Sie sind auf Seite 1von 11

Desalination 244 (2009) 153163

Pretreatments to control membrane fouling in membrane filtration of secondary effluents


Soun-Ok Baek, In-Soung Chang*
Department of Environmental Engineering, Hoseo University, Asan, 336-795, South Korea Tel. +82 (41) 540-5744; email: cis@hoseo.edu Received 5 February 2007; Accepted 30 April 2008

Abstract Membrane processes have been used as a key technology for water reclamation and reuse of secondary effluent discharged from municipal wastewater treatment plants. However, its extensive practices are limited due to membrane fouling. To control and manage the membrane fouling properly, pretreatment methods were compared and evaluated. Prior to direct membrane filtration of secondary effluent, the effect of coagulation with alum and ferric sulfate on membrane fouling was investigated using two different ultrafiltration membranes (YM30 and PM30). Membrane filterability was enhanced by addition of alum and ferric sulfate respectively. This was due to the effective destabilization of colloidal particles, which was confirmed by measuring particle size distribution. Soluble foulants present in secondary effluents were entrapped to coagulated flocs. This reduced the concentration of soluble foulants, which lead to a decrease in Rf values. The hydrophobic membrane (PM30) showed high flux declines, whereas the hydrophilic membrane (YM30) showed relatively low flux decline. For the purpose of controlling membrane fouling, a pretreatment using coagulation is more efficient for hydrophobic than hydrophilic membranes. This could give us a good criterion for selecting membrane materials for water reuse practices. Keywords: Activated sludge; Coagulation; Flux; Membrane; Ultrafiltration; Rc; Rf

1. Introduction Water demands in Korea have been gradually increased, but available water resources are very limited. According to the national plan for the development of water resources [1], 1.8 billion
*Corresponding author.

tons of water will be deficient in 2011. Dam construction is not considered as a good alternative for reserving water resources because it might destroy local ecology. Moreover, deteriorated water quality could not be recovered despite national investments and management of water quality. Therefore, the available water resources

0011-9164/09/$ See front matter 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.desal.2008.04.043

154

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

are getting limited every year. In this situation, water reuse-system is gaining a more attention. Secondary effluent, currently discharged from municipal wastewater treatment plants (WWTP) to the receiving water bodies, could be possibly reused to various purposes such as agricultural irrigation, landscape irrigation, industrial recycling, groundwater recharge, stream flow augmentation, and non-potable reuse [2]. Legislation of mandatory construction of dual water supply system at the new building is one of successful examples encouraging water reuse in municipal areas. There are two ways to begin water reuse: either retrofitting the current WWTP facilities or constructing extra advanced processes after secondary treatments. Many WWTPs are being retrofitted for the purpose of water reuse and nutrients removal. Biological nutrients removal (BNR) processes modifying the current conventional WWTP is a typical example of retrofitting. Ozonation, carbon adsorption and membrane filtration are typically used for extra-installation of the advanced treatment of secondary effluent. The most frequently practiced technique for water reclamation and reuse is membrane separation. Membrane processes can easily satisfy the required water quality for reuse due to its excellent separation capability. Moreover, it makes the process very stable, automatically controllable and compact. Therefore, many applications of membrane techniques have been applied to wastewater reclamation and reuse for the last decades. Membrane bioreactors (MBR) and direct filtration of secondary effluents with microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO) are widely used processes for wastewater reclamation and reuse [36]. Notwithstanding many advantages, the widespread application of membrane processes is constrained by membrane fouling, which is mainly associated with cake layer formation on the membrane surface. Membrane fouling leads

to permeate flux decline, making more frequent membrane cleaning and replacement necessary, which concomitantly increases the maintenance and operating costs. Therefore, a strategy of membrane fouling control should be preceded for efficient and economical operation of membrane processes [710]. Coagulation and adsorption are widely used pretreatment unit processes for water reclamation and reuse [11,12] Therefore, the aim of this study was to investigate effect of pretreatment of secondary effluents on membrane fouling for direct filtration of secondary effluent. Particularly, conventional coagulants such as alum and ferric sulfates were used for the pretreatment of the secondary effluents prior to direct membrane filtration. Membrane filtration characteristics were investigated before and after pretreatment.

