Brewery Wastewater Treatment Using Aerobic Sequencing Batch Reactors with Mixed Culture Activated Sludge by Luqiong Ling M.Sc. (Environmental Biology), University of Liaoning, 1990 B.Sc. (Biology), University of Liaoning, 1985 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE IN THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Chemical & Bio-Resource Engineering) We accepptrus thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA October, 1998 ©Luqiong Ling, 1998 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of CJ-e^mi-zA^t fc I3$p- /?e{t?ut~ce The University of British Columbia Vancouver, Canada Date Oct. IV , t992 DE-6 (2/88) Abstract Laboratory-scale aerobic sequencing batch reactors, in both suspended-growth and attached-growth modes, were used to study the treatment of brewery wastewater. A Ringlaces material was selected and employed for the attached-growth reactors. Experiments were conducted employing a wide range of hydraulic retention times, from 0.56 to 6.06 days. The experimental results demonstrated that brewery wastewater could be successfully treated using both suspended-growth and attached-growth aerobic sequencing batch reactors. Treatment efficiencies in terms of the removals of total organic carbon (TOC), the five days biological demand (BOD 5), chemical oxygen demand (COD), and suspended solids (SS) were consistently maintained over 90%, with the suspended-growth reactors performing significantly better than the attached-growth reactors. As the results of these experiments demonstrated that the performance of suspended-growth SBRs was superior to that of attached-growth SBRs, only the suspended-growth SBR system was selected to study the optimal conditions of HRT and loading rate. The results showed that the maximum removal of TOC and SS could be reached at the optimal of HRT and loading rate. The removal of TOC was more sensitive to variations in the HRT than to variations in the loading rate; however, the effect of loading rate was dominant in the removal of SS compared to the effect of the HRT. The pH remained relatively constant during the aeration stage. The dissolved oxygen concentration changed as aeration proceeded. This may be related to TOC ii degradation and microbial activity. A lower sludge production rate was observed in the aerobic suspended-growth SBRs. in Table of Contents Contents Page Abstract i i Table of Contents iv List of Tables vii List of Figures ix List of Abbreviations xi Acknowledgement xii Introduction 1.1 General characteristics of brewery wastewater 1 1.2 Objectives 4 Literature Review 2.1 Previous researches on brewery wastewater treatments 5 2.1.1 Anaerobic process in brewery wastewater treatment 5 2.1.2 Aerobic treatment of brewery wastewater 7 2.1.3 Other systems 9 2.2 Sequencing batch reactor system 11 2.2.1 Operation of sequencing batch reactor 11 2.2.2 The applications of sequencing batch reactor system in wastewater treatment 12 2.2.3 Advantages of sequencing batch reactor 15 iv Materials and Methods 3.1 Source of brewery wastewater and activated sludge ; 17 3.2 Equipment 17 3.3 Experimental Design 21 3.3.1 The aerobic SBR start-up and operation scheme 21 3.3.2 Studies on comparison of treatment efficiencies under suspended-growth and attached-growth conditions 23 3.3.3 Determination of optimal conditions of hydraulic retention time and loading rate under suspended-growth condition 26 3.3.4 Track studies of pH, TOC, and dissolved oxygen vs aeration time in suspended-growth sequencing batch reactor 28 3.4 Sampling and analytical methods 28 3.5 Data analysis 29 Results and Discussion 4.1 Characteristics of brewery wastewater 31 4.2 The start-up of SBRs 32 4.3 Comparison of performance of suspended-growth and attached-growth aerobic SBR 34 4.3.1 Effects of HRT and loading rate under two growth type conditions.. .34 4.3.2 Effects of growth types on the performance of aerobic SBRs 43 4.3.3 The observation of microorganisms under the microscope 46 4.3.4 Correlation among TOC, BOD 5 , and COD in brewery wastewater....50 4.4 Optimal conditions of HRT and loading rate in suspended growth SBR 51 4.4.1 Analysis of variance among the four reactors 51 4.4.2 Correlation between selected variables 53 4.4.3 Response surface analysis for the effects of HRT and loading rate on TOC removal 55 4.4.4 Response surface analysis for the effects of HRT and loading rate on SS removal 60 4.5 Track studies of pH, TOC, and dissolved oxygen vs aeration time in the suspended-growth aerobic SBR 65 4.5.1 The variation of reactors 65 4.5.2 The results of track studies 68 4.5.3 The relationship between TOC degradation and aeration time 69 4.5.4 Sludge production 75 Conclusions and recommendations 5.1 Conclusions 77 5.2 Recommendations 78 References 80 Appendix 85 vi List of Tables Tables Pages 1.1. The general characteristics of brewery wastewater 2 3.1. Apparatus required in the experiments 19 3.2. Time courses of aerobic SBR under suspended-and attached-growth conditions 25 3.3. Factors in the factorial experiments 27 3.4. Experimental schedule of factorial experiments 27 3.5. Sampling and analytical schedules for comparison of growth types 29 4.1. Characteristics of brewery wastewater 31 4.2. Correlation coefficients between studied variables in the experimental start-up....33 4.3. Effects of HRT on the treatment efficiencies of aerobic SBR under suspended- and attached-growth conditions 38 4.4. Summary of statistical analysis of HRT and loading rate effects on the performance of SBRs at two growth conditions 41 4.5. Summary of the effects of growth types on BOD 5 , COD, and TOC removal 44 4.6. Comparison of means from suspended- and attached-growth SBRs 44 4.7. The ratios of B O D 5 , COD, and TOC 50 4.8. Correlation coefficient for the four reactors 52 4.9. A N O V A for the four reactors in factorial experiments 52 4.10. Correlation coefficient and level of significance between variables in suspended-growth SBRs 54 vii Tables Pages 4.11. Summary of A N O V A in the factorial experiments 55 4.12. Response surface analysis of TOC for the effects of HRT and loading rate 56 4.13. The significance of TOC regression coefficient 58 4.14. Response surface analysis of SS removal for the effects of HRT and loading rate 61 4.15. The regression coefficient for the response surface of SS removalvs HRT vs loading rate 62 4.16. A N O V A the four reactors in the track study 67 4.17. The Tukey-test (LSD) of the four reactors in truck study 67 4.18. A N O V A for the effects of aeration time in the track study 72 4.19. The grouping of loading rate (based on the multiple-comparison) 72 4.20. Sludge production in the suspended-growth SBR 76 Vl l l List of Figures Figures Pages 1.1. Monthly monitoring of brewery wastewater 3 2.1. The scheme of SBR operation 12 3.1. Diagram of the reactors set-up 20 3.2. The operation scheme of SBRs 22 3.3. The photograph of a Ringlaces 24 4.1. TOC removal vs loading rate 35 4.2. TOC removal vs HRT 35 4.3. B O D 5 removal vs loading rate 37 4.4. COD removal vs loading rate 37 4.5. B O D 5 removal vs HRT 39 4.6. COD removal vs HRT 39 4.7. SS removal vs HRT 42 4.8. SS removal vs the loading rate. 42 4.9. Photograph of microbe taken from the suspended-growth SBR 47 4.10. Photograph of microbe taken from the attached-growth SBR 48 4.11. The response surface of TOC removal vs HRT vs loading rate in the suspended-growth SBR 59 4.12. The response surface of SS removal vs HRT vs loading rate in the suspended-growth SBR 63 4.13. Continuous monitoring of TOC vs aeration time in the four reactors 66 ix Figures Pages 4.14. The plots of TOC, DO, and pH vs aeration time in the suspended-growth SBR 70 4.15. TOC vs aeration time under different loading condition in reactor 2 71 4.16. The typical bacterial growth pattern in the pure culture 74 x List of Abbreviations A N O V A : Analysis of variance B O D 5 : Biological oxygen demand in five days COD: Chemical oxygen demand CORR: Correlation analysis DO: Dissolved oxygen LSD: Least significant difference MLSS: Mixed liquor suspended solids HRT: Hydraulic retention time SBR: Sequencing batch reactor SS: Suspended solids T K N : Total kjeldahl nitrogen TOC: Total organic carbon VSS: Volatile suspended solids VS: Volatile solids Acknowledgment I would like to thank my committee members Dr. Richard Branion, Dr. Anthony Lau, and Dr. Victor Lo (my supervisor) for their support and guidance. I'd also like to give my thanks to Juergan Pehlke, Neil Jackson, and Dr. Ping Liao for helping me set-up the equipment, and to my fellow graduate students for their valuable advice. Mr. Tom Saunderes and Anh-long are also highly appreciated for their help to collect the wastewater. Many thanks must go to my husband Hong Qian and my parents (Xiehe Ling and Jiping Li) for their encouragement and support. Without their inspiration, I could not have completed this program. xii Introduction 1.1 General characteristics of brewery wastewater Brewery plants produce large quantities of highly polluting wastewater rich in organic substances. As the scale and production of the brewing industry increases, the amount of wastewater increases substantially, resulting in increasing pollution problems in the environment. Due to wide variations in its strength (in terms of COD, B O D 5 , and TOC), pH and the amount of wastewater discharged, brewery wastewater tends to be very difficult to treat. In view of this situation, there is a need to develop a technology which is capable of efficiently treating the increasing volumes and strength of wastewater from brewery plants. The wastewater discharged from breweries is generally a combined effluent comprising discharges from various sources within the plant. The fermented liquor is the final product in the brewing process. Wastes arise from the separation of grain