OXIDATION R E D U C T I O N P O T E N T I A L (ORP) AS A R E A L - T I M E C O N T R O L P A R A M E T E R IN SWINE M A N U R E T R E A T M E N T PROCESS BY CHANG-SIX RA B.Sc. (Animal Science), Kwang Won National University, 1989 M.Sc. (Animal Science), Kwang Won National University, 1991 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF T H E REQUIRED F O R T H E D E G R E E O F D O C T O R O F PHILOSOPHY in T H E F A C U L T Y O F G R A D U A T E STUDIES (Dept. of Chemical and BioResource Engineering, Waste/Wastewater Management) We accept this thesis as conforming to the required standard T H E UNIVERSITY O F BRITISH C O L U M B I A Oct. 1997 © Chang-Six Ra, 1997 In presenting this thesis in partial fulfillment 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 the 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 fe»^ ^ C2ST Reduced form ne" : number of electrons transferred A mixture of the reduced and oxidized forms of a substance (e.g. Fe 2 + / Fe 3 + ; N A D H / N A D + ) is known as a redox couple. If an inert electrode, such as platinum, is put into a solution of a redox couple, the metal is charged, and the potential difference set up between the metal and the solution can be compared with the steady potential produced by a reference electrode, such as the hydrogen electrode (Wilson and Goulding, 1987). The reference standard is the hydrogen 8 electrode, which is assigned a potential of 0 volts under standard conditions (the pressure is 1 atm, 25° C, and species are at unit activity). A list of the reduction potentials developed can be found in many water chemistry references. In practice, the most common reference electrodes are the Ag/AgCl and the calomel electrode. Redox couples have positive or negative redox potential depending on whether they are more oxidizing or reducing than the standard hydrogen electrode. A negative potential with respect to the hydrogen half-cell has little tendency to exist in the reduced state, having little tendency to go to the right reaction. Such substances prefer the ionic form. In contrast, the positive potential has the tendency to accept an electron from an electron donor, having a strong tendency to go to the right reaction. The higher the potential, the greater the affinity of the acceptor for the electron. A positive potential means that electrons are transferring from reference electrode to sample solution, while a negative potential means the electrons are flowing from the sample solution to the reference electrode (Wareham, 1992). Also, the experimental potential measured depends on the ratio of oxidized to reduced forms and frequently on the pH. The redox potential measured is related to all the pairs of reducible and oxidizable compounds found in a biological or chemical system. The redox potential measurement is nonspecific in that it does not indicate the presence or absence of a particular ion but, rather, indicates the activity ratio of total oxidizing species present to that of total reducing species present. The ORP can be described in mathematical form with the Nernst equation. The Nernst equation can be used to calculate cell potentials for conditions other than standard state. The Nernst equation is: E h = E 0 + 2.303 [RT/nF] log { Oxid / Red } (1) 9 Where: Eh — the measured potential (the voltage difference between the oxidation-reduction half cell and the standard hydrogen electrode E 0 — the standard redox potential (when the activities of all oxidant and reductants are unity, at pH 0 and 25 °C Ox — Oxidized species Re — Reduced species n — number of moles of electrons transferred F — Faraday constant (23,061 cal/mole-volt) T — Temperature (degree Kelvin) R — Universal gas constant (8.315 volt-joules coulombs) {} — the activity of the oxidized and reduced species The form of the Nernst equation described above can be applied if H + is not involved in the change from oxidized to reduced form. If H + is involved in the equation, the equation becomes: E h = E 0 + 2.303 [RT/nF] log { Oxid / Red } + 2.303 [RT/nF] log [H +] a (2) Where: a — the number of protons involved in the reaction 2.3.2 Oxidation-reduction potential in biological treatment systems Most wastewaters are very complex solutions. A variety of chemical and biological reactions are occurring simultaneously during biological treatment. The microorganisms which make up the community of a biological reactor induce the chemical and physical changes. The microorganisms adsorb the nutrients present in wastewater and transport them across the cell wall. For example, carbohydrates are broken down into subunits outside the cell by extracellular hydrolytic enzymes, then transported through the cytoplasmic membrane for metabolism. The 10 compounds transported are degraded to produce energy (catabolism). The major reactions of catabolism are oxidation reactions and biological oxidations generally involve the removal of hydrogen or electrons. Oxidation reactions are associated with the release of free energy. This free energy is converted into biologically utilisable energy by microbial metabolism and stored inside the organism in a chemical form called adenosine 5' - triphosphate (ATP), which serves as a convenient energy carrier. Such oxidation reactions are always coupled with reduction reactions. Electrons removed from the substrate are transferred to electron carriers, such as the coenzymes N A D and FAD, then the electrons are transferred from electron carriers to the final electron acceptor. This is an oxidation-reduction reaction. A l l organisms use similar oxidation-reduction reactions to transfer electrons and similar mechanisms to use the energy released to produce ATP. An electron transport chain consists of a series of electron carriers that are capable of oxidation and reduction. In prokaryotic cells, the electron transport chain is found in the plasma membrane. Two distinct modes of microbial metabolism have evolved to accomplish these tasks, autotrophy and heterotrophy. Autotrophic metabolism does not require organic matter to generate ATP or as a carbon source for the biosynthesis of the cell; rather autotrophs generate ATP from the oxidation of inorganic compounds. The carbon for autotrophs is derived from inorganic carbon dioxide (CO2). For example, a small group of autotrophic bacteria in wastewater treatment system is chemolithotrophs that meet all their energy needs from the aerobic oxidation of an inorganic compound (i.e. nitrification) and use CO2 as their sole source of carbon (Ketchum, 1988). In the Calvin cycle, C 0 2 is reduced to form the carbohydrate (such as glucose) required for biosynthesis of the macromolecules of the cell (Atlas, 1986). The Calvin cycle requires a great deal of energy and reducing power; 18 ATP and 12 N A D P H are consumed for the conversion of CO2 to glucose. Based on the metabolism of chemoautotrophic 11 (chemolithotrophic) microorganisms described above, a biochemical model of a nitrifier is postulated and illustrated in Figure 2.1. Figure 2.1 A biochemical model for nitrifier metabolism (drawn by author) During the initial step in ammonia oxidation (nitrification), a mixed-function oxygenase is involved and the N A D H is used as an electron donor (Brock and Madigan, 1984). The ammonia taken into the bacteria (such as nitrosomonas) is oxidized to nitrite and ATP formation occurs via electron transport phosphorylation, through a cytochrome system. The produced and released nitrite is further oxidized to nitrate by nitrite-oxidizing bacteria, such as nitrobacter. 12 This reaction is carried out by a nitrite oxidase, with the electrons being transported to oxygen via cytochromes. 65 kcal/mole ( A G ' = -65 kcal/mole) and 18 kcal/mole energy (AG' = -8 kcal/mole) are produced in the oxidation of ammonia to nitrite and nitrite to nitrate, respectively. The energy available from oxidation reactions is used for electron transport and operation of the Calvin cycle for the conversion of C O 2 to organic compounds. The generation of reducing power (NADPH) for the reduction of C O 2 to organic compounds also comes from ATP-driven reserved electron transport reactions. In heterotrophic metabolism, an organic carbon substrate is required to generate energy or for the synthesis of cells. The organic carbon compounds taken by heterotrophic bacteria are oxidized to smaller molecules, releasing sufficient energy that is stored as ATP. In this reaction, the coenzyme N A D acting as an oxidizing agent is reduced to N A D H and the N A D H is reoxidized to N A D in subsequent biochemical reactions to ensure the continuous supply of N A D required for the ATP-generating pathway. Thus, the generation of energy required for various metabolic reactions occurs by a balanced oxidation-reduction reaction. Two types of biochemical mechanisms are involved in the ATP generation reaction, respiration (aerobic and anaerobic respiration) and fermentation. In aerobic respiration, the oxygen serves as final electron acceptor. When the inorganic substances such as nitrate ion (N03") or sulfate ions (S042~) serve as the final electron acceptor, the process is called anaerobic respiration. If organic substances serve as the final electron acceptor, the process is called fermentation. In respiration metabolism, the organic carbon compounds, such as polysaccharides (carbohydrate), taken up by bacteria are oxidized to pyruvate via the glycolysis pathway, which, in turn, can be oxidized into Acetyl CoA. Also, the lipids and proteins taken up by bacteria are oxidized into fatty acids and amino acids, respectively, that can be further oxidized into Acetyl CoA. In the Krebs cycle (Tricarboxylic Acid cycle), the Acetyl CoA is oxidized into C 0 2 , 13 releasing ADP. Then, the released energy is transferred to N A D (and/or FAD), forming N A D H ( F A D H 2 ) . Overall, the net reaction of the Krebs cycle, starting with the pyruvate can be explained by Equation 3. 2 pyruvate + 2 ADP + 2 FAD + 8 N A D —=> 6 C02 + 2 ATP + 2 F A D H 2 + 8 N A D H (3) As a final stage of fuel breakdown and processing, electron transport and oxidative phosphorylation occurs. The reduced coenzyme ADP is converted to ATP by oxidative phospholylation. Also, during oxidative phospholylation, electrons from N A D H and F A D H 2 are transferred to a terminal electron acceptor, through the electron transport chain, which involves a series of oxidation reduction reaction. When molecular oxygen present, these reactions result in the formation of water as a result of the reduction of oxygen. However, in anaerobic respiration, nitrogen molecules, H 2 S or other reduced compounds are produced in addition to water, depending on the terminal electron acceptor. The overall reactions for respiratory metabolism of carbohydrate (glucose) in the presence of oxygen is shown in Equation 4. C6 H 1 2 0 6 + 602 =—> 6 C02 + 6H 2 0 AQ 0' = -686 kcal/mole (4) In the absence of an external electron acceptor, balanced oxidation-reduction reactions of organic compounds, with the release of energy, occurs in many microorganisms. In this fermentation process, the organic molecules act as the electron acceptor, as well as the electron donor. Since only partial oxidation of the organic carbon compounds occurs in the fermentation process and the organic molecule must also serve as both the electron donor and electron acceptor, only a small amount of energy is released during the fermentation process. Unlike respiration, oxidative phospholylation for energy generation is not involved in fermentation and 14 the synthesis of ATP is largely restricted to the amount formed during glycolysis (Atlas, 1986). The fermentation of organic compounds begins with glycolysis and terminates with the formation of end products. For instance, a carbohydrate is oxidized into two pyruvate in the glycolytic pathway, generating two N A D H and two ATP per mole of glucose, which goes through glycolysis. Then, the oxidized pyruvates are converted to end products such as ethanol, lactic acid, propionic acid and butyric acid, depending on the species of microorganisms involved in the fermentation. It is expected that mixed acid fermentation is relatively common in wastewater treatment processes which include anaerobic condition. The mixed acid fermentation is carried out by Enterobacteriaceae including Escherichia coli. In this mixed acid fermentation, the pyruvate formed during glycolysis is converted to a variety of products, including ethanol, acetate, formate, hydrogen and carbon dioxide (Atlas, 1986). 2.3.3. Application of ORP in wastewater treatment system Oxidation-reduction potential has been applied for the monitoring or control of various processes in biological nutrient removal systems. The earliest ORP measurement, in connection with wastewater treatment, was made in 1906 and interest in ORP monitoring of wastewater treatment processes flourished in the 1940's (Koch and Oldham, 1985). However, this interest seldom provoked any intensive research, due to the development of the dissolved oxygen probe in a commercial form, the difficulty in obtaining reliable ORP measurements and the possibility of ambiguous interpretations (Koch and Oldham, 1985; Wareham et al, 1993). Since the incorporation of anaerobic and anoxic as well as aerobic zones into biological nutrient removal processes, a tool to monitor the various biochemical processes in the unaerated zone has been needed. This need provided the real impetus for the development of ORP technology and thus, ORP has been the subject of intensive study by a number of researchers (Charpentier et al, 1987 and 1989; Heduit and Thevenot, 1989; Jenkins and Mavinic, 1989; Koch 15 and Oldham, 1985; Koch et al, 1988; Peddie et al, 1988 and 1990; Watanabe et al, 1985). Through these researchers, it has been recognized that ORP can be used as an process control or monitoring parameter. ORP, as a process monitoring and control parameter, has been known to be somewhat specific to the system in which it is measured (Peddie et al., 1990). The absolute values of ORP in the same system can be affected by the factors such as: chemical species, biological activity, pH, and temperature. A l l of these factors may also vary within a given system over time. Thus, wastewater control strategies based on an ORP-time profile, rather than based on absolute ORP values, have been recognized to be more strongly correlated to biologically-mediated reactions (Wareham et al., 1994). ORP has also been used for the purpose of controlling and monitoring of processes in recent time. As a monitoring tool in biological treatment systems, ORP has been applied extensively to fermentation processes, in which the measurement of dissolved oxygen by a commercial DO probe is usually impossible. Many micro-aerobic fermentation processes take place at a concentration of dissolved oxygen which is below detection limits of the oxygen probe (Kjaergaard, 1977). It was also found that ORP varies linearly with the log of oxygen concentration, indicating that ORP is a sensitive parameter at very low oxygen levels (Peddie, 1990). Also, fermentation processes were found to be optimized at a certain range of ORP level. For example, methane production was found to be optimized at ORP levels between -500 and -520 mV (based on a silver-silver chloride), although gas production was observed to occur over a wide range (-495 to -555 mV) (Koch and Oldham, 1985). In research by Radiai et al (1984), the specific ORP values also optimized the production rates of specific amino acids, such as homoserine, valine and lysine. 16 In wastewater treatment processes, ORP has been used for the optimization of oxygen usage and to control the loading rate of external carbon sources. Watanabe et al (1985) used a designated ORP setpoint to control the addition of an external carbon source for complete denitrification. As the carbon source was used for denitrification by bacteria, the ORP value rose above the setpoint and addition of the carbon source (methanol) was initiated. Also, ORP has been used as a indicator for nitrification and denitrification. Sekine et al (1985) established a linear relationship between nitrification and ORP, and proposed using ORP as a control parameter for nitrification and oxygen demand (Peddie et al., 1990). In this research, a circuit converted the ORP values into a nitrification rate and made a correction to the DO level to optimize nitrification. The correlation between oxygen, nitrate, and ORP was also established by Koch and Oldham in 1985. The observed breakpoint in the ORP curve (knee) under anaerobic conditions was quite clearly associated with complete denitrification, meaning the end point of anaerobic respiration. In 1989, tests were carried out, both in the laboratory and in full-scale treatment plants, to define the relationship between ORP and oxygen concentration in activated sludge reactors (Heduit et al., 1989). The results demonstrated the importance of dissolved oxygen concentration in the mechanisms which determine the metal electrode potentials in activated sludge and the role of other electroactive species in the process. Also, they illustrated that the type and concentration of these species depended on parameters such as the sludge loading, the overall oxygen supply, the aeration sequence and the sludge concentration. Charpentier et al., (1987) investigated strategies to optimize nutrient removal and energy costs, using ORP regulation. Variations in ORP were correlated with the effluent ammonia-N and nitrate concentration in a low loaded activated sludge process. Subsequently, two ORP setpoints (-80 to + 120 mV) were targeted and the aeration cycle was controlled by these ORP 17 setpoints. In this study, a constant quality of effluent was obtained, determining the electricity consumption for aeration. Also, Chapentier et al (1989) furthered this investigation, and proposed that the targeting of upper and lower ORP setpoints in the aeration cycle simultaneously optimized the effluent quality and electricity consumption (saving 10 to 20 %). Similar research was carried out by Nakanish, et al (1990). Aeration into a complete-mixed, single reactor was controlled by the On-Off control with ORP as an operational parameter, while maintaining the pH with a pH controller. In their study, when the aeration was kept between 350 and 450 mV in sewage treatment, the nitrification and denitrification were optimized. Also, Moriyama et al (1990) reported that simultaneous nitrification and denitrification can occur by maintaining a proper ORP level (125 mV), using aeration control coupled with ORP measurement. Recently, Lo et al., (1994) discussed the effects of various ORP setpoints on the performance of lab-scale, extended aeration treatment systems. In their research, ORP controlled aeration had no significant impacts on the carbonaceous removal of the system. However, excellent nitrogen and phosphorus removal were achieved under optimum ORP setpoint of 70 -180 mV. They concluded that enhanced biological nutrient removal can be achieved in conventional extended-aeration treatment systems by the incorporation of ORP controlled aeration, which the optimum ORP setpoint for total nitrogen removal being 110 mV. Peddie et al., (1990) described distinctive ORP profiles within aerobic sludge digesters undergoing aerated and nonaerated conditions. Observations were made with automated monitoring of reactor ORP and dissolved oxygen. In their study, they found that the ORP-time profile had a number of distinctive features, directly related to changes in system chemistry and biological activity. They illustrated the ORP-time profile with 5 distinctive features. The ranges of aerobic and anoxic respiration, as well as fermentation, were clearly defined by slope changes on the ORP profile. However, attention was directed toward two predominant features: the 18 dissolved oxygen breakpoint and the nitrate breakpoint. Through this investigation, they confirmed that the ORP is related to low levels of dissolved oxygen and nitrate; also, they concluded that the reproducibility of the ORP-time profile and its sensitivity to changes in biological or chemical activity appeared to make it an ideal parameter for automated monitoring and process control. Real-time control, using specific ORP features, were also conducted recently (Wareham et al., 1993, 1994; Saune et al., 1996; Sasaki et al., 1996). In experiments by Wareham (1993 and 1994), the sludge digestion reactors were operated in sequencing batch reactor mode (aerobic and anoxic). 3 hours of aeration time was provided and the air-off time was determined by computer detection of the nitrate breakpoint on the ORP-time profile. The performance of the real-time reactor was compared to the control reactor (3 hour-aerobic and 3 hour-anoxic). The results of this research showed that the real-time reactor had better nitrogen removal than the control reactor; as well, the real-time reactor could more readily accommodate the stress. In 1996, a control strategy for an anaerobic-aerobic activated sludge process was developed and tested in a pilot plant by Sasaki et al (1996). They used the nitrate break point as the control index to adjust the aerobic, anoxic and anaerobic periods. The DO during the aerobic period was controlled at 2.5 mg/L and the ratio of the anaerobic period was set at 25 % of the total time. After aeration was performed for a predetermined period, the time when the nitrate break point appeared on the ORP curve was calculated and used as a predetermined value for the next cycle. Based on the results, using this control manner, a high degree of nutrient removal was achieved and the nitrogen removal efficiency was much better than the A 2 0 process operated simultaneously. Another real-time control for the removal of nitrogen in single SBR was conducted by Saune et al. (1996). In their study, aeration was terminated at the DO break point on the ORP curve and initiated at the nitrate break point, achieving nitrification and denitrification. However, 19 they did not describe the control technology, using ORP, for detecting the control points. Also, the ORP and DO-time profiles illustrated in the paper are ambiguous, in that the ORP profile is not well matched to the control strategy. 20 CHAPTER 3 EXPERIMENTAL METHOD AND ANALYTICAL TECHNIQUES In this research, a new type of treatment process, composed of two tanks, was designed for the treatment of the swine manure having high concentration of C, N and P. This newly developed process was named the " Two-stage Sequencing Batch Reactor (TSSBR) ". The TSSBR was operated with or without the specially invented Real-Time Process Control technologies (RTPC), to evaluate the usefulness of RTPC in the treatment of wastewater. Also, as a preliminary study, a four-stage process was designed to be operated in a batch-mode (FBM) and the effectiveness of this process for the treatment of swine manure was evaluated. With these designed two unit processes, various investigations were conducted to assess ORP as a control parameter. The operational conditions and strategies used for both of these designed systems are described in the following section. 3.1 Experimental design and setup for FBM 3.1.1 System Configuration and General Description The components of the batch reactor used for this study are shown in Figure 3.1. Five Plexiglas reactors were used, providing a total volume of 22 L (anoxic - 5 L, aerobic - 8 L, denitrification - 3 L, degas 3 L and clarifier 3 L). These separate reactors were designed to achieve carbon oxidation, nitrification and denitrification. Raw swine wastewater was loaded into the first anoxic reactor, into which the sludge plus nitrate was recycled from the clarifier. The carbon present in the wastewater was used to denitrify the recycled nitrate; also, ammonia was assimilated into the bacteria cell in the first anoxic reactor. Reactions occurring in the aerobic reactor were nitrification and carbon oxidation. The nitrified mixed liquor from the aerobic reactor was then passed into the second anoxic reactor, where additional denitrification occurred, using both exogenous and endogenous carbon sources. The second aerobic zone, named the degas reactor, 21 further stripped entrained nitrogen gas produced by denitrification in the second anoxic reactor and nitrified any residual ammonia, prior to clarification. Two sampling ports were installed into each reactor, upper and bottom sides. Peristaltic pumps were used for batch-mode pumping the influent, effluent and mixed liquor to the next reactor. The pumping rates were controlled by a speed controller. The components of the experimental apparatus are described in section 3.2.3. Electronical mixers were installed into each reactor to achieve complete mixing conditions. Air was introduced into the aerobic reactor via an air-stone placed at the bottom of the reactor. A commercial aerator was used and the air flow rate was adjusted to 1 litre per minute. For aeration of the degas reactor, a very low air flow rate was used in the operation. To study the correlation of ORP with loading rate, three oxidation and reduction potential (ORP) probes (Broadley-James Co., Ag/AgCl) were inserted into the aerobic reactor, to measure the ORP values; one DO probe was also inserted into the aerobic reactor. The DO probe was connected to the DO controller, built for this study, and also connected to a personal computer for data acquisition. It should be noted that the probes (DO and ORP) shown in Figure 3.1 were inserted into the FBM to study the correlation of ORP with the influent loading rate, when the process was controlled by the DO as a control parameter. A PC-LabDAS software package (Advantech.Co. Ltd) with amplifier and multiplexer boards was used for the monitoring and collection of ORP and DO data. The operation strategy of F B M is described in section of 3.1.4. In this section, the control method with a DO controller is discussed in detail. 3.1.2. Feed Swine Wastewater and Sludge The swine wastewater that was used for the FBM was obtained from a farm in Aldergrove, B. C. Canada. The wastewater collection and pretreatment methods are described in detail in section 3.2.3.1. 23 Activated sludge from a sequencing batch reactor, operated with swine manure influent, managed by the department of Chemical and Bio-Resource Engineering of University of British Columbia, was seeded into the FBM. The concentration of MLSS of the seeded sludge was approximately 11,000 mg/L. The first anoxic reactor was seeded with 4 litres of sludge, and then filled with influent wastewater and operated under designed operational conditions. For sludge acclimation, a five week period was provided, before taking samples from FBM. 3.1.3 Operation Strategy of F B M To evaluate the effectiveness of the F B M for the treatment of swine manure, two different SRTs were assessed. This process was controlled on the basis of both the solids retention time (SRT) and the hydraulic retention time (HRT). Based on the total volume of the unit process, a mid range 17 days SRT was used for the operation of FBM#1 and a long SRT (47 days) was used for FBM#2 operation. The long SRT in FBM#2 was adopted to achieve more complete nitrification in the aerobic reactor by increasing the aerobic SRT from 3.5 days to 22.7 days. The extended HRT and SRT values are typical in the biological treatment of swine wastewater (Bortone, 1992). The SRT and HRT were calculated according to the equation below. HRT = Total volume of system Loaded Influent volume per day (5) SRT = Total MLVSS in system Wasted MLVSS per day (6) 24 Table 3.1. Operational parameters used in F B M Parameters FBM#1 FBM#2 Reactor volume 22 L 22 L A V G HRT 20.3 d 19.4 d A V G SRT 17.2 d 47 d A V G SRTa* 3.5 d 22.7 d MLSSa* 2321 mg/1 3540 mg/1 *a; based on aerobic reactor The primary operating parameters are shown in Table 3.1. The process was operated in a "batch-mode". After effluent decanting from the clarifier, the mixed liquor in each reactor was transferred into the next reactor in sequence and finally the influent was loaded into the first anoxic reactor. The operation mode was controlled with timed cycles. When the mixed liquor in the aerobic reactor was transferred to the denitrification reactor, 2 ml of methanol as a carbon source was added into the denitrification reactor to enhance the denitrification, yielding a average COD to NOx-N ratio of 4.8:1. Internal recycling was not conducted and the sludge recycle ratio from the clarifier was 0.2 in FBM#1 and 0.25 in FBM#2. 