Coagulation–bubbling–ultrafiltration: Effect of floc properties on the performance of the hybrid process

Desalination 333 (2014) 126–133

Contents lists available at ScienceDirect

Desalination journal homepage: www.elsevier.com/locate/desal

Coagulation–bubbling–ultrafiltration: Effect of floc properties on the performance of the hybrid process Jianwei Liu a,b, Baicang Liu a,⁎, Tao Liu a, Yang Bai a, Shuili Yu b a b

College of Architecture and Environment, Sichuan University, Chengdu 610065, China State Key Laboratory of Pollution Control and Resource Reuse, College of Environmental Science and Engineering, Tongji University, Shanghai 200092, China

H I G H L I G H T S • • • • •

A novel hybrid process of coagulation–bubbling–ultrafiltration was proposed. Relationship of air bubble size and gas flow rate was studied. Bubbles and floc properties were measured to explain membrane performance. Concentration polarization resistance was reduced with air bubbles injected. Floc separation by air bubbles holds great promise in reducing membrane fouling potential.

a r t i c l e

i n f o

Article history: Received 11 August 2013 Received in revised form 19 October 2013 Accepted 13 November 2013 Available online 15 December 2013 Keywords: Coagulation Bubbling Ultrafiltration Hybrid process Floc properties

a b s t r a c t A novel hybrid process of coagulation–bubbling–ultrafiltration was proposed to study membrane fouling phenomena by surface water. Relationship of bubbles, flocs and the hollow fibers was explored. When applying less than 20 mL/min gas flow rate, membrane fouling was accelerated with air bubbles introduced. When gas flow rate increased further to 40 mL/min and 60 mL/min, TMP showed a two-stage development trend, which was a fast development in the first few hours followed with a relatively slow development after about 4 h. Unified membrane fouling index (UMFI) increased from 0.00216 (without bubbles) to 0.00274 m2/L (40 mL/min gas flow rate) and 0.00219 m2/L (60 mL/min gas flow rate). As gas flow rate increased, bubble size became bigger, and its distribution range became wider, resulting in higher shear rate in the ultrafiltration column, which led to severe floc breakage. Flocs of small size and compact structure accelerated membrane fouling, resulting in highest UMFI value under 40 mL/min gas flow rate. However, under 60 mL/min gas flow rate, with largest bubbles and highest shear rate examined in this study, concentration polarization was effectively limited. As a result, TMP development slowed down when pore blockage reached equilibrium. © 2013 Elsevier B.V. All rights reserved.

1. Introduction Membrane process of ultrafiltration (UF) applied in water treatment received much attention in recent years due to its small footprint, highquality finished water and relatively low cost [1,2]. Hollow fiber is particularly popular because of its high packing density and the convenience of modularized installation [3]. But like other membrane separation process, hollow fiber ultrafiltration is faced up with the problem of membrane fouling which usually results from particles accumulating on the membrane surface and colloids plugging in membrane pores [4]. Several methods including pretreatment before membrane filtration and hydrodynamic disturbance during filtration were developed to solve this problem [5,6].

⁎ Corresponding author. Tel.: +86 28 85995998; fax: +86 28 85405613. E-mail address: [email protected] (B. Liu). 0011-9164/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.desal.2013.11.029

Huang et al. reviewed the major pretreatment approaches applied before low-pressure membrane filtration [5]. They concluded that compared with adsorption, preoxidation and prefiltration, coagulation has been the most successful pretreatment for fouling reduction. Coagulant type, source water quality and coagulation condition can influence membrane filtration significantly [7]. When coagulant was added in two stages, internal pore fouling could be remarkably limited, Liu et al. explained this as larger and more irregular flocs formed by the addition strategy [8]. Yu et al. observed a lower TMP increase with flocs formed in breakage and regrowth process [9]. They found that breakage followed by regrowth led to lower fractal dimension of flocs and decreased amount of small micro-flocs. Barbot et al. also emphasized the importance of large flocs with open branches to promote permeate flux [10]. Thus, it seems that floc size distribution and structure characteristic had a significant influence on membrane performance. Among many hydrodynamic disturbance methods such as vibration, pulsating flow and vortex generation, gas sparging has been proved to

