More stringent requirements on nitrogen removal from wastewater are the motivation for this thesis. In order to improve treatment results and enhance cost-efficient operation of wastewater treatment plants, model based control strategies are presented. A Java based simulator for activated sludge processes (JASS) is presented. The graphical user interface, educational experiences and implemented control strategies are discussed. Controlling the addition of an external carbon source is the next topic discussed. A simple model based feedforward controller is derived and evaluated in a simulation study. The controller attenuates process disturbances quickly. Further, two feedforward controllers for adjusting the aeration volume in activated sludge processes are derived. The aim of the volume control strategies was to efficiently dampen the impact of process disturbances without using an excessively high dissolved oxygen concentration. The simulation results are promising and show that the aeration volume may be a feasible control variable. A linearisation method for static input non-linearities is presented. The method gives essentially the same result as the existing standard method, but possesses some implementational advantages. The method is used to linearise the non-linear oxygen transfer function of an activated sludge process in an application study. Multivariable interactions in a process model describing nitrate removal in an activated sludge process are studied using the well known RGA method as well as a relatively novel tool based on Hankel norms. The results of the analysis are compared to conclusions drawn from common process knowledge and are used to design a multivariable control strategy. It was found that process disturbances may be rejected faster using multivariable control. Finally, the operational costs of the denitrification process are investigated and visualised graphically. Cost optimal regions are found by a numerical grid search. Procedures for controlling the denitrification process in a cost-efficient way are described.

Slamsvällning orsakar stora problem i avloppsreningsverk med biologisk rening i aktivt slamprocesser. Slamsvällning orsakas av filamentösa (trådformiga) bakterier, som inverkar negativt på slammets sedimenteringsegenskaper. Himmerfjärdens vattenreningsverk har drabbats av detta problem som leder till ett stabilt lager av slam på ytan av sedimenteringsbassängen som inte sedimenterar. För att lösa detta problem behandlades returslammet från sedimenteringsbassängen med ozon för att minska mängden filamentösa bakterier i returslamflödet. Ozon är en starkt oxiderande gas, som är väl användbar för icke-specifik bekämpning av slamsvällning. När ozon kommer i kontakt med den filamentösa bakteriens cellvägg penetreras det in i cellen, varvid cellen lyserar. Ozonbehandlingen resulterade i en förminskning av antalet filamentösa bakterier. Ozonbehandling av returslam förbättrade sedimenteringsegenskaperna hos svällande slam utan att påverka andra viktiga mikrobiologiska processer t.ex. nitrifikation. / Bulking sludge causes major problems in wastewater treatment plants that deal with biological nutrient removal in activated sludge processes. Bulking sludge is caused by filamentous bacteria, which have a negative impact on the sludge settling properties. Himmerfjärden wastewater treatment plant suffers from this type of problem with bulking sludge which creates a stable layer at the surface that does not settle in the clarifier. In order to solve this problem, on site generated ozone was used to decrease the amount of filamentous bacteria in the return activated sludge flow. Ozone is a strong oxidant is suitable for non-specific bulking control. It stresses the filamentous bacteria causing inactivation through cell wall disintegration. The ozone treatment resulted in decreased abundance of filamentous bacteria. Ozone treatment of the recycled activated sludge improves the settling properties of bulking sludge, without interfering with other important microbiological processes e.g. nitrification.