residues, from spent-hops and yeast from the fermentation processes, from spillage, from possible spoilage of the beer, from fillers as well as from packaging and from washing wastewater. The wastewater production rates from the brewing and packaging sections vary independently of one another. While the packaging process produces a high flow, high pH, weak waste composed primarily of spilled beer and caustic bottle cleaning solutions, the brewing process produces a low flow, neutral pH, and high strength alcohol-carbohydrate-protein waste. A continuous monthly monitoring of the effluent from a brewery plant showed considerable variation in general wastewater characteristics in terms of biological oxygen 1 demand (BOD 5), chemical oxygen demand (COD), and solids concentration. As described in Table 1.1, total COD varied from 87 mg/l to 6550mg/l, and the concentration of suspended solids varied from 16mg/l to 1162mg/l. The ratio of soluble B O D 5 to total BOD5 was about 0.91, which implied that most of the biodegradable materials were in soluble forms. One important characteristic of brewery waste is its fluctuation in flow and quality at night and weekends, compared with the average working day flow. The fluctuations in the quantities of wastewater discharged from a local brewery is depicted in Figurel.l . Table 1.1 The general characteristics of brewery wastewater. Total BOD5 41-4260 mg/l Soluble BOD5 34-3890 mg/l Total COD 87-6550 mg/l Soluble COD 37-4830 mg/l Total Suspended Solids 16-1380 mg/l Volatile Suspended Solids 11-1230 mg/l PH 6.1-9.1 The pH of the wastewater from various processes within the plant was neutralized in a pH tower within the plant. This ensured that the pH did not change significantly. Since brewery wastewater has a poor buffering capacity (Cronin, 1996; Cronin and Lo, 1998), hydrolysis and anaerobic activity usually reduce the pH. The pH tends to drop from 10 to 4 within a day at room temperature, and it will drop from 10 to 5 within 3-4 days in a walk-in cooler at temperature approximately 4 °C. Due to a preponderance of carbonaceous matter it tends to be relatively short of nitrogen-nutrients. Slight seasonal temperature variations in the wastewater ranged between 19 °C in the winter and 31.4 °C in the summer. 2 9000 8000 ^ 7000 -I _ J O) 6000 e Q 5000 9000 4000 JS 3000 O H 2000 1000 0 10 15 20 Time (day) 25 Figure 1.1 Monthly monitoring of brewery wastewater Data for the Figure 1.1 were obtained from a monthly continuous monitoring of brewery wastewater in a local brewery. Samples were collected randomly once a day. The sampling month was July, 1997. 3 1.2 Objectives The objectives of this study were: • to start-up aerobic sequencing batch reactors for brewery wastewater treatment by using mixed culture activated sludge; • to compare treatment efficiencies of suspended-growth and attached-growth in the aerobic SBRs; • to investigate the effects of variations in loading rate and HRT on suspended-growth aerobic SBRs; and • to distinguish the relationships of TOC degradation, the concentration of dissolved oxygen (DO), and the pH with aeration time in suspended-growth SBRs. 4 Literature Review 2.1 Previous research on brewery wastewater treatment D u e to its h igh organic content and high biodegradabi l i ty , brewery wastewater is idea l ly suited to b i o l o g i c a l treatment. A l l treatment methods bas ica l ly i n v o l v e the convers ion by microorganisms o f fa i r ly c o m p l e x stable/unstable organic c o m p o u n d s to carbon d i o x i d e and water. B i o l o g i c a l treatment results in the removal o f B O D 5 , the coagulat ion o f nonsettleable c o l l o i d a l sol ids, and the stabi l izat ion o f organic matter. M o s t o f the previous research i n v o l v i n g brewery wastewater treatment used anaerobic treatment systems. 2.1.1 Anaerobic processes in brewery wastewater treatment It has been w i d e l y recognized that anaerobic treatment o f h igh strength industr ia l wastewaters can be competi t ive w i t h conventional aerobic treatment processes. A m o n g those anaerobic treatment processes, anaerobic digestion, in particular, has been suggested as a suitable process i n brewery wastewater treatment (Sax, 1985; Ince, et a l . , 1986; H u a n g , et a l . , 1989; L o et a l . , 1989; W a r e , et a l . , 1989; A n d e r s o n , 1991; S t r o h w a l d , et a l , 1992; T a n e m u r a e t a l . , 1992; L i a n g , et a l . , 1993; C h u n g , et a l . , 1993; C r o n i n , 1 9 9 6 ) . T h e treatment of brewery effluent by the anaerobic digestion - ultraf i l tration ( A D U F ) process has been studied on a laboratory scale (Strohwald, 1992). T h e results s h o w e d that C O D reductions o f 9 6 % - 9 9 % were possible at a loading rate o f 15 k g C O D m - 3 . d - l and a hydraul ic residence t ime o f 0.5-0.8 days. T h e C O D values o f the U F 5 permeate (final effluent) were generally below 100 mg/l for an 80 day test period. Since the brewery effluent showed a nitrogen deficiency, the addition of urea was found to result in more stable and improved digester performance. Tanemura et al. (1992) investigated the operation conditions for anaerobic treatment of wastewater from breweries, and they compared the efficiency of single- and double-anaerobic fluidized-bed reactors ( A F B R ) . They found that the double-AFBR process was more advantageous in obtaining treated effluent which could be discharged into rivers. The anaerobic pretreatment of brewery wastewater on an industrial scale has been reported by Sax (1985). A Biothane wastewater treatment system employing upflow anaerobic sludge blanket ( U A S B ) technology has been used and has been in nearly continuous operation at a brewery since late 1981. Soluble C O D removal efficiencies of about 90% are routinely achieved at volumetric loading capacities of 5 to 10 kg C O D per cubic meter of digester volume per day and a hydraulic retention time within the digester vessel of about 6 hours. The methane purity of the generated biogas approaches 80%, and the methane yield is close to the theoretical stoichiometric quantity of 0.35 cubic meter per kg of C O D removed. The system has been shown to perform consistently despite fluctuations in day-to-day C O D loading by nearly a factor of three. Normal anaerobic biomass growth within the digester occurs, and sludge losses due to washout are trivial. However, several problems appeared in this system; for example, filamentous bacteria were growing luxuriantly, and some ancillary equipment for odor control and upstream solids removal needed to be improved. The growth of filamentous bacteria was resolved by upstream acidification to insure consumption of dissolved oxygen within the wastewater. 6 The anaerobic process potentially has many advantages. It produces methane gas which may be burnt as fuel resulting in the production of carbon dioxide and water. In addition, anaerobic fermentation produces far less sludge, than aerobic processes. Its wide applications has hindered by the inconvenience and difficulty of sufficiently intensifying the anaerobic process at ambient temperatures and the inability to adequately control the fermentation leading to methane formation. In addition, the resultant methane gas can be nauseating if leaked to the atmosphere. Furthermore, anaerobic treatment is generally considered as a pretreatment and usually requires post-treatment to meet effluent standards. In the specific case of the anaerobic treatment of brewery wastewater, the pH fluctuation, sludge bulking, and longer reaction times are among the problems encountered. 2.1.2 Aerobic treatment of brewery wastewater When the organic strength of a wastewater is not too high, conventional aerobic biological treatments are often a cost-effective form of treatment. In the case of brewery wastes, aerobic treatment has in the past proven to be successful on an industrial scale, as demonstrated by the deep shaft treatment system at Molson Brewery in Barrie, Ontario (Le Clair, 1984). This deep shaft process was capable of producing an effluent having an average TBOD5 and TSS less than 50 mg/l. It was claimed that the potential of deep shaft process would reduce energy and space requirements, have lower capital costs and reduced sludge production compared to conventional technology. However, the operation suffered from several mechanical problems such as the failure of the main downcomer section. In the case of a poorly flocculating sludge, an increased rotifer concentration and 7 nearly zero protozoa levels were seen during periods when yeast levels in the waste were high. Both protozoa and rotifer concentration dropped to low levels when beer levels in the waste were high. The high rate aerobic treatment of brewery wastewater using the Jet Loop reactor has been reported by Bloor et al. (1995). The use of a jet aeration system for the biological treatment of wastewater is becoming more commonplace as a means of combining efficient oxygen transfer with high turbulent mixing. In their studies, a loading rate of 50kg C O D / m 3 / d was achieved with 97% C O D removal for a period of 5 weeks. Although the settleability was found to be acceptable, non-flocculating motile bacteria caused the effluent to be cloudy and to have a high suspended solids concentration in the order of 200-350 mg/l. A Two-stage unitank system has also been developed for the treatment of brewery wastewater (Eyben, 1985). After a series of preliminary treatments including screening, grit-removal, no primary settlement, and eventually buffering, the wastewater is treated in a high-loaded combined aeration sedimentation stage. The BOD 5 -reduction is about 80-85%. The partial purified water then flows by gravity to a second low-loaded combined aeration sedimentation