3.1.4 Operational method for the study of the correlation of loading rate with ORP For this study, the computer system was set-up to monitor the DO and the ORP levels in the aerobic reactor continuously. A DO level of 2.5 mg/L in the aerobic reactor was used as a set point for process control. When the DO level reached 2.5 mg/L, the feed pump was turned on for a period of 15 minutes and the ORP was continuously monitored to identify the minimum ORP value during the loading of the aerobic reactor (Figure 3.2). The monitored minimum ORP values were plotted with the loaded nutrient amount (mg) to assess the correlation. When changes in feed volume were desired, the pump speed was adjusted accordingly, thereby maintaining a constant period of 25 feeding. The ORP values, which corresponded to the designated DO concentration of 2.5 mg/L, were in the range of 297 to 312 mV. During this study, various volumes of swine wastewater with somewhat different strengths were fed into the aerobic reactor (0.57 to 2.11 L / loading). Time Figure 3.2. ORP and DO pattern in aerobic reactor 3.2 Experimental Design and Set-Up for TSSBR 3.2.1 System Configuration A major schematic diagram of the experimental system is shown in Figure 3.3. The reactors were made of Plexiglas. The working volume of these reactors was flexible, depending on the control strategies. The minimum and maximum working volume throughout this study were 9.57 L and 7.78 L (A/O reactor), and 5.07 and 3.28 L (Anoxic reactor), respectively. Spigots for D O probe O R P probe Influent pH prope \ A/O (Anoxic /Oxic) AID converter Board Ampli f ier Effluent Anoxic VT sludge add Digital I/O, ON/OFF control board Relay Box connect to pump, aerator, mixer ere. Final Effluent ^ sludge waste Figure 3.3. Schematic of the TSSBR with the Real-Time control system 27 decanting were placed at 7.5 L in the A/O reactor and 3.0 L in the Anoxic reactor. Wasting of sludge was done manually or automatically, depending on the experimental strategies. Mechanical agitators were installed in both reactors. For complete mixing of the reactor and to avoid electrical interference which affects on ORP reading, a plastic shaft with an appropriate blade was designed. Air for the A/O reactor (1.5 L/min) was provided by an aerator through an air-stone placed at the bottom of reactor. Three ORP probes were inserted into the A/O reactor and one probe into the Anoxic reactor. These ORP probes (Broadley-James Corporation) have Ag/AgCl as a reference electrode and a platinum band as a noble metal. These probes were placed at strategic points in the reactors. The four probes were labeled 0, 1,2, and 3, to check the conditions of the probes at a regular interval and to keep them in a designated place. To monitor the pH trend, pH probes were inserted into both reactors. Also, a dissolved oxygen probe was inserted into the A/O reactor to monitor the changes of DO in the reactor. These DO and pH meters were connected to a junction box, which relayed the signal through an electronic cable into the computer. The particular models of the experimental components are summarized in Table 3.2. 3.2.2 Set-Up of Electronic Devices Usually, the ORP probes generated a low level voltage electrical signal in the range of -400 to + 300 mV in the experimental conditions of this study. Because of the physical characteristics of the ORP probes, they were connected to a custom-built amplifier for the accurate measurement of the voltage. As presented in Figure 3.3, the amplifier output was connected to a junction box. The electrical signals obtained were then relayed into the computer through an electronic cable ribbon. An analog to digital (A/D) card installed in computer converted the signals into binary code which could be processed by the computer. The computer and all other power cords were plugged into a 28 commercial uninterruptible power supplier (UPS) to avoid control failures by power shutdown. A commercial I/O control card was used for the automatic process control. The I/O card was installed into an expansion slot in the motherboard of the computer. The output lines were connected to a solid state relay box, built for this study. The relay box had five solid state relays and five outlets so that each outlet could be controlled independently by a single solid state relay. The pumps, aerator, and agitators were plugged into each designated outlet. A commercial software (LabTECH Control, LabTech. Ltd.) was used for data collection and process control. Process control methods were programmed and the control, originating from software switching bits (1 = ON, 0 = OFF), was achieved. The programmed control methods are described in Section 3.2.5.1. 3.2.3 System Operating Procedures for TSSBR 3.2.3.1 Feed Swine Wastewater and Sludge The swine wastewater used during this study was obtained from a farm in Aldergrove, B.C. The fresh swine manure was collected directly from a pig raising pen. The collected swine manure in 25 L containers was stored in the laboratory cooler at 4°C until required. Prior to utilization, the swine manure was diluted with tap water and screened by a sieve with 0.5 mm mesh opening, to remove larger solids before feeding into the influent container. Since swine manure contains a high proportion of suspended solids that resist biological degradation, the removal of these solids was performed to enhance the rate of biodegradation (Manning and Irvine., 1985). Also, the advantages of separating the suspended solids, before treatment, meant easier handling and a reduced oxygen demand (Burtone, 1992). The strength of the influent was not adjusted during this research. Because of the variability of the swine manure during the lab-scale experiment and also since most wastewater treatment plants are faced with varying changes in wastewater quality, various influent strengths were used during the experiments. 29 a o s-w Ifl ii -E m =3 O H O H < 13 S-i 6 o S c e o a £ o U o a O L H i >0 i O i r<1 • oo I o 6 o S-l cd O , IU Is* : O i o IO I o ; o ! ^° i =tt : -a " o io m a o . s-H—» >. i C X o . o : CL, O C a o H-» >, on H OH . X ! CN I o : IO I ° i n-6 i O a *H—» c i O :H l < i o i CN 13 -t—» '5b ! T 3 ' O . O i O i m ; co i oo ! H I-19 i *o i oo ICQ i H ; oo ; C N IS i O • C N i CN i m S H H—» o X O u H-» io I -o o . >-< i OH io O tH I O H O H i- H—» O H O H s o !U !T3 • & IU C o IU .t i O O H > Q i—1 U &H T3 S3 u 13 H-» '5b 5 o so o 13 c < I oo i OH i ^ c o o 1-o o O H o 'C a < " O H I tZ3 1-o o : OH . >' !0H i m N ; o i o ; m ; IT) : IT) 13 : T3 ' O X 00 cd i ON "o o i O i-d D 0» O H i CO o. a ! O H The seed sludge was taken from a pilot-scale, biological nutrient removal sewage treatment plant, managed by the Department of Civil Engineering at the University of British Columbia (located at the UBC south campus). 3 L and 1 L of sludge were added into the A/O reactor and the Anoxic reactor, respectively. The MLSS concentration of the added sludge was approximately 12,000 mg/1. The reactors were then filled with very dilute swine wastewater and operated under a normal sequence with a time-controller. Each day, the strength of influent was increased. After three weeks, the time-controller was replaced by the constructed automatic process control system. Prior to startup, the ORP pattern in the system was monitored carefully, to ensure an optimal condition for proper control. Sampling for analysis was taken from the system, when the system showed a constant real-time operation and steady state conditions. 3.2.4 General Description of TSSBR with Real-Time Process Control (RTPC) As previously described, the objectives of this study were to develop an effective swine wastewater treatment process and to establish real-time control strategies using ORP. To achieve combined carbon, nitrogen, and phosphorus removal, two series sequencing batch reactors were designed, and various stages for C, N , and P removal were provided. As shown in Figure 3.3, this system was composed of two reactors and each reactor had specific operation modes. The operation cycles used are shown in Figure 3.4 (a, b). Usually, the sludge in each reactor was completely separated, except during the operation strategy # 2. The sludge in the A/O reactor was added into the Anoxic reactor in operation # 2 (TSSBR # 2) to achieve enhanced denitrification without an organic carbon source addition. The excess sludge wastage was done on a weekly basis during the operation of TSSBR # 1. However, in the operation of TSSBR # 2, 3, and 4, a certain volume of settled sludge was wasted every cycle, mainly from the A/ O reactor and the volume was recorded on a daily basis. The wasted sludge from both reactors was collected in a sludge storage tank to calculate the average concentration of 31 1 Feeding Anoxic phase Final effluent jpscharge b) operation cycle of anoxic reactor Draw - — Sludge settle Figure 3.4. Schematic of the operation cycle of T S S B R 32 wasted MLVSS/d. The MLVSS concentration in both reactors was analyzed daily for the calculation of SRT. The solid retention time (SRT) was calculated using Equation 6. It should be noted that the periods of cycle and influent loading rate were determined by the computer in real-time operation. Therefore, the hydraulic retention time (HRT) and the reactor volumes were not constant. A equation for the calculation of the average HRT for the TSSBR system was developed and calculated according Equation 7. HRT = [(Vol of loaded influent /day / cycle No per day) + 10.5 L] Loaded influent vol. per day (7) Various process control strategies, using ORP, were applied to achieve a Real-Time operation in this research. The rationale for adopting ORP control strategies was to always ensure a ready supply of highly efficient electron acceptors to bacteria, thereby achieving better treatment efficiencies (Wareham, 1992) In this research, chemicals such as methanol, acetate or carbonate, which have commonly been used for enhanced nitrogen and phosphorus removal, were not used at all, since one of the objectives of this study was to develop a wastewater treatment process which could achieve complete nutrient removal without an supplemental addition of a carbon source. 3.2.4.1 A/O (Anoxic/Oxic) Reactor As described in Figure 3.4 (a), this reactor had five sequences (influent feed -> anoxic phase -> aerobic phase -> settle -> effluent decant to second Anoxic reactor). These sequences were designed to achieve complete removal of carbon, nitrogen and phosphorus in swine wastewater. The main reactions that occurred in this reactor were carbon oxidation and assimilation, nitrification, denitrification, and phosphorus release and uptake. The lengths of each sequence were 33 controlled by the ORP control system or fixed, depending on the control strategies. The duration of influent feeding and aerobic phases were automatically controlled by the computer, while the length of sludge settling and effluent decanting time were fixed. Based on the Air-ON/OFF, the sequences consisted of a 3 hour air-OFF period but a variable length of time for Air-ON, contingent upon the detection of the control point by the computer. In detail, 15 and 5 minutes were provided for sludge settling and decanting of effluent, respectively, while 160 minutes were provided for both influent feeding and the anoxic reaction. During this study, the influent feedings were usually completed within 20 minutes (5 to 30 min), depending on the influent quality, and thus, the period of anoxic sequence was an average, 140 minutes (varied from 130 to 155 minutes). The control strategy for influent loading by the ORP control system is described in Section 3.2.5.2 in detail. In general, the influent pump was kept ON until the ORP values reached the ORP loading rate control set-points, and then the anoxic phase was maintained for the rest of the time (160 min. minus time period consumed for feeding). Then, the aerobic condition was maintained (air-ON) until the designated control point occurred. The control strategies by ORP are represented in detail in Section 3.2.5. At a designated control point, the air and mixer were turned OFF for sludge settling before both the final effluent decanting from the Anoxic reactor and the liquid transferring to the Anoxic reactor. Then, influent was fed again. 3.2.4.2 Anoxic Reactor This reactor was designed to achieve a relatively complete denitrification of nitrates without carbon source addition. This operating strategy was formulated to demonstrate that an economical wastewater treatment could be achieved by applying ORP as a control parameter into a designed Two-Stage SBR. The principal concept was that the required carbon sources for denitrification could be provided by the residual organic materials in the liquor transferred from the A/O reactor and endogenous respiration of bacteria. 34 When the computer recognized the Real-Time control point, the mixer was turned off for sludge settling and final effluent decanting. Both 10 minutes and 5 minutes were provided for the sludge settling and the final effluent decanting, respectively. 3.2.5 Control Strategies Using ORP On-Line Monitoring System As mentioned in the previous section, the primary objectives of this research were to develop an effective wastewater treatment process for the combined high C, N, and P removal, to establish Real-Time control technology using the ORP as control sensor and to establish a general method to achieve enhanced denitrification, using an internal carbon source. To achieve these objectives, two real-time process control strategies using ORP trends were designed and applied in the Two Stage Sequencing Batch Reactor system (TSSBR). 3.2.5.1 ORP Real-time Control Strategies A system was setup for the real-time process control as portrayed in Section 3.3. The average ORP values from the 3 ORP probes inserted in the A/O reactor was scanned at an interval of 1 minute and processed with a data processing method programmed for this study. The programmed control technology is shown in Chart 4 of Appendix B. Two Real-time control points were designated on the ORP-time profile and programmed to send the ON/OFF signal to a specific relay to be controlled when the computer recognized the designated control points (see Figure 3.5). The ORP curves in the reactor during operation have several specific features. These features of the ORP response to the changes within the system are illustrated in Figure 3.6. * Point A. This point is the rapid response of ORP to the oxygen provided by the start of aeration. The changes of the ORP curve are very high at this point and the rate of ORP change with aeration is dependent upon the aeration rate. A similar ORP response has been observed by Wareham et al., (1990) in their aerobic sludge digestion study. * Point B. The introduction of dissolved oxygen has been the primary influence on the ORP 35 Time (hrs) Figure 3.5. The selected Real-time control points on the ORP curve 36 Figure 3.6. Speci f ic features of the O R P curve 37 change to this point. A further increase in the ORP signal is now due to the ongoing bio-chemical reactions iri wastewater (i.e. organic carbon oxidation, nitrification, and phosphorus uptake). *Point C. This represents the Ammonia-Nitrogen Break Point, which means the end of nitrification in the system. This NBP corresponded to the dissolved oxygen break point and results in another abrupt change in the ORP. Aeration can be terminated at this point when the targeted nutrient is ammonia-nitrogen. *Point D. Further organic carbon oxidation will take place if the wastewater has a high organic carbon content or there is any residual organic carbon in the system. The ORP curve has a relatively constant slope change at this point. Aeration can be terminated at any time past this point, dependent upon wastewater characteristics and control strategy. Any D x point on the ORP profile can be used for the control of the carbon oxidation state, depending on the wastewater characteristics. *Point E is the plateau of the ORP values which means the system has reached a fully oxidized state. Continued aeration past this point is not required under any control strategy. It is not recommended that the aeration phase be allowed to extend to this point, since there will not be any carbon left in solution for denitrification. In this figure, the E point is not clear. This point is also marked as the complete COD oxidation point in Figure 3.5 and has a ORP change of nearly zero. *Point a. The ORP inflection point is due to the stoppage of aeration. This point is mainly influenced by the decrease of dissolved oxygen. The dissolved oxygen in water is consumed by bacteria in the reactor and results in a decrease of DO. *Point F. Dissolved oxygen concentration in the reactor reaches 0 mg/L and denitrification begins. *Point G. A "nitrate knee" that indicates the completion of denitrification. This point can be used for control of denitrification in an anoxic reactor. These interesting features of the ORP curve were used for real-time process control. The C 38 point and D points, which have an abrupt and a constant slope change on the ORP-time profile, were designated as an ORP real-time control point and named as Nitrogen Break Point (NBP) and Residual Carbon Manipulation Point (RCMP), respectively (see Figure 3.5). As described before, any D x point can be used for a real-time process control point (RCMP), depending on the wastewater quality. The RCMP is not a complete carbon break point, but a small amount of organic carbon can be left in solution for denitrification by designating this point as a RTPC, resulting in the complete removal of residual organic carbon during denitrification in the anoxic reactor. This D x point must be selected based on the influent and effluent quality. These selected real-time process control points are shown in Figure 3.5. In this graph, the aeration was continued until the ORP slope change reached nearly zero, to show the complete organic oxidation point (not real-time operation). However, in real-time operation, the aeration must be turned OFF at a designated control point (NBP or RCMP), before the occurrence of the complete COD oxidation point. By selecting these points as a RTPC point, unnecessary extended aeration can be prevented, hence it could result in the optimization of electrical and chemical costs, as well as the optimization of process operation and treatment capacity. The NBP always corresponds to the dissolved oxygen break point, resulting in rapid change of ORP, while the RCMP begins to have a constant ORP change. To prevent erroneous process control, the control strategy was programmed to recognize each feature step by step, until occurrence of the final designated control point. In the process control strategy with the NBP control point, point C was finally recognized by the computer after recognizing the A and B points in order. In a similar way, A, B, and C points were recognized in sequence and process control was done at the onset of D point (RCMP) in another real-time control strategy. 39 In either full- or lab-scale wastewater treatment systems, enormous sources of electrical noise exist. The complete removal of this electrical noise is very difficult to achieve, since most wastewater treatment plants use various electrical device such as aerators, pumps and agitators. The sudden electrical noise problems may result in erroneous process control when the wastewater treatment system is controlled by the reading of dORP/dt. Also, since the ORP reading is usually affected by the change in pH in the wastewater, due to the change in alkalinity in the reverse direction, the abrupt changes in alkalinity may be erroneously recognized as certain control features, if dORP/dt is used for real-time process control. For instance, in the process control with RCMP, A, B, and C should be recognized first, in order, before the recognition of the final D point as mentioned above. If each control feature is designed to be recognized with dORP/dt, under conditions of electrical noise, the positive and negative dORP/dt values which occur repeatedly may be recognized as the A and B or C and D points occur. To avoid these problems, the Moving Range of the ORP was used for the process control in this research. If there is any ORP change at B or D points (whether it is negative or positive) caused by electrical noise or unusual abrupt alkalinity change in the system, a high moving range value should result, since the moving range always gives a positive value. In this way, the ORP change due to noise might not be recognized as a control feature by the computer. Usually electrical noise seldom occurs at points A and C, since the ORP is heavily affected by the introduction of air and the dissolved oxygen changes in the system. Also, since the duration of A and C points is usually short in real wastewater treatment systems, the possibility of erroneous control, due to electrical noise, can be minimized by using the ORP moving range pattern. The ORP moving range pattern in a wastewater treatment plant is shown in Figure 3.7. As found in this graph, several control points on the ORP curve can be easily detected by using an ORP moving range pattern. The selected RTPC point was the RCMP in this graph. 40 t.u;iu'(AUJ) aBuey 6U;AO|/\| 0) E a: m co To o. a> c CO c > o E a. O CD co CD (A"J) d d O An ORP moving range is the difference between the greatest and least ORP values seen in the last r points. Another advantage of using the moving range is that it produces r charts for statistical process control. In this research, the ORP values were monitored at an interval of one minute and r = 7 was used for the calculation of the moving range. 3.2.5.2 Loading rate control The objectives of influent loading rate control, using ORP set-points in this research, was to first, achieve a relatively constant influent loading rate into the wastewater treatment system, despite the variation of wastewater quality; secondly, it was designed to obtain a stable ORP-time profile by eliminating the disturbance of biological activity in the treatment system due to the serious fluctuation in loading rates. Since the ORP values monitored in a specific wastewater treatment system are heavily dependent upon the dissolved oxygen (DO) level, the DO level in the system should be maintained or controlled to achieve an effective control of influent loading rate, using ORP. To eliminate some of these factors, the influent feeding was designed to occur at a zero DO level. The zero DO level condition was achieved by continued bacterial respiration during the settling time provided for the A/O reactor. During the settling period, the residual DO was consumed by bacteria, resulting in non-measurable DO level with the DO probe. In this research, 4 ORP set-points were selected, and each ORP-set-point was used for the control of influent loading rate, depending on operating strategies. 3.2.6 Operating Strategies in the TSSBR System To evaluate the potential of ORP as a process control parameter and to establish an effective process control method using the ORP, various operational strategies were formulated. The study on TSSBR consisted of four operating strategies and each operating strategy included one to two runs. 42 Four ORP loading rate set-points were arbitrary selected to investigate the loading rate control by the ORP. The rationale for adopting these control strategies was: since the ORP is mainly a function of the ratio of oxidized materials to reduced materials, the concentration of MLSS, and the concentration of highly oxidizing matter such as oxygen, the influent loading rate could be controlled by the ORP reading if the unit system is operated under a condition of relatively constant MLSS concentration and if the influent feeding is done at a zero-DO condition. Theoretically, the volume of influent required to decrease the ORP value to a certain ORP set-point may depend upon the strength of influent, because influent feeding also means an increase in reduced material in solution. In this way, the HRTs can be adjusted automatically, depending on the quality of wastewater. Two ORP Real-Time Control Points were designated, through track-studies, and the usefulness of the developed real-time control methods, using a specific ORP point, was evaluated. The detailed explanations about the control strategies and the designated ORP control points are given in section 3.2.5.1 and the control or operating strategies applied in each operation are shown in Table 3.3. Before each run, a minimum of 2 SRT's were provided to let the system reach a steady state condition. Once steady state was reached, intensive study was initiated. A relatively short SRT was used during these experiments (9 to 11 days). Table 3.3. Operational strategies of the TSSBR TSSBR I TSSBR II TSSBR III TSSBR IV RTC * point RCMP ** R C M P NQP *** Not applied Sludge Separated Not separated Separated Separated Run No. RTC 1 & 2 RTC 3 RTC 4 & 5 * Real-time control, ** Residual carbon manipulation point, *** Nitrogen break point * * * * p j x e ( 1 t j m e 43 3.2.6.1 Operation 1, Two-Stage SBR#1 (TSSBR#1) The general operational modes of TSSBR were described in Section (3.2.4). As mentioned, the TSSBR were operated with various operating strategies to get more insight into the importance of control methods using ORP as process control parameters. In this first strategy, two ORP set-points were arbitrarily selected for the flexible influent loading rate control, based on wastewater quality; also, the Residual Carbon Manipulation Point (RCMP) on the ORP-time profile was designated for the real-time control of TSSBR. The evaluation study of this pre-determined ORP control strategy was performed through two runs (1 and 2 runs). In general, the sludge in each reactor was completely separated in this operational strategy. The control parameters of both runs are summarized below. RTC 1) The ORP set-point for influent loading rate control > -190 mV The designated ORP Real-Time control point >RCMP SRT > 10 days HRT > 4.6 days Sludge in two reactors > completely separated RCT 2) The ORP set-point for influent loading rate control > -240 mV The designated ORP Real-Time control point > RCMP SRT > 9.25 days HRT > 4.7 days Sludge in two reactors > completely separated 3.2.6.2 Operation 2, Two-Stage SBR#2(TSSBR#2) In this second operation, a RCMP (residual carbon manipulation point) was used for the real-time control of the TSSBR, as in operation strategy #1 (TSSBR#1). While the sludge in each reactor was completely separated in TSSBR#1, 50 ml of settled sludge in the A/O reactor was 4 4 added into the Anoxic reactor in this operation. In other words, 50 ml of the anoxic reactor's sludge was replaced with same volume of the A/O reactor's sludge at the time of liquor transferring into the anoxic reactor. This sludge seeding strategy was carried out to investigate the effects of sludge seeding on the denitrification efficiency, and thereby establish a method for enhanced denitrification, under a condition of non-available organic carbon source. Theoretically, effective denitrification can be achieved by the sludge seeding into the anoxic reactor for these reasons: a) an enhanced endogenous respiration can be obtained, since the nitrifiers added into the anoxic reactor can not survive for a long term in an anoxic condition, b) the stored carbon inside the seeded biomass (heterotrophic bacteria) can be used for denitrification, when a external carbon source is not available. For the control of influent loading, the same ORP set-point as in RTC 2 of TSSBR #1 was used to determine the effects of sludge addition on denitrification efficiency by comparing this operational strategy to RTC2. The general operational parameters used in TSSBR#2 are listed below; RCT 3) The ORP set-point for influent loading rate control > -240 mV The designated ORP Real-Time control point > RCMP SRT > 9.87 days HRT > 5.8 days Sludge in two reactors > sludge of A/O reactor was added into Anoxic reactor 3.2.6.3 Operation 3, Two-Stage SBR#3(TSSBR#3) For Real-Time control, a Nitrogen Break Point (NBP) was designated in the third operational strategy, while the Residual Carbon Break Point (RCMP) was applied in the first and second strategies (TSSBR#1 and #2). A Real-Time process control method, using the designated 4 5 NBP, was developed as mentioned in Section 3.2.5, and the usefulness of this technology was ( evaluated through two separate runs, RTC 4 and 5. As in the preceding first and second strategies, the influent loading was controlled by the ORP set-point applied for each run. Sludge seeding to the anoxic reactor was not carried out in this operational strategy, hence a complete separation of sludge was achieved during experiments. The control and maintenance parameters applied in each run are summarized as follows; R C T 4) The ORP set-point for influent loading rate control > -300mV The designated ORP Real-Time control point > NBP SRT > 10.2 days HRT >4.1 days Sludge in two reactors > completely separated R C T 5) The ORP set-point for influent loading rate control > -280mV The designated ORP Real-Time control point > NBP SRT > 9.4 days HRT > 4.06 days Sludge in two reactors > completely separated 3.2.6.4 Operation 4, Two-Stage SBR#4(TSSBR#4) It should be noted that the general operating strategies mentioned in Section 3.2 .4 are not applicable to this operation. In this operational strategy, ORP was not provided for the control of the process, but only for the monitoring of the process. In other words, the strategies for Real-Time process control and influent loading rate control, using ORP, were not carried out in this operation. In contrast to the first, second and third operation strategies, the TSSBR # 4 was operated in a fixed-time fashion. Also, a constant HRT was used in this operation via the loading of a fixed-volume of influent, despite the variation in influent quality. 