- Email: [email protected]
Removal of organic pollutants and analysis of MLSS–COD removal relationship at different HRTs in a submerged membrane bioreactor
ARTICLE IN PRESS
International Biodeterioration & Biodegradation 55 (2005) 279–284 www.elsevier.com/locate/ibiod
Removal of organic pollutants and analysis of MLSS–COD removal relationship at different HRTs in a submerged membrane bioreactor Nanqi Rena, Zhaobo Chena,, Aijie Wanga, Dongxue Hub a
School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China b School of Mathematics and Computer Science, Harbin Normal University, Harbin 150080, China
Abstract In order to investigate the influence of hydraulic retention time (HRT) on organic pollutant removal in a submerged membrane bioreactor (SMBR), a laboratory-scale experiment was conducted using domestic sewage as influent. The dissolved oxygen (DO) concentration was controlled at 2.0–3:0 mg L1 during the experimental period. The experiments demonstrated that when HRT was 3, 2 and 1 h, the reduction of chemical oxygen demand (COD) was 89.3–97.2, 88.5–97.3 and 80–91.1%, and the effluent COD was 38.9–11.2, 41.6–10.8 and 63.4–35:8 mg L1 , respectively. It is suggested that an HRT of 1 h could meet the normal standard of discharged domestic sewage, and an HRT of 2 h could meet that of water reclamation. In addition, we use mathematical software MATLAB to analyse the relation of mixed liquor suspended solids (MLSS) and COD removal. The results showed that the optimum MLSS concentration should be maintained at around 6000 mg L1 in the SMBR. The results also showed that the COD removal was related to HRT ðtÞ, influent concentration ðS 0 Þ and sludge loading rate for COD removal (N S ). Moreover, the high COD removal could be achieved through adjusting t, S0 and N S . r 2005 Elsevier Ltd. All rights reserved. Keywords: Submerged membrane bioreactor (SMBR); Domestic sewage; HRT; MLSS; COD removal
1. Introduction It is well known that submerged membrane bioreactors (SMBRs) have the following advantages for wastewater treatment: high sludge concentration (Halil Hasara et al., 2002), high quality of effluent, long contact time between activated sludge and organic pollutants (Brindle and Stephenson, 1996), and complete separation of the hydraulic retention time (HRT) and sludge retention time (SRT) (Ueda et al., 1996; Devies et al., 1998; Guender and Krauth, 1998). Moreover, highly treated water in an SMBR is free from bacteria and has potential for municipal and industrial reuse (Xing et al., 1998). Although there are shortcomings of high-cost and high-energy consumption, SMBR technology has been applied to wastewater treatment and Corresponding author. Fax: +86 451 8628 2009.
E-mail address: [email protected] (Z. Chen). 0964-8305/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibiod.2005.03.003
reclamation previously (Chiemchaisri et al., 1993; Knoblock et al., 1994; Trouve et al., 1994). In Europe, America and Japan, SMBRs are used to rebuild sewage treatment plants and to reclaim wastewater. It is accepted that HRT is the key to further improving the capacity of an SMBR. At present, when an SMBR is used for domestic sewage treatment, the HRT is set at 1.5–7.5 h in laboratory-scale tests and at 2.7–34.2 h in the pilot-scale tests (Makoto et al., 1998; Urbain et al., 1998; Defrance and Jaffrin, 1999; Huang et al., 2000; Gu and He, 2002; Shim et al., 2002). When Rosenberger et al. (2002) used a membrane bioreactor to treat municipal wastewater at HRTs varying from 10.4 and 15.6 h, the concentration of mixed liquor suspended solids (MLSS) gradually increased and the chemical oxygen demand (COD) was reduced by 95%. Until now, the lowest HRT, 1.5 h, has been designed by Stefan and Walter (2001) to treat synthetic wastewater. The organic loading rate (OLR) in their study was in the
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
280
range 6.0–13:0 kg m3 day1 , and COD reduction was 495%. However, little has been reported on the optimum HRT needed to meet reused water quality standard, and how to control the operational conditions of SMBRs in order to reach different water quality standards. The purpose of this study was to investigate the shortest HRT needed in SMBRs in order to reach different water quality standards and the effect of HRT on COD removal. A laboratory-scale experiment was conducted using artificial domestic sewage as influent in an SMBR. The dissolved oxygen (DO) concentration was controlled at 2.0–3:0 mg L1 for the duration of the experimental period, and the HRT at 3, 2 and 1 h. The effect of MLSS on COD removal at different HRTs is examined.
