Volume 8, numéro 4, 1995

FR :

Sur la base d'études microbiologiques expérimentales réalisées sur différents réseaux de distribution d'eau potable en France, un modèle de la dynamique des bactéries et du carbone organique biodégradable (CODB) a été développé ( le modèle SANCHO).

Les processus suivants ont été pris en compte :

- l'hydrolyse bactérienne exoenzymatique de la matière organique et la croissance des bactéries libres et fixées, à partir des produits des hydrolyses ;

- la mortalité des bactéries, source de relargage de matière organique ;

- les interactions entre les surfaces internes des canalisations et les bactéries avec un processus d'adsorption réversible ;

- la consommation de chlore par réaction avec la matière organique et l'effet du chlore libre sur l'activité des bactéries en suspension et fixées qui a été déterminé expérimentalement.

Le modèle calcule, en considérant les cinétiques présentées ci-dessus, les variations spatiales à l'état stationnaire du CODB, du résiduel de chlore libre, de la biomasse bactérienne libre et fixée. Ce calcul s'effectue pour une masse d'eau circulant durant des temps connus dans des canalisations de diamètres décroissants.

Afin de valider ce modèle, des comparaisons entre les résultats expérimentaux et calculés par le modéle ont été effectuées pour deux réseaux de distribution différents. Un bon accord est observé entre les résultats expérimentaux et ceux calculés pour les situations présentées qui sont caractérisées par des teneurs très différentes en chlore et en CODB dans l'eau alimentant les réseaux.

Une fois validé, ce modèle peut être utilisé afin de prédire l'effet de certaines modifications de la qualité de l'eau alimentant un réseau sur la dynamique bactérienne au sein de ce réseau. Il permet de définir la qualité de l'eau requise en sortie de filière afin de respecter une qualité souhaitée dans tout le réseau de distribution.

EN :

On the basis of experimental microbiological studies performed on different french distribution systems, a model of the dynamics of bacteria and biodegradable dissolved organic carbon (BDOC) in distribution networks was developed (the SANCHO model). The following processes are taken into account in the model: exoenzymatic hydrolysis of dissolved organic matter by bacteria and growth of free and fixed bacteria on the hydrolysis products; bacterial mortality which releases organic matter; reversible adsorption of bacteria and their biological attachment to inner pipe surfaces, chemical consumption of free chlorine and impact of free chlorine on the activity of free and fixed bacteria. The kinetics of these processes were all studied experimentally.

The model considers the case of a water body flowing down in successive pipes of decreasing diameters and calculates the spatial variations, at steady state, of BDOC and chlorine concentration, free and fixed biomass.

The SANCHO model was validated by the comparison of calculated and experimental data on various distribution systems which work over a large range of conditions. Good agreement between calculated and measured values of chlorine, BDOC, suspended and fixed bacterial biomass was found for the different distribution networks.

This model can be used to define the water quality required at the outlet of a given treatment plant in order to avoid problems of bacterial growth within the distribution system fed by this plant.

Keywords

.

FR :

La filtration successive des eaux, préalablement filtrées sur membrane de porosité 0,45 µm puis acidifiées à pH 2, sur deux colonnes de résines XAD8 et XAD4 placées en série, caractérise la matière organique en trois grandes classes : les substances hydrophobes, les acides hydrophiles et les hydrophiles non retenus. Cette étude s'est intéressée à déterminer et à analyser l'impact sur la distribution hydrophobe/hydrophile des traitements conventionnels utilisés dans les usines de production d'eau potable.

En ce qui concerne l'effet de la clarification, la distribution hydrophobe/hydrophile du COD a été inchangée dans le cas de sept prélèvements sur dix, indiquant une isoélimination de chaque fraction. Dans le cas des eaux R1.2, R4 et R6.1, une diminution de la fraction hydrophobe est obtenue entre l'eau brute et son eau clarifiée.

L'étude de différents coagulants indique que la distribution de la matière organique des eaux clarifiées dépend de sa composition initiale mais aussi de la nature du coagulant mis en oeuvre. Le chlorure ferrique permet une meilleure élimination de chaque fraction comparativement à un sel d'aluminium.

Un traitement d'oxydation à un taux de 1,5 mg Oxydant/mg COD et quel que soit l'oxydant mis en oeuvre (O3, Cl2, ClO2) diminue significativement les substances hydrophobes et augmente les substances hydrophiles. Ces résultats sont logiques si nous nous référons au mode d'action de ces trois oxydants. L'ozonation d'eaux brute et clarifiée s'accompagne aussi d'une augmentation du CODB, soulignant que les substances hydrophiles générées par l'étape d'ozonation seraient biodégradables.

