APPLICATION OF THE AIR BACKFLUSHING TECHNIQUE IN A MEMBRANE BIOREACTOR FOR SEPTIC WASTEWATER TREATMENT

by

Sudtida Pliankarom

A thesis submitted in partial fulfillment of the requirements for degree of Master of Engineering.

Examination committee

Dr. C. Visvanathan (Chairman) Mrs. Samorn Muttamara Dr. Byung-Soo Yang

Nationality Previous Degree

Thai B.Sc. (Environmental Health Science) Mahidol University Bangkok, Thailand Government of Netherlands

Scholarship Donor

Asian Institute of Technology School of Environment, Resources and Development Bangkok, Thailand August, 1996

Acknowledgements The author wishes to expressed her profound gratitude, great appreciation and indeptness to her advisor Dr. C. Visvanathan for his valuable guidance, encouragement, support and sharing the knowledge throughout the research. Special thanks are extended to Mrs. Samorn Muttamara (Co-chairman) and Dr. Byung-Soo Yang for their valuable suggestions and guidance given for this study as members of the examination committee. The author also thankful to the staff and the friend of the Environmental Engineering Program for their prompt assistance and cooperation during this study. A very grateful acknowledgment is extended to the Government of Netherlands for providing the author with a scholarship to study the master program at AIT. The author wishes to express her sincere gratitude to Prof. R. Ben Aim of Department of Chemical Engineering, University de Technologic de Compiegne, France and Mr. Robert T. Wale of Memtec Ltd., Australia for providing membrane and giving valuable technical information. Finally the author express her profound gratitude to her parents for their strong encouragement and inspiration given to her.

Abstract In this study the possibility of application of air backflushing technique through hollow fiber microfiltration was investigated. The process employed direct solid-liquid separation by immersed two membrane modules with pore size of 0.2 m directly in the activated sludge aeration tank of 80 L volume. This study was conducted with high concentration of activated sludge and divided into short-term and long-term experiments. In short term experimental runs, the optimum air backflushing and filtration cycle was investigated. 15 minutes filtration and 15 minutes air backflushing provided the best result in term of flux improvement and stability. Due to the membrane module stability limitation, the applied compressed air pressure of 1 bar was not sufficient to remove the clogging completely. However this cyclic operation provided higher flux stability compared to operation without air diffusion. In long-term experiments, the initial sludge concentration was 13,000 mg/L. Three different hydraulic retention times (HRT) of 26, 18 and 10.5 hour which corresponds to the permeate flux of 3.08, 4.44 and 7.62 L/m2. h were investigated. Here, it was noted that the filtration pressure related to the MLSS concentration. Whereas the stable operation could be obtained at 26 and 18 hours. All experimental runs provided more than 90% removal of COD, BOD and TKN with final MLSS of 40,000 mg/L in the reactor. Although the operation with daily sludge draining (1.6 L/d), the MLVSS/MLSS values seem slightly decreased. However, such conditions could not effect significantly to the process performance in term of physical, chemical, biological and bacteriological qualities of membrane bioreactor effluent.

List of Abbreviations AS ASP BOD COD CST DO EA ED PHF Eff F-BOD F-COD F/M HRT Inf J k kd KLa Ks MBR MF MLSS MLVSS NF NH3-N NO2-N NO3-N NTU Rd Rg Rg Rm Rmo Rm1 Rm2 RO Rsu SRT SS TKN T-N TP TS - Activated Sludge - Activated Sludge Process - Biochemical Oxygen Demand - Chemical Oxygen Demand - Capillary suction time - Dissolved Oxygen - Extended aeration - Electrodialysis - Polyethylene hollow fiber - Effluent - Filtered biochemical oxygen demand - Filtered chemical oxygen demand - Food / Microorganism ratio - Hydraulic retention time - Influent - Permeate flux - Maximum rate of substrate utilization per unit mass of microorganism - Endogenous decay coefficient - Overall gas transfer coefficient - Half- velocity constant (saturation constant) - Membrane bioreactor - Microfiltration - Mixed liquor suspended solids - Mixed liquor volatile suspended solids - Nanofiltration - Ammonia nitrogen - Nitrite nitrogen - Nitrate nitrogen - Naphelometric turbidity unit - Membrane resistance due to the deposition of solids - Rate of bacterial growth - Net rate of bacterial growth - Apparent membrane resistance - Initial membrane resistance - Membrane resistance after first cleaning - Membrane resistance after second cleaning - Reverse osmosis - Substrate utilization rate - Solids retention time - Suspended solids - Total kjedahl nitrogen - Total nitrogen - Total phosphate - Total solids

List of Abbreviations TVS UF VSS Y m 15:15 15:15* - Total volatile solids - Ultrafiltration - Volatile suspended solids - Sludge growth coefficient - Dynamic viscosity - Maximum specific growth rate - 15 minutes of filtration then 15 minutes of air diffusion - 15 minutes of filtration then stop without sending air for 15 minutes

Table of Contents Chapter Title Title Page Acknowledgements Abstract Table of Contents List of Tables List of Figures Abbreviations 1. 2. Introduction Literature Review 2.1 Fundamentals of Activated Sludge 2.2 Fundamentals of Microfiltration 2.3 Biological Nitrification 2.4 Denitrification 2.5 Phosphorrus Removal from Wastewater 2.6 Microfiltration Membrane in Domestic Wastewater Treatment 2.7 Application of Membrane Bioreactors in Domestic Wastewater Treatment 2.8 Clogging Mechanisms in Microfiltration 2.9 DecloggingTechniques 2.10 Fundamentals of Gas transfer 2.11 Applications of Gas Diffusion through Membrane Experimental Set Up 3.1 Measurement of Gas Transfer Efficiency of Aeration Units 3.2 Measurment of Initial Membrane Resistance 3.3 Short Term Experiments 3.4 Long Term Experiments 3.5 Analytical Methods Results and Discussion 4.1 Gas Transfer Efficiency of Aeration Units and Initial Membrane Resistance 4.2 Short Term Experiments 4.3 Long Term Experiments Page i ii iii iv vi vii ix 1

3 3 4 5 5 7 7 10 11 12 14

3.

17 19 21 24 25

4.

26 30 38

Table of Contents Chapter 5. Title Conclusion and Recommendations 5.1 Conclusions 5.2 Recomendations for Future Work References Page

64 65 66

Appendix A 68 Gas Transfer Efficiency and Membrane Resistance Appendix B 76 Detail Results of Short Term Experiments Appendix C 87 Detail Results of Long Term Experiments Appendix D 98 Detail Results of Typical Coefficients of Activated Sludge Process for Domestic Wastewater Treatment Appendix E 110 Screening and Toxicity Methodololy for Measurement of Biomass Activity Appendix F 112 Detail Results of Nutrient Removal Appendix G 116 Detail Results of Activated Sludge Characteristics Appendix H 121 Detail Results of Membrane Cleaning Efficiency

List of Tables Table 3.1 3.2 4. 1 4.2 4.3 4.4 4.5 4.6 4.7 4.8a 4.8b 4.9a 4.9b 4.9c 4.10 4.11.1.1 4.12 Dynamics of Activated Sludge at Steady State for each Experimental Runs 58 Title Page

Characteristics of Raw Septic wastewater 21 Parameters Analyzed ( long tem experiments) 25 Comparison of Performance of Different Suction Pressure with 15:15* Operation Mode and with Air Diffusion 30 Comparison of Performance of Different Suction Pressure with 15:15 Operation Mode and 1 bar Compressed Air 31 Comparison Performance of Different Mode of Operation with Suction Pressure at 7 kPa and 1 bar Compressed Air 35 Comparison Performance of Different Compressed Air Diffusion with Suction Pressure at 7 kPa and 15:15 Operation Mode 36 Kinetics of Acclimatized Sludge Growth 38 Transmembrane Pressure of each Experimental Runs 42 Average Value of Biological Solids Concentration of each Experimental Runs 45 Nitrogen Mass Balance at Steady State 49 Calculated Data for Nitrogen Mass Balance 50 Phosphate Mass Balance at Steady State 53 Calculated Data for Phosphate Mass Balance 53 Characteristic of total Phosphate in Sludge from Bioreactor 54 Characteristics of Sludge from Conventional Activated Sludge and Membrane Bioreactor Process 55 Microbiological Quality of Permeate and Effluent from Conventional Activated Sludge

List of Figures Figure 2.1 2.2 2.3 2.4 2.5 3.1 3.2 3.3a 3.3b 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 Title Page 4 9 11 12 14 18 18 23

4.10

4.11 4.12 4.13

4.14

Dead-End Filtration and Crossflow Filtration Schematic of Pilot - Scale Membrane Bioreactor Module Backflushing with Gas Effect of Gas Backflushing during Wine Filtration Schematics of Two-Film Theory of Gas Transfer Membrane Module Used for All Experiments Experimental Set Up for Gas Transfer Efficiency Schematic of Membrane Bioreactor Set Up Actual Membrane Bioreactor Set Up 23 Variation of Oxygen Concentration with Time when Using Ambient Air (stone diffusers) Variation of Oxygen Concentration with Time when Using Pure Oxygen (stone diffusers) Plot of Cs - Ct with Time when Using Ambient Air (stone diffusers) Plot of Cs - Ct with Time when Using Pure Oxygen (stone diffusers) Comparison of KLa (20) o C at Different Flow Rates for Ambient Air and Pure Oxygen (stone diffusers) KLa (20) o C at Different Pressure (membrane diffusers) Relationship between Permeate Flux of Clean Water and Transmembrane Pressure (membrane module 1) Relationship between Permeate Flux of Clean Water and Transmembrane Pressure (membrane module 2) Variation of Average Permeate Flux with Time at Each Hour for Different Transmembrane Pressure with 15:15* Operation Mode (* = without sending air) Comparison of Permeate Flux with Time for 40 kPa Suction Pressure between Operation with and without Air Diffusion Comparison of Flux with Different Suction Pressure Variation of Permeate Flux with Time for Long - Run Observation when using Transmembrane Pressure of 7 kPa Comparison of Permeate Flux with Time between 10:10 and 15:15 Operation Modes for 7 kPa Suction pressure and 1 bar Compressed Air Comparison of Permeate Flux with Time between 20:20, 25:25 and 30:30 Operation Modes for 7 kPa Suction pressure and 1 bar Compressed Air

27 27 28 28 29 29 32 32

33

33 34 34

37

36

List of Figures Figure 4.15 4.16a 4.16 b 4.17 4.18 Title Page

4.19 a 4.19 b 4.20 4.21 4.22 4.23 4.24 4.25a 4.25b 4.25c 4.26a 4.26b

Variation of MLVSS and Effluent Filtered-COD with Time after Sludge Acclimatization Variation of Transmembrane Pressure (suction pressure) with Time for Different Experimental Runs Variation of Permeate Flux with Time for Different Experimental Runs Variation of Effluent Turbidity and DO with Time for Different Experimental Runs Variation of COD Concentration with Time for Different Experimental Runs Variation of Biological Solids Concentration with Time for Different Experimental Runs Accumulation of Inert Materials with Time for Different Experimental Runs Variation of F/M-ratio and BOD Concentration with Time for Different Experimental Runs Visual Characteristic of Raw Influent, Sludge in Bioreactor and Effluent Variation of TKN Concentration with Time for Different Experimental Runs Variation of NO3 -N Concentration with Time for Different Experimental Runs Viation of Total Phosphate concentration with Time for Different Experimental Runs Biological Flocs with Safanin-O at steady state of RUN 1, 200x Biological Flocs with Safanin-O at steady state of RUN 2, 200x Biological Flocs with Crystal Violet at steady state of RUN 3, 200x Relationship between Flux of Clean Water and Transmembrane Pressure (membrane module 1) Relationship between Flux of Clean Water and Transmembrane Pressure (membrane module 2)