46 The process sequence and each sequence period in the A/O reactor were exactly identical with other operations, except for the aerobic sequence period. In contrast to the ORP Real-Time operation (TSSBR#1, #2 and #3), which had flexible aerobic periods switched by the computer, a fixed time was provided for the aerobic sequence in this operation. For the aerobic period, the average aerobic period occurring in real-time fashion during the operation of TSSBR #1 (RTC1 and 2) was calculated and used for operational strategy #4. Based on air ON/OFF, the sequences were composed of 360 minute air-ON and 180 minute air-OFF. 10, 5, 5, 20 and 140 min were provided for sludge settling, the final effluent decanting from the anoxic reactor, liquid transferring from the A/O reactor to the Anoxic reactor, influent feeding and anoxic sequence of the A/O reactor, respectively. For control of the sequence in this operation, a commercially available control timer was used. F T 6) The ORP set-point for influent loading rate control > not applied 3.3. Analytical and Sampling Techniques The volume of the influent loaded into the reactor was measured daily for the determination of HRT. The influent and effluent wastewater samples were taken from the influent and effluent storage bucket, respectively. The handling and preservation time of the collected samples before analysis was kept to a minimum. Solid analysis was done at the experimental site immediately after sampling and the other chemical contents were measured within one week. System performance parameters routinely assayed included COD, BOD 5 , TOC, TKN, NH 4 -N, NO x -N, Ortho-phosphate (PO4"3), total phosphorus (TP), mixed liquor suspended solids (MLSS), mixed liquor volatile HRT The designated ORP Real-Time control point Sludge in two reactors SRT -> not applied -> 11.4 days •> 4.83 days -> completely separated 47 suspended solids (MLVSS), total solids (TS), total volatile solids (TVS), pH, alkalinity, and sludge settling volume index (SVI). Majority of the analysis was conducted in accordance with the Standard Methods, 16th Edition (A.P.H.A. et al., 1985). In the F B M operation, samples from the aerobic and anoxic reactors were taken, before and after the loading into aerobic reactor, to determine the correlation of ORP with loading rate. Also, samples were taken from each reactor to investigate the role of each reactor in overall treatment performance. These samples were centrifuged and analyses were then conducted on the supernatant. A number of tracking studies were performed in the two stage SBR with normal operation and in the batch reactor. The sludge used in the batch study was withdrawn from the reactor, and both MLSS and MLVSS measured before testing. In the tracking studies conducted for the determination of specific nitrification, denitrification and ammonia-N air-stripping rate, a nitrification inhibitor or chloroform were added into the batch reactor to suppress nitrification and the nitrogen assimilation by bacteria,. 3.3.1. Nitrogen Analysis 3.3.1.1 NO x-N and NH 4-N Samples for NO x -N and ammonia-N were filtered with filter paper (Whatman No. 4, pore size range of 20 - 25 um) prior to measurement and preserved by adding 3 % H2SO4 and phenyl mercuric acetate, respectively, to lower the pH to 3-4. The preserved samples were stored at 4 0 C and analyzed within one week. The NO x -N was analyzed with the calorimetric automated cadmium reduction method, using a Technicon Auto Analyzer II Continuous Flow Analytical System. The N H 4 - N was measured with the automated phenate method, using Technicon AutoAnalyzer II. 3.3.1.2 Total Kjeldahl Nitrogen (TKN) An appropriate volume of samples (1 ml influent, 20 ml effluent, and calculated volume of standard) with concentrated H2SO4 and K 2 S 0 4 were digested on a BD-40 block digester to liberate 48 all organically bound nitrogen. Then, the Total Kjeldahl Nitrogen was measured with the Technicon AutoAnalyzer. The percent N in the sludge (MLSS) was measured with exactly the same method. For the determination of nitrogen content in the MLSS, a 100 ml mixed liquor sample was taken from the each reactor, and the MLSS was measured immediately. Then, a portion of samples were centrifuged at 1500 rpm for 10 minutes. The N concentration of mixed liquor and supernatant was measured separately, in order to obtain the percent N in MLSS by deducting the N of supernatant from the N of mixed liquor. 3.3.2 Phosphorus Analysis Ortho-phosphate was analyzed on filtered samples with a Technicon AutoAnalyzer II, using the automated ascorbic acid reduction method. In this method, the ortho-phosphate was measured in the form of PO4"3. Samples for total Phosphorus (TP) and percent P of MLSS were prepared and analyzed in the same way as T K N . 3.3.3 Carbon Content Analysis In order to characterize the carbon content of influent and effluent / or evaluate the treatment efficiencies, COD, BOD5, and TOC were analyzed. 100 ml of sample was collected and used for carbon measurement. To measure soluble carbon content, the samples were filtered with filter paper (Whatman No.4), and then analyses were done. For COD analysis, the dichromate reflux method was used (APHA., 1995). The influent samples were diluted twenty times with distilled water for proper COD measurement, while no dilution was done in the effluent sample. Samples for BOD 5 analysis were diluted with seed water and incubated at 20 0 C for 5 days. A DO probe and meter (model No. 59, YSI Inc.) was used to measure the dissolved oxygen concentration in the BOD test bottles. Total Organic Carbon was analyzed with a Shimadzu Total Organic Carbon Analyzer (Model TOC-500) using a series of standards. Before the TOC measurement of influent, the solids 49 in the samples were allowed to settle down at room temperature for 20 minutes to prevent clogging of the syringe of the TOC Analyzer by large solids in the influent, while the effluent sample tubes were mixed completely. 3.3.4 Suspended and Total Solids Analysis Samples for suspended solids (SS) analysis were filtered through Whatman glass microfiber filter (Model No. 934-AH, 90 mm) and dried over night at 105 °C, before the measurement of suspended solids concentration. After measuring the suspended solids, samples were burned at 550 0 C for 30 minutes to measure the concentration of volatile suspended solid (VSS) by weighting the residue. 30 ml of influent and effluent samples was evaporated at 105 0 C for over night for the measurement of total solid (TS) and burned at 550 0 C, to determine the concentration of total volatile solid (TVS). 3.3.5 pH, Alkalinity, Dissolved Oxygen and ORP Measurement For the pH measurement of samples, a digital pH meter (201 ATC) with automatic temperature compensation was used throughout, the study. Before measurement, the pH meter was calibrated using a standard buffer solution (4.0, 7.0 and 10.0). Alkalinity was measured with an end point-pH method described in Standard Methods (APHA., 1985). Samples were titrated with 0.1 N H2SO4 acid to pH 4.5, recording the volume of titrant. A point-4 Oxyguard was used for the measurement of DO levels in the reactor and samples. The membranes of the DO probe were regularly changed to avoid improper reading of the DO. Prior to the ORP measurement, the ORP electrode's accuracy was tested with Quinhydrone because incorrect ORP potentials are occasionally measured and the cause of these errors are usually a contaminated platinum (Pt) surface or contaminated internal reference solution. The pH buffer solutions (pH = 4 and pH = 7) were saturated with Quinhydrone (0.2g /100 ml of pH buffer solution) and each probe was immersed into the solutions to test the reading accuracy. The ideal 50 values of the ORP electrode (Ag/AgCl reference) in pH 7 and 4 buffer solutions saturated with Quinhydrone are + 86 mV and + 263 mV, respectively. The electrode responses were in the range of 10 to 20 mV difference in all cases. During this research, the ORP probes were cleaned with tap water on a regular basis, to prohibit the build up of bacteria on the platinum surface. 3.3.6 Sludge Volume Index (SVI) In order to monitor the sludge settling characteristics in real-time operation of TSSBR, 1.1 litre of mixed liquor was withdrawn from each reactor. 1 litre of this well-mixed liquor was put into a conical cylinder to measure the volume of settled sludge after 30 minutes settling time. For the sludge Volume Index (SVI ) determination, the suspend solids concentration of a well-mixed sample of the suspension was analyzed, and the volume in milliliters occupied by 1 g of a suspended solids was measured. For the calculation of SVI, Equation 8 was used. After measuring the settled sludge volume, the mixed liquor was returned to the reactor where it was taken from. SVI = Settled sludge volume (mL/L) x 1000 Suspended Solids (mg/L) (8) 51 C H A P T E R 4 Four stage-batch mode process (FBM) In this Chapter, results from preliminary research with a four-stage system (FBM) are described. The designed F B M was operated in a batch mode and two different solid retention times (SRT) were assessed, to evaluate the effectiveness of this process for the treatment of high strength wastewater (swine manure). The F B M was first operated at a SRT of 17.2 days (FBM # 1), then operated at a longer SRT (47 days) (FBM # 2). 4.1 Influent Wastewater Characteristics and Loading rate Table 4.1 summarizes the characteristics of wastewater used for the operation of F B M # 1 and # 2. The composition of wastewater, after screening, averaged 5,085 mg/L COD, 985 mg/L BOD 5 , 2018 mg/L T K N and 1254 mg/L NH4 -N in F B M # 1; and 3878 mg/L COD, 1079 mg/L BOD5, 1896 mg/L T K N and 1142 mg/L NH 4 -N in F B M # 2. The bio-degradability of carbon materials in the wastewater was higher in F B M # 2 than in F B M # 1, based on the relative ratio of BOD5 to CODt (1 : 5.16 and 1 : 3.59 in F B M # 1 and 2, respectively). Part of the higher CODt (total COD) and TKNt in F B M # 1 could be due to the higher solid concentration in the influent. The total and suspended solid concentrations were respectively 1.3 and 1.72 times higher in F B M # 1 than in F B M # 2, having 6.79 g/L - TS and 3.378 g/L - MLSS, and 5.243 g/L - TS and 1.967 g/L - MLSS, respectively. The COD and T K N variations, due to the suspended solids content, were calculated by deducting CODs (TKNs) from COD (CODt (TKNt); this produced 104.7 mg-TKN/g SS and 720 mg-COD/g SS. With regard to nitrogen, the influent T K N was composed of an average 68 percent NH4 -N and 32 percent soluble or insoluble organic nitrogen. Since the strength of the influent was not adjusted during this study, to simulate field conditions for wastewater treatment, the variation in influent quality (and influent loading rate) was quite large, as noted by the large standard deviation in Tables 4.1 (and 4.2). 52 Table 4.1. Characteristics of swine manure used for this study Parameters (mg/1) Means FBM#1 Min. - Max. St. Dev. Means FBM#2 Min. - Max. St. Dev. BOD5t** 985 636 - 1215 151 1079 309- 1886 391 BOD5s*** — 624 221 - 1469 246 CODt 5085 2090 - 5838 740 3878 1951-5161 729 CODs — 2462 1748 -3052 352 TKNt 2018 1432-2358 255 1896 1240 - 2498 295 TKNs — 1690 1128 -2413 262 NH4-N 1254 765 - 1480 168 1142 783 - 1499 166 NOx-N 3.18 0.5 - 5.65 1.51 1.48 0.9-2.1 0.30 TS (g/1) 6.79 5.947- 7.801 0.744 5.243 3.995-6.491 1.765 SS (g/1) 3.378 2.520-4.310 0.592 1.967 0.767-3.167 1.697 pH 9.06 8.38 - 9.44 0.32 9.16 8.19-9.77 0.46 Alkalinity* — 5591 4611-6525 530 **t; total ***s; soluble *alkalinity as CaC03 Table 4.2. Loading rate variation during this study Loading FBM#1 FBM#2 St.Dev. rate Mean Min. - Max. St. Dev. Mean Min. - Max. (mg/d) COD 4016 1672 -4691 640 3787 1403 - 6285 1445 BOD 5 761 470 - 960 154 1042 221 -2339 478 T K N 1272 767 - 1602 287 1841 773 - 3402 706 NH 4 -N 997 612-1242 143 1058 489 - 2385 405 53 As shown in Table 4.2, the variability of the loading rate was much larger in F B M # 2 than in F B M # 1. Based on the means of the COD and T K N loading rates, the standard deviations were 15.9 % and 22.6 % in F B M # 1, and 38.2 % and 38.3 % in F B M # 2. Due to this high variability, maintaining a constant hydraulic retention time (HRT) resulted in considerable fluctuations in the food to microorganism ratio (F/M), through changes in loading rate alone, thus resulting in variable treatment. No significant pH variation of the influent wastewater was observed during both operations; the high pH value was attributed to the high content of ammonia-N in the wastewater (Table 4.1). The correlation of alkalinity with NH 4 -N in wastewater is described in Section 4.3. 4.2 System Performance and Operational Characteristics As described in Chapter 3, phase I of the study (FBM # 1) had a system SRT of about 17 days, including a 3.5 days SRT in the aerobic reactor. Phase II (FBM # 2) had about a 47 day system SRT, with a 22.7 day SRT in the aerobic reactor. The long SRT in FBM#2 was adopted to achieve more complete nitrification in the aerobic reactor. With this long SRT, the mixed liquor suspended solids (MLSS) concentration in the aerobic reactor increased approximately 52 %, with respect to FBM#1. However, it appeared that the adopted sludge recycle ratio was low. Although a longer system SRT was achieved by wasting a low volume of sludge from the clarifier, the low recycle ratio resulted in not much increased MLSS levels in the reactor and an accumulation of excess sludge in the clarifier. Also, the actual SRT would be shorter than calculated, because of high solids release in the decant of effluent. As shown in Table 4.3, the average removal efficiencies of BOD5, COD, ammonia-N and T K N in FBM#1 were 92 %, 83.1%, 85.3% and 82.1%, respectively, with 85.2 %, 58.3%, 86% and 76.5%, respectively, in FBM#2. A comparison of the treatment results obtained in this study to the results published by Bortone (1992), Lo (1990) and Ng (1987), for SBR systems, used to 54 treat swine manure, showed both F B M had almost the same nitrogen removal efficiency as reported by Bortone (-88%), but much higher than the others. For organic carbon removal, in the other word, COD removal efficiencies were 48% with a short HRT (1.3 d) and low influent strength (COD = 2,194 mg/L) (Ng, 1987), 40% with low influent strength [Lo, 1990] and 93% with a long HRT (lOd), a long SRT (34d) and a high strength swine waste (COD = 10,580 mg/L) [Bortone, 1992], respectively. This suggests that the F B M system may be an effective treatment method for swine manure; however, further studies, using influents of consistent quality, are required to evaluate the effectiveness of the F B M system with respect to other types of bioreactors. Table 4.3. Effluent quality and removal efficiencies Parameters (mg/1) Means FBM#1 Range % Removal Means FBM#2 Range % Removal B O D 5 75 32 - 147 92.0 160 62 - 355 85.2 COD 858 625 - 1173 83.1 1617 962 - 2169 58.3 N H 4 - N 184 141-215 85.3 159.8 137-188 86.0 T K N 360 312-409 82.1 446.0 281 -631 76.5 TS(g/l) 3.288 2.115-4.215 51.6 4.0061 3.794-4.218 23.6 SS(g/l) 0.106 0.001-0.295 96.8 0.3492 0.210-0.488 82.2 NOx-N 17 0-91.6 1.93 0 - 11.6 Alkalinity 1624 1500-1775 71.1 In Phase II (FBM # 2), with the exception of ammonia-N, the removal efficiencies for all parameters were lower than observed during F B M # 1. The lower T K N removal efficiency observed may be due to a higher suspended solids concentration in the final effluent, since the effluent analysis was not conducted on filtered samples. The relatively lower COD removal efficiency in F B M # 2 is also attributed to the higher solids concentration in the effluent. During 55 7000 6000 5000 ^4000 §3000 2000 1000 •COD L.R. •NH4-N L.R. FBM # 1 2500 2000 1500 1000 500 ^ a - o < o c N c o ^ r o r - - c o a > i o * - I ^ -T ^ C O C O O ) 0 > O t - - t - C N C g c O T T T ! -T - ^ ^ T - T - C M C S I C v J C M C M C N C M r v J Days Figure 4.1. Influent variability 2500 2000 j1500 £ a O1000 500 500 -Effluent COD -Effluent NH4-N ^ J - O C £ ) C M 0 0 - ^ - O r - ~ C 0 0 5 L 0 T — r--N t O C O O ) 0 ) O T - r O J O j n \ t \ t T - T - T - T - T - f ^ C M C N C N J C N J C N C N J C N Days Figure 4.2. Effluent variability 56 F B M # 2 operation, excessive sludge accumulation occurred periodically and signs of fermentation were observed in the clarifier. As presented in Table 4.3, the concentration of suspended solids and total solids in the final effluent were 3.27 and 1.22 times higher, respectively, in F B M # 2 than in F B M # 1. In addition, the operational condition should affect the removal efficiencies in F B M # 2. As shown in Figure 4.1, the changes in loading rate were quite severe during F B M # 2. This daily fluctuation of influent quality might have resulted in a disturbance of bacterial activity in the system, as well as providing a shock-loading condition. Under the severe loading rate change condition, the nitrogen removal efficiency was relatively constant, while the organic carbon removal efficiency deteriorated with time (Figure 4.2). The role of each reactor in treatment performance was tested in the operation of FBM#2, as shown in Figure 4.3. Unfortunately, this test was not conducted in FBM#1. During this operation, samples were collected from each reactor every morning, ten minutes before influent feeding; the average concentrations for each parameter were used in Figure 4.3. Based on the soluble influent, the achieved average removal efficiencies of COD, NH 4 -N and T K N in FBM#2 were 34.4 %, 86 % and 74 %, respectively. The removal efficiencies of ammonia-N and T K N in the anoxic reactor were 23% and 20.2%, and 52.2% and 51% in the aerobic reactor, respectively. An additional 10.6% of ammonia-N and 2.4% of T K N were removed in the second anoxic condition (denitrification to final effluent). Nitrogen removal in the first (anoxic reactor) and second anoxic conditions is attributed to heterotrophic bacteria assimilation and some ammonia stripping. Also, sludge recycle from the clarifier to the anoxic reactor could have resulted in a lower nitrogen concentration in the anoxic reactor, through simple dilution effects. In contrast to nitrogen removal, little or no soluble COD was removed in the aerobic reactor; as such, anoxic conditions played an important role in any soluble COD removal. 57 The 34.4 % CODs removal in the anoxic conditions could be achieved by combined bacterial assimilation and denitrification requirements. The organic carbon removal in the second anoxic condition was much higher than that in the first anoxic reactor, obtaining 24.5% and 9.9% soluble COD removal, respectively. The lower organic carbon removal in the first anoxic reactor could be due to the lower denitrification requirement of bacteria, since the NOx-N levels in the influent and recycled liquor were low. On the other hand, the 24.5 % of organic carbon removal in the second anoxic condition could be attributed to the higher organic carbon requirement for the denitrification of high NOx-N levels. It is speculated that the organic carbon removal in the first reactor could be enhanced, if internal recycle from the aerobic reactor to the anoxic reactor was carried out. Also, the fixed amount of carbon source addition into the denitrification reactor, during the operation of FBM#2, sometimes resulted in deteriorated organic carbon removal. It should be noted that the COD and T K N removal efficiencies obtained in the second anoxic condition must be higher than calculated, since the effluent analysis was not conducted with filtered samples (effluent contained high solids level), while the analysis of samples from each reactor and influent was done with filtered samples. Throughout this study, the NO x -N concentration in the aerobic reactor was high (413 to 665 mg/L). This elevated NO x -N concentration could impact upon the treatment efficiencies in both operations, since a high nitrite concentration can be toxic to bacteria (Anthonisen, et al., 1976)). In spite of this high NO x -N level, relatively complete denitrification was achieved in both operations. The average NO x -N concentration in the effluent of FBM#1 and FBM#2 were about 17 mg/L and 2.0 mg/L, respectively. The wide concentration range of N O x - N in the FBM#1 effluent (0 to 91.6 mg/L) was attributed to improper mixing in the denitrification reactor, caused by occasional mixer failure during operation. 59 4.3 Correlation of Loading Rate with ORP The correlations of loading rate with ORP in aerobic reactor, under the operating condition of a DO set-point, are shown in Figures 4.4 and 4.5. The variation in influent quality fed to the aerobic reactor is also shown in Figure 4.6. The relative ratio of components was not constant, as follows; COD:BOD 5 , 3.86 to 11.2; COD:TKN, 1.2 to 2.8; and, TKN:NH 4 -N, 1.2 to 1.7. Over the course of the study, the pH level in the aerobic reactor was relatively constant, varying between 7.43 and 7.98. As shown in these graphs, the loading rate of carboneous and nitrogenous materials into the aerobic reactor might be controlled by monitoring ORP under a designated level of maximum DO, if the level of pH and MLSS in wastewater treatment processes can be maintained in a close range. Considering the correlation between carbon material and redox potential shown in Figure 4.4, it was found that COD had a higher correlation than BOD 5 with ORP. This can be explained by the fact that the ORP reading respond to the overall status of electron movement in the treatment system, whether the electron involved in bio-chemical reaction originates from biodegradable organic materials or not. Also, when considering the fact that the maximum ORP values (which corresponded to the designated DO level) were in a relatively constant range (297 to 312 mV) during this study, the nutrient loading rate might be controlled without a separate DO level control, by designating the upper ORP value, instead of the actual DO level. Specifically, it was found that the correlation between nitrogen loading rate and ORP was significant. As represented in Figure 4.7, the ORP values directly reflect changes in ammonia-N concentration after loading. This indicates that the concentration of nitrogenous materials in the aerobic reactor can be monitored indirectly by ORP readings. During loading into the extended aerobic reactor, the ORP value continuously decreased until feeding was completed, while the 60 1 5 0 1 5 0 0 -I 1 1 1 1 1 1 1 1 1 1 0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 COD load (mg) Figure 4.4. Organic carbon loading vs O R P values (aerobic reactor) 61 150 0 -I 1 1 1 1 1 1 1 1 1 1 0 200 400 600 800 1000 1200 1400 1600 1800 2000 Ammonia-N (mg) 150 0 -I 1 1 1 1 1 1 0 500 1000 1500 2000 2500 3000 TKN load (mg) Figure 4.5. Nitrogen loading vs O R P values (aerobic reactor) 62 30 25 20 + o o O) 15 e 6 c o ° 10 5 + 1 NOx-N • ammonia-N • BOD5 • COD HTKN 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Sampling times (d) Figure 4.6. Variation of wastewater fed into the aerobic reactor 600 500 4-400 4-Ammonia-N after loading ORP after loading 4- 200 100 + 250 H h H h H h 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Sampling times (d) Figure 4.7. Correlation of ammonia-N level with O R P values after loadi 63 DO level in the reactor quickly reached a base level at the start of feeding. This indicated that ORP was also a function of N-loading to the reactor, not just the concentration of DO. This can be explained by the relationship between the ammonia-N concentration and alkalinity, as well as the oxidation states within the reactor. As the wastewater was fed into the treatment system, the ratio of oxidized material to reduced materials decreased, resulting in a decrease of redox potential. Influent alkalinity was also dependent on the concentration of ammonia-N, as shown in Figure 4.8, and a reverse relationship can be established between ORP and alkalinity. Thus, loading of high concentrations of ammonia-N was accompanied by the addition of alkalinity into the reactor, resulting in a decrease in ORP. However, the alkalinity effects on ORP were considered to be a minor factor during the feeding time, compared to the effects of oxidation levels, since the ORP reading obtained during influent loading was mainly a function of oxidation/reduction status in the reactor, in spite of a small variation of pH. During the influent loading, the pH change in the reactor was within less than half a pH unit variation. Also, under a condition of declining pH, due to low pH-influent feeding, a fast decrease in redox potential was observed. However, to achieve successful process control or to avoid an erroneous control by a large pH variation in the system, the pH effects should be minimized by maintaining the pH levels within a certain range. The compensation of pH variation, into the ORP measurement, could be a useful means to track the oxidation/reduction state in the system, preventing inaccurate process control by large pH variations. In addition, the maintenance of a relatively constant MLSS level could provide better process control, since the ORP values are indirectly dependent upon this variable. In order to achieve constant effluent quality under varying influent conditions, certain design constraints must be identified and established. To achieve organic removal and nitrification control, parameters such as HRT and SRT are commonly used in wastewater 64 1800 --1600 --1400 1200 --~ 1000 --x z 800 - -600 400 -• 200 -• • Influent ammonia-N -Alkalinity H 1 1 1 1 1 1 1 1 1 H —I 1 1 1 1 1 1 1 1 1 r— 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Sample No. Figure 4.8. Correlation of ammonia-N cone, with alkalinity in swine wastewater 65 treatment. However, since the concentrations of organic carbon, nitrogen and suspended solids in the final effluent at any given HRT and SRT are directly a function of feed strength and/or loading rate (Hoeker and Schroeder, 1979; Daigger and Grady, 1977; Benefield, 1977), a process control strategy which maintains a constant SRT and HRT would be unable to produce a consistent effluent quality, when subjected to varying influent conditions. Therefore, to achieve constant effluent quality, despite variability in the influent quality, it is necessary to control the loading rate to any treatment system. The correlation between ORP and pollutant levels found in this study shows promise for the control of influent loading rate. The application of a loading rate control strategy, using ORP-set-points, could result in a flexible influent loading volume and a flexible HRT, as determined by the wastewater strength. Under this strategy, obtaining a constant effluent quality at a given SRT may be possible. Also, various process control strategies, with ORP-set-points, could be designed using these relationships. For instance, a semi-continuous aerobic process could be developed in both batch-mode and continuous flow-mode, in such a way that aeration is finished at the upper ORP-set-point and influent loading is kept down the lower ORP-set-point. Using this strategy, a flexible HRT (or aeration period) could be provided for the complete oxidation of carbon materials and nitrification, depending upon the wastewater characteristics. 