Journal of Membrane Science 498 (2016) 315–323
Contents lists available at ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Ribbed PVC–silica mixed matrix membranes for membrane bioreactors Lisendra Marbelia a, Muhammad Roil Bilad a, Nils Bertels a, Carole Laine b, Ivo F.J. Vankelecom a,n a b
Centre for Surface Chemistry and Catalysis, Faculty of Bioscience Engineering, KU Leuven, Kasteelpark Arenberg 23, Box 2461, 3001 Leuven, Belgium Amer-Sil S.A., Zone Industrielle, L-8287 Kehlen, Luxembourg
art ic l e i nf o
a b s t r a c t
Article history: Received 1 June 2015 Received in revised form 22 September 2015 Accepted 6 October 2015 Available online 13 October 2015
The feasibility of ribbed-PVC–silica mixed matrix membranes (MMM) in a membrane bioreactor (MBR) application was investigated. Three operational aspects were evaluated: inclusion of vibration, use of relaxation (cycle time), and optimization of the cleaning procedure. Results show that the specific ribs on the membranes give a 52% higher surface area compared to the corresponding flat membranes, offering two main benefits: i.e. a higher clean water permeability (increased by 70%) and an increased critical flux (increased by 30%) for activated sludge filtration. Applying vibration increased the critical fluxes about 30–50%. In aeration mode, a filtration time of 8 min and a relaxation time of 1 min are found to keep both reversible and residual fouling minimal. Despite these advantages, the increase of the critical flux and the filtration performance (i.e. fouling resistance) are not completely in proportion with the increase of the surface area. Furthermore, a lower cleaning efficiency for the ribbed membrane was found. These results indicate that the ribs on the membrane of the current system are still not optimized to additionally work to promote local shear near the membrane surface and as turbulence promoter in the vibration mode. & 2015 Elsevier B.V. All rights reserved.
Keywords: PVC–silica membranes MMM MBR Ribs Vibration
1. Introduction Membrane technology recently has gained more attention in wastewater treatment for removing colloidal and suspended solids. Membrane bioreactors (MBRs) are among the most popular applications of micro or ultrafiltration in biological wastewater treatment, where they separate the active biomass in the activated sludge from the treated water [1]. Different membrane materials have been used in MBRs. Most full-scale MBR plants apply phase inverted polymeric membranes [2,3]. These polymers include polyvinylidene fluoride (PVDF), polyethersulfone (PES), polyethylene (PE), polypropylene (PP), polysulfone (PS), and polyacrylonitrile (PAN) [2,4,5]. Membrane materials and their surface properties, such as hydrophilicity, roughness, surface porosity and pore size, are well known to have a role on membrane fouling [6–8]. Despite the fact that MBRs have become a mature technology, researches on membrane materials, specifically to reduce membrane fouling are still going on. Different approaches to modify membranes have been undertaken in order to reduce and control the fouling [9,10]. Recently, surface-patterned membranes, particularly on micro- and nanometer scale have proven to be able to mitigate fouling in cross flow filtration system [11–14]. Local shear n
Corresponding author. Fax: þ 32 16 321998. E-mail address:
[email protected] (I.F.J. Vankelecom).
http://dx.doi.org/10.1016/j.memsci.2015.10.017 0376-7388/& 2015 Elsevier B.V. All rights reserved.
is promoted in the patterned membranes so that bacterial deposition are less favorable [12]. In addition to that, surface patterns also give lower local fluxes due to the higher active surface area compared to flat membranes [13,15]. In addition to membranes properties, operating conditions, such as permeate flux, aeration, and filtration mode (filtration, backwash, and relaxation cycle) also influence the occurrence of membrane fouling [16–18] and its reversibility [19]. Moreover, the application of shear-enhanced or dynamic filtration, such as rotating and vibrating membranes, is also known to reduce fouling and thus enhance filterability [20,21]. Fixing a baffle on the membrane surface can alter the fluid hydrodynamics and acts as a turbulence promoter to enhance the filtration process [10]. A combination of turbulence promoter and vibration was recently studied and found to enhance filterability during the filtration of a baker's yeast suspension [22]. Present study is a continuation of our previous feasibility study where the PVC–silica mixed matrix membrane (MMM) [23] was screened for use in MBRs. In the present study, instead of using the flat membranes [24], the performance of ribbed membrane was evaluated for activated sludge filtration. These MMM are produced by a patented low temperature extrusion process. For the ribbed membranes, the ribbed patterns on both surfaces are given directly by the calendar rolls just after the extrusion [25]. The ribs on the permeate side allows production of a spacer-free panel, while the ones outside might act as turbulence promoter.
