Anda di halaman 1dari 10

Journal of Membrane Science 177 (2000) 7382

Ultraltration membrane bioreactor for urban wastewater reclamation


C.-H. Xing a , E. Tardieu b , Y. Qian a , X.-H. Wen a,
a

State Key Laboratory of Environmental Simulation and Pollution Control, Department of Environmental Science and Engineering, Tsinghua University, Beijing 100084, PR China b Direction Dpartementale de lAgriculture et de la Fort, Centre Administratif Cond, 18013 Bourges Cedex, France Received 26 November 1999; received in revised form 1 May 2000; accepted 1 May 2000

Abstract A 162-day pilot-scale operation for reclamation of urban wastewater was studied by using an ultraltration membrane bioreactor (UMBR). Performance of the UMBR was investigated with a sludge retention time (SRT) of 5, 15, and 30 days, a hydraulic retention time (HRT) of 5 h, and membrane ux between 75 and 150 l m2 h1 , respectively. It was observed that the highest sludge concentration in the reactor viz. a suspended solids (SSs) concentration of 23.1 g l1 and a volatile suspended solids (VSSs) concentration of 13.5 g l1 , respectively, could be reached. The ratio of sludge VSS to sludge chemical oxygen demand (COD) was 1.428 in the study, which approximated to the theoretical value of 1.415. Mass loading rates of the UMBR were close to those of conventional activated sludge processes (CASPs), while the volumetric loading rates were two to ve times those of CASP. Averaged 97% of COD, 96.2% of ammonia nitrogen (NH3 -N), and 100% of SSs were removed. It was found that the bioreactor was responsible for 85% of COD removal, while 12% was due to separation of the membrane module. The reclaimed water could be reused directly for municipal purposes or indirectly for industrial uses after additional treatment. 2000 Elsevier Science B.V. All rights reserved.
Keywords: Activated sludge; Bioreactor; Ceramic membrane; Reclamation; Ultraltration; Urban wastewater

1. Introduction Urban wastewater is usually treated by conventional activated sludge processes (CASPs), which involve the natural biodegradation of pollutants by heterotrophic bacteria (i.e. activated sludge) in aerated bioreactors. Activated sludge could be separated by gravitational setting [1]. The treatment efciency is usually limited by the difculties in separating suspended solids (SSs). The optimal sludge concen-

Corresponding author. Fax: +86-10-6277-1742. E-mail address: xhwen@tsinghua.edu.cn (X.-H. Wen)

tration is generally up to 5 g l1 , which imposes large size of aerated bioreactor [2]. Membrane bioreactor (MBR) is an improvement of the 100-year old CASP, where the traditional secondary clarier is replaced by a membrane unit for the separation of treated water from the mixed solution in the bioreactor [3]. Originated from the use of membrane separation, MBR technology has various advantages [4]. The absolute retention of all micro-organisms insures an increase in sludge concentration and complete disinfection of treated water. It allows a complete separation of the hydraulic retention time (HRT) and sludge retention time (SRT). As a result, a high sludge concentration can be main-

0376-7388/00/$ see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 6 - 7 3 8 8 ( 0 0 ) 0 0 4 5 2 - X

74

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382 Table 1 Characteristics of urban wastewater Items COD (mg l1 ) SS (mg l1 ) NH3 -N (mg l1 ) Coliform (number l1 ) Turbidity (NTU) pH value Temperature ( C) Typical 200800 100600 1030 105 106 5070 7.58.5 1525 Range 502234 801327 1040 105 106 5080 7.58.5 1525

tained in the bioreactor and high-strength wastewater can be treated effectively [5,6]. Due to the absence of secondary clarier and the presence of a high sludge concentration, the overall size of the treatment plant can be reduced signicantly [4]. Furthermore, the contact time between activated sludge and organic pollutants can be enhanced, which facilitates an effective removal of low biodegradable pollutants. A highly treated water (by the MBR) is free from bacteria and has a potential in municipal and industrial reuse [8]. About 200 MBRs are currently in operation for various wastewaters, and 90% of them are employed in municipal treatment [3]. In the application of MBR that is based on polymeric materials, the leading country is Japan, where most MBR systems are used for water recycling in buildings [9,10]. As these organic membranes are normally sensitive to caustic cleaning reagents, the difculty met with in cleaning is often encountered especially when the membrane module is seriously fouled during industrial operation [7]. In order to overcome this difculty, an MBR system equipped with ceramic membranes was rst developed in France [11]. It makes the cleaning of membrane easy and convenient in situ because these inorganic membranes possess a high degree of resistance to chemical abrasion and biological degradation. The membrane has a great chemical stability in a wide range of pH and temperature [12].

