外文原文-利用仿真模型试验规模膜生物反应器.pdf

Available online at ScienceDirect JOURNAL OF ENVIRONMENTAL SCIENCES ISSN 1001 0742 CN 11 2629lX Journal of Enwonmental Sciences lY 2007 129 134 Feasibility and simulation model of a pilot scale membrane bioreactor for wastewater treatment and reuse from Chinese traditional medicine REN Nan qi YAN Xian fengl CHEN Zhao bo HU Dong xue2 GONG Man li GUO Wan qian I School of Municipal and Environmental Etqineering Hurbin Institute of Technology Hurhin 150090 Chinu E muil rnq hit edu cn 2 School oJPouer arid Nuclear Engineering Harhin Engineering Iniversit Harbin 150001 China Received 10 February 2006 revised 16 May 2006 accepted 15 September 2006 Abstract The lack and pollution of water resource make wastewater reuse necessary The pilot scale long term tests for submerged membrane bioreactor wen conducted to treat the effluents of anaerobic or aerobic treatment process for the high strength Chinese traditional medicine wastewater This article was focused on the feasibility of the wastewater treatment and reuse at shorter hydraulic retention time HRT of i 0 3 2 and 2 13 h MLSS growth membrane flux vacuum values and chemical cleaning periods were also investigated The experimental results of treating two phase anaerobic treatment effluent demonstrated that the CODfiit was less than 100 mg L when the influent CClD was between 5W1oooO ma at HRT of 5 0 h which could satisfy the normal discharged standard in China The experimental rc sults to treat cross flow aerobic reactor effluent demonstrated that the average value of CODelt was 17 28 ma when the average value of influent COD was 192 84 mg L at HRT of 2 13 h during 106 d which could completely meet the normal standard for water reuse The maximum MLSS and MLVSS reached 24000 and 14500 ma at HRT of 3 2 h respectively Membrane flux had maximal resume degrees of 94 7 at vacuum value of 0 02 MPa after cleaning Chemical cleaning periods of membrane module were 150 d A simulation model of operational parameters was also established based on the theory of back propagation neural network and linear regrc ssion of traditional mathematical model The simulation model showed that the optimum operational parameters were suggested as follows HRT was 5 0 h SRT was 100 d the range of COD loading rate was between 10 66420 451 kg m3 d thc range of MLSS was between 7543 13694 m a Key words submerged membrane bioreactor SMBR Chinese traditional medicine wastewater hydraulic retention time HRT simulation model pilot scale test Introduction Submerged membrane bioreactors SMBRs an im provement of the conventional activated sludge process in which the traditional secondary clarifier is replaced by a membrane unit for the separation of treated water from the mired solution have emerged as one of the innovative and promising solutions for wastewater treat ment and reclxnation Dufresne et al 1998 Stephenson et al 2000 Clech et al 2003 Originated from the use of membrane separation SMBR technology has the following ad antages high sludge concentration Halil et al 2002 high quality of effluent long contact time between activ ited sludge and organic pollutants Brindle and Stephenson 1996 and complete separation of the hydraulic rete ition time HKT and sludge retention time SRT Devie er a 1998 Guendcr and Krauth 1998 Ueda et al 1996 Moreover highly treated water in a SMBR is fret from bacteria and has the potential for Project supported by the Hi Tech Research and Development Program 863 of China No 21WAA6013 10 Corresponding author E mail rnq municipal and industrial reuse Xing et al 1998 Currently the membrane bioreactor technology has been applied in a series of industrial wastewater treatments since the 1980s These applications include oily wastew ater treatment Knoblock et al 1994 Seo et al 1997 foodstuff wastewater treatment Krauth and Staab 1993 Mallon et ul 1999 tannery wastewater treatment Ya mamoto and Win 1991 Dijk and Roncken 1997 and leachate treatment Mishra et al 1996 Pirbazari et nl 1996 and so on However till now there are no pub lications on laboratory scale and pilot scale experiments on treating high strength Chinese traditional medicine wastewater by using the membrane bioreactor Chinese traditional medicine wastewater has the characteristic of highel strength from COD concentration of 6000 to 19000 mg L and lower BOD5 COD ratio of about 30 So this kind of wastewater belongs to higher strength and hard degradable industrial organic wastewater It is well recognized