Applicability of Nanofiltration and Reverse Osmosis for the Treatment of Wastewater of Different Origin

Transcription

1 Cent. Eur. J. Chem. 6(2) DOI: /s Central European Journal of Chemistry Applicability of Nanofiltration and Reverse Osmosis for the Treatment of Wastewater of Different Origin Edit Cséfalvay 1*, Péter M. Imre 2, Péter Mizsey 1 1 Department of Chemical and Environmental Process Engineering, Budapest University of Technology and Economics, 1521 Budapest, Pf. 91, Hungary 2 Gedeon Richter Nyrt. H-1475 Budapest 1, Pf. 27, Hungary Research Article Received 9 January 28; Accepted 7 March 28 Abstract: Membrane separations are finding greater use in wastewater treatment because of their efficiency. In order to prove the effectiveness of membrane filtration an applicability study is carried out. Nanofiltration and reverse osmosis membranes are tested under quite different conditions to reduce the chemical oxygen demands (COD) of wastewaters to meet the Council Directive 76/464/EEC release limit. Two kinds of real wastewaters were selected for the investigation. The wastewaters represent extreme different circumstances since the difference between their COD is two orders of magnitude. All of the membranes tested can be applied either to the treatment of wastewater of high COD (pharmaceutical wastewater) or wastewater of low COD (dumpsite leachate), since the different conditions do not change the membrane characteristics. The experimental data show that none of the membranes can decrease the COD to the release limit in one step. However, if two-stage filtrations (nanofiltration followed by reverse osmosis) are accomplished for both of the wastewaters, a total COD reduction of 94% can be achieved. With the application of the two-stage filtration the COD of the wastewater of low COD can be decreased below the release limit but in case of wastewater of the high COD further treatment will be required. Keywords: Membrane filtration Pharmaceutical wastewater Dumpsite leachate Chemical oxygen demand Nanofiltration Reverse osmosis Versita Warsaw and Springer-Verlag Berlin Heidelberg. 1. Introduction In the last few decades the use of membrane technology has grown significantly in many fields including wastewater treatment. Nanofiltration has proven to be a very effective method for the removal of a wide variety of organic and inorganic compounds from wastewater. Also reverse osmosis is currently considered to be the state of the art in wastewater treatment. In certain cases the required water quality can be achieved only by the combination of membrane separation processes, e.g. nanofiltration (NF) followed by reverse osmosis (RO). Previous studies have shown that membrane separation processes can be applied with success for the reduction of chemical oxygen demand (COD) of several wastewaters. RO processes for wastewater treatment have been applied in the chemical, textile, petrochemical, electrochemical, pulp and paper, and food industries as well as for the treatment of municipal wastewater [1]. Madaeni et al. [2] reported case studies for the treatment of wastewater coming from an alcohol manufacturing plant. Eight polymeric membranes were tested for COD reduction. Experience showed that none of the membranes were able to ensure COD-rejection above 5%. All of the membranes tested failed to reduce COD to a desirable level in one step, at least two consecutive membrane units were needed. Agricultural wastewaters were examined by Reimann et al. [3]. Organic and inorganic membranes were compared regarding COD reduction of several agricultural wastewaters. It was found that the inorganic membranes had better COD-rejection than the organic * csefalvay@ch.bme.hu 277

2 Applicability of Nanofiltration and Reverse Osmosis for the Treatment of Wastewater of Different Origin Figure 1. Schematic drawing of the membrane equipment. ones. The COD-rejection rates were 87% for treatment of milking plant wastewater, 79% for pig slurry treatment, 72% for cattle slurry filtration, and 28% for potato washwater filtration. Treatment of wastewater from the dairy industry with organic RO membranes was examined by Del Re et al. [4]. About 9% COD rejection was observed. Balannec et al. [5] implemented a dead-end filtration of dairy effluent and tested nine different nanofiltration and reverse osmosis membranes. The initial COD of 36, mg