Conditions chimiques contrôlant l'atténuation naturelle des BTEX et solvants chlorés : un état des connaissances
O. Atteia and M. Franceschi
L'atténuation naturelle des BTEX (Benzène, Toluène, Ethyl-benzène, Xylène) et des solvants chlorés est de plus en plus étudiée en raison des potentialités offertes par cette technique de gestion. Cet article, après avoir présenté les aspects abiotiques de l'atténuation détaille les conditions chimiques nécessaires à la réalisation des réactions de biodégradation des polluants organiques. Les aspects thermodynamiques sont abordés afin de décliner les réactions possibles et celles qui ne le sont pas selon les environnements chimiques. La dégradation des BTEX est focalisée sur le benzène, produit le plus toxique et le moins dégradable sur la plupart des sites. Les détails de la dégradation du benzène sur le terrain sont analysés dans la littérature et leur comparaison permet de décrire les mécanismes responsables de celle-ci. Dans le cas des solvants chlorés, l'attention est portée sur le TCE (Trichloréthylène), produit le plus couramment rencontré sur les sites pollués. Une mise en parallèle des évolutions de teneurs observées et des conditions chimiques locales permet de mettre en évidence les conditions nécessaires à la dégradation du TCE, et de ses congénères, ainsi que les cinétiques de dégradation dans différentes conditions. La mise en évidence du rôle prépondérant des conditions chimiques conduit à remettre en cause l'utilisation répandue des constantes de dégradation du premier ordre et donne des pistes pour les modèles nécessaires à une prédiction plus fine de l'atténuation naturelle.
The increasing reliance on natural attenuation in dealing with contaminated sites in North America is the consequence of:
1. the extremely long duration and high cost of aquifer rehabilitation by classical methods, and
2. the discovery of natural biodegradation in many different situations.
However, the use of this management technique is questionable, as intrinsic biodegradation is highly dependent on chemical conditions and particularly on redox equilibria. This paper describes the role of these chemical conditions on BTEX and chlorinated solvent attenuation and, by analyzing the current research, we try to define current limits of the predictability of natural attenuation in field conditions.
Natural attenuation is defined as the sum of processes able to decrease the pollutant concentration at a sampling point in an aquifer. Several physical processes such as dispersion, retardation and solubility play a role in natural attenuation. However, only biodegradation can significantly reduce the overall amount of pollutants in an aquifer, thereby allowing the pollutant concentration to reach the low levels that are required by regulations. The physical processes cited above can be modelled at a site to account for their effect, but the main focus is on biodegradation.
A detailed analysis of the basic thermodynamics of redox reactions involved in biodegradation is necessary to describe the reactions that can potentially occur. A rough analysis shows that BTEX is mainly degraded by oxidation and therefore is degraded more efficiently in aerobic media. However, toluene (and sometime ethylbenzene and xylene) can be degraded by fermentation and thus degradation occurs even in methanogenic conditions. In contrast, chlorinated solvents are degraded mainly by reduction, with the exception of c-DCE (cis-dichloroethylene) and VC (vinyl chloride), which are degraded by reduction and oxidation, thus having two degradative pathways. An overall comparison of reaction rates obtained from laboratory and field experiments clearly demonstrates that under field conditions the supply of redox reactants is a limiting factor in the reaction kinetics.
Degradation of BTEX under field conditions has been widely documented, and toluene ethylbenzene and xylene degradation occurred in almost all chemical environments. The most persistent product observed in almost all the studies was benzene. Due to its persistence, and also its carcinogenic and toxic properties, we focussed on the results obtained for benzene. The kinetic constant for degradation of benzene under most field conditions ranged from almost no degradation in the reduced parts of the plume to fast degradation at the oxygenated border. Degradation under nitrate, methane or iron reducing conditions was almost insignificant, but degradation did occur under sulphate reducing conditions. A detailed analysis of the data on benzene degradation under sulphate reducing conditions showed that there is a competition between bacterial populations for electron acceptors. Benzene is degraded only if electron acceptors are in excess and if no other easily degradable carbon source is present.