J. Liu et al. / Desalination 333 (2014) 126–133

be an effective, simple and low cost technique in membrane filtration [11]. Researches of gas sparging have been applied in different kinds of membrane modules like flat sheet membrane [12], tubular membrane [11] and hollow fiber membrane [13,14]. Membrane performance was related with bubble characteristics, gas flow rate and sparging frequency in these studies. When bubbling was applied inside the membrane modules, larger bubbles or bubble slugs are preferred due to their ability to occupy most of the channel thus promoting mass transfer. For submerged modules, using particle image velocimetry, Yeo et al. calculated bubble induced shear stress around hollow fibers and revealed that one of the key parameters in promoting membrane performance was the fluctuation in shear stress [15]. In their research, small bubbles with high frequency showed some beneficial against large ones. Tian et al. investigated influence of bubble size on the fouling of immersed hollow fiber membrane and concluded that the smallest bubble with diameter of 3.5 mm was most effective [1]. However, hybrid process of coagulation–ultrafiltration and bubbling–ultrafiltration mentioned above were both conducted independently. Scarcely literature integrated the three processes of coagulation, gas sparging and ultrafiltration to explore the interaction of bubbles, particles/flocs and membrane. When air bubbles are introduced in the process of filtration followed by coagulation, bubble induced hydrodynamic disturbance can show some impact on membrane performance. Moreover, air bubbles may adsorb some small flocs and bring them to water surface by floatation or bubble induced shear stress may lead to floc breakage and size/strength redistribution. It is of paramount necessity to reveal how bubbles influence floc property and membrane fouling potential in such a hybrid process of coagulation–bubbling–ultrafiltration. In this paper, a novel hybrid process containing coagulation, air bubbling and ultrafiltration to deal with lake water was developed. The interaction of bubbles, particles/flocs and membrane was integrally studied.

127

2.3. Floc property measurement Use of fractal dimension to describe structure characteristic of flocs was presented in numerous studies [16–18]. Detailed concept and measurement of fractal dimension were interpreted in Bushell and coworkers' review [19]. In this study, fractal dimension for crosssectional area of flocs was determined by box-counting method using the following equation [16,19]:

D f ¼ limL→0 −

logðN Þ logðLÞ

ð1Þ

where N is the number of pixel needed to cover floc image and L is pixel size. After in-line coagulation, outlet of flocculation pipe was put on a flat microscope slide to collect flocs. Coagulated water flowed slowly onto the slide and less than 1 mL sample was collected each time to ensure no floc breakage occurs. Flocs in the ultrafiltration column were collected in the same way at the end of ultrafiltration circle. Image of flocs was captured by a microscope (CX41, Olympus Co., Ltd., Japan) with CCD camera and further processed by a software package Image Pro Plus (Version 6.0, Media Cybernetics Inc., US). Specifically, flocs were separated from their background by brightness and contrast adjustment, sharp edge detection and random noise filtering, which was the same as Chakrabroti et al. did in their research [17,18]. Tests were performed in advance to reach the best similarity between flocs in source image and the processed one. After preprocessing, the longest side value of flocs, which is the parameter adopted to feature floc size, was determined by Image Pro Plus software directly. Box-counting fractal dimension of flocs was calculated using Eq. (1) on a professional graphic processing platform MATLAB 7.0 (Mathworks Inc., USA).

2.4. Air bubbling

2. Materials and methods 2.1. Source water and coagulant Source water used in this study was obtained from Mingyuan Lake located in Jiang'an campus of Sichuan University. Lake water quality was continuously monitored during the experiment. Water quality was shown in Table 1. Aluminum sulfate hydrate (Al2(SO4)3·18H2O; Sigma Aldrich, analytical grade) was used as coagulant. Stock solution was prepared at a concentration of 0.01 M in a measuring flask and saved in dark for use.