På Käppalaverket i Stockholm står luftningen av de biologiska bassängerna för omkring en femtedel av verkets totala elenergiförbrukning. I ett försök att minska energikostnaden utvärderades under hösten 2010 nya metoder för luftflödesreglering på verket. Grundtanken var att styra luftflödet efter medelvärdet på utgående ammoniumkoncentration under en längre tid, istället för som idag efter momentana värden. Ett vanligt sätt att styra luftflöden på reningsverk idag är att använda återkoppling från utgående ammoniumkoncentration, vilket syftar till att alltid hålla den utgående koncentrationen vid ett valt börvärde. Lagstiftade gränsvärden på ammonium avser dock normalt medelvärden över en längre tid, såsom kvartal eller år. Istället för att anpassa luftflödet efter den inkommande belastningen är det därför möjligt att hålla luftflödet relativt konstant medan istället den utgående koncentrationen tillåts variera. I denna studie visades en energibesparing kunna erhållas om luftflödets variation reduceras. Två strategier utvärderades i vilka luftflödet respektive syrehalten hölls så konstant som möjligt. Dessa jämfördes med den idag använda styrstrategin på Käppalaverket, i vilken luftflödet anpassas efter den inkommande belastningen genom återkoppling. Studien inkluderade både simuleringar i modellen Benchmark Simulation Model no. 1 och fullskaleförsök på Käppalaverket. I både simuleringar och fullskaleförsök resulterade de två utvärderade strategierna i en lägre luftförbrukning per reningsgrad än den idag använda återkopplingsstrategin. I fullskaleförsöken erhölls en luftflödesreduktion på 11 % då luftflödet hölls konstant och 15 % då syrehalten hölls konstant. Båda strategierna genererade dock en kraftigt varierande utgående ammoniumkoncentration. Variationerna var störst då luftflödet hölls konstant och korrelerade inte med den dygnsbaserade belastningsprofilen. Sammanfattningsvis visade studien att en reducering av luftflödets variation resulterar i en lägre luftförbrukning men också i en ökad instabilitet. En konstant syrehalt gav en större energivinst och även en stabilare ammoniumreduktion än ett konstant luftflöde, varför denna metod har störst potential till vidare implementering i fullskala. / The aeration of the bioreactors is responsible for one fifth of the energy consumption at the Käppala wastewater treatment plant (WWTP) in Stockholm. In this report, new methods for aeration control were evaluated in order to reduce the energy costs at the plant. The main idea was to control the effluent ammonia concentration in terms of mean values instead of momentary values. A quite common approach for aeration control is to use feedback from the effluent ammonia concentration, thus aiming to keep the effluent concentration consistently at a certain set point. However, discharge limits normally refer to mean values over longer periods of time, such as months or years. Instead of adjusting the airflow to the incoming load it is therefore possible the keep the airflow fairly constant while allowing a fluctuating effluent concentration. In this paper, it was shown that by reducing the variation of the airflow, energy could be saved. Two methods were evaluated in which the airflow and oxygen concentration respectively was held constant. These methods were compared to the control strategy used today at the Käppala WWTP, where feedback control adjusts the airflow to the influent load. The study consisted of simulations with the Benchmark simulation model no. 1 (BSM1) as well as full scale experiments at the Käppala WWTP. Both the simulations and full scale experiments showed a reduced aeration per nutrient removal for the evaluated methods. In full scale, the total airflow reduction was 11 % when the airflow was held constant and 15 % when the oxygen concentration was held constant. However, the methods resulted in large variations of the effluent ammonia concentration, which did not correlate to the daily influent load. The variations were especially large when the airflow was held constant. In summary, this study showed that a reduced airflow variation results in lower aeration costs but also less stability. A constant oxygen concentration required less aeration and provided a more stable degree of ammonia removal than a constant airflow. For this reason, aeration control with a constant oxygen concentration has the best potential for further use at the Käppala WWTP.