The analysis of experimental data on chlorinated solvents is more difficult because fewer studies exist and the degradation processes are slower and more complex. Significant intrinsic biodegradation occurs mainly by reductive dechlorination, with co-metabolism being important only under modified conditions. In the field, PCE (perchloroethylene) and TCE degradation occurred only under methanogenic and sulphate reducing conditions, while c-DCE was degraded in oxygenated media and finally VC degradation occurred under almost all redox potentials. The kinetics of degradation were slow, with half-lives in the order of 1 to several years. It was shown that the variability of such constants was quite high within the same site. This variability could be explained by the availability of reducing species, particularly hydrogen. By comparing the estimated and real length of solvent plumes it was shown that biodegradation was more important than transport for the sites with the most reducing conditions. At other sites, the necessity of both methanogenic conditions and a sufficient pool of electron donors in the aquifer was demonstrated. The high toxicity of VC, when compared to TCE, was of lower concern since it was shown that the plume size was equal to or smaller than that of TCE. This was due to a fast degradation kinetics for VC observed under aerobic conditions.
In conclusion, the controversy surrounding the use of models based on first-order degradation constants arose because of the strong dependence of this constant on prevailing chemical conditions. If the target at risk is far away, use of the statistics on plume length existing for BTEX seems to be sufficient. However, when the benzene content is high and the target at risk is close, there is a need to predict the size of the reduced plume. The approach is the same for more substituted chlorinated solvents. The most important data, which are often missing, are the amount of total 'easily' degradable carbon (i.e. BTEX, short chain acids or alcohols) delivered by the source that will generate the reduced plume. In order to achieve a more precise prediction, models incorporating the whole redox chain need to be developed and tested against existing field data.
R. Ben Amar, E. Ellouze, B. B. Gupta and A. M. Ayad
Dans ce travail un procédé de traitement des effluents issus de l'unité de préparation des peaux des animaux au tannage (travail en rivière) en tannerie-mégisserie a été étudié en utilisant la technique de microfiltration tangentielle sur membrane minérale en céramique. Les performances de ce procédé en terme de flux de filtration et de rendement épuratoire dépendent aussi bien des paramètres hydrodynamiques de filtration que de la qualité des effluents (collectés en été et en printemps) issus des différents bains de traitement et de rinçage des peaux dans l'atelier de rivière. Le flux de filtration varie entre 15 l/h.m2 pour l'effluent de printemps et 90 l/h.m2 pour l'effluent d'été. Les paramètres hydrodynamiques optimaux ont également été déterminés: la vitesse de circulation U=3 m/s, la pression transmembranaire Ptm=2 bar et la température T=43°C. L'étude de la microfiltration à concentration variable conduit à des facteurs de concentration volumique (FCV) de 6,5 pour l'effluent de l'été et de 2,4 pour l'effluent de printemps.
The leather industry is responsible for the transformation of raw animal skin to a final form as shoes, bags, dresses, etc. This industry was known for centuries as a craft activity, and today with industrial development, environmental regulations and new emerging technologies, it has become necessary to include elaborate processes for its wastewater treatment. These industries consume a great amount of water. In Tunisia, more than 15000 tons of skin are treated per year, and about 600000 m3 per year of effluents are discharged. The waste water contains chemicals, fats, hair and protein, varying in composition depending on the season. Figure 1 represents the preparation of raw skin for the tanning operation and the amount of waste water produced. The amount of water used for the preparation of raw skin is about 70% of the total quantity of water used. This waste water has a significant polluting load (chemicals and organic matter), with 5000 - 7500 mg/l of COD and 100 to 150 mg/l of sulfur. Tunisian legislation and regulations concerning the standards for wastewater disposal are 1000 mg/l for COD, 3 mg/l for sulfur and a pH between 6.5-9. Different techniques for wastewater treatment such as: physico-chemical treatment, treatment by electrochemical oxidation and membrane technology were proposed. Wastewater treatment by microfiltration and ultrafiltration with mineral membranes is advantageous because no chemicals are used and it can be combined easily with other physico-chemical or biological pre-treatments. In this study, we have treated two types of effluents