Environ. Sci. Technol. 2007, 41, 2523-2528
Interrelated Effects of Aeration and Mixed Liquor Fractions on Membrane Fouling for Submerged Membrane Bioreactor Processes in Wastewater Treatment FENGSHEN FAN AND HONGDE ZHOU* School of Engineering, University of Guelph, Guelph, Ontario N1G 2W1, Canada
The interactions of mixed liquor fractions and their impacts on membrane fouling were examined at different sparging aeration intensities for submerged hollow-fiber membrane bioreactors (MBR) in wastewater treatment. The mixed liquor samples were fractioned by size into MLSS, colloids quantified by colloidal TOC, and dissolved solutes. The experimental results showed that their significance in membrane fouling was strongly related to aeration intensity. In the absence of sparging aeration, both MLSS and colloids contributed to membrane fouling which was further enhanced by their interactions. For the tested membrane module operated at the vigorous aeration intensity typically employed in practice, however, the deposition of colloids was identified as the most important mechanism controlling membrane fouling rates. In contrast, much fewer effects were exerted by MLSS: the overall fouling rates were increased initially, and then reduced with increasing concentration of MLSS. Thus, the aeration-induced turbulence should be considered for properly assessing the mixed liquor fouling potential for wastewater MBR processes. Finally, little difference in fouling rates was observed with the use of cyclic aeration mode as compared to continuous aeration mode.
1. Introduction Over the past decade, membrane bioreactor (MBR) processes have experienced unprecedented growth in wastewater treatment owing to excellent solid/liquid separation independent of sludge settleability (1, 2). This can be partially attributed to the introduction of submerged MBR configuration, which results in the elimination of costly, energyintensive flow recirculation essential to external MBR systems (3). The submerged MBR is constructed by placing the membrane modules directly into the bioreactor while injecting coarse bubble air outside the membrane modules. This sparging air produces effective turbulence and fiber movement, scourring the particles and other deposited materials away from membrane surfaces. As a result, submerged MBR processes offer many distinct advantages including excellent effluent quality, smaller footprint, and low sludge production. However, membrane fouling remains a major obstacle to their wider application in wastewater treatment. It reduces membrane permeability, thereby, increasing the membrane * Corresponding author phone: +1-519-824-4120, ext. 56990; fax: +1-519-836-0227; e-mail:
[email protected]. 10.1021/es062035q CCC: $37.00 Published on Web 02/21/2007
2007 American Chemical Society
cleaning frequency, reducing the plant treatment capacity and shortening the membrane module lifespan (1, 2, 4). Numerous studies have been directed to elucidate the fouling behaviors and the effects of key operating conditions and mixed liquor characteristics on membrane fouling. For the submerged MBRs operated at a constant permeate flux mode, the transmembrane pressure increases gradually over time. After the membrane module is fouled severely, the transmembrane pressure could increase almost exponentially, highlighting the importance of short-term physiochemical membrane fouling. Consequently, a critical flux concept was proposed initially below which no fouling would occur (5). An increase in air flow rate at the membrane surface increases the critical flux because of the increased back transport of deposited materials from the membrane surface by turbulent shear at a higher air flow rate (6, 7). A critical air flow rate could exist beyond which any further increase in aeration has little effect on fouling suppression (8, 9). The rate and extent of fouling are also affected by the properties of mixed liquor suspension that the membrane modules contact. Fane et al. (10) examined the different roles of mixed liquor materials by fractioning into two components: mixed liquor suspended solids (MLSS) and dissolved solids after filtration with Whatman 41 filter paper. They found that the fouling resistance caused by MLSS started to become dominant when the MLSS concentration exceeded about 3 g/L. Other research has divided the mixed liquor suspension into three fractions: MLSS, colloids, and solutes. Wisniewski and Grasmick (11) collected the colloids and solutes by settling the mixed liquor samples by gravity for 2 h. They further separated the solutes from the colloids by passing the settled supernatant through a 0.05 µm membrane filter. The results showed that the solutes, colloids, and MLSS accounted for 52%, 25%, and 23% of the total fouling resistance, respectively. Similar