- Email: [email protected]
Fouling control of submerged hollow fibre membranes by vibrations
Journal of Membrane Science 427 (2013) 230–239
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Fouling control of submerged hollow fibre membranes by vibrations Tian Li a,b, Adrian Wing-Keung Law a,b,n, Merve Cetin a,b, A.G. Fane a,b a b
School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore Singapore Membrane Technology Centre, Nanyang Environment and Water Resource Institute, 1 Cleantech Loop, Clean Tech One, Singapore 637141, Singapore
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 May 2012 Received in revised form 10 September 2012 Accepted 15 September 2012 Available online 2 October 2012
In this study, we examine the improvement of fouling control of hollow fibre membranes with mechanical vibrations in the dead-end filtration of an inorganic suspension. Hollow fibres with diameters of 1.7 mm and 2 mm vibrating at moderate frequencies (0–15 Hz) and small amplitudes (0–12 mm) were submerged vertically in a 4 g/L Bentonite solution. Experiments were then conducted at both constant permeate flux and constant suction pressure conditions. The results showed that the membrane performance can be greatly improved when the vibration frequency or the vibration amplitude increases beyond a threshold magnitude. For example, over 90% reduction in the membrane fouling rate was achieved at 8 mm amplitude and 8 Hz frequency vibration compared to no vibration. Experiments were also conducted with 1% and 2% fibre looseness. The results showed that a small looseness can reduce the membrane fouling and increase the permeate flux under vibrations, which can be mainly attributed to the additional lateral movement of the fibres induced by the looseness. A comparison of vibrating the hollow fibres with and without the holding frame was also carried out to determine the effects of turbulence generated by the vibrating holding frame used in the experimental setup. Particle Image Velocimetry (PIV) measurements were performed to quantify the associated turbulence inside the membrane reactors. It was confirmed that the turbulence generated by the vibrating frame was more obvious at a high vibration frequency. However, it had little influence on the membrane filtration performance. Overall, the results from the present study confirm that at moderate frequencies, the cake layer resistance can be reduced substantially by vibration due to the dynamic shear enhancement on the membrane surface. & 2012 Elsevier B.V. All rights reserved.
Keywords: Vibration Hollow fibres Looseness Shear stress Particle image velocimetry
1. Introduction Hollow fibre membranes are now widely used in wastewater treatment processes. In particular, it is becoming the membrane of choice in submerged membrane bioreactors (SMBRs) [1–3] which can produce high quality water in a reduced reactor size and also minimize sludge production [4]. However, its application is still hindered by problems associated with concentration polarization and membrane fouling. For fouling mitigation, the use of hydrodynamic shear stresses on the membrane surface is recognized as one of the most effective techniques [5,6]. Hydrodynamic shear stresses can be generated on the membrane surface by moving either the fluid next to the membrane or the membrane surface itself. In the former case, the use of bubbling has been commonly adopted in SMBRs for industry applications. In addition to inducing wall shear stresses on the membrane surface [7], bubbling also n Corresponding author at: School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore. Tel.: þ 65 67905296. E-mail address: [email protected] (A.W-K. Law).
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.09.031
oxygenates the fluid and disrupts the concentration polarization layer around the hollow fibres [8]. However, the shear stresses induced by bubbling are relatively weak. Also, the flux improvement reaches a limit beyond which further increase in the air flow rate becomes ineffective [9]. Vibrating the membrane module can also induce dynamic shear stresses on the membrane surfaces for fouling mitigation. The concept of Vibratory Shear-Enhanced Processing (VSEP) was first proposed by Armando et al. [10]. In their setup, a stack of circular organic membranes were separated by gaskets and permeate collectors, and mounted on a vertical torsion shaft spun in azimuthal oscillations by vibrating the base at a resonant frequency of about 60 Hz. The shear rate at the membrane surface was generated due to the inertia of the fluid which moved at 1801 out of phase with the membrane. Since then, commercial modules based on VSEP have been made available, with further development in the forms of rotating disks [11], rotating membranes [12], vibrating flat membranes [13], and vibrating hollow fibres [14]. Their applications are extensive, including the removal of humic substances [15], diluted milk [16], natural organic contaminants [17,18] as well as for wastewater treatment [19]. The uses of vibrating flat sheet membranes [20] and rotating disks [16] were
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
contribute to the reduction of membrane fouling [23]. We examined the turbulence induced by the holding frame vibration using the technique of Particle Image Velocimetry (PIV), and compared the filtration performance by the hollow fibres with and without the vibration of the holding frame. The results will be discussed in detail in the following after the description of the experimental setup and procedures.
also reported to be very promising, not only for much higher permeate fluxes, but also for better rejection. Recently, the possibility of using vibration for submerged hollow fibre membranes in MBR was raised by Low et al. [14]. They investigated different mechanical motions, including both cross oscillation and lengthwise oscillation. They found that cross oscillation was not effective because of unequalled shear forces on the membrane bundle, while the lengthwise oscillation provided alternating shear forces that are equal throughout the membrane surface. They concluded that the mechanical vibration helped maintain the MBR membrane in a relatively ‘‘clean’’ condition and keep the permeate flux close to that of the clean membrane. Subsequently, Genkin et al. evaluated the effect of vibration with a range of 0–10 Hz frequency and 0–40 mm amplitude and coagulant addition on the filtration performance of submerged hollow fibre membranes [21]. They discovered that both coagulation and transverse vanes could enhance the critical flux of the vibrating membrane unit. For example, at a vibration frequency of 1.7 Hz, the critical flux increased from 17 to 46 L/ (m2 h) with the addition of 34 mg/L Aluminium Chlorhydrate (ACH). In the presence of combined axial and transverse vibrations, a five-fold enhancement in critical flux to 86 L/(m2 h) was also achieved at 1.7 Hz in the presence of 34 mg/L ACH. They attributed the effect of coagulation to the aggregation of fine particles and evacuation of aggregates away from the membrane surface due to inertial and gravitational forces, and the effect of vanes due to the intensified turbulence arising from both longitudinal and transverse vibrations. Beier et al. [22] carried out experiments with a vibrating hollow fibre membrane module using baker yeast and also confirmed that the higher critical fluxes could be achieved through higher vibration frequency and amplitude. The objective of the present study is to further examine the different factors affecting the fouling mitigation of hollow fibre membranes with vibrations in an inorganic Bentonite solution. The scope is wider than the previous studies of Low et al. [14], Genkin et al. [21] and Beier et al. [22], includes various combinations of vibration frequencies, vibration amplitudes, and degrees of fibre looseness (previous studies involved only tightly held fibres). In addition, the frame that held the hollow fibre membranes could generate turbulence in the fluid upon vibration, which could also
Permeate pump
2. Experimental setup and procedures 2.1. Vibration setup The schematic diagram of the experimental setup is shown in Fig. 1. The test tank was made of Persplex with sizes of 400 mm (L) 400 mm (W) 1200 mm (H). The hollow fibre membrane module was submerged in an inorganic suspension made with Bentonite particles from Sigma-Aldrich. The Bentonite particles had a mean size of 5.8 mm, and the particle concentration was 4 g/L (pH 6.0–9.0). In the experiments, a typically small magnitude of air bubbling (50 mL/min) was maintained in the tank to keep the Bentonite particles suspended in the reactor. The holding frame of the membrane module was driven by a brushless DC motor with a crank moving mechanism. The vibration amplitude could be varied from 0 to 12 mm accurately, while the vibration frequency could be set to 0–15 Hz. The permeate flow was controlled by a master flex peristaltic pump together with a needle valve. The suction pressure was measured with a pressure transducer, and the permeate flux with a digital balance. Two different sizes of Polyacrylonitrile (PAN) hollow fibres made by Ultrapure Pte. Ltd. in Singapore with inner/outer diameters of 1 mm/1.7 mm and 1 mm/2 mm, respectively, were tested in the experiments. The length of each fibre was 40 cm, and the nominal pore size was 0.1 mm. A total of 13 fibres were included in the membrane modules. They were aligned vertically in parallel, with the distance between two adjacent fibres of 15 mm. Both ends of the fibres were connected to a 3 cm hard tubing with epoxy, and then mounted on the cross beam of the holding frame. Four screw nuts located at the border of the cross beam could be adjusted to change the submergence of the membrane module in the reactor.
Pressure transducer Valve
Digital balance
P
Frequency controller
Hollow fibre membranes
Nd: YAG laser
Camera Diffuser
Computer
231
Air pressure pump
Air flow meter
Fig. 1. Experimental setup.