from the leather pre-treatment industry collected in the summer (effluent 1), and the spring (effluent 2) seasons. The physico-chemical characteristics of the two types effluents are given in Table 2. The filtration experiments were made on a test bench (Figure 2) equipped with a feed reservoir, a volumetric pump, a filtration module, flow meter, pressure transducers, a heat exchanger and control valves. Ceramic membranes of tubular geometry (7 channels), 0.08 m2 membrane surface area and of 0.1 µm (mean diameter) pores were used. During the microfiltration experiments, the following physico-chemical parameters were analysed in the permeate and retentate: turbidity, specific conductivity, pH, viscosity, chemical oxygen demand (COD), sulfur (volumetric method), fats (Standard JIS 0102.24.2), protein (using Kjeldahl nitrogen), and organic nitrogen. Hydrodynamic parameters such as temperature (25 < T < 50 °C), transmembrane pressure (1 < Ptm < 2.2 bar) and feed velocity (1 < U < 3 m/s) were fixed for experimentation. The COD concentration in the effluent was adjusted and kept constant at 5000 mg/l. The raw effluent was pre-filtered on a screen filter (150 µm pore size). For experiments with variable concentration, we regularly removed the filtrate and the concentration factor was represented by FCV=Vi / Vr, where Vi was the initial volume and Vr was the volume of the retentate. The performance of the microfiltration (J) was expressed in l/h×m2. The retention rate (TR) was defined by: TR=1 - (Cpermeate) / (Cfeed). The total hydraulic resistance (RT) was defined by Darcy's law: Jf=Ptm / µ RT. After each experiment, the membrane was regenerated following a standard protocol and it was verified by measuring water flux. Figure 3a represents the variation of the filtration flux with time for 4 different temperatures: 25 °C, 43 °C, 45 °C and 50 °C with effluent 1. The flux increased from 90 to 118 l/h×m2 when the temperature increased from 25 °C to 43 °C. After 90 min at 50 °C, the filtration flux was 123 l/h×m2. Table 3 shows that the viscosity of the effluent decreased with temperature, while the turbidity of the filtrate increased from 0.63 NTU at T=25 °C to 1.6 NTU for T=50 °C. The retention rate of COD was always superior to 50 %. On the basis of these results, we chose the optimum temperature of 43 °C for other experiments. Figure 4 summarises the variation of flux with transmembrane pressure at flow velocities of 1 m/s, 2 m/s and 3 m/s. The stabilized fluxes were practically the same for the flow velocities of 1 and 2 m/s (of the order of 80 l/h×m2), but were higher at 3 m/s (110 - 115 l/h×m2 at 2 bar). The physico-chemical characteristics of the raw effluent and the permeate obtained after 90 minutes of filtration are summarised in Table 4. Figure 7a shows the variation of filtration flux for 2 types of effluents. The filtration flux for the same conditions of experimentation and at stabilized conditions (at 90 min) was 118 l/h×m2 for effluent 1 and 20 l/h×m2 for effluent 2. The lower filtration flux for effluent 2 can be explained by high deposits of rejected matter on the membrane and in the pores. Table 5 gives a comparison of the characteristics of effluents 1 and 2 before and after microfiltration. At variable feed concentrations, FCV=6.5 for effluent 1 and FCV=2.4 for the effluent 2 and the stabilized flux was about 90 l/h×m2 for the effluent 1 and 15 l/h×m2 for the effluent 2. The time needed for treatment of effluent 1 was about 6 hours, while more that 16 hours was necessary for effluent 2. Table 6 provides physico-chemical characteristics for the two types of effluents. The contents of fat, protein, nitrogen and sulfur in the effluent were important factors for variation. These results indicate that microfiltration is very sensitive to the quantity of polluting matter present in the effluents, particularly sulfur and fat. Increased polluting matter in effluent 2 could be responsible for the membrane polarization and blocking of pores. The resistance model was used to verify this hypothesis. The irreversible resistance values for effluent 2 were greater, thus confirming the hypothesis that the increased adsorption on the membrane surface and passage of pores by the presence of sulfur and organic polluting matter. These experimental results confirm that the best performance can be obtained at the hydrodynamic conditions of: a temperature of 43 °C; a transmembrane pressure of 2 bar; and a flow velocity of 3 m/s. Seasonal variation changed the quality of effluents, which considerably affects the performances of the microfiltration. Effluent 2, which was obtained from the treatment of sheep skin during the spring season, led to more membrane pore blocking than effluent 1 for the same initial concentration in COD. The interactions of fats and sulfur with the membrane layer appear to play an important role in the formation of a cake layer.