The objective of this study was to investigate the long-term performance of a pilot-scale cross-ow ultraltration membrane bioreactor (UMBR) for urban wastewater reclamation. The impact of operational parameters, such as SRT, HRT, and membrane ux, on efuent quality was evaluated. Contributions of the bioreactor and membrane module to the removal efciency were examined. Moreover, the reuse potential of treated water was discussed by comparing with current water quality standards. 2. Experimental The urban wastewater used in the study was pumped from a local sewage station. As shown in Table 1, the wastewater can represent the medium-strength urban wastewater seen in most cities around the world. A schematic of the UMBR is shown in Fig. 1. The inu-

Fig. 1. Schematic of the pilot UMBR.

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

75

ent was taken from the feed tank to the bioreactor by peristaltic pump 1. The bioreactor that was lled with activated sludge had a working volume of 30 l. An aerator was employed to maintain an aerobic environment for the normal growth of activated sludge. To keep an optimal temperature, a heat exchanger was installed in the bioreactor. A stirrer was used to ensure complete mixing of the inuent and the activated sludge. A level controller together with pump 1 was used in order to maintain a constant working volume. The bioreactor, centrifugal pump 2, and membrane module constituted a loop, where the activated sludge was circulated at a high speed. In addition, centrifugal pump 3 was installed to enhance circulation of the activated sludge. To maintain a stable HRT, the owrate of efuent was automatically controlled by an air-driven valve linked up to an electromagnetic owmeter. The variation of SRT (from 5 to 15 and then 30 days) could be brought about by altering the owrate of excessive sludge discharge. To prevent membrane fouling and simultaneously reduce energy consumption, cross-ow velocity inside membrane channels was set at 4 m s1 . To explore the difference of efuent quality under different uxes, the membrane ux was varied from 75 to 150 l m2 h1 by altering the owrate of efuent or the number of membrane modules in operation. In the cleaning mode, the cleaning tank that was lled with cleaning reagents was substituted for the bioreactor to form a circulation loop with the membrane module. The ultraltration membrane employed in the study was ceramic tubular KerasepTM X3 type (Tech-Sep,

France). The membrane skin layer and support materials were made of zirconia (ZrO2 ) and alumina (-Al2 O3 ). Each membrane had seven channels that had a diameter of 4.5 mm. The membrane was 40 cm in length, while the surface area of each membrane was 0.04 m2 . The membrane pore size was about 0.02 m and the molecular weight cut-off was about 300,000 Da. Initial permeability of the new membrane was about 45 l m2 h1 kPa1 on the basis of tap water test at 25 C. The UMBR system was monitored by daily measurement of redox potential by Monec 8935 and temperature. Turbidity was measured by a turbidity meter (model 965-10, Orbeco analytical System Inc., USA). Samples for supernatant chemical oxygen demand (COD) of activated sludge were taken after centrifugation at 4500 rpm for 15 min. COD, SSs, volatile suspended solids (VSSs), and ammonia nitrogen (NH3 -N), and other items were measured as per the Standard Methods for Examination of Water and Wastewater [13].

3. Results and discussion 3.1. Removal of COD, NH3 -N, and SS The UMBR was operated for 162 days with the inuent and efuent COD values illustrated in Fig. 2. The inuent COD was uctuated from 200 to 800 mg l1 ; however, the efuent COD was maintained

Fig. 2. Evolution of COD concentration as a function of time.

76

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

Table 2 Operation parameters in the UMBR Items Duration (day) HRT (h) SRT (day) Flux (l m2 h1 ) 5 5 5 5 5 5 5 15 30 30 150a 75b 75a 75a 150b

Run 1 12 11 Run 2 78 Run 3 22 39


a b

One tubular membrane was in operation. Two tubular membranes were in operation.