Tank
Motor
232
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
2.2. Experimental setup for particle image velocimetry (PIV) The laser imaging technique of PIV has been used successfully before to measure the local velocities around a hollow fibre array [23,24]. With PIV, a laser sheet is projected onto the flow field at successive time intervals, and the subsequent capturing of images details the position of seeding particles that scatters the laser light [25]. A cross-correlation analysis of the changes in the particle position in the sequence of images reveals the Lagrangian velocity distribution. The details of PIV can be found in our earlier references [23–25]. In the present study, a Dantec flow map system was used for the PIV measurements. The illumination was from a dual-cavity frequency doubled Q-switched pulsed mini Nd:YAG laser. The energy level was 150 mJ and the pulse duration was 6 ns. The emitted light sheet was green, at a wavelength of 532 nm. The thickness of the light sheet was approximately 1 mm with a divergence angle of 321. A Hi-Sense Mk II camera was used to capture the images. The camera had a high-performance progressive scan interlined CCD chip. The chip included 1280 1024 light-sensitive cells and an equal number of storage cells. Seeding particles, with a diameter of 20 mm, were used as the particle markers. The concentration of the seeding particles was set at 0.1 g/L, to make sure that a sufficient number of particle markers were present within an interrogation area (IA). The data acquisition was performed at 6 Hz. The final interrogation cell size was 32 32 pixels with 50% overlap in IA, giving a grid of 83 63 for a total of 5229 velocity vectors. For every experiment, a total of 300 double frame images were typically taken to establish the velocity vector maps, from which the magnitude of the mean velocity and the turbulence fluctuations could be analyzed. 2.3. Experimental procedures Before each experiment, a clean water backwash at 20 mL/min was first performed on the hollow fibre membranes under the condition of 10 Hz vibration for 15 min. The purpose was to eliminate any bubbles trapped inside the hollow fibres. After the backwash, the filtration test was initiated, and the permeate volume was recorded at every 20 s interval by a digital balance. The permeate flux was then calculated as the rate of change of the permeate volume. Two types of experiments were conducted in this study, namely the constant flux and constant pressure experiments. They are described in the following. 2.3.1. Constant flux For the constant flux experiments, the permeate flux was typically maintained at 30 70.3 L/(m2 h), while the corresponding suction pressure was recorded using a pressure gauge at 1-min interval. Each experiment lasted for about 1 h. However, if the suction pressure decreased to around 40 kPa, many bubbles began to be generated from the tubing especially around the needle valve. When this occurred, the experiment was terminated and the duration of the particular experiment would then be shortened to be less than 1 h. To maintain consistency, we reported the average fouling rate of the first 30 min as the experimental fouling rate in this study. 2.3.2. Constant pressure In the constant pressure experiments, the suction pressure was controlled at 2470.1 kPa, while the permeate fluxes varied. Before each experiment, the membrane was backwashed in the same manner as in the constant flux experiments. Subsequently, the experiment would begin with the pump flow
increased so that the suction pressure reached 24 kPa. Over time, the permeate flux would typically decrease progressively which was determined by the measurements of permeate volume recorded at 20 s intervals using the digital balance. Each constant pressure experiment lasted for about 1 h. Since the pressure was maintained constant, the interruption due to air bubble generation as in the constant flux experiments did not occur. 2.3.3. Vibrating hollow fibres with and without holding frame As part of the overall experiments, the effects of holding frame vibration on the filtration performance of the hollow fibre membranes were also examined by comparing two test conditions: (a) vibrating hollow fibres, and (b) stationary hollow fibres. The former condition was used in most experiments. To study the stationary hollow fibres with the vibrating holding frame, the hollow fibres were separately fixed to another stationary holding frame while the original holding frame was driven by the motor. By comparing the two test conditions, the frame effect on the membrane performance could be evaluated.
3. Analysis of results 3.1. Membrane filtration The definition of permeate flux is J¼
1 dV S dt
ð1Þ
where S is the effective membrane area and V is the cumulative volume of permeate. According to the resistance-in-series model [26], the relationship between the permeate flux (J) and transmembrane pressure (DP) can be expressed in the following form: ð2Þ J ¼ DP= mRt Rt ¼ Rm þ Rp þ Ref þ Rif
ð3Þ
where m is the dynamic viscosity of the permeate; Rt is the total resistance; Rm is the intrinsic membrane resistance; Rp is the polarization layer resistance caused by the concentration gradient; Ref is the external fouling resistance formed by a deposited cake layer from physic–chemical interactions of solids with the membrane surface; and Rif the internal fouling resistance due to irreversible adsorption or pore plugging. For clean water, Eq. (2) can be simplified as ð4Þ Jw ¼ DP= mRm where Jw is the clean water flux. In the present study, a clean water permeability (Flux/TMP) test was conducted before each experiment for both the 1.7 mm and 2 mm fibres. Their permeabilities were typically found to be about 2.04 L/(m2 h kPa) and 1.73 L/(m2 h kPa) respectively. Since the clean water viscosity at 20 1C (the typical temperature in the experiments) was 1.005 10 3 Pa s, thus Rm for the 1.7 mm and 2 mm fibres can be calculated to be 1.76 1012 m 1 and 2.07 1012 m 1 respectively. For the Bentonite solution, the nominal particle size of the powder was much larger than the pore size of the hollow fibre. Hence, Eq. (2) can be rearranged as J¼
DP
mðRm þ Rc Þ
ð5Þ
where Rc is the cake resistance (Rc ¼Ref þRif), which for hollow fibre membranes should also vary with the cake thickness dc as [27], Rc ¼ Rc 0 rs |c dc
ð6Þ
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
233
where Rc0 is the specific cake resistance per unit mass, rs the solid density, and |c the solid volume fraction. The mass of cake deposited per unit volume of filtrate, c, can then be expressed as c¼
Sdc |c rs V
ð7Þ
Therefore, for an incompressible cake, integrating (1) with constant DP gives [27] t mRc 0 c mRm ¼ 2 Vþ V SDP 2S DP
ð8Þ
Eq. (8) quantifies the filtration characteristics when a cake layer is formed. 3.2. Estimation of filtration parameters For each experiment, the permeate volume was recorded continuously, and the permeate flux was computed with Eq. (1). By plotting the permeate volume versus time, the longitudinal intercept of Eq. (8) can be obtained as b¼
mRm SDP
Fig. 2. Fouling rate at different permeate fluxes without vibration.
ð9Þ
The membrane resistance Rm can then be determined as Rm ¼
b
m
SDP
ð10Þ
The slope of Eq. (8) can be defined as k¼
mRc 0 c 2S2 DP
ð11Þ
The cake resistance Rc can be calculated from this slope together with Eqs. (6) and (7) as: Rc ¼
2k
m
SDPV
ð12Þ
Examples of the calculation can be found in the later part of results and discussion. 3.3. Vibration enhancement factor Membrane oscillation can potentially lead to an improvement in the fouling mitigation. The enhancement factor, E, in terms of the permeate flux can be used to describe the vibration intensification as follows: ðJÞwith oscillation ð13Þ E¼ ðJÞwithout oscillation t ¼ 1 h The flux values were taken after 1 h which corresponded to the typical time needed to reach the quasi-steady state conditions. This approach has been widely used for membrane vibration filtration analysis [20,28]. In this study, we also use this parameter to evaluate the membrane filtration improvement with vibrations.
4. Results and discussion 4.1. Filtration characteristics Fig. 2 shows the measured fouling rate with different constant permeate fluxes for both the 1.7 mm and 2 mm fibres without vibration. It was obvious that the fouling rate increased as the fibre diameter decreased, which may be attributed to the higher axial flux due to the smaller radius of the 1.7 mm fibres. For example, the fouling rate of the 1.7 mm fibres at a permeate flux over 25 LMH was double of that for the 2 mm fibres. It should be noted that a similar effect (greater fouling for smaller diameter
Fig. 3. Fouling rate versus permeate flux at different vibration frequencies (vibration amplitude¼5 mm, 2 mm fibre).
fibres) was also observed elsewhere [29]. In addition, larger permeate fluxes led to much higher fouling rates, as illustrated by the fact that the fouling rate at a constant permeate flux of 30 LMH was 91.2 kPa/h and 48.6 kPa/h respectively for the 1.7 mm and 2 mm fibres, which was much higher than at a flux of 10 LMH where nearly no fouling was observed. This could be attributed to the increased cake compression and consolidation at higher fluxes. Fig. 3 shows the measured fouling rates with different vibration frequencies, at a fixed vibration amplitude of 5 mm for the 2 mm fibres (note that 0 Hz in the figure represents no vibration). The fouling rate typically decreased significantly when vibration was applied. It dropped from 48.6 kPa/h to 0.8 kPa/h at a permeate flux of 30 LMH with vibration frequency increased from 0 Hz to 10 Hz. It was also found that, with 2 Hz vibration, the suction pressure increased from 0.2 kPa/h to 15 kPa/h when the permeate flux increased from 10 LMH to 30 LMH. The same trend can be observed for other vibration frequencies. This implies more serious membrane fouling occurred when a larger amount of fluid permeated through the membrane as expected. In addition, almost no fouling was observed with 10 Hz vibration for the 2 mm fibres. In other words, vibrating the hollow fibre membrane at 10 Hz increased the critical flux for the 2 mm fibres to beyond 30 LMH in this case. Fig. 4 plots the fouling rate at different vibration frequencies and amplitudes for both the 1.7 mm and 2 mm fibres, at a constant permeate flux of 30 LMH. The enhancement by vibration intensified when both vibration amplitude and frequency
234
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 4. Fouling rate at different vibration frequencies and amplitudes (constant permeate flux¼ 30 LMH): (a) 2mm fibre and (b) 1.7mm fibre.
increased. There was an 85% reduction in the fouling rate when the 2 mm hollow fibres were vibrated with 5 mm amplitude and 5 Hz frequency, comparing to no vibration (Fig. 4(a)). Similar enhancement was obtained for the 1.7 mm fibres at 6 mm amplitude and 8 Hz frequency (Fig. 4(b)). When the vibration amplitude increased to 8 mm with the frequency of 10 Hz, more than 95% fouling rate reduction occurred for both the 1.7 mm and 2 mm fibres. When the vibration amplitude further increased to 10 mm with frequency of 12 Hz, nearly no fouling was observed for either fibre. The plot of t/V versus V and vibration frequency F is presented in Fig. 5 with the cake filtration law. The time needed to obtain the same amount of permeate was longer at lower vibration frequencies, implying that vibration could improve the permeate flux. The membrane resistance was calculated based on Eq. (10). For example, for the 2 mm fibre without vibration (Fig. 5(a)) the results showed that the longitudinal intercept b¼2.63 103 and the slope k¼2.98 103. Thus, the average membrane resistance Rm was 2.26 1012 m 1 by Eq. (10), and the relationship between Rc (1012 m 1) and V (L) was Rc ¼5.12 V by Eq. (12). A similar method was used for the 1.7 mm fibres. The average membrane resistances Rm was 1.97 1012 m 1 and the relationship between Rc (1012 m 1) and V (L) was Rc ¼ 8.33 V. The average membrane resistance calculated by the cake filtration law analysis was nearly the same as that from the clean water permeability tests, which illustrated the consistence of the test results. Fig. 6 plots the cake resistance versus cumulative permeate volume according to the cake filtration law. It can be observed that the cake resistance increased gradually with time at different vibration frequencies for both 1.7 mm and 2 mm hollow fibres. Also, the cake resistance of 1.7 mm fibres without vibration was much
Fig. 5. Evolution of t/V versus cumulative permeate volume at different vibration frequencies with cake filtration law identification (constant suction pressure¼ 24 kPa): (a) 2mm fibre, vibration amplitude¼ 5mm and (b) 1.7mm fibre, vibration amplitude¼ 6mm.