Air flow was generated by an air pump and controlled through a glass rotameter. Air bubbles were introduced by a porous stainless steel disk (Nanjing Institute of Metallic Membrane, China) with diameter of 30 mm. According to the manufacturer, the disk has a pore size of 0.1–5.0 μm and porosity of 25–50%. The image of bubbles in ultrafiltration column was captured by a digital camera. Bubble size was determined using Image Pro Plus by applying diameter of the column as scale. More than ten pictures were captured and at least 60 bubbles were calculated to perform bubble size distribution analysis.

2.5. Experiment set-up 2.2. Jar test Jar test was performed in a floculator (ZR-6, Zhongrun Co., Ltd., China) to determine optimal alum dosage based on UV254 and turbidity removal. Firstly, a rapid mixing with rotation speed of 300 rpm was maintained for 1 min, and then rotation speed was reduced to 60 rpm allowing floc growth for 12 min.

Table 1 Characteristic of Mingyuan lake water. Parameter

Value

pH Turbidity (NTU) Temperature (°C) TOC (mg/L) UV254 (cm−1) SUVA (mg/L·m)

7.95–8.63 1.08–2.27 21.5–24.6 2.16–2.92 0.030–0.052 1.30–1.82

Fig. 1 shows the apparatus of the coagulation–bubbling–ultrafiltration set-up. A constant-level tank was used to maintain hydrodynamic condition. Magnetic stirring apparatus was used to provide hydrodynamic gradient in coagulation tanks. In rapid mixing tank, rotation speed was set at 300 rpm for 1 min, then reduced to 60 rpm for 6 min in both of the two flocculation tanks, which was identical with jar test. After in-line coagulation, feed water entered membrane module directly and was sucked by a peristaltic pump (BT100-2J, Longer Pump, China). The ultrafiltration module consists of inner tube and outer tube. Hollow fibers with mean pores of 0.02 μm (Litree Purifying Technology Co., Ltd., China) were installed between the two tubes. Effective length of the fibers is 220 mm, with inner and outer diameter of 1.0 mm and 1.8 mm, respectively. 16 fibers were installed corresponding to a total area of 0.02 m2. The volume of the annular cylinder is 300 mL. Permeate flux was set at a constant value of 20 L/m2·h. Thus, the hydraulic retention time (HRT) of feed water in the membrane module is 43 min. TMP data was monitored by a pressure gauge.

128

J. Liu et al. / Desalination 333 (2014) 126–133

Fig. 1. Schematic system of the experiment.

2.6. Analysis methods 2.6.1. Liquid sessile drop contact angle analysis A contact angle measuring instrument DSA30 (KRÜSS GmbH., Hamburg, Germany) was used to perform water contact angle measurement. The procedure was similar to that described in a previous study [20]. Briefly, 2 μL of deionized water was placed on the surface of the membrane which was air dried for more than 24 h and the dynamic water contact angle variations were recorded over 30 s. Measurements were repeated for 6 times. 2.6.2. Water quality Other than ultrafiltration water, all samples were prefiltered with 0.45 μm membrane before TOC and UV254 analysis. After prefiltration, TOC was determined with a TOC analyzer Liqui TOC II (Elementar, Germany). UV254 was analyzed by a spectrophotometer UV-1800PC (Mapada, China). Then dividing UV254 by TOC, specific UV-absorbance (SUVA) was determined. Turbidity was measured by a turbidimeter 2100P (Hach, USA). 2.6.3. Unified membrane fouling index Huang et al. developed unified membrane fouling index (UMFI) to quantify the fouling of low-pressure membrane [21,22]. The concept of UMFI is described by the following equation:

ð4Þ

DI water was filtered for 2 h with clean membrane for compaction and Rm determination. Then ultrafiltration of coagulated water was maintained for 8 h and Rt was calculated. After that, ultrafiltration column was emptied and filled with DI water again, ultrafiltration of DI water was conducted until a stable TMP reached and the new total resistance of Rt1 could be calculated. Rp was determined by subtracting Rt1 from Rt. Finally, particles accumulated on the hollow fibers were gently removed and membrane was immersed in the column again for backwashing. Backwashing was conducted with air flush for 5 min and water flush for 10 min under TMP of 20 kPa and 90 kPa, respectively. DI water was filtered once again and Ri was determined. Then subtracting Rm, Rp and Ri from Rt, it was Ra. 3. Results and discussion

