés totaux avant la chloration est préférable afin de limiter la formation de sous-produits de chloration.

EN :

Low concentrations of free and combined amino acids are found at every stage of water treatment (LE CLOIREC and RENAUD, 1984; SCULLY et al., 1988; BERNE et al., 1994). A decrease in amino acid concentrations has been observed after settling (LE CLOIREC et al., 1983; LE CLOIREC and RENAUD, 1984), while ozonation has been shown to increase free amino acid concentrations. Biological activated carbon (BAC) filtration may also decrease the concentration of total amino acids (JADAS-HÉCART, 1989; BERNE, 1994).

Total amino acids represent a small fraction of dissolved organic matter (1 to 3% of DOC), but account for an important part of the chlorine demand of treated water (JADAS-HÉCART, 1989; HUREIKI et al., 1994). Moreover, recent work by HUREIKI and GAUTHIER (1994) has suggested that some amino acids found in surface waters may represent a significant fraction of the precursors of some organohalogenated disinfection by- products (DBP). It is also possible that total amino acids amount to an important fraction of biodegradable organic carbon (BOC). These characteristics of amino acids suggest that removing them by treatment will improve water quality, both from a biological and a chemical (DBP) stability standpoint.

The objectives of the research described in this paper were to:

1. assess the seasonal variability of the removal of free and combined amino acids by different treatment processes in two Canadian water treatment plants located in the city of Laval (Québec, Canada);

2. monitor the fate of total amino acids in the distribution systems of these two plants with the objective of correlating total amino acids with biodegradable dissolved organic carbon (BDOC) and fixed bacterial density.

Sampling was conducted in two water treatment plants. The Ste-Rose water treatment plant (100,000 m³/d) uses conventional treatment (dynamic settling and dual-media filtration on sand and anthracite) followed by ozonation, pH adjustment, and post-chlorination using either chlorine of chlorine dioxide. The second plant studied is the Chomedey water treatment plant (180,000 m³/d) in which the following processes are used: conventional treatment (dynamic settling and dual-media filtration on sand and anthracite) followed by ozonation, second stage filtration on biological activated carbon (BAC), pH adjustment, and post-chlorination using chlorine dioxide. Samples were collected in the two distribution systems according to the residence time of the water calculated by a hydraulic model. Samples were taken directly from small diameter (15 cm internal diameter) ductile iron pipes.

Results of the monitoring of the treatment plants show a very strong decrease of total amino acids by coagulation-flocculation-settling (34-72%). First stage dual-media filtration may increase or decrease the concentrations of total amino acids present, depending on the time of the year. In all but one case ozonation increases the concentration of total amino acids (20-100%). To document the source of this increase, we verified the yields of hydrolysis of amino acids found in natural matrices under different hydrolysis conditions. The objective of this experiment was to verify if more drastic hydrolysis conditions would free some amino acids linked to more complex structures found in natural waters. If such were the case, this would explain why higher concentrations of total amino acids were obtained after ozonation. Results showed that the hydrolysis conditions used by BERNE (1994) are optimal for recovery of amino acids in the natural water studied (sand and anthracite filter effluent).

Total amino acids were decreased by BAC filtration in warm water (24 to 34%) but increased in cold water (+35%). This could either be related to the slower kinetics of the hydrolysis of combined amino acids by the fixed biomass or by the form in which amino acids were present in the winter matrix. The expected effect of chlorination on total amino acids was observed with a decrease of 47% at the post-chlorination step. Some impact of chlorine dioxide was noted although it is believed that chlorine dioxide will not readily react with free and combined amino acids. This effect was more pronounced in cold water and could be related to the fact that chlorine dioxide in full-scale plants is produced in the presence of excess chlorine (less than 10%).