larger than 2 mm fibres due to the smaller diameter which would lead to greater effective fouling [29]. It reduced from 7.86 1012 m 1 to 0.41 1012 m 1 when the vibration frequency of 10 Hz was applied compared to no vibration for the 2 mm fibres at a 1.5 L permeate volume. Hence, a 95% cake resistance reduction was achieved by the 10 Hz vibration. Comparatively, 75% cake resistance reduction at 6 Hz vibration and 83% cake resistance reduction at 10 Hz vibration were obtained for the 1.7 mm fibres, which was also significant but nevertheless less than the 2 mm fibres. In summary, Figs. 3–6 all show improved performance of the membrane under vibration, and the improvement was very significant when a certain level of vibration frequency and/or vibration amplitude was reached. This could be attributed to the fact that at higher vibration frequencies, the particles could be easily shaken loose from the membrane surface under conditions of high shear stress. 4.2. Intensification effects of vibration For all the experiments, the average permeate flux was determined numerically after 1 h, which was typically the time needed to reach a quasi-steady state. The membrane vibrations resulted in an increase of this quasi-steady state permeate flux, with the enhancement factor dependent on the applied vibration amplitude and frequency as well as the TMPs. In general, the enhancement factor would be larger with higher vibration amplitudes and frequencies at certain TMPs. Fig. 7 depicts the effect of vibration enhancement factor for both the 1.7 mm and 2 mm fibres. Vibration enhancement faster increased from 1.5 to 2.6 when the vibration frequency increased
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
235
Fig. 6. Evaluation of cake resistance Rc versus cumulative permeate volume at different vibration frequencies (constant suction pressure¼ 24 kPa): (a) 2mm fibre, vibration amplitude 5mm and (b) 1.7mm fibre, vibration amplitude 6mm. Fig. 7. Vibration enhancement factor versus vibration frequency at different vibration amplitudes (constant suction pressure¼ 24 kPa): (a) 2mm fibre and (b) 1.7mm fibre.
from 2 Hz to 10 Hz respectively with the vibration amplitude of 8 mm for the 2 mm fibres, while it was 2.2 times the quasi-steady state permeate flux at 8 mm vibration amplitude and 10 Hz vibration frequency for the 1.7 mm fibres. This could potentially imply that the larger fibre size could perform better in a vibrating system. The enhancement could again be attributed to the reduction of the deposited cake layer at the membrane surface due to the vibration shear at the membrane surface and the subsequent decrease in the cake resistance to the filtration (Fig. 6).
4.3. Effect of membrane looseness Fig. 8 shows the quasi-steady state permeate flux for the tight, 1% and 2% looseness fibres. The quasi-steady state flux increased with an increase of vibration frequency and was higher for the loose fibres. With 1% looseness, it increased 20% from 17 LMH to 21 LMH at 4 Hz, and 30% at 10 Hz. However, very little improvement was noted for all the vibration frequencies when the fibre looseness increased from 1% to 2%. The vibration enhancement effect is further shown in Fig. 9, and a similar trend is observed. The enhancement factor was higher with a larger vibration frequency. It increased from 1.6 to 2 at 6 Hz and from 2 to 2.5 at 8 Hz with 1% looseness. The factor reached as high as 3 with 1% looseness and 2.2 with the tight fibres at 10 Hz. The results confirmed that the fibre looseness can improve the membrane performance with vibrations due to axial vibrational shear stresses and the extra lateral movement induced by the looseness, and only a looseness of 1% would be required to realize the improvement.
Fig. 8. Quasi-steady state flux for tight and loose fibres at different vibration frequencies (constant suction pressure ¼ 24 kPa, 1.7 mm fibre).
Fig. 10 shows the fouling rate of tight and loose fibres at 6–12 Hz vibration frequency. The fouling rate decreased significantly for tight and loose fibres in this frequency range. It dropped from 29.4 kPa/h to 2 kPa/h from 6 Hz to 12 Hz for the tight fibres, and from 8.9 kPa/h to 1.4 kPa/h for the 1% looseness fibres, respectively. Again the fouling rate for the 1% and 2% looseness was quite similar, confirming that the fibre looseness can decrease the fouling rate, and 1% looseness is sufficient to improve the membrane filtration performance. The improvement
236
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 9. Vibration enhancement factor for tight and loose fibres at different vibration frequencies (constant suction pressure¼ 24 kPa, 1.7 mm fibre).
Fig. 11. Fouling rate for vibrating and stationary fibres with vibrating holding frame at different vibration frequencies (constant permeate flux¼ 30 LMH, vibration amplitude ¼10 mm, 1.7 mm fibre).
Fig. 10. Fouling rate for tight and loose fibres at different vibration frequencies (constant permeate flux¼ 30 LMH, vibration amplitude¼ 6 mm, 1.7 mm fibre).
can be attributed to the additional lateral movement of the loose fibres and the resulting increase in shear stress. Similar effects were also noted in the bubbling studies by Yeo et al. [24] and Wicaksana et al. [6]. However, it is probable that excessive fibre looseness would lead to fibre breakage due to extreme movement of fibres. 4.4. Effect of fibre holding frame Fig. 11 shows the fouling rate at constant flux of the stationary and vibrating fibres with the vibrating holding frame. It was obvious that the fouling rate of the stationary fibres was much higher than the vibrating fibres together with the vibration of the holding frame, decreasing from 85.8 kPa/h to 41.5 kPa/h at 6 Hz, and from 79.6 kPa/h to 2.2 kPa/h at 10 Hz. With the stationary hollow fibres, the fouling rate was a little lower with the vibrating holding frame than no vibration. In Fig. 12, a similar trend can also be observed at constant pressure. The quasi-steady state flux of the vibrating fibres with the vibrating holding frame increased 80% from 17 LMH at 4 Hz to 31 LMH at 10 Hz. Overall, the results showed that the vibration of the holding frame and the associated turbulence did not affect the membrane performance significantly. Thus, it can be concluded that the vibration of the fibres to produce shear at the membrane surface is the primary factor that led to the reduction of membrane fouling. In the present study, the current results were different from Genkin et al. [21] where they intentionally used a special design of the holding
Fig. 12. Quasi-steady state flux of vibrating and stationary fibres with vibrating holding frame (constant suction pressure¼ 24 kPa, vibration amplitude¼ 8 mm, 1.7 mm fibre).
frame to generate turbulence in order to markedly enhance the critical flux of the membrane. However, the present data allows a direct examination of the role of dynamic vibration shear on the membrane filtration. 4.5. Dynamic shear rate Previous researchers had shown that the dynamic shear stress on the membrane surface during vibration could reduce the membrane fouling [5,13]. For the vibrating disk, Akoum et al. [13] suggested the relationship between the maximum shear stress and vibration frequency and amplitude to be
gmax ¼ ð2pFÞ 3=2 An1=2
ð14Þ
For simplicity, Genkin et al. [21] assumed that the same expression could be used to describe the shear rate at the surface of a vibrating fibre. However, the omission of the consideration of radius of curvature of the fibre can lead to significant errors with small diameter fibres as pointed out by Zamani et al. [30]. In this particular study, however, the fibre diameters at 1.7 mm and 2 mm were relatively large, and thus the errors are expected to be small. Experiments with the same theoretical constant shear rate (A*F1.5 ¼134) and with the amplitude varied from 4 mm to 12 mm
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
237
Fig. 14. Influence of the shift speed (4A*F) on fouling rate (constant permeate flux¼ 30 LMH, 1.7 mm fibre). Fig. 13. Fouling rate with the same theoretical shear rate (constant permeate flux¼ 30 LMH, 1.7 mm fibre).
Rev ¼ de V r=m
63cm
2.5cm
30.5cm
were conducted. The results are shown in Fig. 13. The fouling rate increased from 10.8 kPa/h to 18.8 kPa/h when the vibration amplitude increased from 4 mm to 9 mm. It then gradually decreased to 16 kPa/h when the vibration amplitude increased from 9 mm to 12 mm. This shows that the amplitude of 9 mm performed the worst under the same shear rate. The results demonstrated that at higher vibration frequencies, the power ratio of vibration frequency in Eq. (14) can be larger than 1.5. It can also be deduced that the vibration frequency could have much more impact than the vibration amplitude for fouling mitigation, which confirms that vibration frequency is more effective to improve membrane performance than vibration amplitude. Vigo and Uliana [31] suggested that the shear stress was dependent on the shift speed (4A*F) of the membrane surface in a vibrating system. The product 4A*F was a quantity proportional to the so-called vibrational Reynolds number as:
25cm
3cm
12cm
ð15Þ Fig. 15. Location of the image window in the reactor.
where de was the equivalent diameter of an annular vessel, and V the mean vibrational speed ( ¼4A*F). Experiments with the same shift speed were performed in this study with the vibration amplitude ranging from 4 mm to 12 mm (Fig. 14). The results showed that the fouling rate increased from 7.4 kPa/h to 66.7 kPa/h when A*F ¼48. Thus, the fouling rate climbed rapidly with increasing vibration amplitude at the same shift speed, i.e., the membrane performance was strongly affected by the increasing vibration frequency. This influence could be attributed to the much higher maximum shear stress at the membrane surface when the fibre was moving relative to the fluid [31]. Fig. 14 also demonstrates that the effect of vibration frequency was more obvious in a vertical vibrating system.
4.6. PIV measurement With PIV, the two-dimensional velocity distribution can be obtained in the form of vector maps, while the turbulence characteristics can be computed by evaluating the local vector values. For the vibration experiments in our study, the velocity components of 300 vector maps were obtained and the average velocity components u and v were calculated.