66 CHAPTER 5 Two Stage Sequencing Batch Reactor Experiments Prior to the real-time control of TSSBR, the system was operated with a timed cycle and the ORP pattern was monitored carefully, in order to designate the proper control point on the ORP-time profile. Figure 5.1(a) represents the ORP, pH and DO patterns in the A/O reactor of the TSSBR. When the ORP slope change neared zero, the point at which the system reaches a fully oxidized state, aeration was shut down and influent was fed into the reactor. Alternative anoxic and aerobic conditions were then provided, and the nutrient trend and ORP pattern were tracked. Following the cessation of aeration and before the initiation of influent feeding, the DO in the A/O reactor reached a non-measurable level. The DO reduction at this point could have resulted from the consumption of the residual dissolved oxygen by the continuous respiration of bacteria. As the DO level decreased, ORP and pH also started to decline. The decline of pH may be attributed to the solubility of C 0 2 from the atmosphere, while the ORP decline could be due to the decreased DO concentration, since ORP is very sensitive to changes in aeration. As a cessation of CO2 air-stripping process, CO2 level in the liquid increase (Hao and Huang, 1996), resulting in the decline of pH. After influent feeding, both the ORP and pH decreased very rapidly, due to the sudden change of oxidation/reduction status in the reactor and the addition of the low-pH wastewater. During the anoxic phase, which followed influent loading, ORP and pH continued to decrease but at a slower pace. The NOx-N produced by the nitrification of ammonia-N in the aerobic condition was denitrified to nitrogen gas during this anoxic period and denitrification, using the influent organic material as a carbon source, was completed within a very short period (Figure 5.1 (b). After complete denitrification, phosphorus release from bacteria and hydrolysis of 67 r1 Time (hrs) Figure 5.1. Track of A / O reactor 68 particulated organic material into soluble form were observed. As air was re-introduced into the reactor, both the ORP and pH responded rapidly. The initial increase in pH at this point may be the result of air-stripping of free C 0 2 from the wastewater. With the provision of air, bio-chemical reactions such as nitrification, phosphorus uptake and organic carbon oxidation began to take place in the reactor, resulting in reduced concentrations of ammonia-N and phosphorus, as well as increased NOx-N concentrations. The trend here was for pH to decrease (after the initial increase of pH with aeration), probably due to the reduction of alkalinity during nitrification. A detailed description of the pH trend in real-time TSSBR is presented in Section 5.1.1. The DO level in the reactor showed no measurable change at this point, despite aeration. This result is attributed to the oxygen uptake rate by nitrification being equal to or greater than the aeration rate; this resulted in all the oxygen being consumed in nitrification as soon as it became available. This is also demonstrated by the reduction in ammonia-N concentrations and the increase of NOx-N in the reactor. At the point of complete nitrification of ammonia-N to nitrate, an abrupt change was observed in the ORP curve. This nitrogen break point (NBP) also corresponded to a DO break point. The jump of DO level at this point could have resulted from the lowered oxygen uptake rate (OUR) of the bacteria. Since no ammonia-N exists in the wastewater, the OUR of the nitrifiers is reduced; thus, the aeration rate becomes greater than the OUR of the bacteria, and DO levels increase in the reactor. A specific feature which corresponded to the DO and nitrogen break points on the ORP curve was also detected on the pH curve. Following the NBP, the only reaction which occurred was organic carbon oxidation. At this point, a relatively constant slope change began on the ORP curve. This point was named the residual carbon manipulation point (RCMP); small 69 amounts of organic carbon can be left in solution and removed completely during denitrification, which requires carbon as an energy source for heterotropic bacteria. The ORP curve eventually reached the ORP plateau, where the slope change is near zero. The plateau indicates that the system had reached a fully-oxidized condition, which means that there is no biodegradable organic carbon in the reactor. This point can be located in the results for soluble TOC shown in Figure 5.1(b). It is not recommended that aeration be extended to this plateau point, since there would not be sufficient carbon left in the reactor for proper denitrification to occur. In terms of the real-time control of TSSBR # 1, the RCMP was selected as the control point and an operating strategy using this point was then devised and evaluated. The ORP pattern in the anoxic reactor was also tracked and the results shown in Figure 5.2. Since the organic carbon oxidation was maintained until the ORP curve reached a plateau, no readily biodegradable organic carbon was left in the mixed liquor. The soluble organic carbon concentration (TOC) was therefore constant during denitrification. The required organic carbon for denitrification was provided by the endogenous respiration of bacteria. In this manner, approximately 3.5 mg/L of NOx-N was denitrified within 4.5 hours. No phosphorus uptake, using nitrate as the electron acceptor, was observed during denitrification. However, the ammonia-N concentrations in the reactor increased slightly within the range of 1 mg/L. This increase in ammonia-N, although negligible, may have resulted from the reduction of nitrate to ammonia-N. Furthermore, the endogenous respiration of bacteria could have resulted in the release of NH 4 -N into solution. Before the liquor transfer from the A/O reactor, the ORP trend exhibited very low values (lower than -300mV). This low ORP trend may indicate a highly reduced state in the reactor and anaerobic conditions. The feed of highly oxidized liquor to the anoxic reactor resulted in a sharp rise in ORP, due to the fast increase in the ratio of oxidized to reduced materials in the reactor. 70 Time(hrs) •A—Anoxic NH4-N X Anoxic NOx-N •e—Anoxic P04-3 —•— Anoxic TOC Figure 5.2. Track of Anox ic reactor 71 Upon mixing, the ORP dropped again. It should noted that the mixer was turned OFF for sludge settling, the final effluent discharge and the liquor transfer. The drop of ORP at this point can be explained by the effects of solids on the ORP reading, since there is a reverse relationship between ORP and solids concentration in solution. After the reactor was completely mixed, the ORP began to rise again and continued until the end of denitrification. As nitrate respiration continued, the nitrate concentrations declined, finally reaching the nitrate knee point, which indicating the end of denitrification. 5.1 Real-time control of TSSBR with Residual Carbon Manipulation Point (RCMP) The TSSBR was operated with the RCMP (D point) as a Real-Time Process Control (RTPC) point and the ORP and pH pattern was monitored. The moving range pattern of ORP in the operation of the real-time control of the TSSBR, with the RCMP, is shown in Figure 5.3. As indicated, the moving range pattern was uniform for every cycle; for example, very high - low -high - low. This moving range (MRG) pattern was used for the real-time control of the TSSBR, as noted previously (Chapter 3). The designated M R G values, in mV, for the detection of A, B, C and D points were M R G > 50, M R G < 5, M R G > 15 and M R G < 3, respectively. M R G is the difference between the greatest and least values seen in the last r points, and makes a statistical control process possible. The theory of moving range calculation with r = 7 is shown in Figure 5.4. The initial values are a running moving range of the points collected from the beginning of the run until the first r points have been collected. As the next data point is collected, the oldest point is dropped and the moving range of r points recalculated. This process of dropping the oldest point when a new data point is collected, followed by the recalculation of the moving range value, continues for the duration. In this operation, the sampling rate and the parameters (r) were 60 sec and 7, respectively. 72 Figure 5.3. The pattern of O R P moving range in T S S B R 73 data input (sampling rate = 1 min) 1 2 3 4 5 6 7 8 9 10 • ••••••••• X Y Z MRG of point 1-7 MRG of point 3-9 MRG of point 2-8 Fig. 5.4 Moving range with parameter, r = 7 5.1.1 Operational Characteristics and ORP (pH) Profile In this section, some general observations are made regarding the real-time operation of TSSBR using the RCMP as the designated control point. Under this operating strategy, a number of cycles occurred each day, resulting in the real-time process control of the TSSBR. Figure 5.5 represents the 3 cycles of ORP (pH)-time profiles which occurred in the RTPC in the operating mode. At the designated control point, the mixer and aeration were turned off, and nutrient loading then took place. Once an optimal ORP curve was established under conditions of real-time process control, a remarkable consistency in ORP and pH patterns was maintained from cycle to cycle over the course of this study, as illustrated in Figure 5.6. This consistency resulted in successful real-time process control of the TSSBR. Results show that this control method could provide near optimal conditions for bacterial growth and performance, resulting also in the optimization of energy costs as well as the optimization of wastewater treatment efficiencies 74 (H/BUJ) oa "8 E/Hd IT) CD sz 01 CD CO CO I -»*— o o 1_ c o O co Bacteria cell + 54 N 0 2 " + 57 H 2 0 + 104 H 2 C 0 3 (9) 400 N 0 2 " + N H 4 + + 4 H 2 C 0 3 + HCO3- + 195 0 2 —> Cell + 3 H 2 0 + 400 NO3- ( 1 °) Based on this theory, 8.64 mg HC0 3 per mg of ammonia-N oxidized (7.14 mg alkalinity as CaCOs) is required in the conversion process. The reduction in alkalinity and the acid production during nitrification would produce a reduction in pH . The liberation of hydrogen ions during nitrification could also be a factor in pH reduction (Equation 11-13). 2 N H 4 + + 3 0 2 > 2 N 0 2 " + 4 H + + 2 H 2 0 + energy (11) 78 Figure 5.7. Characterist ics of the pH curve, in relation to O R P and dissolved oxygen 79 2 N 0 2 " + 0 2 and overall N H 4 + + 2 0 2 > 2 NC>3' + energy > NO3- + 2 H + + H 2 0 + energy (12) (13) At the point where nitrification is complete, the pH break point occurs (point c). This break point is considered to be an on-line, real-time control point. As shown in Figure 5.7, this pH break point is coincident with the nitrogen break point (NBP) on the ORP curve and with the DO break point. This break point is taken to correspond to a point of zero ammonia-N in the system or the point of equilibrium between ammonia-N oxidation and the production of ammonia-N through the ammoniafication of organic nitrogenous materials. The pH increase beyond NBP (from c to d point) could be caused by the changes in soluble-gas composition in the wastewater. Under constant aeration, various gases such as CO, CO2, H2 and N2 present in the atmosphere are continuously dissolved and extracted from solution. An equilibrium in gas-solubility is maintained. In this state of equilibrium, the dissolution of oxygen, after complete nitrification, may result in the extraction of other gases from the wastewater. This air-stripping of CO, CO2 or other gases, which decrease pH, could lead to a subsequent rise in pH such as occurred in the experiments reported here. Figure 5.8 shows that a high ammonia-N removal rate was maintained until the NH4 -N level reached zero. The result was a high reduction in alkalinity. Beyond the NBP, the pH began to increase, although the alkalinity concentration remained constant, due to lack of further consumption. This track study therefore suggests that the pH increase at this point is not related to a change in alkalinity. It is also significant that changes in the aeration rate at the plateau (from d to e point) affected the pH curve. After the pH curve reached a plateau, a higher aeration rate was provided, resulting in an increase in pH. 80 60 50 j O) ^ 3 0 6 c o o 20 Batch A —•— NOx-N — 6—NH4-N —•—Alkalinity/10 air ON » 4 a i r O F F M l air ON 10 8.5 Time (hrs) 80 -a 70 O o TO 0 IO « 60 _j O) E 1 50 & c ra < 40 Bitch B air ON air OFF airON NBP on pH curve 30 10 Time (hrs) Figure 5.8. pH and alkalinity trends in the reactions 81 With regard to biological reaction, it is conjectured that in this phase (from d to e), the hydrolysis of organic nitrogen, bacterial endogenous respiration and aerobic respiration, using the resulting NO3" as a nitrogen source, may have taken place in the system. When the air was turned off and influent was loaded (from e to f ) , a rapid decrease in pH occurred. This sharp drop in pH may be due to the discontinuation of aeration, which stops the stripping process. The free oxygen in solution is then quickly exhausted from the system. As a result, CO2 is resolublized into solution, lowering the pH. The loading of influent, which has a lower pH than that of the reactor, could be a further factor in the pH decrease which occurred here. The continued pH decline (from f to a) after influent loading could also be the result of anaerobic fermentation. In anaerobic fermentation, complex organics are hydrolyzed into higher organic acids, using the organic compounds as electron acceptors. If anaerobic conditions continue, the higher organic acids are then broken down into short chain fatty acids, causing a continuous pH decrease until aeration recommences. The alkalinity recovery (pH increase), known to occur during denitrification in the anoxic phase (between e and f), was not observed during this experiment, probably due to the rapid denitrification which occurred during the influent loading. However, in the results shown in Figure 5.8, derived from batch tests, the alkalinity was recovered during the period when the air was off (pH increase), as the denitrification continued after influent loading. These specific pH features outlined above can be used for real-time control, due to the well defined control point. A feasible real-time control strategy using specific pH feature was also proposed by Y u et al (1997). Simple monitoring of pH patterns could be used to diagnose the on-going biochemical process mechanisms of the system. Two points can be designated on the pH curve for the real-time control of aeration, nitrogen break point (NBP) and plateau (Figure 82 5.9), depending on the wastewater characteristics. The recommended control points may be easily recognized by monitoring the slope change of the pH curve. It is recommended that the mV value of pH be used for the detection of control points, not the unit-pH values. Since the unit-pH change at pH break point (NBP) is usually kept within a 0.2-unit value for over more than one hour, the use of the unit value (dpH/dt) could make it difficult to detect significant slope change at the NBP. However, each control point on the pH curve can be readily identified by using the d(mV)pH/dt, since the mV-pH values have a large-scale slope change at NBP. Usually, one unit pH change has approximately 60 mV-pH variation, depending upon the pH electrolyte. The monitored mV-pH and the d(mV)pH/dt patterns during both the aerobic and anoxic phase are shown in Figure 5.9. The moving-d(mV)pH/dt (designated the pH-moving slope change) was calculated with r = 10 at one minute of sampling rate. The theory of moving slope change calculation with r = 10 is depicted in Figure 5.10. data input (sampling rate = 1 min.) 11 10 ^ 9 14 12 f 15 3 1 ^ 1 dx1 dx2 dx3 dx4 dx5 dx6 dx7 Fig. 5.10. Moving slope change with parameter, r = 10 83 A T- f T- T- T- N CM T- T- T- T- 1- M N T~ Time (hrs) Figure 5.9. Real-t ime control strategy using pH (mV) change 84 Figure 5.9 demonstrates that mV-pH and moving slope change patterns have an identical form from cycle to cycle. Hence, control points can be easily found on the curves. A zero value of d(mV)pH/dt may be used as a trigger value to achieve real-time control for nitrogen removal by terminating the aeration at NBP. When air-off is desired at the plateau (as is probably the case for the real-time control of high C and P wastewaters), the same value can be used as a trigger value. The feasible control strategies, using both NBP and plateau points on the pH curve, are illustrated in Figure 5.11. It should be noted that once each block is triggered (has yes value), the block must not check the entering values and simply pass them to the next block, in order to achieve real-time control, using the strategies outlined in Figure 5.11. The real-time control strategies developed here, using dynamic pH variation, may be applied to other aerobic treatment processes with different conditions, thereby providing opportunities both to save energy costs and to optimize system treatment capacities. 5.1.2 Real-Time Process Control of TSSBR with RCMP # 1 (RCT 1 and 2) 5.1.2.1 ORP and pH Profde Two patterns of anoxic ORP-time profile were observed in the operation of TSSBR#1. Since ORP and pH patterns in the A/O reactor are discussed in the previous section, this section offers general descriptions of the two different ORP curves observed in the anoxic reactor. In Figure 5.12, the nitrate knee, that is, the point at which the complete removal of nitrate occurs, took place within 7.1 hours; however, the complete denitrification did not occur in the results shown in Figure 5.13, despite the provision of a slightly longer anoxic period (7.34 hours) for denitrification. This difference could be attributed to the influence of daily operating factors such as the endogenous respiration rate of the bacteria, the concentration of bacteria in the reactor, the loading rate and the relative content of C/N in the influent wastewater loaded into the system. As noted in Chapter 3, the anoxic reactor was designed to achieve complete removal of 85 the designated RTC = NBP the designated RTC = plateau Figure 5.11. Real-t ime control strategies using dynamic pH change 86 200 Passing the liquor from A/O R. O ^ C N I O i O C O ^ L O C O C S I O ^ C N O i O C O ^ U O ^ C N O ^ C N T - L O r t ^ U ^ ^ C N J O ^ T C V I r r i - T - T - T - r T - ^ T - T - T - T - T - N C M C M O i t M t M Time Figure 5.12. O R P patterns in A / O and Anoxic reactor ORP (A/O reactor) ORP (Anoxic reactor) -incomplete denitrification h - OO OS O r f A i n W l O S C O O l LO CD oo cn L O O i - C M C O ^ J - C N C O ih N ro 6S d T- co Tf Time Figure 5.13. O R P and pH patterns in A / O and Anox ic reactor 87 NOx-N without the use of a supplemental carbon source. Given this, the organic carbon materials required for proper denitrification had to be provided from the bacterial endogenous respiration and from the residual carbon materials transferred from the A/O reactor. Theoretically, the residual carbon materials unoxidized during aerobic reaction would be dependent upon the initial ratio of carbon /nitrogen / phosphorus in the influent wastewater; the amount of reduced NOx-N to nitrogen gas by bacterial endogenous respiration would vary with time, microbial activity and the sludge concentration in the anoxic reactor. In order to trace the factors which affect denitrification efficiency, the graphs resulting from monitoring daily operating conditions were compared (Figure 5.12-operation day 112, Figure 5.13 - operation day 59). Complete denitrification was achieved on day 112, while incomplete denitrification was obtained on day 59. This may due to the higher bacterial concentrations in the reactor and a higher C/N content on day 112. The concentration of VSS on day 112 (5,050 mg/L) was 2.85 times higher than on day 59 (1,775 mg/L) and the relative C/N content of the loaded wastewater on day 112 and 59 was 3.86 (as TOC/TKN) and 3.09, respectively. Assuming that the denitrification rate (in mg/L. mg VSS) by bacterial endogenous respiration is relatively constant, the concentration of NOx-N removed within the designated period will be a function of the concentrations of bacteria and the amount of residual carbon in the reactor. Although the amount of residual organic materials in a reactor may not be exactly controlled using ORP, more residual carbon could be transferred from the A/O reactor when the wastewater has a higher C/N content, since organic carbon oxidation is artificially stopped when the ORP curve has a constant slope change after the complete removal of ammonia-N in the reactor. 5.1.2.2 General Observation As noted previously, this operating strategy consisted of two runs (RTC 1 and 2). The 88 general operating conditions were kept constant during the RTC 1 and 2, except for the ORP loading rate control set-point. In RTC 1, -190 mV was designated as a loading rate control set-point, while - 240 mV was used for RTC 2. Results of a 7 day tracking of the influent loading rate per cycle and of the treatment capacity are reported herein. A way to increase the treatment capacity in the operation of TSSBR is also suggested. The length of the aerobic phase of the A/O reactor is illustrated in Figure 5.14 and 5.15. Based on the tracked data for the 7 day period, the average length of the aerobic phase in RTC 1 of TSSBR # 1 was 5. 29 hours, with a minimum 3.3 to a maximum 7.3 hours; the average TOC and T K N loading rate were 1558 mg/cycle and 658 mg/cycle, respectively. In RTC 2, with -240 mV as a loading rate set-point, the average length of the aerobic cycle was 7.5 hours (Min. 3.8 hr., Max. 13.2 hr.) and the average loading rates were TOC-1988 mg/cycle and T K N - 915 mg/cycle. The longer aerobic length in RTC 2, compared to RTC 1, could be due to the higher nutrient loading rate per cycle resulting from the designation of a lower ORP loading rate set-point, since a longer aeration time will be required for the complete removal of nutrient under higher loading conditions. However, despite fewer cycles (air-ON + air OFF) per day obtained in RTC 2 (2.29 cycle/d) compared to RTC 1 (2.9 cycle/d), more wastewater was treated in RTC 2 (4518 mg TOCs/day and 1908 mg TKN/day in RTC 1 and 4553 mg TOCs/day and 2095 mg TKN/day in RTC2). These results suggest that the wastewater treatment capacity of TSSBR can be maximized by designating an optimal ORP set-point for the control of the loading rate, simultaneously optimizing the treatment efficiency. A relatively short retention time (10 d SRT and 4.6 d HRT in RTC 1 and 9.25 d SRT and 4.7 d HRT in RTC 2) was maintained during this study, compared to other research performed with swine wastewater reported in the literature. A long HRT (> lOd) and SRT (15-35d) have generally been used to achieve high treatment efficiencies in the treatment of swine manure 89 10 9 8 7 .C S 6 u> c a> 5 0 1 4 ^ 3 2 1 0 ORP loading rate setpoint: -190mV AVG loading rate/cycle: TOC - 1558 mg; TKN - 658 mg AVG aerobic cycle length I 58 59 60 6 J . ays 62 63 64 Figure 5.14. Aerob ic cycle length in A / O reactor for 7 days (RTC1) 14 12 10 I 8 c o> u 5 6 0 1 < ORP setpoint: -240mV AVG loading rate/cycle: TOC - 1988 mg; TKN - 915 mg AVG aerobic cycle length 108 109 110 111 112 113 days 114 Figure 5.15. Aerob ic cycle length in A / O reactor for 7 days ( R T C 2) 90 (Bortone, 1992). The MLVSS concentrations in each reactor were on average 7.2417 g/L (A/O reactor) and 1.958 g/L (Anoxic reactor) in RTC 1 and 6.8167 g/L (A/O) and 2.74 g/L (Anoxic) in RTC 2, indicating higher MLVSS concentrations than the recommended values (1.5 - 5.0 g/L) for SBR by Metcalf and Eddy (1991). The average F /M (in applied mg TOCs / mg MLVSS.d) was 0.062 in RTC 1 and 0.065 in RTC 2, based on MLVSS in the A/O reactor. When calculated with BOD,.^ the average F /M (applied B O D 5 S mg / mg MLVSS.d) were 0.105 and 0.109, based on MLVSS in the A/O reactor, respectively. The achieved F /M ratio was kept within the ratio for typical SBR recommended by Metcalf and Eddy (0.05 to 0.3, applied BOD 5 mg/L / MLVSS.d). Equation 14 is derived from the correlation of TOCs with B O D 5 § in influent (described in the Appendix A) and was used for the calculation of average F /M in the form of B O D 5 § / M L V S S , since the measurement of B O D 5 § was not done during the operation of RTC 1. BOD 5 S mg/d = TOCs (mg/L) - 345.13 x loaded influent volume (L/d) (14) 0.448 5.1.2.3 System Performance 5.1.2.3.1 Wastewater characteristics used in TSSBR #1 Table 5.1 summarizes the characteristics of the influent wastewater used for TSSBR # 1. The average COD, T K N and Total phosphorus concentrations of the influent were 6916 mg/L, 776 mg/L and 572 mg/L, respectively, showing C:N:P (as COD:TKN:TP) of 100:11.2:8.3. Based on the soluble BOD 5 , the relative ratio of C:N:P was higher (100:24.1:16) than the optimal ratio (100:5:1) suggested by Metcalf and Eddy (1991). The average concentration of total solids 91 04 oo oo H & T 3 (U O o IT) > Q C O 00 O N r - i o MD I—i m i IT) MD C N ; * .s CM in 00 1—1 MD 00 o CN XI O N O N ; CN co CN O N o I N MD C N ; O N m CO co co CO M D : MD in o co o MD OO I MD 00 r-- O N r-- co IT) T T M D ; co oo c~- 00 i—i CO CN CNi co C N 1 — 1 in MD MD O N o CN O N co MD 1—1 O N i—< i n t-- O N O N 00 m O N co CN m CN co m co CN > Q -d oo X * .g I OH ~5w t-~ • ! i t - oo O N C O C N m t—H O ! ! ! C N C N OO MD C O ! ! ! O N m MD 1—1 1 r- ; ! : C N r - 00 C N * * * * * * Q tsi * Q Q u O O O o U U m H ON OO : ! : OO MD co C N C N OO ^ : i : C O 1—1 oo ON C N m C N o : ! ; r -C N i ] : C N C N O N o in oo in O N oo o co 00 i ; ! i MD in in O o : MD i—<; 1 i ' JO CN MD M D ; CN i—H : ! i cn O N co ^ - r -1 oo ; ! i O ; ! i O 00 cn cn in in C N oo; ! : C N 00 MD 00 1—1 i—i oo oo ON • ! • *sf '—' i n o >—1 rn •—1 C N MD O N CN O o o O N cn —1 O *-* t-- o oo i—i o m o MD MD MD i in i n i i CN o cn o O N i n 00 i n cn O N 3^-o o o o CN o 00 r- 1—1 ON cn ON cn ON MD OO CN cn CN 3 5" OB oo r/3 > OO H oo C O C O C O 00 MD CN o in C O MD i—i 1—1 in C O : O N CN C O 00 ; oo MD C O i C O i n O O N • t~i CN i n MD C O CN 00 00 > O N and suspended solids in the influent were 4.8865 g/L and 2.7582 g/L, respectively, with an average ratio of 0.84 (VSS/TSS). Since no effort was made to regulate the quality of the influent wastewater, the daily variation in the influent strength was very high and this fact is reflected in the large standard deviation. Comparing the chemical parameters used in the two operations, the nutrient content of the influent was relatively higher in RTC 1 than in RTC 2. Furthermore, the variation in the influent wastewater was larger in RTC 1 than in RTC 2. This could, in part, have resulted from the higher solids concentration in RTC 1. This is evident if a comparison is made between the concentrations of soluble carbon and total carbon of the influent in both operations. The total COD concentration in each operation averaged 8267 mg/L in RCT 1 and 5565 in RTC 2. In spite of the large difference in total COD concentration between the two runs, the soluble COD and TOC concentrations in RTC 1 and RTC 2 were similar, with an average soluble COD of 4111 g/L and 3916 mg/L, and soluble TOC of 1457 mg/L and 1419 mg/L, respectively. The soluble BOD 5 of the influent in RTC 2 averaged 2396 mg/L, and the relative ratio of BOD 5s to CODs and the BOD 5 /TOC was 0.61 and 1.68, respectively. The relative ratio of TOCs/CODs averaged 0.35 in RTC 1 and 0.36 in RTC 2. With regard to the organic carbon content, the swine wastewater had a very similar carbon composition to typical domestic wastewater, in which the BOD/COD ratio varies from 0.4 to 0.8 and BOD5 /TOC varies from 1.0 to 1.6 (Metcalf and Eddy, 1991). The average concentrations of ammonia-N, TKN, Ortho-P and Total P loaded into the system were 326 mg/L, 975 mg/L, 470 mg/L and 760 mg/L in RTC 1, and 250 mg/L, 577 mg/L, 299 mg/L and 383 mg/L in RTC2, respectively. The influent NOx-N was only 3.65 mg/L in RTC 1 and 1.92 mg/L in RTC 2. 93 5.1.2.3.2 Removal efficiencies and General Discussion The effluent quality and the removal efficiencies obtained from the two separate operations (RTC 1 and RTC 2) are summarized in Table 5.2 and 5.3. Despite the serious variations in influent quality during the operation of the TSSBR # 1, constant removal efficiencies were achieved. Since the TSSBR was operated under a condition of hugely varying influent conditions, if operated with fixed-time mode, the system could be faced with very high fluctuations in food to microorganisms ratio, resulting in poor performances and a substantive variance in effluent quality. In general, effluent quality is known to be directly functional to influent quality when the system is operated in a time cycle mode (fixed HRT and SRT). However, by achieving real-time control, a relatively constant final effluent quality was obtained along with high nutrient removal performance. This constant and high system performance indicates the ability of real-time control technology to respond to the on-going biochemical reactions in system. The average removal efficiencies of TSSBR # 1 (RTC 1 and 2) were CODt-97.2 %; CODs-96.2 %; 94.9 %; BOD 5-99.6 %; NH 4-N-98.3%; TKN-96.2 %; Ortho-P-94.6 %; Total-P-94.4 %; TVS-94.4 %; TS-75.9%; SS-97.3 % ; and VSS-97.9 %. Comparing the removal efficiencies in RTC 1 to RTC 2, no significant difference was found, despite different loading rate control ORP-setpoints having been designated. This suggests that the performance of TSSBR, with real-time control technology, might not be significantly affected by the loading rate of the influent, since the bio-chemical reaction under real-time control is always ensured, until nutrient removal is complete. However, it is recommended that a sufficiently low ORP set-point, for influent loading control, be selected, in order to obtain proper phosphorus release with sufficient internal organic matter in the air-off phase and thereby achieve enhanced phosphorus removal under aerobic conditions. The objectives of loading rate control, using an ORP-setpoint, is to maintain near optimal operating conditions for bacteria, by 94 o H c CO 'o c "ca > o S CD c -4—» (3 o —i C N C N ON o 00 C N C N CO SH —* CD J a -ca iri CN ON CN CN © ON ON Tl" f-H 1—1 o m CN oo Tl- O © ON 00 T f m CN ON oo IT) o ON i n r -in OO o CN ON CN CN CN * Q * * * Q o O O o u u H C-- i n i n O o V© 1—H i n CN r-CN 1—* m ON c i r--OO 1 — 1 CN o o 1—H o o O o CN © © o o o o o o o CN o o © © © © CN CN OO ON 1—1 r—I i—< i n m CN 1—< O © © © © 2" 3" —^y w > C/3 H CZ) ON CN OO 00 NO m CO in © ON '—1 CN i n CO CN C/3 00 > 10 ON CN u H Pi a '3 « '3 > o s n u 13 s S > Q —^• t/5 S 05 I ,1 s m MD MD O N o CN O N cn o CN o 00 MD 1—1 O N 1—1 in r- O N O N oo T i - 1—1 O N in O N co CN m • CN cn ro O N CO O N in C O CN MD 00 CN O N r—1 oo H GO CO GO > OS preventing serious fluctuations in the F /M ratio; also, it is used to obtain an optimal ORP pattern, as well as to ensure P release and complete denitrification. It should be noted that the fluctuation of the F/M ratio did have an influence on the ORP curve. To maintain the optimal ORP curve which is indispensable to the achievement of successful real-time control under a constant aeration rate, the loading rate should be kept constant; otherwise, the aeration rate should be changed frequently to balance the required oxygen demand of the system with the oxygen uptake rate of the bacteria. Another evident use of loading rate control by the ORP setpoint is to achieve a flexible loading volume of influent, depending on the influent quality and the oxidation status of the reactor. This loading rate control method, using ORP monitoring, can also be effective for non-real-time control systems, such as any system operated with fixed time cycles. By applying this loading rate control method to fixed-time mode systems, a flexible HRT, which varies with influent strength, can be achieved at certain ranges of SRT, thereby optimizing effluent quality. For instance, under a condition of high influent strength or low oxidation status in the reactor, a low volume of influent would be loaded into the system since the monitored ORP value quickly reaches the designated ORP setpoint at the start of influent loading. In this way, a flexible influent loading rate, capable of responding to variations in the strength of the influent and to the oxidation status of the system, could be obtained. Theoretically, the loading volume /or loading rate of influent will be inversely proportion to the influent strength/ or proportional to the oxidation status of the reactor. In RTC 1, removal efficiencies of COD t , C O D s , T O C s , NH 4 -N, TKN, Ortho-P and Total P obtained were 97 %, 95.5 %, 94.2 %, 98.8 %, 95.7 % , 93.2 % and 93.7 %, respectively. The highest removal efficiency achieved was for NH 4 -N, where the average in the influent was 326 mg/L of NH 4 -N and the average in the effluent declined to 4 mg/L. Very high suspended solids 97 removal efficiencies were also achieved; the average was over 99 %. The removal efficiency for total solids was low (76.4 %), while the total volatile solids removal efficiency was greater than 90 %. This latter result suggests that swine manure contains a high portion of inert colloidal solids which cannot be biodegraded by bacterial action. Profile of the nutrient parameters over time are presented in Figure 5.16 to 5.20. The COD in the influent fluctuated in accordance with the solids concentration of the influent. This is evident if Figure 5.16 is compared to Figure 5.19, as the profiles are similar. The T K N and phosphorus profiles with time (Figure 5.17 and 5.18) show a similar trend, where total COD varies with the solids concentration in the influent. While the soluble phosphorus in the influent fluctuated in accordance with the total phosphorus, the T K N was not dependent upon the concentration of ammonia-nitrogen. This is attributed to the major portion of T K N in the influent being organic-nitrogen, and this can vary with the solids concentration of the wastewater. The percent ammonia-nitrogen of T K N in the influent averaged 33.4 % and the organic-nitrogen averaged 66.6 %. However, the T K N associated with ammonia-nitrogen on day 51, when the influent had a high solids content, was only 18.7 %. A higher denitrification efficiency was achieved between days 41 to 52, where an average 0.75 mg/L of NOx-N was achieved in the effluent. During this period, the effluent contained much higher levels of carbon material than in the days which achieved a lower denitrification efficiency. During this same period, the total COD and soluble TOC concentrations in the effluent were 2.87 and 1.79 times higher than other days, respectively. This indicates that a carbon source sufficient for denitrification was present in solution during this period, resulting in complete NOx-N removal in the anoxic reactor. The parameter profile graphs (Figure 5.16 to 5.20) reveal that under conditions of influent fluctuation, a constant quality of effluent was achieved through Real Time Control of TSSBR. 