316
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
The aims of the present work were to understand and define the optimum operational parameters that should be applied for the ribbed membranes, i.e. the use of vibration, different filtration mode (cycle time of relaxation), and cleaning procedures for improving the competitiveness of PVC–silica MMM as membrane for MBRs. In the end, the comparison with commercial flat-sheet MBR membranes is made in order to better evaluate the feasibility of the ribbed membranes for bigger scale MBR applications.
activated sludge in the MBR slightly fluctuated, with total solid (TS) 5.72 70.87 gL 1 (for experiments in Section 3.2.1) and 3.237 0.47 gL 1 (for experiments in Section 3.2.2) respectively. The MBR was able to remove the COD with a corresponding COD removal above 90%. The values of the COD, total nitrogen, and phosphates can be found in the supplementary information. The aeration in the MBR was kept constant around a flow rate of 10 L/ min. 2.3. Filtration experiments
2. Experimental methods 2.1. Membrane preparation and characterization Membranes, both flat and ribbed, were supplied by Amer-Sil. The properties of the membranes can be found in Table 1. The microstructures of pristine membranes were observed with scanning electron microscopy (SEM, Philips SEM XL30 FEG). Membrane panels with an effective area of 0.012 m2 were prepared following the method by Bilad et al. [26] with a modification for the ribbed membranes. For the flat membranes, a certain size of membrane were cut, fixed to a PVC frame, glued at the edge (UHU-Plus endfest 300, Germany), folded like an envelope after putting spacers and a permeate pipe in between, and left overnight to complete drying process. For the ribbed ones, spacers were not needed in between the membranes as they already have a ribbed pattern on the surface which allows to separate the membrane active area (Fig. 1). Clean water permeability of the pristine membrane panels was measured by filtrating deionized water. The panel was connected to a pump and pressure gauge and immersed in a tank containing demineralized water from where permeate is extracted.
Each membrane panel was immersed in the MBR and sucked for a certain flux through individual permeate line. Each permeate line is equipped with a pressure gauge (Wika, Germany) which was further connected to a multichannel peristaltic pump (Watson-Marlow, UK). The transmembrane pressures showed by the gauge were recorded and the filtration performances were then compared. Membrane permeance and resistances were calculated as:
L=
J ΔP
R tot =
(1)
1 μ . L end
R f = R tot − R m =
R rev = R tot − 2.2. MBR set-up A lab-scale MBR was used in this experiment to treat a synthetic wastewater made of protamylasse solution (kindly provided by Avebe, Veendam, The Netherlands). The characteristics of the
(2)
R res =
1 1 − μ . L end μ . L pristine
1 μ . CWPrinsed
1 1 − μ . CWPrinsed μ . CWPcleaned
Table 1 Membrane properties.
Properties
Ribbed Membranes
Volume porosity (%)
Flat Membranes 70.0
Porosity (cm3/g)
1.26
1.32
Pore size (µm) Median Minimum Maximum
0.22 0.05 1.25
0.19 0.05 1.41
69.7
Measureda Calculatedb Clean water permeance (LMH/bar) Critical Flux (LMH) Without vibration (WV) With vibration (V) a
536±30
918±57
603±36
18c 24-27
21-24 30-33
14-16 20-22
Measured flux (based on projected area = 1+3) Calculated flux (based on actual available membrane surface, including the full area of the ribs = 1+2+3+4) c Results from [24] b
(3)
(4)
(5)
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
Permeate line Effective area for permeation
Retentate Side 5mm
317
Gluing-area
Permeate Side vibration
12mm 1.435mm 3.48mm 0.61mm Fig. 1. Top: Retentate side of ribbed membrane (left), permeate side-membrane panel assembly (right); Bottom: membrane dimensions.
R irrev + irrec =
1 − Rm μ . CWPcleaned
(6)
with L membrane permeability (LMH/bar), J flux (LMH), ΔP transmembrane pressure (bar), Lend the permeability at the end of the filtration test, Lpristine the permeability of the pristine membrane, CWPrinsed the clean water permeance after rinsing the membranes and CWPcleaned the clean water permeance after soaking the membranes in a 2000 ppm NaOCl solution for 2 h. μ is the dynamic viscosity of the permeate, which will be approximated as 1.002 mPa s, i.e. the dynamic viscosity of water at 20 °C. For the resistances, Rtot is the total resistance, Rm the membrane resistance, and Rf the fouling resistance. Reversible fouling (Rrev) is here seen as the fouling that can be removed through physical cleaning and rinsing with tap water, while residual fouling (Rres) requires maintenance cleanings to get removed. Irreversible and irrecoverable fouling (Rirrev þ irrec) was considered not to get removed by the implemented cleaning step. 2.3.1. Critical flux measurement To determine the critical flux of the membranes, flux-stepping tests were performed based on [27] with modified on the step height and length. The initial flux was 3 LMH, which was increased in steps of 3 LMH (step height) to a maximum of 49 LMH, with a step length of 10 min of filtration without relaxation. The critical flux itself was determined evaluating the obtained fouling rates. Arbitrary value of 0.1 kPa min 1 was chosen as the critical threshold fouling rate at which severe fouling started to occur. 2.3.2. Influence of vibration The influence of vibration was studied for the ribbed membrane with comparison to the flat ones, both in short and long term filtration tests.
