Ultrasonic visualization of sub-critical flux fouling in the double-end submerged hollow fiber membrane module

Journal of Membrane Science 444 (2013) 394–401

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Ultrasonic visualization of sub-critical flux fouling in the double-end submerged hollow fiber membrane module Xianhui Li a, Jianxin Li a,n, Jie Wang b, Hong Wang a, Benqiao He a, Hongwei Zhang b a State Key Laboratory of Hollow Fiber Membrane Materials and Processes, School of Materials Science and Engineering, Tianjin Polytechnic University, Tianjin 300387, PR China b School of Environmental and Chemical Engineering, Tianjin Polytechnic University, Tianjin 300387, PR China

art ic l e i nf o

a b s t r a c t

Article history: Received 24 February 2013 Received in revised form 23 May 2013 Accepted 24 May 2013

Ultrasonic time-domain reflectometry (UTDR) was extensively employed to visualize and analyze quantitatively the effect of aeration rate, operating flux and fiber length on sub-critical flux fouling profiles in a double-end submerged hollow fiber membrane (SHFM) module. Five 10 MHz transducers were externally mounted in contact with the outside surface of the double-end SHFM module. A single polyethersulfone hollow fiber membrane was used to filter 5 g/L yeast suspension. Results showed that the double-end SHFM module has better filtration performance by comparison with one-end SHFM module. The acoustic measurements revealed that the membrane near the upper suction end of the double-end SHFM module was more easily suffered from fouling than that near the lower suction end under sub-critical flux operation. Furthermore, the progress of foulant deposition gradually migrated from both ends to middle and reached the plateau finally. Moreover, a low operating flux was more effective to reduce membrane fouling. And a short double-end SHFM module was more easily subjected to membrane fouling than a long one at the same operating conditions. & 2013 Elsevier B.V. All rights reserved.

Keywords: Double-end submerged hollow fiber membrane module Sub-critical flux Membrane fouling Ultrasonic time-domain reflectometry (UTDR)

1. Introduction The submerged hollow fiber membrane (SHFM) module is widely used in wastewater treatment processes. It has become the major choice in membrane bioreactor (MBR) due to high membrane-surface-area-to-footprint ratio and low energy cost [1,2]. For the SHFM module, the effluent can be permeated from either one end or two ends of the fibers. However, by comparison with the double-end SHFM module, the one-end SHFM module is easily subjected to membrane fouling due to the more nonuniform distribution of the local transmembrane pressure (TMP) along the fiber [3]. Nowadays the double-end SHFM modules have been extensively used in most of membrane-based water treatment plants [4]. Nevertheless, the membrane fouling remains the most crucial problem in the large-scale applications of the doubleend SHFM systems. In order to overcome this fouling problem, the double-end SHFM system is generally operated under sub-critical flux conditions to maintain a sustainable permeability. In fact, despite the operation under the sub-critical flux conditions, a perceptible change occurs in membrane permeability as a result of membrane

n

Corresponding author. Tel.: +86 22 83955798; fax: +86 22 8395 5055. E-mail addresses: [email protected], [email protected], [email protected] (J. Li). 0376-7388/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2013.05.052

fouling and clogging [5,6]. Consequently, a number of studies have been conducted to investigate the characteristics of sub-critical flux fouling in SHFM system in recent years. It is generally characterized by two-step fouling phenomenon. That is to say, an initial slow and gradual TMP increase is followed by a sudden transition to a rapid TMP rise. This behavior has been explained with different perspectives based on local critical flux concept [7,8], critical suction pressure [9] and in-homogeneity of fiber bundles [10] etc. Many researchers further studied the effect of different fractions of sludge (including soluble substances, colloids and suspended solids, etc.) on membrane fouling [11,12]. Although those intensive efforts are very helpful to understand the characteristics of sub-critical flux fouling in SHFM system, there is lack of sufficient information on local sub-critical flux fouling behavior, especially for the double-end SHFM module with the more complicated local hydrodynamic. Therefore, the development of non-invasive technique to visualize and analyze the local subcritical flux fouling behavior and hydrodynamics of double-end SHFM module is of great importance. Recently, a number of measurement techniques have been extensively employed to investigate membrane fouling behavior and hydrodynamics. Marselina et al. [13] employed the direct observation (DO) technique to quantify the effects of crossflow velocities and backwashing on particle deposition and removal mechanisms from the hollow fiber membrane under constant flux conditions. The results showed that the fouling layer was removed