In this study, we carried out PIV measurements in the area shown in Fig. 15 just in front of the fibres and plotted the contour map of the velocity deviation U’ from the mean velocity at different frequencies in Fig. 16, where U0 ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 uu þ vv
ð16Þ
From figures, it can be observed that the value of U’ and the generation of turbulence by the holding frame was much more significant at higher vibration frequencies. The induced turbulence can potentially reduce the cake fouling layer on the membrane surface, thus increase the apparent permeability. Gomaa et al. [32] investigated the flux enhancement using oscillatory motion and turbulence promoters. They revealed that with higher frequencies (F45 Hz), the effect of turbulence promoters extended further away from the surface, resulting in a well-mixed flow structure and scouring of the surface between the promoters caused by vortices shedding. The effects of turbulence promoters in Gomaa et al. [32] are similar to the turbulence produced by the holding frame in the present study. However, as
238
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 17. A comparison of operating power consumption between (a) vibration and (b) aeration. Fig. 16. Contour of U’ at different vibration frequencies (no suction pressure, 2 mm fibre): (a) 2Hz and (b) 10Hz.
discussed earlier, the effect here of the holding frame per se was not significant due to the different structural configurations. 4.7. Power consumption comparison with bubbling As reported by Genkin et al., the specific power consumption is defined as the power necessary to oscillate a kilogram of mass with a given amplitude and frequency [21]. In our study, the power consumption of vibration of the motor at the different vibration frequencies and amplitudes was measured by a clamp metre (see Fig. 17(a)). As expected, the vibration power consumption increased with higher vibration frequencies and amplitudes. For example, with 8 mm amplitude and 10 Hz frequency vibration, 16.6 W power was consumed, which from observations shown earlier led to 95% fouling reduction. For comparison, with the same setup, we examined the use of aeration in a preliminary experiment and found that 5 L/min bubbling rate that led to 10% fouling reduction consumed 21 W power (Fig. 17(b)). Clearly, the vibration energy consumption was relatively low due to the fact that only the boundary fluid layers around the fibres were mobilized in the vibrating system. Thus, vibration can have a distinct advantage in terms of operating energy consumption. A similar benefit was also observed by Jaffrin and coworkers [13]. 5. Conclusions The present study confirms that mechanical vibration of vertical hollow fibres can improve the membrane filtration
performance with a feed solution of inorganic Bentonite particles by reducing the fouling on the membrane surface. The permeate flux increased with increasing vibration frequency and amplitude, which could be attributed to the thinning cake layer and lowering cake resistance. The introduction of a small membrane looseness of 1% was found to be able to significantly reduce the membrane fouling and improve the membrane permeate flux. The vibration of hollow fibres with and without the vibrating holding frame suggests that the turbulence generated by the vibrating holding frame played a minor role in enhancing the membrane performance in this study. PIV measurements further revealed that at higher vibration frequencies, the turbulence generated by the holding frame was more obvious. Finally, it was found that the operating power consumption with vibration is significantly less than aeration for a similar fouling rate, which can be due to the fact that only the boundary fluid layers around the fibres are mobilized and thus the energy dissipation is much reduced.
Acknowledgements The authors would like to gratefully acknowledge support from the Singapore National Research Foundation under its Environmental & Water Technologies Strategic Research Programme and administered by the Environment & Water Industry Programme Office (EWI) of the PUB. We also thank Prof. William Bill Krantz from University of Colorado for his helpful suggestions and Dr. Adrian Yeo from Singapore Membrane Technology Centre for his collaboration in the vibration setup.
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Nomenclature [3]
A ACH b c de dTMP/dt E F IA J Jw k MBR PAN PIV Rc Rc’ Ref Rev Rif Rm Rp Rt S SMBR t TMP, DP u u U0 v v V v VSEP
vibration amplitude aluminium chlorhydrate longitudinal intercept of Eq. (8) mass of cake deposited per unit volume of filtrate equivalent diameter of an annular vessel rate of TMP increase (or fouling rate) enhancement factor vibration frequency interrogation area Flux (L/(m2 h), LMH) clean water flux slope of Eq. (8) membrane bioreactor polyacrylonitrile particle image velocimetry hydraulic resistance attributed to the cake layer specific cake resistance per unit mass external fouling resistance vibrational Reynolds number internal fouling resistance hydraulic resistance of the membrane polarization layer resistance total resistance of the membrane effective membrane area submerged membrane bioreactors filtration time transmembrane pressure (kPa) average velocity of each grid position (x direction) in the contour map (m/s) average velocity of all grid positions (x direction) in the contour map (m/s) velocity deviation in the PIV contour map (m/s) average velocity of each grid position (y direction) in the contour map (m/s) average velocity of all grid positions (y direction) in the contour map (m/s) cumulative volume of permeate mean vibrational speed vibratory shear-enhanced processing system
Greek symbols
[4]
[5]
[6] [7] [8]
[9] [10]
[11]
[12] [13]
[14] [15]
[16]
[17]
[18]
[19]
[20] [21]
[22]
[23] [24]
r rs m n gmax dc |c
density of fluid solid density fluid dynamic viscosity fluid kinematic viscosity maximum shear rate cake thickness solid volume fraction
[25]
[26] [27] [28]
[29]
References [1] S. Delgado, R. Villarroel, E. Gonza´lez, Effect of the shear intensity on fouling in submerged membrane bioreactor for wastewater treatment, J. Membr. Sci. 311 (2008) 173–181. [2] A. Viero, T. Demelo, A. Torres, N. Ferreira, G. Santannajr, C. Borges, V. Santiago, The effects of long-term feeding of high organic loading in a submerged
[30] [31]
[32]
239
membrane bioreactor treating oil refinery wastewater, J. Membr. Sci. 319 (2008) 223–230. A. Viero, G. Santannajr, R. Nobrega, The use of polyetherimide hollow fibres in a submerged membrane bioreactor operating with air backwashing, J. Membr. Sci. 302 (2007) 127–135. B. Lesjean, S. Rosenberger, J. Schrotter, A. Recherche, Membrane-aided biological wastewater treatment—an overview of applied systems, Membr. Technol. 2004 (2004) 5–10. M.Y. Jaffrin, Dynamic shear-enhanced membrane filtration: a review of rotating disks, rotating membranes and vibrating systems, J. Membr. Sci. 324 (2008) 7–25. F. Wicaksana, A.G. Fane, V. Chen, Fibre movement induced by bubbling using submerged hollow fibre membranes, J. Membr. Sci. 271 (2006) 186–195. Z.F. Cui, S. Chang, A.G. Fane, The use of gas bubbling to enhance membrane processes, J. Membr. Sci. 221 (2003) 1–35. 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. L. Xia, A.W.K. Law, A.G. Fane, Optimal air sparging for hollow fiber membranes in a bubble column reactor. J. Membr. Sci., submitted for publication. A.D. Armando, B. Culkin, D.B. Purchas, New separation system extends the use of membranes, in: Proceedings of the Euromembrane’92, vol. 6, Lavoisier, Paris, 1992, 459–462. R. Bouzerar, L.H. Ding, M.Y. Jaffrin, Local permeate flux–sheer–pressure relationships in a rotating disk microfiltration module: implications for global performance, J. Membr. Sci. 170 (2000) 127–141. L.F. Liu, B. Gao, J.D. Liu, F.L. Yang, Rotating a helical membrane for turbulence enhancement and fouling reduction, Chem. Eng. J. 181 (2012) 486–493. O.A. Akoum, M.Y. Jaffrin, L. Ding, P. Paullier, C. Vanhoutte, An hydrodynamic investigation of microfiltration and ultrafiltration in a vibrating membrane module, J. Membr. Sci. 197 (2002) 37–52. S.C. Low, H.H. Juan, L.K. Siong, A combined VSEP and membrane bioreactor system, Desalination 183 (2005) 353–362. K.T. Takata, K. Yamamoto, R. Bian, Y. Watanabe, Removal of humic substances with vibratory shear enhanced processing membrane filtration, Desalination 117 (1998) 273–282. M. Frappart, O. Akoum, L. Ding, M. Jaffrin, Treatment of dairy process waters modelled by diluted milk using dynamic nanofiltration with a rotating disk module, J. Membr. Sci. 282 (2006) 465–472. M.D. Petala, A.I. Zouboulis, Vibratory shear enhanced processing membrane filtration applied for the removal of natural organic matter from surface waters, J. Membr. Sci. 269 (2006) 1–14. W. Shi, M.M. Benjamin, Membrane interactions with NOM and an adsorbent in a vibratory shear enhanced filtration process (VSEP) system, J. Membr. Sci. 312 (2008) 23–33. M.R. Bilad, G. Mezohegyi, P. Declerck, I.F. Vankelecom, Novel magnetically induced membrane vibration (MMV) for fouling control in membrane bioreactors, Water Res. 46 (2012) 63–72. H.G. Gomaa, S. Rao, Analysis of flux enhancement at oscillating flat surface membranes, J. Membr. Sci. 374 (2011) 59–66. G. Genkin, T. Waite, A. Fane, S. Chang, The effect of vibration and coagulant addition on the filtration performance of submerged hollow fibre membranes, J. Membr. Sci. 281 (2006) 726–734. S. Beier, M. Guerra, A. Garde, G. Jonsson, Dynamic microfiltration with a vibrating hollow fiber membrane module: filtration of yeast suspensions, J. Membr. Sci. 281 (2006) 281–287. 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. 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. A.W.K. Law, H. Wang, Measurement of mixing processes with combined digital particle image velocimetry and planar laser induced fluorescence, Exp. Therm. Fluid Sci. 22 (2000) 213–229. K.H. Choo, C.H. Lee, Membrane fouling mechanisms in the membranecoupled anaerobic bioreactor, Water Res. 30 (1996) 1771–1780. H.G. Gomaa, S. Rao, A.M. Al-Taweel, Intensification of membrane microfiltration using oscillatory motion, Sep. Purif. Technol. 78 (2011) 336–344. O. Akoum, M.Y. Jaffrin, L.H. Ding, Concentration of total milk proteins by high shear ultrafiltration in a vibrating membrane module, J. Membr. Sci. 247 (2005) 211–220. 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. F. Zamani, A.W.K. Law, A.G. Fane. Hydrodynamic analysis of vibrating hollow fibres membranes. J. Membr. Sci., submitted for publication. F. Vigo, C. Uliana, E. Ravina, A. Lucifredi, M. Gandoglia, The vibrating ultrafiltration module-performance in the 50–1000 Hz frequency-range, Sep. Sci. Technol. 28 (1993) 1063–1076. H.G. Gomaa, S. Rao, M.A. Taweel, Flux enhancement using oscillatory motion and turbulence promoters, J. Membr. Sci. 381 (2011) 64–73.