For the short test, one set of membranes containing a flat and a ribbed membrane were used for activated sludge filtering in the lab-scale MBR using a magnetically membrane vibration system [21]. A high flux filtration at 40 LMH for 16 h was applied to the set of membranes for an accelerated fouling filtration. The flux value was selected deliberately to accelerate membrane fouling and thus shorten testing duration. This experiment was done with and without vibration, and each was done three times (cycle). In between two cycles, a chemical cleaning was operated to all the membranes, for simplicity, regardless of their total resistance/ permeance. The development of the trans-membrane pressure was recorded at certain time interval. To confirm the results from the short-term test, a continuous long-term filtration experiment was carried out. It was done for 34 days; with a filtration and relaxation mode of 8 and 2 min. Two instantaneous fluxes were applied, i.e. 24 LMH (for the first 17 days) and 31 LMH (for the second 17 days). The trans-membrane pressures of the membranes were recorded daily. The fouling resistance and permeance of the membranes were calculated based on Eqs. (1)–(3). 2.3.3. Influence of cycle time and relaxation time To further investigate the operational parameters for the ribbed membrane, variations on cycle and relaxation time were performed in the MBR without any vibration. In each experiment, three membranes panels were used. With regard to cycle time (CT), three different values were used; i.e. 5, 10 and 15 min with constant relaxation time of 10%. This means that for example for the 5minutes cycle time test, the filtration cycle consisted of 4.5 min filtration followed by 0.5 min relaxation. In this experiment, the flux was kept constant at 35 LMH. For the relaxation time (RT), it was varied between 5, 10 and 20% with a constant cycle time of 10 min. For example, the 20% relaxation time test involved cycles with 8 min filtration and 2 min
318
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
Fig. 2. Pore size distribution of pristine Amer-sil membranes: (a) ribbed and (b) flat membrane.
vibration. To give constant net water productivity, consequently the instantaneous flux was varied to 35, 38.5, and 42 LMH respectively Different fouling types were characterized based on Eqs. (3)– (6) and were compared comprehensively. 2.3.4. Cleaning procedure All chemical cleanings involved in this study (every time after filtration) were done using 2000 ppm of NaOCl (Sigma-aldrich, available chlorine 10–15%) for 1 h. The cleaning procedure was optimized by cleaning fouled membranes by rinsing them with tap water, followed by (1) immersing in 2000 ppm of NaOCl for 2 h or (2) immersing in 2000 ppm of NaOCl for 1 h and in 3 g/L citric acid for 1 h. The cleaning performance was then compared by means of ATR-FTIR (Alpha-P, Bruker Corporation).
3. Results and discussion 3.1. Membrane properties Table 1, Figs. 2, and 3 respectively show the main properties, pore size distribution graphs, and microstructures of the flat and
the ribbed membranes. The volume porosities of the membranes are quite similar. The pore sizes are also quite similar, with a slightly bigger median pore size for the flat membrane but a smaller maximum pore size. The membranes have a bimodal pore structure The membranes have a bimodal pore structure: small pores in the range of 0.02–0.1 mm and big ones in the range of 1– 2 mm. The silica particles are precipitated silica, hence porous. The porosity in the membranes is contributed by the silica particles itself (intra-granular porosity), the voids between silica aggregates (inter-granular porosity) and the voids left from the solvent extraction after extrusion and calendaring [28]. However, as listed in Table 1, the volume porosity and also pore size of the two membranes are quite close, it is assumed that the difference between the two membranes is solely on the ribs. Table 1 clearly shows the superiority of the ribbed membranes over the flat ones, i.e. 70% increase in clean water permeance. Considering their similar microstructure, these results indicate that the three sides of the ribs all work as active area, thus providing additional surface for filtration. Based on the calculations from the ribbed membrane dimensions (Fig. 1), the ribs give 52% extra area compared to the flat membranes. The same finding on permeance increase using patterned membranes was found by others but much smaller scale of pattern, i.e. in micrometer scale [13]. With respect to the critical flux, the measured critical flux again indicates the superiority of the ribbed membranes over the flat ones, i.e. 10–30% higher for the ribbed membranes. However, calculated critical fluxes (based on the actual available area including the ribs sides) are actually lower for the ribbed membranes. It means that the increase of the critical flux is not proportional with the additional area, but slightly less, indicating some disadvantages of the ribs. One probable reason of this phenomena is that the foulants deposit in the corners (most likely dead-zone area) of the lower parts of the ribs. This disadvantage of the ribs on the membranes is in contradiction with another study on patterned membranes which found that local shear was promoted in the patterned membranes limiting bacterial deposition [12]. Nevertheless, there are at least two main reasons for the aforementioned fact: (1) the two patterns are very different in size, i.e. micro- vs millimeter size, and (2) the filtration system are different, i.e. side-stream vs submerged system. The former system is characterized by a higher cross-flow velocity than the latter. In the present study, the combination of the millimeter size patterns, the flow regimes and the submerged system (although with vibration) do not seem to facilitate the promotion of the local shear. These findings also indicate that an optimization of operating condition, in particular hydrodynamics for optimizing local mixing, is required to maximize the potential of the ribbed membranes in submerged MBR systems.
Fig. 3. Surface SEM images of pristine Amer-Sil membranes: (a) ribbed and (b) flat membrane surface.
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
319
Fig. 4. Permeance and Rf evolution during the filtration test with vibration (V) and without vibration (WV).
3.2. Operational parameters Three main operational parameters were tested: use of vibration, filtration mode (relaxation cycle time), and cleaning procedure. In the experiment with vibration, the role of the ribs was investigated by comparing the filtration with the flat membranes. Second, in the experiment of relaxation/cycle time, optimization of the filtration mode was carried out. Finally, the evaluation of the applied cleaning procedure is discussed. 3.2.1. Influence of vibration The advantage of membrane vibrations is obvious as vibrations increase the membrane critical flux (Table 1). This agrees with the previous study by Bilad et al. [21]. The critical fluxes of both flat and ribbed membranes increase about 30–50% when vibrating. This increase is due to the enhanced shear from the vibration that changes the turbulence near the membrane surface, thus reducing concentration polarization and cake formation [20,21]. Fig. 4 shows the permeance and fouling resistance evolution during filtration experiments both with (V) and without vibration (WV) on the flat and ribbed membranes. The experiments consist of 6 runs of 16 h filtration with an instantaneous flux of 40 LMH. The graph is arranged based on the performed time. At the end of each run, membranes were chemically cleaned and used again for the next filtration run. The summary of these data is shown in Fig. 5.
