Fouling Elimination of PTFE Membrane under
Precoagulation Process Combined with
Ultrasound Irradiation
Meng-Wei Wan, Ph.D.
1; Hui-Ling Yang, Ph.D.
2; Cheng-Hung Chang
3; Febelyn Reguyal
4;
and Chi-Chuan Kan, Ph.D.
5Abstract: Precoagulation is one of the effective pretreatments in membrane filtration. This process mitigates membrane fouling, which is the major drawback of membrane technology in drinking water and wastewater treatment. This study investigated the effects of precoagulation under different coagulation mechanisms on membrane fouling. Use of ultrasound in polytetrafluoroethylene (PTFE) membrane cleaning was also evaluated. In precoagulation, synthetic raw water was precoagulated using aluminum sulfate at different coagulation mechanisms, named electrostatic patch effect (EPE), charge neutralization (CN), and sweep flocculation (SW). Flocs produced from different coagulation mech-anisms exhibited different sizes, structures, and strengths. Likewise, the fouling type generated from each mechanism was demonstrated as pore blocking for EPE, cake formation for SW, and combination phenomenon for CN. Moreover, this study indicated that the membrane flux was enhanced in the sequence of EPE> CN > SW. The flux recovery rate after ultrasonic cleaning was in the sequence of SW> CN > EPE. Finally, this study evidenced that the floc characteristics from various coagulation mechanisms affected membrane per-formance, fouling types, and ultrasonic cleaning efficiency. DOI:10.1061/(ASCE)EE.1943-7870.0000406. © 2012 American Society of Civil Engineers.
CE Database subject headings: Coagulation; Filtration; Ultrasonic methods; Wastewater management. Author keywords: Coagulation; Microfiltration; Ultrasonic cleaning; Floc characteristic.
Introduction
In recent years, membrane technology has been widely used in water and wastewater treatment, including microfiltration (MF), ul-trafiltration (UF), nanofiltration (NF), and reverse osmosis (RO) (Bergamasco 2009;Kweon et al. 2009). Compared with conven-tional methods, membrane technologies could provide more effi-cient but less economical water treatment process (Strathmann 1976). High operating costs and membrane fouling were consid-ered as the limitations for their applications. Many researchers have recently devoted their efforts in significantly reducing the operation cost for membrane systems. However, the fouling problem remains
to be the major obstacle to membrane technology (Lamminen et al. 2004).
Fouling problem reduces permeability of a membrane attribut-able to the accumulation of colloids, particles, macromolecules, and salts on the membrane surface, which consequently resulted in flux reduction. This phenomenon increases the maintenance and operating costs attributable to more frequent membrane clean-ing and replacement (Lee et al. 2004;Kim et al. 2006;Lee et al. 2009). Moreover, most of the literature in the past two decades fo-cused on fouling rather than cleaning, although a fouling problem may actually be a cleaning problem. Considerable progress has been executed in understanding the interactions between foulants, membrane and operating conditions (Li et al. 2002).
Pretreatments such as coagulation, adsorption, and ozonation are usually utilized before membrane filtration to minimize fouling problems (Chen et al. 2007). Among these, chemical coagulation is the most common one because of its relatively low cost and ease of application (Pikkarainen et al. 2004). Factors that affect coagula-tion performance and characteristics of membrane fouling include coagulant types, coagulation mechanisms, mixing conditions, add-ing dosages, and pH (Choi and Dempsey 2004;Muthulumaran et al. 2004).
Although pretreatments might have been employed to minimize fouling, most membrane cleaning techniques are still practically inadequate for membrane filtration systems. Typical methods of membrane cleaning that have been used in industrial applications are forward flushing (spiral wound and tubular) and backwashing (hollow fiber), which are useful with colloidal suspensions (Li et al. 2002). In the mean time, other techniques, such as hydraulic, chemical, mechanical, and electrical cleaning processes, were also utilized to mitigate fouling problems (Lammien et al. 2004;Chen et al. 2007). However, disadvantages of these techniques including 1Assistant Professor, Dept. of Environmental Engineering and Science,
Chia Nan Univ. of Pharmacy and Science, 60, Erh-Jen Rd., Sec.1, Jen-Te, Tainan, 71710, Taiwan.
2Researcher, Disaster Prevention and Water Environment Research
Center, National Chiao Tung Univ., 1001 Univ. Road, Hsinchu, 300, Taiwan.
