Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigation of nitrous oxide as a nitrification monitoring and aeration system control parameter in… Shiskowski, Dean Michael 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2005-995496.pdf [ 15.68MB ]
Metadata
JSON: 831-1.0063470.json
JSON-LD: 831-1.0063470-ld.json
RDF/XML (Pretty): 831-1.0063470-rdf.xml
RDF/JSON: 831-1.0063470-rdf.json
Turtle: 831-1.0063470-turtle.txt
N-Triples: 831-1.0063470-rdf-ntriples.txt
Original Record: 831-1.0063470-source.json
Full Text
831-1.0063470-fulltext.txt
Citation
831-1.0063470.ris

Full Text

INVESTIGATION OF NITROUS OXIDE AS A NITRIFICATION MONITORING AND AERATION SYSTEM CONTROL PARAMETER IN SEQUENCING BATCH REACTOR WASTEWATER TREATMENT SYSTEMS by Dean Michael Shiskowski Dipl. (Environmental and Water Resources Engineering Technology) Saskatchewan Institute of Applied Science and Technology, 1988 B.A.Sc. (Regional Environmental Systems Engineering) University of Regina, 1993 M.A.Sc. (Civil Engineering) University of British Columbia, 1995 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF CIVIL ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA November, 2004 © Dean M. Shiskowski, 2004 ABSTRACT The provision of aerobic, autotrophic nitrification in biological wastewater treatment systems requires significant mass quantities of oxygen to provide the biologically-catalyzed oxidation of ammonia to nitrite and nitrate. In addition, the relatively high oxygen half-saturation coefficient for nitrifying organisms causes the kinetic rate of these organisms to be very sensitive to oxygen availability at low bioreactor mixed liquor dissolved oxygen (DO) concentrations (i.e. < 2 mg/L). The efficiency of transferring oxygen into mixed liquor is also affected by DO concentration. However, unlike nitrification, where higher DO levels translate into higher kinetic rates (i.e. a positive impact), elevated DO levels negatively impact oxygen transfer efficiency (OTE). Thus, the dilemma in bioreactor operation, with respect to controlling the oxygen supply rate, involves balancing the required ammonia oxidation rate with OTE, such that complete nitrification can be achieved under a given set of operating conditions while maximizing OTE. This situation ultimately impacts the energy cost associated with bioreactor oxygen supply. Ammonia oxidizing bacteria (AOB), such as Nitrosomonas, can possess a nitrite reductase (NiR) enzyme. This enzyme allows the organism to use nitrite as an alternate terminal electron acceptor in oxygen-limited environments, resulting in the reduction of nitrite to gaseous nitrous oxide (N20). For this research, it was hypothesized that reactor off-gas N 2 0 data could be used to monitor the extent of AOB oxygen limitation-availability and provide an indication of the overall nitrification kinetic rate. The main objective of this research was to investigate the feasibility of using reactor off-gas N 2 0 as a aerobic-phase nitrification monitoring and aeration system control parameter in a sequencing batch reactor (SBR) wastewater treatment system. This was accomplished by subjecting an anoxic-aerobic SBR system, operating under oxygen-limited conditions, to wastewater (i.e. ammonia, readily degradable carbon, slowly degradable carbon (SDC)) and aeration rate perturbations, outside the normal baseline operating conditions, and monitoring system response. Specific experiments were also conducted to confirm the source of generated N 2 0, investigate the effects of DO concentration and slowly degradable carbon utilization rate on aerobic-phase heterotrophic N 2 0 reduction-consumption, examine the influence of nitrite and nitrous acid levels on N 2 0 generation, and evaluate the reactor gas mass transfer characteristics. The data support the hypothesis that AOB were responsible for N 2 0 generation, with aerobic-phase 11 heterotrophic denitrification generating: little, if any, N 20. For a given SBR cycle, reactor oxygen supply rate/DO concentration, nitrite concentration, and pH level-nitrous acid concentration were shown to impact the aerobic-phase N 2 0 generation rate. In addition, the availability of biologically utilizable carbon (i.e. SDC), under suitable DO conditions, could provide significant aerobic-phase heterotrophic N 2 0 reduction-consumption rates, and thus affect the observed (i.e. net) N 2 0 generation rate. The N 2 0 reduction rate was sensitive to oxygen availability/DO concentration, likely related to the oxygen sensitivity of the heterotrophs N 2 0 reductase enzyme. Long-term SBR operation, under the oxygen-limited baseline conditions, found that the aerobic-phase generated N 2 0 mass to oxidized ammonia mass ratio changed over time, increasing and decreasing over a range of one order-of-magnitude. At the peak mass ratio, about 25% of the oxidized wastewater ammonia was converted to N 20. Subtle shifts in the microbial population, with respect to the relative fractions of the AOB and nitrite oxidizing bacteria (NOB), were believed to have caused the observed phenomenon. From a process monitoring perspective, it was shown that off-gas N 2 0 information can be used to identify a change in the oxygen-competition dynamic between AOB and NOB that is induced by a change in aeration rate, as well as a change in wastewater characteristics. This phenomenon affects the relative difference in the ammonia and nitrite oxidation rates, induces subtle differences in transient nitrite levels that impacts N 2 0 generation, and ultimately provides an indication of how the ammonia oxidation rate has changed due to the altered oxygen availability, via off-gas N 2 0 data. Alternately, for. many combinations of aeration rate perturbations and wastewater slowly degradable carbon (SDC) utilization rates, DO and pH data alone could not be used to provide an indication of the effect that a change in SBR operating condition had on the ammonia oxidation rate. Furthermore, the ammonia oxidation rate was observed to decrease later in the aerobic-phase with decreasing mixed liquor ammonia concentration. A reduction in N 2 0 generation rate was coincidental with the reduction in ammonia oxidation rate, and was clearly resolvable in the off-gas N 2 0 data, providing advanced indication of the timing of completed nitrification. The DO and pH data could not be used to identify the reduction in ammonia oxidation rate. The research findings confirm the potential feasibility of using off-gas N 20, collected from a covered bioreactor, as a aerobic-phase nitrification monitoring and aeration system control parameter in SBR wastewater treatment systems. To this end, a conceptual, N20-based aeration system control strategy iii was developed. The proposed strategy would utilize a pattern recognition approach, along with three key state variables: N 2 0 generation on-set time (OST), steady-stage N2O generation rate (SSNGR) and ammonia oxidation-N20 generation reducing-rate time (RRT). An artificial neural network (ANN) is proposed as the implementation framework for the control strategy. i v TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS v LIST OF TABLES viii LIST OF FIGURES ix NOMENCLATURE xvi ACKNOWLEDGMENTS xix 1 INTRODUCTION 1 1.1 Background 1 1.2 The Need for Research 2 1.3 Research Rationale and Objectives 6 2 LITERATURE REVIEW 9 2.1 Introduction 9 2.2 Biological Nitrogen and Carbon Conversion 9 2.2.1 Nitrification 9 2.2.2 Denitrification and N 2 0 Generation 13 2.2.3 Biological Carbon Conversion 19 2.3 N 2 0 Properties, Sources and Environmental Fate 22 2.3.1 Properties 22 2.3.2 Sources 22 2.3.3 Environmental Fate 23 2.4 Wastewater Treatment 23 2.4.1 SBR Systems 24 2.4.2 Nitrification and Oxygen 25 2.4.3 N 2 0 Generation and Emission Control 28 2.4.4 On-Line Process Monitoring and Control Parameters 33 2.5 Summary 39 3 MATERIALS AND METHODS 41 3.1 Experimental Apparatus and Operation 41 v 3.1.1 SBR and Clean Water Reactor Systems 41 3.1.2 SBR Start-Up and Operation 43 3.1.3 Synthetic Wastewater 46 3.2 Experimental Program and Design , 46 3.2.1 SBR Experiments 46 3.2.2 Gas Mass Transfer Experiments 50 3.3 Analytical Methods and On-Line Reactor Monitoring 51 3.3.1 Sample Collection, Pretreatment and Preservation 51 3.3.2 Analytical Methods 52 3.3.3 On-Line Reactor Monitoring 54 3.4 Calculation Methods 56 3.4.1 Gas Concentration Unit Conversion 56 3.4.2 Liquid N 2 0 Concentration Calculation 57 4 RESULTS AND DISCUSSION 58 4.1 Introduction 58 4.2 Typical SBR Time-Variant Parameter Profiles 60 4.3 Reactor Gas Mass Transfer Experiments 75 4.3.1 N 2 0 Stripping 75 4.3.1.1 Total (Gas Transfer + Liquid Surface Diffusion) Stripping 75 4.3.1.2 Liquid-Surface Diffusion Stripping....... 76 4.3.2 0 2 Transfer 82 4.3.3 N 2 0 Liquid-Headspace Partitioning 84 4.4 N 2 0 Source Experiment 93 4.5 Wastewater Component Experiments 105 4.5.1 Ammonia Load '. 105 4.5.2 Readily Degradable Carbon Load 120 4.5.3 Slowly Degradable Carbon Utilization Rate 127 4.6 N 2 0 Reduction Experiments 154 4.6.1 DO Concentration 154 4.6.2 Slowly Degradable Carbon Utilization Rate 164 4.7 Aeration Rate Experiments 172 4.7.1 Baseline Wastewater Experiments 174 v i 4.7.2 Slowly Degradable Carbon .Utilization Experiments 193 4.8 Nitrite-Nitrous Acid Experiments 217 4.8.1 Nitrite Spikes •. 217 4.8.2 pH (nitrous acid) 225 4.8.3 Long-Term SBR Operation: N 2 0 Generation and Nitrite-Nitrous Acid Levels 234 4.9 Long-Term SBR Operation: General Observations 241 4.10 Summary of Major Findings 249 4.11 Development of a Conceptual N20-Based SBR Aeration System Control Strategy 252 5 CONCLUSIONS AND RECOMMENDATIONS 260 5.1 Conclusions 260 5.2 Recommendations for Future Research 264 REFERENCES 266 LIST OF APPENDICES ." 277 vii LIST OF TABLES Table 4.1 SBR anoxic-phase and aerobic-phase carbon utilization calculations (SBR-TS16a) 70 Table 4.2 Initial aerobic-phase ammonia concentration, average ammonia oxidation rate and relative statistics for wastewater ammonia load experiments 107 Table 4.3 Average ammonia oxidation rate and relative statistics for slowly degradable carbon utilization rate experiments 134 Table 4.4 Summary of aeration rate experiments 175 Table 4.5 Aeration system operation, average ammonia oxidation rate and relative statistics for baseline wastewater - aeration rate experiments 176 Table 4.6 Aeration system operation, average ammonia oxidation rate and relative statistics for slowly degradable carbon utilization - aeration rate experiments 194 v i i i LIST OF F I G U R E S Figure 1.1 Ammonia oxidation rate and oxygen transfer efficiency versus DO concentration 4 Figure 3.1 Schematic of SBR system 42 Figure 3.2 Schematic of reactor gas mixing system 44 Figure 3.3 Experimental program summary 48 Figure 4.1 SBR DO, pH, ammonia, nitrite and nitrate concentration-time profiles (SBR-TS16a) 61 Figure 4.2 SBR soluble organic nitrogen concentration-time profiles (MISC-15, MISC-20).. 63 Figure 4.3 SBR biomass organic nitrogen concentration-time profiles (MISC-15, MISC-20). 63 Figure 4.4 SBR headspace N 20, off-gas N 2 0 and ammonia concentration-time profiles (SBR-TS 16a) 65 Figure 4.5 SBR mixed liquor and headspace N 2 0 concentration-time profiles (MISC-22).... 66 Figure 4.6 SBR TOC and PFTB concentration-time profiles (SBR-TS 16a) 68 Figure 4.7 SBR anoxic-phase and aerobic-phase wastewater carbon utilization fractions (SBR-TS 16a) 69 Figure 4.8 Reactor liquid N 2 0 concentration-time profile for "total" stripping experiment with air flow rate = 876 mL/min (NOSE-3a) 77 Figure 4.9 Reactor N 2 0 "total" mass transfer coefficient (20°C) versus air flow rate 78 Figure 4.10 Reactor configuration for N 2 0 liquid-surface diffusion stripping experiments 79 Figure 4.11 Reactor headspace and liquid N 2 0 concentration-time profiles for liquid-surface diffusion stripping experiments using air flow rates of 1,372 mL/min (NODSE-la) and 514 mL/min (NODSE-2a) 80 Figure 4.12 Reactor headspace and liquid N 2 0 concentration-time profiles for total stripping (NOSE-2a ) and liquid-surface diffusion stripping (NODSE-la) experiments conducted at an air flow rate = 1,372 mL/min 81 Figure 4.13 Reactor 0 2 mass transfer coefficient (20°C) versus air flow rate 83 Figure 4.14 Reactor N 2 0 mass transfer coefficient versus 0 2 mass transfer coefficient (20°C) for air flow rates = 500 to 1,900 mL/min 85 ix Figure 4.15 SBR headspace (cg) and mixed liquor (C|) N 2 0 concentration-time profdes for a) MISC-22 (aeration rate = 1,040 mL/min, AOR = 11.8 mg N/L/hr), b) M1SC-23 (aeration rate = 1,510 mL/min, AOR = 17.0 mg N/L/hr), and c) MISC-24 (aeration rate = 670 mL/min, AOR = 6.5 mg N/L/hr) experiments.... 87 Figure 4.16 SBR headspace (cg) / mixed liquor (q) N 2 0 ratios versus elapsed time for a) MISC-22 (aeration rate = 1,040 mL/min, AOR = 11.8 mg N/L/hr), b) MISC-23 (aeration rate =1,510 mL/min, AOR = 17.0 mg N/L/hr), and c) MISC-24 (aeration rate = 670 mL/min, AOR = 6.5 mg N/L/hr) experiments... 88 Figure 4.17 SBR headspace (cg) concentration-, cumulative generated N 2 0 mass- and cumulative stripped N 2 0 mass-time profdes for a) MISC-22 (aeration rate = 1,040 mL/min, AOR = 11.8 mg N/L/hr), b) MISC-23 (aeration rate = 1,510 mL/min, AOR= 17.0 mg N/L/hr), and c) MISC-24 (aeration rate = 670 mL/min, AOR = 6.5 mg N/L/hr) experiments 90 Figure 4.18 SBR DO, pH, nitrite and nitrate concentration-time profdes; perturbation = no ammonia in wastewater (NOME-8)....; 95 Figure 4.19 SBR ammonia, nitrite, headspace N 2 0 and headspace NO concentration-time profdes; perturbation = no ammonia in wastewater (NOME-8) 97 Figure 4.20 SBR DO, pH, nitrite and off-gas N 2 0 concentration-time profdes; perturbation = nitrite spikes (NOME-3b) 102 Figure 4.21 SBR ammonia, nitrite, nitrate and pH concentration-time profdes; perturbation = - 50% wastewater ammonia (SBR-TS13f) 109 Figure 4.22 SBR off-gas N 2 0 concentration, mean slope-, perturbation/baseline ratio-and DO.concentration-time profdes; perturbation = - 50%. wastewater ammonia ( S B R - T S l-3f> •:. 110 Figure 4.23 SBR ammonia, nitrite, nitrate and pH concentration-time profiles; perturbation = + 50% wastewater ammonia (SBR-TS 13a) 112 Figure 4.24 SBR off-gas N 2 0 concentration-, mean slope-, perturbation/baseline ratio-and DO concentration-time profiles; perturbation = + 50% wastewater ammonia (SBR-TS 13a) ; 114 Figure 4.25 SBR DO, pH, off-gas N 2 0 and ammonia concentration-time profiles; perturbation = + 96% aeration air flow rate (SBR-TS 12a) 117 Figure 4.26 SBR TOC, PUB, nitrite and nitrate concentration-time profiles; perturbation = - 40% wastewater carbon (SBR-TS 14a) 122 Figure 4.27 SBR DO, ammonia, off-gas N 2 0 and nitrite concentration-time profiles; perturbation = - 40% wastewater carbon (SBR-TS 14a) 123 Figure 4.28 SBR off-gas N 2 0 concentration-, mean slope- and perturbation/baseline ratio-time profiles; perturbation = - 40% wastewater carbon (SBR-TS 14a) 126 x Figure 4.29 Carbonaceous OUR versus aerobic-phase carbon loading rate 131 Figure 4.30 SBR ammonia concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 133 Figure 4.31 The ratio in the ammonia oxidation rate (AOR) relative to the baseline cycle AOR versus the ratio of the induced carbonaceous OUR relative to the baseline cycle ammonia oxidation OUR (AO OUR) 135 Figure 4.32 SBR TOC concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 137 Figure 4.33 SBR PFD3 concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 138 •Figure 4.34 SBR nitrite concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 140 Figure 4.35 SBR nitrate concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 141 Figure 4.36 SBR DO concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 143 Figure 4.37 SBR pH-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS16c) SDC utilization rate experiments 144 Figure 4.38 SBR off-gas N 2 0 concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 146 Figure 4.39 SBR off-gas N 2 0 mean slope-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments '.. 147 Figure 4.40 SBR perturbation/baseline cycle off-gas N 2 0 ratio-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 148 Figure 4.41 SBR cumulative stripped N 2 0 mass-time profiles for a) low (SBR-TS 16a) and b) medium (SBR-TS16d) SDC utilization rate experiments 150 Figure 4.42 SBR DO, pH and off-gas N 2 0 concentration-time profiles (NOME-10a) 157 x i Figure 4.43 SBR mixed liquor N 2 0 concentration-time profdes for a) Baseline Test (DO = 7 mg/L), b) Test #1 (DO = 0.2 mg/L) and c) Test #2 (DO = 0.5 mg/L) (NOME-10a) 158 Figure 4.44 SBR headspace N 2 0 concentration-time profdes for a) Baseline Test (DO = 7 mg/L), b) Test #1 (DO = 0.2 mg/L) and c) Test #2 (DO = 0.5 mg/L) (NOME-10a) 160 Figure 4.45 SBR nitrate concentration-time profdes for a) Test #1 (DO = 0.2 mg/L) and b) Test #2 (DO = 0.5 mg/L) (NOME-lOa) .- 162 Figure 4.46 SBR DO, pH and off-gas N 2 0 concentration-time profdes (NOME-10b) 166 Figure 4.47 SBR mixed liquor N 2 0 concentration-time profdes for a) Baseline Test, b) Test #1 (9.5 mg C/L/hr SDC utilization rate) and c) Test #2 (19 mg C/L/hr SDC utilization rate) (NOME-lOb) 167 Figure 4.48 SBR headspace N 2 0 concentration-time profdes for a) Baseline Test, b) Test #1 (9.5 mg C/L/hr SDC utilization rate) and c) Test #2 (19 mg C/L/hr SDC utilization rate) (NOME-10b) 168 Figure 4.49 SBR nitrate concentration-time profdes for a) Baseline Test, b) Test #1 (9.5 mg C/L/hr SDC utilization rate) and c) Test #2 (19 mg C/L/hr SDC utilization rate) (NOME-10b) 170 Figure 4.50 Relative difference in ammonia oxidation rate versus relative difference in a) oxygen mass transfer coefficient and b) mean DO concentration for baseline wastewater - aeration rate experiments (SBR-TS8a/b, SBR-TS9a/b, SBR-TS 12a) 177 Figure 4.51 The ratio of generated N 2 0 mass to oxidized ammonia mass versus mean DO concentration for a) low K La (SBR-TS9a/b), b) medium K La (SBR-TS8a/b) and c) high K La (SBR-TS 12a) perturbation - baseline wastewater experiments; d) SBR generated N 2 0 / oxidized ammonia mass ratio as a function of time since SBR start up 179 Figure 4.52 SBR ammonia concentration-time profiles for a) + 15% K La (SBR-TS9a), b) + 34% K La (SBR-TS8a) and c) + 68% K La (SBR-TS 12a) baseline wastewater experiments 181 Figure 4.53 SBR nitrite concentration-time profiles for a) + 15% KLa (SBR-TS9a), b) + 34% K La (SBR-TS8a) and c) + 68% KLa (SBR-TS 12a). perturbation - baseline wastewater experiments ' 182 Figure 4.54 SBR off-gas N 2 0 concentration-time profiles for a) + 15% K La (SBR-TS9a), b) + 34% K La (SBR-TS8a) and c) + 68% KLa (SBR-TS 12a) perturbation - baseline wastewater experiments 183 xn Figure 4.55 SBR cumulative stripped N 2 0 mass-time profiles for a) + 15% KLa (SBR-TS9a), b) + 34% K La (SBR-TS8a) and c) + 68% KLa (SBR-TS 12a) perturbation -baseline wastewater experiments 184 Figure 4.56 SBR DO concentration-time profdes for a) + 15% K La (SBR-TS9a), b) + 34% K La (SBR-TS8a) and c) + 68% KLa (SBR-TS 12a) perturbation - baseline wastewater experiments 186 Figure 4.57 SBR pH-time profiles for a) + 15% K La (SBR-TS9a), b) + 34% KLa (SBR-TS8a) and c) + 68% KLa (SBR-TS 12a) perturbation-baseline wastewater experiments 190 Figure 4.58 Relative difference in ammonia oxidation rate versus relative difference in oxygen mass transfer coefficient for low (SBR-TS 18a/b) and medium (SBR-TS 17a/b) SDC utilization rate experiments 195 Figure 4.59 SBR off-gas N 2 0 concentration-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K La perturbation experiments 197 Figure 4.60 SBR cumulative stripped N 2 0 mass-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K La perturbation experiments 198 Figure 4.61 SBR DO concentration-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% KLa perturbation experiments 199 Figure 4.62 SBR pH-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K La perturbation experiments 200 Figure 4.63 SBR off-gas N 2 0 concentration-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS 18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K La - medium SDC utilization rate (SBR-TS 17a) experiments 202 Figure 4.64 SBR cumulative stripped N 2 0 mass-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS 18b), b) + 33% KLa - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K La - medium SDC utilization rate (SBR-TS 17a) experiments 203 Figure 4.65 SBR DO concentration-time profiles for a) + 15% KLa - low SDC utilization rate (SBR-TS18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K La - medium SDC utilization rate (SBR-TS 17a) experiments... 205 xiii Figure 4.66 SBR pH-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS 18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K,a - medium SDC utilization rate (SBR-TS 17a) experiments 207 Figure 4.67 SBR ammonia concentration-time profiles for a) + 33% K La (SBR-TS 19a) and b) + 68% K La (SBR-TS 19b) high SDC utilization rate experiments 208 Figure 4.68 SBR nitrite, off-gas N 20, cumulative stripped N 2 0 mass and DO profiles for + 68% K La - high SDC utilization rate experiment (SBR-TS 19b) 210 Figure 4.69 SBR pH-time profiles for a) + 33% K La (SBR-TS 19a) and b) + 68% K,a (SBR-TS 19b) high SDC utilization rate experiments 211 Figure 4.70 SBR nitrite concentration-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33%> K La - medium SDC utilization rate (SBR-TS17a) experiments 212 Figure 4.71 SBR DO and pH concentration-time profiles for NOME-3 and NOME-3b nitrite spike experiments; perturbation = nitrite spikes 219 Figure 4.72 SBR nitrite and off-gas N 2 0 concentration-time profiles for NOME-3 and NOME-3b nitrite spike experiments; perturbation = nitrite spikes 220 Figure 4.73 SBR nitrous acid concentration-time profiles for NOME-3 and NOME-3b nitrite spike experiments; perturbation = nitrite spikes 221 Figure 4.74 SBR pH, DO, ammonia and nitrate concentration-time profiles for low pH (nitrous acid) experiment (NOME-5b); perturbation = reduced pH 227 Figure 4.75 SBR nitrite, nitrous acid, off-gas N 2 0 concentration and off-gas N 2 0 mean slope profiles for low pH (nitrous acid) experiment (NOME-5b); perturbation = reduced pH 228 Figure 4.76 SBR pH, DO, ammonia and nitrate concentration-time profiles for high pH (nitrous acid) experiment (NOME-5c); perturbation = elevated pH 229 Figure 4.77 SBR nitrite, nitrous acid, off-gas N 2 0 concentration and off-gas N 2 0 mean slope profiles for high pH (nitrous acid) experiment (NOME-5c); perturbation = elevated pH 231 Figure 4.78 SBR off-gas N 2 0 and nitrite concentration-time profiles for a) SBR-TS 13a, b) SBR-TS 13f and c) SBR-TS 15 experiment baseline cycles 235 Figure 4.79 SBR off-gas N 2 0 and nitrous acid concentration-time profiles for a) SBR-TS 13a, b) SBR-TS 13f and c) SBR-TS 15 experiment baseline cycles 237 xiv Figure 4.80 SBR a) generated N 2 0 mass / oxidized ammonia mass ratio, b) mean aerobic-phase nitrite concentration and c) mean aerobic-phase DO concentration as a function of time since SBR start-up; d) generated N 2 0 mass / oxidized ammonia mass ratio versus mean aerobic-phase nitrite concentration 238 Figure 4.81 SBR a) specific ammonia oxidation rate as a function of time since SBR start up; b) generated N 2 0 mass / oxidized ammonia mass ratio Versus specific ammonia oxidation rate 239 Figure 4.82 SBR a) biomass-supernatant interface height after 30 min settling, b) effluent suspended solids concentration, c) mixed liquor suspended solids concentration and d) mixed liquor temperature as a function of time since SBR start up 244 Figure 4.83 SBR a) aerobic-phase final N mass / initial N mass ratio versus generated N 2 0 mass and b) "missing" N mass versus generated N 2 0 mass 246 Figure 4.84 Defined state variables for the conceptual N20-based aeration system control strategy 254 Figure 4.85 Elements of conceptual aeration system control strategy 256 Figure 4.86 SBR C0 2 concentration-time profiles for MISC-13 and MISC-14 experiments.... 258 xv N O M E N C L A T U R E AMO ammonia monooxygenase ANN artificial neural network ANAMMOX anaerobic ammonia oxidation AOB ammonia oxidizing bacteria AOR ammonia oxidation rate BEPR biological excess phosphorus removal BNR biological nutrient removal C carbon °C degrees Celsius CD control decision COD chemical oxygen demand c o 2 carbon dioxide d day DO dissolved oxygen g gram GC gas chromatograph H Henry coefficient HAO hydroxylamine oxidoreductase HC1 hydrochloric acid HN0 2 nitrous acid hr hour IR infra-red K Kelvin K L a volumetric gas mass transfer rate kPa kilopascal K 0 oxygen half-velocity coefficient K s half-velocity coefficient L litre m meter M molar mg milligram xvi min minute MISC miscellaneous experiment mL millilitre mM millimolar NaOH sodium hydroxide MLSS mixed liquor suspended solids MLVSS mixed liquor volatile suspended solids mm millimeter MW molecular weight N nitrogen NH3 unionized ("free") ammonia N H / ammonium NiR nitrite reductase NO nitric oxide NO2 nitrogen dioxide NO2" nitrite NO3" nitrate NOB nitrite oxidizing bacteria NH2OH hydroxylamine N 2 dinitrogen N 2 0 nitrous oxide N 2 O 4 dinitrogen tetroxide NOME nitrous oxide mechanism experiment NOSE nitrous oxide stripping experiment NODSE nitrous oxide diffusion stripping experiment O oxygen OLAND oxygen-limited nitrification-denitrification ORP oxidation-reduction potential OST on-set time OTE oxygen transfer efficiency OUR oxygen uptake rate PHB poly-P-hydroxybutyrate ' • -: ppm parts per million xvi 1 RDC readily degradable carbon RRT reducing rate time RS relative statistic R 2 correlation coefficient s second S substrate SBR sequencing batch reactor SBR-TS SBR tracking study SDC slowly degradable carbon SND simultaneous nitrification-denitrification SSNGR steady-state N 2 0 generation rate SRT solids retention time T temperature TEA terminal electron acceptor Tg teragram (1 x 1012 gram) TKN total Kjeldahl nitrogen TOC total organic carbon TSS total suspended solids M- specific organism growth rate Mm ax ; maximum specific organism growth rate Y biomass yield , XVll l A C K N O W L E D G M E N T S Like any significant pursuit in life, the decision to undertake and then carry out my doctoral studies involved many people. Dr. Donald S. Mavinic, P.Eng., my research supervisor, was a key catalyst in my decision to return to university. Dr. Mavinic supervised my M.A.Sc. research, and upon hearing my Ph.D. contemplation's, enthusiastically encouraged my continued studies. I have known Dr. Mavinic for over a decade and knew that the mentorship he provided early in my career would be an important asset in my most recent endeavor. His enthusiastic support, wise counsel and willingness to lend a listening ear were invaluable, and for such qualities I sincerely thank Dr. Mavinic. I also wish to acknowledge my research committee members, Dr. Ken J. Hall and Dr. Victor Lo from the University of British Columbia, Dr. William K. Oldham, P.Eng. (UBC professor emeritus, Oldham Environmental Engineering Ltd.) and Dr. David J.L. Forgie, P.Eng. (UBC adjunct professor, Associated Engineering (B.C.) Ltd.). I thank them for their willingness to participate in this project in the midst of busy professional schedules, as well as their constructive input and support for this particular research. A thesis cannot be written without first collecting data, and to this end I offer many thanks to Susan Harper and Paula Parkinson of the Environmental Engineering Laboratory, Department of Civil Engineering. This project would have truly not been possible without their expertise and assistance, and I thank them for also providing an enjoyable and safe working environment. I too thank Fred Koch, Research Associate, for his passion for research and challenging discussions. I also thank Scott Jackson and Harald Schrempp, from the Civil Engineering Workshop, for their assistance with the experimental systems. The many hours spent in the laboratory were shared with other graduate students and I am grateful for the company provided by Craig Bradley, Alessandro Monti and others. In particular, I would like to thank Robert Simm, with whom I worked closely over the course of my research, for the many discussions related to our projects, the consulting industry from which we came, and life in general. His friendship and good-natured humour made the long days pass quickly. I also thank Rob for his assistance in data collection and providing hydroxlyamine sample analysis. There was a time when the prospect of pursuing doctoral studies seemed an increasingly remote possibility given personal and professional responsibilities. To this end, I wish to acknowledge and xix thank Mr. Kerry Rudd, P.Eng., Vice President and General Manager, Associated Engineering (B.C.) Ltd., and Mr. J. Richard E. Corbett, M.A.Sc., P.Eng., Vice President - Environmental Engineering, Associated Engineering Group Ltd., for encouraging me to pursue further education, providing a financial means to assist this undertaking, and enthusiastically supporting me throughout this time. My doctoral program was financially supported by scholarships received from various funding agencies. Thus I would like to acknowledge and thank the University of British Columbia (University Graduate Fellowship and Jean MacDonald Graduate Fellowship), Natural Science and Engineering Research Council of Canada, Canadian Council of Professional Engineers - ENCON Group Ltd., and the British Columbia Water and Waste Association for their support. The role of family in ones life pathway cannot be underestimated. To this end, I wish to thank my parents (Alex and Doris), brother (Devin) and his family for their support and encouragement for all the times I told them I was once again returning to school, as well as during these studies. In particular, I would like to thank my wife, Nita, whom I met after I started this adventure. She was alongside me every step of the way and provided continual encouragement and unconditional understanding for the rigors of such pursuits. Finally, looking back at my academic and professional endeavors, I would be remiss if I did not acknowledge the science teachers at Lanigan Central High School who, likely unknowingly, stirred my scientific curiosity and influenced my career choice. For their enthusiastic teaching, I am grateful to Mr. Lome Skoronski, Mrs. Erna Alexander and the late Mr. Allistar Ingham. xx CHAPTER ONE INTRODUCTION 1.1 BACKGROUND Biological waste treatment has been a naturally occurring phenomenon since the dawn of time, as microorganisms participate in the recycling of carbon and nitrogen from the atmosphere to organic substances, and back to the atmosphere. Humans, more recently, have learned to harness these biological processes for wastewater management in order to protect our health and that of the greater ecological environment. In the United States, the first formal biological wastewater treatment process, an intermittent sand filter, was tried in Massachusetts in 1887, with the first activated sludge system commissioned in Texas in 1916 (WEF, 1992). While the biological processes utilized and exploited in engineered wastewater treatment systems have remained unchanged over time, our understanding of these processes has dramatically increased in the past few decades. These gains in process understanding are largely attributable to the development of measurement, analytical and computational techniques that allow enhanced investigation of the basic science of these processes. As a result, wastewater treatment practitioners have been able to expand wastewater treatment from carbon oxidation and removal to include nitrogen and phosphorus removal, further reducing the impact of effluent discharges to sensitive aquatic environments. Control of nitrogenous substances via biological nitrification-denitrification became a practical treatment objective in the 1960s (WEF, 1992), with biological excess phosphorus removal realizing relatively common application in the past two decades. Increasing population and industrial activities, combined with enhanced public sensitivity to environmental issues, continue to drive increased demands on the wastewater management industry for cost effective operation and enhanced performance and reliability of treatment systems. The United States Environmental Protection Agency (USEPA) 1986 Needs Survey projected that the majority of United States treatment plant capacity, over the following 20 years, would be "devoted to treatment performance levels above secondary treatment equivalency" (WEF, 1992). The costs of environmental protection are staggering - the United States government alone spent $50 billion dollars funding the construction of municipal wastewater collection, treatment and disposal systems between 1972 and 1987 1 (WEF, 1992). Recent (1997) US expenditures on wastewater systems were in the order of $10 billion annually, with an estimated $12 billion annual gap in infrastructure needs and current spending (Dessoff, 2001). The energy required for operation of municipal wastewater treatment facilities has been estimated to be as much as 0.3% of the total national energy requirement for the United States (WEF, 1997). To this end, the wastewater industry continues to find ways to provide enhanced treatment performance and reliability at reduced cost. For example, the International Water Association (IWA) Task Group on Mathematical Modelling for Design and Operation of Biological Wastewater Treatment Processes has expended considerable effort over the past few decades to provide treatment system designers with sophisticated tools for process design and optimization. Similarly, continued development of on-line instrumentation, coupled with advances in process control strategies that utilize the collected data, promises to enhance treatment system efficiency, in terms of operating costs, while increasing treatment performance and reliability to meet increasingly stringent effluent quality criteria. As will be shown, these concepts provided the impetus for the current research. 1.2 THE NEED FOR RESEARCH The need for continued research in the field of biological wastewater treatment and nitrogen removal, in the context of on-line process monitoring and control, is motivated by at least three factors, as discussed below. • Nitrification requirements. Under certain conditions, un-ionized ammonia (NH3) is toxic to aquatic organisms, with evidence suggesting that ionized ammonia (NH4+) may also be toxic (USEPA, 1999). Nitrite (NO2) is also toxic to aquatic organisms (British Columbia, 1989). Furthermore, ammonia and nitrite oxidation in aquatic systems consumes oxygen, potentially reducing oxygen concentrations to levels that are inhibitory to aquatic organisms. Many wastewaters contain significant quantities of nitrogen, including domestic wastewater, landfill leachate, agricultural wastes, and wastewater generated by industries such as food processing and fertilizer production. As a result, nitrogen control forms an important component of wastewater management. Implication -> The removal of reduced nitrogen species from wastewater will continue to be an important wastewater treatment objective in many situations. Therefore, treatment 2 system design and operation will continue to evolve to provide the most capital- and operating-cost effective solutions. • The role and importance of oxygen in nitrification and bioreactor oxygen transfer efficiency. Aerobic, autotrophic nitrification has long been recognized as a major energy sink in wastewater treatment facilities due to the large oxygen requirement of the biochemical reactions (i.e. 4.6 g 0 2 per g NH 4 + -N nitrified; USEPA, 1993). As an example, for municipal wastewater treatment systems, nitrification energy requirements related to oxygen supply can comprise up to 30% of the facilities total energy costs (WEF, 1997). Besides the stoichiometric nitrification oxygen requirement, the bioreactor oxygen supply rate/mixed liquor dissolved oxygen concentration has a large impact on nitrification kinetic rates. The relatively high oxygen half-saturation coefficient for nitrifying organisms causes the kinetic rate of these organisms to be very sensitive to mixed liquor dissolved oxygen (DO) concentrations below about 2 mg/L (Grady et al, 1999). As a result, it is a common industry practice to operate bioreactors with DO levels of at least 2 mg/L to ensure adequate nitrification rates. The efficiency of transferring oxygen into mixed liquor is also affected by DO concentration. However, unlike nitrification, where higher DO levels translate into higher kinetic rates (i.e. a positive impact), elevated DO levels negatively impact oxygen transfer by reducing the oxygen transfer efficiency (OTE). Figure 1.1 illustrates the opposing effect that DO concentration has on the ammonia oxidation rate (AOR) and OTE. The AOR is shown as a function of the maximum AOR, calculated using the Monod kinetic equation and a typical nitrifier oxygen half-saturation coefficient value (Grady et al, 1999). The OTE values were calculated using the two-film theory and typical characteristics of a fine-bubble diffuser system in an operating bioreactor (Metcalf and Eddy, 1991). Thus, it can be deduced from Figure 1.1 that the dilemma in bioreactor operation, with respect to controlling the oxygen supply rate, involves balancing the required ammonia oxidation rate with oxygen transfer efficiency, such that complete nitrification can be achieved under a given set of operating conditions, while maximizing oxygen transfer efficiency. As an example of inefficiency, the Figure 1.1 data show that operating the bioreactor with a DO concentration of 2.5 mg/L will result in a 21% reduction in OTE relative to operation at a 0.5 mg/L level, under the assumption that the kinetic rate provided at the lower DO level would indeed be satisfactory for the given situation. 3 D O (mg/L) Figure 1.1: Ammonia oxidation rate and oxygen transfer efficiency versus D O concentration 4 The reduction in OTE contributes to an increase in .energy, and cost, expended to supply the required oxygen. Implication —> Given the significance of the nitrification oxygen requirement in the context of the energy needed to supply it to the biomass, and the large fraction of this energy relative to the total energy required to operate a treatment facility, it can be seen that inefficient bioreactor operation can be costly in economic terms. Therefore, process monitoring and control systems must be able to recognize inefficient bioreactor operation and make control decisions to remove any inefficiencies. • Treatment process monitoring. The ability to monitor biological treatment processes in real-time provides the opportunity to control these processes in real-time. Such control allows more optimized system operation, resulting in reduced capital and operating costs. Dissolved oxygen and pH have seen the most application to on-line monitoring in the wastewater treatment industry. This likely results from the simple measurement techniques and availability of robust pH and DO measurement probes. However, on their own or together, DO and pH data provide limited information with respect to the microbial workings of a bioreactor. More sophisticated instrumentation is now available for on-line measurement of various parameters in wastewater, bioreactor mixed liquor and effluent. For example, wet-chemistry and biosensor instruments can be used to measure carbon and nutrient levels. Respirometers and titrimetric instruments provide information on biomass activity. While the information obtained from such systems can provide significant insight into bioreactor performance, these types of instruments are expensive to purchase, and some consume chemical reagents that induce operating costs and waste disposal issues. Furthermore, system reliability and maintenance requirements are also potentially disadvantageous. Finally, none of these parameters provide "direct" information on the biomass response to oxygen availability-limitation, in the context of nitrification. Implication -> The described constraints of historical, as well as more recently developed, process monitoring parameters suggests the need to pursue the use of alternative, non-traditional parameters for process monitoring and control applications. Ideally, the instruments/sensors used to measure the parameter quantities should be accurate, inexpensive, simple, robust and easy to maintain. 5 1.3 RESEARCH RATIONALE AND OBJECTIVES The possible solution to the aforementioned dilemma of bioreactor operation (i.e. supplying oxygen at a sufficient rate to provide, the required nitrification rate and simultaneously maximizing oxygen transfer efficiency), while avoiding the constraints/difficulties of traditional monitoring approaches, may be available in the biochemistry of ammonia oxidizing bacteria (AOB). Autotrophic AOB use molecular oxygen (i.e. 02) for two key purposes during the oxidation of ammonia to nitrite (Bock and Wagner, 2001). First, molecular 0 2 , not water, is the source of oxygen in hydroxylamine (NH2OH), the product of the first step of ammonia oxidation. Second, molecular 0 2 serves as the "normal" terminal electron acceptor (TEA) for the AOB electron transport chain. However, AOB, such as Nitrosomonas, can possess a nitrite reductase (NiR) enzyme (Poth and Focht, 1985; Bock and Wagner, 2001). This enzyme allows the organism to use an alternate terminal electron acceptor (i.e. nitrite, N02") in addition to molecular oxygen. In this biochemical pathway, hydroxylamine, produced during ammonia oxidation, is oxidized with the concurrent reduction of nitrite to gaseous nitrous oxide (N20). The NiR enzyme is inducible in oxygen-limited environments, and its presence in AOB cells has been partly viewed as a survival mechanism to conserve oxygen for the initial oxidase step of ammonia oxidation (Poth and Focht, 1985). Under aerobic conditions, molecular 0 2 is also the TEA for nitrite oxidizing bacteria (NOB) and heterotrophic, carbon oxidizing bacteria. Thus, in a mixed microbial population that exists in a wastewater treatment bioreactor, the three groups of organisms (i.e. AOB, NOB, heterotrophs) compete for available oxygen, possibly inducing an AOB oxygen limitation and potentially impacting the nitrification rate. For this research, it was hypothesized that reactor off-gas N 2 0 data could be used to monitor the extent of AOB oxygen limitation-availability and provide an indication of the overall nitrification kinetic rate. This information could potentially be used to control the bioreactor aeration system in a manner that maximized oxygen transfer efficiency while simultaneously providing a suitably high nitrification rate. Like nitrite and nitrate, N 2 0 is a biochemical reaction product that, unlike DO, provides information on the state of nitrification. However, unlike nitrate, for example, N 2 0 could provide a direct indication of the AOB oxygen limitation. Furthermore, reactor off-gas N 2 0 has many practical advantages as a monitoring parameter, when compared to more traditional parameters such as ammonia and nitrate: simple and inexpensive measurement technology (i.e. infra-red), requires no chemical reagents, no sample waste disposal issues, and it is essentially maintenance free since there is no contact with 6 bioreactor mixed liquor. The advantages of infra-red gas measurement technology, originally developed for the medical and industrial fields, make it ideally suited to biological wastewater treatment systems. As will be discussed in Section 2, very little research has been conducted to date regarding the potential use of N 2 0 as either a process monitoring parameter or an aeration system control parameter in biological wastewater treatment systems. Furthermore, none of the conducted research focused on sequencing batch reactor (SBR) systems. Therefore, the main objective of this study was as follows: Investigate the feasibility of using off-gas nitrous oxide (N20) as a nitrification monitoring and aeration system control parameter in SBR wastewater treatment systems. The specific research objectives, that were pursued to provide information to support the main objective, can be summarized as follows: 1) Identify/confirm the principle biological N 2 0 generation mechanism during the aerobic-phase of the SBR cycle. 2) Study the reactor N 2 0 mass transfer characteristics under clean water and operating bioreactor conditions. 3) Investigate the effect that wastewater components (i.e. ammonia, readily degradable carbon, slowly degradable carbon) have on SBR response (i.e. nitrification kinetics, N 2 0 generation) when operated using the baseline aeration rate. 4) Evaluate the effect that aeration rate perturbations have on SBR response (i.e. nitrification kinetics, N 2 0 generation) when treating the baseline wastewater, as well as wastewater with modified components. 