Des campagnes de prélèvements ont indiqué que les traitements de finition (ozonation, filtration sur CAG) en configuration industrielle ont une légère influence sur la distribution interne de la matière organique. Les résultats, du fait de taux de traitement plus faibles, sont moins nets que ceux obtenus en laboratoire. Une filtration sur CAG, précédée d'une étape d'ozonation, conduit à une augmentation de la fraction hydrophobe et à une diminution du CODB.

EN :

Dissolved organic matter (DOM) in natural waters is a mixture of compounds. Some authors have fractionated this organic matter into large classes (humic substances, hydrophilic acids . . .). Humic substances have been defined as the fraction of organic matter retained on XAD8 resin at acidic pH (THURMAN and MALCOLM 1981); this isolation procedure is recommended by the International Humic Substances Society. Only a few investigators have dealt specifically with hydrophilic substances which are not adsorbed on XAD8 resin at acidic pH. The "hydrophobic/hydrophilic" distribution can be determined by a simple method of organic matter fractionation, using two superimposed XAD8 and XAD4 resin columns (CROUE et al. 1993). This procedures, carried out at pH 2, consists of first isolating hydrophobic substances (essentially humic acids) on an XAD8 resin and then isolating the hydrophilic acids from the XAD8 effluent on an XAD4 resin.

The aim of this work was to study the evolution of the hydrophobic/hydrophilic distribution during water treatment steps as applied in waterworks. The analytical procedure was first applied to determine the change in the DOM distribution of ten surface waters after clarification. The results obtained after clarification were completed by the study of the effect of the coagulant nature (ferric chloride, aluminium sulphate, prehydrolyzed salt (WAC)). In the second phase, oxidation experiments using ozone, chlorine and chlorine dioxide were conducted on raw and clarified waters to determine their effect on the DOM distribution. Finally the fractionation procedures were carried out on two water treatment plants to compare our laboratory data with results obtained in a working plant and to observe the change in DOM distribution during granular activated carbon (GAC) filtration.

Dissolved organic carbon (DOC) concentrations were analysed using a Dohrman DC 80 apparatus. UV absorbance was measured with a one or five centimeter cell using an Uvikon spectrophotometer. Oxidation experiments were carried out in a batch procedure. The study of different coagulants was made in the laboratory with a Jar Test procedure described elsewhere (LEFEBVRE 1990). Biodegradable dissolved organic carbon (BDOC) was analyzed according to the suspended bacteria method (SERVAIS et al. 1987, 1989).

The hydrophobic/hydrophilic distributions of raw and clarified waters were found to be unchanged by the clarification step in seven of ten test waters. These results indicate that for a studied water, the relative reduction in DOC of one fraction was of the same order of magnitude as the other fraction. The distributions of R1.2, R4 and R6.1 were significantly modified by the clarification treatment. In these three cases, the hydrophobic substances showed the greater DOC reduction. The "non-retained" hydrophilic substances became predominant in the R6.1 clarified water, whereas in the cases of R4 and R1.2 the total hydrophilic fraction and the hydrophilic acids increased, respectively.

The results obtained in the laboratory on three different raw waters, clarified by different coagulants (ferric chloride or aluminium salts), showed that the nature of the coagulant (iron or alum) can influence the hydrophobic/hydrophilic distribution. A marked influence was found in the case of R6 and less significant results were obtained for R1.2. Powdered activated carbon had no real effect on the DOC distribution, under our experimental conditions (applied dosage 25 mg/L). If the removal of each fraction is considered, and if it is assumed that each removal is independent from the others, ferric chloride appears to be the best coagulant. It removes humic substances efficiently as well as hydrophilic acids (> 72%), but is less efficient for the "non-retained" hydrophilic fraction. The DOC distribution of a clarified water depends on the distribution in the original raw water and the nature of the coagulant.

Under our experimental conditions (applied dosage: 1.5 mg oxidant / mg DOC), ozone, chlorine and chlorine dioxide significantly affected the DOM distribution of the R1.2 water. The hydrophobic substances showed the higher relative DOC reduction, which can be correlated with an increase in the hydrophilic fraction. The greatest change was obtained for treatment with ozone. In the case of chlorine and chlorine dioxide, an increase of the "non-retained" hydrophilic fraction was observed whereas for ozone the two hydrophilic fractions increased. These results are in agreement with current knowledge about the action of ozone, chlorine and chlorine dioxide.