fibre membrane module. Aeration was employed to maintain an aerobic environment for the normal growth of activated sludge. The amount of air was adjusted using a gas flow meter and controlled at 0.3–0:5 m3 h1 . The water level in the SMBR was controlled by a ballcock in the water balance tank, which balanced the flux of effluent and influent. 2.2. Substrate The artificial domestic sewage containing ðmg L1 Þ glucose (300–400), beef grease (20–40), peptone (20–40), NH4 Cl (5–10), Na2 HPO4 (5–10) and NaH2 PO4 (5–10) was used as influent. The influent was maintained at pH 7.0 by adding NaOH. Influent COD concentration was 350–500 mg L1 . 2.3. Inoculation and acclimation of activated sludge
2. Materials and methods 2.1. SMBR The plexiglass SMBR (Fig. 1) had a working volume of 7.0 L. The hollow polypropylene fibre membrane module employed in this study was 0.5 m long and had a pore size of 0:1 mm and surface area of 2:0 m2 . The process was maintained at 20–25 C. The height was from liquid surface in the bioreactor to effluent port. The mixed liquor in the bioreactor was driven under the height by gravitation and passed through the hollow
Activated sludge was taken from the aeration pool in Harbin Refinery and was incubated in batch culture. After 7 days, MLSS reached 1858 mg L1 . An old membrane module (operated for 1 year in the same bioreactor) was added to the bioreactor to operate the SMBR continuously. Seven days later, MLSS reached 2820 mg L1 . The old membrane module was taken from the bioreactor and a new membrane module was added to the bioreactor to operate the SMBR. With incubation and acclimation for 14 days, the colour of the flocs changed to a brown colour. The amount of
Backwashing device
Thermometer Ballcock Hollow fibre membrane
Water balance tank
Sample port
Height Pump
Feed tank Air pump
Gas flow meter
SMBR Air diffuser
Effluent Fig. 1. Schematic diagram of the SMBR.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
protozoa, such as rotifer species was observed through a microscope. Effluent COD was around 80 mg L1 . Based on the above, it was judged that the process of incubation and acclimation was completed. 2.4. SMBR operational conditions To maintain a stable HRT, an air-driven valve linked to an electromagnetic flow meter automatically controlled the flow rate of effluent. Dissolved oxygen was maintained at 2.0–3:0 mg L1 by adjustment of a rotary flow meter. Three membrane bioreactors operated synchronously in this study. The flux in the three bioreactors was 38.8, 58.3 and 116:7 ml min1 when the HRT was 3, 2 and 1 h, respectively. 2.5. Analytical methods Standard methods (APHA, 1995) were used to determine COD, MLSS, DO and pH values. The relation between MLSS and COD removal was analysed using the mathematical software program MATLAB, which from the literature does not appear to have been used previously for analysis of this relationship in the SMBRs.
3. Results and discussion 3.1. Effect of HRT on COD removal The changes in COD removal and effluent COD values with time for each of the three HRT settings are illustrated in Fig. 2. When the HRT was 3 h, the highest COD removal was 97.2%, after 50 days operation. It is
281
viewed that the removal of organic pollutants was a cofunction of microbial metabolism and membrane filtration. The effluent COD concentration reached a level of o30 mg L1 , and even fell to 10 mg L1 , although the influent COD fluctuated from 350 to 500 mg L1 . When the HRT was 2 h, the COD removal fluctuated between 88.5% and 97.3% and the effluent COD between 41.6 and 10:8 mg L1 . COD removal was therefore very high when the HRT was either 2 or 3 h, and the effluent COD was less than 30 mg L1 . This is sufficient to meet the reused water quality standard in China. When the HRT was 1 h, COD removal increased with the increase in MLSS. The COD removal fluctuated from 80.0% to 91.1% and effluent COD from 63.4 to 35:8 mg L1 . This meets the water quality standard for discharge in China. It can therefore be concluded that the removal of organic pollutants was high and stable when SMBR technology was applied to treat domestic sewage.
3.2. Effect of MLSS on the COD removal at different HRTs The effect of MLSS on COD removal (Fig. 3) showed the same trend at different HRTs. The mathematical analysis of the relations between MLSS and COD removal was developed based on the experimental data to describe this trend using software of MATLAB. A fourth-order polynomial function (1) was considered to simulate the data in Fig. 3 y ¼ a4 x4 þ a3 x3 þ a2 x2 þ a1 x þ a0 .
(1)
Unknown coefficients of a0 , a1 , a2 , a3 and a4 needed to be calculated by the equation group (2), where a matrix with ten rows and five columns showed the coefficients. 100
100
120
90
100 80
85
60 80 40
COD removal (%)
95
95
Effluent COD(mg L−1)
COD removal (%)
140
90
85 HRT3h
80
HRT2h
75
20
70 30
32
34
36
38
40
42
44
46
48
50
0 52
time (day) COD removal(HRT3h)
COD removal(HRT2h)
COD removal(HRT1h)
Effluent COD(HRT3h)
Effluent COD(HRT2h)
Effluent COD(HRT1h)
Fig. 2. COD removal at different HRTs.