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

with a combination of cake expansion and gradual erosion mechanisms at low crossflow velocities and direct erosion of cake height at higher crossflow velocities during the backwashing operation. In addition, Sun et al. [14] developed confocal laser scanning microscopy (CLSM) and image analysis to characterize and quantify biofouling of SHFM in drinking water treatment system. The results demonstrated that the protein was a dominating component of the membrane biofouling at low flux. And the carbohydrate present in the biofouling had a more important effect on the membrane permeability. Although the CLSM is a valuable tool for obtaining high resolution images and 3D reconstructions, the method relies on membrane autopsy. Buetehorn et al. [15] established the nuclear magnetic resonance (NMR) to non-invasively visualize the local cake growth on the surface of SHFM for a microfiltration of silica suspensions. The results showed that the cake-growth rate increased with an increase in the permeate flux or the solids concentration. And the cake-removal efficiency was enhanced with an increase in aeration pressure or duration of aeration. In addition, ultrasonic time-domain reflectometry (UTDR) as an in situ, noninvasive and real-time technique has been successfully developed to measure membrane fouling and hydrodynamic behavior in a hollow fiber membrane module. For example, Xu et al. [16] developed the extension of UTDR for the real-time measurement of particle deposition in a single hollow fiber membrane. The results presented that UTDR technique could distinguish and recognize the acoustic response signals from various curved surfaces of the housing holder and the hollow fiber within a single hollow fiber membrane module. Furthermore, the UTDR technique with wavelet signal analysis was employed to investigate the oil deposition profile along the hollow fiber membrane and the diffusion behavior of oil droplets [17]. The results showed that the fouling became mitigatory from the inlet to the outlet of the membrane module along the shell-side flow direction during the crossflow filtration of oil wastewater. Moreover, after shutting down the microfiltration system, the relaxation of fouling layer and diffusion of oil droplets into the bulk solution were also observed by UTDR. In a recent study, the UTDR was extensively employed to in situ measure the fouling profile along the one-end SHFM module under sub-critical flux operation [18]. The results indicated that the fouling still deposited at the suction end of the one-end SHFM module and progressively expanded down along the fiber under the operation of the sub-critical flux. However, this previous study focused on the local sub-critical fouling behavior in the dead-end SHFM module rather than double-end SHFM module. In particular, the filtration performance of double-end SHFM module could not be simply recognized as the superposition of two one-end SHFM modules due to the complicated hydrodynamics. The aim of the present study is to extensively use UTDR technique coupled with the signal processing technique of differential signal to investigate the fouling profile and hydrodynamic behavior of the double-end SHFM module under sub-critical flux operation. Five 10 MHz transducers were externally mounted in contact with the outside surface of the double-end SHFM module. A single polyethersulfone hollow fiber membrane was used to filter 5 g/L yeast suspension. The effects of aeration rate, fiber length and operational flux on fouling profile were explored at the constant flux filtration mode. Scanning electron microscopy (SEM) as an independent method was used to corroborate the UTDR measurements.

395

Peristaltic pump P Hollow-fiber membrane 50mm

TD1 200mm

Feed pump

Pulser/Receiver

TD2 200mm

Feed tank

TD3 200mm

Oscilloscope

TD4 200mm

Rotameter

Aerator

TD5 50mm

Test module

Computer

Fig. 1. Schematic of experimental setup: the submerged microfiltration system and the ultrasonic measurement system.

of a 5 L feed tank for storage and supply of the feed solution, peristaltic pump (BT 100-2J, Longer, China), feed pump (TP 10-20, Motimo, China), air pump (ACO-818, Yuting, China), gas rotameter (LZB-3, Huanming, China), tubular test module with 40 mm inside diameter and 1000 mm length and precision vacuum gauge with a pressure range from −100 70.25 to 0 kPa. The permeate from both ends was extracted continuously at a constant flux using a peristaltic pump and recycled back to the feed tank. The level of liquid in the tubular test module was maintained constant by a feed pump. The TMP increase was monitored using a precision vacuum gauge installed on the permeate stream. Air was introduced from the air pump through 1 mm nozzle that generated bubble flow from the bottom of the tubular test module. The aeration was monitored by a gas rotameter and controlled by a needle valve. The ultrasonic measurement system consisted of five 10 MHz ultrasonic transducers (Panametrics V111), a pulser–receiver (Panametrics 5058PR) and a 350 MHz digital oscilloscope (Agilent 54641A) with sweep speeds from 50 s per division to 1 ns per division and 2 mV per division sensitivity as shown in Fig. 1. The oscilloscope connected to the pulser–receiver captured and displayed the data signal as amplitude changes on its front panel. Each set of ultrasonic data generated included 5000 data points which could be stored on a computer's hard drive. The ultrasonic data could be further analyzed by MS Excel. Five transducers (TD1– TD5) were externally mounted in contact with the outside surface of the vertical tubular test module using castor oil as the acoustic couplant. The transducer TD1 was fixed on the part near the upper suction end of the membrane module, and the distance to the surface of circulating fluid was 50 mm. The transducers TD2-4 were located on the middle part of the membrane module. The transducer TD5 was located on the part near the lower suction end of the membrane module, and the distance to the suction end of the fiber was 50 mm. The distance among the five transducers was 200 mm. 2.2. The double-end submerged hollow-fiber membrane module

2. Material and methods 2.1. Submerged microfiltration and UTDR measurement systems Fig. 1 shows the schematic diagram of the submerged microfiltration and UTDR measurement system. The assembly consisted

Three double-end SHFM modules used were made in our laboratory. Each module contained only one hollow fiber membrane. Polyethersulfone (PES) hollow fiber membranes (Tianjin Motimo Membrane Technology Co. Ltd., China) with a nominal pore size of 0.1 μm, inside and outside diameter of 1.0 and 1.6 mm