3M.S. Student, Dept. of Environmental Engineering and Science, Chia
Nan Univ. of Pharmacy and Science, 60, Erh-Jen Rd., Sec.1, Jen-Te, Tainan, 71710, Taiwan.
4M.S. Student, Dept. of Chemical Engineering, Univ. of
Philippines-Diliman, Quezon City, 1101, Philippines.
5Assistant Professor, Dept. of Environmental Engineering and Science,
Chia Nan Univ. of Pharmacy and Science, 60, Erh-Jen Rd., Sec.1, Jen-Te, Tainan, 71710, Taiwan (corresponding author). E-mail: cckanev@mail .chna.edu.tw
Note. This manuscript was submitted on November 10, 2010; approved on April 11, 2011; published online on April 13, 2011. Discussion period open until August 1, 2012; separate discussions must be submitted for in-dividual papers. This paper is part of the Journal of Environmental En-gineering, Vol. 138, No. 3, March 1, 2012. ©ASCE, ISSN 0733-9372/ 2012/3-337–343/$25.00.
noncontinuous operation, high chemical costs, waste disposal prob-lems, potential danger of electrolysis, and high capital cost, have limited their field applications.
Ultrasound, a sound wave traveling through a medium at a fre-quency above 18 kHz, has been considered as an effective tech-nique for membrane cleaning in the past decade. Cavitation phenomenon from ultrasonication can be defined as the formation, growth, and implosion of bubbles, and it can be utilized as an ef-fective energy in fouling elimination. It is widely believed that ultrasonic cavitation, acoustic streaming, ultrasonic-induced vibra-tion of membrane, and ultrasonic heating are the main causes for the enhancement (Zhu and Liu 2000; Lamminen et al. 2004;
Muthulumaran et al. 2004). Recently, a number of researchers have demonstrated the effective use of ultrasound for cleaning fouled membranes or for increasing permeate flux of membranes. Lamminen et al. (2004) have illustrated that ceramic membranes were effectively cleaned using ultrasound at frequencies from 70 kHz up to 620 kHz without damaging the membranes. Increases in power intensity of the ultrasonic system increased the cleaned flux ratio. This increase was attributed to an increase in the number of cavitation bubbles and an increase in acoustic energy in the sys-tem. However, few mechanistic models have been proposed to explain the ultrasonic enhancement on the membrane separation processes. Lim and Bai (2003) indicated that the particles that blocked the membrane pores were not effectively removed by ultra-sound assisted cleaning methods. Therefore, a better understanding on the floc characteristics from various coagulation mechanisms in connections with permeate flux, membrane fouling, and ultrasonic cleaning efficiency is warranted.
This study investigated the extent of membrane fouling from different precoagulation processes and their removals by ultrasound energy. Specifically, it was aimed (1) to determine the optimal alum doses in producing different coagulation mechanisms, named electrostatic patch effect (EPE), charge neutralization (CN), and sweep flocculation (SW); (2) to characterize the flocs from EPE, CN, and SW mechanisms based on their particle diameter, floc strength, and fractal dimension; (3) to develop the plausible mech-anisms of precoagulation processes with regards to intrinsic mem-brane, irreversible fouling and cake resistances; and (4) to determine the cleaning efficiency of fouled PTFE membranes by ultrasound energy.
Materials and Methods
Preparation of Synthetic Raw Water
The synthetic raw water (SRW) was prepared by mixing 1 g of bentonite (Sigma-Aldrich) in 1 L of tap water at 200 rpm for 1 h. The turbidity of the SRW was measured by using a Hach 2100P turbidimeter to be 200 10 NTU.
Preliminary Study for Different Coagulation Mechanisms
Jar-tests were carried out to produce feed waters of different char-acteristics for membrane filtration. Doses of aluminum sulfate (0.5, 1, 3, 5, 7, 10, and 15 mg=L as Al) were varied to determine the most suitable coagulant concentration for each coagulation mechanism.
Two liters of prepared SRW was used for coagulation and floc-culation. The pH was controlled at 7 0.3 using 0.1 N HCl and 0.1 N NaOH. Subsequent to the addition of aluminum sulfate, the mixture was agitated at 200 rpm for 1 min (G¼ 350 s1) followed by
flocculation at 30 rpm for 20 min (G¼ 25 s1). Flocs were allowed to settle for 30 min and were then collected.