The analysis of the composition of total amino acids present in the Mille-Îles River showed that the amino acids most commonly found were glycine, serine, alanine, leucine, lysine, aspartic acid and glutamic acid. We observed that the trends of major amino acids present followed the trends observed for total amino acids. This concordance of trends does not reflect the individual characteristics of each amino acid or the ability of a treatment process to remove or transform them. It most probably reflects the ability of each treatment process to remove or transform complex bound forms of amino acids.

In the Ste-Rose distribution system (DS), we observed stable concentrations of total amino acids regardless of the residence time. In the case of the Chomedey distribution system, concentrations decreased slightly with residence time, suggesting a greater stability of amino acids in the DS fed by biologically treated water. In both cases, levels of amino acids were very low close to the detection limit. In the case of the Ste-Rose DS, these low levels were attributed to biological removal, whereas in the case of the Chomedey DS this was associated with post-chlorination.

No direct correlation between biodegradable organic carbon (BDOC) and total amino acids was observed. This may reflect the fact that total amino acids represent a variable fraction of the total pool of biodegradable organic carbon depending on the source water composition and on the treatment process applied. In the case of the source water studied, total amino acids represented a major fraction of the biodegradable organic carbon pool: more than 42% in raw water and more than 45% in BAC filter effluent. These high proportions may partially be explained by the eutrophic state of our source water and by the method of BDOC measurement used.

In conclusion, amino acids represent an important fraction of biodegradable organic carbon which can be removed most efficiently by optimized coagulation-flocculation and settling, biological treatment and chlorination. It is preferable to remove total amino acids before chlorination in order to limit the formation of undesirable DBPs.

FR :

Identifier tous les processus physiques élémentaires du cycle hydrologique qui peuvent avoir lieu dans un bassin versant et attribuer à chacun d'eux une description analytique permettant la prévision conduisent à des structures complexes employant un nombre élevé de paramètres difficilement accessibles. En outre, ces processus, même simplifiés, sont généralement non linéaires. Le recours à des modèles à faible nombre de variables, capables de traiter la non-linéarité, s'avère nécessaire.

C'est dans cette optique que nous proposons une méthode de modélisation de la relation pluie et débit basée sur l'utilisation de réseaux neuronaux. Les performances de ces derniers dans la modélisation non linéaire ont été déjà prouvées dans plusieurs domaines scientifiques (biologie, géologie, chimie, physique). Dans ce travail, nous utilisons l'algorithme de la rétropropagation des erreurs avec un réseau à 3 couches de neurones. La fonction de transfert appliquée est de type sigmoïde. Pour prédire le débit à un moment donné, on présente à l'entrée du réseau des valeurs de pluies et de débits observés à des instants précédents. La structure du réseau est optimisée pour obtenir une bonne capacité prévisionnelle sur des données n'ayant pas participé au calage.

L'application du réseau à des données pluviométriques et débimétriques du bassin de l'oued Beth permet d'obtenir de bonnes prévisions d'un ou plusieurs pas de temps, aussi bien journalières qu'hebdomadaires. Pour les données n'ayant pas participé au calage, les coefficients de corrélation entre les valeurs observées et les valeurs estimées par les différents modèles sont élevés. Ils varient de 0.72 à 0.91 pour les coefficients de corrélation de Pearson et de 0.73 à 0.95 pour les coefficients de Spearman.

EN :

Identification of the elementary processes of the hydrological cycle in a drainage basin, and the comprehensive description of each of them, lead to hydrological models with a complex structure including a high number of relatively inaccessible parameters. Moreover these processes, even when simplified, are generally non-linear. Using models with a smaller number of parameters, in order to cope with non-linearity, is therefore necessary.

In this perspective, we propose an artificial neural network for rainfall-runoff modeling. Performances of this method in non-linear modeling have been already demonstrated in several scientific fields (biology, geology, chemistry, physics). In the present work, we use the error back-propagation algorithm with a three-layer neural network. The transfer functions belong to the sigmoidal type at each layer. To predict the runoff at a given moment, the input variables are the rainfall and the runoff values observed for the previous time period. The structure of the network (number of hidden nodes, learning coefficient and momentum values) is optimized to guarantee a good prediction of the runoff, using a set of test data (validation set) not used in the training phase.