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

2.3. Experimental procedure of membrane fouling monitoring Before the experiments of membrane fouling monitoring, air bubbles inside the lumen of the membrane after casting were removed thorough the following steps: (i) The hollow fiber membrane was soaked in 30 wt% glycerol and water solution for at least 24 h, in order to increase the hydrophility and remove the air bubbles from the membranes; (ii) The deionized water was gently injected into the fiber lumen from either open-end using a 10 mL syringe; and (iii) The wetted fiber was then carefully loaded into the membrane module. In each run of the fouling monitoring experiments, a new membrane was employed and first conducted deionized water filtration to compress the membrane for 60 min so as to build up a steady flow field. Once the steady state with respect to the permeate flux and TMP was attained, the feed was switched to the yeast suspension to initiate the fouling phase. The fouling experiments were carried out with 5 g/L yeast suspension at the temperature of 20 72 1C in order to generate serious membrane fouling during the submerged microfiltration. The yeast suspension was prepared by mixing the unwashed yeast (mean diameter 4.5 μm) with deionized water for half an hour using a mechanical stirrer. The TMP and ultrasonic signals were collected at regular interval (every 15 min) during the fouling experiments. During the collection of ultrasonic signals, the aeration was stopped to eliminate the scattering and attenuation of the ultrasonic waves caused by air bubbles. And the peristaltic pump was also stopped to avoid particles deposition onto membrane surface due to the drag force of filtration without aeration. The total operating time for the fouling monitoring experiments was 6 h. All of the fouling monitoring experiments were conducted for three times to confirm the reproducibility of the experimental data. After the fouling experiment, the fouled hollow fiber membrane was taken out of the test module carefully. In order to reveal the different morphologies along the double-end SHFM, the fouled fiber was cut into five small segments from the locations corresponding to the detection of the five ultrasonic transducers (TD1TD5) and subjected for the morphological characterization by SEM (QUANTA200, FEI). 2.4. Ultrasonic spectrum analysis The principle of ultrasonic measurement for membrane fouling is based on the propagation of mechanical waves. Once fouling initiated on the membrane surface, the acoustic impedance difference at the feed solution/membrane interface would change, leading to a change in the magnitude of the peak [19]. In this study, owing to the superimposition of the ultrasonic waveforms from the single hollow fiber membrane and the formation of very thin fouling layer under the sub-critical flux operation, a differential signal analysis technique would be used to improve the measurement distinguishability so as to investigate the build-up of the fouling layer and its physical state on the membrane surface. A differential signal as one of the approaches for overlapping signal separation is defined as the difference between reference waveform and test waveform [20]. The waveforms obtained at the pure water state were considered as the references, and a fouled membrane waveform was used as the test waveform.

The differential signal represents an echo signal of the fouling layer on the membrane surface. In our earlier work [18], the differential signal analysis technique was successfully employed to investigate yeasts deposition and profile on a single SHFM. In addition, in order to benefit signal analysis, the ultrasonic measurements were only focused on the peak generated from the interface between the feed solution (water) and front outside surface of the hollow fiber membrane.

3. Results and discussion 3.1. Critical flux determination In the fouling experiment, a double-end SHFM filtration module with a single fiber length of 1000 mm was used for the filtration of 5 g/L yeast suspension with aeration rate of 15 mL/min. A shortterm critical flux has been obtained by flux-step method with constant flux mode as shown in Fig. 2. It can be seen from Fig. 2 that the TMP keeps a constant value of 1.8, 2.7, 3.6, 4.5 kPa during the operation of 30 min at the imposed flux of 6, 10, 14, 18 L/(m2 h), respectively. When the imposed flux reached up to 20 L/(m2 h), the TMP increased rapidly from 4.9 to 5.7 kPa. Consequently, the short-term critical flux in this experiment was determined as 18 L/(m2 h). In addition, an interesting phenomenon was observed that the critical flux value of double-end SHFM module (18 L/(m2 h)) obtained was obviously more than that of one-end SHFM module (14 L/(m2 h)) obtained in our previous study [18] under the same operating conditions. Furthermore, the TMP value obtained in double-end SHFM module was 3.6 kPa, which was lower than that (4.0 kPa) obtained in one-end SHFM module at the same imposed constant flux of 14 L/(m2 h) [18]. This is because the axial profile of local TMP in the double-end SHFM module is more uniform than that in the one-end SHFM module, leading to more symmetrical fouling or dynamic distribution along the fiber [3,21]. This is the main reason why the double-end fiber bundle is becoming the most popular module packing configuration [22]. On the basis of the observations, the critical flux value as a reference would be used to investigate fouling behavior under different operating conditions in the following study. 3.2. Effect of sub-critical operating flux on TMP As an important parameter, TMP was employed to monitor membrane fouling and evaluate membrane permeability in a submerged MBR system. Thus, the effect of operating flux on TMP was carried out at five different sub-critical fluxes of 6, 9, 12, 14 and 16 L/(m2 h) as shown in Fig. 3. Similarly, the fouling operation was conducted at aeration rate of 15 mL/min with 5 g/L yeast suspension and the fouling time was 360 min. It can 6.0

35

TMP Flux

5.0

30 25

4.0

20 3.0

15

2.0 1.0

Flux (L/m2 h)

were used for the experimental study. The inside diameter and length of the tubular test module was 40 and 1000 mm, respectively. One module with length of 1000 mm had 0.005 m2 of membrane area. And the other two membrane modules with the length of 600 and 800 mm were employed to investigate the effect of fiber length on membrane fouling behavior.