Bioprocesses for air pollution control can generally be categorized as bioscrubber, biofilter, and biotrickling filter systems. These processes have been proven to be economical and effective for control of volatile organic compounds (VOCs) with concentrations of <1,000 mg C m-3 in gas streams. First, an activated sludge aeration tank (W x L x H = 40 x 40 x 300 cm) with a set of 2 mm orifice air spargers was utilized to treat gas-borne VOCs (toluene, p-xylene, and dichloromethane) in air streams. The effects of liquid depth (Z), aeration intensity (G/A), the overall mass transfer rate of oxygen in clean water (KLaO2), the Henry¡¦s law constant of the tested VOC (H), and the influent gaseous VOC concentration (C0) on the efficiency of removal of VOCs were examined and compared with a literature-cited model. Results show that the measured VOC removal efficiencies and those predicted by the model were comparable at a G/A of 3.75 ¡V 11.25 m3 m-2 hr-1 and C0 of around 1,000 ¡V 6,000 mg m-3. Experimental data also indicate that the designed gas treatment reactor with KLaO2 = 5 ¡V 15 hr-1, could achieve > 85% removal of VOCs with H = 0.24 ¡V 0.25 at an aerated liquid depth of 1 m, and > 95% removal of dichloromethane with H = 0.13 at a 1 m liquid depth. The model predicts that, for gas treatment in common activated sludge tanks, with KLaO2 = 5 ¡V 10 hr-1, depth = 3 ¡V 4.5 m, G/A = 9 ¡V 18 m3 m-2 hr-1, > 92% VOC removal can be achieved with operating parameters of Z of 3.0 m and KLaVOC/(G/A) of about 0.28 m-1, for VOCs with H < 0.3, such as most oxygen-containing hydrocarbons with low molecular weights, and benzene, toluene, ethylbenzene, and dichloromethane. Second, an activated sludge aeration tank and a sieve-plate column with six sieve plates were utilized to treat gas-borne VOCs in air streams. The tank was used for the biodegradation of the absorbed VOCs from the column which utilized the activated mixed liquor drawn from the tank as a scrubbing liquor. This research proposed a model for VOC absorption to a down-flow activated sludge liquor in a sieve-plate column. The experimental setup consisted of a pilot-scale activated-sludge tank and a sieve-plate tower, as demonstrated. The sieve-plate tower was constructed from a 25 x 25 x 162 cm (W x L x H) acrylic column with six custom-made sieve plates. Each plate has 382 holes which are 3 mm in diameter arranged on a square pitch. The holes give an open area of 3.82% of the whole plate area for gas flow. Two 25 mm-i.d. down-comer pipes were also equipped to allow for the downflow of the activated sludge liquor. Ports were provided at the column inlet, outlet, and each plate for gas and liquid sampling. Experiments were conducted and the model verified based on the results of tests on the removal efficiencies of isopropyl alcohol (IPA), toluene and p-xylene in the system operated at a range of influent VOC concentrations, air application rates, and liquid/gas flow ratios (L/G). The model developed by a material balance for the gaseous- and liquid-VOC over each plate of the column was developed and experimentally verified in this study. Superficial gas velocity over the column plate (U), number of plates (N), volumetric liquid-phase VOC-transfer coefficient (KLaVOC), aerated liquid depth over the plate (Z), volumetric liquid/gas flow-rate ratio (L/G), dimensionless Henry¡¦s law coefficient of the VOC to be absorbed (H), VOC content of the influent scrubbing liquor (xN+1), and the biodegradation rate constant of the VOC in the activated sludge mixed liquor (k) are among the affecting parameters to the effectiveness of the VOC removal. Model application by the model for effects of affecting parameters on the VOC removal effectiveness indicates that L/G, plate number N, biodegradation rate constant k, Henry¡¦s law constant of VOC H are among the important ones. A L/GH of greater than 2 and N of around 6 are enough for the effective (>90%) removal of the influent VOCs with H < 0.01 if no biodegradation occurred in the column. However, a N of over 16 is required for the influent VOCs with H of around 0.2. Biodegradation with a rate constant of around 100 hr-1 in the column greatly improves the column performance.

Bioflocculation happens naturally and microorganisms aggregate into flocs during wastewater treatment. It is critical to understand the mechanisms of bioflocculation and its impact on the following solid/liquid separation process since seperation by settling is one of the key aspects that determine the efficiency and the overall economy of activated sludge systems. Bioflocculation occurs via extracellular polymeric substances (EPS) and cations by creating a matrix to hold various floc components together so the cations become an important part of the floc structure. The main objective of this study is to investigate the effects of monovalent cations specifically potassium and sodium (K and Na) on the bioflocculation, settleability and dewaterability of activated sludge. The particular aim is to grow the mixed culture microorganisms in the presence of specific cation so that the effect of cation on the stimulation of EPS production can be seen. In order to achieve this aim, semi-continuous reactors were separately operated at concentrations of 5, 10, and 20 meq/L of each cation with mixed culture bacteria and fed with synthetic feed medium representing influent to the activated sludge systems. Also, a control reactor at low cation dose was operated for each reactor set. The effective volume of the reactors was 2 L with 8 days of sludge residence time (SRT) and pH was kept at 7.7&plusmn / 0.3. The activated sludge reactors were operated until the reactors reached steady state and then related analyses were conducted. It was found that addition of potassium and sodium ions at increasing concentrations resulted in increase in total polymer concentration. However, potassium ions promoted the synthesis of both polysaccharide and protein type polymers whereas sodium ions tended to stimulate production of protein type polymers and had an affinity to bind more protein within the floc structure. Sodium sludges had lower hydrophobicity and higher surface charges, so sodium ions led to deterioration in flocculation of sludges. Addition of both these ions decreased the dewaterability, sodium ions had more detrimental effect on dewaterability of sludges compared to potassium ions. The examination of data related to settleability showed that potassium ions led to no drastic deterioration in settling characteristics of the activated sludge but the addition of sodium ions deteriorated the settleability. In addition, it was seen that while the addition of potassium ions to the feed led to a decrease in viscosity, increase in sodium concentration correlated with an increase in viscosity. Finally, the comparison of chemical oxygen demand (COD) removal efficiency of these cations showed that sodium is more efficient in COD removal.