38 41 41 44

44 46 46 47 47 51 51 54 59 59 59 63 63

Chapter 1 Introduction 1.1 General The most common and classical wastewater treatment process which has been used to treat domestic wastewater is the activated sludge process (ASP). In this system the organic and inorganic matters present in the suspended solid, colloidal and soluble forms can be removed up to 95%. However, there are some limitations in classical ASP when a high quality of effluent is required. In such situations, large secondary sedimentation tank is required to provide sufficient retention time. Moreover, there are various factors that must be concerned to reach good settling characteristic. Therefore, various types of combination between ASP and membrane unit have been studied and adapted to overcome these problems and to obtain good effluent quality. Membrane separation technology in water and wastewater treatment can be categorized into four classes according to the membrane, namely, reverse osmosis(RO), ultrafiltration (UF), microfiltration (MF) and electrodialysis (ED). UF and MF techniques are useful in removing macromolecule, colloids and suspended solids. Membrane separation technology has been introduced for solid/liquid separation in biological treatment system. The advantages of employing membrane separation are minimum sludge wastage by maintaining low F/M ratio, reducing plant size by maintaining higher biomass concentration in the reactor, and solid free effluent could be obtained. Complete retention condition could be maintained by operation without sludge wastage since the solid/liquid separation could be done regardless of sludge settleability. For domestic wastewater treatment, combined activated sludge/membrane filtration can provide a high degree of treatment in terms of organic oxidation and nitrogen removal. However, the power/energy consumption that has been reported is much higher than the value for conventional ASP. (Yamamoto,1989) As Yamamoto (1989) indicated, the process is not cost effective. The main reason for the high cost is due to the recirculation pump, which connects the main reactor with a membrane unit and maintain high crossflow velocity on the membrane surface to keep the flux undeclined. The solution for this has been investigated by direct membrane separation using hollow fiber in an activated sludge aeration tank which still give a stable operation and good quality of effluent. Considering the process performance, direct membrane separation in ASP with continuous suction operation caused severe clogging of the membrane module whenever transmembrane pressure is increased. Using the intermittent suction

operation enabled a stable flux to be maintained for suitable, particular conditions.(Yamamoto et al., 1989)

Cyclic operation with air diffusion has been investigated by Chiemchaisri (1990). Air backflushing technique was applied to achieved the recovery of permeate flux and net cumulative volume. However, increasing the pressure applied for air backflushing to achieve complete membrane cleaning may damage the membrane. 1.2 Objectives of the study 1. To investigate the possibility of using 0.2 m-membrane pore size for effluent filtration and air diffusion purposes in alternative cycle. 2. To compare the effect of produced gas bubbles by using ambient air and pure oxygen by consideration of gas transfer coefficient. 3. To study the effect of operation cycle ( effluent filtration and air diffusion ) in membrane bioreactor to prolong the operational life of membrane bioreactor. 4. To find out the optimum operating condition for 0.2 m-pore size of membrane. 5. To study the treatment efficiency and operational stability of the membrane bioreactor. 1.3 Scope of the study 1. This study was carried out in laboratory-scale. 2. Polyethylene hollow fiber (PHF) membranes of 0.2 m pore size were used in this study. 3. Actual septic wastewater collected from septic tank of public apartment in the area of Pathumthani province was used as feed substrate. 4. COD, MLSS, MLVSS, Turbidity, DO, pH, Temperature, TKN, NO3--N, NO2--N and total phosphate were monitored regularly to observe the reactor temporal performance. In addition, the permeate flux and transmembrane pressure were also monitored to assess the reactor performance.

Chapter 2 Literature Review 2.1 Fundamentals of Activated Sludge Process The activated sludge process make use of the suspended biomass to stabilize, biochemically, organic waste in wastewater with the presence of oxygen. The aerobic condition is achieved by the use of diffused or mechanical aeration, which also serves to maintain the mixture called mixed liquor in completely mixed regime. After a specific period of time, the conversion of organic wastes to the more stabilized substances take place and provide the desired quality of water. Extended aeration is similar to the conventional activated sludge process with the exception of the operation in endogenous respiration of growth curve. The process operations prefer the low organic loading, long aeration time and low F/M ratios. Due to the stated operations, the sludge problems can be overcome in view of small amount of waste sludge produced which need to be carried and good sludge characteristic for dewatering unit. 2.2 Fundamentals of Microfiltration Microfiltration membranes are applied for separation of particles within the range of 0.02-10m. High pressure driven force allows the passage of water through the membrane at the feed side and the tangential liquid flow promote the membrane cleaning at the inverse direction. Microfiltration membrane process is widely used in water and wastewater treatment, which the present pollutants contain diverse particle sizes of colloids and suspended solids. Due to larger membrane pore size, higher flux is obtained in microfiltration system compare to the RO and UF. However, often MF membrane process faces possible internal and external pore clogging due to colloidal fraction, which lead to significant flux reduction. This problem can be overcome by selecting appropriate membrane pore size and by using appropriate pretreatment techniques. The operational mode of microfiltration can be classified into two types as shown in Figure 2.1. 1. Dead-end or conventional filtration (Vigneswaran et al., 1991) 2. Crossflow filtration

Figure 2.1 Dead- End Filtration and Crossflow Filtration (Vigneswaran et al., 1991) In conventional filtration the flow direction perpendicular to the filter medium, while the crossflow filtration has tangential flow to the membrane, the feed is along the membrane surface and the permeate is perpendicular to the feed. Thus this system is known as cross-flow filtration. 2.3 Biological Nitrification Nitrification is the conversion of ammonia nitrogen (NH4+-N) and some organic nitrogen form to nitrate nitrogen (NO3--N) with nitrite (NO2--N) formation as an intermediate and is performed by either heterotrophic bacteria or autotrophic bacteria. However, the major nitrifying bacteria are the autotrophic species, Nitrosomonas and nitrobacter which are common in soil and aquatic ecosystems. They derive energy for growth from the oxidation of inorganic nitrogen compounds instead of oxidation of organic matter. The stoichiometric reaction of nitrification and assimilation become 55 NH4+ + 5 CO2 + 76 O2 ------ C5 H7 O2 N + 54 NO2- + 109 H+ + 52 H2O (2.1) 400 NO2- + 195 O2 + 5 CO2 + NH4+ + 2H2O------- C5 H7 O2 N + 400 NO3- + H+ (2.2) It is seen that approximately 3.22 mg O2 will be required for each mg of NH4+-N oxidized to NO2--N, and 1.11 mg O2 will be need for each mg of NO2--N oxidized to of NO3--N for a total of 4.33 mg O2 mg of NH4+-N oxidized all the way to NO3--N. It is generally accepted that the specific growth rate of Nitrobacter is higher than the growth rate of Nitrosomonas and hence there is no accumulation of nitrite in the process

and the growth rate of Nitrosomonas will control the overall reaction. (Medcalf & eddy, 1991) 2.4 Denitrification Denitrification is a biochemical reaction which involves the reduction of nitrate or nitrite, present in water, to gaseous nitrogen compounds such as nitrogen gas, nitrous and nitric oxides and is carried out by facultative heterotrophic bacteria under anoxic conditions. There are also certain autotrophic bacteria that denitrify using an inorganic energy source. The principal genera are Pseudomonas, Micrococcus, Achromabacter and Bacillus, which were reported as abundant in sewage. Denitrification offers a mechanism of not only removing nitrogen in a non-polluting form, but also oxidizing organic matters in the process. Thus the oxygen which has been supplied in nitrification can, in principle, be effectively recovered and reused in denitrification. Nitrate readily replaces oxygen as electron acceptor because the pathway for the transfer of electrons from the organic substrate to the final electron acceptor is similar, but the presence of dissolved oxygen acts as a strong inhibitor on denitrification as it prevents the formation of the enzyme necessary for the final electron transfer to nitrate. There are four conditions that are necessary for denitrification : (Medcalf & eddy, 1991) 1. Presence of nitrate 2. Absence of dissolved oxygen 3. Bacterial mass that can accept nitrate and oxygen as electron acceptor 4. Presence of a suitable electron donor ( energy source ) 2.5 Phosphorus Removal from Wastewater 2.5.1 Chemical Phosphorus Removal Chemical P removal can be achieved by addition of cation, which will cause precipitation of phosphorus holding wastewater. Lime is less used now because it produces large quantity of sludge and alkaline effluent. Alum has also been used, however its active pH condition is slightly acidic. The optimum pH for activated sludge is neutral range so pH adjustment is then required prior to precipitate by alum. Metal precipitant, FeCl3 or Fe(OH)3, or coagulant was then selected on the performance of phosphorus removal in wastewater treatment. Jar test have to be conducted in order to determine which chemical and at what dosing level give optimum results. The stoichiometric of ferric salt was shown in the following equations.

FeCl3

Fe+3 + 3 Cl-

(2.4)

Fe+3 + PO4-3

FePO4

(2.5)

Phosphorus can be removed by chemical dosing at: 1. Primary settling tank 2. Activated-sludge aeration tank 3. Prior to secondary sedimentation tank 4. After secondary sedimentation tank Coagulation upstream from primary settling tank results substantially reduction in organic loads on secondary treatment unit because considerable proportion of solids and colloids together with some soluble material were removed. However, greater chemical usage is required and primary sludge production is increased significantly. It has been found that organically-bound phosphorus is not easily precipitated and therefore complete phosphorus removal might not be achieved. Chemical addition as a tertiary treatment is probably a valuable alternatives when high quality of effluent is required to achieved reliable total Phosphorus standards below 0.5 mg-P/L. However, addition of solids polishing facilities are then required in order to reduce solids loss to the effluent. Dosing directly into an aeration tank or prior to a secondary sedimentation tank is likely to be preferred in situations as: (Cooper et at., 1994) 1. It need less chemical 2. Organically-bound phosphorus is oxidized and precipitated in the aeration tank 3. Less excess sludge is produced

2.5.2 Biological Phosphorus Removal The basic mechanisms is to create the alternative conditions of anaerobic and aerobic or oxic stages. Under the anaerobic conditions the growth of particular strains of bacteria such as Acinetobacter is selected. Energy uptake under these condition is gained by hydrolysis of polyphosphates stored in the cells. The hydrolyzed polyphosphates are then released out from the cell into the liquid as Orthophosphates. During the aerobic stage the soluble phosphorus is taken up and stored as polyphosphate in order to produce energy for their cells. The unit processes of biological phosphorus removal can be applied in different ways. Some of the processes include the anaerobic stage within the existing activated sludge process. This is called water-line phosphorus removal process. Other processes, particularly for PhoStrip, create an anaerobic stage outside the existing activated sludge plant where some part of recycled activated sludge is stripped of its phosphorus and then returned to the aerobic activated sludge plant to take up more

phosphorus. This is called the sludge-line phosphorus removal process. (Cooper et at., 1994) 2.6 Microfiltration Membrane in Domestic Wastewater Treatment Many researchers have studied the application of membrane technology in domestic wastewater treatment. In conventional treatment, membranes can be inserted at three locations; namely: (Vigneswaran, 1991) 1. After the primary sedimentation 2. In the activated sludge tank 3. After the secondary sedimentation, in the tertiary treatment with or without pretreatment. Combination of membrane together with the activated sludge process is used for separating of liquid from solids. This process performance provide attractive results such as: 1. SS are totally eliminated through membrane separation 2. Settleability of the sludge has no effect on the quality of the treated water 3. Adequate sludge retention time (SRT) which allow the proliferation of low growth rate microorganisms such as nitrifying bacteria 4. Maintain of high concentrations which over all activities level can be raised due to high concentration of the dispersed microorganism maintaining in the bioreactor as long as possible, high concentrations create a favorable environment for endogenous thereby ensuring high treatment efficiency 5. This method can produce bacteria and virus free treated water. Because of all solids are retained in the bioreactor and long SRT, dissolved organic substances with low molecular weights can be taken up, broken down and gasified by microorganisms or converted to polymers as bacterial cell. Thereby, improving the quality of treated water could be achieved. The retained polymeric substances can be biodegraded, then provide less accumulation of substances within the treatment process. (Yamamoto, 1994)

2.7 Application of Membrane Bioreactors in Domestic Wastewater Treatment Talat (1988) investigated hollow fiber microfiltration for solid-liquid separation from the aeration tank of an activated sludge process. The variation of 3 parameters of pore size (0.1, 0.2 and 0.45m), MLSS in the reactor (5000, 10,000 and 20,000 mg/L) and

suction pressure (1.36, 2.72, and 7.5 m head of water) were conducted during a short term experiment in order to find out the suitable mode of operation for long term experiment. The short term result shown that at 10:10 intermittent operation provided the best condition for the stable flux. In long term experiments, membrane modules were regulated at constant flux of 1.5, 2.5 and 3.5 L/m2.h and the corresponding increase in suction pressure was recorded. Volumetric organic loading of 3 kg COD/m3.d shown critical condition toward the separation process. However, loading of 2 kg COD/m3.d appeared to provide most suitable condition since the COD removal efficiency was upto 95-97%. Nitrification and denitrification was achieved 100% and 30-40% respectively. Under similar operating conditions, the removal efficiency were independent of the membrane pore size. The 0.45 m membranes which operated at lower suction pressure than the 0.1 m membrane under similar operating conditions can provide the highest flux (3.5 L/m2.h or 0.084 m3/m2.d) and similar in clogging characteristic to others. Low value of Y, kd and F/M ratio showed very small sludge production. The 100% removal of fecal coliform can be achieved by using 0.1 and 0.45m membrane filters. Yamamoto et al. (1989) investigated direct membrane separation by using 0.1m pore size hollow fibers which was immersed in the aeration tank, regardless of using secondary sedimentation tank for solid/liquid separation. The treated water was filtered by suction with various operation modes. Continuous suction exhibited dramatic flux decrease as well as high MLSS together with high pressure difference. Intermittent suction at low pressure (13 kPa) provided good result in order to prevent the unrecoverable clogging and to prolong the operating time without cleaning. COD removal can be achieved higher than 95% while the nitrogen removal can be reached up to 60% by investigating the intermittent aeration mode. Chiemchaisri (1990) investigated an activated sludge using 0.1m hollow fiber membrane modules for solid liquid separation. This study was conducted to treat low strength wastewater from AIT domestic wastewater. Comparison of the membrane bioreactor under different operating conditions, such as non-aerated and aerated, with different initial hydraulic retention time (HRT) of 1, 3 and 6 h which provided corresponding permeate flux of 4.17, 1.38 and 0.7 L/m2.h was studied. The process was operated at 10:10 intermittent time. From the experiment, it can be seen that the non-aerated bioreactor has an advantage over the aerated condition at an initial HRT of 3 and 6 h, since lower energy consumption was required while giving similar effluent quality and process stability. However, at lower HRT of 1 hour (or higher permeate flux, 4.17 L/m2.h) aeration is required in order to prevent membrane clogging. This highest flux of 4.17 L/m2.h seem to be a critical value since creating severe clogging condition. At lower flux, no clogging was observed under non-aerated and aerated conditions. The quality of permeate in term of COD was independent of the low volumetric organic loading at the range of 0.2-2 kg COD/m3.d. Because of the long solid retention time (SRT), the process was stable and steady, COD removal efficiency was similar in every experimental conditions.