Modélisation du lessivage des bromures dans des cases lysimétriques par la méthode numérique des lignes
R. Majdoub, J. Gallichand and J. Caron
La modélisation du transport des solutés dans un milieu non-saturé repose habituellement sur l'équation de dispersion-advection (EDA). Un modèle numérique (TSOL) a été développé en couplant l'EDA avec l'équation de Richards et en incluant le prélèvement de l'eau par les plantes. La résolution numérique a été effectuée par la méthode numérique des lignes (MNL) qui présente une grande simplicité de programmation et résulte en une très bonne précision numérique. La précision de TSOL a été testée avec les résultats d'un modèle d'éléments finis (HYDRUS), et avec des données expérimentales (profils de concentration et masses de bromures récupérés) collectées pendant 195 jours dans trois cases lysimétriques installées sur un sol non remanié cultivé en pommes de terre. La comparaison entre TSOL et HYDRUS montre que la solution de la MNL est similaire à celle des éléments finis. Toutefois, pour l'ensemble des cases et des profondeurs, les modèles ont montré une surestimation des valeurs de concentration avec un écart moyen entre les concentrations mesurées et simulées par TSOL variant de 22 à 112 mg/l. Pour les cases B et C, l'erreur moyenne de biais hebdomadaire entre TSOL et les masses de bromures récupérés, était d'environ 5 mg/semaine. Dans le cas de la case A, l'erreur moyenne de biais hebdomadaire de TSOL était de 39 mg/semaine.
Simulation of solute transport under transient unsaturated conditions is generally based on the dispersion-advection equation (DAE); a partial differential equation of the parabolic type under unsaturated conditions. The DAE has been solved by various numerical methods, such as finite elements and finite differences. However, these methods require advanced knowledge in mathematics and computer programming, in addition to specific adaptations to each problem in order to avoid numerical difficulties such as stability and convergence. The numerical method of lines (NML) can solve complex problems while keeping programming to a level accessible to a large number of engineers and scientists. The purpose of this article are (1) to develop and evaluate the numerical performance of a NML model (TSOL) that solves the DAE coupled with Richard's equation under unsaturated conditions; (2) to compare results of the TSOL model with those of a recognized finite elements model (HYDRUS); and (3) to validate the TSOL model with experimental data collected during 195 days under a potato field.
The experimental setup was installed on September 1994 in a potato field located at Saint-Pierre, Île d'Orléans, near Québec City. It consisted on three pan lysimeters (A, B, and C) with a surface area of 0.48 m2 and a depth of 1.00 m, installed in an undisturbed sandy soil. On May 12, 1995, 15 g of KBr, dissolved in 60 ml of water, were applied uniformly over the surface of each pan lysimeter. The applied bromide was monitored until November 23, 1995. The monitoring period was divided into a first phase of 49 days, during which the soil was not cultivated and measures taken daily at 4 pm, and a second phase of 146 days during which measures were taken every Wednesday at 4 pm. The three pan lysimeters were sowded with potatoes on July 5 and harvested on September 5. Monitoring of the pan lysimeters included:
1. the drained water volume;
2. the water volume sampled by the pan lysimeter; and
3. the Br- concentration of all samples.
Numerical solution of the governing equations was obtained by the NML which belongs to the semi-discret methods consisting in discretising all independent variables except time, which is considered continuous for initial conditions problems. The discretisation of the DAE spatial variables was done by finite differences and resulted in a system of ordinary differential equations solved by LSODES (Livermore Solver for Ordinary Differential Equations Sparse); a solver used for systems with a sparse jacobian. The sparse nature of the jacobian results from our numerical procedure which solves simultaneously the DAE and Richard's equation.
The initial simulation time was fixed to May 12, 1995 at 4 pm and final time to November 23, 1995 at 4 pm with an hourly time step. The total depth of the pan lysimeter was simulated with a uniform internodal space of 1.0 cm. For each pan lysimeter, the initial pressure profile was measured by five pairs of tensiometers at depths of 7.5, 22.5, 45.0, 70.0 and 100.0 cm. The total mass of bromide applied was distributed equally over the first three upper nodes, and converted to concentration using the water content. At the soil surface, boundary condition for Richard's equation was taken as the hourly amount of rain fall (from planting to harvest), and as the net hourly water flux (from harvest until the end of the monitoring period). For the DAE, the boundary condition at the soil surface was of the third type. At the bottom of the pan lysimeters, a contant pressure head was assigned, which was the mean pressure measured by the deepest tensiometer; for concentration, a boundary condition of zero gradient was assigned.