An increase in the BDOC fraction was observed with the applied ozone dosage. In the case of R1, ozonation of raw and clarified waters appeared to shift the dissolved organic carbon distribution towards the "non-retained" hydrophilics. A good correlation exists between the BDOC increase (BDOC/BDOCo) and the decrease of hydrophobic acids or the increase of "non-retained" hydrophilics (slopes are respectively 12.3 ; - 16.5 and 15.5). In the case of another sample of R1.2, the comparison of BDOC in the XAD4 effluents of raw and ozonated waters indicated that at least 62% of BDOC produced by ozonation was in this fraction. This result indicates that the "non-retained" hydrophilics of this raw water are not biodegradable compared with those induced by ozonation.

Results obtained on samples taken on two water treatment plants indicate that ozonation and GAC filtration have a small effect on the hydrophobic/hydrophilic DOM distribution. Intermediate ozonation at the industrially applied dose slightly modifies the distribution; a slight decrease of hydrophobic substances is observed. GAC filtration induces an increase of the hydrophobic fraction.

The water treatment process includes clarification (EFS), inter-ozonation (EFSO3), and GAC filtration (EFCAG). GAC filters A and B had been respectively running for one and three years; filter C has been regenerated one month before. Good organic matter removal is obtained during clarification (removal of DOC and UV-absorbance: 75% and 88%). This treatment step changes the DOM distribution: increase of the hydrophobic fraction and decrease of the hydrophilic fraction. BDOC was completely removed. Intermediate ozonation (0.7 mg O3/mg DOC) modifies the DOC distribution and creates BDOC. This BDOC, in absolute value (0.36 mg/l), is equal to the increase of hydrophilic fraction. On the oldest filters (A and B) BDOC was completely removed, but this was not the case for filter C, probably because the bacterial biomass was insufficient. On GAC filters hydrophilic substances were found to be better removed than humic substances. The age of the GAC bed seems to have an effect on DOC distribution: the effluent from the oldest GAC filter contained less of the hydrophobic fraction than did the effluent from the youngest filter.

FR :

L'élimination de la matière organique et la réduction de volume des boues peuvent être obtenues par incinération, par oxydation sous pression en milieu humide ("wet air oxidation") ou par combustion en eau supercritique ("supercritical water oxidation"). Une étude en autoclave agité a permis de comparer sur une même boue d'épuration les performances des deux techniques d'oxydation voie humide et d'oxydation supercritique, en mettant l'accent sur les sous-produits résiduels en phase liquide et la composition de la phase gaz.

Les résultats obtenus montrent que l'élimination de la DCO dépend fortement de la température: l'abattement de la DCO passe de 70 % à 235 °C à 94 % à 430 °C. L'azote organique de la boue est transformé en NH4+ mais seule une élimination limitée de l'azote totale est obtenue à 430 °C. Les sous-produits résiduels dans la phase liquide sont constitués en majorité d'acides gras, d'aldéhydes et de cétones, l'acide acétique étant prédominant.

Hormis le CO2, les sous-produits gazeux majeurs formés par des réactions complexes comme la pyrolyse, le réformage et la méthanation sont CO, H2 et CH4. Dans les conditions supercritiques, tous les sous-produits gazeux sont fortement oxydés. L'augmentation de la température de traitement permet d'obtenir un résidu solide de plus en plus inerte, les cendres obtenues en conditions supercritiques contenant moins de 1 % de matière organique.

Les performances des deux procédés étudiés laissent envisager leur développement à moyen terme comme voies alternatives d'élimination des boues.

EN :

Context

As the number of wastewater treatment plants increases, and the efficiency of treatment improves, the problem of how to dispose of the ever increasing amounts of generated sludge has intensified. For the beginning of the next century 1 million tons of sludge will be produced annually in France; disposal in landfills will be impossible and agricultural use could be limited by tight quality standards. Therefore, the development of effective and acceptable sludge processes is urgently needed.