tests were also conducted by Defrance et al. (12), who reported that the MLSS were responsible for 65% of the total fouling resistance. More recently, Bouhabila et al. (13) centrifuged mixed liquor samples to collect the supernatant for colloids and solutes mixture. Afterward, the solutes were further separated by flocculation followed by 4500 rpm centrifugation for 10 min. Their experimental results showed that membrane fouling was mainly caused by colloids, because the MLSS, colloids, and solutes accounted for 24%, 50%, and 26% of fouling resistance, respectively. The discrepancy among these studies may result from different fractioning methods, membrane systems, and their operation modes, in addition to different mixed liquor samples used. Two key limitations in comparing these studies are the use of different aeration intensities and the lack of quantitative measurement of the concentration of colloids. Both limitations are addressed in this study. The objective of this study was to determine the significance of major mixed liquor fractions and their interactions in short-term physiochemical membrane fouling under different aeration intensity conditions. The presumption is that the deposition of heterogeneous mixed liquor fractions on the membrane surface could be interrelated, the extent of which depends on hydrodynamic conditions (14, 15). The concentration of colloids in mixed liquor samples was represented by measuring the TOC for the particles ranging from 0.04 to 1.5 µm. The constant permeate flux filtration tests were conducted using a bench-scale submerged hollow fiber membrane module operated at different aeration intensities. VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2523
TABLE 1. Specifications of the Laboratory-Scale MBR Apparatus parameter membrane tank diameter × length membrane length internal diameter nominal pore size membrane looseness no. of membrane hollow fibers total membrane area
FIGURE 1. Schematic of the laboratory-scale submerged membrane filtration apparatus.
2. Materials and Methods 2.1 Mixed Liquor Sample Preparation. Mixed liquor samples were collected from a large submerged MBR pilot plant (ZeeWeed 500, GE/Zenon Membrane Solutions, Canada) installed at the City of Guelph Wastewater Treatment Plant. It was fed with the screened raw municipal wastewater and operated at a hydraulic retention time (HRT) of 6 h and a sludge retention time (SRT) of 20 days. A more detailed description of this pilot plant can be found in Fan et al. (16). The collected raw mixed liquor samples were separated by size into three fractions: MLSS, colloids, and dissolved solutes. Analytically, MLSS fractions were intercepted by filtering the resultant supernatant with a 1.5 µm filtration paper (934/AH, Whatman, USA) after the mixed liquor samples were pretreated by centrifugation at 2000g for 10 min to overcome the clogging difficulty. The solutes were then obtained by collecting the permeate samples directly from the pilot hollow fiber membrane modules with a nominal pore size of 0.04 µm. The particles ranging from 0.04 µm to 1.5 µm were designated colloids, similarly to the commonly accepted suggestion (17). Their concentrations were quantified by measuring the difference in TOC between the filtrate passing through a 1.5 µm pore size filter and the permeate collected directly from the membrane pilot plant. Note that such colloidal fraction separated operationally might also contain large-molecular-weight microbial byproducts and other extracellular polymeric substances. Additional mixed liquor samples were then prepared by combining the mixed liquor samples with the permeate or filtrate obtained above in a series of preset volumetric ratios to obtain the concentrations of MLSS and cTOC from about 8 to 20 g/L and 2 to 20 mg/L, respectively. Similar concentration ranges were found for the mixed liquors taken from the MBR systems operated at different SRT and HRT conditions (16). An advantage of this approach is allowing to obtain a series of test samples with desired different proportions of key mixed liquor fractions without changing the surface properties of particles and chemical properties of solutes. 2.2 Laboratory-Scale Membrane Filtration Tests. Figure 1 shows a schematic of the laboratory-scale MBR apparatus used for this study. It consists of a Φ0.088 m × 0.75 m cylindrical vessel immersed with four membrane hollow fibers cut from ZeeWeed 500 modules with a length of 0.4 2524
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
unit
values
m
Φ0.088 × 0.75
m mm µm %
0.4 0.9 0.04 2 4 (installed equally around a diameter of 50 mm) 0.01
m2