L -1 was reduced by 99% following membrane filtration. The reduction of COD in the case of dairy process water was investigated by Aokum et al. [6] using a vibratory shear-enhanced filtration system with various NF and RO membranes. Their results showed that the vibratory shear-enhanced filtration system outperformed conventional cross-flow filtration in NF and in RO both in terms of flux and COD-rejection. A special membrane system for COD removal from the effluent of anaerobic digestion of agricultural waste was developed by Castelblanque et al. [7]. After the biological treatment the effluent consisted of ammoniasalts and non-degradable COD which was filtered through semi-permeable membranes. The membrane system consisted of a dual media sand/anthracite filter, included an ultrafiltration (UF) spiral-wound membrane followed by an NF and/or an RO membrane. CODrejection of about 8% was achieved, therefore attaining the environmental standards. Treatment of dumpsite leachate from several dumps was investigated by Rautenbach et al. [8] With the combination of reverse osmosis, nanofiltration, and high pressure reverse osmosis (in this case the transmembrane pressure was between 12 and 2 bar), a water recovery rate (permeate volume compared to the feed volume) of more than 95% and COD-rejection of above 9% were achieved. The membrane filtration system extended with a crystallization unit resulted in a higher water recovery of 97%. Also higher rejection of organics (COD) of about 92% was achieved. Rautenbach et al. [9] studied the effectiveness of the combination of reverse osmosis - nanofiltration - crystallisation - high pressure reverse osmosis in dumpsite leachate treatment. COD-rejection of approximately 96% was achieved. In treatment of textile wastewater, ultrafiltration membranes were applied as membrane bioreactors. The experiments by Badani et al. [1] showed an average COD reduction of 97%. Ten different types of membranes (microfiltration (MF), UF, NF, and RO) were examined by Bottino et al. [11] for the treatment of textile wastewater. The performances of the membranes regarding the COD retention were: UF: 21-77%, NF: 79-81%, RO: 89-91%. They experienced that the quality of the feed had an effect on the performance of the membranes. From these results, they proposed that at least a two-stage filtration would be required to effect the desired pollutant removal. Artiga et al. [12] applied a Zenon Membrane Bioreactor (MBR) with success in treating winery (test) and tannery (real) wastewaters. A COD removal efficiency above 97% was obtained with winery wastewater, and about 86% with tannery wastewater. In treatment of seafood processing wastewaters Ferjani et al. [13] achieved good results using UF and NF membranes: the COD reduction of UF membranes was 5-65%. In case of NF membranes, COD reductions of up to 93% were achieved. These works all demonstrate that membrane processes are being more frequently and efficiently used for wastewater treatment. Several studies compare the effectiveness of nanofiltration and reverse osmosis membranes for various wastewaters. Some of them show that the required water quality can always be achieved by membrane processes- if necessary in cascaded operation [5,8,9,11]. Since, in certain highly populated areas in Hungary biological treatment is not allowed by local authorities, the use of alternative wastewater treatment options is required. The present work is an experimental study aimed at comparing the efficiency of several nanofiltration and reverse osmosis membranes in treating wastewaters from several origins (pharmaceutical wastewater and dumpsite leachate) and reducing their COD. The previous works deal with the treatment of one wastewater using different membranes, but in this manuscript the treatments of two different wastewaters using different membranes are presented here. 2. Experimental Procedures 2.1. Laboratory scale membrane apparatus The experiments are carried out in a 3DTA type crossflow test-membrane apparatus. Schematic drawing of the cross-flow equipment can be seen in Fig