Destruction of organic matter in sludge and large reductions in sludge volume are achieved either by incineration or by wet air oxidation (WAO), which needs no fuel and generates no smoke, fly ash or emissions of NOx and SO2. Supercritical water oxidation (SCWO) offers an attractive alternative. Water, above its vapor-liquid critical point of 374°C and 221 bar, is an excellent solvent for organic compounds and becomes completely miscible with oxygen. Reported results of sewage sludge SCWO demonstrate rapid and effective treatment. The objective of this study was to compare sub- and supercritical water oxidation of sludge in terms of organic matter destruction and formation of by-products in both gas and liquid phases.

Methodology

Oxidation of sludge was studied in a 0.5 L batch reactor rated for 450°C-300 bar. The raw material was a biological sludge containing 4% solids with a chemical oxygen demand (COD) value of 52 g/L. In the standard experimental procedure, 100 mL sludge were heated up to reaction temperature and oxygen was then introduced in 50% excess with respect to COD. Heating was maintained during 1 hour before slow cooling to room temperature. The overall organic destruction was quantified in terms of total organic carbon (TOC) and COD. Gas and liquid phases sampled at room temperature after reaction were analyzed by gas chromatography (GC). Sulfur and nitrogen species were also analyzed.

Results

When the temperature increased from 210 to 383°C, COD destruction increased significantly (Table 1). At 383°C, a COD destruction efficiency of 94.3% was obtained. However, at 430°C, organic matter oxidation was only marginally improved. In WAO tests, considerable acetic acid was produced and remained in the substrate. The produced acetic acid was oxidized rapidly under SCWO conditions. Surprisingly, the concentrations of the other volatile fatty acids (VFA) remained approximately constant between 310 and 430°C (Table 2). In addition to VFA, which represent ca. 50% of the residual COD, oxygenated organic compounds such as aldehydes, ketones and alcohols were produced (Table 3).

The data in Table 4 show that decomposition of organic nitrogen compounds into ammonia was completed at 383°C, while nitrates were reduced to N2 by reaction with organic matter and ammonia. NOx were not detected in the gas phase. The low reactivity of ammonia in supercritical water had been previously demonstrated. At 430°C, ammonia removal from sludge was limited to 15%. On the other hand, even in WAO conditions all sulfur species were totally converted to sulfate.

Under subcritical conditions, the gas phase contained significant concentrations of hydrogen and carbon monoxide in addition to water, residual oxygen and carbon dioxide. Traces of methane and C2-C3 hydrocarbons were also detected (Figs. 1 and 2). These gases result from a complex chemistry including pyrolysis, steam reforming and water-gas shift. Under supercritical conditions, all these compounds were extensively eliminated by oxidation. Under supercritical conditions the residual solids contained less than 1% organic matter. By X-ray diffraction hydroxyapatite, quartz and kaliophilite were identified in the residual solids.

These results confirm that supercritical water oxidation is a new sludge treatment concept of great interest. The degree of conversion of organic carbon is high, while liquid and especially gaseous by- products are produced in minor amounts compared to subcritical conditions. Temperatures higher than 430°C would be needed for substantial nitrogen removal.

FR :

La charge énergétique en adénylates (CEA) en tant qu'indicateur de pollution sublétale est envisagée pour un invertébré aquatique Asellus aquaticus L. ( Crustacé, Isopode ). Après contamination par du lindane pendant 48 heures aux concentrations de 2, 4, 8, ou 10µg/l, le dosage des différents adénylates ( ATP, ADP, AMP ) est réalisé par la méthode de bioluminescence. La CEA ( ( ATP ) + 1/2 ( ADP ) / ( ATP ) + ( ADP ) + ( AMP ) ) est calculée. Elle est comparée à l'échelle des valeurs établies par Ivanovici pour caractériser le degré de pollution d'un milieu. Sur cette base on peut dire que la CEA du témoin (0,77) et celles des organismes soumis à 2 µg/l de lindane (0,79) ne sont pas significatives d'une modification du milieu. Par contre celles observées chez les aselles contaminées par 4, 8 ou 10 µg/l de lindane, soit respectivement 0,69; 0,65 et 0,72 témoignent d'une perturbation du milieu et permettent de prévoir une récupération possible pour ces organismes lors d'un retour dans un milieu non pollué. Cependant nous notons que seules les valeurs observées pour 4 et 8 µg/l de lindane sont statistiquement différentes de la valeur témoin. Par ailleurs à 10 µg/l de lindane on observe une tendance à l'augmentation de la CEA qui est proche de celle du témoin.

La valeur de la CEA semble rendre compte de la perturbation physiologique consécutive à la phase d'hyperexcitabilité déclenchée par ce type d'insecticide; par contre à la phase la plus avancée (paralysie) de l'intoxication, la CEA prend une valeur proche de celle du témoin.