98 20000 18000 16000 14000 4 3500 5" 12000 O) E 10000 a o o 8000 -6000 4000 -2000 Time (day) Figure 5.16. Track of organic matter in influent and eff luent-RTC1 3000 Time (days) Figure 5.17. Track of phosphorus in influent and eff luent-RTC1 gg 1800 -. Time (days) C M < D T - c o i n N - o > T - c o m f ^ o > T - c o i o Time (days) Figure 5.18. Track of nitrogen in influent and effluent -RTC1 1 0 0 20 18 16 ? 14 D) » 12 T3 5 8 3 5= = 6 4 2 0 •Influent TS •Influent TVS •Influent SS • influent VSS 8 • 0 x • • • • • • • Time (days) Figure 5.19. Track of influent solids in R T C 1 4 3.5 3 S 2.5 in o ? in c 0) 2 1.5 i t iii 1 0.5 Effluent TS effluent TVS Effluent SS effluent VSS • - X - " D ° ° ° ° I X X X X Time (days) Figure 5.20. Track of effluent solids in R T C 1 101 Under real-time control, the bio-chemical reaction in a system is auto-maintained artificially until the system reaches a targeted oxidation status and the optimal conditions for bacteria can be maintained by switching the reaction phases, when no external nutrients exist in solution. This results in high removal efficiencies and a constant quality of effluent, despite very large variations in the influent strength. High removal efficiencies of organic materials were also achieved in RTC 2 of TSSBR # 1, averaging 97.4 % of total COD, 96.9 % of soluble COD, 99.6 % of soluble B O D 5 and 95.6 % of soluble TOC. The effluent quality from the system in RTC 2 is shown in Table 5.3. The highest removal efficiency was found in soluble BOD5, indicating the complete removal of biodegradable organic materials from the wastewater. The average effluent soluble BOD5 was 9.3 mg/L and ranged between 3 to 20.5 mg/L. The high organic carbon removal efficiencies in both RTC 2 and RTC 1 are attributed to the real-time control strategy with the RCMP, an operating strategy which was designed to stop the carbon oxidation in the A/O reactor when relatively complete organic carbon removal was obtained. The residual carbon materials in solution can be utilized as a carbon source by denitrifiers, to convert the NOx-N produced to nitrogen gas; this results in more complete biodegradable organic carbon removal from the anoxic reactor. The influent and effluent organic carbon profiles with time are presented in Figure 5.21 and 5.22. As in RTC 1, the organic carbon material in the influent fluctuated in accordance with the solids concentration in the wastewater. The total COD trend of the influent during day 94 to 100 could be due to the effects of suspended solids on total COD measurement. (The solids content in influent during that period, was different from those of other days, as presented later). While the concentration of total VSS was usually higher than that of SS during the operation, the 102 total VSS concentration in the influent was lower than the SS concentration during those days. This high SS concentration should have resulted in the high total COD concentration. High nitrogen removal was also achieved in RTC 2; 97.8 % of ammonia-N and 96.7 % of T K N removal efficiencies were obtained. The average concentrations of N H 4 - N and T K N in the final effluent were 5.4 mg/L and 19 mg/L, respectively. This high nitrogen removal efficiency achieved from the TSSBR may also be attributed to the real-time operation: the complete removal of nitrogen in a system is always ensured under real-time process control with RCMP. The influent nitrogen and effluent profile with time are presented in Figure 5.23. The N H 4 - N concentration in the influent was relatively constant during the operation, while the T K N content fluctuated with time, due to the measured T K N concentration being partially dependent upon the solids content in a wastewater. Average NOx-N concentrations in the effluent from RTC 2 was 11 mg/L, varying from 3 to 27.1 mg/L. To investigate the effects on denitrification of additions of sludge to the anoxic reactor, 200 mL of settled anoxic reactor sludge was replaced with the same volume of settled sludge from the A/O reactor on day 100 and the change in NOx-N in the effluent was tracked over time. Figure 5.23 shows that high NOx-N concentrations in the effluent (average 21 mg/L) was significantly decreased to less than 5 mg/L. This enhanced denitrification efficiency was maintained during the rest of the period of operation, indicating that the carbon source required for denitrification can be provided by the increased endogenous respiration of bacteria and the energy stored within the bacteria. Particulated organic carbon materials in the sludge or organic carbon trapped in the slime of bacterial floes was probably hydrolyzed to short chain carbon material and this organic material could also be used for enhanced denitrification. This suggests that the addition of sludge can enhance denitrification efficiency. Furthermore, addition of sludge at intervals could provide a means of controlling the NOx-N concentration in the final 103 10000 9000 8000 7000 6000 O) E, 5000 Q O O 4000 3000 2000 1000 -•—Influent CODt -•—Influent CODs -A—Effluent CODt -G— Effluent CODs I I I M I I I I I I H i d 89 91 93 95 97 99 101 103 105 107 109 111 113 115 117 Time (days) Figure 5.21. Track of C O D of influent and eff luent-RTC 2 3500 -, 500 89 91 93 95 97 99 101 103 105 107 109 111 113 115 117 Time (days) Figure 5.22. Track of T O C s and B O D 5 in influent and eff luent-RTC 2 104 1200 -, CT)^-fomi^O)^-coir)r--cx>T-coir)r-o o o ) 0 ) - o > c j > a > o o o o o - < - T - T - T -Time (days) Figure 5.23. Track of nitrogen in influent and effluent - R T C 2 105 effluent from TSSBR. Over 95 percent of phosphorus removal was achieved in RTC 2, with an average ortho-P removal of 96 % and total-P removal of 95 %. The phosphorus removal efficiency in the TSSBR was nearly constant, despite the fluctuations in influent phosphorus concentrations. As shown in Figure 5.24, the ortho-P concentration in the influent during the operating period of day 98 to 102, was not proportional to the total phosphorus concentration, while the ortho-P and total-P concentration in day 103 to day 113 day did correlate. The high total phosphorus during the period day 98 ~ 102 is attributed to the high solids concentration present during this period. The solids profile with time is represented in Figure 5.25 and 5.26. No phosphorus release was expected to occur in the anoxic reactor, since the nitrate concentration in the reactor never 6 0 0 CO OJ Tim e ( d a y s ) Figure. 5.24. Phosphorus in influent and effluent - RTC 2 106 9 8 • Influent TS -Influent TVS 7 A Influent SS 6 • X influent VSS •o 5 C 4 a> £ 3 2 1 0 • • • • A • • A • • * • • • • A • • • X 89 91 93 95 97 101 103 105 107 109 111 113 115 117 Time (days) Figure 5.25. Variation of Sol ids in Influent - R T C 2 2.5 in o in C 0> 3 3= HI 1.5 0.5 89 - X -91 93 - X -• • • • • 0 X A -* • Effluent TS • effluent TVS A Effluent SS X effluent VSS • J L • J L 95 97 99 101 103 105 Time (days) 107 109 111 113 115 117 Figure 5.26. Variation of solids in eff luent-RTC 2 107 reached a zero state during the operation. In contrast to P release, phosphorus uptake, using the nitrate as an electron acceptor may have occurred in the anoxic reactor, resulting in the low phosphorus concentrations in the final effluent. Solids removal shows very similar results to RTC 1, with high TVS, SS and VSS removal efficiencies, but low TS removal efficiency. The average total solid (TS) and total volatile solid (TVS) concentrations in the effluent were 0.61 g/L and 0.39 g/L, achieving 75.3 % and 84.9 % removal efficiencies. The removal efficiencies of SS and VSS were 95.2 % and 96.4 %, with an average concentration of 0.22 g/L and 0.15 g/L, respectively. Unlike the influent, the effluent nutrient levels were not dependent upon the solids concentration, as the parameter profile graphs show. 5.1.3 Real Time Process Control of TSSBR # 2 with RCMP (RTC 3) As noted in Chapter 3, the RCMP on the ORP curve was designated as the control point for the real-time process control of both TSSBR # 1 and 2. The same ORP value of -240 mV as in RTC 2 of TSSBR # 1, was chosen for control of the loading rate. There was, however, a major difference in the sludge conditions between RTC 2 of TSSBR # 1 and TSSBR # 2. While the sludge in each reactor was completely separated in the operation of TSSBR # 1, in TSSBR # 2, 50 mL of the settled sludge in the anoxic reactor was replaced by the same volume of settled sludge from the A/O reactor. The sludge was replaced during every cycle at the time of liquor transfer from the A/O reactor to the anoxic reactor. 5.1.3.1 ORP and pH Profde Optimal ORP, pH and DO curves were consistently obtained throughout the experiment, thereby achieving successful real-time process control of TSSBR # 2. The ORP, pH and DO profile with time in the A/O reactor is shown in Figure 5.27. This figure illustrates that 108 Figure 5.27. Real-time control of TSSBR#2 (ORP, pH and DO in A/O) 109 the DO values corresponding to the ORP real-time control point (RCMP), as well as the DO slope change prior to the DO break point, shifted down with the progress of time. This trend is attributed to the sludge accumulation in the reactor. Sludge accumulated in the A/O reactor until the excess was wasted from the reactor on a weekly basis. The sludge accumulation rate in the A/O reactor averaged 0.42 g/L per day during the operation of TSSBR#2. As sludge concentration increased, the total oxygen uptake rate could increase, resulting in the downward shift of the DO curve. At the start of aeration in the first and second cycle, the DO concentration (DO slope change) in the reactor increased quickly, probably due to the aeration rate being greater than the oxygen uptake rate by bacteria. This DO trend indicates that the aeration rate in the first and second cycles was probably not optimized, resulting in excessive energy usage. On the other hand, during the sixth and seventh cycles, where there is a low DO slope change, the response indicates that an optimal aeration balance was established in the system. During these cycles, the aeration rate was probably equal to or a little higher than the oxygen uptake rate by bacteria. The increase in sludge in the reactor should also have resulted in a decline of the ORP as time progressed, since the ORP has an inverse relation to the concentration of solids in solution. The graph indicates that the ORP value at the RCMP was not significantly changed. However, the B point on the ORP curves was shifting downward as sludge accumulated in the reactor. After wasting of the excess sludge, the ORP and DO curves returned to a higher position again and the decline repeated from that point until the next sludge wastage. Unlike the ORP and DO trends, the pH curve was constant. This may be due to the pH electrolyte response to the hydrogen ion concentration in solution, and not to the overall bio-chemical reaction state in the system. In order to achieve system optimization and real-time control using ORP, the system oxygen demand and the aeration rate must be balanced. Based on the ORP and DO patterns 110 obtained through this study, when the B point on the ORP curve occurred approximately between -200 and -100 mV and the ORP curve beyond B point showed an approximately 50 to 60 0 slope change, a balance between the aeration rate and the oxygen demand of the system was established. This balanced condition resulted in the ideal ORP curves essential to the achievement of real-time control. This suggests that optimal aeration balance can be maintained by carefully monitoring the B point on the ORP curve. Furthermore, a diagnosis can be made as to whether there is insufficient or excess aeration by checking the ORP curve. This is due to the fact that the absolute ORP value and the trend of the ORP curve are usually correlated to the DO level and pattern. As noted in Figure 5.27, when the B point occurred above -100 mV, the system was operating under conditions of excess aeration, while, when the B point occurred between -150 to -100 mV, an optimal rate of aeration was established. Based on these observations, two control strategies for the optimization of the aeration rate may be established. The first is to constantly maintain the optimal sludge concentration at an established aeration rate by wasting sludge on a daily or cyclical basis. The second strategy is to adjust the aeration rate frequently in relation to careful monitoring of the ORP curve. The first method may be difficult to achieve, since the sludge production rate is dependent upon several factors such as the solids concentration in the influent, bacterial activity and the nutrient loading rate. The most evident advantage of real-time process control over a fixed-time cycle mode of operation would be the complete removal of contaminants in wastewater by the achievement of a flexible hydraulic retention time (HRT); this would vary with both the wastewater quality and the bacterial activity in the system. A further advantage would be the optimization of operating costs and system capacity. Figure 5.28 reflects a flexible HRT in the A/O reactor. During the operating period day 154 to 155, more than 6 cycles occurred and the lengths of the aerobic periods needed 111 0 E c g -4—' o CO CD c o CO '(_ CD > 00 CM i r i CD (A"J) dyo to achieve the targeted oxidation status varied from 3.3 hrs to 8.5 hrs, despite the relatively constant composition of the influent loaded into the system during that period. During the real-time control of TSSBR # 1 and # 2, the ORP values, which corresponded to the RCMP control point, never reached more than + 50 mV. This is probably a result of the high sludge concentrations in the reactor. During the operation of TSSBR # 2, a relatively high suspended solids concentration in the A/O reactor was maintained, averaging 10.5 g/L of SS and the pH was kept in a neutral range. According to Hao and Huang (1996), the absolute value of ORP does not impart any process significance, since the ORP values are subject to change according to other factors, such as pH and solids concentration. With respect to biochemical reactions, nitrification, organic carbon oxidation and phosphorus uptake may have occurred simultaneously, from the start of aeration to the nitrogen break point (NBP) on the ORP curve. However, beyond the NBP, the dominant reaction appears to be organic carbon oxidation. Of course, P uptake could have continued if there was any phosphorus remaining in solution. In the second cycle (Figure 5.28), nitrification was completed so quickly (i.e. there was a high nitrification rate) that the period of aerobic treatment provided to achieve the targeted organic carbon oxidation level (RCMP) beyond the NBP was longer than that required for nitrification. This was unlike other cycles. In the third cycle, most of the organic carbon could be simultaneously removed during the quite long nitrification period; also, a shorter aerobic period was maintained beyond the NBP, in order to achieve the targeted carbon oxidation status. These two patterns of reactions suggest the capacity of ORP to respond to variations in biochemical reactions and to system conditions is very good. Figure 5.29 also represents the response of the real-time control system to treatment system conditions. The variation in the aerobic period in the A/O reactor was tracked for 7 days (day 163 to 169). The aerobic cycle lengths were highly variable (2.2 to 14.8 hours), with an 113 Figure 5.29. Variation of aerobic cycle length in A / O for 7 days 114 average of 6.5 hours. During the first 4 days ( day 163 to day 166), the length of the aerobic cycle required to achieve the targeted oxidation status showed quite a high variation, indicating that the biological activity in the system fluctuated considerably from cycle to cycle. Feeding control failure was suspected to be occurring at that time. However, it was found that no variable loading rate condition was provided during those days. Approximately the same quality of influent was loaded into the system and no difference in the reactor volume was observed. It was therefore concluded that the highly variable cycle lengths during this period could be due to some other operational factors, not yet ascertained. This data does reflect, however, the potential of a real-time control system to optimize system performance under highly variable operating conditions. Considering the results in Figures 5.28 and 5.29, it is possible that the process control strategies using fixed HRTs (fixed time cycle) could be lead to unnecessarily high energy usage with extended, over-aeration or incomplete treatment of wastewater, since the HRTs required to obtain complete nutrient removal were highly variable from cycle to cycle. The average TOC and T K N loading rate per cycle in that same period were slightly different from RTC 2, although the same ORP loading rate set-point was used. Although, 1988 mg of TOC/cycle and 915 mg of TKN/cycle were fed into the system in RTC 2, the resultant loading rates were 2353 mg TOC/cycle and 436 mg TKN/cycle. These results can be attributed to the different composition of the influent wastewater between RTC 2 and this operation. The C/N (as TOC/TKN) ratios of the influent were much higher in this operation than in RTC 2: 2.2 in RTC 2 and 5.4 in this operation. Since the influent is fed until the redox potential reaches the designated ORP loading rate set-point and the ORP reading generally indicates the overall oxidation-reduction status in system, the loading rate of any specific nutrient (C, N , or P) will be dependent upon the quality of the wastewater, namely the relative ratio of nutrients in the wastewater. For instance, assuming that the influent wastewater loaded into a system contains a 115 high ratio of C/N, more and more organic carbon over nitrogen will gradually be loaded into the system, as the influent loading is continued, until redox potential has reached an established setpoint value. However, it is thought that specific nutrient (C or N) loading rate control may be possible, if the wastewater quality has a small range of variation. It is interesting to note that, when the C and N loading rates are combined (C plus N), a similar loading rate was achieved in both operations, which had the same loading rate set-point (-240 mV), 2790 mg /cycle in RTC 3 and 2900 mg/cycle in RTC 2, respectively; at the same time, an average 2215 mg /cycle loading rate was achieved with -190 mV, as a loading rate control set-point in RTC 1. To get successful real-time process control, over- or under-aeration should be prevented, since an ideal ORP curve cannot be obtained under these conditions. As mentioned before, an optimal condition of aeration can be established by balancing the aeration rate to the biomass concentration and the nutrient levels in a given system. However, since the amount of air required at any time is a function of many factors (such as temperature) and since the solubility of oxygen into water is dependent upon the level of purity of the wastewater, an on-line monitoring system may be needed to diagnose the status of aeration in a system. In order to get more insight into the relation of ORP to the aeration rate, the effects of the aeration rate on the ORP curve and the system treatment capacity were tested by changing the aeration rate during normal real-time operation. The results are portrayed in Figure 5.30. All operating conditions were kept constant during this study, except for the rate of aeration. The quality of influent was also kept relatively constant (see day 147 and 148 influent). During the first two cycles, 0.7 L/min. (less than half of a normal system aeration rate) was provided and maintained, while a normal system aeration rate (1.5 L/min.) was provided for the other three cycles. Figure 5.30 shows that an average of 14.1 hours of aerobic treatment was required with 116 Figure 5.30. Aeration effects on O R P curve and treatment periods 117 the low aeration rate, and only two cycles occurred over 34 hours. However, when the system was operated with a normal aeration rate, the average aerobic period was 4.9 hours and a total of three cycles was obtained over 23.7 hours. The long aerobic period which occurs with a lower aeration rate may be due to lower rates of nitrification, P-uptake and organic carbon oxidation. Partial inhibition of nitrification in a system can occur when the system is subjected to conditions of oxygen deficiency. A low nitrification rate, attributed to insufficient aeration, can be detected by monitoring ORP or pH. The pH trend in conditions of low aeration is somewhat different from the pH trend in optimal conditions. In these studies, the pH increase at the start of aeration was maintained for a much longer time under conditions of low aeration. In addition, when insufficient aeration was provided, the ORP curve lost its ideal pattern. To be specific, the angle of ORP slope change, beyond B point, was significantly reduced to about 40°. These results suggest that the operating conditions in any unit system can be diagnosed by in situ pH and ORP monitoring. It also strengthens the idea put forward earlier that a means is hereby provided for optimal aeration control strategies using ORP. It is evident from this data that a balanced aeration rate should be maintained in order to optimize the treatment capacity and the operational costs through energy conservation. A lower aeration rate minimizes the treatment capacity by lengthening the treatment time, and an excessively high aeration rate can result in excessively high energy usage. The data also suggests that the loading rate should be controlled to obtain an ideal ORP curve and to optimize the system conditions, since the abrupt fluctuation of loading rate, due to the variability of influent quality, can give rise to undesired effects. 5.1.3.2 Operational Characteristics and General Observation The average system SRT and HRT in this operation were 9.87 and 5.8 days, respectively. The MLVSS concentration in each reactor averaged 8.4167 g/L in the A/O reactor and 3.6333 g/L in the Anoxic reactor. The F/M, as applied mg TOCs / mg MLVSS.d, was 0.075, based 118 on the A/O MLVSS. Comparing HRTs obtained in TSSBR#1 (RTC1 and 2) and TSSBR#2 (RTC 3) which were operated with the same real-time control point (RCMP), it was found that the HRTs maintained were proportional to the F /M ratio; that is, a higher F /M resulted in a longer HRT. The achieved average F/M ratios in RTC 1, RTC 2 and RTC 3 were 0.062, 0.065 and 0.075 and the resultant HRTs were 4.6, 4.7 and 5.8 days, respectively. This suggests that under real-time control, the HRTs can be automatically adjusted to a change of F/M, resulting in flexible HRTs. During the system operation, some failures in real-time control occurred. These were due to a failure of influent feeding and a malfunction in the control system (Figure 5.31 and 5.32). As illustrated in Figure 5.31, influent was not loaded into the system on the 4th cycle, due to an empty influent bucket. This resulted in a real-time control failure in the next (5th) cycle. At the 5th cycle, the computer did not detect the B point on the ORP curve, as the ORP curve reached the plateau very quickly. To re-establish real-time process control, aeration was shut down manually and influent loading was carried out. A 2.5 hour anoxic phase was then provided to mimic the normal real-time sequences and the computer was re-started. By these actions, real-time conditions were recovered in the next cycle and continued real-time control was re-instituted. When the feeding was missed, the ORP and pH curve lost the normal pattern, in that no declining pH curve trend occurred at the time of influent feeding and no B point was observed on the ORP curve. However, unlike other cycles in which influent loading was not missed, a distinctive nitrate knee did occur (4th cycle). This is explained by postulating that a carbon source for denitrification could have been provided by the residual organic carbon materials left in solution, in a stored carbon source within the cells and by bacterial endogenous respiration. It is most likely that the major organic carbon source used for complete denitrification would be the 119 Figure. 5.32. Control failure by wire disconnection 120 residual organic materials in solution and the stored energy source inside cells, considering that the denitrification was completed within a short (2.5 hours) period, despite the lack of an additional external carbon source. Usually, an extended reaction time is required for complete denitrification by the endogenous respiration of bacteria. The denitrification rates using internally available organic carbon sources are described in section 5.1.3.3.2. Further possible reasons for the lack of a nitrate knee under a normal real-time control are described in detail in section 5-1. To eliminate unexpected control failures, electrical devices and equipment should be checked on a regular basis. The control failure in Figure 5.32 resulted from an accidental wire disconnection between the computer and relay. Although the computer recognized the designated real-time control point and sent out the ON/OFF signal to relay, the treatment system was not properly controlled, resulting in unnecessary operational costs due to approximately 12 hour's over-aeration. 