TMP (kPa)

396

10 0

30

60

90

120

5 150

Time (min) Fig. 2. Critical flux measurement by the flux-step method during the fouling experiment carried out with 5 g/L yeast suspension.

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

could lead to the enhanced transport of particles to the membrane surface and the formation of fouling layer, finally resulting in the increase of TMP [23]. Although the observations on overall transmembrane pressure could provide some available information about the changes of membrane performance with the progress of fouling operation time, the TMP history could not serve as a sensitive indicator of the complex variations in membrane fouling distribution under any operation conditions, especially under the sub-critical flux condition. Therefore, the UTDR technique was used to further detect the fouling behavior in the double-end SHFM module under the subcritical flux conditions in the following studies.

be found in Fig. 3 that no obvious variations of TMP were observed during the fouling experiment at the sub-critical flux of 6 and 9 L/(m2 h). However, the TMP slightly increased from 3.9 to 4.0 kPa at the fouling time of 240 min during the fouling experiment at the sub-critical flux of 12 L/(m2 h) (Fig. 3). Upon further increasing the imposed sub-critical flux to 14 L/(m2 h), the TMP kept constant at 3.6 kPa at the initial stage, then started gradually to increase to 4.1 kPa with the fouling time from 90 to 360 min. In a similar way, the TMP gradually increased from 4.0 to 4.7 kPa with the fouling time from 60 to 360 min at the imposed sub-critical flux of 16 L/ (m2 h). These results implied that the membrane fouling still appeared even if the operating was carried out at the normal sub-critical flux in the double-end SHFM module. In short-term test, the initial operating flux has a significant impact on the fouling behavior. This is because that the operation at a high flux 6.0

6L/m2 h 12L/m2 h 16L/m2 h

TMP (kPa)

5.0

3.3. UTDR measurements In order to further investigate the local sub-critical flux fouling behavior, the differential signal based ultrasonic measurement was employed to visualize the fouling profile along the fiber in real time. The differential signals (red line) obtained from the fouled membrane by five ultrasonic transducers (TD1–TD5) after 360 min of fouling operation under the sub-critical flux of 12 L/(m2 h) with the aeration rate of 15 mL/min were illustrated in Fig. 4. Once the yeast particles deposited on the membrane surface, the acoustic impedance difference at the bulk solution/membrane interface would change, leading to the appearance and growth of the fouling echo. The differential signals with the amplitudes of 0.38, 0.15, 0, 0.09 and 0.28 V (V) obtained from TD1 to TD5 after 360 min of fouling operation were observed in Fig. 4. The difference between the amplitudes of the differential signals is

9L/m2 h 14L/m2 h

4.0 3.0 2.0 1.0

0

60

120 180 240 Operation time (min)

300

360

Fig. 3. TMP profiles versus fouling operation time at different sub-critical fluxes.

4

4 3 2

Amplitude(V)

Amplitude(V)

3 2

Fouling

1 0 -1 -2

1 0 -1 -2

-3 -4 2.90E-05

-3 2.92E-05

2.94E-05

2.96E-05

-4 2.90E-05

2.98E-05

2.92E-05

4

3

3

2

2

Amplitude(V)

4

1 0 -1 -2

2.96E-05

2.98E-05

1 0 -1 -2

-3 -4 2.90E-05

2.94E-05

Arrival time(s)

Arrival time(s)

Amplitude(V)

397

-3 2.92E-05

2.94E-05

2.96E-05

2.98E-05

-4 2.90E-05

2.92E-05

2.94E-05

2.96E-05

2.98E-05

Arrival time(s)

Arrival time(s) 4

Amplitude(V)

3 2 1 0 -1 -2 -3 -4 2.90E-05

2.92E-05

2.94E-05

2.96E-05

2.98E-05

Arrival time(s) Fig. 4. Ultrasonic signal responses of the clean hollow fiber membrane (blue line) and differential signals (red line) detected by (a) TD1, (b) TD2, (c) TD3, (d) TD4 and (e) TD5 after 360 min of fouling experiment. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

398

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

attributed to the deposition and growth of the fouling layer [20]. Obviously, no peak of fouling echo could be seen at the middle part of the fiber. However, the observed differential signals implied that the foulant still deposited on the membrane surface of the other parts of the fiber under the sub-critical flux conditions. And more foulant deposited on the parts near both of the suction ends of the fiber. Furthermore, an interesting phenomenon was observed that the amplitude of the differential signal obtained from TD1 (0.38 V) was higher than that obtained from TD5 (0.28 V) (Fig. 4a and e). That is to say, the membrane near the upper suction end of the double-end SHFM module was more easily suffered from fouling than that near the lower suction end under sub-critical flux operation. This phenomenon was ascribed to two reasons. First, the flow of the permeate from the lumen near the upper suction end of the module was suffered from less filtration resistance and fluid resistance or gravity than that near the lower suction end, leading to a higher local TMP near the upper suction end [24]. Second, the shear force generated by the aeration was larger at the part near the lower suction end of the fiber than that near the upper suction end. The shear force would help alleviate the foulant deposition on the part near the lower suction end of the fiber [25]. In addition, to investigate the local flux variation along fiber, the local flux was calculated by Eq. (1) based on the model of local flux distribution proposed by Chang et al. [3]: JðzÞ ¼ λLJ e

eλx þ e−λx ; eλL −e−λL

λ¼

sffiffiffiffiffiffiffiffiffiffiffi 16 r 3i Rm

ð1Þ

where L is the hollow fiber length (m), Je is the average operating flux (L/(m2 h)), ri is the inner radius of the hollow fiber (0.001 m) and Rm is the intrinsic membrane resistance (1.7  1011 m−1). It should be noted that Eq. (1) could only be used in the calculation of local fluxes along the half of fiber in the case of the double-end fibers.