EN :

Context

The objective of this study was to evaluate the applicability of adenosine energy charge (AEC) as an indicator of sublethal pollutant contamination for an aquatic invertebrate Asellus aquaticus L. ( Crustacea, Isopoda ). This study was carried out under laboratory conditions.

Asellus collected in natural ponds were acclimated in the laboratory during a minimum period of 15 days. Individuals between 3 and 7 mg in weight were selected and kept at 15°C for 24 hours before contamination with lindane. Contamination was performed in glass containers in 250 ml of water, and 1 ml of lindane acetone solution. Concentrations of 2, 4, 8 and 10 mg/l were tested. The experimental period was 48 hours. After cold-induced anesthesia, Asellus individuals were rapidly dried and then dipped into liquid nitrogen and ground to a powder at - 80°C. Extraction of adenylates was performed with dimethyl sulfoxide (DMSO). ADP and AMP were converted to ATP with pyruvate kinase and phospho-enol-pyruvate for ADP; pyruvate kinase, myokinase and phospho- enol-pyruvate for AMP. The concentrations of ATP, ADP and AMP were measured using a bioluminescence technique with a luminometer (LKB Wallac 1250). AEC, defined as the ratio between (ATP + 1/2 ADP) and (ATP + ADP + AMP) concentration, was then calculated.

AEC values were 0.77, 0.79, 0.69, 0.65 and 0.72 respectively for control animals and for Asellus specimens exposed to 2, 4, 8 and 10 µg/l lindane. According to IANOVICI (1979), AEC values for Asellus contaminated with 4, 8 or 10 µg/l of lindane were representative of the perturbation of environmental conditions. Nevertheless, these values show that recovery is possible if environmental conditions return to normal. However statistically significant differences (ANOVA, p=0.05) were noted only between control and 4 or 8 µg/l lindane contaminated Asellus.

ATP concentrations were 0.1451, 0.1876, 0.1821, 0.2325 and 0.1570 µmol/mg respectively for control, 2, 4, 8 and 10 µg/l of lindane. No significant difference was noted between control and contamination, except for 8 mg/l of lindane ( p=0.01 ). ADP concentrations were 0.0698, 0.0253, 0.1200, 0.2121 and 0.0679 mmol/mg respectively for control, 2, 4, 8 et 10 mg/l of lindane. Only the ADP concentration for 8 mg/l of lindane was significant of ADP accumulation (ANOVA, p=0.05 ). AMP concentrations were 0.0193, 0.0272, 0.0470, 0.1182 and 0.0416 mmol/mg respectively for control, 2, 4, 8 and 10 mg/l of lindane. The increase of AMP concentration for 8 mg/l of lindane was significant ( risk 0.05 ). Variations of the adenylate pool (ATP + ADP + AMP) were 0.2342, 0.2401, 0.3491, 0.5628 and 0.2665 mmol/mg respectively for control, 2, 4, 8 et 10 mg/l of lindane. The increase of the adenylate pool concentration for 4 and 8 mg/l of lindane was significant (p=0.05 ).

It appeared that the decrease of AEC at lindane concentrations of 4 and 8 mg/l was indicative of the increase of the energetic cost and the metabolism, resulting from the hyperexcitability characteristically induced by this category of contaminant. At 10 mg/l of lindane, the AEC value was approximately equal to that of the control exposure. It appeared correlated to the decrease in metabolic activity and accompanying reduction energy expenditure in response to the paralytic phase of intoxication.

Finally under laboratory conditions AEC values appeared to be indicative of sublethal contamination, for this species and this toxicant. However for acute exposures it does not appear that AEC is a very good indicator.

FR :

Cet article présente l'ensemble des biocapteurs en cours d'étude et proposés pour le contrôle en continu, automatisé et in situ de la qualité des eaux. Le principe des systèmes, étudiés jusqu'ici majoritairement en laboratoire et sur pilote, sera donné avec leurs performances au plan sensibilité et spécificité de détection des polluants hydriques. Ces performances conditionnent leur domaine d'application : les systèmes très sensibles étant affectés au contrôle des eaux d'alimentation et des eaux souterraines, les moins sensibles au contrôle des effluents très contaminés.

