Direct sewage filtration using aerated and vibrated module in submerged filtration system

Transcription

1 Direct sewage filtration using aerated and vibrated module in submerged filtration system M.R. Bilad and I.F.J. Vankelecom Centre for Surface Chemistry and Catalysis, Faculty of Bioscience Engineering, K.U.Leuven, Kasteelpark Arenberg 23, Box 2461, 31 Leuven, Belgium ( (M.R. Bilad), (I.F.J. Vankelecom)) Abstract This study was performed to compare the performance of aerated filtration module (AerFM) and vibrated filtration module (VibFM) to up-concentrate fresh sewage using a flat-sheet membrane in a submerged configuration. The filtration performance was assessed by conducting the improved flux step method (IFM), batch up-concentration filtrations and verifying the recovery of the chemical oxygen demand (COD). Overall results suggest that VibFM is more effective than AerFM judging from the critical flux (J C ), the critical flux for irreversibility (J Cir ) and the TMP profile during batch upconcentration tests. In addition, The VibFM is also a more energy efficient system. The energy consumption was estimated to vary from.7 to.6 kwh/m 3 and.14 to.17 kwh/m 3 for AerFM and VibFM, respectively. No significant difference was found for the filterability of sewage with different up-concentration levels. An over-exposure of feed in the filtration tank leads to solubilisation of particulate COD thus reducing the harvesting efficiency. Further study on long-term and continuous sewage up-concentration is still necessary. Keywords Membrane; submerged filtration; magnetic vibrating module; fouling INTRODUCTION Conventional reuse schemes of discharged water from secondary treatment normally consist of a train of processes, including aerobic biological process (activated sludge process [ASP]), microfiltration (MF), ultrafiltration (UF) and reverse osmosis (RO). This train was recently optimised in membrane bioreactor (MBR) process that combines ASP and MF/UF, followed by RO. Both process trains are costly of up to.8/m 3 and.9/m 3, respectively (Diamantis, ). A new process layout of ZeroWasteWater that enables maximal recovery of water, energy, inorganic and organic fertilizers from domestic discharge was proposed by Verstaete et al. (9). This approach aims to short-cycle the water, energy and valuable materials while adequately abating pathogens, heavy metals and trace organics. It changes the mindset to view sewage not as a wastewater but as used water that contains multi resources to be recovered. In this concept, the diluted used water needs to be up-concentrated to produce permeate and concentrate. The permeate is aimed for reuse and the concentrate is sent to an anaerobic reactor to generate biogas and to recover nutrients. The up-concentration stage is one of the key elements to implement this concept effectively. A high fraction of COD needs to be recovered as feed for the anaerobic reactor and substantial amounts of water are required to be reused. The main challenge of this stage is a high degree of upconcentration of at least times that is required, equivalent to 3 kg/m 3, to meet the economical benefits of the anaerobic treatment (Verstraete et al., 9). The produced energy from the anaerobic treatment should at least be able to provide sufficient heat for the system it self. In this study, the performance of a submerged aerated and a vibrated filtration was compared to upconcentrate the fresh sewage using a MF membrane, further refered to as aerated filtration module (AerFM) and vibrated filtration module (VibFM) respectively. The submerged system normally offers the advantage of lower energy demand and limits the exposure of fresh sewage to enhanced