at a low level. Taking the 162 days of experiment as a whole, 94% of the efuent COD data were lower than 12 mg l1 , while the averaged COD of efuent was only 9.4 mg l1 . An averaged COD removal efciency of 97% was achieved. It can be concluded that the removal of organic pollutants by the UMBR was very high in terms of COD and a good-quality efuent can be achieved during the long-term operation. Note that the inuent COD in Run 3 was once as high as 2234 mg l1 , yet the efuent COD remained lower than 10 mg l1 . This indicated that the UMBR had a potential in treating high-strength urban wastewater. As listed in Table 2, the SRT was sequentially increased from 5 to 15 and then 30 days and the membrane ux was increased from 75 (two tubular membranes were in operation) to 150 l m2 h1 (one tubular membrane was in operation). However, there

were no remarkable changes in efuent COD during Runs 13 (see Fig. 2). This may be attributed to the forced separation of ultraltration membrane because its pore sizes were xed and its mechanical interception to macromolecules from sludge-mixed liquor was no longer affected by the variation of SRT and membrane ux. Thus, the efuent COD was not inuenced by change in the SRT and membrane ux. The highest value of efuent COD was observed at 30 mg l1 on Day 24 because the pilot had been halted for 10 h on Day 23. No nutrient was available for the maintenance and growth of aerobic bacteria, while the aerator still worked normally. As a result, some of the bacteria died of starvation and partial self-hydrolysis of activated sludge occurred in the bioreactor. The sludge activity was therefore dramatically diminished. Though the inuent COD on the following day was only 133 mg l1 , the efuent COD reached the highest record in the study. However, the efuent COD went down gradually to the normal level of about 10 mg l1 soon after the UMBR system was restored. The concentration of NH3 -N was measured weekly during the pilot-scale experiment. Fig. 3 illustrates the evolution of NH3 -N concentration with time. When the inuent NH3 -N concentration was changed from 10 to 40 mg l1 , the efuent concentration was reduced to the low level of 0.21.3 mg l1 . The averaged removal of NH3 -N was 96.2%, indicating that the NH3 -N in the inuent had been deeply nitried in the UMBR system. This was mainly due to two reasons.

Fig. 3. Evolution of NH3 -N concentration as a function of time.

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

77

First, as the nitrifying population was completely conned within the bioreactor, these autotrophic nitriers having long generation times were forced to proliferate speedily without any loss. However, they were unavoidably washed out in the CASP when the SRT was kept too low [14]. Secondly, as the sludge production was low in MBR processes, nitriers in the bioreactor faced less menace from those heterotrophic bacteria which were better competitors for the NH3 -N [15,16]. Consequently, a high nitrication could be achieved in the UMBR system even at an HRT of 5 h and an SRT of 5 days during Run 1 of this study. Similar to the COD removal discussed above, the NH3 -N removal in Runs 13 was stable when the SRT was varied from 5 to 15 and then 30 days (see Fig. 3) and the membrane ux was changed from 75 (Days 1324 and 102123) to 150 l m2 h1 (Days 112 and 124162) (see Table 2), respectively. However, there was no observable difference in the NH3 -N concentration of UMBR efuent. In other words, the efciency of the removal of NH3 -N was irrelevant to the alteration in the SRT and ux. As illustrated in Fig. 4, there was no SS detected in the UMBR efuent during the 162-day experiment even though a great uctuation was observed in the inuent SS concentration. The efciency of the removal of SS remained as high as 100%, which demonstrated the better separation effect of the ultraltration membrane module than that of the secondary clarier in CASP [4,10].

3.2. Evolution of sludge concentration The microbial population in the bioreactor was practically measured by means of easily determined parameters such as the SSs and the VSSs of activated sludge [17]. The information on variation of sludge concentration with time was necessary to understand the actual performance of the bioreactor. As depicted in Fig. 5, the highest sludge concentration in terms of SS and VSS had reached values of 23.1 and 13.5 g l1 on Day 155 in Run 3. The VSS to SS ratio was vacillated around 0.6 during Runs 13 in spite of the fact that the absolute concentration of sludge in the UMBR system showed signicant changes with the sequential extension of SRT. The VSS/SS ratio of the UMBR corresponded to the range of CASP known to be 0.50.8 in most cases. Furthermore, the relatively stable ratio of VSS to SS implicated that the amount of active biomass and inorganic remainders constituted a dynamic balance. There was no accumulation of inert fractions in the bioreactor during the pilot-scale experiment [14]. From Figs. 24, it can be seen that the variation of sludge SS and VSS had no effect on the removal efciencies of COD, NH3 -N, and SS. High treatment efciency could always be maintained regardless of the absolute level of sludge concentration in the bioreactor. The evolution of sludge concentration with time was just an inevitable result of the passive adaptation of micro-organisms to the change

Fig. 4. Evolution of inuent SS as a function of time.