Les biocapteurs peuvent se caractériser par deux de leurs composantes principales :

- le réactif biologique ou biocatalyseur, sensible au(x) polluant(s);

- le détecteur appelé transducteur, qui traduit la réponse biologique du biocatalyseur en un signal électrique. Le transducteur peut être de type optique, électrochimique, ampérométrique principalement, ou piézoélectrique.

Trois grands types de biocapteurs peuvent être distingués selon la nature du biocatalyseur :

- les bioréacteurs, basés sur l'étude des réponses comportementales des vertébrés (poissons) et d'autres organismes aquatiques (microcrustacés, bivalves):

- les biosondes cellulaires reposant sur l'étude des fonctions métaboliques telles que la respiration, la bioluminescence, la photosynthèse de microorganismes immobilisés (bactéries, microalgues, levures) ou libres (boues activées) dans le milieu analysé:

- les biocapteurs "d'affinité" basés sur l'utilisation d'enzymes ou d'anticorps, chargés de détecter respectivement les substrats et inhibiteurs enzymatiques spécifiques, ou les substances antigéniques vis à vis desquelles les anticorps ont été développés. Ces systèmes sont, par principe, les plus spécifiques mais aussi les plus sensibles. Ils ne couvrent, cependant, qu'une gamme encore très limitée de micropolluants hydriques.

Le degré d'autonomie d'un biocapteur, sa facilité d'utilisation et de maintenance et sa fiabilité, sont des éléments qui rentrent en ligne de compte dans les performances. Ces qualités devront être évaluées lors de la phase de validation in situ, essentielle et déterminante pour juger de l'intérêt du système en conditions de fonctionnement réel.

EN :

Context

This paper reviews the use of biosensors for environmental biomonitoring and especially for the detection of water pollutants. These systems are developed in view of on-line applications, continuous and real time analysis. The principle and the design of the different systems proposed for this purpose are described with their performances deduced from pilot or in situ studies carried out up to now. Automation and autonomy, sensitivity and specificity are critical points that will determine the success of their applications in biomonitoring and the kind of application that can be envisaged. It is necessary they require minimal human intervention for maintenance and working . The more sensitive systems can be used for the monitoring of drinking and ground waters, the less sensitive ones for the monitoring of complex effluents, more heavily contaminated.

Biosensors can be distinguished on the basis of the type of biocatalyst associated with thetransducer: the biological signal delivered by the biocatalyst is transmitted to a detector, also called transducer. The transducer, which may be an optical, electrochemical or piezoelectrical detector, transforms the biological response into an electric signal. This signal can be easily amplified and interpreted in terms of the toxicity and level of pollution of the analyzed sample.

Three categories of biosensors can be defined:

- biosensors using aquatic vertebrates and invertebrates: fish, microcrustacea, bivalves. Their behavior in the tested medium is studied as the criterion for toxicity;

- cellular sensors, measuring physiological and biochemical functions such as respiration, bioluminescence, and photosynthesis, in microorganisms immobilized on the transducer (bacteria, yeast, microalgae,..) or suspended in the tested medium (activated sludge);

- biosensors measuring an "affinity" response and a specific binding between enzyme/substrate or antibody/antigen. These systems use enzymes or antibodies immobilized in close contact with the transducer; they may detect the (analogs of) enzymatic substrates and inhibitors, or the (analogs of) antigenic substances binding to the antibody. These systems appear promising on the basis of their sensitivity. At present they can be applied for the detection of triazines and phenols. Such systems need to be developed and extended to other pollutants in order to cover the wide range of aquatic contaminants.

User-friendliness, attendance and maintenance requirements, and service life are other critical aspects affecting the performances of a biosensor. These qualities need to be evaluated during the validation step of the equipment. In situ validation is essential for evaluating the relevance of the system in environmental biomonitoring and its applications. It is probable that among the numerous systems proposed as biosensors, only a few will be considered as suitable tools for on-line monitoring of waters.

FR :

Cet article décrit une procédure de décision destinée à la prévision des crues en temps réel. La méthode proposée permet de combiner les résultats de différents modèles avec une pondération adaptée aux conditions de fonctionnement du système de prévision.

Chaque modèle a lors du calage des performances qui lui sont propres ainsi qu'une sensibilité particulière aux différentes perturbations. Les résultats de ces modèfles étant corrélés, la décision optimale est par conséquence une combinaison linéaire de la décision qui peut être prise dans chaque état du système pondérée par la probabilité de se trouver dans une configuration donnée.