The performance of 0.03 m pore size with 9 m2 surface area of hollow fiber membrane was also investigated in pilot-scale unit. Two hollow fiber membranes module was immersed in an aeration tank which feed with diurnally AIT domestic wastewater. The suction pump was used at 10:10 minute intermittent operation to extract the permeate through the membrane as shown in Figure 2.2. For jet aeration, the effect of jet aeration period (1/2 and 1 h) and jet aeration pattern 15 minutes for two times a day and 30 minutes for once a day was investigated. The jet aeration flow rate used was 20 L/min. In the bioreactor consist of 2 zone: aerobic and anaerobic on top and at bottom of the reactor respectively. consequently, low MLSS in aerobic zone of which it could reduce the clogging problem of membrane.

Figure 2.2 Schematic of Pilot-Scale Membrane Bioreactor (Chiemchaisri, 1990)

The mean hydraulic retention time (HRT) was determined after the permeate flux reached steady state. At the flux 4.17 L/m2.h has reached, average HRT of 1 day was obtained under diurnal varied loading. Diurnal variation in loading play a minor role in the nitrification process since more than 80% nitrification can be observed throughout the experiment. The MLSS in the bioreactor was affected by the aeration flow rate and optimum air flow rate in this experiment was taken as 7.5 L./min. which provided sufficient oxygen for the microorganisms and maintain low MLSS in the aerobic zone. Maythanukhraw (1995) applied 0.1 m hollow fiber membrane directly in the reactor for solid-liquid separation to treat domestic wastewater from AIT campus, Bangkok. For short term experiments, the effect of transmembrane pressures, intermittent mode of operation and duration of air diffusion were investigated to find out the optimum conditions which corresponding to high and constant flux obtained. Variation of transmembrane pressures were studied with values of 13.3, 21.3, 32.0 and 41.0 kPa. It was found that using 13 kPa transmembrane pressure was a limiting pressure for all experiments. The different of

operation modes were studied by varied the duration of effluent filtration and air diffusion: 5:5, 10:10, 15:15, 30:30, 60:60 and 15:15* (15* = 15 minute without sending air). The results shown that at the operational mode of 15:15 provided the best results. Although cyclic operation with air diffusion could not completely remove the clogging, air backflush technique in this mode of experiment could improve the flux by up to 371 % compared to the continuous operation. At 15:15 operation mode, further experimental run was continued to find out the optimum air diffusion duration on that mode. By varied the values of 15:5, 15:10 and 15:15, the best performance can be observed under 15:10 mode. Considering both recovery of permeate flux and net cumulative flux. To study the process performance of long term experiment under the optimum conditions obtained from short term experiment , the effects of HRT at 12, 6 and 3 hr. attributed to variation of volumetric organic loading. At 3 hr.- HRT provided fluctuation of volumetric organic loading in range of 1.9-2.1 kg.COD/m3-d. Transmembrane pressure was increased according to the cake formation on the membrane. So, after the long period of experimental runs the steep increasing in transmembrane pressure can be observed even operated under air backflushing. The periodically chemical membrane cleaning was needed in order to recover the permeate flux. The permeate flux contains good quality in term of the very low SS. Since the infinite SRT was operated, more than 90% of COD with effluent concentration below 20 mg/L was achieved in all runs. The TKN removal was more than 90 %, and total phosphate removal around 50 % in all experimental runs. The MLVSS/MLSS in the bioreactor was in the order of 20-30%. Inorganic mass balance calculation indicated a steady accumulation within the reactor. The lower fraction of active microorganisms in the bioreactor did not show any significant effects on the process efficiency. Nevertheless, it is anticipated that in longer run it might affect the process, thus it is advisable to have periodic sludge draining. To use the membrane as an air diffusers, the compressed air pressure should be high enough to produce steady stream of micro-air bubbles according to the bubble point concept. One way to overcome this problem was to use relatively large pore size membranes. The membrane cleaning process which was adopted in this study was found to be adequate to remove mainly the external membrane resistance. It is necessary to have chemical cleaning procedure for complete elimination of internal and external resistances, which mainly caused by the macromolecular adsorption. Longer air diffusion will improve the recovery of permeate flux. However, by considering both the recovery of permeate flux and net cumulative flux 15:10 operational mode gave better results than the 15:15.

2.8 Clogging Mechanisms in Microfiltration The following four different types of clogging mechanism are observed in a filtration system: 1. Complete blocking occurs when particles plug the capillaries;

2. Cake filtration involves the formation of a porous layer on the membrane surface which poses an additional resistance; 3. Standard blocking occurs when solids adhere to the walls of the capillaries which reducing their internal diameter. Standard blocking was observed to be the most common in the absence of cake filtration. 4. The forth mechanisms was called intermediate blocking because of the rate of blocking falls between cake filtration and standard blocking. 2.9 Declogging Techniques In microfiltration system, particle deposition and internal clogging cause major operational problems in membrane filtration. These depositions cause permeate flux reduction in addition to decrease membrane life span. There are two simple ideas of declogging technique which are described as followed : 1. prevent particles reaching the membrane surface or 2. flush the deposited out

Backflushing technique is used in order to achieved higher membrane process efficiency. The rate of permeate flux would be increased by backflushing. A high pressure of air is applied from the permeate side in order to removed the deposits out of the membrane surface. For gas backflushing, gas is brought to pressure in the lumens from permeate side and then explodes through the membrane wall whereby the boundary layer is released and can easily be transported away as shown in Figure 2.3. This results in a very efficient cleaning of the membrane. The effects of backflushing with gas in permeate flux is shown in Figure 2.4

Figure 2.3 Module Backflushing with Gas ( Peters & Pederson, 1990)

Figure 2.4 Effect of Gas Backflushing during Wine Filtration (Vigneswaran et al., 1991)

2.10 Fundamentals of Gas Transfer Gas transfer is defined as the process by which gas is transferred from gas phase to liquid phase. Oxygen transfer in the biological treatment of wastewater is the most common application in field of wastewater treatment. Due to the low solubility of oxygen, normal surface air-water interfaces can not provide sufficient oxygen. To satisfy the requirement of aerobic waste treatment, aeration devices are used to create additional gasliquid interface. The rate of molecular diffusion of a dissolved gas in the liquid depends on the characteristics of the gas and the liquid, temperature, concentration gradient and the crosssectional area across which diffusion occurs. Equation (2.5) is used to explain Figure 2.5. (2.5) rm = Kg.A (Cs-C)

where rm = rate of mass transfer ( mg/s.) Kg = coefficient of diffusion for gas (L/m2.s.) A = area through which gas is diffusing (m2) Cs = saturation concentration of gas in solution (mg/L)

C = concentration of gas in solution. (mg/L)

Under the conditions of mass transfer encountered in the field

rm = V.dC/dt

(2.6)

So, Equation (2.6) can be written as :

rc = dC/dt = Kg (A/V) (Cs-C)

(2.7)

where Kg(A/V) = KLa

(2.8)

Therefore Equation (2.7) can be written as :

rc = dC/dt = KLa (Cs-C)

(2.9)

where rc

= change in concentration, mg/L.s

KLa = overall mass-transfer coefficient, s-1 Cs = saturation concentration of gas in solution, mg/L C = concentration of gas in solution, mg/L V = volume of gas (L)

Equation (2.9) could be modified to the practical form which is presented as follow :

Log (Cs-CL)

Log (Cs-C0) - (KLa/2.3)* t

(2.10)

where, C0 = initial concentration of gas in liquid phase CL = concentration of gas in liquid phase

For a given volume of water being aerated, oxygen-transfer can be evaluated on the basis of the quantity of oxygen transferred per unit of air, which is introduced to the water with equivalent conditions. (temperature and chemical composition of the water, depth at which the air is introduced, etc.)

To measure oxygen transfer in clean water, the accepted testing procedure involves the removal of oxygen (DO) from a known volume of water by addition of sodium sulfite followed by reoxygenation near the saturation level belonging to the water temperature. The DO of the water volume is monitored during the reaeration period by measuring DO concentration at several different points. The data obtained were then analyzed to estimate the apparent volumetric mass-transfer coefficient, KLa. These estimates of various point of DO are adjusted to standard conditions.

Figure 2.5 Schematic of Two-Film Theory of Gas Transfer ( Metcalf & eddy, 1991) 2.11 Applications of Gas Diffusion through Membrane Semmems et al. (1991) developed a bubbleless hollow-fiber membrane aerator and tested for oxygenation of water . The aerator houses a bundle of sealed, hollow, gas permeable fibers that are filled with pure oxygen under pressure. By considering Equation (2.9) for mass transfer; raising the oxygen transfer efficiency could be accomplished by increasing KLa and Cs values. Eventhough, practically the external surface area of membrane is fixed. Therefore large KLa can be achieved with the action of elongated stationary bubbles of oxygen. Very high value of Cs can also achieved if specially coated membrane are used, of which it can be operated at pressures up to 60 psi with pure oxygen. At temperature of 10 o C and a flow rate of 75 gpm. through the pipe, dissolved oxygen of water is ranged from 023.1 mg/L Pierre et al. (1988) investigated bench-scale experiments using both dense polymer membrane and porous membrane as an aeration units. Bubble-free aeration using membranes has potential applications for wastewater treatment when conventional bubble aeration gives unsatisfactory results, such as toxic volatile organic compound stripping or foam production. In order to find out the design parameters of membrane aerators, both types of membrane were immersed directly in well-mixed biological reactor.

Three systems which were studied for these experiments based on different settings of design parameters such as: 1. Specific oxygenation capacity 2. Mass transfer characteristics of the membranes 3. Type of gas and 4. Operating gas pressure Three operation modes consist of A, B and C were conducted in order to define the essential design parameters for membrane aerator. Cases A and B used 5 and 9 bar. Case C was evaluated by using pressurized industrial oxygen. The mass transfer analysis indicated that the high oxygen flux were based on surface area of the membrane. The optimal value of 38.3 m2-membrane per m3-aeration basin was suggested to meet the aeration requirements of conventional bioreactor (100 g. O2 / m3.h) while the high- rate bioreactor required 213 m2 / m3 . The fraction of oxygen transferred is a design choice. The value of 0.8 was much higher than that possible with conventional diffused or mechanical aeration. As the fraction of oxygen transferred is increased, however, the average driving force decreases. Limitations of membrane aerator were due to the high capital cost according to the membranes. Furthermore, the membrane themselves seem to represent as an additional resistance to oxygen transfer, which could be translated into high energy cost. However, it is suggested that applications should be developed with industrial oxygen in order to reduce the requirement of membranes. Tariq & Semmems (1992) studied the mass transfer in a various pore diameter of hollow fiber membrane aerator. Individually-sealed hollow fibers were filled with oxygen and immersed in a flowing stream of water. Three experiments have been checked by measuring the pressure drop, gas flow velocity and gas composition along the length of the fibers. Pressure drop was measured according to the difference between inlet and outlet. The pressure drop due to friction across the 362 cm length was negligible. By calculating, the optimum operating pressure was below 1-3 psi. Gas flow velocity inside the fiber depend on the mass transfer coefficient. The gas flow velocity was ranged from 0.008 cm/sec to 0.03 cm/sec. The decrease of oxygen partial pressure inside the fiber was observed along the fiber length. Better oxygen transfer could be achieved by pumping oxygen continuously through the hollow fibers in addition to removing accumulated nitrogen due to the back diffusion from outside to inside the fibers. Feeding of pure oxygen encourages nitrogen enter the fiber. Since partial pressure of nitrogen within the fiber is less than external partial pressure of approximately 0.79 atm, back diffusion along the fiber occurred. However, the nitrogen back diffusion rate dose not increase significantly when operated with high water flow velocity outside the fiber and high oxygen feed flow rate inside the fiber in order to decrease the nitrogen concentration gradient and increase oxygen concentration gradient between outside and inside the fiber respectively.