m (GE/Zenon Membrane Solutions, Canada) to obtain the homogeneous turbulence induced by aeration. In addition, 2% membrane fiber looseness was selected in order to be consistent with the typical range of 1-5% used in the design of full-scale membrane modules (18). A peristaltic pump was used to withdraw the permeate by vacuum from the lumen of the membrane hollow fibers at a constant permeate flux. The sparging air was controlled by a rotometer (150-mm PTFE with valve, Cole-Parmer, USA) and then bubbled through a coarse air stone (A-965 Elite Sphere Type, Hagen, Canada) installed at the bottom of membrane hollow fibers for complete mixing and membrane fouling control. Table 1 summarizes the main specifications of this laboratoryscale MBR apparatus. Prior to their use, new membrane hollow fibers were first soaked in 200 mg/L NaOCl at room temperature for 24 h and, after washing with Milli-Q water, then soaked in 2000 mg/L citric acid for another 24 h to remove glycerol on the membrane surface. To stabilize the initial membrane fouling, these pre-washed new membrane fibers were conditioned to filter the mixed liquor sample for 24 h at a permeate flux of 10 L/m2/h and an air flow rate of 4 L/min. Afterward, they were gently wiped with a sponge and rinsed with Milli-Q water thoroughly to remove the accumulated cake. For their subsequent use, the membrane fibers were gently scraped with a sponge, and soaked and flushed in Milli-Q water for 10 min for cake removal. The difference in clean water membrane resistance (Rm) between these cleaned membranes was less than 5%. This membrane preparation procedure was deemed necessary, as our previous experiments showed that there was an initial membrane fouling caused by organic adsorption, pore clogging, etc. in the first 24 h, followed by an almost stable, much slower fouling rate (16). The membrane filtration tests were started by measuring the clean water filtration resistance with Milli-Q water at a permeate flux of 20 L/m2/h. The mixed liquor sample warmed to 20 °C was then poured into the laboratory-scale MBR unit. The aeration intensities varied from 0 to 4 L/min, resulting in the mean velocity gradients estimated for clean water as high as 337 s-1 in order to represent the operating conditions commonly encountered in practice (16). A permeate flux of 35 L/m2/h was selected to provide the reliable measurements for the fouling rates, while falling within the range from about 20 to 50 L/m2/h used in practice to accommodate the annual average, maximum day, and peak hour fluxes (18). During the filtration, the transmembrane pressure was recorded continuously by using a pressure sensor connected to a datalogger. The permeate flux was measured using a digital balance with a precision of 0.01 g (BB2442, Mettler, Switzerland). The withdrawn permeate was recirculated to the membrane tank to maintain a constant composition of mixed liquor sample. To be consistent with submerged MBR operations being typically used in practice, each filtration cycle lasted for 10 min prior to 0.5 min back pulse with permeate (18).
2.3 Analytical Methods. Mixed liquor suspended solids were measured according to Standard Methods (19). Total organic carbon was analyzed using a TOC analyzer (TOCVCSH, Shimadzu, Japan). The particle size distribution was measured using a laser diffraction particle size analyzer (Mastersizer 2000, Malvern Instruments Ltd., UK). 2.4 Data Analysis. The transmembrane pressures collected over filtration time were used to calculate the total filtration resistance according to Darcy’s law (15):
∆p η(Rm + Rc)
J)
(1)
The average fouling rate was then calculated as the difference between initial and final cake resistances divided by the duration of filtration cycle below:
F)
Rc,t2 - Rc,t1 t2 - t1
(2)
To accommodate the effects of pump ramp-up, the initial filtration resistance was taken after 1 min of each filtration cycle. The percentage of fouling rate caused by each mixed liquor fraction was calculated as:
Fsolute × 100% Fmixed liquor
(3)
Fcolloid+solute - Fsolute × 100% Fmixed liquor
(4)
Fmixed liquor - Fcolloid+solute × 100% Fmixed liquor
(5)
Fsolute,% )
Fcolloids,% )
FMLSS,% )
FIGURE 2. Comparison of fouling behaviors between (a) no aeration and (b) 4 L/min sparging aeration (20 °C and permeate flux ) 35 L/m2/h; -9- MLSS 7.0 mg/L and cTOC 21.08 mg/L; -2- MLSS 21.0 mg/L and cTOC 12.38 mg/L).
It should be pointed out that Fcolloid calculated above actually represents the combined contributions of colloids and their interactions with solutes as it is difficult, if not impossible, to completely separate them analytically. Similarly, the calculated FMLSS represents the combined contributions of MLSS and their interactions with colloids and solutes. Note that the negative contribution calculated for a fraction suggests that its presence interfered with the deposition of other fractions on the membrane surface, thus reducing overall membrane fouling.