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Fouling control of submerged hollow fibre membranes by vibrations Tian Li a,b, Adrian Wing-Keung Law a,b,n, Merve Cetin a,b, A.G. Fane a,b a b
School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore Singapore Membrane Technology Centre, Nanyang Environment and Water Resource Institute, 1 Cleantech Loop, Clean Tech One, Singapore 637141, Singapore
a r t i c l e i n f o
a b s t r a c t
Article history: Received 7 May 2012 Received in revised form 10 September 2012 Accepted 15 September 2012 Available online 2 October 2012
In this study, we examine the improvement of fouling control of hollow fibre membranes with mechanical vibrations in the dead-end filtration of an inorganic suspension. Hollow fibres with diameters of 1.7 mm and 2 mm vibrating at moderate frequencies (0–15 Hz) and small amplitudes (0–12 mm) were submerged vertically in a 4 g/L Bentonite solution. Experiments were then conducted at both constant permeate flux and constant suction pressure conditions. The results showed that the membrane performance can be greatly improved when the vibration frequency or the vibration amplitude increases beyond a threshold magnitude. For example, over 90% reduction in the membrane fouling rate was achieved at 8 mm amplitude and 8 Hz frequency vibration compared to no vibration. Experiments were also conducted with 1% and 2% fibre looseness. The results showed that a small looseness can reduce the membrane fouling and increase the permeate flux under vibrations, which can be mainly attributed to the additional lateral movement of the fibres induced by the looseness. A comparison of vibrating the hollow fibres with and without the holding frame was also carried out to determine the effects of turbulence generated by the vibrating holding frame used in the experimental setup. Particle Image Velocimetry (PIV) measurements were performed to quantify the associated turbulence inside the membrane reactors. It was confirmed that the turbulence generated by the vibrating frame was more obvious at a high vibration frequency. However, it had little influence on the membrane filtration performance. Overall, the results from the present study confirm that at moderate frequencies, the cake layer resistance can be reduced substantially by vibration due to the dynamic shear enhancement on the membrane surface. & 2012 Elsevier B.V. All rights reserved.
Keywords: Vibration Hollow fibres Looseness Shear stress Particle image velocimetry
1. Introduction Hollow fibre membranes are now widely used in wastewater treatment processes. In particular, it is becoming the membrane of choice in submerged membrane bioreactors (SMBRs) [1–3] which can produce high quality water in a reduced reactor size and also minimize sludge production [4]. However, its application is still hindered by problems associated with concentration polarization and membrane fouling. For fouling mitigation, the use of hydrodynamic shear stresses on the membrane surface is recognized as one of the most effective techniques [5,6]. Hydrodynamic shear stresses can be generated on the membrane surface by moving either the fluid next to the membrane or the membrane surface itself. In the former case, the use of bubbling has been commonly adopted in SMBRs for industry applications. In addition to inducing wall shear stresses on the membrane surface [7], bubbling also n Corresponding author at: School of Civil and Environmental Engineering, Nanyang Technological University, 50 Nanyang Avenue, Singapore 639798, Singapore. Tel.: þ 65 67905296. E-mail address: [email protected] (A.W-K. Law).
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.09.031
oxygenates the fluid and disrupts the concentration polarization layer around the hollow fibres [8]. However, the shear stresses induced by bubbling are relatively weak. Also, the flux improvement reaches a limit beyond which further increase in the air flow rate becomes ineffective [9]. Vibrating the membrane module can also induce dynamic shear stresses on the membrane surfaces for fouling mitigation. The concept of Vibratory Shear-Enhanced Processing (VSEP) was first proposed by Armando et al. [10]. In their setup, a stack of circular organic membranes were separated by gaskets and permeate collectors, and mounted on a vertical torsion shaft spun in azimuthal oscillations by vibrating the base at a resonant frequency of about 60 Hz. The shear rate at the membrane surface was generated due to the inertia of the fluid which moved at 1801 out of phase with the membrane. Since then, commercial modules based on VSEP have been made available, with further development in the forms of rotating disks [11], rotating membranes [12], vibrating flat membranes [13], and vibrating hollow fibres [14]. Their applications are extensive, including the removal of humic substances [15], diluted milk [16], natural organic contaminants [17,18] as well as for wastewater treatment [19]. The uses of vibrating flat sheet membranes [20] and rotating disks [16] were
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
contribute to the reduction of membrane fouling [23]. We examined the turbulence induced by the holding frame vibration using the technique of Particle Image Velocimetry (PIV), and compared the filtration performance by the hollow fibres with and without the vibration of the holding frame. The results will be discussed in detail in the following after the description of the experimental setup and procedures.
also reported to be very promising, not only for much higher permeate fluxes, but also for better rejection. Recently, the possibility of using vibration for submerged hollow fibre membranes in MBR was raised by Low et al. [14]. They investigated different mechanical motions, including both cross oscillation and lengthwise oscillation. They found that cross oscillation was not effective because of unequalled shear forces on the membrane bundle, while the lengthwise oscillation provided alternating shear forces that are equal throughout the membrane surface. They concluded that the mechanical vibration helped maintain the MBR membrane in a relatively ‘‘clean’’ condition and keep the permeate flux close to that of the clean membrane. Subsequently, Genkin et al. evaluated the effect of vibration with a range of 0–10 Hz frequency and 0–40 mm amplitude and coagulant addition on the filtration performance of submerged hollow fibre membranes [21]. They discovered that both coagulation and transverse vanes could enhance the critical flux of the vibrating membrane unit. For example, at a vibration frequency of 1.7 Hz, the critical flux increased from 17 to 46 L/ (m2 h) with the addition of 34 mg/L Aluminium Chlorhydrate (ACH). In the presence of combined axial and transverse vibrations, a five-fold enhancement in critical flux to 86 L/(m2 h) was also achieved at 1.7 Hz in the presence of 34 mg/L ACH. They attributed the effect of coagulation to the aggregation of fine particles and evacuation of aggregates away from the membrane surface due to inertial and gravitational forces, and the effect of vanes due to the intensified turbulence arising from both longitudinal and transverse vibrations. Beier et al. [22] carried out experiments with a vibrating hollow fibre membrane module using baker yeast and also confirmed that the higher critical fluxes could be achieved through higher vibration frequency and amplitude. The objective of the present study is to further examine the different factors affecting the fouling mitigation of hollow fibre membranes with vibrations in an inorganic Bentonite solution. The scope is wider than the previous studies of Low et al. [14], Genkin et al. [21] and Beier et al. [22], includes various combinations of vibration frequencies, vibration amplitudes, and degrees of fibre looseness (previous studies involved only tightly held fibres). In addition, the frame that held the hollow fibre membranes could generate turbulence in the fluid upon vibration, which could also
Permeate pump
2. Experimental setup and procedures 2.1. Vibration setup The schematic diagram of the experimental setup is shown in Fig. 1. The test tank was made of Persplex with sizes of 400 mm (L) 400 mm (W) 1200 mm (H). The hollow fibre membrane module was submerged in an inorganic suspension made with Bentonite particles from Sigma-Aldrich. The Bentonite particles had a mean size of 5.8 mm, and the particle concentration was 4 g/L (pH 6.0–9.0). In the experiments, a typically small magnitude of air bubbling (50 mL/min) was maintained in the tank to keep the Bentonite particles suspended in the reactor. The holding frame of the membrane module was driven by a brushless DC motor with a crank moving mechanism. The vibration amplitude could be varied from 0 to 12 mm accurately, while the vibration frequency could be set to 0–15 Hz. The permeate flow was controlled by a master flex peristaltic pump together with a needle valve. The suction pressure was measured with a pressure transducer, and the permeate flux with a digital balance. Two different sizes of Polyacrylonitrile (PAN) hollow fibres made by Ultrapure Pte. Ltd. in Singapore with inner/outer diameters of 1 mm/1.7 mm and 1 mm/2 mm, respectively, were tested in the experiments. The length of each fibre was 40 cm, and the nominal pore size was 0.1 mm. A total of 13 fibres were included in the membrane modules. They were aligned vertically in parallel, with the distance between two adjacent fibres of 15 mm. Both ends of the fibres were connected to a 3 cm hard tubing with epoxy, and then mounted on the cross beam of the holding frame. Four screw nuts located at the border of the cross beam could be adjusted to change the submergence of the membrane module in the reactor.
Pressure transducer Valve
Digital balance
P
Frequency controller
Hollow fibre membranes
Nd: YAG laser
Camera Diffuser
Computer
231
Air pressure pump
Air flow meter
Fig. 1. Experimental setup.