Chapter 3 Experimental Set Up The experiments that were carried out in this study can be classified into three parts as follows: 1. Gas transfer efficiency of aeration units and measurement of initial membrane resistance 2. Short term experiments 3. Long term experiments 3.1 Measurement of Gas Transfer Efficiency of Aeration units 3.1.1 Materials Used - Ordinary ceramic porous diffusers - Micro filtration Membrane For this study, a microporous hollow fiber membrane was used. It is produced from high density polyethylene with the pore size of 0.2 m. Each membrane module was assembled in plastic air tight cap to ensure the capability of operation under vacuum pressure as shown in Figure 3.1. - Clean water Tap water was prefiltered by using cartridge ultra-filter to produce clean water for this test. - Chemicals : 60 mg/L i) Sodium sulfite ( Na2SO3 ) : ii) Cobaltous chloride ( CoCl2 ) : 0.2 mg/L - Dissolved oxygen probe One probe was used for measuring dissolved oxygen concentration. The sampling points were at the head, middle and at the end of reactor. The position of each point was at the middle of water depth. - Turbine mixer One turbine mixer was used for promoting turbulence and homogenous mixture. Speed with value of 50 rpm. was used for this test.

3.1.2 Experimental Set up

Figure 3.2 presents the experimental set-up. The porous diffusers and membrane are immersed separately in acrylic-rectangular reactors which held a working volume of 80 L. The reactor was filled with clean water, and the increase in oxygen concentration was measured by dissolved oxygen probe. Turbine mixer was used to promote turbulence.

Figure 3.1 Membrane Module Used for All Experiments

Figure 3.2 Experimental Set up for Gas Transfer Efficiency

3.1.3 Process Operation

The effect of pressure was studied in batch operation in order to define the appropriate KLa. To compare the transfer efficiency of two aeration devices, the membrane modules and ordinary air diffusers, were operated as shown in Figure 3.1. The testing procedure began with the removal of oxygen from water by addition of 60 mg/L sodium sulfite (Na2SO3) with cobalt (CoCl2) as a catalyst. (Pierre, 1989) The increase in oxygen concentration was measured during aeration under specified pressure, and the overall transfer coefficient was calculated from Equation. (2.10) log (CS-CL) = log (CS - C0) - (KLa /2.3)* t (2.10)

when,

C0 CL

= =

initial concentration of gas in liquid phase concentration of gas in liquid phase

The gas, which was sent to the testing unit, come from purified ambient air and pure oxygen. Dissolved oxygen was measured by using dissolved oxygen probe. Three sampling points are at the middle of water depth along the tank; head , middle and end. DO and water temperature were recorded every three minute intervals until the dissolved oxygen reached a constant level. The pressure which were applied in this study were 0.2, 0.4, 0.6, 0.8 and 1.0 bar. The maximum KLa were then indicated the appropriate operating pressure for air diffusion which was used for short term and long term experiments. The effect of gas flow velocity which was sent to aeration devices was studied. By varying the velocity of 9, 14 and 18 L/min., the calculated maximum KLa then indicated the optimum velocity which was used for short term and long term experiments.

3.2 Measurement of Initial Membrane Resistance 3.2.1 Material Used - Membrane Two microporous hollow fiber membranes produced from high density polyethylene with the pore size of 0.2 m was used in this study. Each membrane module is assembled in plastic air tight cap to ensure the capability of operation under vacuum pressure.

3.2.2 Experimental Set up

The measurement of initial membrane resistance was conducted by immersing membranes in the rectangular reactor. Clean water was fed to the reactor. The speed controlled roller pump was used for extracting the permeate. The transmembrane pressure was measured by mercury-filled manometer and the filtered water was recorded and returned to the reactor to keep the volume constant during the experiment. 3.2.3 Process Operation The relationship between the flux and transmembrane pressure is given in the following equation : J = P / Rm (3.1)

where, = flux (L/m2-h.) = transmembrane pressure (kN./m2) = viscosity (kN..s /m2) = apparent membrane resistance = Rmo + Rd

J Rm

where,

Rmo Rd

= initial membrane resistance = membrane resistance due to the deposition of solids

The modified equation to find the initial membrane resistance, when clean water is used, is: P = .Rmo.J + Po Po effect.

(3.2)

= initial transmembrane pressure required to over come the air blocking

So, by modifying the transmembrane pressure and measuring the permeate flux, the value of Rmo could be determined. 3.3 Short Term Experiments 3.2.1 Materials Used

- Membranes
Microporous hollow fiber membranes produced from high density polyethylene with the pore size of 0.2 m were used in this study. The properties of the membrane was similar to that used in measurement of membrane resistance. Each membrane module is assembled in plastic air tight cap to ensure the capability of operation under vacuum pressure application.

- Feed substrate
Glucose solution and tap water were used for concentrating or diluting the wastewater respectively to the desired COD concentration. - Biomass culture The seed microorganisms which was used in these experiments was obtained from the acclimatization of biomass which initially exist in septic wastewater. 3.3.2 Experimental Set-up - Acclimatization unit Acclimatization procedure of these biomass to raw septic wastewater was properly conducted by daily fill and draw operation before being placed in the reactor. Glucose solution was fed in order to concentrate biomass mixture in the acclimatization unit. By considering the initial characteristics of raw influent septic wastewater which presents in Table 3.1, glucose solution was prepared and fed daily in proportion to maintain COD concentration around 5,000 mg/L. Stock solution of 98.2 g-glucose/L is equivalent to 100 gCOD/L. Table 3.1 Characteristics of raw septic wastewater Parameters 1. Settled solid 2. Suspended solids 3. COD 4. Filtered COD 5. BOD 6. Filtered BOD 7. Total nitrogen 8. Total phosphate 9. pH Values 700 2,200-3,500 3,500-5,500 300-400 600-1200 158-560 240-350 30-38 8.88 Units mL/L mg/L mg/L mg/L mg/L mg/L mg/L mg/L

- Membrane bioreactor The schematic diagram of the experimental set up was similar as presented in Figure 3.2. Membranes were immersed directly in the reactor of activated sludge system.

Speed controlled roller pump was used for extracting the permeate while the gas was sent through the membrane module for declogging purpose. This operation was carried out alternatively. The intermittent extraction and air injection was then controlled by intermittent controller and solenoid valves. The transmembrane pressure was measured by mercury manometer. The filtered water was then returned to the reactor to keep the volume and concentration constant during the runs. Compressed air which was sent through membrane had to pass air filter to remove oil vapor and was maintained at 1 bar. Compressed air was also sent to the stone diffusers located at the bottom of the reactor. The air flow rate was controlled at 14 L/min. by air flow meter while the permeate flux was then recorded. The main purpose of aeration were to : 1. Allow homogenous mixture in the reactor 2. Provide additional dissolved oxygen for biomass 3. Remove or shear off solids deposited on the membrane surfaces

Figure 3.3a Schematic of Membrane Bioreactor Set Up

Figure 3.3b Actual Membrane Bioreactor Set Up

3.3 Process Operation The effect of suction pressure was studied with 7, 23, 32 and 40 kPa. The operating conditions for these test were then fixed at 15:15* and 15:15 operation mode. ( 15:15* = 15

minutes of filtration then 15* minutes without effluent filtration or sending any air diffusion while 15:15 = 15 minutes of filtration then 15 minutes of air diffusion) The effect of operation modes with air diffusion and effluent filtration were studied. The suction pressure of 7 kPa and 1 bar compressed air were kept for all the operational modes. All runs were conducted as batch experiments for a duration of five hours. The following five different modes were studied: (1) 30:30 (2) 25:25 (3) 20:20 (4) 15:15 (5) 10:10 ( 30:30 : indicates 30 minutes of filtration then 30 minutes of air diffusion) The effect of compressed air for air diffusion was also studied. The operating conditions for this test were kept at 15:15 operation mode with 7 kPa suction pressure. Five different runs were studied with 0.3, 0.5, 0.7, 1.0 and 2.0 bar 3.4 Long Term Experiments: 3.4.1 Materials Used - Membranes Two microporous hollow fiber membranes were used. The properties of membrane were similar to that used in short term experiments. - Feed substrate Septic wastewater collected by Department of Public Welfare of the Local Authority of Pathumthani Province was used as substrate for long term experiments. - Biomass culture The seed microorganisms which was used in these experiments was obtained from the acclimatization of biomass which originally exist in septic wastewater which obtained from short trm experiments. 3.4.2 Experimental Set-up This experimental set-up was similar to short term experiment. A working volume of 80 L is maintained by using floating valve. Since the mixed level in the reactor dropped due to the extraction of the permeate, then the same amount of substrate was then supplied automatically from the influent storage tank to the reactor by using floating valve. Roller pump was also used for extracting the permeate through one of the membrane modules conducted at the optimum operational frequency, identified in the earlier short term experiments. This operation was carried out alternatively which was controlled by intermittent controller and control valve. (Maythanukhraw, 1995) 3.4.3 Process Operation

The effect of HRT was studied in this experiment. The initial HRT of membrane bioreactor process was varied at 7, 12 and 26 hours by controlling the effluent permeate fluxes. The change in transmembrane pressure was monitored everyday to monitor the clogging of the membrane during the long term operations. The effect of sludge draining was also studied with SRT equal to 50 days. This could be achieved by draining 1.6 liters of mixed liquor out of the bioreactor everyday. 3.5 Analytical Methods The analysis during these experiments were carried out following the procedure stated in APHA, AWWA, JPWCF, 1991 except for NO3--N and NO2- determination. Nitrate and nitrite analysis were analyzed by sodium salicylate and napthylamine method, respectively. Total phosphate in sludge was determined by digestion or oxidation technique, which convert all forms of phosphate such as orthophosphate and condensed phosphate, both soluble and insoluble, and organic and inorganic species to the reactive orthophosphate. Then reactive orthophosphate was measured during colorimetric determination steps. This step involves the formation of intermediates (molybdophosphoric acid), which is reduced to the intensely colored complex substances. Table 3.2 Parameters Analyzed ( Long Term Experiments) Parameters COD pH DO Turbidity Temperature MLSS & MLVSS TKN NO2--N NO3--N Total- Phosphate standard plate count Total coliform Fecal coliform Sludge characteristics : TS, TVS, CST, specific resistance etc. Frequency everyday everyday everyday everyday everyday everyday once in two days once in two days once in two days once in two days once in each run once in each run once in each run once in each run Sampling points Inf / Eff Inf / Reactor / Eff Reactor Eff Reactor Inf / Eff Inf / Eff Inf / Reactor / Eff Inf / Eff Inf / Eff Eff Eff Eff Reactor

Remarks : Inf = Influent ; Eff = Effluent

Chapter 4 Results and Discussion 4.1 Gas Transfer Efficiency of Aeration Units and Initial Membrane Resistance 4.1.1 Gas Transfer Efficiency of Aeration Units Before conducting the experimental runs, the measurement of gas transfer efficiency of stone diffusers and membranes were determined. Their oxygen concentration versus time was recorded and gas transfer coefficient (KLa) values were then calculated corresponding to the Equation (2.10). Figure 4.1 and 4.2 present the relationship between oxygen concentration and time of ambient air and pure oxygen at different flow rate. (Results are summarized in Table A1 and A-3 in Appendix A) It was found that the dissolved oxygen in water was increased with time until it reached the saturation level, which is related to the water temperature. For this experiment, the saturation level of oxygen concentration when using ambient air is 7.423 mg/L (30.9 o C) and 37.36-37.81 mg/L (Metcalf & Eddy, 1991) for pure oxygen. Both using ambient air and pure oxygen, the higher diffusion rate of air to water could be observed with the higher flow rates. Figure 4.3 and 4.4 show the relation of Cs-Ct with time and based on this data the gas transfer coefficient, KLa, was calculated. Detail experimental results and calculations are presented in Table A-2-1 to A-2-3 for ambient air and A-4-1 to A-4-3 for pure oxygen of Appendix A. These results present higher slope with higher flow rate of gas which was sent through diffusers. It can imply that higher diffusion rate of gas phase to liquid phase was obtained. Figure 4.5 presents the comparison of gas transfer coefficient at 20 o C between ambient air and pure oxygen for stone diffusers. From these results, it can be deduced that the gas transfer coefficient of ambient air, from gas phase to liquid phase increases with higher air flow rates. Figure 4.6 presents gas transfer coefficient when using ambient air with membrane diffusers. By comparison these experimental results in Table 4.1 and 4.2, the maximum gas transfer coefficient was 28.05 h-1 when using 1 bar of ambient air. It clearly demonstrated the effectiveness of the membrane module as an air diffuser in comparison to the conventional stone diffusers.