stage where the residual B O D 5 is removal to obtain the high-quality effluent resulting in more than 98% BOD 5 -reduction. This two-stage unitank system had better process performance in terms of high-treatment efficiency and control of sludge bulking. In addition it is a simple and reliable process and is easily controlled by microprocessor. Its flexible operation allows for temporary operation at half capacity and quick restoration to full capacity. It also allows for temporary operation as two high-loaded one-stage systems (during periods of peak production or heavy rainfall) thereby 8 conserving a treatment efficiency of 80-85% BOD5-reduction. Lower capital and operating costs are also an advantage of the two-stage unitank system. Eyben (1995) also studied the use of the unitank system in the biological removal of nitrogen and phosphorus from brewery waste water. They reported biological N and P removal efficiencies of 90-99 and 97%, respectively, using this system. Sequencing batch reactor systems are receiving increasing use in the treatment of municipal, industrial and agricultural wastewater (Irvine and Busch, 1979; Lo et al., 1986). However, reports on the application of the SBR process to brewery wastewater are scarce. The performances of an SBR operated in the conventional mode and an SBR using an alternative mode called the contact-stabilization mode were compared in the treatment of brewery wastewater by Yu et al. (1997). The results showed that this new contact-stabilization operational mode, based on the concept of rapid uptake of organic matter by biomass in a short retention time and subsequent regeneration of biomass after decant, had no negative influence on organic material removal efficiency and biomass settleability. A new operating scheme for a three-tank SBR system was proposed in their study. This alternative scheme was able to expand the three-tank SBR system's capability of withstanding inflow variations. 2.1.3. Other treatment methods of brewery wastewater As an alternative to the biological treatment of brewery effluent arising during the fermentation process, a recovery process for yeast and ethyl alcohol was analyzed in terms of its economic implications (Wysocki, 1973). Such an effluent can contain up to 16% yeast and 2.3% ethyl alcohol results in a BOD5 of 200 g/1. In economic terms, such a 9 recovery process for yeast and alcohol seems to be the most reasonable solution to this particular effluent problem. An aerobic spore-forming Bacillus sp., isolated from a hot spring, was found to produce hydrolytic extracellular enzymes when cultured on opaque brewery wastewater supplemented with defatted soy, spent yeast, and malt flour (Zvauya, 1996). Fe (1974) reported on the efficiency of the biofiltration of brewery waste water treatment. Advantages claimed included high efficiency, low construction costs, fast start-up, insensitivity to environmental conditions, low space and personnel requirements, and production of an easily flocculated protein-rich sludge suitable for sale as animal feed. The treatability of high strength brewery wastewater with stabilized refuse was studied by Fan (1990). A column with a surface area of 790 cm 2 and effective height of 105 cm was used for the study. The column was filled with stabilized refuse compacted to an average density of 500 kg/m3. High strength brewery wastewater with a COD of 6000 mg/l was homogeneously trickled over the top of the refuse at flow rates of 8, 16, 24, and 36 1/day. After six month intensive study, it was demonstrated that the stabilized refuse method was effective in the treatment of the brewery wastewater. The COD was reduced to as low as 60 mg/l after the wastewater had trickled through the 90 cm refuse layer which was equivalent to a removal efficiency of 99%. The variations in major parameters such as pH, alkalinity, and volatile fatty acids gave a strong indication that acidogenesis occurred quickly in the first 15 cm. As the flow rate increased, acidogenesis occurred deeper in the column and its recovery slowed. The oxidation reduction potential 10 (ORP) dropped to as low as -290 mV in the refuse column, indicating the anaerobic condition of the system (Fan, 1990). 2.2 Sequencing batch reactor The mass of contaminants present in domestic and industrial wastewater, in leachates and groundwater, and in soils generally varies with time and space. These natural and sometimes severe variations are coupled with the uncertainties associated with direct exposure to the environment. However, despite such unsteady-state behavior, facilities used for the removal of contaminants are often designed with the potentially unrealistic expectation that they can be operated as steady-state systems (Irvine et al., 1997). The SBR technology may be regarded as one of a number of methods which can be operated periodically and as a controlled unsteady-state system. 2.2.1 Operation of SBR An SBR biological treatment unit operates periodically in a typical cycle of five phases: FILL (inflow of wastewater), REACT (aeration), SETTLE (quiescent sedimentation of biomass and solids), D R A W (outflow of treated effluent) and IDLE. Figure 2.1 illustrates the scheme of a typical SBR process cycle. The filled phase and drawn phase are operated within a defined period of time. After completion of filled phase, variations in the influent of the treatment plant no longer have any effect on the process taking place in the reactor just filled. Before filling, the reactor contains an active and sizable microorganisms population which will biodegrade the influent wastewater. At start-up, microorganisms 11 W a s t e s l u d g e ( w h e n n e e d e d ) IDLE F I L L j Air R E A C T D R A W S E T T L E Figure 2.1 The scheme of SBR operation. must be seeded into the reactor from a suitable source. Once operating, however, the biomass remains in the reactor from cycle to cycle. The cyclic nature of the SBR process allows control over the dilution of operating conditions including the air supply and mixing, and it is thus possible to design a series of specific states in order to achieve a particular set of biochemical reactions on given waste materials. 2.2.2 The applications of SBR in wastewater treatment In recent years, SBRs have emerged as an innovative technology in the wastewater treatment systems. It has been found to be an efficient and flexible method for treating various dilute wastewater (Alleman et al., 1985; Decreon et al., 1985; Tarn et al., 1986; Imura, et al., 1993). Since SBRs are time-oriented, not space-oriented, fill/reaction ratios, aeration periods, and mixing cycles may be altered to accommodate specific operating conditions and to yield desired results. SBRs are therefore uniquely suited to wastewater treatment applications which are characterized by low and/or intermittent flow conditions. Furthermore, advances have occurred in sludge bulking control technologies using selectors mechanisms. Because SBRs have the flexibility to incorporate many of these selectors mechanisms in their operation, they are also suited to applications under high organic/low nitrogen concentration conditions (Sheker et al., 1993). It has been reported that an SBR system, operated in a fill-and-draw mode under sequential anoxic/aerobic conditions in a single tank, yielded a high combined removal rate of organic carbon and nitrogen compounds (Fernandes et al., 1991, 1994). An anoxic fill sequence, rich in exogenous organic carbon, favored denitrification and as a result oxidized nitrogen levels dropped. In the aerated react phase the ammonia accumulated during fill was oxidized to NO2 -NO3 -N. The nitrifiers were not inhibited by the anoxic operation. Some studies have also described SBR processes consisting of two or more tanks. The tanks were operated in a fill-and draw mode, which is the same as uni-tank system. One tank accepts the incoming wastewater while the others are in reaction, settling, draw-down, or idle phases. When the first tank is full, the incoming wastewater stream is diverted to a second tank which has been drawn and is in a standby phase ready to accept wastewater (Irvine et al., 1988; Ketchum et al., 1979). Ketchum et al. (1979) observed that a highly variable oxygen demand was exerted on an SBR system. They stated that the required aeration rate increased from a level needed for endogenous respiration in the standby phase, when the substrate concentration and liquid volume were both low, to a peak at the end of the filling cycle. During the reaction phase the required aeration rate 13 was dropped to a level needed for endogenous respiration and then shut off completely to allow settling and draw down to the minimum level needed to contain the settled solids. Novel approaches have also been developed based on sequencing batch reactor concepts. One was the sequencing batch biofilm reactor (SBBR) technology. Hirl and Irvine (1997) used anaerobic SBBR to dechlorinate reductively PCE (perchloroethylene). The A S B B R was a periodically operated up-flow packed column reactor. It was operated on a cycle consisting of three periods: FILL, REACT, and D R A W , and was constructed from glass columns filled with acid washed pea gravel as a support matrix. The consortium of microbes which can decholorinate was enriched by this ASBBR, and the reductive decholorination always occurred in the presence of methanogenesis. The efficiency of an SBR operation is also affected by factors such as nutrient levels, temperature, and fill strategies. It has been proven that temperature influenced nitrification and denitrification in anaerobic SBR (Fernandes, 1994; Schmit et al., 1994). It has been also proven that higher temperature exert a positive influence on the overall performance of the SBR in the range of 5-21 °C, and the process performance seriously deteriorated at 5 °C (Fernandes, 1994). The effects of anoxic and oxic fill strategies on SBR performance under nitrogen ( N H 4 C I as the nitrogen source) deficiency and rich conditions were evaluated using glucose as the sole substrate (Sheker, 1993). The performance was evaluated on the basis of substrate removal, sludge settleability, supernatant suspended solids, and reactor biomass concentration. Substrate removal efficiencies were found to be independent of the fill strategies adopted under all conditions tested. The incorporation of anoxic selector environment failed to prevent the development of bulking sludge under conditions of nitrogen deficiency, thereby resulting 14 in a gradual depletion of reactor biomass. Under nitrogen rich conditions, the sludge settleability improved significantly when an anoxic fill strategy was adopted. Furthermore, suspended solids readings taken at the end of the settling period were greater with anoxic fill than with oxic fill, indicating that the latter discourages the growth of dispersed bacteria. 