In the precoagulation, different coagulation mechanisms, in-cluding EPE, CN, and SW, were observed. It was concluded that the EPE mechanism occurred at 1 mg=L of Al wherein the turbidity of the SRW was significantly reduced from 200 to 1.04 NTU. The CN and SW mechanisms took place at 5 and 10 mg=L of Al with zeta potential and residual turbidity of almost 0 mV and 0.49 NTU andþ1:66 mV and 0.98 NTU, respectively. The tests of EPE, CN, and SW coagulation mechanisms were called as the EPE, CN, and SW tests in this study.
Floc Characteristics
Characteristics of the flocs collected from the EPE, CN, and SW tests were evaluated. The flocs were characterized on the basis of their particle size, fractal dimension, and floc strength. The particle size (d50) was determined using a particle size analyzer
(Master-sizer 2000, Malvern, UK). The fractal dimension (Df) was
calcu-lated by adding unity to the slope of the log-log plot of the particle size and settling velocity. The floc strength (γ) was the regression slope of floc size (d50) and mixing intensity (G).
Membranes
Previous physical analysis showed that polytetrafluoroethylene (PTFE) membrane could withstand the ultrasound irradiation. The PTFE membrane with a nominal pore size of 0.5 μm and an effective area of 1:96 × 103m2 was used as the microfilter
(MF) membrane in this study based on the particle size distribution of the prepared SRW. Prior to use, the PTFE membrane was soaked in citric acid for 30 min and was then washed with deionized water. Experimental SetUp
The schematic diagram of the dead-end microfiltration set-up is shown in Fig. 1. The set-up was composed of a feed reservoir, the membrane, a pump, a permeate tank, an electronic scale, and an ultrasonic probe. The feed reservoir with a volume of 20 L was used to contain the feed water. The feed water was pumped to the permeate tank by passing through the PTFE mem-brane. The microfiltration was operated at a transmembrane pres-sure of 85 kPa. The permeate tank was placed on the electronic scale (XB 4200C, Precisa, Switzerland) that was connected to a computer that recorded the mass of cumulative permeate.
The ultrasonic probe (Sonicator 3000, Misonix) provided physi-cal vibration to clean the cake formed on the surface of the mem-brane. It was operated at a constant frequency of 20 kHz and 50 W power.
Dead-End Microfiltration
Fifteen liters of precoagulated SRW were fed to the dead-end mem-brane filtration set-up. A series of five MF tests were conducted. For each test, the feed water was subjected to the MF process for 15 min followed by 1.5 min ultrasonic cleaning.
Analysis of Membrane Fouling Resistances
Membrane fouling resistances were evaluated using the resistance-in-series model (Bae and Tak 2005;Chang and Lee 1998)
J¼ΔP
ηRt ð1Þ
in which J = permeate flux (m3=m2-s);ΔP = transmembrane
pres-sure (Pa); η = viscosity of the permeate (Pa-s); and Rt = total
resistance (m1), which can be calculated using Eq. (2):
Rt¼ Rmþ Rfþ Rc ð2Þ
in which Rm= intrinsic membrane resistance (m1); Rf =
irrevers-ible fouling resistance caused by pore blocking in the membrane pores and on the membrane surface (m1); and Rc = resistance
of the cake formed on the surface of the membrane (m1). The Rm, Rf and Rc can be computed using Eqs. (3)–(5):
Jiw¼ηRΔP iw ð3Þ Jf w¼ηRΔP f w Rm ð4Þ Jss¼ηRΔP ss Rm Rf ð5Þ
in which Jiw= initial water flux (m3=m2-s), which was determined
by filtering deionized water until constant flux was achieved; Jf w=
final water flux (m3=m2-s), which was evaluated by passing
deion-ized water through the fouled membrane after removing the cake formed on the membrane surface; and Jss = steady state flux
(m3=m2-s), which was determined when the flux of the membrane
was almost constant or not changing. Fouled Membranes Examination
The membranes after cleaning at the end of each test were subjected to surface observation using a scanning electron microscope (SEM S-4800, Hitachi, Japan).
Results and Discussion
Precoagulation with Aluminum Sulfate
Characteristics of Flocs from Different Coagulation Mechanisms
The flocs collected from different coagulation mechanisms were characterized on the basis of their particle sizes, fractal dimensions, and floc strengths. These parameters are summarized in Table1. The EPE test produced the smallest particle size followed by the CN and SW tests. As mentioned earlier, EPE, CN, and SW occurred
at 1, 5, of Al, respectively. The increasing size of the flocs from the EPE to CN to SW is associated with the increasing alum dose. Ac-cording to Wu et al. (2009), floc size increases as the alum dose increases because precipitates form on the surface of the particle and in the solution. A particle collides with other particles to form precipitation. More and more precipitations result in the formation of larger flocs.