3. Results and Discussion Figure 2 shows typical transmembrane pressure increase as a function of filtration time within each filtration cycle. Two different mixed liquor samples were tested: mixed liquor A had a lower MLSS concentration (7.0 g/L) and a higher cTOC (21.08 mg/L), while mixed liquor B had a higher MLSS concentration (21.0 g/L) and a lower cTOC (12.38 mg/L). In the absence of sparging aeration, mixed liquor A exhibited a slower increase in transmembrane pressure as compared to mixed liquor B. At an air flow rate of 4 L/min, an opposite trend was observed, implying that the deposition behaviors of mixed liquor fractions was strongly related to aeration intensity. 3.1 No Aeration. Figure 3 shows the contributions of different mixed liquor fractions to overall membrane fouling rates in the absence of sparging aeration. Over the wide range of cTOC, the membrane fouling was largely dependent on the concentration of MLSS which accounted for 62-87% of the total membrane fouling rates, while colloids and solutes contributed less than 36% and 13%, respectively. As expected, the relative contribution from MLSS to the overall membrane fouling became more significant as MLSS concentration increased and/or cTOC decreased.
FIGURE 3. Contributions of different mixed liquor fractions to membrane fouling in absence of sparging aeration: (a) MLSS ) 7.0 g/L; (b) MLSS ) 14.3 g/L; (c) vertical bars, MLSS; horizontal bars, colloids; 0 solutes). Figure 4 plots the contour of membrane fouling rates by pooling all the experimental results obtained at different concentrations of MLSS and colloids. In general, the total fouling rates were increased with increasing concentrations of both MLSS and cTOC because both particles were deposited onto membrane surface. The effect of MLSS on VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2525
FIGURE 4. Contours of the overall fouling rates in absence of sparging aeration.
FIGURE 5. Contributions of different mixed liquor fractions to membrane fouling at a sparging air flow rate ) 4 L/min: (a) MLSS ) 7.0 g/L; (b) MLSS ) 14.3 g/L; (c) MLSS ) 21.0 g/L (20 °C, permeate flux ) 35 L/m2/h; vertical bars, MLSS; horizontal bars, colloids; 0 solutes). fouling rate was enhanced with the presence of colloids. At an MLSS concentration of 20 g/L, e.g., the fouling rates were increased by approximately 2.5 times when cTOC was increased from 4 to 11 mg/L. A possible explanation is that the colloids enter the cake layer to block the cake pores and/ or be absorbed on the cake pore walls, thereby reducing the effective cake porosity. 3.2 Sparging Aeration. Figure 5 shows typical percentages of total fouling rates caused by each mixed liquor fraction after introducing the sparging aeration. Unlike the absence of sparging aeration, the vigorous aeration condition made the mixed liquor colloids the predominant fraction responsible for membrane fouling. It is interesting to note that for the mixed liquors with higher MLSS concentrations (14.3 and 21.0 g/L), the fouling rates obtained from the filtrate 2526
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
FIGURE 6. Typical contours of the overall fouling rates with sparging aeration (air flow rate ) 4 L/min). containing only the colloids and solutes were even higher than those from the raw mixed liquor samples, resulting in the negative FMLSS values. In other words, the presence of MLSS could cause a slight reduction in membrane fouling when its concentration became higher than a critical value. Figure 6 shows a typical contour plot in terms of the overall fouling rates under the sparging aeration conditions. In general, the overall fouling rates varied greatly, ranging from 1.5 × 1010 to 3. 5 × 1010 m/min. As the concentration of colloids increased, the overall fouling rates increased. In contrast, much less effect on the fouling rates was posed by the variation in MLSS concentration. Moreover, the effects of MLSS on the overall fouling rates became much more complicated. The overall fouling rates increased as the concentration of MLSS was increased initially. However, further increase in MLSS concentration eventually caused the reduction in membrane fouling rates. To provide a better understanding of the phenomena observed above, the mixed liquor samples were analyzed by measuring the particle size distribution. As expected, these samples contained a wide range of particle sizes (data not shown). The diameters of most particles fell with the range from 20 to 50 µm with the volume weighted mean size being around 42 µm. It has also been reported that the deposited MLSS could act as a secondary membrane, thereby preventing the pore blocking and adsorption of colloids and solutes onto the membrane surface (20). Moreover, it has been suggested that the shear-induced hydrodynamic diffusivity is proportional to the square of particle diameter and the shear rate (21). With the vigorous turbulence caused by sparging aeration, the larger particles become more susceptible to the back-transport. In fact, the shear-induced diffusion could become predominant in particle backtransport for the particles with diameters from 1 to 50 µm (16). Consequently, the smaller colloids could be selectively deposited on the membrane surface over the larger MLSS. Finally, the MLSS could provide a lateral scouring action that scrapes some of the colloidal cake layer off the membrane, resulting in a thinner cake with the reduced membrane filtration resistance. Figure 7 shows the typical variations of the overall fouling rates and the fouling contributions of different mixed liquor fractions as a function of aeration intensity, respectively. Note that three mixed liquor samples were compared, each of which had the same concentration of cTOC, but differed greatly in the concentrations of MLSS. At the low air flow rates, the fouling rates were much greater for the mixed liquor samples with the higher concentration of MLSS than those with the lower concentration of MLSS due to the significant deposition of MLSS. As well, an initial increase in aeration intensity resulted in the drastic drop in the overall fouling rates. When the air flow rate exceeded 2 L/min, however, the fouling rates gradually leveled off especially for the mixed
FIGURE 9. Comparison of the membrane fouling rates between different air sparging modes.