ð2Þ 2

where Vs (unit: L/m ) is permeate volume per membrane area, J′s is normalized membrane specific flux which is defined as: JS J S0

ΔP : J¼  μ Rm þ Ra þ Rp þ Ri

3.1. Membrane hydrophilicity

1 ¼ 1 þ ðUMFIÞV S J′S

J′S ¼

polarization resistance (Rp), intern fouling resistance (Ri) and membrane hydraulic resistance (Rm). Flux is related to TMP (ΔP), dynamic viscosity (μ) of permeate and resistance by the Eq. (4):

ð3Þ

Statistical analysis of membrane contact angle was reported in a box and whisker plot (Fig. 2). Results showed that the membrane was very hydrophilic. The initial contact angle was 56.28 ± 4.30°. As time increased, membrane contact angle decreased almost linearly. The 2 μL water drop disappeared in about 60 ± 20 s. 3.2. Optimal alum dosage

where Js and Js0 is permeate flux normalized to transmembrane pressure at a specific time and time zero respectively. 2.6.4. Resistance Application of resistance in series model in UF process is effective in explaining membrane fouling mechanism [23]. In this model, total resistance (Rt) of membrane composes of adsorption resistance (Ra),

Jar test was performed to decide optimal alum dosage based on UV254 and turbidity removal. As shown in Fig. 3, turbidity removal efficiency was relatively steady when alum dosage reached 0.05 mM. However, UV254 showed a re-increase trend when alum dosage exceeded 0.06 mM. Thus, dosage of 0.05 mM was used during the whole experiment.

J. Liu et al. / Desalination 333 (2014) 126–133

129

Fig. 2. Membrane contact angle versus time.

3.3. Water quality Lake water quality was monitored in different days throughout the experiment period and data was plotted in Fig. 4. Turbidity of lake water was in the range of 1.08–2.27 NTU, but in most days, the turbidity was in a smaller range of 1.61–2.05 NTU. UV absorbance at the wavelength of 254 nm (UV254), which is the indicator of hydrophobic organic matter, ranged from 0.030 to about 0.055. For natural surface water, this fluctuation is inevitable. Quality of coagulated water changed with the change of source water and the trend of both turbidity and UV254 were quite similar to that of source water. 3.4. Gas/liquid two phase flow pattern and bubble size distribution In gas/liquid two-phase flow system, flow patterns are commonly categorized into four levels, which are bubble, slug, churn and annular flow. Discussion about these four flow patterns can be found in Cui and coworkers' review [6]. In this study, flow pattern in the ultrafiltration column was regarded as bubble flow since bubble diameter, which will be discussed in detail later, was significantly smaller than the channel size (13 mm). Gas flow rate was used in many literatures to feature injected air. However, the same sparger under different gas flow rates can generate bubbles of different sizes [24]. Providing actual size of bubble can eliminate this problem and give specific description of gas/liquid two-phase flow as well. Accumulated frequency of bubble size under gas flow rate of 12, 20, 40 and 60 mL/min was plotted in Fig. 5. It is obvious that as gas flow rate increased, bubble size increased. Smallest bubbles under

Fig. 3. Effect of alum dosage on turbidity and UV254 removal.

Fig. 4. Time varied lake water and coagulated water quality.

various gas flow rates had almost the same sizes of 0.6–0.8 mm while largest bubbles under gas flow rates of 12, 20, 40 and 60 mL/min had increasing sizes of 1.8, 2.1, 2.4 and 2.9 mm, which means a wider bubble size range under higher gas flow rate. Shear stress produced by air bubbles is relevant to their rise velocity. Big bubbles rise fast in water, and they can generate higher shear rate, which would greatly enhance local mixing in water. Moreover, bubbles with wider size range lead to strong fluctuation in shear stress, which can be a key factor in promoting membrane performance [15]. 3.5. Membrane performance 3.5.1. TMP development trend When gas flow rate was lower than 20 mL/min, TMP showed a rapid development trend as gas flow rate increased (data not shown). This might attribute to bubble enhanced mass transfer between hollow fibers and flocs. In stationary water, flocs can easily settle to the bottom of ultrafiltration column. With gas sparging, air bubbles carried flocs to the upper side of column. As analyzed in Section 3.4, turbulence was relatively weak under low gas flow rate, resulting in direct interaction of membranes and flocs, adsorption of flocs to hollow fibers was enhanced under such condition. Therefore, gas flow rate was increased to

Fig. 5. Cumulative frequency of bubble size under different gas flow rates.