that HRT is a key factor for further improving the capacity of the membrane bioreactor In industrial wastewater treatment HRT is generally longer ranging from 10 5 to 389 h Stephenson er al 2000 Gao I30 REN Nan qi el al Vol 19 et al 2004 However Sutton proved that longer HRT has no effect on operation of the membrane bioreactor in the oily wastewater treatment Sutton et al 1994 The COD removal rate maintained 90 with HRT of between 1 87 3 74 d and SRT between 50 100 d Similar results were also obtained from tannery wastewater treatment and pulping process wastewater treatment Dufresne et al 1998 The wastewater was treated by two phase anaerobic treatment and cross flow aerobic process as main treatment process in Harbin Chinese traditional medicine plant In this study SMBR was used to treat Chinese traditional medicine wastewater from two phase anaerobic treatment and cross flow aerobic reactor effluent respectively The purpose of this study was to investigate the feasibility of Chinese traditional medicine wastewater treatment and reuse We conducted two pilot scale experiments for SMBR to treat the wastewater effluent from two phase anaerobic treatment and cross flow aerobic reactor respec tively and studied COD removal mixed liquor suspended solids MLSS growth membrane flux vacuum values and chemical cleaning In addition a simulation model of operational parameters was established based on the theory of back propagation neural network BPNN and linear regression of traditional mathematical model The simulation model realized the prediction of the effluent quality and ascertained the key impact factor and the optimum operational parameters of the SMBR 1 Materials and methods 1 1 Experimental apparatus and operational conditions A schematic diagram of the SMBR is shown in Fig 1 The influent is through the feed tank to the bioreactor by peristaltic pump I The effluent of the bioreactor was connected with automatic vacuum effluent system directly by rotor flow meter The automatic vacuum effluent system consists of vacuum pot vacuum pump gas water segre gator level sensor pump 2 electron magnetic valve and power control device When the vacuum pump starts up a part of gas in the vacuum pot was pumped out to form negative pressure and then the wastewater in the bioreactor was drained out through the membrane module and enters into the vacuum pot When the liquid surface in the vacuum pot reached 80 of vacuum pot height the water pump 2 started up and the effluent stored in the pot was drained out The working volume of the SMBR was about 3 2 m in which the membrane module is directly submerged Aeration was employed to maintain the given dissolved oxygen concentration DO of 2 3 m a Air flow rate was adjusted by gas flow meter and controlled at 10 15 m7 h Level controller and level sensor con trolled the water level in the bioreactor The sludge for inoculation wa5 from the bioreactor in a local Chinese traditional medicine wastewater treatment plant Initial concentration of MLSS was about 2140 m a The sludge is continually withdrawn by sludge withdrawing pump according to different SRT HRT was controlled at 8 0 5 0 3 2 and 2 13 h by rotor flow meter under the operational condition of invariable membrane flux which were 8 0 12 8 20 0 and 30 L m2 h in three stages respectively The temperature was not controlled during the expenment The experiments were carried out for 452 d and 106 d for treating the effluents of two phase anaerobic treatment and cross flow aerobic reactor effluent respectively and the operating conditions of SMBR are shoan in Table 1 1 2 Membrane characteristics In this pilot plant test a hollow fiber PVDF micro filtration MF membrane supplied by Tianjin Motian Membrane Engineering and Technology Co Ltd China was used with pore size of 0 22 pm The effective surface area of a MF membrane module was 12 5 m2 and four MF membrane modules were used in this ctudy Membrane flux was between 8 0 and 20 0 L m2 h 1 3 Analytical methods The value of MLSS MLVSS mixed liquor volatile suspended solids COD BOD5 and DO were measured using standard methods APHA 1995 The effluent COD from SMBR was shown as CODhlL Vacuum values were measured by an electro pressure meter Membrane effluent flux Q was measured by a rotor flowmeter Fig 1 Scheme of the pilot scaled SMBR I Valve 2 elcctron magnetic valve 3 pressure control valve 4 flow meter 5 level sensor 6 vacuum meter 7 gas water separator Feasibilit and simulation model of a pilot scale