For the two simulated variables (soil water bromide concentration and recovered mass of bromide), results of TSOL and HYDRUS were similar, showing an over-estimation of bromide concentration profiles but similar drained masses of bromide except lysimeter A, for which a large over-estimation was observed. This over-estimation may be explained by the presence of cracks between the soil and the plastic film surrounding the lysimeter. These cracks may allow a quick surface water flow along the sides during heavy rainfall. Because the solute was initially applied over all of the surface area of the lysimeters, a fraction of the solute might have migrated with the flowing water, and the remaining fraction by the soil matrix. The water flowing through cracks will quickly reach the bottom of the lysimeters. For all pan lysimetres and soil depths, the mean absolute error for weekly soil solution concentration profiles was 96 mg/l and the mean bias error 80 mg/l. For the mass of bromide recovered in lysimeters B and C, the weekly absolute mean error was 7 mg/week and the mean bias error 5 mg/week. For lysimeter A, the weekly absolute mean error and the mean bias error were the same, that is 39 mg/week.
For the simulation period, the numerical mass balance was negligible for the two models. However, the simulation time was longer for TSOL than for HYDRUS (18 versus 13 min). This difference is explained by the completely implicit differentiation scheme used by HYDRUS compared to the Backward Differentiation Formula used by LODES which is more complex, but frees the user from checking timestep precision. Considering the ease of programming and the resulting numerical precision, the NML has proven very effective in solving the solute transport equations in unsaturated conditions.
Étude paléolimnologique de l'histoire trophique du lac Saint-Charles, réservoir d'eau potable de la Communauté Urbaine de Québec
R. Tremblay, S. Légaré, R. Pienitz, W. F. Vincent and R. I. Hall
Cette étude utilise une approche paléolimnologique pour reconstituer l'histoire trophique du réservoir d'eau potable de la Communauté Urbaine de Québec (CUQ), le lac Saint-Charles. Ce lac manifeste présentement un manque d'oxygène près du fond à la fin de la stratification estivale et hivernale. L'étude révèle des changements dans la communauté diatomifère fossile depuis environ les 150 dernières années. L'événement ayant entraîné le plus de changements biologiques et physico-chimiques dans le bassin est la transformation hydrologique engendrée par la construction d'un barrage en 1934 qui éleva le niveau du lac d'environ 1,5 à 2 mètres. Par conséquent, il y eut des changements dans la structure des communautés de diatomées avec des effets sur le ratio espèces planctoniques / benthiques, sur la paléoproductivité et les caractéristiques physico-chimiques des sédiments suivant cette période. Les assemblages diatomifères indiquent que les conditions mésotrophes se sont maintenues pendant toute la période étudiée. L'analyse du phosphore total dans les sédiments et la reconstitution à partir des diatomées fossiles du phosphore total dans l'eau montrent une légère diminution de la concentration en phosphore avec le temps. Ces observations démontrent qu'il n'y a pas eu d'accélération du processus d'eutrophisation engendrée par les activités humaines. Par contre, l'analyse géochimique des sédiments révèle un apport plus important de métaux depuis la fin du 19e siècle, qui atteint un plateau vers la fin des années 70.