Data compiled in our model are a ten year set of rainfall-runoff values collected by the Rabat hydraulic administration (September 1983 to April 1993) in the Beth Wadi catchment. In this study, we develop two types of models according to two different time steps (daily and weekly). The data are subdivided into two sets: a first set to train the model (training set) and a second set to test the model (validation set). For the daily timestep model, we used data of the last two years: April 1991 to April 1993. The initial 365 data (April 1991- April 1992) constitute the training set and the 365 remaining data constitute the validation set. For the weekly data (Monday to Sunday averages), we have 502 pairs of values. We worked by preserving the last 120 values as the validation set and trained the neural network with the remaining data, i.e. 382 pairs of values of weekly rainfall-runoff.

Three types of estimation have been carried out:

1. at instant prediction: prediction of runoff at time t taking into account rainfall values at time t, as well as runoff and rainfall values at preceding times (until t-1);

2. one step ahead prediction: prediction of runoff at time t from rainfall and runoff values at the preceding times (until t-1);

3. multistep prediction: prediction of runoff values for a period from t-jh until t, given that values of the runoff for the period 1 to t-jh-1 and values of the rainfall at times 1 to t are available (h is the timestep).

The step time is daily for the at instant prediction and weekly for one step ahead and multistep predictions. The choice of input variables is determined by autocorrelation function (ACF) and partial autocorrelation function (PACF) analyses on runoff values, and cross-correlation function (CCF) analysis between rainfall and runoff values. For the at instant prediction, the input vector is composed by runoff values of the four days preceding day t, and rainfall values for the three last preceding days as well as its value on day t. For the one step ahead prediction, the input vector is composed of runoff values of the five weeks preceding week t, and rainfall values for the three preceding weeks (without considering the rainfall at time t). Finally, for the multistep prediction, the input vector is the same as for the one step ahead prediction but rainfall values include time t. The runoff values for the week t-jh+1, as well as for the following weeks, are computed by feed backing to the input vector the runoff value predicted for the preceding week.

The rainfall-runoff models allow a good estimation for one or several timesteps, daily as well as weekly. In the validation set, correlation coefficients between observed and estimated values are high. In the at instant prediction, we obtain the Pearson correlation coefficient R=0.772 and the Spearman correlation coefficient CR=0.958. The weak value of R as compared to CR is explained by a few extremely high values of error of prediction. In the one step ahead prediction (R=0.887 and CR=0.782) and multistep prediction (R=0.908 and CR=0.727), the R coefficients are higher that CR. This confirms that predicted values are in good agreement with the peaks of observed values (absence of large exceptional errors). In all cases, the results obtained are better than those obtained with linear methods. The neural network models can thus be recommended for time series studies in environmental sciences.

FR :

L'épuration ou la destination finale des éluats de régénération des résines utilisées en dénitratation reste un problème mal résolu. De nouvelles voies sont recherchées; la dénitrification biologique semble très attrayante car elle peut être mise en oeuvre sous forme d'un réacteur de faible coût et d'exploitation facile.

L'objectif de ce travail a été d'affiner la connaissance du procédé de dénitrification chimioautotrophe sur soufre en milieu salé, dirigé vers l'épuration et/ou la valorisation de ces éluats.

Claus et Kutzner (1985) ont montré l'action inhibitrice de la dénitrification du chlorure de sodium à partir de 20 g/l. Une colonne contenant du soufre est alimentée en continu par une solution comportant des nitrates et des sels nutritifs. La première partie de l'étude permet de relier la charge volumique (Cv) au rendement de dénitrification. Le rendement de dénitrification est supérieur à 85 % pour Cv < 2 kg NO3-/m3.j ; il tombe à 50 % pour Cv ~ 3,5 kg NO3-/m3.j

Des rapports stoechiométriques entre les nitrates réduits et les produits formés/consommés ont été établis. Ainsi : NO3- donne 0,98 SO42- et 1 NO3- consomme 0,90 HCO3-.