According to Eq. (1), the calculated local flux at the upper suction end of the fiber was 21.4 L/(m2 h), which was higher than the average critical flux of 18 L/(m2 h) when the fiber was operated at the average operating flux of 12 L/(m2 h). It implied that the foulants still deposited near the suction ends of the double-end SHFM (Fig. 4). In order to corroborate the acoustic observations, SEM images were taken from the five specimens corresponding to the parts of the fouled membrane detected by the five transducers after 360 min of fouling operation at the constant flux of 12 L/(m2 h) as illustrated in Fig. 5. These SEM images clearly presented that the foulant deposited on the membrane surface unevenly. Obviously, no deposited foulant can be seen on the middle part of the fouled membrane corresponding to the area detected by TD3 (Fig. 5c). Only a few particles aggregated on the membrane surface corresponding to the area detected by TD2 and TD4 (Fig. 5b and d). The parts near both of the suction ends of the fiber were covered with a dense fouling layer (Fig. 5a and e). Furthermore, the coverage of the yeast particles on the membrane surface detected by TD1 was denser than that detected by TD5. Additionally, in order to get the overall growth and development of the fouling layer at different positions of the double-end SHFM module under the condition of sub-critical flux, the amplitudes of the differential signals as a function of operation time during the fouling experiment carried out with constant flux of 12 L/(m2 h) and aeration rate of 15 mL/min were summarized in Fig. 6. It can be seen from Fig. 6 that there was no any growth in the amplitudes of differential signals in the initial 60 min of fouling operation (Fig. 6). It suggested that the foulant deposition on the membrane surface was so slight that no change of the amplitude could be observed. The amplitudes of differential signals obtained by TD1 and TD5 began to increase rapidly after 60 min of fouling operation, and then reached a plateau with the maximum value of 0.38 and 0.28 V after 210 and 180 min, respectively. The growth of amplitude is attributed to the deposition and

Fig. 5. SEM images of the outside surfaces of the fouled hollow fiber membranes obtained from the corresponding areas detected by (a) TD1, (b) TD2, (c) TD3, (d) TD4 and (e) TD5 after 360 min of fouling experiment.

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

growth of the fouling layer [20]. Afterwards the plateau obtained is mainly ascribed to a force balance between the shear force caused by aeration and the adhesive forces generated from local TMP [26]. A similar trend was observed in the amplitudes of differential signals obtained by TD2 and TD4. Nevertheless, the initial growth time of the amplitude postponed to 120 min of operation time. The maximum value of the amplitude obtained by TD2 and TD4 was 0.09 and 0.15 V after the operation time of 360 min. It suggested that the fouling deposition migrated progressively from both the suction ends of the fiber to middle part owning to the self-adjustment of local flux [7]. Furthermore, the amplitude of differential signal obtained from TD3 still kept at 0 V during the

Amplitude (V)

0.4

TD1 TD2 TD3 TD4 TD5

0.3 0.2 0.1 0.0

0

60

120

180

240

300

360

Operation time (min) Fig. 6. Amplitudes of differential signals versus operation time during the fouling experiment at the constant flux of 12 L/(m2 h) and aeration rate of 15 mL/min.

3.4. Effect of aeration rate on membrane fouling Gas sparging has commonly been employed to induce shear forces at the membrane surface so as to remove the accumulated foulants in submerged membrane systems [4]. However, a high aeration rate would increase the operating cost in the full-scale submerged hollow fiber membrane plant [22]. Herein, the different aeration rates were conducted in the fouling experiments in order to assess the effect of the aeration rate on the membrane fouling. Fig. 7 shows the amplitudes of differential signals obtained from the five transducers versus operation time during the fouling experiment at the constant flux of 12 L/(m2 h) with the different aeration rates of 15, 25, 35 and 45 mL/min, respectively. It can be seen from Fig. 7 that the amplitudes of differential signals decreased gradually with the increase of aeration rate owing to the increase of turbulent shear. For example, the steadystate amplitude of differential signal obtained from TD1 decreased from 0.38 to 0.28, 0.21, 0.15 V when the aeration rate increased from 15 to 25, 35 and 45 mL/min, respectively (Fig. 7a). Similarly, the steady-state amplitude of differential signal obtained from TD5 decreased from 0.28 to 0.2, 0.15, 0.12 V (Fig. 7e). Nevertheless, the peak of differential signal obtained from TD2 disappeared with the increase of aeration rate from 15 to 45 mL/min (Fig. 7b). And the amplitude of differential signal obtained from TD3 still remained

0.4

15ml/min 25ml/min 35ml/min 45ml/min

0.3

whole fouling operation. This performance is related to the low local TMP at the middle part of the double-end SHFM [3].