Ultrasonic sludge disintegration &ndash / the most commonly used mechanical pretreatment method- enables the occurrence of cavitation bubbles to extract intracellular material from the cell into aqueous phase. However, there is a lack of information on the volatile solids loading on the anaerobic digestion process performance of ultrasonically treated sludge. In this thesis work, the effect of sonication on disintegration of waste activated sludge (WAS) and an important parameter digester F/M (food to microorganism) ratio on ultrasonically treated WAS were investigated. First, preliminary studies were conducted. It was obtained that when the sonication power and time increased, soluble COD in the supernatant increased as well. Then, batch anaerobic digestion tests were conducted. Effect of F/M ratio in the digesters by using sonicated sludges at different powers was analyzed. For the sludge sonicated at high power, the methane content increased up to 55.1 % at F/M ratio of 10 compared to untreated sludge. On the other hand, methane generation rate slowed down with the increase in F/M ratio. Moreover, 10 % and 15 % increase in the destruction of MLVSS and total COD content was observed for sonicated sludges compared to the untreated sludges, respectively. In summary, both the sonication as a pretreatment method and the increase in digester F/M ratio increased the biogas production and the solids reduction during anaerobic digestion prosess. These results may have important implications for the operation of full scale systems in terms of system efficiency and operation.

The activated sludge process efficiency depends on separation of microbial cells from treated wastewater. Separation can fail due to a number of problems. One of these problems is sludge bulking which is non-settling situation of biomass. Former studies showed that phosphorus deficiency caused filamentous sludge bulking with increasing magnesium ion concentrations. The main objectives of this study are to find out the effect of magnesium ions on sludge properties in phosphorus deficient medium and to determine if there is any bulking. Three different concentrations of magnesium (0.5, 5, 15 meq/L) were added to three bioreactors which contained phosphorus deficient medium. In first set C: N: P ratio was 100:5:0.05. In second set, C:N:P ratio was elevated to 100:5:1. At steady state, physical characteristics including sludge volume index (SVI), viscosity, turbidity and dewaterability were determined. Besides concentration of extracellular polymeric substances (EPS) as well as conductivity was measured. By using API kits, bacterial identification was achieved. In first set phosphorus deficiency and increasing magnesium ion concentration caused filamentous bulking. Carbohydrate content of extracellular polymeric substance significantly increased by magnesium addition. Dewaterability of the system got worse and viscosity decreased. Sludge Volume Index (SVI) indicated severe bulking at all magnesium concentrations. By using biochemical tests microorganisms dominant in the system were determined In second set, all of the parameters indicated healthy flocculation. By magnesium addition, EPSp and EPSc increased. Dewaterability and settleability, improved by the presence of phosphorus with close values measured at different magnesiuim concentrations. Nocardia related genera of Corynebacterium and Enteric microorganisms were identified.