L'incidence de la teneur en NaCl de la solution alimentaire permet de montrer l'inhibition non compétitive du chlorure vis-à-vis de la dénitrification. L'incidence de la concentration en NaCl sur le vitesse de dénitrification apparaît pour une charge de 1 kg NO3-/m3.j. On note que le rendement de 90 % ou plus est obtenu pour une concentration ² 30 g/l. La constante d'inhibition est de l'ordre de 40 g/l.

L'examen de la répartition de la biomasse sur le matériau et une approximation simple d'écoulement en flux piston nous conduisent à prévoir le rendement d'élimination de NO3- par :

r=1 - e[ - 143,5 ] (sur) CV (41+I)

Cette équation traduit les rôles de la charge volumique (Cv) et de la concentration en NaCl (I).

EN :

The treatment and ultimate disposal of saline eluates from ionic exchange resins remain a badly worked out problem. New approaches are needed and biological denitrification seems to be very attractive since it may be implemented in the form of an inexpensively and easily used reactor.

The aim of the present work was to better understand the bacterial denitrification process on sulphur by a chemoautotrophic bacterium, Thiobacillus denitrificans, in a saline medium, and to apply this knowledge to the treatment and/or the recycling of these eluates. Claus and Kutzner (1985) reported an inhibitory effect of NaCl starting at a concentration of 20 g/L.

A column containing elemental sulphur was continually fed by a solution containing nitrate and nutrient salts. In the first step of the study, a relationship between the volumetric load (Cv) and the denitrification efficiency was established. For Cv < 2 kg NO3-/m3/j, the denitrification efficiency was greater than 85%, whereas it decreased rapidly to 50% for Cv 3,5 kg NO3- /m3/j. Stoichiometric ratios between reduced nitrates and formed/consumed products were determined. For example, the reduction of 1 mole of NO3- induces the formation of 0,98 mole SO42- and requires 0,90 moles HCO3-.

The effect of the NaCl concentration was investigated for a volumetric charge of 1 NO3-/m3/j. An uncompetitive inhibition of denitrification by the sodium chloride present in the feed solution was observed. The denitrification efficiency reaches at least 90% for NaCl concentrations up to 30 g/L. The inhibition constant is about 40 g/l. A model linking the denitrification efficiency (r) to the volumetric loading and to the sodium chloride concentration in the regeneration water was developed, based on an examination of the biomass distribution on the fill material and the assumption of plug-flow within the column:

r=1 - e[ - 143,5 ] (over) CV (41+I)

FR :

Dans le cadre de travaux sur le rôle des échinodermes dans le cycle biogéochimique de certains métaux traces en milieu côtier, une étude de l'accumulation et de la rétention d'espèces chimiques du mercure a été menée avec l'étoile de mer Leptasterias polaris sur une période de 40 jours. Les étoiles de mer ont été divisées en deux groupes et nourries avec des moules contaminées soit au méthylmercure (MeHg) (14 mg.kg-1), soit au mercure inorganique (13 mg.kg-1). Les résultats montrent une vitesse d'accumulation dans le système digestif de 0,22 mg.kg-1.j-1 pour le Hg inorganique et 0,17 mg.kg-1.j-1 pour le MeHg. Ces vitesses sont 10 à 15 fois moins importantes dans les gonades et l'endosquelette. Pour le Hg inorganique, la charge relative dans les différents organes analysés se fixe après quelques jours et demeure stable jusqu'à la fin de l'expérience. Pour le MeHg, au contraire, cette charge relative passe progressivement du système digestif vers les gonades et l'exosquelette tout au long de l'expérience. Le pourcentage de rétention (%) demeure constant pour le Hg inorganique (51 ± 13%) mais il augmente régulièrement pour le MeHg, atteignant 90-95% vers la fin de l'exposition. L'application d'un modèle cinétique simple, basé sur un processus d'échange ionique, a permis de calculer des constantes de vitesse d'échange entre le digestat et le tissu digestif. Le MeHg s'échange plus vite entre la solution et les sites, mais semble mettre un peu plus de temps à voyager jusqu'aux organes de bioaccumulation. En conclusion, l'étoile L. polaris, par sa taille, sa longévité et sa grande efficacité à digérer tout ce qui est ingéré, semble en mesure de jouer un rôle important à la fois dans la séquestration du MeHg et la remise en solution du Hg inorganique.