membrane bioreactor for wastewater treatment 131 No 2 Table 1 Operating conditions for treating two phase anaerobic treatment and cross flow aerobic reactor effluent Item Two phase anaeroblc treatment effluent Cross flow aeroblc reactor effluent Stage 1 Stage 2 Stage 3 Duration id 1 15 16is 17 28 mg L when the average value of influent COIl was 192 838 mg L with HRT of 2 13 h during 106 d CODfil of less than 30 mg L could meet the normal st ndard for water reclamation for car washing land watering and so on China CJ25 1 89 2 2 MLSS grqwth The SMBF was operated over 452 d for treating two phase anaerobic treatment effluent The changes of MLSS 3 800 h 600 400 0 c 200 0 14000 100 2 12000 90 5 I Influent CO 80 2 70 50 B g 10000 o COD remov6 rate IU 3 g AODO 6000 6 4000 40 0 20 2000 30 8 0 40 80 120 160 200 240 280 320 360 400 440 Time d Fig 2COD remc dl in thc pilot wale SMBR for tredllng two phase anaerobic treatmen effluent 400 I I O Influent COD x Filtrate COD 0 10 20 30 40 50 60 70 80 90 100 110 Time d Fig 3 COD removal in the pilot scale SMBR for treating cross flow aerobic reactor effluent MLVSS and MLVSS MLSS with time are shown in Fig 4 As shown in Fig 4 MLSS was in the range of 3000 7000 mg L MINSS was 2000 5000 mg L and the average value of MLVSSWLSS was 0 72 at HRT 8 0 h and SRT 50 d during stage 1 MLSS and MLVSS increased over time gradually During the stage 2 as soon as HRT was shortened and SRT was raised volumeh ic loading rate was raised so MLSS and MLVSS began to grow once again MLSS was between 6000 14000 mg L and the average value of MLVSSWLSS was 0 76 at HRT 5 0 h and SRT 100 d During the stage 3 MLSS and MLVSS continued to grow with HRT shortening The maximal MLSS and MLVSS arrived at 24000 and 14500 mg L at HRT 3 2 h respectively and the average value of MLVSSWLSS was 0 62 MLVSS MLSS presented a depressed trend over time from stage I to stage 3 It could be concluded that the sludge activity expressed by MLVSS MLSS decreased over time So MLSS grew with HRT shortening and volumet ric loading rates increasing and was higher than those observed in conventional activated sludge process The bioreactor could be operated biologically at higher vol umetric loading rate and lower sludge loading rate and then sludge yield was low Therefore the volume of the 50000 2 40000 W 30000 m 2 4 LUUUU m 50 lOOT50 200 250 300 350 400 450 1 0 0 8 vl 0 6 3 rn u 4 0 2 5 0 I Time d Fig 4 Changes of MLSS concentration and MLVSS MLSS in the SMBR during stages 1 3 for treating two phase anaerobic treatment cfflucnt 132 REN Nan qi et ul Vol 19 bioreactor could be reduced greatly In general enough MLSS could insure favorable removal of the pollutants and the best effluent quality However the higher MLSS would affect membrane flux Therefore it was necessary to investigate membrane flux attenuation The SMBR was operated over 106 d to treat cross flow aerobic reactor effluent The changes of MLSS MLVSS and MLVSS MLSS with time are shown in Fig 5 MLSS and MLVSS were at 3000 6000 mg L and increased slowly because influent COD of 192 838 mg L was lower and nitrogen and phosphorus of cross flow aerobic reactor effluent were a little Therefore the SMBR was operated at the SRT of infinity in this study The purpose was to increase MLSS and COD removal rate reduce sludge yield into zero From Figs 3 and 5 it could be seen that the difference in this study with previous reports was the shorter HRT of 2 13 h and the longer SRT CODfil could completely meet the normal standard for water recla mation Thus we improved COD removal rate reduced volume and saved engineering investment of SMBR 12000 I 1 0 0 MLvssms 0 0 10 20 30 40 50 60 70 80 90 100 110 Time d Fig 5 Changcs of MLSS concentration and MLVSS MLSS i n the SMBR for treating cross flow aerobic reactor effluent 2 3 Membrane flux vacuum value and chemical clean ings Membrane fouling and membrane flux attenuation were inevitable over a long operational period in despite of the optimum membrane module and operational conditions Here chemical cleaning was needed to resume membrane flux and remove pollutants in membrane apertures Fig 6 represents the changes of vacuum value and membrane flux of membrane module This system was operated for 508 d Membrane fluxes were 400 600 and 1000 Vacuum value 3 v 5 600 G a l 400 4 2 0 0 p 1 0 0 50 100 150 200 250 300 350 400 450 Time d Fig 6 Changes of vacuum value and niembrarie flux from stages 1 3 1000 L h during 452 d respectively Membrane flux in this experiment was