26

4.1.2 Initial Membrane Resistance The relationship between filtration flux and transmembrane pressure was investigated with microfiltered water. The initial membrane resistance of each membrane module was then determined, prior to using them in the actual experimental runs. As shown in Figure 4.7.1 and Figure 4.7.2 ( Results are summarized in Table A-7-1 and A-7-2 in Appendix A), the permeate flux was directly proportional to the transmembrane pressure. This corresponds to the relationship given in Equation (3.2). It was found that the initial membrane resistance (Rmo) of membrane module 1 and module 2 were 8.01 x 1012 m-1 and 9.26 x 1012 m-1 at 24.7 o C respectively. The flux obtained for membrane module 1 and 2 were 310 L/m2.h and 350 L/m2.h respectively. Rm values are useful not only for modeling purposes, but also for evaluating the effectiveness of the cleaning procedure and for designing long-term operation stability of the membrane which is discussed in the next section. 4.2 Short Term Experiments 4.2.1 Effects of Transmembrane Pressure on Permeate Flux This test was conducted without air backflushing operation. Two membrane modules were used with alternative operation of effluent filtration and then stopping the filtration process. The intermittent mode of 15:15* was used for this run. Figures 4.8 and 4.9 show the relationship between permeate flux and time at different transmembrane pressure without sending air through the membrane ( Table B-1 to B-4 in Appendix B). The flux seem to be constant during 5-hour operation. At any rate, the higher flux could be obtained when the transmembrane pressure was increased. The final flux of this experimental performance at different suction pressures is presented in Table 4.1. Table 4.1 Comparison of Performance of Different Suction Pressure with 15:15* Operation Mode and without Air Diffusion Total Average Flux 1 (L/m2.h) 3.42 5.67 5.79 7.46 Final Average Flux 2 (L/m2.h) 3.63 5.54 5.83 7.25
Comment:

Suction Pressure (kPa) 7 23 32 40

Note: 27

15*
1 2

= = =

15 minutes without sending air Average flux at 5th hour Total average flux in 5th hour

Figure 4.10 ( Appendix B) shows the comparison of permeate flux between applying 15 minutes of air diffusion and not applying air diffusion. The flux was almost constant in both applications. Figure 4.11 and Table 4.2 show the comparison of flux at different suction pressure between the application with and without air diffusion. From these results, higher flux could be obtained when increasing suction pressure. Table 4.2 Comparison of Performance of Different Suction Pressure with

15:15 Operation Mode and 1 bar Compressed Air

Suction Pressure (kPa)

Total Flux (L/m .h)

2

Final Flux (L/m .h)

2

Percentage Improvement in Flux with without Air Diffusion 3

7 23 32 40

3.08 5.63 6.13 7.29

3.17 5.50 6.04 7.46

no improvement no improvement 4 3

Note:

Percentage improvement based on final flux

4.2.2 Effects of Operation Time By considering data from Table 4.1 and 4.2, the operation without sending air seem to provide higher flux when compared to the operation with air diffusion. The effect of operation time was then studied by extent the operation time from five hours to ninety hours. This experimental performance was conducted in order to clearly demonstrated the stability of permeate flux between applying and not applying air through membranes. This experimental result is presented in Figure 4.12. By operating without air diffusion, permeate flux declined apparently after ten hours of operation. In turn, operation with air diffusion provide constant permeate flux beyond this critical period of time. Due to the fixed membrane module 2 which caused lower performance ability, different value of permeate flux between module 1 and 2 was obtained and presented in Figure 4.12. 4.2.3 Effects of Intermittent Mode of Operation 28

During short term experiments, the effects of different operation modes on permeate flux were investigated. Two membrane modules were used in alternative function of air diffusion and effluent filtration. In all these experimental runs, transmembrane pressure was maintained at the critical value of 7 kPa. The MLSS values in all cases were maintained in the range of 12,500-12,875 mg/L. Figure 4.13 to 4.14 (Table B-5 to B-9 in Appendix B) show the performance of different modes of operation. There were 12 and 10 percent of flux improvement in 10:10 and 15:15 operation mode respectively. Results are summarized in Table 4.3. Based on this results, it can be concluded that flux was improved with high frequency of intermittent operation. As the study which was conducted by Maythanukhraw (1995), air diffusion improves permeate flux by: (a) It removes the surface deposition or external deposition and prevents the compaction of cake layer under filtration pressure. (b) It removes the particles which clogged the pore of membranes. Among different modes of operational test, 10:10 ( 10 minutes for effluent filtration and 10 minutes for air diffusion) operation mode presents the highest improvement in flux. In any case, accumulation of permeate volume was less because effluent filtration time was wasted during the lag time to have filtered water through the module, after each removal operation. For longer frequency, 60:60, provided longer air diffusion time which will increase the cleaning ability, however this will allow high solids compaction on membrane surface due to longer effluent filtration time. By considering all these factors, 15:15 operation was selected as the optimum condition for long term experiments. Table 4.3 Comparison Performance of different Modes of operation with Suction Pressure at 7 kPa and 1 bar Compressed Air Mode of Operation Total Average Flux (L/m .h)
2

Final Average Flux (L/m .h)

2

Percentage Improvement in Flux compare with Table 4.1

10:10 15:15 20:20 25:25 30:30

4.13 3.83 3.38 3.25 2.83

4.08 4.00 3.54 3.50 2.96

12 10 no improvement no improvement no improvement

4.2.4 Effects of Compressed Air Pressure for Backflushing

29

The effects of compressed pressure air on the permeate flux stability were also studied. As shown in Table 4.4 ( Table B-10 to B-14 Appendix B), the highest flux could be obtained at the performance by backflushing with 1 bar compressed air. Due to the limitation of membrane stability, application with 2 bar compressed air was not completely operated. Although membrane module has repaired with quick-set epoxy resins, it did not function properly as its original status. In any case, increasing compressed air which was sent through membrane can theoretical improving the surface cleaning of membrane. For the long term experiments, 1 barcompressed air was selected to achieve the membrane backflushing. Table 4.4 Comparison Performance of Different Compressed Air Diffusion

with Suction Pressure at 7 kPa and 15:15 Operation Mode

Compressed Air (bar)

Total Average Flux (L/m .h)

2

Final Average Flux (L/m .h)

2

Percentage Improvement in Flux Compare with the Table 4.1

0.3 0.5 0.7 1

3.92 3.75 3.29 3.83

3.92 3.54 3.38 4.00

8 no improvement no improvement 10

30

4.3 Long Term Experiments 4.3.1 Characterization of Acclimatized Sludge Characteristics of sludge were determined after the sludge was acclimatized at the start of the experimental run. For this test, MLSS and COD were recorded with time in order to determined the biomass built up rates and COD reduction or substrate utilization rates. The results were presented in Figure 4.15 and Table 4.5.

1200 1100

6000

5000 1000 Effluent Filtered-COD 900 800 700 2000 600 500 F-COD(mg/L) 400 0 10 20 30 60 Time (min.) 120 180 MLSS(mg/L) 0 1380 1860 1000 3000 MLVSS (mg/L) 4000

Figure 4.15 Variation of MLSS and Effluent Filtered-COD with Time

31

Table 4.5 Kinetics of Acclimatized Sludge Growth

Parameters Rate of bacterial growths , rg Substrate utilization rate2 , rsu Sludge growth coefficient , Y Maximum rate4 , m Maximum rate of substrate utilization per unit mass of microorganism5 , k Half-velocityconstant, substrate concentration at one-half the maximum growth rate6 , Ks Endogenous decay coefficient7 , kd
1 3 1

Bioreactor 0.11 0.022 5.34 79.2

Conventional ASP 0.4 - 0.8 -

Units mg-VSS/ L.s mg-BOD5 /s mg-VSS/ mg- BOD 5 d -1 d -1

specific

growth

14.3

2 - 10

8651.5

25 - 100

mg -BOD 5 /L

2.68

0.025 - 0.075

d -1

= =

rate of bacterial growths, rg ( Xi - Xo)/ (Ti -To) substrate utilization rate, rsu (So - Si)/ (Ti -To) sludge growth coefficient, Y (Xi -Xo)/ (So -Si ) maximum specific growth rate, m (1/Ti) maximum rate of substrate utilization per unit mass of microorganism, k m/Y half-velocity constant, substrate concentration at one-half the maximum growth rate, Ks 32

= =

= =

= =

= =

=>
7

rsu = - [k. Xi. Si ]/ [Ks + Si ] endogenous decay coefficient, kd rg = -Y. rsu - kd . X rg = net rate of bacterial growth, mg/L.s = (Xii -Xo)/ (Tii -To)

= =>

By comparing typical kinetic coefficients of activated sludge process for septic wastewater with domestic wastewater (Appendix D), it was found that the values for septic wastewater are higher than domestic wastewater. These results are reasonable due to septic wastewater contained large portion of stabilized organic substances. It implies that conventional activated sludge process for treating septic wastewater requires longer solid retention time (c ) to meet effluent standard. Based on these results, membrane bioreactor which incorporated with physical and biological processes seem to be more appropriate for treating this wastewater. 4.3.2 Effect of Hydraulic Retention Time In these experiments, three different experimental runs (RUN 1, RUN 2 and RUN 3) were conducted by keeping HRT at 26, 18 and 10.5 hours respectively. Reducing HRT was controlled by increasing permeate flow rate. In this study, septic wastewater was fed to the reactor continuously. Due to increasing permeate flow, average values of volumetric organic loading were fluctuated with 3.38 kg-COD/m2.d for RUN 1, 6.73 kg-COD/m2.d for RUN 2 and 12.2 kg-COD/m2.d for RUN 3. The intermittent operational mode of 15:15 (15 minutes filtration and 15 minutes air diffusion) was used in all runs. In order to maintain log phase of biomass growth, activated sludge was wasted 1.6 L everyday ( SRT = 50 days). 4.3.2.1 Transmembrane Pressure Figure 4.16a (Table C-4 to C-6 in Appendix C) shows the variation of transmembrane pressure with time of membrane module 1 and 2 for different experimental runs. It shows that transmembrane pressures increased with increasing permeate flow (Figure 4.16b). These increased flux caused the increase in solid deposition of membrane surfaces. By considering transmembrane pressures in each run, it shows that transmembrane pressure in RUN 1 and RUN 2 of both membrane modules were almost constant through out the experiments. However, the sharp increasing of transmembrane pressure in RUN 2 can be observed after 14 days of this run. After RUN 2 both membranes were cleaned ( 2 nd cleaning) by air diffusion in clean water to remove the cake layer on the membrane surfaces. Transmembrane pressures of RUN 3 were constant at early stage of this run, however transmembrane pressure began to increase dramatically after 7th day. Table 4.6 presents average values of transmembrane pressure for each experimental runs. Once the membrane get clogged, it increases the transmembrane pressure which, again in 33

turn, promotes membrane clogging. At this point, the air diffusion at 1 bar pressure is insufficient to remove most of the deposited solid material on the membrane surface. This results an unrecoverable dead end operation. Therefore periodic membrane cleaning was then necessary in order to maintain required hydraulic retention time (HRT). Table 4.6 Transmembrane Pressure of each Experimental Runs