78

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

Fig. 5. Evolution of sludge concentration as a function of time.

in the inuent concentration and operation parameters such as HRT and SRT. The higher the inuent concentration, the shorter the HRT and the longer the SRT, the higher the sludge concentration (SS, VSS), and vice versa. This may be due to the fact that the quality of UMBR efuent was no longer restricted by the setting of activated sludge and therefore irrelevant to the sludge concentration in the bioreactor. The relationship between sludge COD and VSS during the long-term experiment is illustrated in Fig. 6. The correlation coefcient (R2 ) was 0.9629,

suggesting the favorably linear relationship between sludge COD and VSS. The ratio of COD to VSS was 1.428, which was close to the theoretical value of 1.415 derived from the bacteria formula (C5 H7 NO2 ) when completely oxidized. The difference between the experimental and theoretical data was only 0.9%. Furthermore, this evinced a new method for quick determination of sludge concentration by measuring its COD concentration. When the CO D/VSS and VSS/SS ratios were taken into account, the sludge SS could be computed [7]. The precision was sound enough to meet the requirements of scientic research

Fig. 6. Relationship between VSS and COD of sludge.

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

79

Fig. 7. Distribution of COD removal efciency in the UMBR system.

and practical operation of a wastewater treatment plant. 3.3. Distribution of removal efciency In the UMBR system, COD removal was accomplished by the two functional units viz. the bioreactor and the membrane module. Their respective contributions to removal efciency are presented in Fig. 7. The total removal of COD was 97% on an average during the whole experiment, out of which, 85% was removed by the bioreactor. Only the residual 12% resulted from membrane separation. It was demonstrated that the removal of organic pollutant in terms of COD was mainly due to the bioreactor unit. The role of membrane module was to conne the biomass within the bioreactor, and thus, maintain a good separation of activated sludge and treated water. From Fig. 7, it can also be seen that COD removal by the bioreactor was as low as 61.9% on an average when the inuent concentration had a sudden decrease on Days 27, 58, 87, 98, and 154. However, COD removal by the membrane module remarkably went up to 32%, which was about one-third that of the bioreactor. On the other hand, if the inuent concentration had a sudden increase, COD removal by the bioreactor exceeded 90% on an average. Removal by the membrane module was generally lower than 10% on Days 11, 35, 66, 80, 98, 111, 136, and 154. These demonstrated that the membrane module played a minor role in COD removal if the inuent concentration was higher, and vice versa.

3.4. Mass loading rate and volumetric loading rate The evolution of sludge mass loading rate over time is shown in Fig. 8. Taking Runs 13 as a whole, the averaged mass loading rate was about 0.54 kg COD kg VSS1 per day, which was in the range of CASP. Due to the uctuation of inuent concentration, the mass loading rates had proportional response during the long-term experiment on the UMBR. The mass loading rate on Day 65 was only 0.49 kg COD kg VSS1 per day. Because of the sudden increase in inuent COD, the mass loading rate went up to the maximum value of 1.99 kg COD kg VSS1 per day on the following day, which was about 520 times that of CASP (it generally ranged from 0.1 to 0.4 kg COD kg VSS1 per day). However, the efuent COD still remained as low as 10 mg l1 . It can be concluded that the UMBR had a strong ability to resist a shock loading. We noticed that a very low or high mass loading rate would inevitably lead to poor efuent quality in CASP. However, the uctuation of mass loading rates would not impose a negative effect on the quality of UMBR efuent. The mass loading rate, one of the most important parameters in CASP design, was no longer crucial to the full-scale design of the UMBR system when applied to urban wastewater reclamation [7]. With an extension of the SRT, the averaged mass loading rates were sequentially diminished from 0.76 to 0.6 and then 0.4 kg COD kg VSS1 per day (see Table 3). This was because a longer SRT induced more of sludge VSS when the HRT and inuent concentration were unchanged. As a result, the mass loading

80

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382

Fig. 8. Evolution of mass loading rates as a function of time.