Amplitude (V)

Amplitude (V)

0.4

0.2 0.1 0.0

15ml/min 25ml/min 35ml/min 45ml/min

0.3 0.2 0.1

0

60

120

180

240

300

0.0

360

0

60

Operation time (min)

0.4

15ml/min 25ml/min 35ml/min 45ml/min

0.3 0.2 0.1

0

60

120

180

240

Amplitude (V)

180

240

300

360

300

360

15ml/min 25ml/min 35ml/min 45ml/min

0.3 0.2 0.1 0.0

0

60

Operation time (min)

0.4

120

Operation time (min)

Amplitude (V)

Amplitude (V)

0.4

0.0

399

120

180

240

300

360

Operation time (min)

15ml/min 25ml/min 35ml/min 45ml/min

0.3 0.2 0.1 0.0

0

60

120

180

240

300

360

Operation time (min) Fig. 7. Amplitudes of differential signals detected by (a) TD1, (b) TD2, (c) TD3, (d) TD4 and (e) TD5 versus operation time during the fouling experiment at the constant flux of 12 L/(m2 h) with different aeration rates of 15, 25, 35 and 45 mL/min, respectively.

400

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

Amplitude (V)

0.4

1000 mm 800 mm 600 mm

0.3

0.2

shorter double-end SHFM module possesses higher efficiency of water production due to the lower pressure drop [22], the entire surface along the shorter double-end SHFM module was easily subjected to fouling at the same operating conditions. Consequently, a longer double-end SHFM module can be suitable to maintain a sustainable operating performance in practical applications.

0.1

3.6. Effect of operating flux on membrane fouling TD1'

TD2'

TD3'

Fig. 8. Amplitudes of differential signals detected by TD1′, TD2′ and TD3′ after 360 min of fouling experiment at the constant flux of 12 L/(m2 h) with different fiber lengths of 600, 800 and 1000 mm, respectively.

0 V during the whole fouling operation (Fig. 7c). Therefore, the monitoring results implied that the membrane fouling could not be completely eliminated even at a high aeration rate and still appeared at the parts near both of suction ends of double-end SHFM module. This is the reason that the adhesive force caused by the local TMP exceeds the surface shear force generated from the dual-phase flow of aeration near the outlet regions [26]. Furthermore, it also can be seen in Fig. 7 that the emergence time of differential signal extended gradually with the increase of the aeration rate. For example, the peak of differential signal detected by TD1 emerged after 60, 120, 180 and 270 min filtration when the aeration rate was 15, 25, 35 and 45 mL/min, respectively (Fig. 7a). Similarly, the emergence time of differential signal detected by TD5 was 60, 90, 150 and 225 min, respectively (Fig. 7e). It suggested that the continue increase of aeration intensity could effectively mitigate membrane fouling and restrain the formation of fouling layer. However, it could not completely eliminate membrane fouling in double-end SHFM module. 3.5. Effect of fiber length on membrane fouling In order to further quantify the effect of fiber length on the fouling distribution, the fouling experiments were carried out by three membrane modules with different fiber lengths of 600, 800 and 1000 mm at the same constant flux of 12 L/(m2 h) and the aeration rate of 15 mL/min. At the same time, only three transducers named TD1′, TD2′ and TD3′ were employed to mounted on the upper, middle and lower part of the test module, respectively. In all the three modules, the distance of TD1′ to the surface of circulating fluid was 50 mm and the distance of TD3′ to the lower suction end of the fiber was 50 mm. The distances between the three transducers (TD1′–TD3′) were 200, 300 and 400 mm in the module lengths of 600, 800 and 1000 mm, respectively. Fig. 8 presents the amplitudes of differential signals obtained by TD1′, TD2′ and TD3′ after 360 min of fouling operation with the different fiber lengths of 600, 800 and 1000 mm, respectively. The amplitude of differential signal obtained from TD1′ decreased from 0.38 to 0.26 V with the decrease of the fiber length from 1000 to 600 mm. Similarly, the amplitude of differential signal obtained from TD3′ decreased gradually from 0.28 to 0.20 V when the fiber length decreased from 1000 to 600 mm. On the contrary, the amplitude of differential signal obtained by TD2′ increased gradually from 0 to 0.12 V as the decease of the fiber length from 1000 to 600 mm. It implied that a shorter double-end SHFM module suffered from more serious fouling than a longer one at the same constant flux of 12 L/(m2 h). The fouling profile became more uniform with the decrease of the fiber length. The reason is that the local TMP at the middle part of double-end SHFM increased with the decrease of the fiber length [21]. Simultaneously, compared with the short fiber, a longer fiber could induce substantial movement that helps control fouling [27]. In a word, although a