Experimental Runs

Transmembrane Pressure (kPa) Module 1 Module 2 20.1 43.9 64.1

RUN 1 RUN 2 RUN 3

13.9 38.6 58.8

By considering biological solids concentration as shown in Figure 4.19a (Table C-7 to C-9 in Appendix C), activated sludge concentration increased with increasing volumetric organic loading. Effect of activated sludge attachment on membrane surfaces was increased when MLSS in the bioreactor became concentrated for each experimental runs. This occurrence caused transmembrane pressure to be increased. In RUN 1 and 2, transmembrane pressure seem almost constant. The attached sludge on external membrane surfaces was detached due to the agitation in the bioreactor and membrane air diffusion. It allows sludge to move and shear off the solidified sludge from membrane which assist the cleaning abilities by membrane air diffusion. The rapid increasing of transmembrane pressure at the end of RUN 2 and in RUN 3were due to sludge concentration in bioreactor became highly concentrated which close to the limit concentration of activated sludge mixture. The limit for membrane filtration of activated sludge mixture is 30,000 - 40,000 mg/L which called as Threshold Limits of Sludge Concentration (Yamamoto, 1994). This caused less abilities of sludge movement and promoted cake compaction on membrane surfaces, especially among hollow fibers. By considering percent suspended solids to total solids which is summarized in Table 4.10, it was found that there were more than 98% of suspended solids to total solids. It imply that most of contained solids in bioreactor has particle size more than 0.45 m which is larger than membrane pore size. In order to maintain the stability of transmembrane pressure and permeate flux, application of periodic jet aeration of clean water in membrane module or increase the air diffusion pressure may assist to detach external deposition of sludge. Figure 4.16b shows the variation of permeate flux with time for each experimental runs. These results have the same pattern with transmembrane pressure. Operating at lower HRT caused higher permeate flux. Permeate flux of RUN 1and 2 were almost constant through experimental runs. For RUN 3 which was operated within threshold concentration of sludge, 34

permeate flux was highly fluctuated. This fluctuation was due to the attachment and compaction of sludge and detachment of solidified sludge. In any case, constant permeate flux can be observed for 5 days at the early stage of this run after membranes were cleaned. 4.3.2.2 Turbidity Figure 4.17 (Table C-1 to C-3 in Appendix C) shows the effluent turbidity and dissolved oxygen concentration for all three experimental runs. Influent turbidity is very high which contained more than 1000. There was no significant different in the permeate turbidity. The average turbidity for RUN 1, 2 and 3 were 1.1, 0.4 and 0.3 NTU, respectively. This low effluent turbidity corresponds to 0 mg/L of SS, thus meeting well below the required Thailand effluent standard of 30 mg/L. When compared to the study with 0.1 m by Maythanukhraw (1995), effluent turbidity by using 0.2 provided lower values for all experiments. The reasons for this result can be delineated based on the operation with high concentration of activated sludge. This operation caused high cake layer formation on the external surface of membranes. Cake layer assist to absorb the macromolecules which contribute to yellow color in the effluent. However, high yellowish color was observed at the initial stage of membrane bioreactor operation, but later it became relatively clear which is presented in Figure 4.21. 4.3.2.3 COD Removal Figure 4.18 (Table C-7 to C- 9 in Appendix C) shows the influent and effluent COD of RUN 1, 2 and 3. The influent COD was maintained in the range of 4000 - 6000 mg/L. The removal efficiency in all runs was maximum from the beginning which was greater than 90%. The effluent COD can be explained by the following equation. S = Ks (1+kd. c) c (yk - kd) -1 (4.1)

S Y c Ks k kd

= = = =

concentration of waste surrounding the microorganisms (mg/L) growth yield coefficient mean cell residence time (d) waste concentration at which rate of waste utilization per unit weigh of microorganism in one-half the maximum rate (mg/L) = maximum rate of utilization per unit weight of microorganisms (d-1 ) = microorganism-decay coefficient (d-1 )

Although 98% of COD removal was achieved by bioreactor, however more than 30 mg/L of COD can be observed in effluent for all experimental runs. The remaining COD in effluent probably presented in form of color.

35

4.3.2.4 BOD Figure 4.20 shows the variation of F/M ratio and BOD concentration for different experimental runs. The average value of F/M ratio for RUN 1, 2 and 3 were 0.25, 0.36 and 0.43 kg-COD/ kg-MLVSS.d. Effluent BOD5 concentration was measured weekly. During all the experimental runs the effluent BOD was maintained below the expected standard of 20 mg/L. This means that effluent BOD5 is independent from the variation of F/M ratio. As this study was operated with long SRT, c = 50 days, therefore adequate sludge retention time allows dissolved organic substances with both high and low molecular weights can be taken up, broken down and gasified by microorganisms or converted into polymers as constituents of bacterial cells, thereby raising the quality of treated effluent. 4.3.2.5 MLSS Figure 4.19a shows variation of biological solids concentration at different HRT. Table 4.7 shows the average value of biological solids concentration of each experimental runs. Table 4.7 Average Value of Biological Solids Concentration of each Experimental Runs

Experimental Runs

Average value of Biological Solids MLSS (mg/L) MLVSS (mg/L) 13,485 18,757 28,781 % MLVSS/ MLSS 79.1 73.3 72.7

RUN 1 RUN 2 RUN 3

17,181 25,675 39,560

By increasing volumetric loading, MLSS and MLVSS were also increased while percent of MLVSS/ MLSS were decreased. The general decreasing trend of MLVSS/ MLSS in the bioreactor could be explained by the following inorganic mass balance. Input Bioreactor Output

Inorganic solid = (SS - VSS) mg/L

Inorganic solid = 0 mg/L

Input of Inorganic = Accumulation of Inorganic 36

Hence, the following Figure 4.19b (Table C-7 to C-9 in Appendix C) presents the accumulation of inert material in the reactor. This figure clearly indicates the slightly increase of inert materials within the reactor. Although sludge draining was conducted to maintain 50 days residence time, accumulation of inorganic matter can be observed. However percent of MLVSS/ MLSS was maintained above 70% , normal condition for healthy conventional activated sludge, for all runs ( Metcalf & Eddy, 1991). It should be suspected that further continuation of experimental runs will probably lead to relatively low MLVSS/ MLSS ratio. Such situation might hinder the biomass growth and possible reduction of MLVSS and reduction of organic removal ability. In order to avoid such phenomenon in long runs, it is advisable to investigate the operation with shorter sludge retention time. The details of method to measure biomass activities in bioreactor is presented Appendix E. 4.3.2.6 Nitrogen Removal Figure 4.22 (Table C-10 to C-12 in Appendix C) shows the influent and effluent of TKN for different experimental runs. It was found that TKN removal efficiency was maintained above 90% for all runs. High degree of nitrification was due to long sludge age condition which promoted the enrichment of low growth rate bacteria such as nitrifying bacteria. Membrane bioreactor also do not allow the washout of nitrifying bacteria. In order to explain the removal of nitrogen inside bioreactor, mass balance of total nitrogen was investigated as followed. 4.3.2.6.1 Mass Balance of Total Nitrogen Three principal mechanisms responsible for the removal of nitrogen are assimilation nitrification, denitrification and sedimentation.

During nitrification process, ammonia nitrogen is oxidized to nitrite and nitrate by nitrifying bacteria under aerobic conditions. As discussed in 2.3 and 2.4 Chapter II, it was found that more than 95% of influent TKN (total ammonia nitrogen) was converted to nitrite and nitrate nitrogen. In order to determine the presence of denitrification reaction, nitrogen mass balance was then calculated based on the following equations: (Thanasupsin, 1995) TKNi + NO2 i+ NO3 i = TKNe + NO2 e + NO3 e + N2 + assimilation by bacterial cell (4.3)

N2 (denitrified- N)

(TKNi - TKNe) + (NO2 i - NO2 e ) + (NO3 i - NO3 e) - (BODi - BODe ) 37 (4.4)

where

conversion factor (ratio of nitrogen to BOD which is required by microorganisms)

The term of (BODi - BODe ) represents as the removal of nitrogen by assimilation. The value of was calculated based on BOD : N = 100 : 5. So, value of 0.05 was applied in Equation (4.4). Mass balance of nitrogen at steady can be derive as presented in the following: Table 4.8 a Table 4.8a Measured Nitrogen Mass Balance at Steady State

Influent (mg/L) Parameters 26 TKN NO2 NO3 BOD 166.8 0.17 1.58 2400 HRT (hours) 18 352.8 0.27 0.063 2200 10.5 146.4 1.32 1.22 3000 26 0.28 0.2 164.2 1.5

Effluent (mg/L) HRT (hours) 18 0 0.53 337.24 12 10.5 0 0.45 30.4 8

Table 4.8b Calculated Data for Nitrogen Mass Balance (BODi - BODe ) 119.93

HRT TKNi -TKNe 26 166.52

NO2i - NO2e - 0.03

NO3i - NO3e - 162.62

N2 - 116.06

18

352.8

- 0.26

- 337.20

109.4

- 94.06

10.5

146.4

0.87

- 29.18

149.2

- 31.11

As shown in Table 4.8a, material balance of nitrogen shows that about 99% of nitrification was obtained by maintaining dissolved oxygen concentration above 1.5 mg/L. 38

(Chiemchaisri et al. 1992) Based on extended aeration of activated sludge process, high degree of nitrification was due to long sludge age condition of the system which allowed nitrifying bacteria to be grown up and membrane biorector which does not allow any loss of nitrifying bacteria from the system. Based on these results, denitrification process occurred. Because this operation contains highly concentrated activated-sludge, agitation inside the bioreactor might not be sufficient. Therefore, there was possible tendency of sludge attachment ,accumulation and settlement in dead-zone which can be observed at the bottom of bioreactor. Besides this occurrence, there was the compaction of cake layer on membrane surface which cause the anoxic conditions and limitation of substrates and nutrients at inner part of the attached sludge among hollow fibers. Chang and Moo Yang (1988) has been studied the oxygen penetration depth for animal cell immobilized in hollow fibers. The oxygen penetration was about 500 - 1,000 m (0.5-1.0 mm). As Chang and Moo Yang indicated, the shortage of oxygen concentration was probably due to limitation of oxygen penetration from both sides of membrane surface. Hence, Nitrate was reduced during denitrification steps. Gaseous products were stripped by air bubbles and released out from the system.

4.3.2.7 Phosphate Removal Figure 4.24 ( Table C-10 to C-12 in Appendix C) shows the variation of influent and effluent total phosphate for each run. The removal efficiency fluctuated with value of 28.3, 48.6 and 97.1% for RUN 1, 2 and 3, respectively. In order to explain total phosphate removal, mass balance of total phosphate was investigated as followed. 4.3.2.7.1 Mass Balance of Total Phosphate Total phosphate removal mechanisms can be categorized into two means. One is the assimilation by microorganisms for cell synthesis and the other is sedimentation of total phosphate inside the reactor. (Thanasupsin, 1995) TPi
i

TPe + (BOD i - BOD e) + (VSSi - VSS e)

(4.5)

where

conversion factor (ratio of phosphorus to BOD which microorganisms require for their growth)

= =

conversion factor (ratio of phosphorus to VSS) influent and effluent, respectively

,e

The term of (BOD i - BOD e) and (VSSi - VSS e) represent the removal of phosphorus by assimilation and sedimentation, respectively. For this study, phosphorus was 39

determined in term of total phosphate. Therefore, term of (VSSi - VSS e) which indicated the effect of biomass sedimentation is neglected. Mass balance of total phosphate at steady state which is used for this study become as followed: TPi = TP e + (BOD i - BOD e) (4.6)

The value of was put as 0.01 based on BOD: P = 100: 1. Mass balance of total phosphate at steady state can be derived as followed. Table 4.9a Data of Total Phosphate Balance

HRT

Influent (mg/L) TP i BOD i 2400 2200 3000

Effluent (mg/L) TP e 12.8 16.6 3.0 BOD e 1.5 12 8

26 18 10.5

36.5 38.5 82.9

Table 4.9b

Calculated Data for Total Phosphate Mass Balance

HRT

TP i (mg/L) [1]

TP e (mg/L) [2] 12.8 16.6 3.0

(BOD i BOD e) [3] 23.99 21.88 29.92

[2] + [3] = (mg/L) [4] 36.79 38.48 32.92

(mg/L) [5]=[1]-[4] -0.29 0.02 49.98

26 18 10.5

36.5 38.5 82.9

In column [5], it is expected to be zero. Based on the above results of HRT 26 and 18 hours, it can be described that phosphorus was removed by biomass assimilation. Only the active soluble phosphorus, ortho-phosphate, was consumed by the activated sludge in the bioreactor for their growth. For HRT 10.5 hours, phosphate removal was not balance. It can be concluded that there were some amounts of phosphorus which retained and then accumulated in the bioreactor. Due to the operation with high concentration of activated sludge in the 40

bioreactor, complex phosphate (pyro- and poly phosphate) and organic - bound phosphate which normally presented in combined sewage wastewater, can be removed by sludge attachment and then settled at the bottom of bioreactor. In order to clearly demonstrate the effects of sludge settlement, it is necessary to find out phosphorus containing sludge and active soluble phosphorus which presented in influent. To determine the amount of phosphorus which is contained in sludge can be accomplished by digesting sludge before phosphorus determination which followed the standard methods. For this study, sludge samples were taken from the bioreactor and settled for 30 minutes before analyzed. Table 4.9c presents characteristics of total phosphate in sludge. The marked increase in sludge containing phosphate of RUN 3 demonstrated that complex and organic-bound phosphorus were settled by sludge. Reactive soluble phosphorus data and related results were summarized in Appendix F.