This study adopted a combined paleolimnological-limnological approach towards evaluating the trophic history of Lake Saint-Charles, the drinking water reservoir for a population of 305,000 in the Québec City (Canada) region. Our limnological measurements indicate that the lake is currently in a state of advanced mesotrophy. Water column profiles during late summer stratification (September) showed that the bottom waters were anoxic, enriched in phosphorus (up to 17 µg total P L-1) and had a pH almost 2 units lower than the surface waters. Surface phytoplankton concentrations were high at this time of year with Chlorophyll a concentrations of up to 12 µg·L-1. At the end of winter stratification, oxygen concentrations were below saturation at all depths, ranging from 72% immediately under the ice to 4% at the bottom of the water column. This tendency towards eutrophic conditions was offset, however, by a rapid flushing rate (mean hydraulic residence time=23 days). Because there are concerns that the lake has experienced accelerated nutrient enrichment due to increased human activities in its drainage basin, the objectives of our paleolimnological approach were to document the recent trophic history of this lake, to estimate the extent of recent changes in trophic status, and to identify critical periods of past anthropogenic disturbances from the fossils of siliceous algae (diatoms; class Bacillariophyceae) preserved in its sediments. Quantitative estimates of past total phosphorus (TP) concentrations in the water column of Lake Saint-Charles were obtained by applying a diatom-TP reconstruction model developed for 54 lakes located in south-eastern Ontario to fossil diatom assemblages from a 28 cm long sediment core. The timing of changes in the fossil diatom record was estimated by210 Pb dating. The study reveals changes in fossil diatom assemblage composition during the past ca. 150 years, with the most striking biological and physico-chemical changes occurring immediately after 1934. This date coincides with the construction of a dam, which raised the lake water level by 1.5-2 m. This modification was accompanied by significant shifts in diatom community structure, especially in the planktonic/benthic ratio (with increases in planktonic diatoms Cyclotella stelligera and Aulacoseira distans), and by changes in the physico-chemical characteristics of the sediments. Paleoproductivity increased at the same time, but remained more or less stable following conservation efforts between 1950 and 1970 (e.g., construction of a sewage treatment system). The organic matter content of the sediments showed an increase in the order of 20% between 1850 and 1950, after which it remained constant. Fossil diatom community structure indicates that mesotrophic conditions have prevailed during the recent history of Lake Saint-Charles, and that diatoms typical of eutrophic conditions never became established in the lake. The geochemical analysis of phosphorus in the sediments as well as the diatom-inferred quantitative reconstruction of lake water total phosphorus reveals a slight decrease in total phosphorus over time, from close to 17 µg·L- prior to 1887 to about 13 µg·L- 1 in recent times. These observations suggest that Lake Saint-Charles has not experienced significant recent changes in trophic status due to increased human activities in its drainage basin. However, our geochemical analyses show a sharp rise in metal concentrations (especially Fe, Mn, Cu, Pb and Zn), beginning in the late 19th century, reaching a plateau by the late 1970s, which may be attributed to increased atmospheric pollution since the beginning of intense human colonization in the lake's catchment and surrounding areas. This in combination with the advanced mesotrophic status of the lake indicates the ongoing need for careful management of the watershed to prevent further changes in this important urban water resource.
L. Paugam, S. Taha, J. Cabon, N. Gondrexon and G. Dorange
Cet article propose une étude de la rétention du nitrate d'ammonium par une membrane commerciale de nanofiltration (Nanomax 50). Les effets de la pression, de la concentration et de la vitesse d'écoulement tangentiel ont été étudiés avec le souci d'une meilleure compréhension du mécanisme de transport des ions nitrate et en vue d'une optimisation de la rétention. Le taux de rétention des ions nitrate augmente dans un premier temps avec la pression, atteint un maximum puis diminue. La rétention, pour des pressions élevées, peut cependant être améliorée en augmentant la vitesse d'écoulement tangentiel. La séparation résulterait d'un rapport de différentes forces : une force d'entraînement radial dans le pore (illustrée par l'effet de la pression transmembranaire), une force d'entraînement tangentiel vers le rétentat (illustrée par l'effet de la vitesse d'écoulement tangentiel) et une force de surface traduisant les interactions membrane-soluté (illustrée par l'effet de la concentration). L'équation de Spiegler et Kedem est proposée en première approche de modélisation pour une valeur limite de pression.
Many water sources deal with the problem of increasing nitrate concentrations above authorised levels for drinking water. In order to minimise this amount of pollution and to achieve high quality of water and reused water in the distribution system, membrane processes are becoming a promising technology. Indeed, they present the major advantages of a small land area requirement, low temperature operation, continuous separation, better effluent quality, little or no sludge production and a large reduction in the quantities of chemical additives. Reverse osmosis has already been used to remove most of the nitrates together with the other solutes, but the disadvantage is that this technique induces a total demineralisation of the treated water. Another possible filtration process, nanofiltration, has been investigated in this study while no extensive research has been carried out on its nitrate removal potential. Theories cannot adequately predict the influence of operating parameters on membrane performance. Consequently, new membranes and modules must be experimentally evaluated for each new application. The main objective of this study was to provide fundamental data for designing an operation of nanofiltration under various operating conditions such as transmembrane pressure, cross-flow velocity and initial feed concentration for drinking water and water reuse purposes.