Dans le cadre d'un système de prévision exploitant le résultat de plusieurs modèles, la procédure se décompose en deux étapes. Pendant la première étape, après analyse de l'ensemble des informations collectées on détermine les états possibles du système. La deuxième étape établit une prévision en calculant des poids optimaux accordés à chaque modèle en fonction des différentes configurations du système.

Un tel système de décision nécessite une étude complète afin de décrire et modéliser toutes les perturbations et situations hydrologiques possibles. Une procédure simplifiée a donc été développée identifiant seulement deux états du système : le premier où tout fonctionne correctement et le deuxième déterminé à partir des derniers pas de temps connus.

Comparativement à des méthodes de changement de modèles, cette méthode simplifiée a montré sur la Vézère de meilleurs résultats.

L'intérêt de cette approche est confirmé même dans le cas où des erreurs accidentelles interviennent en temps réel. On présente les performances de cette procédure, observées depuis son implantation sur le système d'annonce de crues du bassin supérieur de la Garonne.

EN :

Context

This article describes a decision process for real-time flood forecasting. A method is presented which combines the results of different models with weights adapted to the state of the flood forecasting system. Each model has specific performances when it is calibrated, and has a specific sensitivity to each perturbation that can occur. The results (forecasts) of all of these models are cross-correlated. For a wide class of adequacy criteria, the best estimate of the variable to be predicted is a linear combination of these results. In that combination the weight of each model forecast is strongly linked to the model's relative performance. For a specific perturbed environment (we will call it a state of the hydrologic hydrometric system) the performance of some sensitive models decreases but others may remain stable. Optimal weights are therefore specific for each state of the system.

When states can be pre-specified, the decision process splits into two different steps: diagnostic and prognostic. During a first diagnostic step, using various coherence tests among all collected data and possible external information that is not used in the model, the probability of being in each state is estimated. This step may include expert judgement as is the case in practice. During the second prognostic step, optimal weights of the model are computed, using the error covariance matrix for all models in each state of the system and the estimated probabilities of being in each state as determined during the previous diagnostic step.

Such a complete decision rule needs a detailed study that cannot be done for each case, and which is based on the assumption that all perturbations can be described and modeled. A simplified procedure has thus been developed based on the basic idea that only the normal state can be easily identified. Instead of estimating state probabilities, this procedure relies on weights that evolve according to recent errors. The results of this procedure, as applied to the Vérèze catchment, have been quite as good as those obtained with the complete pond.

These weighting decision procedures also proved in this case to be much more effective than any attempt to choose the best model at each time step. This very elementary result is in contradiction with classical approach of switching among multiple models. Switching procedures are in fact designed to estimate the state of the system, assuming that one must decide what is the state. For operational purpose such decisions are useless and switching procedures appear to be sub-optimal, as they lead to a discontinuous forecast decision. The interest of the multimodel approach is confirmed even if uncontrollable perturbations occur in real-time. The performances obtained since the implementation of this decision process in the flood forecasting system of the upper Garonne watershed are presented.

FR :

La station expérimentale pilote d'épuration des eaux usées par lagunage de l'École Inter-États d'Ingénieurs de l'Équipement Rural (EIER), Ouagadougou, Burkina Faso, reçoit les eaux usées des bâtiments administratifs et de l'internat des étudiants. Dans les eaux usées brutes, des oeufs d'helminthes (Ascaris lumbricoides, Ankylostoma duodenale), des kystes de protozoaires (Entamoeba coli, Entamoeba histolytica) et des larves d'Anguillule ont été mis en évidence.

Avec un temps de séjour de 16,4 jours dans les bassins de lagunage (2 à 3 heures dans le décanteur, 3,4 jours dans le premier bassin, 13 jours dans le deuxième bassin), les kystes d'Entamoeba coli et d' Entamoeba histolytica sont éliminés respectivement à 94 et 96 %, les oeufs d'Ascaris lumbricoides à 100%, les oeufs d'Ankylostoma duodenale à 90 % et les larves d'Anguillule à 92 %. Quand on considère tous les parasites confondus, le décanteur a un rendement éliminatoire de 33 %, le premier bassin 62 % (malgré la forte charge appliquée), le deuxième bassin 78 %. Le rendement global obtenu pour l'ensemble des parasites est de 94 %.

Si les évolutions des concentrations des parasites fluctuent dans le temps, il y a une différence très significative entre les concentrations obtenues en sortie du lagunage et celles des eaux usées brutes.