2. Materials and methods 2.1. Sample preparation for pretreatment to membrane filtration Activated sludge was collected from an aeration basin at the Chonan city WWTP in Korea. The sludge was delivered to the laboratory and the sludge was settled for 90 min. Supernatant of the settled sludge represents the secondary effluent. Aluminum sulfate (Al2(SO4)31418H2O) and ferric sulfate (Fe2(SO4)3nH2O) were used for coagulating agents in this study. A series of jar tests with the secondary effluent was carried out to select optimum coagulant dosage and pH. Rapid mixing (300 rpm) for 1 min and subsequently slow mixing for 15 min were given for effective coagulation and flocculation. After coagulation and flocculation, the secondary effluent was settled again and then the supernatant was used as a feed solution for direct membrane filtration. Activated carbon (850 140 mm) was also used to verify if the carbon adsorption could be a possible alternative for

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

155

Fig. 1. Schematic of stirred cell system.

pretreatment to membrane separation. Adsorption was carried out separately and the jar test condition was the same as the coagulation. 2.2. Membrane filtration Fig. 1 shows the stirred batch cell (8200, Amicon, US) system. The permeate flux was determined by weighing permeates on an electronic top-loading balance (LP220S, Sartorious, Germany) connected to a personal computer equipped with an autoreading program. Transmembrane pressure (TMP) was regulated at 1.4 bars using nitrogen gas and the stirring speed was about 180 rpm for all tests in this work. Preliminary studies showed that 1.4 bars of TMP was good enough to produce permeate in a mean time. Two UF membranes with different hydrophobicity (YM30 and PM30, Amicon, US) were used to filtrate the pretreated secondary effluent. Both membranes have the same molecular weight cut-off (MWCO) of 30,000 Daltons. Relative hydrophobicity of each membrane was characterized by measuring the contact angle between water droplets and membrane surface with a goniometer (NRL, Rame Hart, US). Based on the

contact angle measurements, the relative hydrophobicity was found to be PM30 > YM30. The fresh membrane was first rinsed by letting it float skin-side down in ultrapure water for 90 min. The rinsing water was changed three times during this cleaning period. The cleaned membrane was placed in the stirred cell. 2.3. Resistance analysis Before each experiment the pure water flux (Jiw) was measured with deionized water. The stirred cell was then emptied and filled with the pretreated secondary efflunt. UF was performed until the volume concentration ratio (VCR) reached 5: VCR = Vfeed/Vconc where Vfeed = feed volume and Vconc = concentrate volume. Then the stirred cell was emptied again and refilled with the deionized water. The surface rinsing of the tested membrane with the pure water continued for 10 min without applying pressure, and then the rinsing water was discarded. Pure water flux was determined with the ultrapure water right after the surface rinsing (Jfw). The resistance-inseries model [13] is applied to evaluate the filtration characteristics.

156

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

J = PT /(@Rt) Rt = Rm + Rc + Rf

(1) (2)

where J is the permeation flux, PT is the TMP, is the viscosity of the permeate, Rt is the total resistance, Rm is the intrinsic membrane resistance, Rc is the cake resistance formed by the cake layer deposited over the membrane surface, and the fouling resistance, Rf, is the resistance caused by solute adsorption into the membrane pores and walls. Each resistance value can be obtained through Eqs. (3)(5): Rm = PT /(@Jiw) Rf = PT /(@Jfw)!Rm Rc = PT /(@J )!(Rm+Rf) (3) (4) (5)

Fig. 2. Flux decline during ultrafiltration of the secondary effluent with YM30 and PM30.

Jiw, Jfw and J are flux values determined experimentally. Jiw is the initial water flux before UF, Jfw is final water flux after removing cake layer, and J is the stabilzed flux with the pretrerated secondary effluent. 2.4. Analytical methods Water quality of the secondary effluents, pretreated secondary effluent and membrane permeate were determined according to the analytical methods described in Standard Methods [14]. Particle size distribution was analyzed by the particle sizer (Helos, Sympa-TC, Germany), which can scan the particle size from 0.5 to 350 m.