solutes on membrane surface might result mainly from adsorption. It should be pointed out that this does not exclude their potential affecting the long-term membrane fouling because the adsorbed solutes could alter the membrane surface property, thereby affecting the irreversible attachment of other particles and/or biofilm growth.
FIGURE 7. Effects of aeration intensity on membrane fouling (20 °C, permeate flux ) 35 L/m2/h).
3.3 Air Sparging Modes. Because the sparging aeration represents a substantial portion of total operating costs for the submerged MBRs being used in wastewater treatment (4), further tests were conducted to see whether the use of cyclic aeration would affect the fouling rates at different air flow rates. As shown in Figure 9, no substantial difference in fouling rates was found between continuous and cyclic air sparging modes. This suggests that any additional deposition of particles during aeration-off period will be effectively removed by subsequent aeration. Considering the fact that this cyclic air sparging mode only requires 50% of total air volume as compared with a continuous air sparging mode, its use can lead to a substantial cost saving in practice without compromising the fouling control.
Acknowledgments
FIGURE 8. Effects of solutes on average fouling rates. liquor samples with the higher concentration of MLSS. The trend is consistent with previous observations (8, 9) that there may exist a critical air flow rate beyond which any further increase in aeration would have little improvement in fouling suppression. At the highest air flow rate of 4 L/min, the higher MLSS mixed liquor samples had a slightly lower fouling rate than those with a lower MLSS concentration. The phenomenon can again be explained by the fact that MLSS caused possible lateral cake scouring action and/or the formation of secondary membrane. Under this aeration condition, nevertheless, the colloids still played a predominant role in membrane fouling. Consistently, Figure 7(b) showed that the percentage of fouling rate caused by the colloids was increased with the increasing the air flow rate. Unlike the colloids and MLSS, the solutes contributed much less to the membrane fouling. This was further confirmed by conducting the additional membrane filtration tests with the permeates after removing the colloids and MLSS. As shown in Figure 8, the average fouling rates were much lower than those of the raw mixed liquors presented above. Furthermore, they were little affected by the air flow rate because the mechanism underlying the deposition of
We are grateful to Dr. Hadi Husain and Mr. Henry Behmann for their thoughtful advice during planning the experiments, Mr. Edwin Castilla from GE/Zenon Membrane Solutions for operating the MBR pilot plants, and the staff from the City of Guelph Wastewater Treatment Plant for assisting in mixed liquor sampling. We would also acknowledge the financial support provided by Ontario Centres of Excellence-Earth and Environmental Technologies and Natural Sciences and Engineering Research Council of Canada.