As shown in Table1, the EPE test also produced the most com-pact and strongest flocs based on its fractal dimension (Df) and floc
strength (γ). Fractal dimension indicates the structure of the floc or the packing of the aggregates. Flocs with high fractal dimension generate denser flocs that are harder to break compared with the flocs with low fractal dimension. Moreover, stronger flocs have lower floc strength values. Floc size and floc strength have an in-verse relationship with fractal dimension. Thus, the SW test pro-duced the largest particle but the weakest and the loosest flocs. Preliminary Filtration Test under Different Coagulation Mechanisms
The effects of coagulation mechanisms on the permeate flux were evaluated in this study. Fig.2shows the declines of the permeate fluxes under three coagulation mechanisms and the blank test. As expected, the blank test showed the most rapid flux decline, and the membrane pores were readily blocked by the particles. The precoa-gulated SRW under all three coagulation mechanisms reduced the rate of flux decline. The EPE test yielded the slowest flux decline rate. As previously mentioned, the EPE best generated the smallest but the strongest and the most compact flocs. In the EPE mecha-nism, although the negative charges on some particles are neutral-ized, repulsion between particles will result in forming filtration cake with larger pore size on membrane surface that cause less flux decline. The SW test produced the largest but the weakest and the most loosely packed flocs in which the particles are surrounded by large amounts of AlðOHÞ3coagulants to perform cosedimentation
effect, and it will result in forming filtration cake with smaller pore
Fig. 1. Schematic diagram of cut-off MF device
Table 1. Floc Characteristics under Different Coagulation Mechanisms using Aluminum Sulfate as Coagulant
Coagulation mechanism d50(μm) γ Df
EPE 39.54 0.262 2.285
CN 56.57 0.298 2.109
SW 58.90 0.310 1.973
sizes on the membrane surface, causing rapid flux decline. Our re-sults show that coagulation pretreatment improves the permeate flux of the membrane. The permeate flux is enhanced in the order of EPE> CN > SW.
Ultrasonic Cleaning
Five MF and ultrasonic cleaning cycles were performed to deter-mine the effect of ultrasonic cleaning on the initial flux and flux recovery under different coagulation mechanisms. Each cycle was composed of 15 min MF process and 1.5 min ultrasonic clean-ing at 20 kHz frequency and 50 W power. Fig.3illustrates the de-clines of the permeate fluxes for the five MF and ultrasonic cleaning cycles, and Table2shows the percent flux recoveries after the ultrasonic cleaning.
The percent flux recoveries in all tests, including the blank test, decreased after each cycle. Ultrasonic energy was limited to the removal of the foulants on the membrane surface. As the number of filtration and ultrasonic cleaning cycle increased, the amount of particles blocking the membrane pores also increased, which re-sulted in decrease in the percent flux recovery. Moreover, among all the tests, the SW test illustrates the highest flux recovery. Be-cause the SW test produced the largest flocs, few membrane pores were blocked by larger particles, where cake was mostly formed on the membrane surface. The cavitation by ultrasound has sufficient energy to generate high velocity fluid movement to remove the cake on the surface of the polymer membrane. Thus, the percent flux recovery increased as the particle size of flocs increased. Time (s) 0 200 400 600 800 1000 1200 1400 1600 1800 2000 J/J 0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 without precoagulation EPE CN SW
Fig. 2. Membrane flux at different coagulation mechanisms using alu-minum sulfate as coagulant
Time (s) (a) (b) (c) (d) 0 1000 2000 3000 4000 5000 J/ J0 0.00 0.02 0.04 0.06 0.08 0.10 0.12 1st filtration 2nd filtration 3rd filtration 4th filtration 5th filtration Time (s) 0 1000 2000 3000 4000 5000 J/ J0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 1st filtration 2nd filtration 3rd filtration 4th filtration 5th filtration Time (s) 0 1000 2000 3000 4000 5000 J/ J0 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 1st filtration 2nd filtration 3rd filtration 4th filtration 5th filtration Time (s) 0 1000 2000 3000 4000 5000 J/ J0 0.02 0.04 0.06 0.08 0.10 0.12 1st filtration 2nd filtration 3rd filtration 4th filtration 5th filtration
Fig. 3. Flux changes after precoagulation filtration with aluminum sulfate combined with ultrasonic cleaning: (a) without coagulation mechanism; (b) EPE mechanism; (c) CN mechanism; (d) SW mechanism
Fouling Mechanism
The fouling resistances and mechanisms of the blank and the differ-ent precoagulation tests were evaluated. The total resistance (Rt)
for each MF and ultrasonic cleaning cycle was calculated using the resistance-in-series model. The total resistances for the blank, EPE, CN, and SW tests were 2:52–6:32 × 107, 1:04–3:26 × 107, 1:79–2:88 × 107, and 2:19–3:94 × 107 m1, respectively. Blank test showed the highest total resistance, and coagulation pretreat-ment decreased the total resistance in the MF process.