3. Results and discussion 3.1. Filtration characteristics as a function of membrane hydrophobicity The activated sludge delivered from WWTP

was settled for 90 min. The resulting supernatant was collected and used as the feed solution for membrane filtration. The supernatant was considered as secondary effluent in this study. Fig. 2 shows the flux declines during UF of the secondary effluent with YM30 and PM30 membranes. The extent of flux decline is significantly different according to membrane materials. The PM30 membrane, which was more hydrophobic than YM30, showed more abrupt flux decline than the hydrophilic membrane did. Accordingly, the degree of flux decline is intimately related with the membrane hydrophobicity. This results from hydrophobic interaction between the membrane surface and colloidal matters present in the secondary effluent. Microbial cell surfaces are known to exhibit hydrophobic molecules such as proteins or lipids [15]. Therefore, the hydrophobic interaction may result from the attractive forces existing between the complex surface components of colloidal matters and the membrane surface. As a consequence, the hydrophobic membrane (PM30) exhibited a higher fouling tendency than the hydrophilic membrane (YM30). The importance of hydrophobic interaction has been recognized by many other researchers. For example, Ridgway et al. [16] reported that the hydrophobic interaction between bacterial cell surface components and the RO membrane surface plays an important role in bacterial adhesion

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163 Table 1 Calculated resistances after ultrafiltration of the secondary effluent Membrane YM30 PM30 Rm (1011 @ m!1) 21 6.7 Rc (1011 @ m!1) 7.4 14.7 Rf (1011 @ m!1) 0.1 5.0 Rc /(Rc+Rf ) (%) 98.8 74.6

157

Fig. 3. Flux declines during ultrafiltration of the control (I) and the supernatant Alum solution with YM30 and PM30.

and biofilm formation. Some researchers have suggested that the hydrophobic substance such as a humic fraction of organic matter is a major foulant which controls the rate and extent of fouling [17]. However, the hydrophobic interaction might not be the only reason for this phenomenon. The different fouling tendency depends not only on the membrane hydrophobicity but also on the membrane surface properties. The membranes used in this study do not have the same surface properties. For example, the surface porosity of the YM30 membrane (~50%) is quite different from that ( >10%) of the PM30 membrane [18]. To evaluate the characteristics of membrane fouling, resistance values for both membrane were calculated (Table 1). Rc was predominant resistance for both case, indicating the cake layer is mainly responsible for the membrane fouling.

It should be noted that the ratio of the Rc and Rf is different for each membrane. The ratio Rc /(Rc+Rf) of YM30 was 98.8%, whereas it was 75% for PM30. This is because the soluble foulants present in the secondary effluent had a greater affinity to PM30 membrane than to YM30. In other words, the hydrophobic interaction between the soluble foulants and PM30 membrane was stronger than YM30. This might be due to the hydrophobic nature of the soluble foulants originating from the soluble microbial products in the activated sludge. 3.2. Effect of coagulation on the membrane filterability To examine optimum dosage of coagulants and pH, preliminary experiments with the secondary effluent were carried out using a jar-

158

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

tester. Optimum pH and dose of aluminum sulfate (= alum) were found to be 8 and 50 mg/L, 7 and 10 mg/L for ferric sulfate, and 8 and 500 mg/L for activated carbon adsorption. Coagulated secondary effluent was settled and supernatant was collected. Then it was used as a feed solution for direct membrane filtration. Fig. 3 shows membrane filtration characteristics of the secondary effluent with and without alum pretreatment. The legend in the figure, supernatant-Alum means that the feed of membrane filtration was the supernatant from alum pretreatment, and Control I means that the membrane filtration was carried out with the secondary effluent. As described in the previous section, flux of the hydrophobic membrane (PM30) was much lower than that of the YM30. Alum addition resulted in increase in 35 l/m2.h of the stabilized flux for YM30, and 80 l/m2.h for PM30. Since supra-colloidal materials (or suspended solids) were already removed by settling, the colloidal solids and solutes (probably SMP and/or EPS) are considered as main constituents of the membrane fouling. To have a further insight of the fouling phenomenon, several resistances were calculated and summarized in Table 2. Both Rc and Rf decreased after alum pretreatment. The overall reduction of resistance (=Rc+Rf) for PM30 was much larger than that of YM30. However, Rc was reduced much more than Rf was. For the case of YM30, the ratio Rc /(Rc+Rf) expressing the percentage of the cake resistance to overall resistance was not such a large difference between the control and the alum pretreatment. On the other hand, the Rc /(Rc+Rf) ratio of alum pretreated PM30 filtration ranged from 6273% of the control, indicating that the contribution of Rc to total resistance decreased. This is because the colloidal matters which could be formed the cake layer, were coagulated, removed and thereby contribution of the cake layer to overall resistance was decreased. Fig. 4 shows the particle size distribution of