EN :

In spite of a large body of work on the uptake of trace metals in aquatic organisms in the last two decades, very little attention has been devoted to echinoderms (urchins, starfish, and sea cucumbers, among the most important ones), a large group of invertebrates forming a major component of most coastal ecosystems. In the course of our research program on the role of echinoderms in the biogeochemical cycling of some trace metals in coastal environment, an experiment has been conducted on the uptake and the retention of mercury species with starfish Leptasterias polaris over a 40-day exposure period.

Starfish were caught near Pointe-au-Père (Québec, Canada) and were acclimatized to laboratory conditions for 10 days in a flow-through aquarium. Forty stars (84 ± 14 g) were divided in two groups, kept in two aquariums, one for inorganic mercury and the other for methylmercury exposure. Animals were isolated from each other by mean of small 65-mm net cubicles allowing free circulation of seawater at a rate of 2 L.min-1. This flow rate provided a rapid renewal of seawater and ensured the elimination of any soluble mercury excreted during the course of the experiment. Control starfish receiving only uncontaminated mussels were also maintained in both aquariums to monitor the possible uptake of mercury from seawater. Blue mussels were collected in the vicinity of Pointe-au-Père and were contaminated following a technique described elsewhere (PELLETIER and LAROCQUE, 1987). Average levels of inorganic mercury and methylmercury (MeHg) in mussels were 13 ± 3 and 14 ± 4 mg.kg-1, respectively. Each starfish received one contaminated mussel per day over a 40-day period. Mussels were weighed before to be introduced in cages and their empty shells weighed again after ingestion by starfish in order to evaluate the amount of food and mercury taken up daily by each star. Two starfish were sampled in each aquarium every five days and dissected for pyloric caeca, gonads, and endoskeleton. Coelomic fluid was also collected. Both mercury species were analysed as total mercury (assuming a preservation level of ± 85-90% of the speciation during the course of the experiment) in biological tissues by cold vapour atomic adsorption spectrophotometry (detection limit=0.005 mg.kg-1 ww). The coefficient of variation was ± 15% and the recovery yield of MeHg-spiked samples was 92 ± 12%.

The concentration of inorganic mercury reached 7.56 mg.kg-1 (wet weight) in pyloric caeca and the uptake rate was 0.22 mg.kg-1.d-1. Concentrations in gonads and endoskeleton were 10 to 20 times lower than in caeca (Fig. 1). The uptake rate of MeHg (0.17 mg.kg-1d-1) was slightly slower and the maximum concentration reached in caeca was 5.34 mg.kg-1. Mercury concentrations found in coelomic fluid were low and at least 100 times smaller than those in ceaca. No mercury was found in tissues of control starfish indicating that mercury excreted by diffusion in water by contaminated starfish was not re-adsorbed by other starfish in aquariums. The mercury load in each organ of stars was calculated and expressed as a percent (%) of total Hg uptake for each chemical species (Fig. 2). The inorganic Hg loads in pyloric caeca, gonads and endoskeleton reached a steady-state after only 10 days and remained unchanged up to the end of the experiment. The behaviour of MeHg was totally different as the loads in caeca decreased from 95% to 65% but increased from almost zero up to 30% in endoskeleton. Finally, the retention (%) of mercury species was calculated by dividing the actual total amount of mercury in each starfish by the total amount of mercury received from mussels (Fig. 3). The retention of inorganic Hg was about 50% throughout the exposure period whereas the retention of MeHg increased up to 90-95% at the end of the experiment.