2 shear-rates, as is the case cross-flow systems. MATERIALS AND METHODS Sewage characteristics The sewage was taken from the influent stream of a municipal wastewater treatment plant in Leuven operated by Aquafin (Belgium). The characteristics of the sewage and permeates at different up-concentration levels are given in Table 1. The COD, TP and TN were measured according to the standard methods (APHA, 1998) using Hach-Lange cuvetes. The COD harvesting efficiency (η) is defined as: COD F - COD P η = COD F (1) where COD F and COD P are COD of feed 1 and of permeate. Table 1. The characteristics of fresh sewage, feed and permeate of filtrations at different upconcentration levels. Parameter Total COD (mg/l) Soluble COD (mg/l) Particulate COD (mg/l) Total Nitrogen (TN) (mg/l) Total Phosphorous (TP) (mg/l) Fresh sewage () Permeate * Permeate * Permeate *retentate 1 is times up-concentrated of feed 1, retentate 2 is times up-concentrated of retentate 1. Membranes A commercial chlorinated polyethylene flat-sheet membrane from Kubota with average pore size of.4µm was used in this study. Prior to use, the membrane was re-potted from its original A-4 size sheet (.1m 2 ) to a smaller module of.16 m 2, according to Bilad et al. (11a). The flux (J) of the filtration was calculated by using eq. 2. V J = (l/m 2.h) (2) A t where V is volume (l), t time (h), A effective filtration area (m 2 ) and TMP trans-membrane pressure (bar or kpa). Experimental Set-up The tests were performed in a lab-scale filtration set-up, illustrated in Fig. 1. The filtration tank had a working volume of l and was equipped with an air bubble aeration system. In the case of the VibFM, the air bubble diffuser was located at the bottom corner of the tank to mix the liquor, and in the case of the AerFM, the air bubble diffuser was placed underneath the membrane module to scour the membrane surfaces as well as to mix the liquor. Inside the filtration tank, the permeate line was connected through a vacuum gauge to a peristaltic pump (Watson-Marlow U 16 Channel Pump, UK) using isoprene manifold tubes (Watson-Marlow, UK). The filtration fluxes were adjusted by changing the rotational speed of the pump. The vibrating engine was put at the top of the tank as shown in Fig. 1B, and was connected to the tank via the module frame. The filtration was performed at a vibration frequency of 4 Hz and with a vibrating power of 6.4 Watt. It was operated in full vibration mode. A more detailed information about the VibFM set-up and its operation is provided by Bilad et al. (11b).

3 Water circulation Fig. 1 Experimental set-up for (A) AerFM (B) VibFM (C) batch up-concentration using AerFM system. Flux stepping test The filtration profile at different fluxes was obtained by applying the improved flux-step method (IFM) (van der Marel 9). The flux of 7 l/m 2.h was set as low flux (J L ). The high flux (J H ) was started from l/m 2.h and stepwise increased by l/m 2.h for minutes until the maximum speed of the pump at l/m 2.h. Prior to use in the filtration test, all membranes were conditioned by filtering clean water at a flux of l/m 2.h for about 1 h. The performance of the system was evaluated based on their critical flux (J C ) and critical flux for irreversibility (J Cir ) values of the membranes. An arbitrary minimum increase in the TMP of Pa/min was used to determine the J C and J Cir from J H and J L, respectively. Up-concentration For the up-concentration filtration, a fixed volume of feed was put in the storage tank (See Fig. 1C) and the filtration tank: 1. l of sewage in the filtration tank and the remaining in the storage tank. The sewage was up-concentrated by filtering the one in the filtration tank, which was continuously fed with fresh sewage from the storage tank at the same flow rate as the membrane flux. This way, the broth volume in the filtration tank was kept constant to ensure constant tank hydrodynamics. The filtration stopped when all sewage from the storage tank was finished. Two stages of upconcentration filtration were performed : filtration 1, where 6 l of fresh sewage was initially put in the storage tank (with 1. l in the filtration tank present at the start and still left at the end, this corresponds to a times upconcentration) and filtration 2, where 6 l of the already times preconcentrated sewage (retentate of filtration 1) was put in the storage tank, both at a constant flux of 18 l/m 2.h. The fresh sewage, the retentate of filtration 1 and the retentate of filtration 2 are further referred to as feed 1, retentate 1 and retentate 2, respectively. The filtration was stopped when the TMP reached ± kpa and the fouled membrane was then cleaned before continuing the filtration. Meanwhile, an overnight relaxation was applied due to practical limitation of the set-up. Energy consumption The estimation of the energy consumption for the filtration was based on the energy consumption map of a related full-scale submerged MBR applied in municipal wastewater treatment (Fenu et al., ), with an overall energy consumption of.64 kwh/m 3. This included the energy consumption related to both the bioreactor operation and the submerged microfiltration. By excluding the cost related to the bioreactor operation (such as fine aeration, sludge mixing and disposal, pre-treatment and tank recycle), the energy consumption for a submerged filtration only (E FS ), as is the case here for the algae harvesting was.4 kwh/m 3. This number still includes influent pumping (P in ) (.3 kwh/m 3 in the full scale municipal MBR), permeate pumping (P P ) (.7 kwh/m 3 ), coarse bubble