3 E. Cséfalvay, P.M. Imre, P. Mizsey Table 1. Optimal operating conditions of the applied membranes. Type Operating ph Operating Pressure ( 1 5 Pa) Optimal Operating Pressure ( 1 5 Pa ) Maximum operating temperature ( C) NF I NF II RO * * applied for the filtration of monovalent ion solutions Table 2. Parameters of the wastewaters examined. Molecular weight cut-off for uncharged organic molecules Wastewater of Origin ph Conductivity (ms cm -1 ) COD (mg L -1 ) high COD Pharmaceutical industry low COD Dumpsite leachate The wastewater is pumped from the feed tank into the membrane unit with a high pressure pump. Flat sheet NF and RO membranes with an effective area of.15 m 2 are tested. The experiments are carried out in a continuous mode. The wastewater feed is pumped continuously into the membrane unit where it is circulated above the membrane surface to avoid fouling of the membrane and also to investigate crossflow filtration. The velocity of the fluid on the membrane surface is 1.9 m s -1. The permeate leaves the membrane continuously. The retentate flow is also taken away to close the mass balance. Before starting the experiments, the membranes are washed with distilled water. During the experiments, two different types of wastewater are treated: wastewater of high COD and wastewater of low COD. The wastewater of high COD is treated by three membranes, two NF and one RO membrane. Parallel experiments are carried out with each membrane. The two NF membranes are tested at two different pressures: near to the lowest and near to highest applicable pressures recommended by the membrane supplier. The RO membrane is tested at a pressure in the range of optimal operating pressure with two different feed. For wastewater of low COD, the same membranes are tested. Preliminary experiments demonstrated that the membrane rejection did not depend on the applied pressure. Therefore only one pressure in the range of the optimal operating pressure recommended for each membrane by the membrane supplier was utilized Two-stage membrane filtration Experiments were also carried out on a two-stage membrane filtration set. First the original wastewater of high COD is treated by NF, then the permeate of the NF is fed and treated forward by RO. In this way a twostage filtration is carried out, nanofiltration followed by reverse osmosis. The wastewater of low COD also was treated by a two-stage filtration set, that is, after nanofiltration a reverse osmosis was carried out Membranes Two NF membranes (NF I., NF II.) and a RO membrane are tested under controlled circumstances. All three membranes are produced and supplied by Osmonics SA (Vista CA, USA). All of them are thin film composite (TFC) membranes. The optimal operating conditions of the membranes can be seen in Table Composition of wastewaters The characteristics of the high and low COD wastewater streams were measured and reported in Table 2. The exact chemical nature of the contaminants in both wastewater streams was not available. The aim of the experiments was to determine what treatments were required to effect the reduction of the COD-value of different origin wastewaters to the release limits prescribed in the Council Directive 76/464/EEC and to compare the filtration effectiveness of different membranes Experimental conditions and calculated data Both NF and RO experiments are carried out. Permeate flux is measured as a typical parameter of filtration. During all experiments samples are taken from the permeate and analyzed. Temperature, ph, and conductivity are continuously measured by a WTW 34i ph/conductivity meter. COD is determined by the K 2 Cr 2 O 7 standard method. This method is equivalent in technical content and fully corresponds to the International Standard, ISO 66:1989. The applied operating pressures for each membrane for the two wastewaters can be seen in Table 3. During the experiments with wastewater of 279

4 Applicability of Nanofiltration and Reverse Osmosis for the Treatment of Wastewater of Different Origin Table 3. Operating conditions of pharmaceutical wastewater filtration. Wastewater of Parameter NF I. NF II. RO high COD Operating pressure ( 1 5 Pa) 5 and 3 5 and 34 4 low COD Operating pressure ( 1 5 Pa) Permeate fluxes of nanofiltration membranes Permeate fluxes of reverse osmosis membrane Permeate flux (L m -2 hr -1 ) NF I.-5*1^5 Pa NF I.-3*1^5 Pa NF II.-5*1^5 Pa NF II.