Fig. 4. Comparison of particle size distribution of the control (I), the supernatant Alum solution, and the mixed liquor of the coagulated solution with Alum.

three different solutions: the secondary effluent without coagulation was indicated as control I, the coagulated solution was indicated as Alum mixed liquor, and the supernatant after settling of the alum mixed liquor was indicated as supernatant Alum. Mean particle sizes of each solution were 53.1, 102.6 and 21.9 m, respectively. The alum mixed liquor shows the biggest particle size among them, indicating that a successful coagulation was carried out. Since most big particles were removed by settling after the coagulation, only small particles (21.9 m) were remained in the supernatant Alum solution. The particulates smaller than 7 m, which are known to be mainly responsible for severe membrane fouling in MBR processes, were not observed in the supernatant Alum solution. Soluble COD, which is another important factor affecting membrane fouling, was measured before and after alum pretreatment. Soluble COD concentration of the control I and supernatant Alum were 14 and 7 mg/L, respectively. Soluble foulants existed in the secondary effluents might be entrapped to the flocs during the coagulation.

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

159

Fig. 5. Flux decline during ultrafiltration of the control (II) and the supernatant of the coagulation with ferric sulfate. Table 2 Calculated resistances after ultrafiltration of the control (I) and the supernatant Alum solution Rc (1010 @ m!1) Control (I) YM30 Run1 Run2 Run3 Run4 PM30 Run1 Run2 Run3 Run4 23.2 56.0 60.0 57.4 141 167 120 105 Rf (1010 @ m!1) Rc+Rf (1010 @ m!1) Supernatant Alum 8.9 8.9 9.0 7.3 97.3 100.4 69.0 42.6 Rc /(Rc+Rf) (%) Control (I) 81 89 91 93 62 67 67 73 Supernatant Alum 67 94 98 73 26 25 34 38

Supernatant Control Alum (I) 6.0 8.4 8.8 5.3 25.1 25.4 23.2 16.1 5.6 7.0 6.0 4.1 85.8 83.6 58.6 38.7

Supernatant Control Alum (I) 2.9 0.5 0.2 2.0 72.2 75.0 45.8 26.5 28.8 63.0 66.0 61.5 226.8 250.6 178.8 143.7

Notes: Control (I): Secondary effluent used as control for alum coagulation. Supernatant Alum: Supernatant after coagulation with alum. Runs 1, 2, 3, 4: Repeated experiments with the same conditions.

The reduced concentration of soluble foulants resulted in a decrease in Rf values as shown in Table 2. Fig. 5 shows membrane filtration characteristics of the secondary effluent with and without ferric sulfate pretreatment. Flux of the PM30 was lower than that of the YM30. Ferric sulfate addition resulted in increase in 52 l/m2.h of the stabilized flux for PM30, and 32 l/m2.h for YM30. Table 3 shows the resistance calculated from the

membrane filtration. Like alum pretreatment, Rc and Rf decreased after ferric sulfate coagulation and Rc was reduced further than Rf. In the case of PM30, the Rc /(Rc+Rf) ratio of the control (control II) was reduced from 74 to 55% after pretreatment, whereas it was reduced from 92 to 87% in the case of YM30. All of these indicate that pretreatment using coagulation for the purpose of controlling membrane fouling is more efficient for hydrophobic membranes than for hydrophilic

160

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163

Table 3 Calculated resistances after ultrafiltration of the control (II) and the supernatant of the coagulation with ferric sulfate, Fe2(SO4)3 Rm (1011@m!1) YM30 Control (II) Supernatant Fe2(SO4)3 PM30 Control (II) Supernatant Fe2(SO4)3 21.0 19.7 18.5 16.9 Rc (1011@m!1) 3.3 1.3 8.6 3.1 Rf (1011@m!1) 0.3 0.2 3.0 2.5 Rt (1011@m!1) 24.6 21.2 30.1 22.5 Rc + Rf (1011@m!1) 3.6 1.5 11.6 5.6 Rc /(Rc+Rf ) (%) 92 87 74 55

Notes: Control (II): Secondary effluent used as control for iron sulfate coagulation. Supernatant Fe2(SO4)3: Supernatant after coagulation with Fe2(SO4)3.