As a significant operating parameter, the operating flux in the system of membrane filtration is highly related to the stability of the system and the capital costs [28]. Hence, it is essential to investigate the effect of the operating flux on the fouling distribution under the sub-critical operating conditions. Fig. 9 presents the amplitudes of differential signals obtained from TD1 to TD5 after 360 min of fouling operation at the constant fluxes of 6, 9 and 12 L/(m2 h) with the aeration rate of 15 mL/min and the fiber length of 1000 mm. It can be seen in Fig. 9 that the amplitudes of differential signals obtained from TD1 to TD5 decreased sharply from 0.38, 0.15, 0, 0.09 and 0.28 V to 0.15, 0.06, 0, 0 and 0.10 V with the decrease of the imposed operating flux from 12 to 9 L/(m2 h), respectively. Upon further decreasing the operating flux to 6 L/(m2 h), only slight differential signals with amplitudes of 0.09 and 0.07 V obtained by TD1 and TD5 could be observed. It implied that the membrane fouling would be substantially mitigated under the operating flux of 6 L/(m2 h). In addition, the difference between the maximum and minimum amplitudes of differential signals obtained from TD1 to TD5 decreased from 0.38 to 0.09 V when the operating flux decreased from 12 to 6 L/(m2 h) (Fig. 9). It indicated that the asymmetry of fouling distribution decreased with the decrease of the operating flux. In the same way, Eq. (1) could be also used to quantify the effect of operating flux on local flux. When the operating flux decreased from 12 to 6 L/(m2 h), the local fluxes at the upper suction end and middle part of the fiber obtained by Eq. (1) gradually decreased from 21.4 to 7.8 L/(m2 h) to 10.7 and 3.9 L/ (m2 h), respectively. It was attributed to the lower driving force at the lower flux [29]. These results further confirmed that a low operating flux conducted was the main reason to control membrane fouling. Furthermore, an interesting phenomenon was observed that the amplitudes of differential signals obtained by TD1 and TD5 at operating flux of 6 L/(m2 h) were 0.09 and 0.07 V as shown in Fig. 9, which were much lower than that (0.15 and 0.12 V) obtained at the aeration rate of 45 mL/min as shown in Fig. 7a and Fig. 7e. In addition, a slight differential signal with amplitude of 0.02 V obtained from TD4 could be still observed at the aeration rate of 45 mL/min (Fig. 7d). This implied that a low operating flux is in 0.4 12L/m2 h

9L/m2 h

6L/m2 h

TD4

TD5

0.3

Amplitude (V)

0.0

0.2

0.1

0.0

TD1

TD2

TD3

Fig. 9. Amplitudes of differential signals detected by TD1, TD2, TD3, TD4 and TD5 after 360 min of fouling experiment at the aeration rate of 15 mL/min with different operating fluxes of 6, 9 and 12 L/(m2 h), respectively.

X. Li et al. / Journal of Membrane Science 444 (2013) 394–401

favor of membrane fouling control in the double-end SHFM system. 4. Conclusions The sub-critical flux fouling behavior along the double-end SHFM was non-invasively visualized by UTDR during the filtration of yeast suspension. The double-end SHFM module showed better filtration performance compared with the one-end SHFM module under the same filtration conditions. However, the double-end SHFM still easily suffered from the membrane fouling on the parts near both the upper and lower suction ends under the normal subcritical flux operation. And the deposition of the fouling onto the membrane surface of the double-end SHFM gradually migrated from both suction ends to middle part of the fiber and reached the plateau finally. Interestingly, the deposition of particles on the part near the upper suction end of fiber was much more evident than that near the lower suction end. The increase of aeration intensity could effectively mitigate membrane fouling, but could not completely eliminate membrane fouling in the double-end SHFM module. A low operating flux is in favor of membrane fouling control in the double-end SHFM system. Moreover, a shorter double-end SHFM module easily suffered from more serious fouling than a longer one at the same operating conditions. Acknowledgments This work was financially supported by the National Program on Key Basic Research Project of China (no. 2011CB612311), the SA/ CHINA Agreement on Cooperation on Science and Technology (no. 2010DF51090), National Natural Science Foundation of China (no. 20876115) and the Commonweal Project of China National Ocean Bureau (no. 201105025). References [1] A. Robles, M.V. Ruano, J. Ribes, J. Ferrer, Sub-critical long-term operation of industrial scale hollow-fibre membranes in a submerged anaerobic MBR (HFSAnMBR) system, Sep. Purif. Technol. 100 (2012) 88–96. [2] T. Zsirai, P. Buzatu, P. Aerts, S. Judd, Efficacy of relaxation, backflushing, chemical cleaning and clogging removal for an immersed hollow fibre membrane bioreactor, Water Res. 46 (2012) 4499–4507. [3] S. Chang, A.G. Fane, S. Vigneswaran, Modeling and optimizing submerged hollow fiber membrane modules, AIChE J. 48 (2002) 2203–2212. [4] Z.F. Cui, S. Chang, A.G. Fane, The use of gas bubbling to enhance membrane processes, J. Membr. Sci. 221 (2003) 1–35. [5] P. Buzatu, T. Zsirai, P. Aerts, S.J. Judd, Permeability and clogging in an immersed hollow fibre membrane bioreactor, J. Membr. Sci. 421–422 (2012) 342–348. [6] Z.W. Wang, Z.C. Wu, X. Yin, L.M. Tian, Membrane fouling in a submerged membrane bioreactor (MBR) under sub-critical flux operation: membrane foulant and gel layer characterization, J. Membr. Sci. 325 (2008) 238–244. [7] B.D. Cho, A.G. Fane, Fouling transients in nominally sub-critical flux operation of a membrane bioreactor, J. Membr. Sci. 209 (2002) 391–403.