5) Develop a conceptual, N20-based, SBR aeration system control strategy. The research objectives were accomplished by subjecting an anoxic-aerobic SBR system, operating under oxygen-limited conditions, to wastewater (i.e. ammonia, readily degradable carbon, slowly degradable carbon (SDC)) and aeration rate perturbations, outside the normal baseline operating 7 conditions, and monitoring system response. Specific experiments were also conducted to confirm the source of generated N 20, investigate the effects of DO concentration and slowly degradable carbon utilization rate on aerobic-phase heterotrophic N 2 0 reduction, examine the influence of nitrite and nitrous acid levels on N 2 0 generation, and evaluate the reactor gas mass transfer characteristics. 8 C H A P T E R T W O L I T E R A T U R E R E V I E W 2.1 INTRODUCTION The literature review presents information related to the current research. Section 2.2 examines biological nitrogen and carbon conversion processes applicable to wastewater treatment systems, including nitrification, denitrification, N 2 0 generation, and carbon storage and accumulation. Section 2.3 presents the characteristics of N 2 0, including its physical and chemical properties, and environmental fate and effects. Section 2.4 discusses related wastewater treatment topics, including sequencing batch reactor (SBR) treatment systems, nitrification and oxygen issues, N 2 0 generation and emission control, and process monitoring and control parameters. 2.2 BIOLOGICAL NITROGEN AND CARBON CONVERSION Biological nitrogen and carbon removal wastewater treatment systems utilize a diverse microbial population to remove nitrogenous and carbonaceous compounds from wastewater. Nitrogen removal involves the conversion of reduced nitrogen compounds (i.e. ammonia) to a more oxidized species (i.e. nitrate), termed nitrification, followed by the conversion of the oxidized species to reduced gaseous products (i.e. dinitrogen gas, N2), termed denitrification. Carbon removal involves oxidation of reduced, organic carbon species to carbon dioxide, with a portion of the carbon used to synthesize new biomass. 2.2.1 Nitrification Heterotrophic Nitrification Wastewater system nitrification is generally believed to be dominated by autotrophic bacteria (Grady et al, 1999), but it is worthwhile to first examine the role of other organisms in nitrification. Over one hundred heterotrophic species have been identified with the ability to form nitrite from ammonia (Grady et al, 1999), spread across diverse groups of bacteria, fungi and algae (Bock et al, 1991). Fungi are considered to be the most common heterotrophic nitrifying organisms (Wrage et al, 2001). Nitrate production by heterotrophic nitrifying organisms is thought to be less common than nitrite production (Papen et al, 1989). 9 Heterotrophic nitrification involves co-oxidation of ammonia and carbon, where ammonia oxidation is not coupled to energy generation (Bock et al, 1991). It appears that heterotrophic nitrifiers are able to oxidize reduced nitrogen compounds only in the presence of an external organic carbon source, suggesting carbon oxidation generates energy for cell growth (Zhao, 1998), a distinct difference from the autotrophic nitrifiers. Furthermore, while the substrates and products of heterotrophic nitrifiers are the same as those of the autotrophic nitrifiers, the enzymes involved in the conversions (i.e. ammonia monooxygenanse and hydroxylamine oxidoreductase) appear to have differences with respect to substrate inhibition and structure (Wrage et al, 2001). Acidic conditions, high oxygen concentrations, and availability of organic carbon are the main environmental conditions which appear to favour heterotrophic nitrification (Wrage et al, 2001). In addition, Bock et al (1991) noted that heterotrophic nitrifiers predominantly use organic nitrogen substrates, although Papen et al (1989) found that ammonia could be utilized as the sole substrate. Unit heterotrophic nitrification rates are lower than autotrophic rates, but the fact that heterotrophs grow faster than the autotrophs increases the overall heterotrophic nitrification rate relative to the overall autotrophic rate, and could allow the heterotrophs to "... make a significant contribution to the total nitrification under some conditions" (Kuenen and Robertson, 1994). In general, environments unfavourable to autotrophic nitrifiers are thought to preferentially allow heterotrophic nitrification (Schmidt et al, 2003). Therefore, the various environmental conditions required for heterotrophic nitrifier growth may explain why nitrification in wastewater treatment systems is generally considered to be the result of predominantly autotrophic organisms. Aerobic Autotrophic Nitrification Aerobic, chemolithoautotrophic nitrification is characterized by inorganic substrates serving as energy sources for cellular growth, with carbon dioxide fixation providing the required carbon for cell synthesis (Bock et al, 1991). Given its apparent dominance in wastewater treatment systems, with respect to nitrification, it is worthwhile to closely examine the details of this mechanism. Aerobic, autotrophic nitrification can be broken down into two main steps, ammonia oxidation and nitrite oxidation, provided by two distinct groups of bacteria. Overall ammonia oxidation to nitrite (Eqn. 2-1 and 2-2), which exists in equilibrium with nitrous acid (HN02), is believed to be a two-stage process with hydroxylamine (NH2OH) as an intermediate (Bock et al, 1991; Colliver and Stephenson, 2000). 10 Equation 2-3 illustrates the terminal oxidase reaction with molecular oxygen as the electron acceptor. NH 3 + 0.5 0 2 + 2 H + + 2 e" -> NH 2OH + H 2 0 NH 2OH + H 2 0 -> HN0 2 + 4 hf + 4 e" (2-1) (2-2) 0.5 0 2 + 2 H + + 2e ->H 2 0 (2-3) The catalyzing enzyme for ammonia oxidation is thought to be cytoplasmic membrane-bound ammonia monooxygenase (AMO) (Bock and Wagner, 2001). As shown in Equation 2-1, unionized or free ammonia is theorized to be the actual substrate for the ammonia oxidizing bacteria (AOB) for hydroxylamine generation. Molecular oxygen has traditionally been the assumed source of oxygen in hydroxylamine. However, research findings by Schmidt et al (2001) indicate that dinitrogen tetroxide (N204) is the actual oxidizing agent. They suggest that gaseous nitric oxide (NO), in addition to hydroxylamine, is produced during ammonia oxidation. In the presence of oxygen, the NO is oxidized to gaseous nitrogen dioxide (N02), with the dimeric form of nitrogen dioxide (i.e. dinitrogen tetroxide, N 20 4) being the actual oxidizing agent. The key point of their model is that molecular oxygen is still required for the reaction, but in a less direct manner. Although ammonia oxidation generates hydroxylamine, the latter compound remains undetected in the bulk solution under normal, balanced AOB growth (Simm et al, 2004d). However, Simm et al (2004d) did detect measurable bulk solution hydroxylamine in wastewater bioreactors that were subjected to extreme perturbations (i.e. high pH, low dissolved oxygen) outside the normal operating range. The ammonia oxidation step does not release electrons and thus does not contribute to energy production, but in fact requires two electrons to "drive" AMO (McCarty, 1999; Bock and Wagner, 2001). These electrons originate from the oxidation of hydroxylamine to nitrite-nitrous acid (Equation 2-2), with the reaction catalyzed by the hydroxylamine oxidoreductase enzyme (HAO) located in the AOB periplasmic space between the cell wall and the cytoplasmic membrane (Bock and Wagner, 2001). Water is the source of the second oxygen atom in nitrite. A total of four electrons are generated from hydroxylamine oxidation, with the remaining two electrons directed to the terminal oxidase (Eqn. 2-3), normally molecular oxygen, for energy production. 11 It should be noted that AMO is a non-specific enzyme that will catalyze the oxidation of at least forty known organic substrates (McCarty, 1999) that include methane, carbon monoxide, and aliphatic and aromatic hydrocarbons (Bock and Wagner, 2001). Oxidation of such "suicide" substrates (i.e. substrates that will not result in energy production) preludes ammonia oxidation and hydroxylamine production, eliminating the cellular energy generation mechanism provided by hydroxylamine oxidation. Thus, AOB are inhibited due to i) the competition between ammonia and the other substrates for the AMO binding site (McCarty, 1999) and ii) the resulting loss of the energy generation mechanism. Nitrosomonas organisms have traditionally been assumed to be the main AOB in wastewater treatment systems (Grady et al, 1999; Metcalf and Eddy, 1991). The relatively recent application of molecular-based genotypic analysis techniques appears to have confirmed the dominance of Nitrosomonas in wastewater systems (Grady, 2002). Aerobic, chemolithoautotrophic nitrite oxidation refers to the conversion of nitrite to nitrate, with the reactions shown below, where nitric acid (HNO3) exists in equilibrium with nitrate (Hooper, 1989; Bock et al, 1991). As shown in Eqn. 2-4, the additional oxygen atom in nitrate is derived from water (Bock et al, 1991). Equation 2-5 represents the terminal oxidase pathway, where molecular oxygen is the electron acceptor. HN0 2 + H 2 0 -> H N O 3 + 2 H + + 2 e" (2-4) 2 FT + 2 e" + 0.5 0 2 -> H 2 0 (2-5) Nitrite oxidation is catalyzed by the membrane-bound nitrite oxidoreductase enzyme, with this enzyme also capable of reducing nitrate to nitrite in the absence of oxygen (Bock and Wagner, 2001). Similar to free ammonia for the AOB, nitrous acid may be the actual substrate for nitrite oxidizing bacteria (NOB). Grady et al (1999) noted the possibility of nitrous acid, rather than nitrite proper, as the real substrate for Nitrobacter organisms. However, comprehensive biochemical descriptions of nitrite oxidation (e.g. Bock and Wagner, 2001; Bock et al, 1991; Hooper, 1989) do not explicitly identify the true NOB substrate. Although heterotrophic growth of autotrophic AOB has not been observed (Bock et al, 1991), some NOB (i.e. specific Nitrobacter strains) can grow using various carbon sources (e.g. acetate, pyruvate, glycerol, formate, a-oxoglutarate) as electron donors and carbon sources. Such growth has been characterized as "slow" under both aerobic and anaerobic conditions (Bock et al, 1991). 12 Nitrobacter organisms have been traditionally assumed to be the dominant NOB in wastewater treatment systems (Grady et al, 1999; Metcalf and Eddy,. 1991). However, recent molecular-based surveys suggest that Nitrospira organisms may be the dominant NOB in wastewater systems (Grady, 2002). Anaerobic Autotrophic Nitrification Anaerobic ammonium oxidation (ANAMMOX) refers to the chemolithoautotrophic oxidation of ammonium to dinitrogen gas (N2) using nitrite as an electron acceptor under strict anoxic conditions (Jetten et al, 1999, Schmidt et al, 2003). Although the process was predicted to exist in the 1970s, actual discovery was first reported only in 1990 (Jetten et al, 1999). The ANAMMOX reaction has been shown to be carried out by "conventional" aerobic AOB such as Nitrosomonas eutropha. However, the amount of generated energy is believed to be sufficient only for cell survival and not growth (Jetten et al, 1999). Alternately, Strous et al (1999) identified a planctomycete, Brocadia anammoxidans, that appears to be main organism responsible for the observed ANAMMOX reaction (Wrage et al, 2001). A second planctomycete, Kuenenia stuttgartiensis, has also been identified with the ANAMMOX reaction (Schmidt et al, 2003). The research by Jetten et al (1999) found that the at-the-time unidentified planctomycete bacteria had a ANAMMOX reaction rate that was 25 times faster than Nitrosomonas, but still had an extremely slow doubling time of 11 days. 2.2.2 Denitrification and N 2 0 Generation Heterotrophic Processes Heterotrophic denitrification under fully anoxic conditions is a familiar biochemical pathway utilized in wastewater treatment systems. The pathway involves the step-wise reduction of nitrate to N 2 via nitrate respiration, nitrite respiration combined with NO reduction, and N 2 0 respiration (Zumft, 1997). Under these conditions, denitrification can be accomplished by several genera of bacteria including Achromobacter, Aerobacter, Alcaligenes, Bacillus, Brevbacterium, Flavobacterium, Lactobacillus, Micrococcus, Proteus, Pseudomonas and Spirillum (Metcalf and Eddy, 1991). These denitrifying bacteria are heterotrophic organisms; therefore, they require organic carbon for both cell synthesis as well as an electron donor for energy production. In this manner, denitrification also provides oxidation of wastewater organic carbon. These bacteria are also facultative, thus they have the ability to use oxygen, as well as nitrate and nitrite, as electron acceptors during energy production by modifying enzymes in their metabolic systems (USEPA, 1993). Energy production is most efficient when oxygen is 13 the electron acceptor; hence, the bacteria will preferentially utilize oxygen over nitrate or nitrite. Therefore, the absence of oxygen is important for the reduction of nitrate and nitrite. As indicated above, N 20, along with more commonly recognized nitrate and nitrite, is also a denitrification substrate. Wicht (1996) found that the maximum, uninhibited N 2 0 reduction rate was approximately four times faster than the nitrate and nitrite reduction rates. Therefore, the large differences in rates, under normal denitrifying conditions, explains why N 2 0 is usually undetectable and often neglected in discussions on denitrification. However, under atypical denitrifying conditions, measurable N 2 0 can be generated. The literature indicates two main possibilities for this type of N 2 0 generation: (i) a carbon limitation that prevents denitrification from reaching the N 2 endpoint and (ii) selective inhibition of the N 2 0 reductase enzyme that causes the net accumulation of N 20. Examples of relevant studies are discussed below: • Anoxic, heterotrophic denitrification, wastewater experiments conducted by several researchers (Hanaki et al, 1992; Hong et al, 1993; Wicht, 1996; Itokawa et al, 2001) all found increasing N 2 0 generation with reduced wastewater carbon/nitrogen (e.g. COD/NO3"),ratios. Itokawa et al (2001) observed that N 2 6 was generated during endogenous denitrification once the supplied wastewater carbon was utilized, possibly since "... it is expected that electron supply limits reduction reactions." Furthermore, Wrage et al (2001) noted the preference of nitrate over N 2 0 as an electron acceptor in soil denitrification. Low wastewater carbon/nitrogen ratios imply elevated reactor nitrate levels due to incomplete denitrification caused by a carbon limitation, creating the potential situation for preferential use of nitrate, thus resulting in N 2 0 accumulation. Although carbon limitation appeared to be responsible for N 2 0 generation, both Hanaki et al (1992) and Itokawa et al (2001) observed nitrite accumulation under the experimental conditions. In subsequent experiments conducted by Itokawa et al (2001), nitrite spikes resulted in immediate N 2 0 generation, and suggested that endogenous denitrification in the presence of only nitrate would not necessarily generate N 2 0 without the "toxic effect" of nitrite, as described below. Furthermore, from their experimental results, von Schulthess et al (1995) explained that the small increase in N 2 0 generation during carbon-limited denitrification was likely related to the increased nitrite concentration rather than a preference of the organisms to use nitrate-nitrite as electron acceptors over N 20. • Similar to the possible effect of nitrite on anoxic N 2 0 generation, as reported by Itokawa et al (2001), other researchers have considered that N 2 0 generation can be induced by "selective", or more 14 enhanced, inhibition of the N 2 0 reductase enzyme over the other denitrifying enzymes. Nitrite and NO have a high affinity to metal ions (e.g. iron-sulphur, copper) located in the active site of enzymes, possibly causing an inhibitory effect to all the denitrifying enzymes (von Schulthess et al, 1995). The proposed mechanism for selective N 2 0 reductase inhibition, induced by exposure to high nitrite levels, is the inhibition of NO reductase by nitrite, followed by NO accumulation, with the subsequent inhibition of N 2 0 reductase the apparent result of the high NO concentration (von Schulthess et al, 1995). The presence of dissolved oxygen has been found to induce net N 2 0 generation in strictly heterotrophic biomass cultured under normally fully anoxic conditions (von Schulthess et al, 1994). While DO was found to inhibit all denitrifying enzymes, N 2 0 reductase was most significantly affected, resulting in N 2 0 accumulation. Thorn and Sorensson (1996) found that low anoxic zone pH levels induced maximum N 2 0 generation, but provided no hypothesis for the observations. This observation could be the result of the dissociation equilibrium between nitrite and nitrous acid, controlled by pH, and inhibition of N 2 0 reduction by nitrous acid rather than nitrite proper (Wicht, 1996). Schonharting et al (1998) found that low concentrations of hydrogen sulphide can strongly inhibit N 2 0 reduction, with the inhibitory effect related to the pH-controlled "free membrane-permeable" hydrogen sulphide concentration. They found that the inhibitory effect of hydrogen sulphide on N 2 0 reductase was stronger than on nitrate or nitrite reductase. Schonharting et al (1998) also found that nitrate had a strong inhibitory effect on nitrite and N 2 0 reduction in the presence of acetic acid, reportedly consistent with the findings of other researchers. Unlike conventional heterotrophs, heterotrophic nitrifying organisms, or at least some of them, have the ability to denitrify under fully aerobic conditions (Wrage et al, 2001). As a result, N 2 0 would be an intermediate during aerobic, heterotrophic denitrification. The extent of N 2 0 accumulation, under these conditions, would presumably depend on the previously discussed N 2 0 reductase inhibition mechanisms. Studies by Anderson et al (1993) and Otte et al (1996) have demonstrated aerobic N 2 0 generation from the heterotrophic nitrifier Alcaligene faecalis. N 2 0 generation was observed under fully anaerobic to fully aerobic conditions, but was maximized under oxygen-limited conditions (i.e. 5% oxygen saturation) (Otte et al, 1996). The researchers also indicated that the effects of oxygen 15 on denitrification enzyme expression and inhibition appeared to differ between reductase enzymes and organisms. Finally, Itokawa et al (2001), citing the work of Greenberg and Becker (1977) and others, noted that some heterotrophic denitrifying organism may form N 2 0 as their main reaction end product, presumably due to the absence of a N 2 0 reductase enzyme. Based on experimental results for a wastewater treatment system, Hanaki et al (1992) suggested that short solids retention times (i.e. SRT < 10 d), in combination with and depending on the amount of available carbon, may have selected for organisms that generate N 2 0 as their denitrification end product. As a related aside, simultaneous nitrification-denitrification (SND) is a term often utilized in the wastewater treatment industry to describe nitrogen loss, from the mixed liquor, in the aerobic zone of bioreactors. In this context, SND is traditionally explained to be the result of a combination of "conventional" autotrophic nitrification (i.e. oxidation of ammonia to nitrite and nitrate) and heterotrophic denitrification (i.e. reduction of nitrate and nitrite to dinitrogen gas) (Munch et al, 1996); this results from dissolved oxygen (DO) concentration gradients within the reactor and/or within microbial floes (Grady et al, 1999). Provision of this type of SND requires an oxygen transfer system that provides the macro- and/or micro-scale DO gradients, control of oxygen delivery to the bioreactor, and a sufficiently long solids retention time (SRT) that provides full nitrification at very low DO levels (Grady et al, 1999). For the microbial floe SND, the DO gradient within the floes is the result of diffusional limitations, with autotrophic nitrifying organisms existing in regions of high DO levels and the heterotrophic denitrifying organisms present in the regions of low DO (Munch et al, 1996). Beyond the conventional definition of SND, the wastewater treatment industry has begun to acknowledge the existence of other processes (e.g. autotrophic denitrification) that may also contribute to the observed SND (Munch et al, 1996). The following section further examines these processes. Autotrophic Processes A number of different autotrophic processes provide "denitrification", some of which are coupled with N 2 0 generation. Other processes may provide N 2 0 generation but no denitrification per se. This section provides an overview of the various processes: • Section 2.2.1 noted that molecular oxygen was the normal terminal electron acceptor for autotrophic 16 ammonia oxidizing bacteria (AOB), such as Nitrosomonas, functioning in aerobic environments. However, as will be shown, nitrite can serve as an alternate terminal electron acceptor in oxygen-limited environments. • An N-labelling study conducted by Ritchie and Nicholas (1972) found that Nitrosomonas europaea cells could generate N 2 0 via several mechanisms under aerobic and anaerobic conditions, and confirmed the autotrophic N 2 0 generation observed by earlier researchers. However, based on statements made by Poth and Focht (1985), there appeared to be uncertainty with respect to whether "nitrification" was the direct source of all of the generated N 2 0, via a nitrification-unstable intermediate route, or whether the AOB (i.e. Nitrosomonas europaea) were capable of "denitrification" by reducing nitrite to N 2 0 "... under conditions of oxygen stress while it is actively oxidizing ammonium". The subsequent N-labelling and kinetic analysis study conducted by Poth and Focht (1985) was credited as being the first study to demonstrate that nitrite could indeed act as an electron acceptor, during oxygen-limited nitrification, with the resultant production of N 2 0 (Anderson et al, 1993). Their findings suggested that Nitrosomonas contained a nitrite reductase (NiR) enzyme, supporting the earlier enzymatic study of Hooper (1968) that first identified such an enzyme in Nitrosomonas europaea (Wrage et al, 2001). The NiR enzyme is inducible in oxygen-limited environments (Hooper, 1989), and its presence in AOB cells has been partly viewed as a survival mechanism to conserve oxygen for the initial oxidase step of ammonia oxidation (Poth and Focht, 1985). Other postulated benefits of such an AOB pathway include the removal of a potentially toxic metabolic product (i.e. nitrite), and reducing the competition for oxygen by consuming the nitrite oxidizing bacteria substrate (i.e. nitrite) (Poth and Focht, 1985). In this biochemical pathway, hydroxylamine (NH2OH), produced during ammonia oxidation, is oxidized with the concurrent reduction of nitrite to gaseous N 20. Electrons flow from the hydroxylamine oxidoreductase enzyme to the terminal oxidases via cytochrome C554 -» cytochrome C552 -> to terminal oxidase (Bock and Wagner, 2001). The model of Schmidt, shown in Bock and Wagner (2001), indicates cytochrome c 5 5 2 as the electron branch, where electrons can flow to one of two terminal oxidase pathways. The first pathway involves molecular oxygen as the terminal electron acceptor, where it is reduced to water upon electron acceptance. The second pathway involves a linear sequence of nitrite reductase -> NO reductase —> N 2 0 reductase enzymes, thought to be located in the AOB periplasmic space (Bock and Wagner, 2001). Nitrite is first reduced to NO and then to N 20. Experimental studies have shown that Nitrosomonas cells (e.g. Anderson et al, 17 1993) can generate measurable NO under oxygen-limited conditions. However, as noted in Section 2.2.1, NO is oxidized to N0 2 in the presence of oxygen (Schmidt et al, 2001), presumably making it difficult to quantify NO generation in the absence of N0 2 data. The work of Poth (1986) and Bock et al (1995) has shown that some strains of Nitrosomonas can generate dinitrogen (N2) gas. Therefore, the Schmidt model includes an N 2 0 reductase enzyme that provides the reduction of N 2 0 to N 2 . The N 2 0 reductase enzyme has yet to be isolated from these organisms (Bock and Wagner, 2001). It should be noted here that the so-called OLAND (oxygen-limited autotrophic nitrification-denitrification) process (Kuai and Verstraete, 1998; Verstraete and Philips, 1998) is based on an biochemical pathway where N 2 , rather than N 20, is the reaction end product. Schmidt et al (2003) noted that the actual OLAND mechanism was unknown, but when combined with data from Kuai . and Vertraete (1998) that clearly show significant N 2 0 generation in their OLAND experiments, seems to suggest that conventional AOB, with the NiR enzyme, play a role in the OLAND process. Subsequent research confirmed that AOB and ANAMMOX organisms exist in the OLAND process (Wyffels et al, 2004; Pynaert et al, 2004). • In the preceding model, it is important to note, that hydroxylamine is the. electron donor, with nitrite as the electron acceptor and the source of nitrogen in the generated N 20. For this model, the oxidation of hydroxylamine itself does not generate N 2 0 (Bock and Wagner, 2001). However, in contrast to the conclusions of Poth and Focht (1985), Hooper et al (1990) noted the possibility of "direct oxidative production" of N 2 0 during ammonia oxidation, based on earlier work with isolated (i.e. in-vitro) HAO (hydroxylamine oxidoreductase) enzyme. Isolated HAO has been shown to produce NO and N 2 0 during hydroxylamine oxidation. The Nitrosomonas europaea cell N-labelling study conducted by Hooper et al (1990) found that some of the generated N 2 0 was produced by the direct oxidation of ammonia. However, the findings were predicated on an assumption related to the exchange of oxygen between water and nitrite during ammonia oxidation. Thus, the observed results contained some uncertainty. More recently, Beaumont et al (2002) disrupted the gene that encodes the nitrite reductase enzyme in Nitrosomonas europaea, resulting in NiR deficient cells. However, the cells were still able to generate NO and N 20. The researchers indicated the alternative pathway for gas production could involve the HAO enzyme, providing further evidence of a direct oxidative pathway for N 2 0 18 generation. • Anaerobic ammonia oxidation (ANAMMOX), discussed in Section 2.2.1, can also be described as an autotrophic denitrification process. Here, nitrite also acts as a electron acceptor for ammonia oxidation but the reaction product is N 2 rather than N 20. Furthermore, the ANAMMOX organisms perform the conversion in the absence of oxygen. However, such an anoxic environment can be also found in aerated bioreactors, presumably within the biomass floes, based on the reported synergistic growth of aerobic AOB with ANAMMOX organisms (e.g. Sliekers et al, 2002; Schmidt et al, 2003). • Finally, recent reviews of autotrophic nitrifier denitrification (Wrage et al, 2001; Colliver and Stephenson, 2000) have cited the work by Freitag et al (1987) where they reported N 2 0 generation via NOB (i.e. Nitrobacter). The proposed pathway was anaerobic reduction of nitrate to N 20, with pyruvate as the electron donor. Wrage et al (2001) acknowledged the limited available information regarding this N 2 0 generation pathway, consistent with its absence in comprehensive reviews of autotrophic nitrification provided by Bock and Wagner (2001) and Bock et al (1991). 2.2.3 Biological Carbon Conversion General Description Biological treatment processes utilize a mixed population of microorganisms to remove carbon from wastewater. For most biological carbon removal processes, heterotrophic bacteria (i.e. single-celled prokaryotic organisms that utilize organic carbon as both an electron donor for energy production and a carbon source for cell synthesis) are the dominant biomass microorganisms, supplemented by eucaryotic organisms such as fungi, protozoa and rotifers, and sometimes algae (Metcalf and Eddy, 1991). In general terms, under aerobic conditions where oxygen acts as the terminal electron acceptor for energy production, biological treatment provides removal of carbon from wastewater via enzyme catalyzed oxidation of a portion of the organic carbon to carbon dioxide, with the remaining portion of un-oxidized carbon synthesized into new cells (i.e. biomass). Separation of the biomass from the treated wastewater provides the clarified final effluent, which will exert a significantly reduced biochemical oxygen demand (e.g. 90 to 95%) in the receiving aquatic environment. Wastewater carbon oxidation can also occur in an unaerated zone of a bioreactor, where nitrite and 19 nitrate are available as terminal electron acceptors, rather than free oxygen. This process is called denitrification, as described in Section 2.2.2. Carbon Storage and Accumulation Carbon uptake, and intracelluar storage, is recognized as a key mechanism in the provision of biological excess phosphorus removal (BEPR) exploited in wastewater treatment systems (e.g. Mino et al, 1998). However, in non-BEPR systems, storage polymers were first investigated in the context of kinetic selection of non-filamentous organisms to improve the biomass settling and compressibility characteristics in secondary clarifiers (e.g. Cech and Chudoba, 1983; Chudoba et al, 1985). While researchers (e.g. Majone et al, 1996; Beccari et al, 1998) continued to investigate the role of storage polymers in the kinetic selection of non-filamentous organisms, the importance of substrate storage in the overall heterotrophic process has received considerable attention in recent years, particularly in the context of process modelling (e.g. Gujer et al, 1999; Majone et al, 1999; van Loosdrecht and Henze, 1999; Krishna and van Loosdrecht, 1999; Carucci et al, 2001). One of the main features of the Activated Sludge Model No. 3 (ASM3), developed by the International Association on Water Quality (IAWQ) Task Group on Mathematical Modelling for Design and Operation of Biological Wastewater Treatment Processes, was the incorporation of storage processes into the model (Gujer et al, 1999). Therefore, the potential significance of carbon storage, and its possible impact on treatment system performance, requires consideration when contemplating systems that provide biological nitrogen removal in conjunction with carbon removal. Full-scale treatment systems function under transient loading and operating conditions, with wastewater containing a variety of readily and slowly degradable carbon substrates. As a result, during aerobic biological treatment under "feast" conditions, the organisms may take up certain substrates at a rate that exceeds the metabolism rate, resulting in the transient storage and/or accumulation of substrate within the cells. The cells then metabolize this substrate during subsequent "famine" periods. The general theory for substrate storage is based on the idea that organisms possess a relatively large number of substrate uptake enzymes when growing under substrate-limited conditions, since the uptake is limited by the low substrate concentration (van Loosdrecht et al, 1997). When the organisms are then exposed to an environment with high substrate concentrations, the large number of enzymes allow very rapid uptake of the substrate, which exceeds the growth response. The storage response can be considered more rapid than growth response since synthesis of the storage polymers is less complicated than synthesis of the whole cell, requiring less physiological adaptation for storage (Majone et al, 1999). van Loosdrecht et al 20 (1997) suggested that "... if the substrate could not be converted to a polymer the whole cell metabolism would have the risk of getting out of balance." It should be noted that-internal, cellular "accumulation" of substrate differs from internal "storage". Accumulation has been defined as the situation where substrate removed from the wastewater is maintained in the cells "as such and/or transformed into low-molecular weight intermediates and precursors" (Dionisi et al, 2001). Alternately, storage can be defined where substrate removed from the wastewater is "transformed into specialized internal polymers (usually polysaccharides or lipids)" (Dionisi et al, 2001). Poly-B-hydroxybutyrate (PHB) is the commonly identified polymer since many substrates are metabolized within the cell to an acetyl-CoA (CoA = carrier molecule coenzyme A (Tortora et al, 1989)) intermediate compound, which is the precursor for PHB formation (van Loosdrecht et al, 1997). Majone et al (1999) noted that "for thermodynamic reasons (unfavourable gradients, osmotic pressure) this [accumulation] can be done efficiently to a much more limited extent than in the case of storage where molar concentration of substrate or intermediates is reduced by polymerization." Under the feast-famine regime, organisms that have the ability to balance their growth, through substrate storage and subsequent metabolism, have a competitive advantage over organisms that do not possess such ability (van Loosdrecht et al, 1997). This suggests that these organisms will dominate the biomass population, particularly in systems which experience large transient loading variations or substrate gradients (e.g. plug-flow bioreactors, sequencing batch reactors). Furthermore, although relatively little research has been conducted with respect to carbon substrate accumulation/storage under anoxic conditions (Dionisi et al, 2001), anoxic carbon storage can occur and it may have significant implications for nitrification and denitrification processes (van Loosdrecht et al, 1997). One specific implication is that readily degradable carbon, either contained in the wastewater or externally added to the system to enhance denitrification, may be transferred from the anoxic zone to the aerobic zone of the bioreactor as internally stored carbon instead of being used as an anoxic denitrification carbon source. The combination of these factors (i.e. selection of carbon-storing organisms and anoxic carbon storage) could impact heterotrophic behaviour in aerobic environments, potentially affecting autotrophic organisms and nitrification rates, especially in oxygen-limited environments. It should also be noted that autotrophic nitrifying organisms can also utilize organic carbon (Grady et al, 1999), although the assimilation of organic carbon, relative to carbon dioxide, in ammonia and nitrite oxidizing organisms is limited (Bock et al, 1991). These autotrophs have also been shown to store the 21 utilized carbon as PHB (Bock et al, 1991). Grady et al (1999) notes that the amount of carbon uptake is small and is generally ignored in nitrification stoichiometry. By extension, for biomass cultured on wastewater containing both organic carbon and ammonia, autotrophic carbon storage would likely be an insignificant fraction of the observed biomass carbon storage capacity, when compared to the heterotrophic carbon storage fraction. 2.3 N 2 0 PROPERTIES, SOURCES AND ENVIRONMENTAL FATE 2.3.1 Properties Nitrous oxide (N2O) is a colourless gas (boiling point = - 88.5°C; vapour pressure = 5,150 kPa @ 20°C; relative vapour density = 1.53 (air = 1.0)) that has been described as the least reactive of nitrogen oxides at normal temperatures (Mattson et al, 2004; IPCC, 2003). However, N 2 0 is a strong oxidant at temperatures above 300°C, and reacts violently with compounds such as aluminum, hydrazine and tungsten carbide (IPC S, 2003 ). Compared to oxygen, N2O is extremely soluble in water. Under conditions of one atmosphere of pressure (i.e. pure N2O headspace @ ambient pressure) and 0.0 % salinity, approximate equilibrium N 2 0 water concentrations are as follows.(Weiss and Price, 1980): " . • 1,800 mg/L @ 10°C • 1,200 mg/L @20°C • 900 mg/L @30°C 2.3.2 Sources In nature, N 2 0 is generated through biological nitrification and denitrification processes (Section 2.2.2) that occur in the ocean, atmosphere (ammonia oxidation) and in soils (IPCC, 2001). Besides these natural N2O sources, anthropogenic N2O sources include agricultural soils, biomass burning, industrial sources and cattle/feedlots. Current estimates of natural and anthropogenic global N2O generation rates are in the order of 10 Tg N/yr and 8 Tg N/yr, respectively (IPCC, 2001). USEPA (2003) provides a more comprehensive breakdown of the major anthropogenic N 2 0 sources, as shown below: 22 • agricultural soil management via indirect or direct application of soil additives (i.e. nitrogen fertilizer) • stationary and mobile combustion (i.e. power generation and vehicles) • adipic acid production, for use in nylon manufacturing • nitric acid production, for manufacture of synthetic fertilizers, adipic acid and explosives • N 2 0 product usage as an anesthetic • manure management • field burning of agricultural residues • domestic wastewater treatment facilities and effluent receiving environments • waste combustion 2.3.3 Environmental Fate The atmospheric stratosphere is ultimately the sink for generated N 20. Here, photodissociation (90%) and reaction with excited oxygen atoms (10%) transform N 2 0 to NO, with N 2 0 having an estimated atmospheric lifetime (i.e. average length of time an atom or molecule spends in the atmosphere) of 120 yr (IPCC, 2001). Current estimates suggest that about 12 Tg N/yr of these transformations occur in the stratosphere, leaving an imbalance of around 6 Tg N/yr, when considered with the previously discussed 18 Tg N/yr generated from atmospheric and terrestrial sources. The resulting product, NO, catalytically destroys stratospheric ozone (Wrage et al, 2001; Barton and Atwater, 2002). The N 2 0 remaining in the stratosphere efficiently absorbs infra-red radiation emitted by the Earth's surface and clouds (IPCC, 2001). The net effect of the absorption and subsequent re-emission of this radiation is partial trapping of the energy and; therefore, a tendency to warm the earth's surface. Thus, along with carbon dioxide, methane, ozone and water vapour, N 2 0 is one of the earth's primary "greenhouse gases" (IPCC, 2001). Compared to carbon dioxide, N 2 0 is very efficient at trapping infra-red radiation. Therefore, the IPCC (2001) has assigned N 2 0 a global warming potential of 310 units relative to carbon dioxide. 2.4 WASTEWATER TREATMENT The wastewater industry has demonstrated the ability of biological treatment systems to provide very high levels of nitrogen removal, using nitrification and denitrification processes, in a variety of treatment system configurations (e.g. Grady et al, 1999). This holds true for wastewaters ranging from relatively 23 weak domestic sewage with high carbon/nitrogen (C/N) ratios (Randall et al, 1991), to very high ammonia, low C/N wastewaters such as landfdl leachate (Shiskowski and Mavinic, 1998). This section examines several topics related to the current research project, including sequencing batch reactor systems, nitrification oxygen requirements and oxygen influences.on kinetics, N 2 0 generation and emission control, and treatment system process monitoring and control parameters. 2.4.1 SBR Systems Sequencing batch reactor (SBR) wastewater treatment systems have their origins in the earliest attempts to use "activated sludge" (i.e. a suspended-growth culture of microorganisms) to treat wastewater, as the first such application of microorganisms involved batch treatment (Grady et al, 1999). As the name suggests, SBR systems use a sequence of steps, or phases, to receive, treat and discharge an incoming batch of wastewater over the course of an operating cycle. The phases can generally be described as follows: wastewater fill, react, biomass settling, effluent decant, and system idle (Metcalf and Eddy, 1991). All phases occur as a function of time in a single reactor, analogous to what a continuous-flow treatment system provides in space. From their simple batch reactor origin, SBR systems have experienced an ebb and then surge in popularity of application in full-scale wastewater treatment facilities. Their initial decline in use has been suggested to be the result of the need for larger facilities as the industry progressed over time (Grady et al, 1999), implying that continuous-flow systems were more practical and cost-effective at large scale. The popularity of SBR systems surged in the 1970's with the recognition of the benefits provided by their operational flexibility, particularly in smaller scale applications (Grady et al, 1999). However, their increased popularity was also likely related to the advent of robust microprocessor-based software and hardware control systems, and aeration devices, needed to efficiently operate SBR systems (Metcalf and Eddy, 1991). SBR systems can provide significant nitrogen removal through provision of alternating anoxic (i.e. unaerated) and aerobic (i.e. aerated) phases within a single operating cycle. Facilities commonly use an anoxic fill phase, as well as a anoxic react phase prior to an aerobic phase, to take advantage of wastewater carbon for denitrification (Grady et al, 1999). In this manner, the nitrate generated during the aerobic-phase of the previous SBR cycle, and remaining in the SBR, is reduced to N 2 . Such SBR operation and performance over time is analogous to that provided in space by continuous-flow, pre-24 denitrification systems configured in the Modified Ludzack-Ettinger process (Grady et al, 1999). The length of the SBR fill-phase affects the instantaneous process loading factor (Grady et al, 1999). A short fill-phase provides for a more idealized batch reactor, analogous to a perfect plug-flow reactor (i.e. an infinite series of completely-mixed reactors) in continuous-flow systems. Alternately, a long Fill-phase results in SBR behaviour more similar to a single, completely-mixed reactor in a continuous-flow system. From a biological process perspective, the main advantage of an SBR, over a continuous-flow reactor system, is the potential for faster kinetic rates. This potential advantage is most fully realized when the SBR is operated with a short fill-phase. First consider continuous-flow systems. A plug-flow reactor provides a theoretically more efficient kinetic system for substrate utilization compared to a single complete-mix reactor, assuming reaction kinetics are a function of the in-situ concentration of wastewater constituents. However, longitudinal backmixing within the plug-flow reactor may reduce the actual substrate gradient compared to the idealized gradient, thus reducing system efficiency (Scuras et al, 2001). To combat this problem, a series of complete-mix reactors can be used to provide a plug-flow type of substrate gradient. Alternately, an SBR operated with a short fill-phase is essentially an "ideal" batch or plug-flow reactor. 