HRT1h
75
70 500
2500
4500
6500
MLSS (mg
8500
10500
L-1)
Fig. 3. Effect of MLSS on COD removal at different HRTs.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
282
2
y1
3
2
1 x1 6 y 7 61 x 6 27 6 2 7 6 6 6 y3 7 6 1 x 3 7 6 6 7 6 6 6 y4 7 6 1 x 4 7 6 6 6 y5 7 6 1 x 5 7 6 6 7¼6 6 6 y6 7 6 1 x 6 7 6 6 6 y7 7 6 1 x 7 7 6 6 7 6 6 6 y8 7 6 1 x 8 7 6 6 6 y 7 6 1 x9 4 95 4 y10 1 x10
x21 x22
x31 x32
x23 x24
x33 x34
x25
x35
x26 x27
x36 x37
x28 x29
x38 x39
x210
x310
3 x41 x42 7 7 7 4 7 x3 7 2 3 a0 7 x44 7 6 7 7 6 a1 7 x45 7 7 7 6 6a 7 4 76 2 7. x6 7 6 7 7 4 a3 5 x47 7 7 a4 7 x48 7 7 x49 7 5 x410
(2)
MATLAB would output a row vector of the polynomial (Gander, 1999; Ren et al., 2001; Su, 2002). The calculation of coefficients showed as follows: When the HRT was 1 h, a0 ¼ 74:000, a1 ¼ 0:0035, a2 ¼ 3:7e 007, a3 ¼ 1:1e 010 and a4 ¼ 5:9e 015; when the HRT was 2 h, a0 ¼ 77:000, a1 ¼ 0:0023, a2 ¼ 3:3e 006, a3 ¼ 6:2e 010 and a4 ¼ 3:4e 014; and when the HRT was 3 h, a0 ¼ 85:000, a1 ¼ 0:0082, a2 ¼ 7:6e 006, a3 ¼ 1:7e 009 and a4 ¼ 1:1e 013. Consequently, the following three mathematical models could be obtained with regard to the effect of MLSS on COD removal at different HRT:
Z ¼ k2 M 2 k1 M þ DZ.
(4)
(iii) Supposing N s stands for sludge removal loading rate, S 0 for influent COD, S e for effluent COD and t stands for hydraulic retention time. According to the equation for active sludge kinetics, N S ¼ ðS 0 S e Þ=ðM tÞ, it can be deduced that
y1 ¼ 5:9e 015x4 1:1e 010x3 þ 3:7e 007x2 0:0035x þ 74:000, y2 ¼ 3:4e 014x4 6:2e 010x3 þ 3:3e 006x2 0:0023x þ 77:000,
M¼
y3 ¼ 1:1e 013x4 1:7e 009x3 þ 7:6e
S0 Se . NS t
(5)
Accordingly, the mathematical model for COD removal with the change of sludge loading rate:
006x2 0:0082x þ 85:000. The comparison of simulated results and the original data (Fig. 4) demonstrated that the fourth-order polynomial was very stable. In order to analyse the relation of COD removal and MLSS more directly, Z and M respectively replaced y and x, and the following mathematical model was obtained. Z ¼ k4 M 4 k3 M 3 þ k2 M 2 k1 M þ DZ.
Here Z is COD removal, M is MLSS, k1 , k2 , k3 , k4 are constants, and DZ is a constant item, which is defined by COD removal when M is equal to zero. Table 1 shows the mathematical model and correlative mathematical parameters of the effect of MLSS on COD removal at different HRTs in this experiment, from which it is concluded that: (i) Fig. 4 and Eq. (3) show that Z increased with the increase of M. When M increased to 6000 mg L1 , the trend of Z gradually became stable. At the start, the membrane bioreactor MLSS was very low, and consequently, the microbial metabolism and COD removal were quite low. After that, the sludge continued to increase and COD removal also increased. However, when the MLSS arrived at a particular value ðMLSS ¼ 6000 mg L1 Þ COD removal and effluent COD were stable, as a result of a stable gelatin layer at the surface of the membrane. Therefore, the simulation results for this mathematical model showed that the optimum MLSS should be maintained at around 6000 mg L1 in the SMBR. (ii) Because of k4 5k3 5k2 , the fourth degree equation could be replaced by the second degree equation
(3)
Z ¼ k2
ðS0 Se Þ2 S0 Se þ DZ. k1 2 2 NS t NS t
(6)
Formula (6) shows that COD removal ðZÞ was related to HRT ðtÞ, the COD of influent ðS 0 Þ and sludge loading rate ðN s Þ. Perfect COD removal could be gained by adjusting parameters such as t, S 0 and N s . In addition, the coefficients of simulation and norm of residuals at different HRTs were used to validate the
Fig. 4. Curve simulation results at HRT of (a) 1 h (b), 2 h and (c) 3 h.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
283
Table 1 Correlative mathematical parameters and models of COD and MLSS removal at different HRTs HRT (h)
DZ
Mathematical model
k=k1 , k2 , k3 , k4
1 2 3
74 77 85
y1 ¼ 5:9e 015x4 1:1e 010x3 þ 3:7e 007x2 0:0035x þ 74:000 y2 ¼ 3:4e 014x4 6:2e 010x3 þ 3:3e 006x2 0:0023x þ 77:000 y3 ¼ 1:1e 013x4 1:7e 009x3 þ 7:6e 006x2 0:0082x þ 85:000
0:0035 3:7e 007 1:1e 010 5:9e 015 0:0023 3:3e 006 6:2e 010 3:4e 014 0:0082 7:6e 006 1:7e 009 1:1e 013
removal gradually stabilised. It was demonstrated that the optimum MLSS in an SMBR should be maintained at around 6000 mg L1 . (3) Formula, Z ¼ k2
ðS0 Se Þ2 S0 Se þ DZ, k1 2 2 NS t NS t
showed that COD removal ðZÞ was related to HRT ðtÞ, influent COD ðS 0 Þ and sludge removal loading rate ðN s Þ. Perfect removal of COD could be obtained by adjusting these three parameters.