published_or_final_version / Civil and Structural Engineering / Master / Master of Philosophy

Phylogenetic hybridization and phenotypic fingerprinting were applied to the analysis of bacterial communities in wastewater treatment systems. These approaches were aimed at (i) developing monitoring tools able to foresee operational problems, and (ii) providing the rationale to optimize the operation of bioreactors. The work presented is intended to first describe the community found in two reactors treating pulp and paper mill effluent, and second evaluate the possibilities of these techniques with respect to the development of new monitoring tools. / Phylogenetic membrane hybridization showed that the bacterial communities were dominated by Alpha and Beta Proteobacteria, a structure probably linked to the low F:M ratio. Other important factors determining the community structure were the proportion of COD in the high molecular weight fraction, the sludge age, phosphate addition, and the concentration of specific compounds (alcohols, phenols, volatile fatty acids) in the influent. The community structure partly determined the sludge characteristics demonstrating its potential value in the assessment of reactor performance. The results obtained by phylogenetic membrane hybridization suggest that the probes used in a monitoring tool would not need to be targeted to the species level to provide relevant information. However, they also suggest that the technique is more sensitive to changes in population density as opposed to changes in bacterial metabolism. / Phenotypic fingerprinting measured a smaller difference between the communities of the two reactors studied than what was measured by phylogenetic membrane hybridization. However, differences in heterotrophic activities observed between the two communities were linked to differences in influent composition.

The biological treatment of wastewater has evolved significantly from simple single sludge systems practicing organic carbon removal to ones which now include either nitrification/denitrification (N/DN) and / or phosphorus (P) removal. The inclusion of more biological processes have increased the complexity of current wastewater systems which has subsequently led to the development of more complex mathematical models. The operation of plants can be assessed and improved by the use of mathematical modelling tools which require accurate input data. Thus, knowledge of the wastewater characteristics is an important step towards the optimum modelling, design and operation of present and future plants. However, for these tools to be effective, the input data needs to be accurate which is dependent on the current methods used to determine them. Wastewater is a complex substrate consisting of compounds of differing biodegradability. Biokinetically, these compounds have been divided into readily biodegradable (RBCOD), slowly biodegradable (SBCOD) and unbiodegradable substrate groups. Compounds with intermediate biodegradability i.e. compounds which fall between the RBCOD and SBCOD groups, have been termed readily hydrolyzable organic substrates (RHCOD). The organic matter is discussed in terms of chemical oxygen demand (COD). The readily biodegradable and readily hydrolyzable COD fractions of wastewater can be determined by respirometric tests such as the oxygen utilization rate (OUR) and nitrate-N utilization rate (NUR) tests. The principal aim of this project was to investigate the NUR test as a tool for wastewater characterization and to study denitrification kinetics in batch reactors. In addition, an experimental readily biodegradable substrate, acetate, was used to determine the reliability of the NUR tests. Acetate was also used to ascertain utilization profiles and rates of a typical readily biodegradable substrate during denitrification. Biodegradable COD characterizations with enhanced biological phosphorus removal (EBPR) sludges were also investigated to determine the impact of anoxic phosphorus removal on NUR tests. The results obtained from the numerous NUR tests added to the undestanding of the NUR test. Samples from 22 wastewater treatment plants were tested, most of which were located in France. Four South African plants were also tested. Data obtained from the NUR tests were used to calculate the RBCOD and RHCOD fractions. The SBCOD, however, could not be determined directly from the 6 h NUR batch tests. The readily biodegradable COD (RBCOD) fractions ranged between 7 and 25 % of the total COD concentration of raw wastewater, with majority of those results falling within the 10-20 % (of the total COD) range. The results also showed that the initial rapid rate associated with readily biodegradable COD utilization was sometimes followed by a short intermediate phase (i.e. short duration, 2 to 3 h). The intermediate fraction was found to range between 5 and 29 % of the total COD concentration and was classed as a readily hydrolyzable COD component of raw wastewater since the magnitude of the RHCOD fraction was too small to be classed as slowly biodegradable COD which comprises approximately 30 to 60 % of the total COD found in raw wastewaters. The variability of the RHCOD fractions suggests that this fraction is either very variable or that the NUR test does adequately or accurately