5.1.3.3 System Performance 5.1.3.3.1 Wastewater Characteristics used for TSSBR#2 The characteristics of the wastewater used in TSSBR#2 are summarized in Table 5.4. The average concentrations of TOCs, NH 4 -N , T K N , Ortho-P, Total-P and NOx-N were 2518 mg/L, 206 mg/L, 522 mg/L, 122 mg/L, 152 mg/L and 1.77 mg/L, respectively. The relative C/N ratio (TOCs/TKN) was 4.82, which was 2.4 times higher than in TSSBR#1. The percent of ammonia-N relative to total Kjeldahl nitrogen (TKN) in the swine wastewater was 39 % (NH4 -N/TKN = 0.39), indicating that the swine manure has a high portion of organic nitrogen. With regard to solids content, the influent total solid (TS) was 26 % inorganic and 74 % organic matter. The influent suspended solids (SS) content was predominately organic matter: 95 % organic and 5 % inorganic matter. A comparison of the TS and SS content revealed that the majority of inorganic solids existed in colloidal form. The average concentration of TS, 121 Table 5.4. Wastewater charateristics used for TSSBR # 2 (RTC 3) Parameters Means Min. - Max. Std. Dev. (mg/L) T O C s * * 2518 1166 -4380 870 N H 4 - N 206 99 - 328 83 T K N 522 332 - 809 128 NO x -N 1.77 0.41-3.6 0.92 Ortho-P 122 69-163 30.3 T-P 152 94 - 240 39.7 T S (g/L) 5.1167 3.1-9.1 1.687 TVS (g/L) 3.8014 2.4-7.1 1.332 SS (g/L) 3.3875 1.75-6.95 1.490 VSS (g/L) 3.2250 1.75-6.40 1.319 ** s: Soluble Table 5.5. The effluent quality and removal efficiencies in TSSBR # 2 (RTC 3) Parameters (mg/L) Means Min. - Max. Std. Dev. % removal TOCs** 52 38-64 6.57 97.9 NH 4 -N 2.9 0-4.9 1.36 98.6 T K N 9.1 4.9-14.6 1.90 98.3 . NO x -N 2.7 0.1 -5.7 1.62 Ortho-P 6.1 2.8-15.3 2.89 95.0 T-P 7.4 3.1-22.6 4.48 95.1 T S (g/L) 0.7839 0.27- 1.467 0.419 84.7 TVS (g/L) 0.2641 0.11-0.6 0.150 93.1 SS (g/L) 0.0864 0.005 - 0.75 0.209 97.4 VSS (g/L) 0.0864 0.005 - 0.75 0.209 97.3 ** s: Soluble 122 TVS, SS and VSS was 5.1167 g/L, 3.8014 g/L, 3.3875 g/L and 3.225g/L, respectively. Figure 5.33 shows that the variation in pH of influents and effluents measured during the operation of TSSBR#1 and #2 was within a neutral range (average 6.8), ranging from 6.33 to 7.53. The final effluents had slightly alkaline-pH values, with an average of 8.21 and a range of 7.72 to 8.67. Although the influent pH range fell within the neutral range, the influent contained sufficient alkalinity (averaging 1526 mg/L as CaCC>3) for complete nitrification to take place. Therefore, there was no need to supplement the system and no additions or adjustments were made. Figure 5.34 shows the correlation between influent alkalinity and NH4-N data. As noted in this graph, the influent alkalinity trend is similar to that for NH 4 -N. Figure 4.8 (Chapter 4) also illustrates this correlation between ammonia-N concentration and alkalinity for a very different quality of swine wastewater. Since alkalinity is a measure of the aggregate properties of water, the measured alkalinity, using a pH-end point, should include contributions from phosphates and ammonia-N or other bases if these are present in the wastewater, as well as the carbonate, bicarbonate and hydroxide contents as shown in Equation 15: Measured alkalinity = [ C 0 3 2] + [HCO3 ] + [H 2 C0 3 ] + [NH3] + [ H P 0 4 2] + [other bases] - [IT] + [OH] (15) A batch test was conducted in order to estimate the alkalinity contributed by the N H 4 - N levels in the swine wastewater. A toxicant (2 ml of chloroform) and nitrification inhibitor (0.5 g) were mixed with swine wastewater (1L) to inhibit biological reaction during the batch test and then the pH was adjusted to 10.8, using NaOH for ammonia air-stripping. During the air-stripping, samples were taken from the beaker and measured for residual NH4-N and alkalinity concentrations. The measured NH4 -N and alkalinity concentrations were deducted from those at 123 10 9.5 9 8.5 8 £ 7.5 7 6.5 6 5.5 -Influent pH -effluent pH Time (days) Figure 5.33. The variation of pH in influent and effluent -Influent alkalinity -Influent NH4-N 800 700 600 500 "3) C 400 C Z X 300 z 200 100 0 Time (days) Figure 5.34. The correlation of influent N H 4 - N with alkalinity 124 time zero, in order to calculate the amount of air-stripped NH4-N and the alkalinity reduction. The data was plotted and is illustrated in Figure 5.35. As found in the coefficient, 100 mg of ammonia contributes 237 mg of alkalinity (as CaCOs). 1000 1 : 1 900 350 400 Removed NH4-N (mg/L) Figure 5.35. Alkalinity by NH4-N in wastewater 5.1.3.3.2 Removal Efficiencies and General Discussion (RTC 3) The overall removal efficiencies achieved were as high using the operating strategy of TSSBR#2 as in that of TSSBR#1. Real-time control of TSSBR #2 also produced a consistent effluent quality in spite of the influent fluctuation. The small range of standard deviation reflects this consistency. The removal efficiencies and effluent quality are summarized in Table 5.5. A high organic materials removal efficiency was achieved with a 97.9 % TOC removal. The TOC profile for the days of operation is shown in Figure 5.36. The average TOC concentration in the effluent was 52 mg/L, varying from 38 mg/L to 64 mg/L. With respect to biochemical reactions, the organic carbon was removed by bacteria assimilation and oxidation in the A/O reactor and assimilation in the anoxic reactor. There was a concern that phosphorus removal efficiency would deteriorate in the TSSBR#2 reactor due to the release of P from the addition of sludge to the anoxic reactor. 125 5000 -i 143 145 147 149 151 153 155 157 159 161 165 167 169 171 Time (days) Figure 5.36. Track of T O C s in influent and effluent (TSSBR#2) Figure 5.37. Track of phosphorus in influent and effluent (TSSBR#2) 126 However, no significant P release was observed in the anoxic reactor. Instead, continuous P uptake occurred, resulting in high removal efficiencies. The average Ortho-P and total P removal efficiencies were 95 % and 95.1%, respectively. The phosphorus profile over time is presented in Figure 5.37. The highest removal efficiency observed was for NH4 -N, with 98.6 % removal efficiency and an average of 2.9 mg/L NH 4 -N in the effluent (range from 0 - 4.9 mg/L). This high removal efficiency was again achieved by the real-time operation. An average 98.3 % T K N removal efficiency was also observed. The nitrogen was removed by bacterial assimilation, ammonia-N air stripping and nitrification in the A/O reactor and by assimilation in the anoxic reactor. The average NOx-N concentration in the final effluent was 2.7 mg/L, varying from 0.1 to 5.7 mg/L. The NH4 -N, T K N and NOx-N profiles with time are represented in Figure 5.38. The average TS, TVS, SS and VSS removal efficiencies were 84.7 %, 93.1 %, 97.4 % and 97.3%, respectively. While high removal efficiencies were achieved for TVS, SS, and VSS, a relatively low TS removal was observed. This may be due to the fact that the influent wastewater contained such a high portion of non-organic solids (inert solids) in colloidal form. The solids levels in influent and effluent are shown in Figures 5.39 and 5.40. Batch tests were performed with TSSBR system and beakers to ascertain the biological mechanisms at work in both the aerobic and anoxic phases of the A/O reactor. Figures 5.41 and 5.42 represent the trends for phosphate and organic carbon for the anoxic phase. Figure 5.41 shows a low P uptake with the occurrence of simultaneous denitrification. The P uptake is attributed to the accumulation of poly-P or denitrifying bacteria. Based on review paper by Barker and Dold (1996), P uptake by poly-P bacteria, using nitrate as the terminal electron acceptor, occurs under anoxic conditions and the P uptake rate is lower than that under aerobic conditions. A fraction of the poly-P bacteria can also denitrify with simultaneous P uptake and 127 Figure 5.38. Track of nitrogen in influent and effluent - R T C 3 128 12 10 ID o m c O r-• l \ t ( [ 1 0 ) 0 ( N t ( D C O O O I 1 ' c S i c W T f L o ' i ^ c b r ^ r ^ c o o i d i o r r r r r i - r » - T - » - T - T - r r P J ( N N M t M Time Figure 5.47. O R P and pH profile of T S S B R # 3 139 on the ORP curve was masked by the rapid descent of redox potential to -300 mV in RTC 4 and -280 mV in RTC 5, during influent loading. During normal real-time control (TSSBR#1, 2 and 3), the nitrate knee was never observed on the ORP profile over time in the A/O reactor. The nitrate knee was, however, observed on the ORP curve when the influent feeding was completed before the redox potential reached -180 mV (Figure 5.53 discussed later). If complete denitrification did not occur during the air-Off period, instead of P release from bacteria, P uptake using nitrate as an electron acceptor would continue. If this had been the scenario, high phosphorus removal would not have occurred during the operation. In the achievement of real-time control, the ORP moving range pattern was used. The sampling rate and parameters (r) were 1 min. and 7, respectively. The moving range pattern of ORP under real-time control using the NBP is represented in Figure 5.48. With regard to the moving range pattern, a very high ORP moving range was obtained at the start of aeration. The moving range dropped at the B point, then rose higher again at the NBP. This was the point used for real-time control of TSSBR. As aeration was terminated at the NBP and influent loading followed, a very high moving range resulted. This change was quite similar to that between points A and B. Figure 5.48 shows that D point, which was detected under real-time control of TSSBR using RCMP (#1,2 and 3) did not occur, because all aerobic reactions were artificially terminated before this point occurred, when the NBP was used as the real-time control point. The designated moving range values used to make the computer recognize each point in sequence were M R G > 50 (for A point), M R G < 5 (for B) and M R G > 15 (for NBP), respectively. Using this control technology, successful real-time control of TSSBR was achieved. Figure 5.49 shows the consistency of the ORP moving range, ORP and pH patterns obtained under real-time control. The distinctive real-time control point (NBP) was always observed on 140 0 1 i f i ^ S t O ( D N l I ) t O ( D N O ) \ t O l D C > i a 3 t O ( D N ( D t O ( 0 ( N C O \ t O C O f N l t O C O c b c f t d l O O O ^ ^ C N C N ^ ^ T - T - T - T - T - T - T - T - T - T - ^ ^ ^ ^ T - ^ T - T - ^ T - ^ C N C N I C N C N C N C N C N Time (hrs) Figure 5.48. The pattern of O R P moving range (r=7) in T S S B R #3 141 Hd CD 0) C C 1 1 ® CO < <1> ) U . CN C ' > O Q- \ IE \ O \ 82:Z iO'V o o e es I-9t?:0 6e:ez iZZZ tt-iZ LVOZ 0161 60:81. 99 :9t 6^91 Z t - n 8221 t z : u H U l Z0:6 00:8 299 9t>9 ie:e fr22 Z U: I. Ol-O Z0E2 SS:i-J 8fr02 « ; : 8 l Z2:^l 03^91 e r s i 90 > I. 69:21. 29:1.1. IZS 0£8 tZl 919 609 2 0 ^ 99:2 CO *fc 01 CD. CO CO _0> O i _ Q_ X Q_ T3 C CO Q _ or O 5^ o c CD -*—< CO '(f) c o O ai T* uri CD 13 O) CM TT O O CM O O O O O O CM O O CO O O T f 9UIAI?(AW) dUO the ORP, moving range and pH curves. A detailed description of pH and a discussion of real-time control technology using pH trends are given in Section 5.1. Figure 5.50 and 5.51 show the two types of ORP curves which were observed in the anoxic reactor, under real-time control of TSSBR#3. Each figure was monitored during the operating period from day 203 to 204 (Figure 5.50) and from day 209 to 210 (Figure 5.51). Figure 5.50 illustrates that the nitrate knee was not observed for cycles 1, 2, 6 and 7. A relatively high redox potential was maintained at this time. The nitrate knees observed on cycles 3 to 5 occurred at higher ORP values than did those in Figure 5.51. When complete denitrification was observed for every cycle as shown in Figure 5.51, the redox potential in the anoxic reactor was kept between -200 and -400 mV. This lower potential suggests that the reactor achieved a highly reduced status. Under these conditions, a real-anaerobic condition would pertain, resulting in the anaerobiosis of particulated organic carbon materials. The solublized carbon materials could be used to achieve complete denitrification in the subsequent cycle. With regard to the C/N ratio of the influent (as TOCs/TKN), the C/N ratio for the days shown in Figure 5.50 and 5.51 were 4.5 and 5.8 respectively. This indicates that a larger carbon source was available for denitrification during the second period than the first, since the carbon oxidation in the A/O reactor was artificially finished at the point of complete nitrification. However, data analysis reveals that there is no significant correlation between the influent C/N and the effluent NOx-N. Given this, it is surmised that the denitrification efficiency of the anoxic reactor may not just be dependent on the C/N ration of the influent; it may also be dependent on the carbon oxidation level in the A/O reactor, as well as on the biochemical activity of bacteria in both reactors (a daily variable). During the anaerobic conditions maintained after complete denitrification as shown in Figure 5.51, no significant phosphorus release was observed. This could be explained by a 143 100 -| Time Figure 5.50. O R P profile of anoxic reactor(TSSBR#3)-incomplete denitrification 100 -I Time Figure 5.51. O R P profile of anoxic reactor (TSSBR#3)-complete denitrification 144 number of facts. First, since the bacteria for the reactors were completely separated during the operation of TSSBR#3, the phosphorus which was taken up in A/0 reactor could not be released in the anoxic reactor. Second, since an insufficient source of readily biodegradable carbon existed in the anoxic reactor after complete denitrification, the P release rate would be very low. Third, since insufficient P release occurred in the anaerobic condition, the P uptake rate, using nitrate as an electron acceptor, would be low in the subsequent cycle. The low P uptake would again result in low P release. (Figure 5.61, discussed later, indicates that no significant P release can be assumed to have occurred in the anoxic reactor). In the operating period from day 209 to day 210, the NOx-N concentration in the final effluent reached zero and hence P-release conditions were provided. The P content in the final effluent did not increase, however, and a constant quality of effluent was maintained. Except for those two days represented in Figure 5.51, the rare achievement of zero NOx-N in the final effluent from TSSBR # 3 meant that in contrast to P release, a low rate of P uptake or assimilation due to bacterial needs occurred, resulting in lower P concentrations in the final effluent. Fig. 5.52 Ortho-P release in Anoxic reactor 145 In order to examine P release in the anoxic reactor, a batch test using a 500 mL beaker was conducted over 11.5 hours. The P trend is represented in Figure 5.52. As the figure illustrates, P uptake and release in the batch test was less than 1.5 mg/L and the P release rate was much lower than the rate of P uptake, with P uptake-at 0.128 mg/L.h.g (MLVSS) and P release at 0.062 mg/L.h.g (MLVSS). In the anoxic reactor, the P uptake and release was 0.36 mg/L.h.g (MLVSS) and 0.038 mg/L.h.g(MLVSS), respectively. 5.2.2 General Observation It is useful to examine the bio-chemical reactions which occur in each phase of the A/O and the anoxic reactors under real-time control using the NBP. To do this, the nutrients (C, N and P) were tracked, monitoring the ORP curve carefully. It should be noted that control of the loading rate using the ORP set-point was not done during this track study. In its place, a fixed volume of various strengths of influent (IL) was loaded into the system while maintaining real-time operation. This simulated the conditions of a variable loading rate and made it possible to examine the general importance of loading rate control in a combined C, N and P removal process. Figure 5.53 shows the nutrient trends and biochemical reactions which occurred in the A/O reactor, when the process was subjected to the simulated variable loading rate conditions. The figures shows that the ORP values at the time of complete influent feeding were approximately -100 and -160 mV, which confirms the achievement of variable loading rate conditions. The nutrient concentrations after influent feeding were as follows: after the first influent feeding - 170 mg/L (TOCs), 11.2 mg/L (NH4-N) and 35.3 mg/L (Ortho-P); after the second influent feeding - 204 mg/L (TOCs), 13.1 mg/L (NH4-N) and 56.7 mg/L (Ortho-P), respectively. Consideration should be given to the idea that when treatment plants are operated with a fixed-volume of influent loading (or in an operating mode using fixed HRT), the real 146 MLSS = 10.6 g/1, MLVSS = 8.65 g/1. 12:40 12:52 13:04 13:16 13:28 13:40 13:52 14:04 14:16 14:28 14:40 14:52 15:04 15:16 15:28 15:43 15:55 16:07 16:19 16:31 16:43 16:55 17:07 17:19 17:31 17:43 17:55 18:07 18:19 18:31 Time (hours) ORP - s — TOCs/3 Ammonia-N —x— NOx-N - a — Ortho-P Figure 5.53. Track of A / O reactor (TSSBR#3) 147 influent loading rate into system should be a fluctuating one. This would result in variable nutrient levels in the system (depending on the variability of the influent wastewater) as represented in this figure. This is important as variable loading rate conditions affect the biological reactions in a system. For instance, when a high loading occurs on a particular day (or in a particular cycle), a high NOx-N level can occur in the system at the end of the aerobic phase. This increased NOx-N level can result in incomplete denitrification or in a retarded state of denitrification, since a longer time is usually required to obtain complete denitrification of high levels of NOx-N. In addition, when quite a low loading rate occurs, complete denitrification may not happen during the fixed air-off period, due to a deficiency in organic carbon materials in the loaded influent. The occurrence of incomplete and/or retarded denitrification, under fluctuating loading conditions, can result in no or insufficient phosphorus release during the fixed air-off period. This, in turn, can lead to ineffective phosphorus removal during the aerobic period. Figure 5.53 can be seen as a representation of these conditions. The figure shows that when a low strength influent was fed at high NOx-N conditions (over 40 mg/L), not enough phosphorus release was induced during the air-off period. Instead, most of this air-off period was used for denitrification. The continuation of P uptake, using nitrate as an electron acceptor, was observed during denitrification. The P release began after the occurrence of the nitrate knee, which means complete denitrification of NOx-N. As the readily biodegradable organic carbon material was used for denitrification, the TOC concentration decreased during denitrification. However, an increase in TOC concentrations, due to organic carbon hydrolysis, was observed after complete denitrification. This suggests that the carbon hydrolysis rate might be less than the carbon assimilation rate during denitrification. An alternative scenario could be that the nitrate repressed the organic carbon hydrolysis. In terms of the Fuhs and Chen (1975) theory, the nitrate 148 caused the cessation of acidogenesis. The observed lack of change in NH4 -N concentrations during the air-off period suggests that the ammonia-N assimilation rate was equal to the ammonification rate of organic nitrogen. At the start of aeration, organic carbon oxidation, nitrification and phosphorus uptake occurred. When the NH4 -N in the A/O reactor reached zero (was completely nitrified), aeration was terminated by the computer and further organic carbon oxidation was prevented. The residual carbon materials were then available for denitrification in the anoxic reactor. With a low loading rate, the resulting NOx-N levels at the end of the aerobic period were much lower. As sufficient influent was fed at this low NOx-N level, denitrification of NOx-N was completed quickly and sufficient anaerobic conditions were provided for phosphorus release to occur. Based on the batch test described in section 5.1.3 (Figure 5.42), a minimum 2 hours of P release time should be provided under real-time control. This means that, in order to obtain adequate P release during the fixed air-off period (3 hours), the C and N loading rates should be controlled. For successful real-time control, the aerobic period should also be regulated so as to be longer than the air-off period. This could be achieved by controlling the influent loading rate. If excessively low loading rates are maintained under real-time control, the aerobic period will be terminated quickly. This results in the aerobic period being shorter than the air-off period. Since the growth rate of heterotrophic bacteria is much greater than that of autotrophic bacteria (i.e. nitrifier), if these conditions are maintained over a long term, the heterotrophic bacteria will dominate the system. This would result in deterioration of the nitrification rate. Figure 5.54 shows the nutrient trend and the ORP pattern in the anoxic reactor. As liquor from the A/O reactor was fed into the anoxic reactor, the NOx-N, TOCs and Ortho-P concentrations in the reactor increased. The NOx-N levels in the liquor transferred into the anoxic reactor were proportional to the influent loading rate into the A/O reactor achieved in the 149 MLSS = 5.4 g/L, M L V S S = 4.3 g/L 12:40 12:52 13:04 13:16 13:28 13:40 13:52 14:04 14:16 14:28 14:40 14:52 15:04 15:16 15:28 15:43 15:55 16:07 16:19 16:31 16:43 16:55 17:07 17:19 17:31 17:43 17:55 18:07 18:19 18:31 18:43 Time (hours) Figure 5.54. Track of Anox i c reactor (TSSBR#3) 150 earlier cycle. The reaction time required for complete denitrification in the anoxic reactor was also a function of NOx-N levels. A sharp increase in redox potential was observed at the time of the liquor transfer due to the response of ORP to the entry of the highly oxidized solution. A distinctive nitrate knee was observed on ORP curve, indicating the NOx-N level in the reactor reached zero. With respect to the biological reactions, the residual organic carbon material in the reactor was assimilated during the denitrification of NOx-N. After complete denitrification, the organic carbon concentration increased slightly. This can be attributed to the solublization of organic materials trapped in the slime of the bacteria floe or to the hydrolysis of particulated organic materials. No change in N H 4 - N levels was observed. Further batch tests were undertaken in order to get additional insight into the biochemical reactions in TSSBR. The biological nutrient removal (C, N and P) trends during the aerobic phase were tracked and are illustrated in Figure 5.55. The graphs in Figure 5.55 illustrate the results of two different sets of experimental conditions (A and B): In experiment A, a two hour P-release period was provided after complete denitrification. The nutrient levels were then measured over time under aerobic conditions. In experiment B, only 30 minutes P-release time was provided. The phosphorus removal trends in both graphs show the same pattern of kinetics. At the beginning of the aerobic phase, the phosphorus uptake rates were governed by zero-order kinetics. As the pool (poly-phosphate reserves) was filled up, phosphorus removal decreased gradually and appears to have shifted from zero-order to higher order kinetics. The numerical phosphorus removal rates are limited by the phosphorus release levels which were achieved in the earlier anoxic phases. When sufficient phosphorus release was obtained under anoxic conditions, the phosphorus uptake rate in the subsequent aerobic phase was much deeper than when insufficient P release occurred. The N H 4 - N removal patterns reveal similar trends to those 151 Time (hrs) Figure 5.55. C , N and P removal in aerobic phase of A / O 152 of P removal, which is a high NH4-N removal rate governed by zero-order kinetics taking place at the initial stage, followed by a slow down in NH4 -N removal. With regard to the removal of organic carbon material, carbon oxidation continued in the aerobic phase after the complete removal of soluble ortho-P and NH4 -N. Since, under real-time process control, carbon oxidation was artificially terminated at the end of nitrification, the residual (unoxidized) organic carbon material should be used for denitrification of NOx-N in the anoxic reactor. This should result in a lowered organic carbon content in the final effluent. As has already been mentioned, during both anoxic and aerobic conditions, phosphorus removal can occur together with nitrate removal. Nitrate can serve as an electron acceptor for the oxidation of stored PHB, resulting in phosphate uptake, using the energy produced by the oxidation of PHB. However, nitrate may not be as efficient as oxygen for P uptake (Barker and Dold, 1996., Sorm et al., 1996). This is indicated in Figure 5.56, showing phosphorus removal over time. Changes in the slope show that anoxic P uptake appeared to occur at a slower rate under anoxic conditions than under aerobic conditions. Although the P uptake rate under anoxic conditions is strongly influenced by the amount of nitrate (Sorm, 1996) and by the type of readily biodegradable materials (Barker, 1996), it could be postulated that aerobic P uptake rates would be higher than anoxic P uptake rates. This would be attributed to the poly-P bacteria mass being comprised of two groups; one which can utilize either nitrate or oxygen as an electron acceptor, and one only able to use oxygen (Barker, 1996). The observed simultaneous NOx-N reduction which occurred under anoxic conditions could be attributed to denitrifiers and to some poly-P bacteria, which are capable of denitrification. The length of the aerobic phase in the A/O reactor in RTC4 and RTC5 was tracked for 7 days and the results are represented in Figure 5.57 and 5.58, respectively. The lower redox potential was used for influent loading rate control, a higher influent loading rate per cycle was 153 Figure 5.56. Phenomena of phosphate uptake under anoxic and aerobic 154 O) c © o >. o u !5 o 12 10 8 £ 4 a> < ORP setpoint: -300mV AVG loading rate/cycle: TOC - 1709 mg TKN - 365 mg AVG aerobic cycle length (4.12 hrs) oo a> CD o o CN o CM CN O CN CO o CN days Figure 5.57. Aerobic cycle length in A / 0 reactor for 7 d a y s - R T C 4 12 10 c 0> o >» o o !o o k_ re ORP setpoint: -280mV AVG loading rate/cycle: TOC -1620 mg TKN - 309 mg AVG aerobic cycle length (3.75 hrs) in CN in m CM co in CN CO L O C N L O C N O C O C N days Figure 5.58. Aerobic cycle length in A / O reactor for 7 d a y s - R T C 5 155 achieved and the aerobic phase lengths were proportional to the loading rate. In RTC 4, which was operated with -300mV of ORP value as an influent loading rate control point, the average length of the aerobic phase was 4.12 hours. A slightly shorter aerobic period of 3.75 hours was obtained in RTC 5, which had -280mV as the influent loading rate control. On a daily basis, a slightly lower number of operating cycles was obtained in RTC 4 (3.37 cycle/day) than in RTC 5 (3.55 cycle/day). The average TOCs and T K N loading rate were 1709 mg TOCs/cycle and 365 mg TKN/cycle in RTC 4 and 1620 mg TOCs/cycle and 309 mg TKN/cycle in RTC5, respectively. The solids concentrations maintained in the A/O reactor during the operation of RTC 4 and 5 were higher than the ones recommended (1.5 - 5.0 MLSS g/L) by Metcalf and Eddy for typical SBR operation. The solid levels in the anoxic reactor were, however, within the recommended values. The average MLSS concentrations inside each reactor were 6.4712 g/L (A/O) and 3.5988 (anoxic) in RTC 4 and 8.8768 g/L (A/O) and 3.5216 g/L (anoxic) in RTC 5. The VSS/TSS ratio was constant in both runs, 0.83-A/O and 0.81-anoxic in RTC 4; 0.82-A/O and 0.82-anoxic in RTC 5. The F/M ratios were twice as high in RTC 4 as in RTC 5: 0.14 and 0.08 (in applied TOCs/MLVSS.d), respectively, based on the MLVSS in the A/O reactor. Calculated using the applied BOD 5s/MLVSS.d, the F/M ratios were 0.26 d"1 in RTC 4 and 0.13 d"1 in RTC 5. The higher F /M ratio in RTC 4 is attributed to the higher loading rate achieved by using lower redox potential for loading rate control and to the lower MLVSS concentrations inside the A/O reactor. The SRT and HRT maintained during the operation of RTC 4 were 10.2 days and 4.1 days, respectively. A similar retention time pertained for RTC 5; a 9.4 day SRT and 4.06 day HRT. A relatively shorter HRT was obtained in the real-time control of TSSBR with NBP (RTC 4 and 5) than for those with RCMP (RTC 1, 2 and 3). While, the length of the HRT achieved is 156 also dependent on the influent strength (loading rate), the shorter HRTs which were achieved in these operations, can be attributed to the NBP occurring earlier on the ORP curve than the RCMP. During the operation of TSSBR#3, the settling properties of the activated sludge in each reactor were measured by means of SVI. Both are depicted in Figure 5.59, together with those for TSSBR#2. In general, the sludge in both reactors settled very rapidly and no significant sludge blankets were observed. These settling characteristics resulted in low SVI values, which averaged 32 ml/g in the A/O reactor and 52 ml/g in the anoxic reactor. A comparison of the settling properties of the two reactors show that the SVI values of the anoxic reactor were approximately 1.6 times higher than those of the A/O reactor. This might reflect the different characteristics of the sludge in each reactor, since the SVI value usually varies with the concentration of mixed-liquor solids, the specific gravity of the suspended solids and the Figure 5.59. Settling properties of sludge in A/O and anoxic reactor 157 microbial composition of the floe. The same figure also shows that no significant difference was found between TSSBR#2 and #3. The SVI-values were as follows; A V G 53.8 ml/g (Anoxic) and 36.6 ml/g (A/O) in TSSBR#2; 52.6 ml/g (Anoxic) and 27.4 ml/g (A/O) in TSSBR#3. Therefore, it seems that the intermittent addition of sludge from the A/O reactor into the anoxic reactor, which was undertaken during the operation of TSSBR#2, did not significantly affect the settling characteristics of the sludge in the anoxic reactor. 