Acknowledgements The present publication has been made possible through the financial, technical and administrative assistance of China National ‘‘863’’ Hi-Tech R & D Program.
References
Fig. 5. Residuals of curve simulation at HRT of (a) 1 h (b), 2 h and (c) 3 h.
correctness of our mathematical models. Fig. 5 indicates that when the HRT was 3, 2 or 1 h all the norms of residuals were o3:0 (Wen et al., 2000; Su, 2002), proving that the model could simulate the experimental data correctly.
4. Conclusions (1) When the SMBR was used to treat domestic sewage, the experimental results showed that an HRT of 1 h could meet the normal standard for discharged domestic sewage, and an HRT of 2 h could meet that for water reclamation. (2) Mathematical analysis of the experimental data showed that COD removal increased with increase of MLSS. When MLSS increased to 6000 mg L1 , COD
APHA,1995. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, Washington, DC. Brindle, K., Stephenson, T., 1996. Application of membrane biological reactors for the treatment of wastewaters. Biotechnology and Bioengineering 49, 601–610. Chiemchaisri, C., Yamamoto, K., Vigneswaran, S., 1993. Household membrane bioreactor in domestic wastewater treatment. Water Science and Technology 27, 171–178. Defrance, L., Jaffrin, M.Y., 1999. Comparison between filtrations at fixed transmembrane pressure and fixed permeate flux: application to a membrane bioreactor used for wastewater treatment. Journal of Membrane Science 152, 203–210. Devies, W.J., Le, M.S., Heath, C.R., 1998. Intensified activated sludge process with submerged membrane microfiltration. Water Science and Technology 38, 421–428. Gander, W., 1999. Solving Problems in Scientific Computing Using Maple and MATLAB. China Higher Education Press Beijing/ Springer, Berlin. Gu, G.W., He, Y.L., 2002. Study and Application of the Membrane Bioreactor in Wastewater Treatment. China Chemistry Industry Press, Beijing. Guender, B., Krauth, Kh., 1998. Replacement of secondary clarification by membrane separation: results with plate and hollow fibre modules. Water Science and Technology 38, 383–393. Halil, H., Cumali, K., Ayhan, U., 2002. Viability of microbial mass in a submerged membrane bioreactor. Desalination 150, 263–268.
ARTICLE IN PRESS 284
N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
Huang, X., Gui, P., Qian, Y., 2000. Performance of submerged membrane bioreactor for domestic wastewater treatment. Tsinghua Science and Technology 5, 237–240. Knoblock, M.D., Sutton, P.M., Mishra, P.N., Gupta, K., Janson, A., 1994. Membrane biological reactor system for treatment of oily wastewaters. Water Environment Research 66, 133–139. Makoto, O., Yutaka, O., Naoki, O., 1998. Decoloration system using rotation membrane UF module. Water Science and Technology 38, 35–43. Ren, N.Q., Zhang, Y., Chen, Z.B., 2001. The mathematical model analysis of membrane flux by using submerged bioreactor with different membrane forms. High Technology Letters 11, 100–103. Rosenberger, S., Kruger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36, 413–420. Shim, J.K., Yoo, I.K., Lee, Y.M., 2002. Design and operation considerations for wastewater treatment using a flat submerged membrane bioreactor. Process Biochemistry 38, 279–285.
Stefan, H., Walter, T., 2001. Treatment of urban wastewater in a membrane bioreactor at high organic loading rates. Journal of Biotechnology 92, 95–101. Su, X.S., 2002. Engineering Application of MATLAB 6.0. China Science Press, Beijing. Trouve, E., Urbain, V., Manem, J., 1994. Treatment of municipal wastewater by a membrane bioreactor: results of a semi-industrial pilot-scale study. Water Science and Technology 30, 151–157. Ueda, T., Hata, K., Kikuoka, Y., 1996. Treatment of domestic sewage from rural settlements by a membrane bioreactor. Water Science and Technology 34, 189–196. Urbain, V., Mobarry, B., Silva, V., Stahl, D.A., Rittmann, B.E., Manem, J., 1998. Integration of performance, molecular biology and modeling to describe the activated sludge process. Water Science and Technology 37, 223–229. Wen, X., Zhou, L., Wang, D.L., Xiong, X.Y., 2000. Application Design of MATLAB NN. China Science Press, Beijing. Xing, C.H. Qian, Y., Wen, X.H., Meng, Y.B., 1998. Performance of microfiltration membrane bioreactor for domestic wastewater reclamation. In: Proceedings of WEFTEC Asia ’98, January 1998, Singapore.