Fig.4shows the distribution ratios of intrinsic membrane (Rm),
irreversible fouling (Rf), and cake (Rc) resistances during the tests.
In almost all the tests, the Rf decreased whereas the Rcincreased as
the number of filtration cycle increased. The results imply that the membrane fouling mechanism was predominantly pore blocking, but gradually switched to cake formation. Comparing the results
of different tests, the EPE test illustrates the largest Rf ratios
(65% ∼ 73%), which indicates that pore blocking was the major fouling mechanism in the EPE test. In CN test, the ratio of Rfvaried
between 44% to 55% and Rcbetween 38% to 52%. The Rf and Rc
ratios showing small differences indicates the combined effects of pore blocking and cake formation. On the contrary, the Rc in the
SW test ranged from 50% to 58%, which is larger than 39% to 47% of Rf, which indicates that cake formation was the predominant
fouling mechanism. Thus, the smaller but more compact and stronger flocs results in pore blocking, whereas larger, but looser and weaker flocs results in cake formation.
Membrane Surface
The SEM photographs of the fouled and ultrasonic cleaned mem-branes are shown in Fig.5. Fig.5(a)shows the surface morphology of the membrane fouled by the SRW without precoagulation. The membrane was covered by dense foulants. Fig.5(b)is the surface image of the fouled membrane under the EPE test, which is very different from the blank test (i.e., the interlaced membrane fibers could be seen clearly and several colloids was between the fibers). Figs. 5(c)and 5(d) are the surface images of ultrasonic cleaned membrane surface from the CN and SW tests, respectively. Both show large colloids and a layer of cake on the membrane.
The SEM photographs of the ultrasonic cleaned membranes show the phenomenon of floc blocking between the membrane fi-bers. Although the smaller and compact flocs may have a larger membrane flux in the early part of filtration, these flocs would block the pores and cause inefficient cleaning, because they were Table 2. Flux Recovery after MF and Ultrasonic Cleaning Cycle under
Different Coagulation Mechanisms
MF cleaning cycle
Flux recovery percent (%)
Blank EPE CN SW 1 — — — — 2 59.65 83.24 64.97 73.21 3 42.98 56.42 57.63 71.43 4 29.82 40.78 54.24 69.64 5 29.82 33.52 51.98 61.61
Fig. 4. Resistance ratio variation with filtration-clean loops in different coagulation mechanism tests: (a) blank test without precoagulation; (b) EPE mechanism; (c) CN mechanism; (d) SW mechanism
more difficult to be washed out. Hence, precoagulation with the EPE mechanism would generate more irreversible fouling. The CN mechanism produced flocs with a strength and shape similar to the SW mechanism but larger than that from the EPE mecha-nism. The larger colloids resulted in larger permeate flux and a bet-ter restoration ratio of flux afbet-ter membrane cleaning. Moreover, the SW mechanism that produced the largest but the looser and weaker flocs had a higher potential in cake formation, where the quick for-mation of cake resulted in higher Rtand lower permeate flux.
How-ever, the formed foulant layer on the membrane surface could really be removed by ultrasonic cleaning as evidenced from good restoration ratios of permeate flux.