2.4.2 Nitrification and Oxygen Section 1.2, as well as the equations shown in Section 2.2.1, alluded to the significant stoichiometric oxygen requirement for aerobic, autotrophic nitrification. Specifically, the oxygen requirements for ammonia oxidation and nitrite oxidation are approximately 3.2 mg 0 2 / mg NH 3-N and 1.1 mg 0 2 / mg N02"-N oxidized, respectively (Grady et al, 1999). The calculated values allow for the typical yield of ammonia (i.e. Nitrosomonas) and nitrite (i.e. Nitrobacter) oxidizing bacteria. The resulting oxygen demand exerted by nitrification in a typical biological nitrogen removal system treating municipal wastewater, with a carbon/nitrogen (C/N) ratio of around 4/1, may comprise 40% to 50%> of the total oxygen demand exerted by the biomass (Metcalf and Eddy, 1991). The remaining 50% to 60% of the total oxygen demand is assumed to be related to heterotrophic energy requirements for cell synthesis and maintenance. Therefore, for municipal wastewater treatment systems, nitrification-related energy requirements, for oxygen supply, can comprise 15% to 30% of the facilities total energy costs (WEF, 1997). For high-ammonia industrial wastewaters and landfill leachates with much lower C/N 25 ratios, the nitrification oxygen demand will make up an even larger fraction of the total process oxygen demand and facility energy requirements. Thus, as can be deduced, the costs for nitrification oxygen supply are not insignificant. Besides the stoichiometric oxygen requirement for complete oxidation of ammonia and nitrite, the availability of oxygen to nitrifying organisms has a significant impact on kinetic rates. Once transferred into the reactor liquid (i.e. mixed liquor), oxygen must diffuse from the bulk solution through the biological floe structure, to the vicinity of the cell, and into the cells to the enzyme sites (Mahendraker, 2003). The enzyme sites for nitrifying organisms are located inside the cytoplasmic membrane (Bock and Wagner, 2001). The diffusion rate across cell membranes depends on the concentration gradient across the membrane (Bitton, 1994). Therefore, as the bulk solution oxygen concentration decreases, the rate of oxygen diffusion into the cell and the vicinity of the enzyme binding sites slows down. At some bulk solution oxygen concentration the oxygen diffusion rate will become the limiting rate in the ammonia and nitrite oxidation processes, assuming non-rate limiting concentrations of substrates (i.e. ammonia, nitrite) and nutrients (e.g. phosphorus). Furthermore, Mahendraker (2003) noted that bulk solution dissolved oxygen (DO) levels below 0.5 mg/L may increase the relative significance of another limitation to oxygen mass transfer into the cell, namely oxygen diffusion across the liquid film surrounding the floe. Before examining the effect of oxygen (i.e. DO concentration) on nitrification kinetics, it is useful to describe and define the terms that are often used to express such effects. The effect of substrate (i.e. electron donor, electron acceptor, nutrients) concentration on reaction kinetics is often adequately described by the empirical Monod model, where various studies have confirmed its general application to both pure culture and mixed microbial populations (Grady et al, 1999). Equation 2-6 shows the Monod model: H = umax[S/(K i + S)] (2-6) where p m a x = maximum specific organism growth rate, S = substrate concentration, K s = the half-saturation coefficient and p. = actual specific organism growth rate. Multiplication of the calculated u value by the organism concentration, and subsequent division by the biomass yield (Y), provides the estimated substrate utilization rate (i.e. kinetic rate). The K s value provides an indication of the sensitivity of the reaction rate to substrate concentration, and is defined as the substrate concentration at 26 which p equals one-half U m a x (Grady et al, 1999). The review of aerobic, autotrophic nitrification kinetics by Stenstrom and Poduska (1980) found that reported K 0 values (i.e. half-saturation coefficient for oxygen) ranged from 0.1 to 2 mg/L. This wide range in values, over an order-of-magnitude, was attributed to several factors, recognizing that the reviewed studies included both pure culture experiments and those conducted with mixed microbial populations (e.g. wastewater treatment biomass). They postulated that one possible reason for the range in values was the effect of oxygen diffusion within the microbial floes. In other words, the bulk solution DO concentration may not be the same as the concentration in the immediate vicinity of the cells of interest, but instead lower. Therefore, the oxygen gradient across the cell membranes would be lower, slowing the rate of oxygen diffusion into the cell, and resulting in a reduced reaction rate for a given bulk solution DO concentration. Floe size and structure are important variables with respect to oxygen diffusion limitations within a floe (Mahendraker, 2003). Considering the most simple case of a pure culture, reactor conditions such as mixing (i.e. energy that might shear floes and reduce their size) and organism concentration will affect floe size and structure. Thus, it is reasonable to assume some variability in K 0 values even in the basic case of a pure culture, due to the conditions under which it was cultured. The situation becomes much more complicated in the case of a mixed microbial population, given the heterogeneous distribution of different groups of microorganisms within the floe, and the different respiration rates of these microorganisms (Mahendraker, 2003). Consider the case for a wastewater treatment bioreactor with a mixed heterotrophic and autotrophic biomass providing carbon and nitrogen oxidation, respectively. Grady et al (1999) reported typical K 0 values for heterotrophic and autotrophic organisms, functioning in municipal wastewater treatment systems, of 0.1 mg 0 2 /L and 0.75 mg O2/L, respectively. The differences in the K 0 values reflect the difference in the overall diffusional resistance in the transport of oxygen from the bulk solution, through the floe structure, and to the individual cells in the floe structure (Mahendraker, 2003). Furthermore, the influence of heterotrophic activity on the nitrifier K G value depends on the carbon loading to the system and resulting oxidation (Hanaki et al, 1990a, b). The preceding discussion highlights several points regarding the influence of oxygen (i.e. DO) on nitrification kinetic rates. First, the nitrifier K 0 value is much larger than the heterotroph K 0 value. 27 Therefore, at lower DO levels (e.g. < 5 mg/L), the nitrification rate will be much more sensitive to DO concentration than the heterotrophic carbon oxidation rate. Second, the nitrifier K 0 value will depend on the level of heterotrophic activity within the bioreactor, which in itself is a variable and a function of the carbon loading to the bioreactor. This situation makes it difficult to know the effect of a given DO concentration on the nitrification rate, given the uncertainty of the nitrifier K 0 value. Finally, the magnitude of the nitrifier K 0 value is relatively large, and in order to ensure maximum nitrification rates, engineering literature (e.g. Metcalf and Eddy, 1991) often suggests to operate bioreactors with a minimum DO concentration of 2 mg/L. The high energy costs associated with bioreactor oxygen supply is not only due to the large oxygen demand exerted by the biomass, but also from the fact that the relative insolubility of oxygen makes its transfer into water quite inefficient. Although much debate exists over the most appropriate way to mathematically model and estimate oxygen transfer into bioreactor mixed liquor (Mahendraker, 2003), the oxygen gradient, and thus the DO concentration, clearly has a significant impact on oxygen transfer efficiency (Section 1.2). Therefore, operating a bioreactor at the maximum nitrification rate, at times when that rate is not needed, is at odds with energy efficiency in the context of oxygen transfer. This implies a need for nitrification-related aeration system control. 2.4.3 N 2 0 Generation and Emission Control The recognition of the environmental fate and effects of N 2 0 has prompted researchers to study N 2 0 generation and emissions in wastewater treatment systems. Most studies were conducted within the past ten years. Besides trying to quantify the extent of N 2 0 generation in wastewater treatment systems, researchers have also, usually simultaneously, studied ways to reduce N 2 0 generation and thus emissions. Section 2.2.2 described wastewater system research that focused on mechanisms (i.e. carbon limitation, inhibition of N 2 0 reductase enzyme, selection of organisms with a N 2 0 denitrification endpoint) that could induce N 2 0 generation through accumulation of this intermediate during anoxic, heterotrophic denitrification. These studies specifically examined N 2 0 generation, usually in off-line batch experiments, that could potentially occur in the anoxic zone of a continuous-flow bioreactor or during the anoxic-phase in a sequencing batch treatment system. The conclusions from these studies either stated, or implied, various ways to minimize the potential for N 2 0 generation and resulting emissions in real wastewater treatment systems: • avoid reactor nitrite concentrations > 2 mg N/L (von Schulthess et al, 1994) 28 • ensure complete denitrification by providing a sufficient wastewater biologically available carbon to nitrogen ratio (Hanaki et al, 1992) • maintain a long SRT (Hanaki et al, 1992) • maintain somewhat alkaline (e.g. 7.5) reactor pH condition levels (Hanaki et al, 1992; Thorn and Sorensson, 1996) • maintain anoxic reactor DO concentrations < than 0.25 mg/L (von Schulthess et al, 1994) • avoid treating wastewater with high hydrogen sulphide concentrations (Schonharting et al, 1998) Some of the recommended methods to minimize N 2 0 generation during anoxic, heterotrophic denitrification are ultimately related to one another. For example, provision of complete anoxic denitrification will ensure low nitrite concentrations, thus reducing the potential for nitrite/NO-induced enzyme inhibition. However, complete denitrification can only be attained in the presence of sufficient available carbon, with biomass cultured under suitable growth conditions (e.g. sufficient SRT). Most of the aforementioned N 2 0 generation "control" methods were related to bioreactor operation. Provision of a suitable SRT is an operation issue. Similarly, maintaining low anoxic DO levels requires careful control of recycle streams and the mixed liquor DO concentration of the location from where those streams originate. Ensuring that sufficient carbon is available for denitrification could also be considered an operational issue in the case of carbon-limited wastewaters, particularly those of industrial origin, where exogenous carbon is added to the bioreactor to enhance denitrification. However, for carbon-limited domestic wastewaters, external carbon addition to support denitrification would be atypical, except in cases where the regulatory agency final effluent nitrogen criteria was very stringent. The potential problem of wastewater hydrogen sulphide could be addressed by an infrastructure, rather than an operational, solution. In this case, the reduction of wastewater hydrogen sulphide levels could be accomplished using a physical-chemical unit operation that preceded the biological treatment system. For example, hydrogen sulphide could be stripped from the wastewater in an aerated tank, with the off-gas directed to a hypochlorite wet scrubber foul air treatment system (Metcalf and Eddy, 1991). Spector (1998a) proposed an infrastructure method of reducing N 2 0 emissions, rather than using operational techniques to prevent N 2 0 generation. Based on his anoxic denitrification research, he found that provision of a sealed bioreactor would allow denitrification of headspace N 2 0 due to the "rapidity with which nitrous oxide reaches equilibrium between the gas and liquid phases". He suggested use of 29 gas- and liquid-staging in a sealed, plug-flow type reactor to minimize. N 2 0 emissions during denitrification. Barton and Atwater (2002) reported a U.S. patent obtained by the same researcher for this concept. • • . Other wastewater treatment researchers (e.g. Osada et al, 1998; Okayasu et al, 1997; Kimochi et al, 1998; Beline et al, 1999; Park et al, 2000; Itokawa et al, 2001; Park et al, 2001; Burgess et al, 2002; Beline and Martinez, 2002; Noda et al, 2003) have examined N 2 0 generation in the context of operating nitrogen removal treatment systems, either at bench- or full-scale, and investigated alternate ways of controlling N 2 0 emissions. The presence of anoxic and aerobic environments in these systems, and mixed populations of heterotrophic and autotrophic organisms providing nitrification and denitrification, significantly complicates identifying the main N 2 0 generation source or mechanism(s). Few studies provide any conclusive evidence as to the source of generated N 20. Furthermore, the wide variety of treatment systems, wastewaters and operating conditions make it difficult to directly compare study results. However, an overview of conducted research provides an indication of directed efforts. Several studies investigated the effect that bioreactor aeration strategies and/or DO concentration had on N 2 0 generation. For example, Kimochi et al (1998) investigated the effect of intermittent aeration (i.e. alternating aerobic and anoxic periods) on nitrification, denitrification and N 2 0 generation in a full-scale, continuous-flow, municipal wastewater treatment facility. They found that increasing the duration of the anoxic periods, while keeping the aerobic period duration constant, minimized N 2 0 emissions. However, nitrification performance suffered with increased anoxic period duration. Therefore, it appears that the reduced N 2 0 emission, with increasing anoxic period duration, was the combined result of less N 2 0 generation during the aerobic periods, with possibly enhanced N 2 0 reduction during the anoxic periods. Osada et al (1995) also found significantly reduced N 2 0 emissions in a bench-scale, swine waste treatment system that was operated using intermittent, rather than continuous, aeration. Again, the limited data indicate that the reduced nitrification rate under the intermittent aeration scheme was likely responsible for a sizable fraction of the reduced N 2 0 emission, due to less aerobic period generation. Experimental results obtained by Beline and Martinez (2002), from a bench-scale, intermittent aeration, swine wastewater treatment system, also found that increasing the anoxic period duration reduced N 2 0 generation. They suggested that differences in the rates of denitrification enzyme induction, after the switch from aerobic to anoxic periods, resulted in temporary accumulation of N 2 0 and other denitrification intermediates. Hence, the longer anoxic periods allowed denitrification of accumulated N 2 0 prior to the onset of aeration and stripping of N 2 0 from the mixed liquor. 30 Okayasu et al (1997) provides an example of a different approach to investigating the effects of aeration on N 2 0 generation and emissions. Under a constant aeration rate, operation of a bench-scale, anoxic-aerobic, mixed liquor circulating (MLC), "night soil" treatment system with aerobic DO levels > 1 mg/L minimized N 2 0 generation, with an emitted mass that was less than 5% of the removed nitrogen. Lower aerobic DO concentrations reduced nitrification efficiency, and also increased the mass of emitted N 2 0 relative to removed nitrogen, up to about 36% in some cases. The same study also examined anoxic-aerobic SBR operation under constant aeration rate, low aerobic DO levels (i.e. 0.3 mg/L). Ammonia oxidation was complete by the end of the aerobic phase, although some nitrite accumulation did occur late in the aerobic phase. Interestingly, nitrate was not observed during the aerobic phase, suggesting that denitrification was occurring at a high rate simultaneous with nitrification. The emitted N 2 0 mass averaged 32% of the removed nitrogen mass. Zheng et al (1994) conducted a chemostat study using biomass from a municipal wastewater treatment facility and synthetic wastewater. They studied the effect of bioreactor DO concentration on N 2 0 generation, and found that DO concentrations in excess of 1.7 mg/L resulted in about 2.4% of oxidized ammonia being converted to N 20. Alternately, DO concentrations between 0.1 and 0.5 mg/L caused between 5 and 7% of the oxidized ammonia to be converted to N 20. Other research has considered the role of external carbon addition in reducing anoxic and aerobic N 2 0 generation. Park et al (2000), operating a bench-scale, intermittent aeration system treating municipal wastewater, found that addition of methanol to the bioreactor reduced N 2 0 generation under both anoxic and aerobic conditions. In particular, aerobic N 2 0 generation was observed to decrease with reduced effluent nitrate concentration, induced by increased reactor methanol addition, supposedly due to increased heterotrophic denitrification. Without external carbon addition, 4.5% of the wastewater nitrogen was converted to N 2 0, compared to less than 0.2% when methanol was added to the reactor. The same study also found that the addition of cellulose media to the reactor, to provide biofilm formation, appeared to reduce N 2 0 generation when compared to the reactor without the media. The effect of the wastewater fill mode, in an bench-scale SBR system treating domestic wastewater, on N 2 0 generation and emission control was investigated by Park et al (2001). They found that a aerobic wastewater fill mode resulted in reduced N 2 0 generation when compared to an anoxic fill mode. The researchers theorized that the high ammonia concentrations that existed at the onset of the aerobic phase, due to the anoxic fill, accelerated N 2 0 generation when combined with the low DO condition and nitrite accumulation that simultaneously occurred at the start of the aerobic phase. Thus, they speculated that 31 aerobic denitrification, by nitrifying organisms, was the source of N 20. As such, the researchers suggested that N 2 0 generation could be minimized by using a aerobic fill mode, rather than an anoxic fill mode. The previously discussed study by Zheng et al (1994) also examined the influence of SRT on N 2 0 generation during nitrification. They found that N 2 0 generation remained constant (i.e. about 2.5% of oxidized nitrogen) for SRT's of 10 and 20 d, but increased as the SRT was reduced to 5 d (i.e. 7.5%) and then 3 d (i.e. 16%), even though the reactor DO concentration was at least 5 mg/L at all times. Nitrite accumulation was only observed during bioreactor operation at the 3 d SRT. Noda et al (2003) also found that lower SRT's reduced nitrification efficiency and increased aerobic N 2 0 generation in a bench-scale, continuous-flow, anoxic-aerobic system treating domestic wastewater. The aerobic reactor DO concentration was maintained at 2 mg/L. The preceding discussion has shown that, in some cases at least, biological wastewater treatment systems can generate a significant amount of N 2 0 relative to the amount of nitrogen contained in the wastewater and/or converted in the treatment system. These observations naturally lead to two questions: (i) how significant is N 2 0 generation in full-scale wastewater treatment, relative to the mass of nitrogen contained in the incoming wastewater and (ii) how significant are N 2 0 emissions from wastewater treatment facilities in the global context? A recent, comprehensive review of N 2 0 emissions from environmental management systems (i.e. wastewater treatment systems and facilities; solid waste landfills, composting and incineration facilities) by Barton and Atwater (2002) provides some answers to these questions. Regarding question (i), their review of five studies conducted at full-scale wastewater treatment facilities found that actual N 2 0 emissions were very low relative to the nitrogen conversions occurring with the treatment system. For example, Kimochi et al (1998) reported a maximum of 0.08%) of influent wastewater nitrogen was converted to N 20. However, Barton and Atwater (2002) observed an important point with respect to the reviewed full-scale studies: the treatment facilities in question provided either complete nitrification and denitrification, or essentially none at all. They argue that it is incomplete or partial nitrogen conversion that appears to lead to N 2 0 generation, thus the studies were biased as a result of the type of studied facilities. Given the uncertainties in N 2 0 generation, they note that the International Panel on Climate Change (IPCC) and the United States Environmental Protection Agency (USEPA) have adopted a N 2 0 emission coefficient for estimating the N 2 0 release from wastewater. The coefficient assumes 1% of the 32 nitrogen contained in wastewater will be converted to N 2 0 (i.e. N per N basis), either at the treatment system itself or in the environment that receives the effluent discharge (Barton and Atwater, 2002). In terms of question (ii), Barton and Atwater (2002) note that wastewater N 2 0 emissions, using the previously described emission coefficient and assuming a N 2 0 global warming potential-carbon dioxide (C02) equivalence factor of 310 (IPCC, 1995), have been estimated to be approximately 0.1% of the total C 0 2 equivalents generated in the United States. Excluding the C 0 2 equivalents generated by energy production, the wastewater N 2 0 emission fraction increases to about 0.9% of anthropogenic sources of greenhouse gases. 2.4.4 On-Line Process Monitoring and Control Parameters An Overview The application of process monitoring and control strategies to wastewater treatment systems extends back to at least the mid-1970's, with DO being the on-line monitoring parameter (Olsson et al, 1998). Critical evaluation (Islam et al, 1999) of the worth of such applications has shown "there are very real economic benefits to be gained by improving the level of process control", largely related to reduced facility operating costs and deferred capital investment for facility expansion. Yet Islam et al (1999) also found that such strategies were also "significantly under-utilized" in wastewater treatment systems. Part of the reason for the apparent under-utilization of process control strategies in wastewater treatment systems likely relates to what Thornberg et al (1993) called "unstable" (i.e. unreliable) on-line monitoring instrumentation. In this context; biological treatment system process control strategies often utilize information obtained from on-line monitoring of carbon (i.e. chemical oxygen demand) and nutrients (i.e. ammonia, nitrate, phosphorus) levels in influent wastewater, bioreactor mixed liquor and final effluent. The instruments that are typically used to conduct such parameter analyses integrate automated sampling and sample pre-treatment (e.g. filtration) systems with automated chemical analysis systems. The chemical analysis systems often utilize standard wet-chemistry (e.g. photometric) procedures adapted from laboratory methods. Although early versions of such instruments were evidently problematic, Londong and Wachtl (1996) found that extended use of modern systems produced reliable data, provided operations staff were trained in their use and maintenance. Other researchers concur with this assessment (e.g. Thornberg et al, 1993; Losson et al, 1998), with the suggestion that instrument system maintenance requirements were acceptable for treatment facilities sized down to a 33 capacity of about 15,000 person equivalents (Nielsen and Onnerth, 1996). • Evidently, however, not everyone in the wastewater treatment industry agrees with the positive assessment of these instrument systems. Several researchers (e.g. Fu and Pocb, 1995; Hack and Kohne, 1996; Lee and Park, 1999) have developed neural network treatment system models that relate parameters such as DO and pH (i.e. perceived to be reliably and accurately measured) to nutrient levels, due to the cited lack of on-line instrument reliability, as well as costs. Furthermore, Larsen et al (2000) noted the slow response time of most such instrument systems (i.e. 10 to 30 min) limits their application and usefulness in some treatment system configurations (e.g. intermittent aeration). Thus, instrumentation research continues to advance measurement technology. Biosensors are one example of such technology. Larsen et al (2000) developed a "rapid response" nitrate/nitrite biosensor with a 30 second response time. In their sensor, nitrate and nitrite diffuse through a membrane, into a very small chamber, housing denitrifying bacteria that are deficient in the N 2 0 reductase enzyme. The nitrate and nitrite are converted to N 20, which is measured using a built-in N 2 0 microsensor, evidently based on electrode potential. The biosensor unit itself is small, comparable to a DO probe, but the sensor tip is very small and in the order of 100 pm in diameter. The extremely small tip size has allowed researchers to use various types of microsensor to measure chemical gradients within wastewater biofilms and floes (Santegoeds et al, 1998). However, membrane detachment limited the operating-life of the Larsen et al (2000) biosensors to 5 days. Thus, continued development work would be required to produce a robust nitrate/nitrite microsensor for practical wastewater treatment applications. Similar biosensors have been developed for carbon measurement (Bourgeois et al, 2001), but short life spans and fouling still appear to be problematic in all types of biosensors. Optical sensors are also under development, with no requirement for reagents being one of their main advantages compared to wet-chemistry methodologies (Bourgeois et al, 2001). However, the technology is still young, and environmental factors that affect measurement, such as temperature, pH and metal concentration, require further research to allow development of practical instruments. While carbon, nitrogen and phosphorus are obvious candidates for on-line monitoring parameters in biological treatment systems, given these same parameters are ultimately used to define treatment system performance and effluent quality, researchers have also investigated what could be called "non-traditional" monitoring parameters. Some of these non-traditional parameters are essentially used as 34 surrogates for the conventional, chemical parameters. Other non-traditional parameters are used to quantify biomass behaviour or activity. A review of the literature reveals a wide variety of such parameters, examples of which are discussed below. Numerous researchers (e.g. Heduit and Thevenon, 1989; Charpentier et al, 1989; Wareham, 1992; Surmacz-Gorska et al, 1996; Zipper et al, 1998; Zhao et al, 1999; Ra et'al, 1999; Kim and Hao, 2001) have investigated the use of oxidation-reduction potential (ORP) as a process monitoring and control parameter for anoxic and aerobic bioreactor environments. Most studies used sequencing batch reactors, where changes in the concentration-time profiles (e.g. ammonia, nitrate, carbon) were tracked using ORP. Ideally, ORP-time profiles contain distinct changes in slope or inflection points that correspond to changes in redox couples (e.g. ammonia-nitrate) related to the disappearance of specific chemical species (i.e. ammonia, nitrate). In this context, ORP acts as an alternate, or surrogate, parameter for these chemical species. As an on-line monitoring parameter, ORP offers the benefit of being an inexpensive and simple technology, since it is simply a potentiometric measurement obtained with a probe, similar to pH. However, other parameters such as DO, pH and dissolved carbon dioxide can strongly influence ORP, affecting the ORP-time profile and thus complicating the interpretation of the information. Another example of the pursuit of simple probe technology for on-line application relates to monitoring anaerobic phosphorus release in BEPR systems. Potassium, magnesium and calcium are counter-ions that are co-transported across the inner bacterial membrane with phosphate molecules (Comeau et al, 1986). As a result, Maurer and Gujer (1995) found that measurement of ionic conductivity provided information on the start and completion of anaerobic phosphorus release in a batch reactor. However, they acknowledged that the method was limited to low ionic strength mixed liquors, and was also impacted by denitrification, if it occurred in the anaerobic zone, due to the consumption of nitrate and acetate ions, and the resulting production of bicarbonate ions, that impacted conductivity. Measurement of intracellular nicotinamide adenine dinucleotide (NADH) has been investigated as a means to identify changing conditions (e.g. carbon and ammonia levels) in a bioreactor, based on the physiological response of microorganisms to these changes (Farabegoli et al, 2003). The assumption of this approach is that the intracellular NADH/NAD + ratio reflects the balance between the electron donor and electron acceptor. Commercial NADH instruments use an automated sampling system to collect and expose a mixed liquor sample, contained in a small bioreactor, to UV light at a wavelength of 340 nm, and monitor the intensity of the resulting fluorescence at 460 nm. As a process monitoring parameter, 35 NADH provides information on microbial activity, yet it is similar to ORP in that both parameters are used to indirectly provide information about chemical compound concentrations as the measured quantity changes with respect to time. In a recent study, Farabegoli et al (2003) investigated the NADH responses of heterotrophs and autotrophs to the supply of electron donors (i.e. acetate, ammonia) and electron acceptors (i.e. oxygen, nitrate). Contrary to some earlier studies, they found minimal or no NADH response to the imposed biomass perturbations, and suggested that NADH may be of little practical value as a wastewater treatment process monitoring and control parameter. Titrimetric instruments are another type of instrument that provides information on biomass activity or behaviour, and have been used to monitor nitrification and denitrification kinetics (Gernaey et al, 1998). The basis for these sensors is that nitrification and denitrification reactions induce proton (i.e. FT) production or consumption, respectively. Automated mixed liquor sampling systems direct a sample to a small reactor, and utilize a pH controller to add either a base or acid to the sample while keeping the pH constant. The resulting titration curves are then used to estimate reaction kinetics and endpoints. Gernaey et al (1998) noted that commercial titrimetric systems are available, and could be used in full-scale treatment facilities. Other researchers have investigated the use of combined titrimetric-respirometric measurements (Gernaey et al, 2001) and titrimetric-off gas analysis (i.e. carbon dioxide) (Gapes et al, 2003) to enhance the value of collected information. The complicated, non-commercial, instrument systems used in these studies appear to be confined to laboratory research activities, rather than on-line process monitoring and control in full-scale systems. The use of respirometry, for estimating biomass oxygen uptake rate (OUR), has been the focus of much wastewater treatment research in recent years. The IAWQ Task Group on Respirometry (IAWQ, 1998) lists approximately 500 references in its summary report, with 77 references that were specifically related to respirometry-based process monitoring and control strategies. The Task Group authors divided the strategies into seven general approaches, representing the following manipulated variables: oxygen mass transfer (21), return activated sludge flow rate (18), waste activated sludge flow rate (13), influent flow rate (13), influent flow distribution (8), cycle length in periodic processes (3) and sludge storage flow rate (1) (where the values in brackets indicate the number of references that investigated the specified approach). Commercial respirometers are available for use in process monitoring and control applications, and typically integrate automated sample collection systems with the respirometer, which contains a small bioreactor and an oxygen measurement device (e.g. electrochemical DO sensor for liquid analysis, paramagnetic or gasometric method for gas analysis). One of the difficulties of using a 36 respirometer to measure OUR is that the data are actually generated off-line, in the units small bioreactor. For example, Jacquez et al (1989) demonstrated that reactor geometry, liquid sample volume, gas head space volume, and agitation velocity all impact the oxygen transfer rate (OTR) of the reactor. Therefore, accurate OUR measurements are not possible if the actual OUR exceeds the OTR of the respirometric reactor. Recognizing some of these difficulties, Hellinga et al (1996) noted that on-line measurement of bioreactor inlet and off-gas oxygen concentrations provides more direct and accurate measurement of oxidation activity than respirometry, typically conducted as an off-line measurement. Finally, some researchers (e.g. Li, 2001; Hellinga et al, 1996; Buitron et al, 1994; Nogita et al, 1982) have investigated the use of carbon dioxide as an on-line process monitoring and control parameter. For example, Nogita et al (1982) monitored bioreactor off-gas carbon dioxide concentrations for use in automated control of in-situ DO concentrations. The researchers cited the high maintenance requirements of a DO probe-based control system, compared to the low maintenance requirements of a carbon dioxide-based control system using an infra-red carbon dioxide analyzer, as the impetus for the research. Using a pilot-scale, plug-flow treatment system, under diurnal loading conditions, they found they could control DO concentrations within +/- 0.4 mg/L of a desired DO set-point, by regulating the blower air flow rate (i.e. manipulated variable) to maintain exhaust gas carbon dioxide concentrations at a specified setpoint. The carbon dioxide setpoints were based on relationships that were developed using data obtained from bench-scale experiments, correlating carbon dioxide production with organic load. With regards to carbon dioxide measurement, infra-red'analyzers are commonly used in the fermentation industry, and have been found to be robust, accurate and relatively inexpensive (Hellinga et al, 1996), thus making carbon dioxide an appealing parameter. However, the- influence of pH. on the carbonate-bicarbonate system equilibrium complicates the determination of biologically and non-biologically generated carbon dioxide (Sperandio and Paul, 1997). Furthermore, carbon dioxide would likely not be a useful nitrification monitoring parameter, although it could be used to detect significant loss of nitrification under high reactor ammonia loading conditions (Hellinga et al, 1996). In summary, a wide variety of parameters have been investigated as potential candidates for use in wastewater treatment system on-line process monitoring and control strategies. All have advantages and disadvantages with respect to the type or nature of provided information. Furthermore, instrument system complexity, reliability, maintenance requirements and costs are important considerations. 37 N20 as an On-Line Process Monitoring and Control Parameter As noted in Section 1.3, relatively little research has been conducted to date regarding the potential use of N 2 0 as either a process monitoring parameter or an aeration system control parameter in biological wastewater treatment systems. . The research conducted by Nogita et al (1981) was the earliest reference identified in the literature review, with respect to the application of bioreactor off-gas N 2 0 as an on-line process monitoring and control parameter for a wastewater treatment system. The research objective "... was to develop a new way of controlling an activated sludge plant by gas-phase instead of liquid-phase measurement." Two treatment systems were used in the study: (i) a bench-scale, continuous-flow system with a single complete-mix reactor, and (ii) a pilot-scale, continuous flow system with a four compartment bioreactor to provide plug-flow conditions. The systems were treating synthetic wastewater, with wastewater organic carbon load, ammonia load, and aeration system air flow rate being the experimental variables. Reactor off-gas N 2 0 levels were measured with an infra-red analyzer. They found that the N 2 0 generation rate was proportional to the nitrogen load when the carbon load was constant. However, they found that the N 2 0 generation rate was independent of the air flow rate, and although they presented no data, also concluded that reactor DO concentration had little impact on the N 2 0 generation rate. They also found that the N 2 0 generation rate increased with an increasing COD/nitrate ratio. The researchers concluded that heterotrophic denitrification, occurring in the oxygen-depressed inner core of the floe, was the N 2 0 generation mechanism. Of particular interest, they also suggested the possibility of an "unknown [microbial] species" that had the ability to generate N 2 0 from organic nitrogen and ammonia under aerobic conditions. Following up on their earlier work, the same group of researchers obtained a U.S. patent for a wastewater bioreactor aeration control system that utilized both off-gas N 2 0 and C0 2 as on-line monitoring parameters (Saito et al, 1982). As previously discussed, this same research group was investigating the use of C 0 2 as a process monitoring and control parameter (Nogita et al, 1982). It could be that their consideration of N 2 0 was the result of the infra-red measurement technology used for C0 2 monitoring —> both parameters can be measured using infra-red analyzers, but at different wavelengths (i.e. C0 2 = 2,380 cm"1, N 2 0 = 2,260 cm"1). Regardless, their patented system proposed to use both parameters to control bioreactor air flow rate and the return pumping rate of settled biomass from the clarifier to the bioreactor. Carbon dioxide was to be measured at the "upstream" end of the plug-flow reactor, with N 2 0 measured at the "downstream" end. Unlike Nogita et al (1981), they did present data that showed increasing off-gas 38 N 2 0 levels with decreasing DO concentration. Moving ahead two decades, Burgess et al (2002a) investigated the feasibility of using bioreactor off-gas N 2 0 as a process monitoring parameter for nitrification failure. Their hypothesis was based on evidence that AOB could generate N 2 0 in the presence of nitrite, and that nitrite accumulation in a nitrifying system was a potential indicator of system failure. Their experiments were conducted with a bench-scale system consisting of a completely-mixed reactor and a clarifier, operated with very long hydraulic retention (i.e. 3 to 5 d) and solids retention (i.e. 23 d) times. For smaller ammonia shock loads, the off-gas N 2 0 concentration and its rate of increase was found to be positively correlated with the magnitude of the ammonia shock load. Alternately, for larger ammonia shock loads, a positive correlation was found between the rate of increase in nitrite concentration and rate of increase in off-gas N 2 0 concentration. During the ammonia shock loads, the reactor DO levels decreased from typical concentrations of about 5 mg/L to around 1 mg/L. Under simulated aeration system failure, that allowed the DO concentration to rapidly decrease to less than 1 mg/L, a similar response between nitrite accumulation and increasing off-gas N 2 0 levels was observed. In what appeared to be a larger scale study, Burgess et al (2002b) monitored bioreactor off-gas N 2 0 levels in the context of detecting nitrification process failure induced by short-duration ammonia shock loadings, DO depletion and chemically-induced nitrification inhibition in a pilot-scale, continuous-flow treatment system utilizing a completely-mixed bioreactor with an anoxic zone and aerobic zone. The reactor off-gas N 2 0 levels sharply increased during the duration of the process "shocks", returned to steady-state levels after the shock was removed, with changes in aerobic zone ammonia, nitrite and nitrate levels only occurring after about 7 hours elapsed after the end of the shock. The total system hydraulic retention time was approximately 8 hours. In their other study, Burgess et al (2002a) hypothesized that aerobic zone "... N 2 0 and nitrite should precede the appearance of ammonia in the final effluent by approximately one hydraulic retention time of the entire aeration and clarification system", but offered no explicit explanation for their hypothesis. The researchers indicated they had applied for an international patent for a monitoring system for nitrification failure based on measurement of off-gas N 2 0 levels. 2.5 S U M M A R Y The practical, engineering objectives for nitrifying wastewater treatment systems are i) the provision of suitable ammonia and nitrite oxidation rates that will enable the required effluent quality criteria to be 39 met at all times and ii) the provision of the required kinetic rates in the most operating-cost efficient manner. Unfortunately, these two objectives are at odds with each other because of the opposite effect that the bioreactor mixed liquor DO concentration has on nitrification kinetic rates and oxygen transfer efficiency. Furthermore, many application- or site-specific variables influence the effect that "absolute" DO levels have on the nitrification rates. This situation implies a need for the ability to monitor the availability' of oxygen to nitrifying organisms and its effect on the kinetic rates, and then using this information in an aeration system control strategy. The wastewater treatment industry has recognized the value, or at least the potential value, of using on-line monitoring, in the context of providing real-time process control, to increase treatment system efficiency and reduce operating costs. To this end, researchers have investigated a wide variety of potential parameters for use in on-line process monitoring and control strategies. All parameters have advantages and disadvantages with respect to the type or nature of provided information. Furthermore, measurement instrument system complexity, reliability, maintenance requirements and costs are important considerations. The results of much research have extended the traditional wastewater treatment perspective of nitrification, to include several different biochemical pathways whose reactions are catalzyed by enzymes generated by a variety of microorganisms. The discovery of the oxygen-limited, nitrite reduction - N 2 0 generation pathway, for the believed to be dominant AOB found in wastewater treatment bioreactors, offers a potential breakthrough with respect to nitrification process monitoring. Specifically, unlike other investigated parameters, N 2 0 may offer the ability to directly monitor the AOB oxygen availability-limitation in nitrifying bioreactors, and its related impact on kinetic rates. Some research has been conducted in the broad area of using bioreactor off-gas N 2 0 as an on-line process monitoring parameter in the context of an aeration system control strategy. However, the research has been of limited scope and sophistication, and not extended to SBR treatment systems. Therefore, the current research project sought an in-depth understanding of aerobic-phase nitrification performance and N 2 0 generation in a SBR system. Key elements of the research included the investigation of SBR operating conditions that affected these phenomenon, examining how off-gas N 2 0 levels were related to N 2 0 generation via mass transfer from the bioreactor mixed liquor to the off-gas, and developing the application of off-gas N 2 0 data in the context of nitrification monitoring and aeration system control strategies. 40 C H A P T E R T H R E E M A T E R I A L S A N D M E T H O D S 3.1 EXPERIMENTAL APPARATUS AND OPERATION 3.1.1 SBR and Clean Water Reactor Systems SBR System A bench-scale, sequencing batch reactor (SBR) system was used for the biological wastewater treatment aspect of the research project (Figure 3-1). The SBR was constructed from acrylic plastic, with a 10.6 L operating liquid volume and 330 mm operating depth. SBR mixing was provided by a variable-speed electric motor, operated at 90 rpm, connected to shaft that contained two sets of impellers. The shaft and impellers were also constructed from acrylic plastic, as were the vertical baffles (four in total) that were fixed to the inside of the reactor. The reactor had a gas-tight headspace (volume = 3.3 L), with the off-gas directed to a beaker of water (50 mm water depth) to ensure a gas-tight seal at the reactor top plate. The reactor top plate included headspace and liquid sampling ports, as well as locations for mounting monitoring probes (i.e. DO, pH). Synthetic wastewater (refer to Section 3.1.3), contained in a room temperature 20 L pail, was supplied to the SBR using a variable-speed Cole-Parmer MasterFlex™ pump.: The wastewater entered the side of the reactor at the same elevation as the lower mixer impeller. Effluent, collected in a 20 L pail, was decanted from the reactor via a side-mounted, open ball valve that was connected to a solenoid valve. Air was supplied to the SBR from the main laboratory air supply system, via a pressure regulating valve (setpoint =15 psi); this was followed by a Cole-Parmer rotameter with precision needle valve, a solenoid valve, and entering the reactor through a porous stone, fine-bubble diffuser mounted in the reactor bottom plate. An in-line pressure gauge allowed confirmation of line pressures during air on and off periods. The wastewater pump, mixer motor, air supply and effluent solenoid valves were connected to a programmable, electronic, ChronTrol™ timer that provided on-off control of these elements. 