4 aeration (A C ) (.23 kwh/m 3 ), cleaning in place (CIP) (.4 kwh/m 3 ) and compressing the air (C a ) (.2 kwh/m 3 ). All these operations would also be present in an industrial scale algae harvesting installation. The energy consumption for direct sewage filtration was estimated by assuming the similar plant scale as for the reference MBR. Since it has a similar capacity, the energy consumption for P in and P P would also be similar. On the other hand, A C, C a and CIP are related to the membrane area and thus a function of the membrane type and its flux for certain feed. Therefore, the ratio (r A ) of membrane area needed (A n ) to the referenced municipal MBR area (A ref ), was used to estimate the energy consumption of these three components. Since In the referenced MBR, the operational flux was set at sub-critical value, the applied fluxes (J) (l/m 2.h) for algae harvesting, and the overall energy consumption were calculated using eq. 4 and 6. J =.8 J C r A E v = A A n ref = P E E = w J = J + P ref + r ( A + C in P A c a + v ρ η a CIP) where J ref is the referenced flux of 22 (l/m 2.h), E v the estimated energy consumption based on the amount of permeating volume (kwh/m 3 ), E w the estimated energy consumption based on the dry weight of harvested COD (kwh/kg) and ρ the solid concentration of microalgae in the feed stream (kg/m 3 ). (4) () (6) (7) RESULTS AND DISCUSSION Filtration performance Improved flux-stepping test Fig. 2 shows the summary of J C and J cir for all IFM tests and the TMP profiles during the IFM tests. The J H and J L reflect the TMPs just before the end of J H and J L in each flux step. As a reference, the IFM for a static system without aeration nor vibration was also presented. As expected, the J C and J cir for static system were far below the ones for the AerFM and VibFM operations. During the filtration, the solute is then transported onto the membrane surface without any back transport induced from shear-rates. The absence of shear-rates also causes a very low degree if anyreversible fouling, indicating by the equal values of J C and J Cir. A significant difference in critical fluxes was observed for the AerFM and VibFM systems, both for J C and J Cir. This is due to the higher shear-rate that is experienced by the liquid-membrane interface in the VibFM system. However, the J C of both systems was equal for filtration of permeate 2. This is due to a relatively high applied flux step in the IFM of l/m 2.h, which excludes detection of small differences in filtration filterability. A more distinct difference can be observed by directly comparing Fig. 2B and 2C at the corresponding up-concentration levels. The results show that the degree of up-concentration is not significantly affecting the performance of both tested systems. It is only obvious for retentate 2, which was up-concentrated for times. This can be explained as the main foulants are expected to be the extracellular polymeric substances (EPS) and not the particulates.

5 4 4 Critical flux (l/m 2.h) J H J L 3 3 J H J L Static AFM VFM Static AFM VFM Filtration system Time (min) Time (min) Fig. 2 Critical flux (A) and TMP during the flux stepping profile of (B) AerFM and (C) VibFM at different concentration level. Up-concentration The TMP profile during the up-concentration is shown in Fig. 3. The up-concentration filtrations were performed in batch mode, starting from the initial volume to finally reach the requested upconcentration target. Fig. 3 shows a clear advantage of VibFM over the AerFM system. No cleaning cycle was required during the filtration of VibFM. Meanwhile, two and respectively one cleaning cycle were required for the AerFM for filtrations 1 and 2. Relatively low irreversible fouling observed after relaxation for the VibFM system. This is probably due to no vibration was applied during the relaxation period. A surprising result is observed when comparing the filtration profile of filtrations 1 and 2. Referring to the IFM results, a similar fouling propensity was expected for both filtrations due to a relatively similar corresponding critical flux value. This unexpected result is most probably due to the presence of colloidal EPS (ceps) that dominates the fouling. During filtration 1, cepss were retained and got stuck to the membrane surface. They were then removed during the membrane cleaning, thus recuing their presence in the retentate. Consequently, lower amounts of these substances were present in retentate 1 and even less in retentate 2. Therefore, less fouling by ceps was experienced by the membranes during filtration 2. In practice, cepss will always present in the fresh sewage and diminishes the filtration performance. The filtration behaviour of an up-concentration filtration in a continuous system using fresh feed over time is strongly suggested as a future direction of the current study. AFM VFM AFM VFM Time (h) Time (h) Fig. 3 The TMP profile during the up-concentration for (A) filtration 1 and (B) filtration 2. ([ ] indicates that chemical cleaning was performed for AerFM and [ ] indicates overnight relaxation for the VibFM) Harvesting efficiency Since the ultimate objective of direct sewage filtration is to harvest the COD, the harvesting