Conclusions
Flocs generated from three different coagulation mechanisms dif-fered in size, structure, and strength. EPE mechanism produced the smallest but the densest and the strongest flocs, whereas the SW mechanism generated the largest, but the loosest and the weakest flocs. Differences in floc characteristics resulted in different types of fouling types. The predominant mechanisms of the EPE and SW tests are pore blocking and cake formation, respectively. The CN test is in between pore blocking and cake formation. Precoagulation improved the permeate flux. The permeate flux was enhanced in the sequence of EPE> CN > SW, in which floc fractal dimension in-creased as the floc strength and particle size dein-creased. The flux recovery after ultrasonic cleaning decreased in the sequence of SW> CN > EPE. This study evidenced that the different coagu-lation mechanisms affect the types of fouling of membrane and ex-tent of flux recovery by ultrasonic cleaning.
References
Bae, T. H., and Tak, T. M. (2005).“Interpretation of fouling characteristics of ultrafiltration membranes during the filtration of membrane bio-reactor mixed liquor.”J. Membr. Sci., 264(1–2), 151–160.
Bergamasco, R., Bouchard, C., Silva, F. V., Reis, M. H. M., and Fagundes-Klen, M. R. (2009). “An application of chitosan as a coagulant/ flocculant in a microfiltration process of natural water.”Desalination, 245(1–3), 205–213.
Chang, I. S., and Lee, C. H. (1998).“Membrane filtration characteristics in membrane-coupled activated sludge system—the effect of physio-logical states of activated sludge on membrane fouling.”Desalination, 120(3), 221–233.
Chen, Y., Dong, B. Z., Gao, N. Y., and Fan, J. C. (2007).“Effect of coagu-lation pretreatment on fouling of an ultrafiltration membrane.” Desalination, 204(1–3), 181–188.
Choi, K. Y., and Dempsey, B. A. (2004).“In-line coagulation with low-pressure membrane filtration.”Water Res., 38(19), 4271–4281. Kim, H. C., Hong, J. H., and Lee, S. (2006).“Fouling of microfiltration
membranes by natural organic matter after coagulation treatment: A comparison of different initial mixing conditions.” J. Membr. Sci., 283(1–2), 266–272.
Kweon, J. H., et al. (2009).“Evaluation of coagulation and PAC adsorption pretreatments on membrane filtration for a surface water in Korea: A pilot study.”Desalination, 249(1), 212–216.
Lamminen, M. O., Walker, H. W., and Weavers, L. K. (2004). “Mecha-nisms and factors influencing the ultrasonic cleaning of particle-fouled ceramic membranes.”J. Membr. Sci., 237(1–2), 213–223.
Lee, B., Choo, K., Chang, D., and Choi, S. (2009).“Optimizing the coagu-lant dose to control membrane fouling in combined coagulation/ultra-filtration systems for textile wastewater reclamation.”Chem. Eng. J., 155(1–2), 101–107.
Lee, N., Amy, G., Crouc, J. P., and Buisson, H. (2004).“Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM).” Water Res., 38(20), 4511–4523.
Fig. 5. SEM observation of membrane surface after precoagulation filtration with aluminum sulfate combined with ultrasonic cleaning: (a) no precoagulation; (b) EPE mechanism; (c) CN mechanism; (d) SW mechanism
Li, J., Sanderson, R. D., and Jacobs, E. P. (2002).“Ultrasonic cleaning of nylon microfiltration membranes fouled by Kraft paper mill effluent.” J. Membr. Sci., 205(1–2), 247–257.
Lim, A. L., and Bai, R. (2003).“Membrane fouling and cleaning in micro-filtration of activated sludge wastewater.”J. Membr. Sci., 216(1–2), 279–290.
Muthulumaran, S., Yang, K., Sueren, A., Kentish, S., Ashokkumar, M. G. W., and Grieser, F. (2004). “The use of ultrasonic cleaning for ultrafiltration membranes in the dairy industry.”Sep. Purif. Technol., 39(1–2), 99–107.
Pikkarainen, A. T., Judd, S. J., Jokela, J., and Gillberg, L. (2004). “Pre-coagulation for microfiltration of an uplandsurface water.” Water Res., 38(2), 455–465.
Strathmann, H. (1976).“Membrane separation processes in advanced waste water treatment.”Pure Appl. Chem., 46(2–4), 213–220.
Wu, X., Ge, X., Wang, D., and Tang, H. (2009).“Distinct mechanisms of particle aggregation induced by alum and PACl: Floc structure and DLVO evaluation.”Colloids Surf., 347(1–3), 56–63.
Zhu, C., and Liu, G. L. (2000).“Modeling of ultrasonic enhancement on membrane distillation.”J. Membr. Sci., 176(1), 31–41.