Tank
Motor
232
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
2.2. Experimental setup for particle image velocimetry (PIV) The laser imaging technique of PIV has been used successfully before to measure the local velocities around a hollow fibre array [23,24]. With PIV, a laser sheet is projected onto the flow field at successive time intervals, and the subsequent capturing of images details the position of seeding particles that scatters the laser light [25]. A cross-correlation analysis of the changes in the particle position in the sequence of images reveals the Lagrangian velocity distribution. The details of PIV can be found in our earlier references [23–25]. In the present study, a Dantec flow map system was used for the PIV measurements. The illumination was from a dual-cavity frequency doubled Q-switched pulsed mini Nd:YAG laser. The energy level was 150 mJ and the pulse duration was 6 ns. The emitted light sheet was green, at a wavelength of 532 nm. The thickness of the light sheet was approximately 1 mm with a divergence angle of 321. A Hi-Sense Mk II camera was used to capture the images. The camera had a high-performance progressive scan interlined CCD chip. The chip included 1280 1024 light-sensitive cells and an equal number of storage cells. Seeding particles, with a diameter of 20 mm, were used as the particle markers. The concentration of the seeding particles was set at 0.1 g/L, to make sure that a sufficient number of particle markers were present within an interrogation area (IA). The data acquisition was performed at 6 Hz. The final interrogation cell size was 32 32 pixels with 50% overlap in IA, giving a grid of 83 63 for a total of 5229 velocity vectors. For every experiment, a total of 300 double frame images were typically taken to establish the velocity vector maps, from which the magnitude of the mean velocity and the turbulence fluctuations could be analyzed. 2.3. Experimental procedures Before each experiment, a clean water backwash at 20 mL/min was first performed on the hollow fibre membranes under the condition of 10 Hz vibration for 15 min. The purpose was to eliminate any bubbles trapped inside the hollow fibres. After the backwash, the filtration test was initiated, and the permeate volume was recorded at every 20 s interval by a digital balance. The permeate flux was then calculated as the rate of change of the permeate volume. Two types of experiments were conducted in this study, namely the constant flux and constant pressure experiments. They are described in the following. 2.3.1. Constant flux For the constant flux experiments, the permeate flux was typically maintained at 30 70.3 L/(m2 h), while the corresponding suction pressure was recorded using a pressure gauge at 1-min interval. Each experiment lasted for about 1 h. However, if the suction pressure decreased to around 40 kPa, many bubbles began to be generated from the tubing especially around the needle valve. When this occurred, the experiment was terminated and the duration of the particular experiment would then be shortened to be less than 1 h. To maintain consistency, we reported the average fouling rate of the first 30 min as the experimental fouling rate in this study. 2.3.2. Constant pressure In the constant pressure experiments, the suction pressure was controlled at 2470.1 kPa, while the permeate fluxes varied. Before each experiment, the membrane was backwashed in the same manner as in the constant flux experiments. Subsequently, the experiment would begin with the pump flow
increased so that the suction pressure reached 24 kPa. Over time, the permeate flux would typically decrease progressively which was determined by the measurements of permeate volume recorded at 20 s intervals using the digital balance. Each constant pressure experiment lasted for about 1 h. Since the pressure was maintained constant, the interruption due to air bubble generation as in the constant flux experiments did not occur. 2.3.3. Vibrating hollow fibres with and without holding frame As part of the overall experiments, the effects of holding frame vibration on the filtration performance of the hollow fibre membranes were also examined by comparing two test conditions: (a) vibrating hollow fibres, and (b) stationary hollow fibres. The former condition was used in most experiments. To study the stationary hollow fibres with the vibrating holding frame, the hollow fibres were separately fixed to another stationary holding frame while the original holding frame was driven by the motor. By comparing the two test conditions, the frame effect on the membrane performance could be evaluated.
3. Analysis of results 3.1. Membrane filtration The definition of permeate flux is J¼
1 dV S dt
ð1Þ
where S is the effective membrane area and V is the cumulative volume of permeate. According to the resistance-in-series model [26], the relationship between the permeate flux (J) and transmembrane pressure (DP) can be expressed in the following form: ð2Þ J ¼ DP= mRt Rt ¼ Rm þ Rp þ Ref þ Rif
ð3Þ
where m is the dynamic viscosity of the permeate; Rt is the total resistance; Rm is the intrinsic membrane resistance; Rp is the polarization layer resistance caused by the concentration gradient; Ref is the external fouling resistance formed by a deposited cake layer from physic–chemical interactions of solids with the membrane surface; and Rif the internal fouling resistance due to irreversible adsorption or pore plugging. For clean water, Eq. (2) can be simplified as ð4Þ Jw ¼ DP= mRm where Jw is the clean water flux. In the present study, a clean water permeability (Flux/TMP) test was conducted before each experiment for both the 1.7 mm and 2 mm fibres. Their permeabilities were typically found to be about 2.04 L/(m2 h kPa) and 1.73 L/(m2 h kPa) respectively. Since the clean water viscosity at 20 1C (the typical temperature in the experiments) was 1.005 10 3 Pa s, thus Rm for the 1.7 mm and 2 mm fibres can be calculated to be 1.76 1012 m 1 and 2.07 1012 m 1 respectively. For the Bentonite solution, the nominal particle size of the powder was much larger than the pore size of the hollow fibre. Hence, Eq. (2) can be rearranged as J¼
DP
mðRm þ Rc Þ
ð5Þ
where Rc is the cake resistance (Rc ¼Ref þRif), which for hollow fibre membranes should also vary with the cake thickness dc as [27], Rc ¼ Rc 0 rs |c dc
ð6Þ
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
233
where Rc0 is the specific cake resistance per unit mass, rs the solid density, and |c the solid volume fraction. The mass of cake deposited per unit volume of filtrate, c, can then be expressed as c¼
Sdc |c rs V
ð7Þ
Therefore, for an incompressible cake, integrating (1) with constant DP gives [27] t mRc 0 c mRm ¼ 2 Vþ V SDP 2S DP
ð8Þ
Eq. (8) quantifies the filtration characteristics when a cake layer is formed. 3.2. Estimation of filtration parameters For each experiment, the permeate volume was recorded continuously, and the permeate flux was computed with Eq. (1). By plotting the permeate volume versus time, the longitudinal intercept of Eq. (8) can be obtained as b¼
mRm SDP
Fig. 2. Fouling rate at different permeate fluxes without vibration.
ð9Þ
The membrane resistance Rm can then be determined as Rm ¼
b
m
SDP
ð10Þ
The slope of Eq. (8) can be defined as k¼
mRc 0 c 2S2 DP
ð11Þ
The cake resistance Rc can be calculated from this slope together with Eqs. (6) and (7) as: Rc ¼
2k
m
SDPV
ð12Þ
Examples of the calculation can be found in the later part of results and discussion. 3.3. Vibration enhancement factor Membrane oscillation can potentially lead to an improvement in the fouling mitigation. The enhancement factor, E, in terms of the permeate flux can be used to describe the vibration intensification as follows: ðJÞwith oscillation ð13Þ E¼ ðJÞwithout oscillation t ¼ 1 h The flux values were taken after 1 h which corresponded to the typical time needed to reach the quasi-steady state conditions. This approach has been widely used for membrane vibration filtration analysis [20,28]. In this study, we also use this parameter to evaluate the membrane filtration improvement with vibrations.
4. Results and discussion 4.1. Filtration characteristics Fig. 2 shows the measured fouling rate with different constant permeate fluxes for both the 1.7 mm and 2 mm fibres without vibration. It was obvious that the fouling rate increased as the fibre diameter decreased, which may be attributed to the higher axial flux due to the smaller radius of the 1.7 mm fibres. For example, the fouling rate of the 1.7 mm fibres at a permeate flux over 25 LMH was double of that for the 2 mm fibres. It should be noted that a similar effect (greater fouling for smaller diameter
Fig. 3. Fouling rate versus permeate flux at different vibration frequencies (vibration amplitude¼5 mm, 2 mm fibre).
fibres) was also observed elsewhere [29]. In addition, larger permeate fluxes led to much higher fouling rates, as illustrated by the fact that the fouling rate at a constant permeate flux of 30 LMH was 91.2 kPa/h and 48.6 kPa/h respectively for the 1.7 mm and 2 mm fibres, which was much higher than at a flux of 10 LMH where nearly no fouling was observed. This could be attributed to the increased cake compression and consolidation at higher fluxes. Fig. 3 shows the measured fouling rates with different vibration frequencies, at a fixed vibration amplitude of 5 mm for the 2 mm fibres (note that 0 Hz in the figure represents no vibration). The fouling rate typically decreased significantly when vibration was applied. It dropped from 48.6 kPa/h to 0.8 kPa/h at a permeate flux of 30 LMH with vibration frequency increased from 0 Hz to 10 Hz. It was also found that, with 2 Hz vibration, the suction pressure increased from 0.2 kPa/h to 15 kPa/h when the permeate flux increased from 10 LMH to 30 LMH. The same trend can be observed for other vibration frequencies. This implies more serious membrane fouling occurred when a larger amount of fluid permeated through the membrane as expected. In addition, almost no fouling was observed with 10 Hz vibration for the 2 mm fibres. In other words, vibrating the hollow fibre membrane at 10 Hz increased the critical flux for the 2 mm fibres to beyond 30 LMH in this case. Fig. 4 plots the fouling rate at different vibration frequencies and amplitudes for both the 1.7 mm and 2 mm fibres, at a constant permeate flux of 30 LMH. The enhancement by vibration intensified when both vibration amplitude and frequency
234
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 4. Fouling rate at different vibration frequencies and amplitudes (constant permeate flux¼ 30 LMH): (a) 2mm fibre and (b) 1.7mm fibre.
increased. There was an 85% reduction in the fouling rate when the 2 mm hollow fibres were vibrated with 5 mm amplitude and 5 Hz frequency, comparing to no vibration (Fig. 4(a)). Similar enhancement was obtained for the 1.7 mm fibres at 6 mm amplitude and 8 Hz frequency (Fig. 4(b)). When the vibration amplitude increased to 8 mm with the frequency of 10 Hz, more than 95% fouling rate reduction occurred for both the 1.7 mm and 2 mm fibres. When the vibration amplitude further increased to 10 mm with frequency of 12 Hz, nearly no fouling was observed for either fibre. The plot of t/V versus V and vibration frequency F is presented in Fig. 5 with the cake filtration law. The time needed to obtain the same amount of permeate was longer at lower vibration frequencies, implying that vibration could improve the permeate flux. The membrane resistance was calculated based on Eq. (10). For example, for the 2 mm fibre without vibration (Fig. 5(a)) the results showed that the longitudinal intercept b¼2.63 103 and the slope k¼2.98 103. Thus, the average membrane resistance Rm was 2.26 1012 m 1 by Eq. (10), and the relationship between Rc (1012 m 1) and V (L) was Rc ¼5.12 V by Eq. (12). A similar method was used for the 1.7 mm fibres. The average membrane resistances Rm was 1.97 1012 m 1 and the relationship between Rc (1012 m 1) and V (L) was Rc ¼ 8.33 V. The average membrane resistance calculated by the cake filtration law analysis was nearly the same as that from the clean water permeability tests, which illustrated the consistence of the test results. Fig. 6 plots the cake resistance versus cumulative permeate volume according to the cake filtration law. It can be observed that the cake resistance increased gradually with time at different vibration frequencies for both 1.7 mm and 2 mm hollow fibres. Also, the cake resistance of 1.7 mm fibres without vibration was much
Fig. 5. Evolution of t/V versus cumulative permeate volume at different vibration frequencies with cake filtration law identification (constant suction pressure¼ 24 kPa): (a) 2mm fibre, vibration amplitude¼ 5mm and (b) 1.7mm fibre, vibration amplitude¼ 6mm.