-34*1^5 Pa Permeate flux (L m -2 hr -1 ) RO-4*1^5 Pa RO-4*1^5 Pa,NF II. permeate feed Figure 2. Permeate fluxes of NF membranes vs. operation time (wastewater of high COD). high COD using NF membranes pressures near to the lowest and the highest applicable values are used; using a RO membrane a pressure in the range of the optimal operating pressure is applied. In case of wastewater of low COD filtration, during the experiments, pressures within the range of the optimal interval are used. The rejection of the organic compounds is calculated according to Equation (1): c p 1 1 c R = (1) f where R is the rejection (%), c p and c f are the COD values of the permeate and the feed, respectively. CODrejection is calculated for each experiment. 3. Results and Discussion 3.1. Permeate flux Wastewater of high COD The results of flux measurements of the NF membranes can be seen in Fig. 2. Fluxes regarding both to the lowest and the highest applied pressures are indicated. At the beginning of the filtration, permeate flux decreases sharply as expected. Since membrane filtration is a pressure driven process, the flux increases with increasing pressure. In every case the system reaches steady state after one hour. Comparing the two NF membranes at the same low pressure (5 1 5 Pa) the NF II (35 L m -2 h -1 ) membrane shows higher flux than NF I. (15 L m -2 h -1 ), in contrast, at higher operating pressure the sequence is reversed and the NF I. membrane has higher flux (7 L m -2 h -1 ) than NF II (57 L m -2 h -1 ). It can be also concluded that the flux of the NF II. membrane is less pressure dependent than that of NF I Figure 3. Permeate fluxes of reverse osmosis membrane vs. operation time (wastewater of high COD). Fig. 3 shows the results of flux measurements of the RO membrane. It is tested within the pressure range of optimal operating pressure but with two input-flows. Not only the wastewater but the nanofiltration permeate obtained with NF II are processed. The permeates of NF I and NF II filtrations have similar character regarding their COD values. It can be stated that, in the current experiments, the permeate flux of the RO membrane does not depend on the feed composition: the fluxes are almost the same in steady state (in case of the original wastewater with a COD value of 165 mg L -1, and also in case of NF II. permeate with a COD value of 93 mg L -1 ). The decrease of the initial permeate flux can be observed also at RO membranes. Steady state can be reached after about one hour operation Wastewater of low COD Fig. 4 shows the measured permeate fluxes both for NF and RO membranes. Since the experiments carried out on the wastewater of high COD show that the COD-rejections of the membranes do not depend significantly on the applied pressure, therefore in the case of the wastewater of low COD filtration the optimal operating pressures are applied for each membrane. It can be clearly seen that NF membranes have higher fluxes than RO membranes. The two NF membranes have very similar fluxes of 14 L m -2 h -1 in the steady state phase. This value is 2-4 times higher than for filtration of wastewater of high COD. The RO membrane shows surprisingly high flux (8 L m -2 h -1 ) compared to the flux of the first wastewater filtration (3-5 L m -2 h -1 ) in spite of the lower pressure applied. These differences can be explained by the very different content of the wastewaters. The initial conductivity is at least five times higher and the initial COD is 1 times higher in the 28

5 E. Cséfalvay, P.M. Imre, P. Mizsey Permeate flux (L m -2 hr -1 ) Permeate fluxes Figure 4. Permeate fluxes vs. operation time for wastewater of low COD. COD (mg L -1 ) NF I.-15*1^5 Pa NF II.-11*1^5 Pa RO-27*1^5 Pa-NF II. permeate feed Figure 5. Permeate COD vs. operation time (wastewater of high COD). COD (mg L -1 ) COD rejection (%) Chemical Oxygen Demand Chemical Oxygen Demand Average COD rejection NF I.-5bar NF I.-3bar NF II.-5bar NF II.-34bar RO-4bar RO-4bar- NF II. permeate feed NF I.-15 bar NF II.-11 bar RO-27 bar-nf II. permeate feed Figure 6. Permeate COD vs. operation time (wastewater of low COD). NF I.-5bar NF I.-3bar NF II.-5bar NF II.