Fig. 6. (A) Fouling resistance and (B) permeance development of the flat and ribbed membranes during long-term filtration test.
Fig. 4B shows in general that the ribbed membranes provide (slightly) higher permeances, regardless the use of vibration. It is obvious since the initial permeance for the ribbed membrane is also much higher. During the filtration, membrane permeance decreased steeply in the beginning. It continued to decrease steeply for the filtration without vibration (WV1, WV2, and WV3) but remained fairly constant for the one with vibration (especially V2 and V3). These results confirm the advantage of vibration to decrease the membrane fouling. The evolution of fouling resistance is shown in Fig. 4A and summarized in Fig. 5B. In general, the Rf values for the flat and ribbed membranes are not different. It means that for the current conditions (instantaneous flux of 40 LMH with this type of sludge), the ribs did not give any benefit from the point of membrane vibration. The vibration itself, indeed, gave advantage on the filtration process by decreasing the Rf values, but not specifically more profound on the ribbed membrane. Fig. 6 shows the permeances and Rf evolution during the longterm filtration experiments. Compared to the short tests which were performed at 40 LMH, the long-term tests were performed at lower fluxes (i.e. 25 and 31 LMH), closer to the critical flux of each membrane, i.e. 21–24 LMH and 30–33 LMH for the flat and ribbed membrane respectively. At 25 LMH, the Rf values of both membranes are more or less the same and stable around 0.5 109 m 1.
Fig. 5. (A) Permeances: initial and after filtration and (B) fouling resistances after 16 h of filtration with vibration (V) and without vibration (WV).
320
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
When applying a flux of 31 LMH, the Rf of the flat membrane increased significantly while the ribbed one still kept its stable Rf value. The finding above confirms the importance of critical flux values. The first flux, 25 LMH, is below the critical flux of the ribbed membrane and is (still) close to that of the flat one. Both membranes then show similar fouling values (low fouling rates). The second flux, 31 LMH, is above the critical flux of the flat membrane and is close to that of the ribbed one. Thus, Rf for the flat membrane increases significantly while the Rf for the ribbed one remains stable. The flux for the short term test, i.e. 40 LMH, was above the critical flux of both membranes. Again, the difference between the two membranes can thus not be distinguished. Due to the fact that the ribbed membranes have a larger actual available area of filtration, and thus lower the calculated flux (Section 3.1.), one could expect to have smaller Rf values for the ribbed membranes. However, the Rf values for both membranes are more or less the same, indicating again the disadvantage of having the ribs. Because the ribs promote membrane fouling, thanks to the additional area, the ribbed membranes still offer two main advantages, namely higher permeability and critical flux, regardless the use of vibration. Thus they can be operated at a higher specific productivity (operational flux) that flat membranes. And finally, a lower area of a ribbed membrane is required than flat one for a similar filtration capacity. This specific productivity is indeed the ultimate parameter to judge the overall membrane performance, in addition to capital and operational expenditure for more extended analyses. Limited advantage of ribbed membranes under vibration are somehow surprising, since it was expected that the ribbed membranes (in vibration mode) can promote turbulence near the membrane surface which can further reduce the fouling [22]. The direction of the vibration is similar; i.e. perpendicular to the ribs. However, the spacing between the ribs is much smaller; i.e. 6 mm for this study and between 15 and 60 mm for Gomma et al. The small spacing of the ribbed membrane in present study in combination with the perpendicular vibration direction might affect the exposure of the rubbed surface to shear. The upper parts could act as shield for the lower parts which limits the lower parts' exposure to shear [29]. However, it is worth to mention that comparing the two cases might not be fair, because the ribs not only act as turbulence promoter in the present case, but also as extra filtering active area. In comparison to other patterned membranes which are able to really mitigate fouling via additional surface area and local shear promotion [12,13], the present membranes are very different in size (much larger pattern). Thus, scaling down the pattern is probably one interesting way to go for further study. By tuning the extrusion process, ribbed membranes with different spacing and size (height and thickness) can be produced and further tested. Another explanation is the relative difference in operation parameters. The aforementioned references used cross flow velocities of 0.05–0.2 m/s, i.e. much higher than the one applied in this study which merely originates from secondary flow caused by upward Table 2 Filtration run details for CT and RT experiments. Run
Filtration (min)
Relaxation (min)
Instantaneous flux (LMH)
CT CT CT RT RT RT
4.5 9.0 13.5 9.5 9.0 8.0