Fig. 6. Flux declines during ultrafiltration of the control (III) and the supernatant after carbon adsorption.

membranes. This could give us a good criterion for selecting membrane materials for water reuse practices. Comparing the flux enhancement of ferric sulfate and alum, alum pretreatment showed a slightly higher efficiency than ferric sulfate. The optimum dosage for iron sulfate in this study (10 mg/L) was lower than that of the alum (50 mg/L), but iron salt usually produces precipitation sludge heavier than aluminum salt does. Further studies on the economic and efficient pretreatment for controlling membrane fouling are needed. Fig. 6 shows flux profile of the secondary effluent and the supernatant after carbon adsorp-

tion. Like the previous results, flux of the PM30 was lower than that of the YM30. Carbon adsorption led to increase in flux of 18 l/m2.h for PM30, and 10 l/m2.h for YM30. Table 4 compares the calculated resistance values before and after carbon adsorption. Opposite to the case of pretreatment using coagulating agents, reduction in Rf was greater than that of Rc. In the case of YM30, the Rc /(Rc+Rf ) ratio of the control (control III) increased from 79 to 93% after adsorption, and the ratio for PM30 was not changed. Considering activated carbon could remove the soluble foulants rather than particulates, it could be a feasible explanation for the meaningful decrease in Rf. After all, carbon adsorption is suitable for reduc-

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163 Table 4 Calculated resistances after ultrafiltration of control (III) and supernatant of carbon adsorption Rm (1011@m!1) YM30 Control (III) Supernatant GAC PM30 Control (III) Supernatant GAC 18.7 18.7 15.8 14.3 Rc (1011@m!1) 8.0 7.9 13.3 11.8 Rf (1011@m!1) 2.1 0.6 3.3 2.9 Rt (1011@m!1) 28.8 27.2 32.4 29.0 Rc + Rf (1011@m!1) 10.1 8.5 16.6 14.7 Rc /(Rc+Rf ) (%) 79 93 80 80

161

Notes: Control (III): Secondary effluent used as control for carbon adsorption. Supernatant GAC: Supernatant after GAC adsorption.

(a)

(b)

Fig. 7. Water qualities of the secondary effluent and their variation along the pretreatment and membrane filtration: (a) COD and (b) turbidity.

ing Rf rather than Rc. However, flux enhancement by carbon adsorption should be optimized because the effect of carbon type (size and raw materials etc.) and operation mode (column or dispersion) should be investigated further. On the other hand, the water qualities of the secondary effluent and their variation along the pretreatment and membrane filtration are important for consideration water reuse. Thus, COD and turbidity were measured and compared along the treatment as shown in Fig. 7. COD of the secondary effluent was 24 mg/L, but decreased to 14 mg/L after alum coagulation. The permeate COD from YM30 membrane without alum pretreatment was 7 mg/L, but 5 mg/L with alum pretreatment, indicating that the pretreatment enhanced the water quality as well as membrane filtration performance. COD from the hydro-

phobic membrane (PM30) was lower than the YM30 membrane. This is due to the greater fouling tendency of the PM30 than YM30. Turbidity of the secondary effluent was 6.2 NTU, but decreased to 0.17 NTU after alum coagulation. Turbidities of the permeate from YM30 and PM30 membranes were under 0.2 NTU regardless of pretreatment and membranes, which is good enough water quality to be reused.