2.2.3 Advantages of an SBR system as a wastewater treatment Compared with continuous systems, SBR systems are more dynamic and flexible in terms of operation and more advantageous kinetically (Irvine et al., 1979, 1989, 1997). The SBR process covers a range from feast to famine during the reaction cycle and the different aerobic/anoxic/anaerobic conditions imposed. Since SBRs impose a diverse array of operating conditions and selective pressures, they can be used as a versatile tool for the enrichment of specific consortia and the induction of desired metabolic pathways. By adding the system's own periodicity of forcing function, the potentially negative impact of those forcing function associated with variations in contaminant concentration, and other environmental uncertainties, can be migrated (Irvine etal., 1997). Research concerning sequencing batch operations indicated that the SBR concept is a viable and economically attractive alternative to the conventional continuous flow activated-sludge process in BOD 5 , SS and nitrogen removal, as well as in the chemical precipitation of phosphorus. The dynamic and flexible nature of SBR systems allows ample room for expansion and operational adjustments at minimal costs. 15 The advantages of an SBR wastewater treatment system are summarized below: • Less react volume because both aeration and settling are in the same tank; • Batch discharge of only treated wastes meeting effluent limitations (this is possible by monitoring the wastes and providing additional treatment if its quality is poor); • High oxygen transfer efficiency; • Facilitated design of a series of specific operating states; • Anaerobic periods enhance denitrification and nitrogen removal, control for filamentous organisms and reduce power consumption; and • Cost effective 16 Materials and Methods 3.1 Sources of brewery wastewater and activated sludge Brewery wastewater was collected weekly from a local brewery and stored in a cold room at 4 °C. The wastewater was allowed to reach ambient room temperature (20 ± 2 °C ) before being fed into the SBRs. An acclimation period was provided for the microorganisms to adjust to the new environment. The characteristics of brewery wastewater are depicted in Section 4.1. A mixed culture activated sludge, taken from the municipal wastewater treatment pilot-plant at the University of British Columbia campus, was used as seed for all the SBRs. Before seeding the reactors, the mixed culture activated sludge was settled, and only the condensed sludge was used. The range of B O D 5 of the activated sludge used to seed the reactors was from 6.66 to 15.6 g/1, and the average of suspended solids was about 4.93 g/1. 3.2 Equipment Four lab-scale reactors made of acrylic plastic pipe, with 62 cm in height and 19 cm in inside diameter were used in these experiments. The total volume of each reactor was 15 1, and the working volume of each reactor was 12 1. A computer system with L A B T E C H Control software (Laboratory Technology Co., 1994) was used to control the sequencing cycle as well as to monitor the pH and dissolved oxygen concentration in the reactors. A PC-711 board was installed in the computer and connected with a relay box, which was linked to the pumps and aerators. 17 Signals coming from computer switched the pumps and aerator on and off. Other instruments used in this research are listed in Table 3.1. Ringlaces was selected to set up immobilized growth aerobic SBR due to its ability in rapid entrapment of microbial biomass. This Ringlace is a synthetic fibrous rope product with loops approximately 1.5 cm long protruding from the center all along the length of Ringlace. It is manufactured from polyvinylidene chloride fiber 100 microns in diameter. It was reported that the material is water resistant and chemically very stable. The fibers are knitted and twisted into individual strands. The flexible fibrous loops are supposed to provide a large surface area for the attached growth and in combination with the swaying motion of the flexible rope tend to improve shedding action to prevent clogging in the process (Setter, 1995). Two peristaltic pumps (each with four pump heads) were used to feed influent and withdraw effluent from the four reactors. Therefore, the same flow rate of feeding and discharging were carried out for the four reactors. Aeration was controlled by adjusting the aerator setting so that the air supply was kept at the same amount for these four reactors. The set-up of a suspended-growth reactor, and a reactor with Ringlace (attached-growth) is shown in Figure 3.1. 18 Table 3.1. Apparatus used in the experiments. Apparatus Quantity Description Reactor 4 Acrylic plastic colume, 19 cm x 62 cm. Working volume: 12 1 Computer 1 386 IBM Input/Output control card 1 PCL-711 Peristaltic pump 1 Cole-Parmer, 6-600 rpm Peristaltic pump 1 Masterflex, 1-100 rpm, Model #661063 Pump speed controller 2 Masterflex, Model #7553-71, 50/60 Hz, 3 amp DO probe 2 Oxyguard, PT4 system Inc. DO meter 1 PT4 oxygen meter, PT4 system Inc. pH probe 2 V W R scientific (Ag/AgCl), Cat. #34105-023 pH meter 1 Good-Digital , 201 A T C Aerator 2 Maxima/Optima, Rolf C. Hagen Inc. 19 (a) . Effluent pump (b) . Influent pump (c) . Control system (d) . Air stone (e) . Aerator (f) . Ringlace (g) . Reactor (1) (2) Figure 3.1. Diagram of the reactor set-up: (1) a suspended-growth reactor; (2) a attached-growth reactor (i.e. reactor with Ringlace). 20 3.3 Experimental Design 3.3.1 The aerobic SBR start-up and operation scheme Four reactors were seeded with equal volumes of condensed activated sludge (2.5 1), and anoxically fed with brewery wastewater. The activated sludge was allowed to acclimate after innoculation in brewery wastewater for 3 to 4 weeks. Sludge retention time (SRT) was fixed at 21-28 days through the experiments. An anoxic fill strategy was adopted in order to prevent the development of bulking sludge (Sheker et al., 1993). The reactors were operated following the basic SBR operation scheme as described in Figure 3.2. Sequencing time length of cycles was set according to the purposes of the experiments. During the fill period, 4 1 of feed (brewery wastewater) was introduced to each of the reactor and the total working volume was brought from 8 to 12 1. Aeration was discontinued during the settle period, and sludge was allowed to settle under relatively quiescent conditions. In discharge stage, 4 1 of the clarified effluent was withdrawn and the liquid volume was decreased to 8 1. The idle period was set as a time to prepare and maintain the reactor for the next cycle. To study parameters related to the aerobic SBR start-up, two reactors were employed with the suspended-growth mode. One of them had a mixer installed and the other reactor was run without a mixer. The hydraulic retention time (HRT) was maintained at 1.56 days (including 12 hours for aeration, 20, 4, 3, and 3 minutes for settling, discharging, idle, and feeding time, respectively) throughout the start-up stage. 21 Waste sludge (when needed) IDLE (3 min) Influent (4 1) FILL (3 min, anoxic) REACT (4-48 hrs) Effluent (4 1) Working volume: 12 1 SRT: 21-27 days DRAW (4 min) SETTLE (20 min) Figure 3.2 The operation scheme of SBR. 2 2 3.3.2 Studies on comparison of treatment efficiencies under suspended-growth and attached-growth conditions In this study, two of the four reactors were used for suspended-growth, and the other two reactors were used for attached-growth. To set-up the attached-growth reactors, precisely tensioned strands of Ringlace, 5 cm apart, were attached between two horizontal plastic support rods. The support rods were then affixed to rigid frames installed in the reactors. The support rods were also spaced 5 cm apart, center to center, thus creating a matrix of Ringlace strands. Figure 3.3 is a photograph showing the set-up of Ringlace media. The operations of these reactors were the same as those of the suspended growth reactors. From our literature review and preliminary experiments, we found that the performance of these aerobic SBRs may be limited by factors such as inadequate oxygen transfer, settling of biomass, stability under different loadings, and hydraulic retention time. In this study, the HRT and loading rates were varied over a wide range of wastewater concentrations. Experiments were carried out with HRTs over a range of 0.56, 0.81, 1.06, 1.56, 3.06, and 6.06 days (Table 3.2). During these experiments, the sequencing cycles were adjusted according to the changes of aeration time, but the time for FEED, SETTLE, DRAW, and IDLE, was fixed at 3, 20, 4, and 3 minutes, respectively. Therefore, the difference of HRT was mainly due to the aeration time, and the time courses of the other stages had no effects on HRT. 23 Figure 3.3 The photograph of a Ringlace. 