6 efficiency is a crucial parameter. Basically, a total retention of particulate COD is expected, leaving only the soluble COD to pass the membrane. The COD harvesting efficiencies of the different feeds in this study are given in Table 2. Results show that a lower COD harvesting efficiency was achieved for the filtration of retentate 1 and retentate 2. This can be explained by solubilisation of particulate COD during processing, thus increasing the concentration of soluble COD in the filtration feed that can easily pass to the permeate stream. No significant removal efficiency was found for TN and TP, since the MF membrane was expected to retain only the particulate COD. Energy consumption An estimation of the energy consumption for both AerFM and VibFM system is given in Table 2. The energy consumption for the vibration engine is estimated to be.2 kwh/m 2 membrane. This value is obtained by assuming that one vibrating engine is use to vibrate one train of modules. One train contains 1 modules and one module has an effective filtration area of 4 m 2 (in one sheet 2 faces x 1 m width x 2 m height), resulting in m 2 membrane area mounted per vibrating engine. It is clear that the VibFM system is more energy efficient than the AerFM system. The main component of the energy consumption for the AerFM was the coarse bubble aeration, which was significantly reduced but still needed for mixing the liquid in the case of VibFM. It has to be noticed that the calculation of the energy consumption for the VibFM system is strongly affected by the estimation value for the energy consumption of the vibrating engine. Bilad et al. (11b) found that the energy consumption for the applied VibFM system is very low. The energy consumption obtained for a lab-scale VibFM was much lower than the pilot scale AerFM system in an MBR application. For E W, results show that the harvesting efficiency becomes a very significant factor. As discussed earlier, since a complete particulate COD retention is expected, solubilisation of particulate COD should be minimized. This can be achieved by avoiding a direct exposure of bulk feed to an enhanced shear-rate. Table 2. Harvesting efficiency and estimation of energy consumption. System Feed COD harvesting efficiency Applied flux (l/m 2.h) Vibrating engine (kwh/m 3 ) E V (kwh/m 3 ) E W (kwh/kg) AerFM VibFM 91.1% % Permeate % % % Permeate % CONCLUSIONS This study reveals the possibility to apply sewage up-concentration using both AerFM and VibFM systems. Results show that VibFM is more effective, judging from the results of IFM and up-concentration tests. In addition, the VibFM is also a more energy efficient system. The over-exposure of bulk feed in the filtration tank to coarse air bubbles or mixing leads to solubilisation of particulate COD, which leads on its turn to a reduced harvesting efficiency. Further studies on long-term and continued sewage up-concentration is still necessary. REFERENCES Bilad M.R., Declerck P., Piasecka A., Vanysacker L., Yan X.and Vankelecom I.F.J. (11a). Development and Validation of a high-throughput membrane bioreactor (HT-MBR). Journal of Membrane Science, 379, Bilad M.R., Mezohegyi G., Declerck P. and Vankelecom I.F.J. (11b) Novel magnetic vibrating

7 membrane for fouling control in membrane bioreactors. Water Research, (submitted) Diamantis V.I., Antoniou I., Athanasoulia E., Melidis P. and Aivasidis A. (). Recovery of reusable water from sewage using aerated flat-sheet membranes. Water Science and Technology, 62, Fenu A, Roels J, Wambecq T, De Gussem K, Thoeye C, De Gueldre G, Van De Steene B () Energy audit of a full scale MBR system. Desalination, 262, van der Marel P., Zwijnenburg A., Kemperman A.J.B., Wessling M., Temmink H., van der Meer W.G.J., (9). An improved flux-step method to determine the critical flux and the critical flux for irreversibility in a membrane bioreactor. Journal of Membrane Science, 332, Verstraete W, de Caveye PV, Diamantis V. (9). Maximum use of resources present in domestic used water. Bioresource Technology, (23), 37 4.