larger than 2 mm fibres due to the smaller diameter which would lead to greater effective fouling [29]. It reduced from 7.86 1012 m 1 to 0.41 1012 m 1 when the vibration frequency of 10 Hz was applied compared to no vibration for the 2 mm fibres at a 1.5 L permeate volume. Hence, a 95% cake resistance reduction was achieved by the 10 Hz vibration. Comparatively, 75% cake resistance reduction at 6 Hz vibration and 83% cake resistance reduction at 10 Hz vibration were obtained for the 1.7 mm fibres, which was also significant but nevertheless less than the 2 mm fibres. In summary, Figs. 3–6 all show improved performance of the membrane under vibration, and the improvement was very significant when a certain level of vibration frequency and/or vibration amplitude was reached. This could be attributed to the fact that at higher vibration frequencies, the particles could be easily shaken loose from the membrane surface under conditions of high shear stress. 4.2. Intensification effects of vibration For all the experiments, the average permeate flux was determined numerically after 1 h, which was typically the time needed to reach a quasi-steady state. The membrane vibrations resulted in an increase of this quasi-steady state permeate flux, with the enhancement factor dependent on the applied vibration amplitude and frequency as well as the TMPs. In general, the enhancement factor would be larger with higher vibration amplitudes and frequencies at certain TMPs. Fig. 7 depicts the effect of vibration enhancement factor for both the 1.7 mm and 2 mm fibres. Vibration enhancement faster increased from 1.5 to 2.6 when the vibration frequency increased
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
235
Fig. 6. Evaluation of cake resistance Rc versus cumulative permeate volume at different vibration frequencies (constant suction pressure¼ 24 kPa): (a) 2mm fibre, vibration amplitude 5mm and (b) 1.7mm fibre, vibration amplitude 6mm. Fig. 7. Vibration enhancement factor versus vibration frequency at different vibration amplitudes (constant suction pressure¼ 24 kPa): (a) 2mm fibre and (b) 1.7mm fibre.
from 2 Hz to 10 Hz respectively with the vibration amplitude of 8 mm for the 2 mm fibres, while it was 2.2 times the quasi-steady state permeate flux at 8 mm vibration amplitude and 10 Hz vibration frequency for the 1.7 mm fibres. This could potentially imply that the larger fibre size could perform better in a vibrating system. The enhancement could again be attributed to the reduction of the deposited cake layer at the membrane surface due to the vibration shear at the membrane surface and the subsequent decrease in the cake resistance to the filtration (Fig. 6).
4.3. Effect of membrane looseness Fig. 8 shows the quasi-steady state permeate flux for the tight, 1% and 2% looseness fibres. The quasi-steady state flux increased with an increase of vibration frequency and was higher for the loose fibres. With 1% looseness, it increased 20% from 17 LMH to 21 LMH at 4 Hz, and 30% at 10 Hz. However, very little improvement was noted for all the vibration frequencies when the fibre looseness increased from 1% to 2%. The vibration enhancement effect is further shown in Fig. 9, and a similar trend is observed. The enhancement factor was higher with a larger vibration frequency. It increased from 1.6 to 2 at 6 Hz and from 2 to 2.5 at 8 Hz with 1% looseness. The factor reached as high as 3 with 1% looseness and 2.2 with the tight fibres at 10 Hz. The results confirmed that the fibre looseness can improve the membrane performance with vibrations due to axial vibrational shear stresses and the extra lateral movement induced by the looseness, and only a looseness of 1% would be required to realize the improvement.
Fig. 8. Quasi-steady state flux for tight and loose fibres at different vibration frequencies (constant suction pressure ¼ 24 kPa, 1.7 mm fibre).
Fig. 10 shows the fouling rate of tight and loose fibres at 6–12 Hz vibration frequency. The fouling rate decreased significantly for tight and loose fibres in this frequency range. It dropped from 29.4 kPa/h to 2 kPa/h from 6 Hz to 12 Hz for the tight fibres, and from 8.9 kPa/h to 1.4 kPa/h for the 1% looseness fibres, respectively. Again the fouling rate for the 1% and 2% looseness was quite similar, confirming that the fibre looseness can decrease the fouling rate, and 1% looseness is sufficient to improve the membrane filtration performance. The improvement
236
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 9. Vibration enhancement factor for tight and loose fibres at different vibration frequencies (constant suction pressure¼ 24 kPa, 1.7 mm fibre).
Fig. 11. Fouling rate for vibrating and stationary fibres with vibrating holding frame at different vibration frequencies (constant permeate flux¼ 30 LMH, vibration amplitude ¼10 mm, 1.7 mm fibre).
Fig. 10. Fouling rate for tight and loose fibres at different vibration frequencies (constant permeate flux¼ 30 LMH, vibration amplitude¼ 6 mm, 1.7 mm fibre).
can be attributed to the additional lateral movement of the loose fibres and the resulting increase in shear stress. Similar effects were also noted in the bubbling studies by Yeo et al. [24] and Wicaksana et al. [6]. However, it is probable that excessive fibre looseness would lead to fibre breakage due to extreme movement of fibres. 4.4. Effect of fibre holding frame Fig. 11 shows the fouling rate at constant flux of the stationary and vibrating fibres with the vibrating holding frame. It was obvious that the fouling rate of the stationary fibres was much higher than the vibrating fibres together with the vibration of the holding frame, decreasing from 85.8 kPa/h to 41.5 kPa/h at 6 Hz, and from 79.6 kPa/h to 2.2 kPa/h at 10 Hz. With the stationary hollow fibres, the fouling rate was a little lower with the vibrating holding frame than no vibration. In Fig. 12, a similar trend can also be observed at constant pressure. The quasi-steady state flux of the vibrating fibres with the vibrating holding frame increased 80% from 17 LMH at 4 Hz to 31 LMH at 10 Hz. Overall, the results showed that the vibration of the holding frame and the associated turbulence did not affect the membrane performance significantly. Thus, it can be concluded that the vibration of the fibres to produce shear at the membrane surface is the primary factor that led to the reduction of membrane fouling. In the present study, the current results were different from Genkin et al. [21] where they intentionally used a special design of the holding
Fig. 12. Quasi-steady state flux of vibrating and stationary fibres with vibrating holding frame (constant suction pressure¼ 24 kPa, vibration amplitude¼ 8 mm, 1.7 mm fibre).
frame to generate turbulence in order to markedly enhance the critical flux of the membrane. However, the present data allows a direct examination of the role of dynamic vibration shear on the membrane filtration. 4.5. Dynamic shear rate Previous researchers had shown that the dynamic shear stress on the membrane surface during vibration could reduce the membrane fouling [5,13]. For the vibrating disk, Akoum et al. [13] suggested the relationship between the maximum shear stress and vibration frequency and amplitude to be
gmax ¼ ð2pFÞ 3=2 An1=2
ð14Þ
For simplicity, Genkin et al. [21] assumed that the same expression could be used to describe the shear rate at the surface of a vibrating fibre. However, the omission of the consideration of radius of curvature of the fibre can lead to significant errors with small diameter fibres as pointed out by Zamani et al. [30]. In this particular study, however, the fibre diameters at 1.7 mm and 2 mm were relatively large, and thus the errors are expected to be small. Experiments with the same theoretical constant shear rate (A*F1.5 ¼134) and with the amplitude varied from 4 mm to 12 mm
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
237
Fig. 14. Influence of the shift speed (4A*F) on fouling rate (constant permeate flux¼ 30 LMH, 1.7 mm fibre). Fig. 13. Fouling rate with the same theoretical shear rate (constant permeate flux¼ 30 LMH, 1.7 mm fibre).
Rev ¼ de V r=m
63cm
2.5cm
30.5cm
were conducted. The results are shown in Fig. 13. The fouling rate increased from 10.8 kPa/h to 18.8 kPa/h when the vibration amplitude increased from 4 mm to 9 mm. It then gradually decreased to 16 kPa/h when the vibration amplitude increased from 9 mm to 12 mm. This shows that the amplitude of 9 mm performed the worst under the same shear rate. The results demonstrated that at higher vibration frequencies, the power ratio of vibration frequency in Eq. (14) can be larger than 1.5. It can also be deduced that the vibration frequency could have much more impact than the vibration amplitude for fouling mitigation, which confirms that vibration frequency is more effective to improve membrane performance than vibration amplitude. Vigo and Uliana [31] suggested that the shear stress was dependent on the shift speed (4A*F) of the membrane surface in a vibrating system. The product 4A*F was a quantity proportional to the so-called vibrational Reynolds number as:
25cm
3cm
12cm
ð15Þ Fig. 15. Location of the image window in the reactor.
where de was the equivalent diameter of an annular vessel, and V the mean vibrational speed ( ¼4A*F). Experiments with the same shift speed were performed in this study with the vibration amplitude ranging from 4 mm to 12 mm (Fig. 14). The results showed that the fouling rate increased from 7.4 kPa/h to 66.7 kPa/h when A*F ¼48. Thus, the fouling rate climbed rapidly with increasing vibration amplitude at the same shift speed, i.e., the membrane performance was strongly affected by the increasing vibration frequency. This influence could be attributed to the much higher maximum shear stress at the membrane surface when the fibre was moving relative to the fluid [31]. Fig. 14 also demonstrates that the effect of vibration frequency was more obvious in a vertical vibrating system.
4.6. PIV measurement With PIV, the two-dimensional velocity distribution can be obtained in the form of vector maps, while the turbulence characteristics can be computed by evaluating the local vector values. For the vibration experiments in our study, the velocity components of 300 vector maps were obtained and the average velocity components u and v were calculated.