0.5 1.0 1.5 0.5 1.0 2.0
35.0 35.0 35.0 35.0 38.5 42.0
5 10 15 5 10 20
Table 3 Membrane (Rm) and fouling (Rf) resistance and relative contributions of reversible (Rrev/Rf), residual (Rres/Rf), irreversible and irrecoverable resistances (Rirrev þ Rirrec/ Rf) to Rf. Resistance
CT 5
0.47 Rm (109m 1) Rf (109m 1) 0.77 Rrev/Rf (%) 11.00 Rres/Rf (%) 36.61 (Rirrev þ Rirrec)/Rf (%) 66.46
7 7 7 7 7
Resistance 9
1
Rm (10 m ) Rf (109 m 1) Rrev/Rf (%) Rres/Rf (%) (Rirrev þ Rirrec)/Rf (%)
CT 10
0.16 0.47 0.23 1.05 9.17 12.27 32.45 19.38 8.89 68.35
7 7 7 7 7
RT 5 0.47 0.71 4.05 24.17 71.77
7 7 7 7 7
0.16 0.24 1.57 3.87 3.97
0.16 0.30 3.98 4.79 8.77
CT 15 0.47 0.89 23.41 41.26 35.34
7 7 7 7 7
RT 10 0.47 0.76 12.58 30.73 67.95
7 7 7 7 7
0.16 0.47 0.23 0.68 6.52 1.00 15.34 29.65 12.86 63.66
0.16 0.24 4.76 13.91 18.48 RT 20
7 7 7 7 7
0.16 0.20 0.26 8.23 13.15
air bubble movements which has a very small influence on crossflow velocity and Reynold number [30]. The impact of surface pattern is strongly influenced by the cross-flow velocity near the membrane surface [29]. For the vibration system, various parameters can be still exploited in the future, i.e. the optimization of the frequency, amplitude, and also movement direction, are all related to local flow dynamic near the membrane surface. 3.2.2. Influence of cycle time (CT) and relaxation time (RT) Table 2 summarizes the details of the CT and RT experiments. In the CT filtration tests, it was studied how the balance between length of filtration and length of relaxation affected membrane fouling. In other words, it was studied whether shorter filtration times coupled with shorter relaxation times, i.e. more frequent but less extensive relaxation periods (5 min cycle time), resulted in different fouling propensity than longer filtration and relaxation times (15 min cycle time), or if there was an optimum (10 min cycle time). In the RT filtration tests, it is important to keep the net water productivity constant by adjusting the instantaneous flux during filtration. It is therefore studied how the balance of instantaneous flux and relaxation time affects membrane fouling and, consequently, which of these two factors plays the biggest role. To compare the amount of fouling, Rf should be taken as the reference value. Based on Table 3, a clear trend cannot be distinguished easily for both CT and RT experiments. In general, the Rf values are almost the same (in the range of 0.68–1.05 109 m 1) with quite a high standard deviation (around 30%). A closer look into the different types of fouling (Table 3) indicates the influence of the operational parameters on the reversibility of the fouling. From the CT experiment, it is obvious that Rrev increases with increasing cycle time. This fact is also confirmed by comparing runs RT 20 and CT 15. Run CT 5 and RT 20 were carried out with the two shortest filtration times (i.e. 4.5 min filtration and 0.5 min relaxation; and 8 min filtration and 2 min relaxation, respectively) and showed the lowest Rrev percentage. Run CT 15 was performed with the longest filtration time (i.e. 13.5 min filtration/1.5 min relaxation) and showed the highest percentage of Rrev. It means that longer filtrations accumulate reversible fouling, which can be removed easily by physical cleaning (in this case: rinsing). For Rres in the CT experiment, an optimal value (minimum) seems to exist at CT 10. It indicates that the filtration mode in CT 5 and CT 15 provides permeation and relaxation combinations which are either too short or too long. For the former, a too short relaxation time (0.5 min) may not provide enough time for particle
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
321
Fig. 7. Photos of the ribbed membranes in a (a) fouled state, (b) after rinsing and (c) after washing.
back transport [17]. For the latter, a too long filtration (13.5 min) causes cake consolidation, which might transform reversible fouling become irreversible [31]. In the RT experiments in which filtration times are almost similar (i.e. ranging from 8 to 9.5 min), the Rres values are relatively close to each other. Thus, for the current system, a filtration time of maximum 8 min and a relaxation time of minimum 1 min are suggested to be minimum values with respect to both reversible and residual fouling. These values are comparable to these of the commercial MBR membranes available, e.g. from Kubota and Toray (applying 8 min on and 2 min off) [2]. Regarding instantaneous flux, CT 10 and RT 10 were actually performed with the same time mode (i.e. 9 min filtration and 1 min filtration) but different fluxes: 35 and 38.5 LMH respectively. The influence of the instantaneous fluxes can then be seen from these two runs. Comparing the values of Rres and Rrev, it is obvious that an increasing flux will result in higher Rres compared to Rrev. It means that a higher instantaneous flux allows more reversible fouling to consolidate and become more irreversible (in this case residual fouling). The values of Rirrev þRirrec are relatively high for all runs, indicating that the cleaning procedures applied in between the experiments are not enough to remove the foulants. The CT experiments (especially CT 5) show slightly smaller (Rirrev þRirrec) values
Fig. 8. FTIR spectra of pristine, fouled, rinsed, and washed membranes.