41 Off-Gas Pre-Treatment System atmosphere IR N 2 0 unit condensor impmger with 5 N K O H pump reactor off-gas ,— (beaker of water) H-j DO and pH Data Logging System electronic timer syn hetic wastewater pail computer S B R soleniod valve pressure gauge rotameter/ flow meter effluent pail 4* pressure regulator laboratory air supply Air Supply System Figure 3.1: Schematic of S B R system 42 Clean Water Reactor System Clean water, gas mass transfer experiments were conducted in a second reactor, identical.to the reactor used in the SBR system. However, since the mass transfer experiments were done on an individual batch basis, water was manually added and removed from the "clean reactor". Gas-Mixing System A simple, manually operated, gas-mixing system (Figure 3-2) was assembled to allow introduction, and at times blending, of various gases (i.e. air, N 2 , N 20) to either the SBR or clean reactor systems. The gas-mixing system was not used for normal SBR operations. The gas-mixing system consisted of several rotameters (i.e. one for each gas), a 3-way valve that could direct gas flow to a bubble flow meter for flow measurements, and a gas-mixing chamber made from a short section (= 140 mm) of 12 mm diameter steel pipe. Pure N 2 , pure N 2 0, and 0.2% N 2 0 were supplied to the gas mixing system from compressed gas cylinders, with the pressure regulators set at approximately 15 psi. Gas flows to both reactor systems were set to the desired rates by adjusting the rotameters, based on the manufacturer's calibration tables. A bubble flow meter was also regularly used to manually check gas supply and reactor off-gas flow rates, and confirm rotameter calibrations. 3.1.2 SBR Start-Up and Operation The SBR was originally seeded and started on December 16, 2002, with biomass obtained from another, similar SBR system operating in the Environmental Engineering Laboratory. The SBR was shut down on March 4, 2003 due to poor biomass settling and excessive "autowasting" of biomass; it was re-seeded the same day using previously wasted biomass that was stored in a 4°C refrigerator. Not surprisingly, given the biomass history, poor biomass settlability returned, and the SBR was again shut down on March 26, 2003. The SBR was restarted on April 10, 2003, using seed biomass obtained from the Agassiz, B.C. municipal wastewater treatment facility. This facility utilizes an SBR system that operates under continuous influent feeding and intermittent effluent withdrawal (Louzeiro et al, 2003). SBR operation continued until July 29, 2003, after which it was taken off-line. The biomass was stored in the 4°C refrigerator, and used to restart the SBR on August 18, 2003. The SBR continued operation until the end of the experimental program, December 17, 2003. 43 to reactor air supply system gas-mixing chamber from laboratory air supply system Figure 3.2: Schematic of gas-mixing system 44 The SBR was operated with six, 240 minute (4 hr) cycles per day. The operating sequence, for one cycle, was as follows: • 4 minute (min) mixed, unaerated (i.e.anoxic) wastewater fill-phase (3.3 L wastewater volume) • 16 min, mixed, unaerated anoxic-phase • 160 min, mixed, aerated aerobic-phase • 60 min, unmixed, unaerated, biomass settle + effluent decant + idle phase The daily volume of treated wastewater was approximately 20 L (i.e. 3.3 L fill volume/cycle x 6 cycles/d). Under normal, "baseline" operating conditions, the aerobic-phase aeration rate (i.e. 1,000 to 1,200 mL/min) was set such that all oxidation reactions were complete by about elapsed time =150 min (i.e. 30 min remaining in the aerobic-phase). This operation resulted in average, aerobic-phase DO levels that varied between 0.2 and 0.5 mg/L. The SBR solids retention time (SRT) was maintained at 10 d throughout the study. Biomass was manually wasted from the SBR once per day, usually near the end of the aerobic-phase of a cycle. The waste volume was calculated based on the measured mass of solids contained in the SBR, and adjusted for the mass of solids lost in the effluent (i.e. captured in the effluent bucket) during the preceding 24-hr period. Given the day-to-day consistency in reactor and effluent solids levels, solids sampling and analysis were usually conducted 3 to 4 times per week. The 10 d SRT resulted in mixed liquor suspended solids (MLSS) levels that were in the range of 2,000 to 2,500 mg/L. The SBR system was operated in the main Environmental Engineering Laboratory at ambient room temperatures. The reactor mixed liquor temperature was typically around 23°C, varying over the long- • term by about +/- 2°C as the room temperature changed over the seasons. SBR system maintenance activities included frequent brush scrubbing of the inside of the reactor, cleaning of the DO and pH probes with distilled water and KimWipes™, replacement of the influent wastewater tubing, and daily cleaning (i.e. bleach or soap) and brush scrubbing of the influent wastewater feed pail. 45 3.1.3 Synthetic Wastewater Twenty litres of the "baseline" synthetic wastewater was made daily using Greater Vancouver Regional District potable water, with the following key items of composition: acetate (374 mg/L), yeast extract (100 mg/L), ammonium chloride (90 mg N/L), and sodium bicarbonate (300 mg/L as CaC03). The wastewater total organic carbon (TOC) concentration was approximately 200 mg C/L; the chemical oxygen demand (COD) concentration was calculated to be about 520 mg/L, based on acetate and cell COD equivalents from Grady et al (1999). The wastewater organic N concentration was approximately 12 mg N/L. Phosphorus, sulfur and trace elements were added in quantities needed to support biological growth, based on Grady et al (1999). A slight change was made to the wastewater composition on March 17, 2003. Subsequent to this date, sodium citrate was added to the wastewater, as a chelating agent, to minimize precipitation of some metals (e.g. iron) in the feed pail. The synthetic wastewater was made daily from several stock solutions that were previously prepared and stored at 4°C. The potable water was allowed to come to room temperature prior to making the synthetic wastewater. Appendix A contains detailed information on the composition of the various stock solutions, as well as the prepared "baseline" synthetic wastewater. 3.2 EXPERIMENTAL PROGRAM AND DESIGN 3.2.1 SBR Experiments The SBR system was operated throughout the study under was has been termed "baseline" conditions, essentially steady-state with respect to wastewater composition, aeration rate, SRT, etc. as described in Section 3.1. During the study, the SBR was utilized to conduct a series of "batch" experiments to investigate specific topics related to the research objectives. Most of the experiments involved subjecting the SBR, for one cycle, to an operational perturbation (e.g. change in aeration rate, wastewater ammonia concentration) that was atypical of normal, baseline operating conditions. The utilized approach, for all such experiments, involved two back-to-back SBR cycles. The "baseline cycle" was used as the experimental control, with the following cycle subjected to the specific perturbation, and thus termed the "perturbation cycle". Both cycles were subjected to the same level of monitoring and data collection. Samples for most parameters were collected every 15 min, with on-line parameters sampled every 1 minute. 46 Several actions were taken to ensure operational consistency between the back-to-back SBR cycles. First, the synthetic wastewater was freshly prepared before the experiment, and where the experimental design allowed, the same batch of wastewater was used for both cycles. Second, the biological solids generated during the baseline cycle were wasted from the SBR prior to the start of the perturbation cycle. The waste volume was calculated as per the previously described method, accounting for SBR operation that utilized six cycles per day. Reactor and effluent solids samples were collected and analyzed prior to the start of the baseline cycle. These experiments were also usually conducted in groups of two, on adjacent days. Thus, the baseline cycle data from two adjacent days could be compared to one another and used to confirm the consistency-repeatability of SBR performance under the baseline (i.e. control) conditions. For a limited number of experiments (i.e. N 2 0 reduction experiments), the SBR was taken off-line from its usual operating sequence and used a batch reactor. Although these experiments did not utilize the baseline cycle-perturbation cycle approach, control data were collected from the system prior to conducting the desired experiment. These experiments were also conducted as a group of experiments, on adjacent days, allowing the comparison of control data. For ease of reference, the specific purpose and design of each experiment is contained in the appropriate sub-section of Section 4, immediately preceding the related presentation of results and discussion. Figure 3-3 illustrates the categories of SBR experiments and their variables, with general information as follows: • The N20 source experiment was conducted to confirm the source of generated aerobic-phase N 20; specifically that AOB were generating the majority of N 2 0 rather than heterotrophic organisms. • The wastewater component experiments investigated the impact that various wastewater components had on nitrification performance and the resulting N 2 0 generation response, and how off-gas N 2 0 data could be related to nitrification kinetic rates. Frequent changes in wastewater composition are a reality in full-scale treatment system operation, they impact nitrification kinetics, and are the main reason for the need for process monitoring and control systems. • The aeration rate experiments investigated the impact that aeration (i.e. oxygen supply) rate had on 47 c SBR Experiments N 20 Source Experiment Wastewater Component Experiments Aeration Rate Experiments Ammonia Oxidation Nitrite-Nitrous Acid Experiments Nitrite Level 1 pH level Ammonia Load Readily Degradable Carbon Load Slowly Degradable Carbon Utilization Rate Air Flow Rate Slowly Degradable Carbon Utilization Rate I i N 20 Reduction Experiments DO Level 1 Slowly Degradable Carbon Utilization Rate ure 3.3: Experimental program summary N 20 Stripping Experiments (Clean Reactor) Reactor Gas Mass Transfer Experiments^^ 0 2 Transfer Experiments (Clean Reactor) N 20 Liquid-Headspace Partitioning Experiments (SBR) Air Flow Rate Air Flow Rate Air Flow Rate N 20 Generation Rate r r Gas Transfer Liquid Surface Diffusion Figure 3.3: Experimental program summary (con't) nitrification performance and the resulting N 2 0 generation response. In full-scale treatment systems, aeration rate is the main operational tool, or manipulated variable, for responding to changes in nitrification rates induced by changes in wastewater composition. • The N20 reduction experiments were conducted to investigate the significance and extent of heterotrophic biochemical reduction (i.e. denitrification) of AOB-generated N 20, and how this mechanism affected the measurable N 2 0 generation response and the interpretation of off-gas N 2 0 data. • The nitrite-nitrous acid experiments were conducted to investigate the sensitivity of N 2 0 generation on nitrite and nitrous acid concentrations in the bioreactor, and how these parameters may impact interpretation of off-gas N 2 0 data. 3.2.2 Reactor Gas Mass Transfer Experiments Several types of reactor gas mass transfer experiments were conducted using the "clean reactor" system. These batch experiments investigated the stripping of N 2 0 dissolved in the reactor liquid, as well as the oxygen transfer characteristics of the reactor. All such experiments were conducted at ambient room temperature, with estimated mass transfer coefficients corrected to a temperature of 20 °C using the method described in ASCE (1992). All experiments were done using the baseline synthetic wastewater as the test "clean" water. Similar to the SBR, the clean reactor was operated with a 90 rpm mixing speed during the mass transfer experiments. Other N 2 0 mass transfer experiments were conducted in the SBR to elucidate in-process mass transfer characteristics under different aeration rates and N 2 0 generation rates. The specific purpose and design of each experiment is contained in the appropriate sub-section of Section 4, immediately preceding the related presentation of results and discussion. Figure 3-3 illustrates the categories of reactor gas mass transfer experiments and their variables, with general information as follows: • The N20 stripping experiments were conducted to determine the primary N 2 0 liquid stripping mechanism from the experimental reactors. This information was important in the context of concept scale-up, when measuring off-gas N 2 0 levels from full-scale bioreactors that have different gas mass transfer characteristics than those of bench-scale reactors. 50 • The 02 transfer experiments provided data that assisted in planning the SBR aeration rate perturbation experiments. • The N20 liquid-headspace partitioning experiments were conducted to assist in the understanding of the dynamics of N 2 0 evolution from the bioreactor mixed liquor into the headspace and off-gas, as well as relating the N 2 0 generation rate to the N 2 0 stripping rate. Both of these issues are important when relating changes in nitrification rate to changes in measured off-gas N 2 0 levels. 3.3 ANALYTICAL METHODS AND ON-LINE MONITORING 3.3.1 Sample Collection, Pretreatment and Preservation SBR mixed liquor samples, for dissolved parameter analysis (i.e. ammonia, nitrite/nitrate, ortho-phosphate, TKN, TOC), were collected using a 60 mL plastic syringe attached to a small-diameter (inside diameter = 2 mm) Tygon™ tube that extended into the reactor through the reactor top plate. The collected samples were immediately placed into plastic centrifuge tubes and spun for approximately 4 min in a centrifuge. The tube supernatant was then decanted into a 60 mL plastic syringe, that was attached to a plastic 47 mm diameter filter holder, and filtered through a 1.5 pm VWR™ 691 Glass Microfiber Filter (Whatman™ 934-AFf equivalent). Sample preservation and storage was as follows: • Ammonia samples were placed into Lachat™ tubes, acidified to pH = 3 with 5% H2S04 and stored at 4°C prior to analysis. • Nitrite/nitrate and ortho-phosphate samples were placed into Lachat™ tubes, preserved with 1 drop of phenyl mercuric acetate (0.1 g phenyl mercuric acetate in 20 mL acetone and 80 mL distilled water) and stored at 4°C prior to analysis. • Dissolved total Kjeldahl nitrogen (TKN) samples were placed into BACH™ glass COD vials, acidified to pH = 2 with 5% H2S04 and stored at 4°C prior to analysis. • Dissolved total organic carbon (TOC) samples were placed into COD vials, acidified to pH = 3 with 50% H 3 P O 4 and stored at 4°C prior to analysis. Dissolved hydroxylamine (NH2OH) samples were preserved and stored as described by Simm (2004). 51 Synthetic wastewater and effluent samples, for dissolved parameter analysis, were subjected to the same pretreatment and preservation methods as the mixed liquor samples, except they were not centrifuged prior to fdtration. Mixed liquor samples, for biomass TKN analysis, were collected from the reactor using a wide-mouth 25 mL pipette. The samples were preserved and stored in the same manner as the dissolved TKN samples. Mixed liquor samples, for biomass poly-B-hydroxybutyrate (PHB) analysis, were collected from the reactor using a wide-mouth 25 mL pipette and placed into COD vials. The vials were immediately spun for approximately 4 min in a centrifuge. The vial supernatant was then decanted, and the vials frozen prior to freeze drying the samples. Mixed liquor samples, for mixed liquor suspended solids (MLSS) and mixed liquor volatile suspended solids analysis (MLVSS), were collected from the reactor using a wide-mouth 25 mL pipette. Effluent samples were collected from the effluent pail for solids analysis. All solids analyses were conducted immediately after sample collection. Mixed liquor N 2 0 concentrations were measured using a modified version of the procedure described by Kimochi et al (1998). Twenty mL of mixed liquor was collected in a 60 mL plastic syringe containing 30 mL of air, capped with a rubber septum, followed by addition of 1 mL of 2 N H 2 S0 4 to the syringe to eliminate biological activity. The syringe was vigorously hand shaken for 1 minute to partition the N 2 0 between the mixed liquor and syringe headspace. After at least a five minute waiting period, a 100 uL headspace sample was collected in a gas-tight, 100 uL Hamilton™ syringe for immediate gas chromatograph (GC) injection. Reactor headspace N 2 0 samples were manually collected in a gas-tight, 100 uL Hamilton™ syringe for immediate gas chromatograph (GC) injection. Headspace NO and C0 2 samples were manually collected in 1 mL Hamilton™ syringes for immediate NO analyzer and gas partitioner injection, respectively. All headspace samples were collected from a gas-tight sampling septum located in the reactor top plate. 3.3.2 Analytical Methods A Lachat™ QuikChem Automated Ion Analyzer was used to analyze ammonia (Method 10-107-06-1-Z) 52 and nitrite/nitrate (Method 10-107-04- 1-Z) samples, in accordance with Methods 4500-NH3 H and 4500-N0 31 of APHA (1998), respectively, and ortho-phosphate (Method 10-115-01-1-7) samples. TKN samples were digested and analyzed in general accordance with APHA (1998) Method 4500-N Org D. The samples were digested with 5 mL of digestion reagent (200 mL H 2 S0 4 and 134 g K 2 S0 4 made up to 1 L with distilled water) for 3.5 hr at 140 °C and 3.5 hr at 365 °C. Following digestion, 30 mL of distilled water was added to the digestion vials and mixed with the samples, with the samples transferred into Lachat™ tubes and stored in a 4°C refrigerator prior to analysis. The prepared samples were analyzed using the Lachat™ analyzer (Method 10-115-01-1-Z). TOC samples were analyzed using a Shimadzu™ TOC-500 analyzer in accordance with the Combustion-Infrared Method 5310B of APHA et al (1998). Purified air, at a flow rate of 150 mL/min, was used as the carrier gas. Samples were automatically pre-sparged by the analyzer, to remove dissolved carbon dioxide, prior to analysis. The freeze dried PHB samples were prepared and analyzed in accordance with the methods of Comeau et al (1988) and Braunegg et al (1978). To eliminate potential leakage of volatile substances from the COD vials during incubation, due to differential expansion between the plastic caps and glass vials, the vials were incubated in a HACH™ COD digestion block at the required 100 °C temperature. The derivatized and extracted samples, containing a benzoic acid internal standard, were analyzed using a Hewlett-Packard™ 5890 gas chromatograph (GC) with a Megabore™ DB-Wax capillary column and a flame ionization detector (FID). The injection port and FID temperatures were set at 120 °C and 260 °C, respectively. The column temperature profile was as follows: 1 min @ 45 °C, a 5 °C/min ramp up to 100 °C, followed by 1 min @ 100 °C. The gas flow rates were 20 mL/min for helium, 30 mL/min for hydrogen and 400 mL/min for air. Reactor and effluent solids analysis were conducted in accordance with Methods 2540 D and 2540 E of APHA et al (1998). Samples were filtered through a 1.5 pm VWR™ 691 Glass Microfiber Filter (Whatman™ 934-AH equivalent). Mixed liquor, dissolved hydroxylamine (NH2OH) samples were analyzed in accordance with the method described by Simm et al (2004b). 53 N 2 0 samples, obtained from the reactor headspace or mixed liquor samples, were manually injected into a Hewlett-Packard™ 5880 gas chromatograph (GC) for analysis. The GC used nitrogen as the carrier gas (flow rate = 25 mL/min) and a Hayesep™ DB packed column, with injector, column and electron capture detector temperatures of 150 °C, 100 °C and 250 °C, respectively. Mixed liquor N 2 0 concentrations were estimated from the obtained data using the Bunsen Adsorption Coefficient, based on the method described by Tiedje (1982) and discussed in Section 3.4.2. Reactor headspace NO samples were injected into a Sievers™ 280i Chemiluminescent Analyzer for NO analysis. The analytical method is based on a gas-phase chemiluminescent reaction between ozone and NO, with a photomultiplier tube used to detect the photon emissions from the generated, and activated, nitrogen dioxide. The analysis was conducted in accordance with the manufacturers high-sensitivity method. Reactor headspace C 0 2 samples were injected into a Fisher/Hamilton™ Model 29 gas partitioner GC, connected to a Spectraphysics™ SP 4290 integrator, for analysis in accordance with the Gas Chromatographic Method 2720C of APHA et al (1998). Helium was the carrier gas, with a flow rate of 46 mL/min. Instrument calibration was performed using a 0.93% volume/volume C0 2 standard prepared from pure C0 2 , obtained from a compressed gas cylinder, and ambient air. 3.3.3 On-Line Reactor Monitoring DOandpH Dissolved oxygen (DO) and pH measurements were obtained using in-situ reactor probes (DO = Oxyguard™, pH = VWR™ SympHony). The probe-generated electrical signals (i.e. mV) were directed to a personal computer (PC) equipped with a 128-bit, analogue-to-digital conversion card (PLC 812). Labtech™ Notebook/XE data acquisition software was used to process the electrical signals, convert the data to the proper measurement units, and display and log the data. The SBR DO and pH data were sampled at 5 s intervals and converted to a 1 min moving mean, with the mean data being logged. During the oxygen transfer tests, the reactor DO data were sampled and logged using a 5 s interval. Off-Gas N20 Besides the manual collection of discreet reactor headspace samples for N 2 0 determination by GC, the 54 reactor off-gas was also continuously monitored forN 20 using a Bacharach™ Model 3010 N 2 0 analyzer. The hand-held sized unit contained a sampling pump, dual wavelength infra-red (IR) sensor, LCD display and internal data logger. Adjustments to instrument configuration and data retrieval were accomplished using the supplied BACH-COM software, installed on a PC, with the PC connected to the Bacharach™ unit via an optical data link. Several technical issues precluded the direct use of the Bacharach unit for this application. First, the Bacharach™ literature noted that-C02 imparts a 5 to 10 ppm N 2 0 interference per 1,000 ppm C0 2 . Second, the measured gas could not contain condensable moisture. Third, the internal sampling pump was not designed for extended, continuous operation. As a result, an off-gas "pre-treatment" system (Figure 3-1) was used to overcome these issues. The pre-treatment system consisted of the following elements: • A positive-displacement, MasterFlex™ pump was used to provide bioreactor off-gas to the pre-treatment system at the flow rate (= 100 to 120 mL/min) required for the Bacharach™ unit. The manufacturer confirmed this mode of operation would not harm the internal pump. The MasterFlex™ pump speed was set using a bubble flowmeter for flow calibration. • The off-gas was bubbled through 25 mL of a 5 N KOH solution, contained in a 30 mL impinger flask (total flask volume = 50 mL), to remove C0 2 from the gas. Strong basic solutions are often used to scrub C0 2 from ambient air for fermenter monitoring ( Sperandia and Paul, 1997). The post-KOH off-gas contained essentially no C0 2 , since collected samples analyzed in the gas partitioner showed C 0 2 levels that were less than ambient air concentrations (i.e. < 300 ppm). The KOH solution was changed after every second experiment, with C0 2 monitoring data confirming the very conservative nature of this change-out frequency. • The off-gas was next directed to an ice-chilled, impinger flask (total volume = 60 mL) to condense water vapour from the gas. • Finally, the off-gas entered the Bacharach™ unit for N 2 0 measurement. The data logger was configured to record data at 1 min intervals. The "average" storage mode was used, 55 which directed the unit to record the average of all N 2 0 concentrations measured during the 1 min sampling interval. The Bacharach™ Model 3010 specifications indicate a measurement range of 0 to 1,000 ppm N 2 0, with an absolute accuracy of approximately +/- 10%. Headspace N 2 0 data (i.e. manual sampling @ 15 min intervals, GC analysis) were collected for each experiment along with off-gas N 2 0 data (i.e. on-line sampling @ 1 min intervals, IR analysis via Bacharach™ unit). As will be shown in Section 4.2, there was generally very good agreement in absolute N 2 0 values provided by the two sampling and measurement methods. There were a few occasions when the Bacharach unit experienced some baseline drift, resulting in either an overestimation or underestimation of N 2 0 levels, when compared to the GC data. In these situations, only the GC data were used in the experiment analysis. 3.4 CALCULATION METHODS 3.4.1 Gas Concentration Unit Conversion The reactor headspace and off-gas N 2 0 data, as well as plastic syringe headspace data for mixed liquor N 2 0 samples, were converted to units of nitrogen mass per unit volume (i.e. g N/m3), as opposed to the volume per volume unit (i.e. ppm = (v/v) x 1,000,000) provided by the GC and IR measurement methods, to allow calculation of N 2 0 mass generation and stripping rates, and for use in estimating the total mass of N 2 0 generated during the aerobic-phase of an SBR cycle. Equation 3-1 shows the conversion equation: N 2 0 (g N 2 0 /m3) = [[ N 2 0 (ppm) x P x MW] / [UGC x T]] / 1,000 (3-1) where: P = reactor headspace/off-gas pressure = 1 atm MW = molecular weight of N 2 0 = 44.02 g / mole UGC = universal gas constant = 0.082 atm' L / mole K T = reactor headspace/off-gas temperature (K) The N 2 0 concentration, in terms of (g N/m3), can be obtained by multiplying the calculated (g N20/m3) value by the N to N 2 0 mass ratio (i.e. [2 x 14.01 g/mole] / [44.02 g/mole] = 0.637). The same equation was used to convert C0 2 data, using the appropriate MW of C0 2 (44 g/mole). 56 3.4.2 Liquid N 2 0 Concentration Calculation As described in Section 3.3.1, mixed liquor N 2 0 samples were prepared in a 60 mL plastic syringe. The N 2 0 contained in the original mixed liquor sample partitions between the mixed liquor contained in the plastic syringe and the plastic syringe headspace. The method of Tiedje (1982) assumes that the partitioning can be described by the Bunsen Adsorption Coefficient, defined as the volume of gas (mL) adsorbed by 1 mL of water at a given temperature, under atmospheric pressure. Equation 3-2 shows the equation for determining the mass of N 2 0 contained in the plastic syringe, based on GC analysis of the plastic syringe headspace gas: M = C g [V g + (V,xa)] (3-2) where: M = total mass of N 2 0 contained in the plastic syringe (ug N 20) C g = plastic syringe headspace N 2 0 concentration (pg N 2 0 /m3) V g = plastic syringe headspace volume (m3) V| = mixed liquor sample volume in plastic syringe (m3) a = Bunsen Adsorption Coefficient @ T (°C) The estimated mixed liquor N 2 0 concentration is obtained by dividing the calculated M value by the mixed liquor sample volume (Vi). 57 CHAPTER FOUR RESULTS AND DISCUSSION 4.1 INTRODUCTION Chapter Four presents data collected during the research program and provides discussions related to the obtained results. Chapter 4 contains the following sections: *•< • Section 4.2 • Section 4.3 • Section 4.4 • Section 4.5 Typical SBR Time-Variant Parameter Profiles This section identifies and describes the typical operating characteristics of the SBR, when treating the baseline wastewater to achieve complete nitrification by cycle end, as manifested in the various parameter concentration-time profiles. This information familiarizes the reader with general SBR operation and response prior to presenting the various experiments. Reactor Gas Mass Transfer Experiments Section 4.3 discusses the N 2 0 and oxygen mass transfer characteristics of the reactors used in this research. These data provide insight into the reactors dynamic N 2 0 mass transfer response when subjected to physical and biological-physical processes. N 2 0 Source Experiment Section 4.4 presents the investigation to identify the aerobic-phase N 2 0 source for the SBR biomass and operating conditions used in the research. Wastewater Component Experiments This section examines the effect that various wastewater components had on nitrification performance and the resulting N 2 0 generation response when the SBR was operated with the same aeration rate for both the baseline and perturbation cycles. The examined wastewater components included ammonia load, readily degradable carbon load, and slowly degradable carbon, the latter expressed in terms of utilization rate. • Section 4.6 N 2 0 Reduction Experiments This section presents the experimental findings related to aerobic-phase N 2 0 biochemical reduction (i.e. denitrification) induced by the availability of slowly degradable carbon under oxygen-limited conditions, and how this mechanism affects the measurable N 2 0 generation response. Variable parameters included DO concentration and slowly degradable carbon utilization rate. 58 • Section 4.7 Aeration Rate Experiments Section 4.8 • Section 4.9 • Section 4.10 • Section 4.11 Section 4.7 investigates the effect that aeration rate had on nitrification performance and the resulting N 2 0 generation response when the SBR was treating the same wastewater for both the baseline and perturbation cycles. The wastewater components were also a variable parameter in related experiments. Nitrite-Nitrous Acid Experiments Section 4.8 examines the sensitivity of N 2 0 generation to changes in available nitrite-nitrous acid between the baseline and perturbation cycle. Nitrite-nitrous acid levels were manipulated through nitrite spikes and pH adjustment. Long-Term SBR Operation: General Observations Sections 4.2 to 4.8 present results and discussion for specific sets of experiments. Alternately, Section 4.9 discusses a range of findings obtained from long-term SBR operation under the baseline conditions. Summary of Major Findings Section 4.10 provides a summary of major research findings, the inter-related nature of the findings, and their significance to the main research objective. Development of a Conceptual N20-Based SBR Aeration System Control Strategy In keeping with the main objective of this research project, the experimental results presented in the preceding sections provide insight into the type and utility of information provided by off-gas N 2 0 data, as related to monitoring SBR nitrification kinetics. This section describes the basic concept, identifies key state variables, presents the conceptual control strategy, proposes an implementation framework and discusses the potential constraints of a N 2 0-based aeration system control strategy. Sections 4.3 to 4.8 and their sub-sections, which describe specific experiments, are generally structured under the headings of experimental design, results, and discussion. The nomenclature for the main types of experiments is as follows: • NOSE • NODSE • NOME • SBR-TS • MISC nitrous oxide stripping experiment nitrous oxide diffusion stripping experiment nitrous oxide mechanism experiment SBR tracking study miscellaneous experiment 59 4.2 T Y P I C A L S B R T I M E - V A R I A N T P A R A M E T E R P R O F I L E S This section identifies and describes the typical operating characteristics of the SBR, when treating the baseline wastewater to achieve complete nitrification by cycle end, as manifested in the various parameter concentration-time profiles. This information familiarizes the reader with general SBR operation and response prior to presentingThe various experiments. * - ' As noted in the following discussion, the presented data were obtained from several "baseline" cycles since experimental logistics prevented sampling for all parameters during each experiment. SBR Cycle Sequence Prior to presenting the data it is worthwhile to reiterate the 240 minute SBR cycle sequence. The SBR was at the end of the previous cycle's idle-phase immediately prior to elapsed time = 0 min. The wastewater feed pump starts at elapsed time = 0 min, with the unaerated (i.e. anoxic) fill-phase ending at elapsed time = 4 min. The anoxic-phase ends at elapsed time = 20 min with the start of aeration, coinciding with the start of the aerobic-phase of the cycle. The aerobic-phase continued from elapsed time = 20 min to elapsed time = 180 min. Biomass settling, effluent decant and an idle period comprised the remaining sixty minutes of the cycle (i.e. elapsed time = 180 to 240 min). DO, pH, Ammonia, Nitrite and Nitrate Profiles Figure 4-1 illustrates the dissolved oxygen (DO), pH, ammonia, nitrite and nitrate concentration-time profiles obtained during SBR-TS 16a. Initially, the DO probe was hanging in air within the reactor; therefore, the recorded DO levels were at air saturation until the mixed liquor contacted the probe. The DO levels rapidly decreased to below 0.1 mg/L for the remainder of the anoxic-phase until the start of the aerobic-phase. The very low, but measurable, anoxic-phase DO levels were likely attributable to a slight inaccuracy in the probe calibration offset, combined with some mixing-induced air entrainment into the mixed liquor. In contrast to DO, pH levels quickly increased, once the probe contacted the mixed liquor, in response to the rapid reduction of nitrate and nitrite (i.e. denitrification) and resulting alkalinity generation. Data for samples collected at elapsed time = 19 min indicate that complete denitrification was achieved by the end of the anoxic-phase. However, the sharp leveling of the pH profile at approximately elapsed time = 10 min suggests an earlier denitrification endpoint. 60 Figure 4.1: SBR DO, pH, ammonia, nitrite and nitrate concentration-time profiles (SBR-TS 16a) 61 After the start of aeration at elapsed time = 20 min, DO levels remained low (i.e. essentially zero) for another ten min. Then, at elapsed time = 30 min, the DO concentration jumped to approximately 0.3 mg/L and slowly climbed to a maximum value of about 0.5 mg/L just prior to what this research termed "DO breakthrough". The DO breakthrough occurred immediately after ammonia and nitrite oxidation was completed, coincidental with the large reduction in oxygen uptake rate (OUR), as the DO concentration increased to a new level in equilibrium with the endogenous OUR and oxygen transfer rate. SBR pH and ammonia levels started to decrease at elapsed time = 30 min, indicating the onset of nitrification. Ammonia removal proceeded at a zero-order rate until ammonia levels reached a concentration of approximately 5 mg N/L. pH levels rapidly declined between elapsed time = 30 to 60 min, after which point the pH continued to slowly decrease until the breakthrough point. Here, the pH began to rise as the C0 2 stripping rate exceeded the post-nitrification C0 2 production rate, related to the cessation of nitrification-induced alkalinity consumption. Typical of all cycles, a small amount of nitrite accumulated during the aerobic-phase. The residual nitrite was quickly oxidized to nitrate as the ammonia was depleted at cycle end. Organic Nitrogen Profiles Figure 4.2 illustrates two soluble organic nitrogen data sets, where data were obtained for triplicate samples. Anoxic-phase (i.e. elapsed time = 5 and 19 min) samples contained about 4 mg N/L of soluble organic nitrogen. These levels were reduced to about 1 mg N/L during the aerobic-phase of the cycle (i.e. elapsed time = 120 min). Figure 4.3 shows the biomass organic nitrogen concentration for the same two data sets shown in Figure 4.2. Although biomass growth would occur during the course of an SBR cycle, the differences in mean biomass organic nitrogen concentration for samples collected at elapsed times = 5, 19 and 120 min were not statistically significant at a 5% level of significance. This finding was not unexpected, given the small (i.e. 1.7%) incremental increase in biomass N content over the course of one cycle, when the SBR was operated with a 10 d SRT and six cycles per day. The biomass organic nitrogen content was approximately 11% based on the biomass volatile suspended solids (VSS) concentration. This value was close to the 12% theoretical N content of cells, when assuming a cellular composition of C 5 H 7 N0 2 and the VSS measurement being representative of cell mass in the SBR. 62 6.0 _ l z 5.0 Ul E. 4.0 z o 'E 3.0 O) 6 2.0 n o 1.0 tf> 0.0 EMISC-15 UMISC-20 m 19 19 19 120 120 120 Elapsed Time (min) Figure 4.2: SBR soluble organic nitrogen concentration-time profiles (MISC-15, MISC-20) 250 _1 z o> 200 E_ z o 150 'E D> s Or 100 mas 50 o 3 0 EMISC-15 BMISC-20 19 19 19 Elapsed Time (min) 120 120 120 Figure 4.3: SBR biomass organic nitrogen concentration-time profiles (MISC-15, MISC-20) 63 Off-Gas/Headspace N20 Profiles Figure 4.4 illustrates the off-gas/headspace N 2 0 profiles for SBR-TS 16a. It shows off-gas data obtained by the on-line infra-red (IR) analyzer that sampled the reactor off-gas at a 60 s frequency, as well as data for the manually collected reactor headspace samples, which were obtained at 15 min intervals for subsequent gas chromatograph (GC) injection. The off-gas/headspace N 2 0 profiles demonstrate the smooth shape characteristic of normal operating cycles, with good agreement in absolute values provided by the two sampling and measurement methods. When viewed with the ammonia data, it can be seen that N 2 0 appeared subsequent to the start of nitrification, with N 2 0 levels rapidly declining after nitrification was complete. Furthermore, as shown in Figure 4.4, the off-gas N 2 0 concentration was observed to decrease in advance of its cessation with completed nitrification. Liquid N20 Profile Figure 4.5 contains the SBR liquid (i.e. mixed liquor) and headspace N 2 0 profiles for MISC-22. The headspace data have been converted from a volume/volume unit (i.e. ppm) to a mass unit (i.e. g N/m3) to allow direct comparison to the liquid concentrations, using common units. The data show that the headspace and liquid nitrous oxide-time profiles share the same general shape, suggesting that the headspace and liquid were in relative equilibrium with respect to N 2 0 distribution between the two phases. Carbon Utilization and TOC and PHB Profiles The baseline synthetic wastewater organic carbon sources included acetate and yeast extract. The soluble acetate would immediately be available for use by heterotrophic organisms, and thus could be described as "readily degradable carbon" (RDC). Alternately, the carbon contained in the yeast extract would not be available to microorganisms until the carbon was released from its complex form (i.e. hydrolyzed) by the action of extracellular enzymes. As a result, the yeast extract could be considered "slowly degradable carbon" (SDC). Carbon utilization, or more specifically, its conversion into biomass (i.e. synthesis) and carbon dioxide (i.e. energy production), would occur via denitrification during the anoxic-phase of the cycle. Here, the preferred carbon substrate would be acetate given its readily degradable nature. Alternately, during the 64 60 80 100 120 Elapsed Time (min) 180 700 -600 -E dd) 500 -O 400 -z in 300 -ro O 200 -8= O 100 -. J . "* 1 onia r —m—N2U 30 25 z 20 o> 15 I z 10 + X 5 z 60 80 100 120 Elapsed Time (min) 140 160 180 Figure 4.4: SBR headspace N20, off-gas N 2 0 , and ammonia concentration-time profiles (SBR-TS 16a) 65 Figure 4.5: S B R mixed liquor (cO and headspace (cg) N 2 0 concentration-time profiles ( M I S C - 2 2 ) 66 anoxic-phase, the yeast extract would undergo some hydrolysis. Most of the yeast extract carbon was likely carried over into the aerobic-phase for continued hydrolysis and subsequent utilization. Acetate, in excess of denitrification requirements, would also be carried over into the aerobic-phase for utilization. Figure 4.6 illustrates the dissolved total organic carbon (TOC) and biomass poly-B-hydroxybutyrate (PHB) profiles, with both parameters expressed in terms of carbon mass per unit volume of mixed liquor, for SBR-TS 16a, with several notable observations as follows: • The presence of 12 mg C/L of TOC near the end of the anoxic-phase (elapsed time =19 min) indicates that not all of the available wastewater-supplied carbon was required for denitrification, since denitrification was in fact complete by elapsed time = 19 min (Figure 4.1). • The residual anoxic-phase TOC was rapidly utilized during the aerobic-phase, and was complete within ten minutes (i.e. elapsed time = 30 min). This timing approximately corresponds with the onset of nitrification, suggesting that the nitrifying organisms could not complete with the heterotrophs for available oxygen during this period of high-rate carbon utilization. Subsequently, the soluble TOC levels remained essentially constant for the remainder of the aerobic-phase. Soluble microbial products, rather than the original wastewater carbon, likely comprised the measured TOC. • The PHB profile illustrates anoxic-phase PHB storage, with its subsequent utilization during the aerobic-phase. Other researchers (e.g. Dionisi et al, 2001) have also found anoxic-phase acetate storage, as PFTB, in anoxic/aerobic systems. This mechanism is generally believed to be the result of a "feast-famine" regime caused by the exposure of heterotrophic organisms.to extended aerobic periods with little readily degradable carbon substrates. • The PHB profile showed the typical, and significant, slope change at about elapsed time = 105 min, indicating the marked reduction in the rate of use of PHB carbon for cell synthesis and energy production. Figure 4.7 summarizes the anoxic-phase and aerobic-phase carbon utilization for SBR-TS 16a, with the data and calculations contained in Table 4.1. The carbon utilization in the two phases was calculated using TOC and PHB data, with two main assumptions. The first assumption was that the wastewater yeast extract (i.e. slowly degradable carbon) was not utilized as a denitrification carbon source, given the 67 30 25 5 20 o> £ 15 • o O in iO e; 0 -( -A 1 — A - — £ — i • ) 2 0 .4 0 . 6 0 . 8 Elaps 0 . 1( ed Time 30 1, (min) >0 140 1( 50 . 