5.2.3 System Performance of TSSBR#3 5.2.3.1 Wastewater Characteristics The characteristics of the influent swine wastewater used for TSSBR#3 are summarized in Table 5.6. Selected statistics for the influent organic carbon, nitrogen, phosphorus and solids levels measured during both operations (RTC 4 and 5) are presented in this table. The influent total soluble organic carbon concentration (TOCs) was higher in RTC 4 than in RTC 5, with an average of 2126 mg/L in RTC 4 and 1327 mg/L in RTC 5. Table 5.6 shows that a larger standard deviation was observed in RTC 4. However, the standard deviations in the operation of TSSBR #3 (RTC 4 and 5) were much smaller than those in TSSBR#1 and #2, indicating that the influent quality was less variable in TSSBR #3 than in TSSBR#1 and #2. Average influent T K N was 391 mg/L in RTC 4 and 287 mg/L in RTC 5, with about 41.7 and 42.9 percent, respectively, existing in the form of NH4 -N (the ammonia-N levels in RTC 4 and RTC 5 were 163 and 123 mg/L, respectively). The nitrate levels in the influent were relatively constant during both runs, approximately 1.6 mg/L. The total phosphorus concentration averaged 121 mg/L in RTC 4 and 88 mg/L in RTC 5, respectively. In RTC 4, approximately 81 % of the total-P was in the form of soluble ortho-P (average 98 mg/L), while in RTC 5, approximately 73 % of T-P was in the form of soluble ortho-P (average 73 %). The composition of TSSBR # 3 influent was quite different from TSSBR#1, 158 c o H 1) =3 0) S3 o s-1) >n > Q in H ^ U '. OS .s Q m * . s u 1) i 1 O N i n O N MD MD 1—H i od | C O O N © O N o CN 1.02 0.91 1.19 1.12 CN MD T l -O T T cn: i n CN r- I >n cn \ © O cn; C N < CN j ~ MD c o CN MD O N i *—< C O i © C O o O C N o o o o i n i n o o O N C N C O CN C O o •n MD o o <—1 o O T f MD MD C N T—1 © © C O O o o \ m h i H i m MD T f C O MD O N C O MD O O i C O O O O : C O i n i n i n ; i n r-; O N 00 00 MD CN CN 0 f - i n i n ! O MD r- r-! °) C O 0 O \ O O i ,_ 0 C O MD 1 C N O N C O : »—< 0O O O MD ! 0 00 T f ! ^ CN ~5x)| m 1/1 > ON IT) but similar to that of TSSBR#2. The relative C:N:P ratio (as TOCs:TKN:T-P) were approximately 100:18.4:5.7 in RTC 4 and 100:21.6:6.6 in RTC 5, while the ratio in TSSBR#1 were 100:66.9:52.2 in RTC 1 and 100:40.7:21.1 in RTC 2. Based on BOD 5 calculated using equation 14, the relative N and P ratio to BOD5 was much higher than the ratio recommended for proper biological treatment by Metcalf and Eddy (1991). It was 100:9.8:3.0 (C:N:P) in RTC 4 and 100:13.4:4.1 in RTC 5, respectively. The total solids in the influent was composed of 37 % SS and 63 % colloidal solid in RTC 4, and 40 % SS and 60 % colloidal solids in RTC 5. The TS contained high levels of inorganic matter which could not be volatilized: 0.72 (TVS/TS) in RTC 4 and 0.74 in RTC 5. The majority of SS in the influent was, however, organic material: 0.99 (VSS/SS) in RTC 4 and 0.93 in RTC 5. These solids characteristics suggest that a high portion of the inorganic solids was in the form of micro-particles (colloidal solids) which could pass the filter paper. 5.2.3.2 Removal Efficiencies and General Discussion The treatment efficiencies and effluent quality in RTC 4 are presented in Table 5.7. A high organic carbon removal efficiency was achieved (approximately 98 %). TOC levels dropped from 2126 mg/L in the feed to 53 mg/L in the effluent. Prior to the operation of this treatment system, there was concern that the effluent organic carbon levels would fluctuate with the influent C/N ratio, since the organic carbon oxidation in the A/O reactor was artificially terminated at the NBP. No correlation between effluent TOCs levels and influent C/N ratio was found, however, and relatively low effluent TOC levels were obtained during the whole operation. Also, there was no correlation between influent C/N ratio and the NOx-N levels in the effluent. These observations indicate that the residual organic carbon levels in the liquor transferred into the anoxic reactor, from the A/O reactor at the NBP, did not significantly exceed the carbon requirements for complete denitrification within the influent C/N variation range of 160 % removal 97.5 99.6 98.1 | 97.4 97.9 77.8 90.2 ?6:8 96.8 Std. Dev. 14.37 0.52 1.81 0.94 0.55 1.03 0.20 0.08 0.02 0.02 Effluent Min. - Max. 37-75 O O 1 o 3.9-11.6 1.6-5.5 oo O N O O 0.583 - 1.167 0.140-0.400 0.020 - 0.090 0.020 - 0.090 Means m r>-o m r-^ CN CN CN 0.8924 0.2825 0.0479 0.0475 Influent Means 2126 m ON m 1.60 00 O N CN 4.0121 2.8890 1.4833 1.4646 Parameters (mg/L) * * O O H TKN I X o Ortho-P T-P TS (g/L) TVS (g/L) SS (g/L) VSS (g/L) 4.2 to 6.9 tested here. However, as shown in Figure 5.60, which shows effluent quality under varying influent conditions, the trend was for effluent TOC levels to increase slowly with the progress of time. This effluent TOCs trend is understood to correlate with the effluent solid's trend. This could also be attributable to the accumulation of organic carbon in the anoxic reactor. Since the real-time process control was achieved using the NBP, more organic carbon material than was required for denitrification could have been left in the reactor and resulted in the trend for TOC levels to increase in the final effluent. Figures 5.61 and 5.62 illustrate the phosphorus and nitrogen trends in the influent and effluent. In general, the phosphorus and T K N contents in the influent follow a trend similar to that of TOCs. As can be seen, relatively low and constant levels of T K N and phosphorus were obtained in the effluent, despite variable levels of influent T K N and phosphorus. The average soluble ortho-P and total-P levels in the final effluent were 2.5 and 2.6 mg/L. This resulted in very high removal efficiencies (97.4 % and 97.9 %, respectively). The removal efficiencies for T K N and NH 4 -N were 98.1 and 99.6 %, respectively. Levels of less than 1 mg/L of NH4-N were obtained in the final effluent (ranged from 0 mg/L to maximum 1.8 mg/L). This indicates that the complete removal of ammonia-N should be possible using real-time control strategies, since complete nitrification is always ensured in such a system. The average T K N in the effluent (7.3 mg/L) may in part be due to solids in the effluent, as the T K N measurement was not done using a filtered sample. During these experiments, the final effluent contained an average of 47.9 mg/L suspended solids and 99 % of the SS was VSS. Average NOx-N levels in the effluent were lower than those obtained in TSSBR#1 and #2 where RCMP was used for real-time control. The TSSBR #3 levels averaged 2.7 mg/L of NOx-N concentration in final effluent. The lower NOx-N levels may indicate that more organic carbon matter was available for denitrification when the process was 162 Figure 5.60. Track of T O C s in influent and effluent (RTC4) 163 6 i Time (days) Figure 5.62. Track of nitrogen in influent and efffluent-RTC#4 164 controlled using the NBP, than when controlled with RCMP; this expectation reflects the fact that carbon oxidation in the A/O reactor was artificially terminated at the NBP in TSSBR#3, while organic carbon oxidation continued beyond the NBP in TSSBR#1 and #2. Also, the higher C/N ratio in TSSBR#3, compared to TSSBR#1 and 2, could have contributed to high levels of denitrification. With regard to NOx-N removal in the anoxic reactor, the denitrfication efficiencies could have been affected by several factors, such as the amount of organic carbon material, the biomass level, rate of biological activity, and the hydraulic residence time in the anoxic reactor. The variations in solids levels in both the influent and effluent are presented in Figure 5.63 and 5.64. The TS in the effluent was composed of 5 % SS and 95 % of colloidal solids, which passed through the filter paper. Since the colloidal solids in the influent contained high proportions of inorganic solids (inert solids), which would not biodegrade but would instead pass through the treatment process, the result would have been high levels of inert solids in the final effluent. This could have led to the lower TS removal efficiencies (approximately 78 %). While the effluent TS contained a high proportion of inert solids (TVS/TS = 0.32), the SS was all organic solids which could be volatilized at burning temperatures (VSS/SS = 0.99). The removal efficiencies of TVS, SS and VSS were 90.2 %, 96.8 % and 96.8 %, respectively. Table 5.8 documents the effluent quality and removal efficiencies in RTC 5. The removal efficiencies obtained for organic carbon, nitrogen and phosphorus were very high, as in RTC 4. With TOCs removal efficiencies of over 96 % being obtained, quite low effluent TOCs levels were achieved and maintained during the operation, averaging 49 mg/L (with a range from 19 to 73 mg/L). The effluent and influent TOCs levels were tracked on a daily basis and the results are represented in Figure 5.65. The results show that the TOCs levels in the final effluent increased over time, as observed in RTC 4. The effluent T K N trend was correlated with the effluent TOCs 165 10 • Influent TS • Influent TVS A Influent SS X influent VSS • • X • X • • • X • X • • • • • • X X • • X - X -l O C D r ^ C O O O T - C N C O T T L O C D r ^ O O C n O T - C M C O ^ - L O t D r ^ O O C n O T - t N C O < D C n o > 0 ) C 5 0 0 0 0 0 0 0 0 0 0 T - T - T - ^ ^ T - T - T - T - T - C M C N t N C N i < - - I - T - T - T - C \ 1 C N C N C \ I C N C N J C N C \ | C N C N C N C N C \ I C \ I C N C N O I C S I C N Time (days) Figure 5.63. Variation of influent sol ids -RTC#4 • Effluent TS • effluent TVS A Effluent SS X effluent VSS • - X -• • • ft ft , - CN CO LO CN CN CN CN CN JJL • _X K • X ^ - C N C 0 " * t L 0 C D r ^ C 0 C T ) O ^ - C N C 0 T - ^ - T - T - T - T - x - T - T - C N C N C N C N C N C N C N C N C N C N C N C N C N C N C N C N C N Time (days) Figure 5.64. Variation of effluent sol ids-RTC#4 166 i n O H & co & C Q GO GO H _c _ o a a* O s ii > Q +-» GO e , 3 E e s C O r~- O N ! CN C N 0 0 M O ; C O »—< C N 1 — 1 O N C O i n C O O N r - T f i n O N o MD O N r--T-H C O © CN C O C O C O o o C O O O r— r -O N i n o o © © © o C O o i n m C O i n C O C O MD o o © © © © o 0 0 • T—< MD i r- C O m C O o © I °' © © © ; r- O N © i MD CN m © T-H O N C O © | CN C O O N o o C O CN •X-Ui O! H i O o o H H O f i 00; GO; >! H ; ^) | GO GO > NO level. The phosphorus and nitrogen trends tracked in RTC 5 are presented in Figure 5.66 and 5.67, respectively. The use of a real-time control strategy meant that complete ammonia-N removal was always ensured. The average NH4 -N levels in the final effluent were less than 1 mg/L (0.7 mg/L) and the removal performance was over 99 %. The effluent T K N level and removal efficiency was 8.2 mg/L and 97 %, respectively. Denitrification levels, using residual carbon materials in the anoxic reactor, were quite high and resulted in low NOx-N levels in the effluent ranging between 1 to 4 mg/L. High P removal was also obtained, averaging 95.2 % soluble ortho-P removal and 96 % total-P removal. The low phosphorus levels in the effluent may indicate that the P uptake by bacteria was relatively complete at the NBP, where nitrification ends. The soluble ortho-P and total-P levels in effluent was in the range of 1 to 5 mg/L during the operation (Figure 5.66). The variations in influent and effluent solids contents are shown in Figure 5.68 and 5.69. The solids contents in the final effluent had similar characteristics to those in RTC 4 (TVS/TS = 0.43 and VSS/SS = 0.93). The average removal efficiencies for TS, VSS, SS and VSS were 76 %, 86.2 %, 97.1 % and 97.3 %, respectively. The feed wastewater quality and strength varied during the operation, since no effort was made to regulate influent quality. Despite this, consistently high system performance was obtained in each run. Although two different loading rate set-points were used, no significant difference was found between the runs. This suggests that nutrient removal in TSSBR#3 was ensured by use of the real-time control strategy, despite the different loading rates. As already stated in section 5.1.2.3, this is attributable to the reaction period, under real-time control, being self-adjusting to changes in the oxidation/reduction status of the system. The overall system performance of TSSBR#3 was as follows; 97 % - TOCs, 99.5 % - NH 4 -N, 98 % - T K N , 96.3 % -ortho-P, 97 %-total P, 77 %-TS, 88.2 %-TVS, 97 %-SS and 97 %-VSS. 168 Figure 5.65. Track of T O C s in influent and effluent ( R T C 5) Figure 5.66. Track of phosphorus in influent and effluent ( R T C 5) 169 Figure 5.67. Track of nitrogen in influent and eff luent-RTC#5 170 0 X • US • • • • Influent TS • Influent TVS A Influent SS X influent VSS • • X a 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 Time (days) Figure 5.68. Variation of influent solids -RTC#5 • Effluent TS • effluent TVS A Effluent SS X effluent VSS Time (days) • • • JL • J L • Figure 5.69. Variation of effluent solids -RTC#5 171 Figure 5.70 shows the percent N and P values of MLSS during the operation of TSSBR#3 (RTC4 and 5). The MLSS in the system contains high levels of N and P. The N and P contents of MLSS in the A/O reactor were higher than in the anoxic reactor; with the average T K N and T-P of MLSS in A/O being 7.06 % and 4.26 %. The A V G T K N and T-P of MLSS in the anoxic reactor were 6.66 and 3.49 %, respectively. Based on the results obtained through real-time control of TSSBR (TSSBR #1 to #3), this research has established the important role which control of the loading rate, using redox potential, can play in the treatment of wastewaters. This can be summarized as follows: 1) the biological activity in a system can be optimized by preventing the occurrence of a serious fluctuation of F /M ratio, as can happen in fixed time modes of operation; 2) the designation of low redox potential as a loading rate control set-point allows for sufficient organic carbon matter to be loaded, to accelerate NOx-N denitrification and P release; 3) by selecting the optimal loading rate point, the system capacity can be maximized; 4) an optimal balance between aeration rate and system oxygen demand can be achieved by controlling the loading rate; 5) under conditions of frequently fluctuating loading rates, the ideal ORP curve essential to real-time control cannot be maintained; 6) by controlling the loading rate, the length of the aerobic phase in the A/O reactor can be kept longer than the air-off phase; 7) by controlling the loading rate in a fixed-time mode of operation, a flexible HRT can be obtained, the length dependent on the wastewater characteristics. The advantages of operating TSSBR with real-time control strategies can be summarized as follows: 1) no organic carbon sources are required for denitrification; 2) no volatile acids such as acetate are needed for P release; 3) complete nutrient removal is always ensured; 4) optimization of energy usage should be possible; 5) a constant quality of effluent can be achieved, despite variations in the characteristics of the influent wastewater; 6) a specific 172 / 197 205 213 221 248 255 262 Date Figure 5.70. The content of nitrogen and phosphorus in sludge 173 nutrient can be targeted for removal; 7) a flexible HRT can be achieved, depending on the biological activity inside the system and the influent characteristics; 7) the sequence followed in the treatment process can easily be modified according to treatment objectives; 8) treatment capacity can be optimized; 9) the treatment efficiency is not dependent upon the system SRT. 5.3 Fixed-time Operation of TSSBR (TSSBR#4) 5.3.1 Operational Characteristics and ORP (pH) profile As described in the section of material and methods, real-time control using a specific point on ORP curve and influent loading rate control by an ORP set-point, were not conducted in this operation. The operation sequences were controlled with a fixed-time cycle, monitoring the ORP and pH patterns continuously. A constant volume of influent wastewater was fed into system, similar to a conventional wastewater treatment system, despite the variation of wastewater characteristics. The detail descriptions on operational mode was provided in Section 3.2.6.4. During operation of TSSBR#4, an ideal ORP curve was rarely observed. The monitored ORP and pH profiles with time are shown in Figures 5.71 and 5.72. These patterns of ORP and pH profiles should be caused by an imbalance between aeration rate and oxygen requirement, since an ideal ORP curve can not be obtained under such conditions. According to one research project aimed at the analysis of the circumstances wherein NBP occurs (Wasiak et al, 1994), the NBP was not detected on the ORP curve under a over-aeration rate or under-aeration rate. Also, as found in Figure 5.72, the responses of ORP to aeration was a function of influent loading rate; under a fixed aeration rate, as the nutrient levels in system increased, the B point, which reflects the influence of the introduction of aeration, occurred at lower values. With a constant influent volume loading into the system, the actual nutrient loading rate was dependent upon the influent wastewater strength. This fluctuating loading rate of influent, at a pre-adjusted aeration rates, could result in excessive over-aeration, when the influent strength was low /or excessive under-aeration, when the influent 1 7 4 Time Figure 5.71. O R P and pH profile with time under fixed-time fashion operation 175 Time > E. a. a. o m u ^ c O T - ^ ' i - ^ r r ^ o c ^ c j ) C N i n c o i n c O ' i - ^ r r - - - o c o u j r o ^ T - O O U ^ U ^ ^ C O C O C N ^ T - O i n u l T r ^ C T l C ^ C O C N * - * -T - T - 1 - C M C N C N T - T - T - T - T - O J C N CM co i- T r - o Ifi M- T CO CN CN ri u i N o i ^ t- »- r- T - CM OJ O CO CD r- O LO ri i n cb Time O P H effects (pH bending point) > a: o Time Figure 5.72. O R P and pH profile under a variable loading rate condition 176 contained a high level of nutrients. As low nutrient loading rates occurred due to a low wastewater strengths, as in Figure 5.71 (loaded TOC = 1275 mg/cycle and T K N = 200 mg/cycle), the distinctive NBP on the ORP curves disappeared. Although the DO pattern was not monitored in this graph, the pre-fixed aeration rate might have been high under the achieved low influent loading rate. Initially, the NBP was detected on the ORP curve, but the NBP totally disappeared as the low loading rate condition was continued. Figure 5.72 may also show the overall ORP and pH profile under a variable loading rate condition. Since the influent loading rate control was not conducted in this operation, the nutrient loading rate into the system was variable from cycle to cycle and the loading rate variation resulted in an unstable system condition, finally resulting in an unstable system performance. Under low loading rate condition (Figure 5.71 and first two cycles in Figure 5.72a), complete nutrient removal was achieved. However, when a higher loading rate was followed (TOCs = 2200 mg and T K N = 352 mg) by a sudden influent strength change, the nutrient removal was not completed during the fixed aerobic periods and this resulted in the accumulation of nutrients and a low redox potential (below -250 mV) in the system. As the low loading rate was achieved (approximately 1400 mg TOCs and 265 mg TKN) at this status, the ORP curve began to recover (Figure 5.72b); however the system condition was not restored to the optimal state, in spite of the maintenance of low loading rates for more than three days. Thus, complete nutrient removal was not obtained. Again, as a high loading rate condition was provided, due to high nutrient levels in influent (loaded TOCs = 3800 mg/cycle and T K N = 602 mg/cycle), the system performance completely deteriorated (Figure 5.72c). Under a condition of a fluctuating loading rate, a false NBP was observed during the fixed aeration period. As found in Figure 5.72 (b), a similar point to the NBP was observed on the ORP curve (6th, 7th and 8th cycles), but this point was identified as a pH bending point. 177 Figures 5.71 and 5.72 show the disadvantages of fixed-time fashion operation. In Figure 5.71, assuming the target nutrient was nitrogen, an extended aerobic condition beyond the NBP could result in higher operational costs, whereas in Figure 5.72, the nutrient removal was not completed during the fixed aerobic period, resulting in high nutrient levels in the final effluent. It is expected that the over oxidation or poor oxidation, under a fixed-time fashion operation, can be prevented through influent loading rate control. By controlling the influent loading rate into the system, a serious variation in the F/M ratio can be avoided and hence a near optimal condition for bacteria can be maintained under a condition of varying influent quality. Thus, a relatively constant removal efficiency may be obtained under a fixed-time operation, by achieving flexible HRT, depending on influent quality. 5.3.2 General Observation The system SRT and HRT maintained during the operation of TSSBR#4 were 11.4 days and 4.83 days, respectively. The obtained system SRT and HRT were longer than those in real-time control (TSSBR #1, 2, 3 and 4), except for the HRT in TSSBR#3. The average SS concentration in the A/O and the anoxic reactor were 7.4509 g/L and 4.6270 g/L, respectively. The solids concentration inside the A/O reactor was higher than those recommended by Metcalf and Eddy for typical SBR (1.5 to 5.0 g/L). However, the average solids concentration in the anoxic reactor was kept within the recommended range. The VSS/SS ratio was 0.82 in both reactors. The daily variation of SS and VSS levels in each reactor was tracked during operation and the results are presented in Figure 5.73 and 5.74. As found in Figure 5.73 and 5.74, the SS levels inside the A/O and the anoxic reactor varied from 4.28 g/L to 9.85 g/L and 2.3 to 7.7 g/L, respectively. The daily sludge wastage rate from the A/O reactor during days 293 to 297 was higher than the daily sludge production rate, resulting in a decrease of solids levels in the reactor. Upon observation in solids trend, the sludge wasting rate was reduced slightly to 178 Time (days) Figure 5.73. Daily variation of sol ids in A / O reactor -Anoxic SS -Anoxic VSS 0> Oi Oi Oi CN CM CN CN Time (days) Figure 5.74. Daily variation of sol ids in Anox ic reactor 179 prevent further decrease in the solids levels in the reactor. The decline in solids concentration between days 307 to 314, under the reduced wasting rate, was considered to have occurred due to poor settling. During those days, poor treatment was achieved due to the increased nutrient loading rate. The solids in the A/O reactor carried over into the anoxic reactor, probably due to the poor sludge settling. This can be found in the increased sludge levels on those days. Thus, complete sludge separation between the A/O and the anoxic reactor was not achieved during those days. With the passing of high nutrient concentration and poorly settleable solids from the A/O reactor, the solids settling in the anoxic reactor deteriorated and resulted in increased solids levels in the final effluent. Since any regulation of influent strength was not conducted and the system MLVSS levels varied, the F/M ratio must have been highly variable on a daily basis. The average F/M ratio during this operation was 0.134 d"1 (in applied TOCs/MLVSS.d), based on the MLVSS in A/O reactor. Based on the applied B O D 5 S (equation 6 was used for the calculation), the average F/M ratio was 0.248 d"1, revealing that the achieved F/M ratio was kept within the design range recommended by Metcalf and Eddy for typical SBR (0.05 to 0.3, applied BOD 5s/MLVSS.d). The C, N and P concentrations in an influent bucket, which was at room temperature, were tracked for four days to determine the biochemical reactions occurring by the indigenous bacteria in swine manure. The influent bucket was replenished with stored swine slurry from the cold room every 3rd or 4th day during the experiments (TSSBR#1, 2, 3 and 4). The feed bucket was usually kept in an anaerobic condition by mixing the contents only during the influent feeding period. However, it is evident that a small amount of air must have entered the mixture during the mixing period, since the influent bucket was not completely air-tight. During the tracked four days (Figure 5.75), the soluble organic carbon and ortho-P contents in the feed bucket declined, while the NH4-N concentration increased with time. The increase of NH4-N is considered to occur by 180 ammonification of organic matter. In theory, ammonia is produced from organically bound nitrogen by extra-cellular biochemical action on animal tissue and animal fecal matter such as proteins, urea and amino acids, by endogenous respiration of living bacterial cells, and from dead and lysed cells. For instance, proteins are firstly broken down to the constituent amino acids and the subsequent deamination of these amino acids results in the release of the NH4-N. The ammonia-N produced by bacterial decomposition and hydrolysis of organic nitrogen is assimilated into the new bacterial cells and the excess is released as NH4-N. In general, assimilation is known to be responsible for removing up to 30% of influent T K N in biological treatment of municipal wastewaters at conventional (non-nitrifying) loading rates (Barnes and Bliss., 1983). The decline of the organic carbon content may have been caused by the fact that the organic carbon may serve both as energy and synthesis substrate for the degradation of complex nitrogenous organic compounds such as proteins and assimilation of ammonia. Also, the inorganic phosphates would be used for the bacterial cell primarily in nucleic acids and phospholipids for growth, resulting in the small decrease of soluble ortho-P in the bucket. 150 2 days Figure 5.75. Nutrient trends in the influent bucket 181 5.3.3 System Performance of TSSBR#4 5.3.3.1 Wastewater Characteristics The influent wastewater characteristics used for TSSBR#4 are presented in Table 5.9. The total soluble organic carbon (TOCs) level averaged 2057 mg/L, varying from 1275 to 3816 mg/L. The influent average T K N and NH 4 -N were 357 mg/L and 141 mg/L, respectively; 39.5 percent of T K N was in the form of NH4-N. The NOx-N levels were relatively low, average 1.78 mg/L (ranged from 0.8 to 2.9 mg/L). The influent total phosphorus concentration averaged 118 mg/L, with approximately 81 percent being in the form of soluble ortho-P (the ortho-P levels was average 96 mg/L). The relative C:N:P ratio (as TOCs:TKN:T-P) was 100:17.4:5.7, showing similar composition characteristics to those of RTC 4 (100:18.4:5.7). The total solids in the influent were composed of 33 % SS and 67 % colloidal or micro-particle solids which could pass the filter paper (SS/TS = 0.33). The relative ratios of TVS/TS and VSS/SS were 0.71 and 0.69, respectively. While the SS of influent was mostly composed of organic material in other operations (above 90 %), the SS contained high inorganic material in this operation. 5.3.3.2 Removal Efficiencies and General Discussion The treatment efficiencies and effluent quality in TSSBR#4 are summarized in Table 5.10. In general, the observed overall removal efficiencies in TSSBR#4 were much lower than those of any real-time control strategies (TSSBR #1, 2 and 3). The removal efficiencies of TOCs, NH4 -N, TKN, soluble ortho-P and total-P were 90.3 %, 74 %, 87.5 %, 84.3 % and 82.8 %, respectively. Also, compared to other real-time operations, the final effluent quality obtained in this operation was much more variable than those obtained in real-time control. The large variability of the final effluent in fixed-time operation can be found by comparing the standard deviation. Figure 5.76 shows the influent and effluent TOCs trends. From this graph, the dynamic 182 Table 5.9. Wastewater charateristics used for TSSBR # 4 Parameters (mg/L) Means Min. - Max. Std. Dev. T O C s * * 2057 1275 -3816 844 NH4 -N 141 65 - 262 53 T K N 357 195-718 137 NO x -N 1.78 0.8 - 2.94 0.54 Ortho-P 96 46 - 199 46.6 T-P 118 59 - 241 52.8 T S (g/L) 4.1409 2.067 - 7.567 2.009 TVS (g/L) 2.9383 1.550-5.683 1.520 SS (g/L) 1.3663 0.633 -2.533 0.729 VSS (g/L) 0.9407 0.200 - 2.250 0.639 ** s: Soluble Table 5.10. The effluent quality and removal efficiencies in TSSBR #4 Parameters (mg/L) Means Min. - Max. Std. Dev. % removal TOCs** 205 37- 1002 278 90.3 NH 4 -N 36.6 0-101 34 74.0 T K N 44.8 3.9-161 42 87.5 NO x -N 2.9 0.1-8.2 2.73 Ortho-P 15.1 1.9-32.4 7.87 84.3 T-P 20.3 5.1 - 67 12.36 82.8 T S (g/L) 1.0208 0.675 - 1.55 0.319 75.3 TVS (g/L) 0.7758 0.217-1.50 0.486 73.6 SS (g/L) 0.0929 0.035 - 0.26 0.075 93.2 VSS (g/L) 0.0783 0.025 - 0.24 0.075 91.7 ** s: Soluble 183 Figure 5.76. Track of T O C s in influent and effluent Figure 577. Track of phosphorus in influent and effluent 184 response of the system, under a variable loading rate condition, can be found. When an average 1627 mg/cycle (varied from 1224 to 2258 mg per cycle) was loaded into the system, the soluble organic matter was almost completely oxidized during the provided aerobic period and the effluent concentration remained at 37 to 74 mg/L. However, when an average 2.3-fold TOCs loading rate (3716 mg/cycle) was imposed upon the system for 5 days due to the influent strength change, a rapid increase (up to 1002 mg/L) in the effluent TOCs was observed. The effluent TOC level started to decrease as the influent loading rate was lowered to an average 1731 mg/cycle; it took more than 7 days for the system to return to its preshock loading condition. This aspect may indicate the importance of loading rate control. The variability of phosphorus and nitrogen in the influent and effluent is illustrated in Figures 5.77 and 5.78. Under a variable loading rate condition, the P and N levels in the final effluent were not constant; the soluble ortho-P and total-P concentrations in the effluent varied from 1.9 to 32 mg/L and 5.1 to 67 mg/L, respectively, while the effluent levels of T K N and NH4 -N varied from 3.9 to 161 mg/L and 0 to 101 mg/L, respectively. From these graphs (Figure 5.77 and 5.78), it can be seen that the TKN and NH4-N levels in the effluent were exactly a function of the influent nitrogen loading rate; however, the effluent P did not correlated to the influent loading rate, under a provided parameters (HRT and SRT). Low effluent NOx-N levels were achieved without a supplemental organic carbon source, averaging 2.9 mg/L. The effluent NOx-N tracked with time is shown in Figure 5.78. The effluent NOx-N levels fluctuated between 0 and 8 mg/L. Less than 1 mg/L of NOx-N levels was maintained during the operation days of 308 to 318, since more organic carbon matter was available for denitrification during that period. The fluctuation of solids in the influent and effluent is represented in Figures 5.79 and 5.80. In general, the solids concentration in the influent were proportional to the nutrient levels. The total 185 Figure 5.78. Track of nitrogen in influent and effluent 186 _ J D) in "D O m C o 3 10 g 8 7 6 5 4 3 2 1 0 • Influent TS • Influent TVS A Influent SS X influent VSS • 8 • • • • * • • • A X A J L A X x • A X r- CN CM CM c O T j - L o c D r ^ c o o j o * -c o o o j o j c n o o i o o C M C M C N C N C M C M C M C O CO CO CO CO CO CO Time (days) i ^ o o o j O T - C M c o ^ t i n c o f ^ - c o O O O - r - T - T - T - T - " - — T - T -C O C O C O C O O O C O C O C O C O C O C O C O Figure 5.79.Variation of influent sol ids 1.8 1.6 1.4 3 5>12 in T3 3 1 to g |U.O 3 it UJQ.6 0.4 • Effluent TS • effluent TVS A Effluent SS X effluent VSS • • • • 0.2 A. J 5 i . ^ - C N C O ^ f L O C D h - O O C D O T - C N C O T r 0 > 0 ) 0 1 0 J O J O J C T ) 0 1 0 ) 0 0 0 0 0 C N C N C N C V I C M C N C V l C M C N I C O t O C O C O C O c o r ^ - c o a > O T - c N c o ^ r m c D h * - c o O O O O T - ^ - ^ - T - T - T - T - ^ - T -c o c o c o c o c o c o c o c o c o c o c o c o c o Time (days) Figure 5.80. Variation of effluent solids 187 solids and SS levels in final effluent averaged 1.02 g/L and 0.093 g/L, varying from 0.675 to 1.55 g/L and 0.035 to 0.26 g/L, respectively. Approximately 9 percent of the effluent total solids was composed of suspended solids. The effluent TVS/TS and VSS/SS ratios were 0.76 and 0.84, respectively. Overall, the average removal efficiencies of TS, VSS, SS and VSS were lower than those observed in other real-time system. 