A kinetic model, based upon the assumption that the uptake process of Hg species in the digestive system is quite similar to an ion-exchange adsorption mechanism between a contamination solution and a solid surface, was developed. The integrated equation of the rate law was expressed as:

(ln(C₀-q))/q₀-q = K'(C₀m-q₀)t + ln (C₀m/q₀)

were Co is the initial concentration of mercury in the prey, qo is the maximun amount of mercury being absorbed, and m stands for the wet weight of the animal. This equation allowed the plot of its left-hand side against time and the slope provided an estimation of an apparent exchange rate constant for each mercury species (Fig. 4). The rate constant k'MeHg was slightly higher than k'Hg, indicating a faster exchange rate for MeHg between digested mussel tissues and binding sites (and also between sites) in caeca. However, a faster exchange rate do not mean a faster uptake rate because transport rate towards other organs also play an important role in the whole bioaccumulation process.

In conclusion, starfish, by its ability to digest all ingested tissues, seems to be in a position to play a major role in the sequestration of methylmercury (high retention) from mussels and from other potential preys and in recycling inorganic mercury in solution (low retention). Starfish can be seen as "marine digester" which engulfs bivalves and other invertebrates (following species) and sends back soluble metabolites and trace metals which have not been bioaccumulated.

FR :

La réutilisation des eaux usées est reconnue comme une technique importante dans les régions arides et /ou grandes consommatrices d'eau. L'une des méthodes actuellement très employée consiste à recharger la nappe phréatique avec des effluents secondaires via des bassins d'infiltration. L'épuration biologique et / ou chimique à travers la zone non-saturée représente une caractéristique importante de cette technologie. Les procédés de ce type sont connus sous l'appellation de Soil Aquifer Treatment (SAT) ou géofiltration. Dans ce travail, le procédé a été étudié comme méthode de réhabilitation d'un effluent secondaire d'eaux usées jusqu'au stade d'eau potable. Cette recherche a été principalement axée sur le comportement, le transport des matières organiques (MO) de l'effluent et particulièrement sur leur rôle de précurseurs potentiels de sous-produits de désinfection lors de la réutilisation de la nappe. Dans la zone vadose, la matière organique est principalement éliminée par biodégradation, et à un degré moindre, par adsorption. Les simulations du procédé, en laboratoire, ont été réalisées en réacteurs recirculés aérobies, en mode cuvée, avec un biofilm acclimaté sur des particules de sable siliceux, afin de déterminer la fraction biodégradable des MO. L'évaluation de celle-ci est essentielle pour prédire leur potentiel de dégradation par la biomasse de la zone vadose. L'effluent mis en oeuvre est issu d'une station d'épuration de l'Arizona (États-Unis) avec biofiltre (lit filtrant à support plastique); sa concentration en carbone organique dissous (COD) se situe entre 10 et 15 mg/L.