1i 30 -25 " d 90. o 2 0 O ) C 1 "". fc. CO X 10 CL i 0 -( ) 2 0 4 0 6 0 8 Elaps 0 1( ed Time )0 1. (min) >0 U to « 50 180 Figure 4.6: SBR TOC and PHB concentration-time profdes (SBR-TS 16a) 68 O 40% -i = 35% o n £ 30% o r ~\-— —i | 25% g 20% g "2 15% • N 3 10% o § 5% | 0% I 1 1 1 1 Denitrification P H B Enmeshed RD Anoxic-Phase Aerobic-Phase Figure 4.7: SBR anoxic-phase and aerobic-phase carbon utilization fractions (SBR-TS 16a) 69 Table 4.1: S B R anoxic-phase and aerobic-phase carbon ut i l izat ion calculations ( S B R - T S 16a) [TOC] distribution @ cycle start 1 : residual [TOC] from previous cycle = < 3 mg C /L A wastewater acetate 2 = 47 mg C/L B wastewater yeast extract 2 = 17 mg C /L C 67 mg C/L [TOC] @ end of anoxic-phase = 12 mg C /L D [PHB] @ end of anoxic-phase = 25.3 mg C /L E [TOC] @ end of aerobic-phase = mg C/L F [PHB] @ end of aerobic-phase = 7.2 mg C/L G Relative Wastewater Anoxic-Phase Summary Fraction Stored P H B m a s s 3 = 18 mg C/L H Adsorbed yeast m a s s 4 = 17 mg C /L I Residual R D C mass at phase end = 9 mg C /L J Calculated R D C mass utilized for denitrification = 20 mg C /L K 3 1 % o 64 mg C/L Aerobic-Phase Summary Oxidized P H B mass = 18 mg C /L L 28% P Oxidized enmeshed yeast m a s s 4 = 17 mg C /L M 27% Q Oxidized R D C mass = 9 mg C /L N 14% R 44 mg C/L Notes: Calculations: 1. Assumes instantaneous wastewater fill @ cycle start. H = E - G M = I 2. Calculated values based on wastewater composition. l = C N = J 3. Assumes initial anoxic-phase [PHB] is the same as J = D - A 0 = K/(B+C) the [PHB] at end of previous aerobic-phase. K = (B+C) (H+l+J) P = L/(B+C) 4. Assumed value; refer to text. L = H Q = M / (B+C) R = N / (B+C) 70 presence of acetate (i.e. readily degradable carbon). The second assumption was that the yeast extract was enmeshed, or adsorbed, into the biomass during the anoxic-phase and did not comprise a significant fraction of the measured TOC. The data illustrate several key points: • Approximately 31% of the wastewater carbon was utilized (i.e. biomass synthesis, energy production) during the anoxic-phase for denitrification. • The remaining 69% of the wastewater carbon was utilized during the aerobic-phase: 28% as PHB, 27%) as yeast extract (i.e. slowly degradable carbon), and 14% during the first 10 min of the aerobic-phase (i.e. readily degradable carbon). The distribution of wastewater carbon utilization across the anoxic-phase and aerobic-phase was a function of N 2 0 generation and the resulting mass of nitrate available for anoxic-phase denitrification. SBR cycles with reduced N 2 0 generation allowed a greater nitrate concentration at cycle end, causing a requirement for more carbon for denitrification during the subsequent cycle. Therefore, less carbon was stored as PHB during the anoxic-phase, with less carbon subsequently oxidized during the aerobic-phase. Overall Biomass Yield and Nitrogen Assimilation The calculated SBR overall biomass yield, for the operating 10 d solids retention time (SRT), was in the range of 0.26 to 0.33 g COD/g COD for the observed mixed liquor suspended solids (MLSS) concentration range of about 2,000 to 2,500 mg/L. The synthetic wastewater and biomass chemical oxygen demand (COD) equivalents were estimated using oxidation state and assumed cellular composition data obtained from Grady et al (1999). The theoretical nitrogen requirement for cell synthesis was estimated to be 3.8 to 4.7 mg N/L per SBR cycle, based on the observed yield and assumed cellular composition (Grady et al, 1999), for the range of observed MLSS levels. The theoretical nitrogen requirements compared well to estimates of 3.4 to 4.3 mg N/L per SBR cycle, calculated using the previously presented biomass organic nitrogen data. Ammonia Oxidation Rate versus Removal Rate Two main mechanisms, ammonia oxidation and ammonia assimilation, contribute to the reduction in reactor ammonia concentration during simultaneous carbon oxidation and nitrification. As a result, 71 cellular synthesis-ammonia assimilation requires consideration when using ammonia data to estimate the ammonia oxidation rate. Examination of the Figure 4.1 ammonia data for SBR-TS16a revealed an essentially constant aerobic-phase ammonia removal rate from the onset of nitrification (i.e. approximately elapsed time = 30 to 40 min) to elapsed time = 135 min. Nitrification proceeded ^ simultaneously with PHB and yeast extract carbon utilization for at least part of the aerobic-phase. The large reduction in the SBR-TS 16a PHB-time profile slope (Figure 4.5), at approximately elapsed time = 105 min, indicated that cell growth using PHB as a carbon source essentially stopped at that time. Ammonia assimilation, related to PHB induced growth, would have also ceased at this time. Assuming that growth on the adsorbed yeast extract carbon followed a similar trend, the data suggest two scenarios: • The ammonia oxidation rate was lower during the period leading up to elapsed time = 105 min, compared to the oxidation rate after elapsed time = 105 min. Simultaneously, the ammonia assimilation rate during the period up to elapsed time = 105 min made up the exact difference between the ammonia oxidation rate and observed ammonia removal rate. After elapsed time =105 min, when cell growth basically ceased, the ammonia oxidation rate was equal to the ammonia removal rate. This scenario seems unlikely, given the required coincidental response of the two mechanisms. • The ammonia assimilation rate was very small compared to the oxidation rate; therefore, the actual ammonia oxidation rate was essentially identical to the observed ammonia removal rate. The organic nitrogen content of the baseline synthetic wastewater, provided by yeast extract, provided a theoretical in-situ SBR concentration of 3.9 mg N/L at the start of the anoxic-phase. The organic nitrogen provided in the wastewater closely matched the actual cellular synthesis requirement (i.e. 3.4 to 4.3 mg N/L per SBR cycle). As a result, from a nitrogen balance perspective on a per SBR cycle basis, the "additional" nitrogen (i.e. organic) supplied in the wastewater compensated for the ammonia used for cell synthesis. As a result, the ammonia assimilation rate was essentially decoupled from the observed ammonia removal rate, allowing the ammonia oxidation rate to equal the ammonia removal rate. Therefore, this second scenario provides a more plausible 72 explanation for the observed constancy in the ammonia removal rate after the cessation of significant cell growth. Therefore, for the remainder of this document, the aerobic-phase ammonia oxidation rate was calculated using the ammonia data when the SBR was treating the baseline synthetic wastewater. Thus, for this case, the ammonia oxidation rate was assumed to be identical to the ammonia removal rate. However, when the SBR was subjected to a significant slowly degradable carbon utilization rate, beyond that provided by the baseline wastewater, the ammonia oxidation rate was not directly calculated using the ammonia data. Instead, the ammonia oxidation rate was estimated from the ammonia removal rate by adjusting the ammonia data for cell synthesis and ammonia assimilation (Section 4.5.3) Aerobic-Phase Oxygen Competition The previous section indicated that the ammonia removal rate, thus the ammonia oxidation rate, remained relatively constant throughout the aerobic-phase of the cycle regardless of PHB concentration. A closer examination of the PHB data (Figure 4.6) revealed a decreasing PHB utilization rate (i.e. 18 to 12 mg C/L/hr) between elapsed time = 45 to 105 min. However, after elapsed time = 105 min, the PHB utilization rate decreased to about 2.5 mg C/L/hr, a five-fold rate reduction. Therefore, the data suggest that the heterotrophic oxygen uptake rate (OUR), associated with PHB utilization, did not induce a significant competition for oxygen that could have impacted the autotrophic ammonia oxidation rate. A similar analysis regarding the yeast extract carbon utilization was conducted, although the yeast extract hydrolysis and subsequent carbon utilization rates were unknown. Assuming that all 17 mg C/L of carbon provided in the wastewater yeast extract was utilized only during the aerobic-phase, and by elapsed time = 105 min of the SBR cycle, a carbon utilization rate of 12 mg C/L/hr is realized. This rate is comparable to the PHB carbon utilization rate. However, the oxygen equivalent (i.e. COD) of yeast extract on a carbon basis (i.e. mg COD / mg C) is only about 25% of the oxygen equivalent of PHB on a carbon basis. Thus, the heterotrophic OUR required for yeast extract carbon utilization would be much lower than the PHB OUR. Therefore, by extension, any changes in yeast extract carbon utilization rate throughout the aerobic-phase would have likely had an insignificant impact on oxygen availability and the ammonia oxidation rate. In conclusion, the heterotrophic oxygen requirement for carbon utilization during the aerobic-phase, when the SBR was treating the baseline wastewater, did not impact the ammonia oxidation rate. This 73 finding indicates that the competition for oxygen between heterotrophic utilization of exogenous carbon and autotrophic ammonia oxidation was not significant during treatment of the baseline wastewater. 74 4.3 REACTOR GAS MASS TRANSFER EXPERIMENTS Several sets of experiments were conducted to provide data on the N 2 0 and oxygen mass transfer characteristics of the reactors used in this research. These data provided insight into the dynamic mass transfer response of the reactors when subjected to physical and biological-physical processes. 4.3.1 N 2 0 Stripping 4.3.1.1 Total (Gas Transfer + Liquid-Surface Diffusion) Stripping The removal of a dissolved gas from a liquid contained in a reactor system involves two main processes: gas transfer and liquid-surface diffusion. This research defines "gas transfer" as the transfer of the dissolved gas of interest (i.e. N 20) from the reactor liquid into the gas (i.e. air) introduced into the reactor via the submerged, porous media, fine-bubble diffuser. The resulting blended gas stream exits the liquid at the liquid-headspace interface and subsequently leaves the reactor headspace as the off-gas. Alternately, N 2 0 dissolved in the reactor liquid can also be removed from the liquid by molecular diffusion across the liquid surface (i.e. liquid-headspace interface) into the reactor headspace, termed "liquid-surface diffusion" in this research. The extent of this process depends on replacement of the headspace air with N20-free air. Both of the described processes induce N 2 0 mass transfer when air is introduced into the reactors diffuser system; therefore, the combined rate of these processes yields the total stripping rate. Several N 2 0 stripping experiments were conducted to provide data on the relationship between air flow rate and N 2 0 stripping rate, and to estimate the total liquid N 2 0 mass transfer coefficient (i.e. kLa). Experimental Design The N 2 0 stripping experiments were conducted using a "clean" reactor that was identical in configuration to the reactor used in the SBR system. All stripping experiments were conducted using the baseline synthetic wastewater as the test fluid. Four stripping experiments were conducted, using air flow rates of 514, 876, 1,372 and 1,906 mL/min; these rates spanned the range used for SBR operation and perturbation. All N 2 0 data used in the analyses were measured using the gas chromatograph (GC) method. 75 The experimental procedure involved blending air (e.g. air flow rate = 280 mL/min) with a compressed air-N20 gas mixture (2,000 ppm N20), with the final gas stream introduced to the reactor via the fine bubble diffuser system. After a period of approximately 90 minutes, when the headspace and liquid N 2 0 concentrations were relatively constant over time, the test began with turning off the compressed air-N20 gas mixture flow and increasing the reactor air flow rate to the desired level. Results and Discussion^ - -The upper panel of Figure 4-8 illustrates typical data for a clean water (i.e. synthetic wastewater) total N 2 0 stripping experiment. The lower panel shows the linearized data (Matter-Muller et al, 1981) used to estimate the coefficients for predicting the liquid N 2 0 concentrations as a function of elapsed time and also the mass transfer coefficient. The models, developed using least squares regression analysis and shown as lines in the graphs, provide an excellent visual fit to the data. The data obtained from these experiments are further discussed in Section 4.3.1.2, in the context of the magnitude of the gas transfer stripping rate compared to the liquid-surface diffusion stripping rate. Figure 4.9 summarizes the calculated total N 2 0 mass transfer coefficients as a function of air flow rate for all conducted stripping tests. A linear model was fit to the data using least squares regression analysis. These data are further discussed in the context of the oxygen mass transfer rate in Section 4.3.2. 4.3.1.2 Liquid-Surface Diffusion Stripping Bench-scale reactors, due to their shallow depth, require very high air flow rates supplied to their diffuser systems, relative to a full-scale reactor, to provide specific oxygen transfer rates. This condition could produce a disproportionate N 2 0 stripping response in a bench-scale reactor compared to full-scale reactor. Specifically, any disproportionality would most likely be caused by the effects of liquid-surface diffusion, rather than gas transfer. Experimental Design Two experiments were conducted to evaluate the significance of liquid-surface diffusion to the total (i.e. combined) N 2 0 stripping mass transfer rate. In these experiments, subsequent to introducing N 2 0 into the reactor via the diffuser system as described in the previous section, air was supplied directly to the reactor headspace by introducing air into the top plate of the reactor, with no air (or N 20) supplied to the 76 Figure 4.8: Reactor liquid N 2 0 concentration-time profiles for total stripping experiment with air flow rate = 876 mL/min (NOSE-3a) 77 12 , 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 A i r F l o w Rate (mL/min) Figure 4.9: Reactor N 2 0 "total" mass transfer coefficient (20°C) versus air flow rate 78 submerged diffuser. A plastic baffle was fitted to the underside of the reactor top plate to reduce short-circuiting of air between its inlet and outlet locations (Figure 4.10). off-gas exhaust Figure 4.10: Reactor configuration for N 2 0 liquid-surface diffusion stripping experiments air supply Results and Discussion Figure 4.11 shows the data obtained from the two experiments, where the variable was air flow rate (514 mL/min and 1,372 mL/min). These air flow rates were selected as they coincided with rates used in the "total" stripping experiments, and also represent a wide range in magnitude. The reactor headspace- and liquid N 2 0 concentration-time profiles were essentially identical for both air flow rates. The decreasing rate of change in the liquid concentration was reflected in the slowly declining headspace concentration, after the initial large reduction in concentration. Figure 4.12 comparatively illustrates data obtained from liquid-surface diffusion and total stripping experiments that were conducted at the same air flow rate (1,372 mL/min). Although the headspace N 2 0 concentration-time profiles were essentially identical for both types of experiments, the liquid profiles were distinctly different. The latter case was due to the much slower stripping of N 2 0 from the liquid, to the headspace, provided by. liquid-surface diffusion, relative to the combined (i.e. total) stripping mechanism. The data indicate that the liquid-surface diffusion stripping rate was about one order-of-magnitude lower than the combined stripping rate. The liquid-surface diffusion experimental procedure likely underestimated the diffusion rate under normal reactor operation due to the absence of liquid-surface turbulence that results as gas bubbles leave the reactor liquid. However, the turbulence appeared minimal and thus would likely have only a slight impact on the actual diffusion rate. 79 n Q. IT) TO 10 a> X 350 300 250 200 150 100 50 0 I I ! • 1,372 mL/min a 514 mL/min 60 120 180 240 300 360 Elapsed Time (min) 420 480 540 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 — • — 1 , 3 7 2 mL/min 1 • -m— o i •t m u m in 60 120 180 240 300 360 Elapsed Time (min) 420 480 540 Figure 4.11: Reactor headspace and liquid N 2 0 concentration-time profiles for liquid-surface diffusion stripping experiments using air flow rates of 1,372 mL/min ( N O D S E - l a ) and 514 mL/min (NODSE-2a) 80 ro Q. tn T3 ro a> I 350 300 250 200 150 100 50 0 « liquid-surface diffusion total stripping * WF*= » • • • • * • 60 120 180 240 300 360 420 480 540 E l a p s e d T ime (min) 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 i i i • —•—liquid-surface diffusion -—H— total stripping V V ^^^^^^ • 60 120 180 240 300 360 420 480 540 E l a p s e d T ime (min) Figure 4.12: Reactor headspace and liquid N 2 0 concentration-time profiles for total stripping (NOSE-2a) and liquid-diffusion stripping ( N O D S E - l a ) experiments conducted at an air flow rate = 1,372 mL/min 81 Process Monitoring and Control Implications Based on the obtained data, liquid-surface diffusion stripping appears to be insignificant, compared to gas transfer stripping, for the reactors used in this research. Assuming a similar liquid surface area to liquid volume ratio in a full-scale reactor, the data indicate that the high aeration rates needed to provide the required oxygen mass transfer rates in the bench-scale reactor should not alter the shape of the N 20-time profile, compared to the same profile for a full-scale reactor. Specifically, the low oxygen transfer efficiency of the bench-scale reactors also translates into a low N 2 0 stripping efficiency. A full-scale reactor, with much higher oxygen transfer-and N 2 0 stripping efficiencies, will require a much lower aeration rate to provide the same mass transfer rates. Therefore, in a full-scale reactor, the shape of the N20-time profile should be the same as that of the bench-scale SBR, but with higher N 2 0 concentrations due to the reduced dilution of the off-gas with air. 4.3.2 G 2 Transfer Several clean water (i.e. synthetic wastewater) oxygen transfer tests were conducted to provide data on the oxygen mass transfer characteristics of the reactor for the range of air flow rates used in the SBR experiments. The data were used in planning the various SBR aeration rate perturbation experiments discussed in Section 4.7. Experimental Design The oxygen transfer experiments were conducted using the "clean" reactor. All experiments were performed using the baseline synthetic wastewater as the test fluid. Tests were conducted using air flow rates of 514, 724, 952, 1,163, 1,372, 1,645 and 1,906 mL/min. Duplicate experiments where conducted at four of the seven air flow rates. The general experimental and data reduction methods followed the procedures outlined in ASCE (1992). For simplicity, the log deficit parameter estimation method was used for the analysis, rather than the non-linear regression procedure. This procedure was deemed appropriate given the context of use for the developed information. Results and Discussion Figure 4.13 summarizes the calculated oxygen mass transfer coefficients as a function of air flow rate for all conducted oxygen transfer tests. A non-linear model was fit to the data given the expected 82 -- 1 1 1 y = (10- 1 4 3 7 ) (x 0 7 7 2 ) R 2 = 0.98 • 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 Air Flow Rate (mL/min) Figure 4.13: Reactor 0 2 mass transfer coefficient ( 2 0 ° C ) versus air flow rate 83 performance characteristics of the reactors diffused aeration system. The oxygen mass transfer coefficient can be used to estimate the mass transfer coefficient of another compound if the molecular diffusion coefficients (D) for the compound of interest and oxygen are known (Matter-Muller et al, 1981). Specifically, the ratio of the mass transfer coefficients equals the ratio of the diffusion coefficients. The D values for oxygen and N 2 0 are 2.04 x 10"5 cm2/s (Metcalf and Eddy, 1991) and 1.77 x 10"5 cm2/s (von Schulthess et al, 1995), respectively, for a temperature of 20°C. Therefore, the D N 2 0 / D 0 2 ratio = 0.87. • . Figure 4.14 illustrates the relationship between the N 2 0 and oxygen mass transfer coefficients for the reactor used in this research. The data points were generated using the models that were developed for the experimental data, representing the range of air flow rates used in the experiments. The slightly non-linear characteristic of the generated data resulted from the non-linear model that was developed for the oxygen mass transfer coefficient. The line shown in the plot represents a linear model fit to the generated oxygen and N 2 0 mass transfer coefficient data. The slope of the line contained in Figure 4.14 equals 0.76, which represents the KLaN 2o / KLao2 ratio. The difference between the experimentally determined K L a N 2 0 / KLao2 ratio and the theoretical D N 2 0 / D 0 2 ratio, relative to the KLaN 2o / KLao2 ratio, was about 13%. The good agreement between the ratios confirms the quality of the collected oxygen and N 2 G mass transfer data. 4.3.3 N 2 0 Liquid-Headspace Partitioning Experimental logistics precluded the high frequency collection of SBR mixed liquor N 2 0 samples, simultaneous with headspace samples, in many of the experiments conducted for this project. Therefore, several SBR experiments were specifically conducted to investigate N 2 0 partitioning between the reactor liquid and headspace when N 2 0 was biologically generated during the aerobic-phase of the cycle. The SBR cycles represent a wide range of aeration rates, purposely manipulated to alter the ammonia oxidation and N 2 0 generation rates when the SBR was treating the baseline synthetic wastewater. Complete nitrification was obtained prior to cycle end for experiments MISC-22 and MISC-23, with ammonia and nitrite oxidation remaining incomplete at cycle end for the MISC-24 experiment. The SBR headspace N 2 0 data presented in this section are shown in units of mass per unit volume (i.e. g N/m3), as opposed to the volume per volume unit (i.e. ppm = (v/v) x 1,000,000). This allowed direct 84 y = 0.759x •-.2 „ « n r\ -If1 *-* 0 2 4 6 8 10 12 14 0 2 k[_a @ 20°C (hr1) Figure 4.14: Reactor N 2 0 mass transfer coefficient versus 0 2 mass transfer coefficient (20°C) for air flow rates = 500 to 1,900 mL/min 85 comparison of headspace N 2 0 levels with liquid N 2 0 concentrations, also shown below in units of g N/m3. All N 2 0 levels were measured using the GC method. The information provided by these experiments provided insight into N 2 0 gas dynamics in the SBR, including the potential to estimate N 2 0 liquid concentrations and generation rates using only the headspace data. General Shape of Liquid- and Headspace-Time N20 Profiles Figure 4.15 illustrates headspace (cg) and mixed liquor (ci) N 2 0 concentration-time profiles for the three different SBR cycles. In general, the data show that the headspace and liquid N 2 0 profiles shared the same general shape under the various SBR operating conditions. The profiles in Figure 4.15c do not have the bell shape shown in Figures 4.15 a and b, since MISC-24 (Figure 4.15c) was purposely conducted with a low aeration rate. The low aeration rate reduced the ammonia and nitrite oxidation rates to the point that nitrification remained incomplete at the end of the aerobic-phase. The profiles were smooth with no evidence of large, short-duration "spikes" in liquid N 2 0 levels. In addition, although not shown, the SBR off-gas N 2 0 data obtained by the IR analyzer (60 s sampling frequency) demonstrate the same smooth shape as the headspace N 2 0 data (15 min sampling frequency) shown in Figure 4.15. N20 Partitioning and Estimation of Liquid N20 Concentrations Although the headspace and mixed liquor N 2 0 profiles shared the same general shape, the Figure 4.15 data indicate that the ratio between headspace and mixed liquor concentrations varies throughout the SBR cycle. Figure 4.16 plots the same data in terms of the Cg/q ratio against elapsed time for the various SBR cycles. Examination of the Figure 4.16 data revealed several key points. First, the value of the cg/ci ratio remained well below the value of the dimensionless N 2 0 Henry coefficient (H = 1.6 at 20°C) (von Schulthess et al, 1995), that describes the static equilibrium for the headspace and liquid partitioning, at all times during the cycle. This indicates that the partitioning of the N 2 0 mass contained in the reactor never reached the static equilibrium. In other words, the off-gas was not completely saturated with N 20. Matter-Muller et al (1981) indicated that conditions that favour an unsaturated off-gas include compounds with high H values (i.e. H > 0.1) and reactors with short bubble-liquid contact times. Both of these conditions apply to the utilized reactor system. 86 E z 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 — — • — e g -Wt-c\ 0.8 0.7 0.6 0.5 £ 0.4 S3) 0.3 ' 0.2 0.1 0 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) MISC-22 (aeration rate = 1,040 mL/min, A O R = 11.8 mg N/L/hr) 0.45 0.40 0.35 „~ 0.30 | 0.25 0.20 o* 0.15 0.10 0.05 0.00 _ J u J — • — c g r - | - * - C l | --- A 0.8 0.7 0.6 0.5 £ 0.4 2 0.3 o 0.2 0.1 0 0 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) MISC-23 (aeration rate =1,510 mL/min, A O R = 17.0 mg N/L/hr) 0.30 0.25 a* 0.20 E S 0.15 s o* 0.10 0.05 0.00 .  i — • — e g — A — e l -—ft— -0 20 40 60 80 100 120 140 Elapsed Time (min) 0.50 0.45 0.40 0.35 , 0.30 "E 0.25 z 0.20 3 0.15 0.10 0.05 0.00 160 180 c) MISC-24 (aeration rate = 670 mL/min, A O R = 6.5 mg N/L/hr) Figure 4.15: S B R headspace (c g) and liquid (C]) N 2 0 concentration-time profiles for a) MISC-22 (aeration rate = 1,040 mL/min, A O R = 11.8 mg N/L/hr) , b) MISC-23 (aeration rate =1,510 mL/min, A O R = 17.0 mg N/L/hr) and c) MISC-24 (aeration rate = 670 mL/min, A O R = 6.5 mg N/L/hr) experiments 87 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) MISC-22 (aeration rate = 1,040 mL/min, AOR = 11.8 mg N/L/hr) s*—' 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) MISC-23 (aeration rate = 1,510 mL/min, AOR = 17.0 mg N/L/hr) J+—4 * N >—* 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) MISC-24 (aeration rate = 670 mL/min, AOR = 6.5 mg N/L/hr) Figure 4.16: S B R headspace (c g) / l iquid (C | ) N 2 0 ratios versus elapsed time for a) MISC-22 (aeration rate = 1,040 mL/min, A O R = 11.8 mg N/L/hr) , b) M I S C -23 (aeration rate =1,510 mL/min, A O R = 17.0 mg N/L/hr) and c) MISC-24 (aeration rate = 670 mL/min , A O R = 6.5 mg N/L/hr) experiments 88 Second, during the period of N 2 0 generation, the reactors "steady-state" cg/ci ratio was approximately 0.6, regardless of aeration rate, ammonia oxidation rate or N 2 0 generation rate. Third, after the cessation of N 2 0 generation, simultaneous with completed ammonia oxidation, the cg/q ratio sharply increased as the overall N 2 0 mass transfer process shifted from generation + stripping to only stripping. The presented data suggest that the mixed liquor N 2 0 concentration could be reasonably estimated using head-space N 2 0 information and a partition coefficient of 0.6, but only during the period of constant N 2 0 generation. As discussed in the following section, using the MISC-22 experiment as an example, the N 2 0 generation rate was constant between elapsed times = 75 and 135 min (Figure 4.17). This time period corresponds with the period of constant cg / Ci ratio, shown in Figure 4.16a. Matter-Muller et al (1981) developed.a batch reactor model for stripping of volatile gases. The model allowed estimation of liquid contaminant concentrations by modifying "the Henry coefficient, through application of a constant parameter value, based on the experimentally determined reactor mass transfer characteristics. However, the data shown in Figure 4.16 clearly illustrate a changing cg / C| ratio during SBR operation. Therefore, application of the Matter-Muller et al (1981) model to the current research would not provide accurate estimations of liquid N 2 0 concentrations. Other N 2 0 studies (e.g. von Schulthess et al, 1994 and 1995) have used the Matter-Muller et al (1981) model to estimate liquid concentrations and ultimately calculate N 2 0 generation rates. However, the suitability of this approach may be questionable, particularly in the case where the N 2 0 was being generated as well as stripped. Comparison of N20 Generation and Stripping Rates Along with the SBR headspace N 2 0 data (cg), Figure 4.17 shows the generated and stripped cumulative N 2 0 mass as a function of elapsed time for the same three SBR cycles. The "stripped" cumulative mass was calculated using only the headspace N 2 0 data, headspace volume and aeration rate, and reflects the mass of N 2 0 stripped from the liquid over the course of a cycle. Alternatively, the "generated" cumulative mass curve includes the stripped N 2 0 mass, plus the mass of N 2 0 contained in the mixed liquor. For the MISC-22 experiment shown in Figure 4.17a, the generated and stripped cumulative N 2 0 mass curves display an essentially linear region between elapsed times of 75 and 135 min, indicating a constant N 2 0 generation rate. Linear models fitted to the data using least squares regression analysis 89 I a 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 •eg • stripped • generated 60 80 100 120 140 160 180 Elapsed Time (min) a) MISC-22 (aeration rate = 1,040 mL/min, AOR = 11.8 mg N/L/hr) 0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 ' ' h — • — e g — • — stripped A generated • _ —m— -• 50 45 40 35 30 25 20 15 10 5 0 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) MISC-23 (aeration rate =1,510 mL/min, AOR = 17.0 mg N/L/hr) 0.30 n 0.25 £ 0.20 Z 0.15 3 O 0.10 0.05 0.00 -eg - stripped - generated 15 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) MISC-24 (aeration rate = 670 mL/min, AOR = 6.5 mg N/L/hr) Figure 4.17: S B R headspace N 2 0 (c g) concentration-, cumulative generated N 2 0 mass- and cumulative stripped N 2 0 mass-time profiles for a) MISC-22 (aeration rate = 1,040 mL/min, A O R = 11.8 mg N/L/hr) , b) MISC-23 (aeration rate =1,510 mL/min, A O R = 17.0 mg N/L/hr) and c) MISC-24 (aeration rate = 670 mL/min , A O R = 6.5 mg N/L/hr) experiments 90 provided calculated N 2 0 generation and stripping rates of 0.50 mg N/min and 0.46 mg N/min, respectively. The ratio of N 2 0 stripping rate to generation rate was 0.92, suggesting that stripping data alone could provide a reasonably accurate estimate of the N 2 0 generation rate for this particular SBR cycle. The MISC-23 data shown in Figure 4.17b also show linear regions, although for shorter durations, between elapsed times 75 and 105 min for N 2 0 generation, and between times 75 and 120 min for N 2 0 stripping. For this cycle, the calculated N 2 0 generation and stripping rates were 0:64 mg N/min and 0.59 mg N/min, respectively, with the stripping rate / generation rate ratio = 0.92. Again, the stripping data alone could be used to fairly accurately estimate the N 2 0 generation rate. However, the MISC-24 data shown in Figure 4.17c illustrate a different scenario. For this cycle, nitrification was incomplete at cycle end due to the low kinetic rates induced by the reduced aeration rate. Linear models fitted to data between elapsed times 150 and 179 min provide calculated N 2 0 generation and stripping rates of 0.20 mg N/min and 0.16 mg N/min, respectively. The ratio of N 2 0 stripping rate to generation rate was only 0.80, indicating a reduced accuracy in estimating N 2 0 generation rate using only stripping data under this specific set of operation conditions. Steady-State Headspace N20 Concentration and N20 Generation Rate An alternate method to estimate N 2 0 generation rate, without collection of liquid samples for subsequent N 2 0 analyzes, involves the use of the "steady-state" headspace N 2 0 concentration. First consider Figure 4.15a. The data show that the liquid N 2 0 concentration was increasing during the initial portion of the aerobic-phase of the SBR cycle. This observation indicates that N 2 0 was generated at a rate that initially exceeded the reactors N 2 0 stripping rate, resulting in the accumulation of N 2 0 in the mixed liquor. Then, at around elapsed time = 105 min, the liquid N 2 0 concentration leveled off and remained fairly constant until just after elapsed time = 140 min when N 2 0 generation subsequently slowed down and then stopped (Figure 4.17a). Reactor headspace N 2 0 levels were also fairly constant during this same time period. During this "steady-state" period the N 2 0 generation rate was equal to the reactors stripping rate, allowing no further increase in liquid concentration. This indicates that the liquid N 2 0 concentration was high enough such that the concentration gradient between the liquid and air bubbles passing through the liquid, for the provided aeration rate, allowed the N 2 0 mass transfer rate (i.e. stripping rate) to equal the generation rate. 91 The reactor headspace concentration during the steady-state period was approximately 0.47 g N/m . Multiplying this value by the 1,040 mL/min aeration rate, and correcting for units, translates into a stripping rate of 0.49 mg N/min. This calculated stripping rate is essentially identical to the 0.50 mg N/min generation rate value that was previously calculated for the linear region of the cumulative N 2 0 mass generation curve. Process Monitoring and Control Implications The presented data indicate that the -N20 generation rate could be estimated with reasonable accuracy using only measured reactor headspace or off-gas concentrations when combined with the known parameters of aeration rate and reactor headspace volume. Thus, the use of the N 2 0 stripping rate as a surrogate parameter for N 2 0 generation rate greatly simplifies implementation of a N20-based monitoring system in a full-scale reactor. Estimating the N 2 0 generation rate using only the "steady-state" headspace or off-gas N 2 0 concentration can only be done if the headspace and liquid N 2 0 levels reach a steady-state condition. The method cannot be used if nitrification remains incomplete at cycle end. In addition, the method cannot be used if nitrification is completed prior to reaching the point of N 2 0 equilibrium in the reactor. 92 4.4 N z O SOURCE EXPERIMENT Section 2 discussed the possible sources of N 2 0 generated in biological wastewater treatment systems. This project specifically investigated the source of aerobic-phase N 2 0 for the SBR biomass and operating conditions used in the research. The following sections present and discuss the experimental design and resultant findings. Experimental Design Early attempts to investigate the source of aerobic-phase N 2 0 were conducted using waste SBR biomass in a small, off-line, gas-tight, respirometry cell. DO control was difficult in the small cell. Furthermore, the different configuration, gas supply and mixing characteristics of the small cell impacted reaction kinetics, making comparisons to the normal operating characteristics of the SBR difficult. Therefore, the NOME-8 experiment, discussed below, was conducted using the SBR as the test reactor. The NOME-8 experiment involved two back-to-back SBR cycles. The first cycle was the "baseline" cycle, where the SBR was operated as usual (aeration rate = 1,040 mL/min) while treating a slightly modified version of the baseline synthetic wastewater. For reasons discussed below, the synthetic wastewater did not contain yeast extract, as the yeast was a source of organic nitrogen and ultimately ammonia. For the perturbation cycle, the SBR was fed synthetic wastewater that did not contain ammonia or yeast extract. The rationale for this approach was based on the model for ammonia oxidizing bacteria (AOB) nitrite reduction. Without feed ammonia, the AOB would not have a significant source of hydroxylamine, the required electron donor for autotrophic nitrite reduction and resulting N 2 0 generation. Other parameters (i.e. nitrite, DO, pH) were controlled during the perturbation cycle such that all SBR operating conditions, processes (with the exception of ammonia oxidation) and parameter profiles were similar to those of the baseline cycle: • A concentrated nitrite solution (2.6 g sodium nitrite/500 mL) was continuously pumped (2.4 mL/min) to the SBR after the start of the aerobic-phase (elapsed time = 25 min to 125 min) to provide a source of nitrite to the nitrite oxidizing bacteria (NOB), simulating AOB nitrite generation. The nitrite 93 addition rate was purposely set at a rate that was slightly higher than the current biomass nitrite oxidation rate to ensure the presence of some residual nitrite in the mixed liquor. Continuous addition of nitrite to the SBR, instead of discreet nitrite spikes, provided an improvement in the procedure reported in Shiskowski et al (2004). A 5 mg N/L nitrite spike was also added to the SBR at elapsed time = 103 min while the SBR continued to receive the pumped nitrite solution. The spike was added to provide a contingency in the case where the pumped nitrite addition was not leaving any measurable residual nitrite in the mixed liquor. • The DO was manually controlled during the aerobic-phase to maintain similar DO concentrations as those of the baseline cycle. Compressed nitrogen was blended with air so that the total gas flow rate (i.e. 1,040 mL/min) supplied to the SBR during the perturbation cycle was the same as the baseline cycle. Preliminary in-situ oxygen uptake tests were conducted in the days preceding the NOME-8 experiment to refine the DO control strategy in terms of the required aeration rate in the absence of ammonia oxidation. • The pH was controlled during the aerobic-phase to maintain similar pH levels to those of the baseline cycle. The pH control scheme involved continuously pumping (2.2 mL/min) a 0.1 N HC1 solution to the SBR between elapsed time = 29 min to 149 min. Discreet volumes of 0.1 N HC1 were manually added to the SBR to refine the pH control (15 mL at 65 min, 10 mL at 124 min). Preliminary alkalinity titrations were conducted to refine the pH control strategy, in terms of selecting the most appropriate acid normality and HC1 solution addition rate to provide the desired change, and rate of change, in bioreactor pH level. Unfortunately, the electrical signal received by the data logging system from the in-situ pH probe was erratic on the day the NOME-8 experiment was conducted. Therefore, mixed liquor samples were manually withdrawn from the SBR for off-line pH measurement, with the mixed liquor then immediately returned to the SBR. Baseline and Perturbation Cycle Process Control Figure 4.18 shows the concentration-time profiles for the directly (i.e. DO, pH) and indirectly (i.e. nitrite, nitrate) controlled parameters. The data indicate that the perturbation cycle DO and pH control strategies 94 Figure 4.18: SBR DO, pH, nitrite and nitrate concentration-time profiles; perturbation = no ammonia in wastewater (NOME-8) 95 were successful in controlling these parameters to levels very similar to those of the baseline cycle. Similarly, the Figure 4.18 nitrate data confirm that the nitrite oxidation rate for both cycles was virtually identical. Prior to the nitrite spike at elapsed time = 103 min, the perturbation cycle nitrite (Figure 4.18) concentrations were occasionally higher, but not excessively, than those of the baseline cycle. This response was likely the result of adding nitrite to the SBR at a slightly higher rate than the usual biological nitrite generation rate. Elapsed Time = 20 to 103 min - Results and Discussion As per usual SBR operating procedures, the aerobic-phase portion of the cycle started at elapsed time = 20 min. The baseline cycle ammonia concentration-time profile shown in Figure 4.19 illustrates the zero-order ammonia oxidation kinetics typical of the SBR. The baseline cycle ammonia oxidation rate was 13.5 mg N/L/hr. The perturbation cycle ammonia levels were expectedly low, and less than 0.1 mg N/L at all times. The baseline cycle headspace N 2 0 profile (Figure 4.19) possessed the typical shape, with significant levels (i.e. 12 ppm) recorded by elapsed time = 45 min., ultimately reaching approximately 50 ppm by elapsed time = 103 min. Alternately, the data clearly show that no notable N 2 0 was generated during the first 103 minutes of the perturbation cycle. Although not shown, N 2 0 data obtained using the IR analyzer confirm the GC data. The data suggest several key points: • First, assume that heterotrophic organisms were responsible for N 2 0 generation during the baseline cycle via the "conventional" denitrification pathway. Lloyde et al (1987) has reported continued denitrification under aerobic conditions following anoxic conditions. The possibility exists that heterotrophs were reducing nitrite all the way to N 2 , rather than stopping short at N 20, during the perturbation cycle. This situation implies that the heterotrophs N 2 0 reductase enzyme was "poisoned" (i.e. inhibited) during the baseline cycle, resulting in N 2 0 accumulation, but somehow this inhibition was removed during the perturbation cycle. DO (von Schulthess et al, 1994), nitrite and nitric oxide (NO) (von Schulthess et al, 1995), low pH (Thorn and Sorensson, 1996) and pH-controlled free hydrogen sulphide (Schonharting et al, 1998) all reportedly can inhibit the N 2 0 reductase enzyme. However, DO and pH levels were very similar during both cycles, eliminating 96 Figure 4.19: SBR ammonia, nitrite, headspace N 2 0 and headspace NO concentration-time profdes; perturbation = no ammonia in wastewater (NOME-8) 97 these parameters as impacting enzyme inhibition. Compared to the baseline cycle, nitrite levels were slightly higher during the perturbation cycle (Figure 4.19). Headspace NO (Figure 4.19), thus liquid NO, levels lower during the perturbation cycle. Based on the literature data, the low nitrite and NO levels during both cycles would likely not impact enzyme activity. Therefore, the removal of N 2 0 reductase inhibition during the perturbation cycle, and its cause for the lack of N 2 0 generation during the cycle, appears highly improbable. Second, during anoxic heterotrophic denitrification, Itokawa et al (2001) and Wicht (1996) found that N 2 0 production was maximized under carbon-limited (e.g. endogenous) conditions. Similar to the enzyme inhibition situation, no generation of N 2 0 during the perturbation cycle implies the removal of a carbon limitation, if this was the actual reason for N 2 0 generation. Samples for soluble TOC and cellular PHB were not collected during the NOME-8 experiment. However, as discussed in Section 4.5.2, available TOC data for other SBR cycles treating the baseline wastewater indicate that, within 10 min after the start of the aerobic-phase, soluble TOC was reduced to a low level (i.e. < 5 mg/L) with concentrations remaining at that level for the remainder of the cycle. Therefore, both the baseline and perturbation cycles were carbon-limited with respect to soluble carbon. Similarly, PHB data available for other cycles shows the storage of some acetate during the anoxic-phase, with subsequent oxidation during the aerobic-phase. The data indicate that the PHB-time profile during the aerobic-phase was relatively insensitive to operating conditions, making it unlikely that there were significant differences in PHB levels between the cycles. Thus, both the baseline and perturbation cycles were likely operating under the same stored carbon regime. As a result, differences in carbon levels between the baseline and perturbation cycles were likely small and therefore, not expected to have been related to the lack of N 2 0 generation during the perturbation cycle. • Given the data and preceding elimination of heterotrophic related factors, it is clear that significant N 2 0 generation only occurred in the presence of ammonia and its oxidation. Based on this observation, a number of N 2 0 generation mechanisms are possible. Chemical decomposition of hydroxylamine, if it was expelled from the AOB cells (autotrophic and/or heterotrophic), could theoretically generate N 2 0 (Chalk and Smith, 1983). However, research conducted by Simm et al, (2004a) found that hydroxylamine chemical decomposition, even at elevated liquid hydroxylamine concentrations, produced N 2 0 in quantities and at rates far below 98 levels observed for the SBR. Their experiments were conducted over a range of pH and DO levels using various liquid media (bioreactor effluent, synthetic wastewater), with live biomass and with killed biomass (via heat treatment). Furthermore, although no samples were collected during the NOME-8 experiment, available hydroxylamine data for other SBR cycles showed levels below the method detection limit (i.e. 0.05 mg/L), as per the hydroxlyamine method described in Simm et al (2004b). Therefore, based on the available SBR hydroxylamine data, combined with the results of Simm et al (2004a), this chemical decomposition mechanism was ruled out as a significant contributor to the observed SBR N 2 0 generation. The remaining possible N 2 0 generation mechanisms are related to ammonia and its biological oxidation by AOB. Heterotrophic AOB (i.e. heterotrophic nitrifiers) can generate N 2 0, since they share the general biochemical pathways of autotrophic AOB (Wrage et al, 2001). Bock et al (1991) described heterotrophic nitrification as a co-oxidation process where nitrification was not an energy generating mechanism. As such, heterotrophic nitrifiers use organic carbon as a source of carbon and energy (Wrage et al, 2001). The lack of available soluble aerobic-phase carbon would preclude the proliferation of heterotrophic AOB in the SBR. Similarly, heterotrophic nitrifiers also prefer acidic environments (Schmidt et al, 2003), contrary to the neutral to basic conditions within the SBR. Therefore, their contribution to ammonia oxidation and N 2 0 generation, relative to autotrophic AOB, was likely very limited in the SBR. Thus, autotrophic AOB form the context of the remaining discussion. Hooper et al (1990) indicated the possibility of "direct oxidative production" of N 2 0 by the hydroxylamine oxidoreductase (HAO) enzyme, as isolated HAO (i.e in-vitro) has been shown to produce NO and N 2 0 during hydroxylamine oxidation. However, more recent work noted that the operation of this mechanism in-vivo remain unclear (Beaumont et al,. 2002). Therefore, the significance of this mechanism in the observed SBR aerobic-phase N 2 0 generation are unknown. Finally, recent reviews of nitrifier denitrification (Wrage et al, 2001; Colliver and Stephenson, 2000) refer to the "classic" pure culture labelling studies conducted by Ritchie and Nicholas (1972) and Poth and Focht (1985). Their work with Nitrosomonas europaea showed that the nitrogen contained in N 2 0 originated from nitrite (i.e. not ammonia or hydroxylamine) during hydroxylamine oxidation -nitrite reduction. Similar to the NOME-8 experimental conditions, low DO levels appear to be the most important condition for inducing this pathway (Colliver and Stephenson, 2000). Based on the 99 current literature, this mechanism most likely explains the aerobic-phase generation of N 2 0 observed in the SBR. However, even this mechanism may not be the definitive process. Beaumont et al (2002) disrupted the gene that encodes the nitrite reductase enzyme in Nitrosomonas europaea, resulting in nitrite reductase deficient cells. However, the cells were still able to generate NO and N 20. The researchers suggested the alternative pathway for gas production could involve the previously described HAO enzyme. NO is an intermediate compound generated under the oxygen-limited conditions that induce AOB to reduce nitrite to N 2 0 (Wrage et al, 2001.);' therefore, its appearance during:N20 generation was expected (Figure 4.19). Baseline cycle NO concentrations (i.e. 0.4 ppm) were typicafof those for other SBR cycles, also showing the usual characteristic where the shape of the NO-time profile closely follows the nitrite profile. NO concentrations were low, possibly since NO is chemically oxidized to nitrogen dioxide (N02) in the presence of oxygen (Schmidt et al, 2001). Very low NO levels (i.e. < 0.1 ppm) were also measured during the perturbation cycle. Chemical nitrite decomposition can generate NO (Chalk and Smith, 1983). For this project an experiment (NOME-6), conducted in a clean reactor identical to the SBR, specifically investigated the rate of NO and N 2 0 formation when aerating (1,022 mL/min) synthetic wastewater (pH varied between 7.2 and 8.5) containing 90 mg N/L ammonia and 10 mg N/L nitrite. Measured reactor headspace NO concentrations were between 0.05 and 0.15 ppm. Measured headspace N 2 0 levels were constant at typical background levels (i.e. 0.2 to 0.3 ppm). While the perturbation cycle NO levels were close to those that could be chemically generated, very limited biological nitrite reduction processes could also have generated NO during this period of the perturbation cycle. Elapsed Time = 103 to 180 min - Results and Discussion As previously discussed, a 5 mg N/L nitrite spike was added to the SBR at elapsed time = 103 min while the SBR continued to receive the pumped nitrite solution. The spike was added to provide a contingency in the case where the pumped nitrite addition was not leaving any measurable residual nitrite in the mixed liquor. Following the nitrite spike, the headspace NO concentration (Figure 4.19) rapidly increased, with the shape of the concentration-time profile closely following the nitrite profile. Furthermore, the post-nitrite spike NO levels were more than three times higher than the levels attained during the baseline cycle. 