In this study, we carried out PIV measurements in the area shown in Fig. 15 just in front of the fibres and plotted the contour map of the velocity deviation U’ from the mean velocity at different frequencies in Fig. 16, where U0 ¼
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 uu þ vv
ð16Þ
From figures, it can be observed that the value of U’ and the generation of turbulence by the holding frame was much more significant at higher vibration frequencies. The induced turbulence can potentially reduce the cake fouling layer on the membrane surface, thus increase the apparent permeability. Gomaa et al. [32] investigated the flux enhancement using oscillatory motion and turbulence promoters. They revealed that with higher frequencies (F45 Hz), the effect of turbulence promoters extended further away from the surface, resulting in a well-mixed flow structure and scouring of the surface between the promoters caused by vortices shedding. The effects of turbulence promoters in Gomaa et al. [32] are similar to the turbulence produced by the holding frame in the present study. However, as
238
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Fig. 17. A comparison of operating power consumption between (a) vibration and (b) aeration. Fig. 16. Contour of U’ at different vibration frequencies (no suction pressure, 2 mm fibre): (a) 2Hz and (b) 10Hz.
discussed earlier, the effect here of the holding frame per se was not significant due to the different structural configurations. 4.7. Power consumption comparison with bubbling As reported by Genkin et al., the specific power consumption is defined as the power necessary to oscillate a kilogram of mass with a given amplitude and frequency [21]. In our study, the power consumption of vibration of the motor at the different vibration frequencies and amplitudes was measured by a clamp metre (see Fig. 17(a)). As expected, the vibration power consumption increased with higher vibration frequencies and amplitudes. For example, with 8 mm amplitude and 10 Hz frequency vibration, 16.6 W power was consumed, which from observations shown earlier led to 95% fouling reduction. For comparison, with the same setup, we examined the use of aeration in a preliminary experiment and found that 5 L/min bubbling rate that led to 10% fouling reduction consumed 21 W power (Fig. 17(b)). Clearly, the vibration energy consumption was relatively low due to the fact that only the boundary fluid layers around the fibres were mobilized in the vibrating system. Thus, vibration can have a distinct advantage in terms of operating energy consumption. A similar benefit was also observed by Jaffrin and coworkers [13]. 5. Conclusions The present study confirms that mechanical vibration of vertical hollow fibres can improve the membrane filtration
performance with a feed solution of inorganic Bentonite particles by reducing the fouling on the membrane surface. The permeate flux increased with increasing vibration frequency and amplitude, which could be attributed to the thinning cake layer and lowering cake resistance. The introduction of a small membrane looseness of 1% was found to be able to significantly reduce the membrane fouling and improve the membrane permeate flux. The vibration of hollow fibres with and without the vibrating holding frame suggests that the turbulence generated by the vibrating holding frame played a minor role in enhancing the membrane performance in this study. PIV measurements further revealed that at higher vibration frequencies, the turbulence generated by the holding frame was more obvious. Finally, it was found that the operating power consumption with vibration is significantly less than aeration for a similar fouling rate, which can be due to the fact that only the boundary fluid layers around the fibres are mobilized and thus the energy dissipation is much reduced.
Acknowledgements The authors would like to gratefully acknowledge support from the Singapore National Research Foundation under its Environmental & Water Technologies Strategic Research Programme and administered by the Environment & Water Industry Programme Office (EWI) of the PUB. We also thank Prof. William Bill Krantz from University of Colorado for his helpful suggestions and Dr. Adrian Yeo from Singapore Membrane Technology Centre for his collaboration in the vibration setup.
T. Li et al. / Journal of Membrane Science 427 (2013) 230–239
Nomenclature [3]
A ACH b c de dTMP/dt E F IA J Jw k MBR PAN PIV Rc Rc’ Ref Rev Rif Rm Rp Rt S SMBR t TMP, DP u u U0 v v V v VSEP
vibration amplitude aluminium chlorhydrate longitudinal intercept of Eq. (8) mass of cake deposited per unit volume of filtrate equivalent diameter of an annular vessel rate of TMP increase (or fouling rate) enhancement factor vibration frequency interrogation area Flux (L/(m2 h), LMH) clean water flux slope of Eq. (8) membrane bioreactor polyacrylonitrile particle image velocimetry hydraulic resistance attributed to the cake layer specific cake resistance per unit mass external fouling resistance vibrational Reynolds number internal fouling resistance hydraulic resistance of the membrane polarization layer resistance total resistance of the membrane effective membrane area submerged membrane bioreactors filtration time transmembrane pressure (kPa) average velocity of each grid position (x direction) in the contour map (m/s) average velocity of all grid positions (x direction) in the contour map (m/s) velocity deviation in the PIV contour map (m/s) average velocity of each grid position (y direction) in the contour map (m/s) average velocity of all grid positions (y direction) in the contour map (m/s) cumulative volume of permeate mean vibrational speed vibratory shear-enhanced processing system
Greek symbols
[4]
[5]
[6] [7] [8]
[9] [10]
[11]
[12] [13]
[14] [15]
[16]
[17]
[18]
[19]
[20] [21]
[22]
[23] [24]
r rs m n gmax dc |c
density of fluid solid density fluid dynamic viscosity fluid kinematic viscosity maximum shear rate cake thickness solid volume fraction
[25]
[26] [27] [28]
[29]
References [1] S. Delgado, R. Villarroel, E. Gonza´lez, Effect of the shear intensity on fouling in submerged membrane bioreactor for wastewater treatment, J. Membr. Sci. 311 (2008) 173–181. [2] A. Viero, T. Demelo, A. Torres, N. Ferreira, G. Santannajr, C. Borges, V. Santiago, The effects of long-term feeding of high organic loading in a submerged
[30] [31]
[32]
239
membrane bioreactor treating oil refinery wastewater, J. Membr. Sci. 319 (2008) 223–230. A. Viero, G. Santannajr, R. Nobrega, The use of polyetherimide hollow fibres in a submerged membrane bioreactor operating with air backwashing, J. Membr. Sci. 302 (2007) 127–135. B. Lesjean, S. Rosenberger, J. Schrotter, A. Recherche, Membrane-aided biological wastewater treatment—an overview of applied systems, Membr. Technol. 2004 (2004) 5–10. M.Y. Jaffrin, Dynamic shear-enhanced membrane filtration: a review of rotating disks, rotating membranes and vibrating systems, J. Membr. Sci. 324 (2008) 7–25. F. Wicaksana, A.G. Fane, V. Chen, Fibre movement induced by bubbling using submerged hollow fibre membranes, J. Membr. Sci. 271 (2006) 186–195. Z.F. Cui, S. Chang, A.G. Fane, The use of gas bubbling to enhance membrane processes, J. Membr. Sci. 221 (2003) 1–35. 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. L. Xia, A.W.K. Law, A.G. Fane, Optimal air sparging for hollow fiber membranes in a bubble column reactor. J. Membr. Sci., submitted for publication. A.D. Armando, B. Culkin, D.B. Purchas, New separation system extends the use of membranes, in: Proceedings of the Euromembrane’92, vol. 6, Lavoisier, Paris, 1992, 459–462. R. Bouzerar, L.H. Ding, M.Y. Jaffrin, Local permeate flux–sheer–pressure relationships in a rotating disk microfiltration module: implications for global performance, J. Membr. Sci. 170 (2000) 127–141. L.F. Liu, B. Gao, J.D. Liu, F.L. Yang, Rotating a helical membrane for turbulence enhancement and fouling reduction, Chem. Eng. J. 181 (2012) 486–493. O.A. Akoum, M.Y. Jaffrin, L. Ding, P. Paullier, C. Vanhoutte, An hydrodynamic investigation of microfiltration and ultrafiltration in a vibrating membrane module, J. Membr. Sci. 197 (2002) 37–52. S.C. Low, H.H. Juan, L.K. Siong, A combined VSEP and membrane bioreactor system, Desalination 183 (2005) 353–362. K.T. Takata, K. Yamamoto, R. Bian, Y. Watanabe, Removal of humic substances with vibratory shear enhanced processing membrane filtration, Desalination 117 (1998) 273–282. M. Frappart, O. Akoum, L. Ding, M. Jaffrin, Treatment of dairy process waters modelled by diluted milk using dynamic nanofiltration with a rotating disk module, J. Membr. Sci. 282 (2006) 465–472. M.D. Petala, A.I. Zouboulis, Vibratory shear enhanced processing membrane filtration applied for the removal of natural organic matter from surface waters, J. Membr. Sci. 269 (2006) 1–14. W. Shi, M.M. Benjamin, Membrane interactions with NOM and an adsorbent in a vibratory shear enhanced filtration process (VSEP) system, J. Membr. Sci. 312 (2008) 23–33. M.R. Bilad, G. Mezohegyi, P. Declerck, I.F. Vankelecom, Novel magnetically induced membrane vibration (MMV) for fouling control in membrane bioreactors, Water Res. 46 (2012) 63–72. H.G. Gomaa, S. Rao, Analysis of flux enhancement at oscillating flat surface membranes, J. Membr. Sci. 374 (2011) 59–66. G. Genkin, T. Waite, A. Fane, S. Chang, The effect of vibration and coagulant addition on the filtration performance of submerged hollow fibre membranes, J. Membr. Sci. 281 (2006) 726–734. S. Beier, M. Guerra, A. Garde, G. Jonsson, Dynamic microfiltration with a vibrating hollow fiber membrane module: filtration of yeast suspensions, J. Membr. Sci. 281 (2006) 281–287. 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. 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. A.W.K. Law, H. Wang, Measurement of mixing processes with combined digital particle image velocimetry and planar laser induced fluorescence, Exp. Therm. Fluid Sci. 22 (2000) 213–229. K.H. Choo, C.H. Lee, Membrane fouling mechanisms in the membranecoupled anaerobic bioreactor, Water Res. 30 (1996) 1771–1780. H.G. Gomaa, S. Rao, A.M. Al-Taweel, Intensification of membrane microfiltration using oscillatory motion, Sep. Purif. Technol. 78 (2011) 336–344. O. Akoum, M.Y. Jaffrin, L.H. Ding, Concentration of total milk proteins by high shear ultrafiltration in a vibrating membrane module, J. Membr. Sci. 247 (2005) 211–220. 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. F. Zamani, A.W.K. Law, A.G. Fane. Hydrodynamic analysis of vibrating hollow fibres membranes. J. Membr. Sci., submitted for publication. F. Vigo, C. Uliana, E. Ravina, A. Lucifredi, M. Gandoglia, The vibrating ultrafiltration module-performance in the 50–1000 Hz frequency-range, Sep. Sci. Technol. 28 (1993) 1063–1076. H.G. Gomaa, S. Rao, M.A. Taweel, Flux enhancement using oscillatory motion and turbulence promoters, J. Membr. Sci. 381 (2011) 64–73.