kept constant by adjusting vacuum value Membrane flux decreased over time under some constant vacuum value Two reasons causing the decrease of membrane flux might be that 1 pore size due to conformational changes decreasing 2 gelatinized layer at the membrane surface forming In order to keep con stant membrane flux vacuum value must be increased Stabilization of membrane flux could be achieved by adjusting the vacuum value It can be seen from Fig 6 that vacuum value with time was obviously divided into three stages There were two declines as a result of the first and second chemical cleaning These indicated that chemical cleaning could effectively resume membrane flux Fig 6 also shows that at the startup stage of the experiment the needed membrane flux could be met under the very low pressure vacuum value is 6 0 kPa It can be expected that high pressure will immediately lead to membrane fouling Therefore its vacuum value would be set at the low value of 6 0 kPa for operation The results demonstrated that the operational condition of the SMBR could be starting up at low vacuum value and operating at constant membrane flux Resume of membrane flux at variation of vacuum value after three chemistry cleanings is illustrated in Fig 7 Mem brane flux attenuation in the SMBR was investigated to confirm optimum chemical cleaning periods and frequen cy Cleaning methods of membrane module in the SMBR were physical cleaning at first Secondly hypochlorite natrium was used Finally chemical cleaning was finished by means of ethanol cleaning Membrane f u x had different resume degree under different operational pressure and had maximal resumes at 0 02 MPa of vacuum value Resume degrees of membrane flux were 94 7 83 7 and 70 at 0 02 MPa of vacuum value respectively Cleaning periods of membrane module were 150 d 2 4 Simulation model of COD loading rate MLSS and In order to optimize operational parameters and predict the effluent quality of the SMBR for Chinese traditional medicine wastewater treatment a three dimensional sim ulation model of COD loading rate MLSS and CODfilt CODfilt 2 400 n 500 0 New membrane flux 0 Membrane flux after chemical cleaning 0 Membrane flux after Membrane flux after the first the secon the third J 0 02 0 01 0 002 Vacuum value lo6 Pa Fig 7 Changcs of water flux at variation of vacuum value after three chemistry cleanings No 2 Feasibility and simulation model of a pilot scale membrane bioreactor for wastewater treatment 133 was established based on back propagation neural network BPNN and linear regression of traditional mathematical model using the mathematical software program MAT LAB These methods from the literature had never been applied prev ously for modeling the SMBRs BP Neural Network had stronger approximation and generalization ability in nonlinear systems At the same time the per fection and precision of data determined the applicability of the model This provided a feasible way for on line control in process The topological architecture of BPNN is illustrated n Fig 8 5 val ratc Input 1ayc r Hidden layer Output layer F g 8 Topological architecture of BPNN As shown in Fig 8 there were five nodes on input layer four nodes on hidden layer and two nodes on output layer Each rode was a BP neuron Epoch was 5026 For debugging and perfecting program comparison of expenmental d u e with BPNN value is shown in Fig 9 250 2 200 E 3 150 g 100 50 0 c Expenmental valuc BPNN value 1 1 9 13 17 21 25 29 33 37 41 45 49 Expenmental times Fig 9 Compdi iwn between expenmental value and BPNN value As shown n Fig 9 the program was comparatively perfect The rwlts of debugging program showed that the amounts oi hidden layer and node could better reali7e approximation and generalization effect and the mean square error t LISE was 0 001365 Thus the program could be applizd to establish the simulation model of COD loading ate MLSS and CODfil Simulation model of back propa ation neural network based on the four limiting factor IHRT SRT COD loading rate and MLSS were illustrate in Fig 10 In the model HRT of 5 h SRT of 100 d and re nperature of 20 C were constant As shown in Fig 10 when HRT SRT and temperature were 5 h lo d and 70 C respectively the optimum MLSS and COD loading rate were less than 13694 mg L and 30 45 kg irn d for obtaining good CODfil The final objective of thi study was not to establish the simulation model but to av ertain the optimum operational parameters and criteria for the design and the operation of the SMBR F