24 Due to the characteristics of the raw wastewater, variations of the influent concentration under each HRT would result in changes of the loading rate. All other experimental parameters, such as anoxic feeding, mean cell residence time (MCRT), influent and effluent flow rates, were maintained the same way in both suspended-growth and attached-growth reactors. The operations of the SBR were the same as described in Section 3.3.1. Table 3.2 Time courses of aerobic SBR under suspended growth and immobilized growth condition. Run# Aeration Settling Discharge Idle time Fill time HRT time (hrs) time (min) time (min) (min) (min) (day) #1 48 20 4 3 3 6.06 #2 24 20 4 3 3 3.06 #3 12 20 4 3 3 1.56 #4 8 20 4 3 3 1.06 #5 6 20 4 3 3 0.81 #6 4 20 4 3 3 0.56 Experiments progressed from the longest HRT to the shortest HRT, and each of the HRTs was continuously run at least 5 cycles (i.e. replication 5 times). When conducting shorter HRT runs (HRT at 0.56, 0.81, and 1.06 days), a time interval of 2-5 days was given between the change of each HRT. If the organic contents were not 25 completely degraded under a shorter HRT, the microorganisms would continue to utilize the remaining nutrients. The given time interval was to minimize the effects of interactions among HRTs. HRT was calculated by the following equation: HRT = V/Q (3.1) where, V represents the working volume of reactor, which was 12 1 in this study; and Q is the influent flowrate, which was calculated by: Q = 24/cycle time a day. The loading rate was calculated by: Where, the concentration of wastewater could be any of the parameters analyzed (TOC, B O D 5 , COD, and SS). 3.3.3 Determination of optimal conditions of HRT and loading rate under suspended-growth condition Factorial experiments were performed, with HRT and loading rate being two factors studied (Table 3.3). Three levels of HRT were used based on the previous experimental results (results from comparison of efficiencies of suspended growth and immobilized growth). Loading rates were mainly based on the different batches of wastewater Loading rate = Concentration of wastewater x Q (3.2) 26 Table 3.3 Factors in the factorial experiments. Factors Aeration time (hr) 4 8 16 HRT (day) 0.56 (00 1.06 (02) '2.06 (03) Concentration of wastewater (TOC) High (C2) 6 i x C 2 G 2 x C 2 0 3 x C 2 Low (CO 0i x C i 0 2 x C , 03 x C , Table 3.4 Experimental schedules of factorial experiments. Reactor Run 1 2 3 4 1 0i x C i 0 3 x C i 0 i x C 2 0 3 x C 2 2 03 x Q 0, x C , 0 3 x C 2 0 i x C 2 3 0 i x C 2 e 3xc 2 0i x C , 03 x C , 4 0 3 x C 2 Gix C 2 03 xC, 0, x C , 5 0 2 x C i 0 2 x C i 02 xC, 0 2 x C i 6 0 2 x C 2 0 2 x C 2 0 2 x C 2 0 2 x C 2 • 0i =0.56 days ;0 2= 1.06 days; 0 3 = 2.06 days; • C i = Low loading rate; C 2 = High loading rate. 27 collected. As shown in Table 3.3, "Low" concentration (TOC) of wastewater was obtained by diluting raw brewery wastewater with tap water. "High" concentration of wastewater was the original brewery wastewater. Experimental schedules are given in Table 3.4. Start-up of the aerobic SBRs in this study was achieved by seeding the four reactors with equal volume of activated sludge acclimated by brewery wastewater. The operation of the SBR was the same as described in Section 3.3.1. A complete factorial experiment was replicated 4 times (4 replicates). 3.3.4 Track studies of p H , T O C , and dissolved oxygen vs aeration time in suspended-growth S B R s In this study, four reactors acted as four replications for each run. Before starting sampling, the reactors were reseeded, and activated sludge was acclimated by growing in brewery wastewater for a week. The operation of SBR was primarily the same as in Section 3.3.1, except for oxic feeding. Sampling was started from the first minute of feeding, and the time interval was 30 minutes. The aeration setting was fixed, and it allowed a dissolved oxygen level of 2.5-3 mg/l. Dissolved oxygen concentration was read by directly put the DO probes into the reactors. 3.4 Sampl ing and analytical methods Influent, effluent, and sludge samples were taken for each run, and analyzed for TOC, BOD5, COD, total solids, suspended solids, volatile solids (VS), suspended volatile solids (VSS) according to the Standard Methods (APHA, 1995). Total Kjeldahl 28 nitrogen(TKN), ammonia nitrogen (NH 3-N), nitrate nitrogen (NO3-N), and orthophosphorus (orthol-P) were determined periodically using a autoanalyzer (Technicon) (Appendix table A-10). To study microbial populations under suspended and attached-growth conditions, samples were taken periodically to examine the quantitative change of the microbial population under a microscope, in order to estimate the dominant and sub-dominant microbes. Table 3.5 Sampling and analytical schedules for comparison of growth types. Growth type TOC COD BOD5 SS VSS Suspended 2 x 4 * 2 x 3 2 x 2 2 x 4 2 x 4 Attached 2 x 4 2 x 3 2 x 2 2 x 4 2 x 4 *: Analytical replication x sampling time 3.5 Data analysis Statistical analyses were conducted by using the SAS software package (Anon, 1988). Correlation analysis was used to determine significance of relationships between variables (using the Pearson product-moment correlation, SAS procedure CORR). Statistical comparison of the results from suspended-growth and attached growth was performed by analysis of variance (ANOVA) and r-test (TTEST) was applied to detect the significance of differences between compared means. The response surface regression (RSREG) procedure was employed. The RSREG procedure fits the parameters of a complete quadratic response surface and then 29 determine critical values to optimize the response with respect to the factors in the model. Many experiments were conducted to discover which factor values optimized a response. If a factor variable is measured at three or more values, a quadratic response surface can be estimated by least-square regression. The predicted optimal value can be found from the estimated surface if the surface is shaped appropriately. As mentioned in Section 3.3.3, two factors were selected and each of them had three (for HRT) or more (for loading rate) levels. Unless stated otherwise, the level of significance used throughout these studies was set at p < 0.05, and the results of data analysis of means were reported as Mean ± SD. 30 Results and Discussion 4.1 Characteristics of brewery wastewater Characteristics of the wastewater used in the experiments are summarized in Table 4.1. The wastewater, which was collected from a local brewery, was continuously discharged from various sources within the plant. The composition of the wastewater was typical of brewery wastewater which contained spent-grain, dead (or small amounts of live) yeast, spilled beer from fillers and packaging, and washing wastewater. The amount of wastewater discharged changed both diurnally and seasonally (Figure 1.1). This indicated that there were wide variations in the strength of the wastewater discharged, as shown in Table 4.1. These variations would undoubtedly affect the performance of treatment systems. Table 4.1 Characteristics of brewery wastewater. Parameter Range of concentration TOC 677-1720 ppm B O D 5 671-4200 mg/l COD 1038-4524 mg/l TSS 450-1044 mg/l TP 6.3-56 mg/l Orthol-P 2-60 mg/l T K N 28-343 mg/l N H 3 - N 0-4 mg/l pH 6.1-9.5 Even though the inorganic nitrogen-nutrients were very low in the brewery wastewater (NH3-N being around 0-4 mg/l), the nitrogen-nutrient was not a limiting nutrient, because the average T K N of mixed wastewater was about 343 mg/l. This average of T K N was higher than that of the supernatant wastewater (after solids settlement), which was about 34 mg/l. This is due to the spent grain and dead (or small amount of alive) yeast from fermentation contained in the solids contents of the wastewater. This suggested that most of the nitrogen-nutrient was in organic forms. Orthol-phosphate and total phosphate were sufficient for microbial growth. It has been reported that a suitable ratio of BOD:N:P is 160:4:1 for aerobic brewery wastewater treatment (Smith, 1986). Obviously, this ratio was attained in our experiments, and it was not necessary to add nitrogen-nutrients or phosphorus-nutrients to the reactors. 4 .2 The start-up of SBRs During the reactor start-up, the deterioration of biomass was observed. It was caused by fed with influent containing wastewater with low levels of organic content (BOD 5 being about 864mg/l) from the washing process. When these conditions pertained, sludge settling was poor and effluent suspended solids removal was lower than 75%. This suggests that the reactors would need to be re-seeded with fresh activated sludge in order to promote the recovery of biomass in the reactors. This would in turn affect the treatment efficiency of the SBR. The correlation coefficients for the variables, which includes the effect of mixer, mixed liquor suspended solids (MLSS), and activated sludge concentration, clearly 32 indicate the importance of MLSS and activated sludge concentration for the performance of the reactors. As shown in Table 4.2, the correlation coefficient of MLSS and activated sludge concentration to B O D 5 removal was 0.887 and 0.853, respectively. It is understandable why the amount of MLSS and activated sludge were strongly correlated with BOD5 removal and with reactor performance. Since SBR start-up is the process of biomass acclimating and accumulating, low concentrations of activated sludge and MLSS would lead to the biomass being washed out. As a result of insufficient biomass inside the reactor, the decomposition of organic compounds would not be achieved efficiently and therefore reactor performance would be poor. Table 4.2. Correlation coefficients between