L'effluent mis 5 jours durant en contact avec le biofilm acclimaté du réacteur montre un abattement de 50-60 % du COD. Il a ainsi été déterminé qu'environ 80 % de l'élimination des MO de l'effluent survient dans les premières 24 heures d'expérimentation, alors que le reste, près de 20 %, est éliminé durant les 48 heures suivantes. Dans ces conditions, le délai de 5 jours apparait suffisant pour dégrader les MO présentes dans ces effluents. Les rendements observés augurent bien de la dégradation dans la zone vadose si l'on tient compte de la combinaison des taux et de la hauteur d'infiltration avec des temps de résidence de 2 à 14 jours ainsi qu'il est proposé dans le procédé. Afin d'accroître la biodégradabilité des MO, une ozonation a été effectuée, en amont du bio-traitement, avec un générateur d'ozone à l'échelle du banc d'essai fonctionnant en mode semi-continu (admission continue de gaz, volume stable de liquide). La pré-ozonation a permis d'accroître la biodégradation de 60-70 %. Bien qu'un fort pourcentage de MO soit éliminé dans ce schéma, il ressort que l'ozone n'a qu'un effet modeste sur la transformation des MO dissoutes non-biodégradables en matières biodégradables par rapport à des expériences similaires effectuées avec des matières organiques naturelles (MON) des eaux de surface. L'eau usée ainsi traitée présente des niveaux de COD comparables à ceux d'une eau de surface employée à des fins de consommation. Les caractéristiques des MO de l'effluent ont été comparées à celles des MON. Une ultrafiltration de l'effluent pour déterminer le poids moléculaire apparent des MO, donne une distribution bimodale de leur poids moléculaire par rapport à une distribution logarithmique normale observée avec des MON typiques. En utilisant des résines non- ioniques pour séparer les fractions hydrophobes et hydrophiles des MO, il ressort que l'ozonation ne transforme pas de façon significative la fraction hydrophobe des MO de l'effluent en fraction hydrophile, tel que cela a été observé durant l'ozonation des MON. Ces eaux ont été chlorées en pilote, selon des conditions similaires à celles des réseaux de distribution (CI2:COD=1:1mg/mg, période d'incubation=24 heures) afin de simuler la post-désinfection après récupération. Les sous-produits réglementés (Trihalométhanes THM) et ceux proposés (Acides holoacétiques, HAA6) ont été formés à des taux inférieurs ou proches des normes en vigueur (ou de celles proposées pour HAA6) pour l'eau potable aux États-Unis. Cependant, une nitrification significative a été observée dans nos simulations de traitement par le sol avec un effluent non-nitrifié, conduisant à des teneurs en nitrates supérieures à la norme américaine pour l'eau potable (10 mg/L).

EN :

Soil Aquifer Treatment (SAT) is being studied as a means of water quality renovation of a secondary effluent for potable reuse. During SAT, infiltration of effluent through the vadose zone provides removal of effluent organic matter primarily through biodegradation, and to a lesser extent, through sorption. A major concern is that effluent organic matter can function as disinfection by-product (DBP) precursors upon recovery and post-disinfection. Through laboratory simulations of SAT, we have found that 50 - 60 % of the dissolved organic carbon (DOC) present in secondary effluent can potentially be removed by biodegradation, with preozonation increasing removal to 60 - 70 %. The resultant water contains DOC at levels comparable to a typical surface water used for potable purposes; moreover, upon chlorination, regulated DBPs are formed at levels below or near the United States of America's drinking water standards. However, significant nitrification was observed in our SAT simulations with an un-nitrified effluent, yielding levels of nitrate above the USA drinking water standard.

FR :

La prolifération bactérienne en réseaux de distribution d'eau potable est un souci majeur des distributeurs d'eau. La complexité des phénomènes impliqués dans la croissance bactérienne en réseaux nécessite une modélisation mathématique pour définir l'impact des différents paramètres de la qualité de l'eau et généraliser ces résultats à l'échelle du réseau de distribution.

Une approche déterministe a été choisie pour développer cette modélisation prédictive de la croissance bactérienne dans les systèmes de distribution. Le modèle prend en compte : la croissance de fa biomasse libre et de la biomasse fixée, la consommation en nutriments exprimés par le CODB, l'action bactéricide du chlore sur la flore libre et la dore fixée, la déposition des bactéries en suspension et le détachement des bactéries fixées. Le modèle propose une approche originale pour la modélisation de l'action bactéricide du chlore. Par ailleurs, différentes formulations du détachement ont été testées algébriquement pour définir la modélisation la plus adaptée à notre système d'équations.

Ce modèle a été couplé au logiciel de modélisation hydraulique IMCCOI.O développé par la SAFEGE. Utilisant les données hydrauliques et de géométrie générées par PICCOLO, le modèle prédit les numérations bactériennes en chaque noeud et sur chaque arc du réseau de distribution. Utilisant l'interface graphique de PICCOLO, le modèle permet une visualisation de l'évolution de la qualité bactérienne par cartographie.