Table 3 Mass loading rates in various operation periods Items Run 1 Run 2 Run 3 Overall Minimum (kg COD kg VSS1 per day) 0.52 0.11 0.08 0.08 Average (kg COD kg VSS1 per day) 0.76 0.6 0.4 0.54 Maximum (kg COD kg VSS1 per day) 1.09 1.99 1.5 1.99

rates were decreased thereafter. The lower the sludge mass loading rates, the better the purication effect of the UMBR. Therefore, longer SRT was recommended in the UMBR operation. The evolution of volumetric loading rates is illustrated in Fig. 9. The maximum volumetric loading

rate of 10.72 kg COD m3 per day occurred on Day 112, which was 1327 times that of CASP (0.40.8 kg COD m3 per day). However, the corresponding efuent COD on that day (see Fig. 2) was only 21 mg l1 . It was proven once again that the UMBR system had the potential to tolerate a shock loading.

Fig. 9. Evolution of volumetric loading rates as a function of time.

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382 Table 4 Volumetric loading rates in various operation periods Items Run 1 Run 2 Run 3 Overall Minimum (kg COD m3 per day) 0.83 0.45 0.99 0.99 Average (kg COD m3 per day) 1.38 1.63 3.15 2.1

81

Maximum (kg COD m3 per day) 2.84 6.85 10.72 10.7

As shown in the gure, the minimum volumetric loading rate was obtained at 0.45 kg COD m3 per day on Day 28, which approached the lower limit of CASP. The average volumetric loading rate was 2.1 kg COD m3 per day that was about two to ve times that of CASP. This indicated that, treating the same wastewater, use of UMBR could not only eliminate the secondary clarier but also reduce the size of the bioreactor by two to ve times in comparison with the CASP. Therefore, the UMBR was treated as one of the cost-effective solutions [7]. As shown in Table 4, the average volumetric loading rates were increased steadily from 1.38 to 1.63 and then 3.15 kg COD m3 per day during Runs 13, respectively. This was caused by increase in the inuent concentration, while the HRT was kept constant in the study. 3.5. Reuse potential of UMBR efuent The efuent quality of the UMBR is listed in Table 5. It shows that the UMBR system can provide a good-quality efuent that is completely acceptable for reuse. The reclaimed water in this study can be directly reused for municipal watering, toilet ushing, and car washing. After the softening treatment, the reclaimed water could be used as cooling supply and processing water. Therefore, lots of urban wastewater can be effectively harnessed, and moreover, large quantities of water could be saved. The development of water industry would be more sustainable [18].

municipal purposes and indirectly for industrial water supply. The removal efciency of COD was on the average as high as 97%, in which 85% was attributed to the bioreactor and the residual 12% resulted from membrane separation. The averaged removal of NH3 -N and SS could reach 96.2 and 100%, respectively. The highest sludge concentration in terms of SS and VSS was 23.1 and 13.5 g l1 , respectively. The VSS to SS ratio of activated sludge was vacillated around 0.6,
Table 5 Comparison between efuent quality and drinking water standards Items Color Turbidity pH value Chloride as Cl Fluoride as F Nitrate as N Nitrite as N Hardness as CaCO3 Phenols Cyanide as CN Sulfate as SO4 2 Arsenic Mercury Chromium as Cr6+ Manganese Lead Iron Total coliforms Units TCUb NTUb mg l1 mg l1 mg l1 mg l1 mg l1 mg l1 mg l1 mg l1 mg l1 g l1 mg l1 mg l1 mg l1 mg l1 UMBR efuent <2.5 <2 8.2 45.4 0.3 19.0 0.1 325 <0.002 <0.002 23.0 <0.001 <0.2 <0.004 <0.05 <0.01 <0.05 d Water reusea 30 10 6.09.0 350
c c c

450
c c c c c c

0.1
c

0.4 3 number 100 ml1

4. Conclusions From the technical point of view, reclamation of urban wastewater by the UMBR was basically applicable. The reclaimed water could be reused directly for

a Water reuse standard for car washing, land watering etc., PR China (CJ25.1-89). b TCU: true color units; NTU: nephelometric turbidity units. c No requirement. d The ultraltration membrane could effectively retain the bacteria (size from 0.5 to 5 m) and viruses (size from 0.01 to 0.3 m). As a result, there were no coliforms and MS-2 virus detected [4,19].