Nomenclature cTOC
colloidal total organic carbon (mg/L)
F
average membrane fouling rate (min/m)
J
permeate flux (m3/m2/h or L/m2/h)
HRT
hydraulic retention time (h)
MLSS
mixed liquor suspended solids (g/L)
∆p
transmembrane pressure (Pa)
Rm
membrane resistance (1/m)
Rc
cake resistance (1/m)
SRT
sludge retention time (day)
η
viscosity (N‚s/m2)
VOL. 41, NO. 7, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2527
Literature Cited (1) Wintgens, T.; Rosen, J.; Melin, T.; Brepols, C.; Drensla, K.; Engelhardt, N. Modelling of a membrane bioreactor system for municipal wastewater treatment. J. Membr. Sci. 2003, 216 (12), 55-65. (2) Zhou, H.; Smith, D. W. Advanced treatment technologies in water and wastewater treatment. Can. J. Civil Eng. 2001, 28 (S1), 49-66. (3) Yamamoto, K.; Hiasa, M.; Mahmood, T.; Matsuo, T. Direct solidliquid separation using hollow fibre membrane in an activated sludge aeration tank. Water Sci. Technol. 1989, 21 (1), 4354. (4) Gauder, M.; Jefferson, B.; Judd, S. Aerobic MBRs for domestic wastewater treatment: a review with cost consideration. Sep. Purif. Technol. 2000, 18, 119-130. (5) Field, R. W.; Wu, D.; Howell, J. A.; Gupta, B. B. Critical flux concept for microfiltration fouling. J. Membr. Sci. 1995, 100 (3), 259-272. (6) Chang, S.; Fane, A. G. Filtration of biomass with laboratoryscale submerged hollow fibre modules - effect of operating conditions and module configuration. J. Chem. Technol. Biotechnol. 2002, 77, 1030-1038. (7) Le-Clech, P.; Jefferson, B.; Judd, S. J. Impact of aeration, solids concentration and membrane characteristics on the hydraulic performance of a membrane bioreactor. J. Membr. Sci. 2003, 218 (1-2), 117-129. (8) Fane, A. G.; Chang, S.; Chardon, E. Submerged hollow fibre membrane module - design options and operational consideration. Desalination 2002, 146, 231-236. (9) Ueda, T.; Hata, K.; Kikuoka, Y.; Seino, O. Effect of aeration on suction pressure in a submerged membrane bioreactor. Water Res. 1997, 31, 489-497. (10) Fane, A. G.; Fell, C. J. D.; Nor, M. T. Ultrafiltration/activated sludge system - development of a predicted model. In Ultrafiltration Membranes and Applications; Cooper, A. R., Ed.; Plenum Press: New York, 1980; pp 631-648.
2528
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 7, 2007
(11) Wisniewski, C.; Grasmick, A. Floc size distribution in a membrane bioreactor and consequences for membrane fouling. Colloids Surf., A 1998, 138, 403-411. (12) Defrance, L.; Jaffrin, M. Y.; Gupta, B.; Paullier, P.; Geaugey, V. Contribution of various constituents of activated sludge to membrane bioreactor fouling. Bioresour. Technol. 2000, 73, 105-112. (13) Bouhabila, E. H.; Aim, R. B.; Buisson, H. Fouling characterization in membrane bioreactors. Sep. Purif. Technol. 2001, 22-23, 123-132. (14) Kimura, K.; Yamato, N.; Yamamura, H.; Watanabe, Y. Membrane fouling in pilot-scale membrane bioreactors (MBRs) treating municipal wastewater. Environ. Sci Technol. 2005, 39, 62936299. (15) Belfort, G.; Davis, R. H.; Zydney, A. L. The behaviour of suspensions and macromolecular solutions in crossflow microfiltration. J. Membr. Sci. 1994, 96, 1-58. (16) Fan, F.; Zhou, H.; Husain, H. Identification of wastewater sludge characteristics to predict critical flux for membrane bioreactor processes. Water Res. 2006, 40 (2), 205-212. (17) Metcalf & Eddy, Inc. Wastewater Engineering: Treatment and Reuse, 4th ed.; McGraw Hill: New York, 2003. (18) Husain, H. Personal communication, GE/Zenon Membrane Solutions, Oakville, Ontario, Canada, 2005. (19) APHA; AWWA; WEF. Standard Methods for the Examination of Water and Wastewater, 19th ed.; APHA: Washington, DC, 1995. (20) Lee, J.; Ahn, W. Y.; Lee, C. H. Comparison of the filtration characteristics between attached and suspended growth microorganisms in submerged membrane bioreactor. Water Res. 2001, 35 (10), 2435-2445. (21) Eckstein, E. C.; Bailey, P. G.; Shapiro, A. H. Self-diffusion of particles in shear flow of a suspension. J. Fluid Mech. 1977, 79 (1), 149-164.
Received for review August 24, 2006. Revised manuscript received January 23, 2007. Accepted January 24, 2007. ES062035Q