Des simulations ont été réalisées sur de nombreux réseaux présentant des tailles et des niveaux de complexité variables. Le modèle a été validé à partir de campagnes de prélèvements sur sites.

Ce modèle permettant de simuler l'évolution de la qualité bactériologique à l'échelle du réseau est un outil unique pour le diagnostic et la gestion qualitative des systèmes de distribution d'eau potable.

EN :

Of the many causes of distributed water quality deterioration, biological phenomena are undoubtedly the subject of the most study, and are also the most closely monitored because of short-term public health risks. Although high heterotrophic bacterial counts do not necessary constitute a health risk, they are the sign that a particular network is subject to biological disorders which can protect pathogenic species. What is more, the evolution of the bacterial biomass in the network also affects other aspects of distributed water quality, such as tastes and odours, the development macro-invertebrates, the appearance of colour and turbidity and the appearance of biocorrosion phenomena.

Qualitative management of distribution networks is therefore to ensure that the quality of the product is kept as constant as possible up to the farthest points of the distribution. With this in mind, it is essential to understand, describe and model the various phenomena which lead to the evolution of water quality during distribution. Mathematical modelling is necessary in order to take ail parameters into account in view of the complexity of the different phenomena involved.

A determinist type modelling was developed to predict bacterial variations (viable and total bacteria) during distribution. The model takes into account:

- the fate of available nutrients consumed for the growth of suspended and fixed bacteria,

- the influence of temperature on bacterial dynamics,

- the natural mortality of bacteria by senescence and grazing,

- the mortality resulting from the presence of chlorine disinfectant, with a differentiation between the action on free et fixed bacteria,

- the impact of different forms of chlorine in water (HCIO/CIO-) dépending on pH on the mortality rate,

- the deposition of suspended bacteria and the detachment of fixed bacteria,

- the chlorine decay kinetics onder the influence of temperature, hydraulics and pipe materials.

The modelling of the fixed biomass as a layer uniformly distributed over the pipe surface, expressed as an équivalent thickness of carbon, has been adopted. By this way, a differentiation between the mathematical expression of the free and that of the fixed biomass was made in the model. This mean it is possible to distinguish between phenomena depending on their locations: reactions in solution, réaction at the water/biofilm surface interface and within the biofiJm.

This model proposes also an original approach for chlorine bactericidal action on suspended and fixed biomass. To model the action of chlorine on the fixed biomass and its stronger résistance compared with the free biomass, the diffusion of the chlorine through the boundary layer and the biofilm has been taken into account. This calculation of the average penetration depth of the chlorine front into the biofilm enables the identification of two layers: a chlorinated layer and a layer not attained by the chlorine which provides a material indication of the better resistance of the fixed biomass.

As detachment is a key phenomenon in the modelling of bacterial dynamics in distribution Systems, the influence of different formulas of detachment kinetics on the mathematical expression of model variables were determined by soiving model equations.

The model has been interfaced with PICCOLO software, the SAFEGE hydraulic calculation model. It is constructed by using hydraulic results previously generated by PICCOLO and a numerical scheme to predict bacterial count at each node and on each link of a network. Installed on a PC type computer, the model uses the graphic interface of PICCOLO and provides an effective and easy way to visualise on a computer screen water quality variations in the network, using a colour code for bacterial count, nutrient concentration and chlorine residual.

The first model calibration was done using data from our pipe loop pilot under various operating conditions. The model has been also used to simulate a variety of distribution Systems of different sizes and levels of details and a validation of the model has been carried out by means of measurement campaigns on different distribution Systems.

Animating and visualising variations of bacteria counts in distribution system is an unique approach to study the changes in water quality. This tool is helpful to propose strategies for the management of distribution Systems and treatment plants and define the different zones of bacterial regrowth in relation with hydraulic conditions.