-34bar RO-4bar RO-4 bar, NF II. permeate feed Figure 7. Average COD rejection of the applied membranes in case of pharmaceutical industry (wastewater of high COD). case of pharmaceutical wastewater (wastewater of high COD) than in the case of dumpsite leachate (wastewater of low COD), showing a higher contamination Permeate conductivity Conductivity can be used to indicate membrane rejection if wastewaters with salt content are filtered. Because the conductivity can be easily measured, it is continuously recorded during the experiments. For the wastewater of high COD, the NF membranes decreased the conductivity from 5 ms cm -1 to ms cm -1 then the conductivity remained constant. Consequently this wastewater probably contained monovalent ions, which cannot be rejected by NF membranes. Salt retention by the RO membrane was the highest out of the tested membranes, the conductivity decreased to 2% of the initial value. For wastewater of low COD, the membranes showed similar results. Using NF membranes the conductivity did not change significantly. Probably this wastewater also contained monovalent ions, which pass through the NF membranes. The RO membrane, however, decreased the conductivity to 9% of the initial value. The difference between the salt rejections of the RO membrane can be explained by the large difference in the initial conductivity. If the salt concentration of the feed is higher, the rejection is lower, since the increased pressure makes the ions pass through the membrane Permeate COD Since the latest environmental regulations urge industrial companies to pay more attention to environmental protection, appropriate wastewater treatment is needed to meet the emission level limit for each pollutant. In Hungary the wastewater release limit for COD is 1 mg L -1. (28/24. (XII. 25) KvVM Degree, Appendix 4, which harmonized the Council Directive 76/464/EEC). On the other hand, biological treatment to reduce the COD is not always allowed and also may not be effective. Consequently membrane filtration has been tested for this issue Wastewater of high COD Permeate COD versus operation time can be seen in Fig. 5. During the experiments COD decreases to a certain value and then remains practically constant. The NF membranes reduce the COD to a level between 8-1 mg L -1. The RO membrane can decrease the COD from the initial value of 165 mg L -1 to 16 mg L -1, that is to 1% of the feed. If the permeate of the NF II. membrane is fed onto the RO membrane the COD value is decreased from 93 mg L -1 to 1 mg L -1, effecting the same 9% COD rejection. As expected, the RO membrane was the best performing of the tested membranes at reducing the COD. 281

6 Applicability of Nanofiltration and Reverse Osmosis for the Treatment of Wastewater of Different Origin COD Rejection (%) Average COD Rejection NF I.-15bar NF II.-11bar RO-27bar-NF II. permeate feed Figure 8. Average COD rejection of the applied membrans in case of dumpsite leachate (wastewater of low COD) Wastewater of low COD There is a special regulation for the content of dumpsite leachate (wastewater of low COD) flowing into surface water. The release limit for dumpsite leachate can be found in Hungarian regulation in 28/24. (XI.12) KvVM Degree, Part 3, Chapter 35 (Harmonisation of Council Directive 76/464/EEC). The recommended COD-value is 2 mg L -1. Permeate COD versus operation time can be seen in Fig. 6. Neither of the two NF membranes could decrease the COD value to the release limit, therefore the permeate of one nanofiltration membrane, namely the permeate of NF II. was fed onto the RO membrane. Due to this cascaded operation, the COD can be reduced to ~7 mg L -1, significantly under the emission limit Average COD rejection Wastewater of high COD The average COD rejection of the investigated membranes applied to wastewater of high COD can be seen in Fig. 7. NF membranes have rejection rates from 35 to 5%. NF I. and NF II. membranes are characterized by an approximate molecular weight cutoff of 15-3 Da for uncharged organic molecules. Since the wastewater contains about 1 wt% of organic and inorganic salt, the charged components probably change the filtration behaviour of the membrane. Therefore the COD rejection rate of the membrane declines. With increasing operating pressure the CODrejection rate very slightly increases. This does not press the application of high operating pressure. The RO membrane shows the best COD-rejection (about 9%) out of the tested membranes. The use of two step filtration, NF II. followed by RO, showed little improvement when compared to the one-step reverse osmosis alternative. Therefore, the application of a two step filtration seems to be unwarranted for the pharmaceutical wastewater. That is, further action will be required to meet the required limit, e.g. preliminary treatment of this kind of