because they were done earlier. With time, accumulation of irreversible and irrecoverable fouling occurred and stabilized, as shown in the relatively stable value of Rirrev þRirrec. The increasing value of Rirrev þRirrec in the beginning of the experiment shows that there is initial process of pore blocking accumulation. It is believed that foulants blocks the big pore mouth at this initial stage [24]. 3.2.3. Cleaning efficiency In the previous study [24], the chemical cleaning was performed by NaOCl 1000 ppm and found to be not optimal yet. In this study, NaOCl 2000 ppm was used. This concentration is also commonly used, and is still safe for the silica. Higher concentration of NaOCl is not possible due to high pH (above 9.5) that can cause dissolution of silica. Fig. 7 shows the appearance of the ribbed membrane in its fouled state, rinsed with water, and washed with chemicals, respectively. During its use in the filtration experiment, the existence of the ribs may actually worsen the fouling as the foulants may be deposited in the corner line of the ribs. These corners are then also more difficult to get cleaned (dead zones). The cleaning efficiency for the ribbed membranes is lower compared to that of the flat ones. This is shown by higher values of Rf in the beginning of every filtration run in experiment 3.2.1 (Fig. 4) for the ribbed membranes compared to the values of the flat ones. Considering the fact that both membranes have a similar microstructure, this may indicate that the ribs are the cause for this lower cleaning effectiveness. The ATR-FTIR spectra of the ribbed membranes in their pristine, fouled, rinsed and cleaned form are shown in Fig. 8. The peaks at 485, 798, and 1200–1000 cm 1 belong to silica corresponding to the Si–O bending, the Si stretching and the Si–O stretching, respectively [32]. The peaks at 691 cm 1 and 1434 cm 1 belong to PVC which correspond to the C–Cl stretching and CH2 deformation respectively [33]. For the characteristic peaks of the foulant material, the peaks at 1200–1000 cm 1, 1540 cm 1 and 1650 cm 1 are interesting to be analyzed since they correspond to the symmetric and asymmetric C ¼O stretching in polysaccharides and polysaccharide-like substances, amide II and amide I respectively [34]. It is important to note that there is an overlap between the peaks of Si–O stretching from silica and polysaccharides at 1200–1000 cm 1, making it impossible to quantitatively evaluate this in terms of foulant removal. Another possible way of interpreting this peak is to look at the reduction of absorbance. The fouled membrane clearly shows
322
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
Table 4 Normalised band heights of the fouled, rinsed and cleaned membrane fragments, together with the removal after rinsing (Rinsed), cleaning with 2000 ppm NaOCl for 2hours (NaOCl) and cleaning with 2000 ppm NaOCl for 1 h and 1 h with 3 g/L citric acid. Wavelength
Normalised band height
Removal (%)
(cm 1)
Fouled
Rinsed
NaOCl
NaOClþ citric acid
Rinsed
NaOCl
NaOClþcitric acid
1540 1650
0.071 0.194
0.069 0.193
0.022 0.127
0.023 0.127
2.40 0.61
69.0 34.6
68.0 34.6
the smallest absorbance while the pristine shows the highest absorbance for this peak. This indicates that fouling results in coverage of the membrane surface which causes a decrease of the absorbance of the peak corresponding to the silica particles. The only peaks that can therefore be used as a measure of foulant removal are the peaks at 1540 cm 1 and 1650 cm 1. To measure the actual foulant removal, the band heights of the fouled, rinsed and both cleaned membrane fragments were normalised to the 485 cm 1 band height corresponding to the Si–O bending. This peak was chosen as it was assumed not to change over the course of the experiments [24]. Table 4 shows the percentage removal of the membranes cleaned with NaOCl for 2 h, membranes cleaned through consecutive soaking in NaOCl and citric acid, and rinsed membranes, respectively. Table 4 shows that the removal efficiency of the treatment with NaOCl in combination with citric acid was almost the same as for the treatment using NaOCl only. The small difference between the 2 h NaOCl and 1 h NaOCl þ1 h citric acid treatments thus shows that soaking the membranes in NaOCl for 2 h did not result in a much higher removal of the foulant material than the 1 h treatment. In respect to the NaOCl concentration, comparing with the previous experiment which used 1000 ppm of NaOCl [24], the current cleaning with 2000 ppm of NaOCl also does not show better removal. Thus, 1000 ppm of NaOCl for 1 h is suggested for the chemical cleaning. 3.3. Membrane panels and modules Building panels and modules in MBR is important as it defines the cost; either capital cost at the moment of installation or operational cost whenever membrane have to be replaced. Most of the flat-sheet panels available in the market have thickness in the