Table 4.8a Measured Nitrogen Mass Balance at Steady State

Influent (mg/L) Parameters 26 TKN NO2 NO3 BOD 166.8 0.17 1.58 2400 HRT (hours) 18 352.8 0.27 0.063 2200 10.5 146.4 1.32 1.22 3000 26 0.28 0.2 164.2 1.5 Effluent (mg/L) HRT (hours) 18 0 0.53 337.24 12 10.5 0 0.45 30.4 8

Table 4.8b Calculated Data for Nitrogen Mass Balance

HRT TKNi -TKNe NO2i -NO2e NO3i - NO3e (BODi - BODe ) 26 166.52 - 0.03 - 162.62 199.93 N2 -196.06

18

352.8

- 0.26

- 337.20

109.4

- 94.06

10.5

146.4

0.87

- 29.18

149.2

- 31.11

41

Table 4.9a
HRT

Data of Total Phosphate Balance

Effluent (mg/L) TP e 12.8 16.6 3.0 BOD e 1.5 12 8

Influent (mg/L) TP i BOD i 2400 2200 3000

26 18 10.5

36.5 38.5 82.9

Table 4.9b

Calculated Data for Total Phosphate Mass Balance

HRT

TP i (mg/L) [1]

TP e (mg/L) [2] 12.8 16.6 3.0

(BOD i BOD e) [3] 23.99 21.88 29.92

[2] + [3] = (mg/L) [4] 36.79 38.48 32.92

(mg/L) [5]=[1]-[4] -0.29 0.02 49.98

26 18 10.5

36.5 38.5 82.9

42

4.3.3 Sludge Characterization As this study was conducted with high sludge concentration and SRT , its characteristic were analyzed and compared with conventional activated sludge (return sludge which was taken from Navanakorn wastewater treatment plant). Table 4.10 Characteristics of Sludge from Conventional Activated Sludge and Membrane Bioreactor Processes Parameters Conventional Activated Sludge (returned sludge) Membrane Bioreactor RUN 2

RUN 1

RUN 3

1.Settled volume (mL/1000 mL) 2.Suspended matter (mg/L), SS 3.Volatile Suspended Solids (mg/L), VSS 4.Sludge Volume Index, SVI 5.Sludge Density Index, SDI 6.Capillary Suction Time, CST (s.) 7.Specific Resistance x 10-7 (gr.s.2/g.2) 8. Total Solids (mg/L), TS 9. Total Volatile Solids (mg/L), TVS 10. SS/ TS

870

965

985

992

8,400 4,150

21,175 16,400

27,150 18,750

38,650 25,700

103.6

45.5

36.1

21.7

0.97 12.4

2.2 81.3

3.79 90.9

4.61 84.4

41.1

413.2

487.8

389.6

10,222

21,191

28,800

41,248

4,622

17,896

23,470

33,205

82.2

96.7

94.3

93.7

57

Based on the results in Table 4.10, It was found that characteristics of activated sludge from bioreactor have higher values than the conventional activated sludge. Operation with highly concentrated activated sludge caused high sludge density but less settleability. By considering the characteristics of activated sludge for different runs, reducing HRT caused the increasing of biological solids concentration. By considering the sludge characteristic such as Capillary Suction Time (CST) which was monitored as a simple, useful and rapid way to evaluate filterability of sludge, it presented higher values than the conventional activated sludge. Specific resistance which was analyzed for all runs. They also confirmed the same trend with CST values. High values of such parameters imply less dewaterability of sludge . While sludge retention time (c ) of bioreactor was maintained at 50 days, typical value of sludge retention time of conventional activated sludge is shorter, around 10 days. By considering all these results, there is a particular advantage of biorector in term of less sludge wastage. However, it has to face with handling difficulties problem of such volume of wasted sludge. To obtain sludge which can be trucked or shoveled, additional dewatering units are necessary. Conditioning is generally essential to prepare the sludge for the various equipment. Difficulties are often encountered in the processes used for reduction of the volume of liquid sludge; these difficulties are closely linked with the hydrophilic colloidal mass, of which sludge suspension are mainly formed. However, to increase removal efficiency of total phosphate, addition of metal coagulant (FeCl3 ) was planned to dose directly to bioreactor. This application may assist to modified the internal cohesive forces of the sludge, to break down colloidal stability and to artificially increase the size of the particles. In the case of hydrophilic sludge, part of the bound water must be freed and transformed into free water which is the basis of efficient operation of dewatering units. To find out the optimum, FeCl3 dosage was determined by jar test process. Sludge characteristics after adding FeCl3 were also analyzed. All these data are summarized in Appendix F. Configuration of biological flocs were also observed regularly by using slide microscope. Figure 4.24 a, b and c show biological flocs at steady state of HRT 26, 18 and 10.5 hours. 4.3.4 Microbiological Quality By using 0.2 m pore size, effect of bacterial rejection was investigated in this system. The total and fecal coliform in the permeate from membrane bioreactor and effluent from secondary sedimentation tank of conventional activated sludge were detected by using MPN method (Multiple Tube Fermentation Technique).

58

From the Table 4.11, it presents 100% bacterial removal in all experimental runs. By considering size of membrane pore, it is expected that no bacteria is allowed to pass through this membranes. As Escherichia coli which is a bioindicator for polluted water, the absence of fecal coliform could imply that this permeate is safe from further pathogenic organisms. Table 4.11 Microbiological Quality of Permeate and Effluent Activated Sludge No. Colonies / 100 mL Conventional Aactivated Sludge
5

from Conventional

Parameters

RUN 1

RUN 2

RUN 3

1. Total Coliform 2. Fecal Coliform

1.4 x 10

<2 <2

<2 <2

<2 <2

2.2 x 10 4

4.3.5 Dynamic Population of microorganisms in the Bioreactor Microorganisms population were observed regularly by using microscope. Table 4.12 shows dynamics of activated sludge at steady state for each experimental runs. Based on these results, predominated microorganisms which constantly found in all experimental runs were bacterial flocs (mixed culture of bacteria). Filamentous organisms can not be found for all runs. The absent of filamentous bacteria in bioreactor was due to the operation with: a) high concentration of substrate b) narrow range of pH-variation c) sufficient dissolved oxygen (DO) When concentration of substrate was controlled, the effects of flow rate variation on microorganisms dynamic were then studied. It was found that the variation of influent flow rate caused less effects on the population of bacterial flocs and free-swimming protozoa. Reducing HRT caused the increase of incoming substrates to bioreactor which promoted bacterial growth and created larger size of bacterial flocs. These results presented in Figure 4.25a, b and c. Free-swimming protozoa which constantly presented in all runs was the ciliate protozoa such as Colpoda, Stylonychia and Aspidisca. Because plenty of bacteria constantly presented in all runs. This apparent caused less competition of free swimming protozoa and less necessity of food searching by protozoa. Based on this result, it might caused the absent of flagellated protozoa which normally used its flagella for feeding. 59

The present of rotifer was observed in run 1 (HRT = 26 hours). This result was reasonable due to the operation with low F/M ratio (0.25 kg-COD/ kg-MLVSS. d) and long hydraulic retention time (HRT). Table 4.12 Dynamics of Microrganism in the Bioreactor at Steady State for each Experimental Runs Conditions Phylum prokaryotae protozoa protozoa protozoa rotaria Class Microorganisms Area found all bact. flocs. bact. flocs. bact. flocs. bact. flocs.

Name bact. flocs. Colpoda Stylonychia Aspidisca ?Philadina ? Proales ? Polyarthar bact. flocs. Colpoda Aspidisca bact. flocs. Colpoda Aspidisca

HRT = 26 hours

ciliata ciliata ciliata rotifer (spine)

Degree of Apparent ***** ** * * <

HRT = 18 hours

prokaryotae protozoa protozoa prokaryotae protozoa protozoa

ciliata ciliata

all bact. flocs. bact. flocs. all bact. flocs. bact. flocs.

***** ** * ***** ** *

HRT = 10.5 hours

ciliata ciliata

Note :

It was not possible to make clear identification.

Column

1 2 3 4 5 Row

Area found : Column 3/5 to 5/5 3/5 to 5/5 2/5 to 3/5 2/5 to 3/5 1/5 to 2/5 < 1/5 4.3.6 Membrane Cleaning Efficiency

and Row 3/5 to 5/5 or 3/5 to 5/5 and 2/5 to 3/5 or 2/5 to 3/5 and/ or 1/5 to 2/5 and/ or < 1/5

= = = = = =

***** **** *** ** * <

60

When the membranes get completely clogged by the biomass, the filtration runs were stopped for cleaning purposes. Cleaning processes of membranes were conducted after short term experiments and after the operation with 18 hour HRT of long term experiments. The membranes were cleaned by the following 2 steps:

1 st step : The membrane modules were immersed in a clean water bath and the external surfaces were manually washed by using water jet diffuser. 2 nd step : The membrane modules were then cleaned by sending air through each membrane for 1 hour. In order to find out the effect of the above membrane cleaning processes, membrane resistance was calculated by measuring the ultra-clean water flux. these results are presented in Figure 4.26 a and b (Table H-1 to H-4 in Appendix H) and Table 4.13 show membrane resistances after different stages of cleaning compared with that of initial membrane resistance (Rmo). Table 4.13 Membrane Resistance after Cleaning Procedure Rmo (m-1 ) 8.01 x 10 12 9.26 x 10 12 Rm1 (m-1 ) 1.04 x 10 13 2.13 x 10 13 Rm2 (m-1 ) 1.32 x 10 13 3.63 x 10 13

Membrane Module 1 Module 2

The mechanical cleaning procedures, which were introduced in this study, are expected to remove both external (Ref) and internal membrane resistance (Rif). After first cleaning, it was found that membrane resistances (Rm1 ) for module 1 and 2 increased 29.8 % and 130 %, respectively. This increasing resulted from internally or externally absorption of macromolecules. Due to the damage of membrane module 2, which was fixed with epoxy resins, sludge penetrated the damaged area and created a blockage at the point of damage. This blockage caused marked increase in membrane resistance. After second cleaning, membrane resistances (Rm2) for module 1 and 2 increased 26.9% and 70%, respectively. These increasing rate are lower than the values of first cleaning. Because the sludge penetrated the damaged area, producing a kind of self-repair function of membrane module 2. ( Yamamoto et. al, 1990) These might probably caused lower value of increasing rate of membrane resistance (Rm2).

61

Thus, it can be concluded that periodic mechanical cleaning with air diffusion could not prevent total internal resistances. Therefore, the chemical membrane cleaning might provide an effective cleaning in term of removing internal clogging which caused an internal membrane resistance (Rif) .