100 Headspace N 2 0 levels also increased after the nitrite spike, but the measured concentrations remained low and were approximately one order-of-magnitude lower than peak baseline cycle levels. The perturbation cycle response suggests that the post-nitrite spike NO and N 2 0 formation was largely biologically mediated. Again, the question becomes one of autotrophic and/or heterotrophic processes, von Schulthess et al (1995) found that adding a large nitrite spike (i.e. 22 mg N/L) to an excess carbon, denitrifying heterotrophic biomass resulted in the immediate generation of NO and N 20. The proposed response mechanism was nitrite inactivation of the NO reductase enzyme, resulting in NO accumulation, followed by NO inhibition of the N 2 0 reductase enzyme. The small nitrite spike used in the NOME-8 experiment relative to that of von Schulthess et al (1995) resulted in liquid NO levels that were estimated to be two orders-of-magnitude lower than those reported by the other researchers. Therefore, even if heterotrophic organisms could potentially generate N 2 0 under the SBR operating conditions, it appears unlikely that the nitrite spike could induce N 2 0 generation via the inhibition of the NO and N 2 0 reductase enzymes. ' •* .. Based on the preceding discussion, autotrophic reduction of nitrite again was the most likely source of the post-nitrite spike NO and N 20. Prior to the nitrite spike, the nitrite concentration inside the AOB periplasmic space would have been extremely low since these organisms were generating very little, if any, nitrite at this time. However, after the nitrite spike, the much higher bulk solution nitrite levels would have produced a nitrite gradient across the cell colonies and into the cells; this would suddenly allow a much higher nitrite concentration within the AOB periplasmic space and thus the electron acceptor for AOB denitrification (refer to Section 4.8.1 for further discussion on this topic). Hydroxylamine acts as the electron donor for AOB denitrification (Wrage et al, 2001), but only a very limited amount would have been available through breakdown of cellular material and nitrogen release, allowing some ammonia and hydroxylamine oxidation. As a result, the nitrite reductase (nitrite to NO) and NO reductase (NO to N 20) enzymes were fueled to generate a small amount of NO and N 20. The comprehensive model provided by Schmidt (Bock and Wagner, 2001) might explain the disproportionately high rate of NO generation relative to the N 2 0 generation, given the limited electron availability to drive these reductase reactions. Figure 4.20 illustrates the dramatic SBR response to small nitrite spikes during normal operating conditions, with the data supporting the hypothesis of the preceding discussion. For the NOME-3b experiment, the perturbation cycle received a 3 mg N/L nitrite spike at elapsed time = 44 min, and a 2.5 101 Figure 4.20: SBR DO, pH, nitrite and off-gas N 2 0 concentration-time profdes; perturbation = nitrite spikes (NOME-3b) 102 mg N/L spike at elapsed time =119 min. The SBR was fed the same baseline wastewater for both cycles. The aeration rate for both cycles was 1,040 mL/min, with the DO and pH profiles identical for both cycles. The data clearly show an immediate and significant response to the nitrite spike in the generated N 20. Data from a similar experiment (NOME-3) also support these results. Section 4.8 further discusses the findings of these specific experiments. Summary The utilized experimental procedure provided satisfactory control over DO, pH and nitrite levels for the SBR perturbation cycle relative to the baseline cycle. By eliminating DO, pH and nitrite as experimental variables, the influence of ammonia, and ammonia oxidation, on N 2 0 generation could be elucidated with a relatively high degree of confidence. The results strongly suggest that ammonia, in the presence of nitrite under low DO conditions, was required to generate N 20. As a result, the data support the hypothesis that autotrophic AOB were generating N 2 0 via hydroxylamine oxidation-nitrite reduction, since a significant quantity of N 2 0 was only generated when the AOB had a source of hydroxylamine, obtained from ammonia oxidation, to act as an electron donor to drive the reaction. In addition, for the studied system, it appears that aerobic-phase heterotrophic denitrification generated little, if any, N 20. These results are consistent with those obtained by Shiskowski et al (2004) for an SBR system operating under similar conditions, but with higher wastewater ammonia levels. The presented findings are significant in that the reviewed wastewater treatment literature reported few studies that attempted to specifically investigate the significance of autotrophic, relative to heterotrophic, N 2 0 generation in a mixed population biomass (i.e. autotrophs and heterotrophs) within an aerated environment. This could explain an earlier comment by Wicht (1996) that "... nitrification has not yet proved to be an important source of N 2 0 from wastewater treatment". More recently, N-labelling studies conducted by Beline et al (2001), Pahl et al (2001) and Gaul et al (2002) have investigated the relative significance of autotrophic and heterotrophic N 2 0 generation in mixed "population, aerated wastewater treatment bioreactors. For example, Pahl et al (2001) utilized a laboratory-scale, continuous-flow, completely mixed bioreactor to treat a piggery waste slurry (COD = 17,000 mg/L, total solids = 1.5%). As an example of one set of operating conditions, continuous aeration of the bioreactor, at a sufficiently high rate, provided 97% 103 ammonia removal. In this case, 24% of the removed ammonia was recovered as N 20. Their mass balance calculations, performed using the l 5 N data (i.e. 15N-labelled nitrate was added to bioreactor) as well as conventional nitrogen analysis data, showed that only 1% of the generated N 2 0 originated from heterotrophic denitrification. The remaining 99% of generated N 2 0 was attributed to autotrophic nitrification. Conversely, bioreactor operation with low aeration rates resulted in low ammonia removal (e.g. 34%), with 2% of removed ammonia recovered as N 20. In this situation, the generated N 2 0 was shown to originate solely from the induced heterotrophic denitrification, believed to have been the result of oxygen poisoning of the N 2 0 reductase enzyme that prevented conversion of N 2 0 to N 2 . The work of Beline et al (2001) was similar to Pahl et al (2001), where N 2Q generation was investigated under low and high bioreactor aeration rates for the treatment of piggery waste slurry. High bioreactor aeration rates provided nitrification and N 2 0 generation, but no N 2 0- 1 5 N was recovered in the off-gas. This indicates that none of the added l5N-labelled nitrate was reduced to N 2 0 via heterotrophic denitrification. Alternately, under low aeration rate conditions, 27%> of the added l5N-labelled nitrate was recovered as N 20-' 5N. In addition, N 2 0- 1 5 N made up about 88% of the total generated N 2 0, indicating that heterotrophic denitrification was likely the primary N 2 0 generation mechanism for this operating condition. This latter conclusion requires caution, as Pahl et al (2001) noted the potential for the heterotrophic biochemical reduction of 15N-labelled nitrate to l5N-labelled nitrite, followed by autotrophic AOB reducing the labelled nitrite to N 2 0- 1 5 N. However, the results of Beline et al (2001) confirm the findings of Pahl et al (2001), where nitrification can be the primary source of N 2 0 depending on bioreactor operating conditions. Thus, the findings of Beline et al (2001) and Pahl et al (2001) verify the results of the current research, where autotrophic AOB can be the primary source of N 2 0 generated in an aerated bioreactor. 104 4.5 WASTEWATER COMPONENT EXPERIMENTS This section examines the effect that various wastewater components had on nitrification performance and the resulting N 2 0 generation response (i.e. off-gas N20-time profile) when the SBR was operated with the same aeration rate for both the baseline and perturbation cycles. The examined wastewater components included ammonia load, readily degradable carbon load, and slowly degradable carbon, the latter expressed in terms of the slowly degradable carbon utilization rate. 4.5.1 Ammonia Load The ammonia load experiments presented in this section were unique, with respect to all other experiments discussed in this document, in that they investigated the SBR response to a type of perturbation that did not purposely induce a change in biomass oxygen supply/availability or alter the competition for oxygen among several groups of organisms. These experiments were conducted to investigate-the phenomenon noted in Section 4.2; here the N 2 0 generation rate was always observed to slow down during the latter part of the aerobic-phase, as indicated by the decreasing off-gas N 2 0 concentrations, in advance of its cessation with completed nitrification. The data presented in Sections 4.2 and 4.4 suggest that ammonia oxidation, when the SBR was treating the baseline wastewater, proceeded at a essentially constant rate subsequent to the onset of nitrification, and as described in Section 4.2, was not impacted by heterotrophic carbon utilization and the related oxygen demand. However, it was hypothesized that the observed reduction in N 2 0 generation rate was, indeed, related to a reduction in the ammonia oxidation rate. Therefore, several sets of experiments were conducted to investigate the effect that wastewater ammonia load and concentration had on nitrification kinetics and the N 2 0 generation response. Experimental Design The ammonia load experiments were conducted using two back-to-back SBR cycles. The SBR was fed the regular baseline synthetic wastewater during the baseline cycle. During the perturbation cycle, the SBR received synthetic wastewater where the ammonium chloride concentration was adjusted to provide the desired ammonia concentration. The bicarbonate concentration of the perturbation cycle wastewater was adjusted in direct proportion to the ammonia concentration in order to maintain a similar pH-time profile between cycles. The aerobic-phase aeration rate was set at 1,040 mL/min for both cycles. 105 A total of six ammonia load experiments were conducted for this phase of the research (Table 4.2). The six experiments were conducted in three groups during different weeks. The two experiments in each group were conducted on adjacent days. Initial Aerobic-Phase Ammonia Concentrations and Average Ammonia Oxidation Rates Table 4.2 summarizes the in-situ SBR ammonia concentration at the beginning of the aerobic-phase (i.e. "initial" concentration) for each experiment cycle. The initial perturbation cycle ammonia concentrations varied from about 11 mg N/L to 35 mg N/L, a three-fold range in concentration. The baseline cycle initial ammonia concentrations were approximately.24,mg N/L. Table 4.2 also contains the calculated reactor and specific average ammonia oxidation rates for each cycle. These rates were calculated using least squares regression to fit. a linear model to the ammonia data. As discussed in Section 4.2, the ammonia oxidation rate was assumed equal to the ammonia removal rate. The ammonia database for each cycle included the sample collected at elapsed time = 45 min (i.e. nitrification underway at this point) and ended using the last sample where the concentration was in excess of 1.0 mg N/L. The calculated oxidation rate was, in effect, an average for the aerobic-phase. The relative statistics (RS) shown in Table 4.2 highlight several key points. First, RS C indicates that the variability in specific ammonia oxidation rate (SAOR) between "grouped" baseline cycles, from adjacent days, was quite low and in the order of 1% to 2%. Second, increasing the perturbation cycle initial ammonia concentration by 16% (SBR-TS 13c) and 34% (SBR-TS 13a) above the baseline cycle concentration (RC A) induced a small increase in the SAOR (RC B) that could be accounted for in the cycle-to-cycle variability as given by RS C. A 36% increase in the initial ammonia concentration (SBR-TS 13e) did yield a slightly higher relative change in SAOR at 4.3%. The data indicate that the average ammonia oxidation rate was essentially independent of the initial ammonia concentration for the utilized increases in concentration. However, reducing the perturbation cycle initial ammonia concentration by 34% (SBR-TS 13b) and 50% (SBR-TS 13f), relative to the baseline cycle, caused a relative reduction in the SAOR of about 9% and 11%), respectively. This magnitude of SAOR reduction significantly exceeded the expected cycle-to-cycle variability, suggesting the possibility that the lower "operating" ammonia concentrations during 106 Table 4.2: Initial aerobic-phase ammonia concentration, average ammonia oxidation rate and relative statistics for wastewater ammonia load experiments Experiment Cycle Initial Average Ammonia Oxidation Rate Relative Statistics 2 3 ' 4 and Aerobic-Phase Grouping Ammonia Reactor Specific A B C Concentration1 (mg N/L) (mg N/L/hr) (mg N/L/hr per g MLSS/L) S B R - T S 1 3 a baseline 23.9 13.5 6.06 perturbation 31.9 13.8 6.22 34% 2.6% SBR-TS13D baseline 24.3 13.4 6.20 2.3% perturbation 16.1 12.2 5.63 -34% -9.2% S B R - T S 13c baseline 24.3 12.2 5.93 perturbation 28.1 12.4 6.04 16% 1.9% SBR-TS13d baseline 24.2 12.0 6.06 2.2% perturbation 18.5 11.8 5.92 -24% -2.3% SBR-TS13e baseline 25.4 12.8 6.56 perturbation 34.5 13.3 6.84 36% 4.3% SBR-TS13f baseline 22.8 12.3 6.49 -1 .1% perturbation 11.4 10.9 5.76 -50% -11.2% Notes: 1. Initial ammonia concentration = mean for samples collected at elapsed times = 19 and 30 min 2 A = (perturbation - baseline) / baseline x 100% for initial ammonia concentration data 3. B = (perturbation - baseline) / baseline x 100% for specific ammonia oxidation rate 4. C = (2nd baseline - 1st baseline) / 1st baseline x 100% for specific ammonia oxidation rate 107 these cycles reduced the average aerobic-phase SAOR. The following sections provide an examination of two specific scenarios to elucidate SBR behaviour and N 2 0 response and to highlight key observations that were consistently observed in the various experiments. The first scenario (SBR-TS 13f) examined the case where a reduction in wastewater ammonia load was accompanied by a reduction in the average ammonia oxidation rate. Alternately, the second scenario (SBR-TS 13a) considered the situation where an increase in ammonia load did not cause an increase in the average ammonia oxidation rate. Scenario 1 - Significant Change in Average Ammonia Oxidation Rate (SBR-TS13J) Before presenting the SBR-TS 13f data, it is worthwhile to first introduce two alternate formats for the off-gas N 2 0 data. These formats, described below, highlight the differences in the baseline and perturbation cycle N 2 0 concentration-time profiles and were used throughout this document when discussing experimental results: • Off-gas N20 mean slope. This parameter is the first derivative of the off-gas N 2 0 concentration-time profde. The mean slope value was calculated using a five minute moving mean, centered on the elapsed time of interest. The mean slope (MS) plot clearly indicates inflection points and locations of significant slope changes in the concentration-time profile. • Perturbation / baseline off-gas N20 ratio. The second data format plots the perturbation cycle / baseline cycle off-gas N 2 0 concentration ratio against elapsed time. The perturbation / baseline cycle off-gas N 2 0 ratio (P/B ratio) readily illustrates significant differences between cycles, since cycles with identical N 2 0 concentration-time profiles would have a constant P/B ratio of one. Figure 4.21 illustrates the ammonia, nitrite, nitrate and pH profiles for the SBR-TS13f experiment. Figure 4.22 shows the off-gas N 2 0, mean slope and P/B ratio profiles, as well as the DO profile. Key observations include: • The linear ammonia concentration models are shown as lines in the ammonia plot (Figure 4.21), and indicate an excellent visual fit to the aerobic-phase data. The perturbation cycle SAOR was 11% lower than the baseline cycle SAOR. 108 Figure 4.21: SBR ammonia, nitrite, nitrate and pH concentration-time profiles; perturbation = - 50% wastewater ammonia (SBR-TS 13f) 109 700 • 600 • ? (pp 500 • q 400 -z in 300 -a ff-G 200 • O 100 -I I I Baseline • Perturbation / >< /J // \ > 16 12 8 . 4 0 -4 • -8 :-12 -16 -20 -24 2.0 1.8 1.6 _ 1.4 *™ f . 1-0 O 0.8 ° 0.6 0.4 0.2 0.0 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 -Baseline -Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 60 80 100 120 Elapsed Time (min) 180 D t ^ A 1 i n A - on J < • V 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 Figure 4.22: SBR off-gas N 2 0 concentration-, mean slope-, perturbation/baseline ratio and DO concentration-time profdes; perturbation = - 50% ammonia (SBR-TS 13f) 110 • The baseline and perturbation cycles had virtually identical nitrite and nitrate profiles (Figure 4.21), even though the perturbation cycle ammonia oxidation rate was notably lower than the baseline cycle rate. The difference in the nitrogen mass appears to be accounted for, at least to some extent, in the reduced perturbation cycle off-gas N 2 0 concentration (Figure 4.22), as well as the mixed liquor N 2 0 concentration, during the first 90 minutes of the cycle (i.e. more ammonia but less N20). The headspace N 2 0 profiles, using the GC data, indicate the same trend (data not shown). A similar response was also evident in SBR-TS 13b (data not shown) where the perturbation cycle ammonia oxidation rate was 9% lower than the baseline cycle rate. • The N 2 0 mean slope and P/B ratio profiles (Figure 4.22) emphasize the differences in the baseline and perturbation cycle off-gas N 2 0 profiles. In particular, the P/B ratio profile provided a clear indication of the difference in the N 2 0 generation rate between the cycles that corroborated the difference in ammonia oxidation rates. For the perturbation cycle, the abrupt off-gas N 2 0 mean slope change at approximately elapsed time = 85 min (Figure 4.22) corresponded to the sudden reduction in N 2 0 generation as ammonia oxidation was nearing completion. The rapid rate of change in the mean slope indicated that the SBR headspace N 2 0 concentration was well below the "equilibrium concentration" for the given N 2 0 generation rate (Section 4.3.3). • The baseline and perturbation cycle pH (Figure 4.21) and DO profiles (Figure 4.22) were indistinguishable from one another even though there was a significant difference in ammonia oxidation rate between the cycles. The presented data indicate that differences in the initial aerobic-phase ammonia concentration could impact the average aerobic-phase ammonia oxidation rate, in turn also affecting the N 2 0 generation rate. Differences in ammonia oxidation rate could be identified using the off-gas N 2 0 data, but not the pH and DO data. Scenario 2 - No Change in Average Ammonia Oxidation Rate (SBR-TS 13a) Figure 4.23 illustrates the ammonia, nitrite, nitrate and pH profiles for the SBR-TS 13a experiment, with key observations discussed below: 111 35 _ 30 2 25 J L 20 | 15 *- 10 X z 5 0 • « • I I A Baseline • A ^ • He -turbatio n ^» 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 5 i 4 3 1 0 • 0 20 40 60 80 100 120 140 ' 160 180 Elapsed Time (min) I I 1 —mm—Baseline — • — Perturbation -* \ \ -A, 4 25 20 15 10 5 0 —A—Baseline — • — Perturbation ^—i 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 9.0 8.5 8.0 , 7 5 7.0 6.5 6.0 Baseline — • — Perturbation t 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) Figure 4.23: SBR ammonia, nitrite, nitrate and pH concentration-time profiles; perturbation = + 50% wastewater ammonia (SBR-TS 13a) 112 • Similar to the SBR-TS 13f experiment, the linear ammonia concentration models for the SBR-TS 13a experiment provide an excellent visual fit to the data. However, for the SBR-TS 13a experiment, both cycles had the "same" average ammonia oxidation rate. • Like the ammonia profiles, the nitrite profiles were virtually identical between cycles and differed only in the oxidation duration caused by the increase in wastewater ammonia load. In addition, during both cycles, the nitrite concentration remained constant from approximately elapsed time = 75 min until the completion of ammonia oxidation. • The nitrate profiles were similar for both cycles, but the perturbation cycle nitrate concentrations were always slightly lower than the baseline cycle concentrations for a given elapsed time. The significance of this observation is discussed later in the section. • The pH profiles were also very similar between cycles. Figure 4.24 shows the off-gas N 20, mean slope and P/B ratio profiles, as well as the DO profile. The data illustrate several important points: • Consistent with the observed equal average ammonia oxidation rate between the cycles, the off-gas N 2 0 concentration and mean slope profiles were virtually identical up until elapsed time = 70 min. Correspondingly, the P/B ratio remained close to 1.0 during this period, with the exception of the period prior to elapsed time = 50 min that was impacted by data noise due to the low N 2 0 concentrations. • After elapsed time = 70 min, and up until elapsed time =110 min, the perturbation cycle N 2 0 levels were slightly higher than the baseline cycle levels. The headspace N 2 0 profiles, using the GC data, indicate the same trend (data not shown). The small difference in peak N 2 0 levels was unexpected given the consistent ammonia oxidation rates between the cycles. However, the slightly higher N 2 0 levels for the perturbation cycle correspond to the slightly lower perturbation cycle nitrate levels (Figure 4.23). The data suggest that during the perturbation cycle the NOB were not able to compete with the AOB for available nitrite as effectively as during the baseline cycle, allowing the conversion of nitrite to N 2 0 rather than nitrate. Therefore, it appears there were very subtle differences in the overall biochemical environment between the cycles. 113 Figure 4.24: SBR off-gas N 2 0 concentration-, mean slope-, perturbation/baseline ratio and DO concentration-time profdes; perturbation = + 50% wastewater ammonia (SBR-TS 13a) 114 • The N 2 0 mean slope became zero at about elapsed times =110 and 120. min for the baseline and perturbation cycles, respectively. At these times, the ammonia concentrations were about 7 mg N/L (baseline cycle) and 12 mg N/L (perturbation cycle). Subsequently, the off-gas N 2 0 concentrations began to decrease in both cycles with continued ammonia oxidation. For the baseline cycle, the N 2 0 concentration was reduced from 675 ppm (elapsed time =110 min) to 600 ppm (elapsed time = 135 min, ammonia concentration = 1.5 mg N/L). During the perturbation cycle, the N 2 0 concentration was reduced from 740 ppm (elapsed time = 120 min) to 650 ppm (elapsed time = 165 min, ammonia concentration = 2.0 mg N/L). Thus, both cycles experienced an approximate 11% reduction in off-gas N 2 0 levels from the point of their decline until the point were ammonia oxidation was nearing completion. • As with the pH profdes, the DO profdes for the baseline and perturbation cycles were indistinguishable from each other, except for the time required for the occurrence of DO breakthrough. For this scenario, the data show that sufficiently high initial SBR in-situ ammonia levels, for different SBR cycles, induce the same average aerobic-phase ammonia oxidation rate and N 2 0 generation rate. However, during the latter portion of the aerobic-phase of both cycles, there was a reduction in N 2 0 generation rate, manifested in reduced off-gas N 2 0 concentrations. The SBR-TS 13f data suggested that the reduction in N 2 0 generation rate was related to a reduction in ammonia oxidation rate, induced by diminishing ammonia levels, during the progression of the aerobic-phase. The following section further investigates this hypothesis. Reduction in Off-Gas N20 Concentration and Ammonia Oxidation Rate Assessing the relationship between ammonia oxidation rate and reduction in off-gas N 2 0 concentration required a data set where the nitrite levels were essentially constant. This is.due to the influence that the nitrite concentration had on the N 2 0 generation rate under constant aeration and ammonia oxidation rates (refer to Section 4.8 for further discussion on this topic). Furthermore, evaluation of the ammonia oxidation rate before, and after, the time when the off-gas N 2 0 mean slope crossed the "zero point" was most suitably conducted with an ammonia data set that included more than two samples for the time period after the zero point. Based on these criteria the SBR-TS 13a perturbation cycle provided the most comprehensive data set. 115 As previously discussed, the off-gas N 2 0 mean slope value crossed the zero point at elapsed time = 120 min (Figure 4.24). The ammonia oxidation rate was assumed to be constant between elapsed time = 60 min and elapsed time = 120 min, with a linear model fit to the five available data points using least squares regression analysis. Three ammonia data points were available after elapsed time = 120. For simplicity, a linear model was also fit to these data, although the ammonia oxidation rate was likely continuing to slow down during this period. The relatively constant change in the off-gas N 2 0 mean slope between elapsed times = 135 to 165 min (Figure 4.24), corresponding to the period of utilized ammonia data, indicated that the selection of a linear model to represent the ammonia data was reasonable. The calculated ammonia oxidation rate for the period preceding the off-gas N 2 0 mean slope zero point was 14.6 mg N/L. The calculated rate for the period after the zero point was 13.8 mg N/L, resulting in a reduction in the ammonia oxidation rate of approximately 6%. Though neither an exhaustive or comprehensive analysis, the obtained results are consistent with data from other SBR cycles where only two ammonia data points were available after the zero point. The data presented strongly suggest that the reduction in off-gas N 2 0 concentration, and thus a reduction in the N 2 0 generation rate, during the latter portion of the aerobic-phase was the result of a reduction in ammonia oxidation rate with decreasing ammonia concentration. However, an alternate hypothesis for the reduced N 2 0 generation rate was postulated. DO levels were typically observed to increase over time as the aerobic-phase progressed. Therefore, it was hypothesized that the reduction in the N 2 0 generation rate was related to reaching a DO "threshold" concentration where oxygen became less limiting after that point in time for the given cycle, resulting in a reduced requirement for nitrite to act as an electron acceptor and thus less N 2 0 generation. The following section further examines this hypothesis. DO Threshold Concentration? The SBR-TS 12a data, shown in Figure 4.25, were used to investigate the hypothesized phenomenon of the DO threshold concentration due to the large imposed differences in DO concentration between the baseline and perturbation cycles. The differences in cycle DO concentrations resulted from an increase in the perturbation cycle aeration rate, relative to the baseline cycle rate. The baseline cycle off-gas N 2 0 reached its peak concentration of 560 ppm at approximately elapsed time = 110 min. The ammonia concentration at elapsed time = 105 min was 8.9 mgN/L. Subsequently, the 116 4.0 -3.5 -3.0 -2.5 "5) E. 2.0 -O o 1.5 1.0 -0.5 • 0.0 Baseline — • — Perturbation < > < • 20 40 .60 80 100 120 Elapsed Time (min) 140 160 180 30 =T 25 o> 20 E — A—Basel ine — • — Perturbation i -A i 20 40 60 80 100 . 120 Elapsed Time (min) 140 160 180 600 — 500 400 300 q z in S> 200 ar ° 100 I I I i Baseline — • — Perturbation 7 20 60 80 100 120 Elapsed Time (min) 140 160 180 i seline •turbation • D d — • — P e \ \ \ \ r — * - - * — v. \ \ -* i 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 Figure 4.25: SBR DO, ammonia, off-gas N 2 0 and nitrite concentration-time profdes; peturbation = + 96% aeration air flow rate (SBR-TS 12a) 117 off-gas N 2 0 level started to decrease and was 500 ppm at elapsed time = 135 min, an approximate 10% reduction in magnitude. At this point the ammonia concentration was still relatively high, at 2.1 mg N/L. The ammonia was completely oxidized just after elapsed time = 140 min, with the off-gas N 2 0 concentration reduced to 480 ppm by this time. The data indicate that the N 2 0 generation rate was clearly slowing down in the last 30 min prior to complete ammonia oxidation. The DO levels were increasing slightly during the 30 minute period (0.47 mg/L at elapsed time = 110 min to 0.61 mg/L at elapsed time = 140 min), as well as the entire aerobic-phase (although the aeration rate remained unchanged). It should be noted that, during this time, the nitrite levels (Figure 4.25) were essentially constant, eliminating this parameter as a causative agent in the reduced N 2 0 generation rate (refer to Section 4.8 for further discussion on this topic). The aeration rate for the perturbation cycle was set at 2,040 mL/min, a very large perturbation above the 1,040 mL/min rate used for the baseline cycle. Not surprisingly, the higher, aeration rate induced much higher DO levels during the entire aerobic-phase of the perturbation cycle. The generated N 2 0 mass was reduced by about 50% during the perturbation cycle, compared to the baseline cycle, due to the reduction in oxygen limitation that was imposed during the entire duration of the aerobic-phase. Again, this was not an unexpected result, and this phenomenon is discussed at length in Section 4.7. However, what is important in the context of the current hypothesis and discussion is the combination of much higher DO levels combined with a much steeper slope of the DO-time profile during the perturbation cycle, relative to the baseline cycle. These two factors should have magnified the effects of a DO "threshold", if it existed, after which point in time the N 2 0 generation rate would be expected to decrease, in a larger relative amount, with simultaneous reduction in off-gas N 2 0 concentrations. However, the difference in off-gas N 2 0 concentration between the point of maximum concentration (i.e. 245 ppm at elapsed time = 75 min) and when there was still about 2 mg N/L of ammonia (i.e. 210 ppm at elapsed time = 90 min) was about 14%, comparable to the baseline cycle, even though the DO concentration passed through a much wider range of values (1.24 mg/L to 1.84 mg/L for the perturbation cycle versus 0.47 mg/L to 0.61 mg/L for the baseline cycle). Therefore, the data do not suggest an obvious DO threshold concentration, during the course of the aerobic-phase for a given aeration rate and "average" DO concentration, that triggered the marked reduction in N 2 0 generation after that point in time when that DO concentration was reached. Discussion Lipid soluble compounds, such as ammonia, enter bacterial cells by passive diffusion across cell 118 membranes (Bitton, 1994). In the case of AOB, the ammonia has to reach the ammonia monooxygenase (AMO) enzyme for oxidation to hydroxylamine. The AOB enzyme is physically located in the cytoplasmic membrane of the AOB, with a conceptual AOB biochemical model showing the AMO enzyme site (i.e. where ammonia actually binds with AMO) located on the cytoplasm side of the cytoplasmic membrane (Bock and Wagner, 2001). Thus, to reach the AMO enzyme binding site, ammonia must diffuse through the outer cell wall and the cytoplasmic membrane. The diffusion rate across cell membranes depends on the concentration gradient across the membrane (Bitton, 1994). Therefore, as the bulk solution ammonia concentration decreases, the rate of ammonia diffusion into the cell and the vicinity of the AMO binding site slows down. At some bulk solution concentration the ammonia diffusion rate will become the limiting rate in the overall ammonia oxidation process. Unionized ammonia (i.e. dissolved gas, NH3), rather than the ionized form (i.e. NH 4 +), is generally believed to be the true AOB AMO substrate (Bock et al 1991; Bock and Wagner, 2001). Cell membranes are very permeable to unionized ammonia (Bock et al, 1991), and when combined with the small size of AOB (Grady et al, 1999) that enhances the diffusion transport of material into and out of the cell (Koch, 1996), allows ammonia oxidation to proceed at a zero-order rate down to very low bulk solution, unionized ammonia concentrations. Grady et al (1999) reports a range in Nitrosomonas ammonia half-saturation coefficients of 0.06 to 5.6 mg N/L: The two order-of-magnitude range in the half-saturation coefficient likely reflects differences in experimental conditions used to collect such data, along with the influence of pH on the distribution of ammonia between the ionized and unionized forms. The preceding discussion provides the theoretical framework that supports the observed, but relatively subtle, reductions in the average aerobic-phase ammonia oxidation rate when the initial ammonia concentration in the SBR was significantly reduced from normal baseline conditions. Furthermore, the same theory also supports the observed reduction in ammonia oxidation rate near the end of the aerobic-phase as nitrification was nearing completion. Process Monitoring and Control Implications The presented data indicate the potential of off-gas N 2 0 information for identifying changes in the ammonia oxidation rate that were induced by the SBR aerobic-phase ammonia concentration. From a process monitoring and control perspective, this information could be beneficial in two ways: 119 • First, feeding the SBR with wastewater that contained significantly lower ammonia concentrations than the baseline wastewater was coincidental with reduced ammonia oxidation rates. The N 2 0 data show potential for discerning large differences in the in-situ reactor concentration at the start of the aerobic-phase of the cycle. Under the same aeration rate and operating DO concentration, the off-gas N 2 0 data, particularly in the form of the P/B ratio, were able to resolve differences in ammonia oxidation rate that were grossly correlated to the differences in the N 2 0 generation rate. Notably, changes in ammonia oxidation rate between the baseline and perturbation cycles could not be identified using pH or DO data. It was possible that other factors, unrelated to in-situ ammonia levels, contributed to the observed reduction in ammonia oxidation, rate. The most likely factors include aeration rate and DO level. However, the "consistency" in the baseline and perturbation cycle DO data, when the SBR was operated with the same aeration rate during both cycles, suggest that this parameter was unlikely to have affected the ammonia oxidation rate. Similarly, the low variability, in the ammonia oxidation rate between adjacent-day baseline cycles, when the SBR was operated with the same aeration rate, indicate it was very unlikely that differences in aeration rate were the cause for the different ammonia oxidation rates. • Second, the presented data, as well as data from all SBR cycles, illustrated the reduction in off-gas N 2 0 concentrations, and thus a reduction in the N 2 0 generation rate, later in the aerobic-phase as nitrification was reaching completion. These results indicated that the reduced N 2 0 generation rate was related to a reduction in ammonia oxidation rate as the cycle progressed. Therefore, the off-gas N 2 0 data provided "advanced" indication of the timing of complete ammonia oxidation. The possibility that the reduction in N 2 0 generation rate was related to a slowly increasing DO concentration, as the aerobic-phase progressed, was eliminated as a causative mechanism. 4.5.2 Readily Degradable Carbon Load Section 4.2 identified the impact that readily degradable carbon (RDC) had on nitrification during the early stage of the aerobic-phase. The presence of RDC at the start of the aerobic-phase, when treating the baseline synthetic wastewater, was the result of the wastewater containing acetate in excess of anoxic-phase denitrification requirements. Experiment SBR-TS 14a was conducted to investigate the impact of RDC on the initiation of nitrification and the related N 2 0 generation response. 120 Experimental Design The SBR was fed the usual wastewater for the baseline cycle. Alternatively, for the perturbation cycle, the SBR was fed wastewater that contained only 60% of the carbon present in the baseline wastewater. The carbon reduction included proportional reductions in both acetate and yeast extract (note: yeast extract, a slowly degradable substrate, comprised only about 20% of the total wastewater COD concentration). The magnitude of carbon reduction was selected such that complete denitrification would still be achieved during the anoxic-phase. The magnitude of carbon reduction was estimated using TOC, PHB and nitrate data obtained from previous experiments. The SBR was operated with a 1,040 mL/min aeration rate for both cycles. Anoxic-Phase Carbon Requirement and TOC and PHB Profiles Figure 4.26 illustrates the TOC, PHB, nitrite and nitrate profiles for SBR-TS 14a. Anoxic-phase (i.e. elapsed time = 0 to 20 min) TOC levels were much lower for the perturbation cycle, compared to the baseline cycle, reflecting the large reduction in wastewater carbon concentration. The TOC concentration at the end of the perturbation cycle anoxic-phase was 5 mg/L, the minimum level typically obtained during the remainder of a cycle, compared to 18 mg/L at the same point in the baseline cycle. However, sufficient carbon was available during the perturbation cycle anoxic-phase to ensure complete denitrification, as both the nitrite and nitrate levels at elapsed time = 19 min were essentially zero. As expected, the aerobic-phase perturbation cycle PHB concentrations (Figure 4.26) were notably lower than those of the baseline cycle due to the reduced availability, during the anoxic-phase, of carbon in excess of denitrification requirements. Overall Nitrification Performance and N20 Generation Subsequent to the onset of nitrification, both cycles demonstrated essentially identical nitrification rates and N 2 0 response. Linear regression analyses performed on the ammonia data (Figure 4.27) found that the baseline and perturbation cycle average ammonia oxidation rates were virtually identical at 12.6 and 12.7 mg N/L/hr, respectively. Similarly, the total generated N 2 0 mass for both cycles was approximately 44 mg, after adjusting the slightly lower perturbation cycle anoxic-phase ammonia concentrations to reflect those of the baseline cycle (i.e. equal ammonia loads at start of aerobic-phase). SBR DO (Figure 4.27) and nitrite (Figure 4.26) levels, after the start of nitrification, were also essentially the same for 121 Figure 4.26: SBR TOC, PHB, nitrite and nitrate concentration-time profdes; perturbation = - 40% wastewater carbon (SBR-TS 14a) 122 2. 3" , oi • £ 1-O 0. ° 0 . 0. 0. 0. i i Baseline — • — Perturbation • 4 9.0 8.5 8.0 1.7.5 7.0 6.5 6.0 450 400 E 350 £ 300 O 250 « 200 S) 150 o 1°° 50 0 30 j 25 z o> 20 E i £ 1 5 z + 10 I Z 5 0 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 I I I Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 \ Baseline — • — Perturbation \ \ _^ X 20 60 80 100 120 Elapsed Time (min) 140 160 180 A . k n seline » r t n r h a t i n n K De -* i 20 40 60 80 100 120 Elapsed Time (min) 140 180 Figure 4.27: SBR DO, pH, off-gas N 2 0 and ammonia concentration-time profiles; perturbation = - 40% wastewater carbon (SBR-TS 14a) 123 both cycles and confirm the observed consistency in nitrification performance between the two cycles. Onset of Nitrification The biggest difference in SBR response due to the wastewater composition, manifested in the absence of RDC at the start of the perturbation cycle aerobic-phase, was related to the timing of the onset of nitrification. Careful examination of the ammonia profiles (Figure 4.27), and the generated linear ammonia models, showed that ammonia oxidation was proceeding at its maximum rate within about 2 minutes of the start of the aerobic-phase during the perturbation cycle. However, this same situation was delayed until about 15 to 20 minutes into the aerobic-phase of the baseline cycle. The presence of almost 1 mg N/L of nitrite at elapsed time = 30 min of the perturbation cycle (Figure 4.26), when the baseline cycle concentration was basically zero at the, same time, further corroborates the earlier onset of nitrification during the perturbation cycle. ' * ' The nitrification observations coincide with the SBR TOC levels (Figure 4.26). The results indicate that, by the end of the perturbation cycle anoxic-phase (i.e. elapsed time = 19 min), the RDC, as measured using the TOC parameter, was reduced to essentially its lowest concentration. Alternately, this TOC level was not reached until about 10 minutes (i.e. elapse time = 30 min) into the aerobic-phase of the baseline cycle. The data suggest that the heterotrophic organisms were out-competing the nitrifying organisms for oxygen during the initial period of very high carbon utilization rates, precluding the start of nitrification. N20 Response The off-gas N 2 0 profiles (Figure 4.27) dramatically illustrate the lag in onset of nitrification in the baseline cycle relative to the perturbation cycle. Twenty minutes after the start of the perturbation cycle aerobic-phase (i.e. elapsed time = 40 min) the off-gas N 2 0 concentration was already 80 ppm, compared to essentially zero for the baseline cycle. Alternately, there were small differences in DO levels during the first ten minutes of the aerobic-phase (i.e. elapsed time = 20 to 30 min). Here, the average perturbation cycle DO concentration was 0.27 mg/L, compared to 0.15 mg/L for the baseline cycle. The reduced baseline cycle DO concentration was indicative of the high carbonaceous oxygen uptake rate (OUR). Beyond elapsed time = 35 min, the DO levels for both cycles were basically identical. The delay in the onset of nitrification was reflected in the DO profiles in a much more subtle manner, compared to the N 2 0 profiles. 