studied variables (the upper value is the correlation coefficient, and the lower symbol represents the level of significance). Variable Mixing MLSS Act.sludge Concen. B O D 5 MLSS 0.000 n.s. Act.sludge 0.000 n.s. 0.958 *** BOD5 0.125 -0.774 -0.834 n.s. *** *** Removal -0.053 0.887 0.853 -0.768 n.s. *** *** *** ***p < 0.001, n.s. = not significant (p > 0.05). The results from correlation analysis revealed no correlation between the function of the mixer and reactor performance; the correlation coefficient of the mixing effect and 33 B O D 5 removal was -0.053. This suggests that the reactors could be run without using a mixer. In order to verify this suggestion, an additional t-test was conducted to compare the mean values of B O D 5 removal from a reactor with a mixer and from a reactor without a mixer. As expected, results of the t-test demonstrated that the difference in performance between the reactor with mixer and the reactor without mixer was not significant (F = 1.17, p = 0.794). It seemed that sufficient mixing could be achieved inside the reactors by the force of aeration. A higher air flow rate (> 4 1/min) was set after the mixer was removed. This resulted in the disruption of sludge particles. A large amount of sludge in a fine particle form was suspended during the aeration and failed to settle in the serttling stage. Under these conditions, the effluent SS removal was about zero. The reactors were then re-seeded with acclimated sludge and air flow rate was reduced to 2.5-3 1/min. This aeration rate was then retained throughout the subsequent experiments. After the sludge had built-up in the reactors, the operation of SBRs attained a quasi-steady state. This was distinguished by a consistent concentration of TOC in the effluent and the visual observation of healthy sludge conditions in the reactors. 4.3 Comparison of the performance of suspended-growth and attached-growth aerobic SBRs 4.3.1 The effects of HRT and loading rate under two types of growth conditions Figure 4.1 shows the percentage removal of TOC influent brewery wastewater for different loading rates and under conventional suspended-growth and attached-growth biofilm conditions. 34 ioo ^ 96 92 88 H 84 80 H 76 72 Suspended-growth Attached-growth 4 —«• 0 8 12 16 Loading rate (g/l/day) 20 Figure 4.1 TOC removal vs loading rate. 100 96 92 88 84 80 H 76 72 >^—r / • Suspended-growth A Attached-growth 0 3 4 HRT (day) Figure 4.2 TOC removal vs HRT. 35 As shown in Figure 4.1, the TOC removal trends in suspended-growth conditions differed from that in attached-growth conditions. When the loading rate was about or below 10 g/l/day, the overall percentage of TOC removal attained was over 90% under both growth conditions. The TOC percentage removal decreased in the attached-growth reactors as the loading rate increased. However, little change was detected in suspended-growth reactors in term of TOC removal at different loading rates. The effects of HRT on the performance of the reactors under different growth conditions are shown in Figure 4.2. The TOC removal trends as a function of HRT in suspended-growth were also different from that in attached-growth reactors. The effects of HRT on TOC removal in suspended-growth conditions were not as significant as they were under attached-growth conditions. Under attached-growth conditions, the TOC percentage removal increased with increasing HRT until the latter reached a certain point (within 2-3 days of HRT as shown in Figure 4.2). After that point, TOC removal levelled out with increasing HRT. The experimental results of the effects of HRT on the treatment efficiencies are summarized in Table 4.3. Parallel to the TOC analysis, B O D 5 and COD analyses were conducted for these same samplings. Similar trends were found for the effects of HRT and loading rate on B O D s and COD percentage removals. Figure 4.3 displays BOD5 removal as a function of loading rate, and Figure 4.4 plots COD removal vs loading rate. Comparison of these two figures with Figure 4.1 reveals a general similarity. TOC, B O D 5 and COD removal decreased as the loading rate 36 96 j 1 >• a—m- • j 1 £ X T f-r- i _ ! f \ • Suspended-growth A Attached-growth 2 4 6 8 10 12 14 16 18 20 Loading rate (g/l/day) Figure 4.3 B O D 5 removal vs loading rate. a O O 80 H 75 70 Suspended-growth Attached-growth 10 20 30 40 50 Loading rate (g/l/day) 60 70 Figure 4.4 COD removal vs loading rate. 37 Table 4.3 Effects of HRT on the treatment efficiencies of aerobic SBR under suspended-growth and attached-growth conditions. HRT (day) Factor Influent (mg/l) Sampling No. Effluent (mg/l) Percentage removal Suspended Attached Suspended Attached 0.56 TOC 896.8±14.13 16 23.1+4.6 164.1±14.5 0.97+0.00 0.82+0.01 COD 2837±317.85 12 61.53+17.81 551.4+55.01 0.98+0.01 0.81+0.01 B O D 5 929.5±88.39 8 10.92±1.41 98.43±8.84 0.99+0.00 0.89+0.02 SS 585±183.85 16 10.78+3.55 140.28+21.66 0.98+0.01 0.74±0.09 0.81 TOC 852.3±120.00 16 27.57±6.17 108.61±18.60 0.97+0.00 0.87+0.01 COD 2449.3±439.83 12 66.3+20.94 273.3±70.13 0.96+0.02 0.84±0.04 B O D 5 1018.0+490.73 8 36.30+33.67 43.81±26.69 0.97+0.02 0.96+0.01 SS 597.5±17.68 16 9.91+3.83 100.7±25.23 0.98+0.01 0.83+0.04 1.06 TOC 905±62.00 16 32.55+31.07 79.17+18.81 0.96+0.03 0.91+0.02 COD 2853.3+661.99 12 89.43+81.08 223.5±124.9 0.97+0.03 0.93+0.03 BOD5 1548.0+483.66 8 20.30+28.34 43.12±12.29 0.99+0.01 0.97+0.01 SS 706.3+249.04 16 25.03+37.95 128.3+58.83 0.97+0.04 0.80+0.11 1.56 TOC 904.3±122.83 16 37.12+28.77 63.46±13.58 0.96+0.03 0.920+.01 COD 2962.3±1841.9 12 142.2+112.0 236.9±135.1 0.94+0.03 0.89+0.03 B O D 5 1327.5+354.97 8 0.53±1.13 12.07+5.17 1.00+0.00 0.99+0.01 SS 638.7±292.82 16 18.88±21.48 72.63±42.78 0.97+0.02 0.87+0.07 3.06 TOC 965.5±0.577 16 34.81+18.34 63.3±8.81 0.96+0.02 0.93±0.01 COD 2493.7+64.66 12 44.47±60.21 128.21±61.16 0.98+0.02 0.95±0.03 B O D 5 1237.5±180.31 8 3.34±3.21 12.83+4.72 1.00+0.00 0.99+0.00 SS 1044 16 41.44±38.36 54.25+39.8 0.96+0.04 0.95+0.05 6.06 TOC 966.5±20.23 16 27.23±7.90 58.09+3.14 0.970.0+1 0.94+0.00 COD 2359.3±222.86 12 21.18+25.91 104.3+51.16 0.99+0.01 0.95+0.03 BOD5 1489.0±154.15 8 10.33±7.67 10.27+7.88 0.99+0.01 0.99+0.00 SS 1044 16 30.81+12.08 33.69+14.48 0.97+0.01 0.97+0.01 38 i1 • Suspended-growth ± Attached-growth 3 4 HRT (day) Figure 4.5 B O D 5 removal vs HRT. 1 0 0 H 75 4 1 1 1 , , , 1 0 1 2 3 4 5 6 7 HRT (day) Figure 4.6 COD removal vs HRT. 39 increased in the attached-growth reactors, while in the suspended-growth reactor, the changes of loading rate were not reflected in changes in TOC, B O D 5 , and COD removal. COD removal was more sensitive to the effects of loading rate in both types of growth conditions than were B O D 5 and TOC removal. A linear decrease of COD removal was seen in attached-growth SBRs as loading rate increased (Figure 4.4). Figure 4.5 and Figure 4.6 show B O D 5 and COD removal as a function of HRT. As HRT increased, the COD removal in the attached-growth SBRs increased. However, it could never reach the same level as that in the suspended-growth SBRs. On the other hand, it appears possible for B O D 5 removal to be increased to the same levels for both growth regimes if the HRT was longer than 3 days. Based on the above results, it was inferred that the effect of HRT was more pronounced than that of loading rate on the performance of an aerobic SBR. This could be attributed to the fact that loading rate is determined by either the wastewater concentration or the influent flow rate, and HRT is related to loading rate by the influent flow rate (see Equation 3.1 and Equation 3.2). This inference was proven by the following results of statistical analysis. Table 4.4 contains the statistical results of HRT and loading rate effects on TOC percentage removal. Both HRT and loading rate had significant effects (at 90% confidence level). The effect of HRT was more sensitive to the growth regime than was the effect of changes in the loading rate. The effects of both HRT and loading rate on suspended solids removal were generally similar to those of BOD 5 , TOC, and COD. The percentage removal of 40 suspended solids decreased slightly as HRT increased in suspended-growth reactors (Figure 4.7). Increased HRT facilitated solid removal in the attached-growth reactors but longer HRT (particularly over 4 days) would not stimulate higher solid removal as shown in Figure 4.7. As would be expected, suspended solids removal decreased as loading rate increased in the attached-growth reactors, and it remained at nearly the same levels in the suspended-growth reactors (Figure 4.8). Table 4.4. Summary of statistical analysis of HRT and loading rate effects on the performance of SBRs under different growth conditions. Variable Effects (TOC) Suspended-growth Attached-growth F value p value F value p value HRT 1.89 0.10* 239.4 0.000*** Loading rate 3.21 0.000*** 3.14 0.001*** Level of significance is 0.1. *p<0.1, ***/?<0.001. It was known that the influent wastewater contained rather high levels of both organic carbon-nutrients and nitrogen-nutrients. Microbial activity degraded nutrients inside the reactors and available nutrients were exhausted as the HRT increased. The depletion of nutrients led to sludge bulking and poor sludge settling. Suspended solids removal therefore decreased, as shown in Figure 4.7. Experimental results also revealed that a portion of the suspended solids could be removed by attachment to the Ringlace in the attached-growth reactors. 