range of 4–7 mm with a separation distance around 6–8 mm [35]. For the ribbed membranes, two layers of membranes will already give 6.96 mm and may reach 8–10 mm (depending on the additional frame). However, it is possible to prepare a module from one sheet of 2-sided ribbed membranes and one sheet of 1-sided ribbed membranes, thus the membrane module can be thinner. Thicker panels is definitely not wanted, as it will decrease the packing density (m2/m3) and further increase the related cost. While most commercial membranes panels are welded into a plate spacer for mechanical integrity, such as an acrylonitrile butadiene styrene (ABS) support for Kubota and Toray or polypropylene (PP) for Brightwater Engineering, the ribbed membranes might be arranged in a different way into, a slimmer panel. As mentioned earlier in Section 3.1 and Fig. 1, the ribbed membranes can be potted into a panel without a spacer. Moreover, these membranes are thicker (0.61 mm for the membranes itself, and 3.48 mm including the ribs) compared to the commercially used phase-inversed membranes, such as PES (Brightwater Engineering), PVDF (Toray), and chlorinated PE (Kubota) [2]. A thin frame can be added only at the edge of the membranes to make the panels more rigid and also for permeate pipe installation. The combination of thicker membranes with thinner required frames results in a comparable overall panel size (as calculated and summarized in Table 5). Table 5 lists five suppliers for flat-sheet MBR with module characteristics, such as membrane materials, panel separation, panels packing density, and operational parameters (flux, permeability, and specific aeration demand) for a full-scale plant. The aforementioned parameters were also calculated and estimated for the studied membrane for comparison. In general, if the membrane separation was taken as 8 mm, the panel packing density values are comparable, being 111 m2/m3. The value of flux,
Table 5 Comparison of panel arrangements and operational parameters of five commercial flat-sheet MBR [2] and the present ribbed MMM. Supplier
Brightwater Kubota Toray Huber Colloide Amer-Sil (this study)a a
Membrane
PES PE chlorinated PVDF PES PES Ribbed PVC/ Silica
Panels /Modules
Operational parameters
Separation (mm)
Panel packing density (m2/m3) Flux (LMH)
Permeability (LMH/ Bar)
Specific aeration demand (Nm3/m2)
9 7 6 n.a. n.a. 8
110 115 135 160 133 111b 169c
150 330–680 208 250 62.5 200-300f
1.28 0.56–1.06 0.4–0.54 0.35 0.5 2.27g 0.796h
27 20–33 25 24 25 o21d o 30e
panel thickness: 10 mm (6.96 mm for two membrane layers (including ribs) plus additional frame) calculated with membrane area and with additional ribs area (accounts for 52% additional area) d taken from critical flux values for aeration mode and e for vibration mode f sustainable permeability maintained by the membranes on the short and long-term filtration section 0 g aeration in the lab-scale MBR (calculated based on air flowrate of 10 L/min for 20 membrane panels) h calculated based on equation SADm¼ 0.0044Jnetþ 0.708 [2]. b c
L. Marbelia et al. / Journal of Membrane Science 498 (2016) 315–323
permeability and the calculated aeration demand are also as good as for the commercial ones. Obviously, translating results from labscale to pilot and full-scale might not be as straightforward as assumed here. Indeed, adjustments with respect to membrane size and feed hydrodynamics need to be taken into account.
4. Conclusions This study proves the competitiveness of the ribbed PVC/silica MMMs in comparison to available commercial membranes for activated sludge filtration in MBRs. A number of advantages are offered by the ribbed membranes: (1) the ribs work as an additional active filtering area, thus giving higher clean water permeability and higher critical flux, and (2) the membranes are relatively thick, offering a good mechanical integrity for building stand-alone and (thanks to the ribs on the permeate side) spacerfree panels. Despite these advantages, the ribs seem to be more prone for foulant deposition which is harder to remove chemically, thus causing slightly lower cleaning efficiency. Scaling down the dimension of the ribs and optimization of the cross-flow velocity near the membrane surface are interesting for future study in order to further promote the local shear near the membrane surface. In the aeration mode, filtration during maximum 8 min and relaxation during minimum 1 min were found to be optimal to keep the reversible and residual fouling minimal for the studied wastewater. Although the irreversible fouling seemed to be dominating, the maintained permeate flux during the filtration was kept quite reasonable and comparable to that of commercial membranes. In the vibration mode, membrane fouling was decreased both with the flat and the ribbed membrane. In order to allow the ribs to work as turbulence promoters, several parameters still need to be optimized, such as vibrational parameters (frequency, amplitude, etc) and rib dimensions (height, distance, etc).
Acknowledgment We wish to thank KU Leuven for the financial support in the frame of the FWO (G.0808.10N), FWO (G.A.043.12N ERA-Net New INDIGO), OT (11/061), IOF-KP (IOFKP/13/004), and the Belgian Federal Government for an IAP grant. And Amer-Sil for providing the tested membranes.
Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.memsci.2015.10. 017.
References [1] Wastewater Engineering, Treatment and Reuse, McGraw-Hill Education, United States, 2004. [2] S. Judd, The MBR Book: Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment, Elsevier, Amsterdam, 2010. [3] A.K. Hołda, I.F.J. Vankelecom, Understanding and guiding the phase inversion process for synthesis of solvent resistant nanofiltration membranes, J. Appl. Polym. Sci. 132 (2015). [4] C. Visvanathan, R. Ben Aim, K. Parameshwaran, P. Taylor, R. Ben Aim, Membrane separation bioreactors for wastewater treatment, Crit. Rev. Environ. Sci.