International Biodeterioration & Biodegradation 55 (2005) 279–284 www.elsevier.com/locate/ibiod
Removal of organic pollutants and analysis of MLSS–COD removal relationship at different HRTs in a submerged membrane bioreactor Nanqi Rena, Zhaobo Chena,, Aijie Wanga, Dongxue Hub a
School of Municipal and Environmental Engineering, Harbin Institute of Technology, Harbin 150090, China b School of Mathematics and Computer Science, Harbin Normal University, Harbin 150080, China
Abstract In order to investigate the influence of hydraulic retention time (HRT) on organic pollutant removal in a submerged membrane bioreactor (SMBR), a laboratory-scale experiment was conducted using domestic sewage as influent. The dissolved oxygen (DO) concentration was controlled at 2.0–3:0 mg L1 during the experimental period. The experiments demonstrated that when HRT was 3, 2 and 1 h, the reduction of chemical oxygen demand (COD) was 89.3–97.2, 88.5–97.3 and 80–91.1%, and the effluent COD was 38.9–11.2, 41.6–10.8 and 63.4–35:8 mg L1 , respectively. It is suggested that an HRT of 1 h could meet the normal standard of discharged domestic sewage, and an HRT of 2 h could meet that of water reclamation. In addition, we use mathematical software MATLAB to analyse the relation of mixed liquor suspended solids (MLSS) and COD removal. The results showed that the optimum MLSS concentration should be maintained at around 6000 mg L1 in the SMBR. The results also showed that the COD removal was related to HRT ðtÞ, influent concentration ðS 0 Þ and sludge loading rate for COD removal (N S ). Moreover, the high COD removal could be achieved through adjusting t, S0 and N S . r 2005 Elsevier Ltd. All rights reserved. Keywords: Submerged membrane bioreactor (SMBR); Domestic sewage; HRT; MLSS; COD removal
1. Introduction It is well known that submerged membrane bioreactors (SMBRs) have the following advantages for wastewater treatment: high sludge concentration (Halil Hasara et al., 2002), high quality of effluent, long contact time between activated sludge and organic pollutants (Brindle and Stephenson, 1996), and complete separation of the hydraulic retention time (HRT) and sludge retention time (SRT) (Ueda et al., 1996; Devies et al., 1998; Guender and Krauth, 1998). Moreover, highly treated water in an SMBR is free from bacteria and has potential for municipal and industrial reuse (Xing et al., 1998). Although there are shortcomings of high-cost and high-energy consumption, SMBR technology has been applied to wastewater treatment and Corresponding author. Fax: +86 451 8628 2009.
E-mail address: [email protected] (Z. Chen). 0964-8305/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.ibiod.2005.03.003
reclamation previously (Chiemchaisri et al., 1993; Knoblock et al., 1994; Trouve et al., 1994). In Europe, America and Japan, SMBRs are used to rebuild sewage treatment plants and to reclaim wastewater. It is accepted that HRT is the key to further improving the capacity of an SMBR. At present, when an SMBR is used for domestic sewage treatment, the HRT is set at 1.5–7.5 h in laboratory-scale tests and at 2.7–34.2 h in the pilot-scale tests (Makoto et al., 1998; Urbain et al., 1998; Defrance and Jaffrin, 1999; Huang et al., 2000; Gu and He, 2002; Shim et al., 2002). When Rosenberger et al. (2002) used a membrane bioreactor to treat municipal wastewater at HRTs varying from 10.4 and 15.6 h, the concentration of mixed liquor suspended solids (MLSS) gradually increased and the chemical oxygen demand (COD) was reduced by 95%. Until now, the lowest HRT, 1.5 h, has been designed by Stefan and Walter (2001) to treat synthetic wastewater. The organic loading rate (OLR) in their study was in the
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
280
range 6.0–13:0 kg m3 day1 , and COD reduction was 495%. However, little has been reported on the optimum HRT needed to meet reused water quality standard, and how to control the operational conditions of SMBRs in order to reach different water quality standards. The purpose of this study was to investigate the shortest HRT needed in SMBRs in order to reach different water quality standards and the effect of HRT on COD removal. A laboratory-scale experiment was conducted using artificial domestic sewage as influent in an SMBR. The dissolved oxygen (DO) concentration was controlled at 2.0–3:0 mg L1 for the duration of the experimental period, and the HRT at 3, 2 and 1 h. The effect of MLSS on COD removal at different HRTs is examined.