401

[8] S. Ognier, C. Wisniewski, A. Grasmick, Membrane bioreactor fouling in subcritical filtration conditions: a local critical flux concept, J. Membr. Sci. 229 (2004) 171–177. [9] S. Chang, A.G. Fane, T.D. Waite, Effect of coagulation within the cake-layer on fouling transition with dead-end hollow fiber membranes, in: Proceedings of International Congress on Membranes and Membrane Processes (ICOM), Seoul, Korea, 2005. [10] A.P.S. Yeo, A.W.K. Law, A.G. Fane, Factors affecting the performance of a submerged hollow fiber bundle, J. Membr. Sci. 280 (2006) 969–982. [11] W. Lee, S. Kang, H. Shin, Sludge characteristics and their contribution to microfiltration in submerged membrane bioreactors, J. Membr. Sci. 216 (2003) 217–227. [12] F.G. Meng, F.L. Yang, Fouling mechanisms of deflocculated sludge, normal sludge, and bulking sludge in membrane bioreactor, J. Membr. Sci. 305 (2007) 48–56. [13] Y. Marselina, P. Lifia, R.M. Le-Clech, V. Stuetz, Chen, Characterisation of membrane fouling deposition and removal by direct observation technique, 341 (2009) 163–171, J. Membr. Sci. 341 (2009) 163–171. [14] C. Sun, L. Fiksdal, A. Hanssen-Bauer, M.B. Rye, T. Leiknes, Characterization of membrane biofouling at different operating conditions (flux) in drinking water treatment using confocal laser scanning microscopy (CLSM) and image analysis, J. Membr. Sci. 382 (2011) 194–201. [15] S. Buetehorn, L. Utiu, M. Küppers, B. Blümich, T. Wintgens, M. Wessling, T. Melin, NMR imaging of local cumulative permeate flux and local cake growth in submerged microfiltration processes, J. Membr. Sci. 371 (2011) 52–64. [16] X.C. Xu, J.X. Li, H.S. Li, Y. Cai, Y.H. Cao, B.Q. He, Y.Z. Zhang, Non-invasive monitoring of fouling in hollow fiber membrane via UTDR, J. Membr. Sci. 326 (2009) 103–110. [17] X.C. Xu, J.X. Li, N.N. Xu, Y.L. Hou, J.B. Lin, Visualization of fouling and diffusion behaviors during hollow fiber microfiltration of oily wastewater by ultrasonic reflectometry and wavelet analysis, J. Membr. Sci. 341 (2009) 195–202. [18] X.H. Li, J.X. Li, J. Wang, H.W. Zhang, Y.D. Pan, In situ investigation of fouling behavior in submerged hollow fiber membrane module under sub-critical flux operation via ultrasonic time domain reflectometry, J. Membr. Sci. 411–412 (2012) 137–145. [19] A.P. Mairal, A.R. Greenberg, W.B. Krantz, L.J. Bond, Real-time measurement of inorganic fouling of RO desalination membranes using ultrasonic time-domain reflectometry, J. Membr. Sci. 159 (1999) 185–196. [20] J.X. Li, V.Y. Hallbquer-Zadorozhnaya, D.K. Hallbauer, R.D. Sanderson, Cakelayer deposition, growth, and compressibility during microfiltration measured and modeled using a noninvasive ultrasonic technique, Ind. Eng. Chem. Res. 41 (2002) 4106–4115. [21] S. Chang, A.G. Fane, T.D. Waite, Analysis of constant permeate flow filtration using dead-end hollow fiber membranes, J. Membr. Sci. 268 (2006) 132–141. [22] S. Judd, The M.B.R. Book: Principles and Applications of Membrane Bioreactors in Water and Wastewater Treatment, Elsevier, London, U.K, 2006. [23] J. Zhang, H.C. Chua, J. Zhou, A.G. Fane, Factors affecting the membrane performance in submerged membrane bioreactors, J. Membr. Sci. 284 (2006) 54–66. [24] A.S. Kim, Y.T. Lee, Laminar flow with injection through a long dead-end cylindrical porous tube: application to a hollow fiber membrane, AIChE J. 57 (2011) 1997–2006. [25] B.G. Fulton, J. Redwood, M. Tourais, P.R. Bérubé, Distribution of surface shear forces and bubble characteristics in full-scale gas sparged submerged hollow fiber membrane modules, Desalination 281 (2011) 128–141. [26] K.J. Hwang, H.C. Chen, Selective deposition of fine particles in constant-flux submerged membrane filtration, Chem. Eng. J. 157 (2010) 323–330. [27] F. Wicaksana, A.G. Fane, V. Chen, Fibre movement induced by bubbling using submerged hollow fibre membranes, J. Membr. Sci. 271 (2006) 186–195. [28] R.W. Field, G.K. Pearce, Critical, sustainable and threshold fluxes for membrane filtration with water industry applications, Adv. Colloid Interface Sci. 164 (2011) 38–44. [29] T. Carroll, The effect of cake and fibre properties on flux declines in hollowfibre microfiltration membranes, J. Membr. Sci. 189 (2001) 167–178.