The retention rate rises with an increase of the applied pressure, reaches a maximum and then decreases. Such a result is quite different from those usually mentioned in the literature where the retention increases and reaches a plateau when the pressure grows. The singular decrease of the retention rate observed in this study could be explained in terms of a concentration polarization phenomenon. However, since the volumetric flux increased linearly with the pressure and remained close to the pure water flux, it might be thought that such an assumption is not valid in the case of this work. Therefore, another hypothesis has to be provided to explain the variation of the retention with transmembrane pressure. As the size of NH4+ ion (ionic radius=0.148 nm) is lower than that of the pore of membrane (diameter=1 nm), cations can enter the pores where they are partially retained due to surface forces (electrostatic and friction forces). When the pressure increases, these forces remain constant while drag forces increase due to the flux in the pore. At low pressure (∆P < 5 bars), the surface forces are stronger than the drag forces. Therefore, the solute flux remains low while the solvent flux increases with the pressure, leading to an increase in the solute retention. Above a given pressure (≅ 5 bars), the drag forces become higher than the surface forces. Consequently, the retention rate decreases.
As can be observed in the obtained results, the retention rate decreased when the feed concentration was increased regardless of the operating pressure. This effect is mainly attributed to the cation shielding of the effective charge of the membrane. This characteristic can be explained by the fact that the electric repulsion becomes less efficient at higher concentration. It has been recognized that the effective charge density of the membrane decreases with an increase in the feed concentration of an ionic solution. Consequently, the retention rate of the co-ion due to charge effect is reduced. It follows that a greater amount of nitrate ions could permeate when feed solutions of higher concentration are applied.
The effect of cross-flow velocity on the fluxes is insignificant since the permeate flux depends only on transmembrane pressure. However, the retention performance increases with velocity. The lower the cross-flow velocity, the higher the interaction between the solute and the membrane. Therefore, at low cross-flow rate, the solute amount that enters the membrane pores is high. When the drag forces become stronger than the surface forces, as explained above, the retention sharply decreases. At high cross-flow velocity, the feed circulation transports a large solute amount and therefore, the solute amount that enters the pores is reduced and is less sensitive to operating pressure. In consequence, the sensitivity of the retention to transmembrane pressure is not so marked. It might be thought that for a very high cross-flow velocity, the retention increases and then remains constant.
It was demonstrated in this work that nanofiltration can be successfully used to remove nitrates from water. The retention was shown to depend strongly on operating parameters such as feed solution concentration, applied pressure and circulation cross-flow rate. In fact, the retention is mainly determined by the intensity of the solute / membrane interaction. This interaction comes from two main forces: a tangential one due to the feed solute flow (illustrated by the cross-flow velocity effect) and a radial one in the pores due to drag forces (illustrated by the transmembrane pressure effect). Moreover, it was observed that the valence of the associated ions is an important factor that can affect nitrate retention. It can be expected that the optimization of the separation performance will result of the best combination of all these parameters. Therefore, with a view to a future industrial application, it will be necessary to take into consideration the chemical composition of the resource and to adapt the operating conditions to the desired objectives.
K. Beven, A. Musy and C. Higy
C'est avec plaisir que nous avons récemment relu les "Tribunes Libres" de Ghislain de MARSILY (1994) et Jacques GANOULIS (1996), plus particulièrement leur discussion relative à une nouvelle typologie des modèles hydrologiques ainsi que leurs réflexions concernant l'analyse des incertitudes. Il nous apparaît toutefois, à la lecture de ces deux articles, qu'il subsiste encore quelques confusions ou interprétations alternatives concernant la modélisation hydrologique. Il est donc important et ceci malgré le fait que nous adhérons à beaucoup de points examinés par ces deux auteurs, de discuter encore quelques éléments de la modélisation hydrologique afin de lever certaines ambiguïtés.