130

J. Liu et al. / Desalination 333 (2014) 126–133

40 mL/min and 60 mL/min to promote turbulent fluctuation. As Fig. 6 shows, under high gas flow rate, TMP development showed a two stage trend, which was a fast development in the first few hours followed with a relatively slow development after about 4 h. Averaged TMP increase ratio in the first stage was 0.70 kPa/h, 0.90 kPa/h and 0.80 kPa/h respectively for 0, 40 and 60 mL/min gas flow rate. In the second stage, the parameters were 0.59 kPa/h, 0.57 kPa/h and 0.30 kPa/h respectively for 0, 40 and 60 mL/min gas flow rate, demonstrating improved membrane performance in long-term gas sparging. 3.5.2. UMFI Initial TMP under different gas flow rates varied slightly in Fig. 6. To eliminate impact of this difference, UMFI was calculated to provide comparable results of membrane fouling potential during the whole ultrafiltration circle. As Fig. 7 shows, when lake water was directly filtered, UMFI value was 0.00291 m2/L. After coagulation pretreatment, UMFI value decreased to 0.00216 m2/L, improved membrane performance was related to contaminant removal by coagulation. Highest UMFI value was 0.00274 m2/L under gas flow rate of 40 mL/min. Under 60 mL/min gas flow rate, UMFI value decreased to 0.00219 m2/L, which was still higher than directly ultrafiltration of coagulated water. Results derived from UMFI value indicate deterioration of hollow fiber performance in the hybrid process of coagulation–bubbling–UF compared with the single process of coagulation–UF. Thus, conclusion based on UMFI value is not consistent with that from TMP development trend. The mechanism of membrane fouling considering both floc property and bubble characteristic needs to be further explained. 3.5.3. Resistance analysis Analysis of resistance can give some in-depth sight to membrane fouling mechanism. Membrane resistance without bubbles and under gas flow rate of 40 and 60 mL/min was plotted in Fig. 8. When no gas bubbles were injected, Ra and Ri comprised only a small part of the total resistance, indicating weak interaction of flocs with hollow fibers, which was consistent with results derived from TMP development and UMFI value. However, contaminants that were not well removed by coagulation pretreatment accumulated in the column and led to high concentration polarization resistance without bubble enhanced hydraulic turbulence. Under 60 mL/min gas flow rate, Rp decreased to less than a quarter of the value without bubbles, confirming bubbles' effectiveness in alleviating concentration polarization. Both Ra and Ri increased obviously when bubbles were introduced, due to bubble enhanced interaction between hollow fiber and flocs. The change of Ra and Ri can also be related to NOM fraction in the source water and membrane hydrophilicity. SUVA of source water in this study ranged from 1.30 to 1.82 mg/L·m. A SUVA value lower than