82

C.-H. Xing et al. / Journal of Membrane Science 177 (2000) 7382 [6] K. Brindle, T. Stephenson, The application of membrane bioreactors for the treatment of wastewaters, Biotechnol. Bioeng. 49 (1996) 601. [7] C.-H. Xing, Ceramic membrane bioreactor for urban wastewater treatment and membrane fouling mechanism, Ph.D. Thesis, Department of Environmental Science and Engineering, Tsinghua University, 1998 (in Chinese). [8] C.-H. Xing, Y. Qian, X.-H. Wen, Y.-B. Meng, Performance of microltration membrane bioreactor for domestic wastewater reclamation, in: Proceedings of WEFTEC Asia98, Singapore, Vol. 1, 1998, p. 119. [9] K. Yamamoto, M. Hiasa, T. Mahmood, Direct solidliquid separation using hollow ber membrane in an activated sludge aeration tank, Water Sci. Technol. 21 (1994) 43. [10] C. Chiemchaisri, K. Yamamoto, S. Vigneswaran, Household membrane bioreactor in domestic wastewater treatment, Water Sci. Technol. 27 (1994) 171. [11] F.W. Trouve, V. Urbain, J. Manem, Treatment of municipal wastewater by membrane bioreactor: results of a semi-industrial pilot scale study, Water Sci. Technol. 30 (1994) 151. [12] D.A. White, M. Asaadi, Fouling behavior in inorganic tubular membranes, in: Proceedings of the International Technical Conference on Membrane Separation Process, Brighton, UK, 2426 May 1989, p. 143. [13] American Public Health Association/American Water Works Association/Water Environment Federation, Standard Methods For The Examination Of Water And Wastewater, Washington, DC, USA, 1992. [14] E.B. Muller, A.H. Stouthamer, H.W. van Verseveld, D.M. Eikeiboom, Aerobic domestic wastewater treatment in a pilot plant with complete sludge retention by cross-ow ltration, Water Res. 29 (1995) 1179. [15] K. Hanaki, C. Wantawin, S. Ohgaki, Effects of the activity of heterotrophs on nitrication in a suspended growth reactor, Water Res. 24 (1990) 289. [16] S. Chaize, A. Huyard, Membrane bioreactor on domestic wastewater treatment: sludge production and modeling approach, Water Sci. Technol. 23 (1991) 1591. [17] N.J. Horan, Biological Wastewater Treatment Systems: Theory and Operation, Wiley, Chichester, England, 1990. [18] Z. Zhang, Y. Qian, Water saving and wastewater reuse and recycle in China, Water Sci. Technol. 23 (1991) 2135. [19] N. Cicek, H. Winnen, M.T. Suidan, B.E. Wrenn, V. Urbain, J. Manem, Effectiveness of the membrane bioreactor in the biodegradation of high molecular weight compounds, Water Res. 32 (1998) 1553.

which corresponded to the range of CASP known to be 0.50.8. The relationship between VSS and COD of activated sludge was linear and the VSS to COD ratio was about 1.428, which was close to the theoretical value of 1.415. The UMBR system had a strong ability for resisting shock loading. Its mass loading rates corresponded to those of the CASP, but the volumetric loading rates were two to ve times those of CASP. This demonstrated that the volume of the bioreactor by UMBR could be greatly reduced in comparison with that by CASP if the same wastewater was treated. Therefore, a large amount of space and investment could be saved. Acknowledgements The authors wish to acknowledge the equipment and nancial support from CIRSEE Lyonnaise des Eaux, France and the State Key Laboratory of Environmental Simulation and Pollution Control at Tsinghua University, China. References
[1] Metcarlf & Eddy, Wastewater Engineering: Treatment, Disposal and Reuse, 3rd Edition, McGraw-Hill, New York, 1991. [2] L. Defrance, M.Y. Jaffrin, Comparison between ltrations at xed transmembrane pressure and xed permeate ux: application to a membrane bioreactor used for wastewater treatment, J. Membr. Sci. 152 (1998) 203. [3] Anonymous, New development in membrane bioreactors: increasing the capacity of existing biological treatment plants, Filtration Sep. 35 (1998) 792. [4] J. Manem, R. Sanderson, Membrane bioreactors. In: J. Mallevialle, P.E. Odendaal, M.R. Wiesner (Eds.), Water Treatment Membrane Process, McGraw-Hill, New York, 1996. [5] A. Nagano, E. Arikawa, H. Kobayashi, The treatment of liquor wastewater containing high strength suspended solids by membrane bioreactor system, Water Sci. Technol. 26 (1992) 887.

Anda mungkin juga menyukai