characterize it. Another possibility is that the RHCOD (or second biodegradable fraction) calculated from the NUR test is a component of the RBCOD of the influent wastewater. In this case, the bacteria may have used some of the RBCOD directly for energy and accumulated or stored the rest as part of a survival mechanism which allows them to be more competitive under dynamic operating conditions. Once the readily biodegradable COD becomes limiting, the bacteria will use the accumulated or stored compounds. This hypothesis is substantiated by tests done with acetate as substrate. An intermediate phase was also observed when acetate was the sole substrate. Thus, it was possible with the 3-phase profiles to calculate a second biodegradable fraction. Results suggest that a significant part of the added acetate (as COD) was stored and the second phase is in fact an 'apparent or residual' phase brought about by the consumption of the stored or accumulated acetate products. This is suggested in two ways: (1) the calculation of the yield coefficient is lower and closer to the 0.5 mg/l values, cited in the literature, when the COD calculated from phases 1 and 2 are considered, and (2) the acetate mass balances were found to be approximately 100 % when phases 1 and 2 were used to calculate the amount of acetate utilized under anoxic conditions. The results obtained with sodium acetate as a readily biodegradable substrate were used to formulate several conclusions on acetate utilization during denitrification. Firstly, from acetate mass balances it was found that acetate may be used exclusively for denitrification (100 % acetate was accounted for). In this case, the sludge contains a significant proportion of denitrifiers and little or no polyphosphate accumulating organisms. This observation was made only when non-EBPR (enhanced biological phosphorus removal) sludges were used. Secondly, acetate mass balances which were found to be < 100 % suggest that acetate could be used for denitrification and the production of storage products like polyhydroxyalkanoates (PHA's). These sludges probably contained a higher proportion of polyphosphate accumulating organisms which competed for the available acetate in the bulk liquid. This observation was made for both EBPR and non-EBPR sludges. Thirdly, acetate could be used for denitrification by denitrifiers and for polyhydroxyalkanoate synthesis by denitrifying polyphosphate accumulating organisms. The stored PHA's in the denitrifying polyphosphate accumulating organisms are subsequently utilized during denitrification. This secondary utilization is manifested in the second denitrification phase and is supported by the observation of phosphorus uptake. These results showed that wastewaters high in volatile fatty acids (VFA's) were also subject to denitrifying polyphosphate accumulating organism activity even though the sludge was sampled from non enhanced biological phosphorus removal systems (non EBPR). Several of the NOx profiles revealed either 2 or 3 rates due to the control of the substrate to biomass ratio (S/X: :<_0.1 mgO2 / mgO2). Majority of the samples (i.e. 85%) tested produced initial maximum specific denitrification rates (k1) between 3 and 6 mgN/gVSS.h. The intermediate denitrification rate (k2) was found to vary between 2 and 3 mgN/gVSS.h. Denitrification rates (k3) obtained from utilization of influent and. endogenous slowly biodegradable COD (SBCOD) varied between 1.0 and 1.5 mgN/gVSS.h. This latter rate is significantly higher than the endogenous denitrification rates cited in the literature. One of the reasons for these higher rates could be be linked to the the reuse of stored or accumulated products by the microorganisms. In addition, a comparative study on RBCOD determination of wastewaters with enhanced biological phosphorus removal and non-EBPR sludges. It was found that the RBCOD values derived by NUR tests with EBPR sludge were consistently lower (4 to 5 %) than those with non-EBPR sludge. Thus, the NUR tests with EBPR sludge resulted in a 4 to 5 % underestimation of the RBCOD fraction of raw wastewaters. This loss in RBCOD to polyphosphate accumulating organisms appears to be linked to the influent raw wastewater acetate concentration. These tests showed that the RBCOD fraction could be adequately characterized using the NUR method. The accuracy of the tests appears to be compromised when enhanced biological phosphorus removal sludges are used in the NUR tests. Moreover, it was found that non-EBPR sludges can also consume some of the acetate that is present in the system for the production and replenishment of storage compounds. Fortunately, for the wastewaters tested, the acetate component of the RBCOD fraction was small and therefore, did not significantly affect the results. Mechanisms such as substrate accumulation and storage may also impact on substrate removal and hence, the determination of the readily biodegradable COD concentration of municipal wastewaters. Thus, while the results showed that the NUR is a useful characterization tool for wastewaters, it will continue to be a more tedious characterization tool than the oxygen utilization rate test, until a suitable nitrate/nitrite electrode is developed to automate the test. / Thesis (Ph.D.)-University of Natal, Durban, 1999.