3 mg/L·m usually indicates that NOMs in water are mainly hydrophilic material [25]. Lee et al. studied fouling of UF/MF membrane with different hydrophobicity/hydrophilicity by various source water, and their results showed that with hydrophilic UF membrane, fouling potential increased as SUVA decreased [26]. Bessiere et al. found that hydrophilic fraction caused significant fouling of membrane and they supposed that fouling phenomena were more attributed to internal blocking or adsorption based on their observation [27]. These results were quite similar to what we have observed in this study. When gas flow rate increased, as would be analyzed in Section 3.6.1, floc breakage became severe. Thus, more NOMs would be released from flocs into water. Bubble enhanced mass transfer led to adsorption of hydrophilic NOMs to membrane surface or internal pores. Therefore, both Ra and Ri increased. 3.6. Floc properties 3.6.1. Longest side value of flocs Bubbles in the ultrafiltration column may have dual effect on floc breakage. Firstly, air bubbles of large diameter rise fast in the column and when they strike on flocs, weak bonding of flocs can be destroyed under direct attack, leading to cleavage of flocs into smaller pieces, which was called as large-scale fragmentation. Secondly, bubble introduced shear stress may act tangentially to the floc surface, leading to release of primary colloids, which was another floc breakage mode called surface erosion. Graphic illustration of these two modes was presented in Jarvis and co-workers' review [28]. In this study, when no air bubbles were introduced into the ultrafiltration column, flocs were considered under no shear rate and no breakage happened. As shown in Fig. 9, peak value of floc size under such circumstance was about 440 μm, and there were big flocs of size exceeding 1000 μm. Applying 60 mL/min and 40 mL/min gas flow rate led to decreased value of peak floc size and there were no flocs bigger than 800 μm, suggesting floc breakage dominated by large-scale fragmentation. Howe et al. investigated the effect of various size fractions in the coagulated water on the fouling of UF process [29]. They found that the colloid of 1 μm to 100 kDa is the main fouling source to UF membrane. Due to limited resolution of floc image, flocs of this size range were not measurable in this study. But according to resistance analysis, Ri which is caused by small colloids plugging in membrane pores increased a little, indicating that surface erosion is happening. 3.6.2. Fractal dimension of flocs Barbot et al. concluded that large flocs of high resistance allowing water flow between the structures are favorable in the coagulation–ultrafiltration process [10], which is the case in this study when no air bubbles were injected in the ultrafiltration column. They also declared that flocs of high fractal dimension and small size were not favorable as they would form tight cake layer on the membrane surface, making the membrane less permeable. As discussed in Section 3.4, increased gas flow rate led to increased shear rate. As a result, floc size decreased (analyzed in Section 3.6.1) while floc fractal dimension increased (Fig. 10), leading to deteriorated membrane performance. The same result was also presented in other literature [30] and can be reflected in resistance analysis in Section 3.5.3. 3.7. Membrane performance considering both bubbles and floc properties

Fig. 6. Comparison of TMP development under gas flow rates of 0, 40 and 60 mL/min.

From viewpoint of both air bubbles and floc properties, it seems that membrane performance can be well explained. When no air bubbles were injected, the single process of coagulation–ultrafiltration showed lowest UMFI value (Fig. 11). This was attributed to limited interaction of hollow fibers and flocs, due to the fact that flocs could easily settle to the bottom of the ultrafiltration column. When air bubbles were applied, bubbles acted on both hollow fibers and flocs. For hollow fibers, bubbles induced shear rate can limit concentration polarization and vibrate hollow fibers, which promoted membrane performance and can

J. Liu et al. / Desalination 333 (2014) 126–133

131

Fig. 7. UMFI (unit: m2/L) of hollow fibers in the ultrafiltration of (a) lake water, (b) coagulated water, (c) coagulated water under 40 mL/min gas flow rate and (d) coagulated water under 60 mL/min gas flow rate.

be labeled as positive effect of bubbles. For flocs, bubbles induced shear rate led to floc breakage and restructuring. Floc size reduced and primary colloids were released from the aggregates. Bubble enhanced mass transfer resulted in strong interaction between hollow fibers and small flocs or primary colloids, which increased membrane resistance and UMFI value. This is labeled as negative effect of bubbles. The reason why UMFI value under 60 mL/min gas flow rate was still lower than that under 40 mL/min means positive effect of bubbles dominated. However, all the conclusions drawn above were based on the whole experiment circle. Considering TMP development trend, there are encouraging results. As discussed in Section 3.5.1, two stages TMP development trend indicated that when air bubbles were injected, membrane fouling was accelerated in the initial stage of ultrafiltration and slowed down in the second stage. Meanwhile, TMP development

Fig. 8. Membrane resistance with various gas flow rates.

Fig. 9. Histogram of floc size under gas flow rates of 60 mL/min, 40 mL/min and no gas bubbles.