fibre membrane module. Aeration was employed to maintain an aerobic environment for the normal growth of activated sludge. The amount of air was adjusted using a gas flow meter and controlled at 0.3–0:5 m3 h1 . The water level in the SMBR was controlled by a ballcock in the water balance tank, which balanced the flux of effluent and influent. 2.2. Substrate The artificial domestic sewage containing ðmg L1 Þ glucose (300–400), beef grease (20–40), peptone (20–40), NH4 Cl (5–10), Na2 HPO4 (5–10) and NaH2 PO4 (5–10) was used as influent. The influent was maintained at pH 7.0 by adding NaOH. Influent COD concentration was 350–500 mg L1 . 2.3. Inoculation and acclimation of activated sludge
2. Materials and methods 2.1. SMBR The plexiglass SMBR (Fig. 1) had a working volume of 7.0 L. The hollow polypropylene fibre membrane module employed in this study was 0.5 m long and had a pore size of 0:1 mm and surface area of 2:0 m2 . The process was maintained at 20–25 C. The height was from liquid surface in the bioreactor to effluent port. The mixed liquor in the bioreactor was driven under the height by gravitation and passed through the hollow
Activated sludge was taken from the aeration pool in Harbin Refinery and was incubated in batch culture. After 7 days, MLSS reached 1858 mg L1 . An old membrane module (operated for 1 year in the same bioreactor) was added to the bioreactor to operate the SMBR continuously. Seven days later, MLSS reached 2820 mg L1 . The old membrane module was taken from the bioreactor and a new membrane module was added to the bioreactor to operate the SMBR. With incubation and acclimation for 14 days, the colour of the flocs changed to a brown colour. The amount of
Backwashing device
Thermometer Ballcock Hollow fibre membrane
Water balance tank
Sample port
Height Pump
Feed tank Air pump
Gas flow meter
SMBR Air diffuser
Effluent Fig. 1. Schematic diagram of the SMBR.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
protozoa, such as rotifer species was observed through a microscope. Effluent COD was around 80 mg L1 . Based on the above, it was judged that the process of incubation and acclimation was completed. 2.4. SMBR operational conditions To maintain a stable HRT, an air-driven valve linked to an electromagnetic flow meter automatically controlled the flow rate of effluent. Dissolved oxygen was maintained at 2.0–3:0 mg L1 by adjustment of a rotary flow meter. Three membrane bioreactors operated synchronously in this study. The flux in the three bioreactors was 38.8, 58.3 and 116:7 ml min1 when the HRT was 3, 2 and 1 h, respectively. 2.5. Analytical methods Standard methods (APHA, 1995) were used to determine COD, MLSS, DO and pH values. The relation between MLSS and COD removal was analysed using the mathematical software program MATLAB, which from the literature does not appear to have been used previously for analysis of this relationship in the SMBRs.
3. Results and discussion 3.1. Effect of HRT on COD removal The changes in COD removal and effluent COD values with time for each of the three HRT settings are illustrated in Fig. 2. When the HRT was 3 h, the highest COD removal was 97.2%, after 50 days operation. It is
281
viewed that the removal of organic pollutants was a cofunction of microbial metabolism and membrane filtration. The effluent COD concentration reached a level of o30 mg L1 , and even fell to 10 mg L1 , although the influent COD fluctuated from 350 to 500 mg L1 . When the HRT was 2 h, the COD removal fluctuated between 88.5% and 97.3% and the effluent COD between 41.6 and 10:8 mg L1 . COD removal was therefore very high when the HRT was either 2 or 3 h, and the effluent COD was less than 30 mg L1 . This is sufficient to meet the reused water quality standard in China. When the HRT was 1 h, COD removal increased with the increase in MLSS. The COD removal fluctuated from 80.0% to 91.1% and effluent COD from 63.4 to 35:8 mg L1 . This meets the water quality standard for discharge in China. It can therefore be concluded that the removal of organic pollutants was high and stable when SMBR technology was applied to treat domestic sewage.
3.2. Effect of MLSS on the COD removal at different HRTs The effect of MLSS on COD removal (Fig. 3) showed the same trend at different HRTs. The mathematical analysis of the relations between MLSS and COD removal was developed based on the experimental data to describe this trend using software of MATLAB. A fourth-order polynomial function (1) was considered to simulate the data in Fig. 3 y ¼ a4 x4 þ a3 x3 þ a2 x2 þ a1 x þ a0 .
(1)
Unknown coefficients of a0 , a1 , a2 , a3 and a4 needed to be calculated by the equation group (2), where a matrix with ten rows and five columns showed the coefficients. 100
100
120
90
100 80
85
60 80 40
COD removal (%)
95
95
Effluent COD(mg L−1)
COD removal (%)
140
90
85 HRT3h
80
HRT2h
75
20
70 30
32
34
36
38
40
42
44
46
48
50
0 52
time (day) COD removal(HRT3h)
COD removal(HRT2h)
COD removal(HRT1h)
Effluent COD(HRT3h)
Effluent COD(HRT2h)
Effluent COD(HRT1h)
Fig. 2. COD removal at different HRTs.