323
Technol. 30 (2000) 1–48. [5] M. Gander, B. Jefferson, S. Judd, Aerobic MBRs for domestic wastewater treatment: a review with cost considerations, Sep. Purif. Technol. 18 (2000) 119–130. [6] A. Drews, Membrane fouling in membrane bioreactors—characterisation, contradictions, cause and cures, J. Memb. Sci. 363 (2010) 1–28. [7] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used in wastewater treatment, J. Memb. Sci. 284 (2006) 17–53. [8] H.H.P. Fang, X. Shi, Pore fouling of microfiltration membranes by activated sludge, J. Memb. Sci. 264 (2005) 161–166. [9] F. Meng, S.-R. Chae, A. Drews, M. Kraume, H.-S. Shin, F. Yang, Recent advances in membrane bioreactors (MBRs): Membrane fouling and membrane material, Water Res. 43 (2009) 1489–1512. [10] N. Hilal, O.O. Ogunbiyi, N.J. Miles, R. Nigmatullin, Methods Employed for Control of Fouling in MF and UF Membranes: A Comprehensive Review, Sep. Sci. Technol. 40 (2005) 1957–2005. [11] S.H. Maruf, A.R. Greenberg, J. Pellegrino, Y. Ding, Critical flux of surface-patterned ultrafiltration membranes during cross-flow filtration of colloidal particles, J. Memb. Sci. 471 (2014) 65–71. [12] Y.-J. Won, J. Lee, D.-C. Choi, H.R. Chae, I. Kim, C.-H. Lee, et al., Preparation and application of patterned membranes for wastewater treatment, Environ. Sci. Technol. 46 (2012) 11021–11027. [13] Y. Gençal, E.N. Durmaz, P.Z. Çulfaz-Emecen, Preparation of patterned microfiltration membranes and their performance in crossflow yeast filtration, J. Memb. Sci. 476 (2015) 224–233. [14] R. Jamshidi Gohari, W.J. Lau, T. Matsuura, a F. Ismail, Effect of surface pattern formation on membrane fouling and its control in phase inversion process, J. Memb. Sci. 446 (2013) 326–331. [15] J.A. Kharraz, M.R. Bilad, H.A. Arafat, Simple and effective corrugation of PVDF membranes for enhanced MBR performance, J. Memb. Sci. 475 (2015) 91–100. [16] J. Wu, P. Le-Clech, R.M. Stuetz, A.G. Fane, V. Chen, Effects of relaxation and backwashing conditions on fouling in membrane bioreactor, J. Memb. Sci. 324 (2008) 26–32. [17] Z. Wang, J. Ma, C.Y. Tang, K. Kimura, Q. Wang, X. Han, Membrane cleaning in membrane bioreactors: a review, J. Memb. Sci. 468 (2014) 276–307. [18] T. Maqbool, S.J. Khan, C.-H. Lee, Effects of filtration modes on membrane fouling behavior and treatment in submerged membrane bioreactor, Bioresour. Technol. 172 (2014) 391–395. [19] L. Defrance, Reversibility of fouling formed in activated sludge filtration, J. Memb. Sci. 157 (1999) 73–84. [20] M.Y. Jaffrin, Dynamic filtration with rotating disks, and rotating and vibrating membranes: an update, Curr. Opin. Chem. Eng. 1 (2012) 171–177. [21] M.R. Bilad, G. Mezohegyi, P. Declerck, I.F.J. Vankelecom, Novel magnetically induced membrane vibration (MMV) for fouling control in membrane bioreactors, Water Res. 46 (2012) 63–72. [22] H.G. Gomaa, S. Rao, M.A. Taweel, Flux enhancement using oscillatory motion and turbulence promoters, J. Memb. Sci. 381 (2011) 64–73. [23] L. Carole, C. Khalauche, I. Vankelecom, M. Roil Bilad, Organic Filtration Membrane, EP 2 902 094 A1, 2015. [24] M.R. Bilad, L. Marbelia, C. Laine, I.F.J. Vankelecom, A PVC–silica mixed-matrix membrane (MMM) as novel type of membrane bioreactor (MBR) membrane, J. Memb. Sci. 493 (2015) 19–27. [25] V. Toniazzo, Amersorb: a new high-performance polymeric separator for lead–acid batteries, J. Power Sour. 144 (2005) 365–372. [26] M.R. Bilad, P. Declerck, A. Piasecka, L. Vanysacker, X. Yan, I.F.J. Vankelecom, Development and validation of a high-throughput membrane bioreactor (HTMBR), J. Memb. Sci. 379 (2011) 146–153. [27] P. Le Clech, B. Jefferson, I.S. Chang, S.J. Judd, Critical flux determination by the flux-step method in a submerged membrane bioreactor, J. Memb. Sci. 227 (2003) 81–93. [28] V. Toniazzo, New separators for industrial and specialty lead acid batteries, J. Power Sour. 107 (2002) 211–216. [29] Y.K. Lee, Y.-J. Won, J.H. Yoo, K.H. Ahn, C.-H. Lee, Flow analysis and fouling on the patterned membrane surface, J. Memb. Sci. 427 (2013) 320–325. [30] F. Meng, B. Shi, F. Yang, H. Zhang, New insights into membrane fouling in submerged membrane bioreactor based on rheology and hydrodynamics concepts, J. Memb. Sci. 302 (2007) 87–94. [31] C. Huyskens, E. Brauns, E. Vanhoof, H. Dewever, A new method for the evaluation of the reversible and irreversible fouling propensity of MBR mixed liquor, J. Memb. Sci. 323 (2008) 185–192. [32] E.R. Lippincott, A. Van Valkenburg, C.E. Weir, E.N. Bunting, Infrared studies on polymorphs of silicon dioxide and germanium dioxide, J. Res. Natl. Bur. Stand. 61 (1934) 61–70 (n.d.). [33] L.M. Matuana, D.P. Kamdem, J. Zhang, Photoaging and stabilization of rigid PVC/wood-fiber composites, J. Appl. Polym. Sci. 80 (2001) 1943–1950. [34] F. Meng, B. Liao, S. Liang, F. Yang, H. Zhang, L. Song, Morphological visualization, componential characterization and microbiological identification of membrane fouling in membrane bioreactors (MBRs), J. Memb. Sci. 361 (2010) 1–14. [35] A. Santos, W. Ma, S.J. Judd, Membrane bioreactors: Two decades of research and implementation, Desalination. 273 (2011) 148–154.