132

J. Liu et al. / Desalination 333 (2014) 126–133

4. The single process of coagulation–UF showed the best performance when considering the whole experiment circle. We attribute this to floc settlement in the bottom of column. However, with bubbling, TMP slowed down after equilibrium of colloid blockage, proving effectiveness of bubbling in the long-term operation. Thus, if the apparatus could be improved to realize separation of flocs from the column by bubbles, the hybrid process would take advantage of both coagulation and bubbling. Acknowledgments The work was supported by the Litree Purifying Technology Co., Ltd., and the National Natural Science Foundation of China under Award number 51278317. Fig. 10. Cross-sectional fractal dimension of flocs.

trend straightly increased regardless of operating time for coagulation– ultrafiltration process. Considering cake formation and polarization resistance would finally reach a certain extent and no air bubbles to eliminate these effects in the process of coagulation–ultrafiltration. The hybrid process of coagulation–bubbling–ultrafiltration is still promising and more works are needed to be done to explain the mechanism in depth. 4. Conclusions Gas bubbles were introduced to enhance hollow fiber ultrafiltration process after coagulation pretreatment. Interaction of bubbles, flocs and membrane was investigated. The main findings of this research were as follows: 1. Bubble size becomes bigger, bubble distribution range becomes wider as gas flow rate increases, resulting in higher shear rate in the ultrafiltration column; 2. As gas flow rate increases, floc breakage becomes severe, leading to the release of small colloids and NOMs from flocs, which could accelerate membrane fouling; 3. UMFI value under 40 mL/min gas flow rate is the highest due to floc breakage and bubble enhanced interaction between membrane and flocs. Under 60 mL/min gas flow rate, UMFI value decreased, this was attributed to limited concentration polarization caused by bubbles.

Fig. 11. Influence of gas flow rate on floc properties and UMFI.