wastewater and/or a third membrane filtration step Wastewater of low COD Fig. 8 shows the average COD rejection for wastewater of low COD filtration. It can be clearly seen that the nanofiltration membranes reduce the COD value by about 2%. In the case of two stage filtration, when the permeate of the NF II. is fed onto the RO membrane, the COD-rejection is 85%. Out of the three tested membranes, the RO membrane shows the best performance by far. Components influencing the COD of the NF membranes are probably molecules smaller than the molecular weight cut-off of the applied membranes which present in the permeate. A two step filtration (NF+RO) can be recommended in this case. 4. Conclusions Two wastewaters of different origin (wastewater from pharmaceutical industry and dumpsite leachate) are treated by membrane filtration with the aim of reaching the recommended chemical oxygen demand release limit and to test the membranes to determine if they can be applied under extremely different circumstances. Permeate flux, conductivity, and COD are measured during the experiments. We conclude that: nanofiltration membranes have higher fluxes than reverse osmosis membranes; the applied pressure influences the permeate flux; for NF membranes the higher the applied pressure, the higher the rejection of COD independently the initial COD of the wastewaters; COD rejection of reverse osmosis membranes is higher than nanofiltration membranes; COD rejection (in percent) of the reverse osmosis membrane does not depend on the initial COD; the composition of the feed (and in particular the ions present) affects not only the permeate flux but the COD rejection as well; two-step filtration (NF+RO) seems to be unwarranted for the treatment of the pharmaceutical wastewater (wastewater of high COD) and alternative actions will be required to meet the regulatory release limit; two-step filtration can be, however, recommended for the dumpsite leachate (wastewater of low COD); all of the membranes tested can be applied for the treatment of wastewaters with very different COD. Since the COD can not be reduced to the desirable level in the case of pharmaceutical wastewater, another type of cascaded operation (e.g. RO followed by RO) or/and prior wastewater treatment might be needed. Another efficient solution could be the ozonation of the wastewater, which is a promising alternative in wastewater treatment. 282

7 E. Cséfalvay, P.M. Imre, P. Mizsey The experiments prove that membrane filtration can be successfully applied to wastewater treatment in quite different areas, especially for dumpsite leachate, and it should be seriously considered for further applications. The experiments also proved that the membranes are stable and can be applied under extremely different circumstances. 5. Nomenclature Acknowledgements This study was partly supported by the grants OTKA T46218 and OTKA T426 of the Hungarian Scientific Foundation and by the grant of the Hungarian Academy of Sciences, F513/1/22/ The authors wish to thank to Gedeon Richter Pharmaceutical Industry for their contribution. MBR COD (mg L -1 ) R (%) c p (mg L -1 ) c f (mg L -1 ) membrane bioreactor chemical oxygen demand determined by the K 2 CrO 7 standard method (International Standard ISO 66:1989) the rejection of the applied membrane the COD value of the permeate the COD value of the feed References [1] W. Ho, K. Sirkar, Membrane Handbook, (Chapman&Holl, New York, USA, 1992) [2] S. S. Madaeni, Y. Mansourpanah, Filtr. Separat. 4, 4 (23) [3] W. Reimann, I.Yeo, Desalination 19, 263 (1997) [4] G. Del Re, G. Di Giacomo, L. Aloisio, M. Terreri, Desalination 119, 25 (1998) [5] B. Balannec, M. Vourch, M. Rabiller-Baudry, B. Chaufer, Sep. Purif. Technol. 42, 195 (25) [6] O. Akoum, M. Y. Jaffrin, L. H. Ding, M. Frappart, J. Membrane Sci. 235,111 (24) [7] J. Castelblanque, F. Salimbeni, Desalination 126, 293 (1999) [8] R. Rautenbach, Th. Linn, Desalination 15, 63 (1996) [9] R. Rautenbach, K. Vossenkaul, T. Linn, T. Katz, Desalination 18, 247 (1996) [1] Z. Badani, H. Ait-Amar, A. Si-Salah, M. Brik, W. Fuchs, Desalination 185, 411 (25) [11] A. Bottino, G. Capannelli, G. Tocchi, M. Marcucci and G. Ciardelli, Membrane Technology, 21, 9 ( 21) [12] P. Artiga, E. Ficaram, F. Malpei, J.M. Garrido, R. Méndez, Desalination 179, 161, (25) [13] E. Ferjani, E. Ellouze, R. Ben Amar, Desalination 177, 43 (25) 283