HRT1h
75
70 500
2500
4500
6500
MLSS (mg
8500
10500
L-1)
Fig. 3. Effect of MLSS on COD removal at different HRTs.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
282
2
y1
3
2
1 x1 6 y 7 61 x 6 27 6 2 7 6 6 6 y3 7 6 1 x 3 7 6 6 7 6 6 6 y4 7 6 1 x 4 7 6 6 6 y5 7 6 1 x 5 7 6 6 7¼6 6 6 y6 7 6 1 x 6 7 6 6 6 y7 7 6 1 x 7 7 6 6 7 6 6 6 y8 7 6 1 x 8 7 6 6 6 y 7 6 1 x9 4 95 4 y10 1 x10
x21 x22
x31 x32
x23 x24
x33 x34
x25
x35
x26 x27
x36 x37
x28 x29
x38 x39
x210
x310
3 x41 x42 7 7 7 4 7 x3 7 2 3 a0 7 x44 7 6 7 7 6 a1 7 x45 7 7 7 6 6a 7 4 76 2 7. x6 7 6 7 7 4 a3 5 x47 7 7 a4 7 x48 7 7 x49 7 5 x410
(2)
MATLAB would output a row vector of the polynomial (Gander, 1999; Ren et al., 2001; Su, 2002). The calculation of coefficients showed as follows: When the HRT was 1 h, a0 ¼ 74:000, a1 ¼ 0:0035, a2 ¼ 3:7e 007, a3 ¼ 1:1e 010 and a4 ¼ 5:9e 015; when the HRT was 2 h, a0 ¼ 77:000, a1 ¼ 0:0023, a2 ¼ 3:3e 006, a3 ¼ 6:2e 010 and a4 ¼ 3:4e 014; and when the HRT was 3 h, a0 ¼ 85:000, a1 ¼ 0:0082, a2 ¼ 7:6e 006, a3 ¼ 1:7e 009 and a4 ¼ 1:1e 013. Consequently, the following three mathematical models could be obtained with regard to the effect of MLSS on COD removal at different HRT:
Z ¼ k2 M 2 k1 M þ DZ.
(4)
(iii) Supposing N s stands for sludge removal loading rate, S 0 for influent COD, S e for effluent COD and t stands for hydraulic retention time. According to the equation for active sludge kinetics, N S ¼ ðS 0 S e Þ=ðM tÞ, it can be deduced that
y1 ¼ 5:9e 015x4 1:1e 010x3 þ 3:7e 007x2 0:0035x þ 74:000, y2 ¼ 3:4e 014x4 6:2e 010x3 þ 3:3e 006x2 0:0023x þ 77:000,
M¼
y3 ¼ 1:1e 013x4 1:7e 009x3 þ 7:6e
S0 Se . NS t
(5)
Accordingly, the mathematical model for COD removal with the change of sludge loading rate:
006x2 0:0082x þ 85:000. The comparison of simulated results and the original data (Fig. 4) demonstrated that the fourth-order polynomial was very stable. In order to analyse the relation of COD removal and MLSS more directly, Z and M respectively replaced y and x, and the following mathematical model was obtained. Z ¼ k4 M 4 k3 M 3 þ k2 M 2 k1 M þ DZ.
Here Z is COD removal, M is MLSS, k1 , k2 , k3 , k4 are constants, and DZ is a constant item, which is defined by COD removal when M is equal to zero. Table 1 shows the mathematical model and correlative mathematical parameters of the effect of MLSS on COD removal at different HRTs in this experiment, from which it is concluded that: (i) Fig. 4 and Eq. (3) show that Z increased with the increase of M. When M increased to 6000 mg L1 , the trend of Z gradually became stable. At the start, the membrane bioreactor MLSS was very low, and consequently, the microbial metabolism and COD removal were quite low. After that, the sludge continued to increase and COD removal also increased. However, when the MLSS arrived at a particular value ðMLSS ¼ 6000 mg L1 Þ COD removal and effluent COD were stable, as a result of a stable gelatin layer at the surface of the membrane. Therefore, the simulation results for this mathematical model showed that the optimum MLSS should be maintained at around 6000 mg L1 in the SMBR. (ii) Because of k4 5k3 5k2 , the fourth degree equation could be replaced by the second degree equation
(3)
Z ¼ k2
ðS0 Se Þ2 S0 Se þ DZ. k1 2 2 NS t NS t
(6)
Formula (6) shows that COD removal ðZÞ was related to HRT ðtÞ, the COD of influent ðS 0 Þ and sludge loading rate ðN s Þ. Perfect COD removal could be gained by adjusting parameters such as t, S 0 and N s . In addition, the coefficients of simulation and norm of residuals at different HRTs were used to validate the
Fig. 4. Curve simulation results at HRT of (a) 1 h (b), 2 h and (c) 3 h.
ARTICLE IN PRESS N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
283
Table 1 Correlative mathematical parameters and models of COD and MLSS removal at different HRTs HRT (h)
DZ
Mathematical model
k=k1 , k2 , k3 , k4
1 2 3
74 77 85
y1 ¼ 5:9e 015x4 1:1e 010x3 þ 3:7e 007x2 0:0035x þ 74:000 y2 ¼ 3:4e 014x4 6:2e 010x3 þ 3:3e 006x2 0:0023x þ 77:000 y3 ¼ 1:1e 013x4 1:7e 009x3 þ 7:6e 006x2 0:0082x þ 85:000
0:0035 3:7e 007 1:1e 010 5:9e 015 0:0023 3:3e 006 6:2e 010 3:4e 014 0:0082 7:6e 006 1:7e 009 1:1e 013
removal gradually stabilised. It was demonstrated that the optimum MLSS in an SMBR should be maintained at around 6000 mg L1 . (3) Formula, Z ¼ k2
ðS0 Se Þ2 S0 Se þ DZ, k1 2 2 NS t NS t
showed that COD removal ðZÞ was related to HRT ðtÞ, influent COD ðS 0 Þ and sludge removal loading rate ðN s Þ. Perfect removal of COD could be obtained by adjusting these three parameters.