References [1] J. Tian, Y. Xu, Z. Chen, J. Nan, G. Li, Air bubbling for alleviating membrane fouling of immersed hollow-fiber membrane for ultrafiltration of river water, Desalination 260 (2010) 225–230. [2] Y. Zhang, C.Y. Tang, G. Li, The role of hydrodynamic conditions and pH on algal-rich water fouling of ultrafiltration, Water Res. 46 (2012) 4783–4789. [3] K.J. Howe, A. Marwah, K.-P. Chiu, S.S. Adham, Effect of membrane configuration on bench-scale MF and UF fouling experiments, Water Res. 41 (2007) 3842–3849. [4] X. Zhu, M. Elimelech, Colloidal fouling of reverse osmosis membranes: measurements and fouling mechanisms, Environ. Sci. Technol. 31 (1997) 3654–3662. [5] H. Huang, K. Schwab, J.G. Jacangelo, Pretreatment for low pressure membranes in water treatment: a review, Environ. Sci. Technol. 43 (2009) 3011–3019. [6] Z.F. Cui, S. Chang, A.G. Fane, The use of gas bubbling to enhance membrane processes, J. Membr. Sci. 221 (2003) 1–35. [7] K.J. Howe, M.M. Clark, Effect of coagulation pretreatment on membrane filtration performance, J. Am. Water Works Assoc. 98 (2006) 133–146. [8] T. Liu, Z. Chen, W. Yu, J. Shen, J. Gregory, Effect of two-stage coagulant addition on coagulation–ultrafiltration process for treatment of humic-rich water, Water Res. 45 (2011) 4260–4268. [9] W. Yu, T. Liu, J. Gregory, L. Campos, G. Li, J. Qu, Influence of flocs breakage process on submerged ultrafiltration membrane fouling, J. Membr. Sci. 385–386 (2011) 194–199. [10] E. Barbot, S. Moustier, J.Y. Bottero, P. Moulin, Coagulation and ultrafiltration: understanding of the key parameters of the hybrid process, J. Membr. Sci. 325 (2008) 520–527. [11] Q.Y. Li, Z.F. Cui, D.S. Pepper, Effect of bubble size and frequency on the permeate flux of gas sparged ultrafiltration with tubular membranes, Chem. Eng. J. 67 (1997) 71–75. [12] Q.Y. Li, R. Ghosh, S.R. Bellara, Z.F. Cui, D.S. Pepper, Enhancement of ultrafiltration by gas sparging with flat sheet membrane modules, Sep. Purif. Technol. 14 (1998) 79–83. [13] S.R. Bellara, Z.F. Cui, D.S. Pepper, Gas sparging to enhance permeate flux in ultrafiltration using hollow fibre membranes, J. Membr. Sci. 121 (1996) 175–184. [14] Y. Lu, Z. Ding, L. Liu, Z. Wang, R. Ma, The influence of bubble characteristics on the performance of submerged hollow fiber membrane module used in microfiltration, Sep. Purif. Technol. 61 (2008) 89–95. [15] A.P.S. Yeo, A.W.K. Law, A.G. Fane, The relationship between performance of submerged hollow fibers and bubble-induced phenomena examined by particle image velocimetry, J. Membr. Sci. 304 (2007) 125–137. [16] A. Vahedi, B. Gorczyca, Application of fractal dimensions to study the structure of flocs formed in lime softening process, Water Res. 45 (2011) 545–556. [17] R.K. Chakraborti, K.H. Gardner, J.F. Atkinson, J.E. Van Benschoten, Changes in fractal dimension during aggregation, Water Res. 37 (2003) 873–883. [18] R.K. Chakraborti, J.F. Atkinson, J.E. Van Benschoten, Characterization of alum floc by image analysis, Environ. Sci. Technol. 34 (2000) 3969–3976. [19] G.C. Bushell, Y.D. Yan, D. Woodfield, J. Raper, R. Amal, On techniques for the measurement of the mass fractal dimension of aggregates, Adv. Colloid Interf. Sci. 95 (2002) 1–50. [20] B. Liu, C. Chen, W. Zhang, J. Crittenden, Y. Chen, Low-cost antifouling PVC ultrafiltration membrane fabrication with Pluronic F 127: effect of additives on properties and performance, Desalination 307 (2012) 26–33. [21] H. Huang, T. Young, J.G. Jacangelo, Novel approach for the analysis of bench-scale, low pressure membrane fouling in water treatment, J. Membr. Sci. 334 (2009) 1–8. [22] H. Huang, T.A. Young, J.G. Jacangelo, Unified membrane fouling index for low pressure membrane filtration of natural waters: principles and methodology, Environ. Sci. Technol. 42 (2007) 714–720. [23] A.F. Derradji, S. Taha, G. Dorange, Application of the resistances in series model in ultrafiltration, Desalination 184 (2005) 377–384. [24] S. Khirani, P. Kunwapanitchakul, F. Augier, C. Guigui, P. Guiraud, G. Hébrard, Microbubble generation through porous membrane under aqueous or organic liquid shear flow, Ind. Eng. Chem. Res. 51 (2011) 1997–2009. [25] A. Matilainen, E.T. Gjessing, T. Lahtinen, L. Hed, A. Bhatnagar, M. Sillanpää, An overview of the methods used in the characterisation of natural organic matter (NOM) in relation to drinking water treatment, Chemosphere 83 (2011) 1431–1442. [26] N. Lee, G. Amy, J.-P. Croué, H. Buisson, Identification and understanding of fouling in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM), Water Res. 38 (2004) 4511–4523.

J. Liu et al. / Desalination 333 (2014) 126–133 [27] Y. Bessiere, B. Jefferson, E. Goslan, P. Bacchin, Effect of hydrophilic/hydrophobic fractions of natural organic matter on irreversible fouling of membranes, Desalination 249 (2009) 182–187. [28] P. Jarvis, B. Jefferson, J. Gregory, S.A. Parsons, A review of floc strength and breakage, Water Res. 39 (2005) 3121–3137.

133

[29] K.J. Howe, A. Marwah, K.-P. Chiu, S.S. Adham, Effect of coagulation on the size of MF and UF membrane foulants, Environ. Sci. Technol. 40 (2006) 7908–7913. [30] W. Xu, B. Gao, Effect of shear conditions on floc properties and membrane fouling in coagulation/ultrafiltration hybrid process—the significance of Alb species, J. Membr. Sci. 415–416 (2012) 153–160.