124 The pH data shown in Figure 4.27 indicate that the perturbation cycle had notably lower pH levels during both the anoxic-phase and the early part of the aerobic-phase. This pH difference was surprising given that both cycles achieved the same amount of anoxic-phase denitrification and thus alkalinity formation, but could have resulted from problems with the electrical signal generated by the pH probe, resulting in inaccurate readings. Regardless of the cause of the pH difference, pH does influence the nitrite-nitrous acid equilibrium. The nitrous acid fraction increases with decreasing pH levels. Therefore, if nitrous acid, rather than nitrite proper, was the actual electron acceptor for autotrophic reduction of "nitrite" to N 20, the lower pH of the perturbation cycle may have also impacted N 2 0 generation along with the earlier onset of nitrification. Section 4.8 describes the experiments that were conducted to specifically investigate this issue. Both cycles shared very similar N 2 0 concentration profiles, as well as N 2 0 mean slope and P/B ratio profiles (Figure 4.28), subsequent to the onset of nitrification, offset in time by an interval approximately equal to the nitrification offset. This observation was consistent with the previously discussed similarity in nitrification kinetics. Discussion The baseline cycle RDC utilization rate, calculated using the TOC data for samples collected at elapsed times = 19 and 30 min, was at least 71 mg C/L/hr. As will be shown in Section 4.5.3, an extended-duration slowly degradable carbon utilization rate of only 38 mg C/L/hr resulted in a 95% reduction in the ammonia oxidation rate when the SBR was operated with the same aeration rate and similar DO levels. As a result, the characteristic delay in the onset of aerobic-phase nitrification, when treating the baseline wastewater, was clearly induced by the carry-over of RDC in excess of anoxic-phase denitrification requirements. Heterotrophic competition for limited oxygen, under the high RDC utilization rate, precluded the onset of nitrification until the RDC was consumed. The larger oxygen half-saturation coefficient of heterotrophs, relative to the autotrophic AOB, provides the former group of organisms, with a competitive advantage in an oxygen-limited environment (Grady et al, 1999). Following heterotrophic RDC utilization, nitrification then proceeded at a rate that was unaffected by the earlier RDC utilization "event". Section 4.5.3 provides further discussion regarding the competition for oxygen between heterotrophs and autotrophs in oxyen-limited environments. 125 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 10 0 20 4 0 60 80 100 120 140 160 180 Elapsed Time (min) Figure 4.28: SBR off-gas N 2 0 concentration-, mean slope- and perturbation/baseline ratio-time profiles; perturbation = - 40% wastewater carbon (SBR-TS 14a) 126 Process Monitoring and Control Implications From a process monitoring perspective, the completion of RDC oxidation and initiation of nitrification manifested itself in a resolvable, but small, increase in DO concentration. Alternatively, the N 2 0 generation response, provided by the off-gas N 2 0 profiles, demonstrated a significantly increased sensitivity to the onset of nitrification when compared to the DO data. 4.5.3 Slowly Degradable Carbon Utilization Rate As discussed in Sections 4.2 and 4.5.2, heterotrophic utilization of soluble readily degradable carbon (RDC) was limited to the very early stage of the aerobic-phase, prior to the onset of nitrification, during SBR treatment of the baseline wastewater. Thus, there was essentially no oxygen demand exerted by heterotrophic organisms, for RDC utilization, that occurred simultaneous with nitrification. Heterotrophic utilization of internally stored carbon (i.e. PHB), as well as adsorbed/enmeshed slowly degradable carbon (SDC) originating from the wastewater (i.e. yeast extract), did proceed simultaneous with nitrification. However, as the data presented in Section 4.2 indicate, the ammonia oxidation rate was not observed to increase with the marked slowing down of PHB oxidation. Thus the heterotrophic oxygen demand related to utilization of these carbon compounds appeared to insignificantly impact ammonia oxidation, when considered in the context of the total biomass oxygen demand. "Real" wastewater often contains a significant fraction of complex carbon compounds, with the term generally applied to particulate and high molecular weight organic matter (Grady et al, 1999). The carbon in these compounds is not biologically available to microorganisms-until the compounds are broken down into smaller and simpler forms by extracellular enzymatic mechanisms, collectively referred to as hydrolysis (Grady et al, 1999). As a result, the rate of complex carbon hydrolysis controls the rate of biologically available carbon utilization and, therefore, the resulting oxygen demand associated with its use. Given the need for the preliminary hydrolysis step, these complex carbon compounds are generally defined as slowly degradable carbon (SDC). This section presents the experimental results where the SBR was subjected to varying rates of aerobic-phase SDC utilization and thus oxygen demand, beyond the typical PHB and yeast extract carbon utilization and oxygen demand rates provided by treatment of the baseline wastewater. This operating condition induced a significant oxygen competition between heterotrophic and autotrophic organisms, and provided information on SBR nitrification performance and N 2 0 generation response. 127 Experimental Design The SDC utilization rate experiments were conducted using two back-to-back SBR cycles. The SBR was operated using the same aerobic-phase aeration rate (1,040 mL/min) for both the baseline and perturbation cycles. The SBR was fed the regular synthetic wastewater for the baseline cycle. The SBR also received the regular baseline synthetic wastewater for the perturbation cycle. However, the oxygen demand associated with an "additional" slowly degradable carbon wastewater component was simulated by pumping a concentrated sodium acetate solution to the SBR during the aerobic-phase of the perturbation cycle. The sodium acetate solution was added to the SBR at a very low flow rate (i.e. 1 mL/min) to minimize reactor dilution effects. The carbon loading rate, and resulting "SDC" utilization rate and oxygen uptake rate (OUR), was controlled by adjusting the rate of sodium acetate addition by changing the solution concentration. This approach allowed the study of SBR response to complex carbon substrates with varying rates of degradation, while eliminating the disadvantages of working with more complex substrates; this would include the need to shift the biomass population in response to the new wastewater and the resulting time for the shift to occur. The carbon solutions were pumped to the SBR for 120 minutes during the aerobic-phase. Carbon addition began at approximately elapsed time = 30 min (i.e. 10 minutes after the start of the aerobic-phase), the point when residual anoxic-phase RDC had been consumed and close to the typical time of the onset of nitrification. The SDC utilization rate experiments were conducted using three different carbon loading rates, expressed below in terms of carbon (i.e. C) and COD loading: • low rate = 9.5 mg C/L/hr = 25 mg COD/L/hr • medium rate = 19 mg C/L/hr = 50 mg COD/L/hr • high rate = 3 8 mg C/L/hr = 100 mg COD/L/hr The range of carbon loading rates were selected such that nitrification would proceed to some extent when the SBR was operated with the same aeration rate for both the perturbation and baseline cycles. The following section presents the data and methods used to estimate the induced OUR provided by the 128 various carbon loading rates. Slowly Degradable Carbon OUR Estimates Two methods were used to estimate the induced carbonaceous OUR provided by the SDC utilization rates. The first method used a theoretical, stoichiometric approach. The second method used experimental data, and provided confirmation of the theoretical estimates. • The first method utilized the half-reaction, stoichiometric approach described in Grady et al (1999). The true growth yield (Y) was calculated using acetate as the electron donor, oxygen as the electron acceptor, ammonia as the nitrogen source, and assuming a cellular composition of C 5H 7N02. The calculated Y was 0.70 mg COD of biomass / mg COD acetate. The fraction of the electron donor (i.e. acetate) used for energy production (i.e. becomes carbon dioxide) equaled 1 - Y = 0.30 g COD / g COD acetate. Therefore, the carbonaceous OUR induced by a given SDC utilization, rate equaled the utilization rate, expressed in terms of mg COD/L/hr, multiplied by 0.30. • The second method used to estimate the Y and carbonaceous OUR values utilized experimental data, as described below. Duplicate, in-situ OUR experiments were conducted at each aerobic-phase SDC utilization rate. These experiments were conducted at the end of several regular SBR cycles following complete nitrification and DO breakthrough, and involved two steps. First, the DO was allowed to rise to about 6 mg/L, then the aeration system was turned off and the rate of DO depletion was recorded as the concentration dropped to about 1 mg/L. These data were used to estimate the endogenous OUR. Then, the aeration system was turned on again to allow the DO level to rise to 6 mg/L, the aeration system was turned off, and the carbon pumping system turned on. The rate of DO depletion was recorded as the concentration fell to about 1 mg/L. These data were used to estimate the [carbonaceous + endogenous] OUR. The OUR data were found to be highly linear; therefore, linear models were fit to the data using least squares regression analysis. The difference between the [carbonaceous + endogenous] OUR and 129 endogenous OUR value was the carbonaceous OUR, due to the applied carbon load. Figure 4.29 illustrates the calculated carbonaceous OUR values as a function of aerobic-phase SDC utilization rate. The slope of the line, 0.33 mg 0 2 / mg COD, represents the fraction of supplied carbon used for energy production. Therefore, Y = 1 - 0.33 = 0.67 mg COD / mg COD. However, during the OUR experiments only nitrate, as opposed to ammonia, was available for cell synthesis, requiring the cells to expend energy to reduce the nitrogen from a +V to -III state prior to assimilation (Grady et al, 1999). This situation reduces the fraction of carbon used for cell synthesis. Under these conditions, the calculated Y, again obtained using the half-reaction approach, was 0.57 mg COD / mg COD. The calculated Y was lower than the Y value obtained from the OUR data (i.e. 0.67). Since the OUR experiments, were.conducted at the end of the-SBR cycle the biomass PHB levels were likely depleted. As a result, some of the supplied acetate was "probably stored as PHB during the OUR experiments, as suggested by aerobic storage data of Dircks et al (2001) and Carucci et al (2001). Although the COD / C ratios for'acetate, PHB and cell mass.(i.e. C 5H 7N0 2) are all approximately 3 mg COD / mg C, if the energy required to store acetate as PHB is lower than the energy required to synthesize new biomass, the actual yield would be expected to be somewhat higher than the theoretical yield. Thus, when allowing for the PHB storage phenomenon, the experimental yield estimate provides confirmation of the calculated, theoretical yield estimate. Therefore, the calculated theoretical value of 0.30 mg COD / mg COD, representing the fraction of supplied carbon used for energy production, provided a suitable estimate of the carbonaceous OUR. The estimated SDC OUR values for the applied carbon loading rates are shown below: • low rate = 25 mg COD/L/hr utilization rate x 0.30 mg 0 2 / mg COD = 7.5 mg 02/L/hr • medium rate = 50 mg COD/L/hr utilization rate x 0.30 mg 0 2 / mg COD = 15 mg 02/L/hr • high rate = 100 mg COD/L/hr utilization rate x 0.30 mg 0 2 / mg COD = 30 mg 02/L/hr Ammonia Assimilation and Estimation of Ammonia Oxidation Rate The aerobic-phase carbon loading provided an additional growth opportunity for heterotrophic organisms. This growth would induce the assimilation of ammonia into newly synthesized biomass, 130 0 25 50 75 100 125 Aerobic-Phase Carbon Loading Rate (mg COD / L' hr) Figure 4.29: Carbonaceous O U R versus aerobic-phase carbon loading rate 131 providing an additional aerobic-phase ammonia removal mechanism. As a result, the perturbation cycle ammonia oxidation rate calculation had to account for assimilation when using the ammonia data to estimate the oxidation rate. The ammonia nitrogen requirements for cell synthesis were calculated using the previously discussed theoretical, stoichiometric half-reaction method. The resulting ammonia assimilation rate estimates, as a function of aerobic-phase carbon loading rate, expressed in terms of C rather than COD, are shown below: • low rate = 9.5 mg C/L/hr utilization rate x 0.165 g N / g C = 1.6 mg N/L/hr • medium rate = 19 mg C/L/hr utilization rate x 0.165 gN / g C = 3.2 mg N/L/hr • high rate = 38 mg C/L/hr utilization rate x 0.165 g-N / g C = 6.4 mg N/L/hr The aerobic-phase ammonia oxidation rate data presented in the following sections represent the "average" rate for the cycle of interest. The average rate was estimated from the ammonia data using least squares regression analysis to fit linear models to the ammonia data. As discussed in Section 4.2, for the baseline cycles, the ammonia removal rate was assumed equal to the ammonia oxidation rate. However, for the perturbation cycles subjected to SDC utilization, the ammonia oxidation rate was defined as the ammonia removal rate (i.e. determined using regression analysis) minus the assumed ammonia assimilation rate. Ammonia Oxidation Figure 4.30 contains the baseline and perturbation cycle ammonia profiles for the three SDC utilization rate experiments. The data illustrate the significant reduction in ammonia removal rate with increasing SDC utilization rate. However, once ammonia assimilation was taken into account, the impact of SDC utilization on ammonia oxidation became even more dramatic. Table 4.3 summarizes the average aerobic-phase ammonia oxidation rates for the three SDC utilization rate experiments. The low, medium and high SDC utilization rates caused a reduction in the perturbation cycle ammonia oxidation rates, relative to the baseline cycle rates, of 14%, 51% and 95%, respectively. Figure 4.31 graphically illustrates the effect of SDC utilization rate on the ammonia oxidation rate in terms of oxygen uptake rates (OUR). The y-axis represents the ratio in the ammonia oxidation rate (AOR) for an SBR cycle, with a given SDC utilization rate, relative to the baseline cycle AOR where 132 ? 2 5 z 0 20 E. x* 1 5 z + 10 1 5 0 1 1 —A—Baseline •—3 ^ 1 —•— F 3erturba tion 20 40 60 80 100 120 Elapsed Time (min) 160 180 a) Low S D C utilization rate ( S B R - T S 1 6 a ) 30 1 z i z 15 10 5 20 40 I I —A—Baseline CI LUI Uet HUM 60 80 100 120 Elapsed Time (min) 140 160 180 b) Medium S D C utilization rate (SBR-TS16d ) 60 80 100 120 Elapsed Time (min) c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.30: SBR ammonia concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 133 Table 4.3: Average ammonia oxidation rate and relative statistics for wastewater slowly degradable carbon utilization rate experiments Experiment Cycle Aerobic-Phase Carbon Loading Average Ammonia Oxidation Rate 2 Reactor Specific (mg N/L/hr) (mg N/L/hr per g M L S S / L ) Relative Statistics 1 A S B R - T S 1 6 a baseline perturbation low 13.4 6.52 11.5 5.60 -14.2% S B R - T S 1 6 d basel ine perturbation medium 13.3 6.19 6.5 3.03 -51 .1% S B R - T S 1 6 c baseline perturbation high 13.7 6.09 0.7 0.31 -94.9% Notes: 1. A = (perturbation - baseline) / basel ine x 100% for specif ic ammonia oxidation rate 2 Perburbation cycle rates adjusted to account for ammonia assimilation: low carbon rate =' • 1.6: mg N/L/hr medium carbon rate = M .. 3.2; mg N/L/hr high carbon rate = L _ J m 9 N/L/hr 134 Figure 4.31: The ratio in the ammonia oxidation rate (AOR) relative to the baseline cycle AOR versus the ratio of the induced carbonaceous OUR relative to the baseline cycle ammonia oxidation OUR (AO OUR) 135 there was no external carbon loading. The utilized baseline cycle AOR was the mean of the baseline AOR values shown in Table 4.3, an appropriate approach given the small variation (i.e. 6%) in specific AOR rates for the experiments caused partly by slightly different MLSS concentrations. The x-axis represents the ratio of the carbonaceous OUR induced by a given SDC utilization rate relative to the baseline cycle ammonia oxidation OUR (AO OUR). The AO OUR values were calculated by multiplying the ammonia oxidation rates by the stoichiometric oxygen requirement for ammonia oxidation (i.e. 3.3 mg O2 / mg N; Grady et al, 1999). This calculation method does not account for the oxygen "saving" where nitrite, rather than oxygen, serves as the terminal electron acceptor for a portion of the oxidized ammonia-hydroxylamihe. However, even if 30% of the original ammonia mass in the reactor was converted to N 2 0 via nitrite reduction, the resulting reduction in consumed oxygen would only be approximately 8%, based on the stoichiometry for the various reactions. Therefore, for the purposes of this discussion, the exclusion was deemed insignificant. The Figure 4.31 data illustrate the level of competition for oxygen between the heterotrophic bacteria and the AOB. For example, when the SDC utilization rate induced a carbonaceous OUR that was 50% of the normal, baseline AO OUR, the ammonia oxidation rate was reduced to only 32% of the baseline ammonia oxidation rate. Thus, the reduction in ammonia oxidation rate was not directly proportional to the change in the ratio of the OUR values, indicating the competitive advantage of the heterotrophs in scavenging available oxygen. Carbon Utilization The TOC and PHB data provide insight into biomass exogenous carbon utilization and how this mechanism could impact the assumed ammonia assimilation rates and resulting estimated ammonia oxidation rates. In all three experiments, the applied carbon loading did not increase the soluble TOC concentrations above the levels normally present during the baseline cycles (Figure 4.32). Therefore, the biomass immediately utilized the supplied carbon at a rate that equaled the supply rate. The baseline and perturbation cycles shared similar PHB profiles for the low SDC utilization rate (Figure 4.33a) experiment. (Note: The initial difference in PHB concentrations between the two cycles of the low carbon loading rate experiment (Figure 4.33a) was the result of a reduced amount of carbon (i.e. 136 1 1 1 —•—Baseline — • — Perturbation 1 \ \ \ ^ •"•—^ "1 • — 0 20 40 ' . 60 80 100 120 140 160 180 , ; \ Elapsed Time (min) a) Low S D C utilization rate ( S B R - T S 1 6 a ) 16 • 14 • — 12 • O 10 -i seline rturbation * , m O e •—tc —&— 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) Medium S D C utilization rate ( S B R - T S 1 6 d ) i A r baseline Perturbation \ —•—f 1 \ \ \ A 1 Y * ^ k —• 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.32: SBR TOC concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments. 137 35 30 - B a s e l i n e - P e r t u r b a t i o n 60 80 100 -120 Elapsed Time (min) 140 160 180 a) Low SDC utilization rate (SBR-TS16a) 35 30 j 25 o 20 E CQ IE 10 5 0 i i i — A — B a s e l i n e • P e r t u r b a t i o n 1 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) Medium SDC utilization rate (SBR-TS16d) 35 30 d 2 5 ro20 E CD a. 10 5 0 I i e ) a t i o n —m*— Baselir tm pprti ir w r c i IUI -• < 20 40 60 80 100 120 140 160 Elapsed Time (min) 180 c) High SDGutilization rate (SBR-TS16c). *>'.., Figure 4.33: SBR PHB concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 138 acetate) in the perturbation cycle wastewater. The carbon removed from the wastewater was subsequently added to the SBR during the aerobic-phase of the perturbation cycle. This procedure was unique to this particular experiment, and a one-time deviation from the method described earlier under "Experimental Design"). For this experiment, the biomass did not appear to be storing the added carbon as PHB, nor did the carbon loading induce an obviously reduced rate of PHB utilization. The medium SDC utilization rate experiment (Figure 4.33b) appeared to induce some additional PHB storage during the period from elapsed time = 45 min to 90 min. However, for the high SDC utilization rate experiment (Figure 4.33c) the PHB levels remained notably elevated during the entire carbon loading period (i.e. elapsed time = 30 min to 150 min) of the perturbation cycle when compared to the same period of the baseline cycle. This observation suggests that some of the added carbon was stored as PHB, rather than being used to synthesize new biomass. In addition, the perturbation cycle displayed a reduced rate of PHB consumption when compared to the baseline cycle. This phenomenon indicates the biomass was preferentially utilizing the added acetate, over PHB, for "normal" cell synthesis and/or maintenance energy generation. These processes could impact the ammonia assimilation rate. In particular, biomass utilization of the acetate for cell maintenance energy requirements would reduce the yield and thus reduce the amount of ammonia assimilation, resulting in an underestimation of the perturbation cycle ammonia oxidation rate, given the calculation method. The uncertainty in the significance of the described mechanisms precluded assessment of their impact on the estimated ammonia oxidation rate. Therefore, the ammonia oxidation rate estimation procedure was not altered from its original design. Nitrite and Nitrate Levels Figures 4.34 and 4.35 illustrate the nitrite and nitrate levels, respectively, for the low, medium and high SDC utilization rate experiments. The baseline and perturbation cycle nitrite (Figure 4.34a) profiles for the low SDC utilization rate experiment were very similar to one another, as were the nitrate (Figure 4.35a) profiles. The similarity in the baseline and perturbation profiles becomes even more apparent if they are shifted in time such that the time of the onset of nitrification was the same for both cycles. Alternately, the nitrite and nitrate profiles were dramatically different for the baseline and perturbation cycles for the medium (Figure 4.34b, Figure 4.35b) and high (Figure 4.34c, Figure 4.35c) SDC utilization rate experiments. The perturbation cycle, for the high SDC utilization rate experiment, resulted in essentially no measurable nitrate concentrations in the SBR. This observation was consistent with the 139 Figure 4.34: SBR nitrite concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 140 16 14 _ 12 _i z 10 ut E 8 5 6 z 4 2 I A — —*—Baseline fy — • — Pe irturbati // -*—i I—*-= 0 20 . 40 60 80 100 120 140 160 180 Elapsed Time (min) a) Low S D C utilization rate ( S B R - T S 1 6 a ) 16 14 12 10 o z l £ seline lurbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) Medium S D C utilization rate (SBR-TS16d ) 16 14 12 • z 10 8 -E. O 6 z 4 -2 0 -I I —A—Baseline —•—Perturbation —•— 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.35: SBR nitrate concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 141 assumed ammonia assimilation rate and the estimated reduction in ammonia oxidation rate for the high SDC utilization rate experiment (Table 4.3). The potential existed for heterotrophic biochemical reduction (i.e. denitrification) of generated nitrite and nitrate during the SDC utilization experiments given the low DO levels in the reactor and the availability of carbon. Section 4.6 presents the results of experiments that were specifically conducted to investigate this issue. SBR Response with Respect to DO and pH Profiles Before examining the off-gas N 2 0 data with respect to SDC utilization rate, it is worthwhile to first evaluate the information provided in the SBR DO and pH profiles. Figures 4.36 and 4.37 contain the DO and pH profiles, respectively, for the three SDC utilization rate experiments. Key observations include: • For the low SDC utilization rate experiment, the DO (Figure 4.36a) and pH (Figure 4.37a) baseline and perturbation cycle profiles were essentially identical to each other after the onset of nitrification in each cycle. The earlier onset of nitrification during the perturbation cycle was the result of little RDC at the end of the anoxic-phase (Figure 4.32), due to the previously noted procedure for this particular experiment. The resulting "shift" in the onset of nitrification, by approximately 15 minutes based on the DO and pH profiles, was consistent with previously described findings (Sections 4.2 and 4.5.2). • The DO profiles for the medium (Figure 4.36b) and high (Figure 4.36c) SDC utilization rate experiments highlight several interesting points. First, between elapsed times = 40 to 140 min, the mean DO concentrations for the baseline cycles (i.e. 0.27 mg/L and 0.29 mg/L) were more than 50% larger than the mean concentrations for the perturbation cycles (i.e. 0.18 mg/L and 0.16 mg/L). The differences in the mean values were statistically significant at a 5%> level of significance. Second, for both experiments, the DO profiles for the perturbation cycles were notably flatter than the baseline cycle profile slopes. The DO information suggests that the perturbation cycles induced slightly higher total OURs than the OURs of the baseline cycles. However, for both SDC utilization experiments, the total OUR of the perturbation cycles was estimated to be lower than the baseline cycle OUR. Consider the medium SDC utilization rate experiment (SBR-TS 16d) as an example. In this case, the estimated baseline 142 2.0 -1.8 • 1.6 • 1.4 -_J O) 1.2 -E. 1.0 • O 0.8 -o 0.6 0.4 -0.2 0.0 --i seline Ba > — • — Perturbation • > 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) Low S D C utilization rate ( S B R - T S 1 6 a ) 2.0 • 1 Q I .O 1.6 • 1.4 -_J 1.2 • E, 1.0 • O 0.8 • Q 0.6 -0.4 • 0.2 • 0.0 -< I I Baseline — • — Perturbation -20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) Medium S D C utilization rate (SBR-TS16d ) 2.0 • 1.8 -1.6 • 1.4 • _1 OJ 1.2 • E_ 1.0 -O 0.8 -Q 0.6 • 0.4 • 0.2 -0.0 -seline Ba — • — Perturbation • 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.36: SBR DO concentration-time profiles for a) low carbon loading rate (SBR-TS 16a), b) medium carbon loading rate (SBR-TS 16d) and c) high SDC (SBR-TS 16c) SDC utilization rate experiments 143 Figure 4.37: SBR pH-time profiles for a) low carbon loading rate (SBR-TS 16a), b) medium carbon loading rate (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 144 cycle ammonia oxidation OUR was 44 mg/L/hr. The perturbation cycle ammonia oxidation and carbonaceous OUR values were estimated to be 21 and 15 mg/L/hr, respectively, for a total OUR of 36 mg/L/hr. The analysis assumes both cycles had similar PHB and endogenous OUR values, and even neglects the nitrite oxidation OUR component of the total OUR. Therefore, the data indicate that DO information, alone, could not be used to assess the impact of carbon loading on the nitrification rate for the given experimental conditions. • The pH profiles for the medium (Figure 4.37b) and high (Figure 4.37c) SDC utilization rate experiments more obviously suggest the negative impact that the carbon loading rate had on the nitrification rate. However, the increased carbon dioxide generation rate with increased carbon loading rate would reduce the pH, due to the production of carbonic acid. This mechanism, the pH buffering provided by the carbonate system, and the logarithmic nature of pH measurement makes it difficult to interpret the pH data in the context of the nitrification rate. Again, consider the medium SDC utilization rate experiment. The baseline cycle ammonia oxidation rate was about 50% less than the baseline cycle rate, yet the relative differences in the cycle pH values were nowhere as large. SBR Response with Respect to Off-Gas N20 Information The SBR off-gas N 2 0 concentration-time profiles are shown in Figure 4.38 for all three SDC utilization rate experiments. As shown in Figure 4.38, the baseline cycle N 2 0 profiles for all three experiments were very similar in shape and had comparable peak concentrations. Therefore, these data indicate similar N 2 0 generation rates for the three baseline cycles. Furthermore, given the limited variability in the baseline cycle ammonia oxidation rates (refer to Table 4.3), the data indicate comparable N 2 0 generation / oxidized ammonia mass ratios for all three baseline cycles. Thus the biomass was functionally stable with respect to N 2 0 generation over the period of time that the three experiments were conducted. The perturbation cycle data for the low SDC utilization rate experiment (Figure 4.38a) were shifted ahead by 15 minutes, as previously discussed, to enable a more consistent comparison of trends with the other experiments. Figures 4.39 and 4.40 contain the off-gas N 2 0 mean slope and P/B profiles, respectively, for all three SDC utilization rate experiments. The off-gas N 2 0 data, whether in its raw (Figure 4.38) or manipulated forms (Figures 4.39 and 4.40), 145 700 _ 600 a 5 0 0 O 400 Z in 300 200 100 0 3= o Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) Low S D C utilization rate ( S B R - T S 1 6 a ) . Perturbation cycle data shifted ahead 15 min. 700 _ 600 B 5 0 0 O 400 z in 300 200 100 0 o O Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) b) Medium S D C utilization rate ( S B R - T S 1 6 d ) 700 _ 600 E Q. O 400 z in 300 200 100 0 I I \ Baseline • Perturbation \ v \_j y 1 L , / 20 40 60 80 100 120 Elapsed Time (min) 180 c) High S D C utilization rate (SBR-TS16C) Figure 4.38: SBR off-gas N 2 0 concentration-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 146 20 -o Q. 16 • O 12 CO 8 • c (Q "c 4 -a E 0 -o E -4 -z (A dd) -8 --12 • Ff-Ga: -16 --20 O -24 --28 -I Baselin Perturb —•— ation I ' 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) Low S D C utilization rate ( S B R - T S 1 6 a ) . Perturbation cycle data shifted ahead 15 min. o E 20 16 12 8 1 2 c -4 •S-12 -16 -20 -24 -28 -Baseline -Perturbation 60 80 100 120 Elapsed Time (min) 140 160 180 b) Medium S D C utilization rate (SBR-TS16d ) 20 g. 16 o 12 to 8 S 4 „ ._ -O E 0 c -4 -16 -20 O it O -24 -28 I / — / / J —^ / hi ma -...\ f I 1 baseline —•— Perturbation \J \ \ 20 60 80 100 120 140 160 180 Elapsed Time (min) c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.39: SBR off-gas N 2 0 mean slope-time profiles for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 147 0) c — o «) n If ¥ o 1.0 0.8 0.6 0.4 0.2 0.0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) Low S D C utilization rate ( S B R - T S 1 6 a ) . Perturbation cycle data shifted ahead 15 min. o C — O © -.=. in ra ii J2 (3 « o 0.4 0.2 0.0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) Medium S D C utilization rate ( S B R - T S 1 6 d ) 1 •£ 0.8 •w ra 5 O 0.6 =. 8 0.4 | O 0.2 0.0 I 60 80 100 120 Elapsed Time (min) 140 160 180 c) High S D C utilization rate ( S B R - T S 1 6 c ) Figure 4.40: SBR perturbation/baseline off-gas N 2 0 ratio-time profile for a) low (SBR-TS 16a), b) medium (SBR-TS 16d) and c) high (SBR-TS 16c) SDC utilization rate experiments 148 clearly indicate that N 2 0 generation, as manifested in the off-gas N 2 0 levels, was significantly impacted by an increasing SDC utilization rate. Figure 4.41 shows the cumulative stripped N 2 0 mass profiles (refer to Section 4.3.3 for the calculation procedure) for the low (Figure 4.41a) and medium (Figure 4.41b) SDC utilization rate experiments. The high SDC utilization rate experiment was excluded from Figure 4.41 due to the lack of measurable N 2 0 during the perturbation cycle. Key points include: • Linear models were fit, using least squares regression analysis, to the baseline and perturbation cycle data for the low SDC utilization rate experiment, using the data between elapsed time =100 min and 150 min (i.e.. linear range of data). The perturbation cycle N 2 0 mass stripping rate (0.68 mg N/min) was 12% lower than the baseline cycle rate (0.77 mg N/min). On the assumption that the N 2 0 mass stripping rate was a suitable surrogate parameter for the N 2 0 generation rate (Section 4.3.3), the data indicate that the 14% reduction in ammonia oxidation rate (Table 4.3) induced by the low SDC utilization rate was fairly well correlated to the 12% reduction in the N 2 0 generation rate. • Models were also fit to the linear range of the medium SDC utilization rate data, between elapsed time 100 min and 150 min for the baseline cycle, and elapsed time =120 min and 150 min for the perturbation cycle. The medium SDC utilization rate induced a 75% reduction in the estimated N 2 0 generation rate (i.e. 0.17 mg N/min versus 0.69 mg N/min), compared to a 51% reduction in the ammonia oxidation rate (Table 4.3). Under the higher SDC utilization rate, the change in N 2 0 generation rate was again fairly well correlated with the change in ammonia oxidation rate. However, the evidently larger impact on the N 2 0 generation rate, compared to the ammonia oxidation rate, suggested another mechanism was possibly impacting N 2 0 generation. Discussion In the more distant past (e.g. > 25 years ago), there was the belief that nitrification could not be maintained in the presence of high carbon loadings, and that perhaps the organic compounds had a "general toxicity" to the nitrifying bacteria (Grady et al, 1999). Eventually, however, studies (e.g. Hockenbury et al, 1977) conducted using "real" wastewater and biomass from full-scale municipal treatment facilities found that ammonia and nitrite oxidation rates were not negatively impacted in the presence of carbon oxidation, as long as suitably high DO levels were maintained in the bioreactor. 149 Figure 4.41: SBR cumulative stripped N 2 0 mass-time profdes for a) low (SBR-TS 16a) and b) medium (SBR-TS 16d) SDC utilization rate experiments 150 Alternately, the data presented clearly illustrated the sensitivity of the ammonia oxidation rate to the SDC utilization rate when the SBR was operated in an oxygen-limited, low DO state under a constant aeration rate. The SDC utilization rates were not excessive when viewed in the context of the estimated induced carbonaceous OUR and the baseline cycle ammonia oxidation OUR. For example, the high SDC utilization rate provided a carbonaceous OUR that was estimated to be only about 70% of the baseline cycle ammonia oxidation OUR (Figure 4.31). However, this same SDC utilization rate essentially shut down nitrification for the given oxygen supply and bulk solution DO levels. As noted in Section 4.5.2, the larger oxygen half-saturation coefficient (K0) for heterotrophic organisms, relative to the autotrophic AOB, provides the former group of organisms with a competitive advantage in an oxygen-limited environment. The differences in the K 0 values reflect the difference in the overall diffusional resistance in the transport of oxygen from the bulk solution, through the floe structure, to the individual cells and into the enzyme sites within the cells (Mahendraker, 2003). Grady et al (1999) reported typical K 0 values for heterotrophic and autotrophic organisms, functioning in municipal wastewater treatment systems, of 0.1 mg 0 2 /L and 0.75 mg 0 2/L, respectively. For SBR operation using the baseline wastewater, the AOB K 0 value was roughly estimated to be in the range of 0.8 to 1.5 mg 0 2 /L using data from aeration rate perturbation experiments (Section 4.7.1). This range of K 0 values compares favourably with the typical 0.75 mg 0 2 /L value reported by Grady et al (1999). Assuming a DO concentration of 0.3 mg/L, typical of the experimental level, and the K 0 values reported by Grady et al (1999) in conjunction with the Monod model, shows that the heterotrophs would be functioning at about 75%) of their maximum rate, compared to only 30% for the autotrophs. The TOC data indicate that the heterotrophic carbon utilization rate was ultimately controlled by carbon supply rate, not DO concentration, since the TOC levels were independent of carbon loading rate. Alternately, due to the relatively high AOB K 0 value, even a very small reduction in DO concentration (e.g. 0.1 mg/L) or oxygen "availability", induced by heterotrophic activity, would significantly reduce the ammonia oxidation rate. Numerous studies (e.g. Stenstrom and Song, 1991; Charley et al, 1980; Stenstrom and Poduska, 1980) have examined the impact of oxygen levels on nitrification. However, the work of Hanaki et al (1990) is more relevant to the current research, as they examined the effect of heterotrophic activity, under different carbon loading rates, on nitrification at low DO levels (i.e. 0.5 mg/L). As an example of their results obtained for a continuous-flow system operated with a 0.5 mg/L DO concentration, they found that addition of 160 mg COD/L of glucose to the synthetic wastewater caused an increase in the overall ammonia oxidation K 0 value to 9.4 mg/L, a substantially larger value than the 1.1 mg/L value for the case 151 where glucose was excluded from the wastewater. These data clearly highlight the significant impact that heterotrophic competition for oxygen, under oxygen-limited conditions, has on nitrification kinetics; they also corroborate the findings of the current research. Process Monitoring and Control Implications From a process monitoring perspective, the impact of SDC utilization rate on the ammonia oxidation rate, and thus the competition for available oxygen between heterotrophic organisms and AOB bacteria, was most obviously reflected in the off-gas N 2 0 data, when compared the DO and pH data. In addition, the relative differences in the baseline and perturbation cycle N 2 0 mass stripping rates, as a surrogate for the N 2 0 generation rates, closely approximated the relative differences in the cycle ammonia oxidation rates given the functional performance (i.e. N 2 0 generation rate) of the biomass at the time the experiments were conducted. Both of these observations indicate the potential value of off-gas N 2 0 data for identifying changes in wastewater composition that significantly impact SBR nitrification performance. Furthermore, the data confirm the findings discussed in Section 4.5.1, where a reduction in ammonia oxidation rate manifests itself in a reduction in N 2 0 generation rate and an altered off-gas N 2 0 concentration-time profile, regardless of the cause of the reduced ammonia oxidation rate (i.e. lower in-situ ammonia concentration, heterotrophic-autotrophic oxygen competition). Section 4.7 further investigates the impact of oxygen supply/concentration/limitation on the nitrification rate and the resulting impact on N 2 0 generation, and provides a comprehensive discussion on these related issues and process monitoring/control implications. However, the observed reductions in N 2 0 generation rate induced by increased heterotrophic oxygen competition (i.e. increased SDC utilization rates) could be partially due to a mechanism independent of the AOB and the ammonia oxidation rate. Specifically, the potential mechanism is heterotrophic biochemical reduction of autotrophically generated N 20, in the presence of biologically available carbon, in an oxygen-limited environment. The heterotrophic denitrification pathway involves the step-wise reduction of nitrate to N 2 via nitrate respiration, nitrite respiration combined with NO reduction, and N 2 0 respiration (Zumft, 1997). Thus N 20, along with more commonly recognized nitrate and nitrite, is also a denitrification substrate. Wicht (1996) found that the maximum, uninhibited N 2 0 reduction rate was approximately four times faster than the nitrate and nitrite reduction rates. Therefore, the large differences in rates, under normal denitrifying conditions, explains why N 2 0 is usually undetectable and often neglected in discussions on denitrification. 152 Section 4.6 presents and discusses the experiments that were specifically conducted to investigate heterotrophic reduction of autotrophically-generated N 20. 1 5 3 4.6 N 2 0 REDUCTION EXPERIMENTS Section 4.5.3 identified the possibility of aerobic-phase, heterotrophic biochemical reduction of N 2 0 during the SDC utilization rate experiments. This section presents the results of experiments that were specifically conducted to investigate this mechanism. Section 4.6.1 examines the effect of DO concentration on N 2 0 reduction, with Section 4.6.2 assessing N 2 0 reduction as a function of SDC utilization rate. It should be noted that N 2 0 "reduction" refers to biochemical reduction of N 2 0 to dinitrogen gas (N2) via a change in the N oxidation state. It should also be noted that such a biochemical reaction also causes a "reduction" in actual N 2 0 mass, given the conversion of N 2 0 to N 2 . 4.6.1 DO Concentration This section presents and discusses the results obtained from a series of tests used to investigate the effect of DO concentration on N 2 0 reduction during the NOME-1 Oa experiment. Experimental Design The NOME-10a experiment was conducted using the SBR as the "test" reactor. However, prior to starting the experiment, the SBR was taken off-line from its normal operating sequence at the end of the aerobic-phase of a cycle. The subsequent experiment was divided into three sub-experiments or tests, as described below: • The Baseline Test was started immediately at the end of the aerobic-phase of the cycle. The purpose of the Baseline Test was to confirm the absence of biological N 2 0 reduction under very high DO levels and in the absence of external carbon loading to the reactor. The Baseline Test also confirmed the accuracy of the N 2 0 mass balance methodology used for the subsequent tests. Air was supplied to the reactor at 1,040 mL/min (i.e. typical baseline aeration rate) through the gas blending system, with the reactor DO concentration allowed to rise to its equilibrium concentration (i.e. approximately 7 mg/L) that balanced the oxygen transfer rate with the endogenous oxygen uptake rate. ' V • The reactor pH was maintained at approximately 7.5, the typical mid-range pH level of the aerobic-phase of an SBR cycle, through discreet manual addition of 0.2 N HC1 to the reactor. 154 The biologically generated N 2 0 from the preceding SBR cycle was quickly stripped from the mixed liquor. Therefore, before the Baseline Test was started, compressed N 2 0 (100% N 20) was also briefly added to the reactor, via the gas blending system, at a rate of approximately 17 mL/min. The compressed N 2 0 was added to the reactor until the off-gas concentration exceeded 1,000 ppm as measured using the IR N 2 0 analyzer. After the compressed N 2 0 flow was shut off, the off-gas concentration started to decrease. The Baseline Test started at the point in time (i.e. elapsed time = 0 min) when the off-gas N 2 0 concentration was about 1,000 ppm (i.e. the upper concentration of reliable measurement for the IR unit). • Test #1 was conducted immediately after the Baseline Test while the SBR was still off-line from its normal operating sequence. The purpose of Test #1 was to investigate the N 2 0 reduction rate under the "medium" SDC utilization rate (i.e. 19 mg C/L/hr, Section 4.5.3) with reactor DO levels that were typical (i.e. 0.2 to 0.3 mg/L) of those for the various SBR cycles discussed in Section 4.5.3. Test #1 used the same pH control strategy as that described for the Baseline Test. A concentrated sodium acetate solution was pumped into the reactor, at approximately 1 mL/min, to provide the desired carbon loading rate. Compressed N 2 was combined with air using the gas blending system, with the blended gas stream supplied to the reactor. The desired DO concentration was maintained in the reactor by adjusting the relative proportions of N 2 and air in the blended gas stream, while maintaining the same total gas flow rate to the reactor of approximately 1,040 mL/min. Finally, before the start of Test #1, N 2 0 was added to the reactor in a similar manner as that described for the Baseline Test. The start time of Test #1 was also assigned elapsed time = 0 min, although, of course, the actual start time was different than that of the Baseline Test. • Test #2 was conducted immediately after Test #1 while the SBR was still off-line from its normal operating sequence. The purpose of Test #2 was to investigate the N 2 0 reduction rate under the medium SDC utilization rate with reactor DO levels that were slightly higher-than-typical (i.e. 0.4 to 0.6 mg/L). With the exception of the DO levels, all other Test #2 experimental procedures were identical to 155 those of Test #1. Each of the three tests were conducted over a 20 min duration. Reactor pH, DO and off-gas N 2 0 levels were logged every minute. Reactor headspace and mixed liquor N 2 0 samples were collected at elapsed times = 0, 10 and 20 min. Duplicate gas samples were collected from the prepared mixed liquor N 2 0 samples for gas chromatograph (GC) analysis. For analytical procedure consistency, the N 2 0 mass balances were calculated using the headspace N 2 0 data (i.e. GC) rather than the off-gas N 2 0 data (i.e. IR analyzer). Duplicate mixed liquor nitrite and nitrate samples were collected at elapsed times = 0 and 20 min for Test #1 and Test #2. Process Control Figure 4.42 illustrates the DO data for Test #1 and Test #2. Although not shown in Figure 4.42, the Baseline Test DO concentration was about 7 mg/L for the entire test. The manual DO control procedure used for Tests #1 and #2 provided remarkably good control, with mean DO concentrations of 0.24 mg/L and 0.46 mg/L, respectively. The difference in the mean values was significant, at a 5% level of significance, when subjected to a paired t-test (Kennedy and Neville, 1986). -j , . Figure 4.42 also shows the pH data for all three tests. Again, the manual pH control procedure provided adequate and consistent control over reactor pH levels. Off-Gas and Mixed Liquor N20 Profiles The SBR off-gas N 2 0 data shown in Figure 4.42 provide the first indication of the significance of DO concentration on N 2 0 reduction. The Test #2 N 2 0 profile was very similar to the Baseline Test profile, suggesting that the Test #2 DO concentration of about 0.5 mg/L largely inhibited N 2 0 reduction. However, the reduced Test #1 DO concentration (0.2 mg/L) clearly allowed some N 2 0 reduction given the obvious difference in the Test #1 profile relative to the other test profiles. The mixed liquor N 2 0 profiles, shown in Figure 4.43, corroborate the off-gas N 2 0 data with respect to the N 2 0 reduction rate. As illustrated in Figure 4.43b, the Test #1 N 2 0 reduction rate was so high that no 156 2 . 