Acknowledgements The present publication has been made possible through the financial, technical and administrative assistance of China National ‘‘863’’ Hi-Tech R & D Program.
References
Fig. 5. Residuals of curve simulation at HRT of (a) 1 h (b), 2 h and (c) 3 h.
correctness of our mathematical models. Fig. 5 indicates that when the HRT was 3, 2 or 1 h all the norms of residuals were o3:0 (Wen et al., 2000; Su, 2002), proving that the model could simulate the experimental data correctly.
4. Conclusions (1) When the SMBR was used to treat domestic sewage, the experimental results showed that an HRT of 1 h could meet the normal standard for discharged domestic sewage, and an HRT of 2 h could meet that for water reclamation. (2) Mathematical analysis of the experimental data showed that COD removal increased with increase of MLSS. When MLSS increased to 6000 mg L1 , COD
APHA,1995. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, Washington, DC. Brindle, K., Stephenson, T., 1996. Application of membrane biological reactors for the treatment of wastewaters. Biotechnology and Bioengineering 49, 601–610. Chiemchaisri, C., Yamamoto, K., Vigneswaran, S., 1993. Household membrane bioreactor in domestic wastewater treatment. Water Science and Technology 27, 171–178. Defrance, L., Jaffrin, M.Y., 1999. Comparison between filtrations at fixed transmembrane pressure and fixed permeate flux: application to a membrane bioreactor used for wastewater treatment. Journal of Membrane Science 152, 203–210. Devies, W.J., Le, M.S., Heath, C.R., 1998. Intensified activated sludge process with submerged membrane microfiltration. Water Science and Technology 38, 421–428. Gander, W., 1999. Solving Problems in Scientific Computing Using Maple and MATLAB. China Higher Education Press Beijing/ Springer, Berlin. Gu, G.W., He, Y.L., 2002. Study and Application of the Membrane Bioreactor in Wastewater Treatment. China Chemistry Industry Press, Beijing. Guender, B., Krauth, Kh., 1998. Replacement of secondary clarification by membrane separation: results with plate and hollow fibre modules. Water Science and Technology 38, 383–393. Halil, H., Cumali, K., Ayhan, U., 2002. Viability of microbial mass in a submerged membrane bioreactor. Desalination 150, 263–268.
ARTICLE IN PRESS 284
N. Ren et al. / International Biodeterioration & Biodegradation 55 (2005) 279–284
Huang, X., Gui, P., Qian, Y., 2000. Performance of submerged membrane bioreactor for domestic wastewater treatment. Tsinghua Science and Technology 5, 237–240. Knoblock, M.D., Sutton, P.M., Mishra, P.N., Gupta, K., Janson, A., 1994. Membrane biological reactor system for treatment of oily wastewaters. Water Environment Research 66, 133–139. Makoto, O., Yutaka, O., Naoki, O., 1998. Decoloration system using rotation membrane UF module. Water Science and Technology 38, 35–43. Ren, N.Q., Zhang, Y., Chen, Z.B., 2001. The mathematical model analysis of membrane flux by using submerged bioreactor with different membrane forms. High Technology Letters 11, 100–103. Rosenberger, S., Kruger, U., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2002. Performance of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Research 36, 413–420. Shim, J.K., Yoo, I.K., Lee, Y.M., 2002. Design and operation considerations for wastewater treatment using a flat submerged membrane bioreactor. Process Biochemistry 38, 279–285.
Stefan, H., Walter, T., 2001. Treatment of urban wastewater in a membrane bioreactor at high organic loading rates. Journal of Biotechnology 92, 95–101. Su, X.S., 2002. Engineering Application of MATLAB 6.0. China Science Press, Beijing. Trouve, E., Urbain, V., Manem, J., 1994. Treatment of municipal wastewater by a membrane bioreactor: results of a semi-industrial pilot-scale study. Water Science and Technology 30, 151–157. Ueda, T., Hata, K., Kikuoka, Y., 1996. Treatment of domestic sewage from rural settlements by a membrane bioreactor. Water Science and Technology 34, 189–196. Urbain, V., Mobarry, B., Silva, V., Stahl, D.A., Rittmann, B.E., Manem, J., 1998. Integration of performance, molecular biology and modeling to describe the activated sludge process. Water Science and Technology 37, 223–229. Wen, X., Zhou, L., Wang, D.L., Xiong, X.Y., 2000. Application Design of MATLAB NN. China Science Press, Beijing. Xing, C.H. Qian, Y., Wen, X.H., Meng, Y.B., 1998. Performance of microfiltration membrane bioreactor for domestic wastewater reclamation. In: Proceedings of WEFTEC Asia ’98, January 1998, Singapore.