41 «3 > O E o i— CO CO 100 -f 95 90 85 80 H 75 70 H 65 60 0 Jt • Suspended-growth A Attached-growth 3 4 HRT (day) Figure 4.7 SS removal vs HRT. re > o •E F 0.2105 5 2 Experiments were performed in these four reactors, and each of the pair-wise combination of two factors was assigned to one of the reactors as randomly as possible. The A N O V A was performed to analyze the variance of the reactors, and results are listed in Table 4.9. With F = 1.52, and the significance probability value associated with this F value 0.2105, the conclusion drawn above was confirmed. That is, the variances among the reactors were not statistically significant. Based on the above conclusion, the variations among the four reactors could be ignored. The results from the four reactors could be treated as replicates under their corresponding experimental conditions according to the following analysis. 4.4.2 Correlation between variables The effects of different variables on the values of TOC removal were tested, and the results were described in the following Table 4.10. A N O V A was used to determine the significance of these variables (Table 4.11). As shown in the results (Table 4.10 and 4.11), samplings were not correlated with TOC removal, nor was the variance in samplings significant (a = 0.05). Even though the Pearson correlation analysis showed that the correlation coefficient between replication and TOC removal was not significant, A N O V A indicated a slight significance in variances of replications at (p - 0.102). TOC removal had a positive correlation with HRT, but had a negative correlation with loading rate and the concentration of influent. Thus, particular attention was paid to the relative dominance of HRT and loading rate effects. The HRT and loading rate were strongly correlated (r = -0.811). The relationship between HRT and loading rate can be described by Equation 3.2. 53 Table 4.10 Correlation coefficient (upper row) and level of significance (lower row) between variables under suspended-growth conditions Variable Replication HRT Loading Concentration Sample HRT 0.000 n.s. Loading -0.003 n.s. -0.811 *** Concen 0.0195 0.053 0.427 n.s. n.s. *** Sample 0.000 0.000 0.000 0.000 n.s. n.s. n.s. n.s. Removal 0.0608 0.354 -0.578 -0.224 0.016 n.s. *** *** ** n.s. *p < 0.05, **p < 0.01, ***p < 0.001, and n.s. = not significant. 54 Table 4.11 Summary of A N O V A in the factorial experiments Variable DF A N O V A SS F value Replication 3 0.0202 2.10' HRT 3 0.3226 100.38"" Loading rate 23 0.437 16.9"" Concentration 13 0.2218 7.51"* Sample 1 0.0001 0.05 n s * p < 0.1, *** p < 0.001, n.s. = not significant. 4.4.3. Response surface analysis for the effects of H R T and loading rate on T O C removal Table 4.12 shows the response surface analysis of TOC removal vs loading rate and HRT. The response variable (dependent variable) was TOC removal; HRT and loading rate were treated as controlled variables. The mean value of response was 0.91 (TOC removal), and R-square was about 0.52 as displayed in Table 4.12. The relatively low value of the coefficient of variation (CV = 4.38) indicated good precision and great reliability of the experiments carried out. An optimum point (maximum point as well) of TOC removal was obtained by this response surface analysis, which was around 97% of TOC removal under HRT = 1.44 days with 3122 ppm/day of TOC loading rate. 55 Table 4.12 Response surface analysis of TOC for the effects of HRT and loading rate. (a) Parameters of TOC removal response surface Variable Response mean Root M S E R-Square TOC percentage removal 0.913812 0.040004 0.5247 Coef. of Variation 4.3777 (b) Critical Value Factor HRT (day) L O A D I N G Coded 0.170594 -1.101631 (ppm/day) Predicted value at stationary point* Uncoded 1.437945 3121.550113 0.979589 (c) Eigenvectors Eigenvalues -0.010578 -0.077435 HRT -0.043942 0.999034 LOADING 0.999034 0.043942 * Stationary point is a maximum 56 The application of response surface methodology yielded the following regression equation which is an empirical relationship between the dependent variable and the test variables in coded unit*: Y=0.965+0.020 X , -0 .023 X 2 -0 .077 X , 2 - 0 . 0 0 6 X , X 2 - 0 . 0 1 1 X 2 2 (4.1) Where Y is the response, (i.e. the TOC removal), and X i and X 2 are the coded values of the test variable HRT and loading rate, respectively. TOC removal with different HRT and loading rates can be predicted by using this regression equation. The significance of each coefficient, which are listed in Table 4.13, was determined by ?-test. The larger the magnitude of r-value and smaller the p-value, the more significant is the corresponding coefficient. This implies that the quadratic main effect of HRT (p < 0.0002) on the TOC removal was more significant than its first order effect. Both first order and quadratic effects of HRT were highly significant (p < 0.05). The loading rate had less effects on the TOC removal than did HRT, and the interaction between HRT and loading rate did not have a significant influence on the TOC removal (p < 0.90). This result was consistent with that given by the eigenvectors** (Table 4.12). As shown in Table 4.12, the HRT has higher eigenvectors than does loading rate. *Coded units/coded data: are standardized data. In this case, H R T C O d e d = (HRT u n c o ( j e ( j — 1.31)/0.75; T O C loading ra te c o d ed = (TOC loading r a t e u n c o d e d - 20719)/15974. **Eigenvalue/eivectors: are from the matrix of quadratic parameter estimates, which determine the curvature of the response surface. 5 7 A 3-dimensional graph was constructed to show the relationship between TOC removal vs HRT and loading rate. This graph was based on the regression data, with HRT on the abscissa, loading rate on the ordinate, and TOC removal perpendicular to the plane (Figure 4.11). This is a quadratically smoothed surface. The relatively flat response surface along the axis of ordinate, which represents the loading rate, indicates that a relatively wide variation can be tolerated without TOC removal being seriously affected. The very pointed surface seen along the axis of abscissa, which represents HRT, suggests that the TOC removal would be sensitive to changes of HRT. Table 4.13 The significance of regression coefficient (TOC). Model term Coefficient (coded) Standard Error r-value p-value Intercept 0.965 0.140 4.933 0.0000 X I 0.020 0.144 2.765 0.0063 X2 -0.023 0.000 0.127 0.8988 X I * -0.077 0.036 -3.799 0.0002 X 2 2 -0.011 0.000 -0.393 0.6950 X1X2 -0.006 0.000 -0.130 0.8967 58 Figure 4.11 Response surface of TOC removal vs HRT vs loading rate vs in suspended-growth reactors. 59 4.4.4 Response surface for the effects of HRT and loading rate on SS removal The HRT and loading rate effects on the suspended solids removal in aerobic SBRs was also analyzed using the response surface analysis. The results are shown in Table 4.14. In this analysis, SS removal was the response variable, which is a function of HRT and loading rate. The mean of SS removal on the response surface was 91.6%, and R-square was 0.389. With the coefficient of variation being about 5.16, the experiments were considered to be reliable. To achieve the maximum percentage removal of SS in this response regression, the predicted critical values of HRT and loading rate are 1.44 days and 13,838 mg/l/day SS, respectively. The regression equation of this response surface is generated as the following Equation 4.2 which is in coded units*: Y=0.996+0.017 Xi+0.046 X2-0.090 X, 2-0.092 X,X 2-0.142 X 2 2 (4.2) Where Y is the response (i.e. the SS removal), and X i and X 2 are the coded values of the test variable HRT and loading rate, respectively. The SS removal with different HRTs and loading rates can be explained using this regression equation. The significance of each coefficient of variables to SS removal was determined by Mests, which are listed in Table 4.15. Both the first order main effect and the quadratic main effect of HRT are highly significant (p < 0.05). This is in good accordance with the *Coded units/coded data: are standardized data. In this case, H R T C 0 ( i e C | — ( H R T u n c 0 d e d — 1.31)/0.75; SS loading r a t e c o d e d= (SS loading rate l i n c o d e d - 12559)78174. 60 Table 4.14 Response surface results of SS removal for the effects of HRT and loading rate. (a) Parameters of SS removal response surface Variable SS percentage removal Response Mean 0.916292 Root M S E 0.047314 R-Square Coef. of Variation 0.3890 5.1637 (b) Critical value Factor Coded Uncoded HRT (day) 0.014917 1.321188 L O A D I N G (mg/l/day) 0.156940 13838 Predicted value at stationary point* 1.00000 (c) Eigenvectors Eigenvalues -0.063279 -0.169217 HRT 0.863852 0.503746 LOADING -0.503746 0.863852 Stationary point is a maximum. 61 results from section 4.3.2. However, the SS removal was significantly affected by the loading rate, and the influence of the interaction between HRT and loading rate was not statistically significant (a = 0.05). The eigenvalues shows that the loading rate had a slightly stronger effect on SS removal than did the HRT (Table 4.14). Table 4.15 The regression coefficient of response surface of SS removal vs HRT and loading rate. Model term Coefficient (coded) Standard error r-value p-value Intercept 0.996 0.248 0.136 0.8920 X I 0.017 0.165 3.840 0.0002 X2 0.046 0.000 2.863 0.0052 X I 2 -0.090 0.035 -4.646 0.0000 X2l -0.092 0.000 -1.821 0.0719 X1X2 -0.142 0.000 -2.878 0.0050 Figure 4.12 is a 3-dimensional graph of the response surface showing the relationships between SS removal, HRT, and loading rate. Here, loading rate is the abscissa, HRT is on the ordinate, and response variable ( SS removal is perpendicular to the plane (HRT and loading rate). This response surface is a rather pointed surface, especially in the direction of the loading rate, which suggests that the SS removal is very sensitive to variations in the loading rate. 62 Loading Rate (g/l/day) Figure 4.12 Response surface of suspended solids vs HRT vs loading rate in suspended-growth reactor. 