188 C H A P T E R 6 C O N C L U S I O N AND R E C O M M E N D A T I O N S 6.1 Conclusions In this research, the newly developed real-time control technique has been successfully applied, without a supplemental carbon source, to a novel bench-scale two-stage SBR system (TSSBR) for the treatment of swine farm wastewaters. The results from this research show that nutrient and organic carbon removal efficiencies, equal to or greater than those achieved by tertiary treatment, can be realized using a much simpler treatment method. This technology can be applied to a variety of wastewaters and in a wide range of wastewater treatment processes. Based on the data obtained in this research, specific conclusions particular to the operating strategies of the new developed treatment system (TSSBR) and to real-time control technology have been drawn as follows: 1) Results show that the newly designed TSSBR are an effective treatment process for animal wastewater (high strength wastewater). In TSSBR#4, operated in a fixed-time fashion with a constant loading volume of influent (fixed-HRT), high removal efficiencies of combined C , N and P were achieved, although these were lower than those achieved with real-time operating strategies. 2) The loading rate of carbonaceous and nitrogenous materials into a treatment system can be controlled by monitoring the ORP. The concentrations of nitrogenous materials can be monitored indirectly by the ORP reading. Improved performance, through control of the loading rate, can be attained by achieving flexible influent loading and HRT, determined on the basis of wastewater strength. 3) Loading rate control, using proper ORP-setpoints, can ensure fast denitrification, followed by phosphorus release during the air-off period. This prevents the accumulation of 189 NOx-N and guarantees relatively complete phosphorus removal during the aerobic period. 4) Control of the loading rate contributes to the maintenance of a balance between the aeration rate and the systemic oxygen demand. This is essential for obtaining the ideal ORP curves required for real-time process control using ORP. 5) Real-time process control technology makes it possible to obtain nearly constant effluent quality, despite high fluctuations in the quality of the influent wastewater. Based on system performance and effluent quality in this research, no significant differences were found between the real-time control (RTC) strategy of TSSBR#3, using NBP as the real-time control point, and real-time control strategy of TSSBR#1 and 2, which used the RCMP as the control point. With regard to organic carbon removal efficiencies, the average removal efficiencies of TOCs in TSSBR#1 were slightly lower than those achieved in TSSBR#3. On the other hand, the average removal efficiency of TOCs in TSSBR#2 (RTC 3), which operated with the same real-time control point as TSSBR#1, was slightly higher than that of TSSBR#3. The lower TOCs removal performance in TSSBR#1, which occurred despite the existence of organic carbon oxidation conditions beyond the NBP, can be attributed to lower biodegradability of the influent organic carbon content than was the case in TSSBR#2 and 3. Significant differences can, however, be found between real-time and fixed-time operations. Under variable influent conditions, a constant quality of effluent could not be obtained during the operation of TSSBR using fixed-time control strategies (TSSBR#4). This was indeed possible, with real-time operation. Furthermore, the nutrient and solids removal efficiencies in the fixed-time operation (TSSBR#4 ) were much lower than those achieved with real-time operating strategies. For the ease of comparison, the system performances of different control strategies are summarized in Table 6.1. 190 '5b 3 O -4-» c o o ,11) • <+-< O /->! ! i ro oo cn MD CN 1 © i o\ T t t--KI oo; ! i CN OO vi cn t--r -a 6) The most distinctive advantages of the real-time TSSBR system was its self-adjusting ability. It was able to optimize biological nutrient removal and energy consumption by delaying the termination of aeration until nutrient removal was complete. 7) The oxidation-reduction potential (ORP) and/or pH pattern in a given wastewater treatment process can be used to identify specific control points which can, in turn, be utilized in the optimization of organic carbon matter, nitrogen and phosphorus removal efficiencies. 8) Real-time control, using pH, may possesses advantages over real-time control using ORP. The pH patterns in the aerobic phase are uniform and reproducible from cycle to cycle, despite imbalances between aeration and oxygen demand; at the same time, the nitrogen break point (NBP) on the ORP curve cannot be discerned under conditions of excessive-aeration, which can result in control failure. With real-time control using pH, there is no need for concern over control failure, because the distinctive control point is easily detected on the pH-time profile, even under conditions of excess aeration. 9) Real-time process control, using ORP, does have practical advantages, over real-time control using pH, in terms of operating costs and system performance. When the nitrogen break point (DO break point) cannot be observed on the ORP curve, for example, excessive or deficient aeration conditions can be suspected. In general, optimization of a given system's capacity is impossible under conditions of under-aeration, due to the retardation in nitrification, while the maintenance of over-aeration results in excessive aeration costs. These conditions can be diagnosed by monitoring the ORP pattern, thereby providing opportunities to maximize the treatment capacity of a given system while minimizing aeration costs. 10) The specific control point named the Residual Carbon Manipulation Point (RCMP) and the Nitrogen Break Point (NBP) can be used in a process control strategy to treat high C/N (or high P/N) and low C/N (or P/N) wastewaters, respectively. 192 11) With real-time control strategies using either the NBP or the RCMP, depending on the relative C:N:P ratio of the influent wastewater, a treatment process can be effectively operated without the addition of an external carbon source to enhance denitrification. 12) The sludge addition strategy used in the operation of TSSBR#2 enhanced the denitrification efficiency. It is suggested that the carbon source for denitrification could have come from the residual organic carbon in the liquor and from the hydrolysis of macro- or particulate organic material trapped in the transferred bacterial slime. Some stored carbon sources within the transferred cells might have been be utilized for denitrification in the absence of an external carbon source. The sludge addition strategy is considered to have enhanced endogenous respiration. 13) A clearly defined Nitrate Break Point, which can be correlated with the end of denitrification, was observed in the anoxic reactor. This point can be used as an indicator for the anoxic reactor. However, this Nitrate Break Point was not usually detectable on the ORP curve in the A/O reactor, since it occurred rapidly and was masked by the fast decline in redox potential during the influent loading period. 6.2 Recommendations The results from this study suggest a number of avenues for further research. 1) Specific measurement of the denitrification rate, using the carbon stored within cells as well as by endogenous respiration is necessary, to ascertain whether or not a supplemental carbon source is needed for any unit process, and, if required, how much should be added for complete removal of nitrate. 193 2) Compensation technology of pH and solids level variations in the ORP measurement is needed for more effective loading rate control using the ORP setpoints. Such a strategy may prevent the ORP reading variations due to variable pH and solid concentrations in the system and influent wastewaters. Development of this technology may help maintain the relatively constant oxidation/reduction state at the loading period from cycle to cycle, thus avoiding shock loading. 3) Evaluation of the process developed here should be conducted at the pilot-plant scale under real field conditions. 4) To prevent the spread of infectious disease, most swine farms are using disinfectants which could be lethal to the biological treatment process. An overdose of disinfectant can stop all biochemical reactions by microorganisms, and if influent wastewater contains even a little of the disinfectant, biochemical reactions could be retarded. Furthermore, the discharge of effluent containing any type of disinfectants could lead to long term environmental problems. Investigations in this area would therefore be quite valuable. They could provide a useful guide for the use of disinfectants in the industry, as well as methods for their role in waste treatment strategies. 194 References Altas, R. M . , 1986. "Basic and Practical Microbiology". Macmillan Publishing Company., New York. American pubulic health association (A.P.H.A), 1985. Standard methods for the examination of water and wastewater. 16th Ed., Washington, D. C. Anthonisen. A. C , Loehr. R. C , Prakasam. T. B. S. and Srinath. E. G., 1976. "Inhibition of nitrification by ammonia and nitrous acid". Journal Water Pollution Control Federation, 48: 835. Arora, M . L., Barth, E. F. and M . B. Umphries, 1985. "Technology evaluation of sequencing batch reactors". Journal Water Pollution Control Federation, 57(8):867. Barker, P. S. and Dold, P. L., 1996. "Denitrification behaviour in biological excess phosphorus removal activated sludge systems". Water Research., 30(4): 769. Barnes, D. and Blis, P. J, 1983. "Biological control of nitrogen in wastewater treatment". E. & F. N. Spon. London. Benefield L.D and Raudoll, C .W. , 1977. "Evaluation of a comprehensive kinetic model for the activated sludge process". Journal Water Pollution Control Federation, 49: 1636. Boopathy, R. and Sievers, D. M . , 1991. "Performance of a modified anaerobic baffled reactor to treat swine waste". ASAE. , 34(6):2573. Bortone G., Gemelli S., Rambaldi A. and Tilche A., 1992. "Nitrification, denitrification and biological phosphate removal in sequencing batch reactors treating piggery wastewater". Water Science and Technology., 26: 977. Brock, T. D. and Madigan, M . T., 1988. "Biology Microorganisms". 4th., Prentice-Hall International Editions. Charpentier, J., Godart, H., Martin, G. and Y. Mogno, 1989. "Oxidation-reduction potential (ORP) regulation as a way to optimize aeration and C, N, and P removal: experimental basis and various full-scale examples". IAWPRC, Brighton, Water Science and Technology, 21:1209. Charpentier, J., Florentz, M . and G. David, 1987. "Oxidation-reduction potential (ORP) regulation: A way to optimize pollution removal and energy saving in the low load activated sludge process". IAWPRC, Rio, Water Science and Technology, 19:645. Chen, M. and Fuhs. G. W., 1975. "Phosphate metabolism in the activated sludge process". Microbial. Ecology., 2:119. 195 Comeau, Y., Rabionwitz, B., Hall, K. J. and Oldham, W. K., 1987. "Phosphate release and uptake in enhanced biological phosphorus removal from wastewater". Journal Water Pollution Control Federation, 59:707. Daigger G. T. and Grady J., 1977. "Factors affecting effluent quality from fill-and-draw activated sludge reactors". Journal Water Pollution. Control Federation., 49: 2390. Hao, O. J. and Huang, J., 1996. "Alternating aerobic-anoxic process for nitrogen removal: process evaluation". Water Environment Research., 68(1):83. Heduit, A. and D. R. Thevenot, 1989. "Relation between redox potential and oxygen levels in activated-sludge reactors". IAWPRC, Brighton, Water Science and Technology, 21:947. Hoeker. E.C. and Schroeder. E. D., 1979. "The effect of loading rate on batch activated sludge reactors". Journal Water Pollution Control Federation, 51: 264. Irvine, R. L., Ketchum, L.H. Jr., Breyfogle, R. and E. F. Barth, 1983. "Municipal application of sequencing batch treatment". Journal Water Pollution Control Federation, 55(5):484. Irvine, R. L., Ketchum, L.H. Jr., Arora, M.L. and E. F. Barth, 1985. "An organic loading study of full scale sequencing batch reactors". Journal Water Pollution Control Federation, 57(8):847. Irvine, R. L. , Murthy, D. V. S., Arora, M . L. , Copeman, J. L. and J. A. Heidman, 1987. "Analysis of full-scale SBR operation at Grundy Centre, Iowa". Journal Water Pollution Control Federation, 59(3): 132. Irvine, R. L. and A. W. Busch, 1979. "Sequencing batch biological reactor - an overview". Journal Water Pollution Control Federation, 51(2): 235. Jenkins, C. J. and D. S. Mavinic, 1989a. "Anoxic-aerobic digestion of waste activated sludge: Part 1- solids reduction and digested sludge characteristics". Environmental Technology Letters, 10:355. Jenkins, C. J. and D. S: Mavinic, 1989b. "Anoxic-aerobic digestion of waste activated sludge: Part II - Supernatant characteristics, ORP monitoring results and overall rating system". Environmental Technology Letters, 10: 371. Ketchum, P. A, 1988. "Microbiology Concepts and Applications". JoHn Wiley & Sons, Inc. Kjaergaard, L., 1977. "The redox potential: its use and control in biotechnology". In advances in biochemical engineering, 7 Ghose, T. K., Fiechter, A. and N. Blakebrough (eds.), Springer - Verlag, Berlin, p. 131. Koch, F. A. and W. K. Oldham, 1985. "Oxidation-reduction potential - A tool for monitoring, control and optimization of biological nutrient removal system". IAWPRC, Paris. Water Science and Technology, 17:259. 196 Koch, F. A., Oldham, W. K. and H. Z. Wang, 1988. "ORP as a tool for monitoring and control in bio-nutrient removal system". In proceedings of the joint ASCE/CSCE Environmental Engineering Conference, July 12-19, Vancouver, B. C , p. 162. Lo, K. V., Bulley, N. R. and Kwong, E, 1985. "Sequencing aerobic batch treatment of milking parlour wastewater". Agricultural Wastes., 13:131. Lo, K. V. and Liao, P. H., 1986. "Anaerobic-aerobic biological treatment of screened dairy manure". Biomass., 10:187. Lo. K.V, Liao. P.H. and Van Kleeck. R.J., 1990. "A full-scale sequencing batch reactor treatment of diluted swine wastewater". Canadian Agricultural Engineering., 33: 193. Lo, C. K., Yu, C. W., Tarn, N. F. Y and Traynor, S., 1994. "Enhanced nutrient removal by oxidation-reduction potential (ORP) controlled aeration in a laboratory scale extended aeration treatment system". Water Research., 28(10) 2087. Maekawn, T., Liao, C. and Feng, X., 1995. "Nitrogen and phosphorus removal for swine wastewater using intermittent aeration batch reactor followed by ammonium crystallization process". Water Research., 29(12): 2643. Manning, J. F. and R. L. Irvine, 1985. "The biological removal of phosphorus in a sequencing batch reactor:. Journal Water Pollution Control Federation, 57(1):87. Metcalf and Eddy., 1991., "Wastewater Engineering: Treatment, Disposal and Reuse". 3rd. McGraw-Hill, Inc. Michael, R. O., Humenik, F. J. and Miner, J. R., 1983. "Livestock Waste Management". Boca Raton, Fla. CRC Press. Moriyama, K., Sato, K., Harada, Y., Washiyama, K. and Okamoto, K., 1990. "Renovation of an extended aeration plant for simultaneous biological removal of nitrogen and phosphorus using oxic-anaerobic-oxic process. Water Science and Technology., 22(7/8): 61. Nakanishi, H., Fukagawa, M. , Takeuchi, M . , Ishikawa, M . , Ukita, M . and Murakami, S., 1990. "Biological nitrogen removal in a complete mixing type aerator with ORP control". Water Science and Technology., 22(7/8):269. Ng. W.J., 1987. "Aerobic treatment of piggery wastewater with the sequencing batch reactor". Biological Wastes., 22: 285. Ng. W.J., 1989. "A sequencing batch anaerobic reactor for treating piggery wastewater". Biological Wastes., 29:39. Osada, T., Haga, K. and Harada, Y., 1991. "Removal of nitrogen and phosphorous from swine wastewater by the activated sludge units with the intermittent aeration process". Water Research., 25:1377. 197 Paca, J., 1980. "Bioengineering aspects of aerobic microbial piggery waste treatment". Process Biochemistry, Feb. Peddie, C. C , Mavinic, D. S. and Jenkins, C. J., 1990. "Use of ORP for monitoring and control of aerobic sludge digestion". Journal of Environmental Engineering., A S A E 116(3):461. Peddie, C. C , Koch, F. A., Jenkins, C. J. and D. S. Mavinic, 1988. "ORP as a tool for monitoring and control of SBR systems for aerobic sludge digestion'. In proceedings of the joint ASCE/CSCE Environmental Engineering Conference, July 12-19, Vancouver, B. C.,p. 171. Radjai M . , Hatch R. T. and T. W. Cadman, 1984. "Optimization of amino acid production by automatic self-tuning digital control of redox potential". Biotechnology and Bioengineering Symposium No. 14, John Wiley and Sons, p. 657. Roberson, A. M . , 1977. "Farm Wastes Handbook: Problems, Properties, Storage, Handling and Use of By-Products from Housing Farm Livestok". Aberdeen; Scottish Farm Building Investigation Units. Sasaki, K., Yamamoto, Y., Tsumura, K., Ouchi, S. and Mori, Y., 1996. "Development of 2-reactor intermittent-aeration activated sludge process for simultaneous removal of nitrogen and phosphorus". Water Science and Technology., 34(1-2): 111. Saune, S. P., Capdeville, M . , Mauret, M . , Deguin, A. and Baptiste, P., 1996. "Real-time control of nitrogen removal using three ORP bending-points: signification, control strategy and results". Water Science and Technology., 33(1):275. Sekine, T., Iwahorik, K., Fujimoto, E. and Y. Inamori, 1985. "Advanced control strategies for the activated sludge process". Instrumentation and control of water and wastewater treatment and transport systems, In Proceddings of the 4th IAWPRC Workshop, 27 April - 4 May, Houston and Denver, USA., p269. Siverstein, J. and E. D. Schroeder, 1983. "Performance of SBR activated sludge processes with nitrification denitrification". Journal Water Pollution Control Federation, 55(4): 377. Sorm, R., Bortone, G., Saltarelli, R., Jenicek, P., Wanner, J. and Tilche, A., 1996. "Phosphate, uptake under anoxic conditions and fixed-film nitrification in nutrient removal activated sludge system". Water Research., 30(7): 1573. Totora, G, J., Funke, B. R. and Case, C. L. , 1986. "Microbiology: An Introduction" 2nd., The Benjamin/Cummings Publishing Company, Inc. U.S. Environmental Protection Agency, 1986. "Summary Report: Sequencing Batch Reactors". Report No. EPA-625-8-86-011, Office of Technology Transfer, Cincinatti, Ohio, 23p. Wasiak, K. W., Heduit, A., Audic, J. M . and Lefevre, F., 1994. "Real-time control of nitrogen removal at full-scale using oxidation reduction potential". Water Science and Technology., 30(4): 207. 198 Wareham, D. G, 1992. "The use of oxidation-reduction potential (ORP) as a process control parameter in wastewater treatment systems". Ph D. thesis. The University of British Columbia, Vancouver, Canada. Wareham, D. G., Hall, K. J. and Mavinic, D. S.,1993. "Real-time control of aerobic-anoxic sludge digestion using ORP". Journal of Environmental Engineering., A S A E , 119: 120. Wareham, D. G., Mavinic, D. S. and Hall, K. J., 1994. "Sludge digestion using ORP-regulated aerobic-anoxic cycles". Water Research., 28(2):373. Watanabe, S., Baba, K. and S. Nogita, 1985. "Basic studies of an ORP/external carbon spurce control system for the biological denitrification process". Instrumentation and control of water and wastewater treatment and transport systems, In Proceedings of the 4th IAWPRC Workshop, 27 April - 4 May, Houston and Denver, USA, p. 641. Wilson, K. and Goulding, K. H., 1987. "A Biologist's Guide to Principles and Techniques of Practical Biochemistry". 3rd. Ed, Edward Arnold Ltd. Wong, S. H., Wu, M . W. and Choi, C. C , 1990. "Upgrading an aerated lagoon to a sequencing batch reactor for piggery waste treatment". Biological Waste.,34:113. Yu, R. F., Chang, C. N., Lu, H. J. and Cheng, W. Y., 1997. "Monitoring and control using on-line ORP on the continuous-flow activated sludge batch reactor system. Water Science and Technology., 35(1): 57. 199 APPENDICES Appendix A Parameter and ORP data Appendix B Real Time Control Strategies (Flow charts) 200 APPENDIX A Parameter and ORP data Figure A - l Daily variation of solid concentration in A/O reactor (RTC 4) A-2 Daily variation of solid concentration in Anoxic reactor (RTC 4) A-3 Daily variation of solids in A/O reactor (RTC 5) A-4 Daily variation of solids in Anoxic reactor (RTC 5) A-5 Correlation of influent CODs with TCs, TOCs and ICs A-6 Correlation of effluent CODs with TCs, TOCs and ICs A-7 Correlation of effluent BOD 5s with CODs and TOCs A-8 Correlation of influent BOD 5s with CODs and TOCs A-9 Specific features of ORP curve (TSSBR#3) A-10 Carbon composition in influent (TC, TOC and IQ-RTC 1 A - l 1 Carbon composition in effluent (TC, TOC and IQ-RTC 1 A-12 Carbon composition in influent (RTC 2) A-13 Carbon composition in effluent (RTC 2) A-14 Carbon composition in influent (RTC 3) A-15 Carbon composition in effluent (RTC 3) A-16 Carbon composition in influent (RTC 4) A-17 Carbon composition in effluent (RTC 4) A-l8 Carbon composition in influent (RTC 5) A-19 Carbon composition in effluent (RTC 5) A-20 Carbon composition in influent (RTC 6) A-21 Carbon composition in effluent (RTC 6) A-22 Correlation of influent alkalinity with NH4 -N level A-23 Correlation of influent pH with alkalinity (as CaCOs) A-24 Variation of alkalinity in influent and effluent A-25 Correlation of effluent alkalinity with N H 4 - N 201 12 o co o CM Time (days) F igu re A -1. Da i l y var ia t ion of so l i d concen t ra t i on in A / 0 reac to r ( R T C 4 ) Time (days) F igu re A-2. Da i l y var ia t ion of so l id concen t ra t i on in A n o x i c reac to r ( R T C 4 ) 202 in •a o CO Time (days) F igu re A-3. Da i l y var ia t ion of so l i ds in A / O reac tor ( R T C 5) S 4 in •a =5 3 CO -Anoxic SS -Anoxic VSS o Time (days) F igu re A-4. Da i l y var ia t ion of so l id concen t ra t i on in A n o x i c reac to r ( R T C 5) 203 3500 3000 2500 _J 2000 E 1500 o 1-1000 500 0 1000 2000 3000 4000 5000 6000 7000 8000 3500 3000 2500 ^ 2000 V) O O (j 1500 1000 500 0 500 450 400 350 d 300 E 250 o 200 150 100 50 0 y = 0.3448X + 56.584 R2 = 0.878 1000 2000 3000 4000 5000 6000 7000 8000 y = -0.0078x+ 172.46 R2 = 0.0183 1000 2000 3000 4000 5000 6000 7000 8000 CODs (mg/L) Figure A-5. Correlation of influent CODs with TCs, TOCs and ICs (a: CODs vs TCs b: CODs vs TOCs c: CODs vs ICs) 2 0 4 0 •! 1 1 H 1 ! 1 i 1 1 0 50 100 150 200 250 300 350 400 450 180 | 1 160 0 -I 1 1 1 1 • 1 1 1 1 1 0 50 100 150 200 250 300 350 400 450 350 50 -0 -I 1 1 1 1 1 1 1 1 0 50 100 150 200 250 300 350 400 450 CODs (mg/L) F igu re A-6. Co r re la t i on of ef f luent C O D s with T C s , T O C s a n d ICs (d: C O D s v s T C s e: C O D s v s T O C s f: C O D s v s ICs) 205 20 0 -I 1 1 1 : 1 1 0 5 10 15 20 25 Effluent BOD5s(mg/l) y = 2.7325X + 95.157 R2 = 0.3935 10 15 Effluent BODss (mg/L) 20 25 Figure A-7. Correlation of effluent BOD5s with CODs and TOCs 206 3000 2700 2400 -=d 2100 O) 600 300 -0 -I i i ' i ! 1 ! : i 1 1 1500 1700 1900 2100 2300 2500 2700 2900 3100 3300 3500 Influent BOD 5 (mg/L) 7000 "i Figure A-8. Correlation of Influent BOD5s with CODs and TOCs 207 M i n N O i T - n w s c n ' - n i n N O i ' - n i f i N O i ' - n i f l K o i c N ' r c o c o o N ^ c D Time Effects of solid (MLSS) settling T - I - T - T - T - - I - - < - - < - T - T - * - . - T - I - * - T - C M C M < N C M < N < M C N C \ 1 C N Time(hrs) Figure A-9. Specific features of ORP (TSSBRs#3) 208 •a C RJ o o 3500 3000 2500 2000 1500 1000 500 •Influent TOCs •Influent ICs •Influent TCs Time (days) Figure A-10. Carbon composition in influent (TC, TOC, and IC)-RTC1 500 i 450 400 E. 350 o 300 d o 1- 250 d 1- 200 C A> 3 150 !T 100 LU 50 0 -Effluent TOCs -Effluent ICs -Effluent TCs Time (days) Figure A-11. Carbon composition in effluent (TC, TOC, and IC)-RTC1 209 3000 5" 2500 at E. ^ 2000 C to O 1500 o -Influent TOCs -Infljuent ICs -Influent TCs o 1000 500 Time (days) Figure A-12. Carbon composition in influent (RTC 2) 300 -Effluent TOCs -Effluent ICs -Effluent TCs O 1- T - T -Time (days) Figure A-13. Carbon composition in effluent (RTC 2) 210 Figure A-14. Carbon composition in Influent (RTC 3) 300 | 0 I . 143 145 147 149 151 153 155 157 159 161 165 167 169 171 Time (days) Figure A-15. Carbon composition in effluent (RTC 3) 2 l l 6000 i m r ^ a 3 T - c o m t ^ o ) T - c o u i r ^ . o j r - < o O J O l C n O O O O O ^ T - T - T - T - C M C M T - T - * - C N C M C M C M C M C M C M C M C M C M C M C M Time (days) Figure A-16. Carbon composition in Influent (RTC 4) Figure A-17. Carbon composition in Effluent RTC 4) 212 3000 2500 c o J3 -Influent TOCs -Infljuent ICs -Influent TCs Time (days) Figure A-18. Carbon composition in Influent (RTC 5) 300 250 O) _ _ . E 200 c o c 0) 3 E LLI -Effluent TOCs -Effluent ICs - Effluent TCs Time (days) Figure A-19. Effluent carbon composition (RTC 5) 213 4500 Time (days) Figure A-20. Influent carbon composition (RTC 6) 1400 i 291 293 295 297 299 301 303 305 307 309 311 313 315 317 Time (days) Figure A-21. Effluent carbon composition (RTC 6) 214 Figure A-22. Correlation of influent alkalinity with NH4-N 4000 IT 3500 ~ 3000 0 2500 J , 2000 E 1 1500 1000 y = 268.4x - 290.96 R 2 = 0.0241 • • * * A • • • • c 500 6.2 6.4 6.6 6.8 7 Influent pH 7.2 7.4 7.6 7.8 Figure A-23. Correlation of influent pH with alkalinity 215 4500 4000 3500 o o ra O 3000 co 5" 2500 E. 2000 < i> kalin 1500 kalin < 1000 X 500 0 -Influent alkalinity -Effluent alkalinity Time (days) Figure A-24. The variation of alkalinity in influent and effluent in RTC -Effluent alkalinity -Effluent NH4-N Time (days) Figure A-25. The correlation of effluent alkalinity with NH4-N 10 9 8 7 6 5 4 3 , 2 1 0 216 APPENDIX B Real-Time Control Strategies (Flowcharts) Chart 1 Real-time control strategy for typical SBR with RCMP (or NBP) and Nitrate Knee Chart 2 Real-time control strategy for aerobic tank of other process with RCMP or NBP Chart 3 Full Real-time control strategy for TSSBR with RCMP (or NBP) and Nitrate Knee Chart 4 Real-time control strategy for new designed TSSBR with RCMP or NBP 217 \ A V G | Anoxic codition [or variable time If P removal required |(for low C/N or P, low P/N wastewater) On/Off signal WW) Chart 1. R e a l - T i m e Con t ro l s t ra tegy for typ ica l S B R s 2 1 8 1 Calculate Moving AVG r = x * when no C point occur by usual factor (two high aeration rate etc.) or no N in W W Mixer ON Anoxic phase If P removal is required When denitrification is not required The target nutrient = C Chart 2. Real-Time control strategy for aerobic tank of any process 2 1 9 0 R P 2 \ » R P 3 | AVG 1 Calculate Moving AVG < r = x Detect A point | No MRG > 50 0 IDetect B point MRG < 5 Mixer Anoxic ON phase(if i i Influent Feeding up to ORP set P. A i effluent discharge Liquor transfer removal required) Detect C point MRG > 15 yes No when the designated control point = C (for [owi & low P/N WW) IDetect D or D, MRG < t yes when the designated control point = D (for high C/N or P & high P/N WW) Air & mixer OFF C/N or P ORP ^ l i ^ X X \ \ / / / / | AVG | Calculate Moving AVG ^1 Reset I Detect a point MRG > 100 No yes Detect F point MRG <5 No yes Detect G point MRG > 30 No yes Mixer OFF & effluent discharge Denitrification tank control Aerobic tank control Chart 3. Fu l l R e a l - T i m e Con t ro l s t ra tegy for T S S B R s 220 ^ R P ^ ^ R P ^ ^ R P ^ Mixer Anoxic On phase 1 k Influent Feeding up to ORP set P. A k effluent discharge . Liquor transfer Chart 4. R e a l - T i m e cont ro l s t ra tegy for n e w d e s i g n e d T S S B R s with R C M P or N B P cont ro l point 221