La distinction effectuée par de MARSILY entre les modèles conditionnés par les phénomènes observables et les modèles à base physique utilisés lorsque aucun phénomène n'a été constaté, invite à la critique eu égard aux pratiques réalisées. Par ailleurs, l'argument de GANOULIS affirmant que les modèles à base physique peuvent fournir une description valable des processus si l'on utilise des coefficients empiriques à différentes échelles spatiales et temporelles, ne résiste pas non plus à une analyse détaillée des effets d'échelle. En d'autres termes, la question soulevée par ces auteurs réside dans l'impossibilité d'utiliser une modélisation à base purement physique pour des applications pratiques, en raison de la difficulté de prendre en compte et de transcrire les caractéristiques et le comportement unique de chaque unité du paysage ou chaque sous-bassin versant. Face à cette attitude, nous pouvons affirmer aujourd'hui qu'il existe d'autres voies de réflexion en ce qui concerne l'usage de modèles dits à base physique. Affirmer que tous les lieux concernés par une modélisation distribuée ont des caractéristiques uniques est une évidence géographique. Il n'en reste pas moins que les limitations de la modélisation, exprimées par de MARSILY (1994) dans le contexte des trois principes d'unité de lieu, d'action et de temps, peuvent être mieux définies en procédant à une analyse plus fine dans le contexte de l'unicité. Les unicités expliquent en partie le développement très répandu de la modélisation par rapport à la théorie et des outils propres à des applications particulières. Force est de constater que les attentes face aux prévisions quantitatives en hydrologie ont augmenté parallèlement à l'évolution de la disponibilité et la puissance des ordinateurs. Cette évolution est toutefois due essentiellement aux avancées technologiques plutôt qu'à de réels progrès scientifiques. Pourquoi? En raison principalement des caractéristiques uniques des bassins versants. Celles-ci transcendent, à notre point de vue, toutes les théories disponibles en matière de modélisation hydrologique. De surcroît, cet aspect ne change pas si on émet de meilleures hypothèses physiques ou si on réalise des prévisions pour les variables ou "phénomènes non-observables" discutés par de MARSILY.
Dans cette communication, nous tenterons une évaluation de ces questions et nous suggérerons une approche pertinente de la modélisation hydrologique pour prendre en compte le caractère unique des bassins versants.
We recently had the pleasure of re-reading the Tribunes Libres of Ghislain de MARSILY (1994) and Jacques GANOULIS (1996), especially their discussions of a new typology for hydrological models and the analysis of uncertainty. It appears, however, that some confusion and alternative interpretations of hydrological modelling still persist. It is therefore important, notwithstanding our agreement with many of the authors' points, to re-examine some aspects of hydrological modelling in order to clarify certain ambiguities.
A distinction made by de MARSILY, between models conditioned by observable phenomena and the physically-based models employed when no phenomena have been observed, invites criticism in terms of the practices to which it leads. GANOULIS' argument, that physically-based models can provide a viable description of processes if differing spatial and temporal empirical coefficients are used, does not stand up to a detailed analysis of the effects of scale. In other words, the issues addressed by these authors arise from the impossibility of using purely physically-based modelling in practical applications due to the difficulty of taking into account and transcribing the characteristics and unique behaviour of each unit of landscape or sub-catchment. To this we can now respond that there are now other lines of thought concerning what are known as physically-based models.
Where distributed modelling is concerned, that all places have unique characteristics is a geographical aphorism. The fact remains that the limitations of modelling, expressed by de MARSILY (1994) as the three principles of uniqueness of place, action and time, can be better defined by performing more detailed analysis in the context of uniqueness. Uniqueness limitations partly explain the wide-ranging developments in modelling in respect of both the theory and tools specific to particular applications. One cannot help but notice that expectations of quantitative prediction in hydrology have increased in parallel with the availability and power of computers. This evolution, however, is essentially due to technological advances rather than real scientific progress. Why? Principally, because of the unique characteristics of catchments: in our view, catchments transcend all available theories concerned with hydrological modelling. Moreover, this does not change if better physical hypotheses are proposed, nor if predictions are made for the variables or "non-observable phenomena" discussed by de MARSILY.
In this paper, we address these questions and suggest a relevant approach to hydrological modelling for taking into account the unique character of catchments.
Tous droits réservés © Revue des sciences de l'eau, 2001