4. Conclusions The effect of pretreatment prior to direct membrane filtration of secondary effluent for the purpose water reclamation and reuse on membrane fouling was investigated. The following conclusions could be drawn:

162

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163 treatment plant, J. Kor. Soc. Environ. Eng., 25 (2003) 946954. Eddy and Metcalf, Wastewater Engineering Treatment and Reuse, 4th ed., McGraw-Hill, 2003. S Lee, W.S. Ang and M. Elimelech, Fouling of reverse osmosis membranes by hydrophilic organic matter: implications for water reuse, Desalination, 187 (2006) 313321. S.-K. Yim, W.-Y. Ahn, G.-T. Kim, G.-W. Koh, J. Cho and S.-H. Kim, Pilot-scale evaluation of an integrated membrane system for domestic wastewater reuse on islands, Desalination, 208 (2007) 113124. D. Bixio, C. Thoeye, J. De Koning, D. Joksimovic, D. Savic, T. Wintgens and T. Melin, Wastewater reuse in Europe, Desalination, 187 (2006) 89101. L. Defrance and M.Y. Jaffrin, Reversibility of fouling formed in activated sludge filtration, J. Membr. Sci., 157 (1999) 7378. R.S. Trussel, R.P. Merlo, W. Hermanowicz and D Jenkins, The effect of organic loading on process performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater, Water Res., 40 (2006) 26752683. I.S. Chang, P. Le-Clech, B. Jefferson and S. Judd, Membrane fouling in membrane bioreactors for wastewater treatment, J. Environ. Eng., 128 (2002) 10181029. H.P. Chu and X.Y. Li, Membrane fouling in a membrane bioreactor (MBR): Sludge cake formation and fouling characteristics, Biotech. Bioeng., 90 (2005) 323331. T. Hamada and Y. Miyazaki, Reuse of carwash water with a cellulose acetate ultrafiltration membrane aided by flocculation and activated carbon treatments, Desalination, 169 (2004) 257267. D Abdessemed and G. Nezzal, Treatment of primary effluent by coagulationadsorption ultrafiltration for reuse, Desalination, 152 (2002) 367373. I.S. Chang and C.H. Lee, Membrane filtration characteristics in membrane coupled activated sludge systemthe effect of physiological states of activated sludge on membrane fouling, Desalination, 120 (1998) 221233. APHA, Standard Methods for the Examination of Water and Wastewater, 19th ed., American Public Health Association, Washington, DC, 1995.

1. Regardless of the characteristics of the feed solution and pretreatment methods employed, membrane fouling of the hydrophobic membrane (PM30) was always greater than the hydrophilic membrane (YM30), and Rc was always greater than Rf, indicating that cake layer formation is mainly responsible for the membrane fouling. 2. Membrane filterability was enhanced by the pretreatment using coagulating agents. Both of Rc and Rf decreased. Colloidal particles which are mainly responsible for membrane fouling were coagulated and removed. That was verified by the measurement of particle size distribution. 3. Flux enhancement of the hydrophobic membrane (PM30) by coagulation was always greater than that of the hydrophilic membrane (YM30). Thereby, for the purpose of membrane fouling control, pretreatment using coagulation could be more beneficial for hydrophobic membranes than for hydrophilic membranes. This could give us a good criterion for selecting membrane materials for water reuse practices. 4. Unlike the pretreatment using coagulants, carbon adsorption did not reduce Rc; instead, Rf decreased because the activated carbon could remove the soluble foulants rather than particulates, indicating that the carbon adsorption is not suitable for reducing Rc, which is mainly responsible for the membrane fouling.

[3] [4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

Acknowledgement The authors would like to thank the ChungNam Environmental Technology Development Center for financial support. References
[1] Ministry of Construction and Transportation, Longterm plan for water resources in Korea, 2001. [2] Y Kim, I.S. Chang and D.-R. Lee, Design considerations of reuse treatment facility for the secondary effluent from municipal wastewater [12]

[13]

[14]

S.-O. Baek, I.S. Chang / Desalination 244 (2009) 153163 [15] V. Urbain, J.C. Block and J. Manem, Bioflocculation in activated sludge: An analytical approach, Water Res., 27 (1993) 829838. [16] H.F. Ridgway, M. Rigby and D. Argo, Bacterial adhesion and fouling of reverse osmosis membranes, J. AWWA, 77 (1985) 97106.

163

[17] W. Yuan and A.L. Zydney, Humic acid fouling during ultrafiltration, Environ. Sci. Technol., 34 (2000) 50435050. [18] A.G. Fane, C.J.D. Fell and K.J. Kim, The effect of surfactant pretreatment on the ultrafiltration of proteins, Desalination, 53 (1985) 3755.

Das könnte Ihnen auch gefallen