On a constaté l'absence des oeufs d'Ascaris lumbricoides pendant toute la période de l'étude ; il en a été de même pour les oeufs d'Ankylostoma duodenale durant une période de 10 mois. Par conséquent, les eaux usées épurées rejetées répondent aux recommandations de l'OMS quant à leur réutilisation agricole.

La mise en place des systèmes rustiques d'épuration des eaux usées dans les pays en voie de développement, surtout dans les zones où les ressources en eau sont limitées, pourrait contribuer sensiblement à la diminution des risques sanitaires liés aux pratiques courantes de réutilisation agricole des eaux usées en agriculture.

EN :

Experimental waste stabilization ponds to purify domestic wastewater from the École Inter-États d'Ingénieurs de l'Équipement Rural (EIER) were built in Ouagadougou, Burkina Faso. The waste stabilization ponds system consists of a raising tank, a decanter, and two ponds in series. The first pond is 1.22 m deep with a surface area of 62 m²; the second pond is 1.07 m deep and covers an area of 340 m². The inflow averaged 22 m³/d, giving a normal retention time of 3.4 days in pond number 1 and 13 days in pond number 2. The average daily BOD5 load was 1500 kg/ha/day for pond number 1 and 200 kg/ha/day for pond number 2.

Different parasites were identified in the raw wastewater, such as cysts of Entamoeba coli and Entamoeba histolytica, eggs of Ascaris lumbricoides and Ankylostoma duodenale and larval stages of Anguillula . The quantitative estimation of cysts, eggs and larvae was obtained using a Mac Master cell after concentration by a formol-ether technique (RITCHIE, 1948). Of 24 samples of raw wastewater analyzed, all contained cysts of E. coli and E. histolytica and Anguillula larvae. Similarly, 100 % of final effluent samples (Fig. 2-B) contained cysts of E. coli and E. histolytica, but the concentrations were respectively 16 and 22 times lower than in raw wastewater. Six of the 24 samples (25 %) contained Anguillula larvae (Fig. 9), 17 % contained Ankylostoma duodenale eggs and no samples (0 %) contained Ascaris lumbricoides eggs (< 1 egg/L).

This study were not carried out on Ouagadougou urban wastewater, due to the lack of a wastewater collection network, but the raw wastewater of the central market of Ouagadougou was analyzed for comparison because it represented a larger population than that in the EIER. In the raw wastewater of the central market, the concentration of parasites was high: Ascaris lumbricoides (110 eggs/L), Taenia saginata (53 eggs/L), Ankylostoma duodenale (39 eggs/L), Trichuris trichiura (19 eggs/L), Entamoeba coli (552 cysts/L), Entamoeba histolytica (479 cysts/L), Anguillula (62 larvae/L). The concentration of parasites identified in raw wastewater from EIER (Entamoeba coli (395 cysts/L), Entamoeba histolytica (269 cysts/L), Ascaris lumbricoides (4 eggs/L), Ankylostoma duodenale (9 eggs/L), Anguillula (26 larvae/L)) seemed consistent with the contamination level of the users of EIER stabilization ponds system.

With a total retention time of 16.4 days in the stabilization ponds, removals of E. coli and E. histolytica cysts were respectively 94% and 96%. Ascaris lumbricoides and Ankylostoma duodenale eggs were 100 and 90% removed, respectively, and removal of Anguillula larvae was 92%. Grimason et al. (1995), suggest that cumulative retention time of up to 55.3 days at Eldoret, Kenya and 40 days at Meze, France were sufficient to remove between 99.1 % and 99.7 % of cysts of Giardia lamblia, respectively. Global efficiency of the system for all parasites was 94%. In the decanter, where retention time was 2-3 hours, the removal efficiency was 33%. With 3.4 days retention time, the first pond (anaerobic due to the high organic load applied) increases removal to 62%. The second pond (optional) afforded a 78% removal efficiency with a 13 days retention time.

If the number of parasites fluctuated with time, the number observed in the effluent was significantly lower than those observed in the influent.

The effluent of EIER waste stabilization ponds system was free of Ascaris lumbricoides eggs throughout the study period and free of Ankylostoma duodenale eggs during 10 months; the effluent thus satisfied the WHO recommendations for reuse in agriculture. Waste stabilization pond systems would seem to have a significant place in Sudan- Sahara area for removal of parasite ova and cysts, being economical and advantageous. They would also contribute to reduce sanitary risks associated with reuse of untreated wastewater in agriculture.