62

Table 4.10 Characteristics of Sludge from Conventional Activated Sludge and Membrane Bioreactor Processes Parameters Conventional Activated Sludge
(returned sludge) 1.Settled volume (mL/1000 mL) 2.Suspended matter (mg/L), SS 3.Volatile Suspended Solids (mg/L), RUN 1

Membrane Bioreactor
RUN 2 RUN 3

870

965

985

992

8,400 4,150

21,175 16,400

27,150 18,750

38,650 25,700

VSS
4.Sludge Volume Index, 103.6 45.5 36.1 21.7

SVI
5.Sludge Density Index, SDI 6.Capillary Suction Time, CST (s.) 7.Specific Resistance x 10 7 (gr.s.2/g.2) 8. Total Solids (mg/L), TS 9. Total Volatile Solids (mg/L), 0.97 12.4 2.2 81.3 3.79 90.9 4.61 84.4

41.1

413.2

487.8

389.6

10,222

21,191

28,800

41,248

4,622

17,896

23,470

33,205

TVS
10. SS/ TS 82.2 96.7 94.3 93.7

63

Table 4.11 Microbiological Quality of Permeate and Effluent from Conventional Activated Sludge
Paramete rs Conventional Aactivated Sludge
5

No. Colonies / 100 mL RUN 1 RUN 2 RUN 3

1. Total Coliform 2. Fecal Coliform

1.4 x 10

<2 <2

<2 <2

<2 <2

2.2 x 10 4

Table 4.13 Membrane Resistance after Cleaning Procedure

Membrane Module 1 Module 2 Rmo (m-1 ) 8.01 x 10 12 9.26 x 10 12 Rm1 (m-1 ) 1.04 x 10 13 2.13 x 10 13 Rm2 (m-1 ) 1.32 x 10 13 3.63 x 10 13

64

Table 4.12 Dynamics of Microrganism in the Bioreactor at Steady State for each Experimental Runs
Conditions Phylum prokaryotae protozoa HRT hours = 26 protozoa protozoa rotaria ciliata ciliata ciliata rotifer (spine) prokaryotae HRT hours = 18 protozoa protozoa ciliata ciliata all bact. flocs. bact. flocs. bact. flocs. Colpoda Aspidisca Microorganisms Class Area found all bact. flocs. bact. flocs. bact. flocs. bact. flocs. Name bact. flocs. Colpoda Stylonychia Aspidisca Philadina

prokaryotae HRT = 10.5 hours protozoa protozoa ciliata ciliata

all bact. flocs. bact. flocs.

bact. flocs. Colpoda Aspidisca

65

Chapter 5 Conclusions and Recommendations 5.1 Conclusions 1. Membrane air diffusion/aeration process plays an important role in the improvement of permeate flux stability by removing external deposits on the membrane surface, preventing the compaction of cake layer and improving internal pore clogging of the membranes. 2. To use the membrane as an air diffuser, the compressed air pressure should be high enough to produce steady stream of micro-air bubbles according to the bubble point concept. For 0.2 m microfiltration membranes, the highest gas transfer coefficient (KLa) could be obtained when compressed air was sent through membrane at 1 bar, was 28.05 h-1 , at 20 o C . 3. Although variation of permeate flux in five hours operation of cyclic operation with air diffusion was not substantially different to cyclic operation without sending air, the flux stability of cyclic operation with air diffusion could be observed beyond ten hours of operation. 4. By considering the recovery of permeate flux and lag time to overcome air blocking in membrane pore, 15:15 (15 minutes of membrane filtration and 15 minutes of air diffusion) operation was found to be the optimum operation mode. 5. Higher compressed air pressure will increase the backflushing ability in order to remove external deposits on membrane surface and prevent the compaction of cake layer. However, maximum limits of compressed air pressure in term of membrane module stability should be considered. For this experiment, 1 bar compressed air provided the highest flux improvement compared to the cyclic operation without air diffusion. 6. COD removal in all experimental runs were greater than 95%. Due to high solids retention time (SRT), around 50 days, and biomass concentration in the bioreactor, there was no significant effects on COD removal efficiency for various volumetric organic loading. The effluent quality in term of SS was extremely good. There were no solids loss (approximately 0 mg/L) in the effluent stream. 7. Due to aerobic condition constantly maintained in the bioreactor, TKN removal efficiency was more than 95%. Total phosphate removal efficiency of more than 35% could be observed in all experimental runs. 8. Due to high concentration of activated sludge was maintained in bioreactor, therefore the limitation of oxygen concentration inside the biological flocs, the settled sludge within dead zones in bioreactor and the compact cake layer among hollow fibers probably caused high degree of denitrification.

64

9. The MLVSS/MLSS in the bioreactor was in the order of 70-75%. Although 1.6 liters of sludge was drained out every day to meet 50 days SRT, inorganic portion in the bioreactor was slightly increased with time. In any case, the increase in inorganic portions in the bioreactor did not show any significant effect on process efficiency. Nevertheless, it is anticipated that in longer run it might affect the process. The increase in inorganic portions to a certain level could limit the biomass activities. Thus it is advisable to conduct further experiment to measure active microorganisms in the bioreactor. 10. Operation with low HRT, the membrane cleaning process adopted in this study was found to be not sufficient to remove external membrane resistance. 11. From the process efficiency point of view, the membrane bioreactor produced extremely good quality over the conventional activated sludge process. Because the common operational problem of settleability and bulking of sludge could be utterly eliminated by the membrane bioreactor. 5.2 Recommendations for Future Work 1. Experimental runs should be conducted for longer duration to identify the accumulation of inorganic portions and to investigate the measurement of microorganisms activities. 2. Due to the air blocking inside membrane pores, the possibility of using larger membrane pore size should be explored. Therefore, operating at low transmembrane pressure will reduce lag time to overcome air blocking inside membrane pores. As well as operating with higher compressed air could probably improve the external deposition of hollow fibers. 3. Nutrient removal, especially for total phosphate should be researched. The possibility of using chemical phosphorus removal (directly in bioreactor, or after bioreactor) should be conducted based on economic comparison. 4. In order to improve phosphorus removal efficiency by adding coagulant directly to activated sludge, further studies should be focused on sludge characterization in term of its dewaterability. 5. It is important to understand what really happens at the membrane/liquid interface, in order to identify the causes of flux reduction. The present study indicated that one of major causes of flux reduction is due to macro-molecular adsorption of the membrane surfaces. Thus, a detailed study should be conducted to investigate the role of macro-molecules in bioreactor flux reduction.

64

CONCLUSIONS
1. Membrane air diffusion/aeration process plays an important role in the improvement of permeate flux stability. 2. To use the membrane as an aerator, the compressed air pressure at 1 bar provided the highest gas transfer coefficient (KLa). 3. The flux stability of cyclic operation with air diffusion could be observed beyond ten hours of operation. 4. 15:15 (15 minutes of membrane filtration and 15 minutes of air diffusion) operation was found to be the optimum operation mode. 5. Higher compressed air pressure will increase the backflushing ability in order to remove external deposits on membrane surface and prevent the compaction of cake layer. 6. Due to high solids retention time (SRT), around 50 days, and biomass concentration in the bioreactor, there was no significant effects on COD removal efficiency. The effluent quality in term of SS was extremely good. 7. Due to aerobic condition constantly maintained in the bioreactor, TKN removal efficiency was more than 95%. Total phosphate removal efficiency of more than 35% could be observed in all experimental runs. 8. Due to high concentration of activated sludge was maintained in bioreactor, therefore the limitation of oxygen concentration inside bioreactor probably caused high degree of denitrification. 9. The MLSS/MLVSS in the bioreactor was in the order of 70-75%. Nevertheless, it is anticipated that in longer run it might affect the process Thus it is advisable to conduct further experiment to measure microorganisms activity in the bioreactor. 10. Operation with low HRT, the membrane cleaning process adopted in this study was found to be not sufficient to remove external membrane resistance. 11. From the process efficiency point of view, the common operational problem of settleability and bulking of sludge could be utterly eliminated by the membrane bioreactor.
64

Recommendations for Future Work

1. Experimental runs should be conducted for longer duration to identify the accumulation of inorganic portions and to investigate the measurement of microorganisms activities.

2. Due to the air blocking inside membrane pores, the possibility of using larger membrane pore size should be explored. Therefore, operating at low transmembrane pressure will reduce lag time to overcome air blocking inside membrane pores. As well as operating with higher compressed air could probably improve the external deposition of hollow fibers.

3. Nutrient removal, especially for total phosphate should be researched. The possibility of using chemical phosphorus removal (directly in bioreactor, or after bioreactor) should be conducted based on economic comparison.

4. In order to improve phosphorus removal efficiency by adding coagulant directly to activated sludge, further studies should be focused on sludge characterization in term of its dewaterability.

5. It is important to understand what really happens at the membrane/liquid interface, in order to identify the causes of flux reduction. Thus, a detailed study should be conducted to investigate the role of macro-molecules in bioreactor flux reduction.

64

References Adam, C. E. Ford, D. C, and Eckenfelder, W. W., Jr. (1981) Development of Design and Operational Criteria for Wastewater Treatment. Envir Press, Inc. Nashville, TN APHA, AWWA and JPWCF, (1991) Standard Methods for the Examination of Water and Wastewater, Inc., Newyork, 1991 Ben, A. R., Liu, M.G. and Vigneswaran, S. (1993), Recent Development of Membrane Process for Water and Wastewater Treatment, Water and Science Technology, 27: 141-149 Chang, H. N. and Yang, M. M. (1988), Appl Micro Biotechnol, 29: 107 Chiemchaisri, C. (1990), Design Consideration of Membrane Bioreactor in Domestic Wastewater Treatment, AIT Master Thesis, EV 90-97, Bangkok, Thailand Chiemchaisri, C., Wong, Y. K., Urase, T. and Yamamoto, K. (1992), Organic Stabilization and Nitrogen Removal in Membrane Separation Bioreactor for Domestic Wastewater Treatment, Water and Science Technology, 25: 231-240 Chiemchaisri, C., Yamamoto, K. and Vigneswaran, S. (1993), Household Membrane Bioreactor in Domestic Wastewater Treatment, Water and Science Technology, 27: 171-178 Cooper, P. and Thomas, V., (1994), Process Option for Phosphorus and Nitrogen Removal from Wastewater, Journal of Institution of Water and Environmental Management, 8: 84 - 93 Curds, C. R. (1973), A Theoretical Study of Factors Influencing the Microbial Population Dynamics of the Activated-Sludge Process-I, Water Research Pergamon Press, 7: 1269-1284 Curds, C. R. (1973), A Theoretical Study of Factors Influencing the Microbial Population Dynamics of the Activated-Sludge Process-II. A Computer Simulation Study to Compare Two Methods of Plant Operation, Water Research Pergamon Press, 7: 1439-1452 Hooijmans, C. M., Abdin, T. A. and Alaerts, G. J. (1995) Quantification of Viable Biomass in Biofilm Reactors by Extractable Lipid Phosphate, Appl Microbiol Biotechnol, 43: 781-785 Leslie, Grady, C.P., Jr. and Henry, C. L. (1980), Biological Wastewater Treatment, Marcel Deckker, Inc., Newyork. Maythanukhraw, R. (1995), Application of Air Backflushing Technique in Membrane Bioreactor, AIT Master Thesis, EV 95-33, Bangkok, Thailand Metcalf and Eddy (1991), Wastewater Engineering : Treatment, Disposal and Reuse, 3rd ed., McGraw Hill Inc., Newyork, U.S.A. Michale, J.S. (1991), Bubbleless aeration, Water Engineering and Management, 28: 18-19 Okihiko, H., Hisao, I. and Aizo, Y.(1991), Preparation of New Support for Immobilization of Activated Sludge, Journal of Fermentation and Bioengineering, 71: 376-378 1

Peter, T.A. and Pederson, F.S. (1990), MEMCOR-Crossflow-Microfiltration with Gasbackwashing: Design and Different Applications for a New Technical Concept, Proceedings of the 5th World Filtration Congress, Nire, France, pp. 473-478. Pierre, C., Bersillon, J.L., Huyard, A., and Faup, G. (1988), Bubble-free Aeration Using Membranes Process Analysis, Journal of Water Pollution Control Federation, 60: 1986-1991 Pierre, C. (1989) Bubble-free aeration Using Membranes: Mass Transfer Analysis, Journal of Membrane Science, 47: 91-106 Semmems, M. J. and Tariq, A. (1991), Use of Sealed End Hollow Fibers for Bubbleless Membrane Aeration: Experimental Studies, Journal of Membrane Science, 69: 1-10. Sopajaree, K. (1989), Domestic Wastewater Treatment using Membrane Bioreactor, AIT Master Thesis, EV 89-28, Bangkok, Thailand Sawyer, N. C, Mccarty, L. P. (1978), Chemistry for Environmental Engineering, 3 McGraw-Hill Inc., U.S.A.
rd

ed.,

Talat, M. (1988), Application of Direct Membrane Separation to Activated Sludge Process, AIT Master Thesis, EV 88-25, Bangkok, Thailand Thanasupsin, K., (1995), Treatment of Domestic Wastewater by Contact Media, AIT Master Thesis, EV 95-17, Bangkok, Thailand Vigneswaran, S., Vigneswaran, B. and Ben, A. R. (1991), Application of Microfiltration for Water and Wastewater treatment, Environmental Sanitation Reviews, ENSIC, AIT, Thailand, No. 31, 1991. William, W. (1989) Industrial Water Pollution Control, 2 nd ed., McGraw-Hill, Singapore Yamamoto, K., Hiasa, H., Talat, M., and Matsuo, T., (1989), Direct Solid Liquid Separation using Hollow Fiber Membranes in an Activated Sludge aeration Tank, Water and Science Technology, Vol. 21, pp. 43-54 Yamamoto, K. (1994), Advanced Treatment by Highly Concentrated Activated-Sludge and Membrane Separation Process, Contemporary Studies in Urban Planning and environmental Management in Japan, Department of Urban Engineering, The University of Tokyo, Japan