63 Differences in the effects of HRT and loading rate on TOC removal and SS removal may be explained by the differences in the characteristics of the wastewater. As mentioned in Section 4.1, brewery wastewater contains a large amount of solids waste. These solids contain "non-volatile" materials, including spent grain. While these solids ( except for yeast) contain no biomass, they do take up room in the reactor. Since only small molecules can be absorbed into microorganisms and afterwards digested, the large particles would be removed by surface phenomena, and the removal of solids by surface phenomena can be limited. As mentioned in Section 4.3.1, the effect of loading rate on the performance of suspended-growth SBR can be related to HRT, as well as to the concentration of the wastewater. Microbial growth would flourish with sufficient nutrients, although a higher loading rate could stop microbial growth due to the impact of shock and to a choking effect caused by the overloading of the solids. In our study, we observed that the increase in the concentration of influent (especially with higher solids concentrations) did result in a decrease in treatment efficiency. In general, a longer HRT gave more time for microbial action to degrade the organic content of the wastewater. Treatment efficiencies therefore increased with an increasing HRT. This did not, however, imply that the longer HRT, the higher the efficiency. As shown in the results of Section 4.3.2, an optimal HRT does exist. Since the SBR is run in batch mode, the nutrients are fed to the reactors at the beginning of the sequencing cycle. As the reaction stage goes on, the nutrients will be exhausted, and the concentration of any toxic compounds will increase in the reactor with a longer HRT 64 (Schmit et al., 1994). Thus, a too-long HRT can result in a decrease in treatment efficiency. 4.5 Track studies of TOC, DO, and pH vs aeration time in suspended-growth SBRs 4.5.1 Variations among the reactors The four reactors were run as replications for this continuous track study of TOC, dissolved oxygen, and pH. Figure 4.13 pictures the patterns of TOC degradation vs aeration time in the four reactors respectively. It was showed that reactors 1 and 4 were more similar than were the other two reactors. Reactor 2 may have had a higher degradation efficiency. It was demonstrated that the variation of the four reactors was statistically significant (Table 4.16). To minimize the variance from the four reactors, the four reactors were grouped based on the results from statistical multiple comparison (Fisher's Least Significant Difference — LSD). In Table 4.17, the means with the same letter are not significantly different, which suggests that they can be grouped together. The results of LSD shows that reactor 2 was certainly in a different group than reactor 1 and 4, while reactor 3 could be grouped with reactor 2 or with reactors 1 and 4 (Table 4.17). Based on this result, reactor 2 was excluded in the final data analysis when averaging TOC concentration, and the average of the data were calculated only from reactors 1, 3, and 4. The LSD results also indicated that reactor 2 had significantly higher performance in terms of TOC degradation, which statistically proved the observations made in Figure 4.13. 65 500 Aeration Time (hrs) Figure 4.13 Continuous monitoring of TOC vs aeration time in four reactors. 66 Table 4.16.Summary of A N O V A for the four reactors in track study of TOC degradation, DO and pH. Sources of Variance Reactors df 3 C.V. 64.92 Root MSE 104.18 F value 2.99 p value 0.04 Dependent variable was concentration of T O C . a = 0.05 Table 4.17 Grouping of the four reactors in track study of TOC degradation, DO, and pH (Using fisher's Least Significant Difference methods) REACTOR Mean* 1 191.21 (A) 2 108.24 (B) 3 155.14 (AB) 4 187.35 (A) *Means with the same letter are not significantly different. 67 4.5.2 The results of track study of pH, dissolved oxygen, and TOC in the suspended-growth aerobic SBRs Environmental conditions pertaining to pH, temperature, dissolved oxygen concentration, and nutrient content have an important effect on the survival and growth of microorganisms. In this research, temperature was fixed at room temperature, which was about 20 °C. As depicted in Figure 4.14, the pH within the four reactors fluctuated between 6.7 to 7.8, which is the optimal pH range for most microbial growth. The degradation of organic compounds by microorganisms inside the reactors produced large amounts of C 0 2 , which, together with other small carbohydrate molecules (such as acetic acids) plays an important role in maintaining the pH of reactors. It was not necessary to add buffer to the aerobic SBRs even though raw brewery wastewater has a poor buffering capacity (Cronny, 1996). The dissolved oxygen concentration, however, changed as aeration time increased, as did the concentration of TOC. Figure 4.14 shows that the dissolved oxygen concentration dropped slightly at the beginning of the reaction period after feeding with fresh wastewater. It decreased quickly from 9.7 mg/l to nearly zero (0.3 mg/l) within 6 hours. The start point of dissolved oxygen concentration in this figure was recorded when oxic feeding concluded. The reactors were saturated with DO by aeration at this point. The level of dissolved oxygen increased greatly at an aeration time of 6 -7 hours, and reached a platform at DO -9.3 mg/l. Meanwhile, the concentration of TOC inside the reactors was reduced from 441.9 to 126.05 ppm in 11.5 hours, most of the organic carbon 68 c o m p o u n d s were spent w i t h i n an aeration t ime 0-6 hours, and the r e m a i n i n g T O C cont inuous ly decreased as aeration t ime increased. 4.5.3 Relationship between T O C degradation and aeration time A n A N O V A and L S D procedure were run to statistically analyze the aeration t ime effects on T O C degradation. A s shown i n Table 4.18, aeration t ime had signif icant effect on T O C degradation. T O C degradation rate was calculated by E q u a t i o n 4.3 for each o f the groups (grouped based on the T-grouping) : T O C degradation rate (ppm/min) = A [ T O C ] / A T i m e (4.3) W h e r e , A [ T O C ] is an increase of T O C w i t h a certain A T i m e . A s s h o w n i n Table 4.19, the degradation rate o f T O C (3.37 ppm/min) was m u c h greater i n the first hal f an hour after feeding w i t h fresh wastewater. It var ied between 0.40-0.82 p p m / m i n unt i l aeration t ime reached 6.0 hours. T h e degradation rate o f T O C was reduced to about 0.19-0.27 ppm/1 as aeration proceeded past 6 hours. F i g u r e 4.15 shows the concentration of T O C vs aeration t ime at different influent concentrations o f T O C i n reactor 2. T h e trend for the two curves is a l ike , w i t h the o n l y difference be ing beg inning point. A t a higher influent concentrations ( T O C = 1539.5 p p m ) , there was a sharp decrease o f T O C after the first hal f an hour. T h i s steep drop d i d not occur w i t h l o w e r influent T O C concentrations ( T O C = 1109 ppm). 69 0 2 4 6 8 10 12 14 Aeration Time (hrs) Figure 4.14 The plot of TOC (average of reactors #1, #3, and #4), dissolved oxygen, and pH vs aeration time in track study. 70 500 Aeration Time (hrs) Figure 4.15 TOC vs aeration time under different influent concentrations in reactor 2. 71 Table 4.18. A N O V A analysis of aeration time effect in track study. Sources of Variance Aeration time df 21 C . V . 26.28 Root M S E 42.19 F value 23.88 p value 0.0001 Dependent variable was concentration of TOC. a=0.05 Table 4.19 The grouping of loading rate (based on the multi-comparison of the means). T Grouping Mean T I M E T O C degradation rate (ppm/min) A 441.90 0 3.37(A) B 340.70 0.5 0.82(B) C B 315.97 1 0.78(C) C D 292.57 1.5 0.82(D) E D 267.80 2 0.68(E) E F 247.30 2.5 0.65(F) G F 227.70 3 0.48(G) G 212.57 3.5 G H 198.67 4 0.72(H) I H 177.10 4.5 0.40(1) I J 165.07 5 0.72(J) K J 143.43 5.5 0.49(K) K L 126.05 6 0.27(L) K L M 113.97 6.5 0.25(M) L M 110.10 7 N L M 98.28 7.5 0.21(N) N 0 M 85.74 8 0.16(0) N 0 M 83.98 8.5 N O 69.81 9 N 0 66.97 11.5 N O 66.72 9.5 O 61.40 10 72 The stabilization of organic matter was achieved by the metabolic activity of microorganisms. As mentioned previously, nutrients and energy are used for microbial growth , and in the batch mode reactor, the changes in nutrient condition would therefore result in changes in microbial growth. A general bacterial growth pattern in pure culture is shown in Figure 4.16. A lag phase occurred in the beginning. It represents the time required for the organism to acclimate to their new environment. During this lag growth phase, there is always an excessive amount of food surrounding the microorganisms, and the metabolic rate and growth is only a function of the ability of the microorganisms to process the substrate. This lag-growth phase was followed by a exponential growth phase, in which cell mass and cell number density increase exponentially with time, and nutrient concentrations are large. The next growth phase is stationary phase (also called declining growth phase), in which the cells have exhausted the substrate or nutrients necessary for growth. Thus the growth rate of the biomass declined because of limitations in the food supply, and the growth of new cells is offset by the death of old cells. In the death phase, the bacterial death rate exceeds the production of new cells. Even though this growth model is the general bacterial growth pattern, it can be used to explain the growth of more complex microorganisms as well . The changes in T O C and dissolved oxygen concentrations may be explained by this growth model. As aeration time increased, the T O C concentration decreased because of the T O C degradation by microbial action. A n increase of biomass and oxygen demanded by the oxidation activities of the vigorously growing microbes resulted in a decrease in the 73 u a in S SN cu +J * J a &, O OJD "a • PM IN