0 1 .3 1 .6 1 .4 1 . 2 1 . 0 0 . 8 0 . 6 0 . 4 0 . 2 0 . 0 — A — T e s t #1 — • — Test #2 k-A—A-^_^. i r - A - A - A - A - j i i — A — i — T - - • 10 Elapsed Time (min) 1 5 20 SO 7.8 7.6 7.4 7.2 7.0 —M— Baseline —A—Test#1 — • — T e s t #2 • • 10 Elapsed Time (min) 15 20 1000 900 800 700 600 500 400 300 200 100 0 I — • — Baseline —A—Test #1 — • — T e s t #2 * A A i 10 Elapsed Time (min) 1 5 20 Figure 4.42: SBR DO, pH and off-gas N 2 0 concentration-time profdes (NOME-10a) 157 1.00 z at 0.80 E. q 0.60 z w O 3 0.40 CT '3 •D 0.20 X i 0.00 10 10 Elapsed Time (min) 20 20 a) Basel ine Test (DO = 7 mg/L) _ 1.00 -i a 0.80 £. 9, 0.60 z 3 0.40 cr •D 0.20 © S 0.00 10 10 Elapsed Time (min) 20 20 b)Tes t#1 (DO = 0.2 mg/L) _ 1.00 _j z ro 0.80 E. 9, 0.60 z § 0.40 #CT •o 0.20 © x E 0.00 10 10 Elapsed Time (min) 20 20 c ) T e s t # 2 ( D O = 0 .5mg/L) Figure 4.43: SBR mixed liquor N 2 0 concentration-time profiles for a) Baseline Test (DO = 7 mg/L), b) Test #1 (DO = 0.2 mg/L) and c) Test #2 (DO = 0.5 mg/L) (NOME-10a) 158 measurable N 2 0 was detected in the mixed liquor samples collected at elapsed time = 20 min. The various sets of duplicate mixed liquor N 2 0 data demonstrate the high quality of the data. N20 Reduction Rates The N 2 0 reduction rates were calculated using the mixed liquor (Figure 4.43) and headspace (Figure 4.44) N 2 0 data, the aeration rate (1,040 mL/min), and the reactor headspace (3.3 L) and mixed liquor volumes (10.6 L). The total N 2 0 mass contained in the reactor at elapsed time = 0 min was equal to the mass contained in the reactor headspace and liquid. However, for elapsed times =10 and 20 min, the total N 2 0 mass also included the mass that exited the reactor in the off-gas between sampling times. The "emission" mass for each time step was calculated using the headspace N 2 0 concentrations and the aeration rate. As expected, the Baseline Test did not induce any discernible N 2 0 reduction. The N 2 0 mass at the end of each time step was within -2.5% (elapsed time = 0 to 10 mini) and -6% (elapsed time = 10 to 20 min) of the mass at the start of the time step. The high "recovery" of the N 2 0 mass indicated that the measurement and mass balance procedures were suitably accurate. The calculated Test #1 and Test #2 N 2 0 reduction rates are shown below: • Test #1 = 20.5 mg N/hr = 1.93 mg N/hr per L of reactor liquid volume • Test #2 =3.6 mg N/hr = 0.34 mg N/L per L of reactor liquid volume The Test #1 reduction rate was calculated using data from only the elapsed time = 0 to 10 min time step because of the aforementioned rapid rate of N 2 0 reduction. For Test #2, data from both time steps were used to calculate the mean reduction rate, shown above. The mean rate provided a reliable estimate of the actual rate, since the rates for the two time steps showed little variation, at 3.5 mg N/hr (elapsed time = 0 to 10 min) and 3.7 mg N/hr (elapsed time = 10 to 20 min). The N 2 0 reduction rate data clearly indicate two main points: • The N 2 0 reduction rate was very sensitive to DO concentration, given the 82% reduction in the Test #2 rate relative to the Test #1 rate. 159 I 0.80 q o.60 z S 0.40 a. in •g 0.20 a> X 0.00 lllIB IBB 10 Elapsed Time (min) 20 a) Basel ine Test (DO = 7 mg/L) 1.00 "E 2 0.80 q o.60 z S 0.40 Q. U) "8 0.20 a> X 0.00 10 Elapsed Time (min) 20 b) Test #1 (DO = 0.2 mg/L) 1.00 N/m 0.80 3 q 0.60 z 0) u 0.40 a. in •o ra 0.20 u X 0.00 10 . Elapsed Time (min) 20 c) Test #2 (DO = 0.5 mg/L) Figure 4.44: SBR headspace N 2 0 concentration-time profiles for a) Baseline Test (DO = 7 mg/L), b) Test #1 (DO = 0.2 mg/L) and c) Test #2 (DO = 0.5 mg/L) (NOME-10a) 160 • The N 2 0 reduction rate for SBR DO levels that were typical to those of the experiments presented in Section 4.5.3 (i.e. Test #1) was significant, when compared to the baseline cycle N 2 0 generation rate. The steady-state off-gas N 2 0 concentration was approximately 600 ppm for the baseline cycles of the wastewater SDC utilization rate experiments (i.e. SBR-TS 16a,c,d). Using this concentration and an aeration rate of 1,040 mL/min, the calculated N 2 0 generation rate was about 43 mg N/hr. Thus, assuming the medium carbon loading rate did not affect the ammonia oxidation rate (although it clearly did), the "net" or observed N 2 0 generation rate would be reduced by approximately 50% because of heterotrophic N 2 0 reduction to N 2 . Nitrate Reduction Rates Figure 4.45 illustrates the duplicate nitrate sample data for Tests #1 and #2.. The data indicate that the relative change in the nitrate concentration across the duration of the tests was very low. In addition, relative to the change in concentration across the test duration, the variability in duplicate sample concentration was quite large. In particular, the Test #2 data were not usable for estimating the nitrate reduction rate because of the concentration variability. For the Test #1 data, the difference in the mean nitrate concentrations across the 20 min test duration was 0.4 mg N/L. This difference translated into a nitrate reduction rate of 1.2 mg N/L/hr. Although there was some uncertainty with respect to the absolute accuracy of the calculated nitrate reduction rate, the rate did fall in the same order-of-magnitude as the average N 2 0 reduction rate (i.e. [1.93 + 0.34] 12 = 1.14 mg N/L/hr). Although not shown, measurable levels of nitrite were not recorded for any nitrite samples collecting during any of the tests. Discussion The slightly elevated Test #2 DO concentrations (i.e. 0.5 mg/L), relative to Test #1 (i.e. 0.2 mg/L), clearly impacted the N 2 0 reduction rate. At least two explanations are possible: • One possible explanation for the observed phenomenon relates to the preferred terminal electron acceptor (TEA) of the heterotrophic organisms. Molecular oxygen is the preferred TEA for these 161 — 16.0 |> 15.! O 15.6 z § 15.4 CT 13 15.2 0> X ^ 15.0 0 20 Elapsed Time (min) a) Test #1 (DO = 0.2 mg/L) — 15.0 z |> 14.8 O 14.6 z § 14.4 •o 14.2 X S 14.0 0 20 Elapsed Time (min) 20 20 b) Test #2 (DO = 0.5 mg/L) Figure 4.45: SBR nitrate concentration-time profiles for a) Test #1 (DO = 0.2 mg/L) and b)Test #2 (DO = 0.5 mg/L) (NOME-lOa) 162 organisms, given the higher oxidation-reduction potential (ORP) of the oxygen/water couple compared to the nitrate/nitrite couple (Grady et al, 1999). In theory, it is possible that the heterotrophs were preferentially utilizing more oxygen and less nitrate/N20, since oxygen was supplied to the biomass at a higher rate and available at a higher concentration during Test #2 relative to Test #1. • The second, and more likely, explanation relates to the oxygen sensitivity of the N 2 0 reductase enzyme. Although heterotrophic denitrification has been shown to persist in highly aerobic environments (i.e. near or at oxygen saturation) following a switch from anoxic conditions (Lloyd et al, 1987), other studies (e.g. von Schulthess et al, 1994) demonstrated the negative impacts of oxygen on enzyme activity and denitrification rates for wastewater biomass. Furthermore, von Schulthess et al (1994) found that the N 2 0 reductase enzyme activity was very susceptible to oxygen inhibition, particularly when compared with other denitrifying reductase enzymes. As discussed in Section 4.4, the reported sensitivity of N 2 0 reductase to oxygen inhibition, relative to the other reductase enzymes, highlights another related N 2 0 generation mechanism. Specifically, heterotrophic nitrate reduction can induce N 2 0 generation if the N 2 0 reductase turnover rate is lower than the turnover rate of the previous enzymes (von Schulthess et al, 1994). This could result in some "indirect" heterotrophic N 2 0 generation, in addition to consumption, during the stepwise reduction of. nitrate in the presence of low DO concentrations. The mass of generated N 2 0, relative to the denitrified nitrogen mass, increases with increasing DO concentration. However, von Schulthess et al (1994) found that, at a DO concentration of 0.5 mg/L (i.e. Test #2 DO level), the mass of generated N 2 0 was only about 1% of the denitrified nitrogen mass. Therefore, under the experimental conditions for the current project, the potential for significant heterotrophic N 2 0 generation during simultaneous N 2 0 reduction-consumption appeared to be extremely limited. Process Monitoring and Control Implications The heterotrophic N 2 0 reduction mechanism has two main process monitoring and control implications: • Oxygen availability affects the ammonia oxidation rate and thus AOB N 2 0 generation rate (Section 4.7), as well as impacting the heterotrophic N 2 0 reduction rate via N 2 0 reductase inhibition. Thus, the "net" (i.e. measurable) N 2 0 generation rate will be impacted by oxygen availability to both groups of organisms and; therefore, requires consideration when attempting to use off-gas N 2 0 data 163 to identify changes in the ammonia oxidation rate. • The experimental data has shown that the availability of biologically utilizable carbon, under suitable DO conditions, can provide significant N 2 0 reduction rates. This factor also complicates the interpretation of off-gas N 2 0 data, with respect to identifying changes in ammonia oxidation rate that are induced by changing wastewater composition. When considered from a more extreme perspective, the data indicate that a significant amount of autotrophically generated N 2 0 might not even be evident (i.e. absent from the off-gas) or measurable, given a sufficiently high rate of heterotrophic N 2 0 reduction. Section 4.6.2 further investigates the effect of the SDC utilization rate on the N 2 0 reduction rate. 4.6.2 Slowly Degradable Carbon Utilization Rate This section presents and discusses the results obtained from a series of tests used to investigate the effect of SDC utilization rate on the N 2 0 reduction rate during the NOME-lOb experiment. Experimental Design The NOME-lOb experiment was conducted in a similar manner as the NOME-lOa experiment described in Section 4.6.1. The NOME-lOb tests are described below, with notable differences from the NOME-10a tests also identified: • The Baseline Test purpose and experimental procedure were identical to those of the NOME-lOa Baseline Test. • The purpose of Test #1 was to investigate the N 2 0 reduction rate under the "low" SDC utilization rate (i.e. 9.5 mg C/L/hr, Section 4.5.3) with reactor DO levels that were typical (i.e. 0.2 to 0.3 mg/L) of those for the various SBR cycles discussed in Section 4.5.3. • Similar to the NOME-lOa experiment, the purpose of Test #2 was to investigate the N 2 0 reduction rate under the "medium" SDC utilization rate (i.e. 19 mg C/L/hr, Section 4.5.3) with reactor DO levels that were typical (i.e. 0.2 to 0.3 mg/L) of those for the various SBR cycles discussed in Section 4.5.3. 164 All other elements of the NOME-lOb tests, with respect to test duration, sampling and analysis, and mass balance calculations, were identical to those described in Section 4.6.1 for the NOME-lOa tests. Process Control Figure 4.46 shows the DO data for Test #1 and Test #2. Although not shown in Figure 4.46, the Baseline Test DO concentration was about 6.9 mg/L during the entire test. The manual DO control procedure again provided good control, although there was more variation in the DO concentration than the variation in the NOME-lOa tests (Section 4.6.1). Over the entire 20 min duration, the mean Test #1 and Test #2 DO concentrations were 0.26 mg/L and 0.29 mg/L, respectively. The difference in the mean DO concentrations was statistically significant at a 5% level of significance when subjected to a paired t-test. However, between elapsed times = 0 and 10 min, the differences in the mean DO concentrations (Test #1 = 0.27 mg/L, Test #2 = 0.30 mg/L) was not statistically significant. As discussed in the following section, only data from the elapsed time = 0 to 10 min periods were used to calculate the N 2 0 reduction rates. Figure 4.46 also illustrates the pH profiles for the three tests, with the data again demonstrating the adequate pH control provided by the manual control method. Off-Gas and Mixed Liquor N20 Profiles The SBR off-gas data shown in Figure 4.46 provide an indication of the impact of carbon loading on the N 2 0 reduction rate. The Test #1 and #2 profiles were very different than the Baseline Test profile, although the differences between the Test #1 and #2 profiles were more subtle. The SBR mixed liquor N 2 0 profiles are shown in Figure 4.47. Similar to the NOME-lOa experiment (Section 4.6.1), the duplicate samples demonstrate the high quality of the data. Also, the N 2 0 reduction rates induced by both the low and SDC utilization rates eliminated measurable mixed liquor N 2 0 by elapsed time = 20 min. N20 Reduction Rates The N 2 0 reduction rates were calculated using the mixed liquor (Figure 4.47) and headspace (Figure 4.48) N 2 0 data as per the procedure described in Section 4.6.1. 165 2 0 1.8 1.6 1.4 1.2 1.0 0 8 0.6 0 4 0 2 0.0 A T . . I Jf A !_ m I est ff 1 f — • — T e s t #2 f" 5 10 15 Elapsed Time (min) 2 0 8.0 7.8 7 6 7 4 7.2 7.0 —•—Base l i ne —A—Test #1 — • — T e s t #2 r l 10 Elapsed Time (min) 15 2 0 5 10 Elapsed Time (min) 15 2 0 Figure 4.46: SBR DO, pH and off-gas N 2 0 concentration-time profdes (NOME-10b) 166 a) Basel ine Test _ 1.00 _j z o> 0.80 E. 9, 0.60 0.40 0.20 0.00 10 10 20 Elapsed Time (min) 20 b) Test #1 (9.5 mg C/L/hr carbon loading rate) 1.00 • z O ) 0.80 E, O 0.60 z uor 0.40 • CT ! J • D 0> 0.20 X i 0.00 10 10 Elapsed Time (min) 20 20 c) Test #2 (19 mg C/L/hr carbon loading rate) Figure 4.47: SBR mixed liquor N 2 0 concentration-time profiles for a) Baseline Test, b) Test #1 (9.5 mg C/L/hr carbon loading rate) and c) Test #2 (19 mg C/L/hr carbon loading rate) (NOME-lOb) 167 1.00 i 0.80 at q 0.60 z S 0.40 (Q Q. in •g 0.20 o x 0.00 10 Elapsed Time (min) a) Basel ine Test 1.00 "E 2 0.80 3 q 0.60 z 8 0.40 a in "g 0.20 a> X 0.00 10 Elapsed Time (min) b) Test #1 (9.5 mg C/L/hr carbon loading rate) 1.00 "E 2 0.80 q 0.60 z S 0.40 Q. in •° 0.20 w X 0.00 10 Elapsed Time (min) 20 20 c) Test #2 (19 mg C/L/hr carbon loading rate) Figure 4.48: SBR headspace N 2 0 concentration-time profiles for a) Baseline Test, b) Test #1 (9.5 mg C/L/hr carbon loading rate) and c) Test #2 (19 mg C/L/hr carbon loading rate) (NOME-lOb) 168 The average N 2 0 mass recovery for the two Baseline Test time steps was -3.5%. This recovery efficiency was consistent with the -4.3% mean efficiency for the NOME-lOa Baseline Test (Section 4.6.1). The calculated Test #1 and Test #2 N 2 0 reduction rates are shown below: • Test#l = 12.8 mgN/hr = 1.20 mgN/hr per L of reactor liquid volume • Test #2 = 26.9 mg N/hr = 2.54 mg N/L per L of reactor liquid volume As previously discussed, only the data for the elapsed time = 0 to 10 min time step were used to calculate the Test #1 and #2 reduction rates. The N 2 0 reduction data indicate two key points: • The medium SDC utilization rate (Test #2) provided an N 2 0 reduction rate that was about 2.1 times larger than the rate induced by the low carbon loading rate (Test #1). This difference in reduction rates was expected, given the two-fold difference in the carbon loading rates (i.e. low = 9.5 mg C/L/hr, medium = 19 mg C/L/hr). Therefore, the data confirm the general linearity of N 2 0 reduction rate with SDC utilization rate. • The Test #2 N 2 0 reduction rate of 26.9 mg N/hr compares fairly well with the 20.5 mg N/hr rate for NOME-lOa Test #1 (Section 4.6.1). Although conducted on adjacent days, both of these tests were conducted under the same SDC utilization rate and DO conditions, and thus provide "duplicate" experimental data. Nitrate Reduction Rates Figure 4.49 shows the duplicate nitrate sample data for Tests #1 and #2. The mean nitrate values were used to calculate the low and medium SDC utilization rate nitrate reduction rates of 0.3 mg N/L/hr and 1.5 mg N/L/hr, respectively. The notable variability in the Test #2 duplicate sample data suggests caution in their interpretation. However, the medium SDC utilization rate (i.e. Test #2) value was consistent with the 1.2 mg N/L/hr value presented in Section 4.6.1, also for the medium SDC utilization rate. 169 — 16.0 z |> 15.8 O 15.6 z § 15.4 c* ^ 15.2 o X S 15.0 0 20 Elapsed Time (min) a) Test #1 (9.5 mg C/L/hr carbon loading rate) — 15.0 z |> 14.8 O 14.6 . Z -§ 14.4 o* •o 14.2 X S 14.0 0 20 Elapsed Time (min) 20 20 b) Test #2 (19 mg C/L/hr carbon loading rate) Figure 4.49: SBR nitrate concentration-time profiles for a) Test #1 (9.5 mg C/L/hr carbon loading rate) and b) Test #2(19 mg C/L/hr carbon loading rate) (NOME-lOb) 170 Consistent with the NOME-lOa tests (Section 4.6.1), measurable levels of nitrite were not recorded for any samples collected during the NOME-lOb tests. Discussion, and Process Monitoring and Control Implications The rate of aerobic-phase hydrolysis of complex carbon compounds (i.e. slowly degradable carbon) contained in "real" wastewater controls the rate of biologically available carbon utilization, assuming all other reactor conditions are equal. The main finding of the NOME-10b tests was the confirmed linearity of the N 2 0 reduction rate with the carbon loading rate, where the carbon loading rate was simulating the biologically available carbon "supply" and utilization rate. Similar to the issues discussed in Section 4.6.1, the rate of carbon supply/utilization will affect the "net" N 2 0 generation rate and thus the off-gas N 2 0 profile; therefore, it requires consideration when using off-gas N 2 0 data to identify changes in ammonia oxidation rate. 171 4.7 A E R A T I O N R A T E E X P E R I M E N T S Section 4.5.3 discussed the impact that the heterotrophic "simulated" slowly degradable carbon (SDC) utilization rate had on nitrification kinetics and the N 2 0 generation response, as manifested in the off-gas N 2 0 concentration-time profile. The nitrification rate was observed to decrease with increased SDC utilization rate when the SBR was operated with the same aeration rate for both the baseline and perturbation cycles, indicating that the heterotrophic organisms were more efficient than the autotrophs in scavenging available oxygen. Simultaneously, the net N 2 0 generation rate was clearly observed to decrease with the reduction in ammonia oxidation rate. As discussed in Section 4.6, part of the reduced N 2 0 generation rate could be attributed to heterotrophic reduction of N 2 0 under appropriate carbon availability and DO conditions. However, the remainder of the reduced N 2 0 generation rate was hypothesized to be related to the reduced ammonia oxidation rate. Section 4.7 further examines autotrophic "oxygen availability", or conversely oxygen limitation, and the potential use of N 2 0 data in identifying this phenomenon. The presented experiments were used to investigate the effect that the aerobic-phase aeration rate had on nitrification performance and N 2 0 generation response, when the SBR was treating the same wastewater for both the baseline and perturbation cycles. Experimental Design : The aeration rate experiments were conducted using two back-to-back SBR cycles, with the SBR fed the same wastewater during the baseline and perturbation cycles. Two types of aeration rate experiments were conducted. The first type utilized the baseline synthetic wastewater for both the baseline and perturbation cycles, and these experiments are presented in Section 4.7.1. The second type of experiment, presented in Section 4.7.2, utilized the baseline synthetic wastewater for both cycles, but additional carbon loading, as described in Section 4.5.3, was also applied to both cycles to simulate increases in the slowly degradable carbon wastewater component. For the low and medium SDC utilization rate experiments described in Section 4.7.2, the SBR received the carbon solution for a 120 min duration (i.e. elapsed time = 30 min to 150 min). Alternately, for the high SDC utilization rate experiments, the SBR received the carbon solution for 90 min, from elapsed time = 30 min to 120 min. The shorter duration of carbon loading, during the high SDC utilization rate experiments, was to allow sufficient time after carbon loading ceased to permit nitrification to be completed by the end of the aerobic-phase of the baseline cycle. 172 The SBR was operated with constant aerobic-phase aeration rates during both cycles. However, for the perturbation cycle, the aeration rate was altered from the baseline cycle rate. For the purposes of data presentation and discussion, the change in the aeration rate was expressed in terms of the relative change in the reactor clean water oxygen mass transfer coefficient (KLa). For example, the oxygen KLa values for aeration rates of 1,040 and 1,248 mL/min were 7.8 hr"' and 9.0 hr"', respectively. Therefore, the relative change in aeration rate was defined as (9.0 - 7.8) / 7.8 = 0.15 = + 15% KLa. This approach was adopted based on the reactor oxygen mass transfer characteristics described in Section 4.3.2, where relative changes in air flow rate did not result in the same magnitude of change in the oxygen mass transfer rate. It is recognized that the absolute values of the clean water mass transfer coefficients were not applicable to the "in process" situation. The method of expression was adopted solely to make it easier to define the magnitude of change in the aeration rate, for the following discussions, while recognizing that changes in aeration rate do not induce a similar magnitude of change in the oxygen mass transfer rate. The magnitude of the perturbation cycle oxygen mass transfer rates (i.e. K La perturbations), relative to the baseline cycle rates, were as follows: , • • small = + 15% and - 12% K La • medium = + 34% and - 26% K La • large = + 68% KLa Due to limitations in nitrification performance, experiments conducted with the increased SDC utilization rates did not include the "negative" K La perturbations. For experiments that utilized the baseline wastewater (Section 4.7.1), the aerobic-phase ammonia oxidation rates were calculated using the ammonia data under the assumption that the ammonia removal rate equaled the ammonia oxidation rate (Note: refer to Sections 4.2 and 4.5.2 for explanation). Alternately, for experiments involving the increased SDC utilization rates (Section 4.7.2), ammonia synthesis was accounted for in the ammonia oxidation rate calculations as discussed in Section 4.5.3. The least squares regression analysis procedure was used to estimate the average aerobic-phase ammonia oxidation rate (Sections 4.5.2 and 4.5.3). Section 4.7 presents the results of five experiments that were conducted using the baseline wastewater, 173 along with six experiments that were conducted using wastewater with increased SDC utilization rates (i.e. increased carbon loading-feed rates). The eleven experiments were conducted in six groups during different weeks. The two experiments in each group were conducted on adjacent days. Table 4.4 summarizes the various experiments. The following sections present and discuss the detailed findings of the baseline wastewater (Section 4.7.1) and SDC utilization rate (Section 4.7.2) K La perturbation experiments. 4.7.1 Baseline Wastewater Experiments The baseline wastewater K La perturbation experiments illustrate the general SBR nitrification response to increased aeration (i.e. oxygen mass transfer) rates without the complexity of the simulated, and additional, SDC utilization component. Aeration System Operation and Average Ammonia Oxidation Rates Before presenting detailed data for the various experiments, this section presents a collective examination of the effect that KLa perturbations had on ammonia oxidation kinetic rates. Table 4.5 summarizes the aeration rates, along with the related oxygen mass transfer coefficients, mean aerobic-phase DO concentrations and average cycle ammonia oxidation rates for the experiments. The mean aerobic-phase DO concentration for each experiment was calculated using data from elapsed time = 40 min to a point in time approximately 10 mins prior to DO breakthrough. The relative statistics (RS) contained in Table 4.5 highlight the relative differences between the baseline and perturbation cycle oxygen mass transfer rates, mean DO concentrations and average cycle ammonia oxidation rates for the various experiments. The relative statistics values highlight several important concepts: • The difference in the perturbation cycle ammonia oxidation rates (RS C), relative to the baseline cycle rates, was highly correlated to the relative differences in the oxygen mass transfer coefficients (RS A). Figure 4.50a illustrates these data, as well as a linear model fit to the data using least squares regression analysis. The data do exhibit some non-linearity; this.was expected since the removal "of the oxygen limitation (i.e.1 higher relative differences in'the oxygen mass transfer coefficient) will induce an increase in ammonia oxidation rate, but only up to a finite point (Grady et al, 1999). However, the linear model offers an excellent visual fit to the data with small residual 174 Table 4.4: Summary of aeration rate experiments + 68% K La i SBR-TS12a - - SBR-TS19b + 34% K La SBR-TS8a SBR-TS18a SBR-TS17a SBR-TS19a + 15% K La SBR-TS9a SBR-TS18b SBR-TS17b -0% K La --12% K La SBR-TS9b - - -- 2 6 % K L a i SBR-TS8b - - • m. Baseline Baseline Baseline Baseline Wastewater Wastewater Wastewater Wastewater + + + Low SDC Medium SDC High SDC Utilization Rate Utilization Rate Utilization Rate 175 Table 4.5: Aeration system operation, average ammonia oxidation rate and relative statistics for baseline wastewater -aeration rate experiments Experiment Cycle Aeration Oxygen Mean Average Ammonia Oxidation Rate Relative Statistics 2 and Rate K L a 1 Aerobic-Phase Grouping DO Reactor Specific A B C D (mL/min) (hr'1 @ 20°C) (mg/L) (mg N/L/hr) (mg N/L/hr per g MLSS/L) SBR-TS8a baseline 1,248 9.0 0.42 11.1 5.16 perturbation 1,815 12:0 0.67 15.3 7.12 34% 60% 38% SBR-TS80 baseline 1,248 9.0 0.40 10.7 5.02 -2.7% perturbation 838 6.6 0.25 7.3 3.43 -26% -38% -32% SBR-TS9a baseline 1,040 7.8 0.48 12.6 5.63 perturbation 1,248 9.0 0.54 14.6 6.53 15% 13% 16% SBR-TS9D baseline 1,040 7.8 0.44 12.6 5.61 -0.4% perturbation 878 6.8 0.38 11.3 5.04 -12% -14% -10% SBR-TS12a baseline 1,040 7.8 0.43 13.3 6.12 perturbation 2,040 13.1 1.11 22.0 10.1 68% 158% 65% Notes: 1. "Clean water" tests conducted using synthetic wastewater (Section 4.3.2) 2. A = (perturbation - baseline) / baseline x 100% for oxygen mass transfer coefficient data B = (perturbation - baseline) / baseline x 100% for mean DO concentration C = (perturbation - baseline) / baseline x 100% for specific ammonia oxidation rate D = (2nd baseline -1 st baseline) /1 st baseline x 100% for specific ammonia oxidation rate 176 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Relative Difference in K L a a) o 0.2 5 o o CP I I — 4 y = -0.016 +0.47 x R = 0.92 < • -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 Relative Difference in Mean DO Concentration b) • . Figure 4.50: Rela t ive difference in ammonia oxidat ion rate versus relative difference in a) oxygen mass transfer coefficent and b) mean D O concentration for baseline wastewater - aeration rate experiments ( S B R - T S 8 a / b , S B R - T S 9 a / b , S B R - T S 12a) 177 values. Furthermore, the model slope equals 1.0, indicating that relative changes in the oxygen mass transfer rate induced an equal relative change in the ammonia oxidation rate for the experimental conditions. • Figure 4.50b shows the relative difference in ammonia oxidation rate (RS C) as a function of the relative difference in the mean DO concentration (RS B). Unlike the K La data, these data follow a distinctly non-linear trend, typical of a Monod-type relationship that expresses the ammonia oxidation rate as a function of DO concentration (Grady et al, 1999). Figure 4.50b contains a linear model for comparative illustration purposes. Generated N20 Mass as a Function of DO Concentration Figure 4.51 illustrates the ratio of generated N 2 0 mass to oxidized ammonia mass as a function of mean aerobic-phase DO concentration. For the experimental conditions, the data indicate a range of maximum N 2 0 generation that corresponded to a mean DO concentration of approximately 0.5 mg/L. Figure 4.51 also shows that the generated N 2 0 / oxidized ammonia mass ratio was approximately one order-of-magnitude higher for the low and high K La perturbation experiments relative to the medium K La perturbation experiments. Figure 4.5 Id shows a time-series of data obtained for a series of baseline SBR cycles following a particular SBR start up event, and illustrates where the data obtained from the various experiments discussed in this section fall. Section 4.9 specifically discusses this phenomenon, related to this particular SBR operating period, which was believed to be the result of a shift in the relative fractions of the AOB and NOB populations contained in the biomass. Clearly, the baseline biomass N 2 0 generation rate changed over the 50 day period that the baseline wastewater - aeration rate experiments were conducted. This contrasts the other SBR operating periods where the "baseline" biomass N 2 0 generation rate, as reflected in the generated N 2 0 / oxidized ammonia mass ratio, remained essentially constant during the period when the SBR was subjected to various "operational perturbations" (i.e. Section 4.5.3 - effect of wastewater SDC utilization rate on nitrification performance under a constant aeration rate; Section 4.7.2 - effect of wastewater SDC utilization rate on nitrification performance under an increased aeration rate). The Discussion section at the end of Section 4.7.1 further discusses the significance of this phenomenon with respect to data interpretation. It should also be noted here that although the generated N 2 0 / oxidized ammonia mass ratio significantly changed 178 0.120 « Jj 0.100 S c Z 0.080 q o cn z 1 | 0.060 ® ^ o • ' E 0.040 o N ~ ° 5 0.020 0.000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 Mean DO (mg/L) a) Low (+ 15%, -12%) K La perturbations (SBR-TS9a/b) 0.012 (fl i I 0.010 fj c Z 0.008 q o o> z i | 0.006 S < O) 5 ' E 0.004 <s S ~ o is 0.002 w 3 0.000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Mean DO (mg/L) 1.1 1.2 b) Medium (+ 34%, -26%) K La perturbations (SBR-TS8a/b) o o> E E 0.200 0.160 § 0.120 S o) 0.080 0.040 0.000 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Mean DO (mg/L) 1.1 1.2 c) High (+ 68%) K La perturbation (SBR-TS12a) 0.30 2 S 0.25 O 5 J? 5J E 0.20 0.15 0.10 0.05 0.00 0 10 20 30 40 50 60 70 80 90 100 110 120 Time Since SBR Start Up (d) SBR-TS12 SBR-TS9a/b SBR-TS8a/b d) Generated N 2 0 / oxidized ammonia mass ratio versus time since SBR start up Figure 4.51: The ratio of generated N 2 0 mass to oxidized ammonia mass versus mean DO concentration for a) low K L a (SBR-TS9a/b), b) medium K L a (SBR-TS8a/b) and c) high K L a (SBR-TS 12a) perturbation - baseline wastewater experiments; d) generated N 2 0 / oxidized ammonia mass ratio versus time since SBR start up _ over the particular 105 day monitoring period shown in Figure 4.5Id, the variability in adjacent day baseline cycles was extremely small, as illustrated in the day 54/56 and 75/76 experiment sets. This limited variability was observed in all baseline cycle data sets obtained during this research project, where experiments were conducted on back-to-back days, and thus provided confidence in the repeatability of N 2 0 generation under the baseline conditions in the context of interpreting perturbation cycle responses. The following sections present and discuss the detailed data for the various KLa perturbation experiments. Ammonia and Nitrite Oxidation Response Consistent with the ammonia oxidation rate data shown in Table 4.5, Figure 4.52 graphically illustrates the increasing ammonia oxidation rate, relative to the baseline cycle rate, with increasing relative difference in the oxygen mass transfer rate (i.e. KLa) for a select set of experiments (SBR-TS9a, SBR-TS8a, SBR-TS 12a). Figure 4.53 illustrates the nitrite profiles for the same experiments, and shows increased nitrite accumulation with the increased difference in perturbation cycle K La rate relative to the baseline KLa. The data indicate that the ammonia oxidizing bacteria (AOB) had a kinetic advantage over the nitrite oxidizing bacteria (NOB) when the biomass was supplied with additional oxygen. N20 Response The relative changes in nitrification rate induced by specific relative changes in the KLa clearly impacted the SBR off-gas N 2 0 profiles (Figure 4.54) (Note #1: The headspace N 2 0 data, rather than the off-gas N 2 0 data, are shown for SBR-TS8a, due to difficulties with the IR unit during the experiment that prevented reliable data collection. Note #2: As previously identified, and further discussed in Section 4.9, the baseline N 2 0 generation rate changed over time. This affected the absolute off-gas N 2 0 concentrations for the various sets of presented experiments.). To account for the different aeration rates used for the perturbation and baseline cycles, and thus dilution effects on N 2 0 concentrations, Figure 4.55 shows the cumulative stripped N 2 0 mass profiles for the cycles. The Figure 4.55 data highlight several observations: 180 30 J 25 z o> 20 E. •» 15 x z + 10 i s I 1 —£—Baseline — • — Perturbation s s »* * 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 15% K u a ( S B R - T S 9 a ) 30 -i 0 20 40 , 60 80 100 120 . 140 160 180 Elapsed Time (min) b) + 3 4 % K L a ( S B R - T S 8 a ) 30 ? 2 5 z ro 20 E, I " 1 5 z + 10 i s 0 -A — Baseline • ' Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) + 6 8 % K L a ( S B R - T S 1 2 a ) Figure 4.52: SBR ammonia concentration-time profiles for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation -baseline wastewater experiments 181 7.0 • 6.0 • 5.0 -z D> 4.0 • E " • 3.0 o 2.0 -z 1.0 • 0.0 • i i i —m\—Baseline • Perturbation \ \ \ \ \ 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 15% K L a ( S B R - T S 9 a ) 7.0 6.0 rr 5.0 4.0 E — 3.0 d Z 2.0 1.0 0.0 —A—Basel ine • Perturbati \ 3n \ \ \ \ IN 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) + 3 4 % K L a ( S B R - T S 8 a ) 7.0 6.0 3 5.0 * 4.0 E r - 3.0 i 2.0 1.0 0.0 I I .A Baseline — • — Perturbation \ \ \ \ y \ \ -A i 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) + 6 8 % K u a ( S B R - T S 1 2 a ) Figure 4.53: SBR nitrite concentration-time profiles for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation • baseline wastewater experiments 182 300 Z 150 in S 100 B a s e l i n e — • — P e r t u r b a t i o n 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 15% K L a ( S B R - T S 9 a ) Figure 4.54: SBR off-gas N 2 0 concentration-time profiles for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation - baseline wastewater experiments 183 Figure 4.55: SBR cumulative stripped N 2 0 mass-time profdes for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation - baseline wastewater experiments 184 • Regardless of the magnitude of the KLa perturbation, increased ammonia oxidation rates always occurred coincidental with increased N 2 0 generation rates, as represented by the cumulative stripped N 2 0 mass profile, during the early portions of the aerobic-phase. • Furthermore, the increasing difference in the perturbation cycle KLa, relative to the baseline cycle KLa, magnified the differences in the perturbation and baseline cycle cumulative stripped N 2 0 mass profiles during the early portion of the aerobic-phase of the SBR cycle. Liquid N 2 0 samples were not collected and analyzed for the presented experiments; thus, the cumulative generated N 2 0 mass curves could not be constructed and used to confirm that the baseline and perturbation cycles did, indeed, yield different N 2 0 generation rates. However, "complete" data from other experiments confirm that the trends in the cumulative stripped N 2 0 mass profiles reflect the trends of the cumulative generated N 2 0 mass profiles. The reader is referred to Section 4.3.3 and Figures 4.17 b and c, which illustrate the MISC-23 (aeration rate = 1,510 mL/min) and MISC-24 (aeration rate = 670 mL/min) experimental data. These SBR aeration rate experiments were conducted on adjacent days, and clearly illustrate the impact that the aeration rate had on nitrification kinetics (AOR = 17.0 mg N/L/hr for MISC-23; AOR = 6.5 mg N/L/hr for MISC-24), and the cumulative generated and stripped N 2 0 mass profiles. These data confirm that the conclusions drawn in the present section with respect to N 2 0 generation, on the assumption that the cumulative stripped N 2 0 mass profiles represent the trends in the generation profiles, were indeed valid. • Although the small K La perturbation (i.e. + 15% KLa) notably increased the ammonia oxidation rate (i.e. 16%) relative to the baseline cycle, it had essentially no impact on the ratio of generated N 2 0 mass per oxidized ammonia mass. The baseline and perturbation cycle mass ratios were 0.095 mg N/mg N and 0.097 mg N/mg N, respectively, as shown in Figure 4.51. This result was likely due to the small differences in the DO concentrations during the cycles. Figure 4.56a illustrates the DO profiles for the two cycles, with the calculated increase in the perturbation cycle mean DO concentration being about 16% (Table 4.5) higher than the baseline cycle average DO value. Alternately, the medium (i.e. + 34%) KLa) and large (i.e. + 68% KLa) K La perturbations induced significant reductions in the total mass of generated N 2 0 per oxidized ammonia mass (Figure 4.51c,d). For the + 34% K La and + 68% K La perturbation experiments, the reductions in the perturbation cycle N20/ammonia mass ratios, relative to the baseline cycle, were 43% and 48%, 185 2.0 1.8 1.6 _ 1-4 E. 1.0 O 0.8 ° 0.6 0.4 0.2 0.0 seline Bs • Perturbation • r i ^^^^ r 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) + 15% K L a ( S B R - T S 9 a ) 2.0 1.8 -1.6 1.4 • 1.2 E, 1.0 -O 0.8 -Q 0.6 -0.4 0.2 0.0 -I seline Be —•— Perturbation I 20 60 80 100 120 Elapsed Time (min) 140 b) + 3 4 % K L a ( S B R - T S 8 a ) 2.0 1.8 1.6 _ 1.4 O) £ io O 0.8 D 0.6 0.4 0.2 0.0 line Bas€ » Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 c) + 6 8 % K L a ( S B R - T S 1 2 a ) Figure 4.56: SBR DO concentration-time profiles for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation - baseline wastewater experiments 186 respectively. Figures 4.56b and c illustrate the DO profdes for these experiments and shows the large differences in the baseline and perturbation cycle DO levels. Although the + 68% KLa perturbation caused a very large increase (i.e. 158%>, Table 4.5) in the perturbation cycle mean DO concentration (1.11 mg/L) relative to the baseline cycle (0.43 mg/L), it did not notably reduce the total generated N 2 0 mass beyond the reduction provided by the + 34% K La perturbation experiment. As previously explained and illustrated in Figure 4.51, the N 2 0 generation "potential" was maximized at a DO concentration of about 0.5 mg/L. However, the data show that the effect of DO concentration on N 2 0 generation was minimal between the fairly wide ranging mean DO values of 0.67 mg/L and 1.11 mg/L for the + 34%> K La and + 68% KLa perturbation experiments, respectively. Discussion In theory, it would seem reasonable to presume that provision of more oxygen to the biomass should reduce the need for the AOB to utilize nitrite as an alternate terminal electron acceptor (TEA), thus reducing N 2 0 generation in the presence of increasing DO levels. However, the baseline wastewater aeration rate experimental data do not support this hypothesis. The ratio of generated N 2 0 mass per oxidized ammonia mass did not decrease with increased DO concentration. Rather, this mass ratio was observed to increase to a maximum point that corresponded with a DO concentration of about 0.5 mg/L. Subsequently, the mass ratio did significantly decrease with higher DO levels, but was not strongly correlated with DO concentration across the range of available experimental conditions. A review of wastewater literature found that the work of Zheng et al (1994) was the only study that provided somewhat comparable information to the findings of the current research, with respect to an "optimal" DO concentration for AOB N 2 0 generation. Their chemostat study used biomass from a municipal wastewater treatment facility. The synthetic wastewater used for the experiment contained no organic carbon, and given the long duration of the experiment, the biomass likely contained only a small population of heterotrophic organisms. Therefore, the generated N 2 0 most likely originated from AOB. The researchers found that the N 2 0 generation rate increased as the DO concentration increased from 0.1 to 0.2 mg/L, and then decreased as the DO concentration increased to 0.5 mg/L and 1.7 mg/L. Of particular note, the 30% increase in the N 2 0 generation rate, as the DO was increased from 0.1 to 0.2 mg/L, was accompanied by a 10% increase in the ammonia oxidation rate. The absolute values of the DO levels are not important, as they will vary with experimental conditions. However, similar to the current research findings, the Zheng et al (1994) data show increases in ammonia oxidation rate and ~N20 187 generation rate, up to a point, with increased DO concentration. Many researchers (e.g. Shrestha et al, 2001; Kester et al, 1997; Bock et al, 1995; Anderson et al, 1993; Hynes and Knowles, 1985; Goreau et al, 1980) have demonstrated reduced AOB N 2 0 generation with increased DO concentration in pure culture studies. However, only a few reviewed pure culture studies presented data that showed an increasing N 2 0 generation rate with increasing DO concentration, as discussed below: • For Nitrosomonas europaea cultured in previously sterile soil, the N 2 0 generation rate increased by about 10% as the oxygen concentration increased from 0.75% 0 2 to 2.5% 0 2, with a relative increase in the ammonia oxidation rate of approximately 40% (Hynes and Knowles, 1985). Beyond the 2.5% 0 2 concentration, the N 2 0 generation rate rapidly decreased but the ammonia oxidation rate continued to increase. • In the study conducted by Anderson et al (1993), the Nitrosomonas europaea N 2 0 generation rate increased by about 40% as the sparging gas oxygen content increased from 0.12% 0 2 to 0.2% 0 2 . The N 2 0 generation rate then remained constant until 0.4% 0 2, followed by a rapid decline in rate as the oxygen level reached 0.6% 0 2. No ammonia oxidation rate data were provided with the N 2 0 data, but the researchers made a notable observation in that'similar tests conducted with washed cells, where the media did not contain any nitrite (compared to 0.2 to 0.5 mM for the unwashed cells), resulted in no N 2 0 generation. Further experiments revealed that increasing the nitrite concentration to 1 mM (i.e. 14 mg N/L) and to 20 mM (i.e. 280 mg N/L) significantly increased the N 2 0 generation rate, while causing the 0.2 to 0.4% 0 2 optimum oxygen concentration range to "disappear". In the baseline wastewater aeration rate experiments, increasing the SBR oxygen mass transfer rate increased the ammonia oxidation rate, and at the same time, increased the N 2 0 generation rate during the early portion of the aerobic-phase. As well, measurable quantities of N 2 0 appeared in the off-gas earlier during the aerobic-phase of the cycle. It was earlier noted that the baseline cycle generated N 2 0 / oxidized ammonia mass ratio significantly changed over the course of this particular data collection period. Thus, it is possible that such a change could impact the relative change in the induced N 2 0 generation rate in response to additional oxygen supplied to the biomass during the perturbation cycles. 188 However, the Figure 4.55 a and c data (i.e. where baseline SBR cycles were generating relatively comparable amounts of N 20, compared to the situation illustrated in Figure 4.55b where N 2 0 generation was one order-of-magnitude smaller) clearly show that the larger the relative increase in the perturbation cycle K L a , relative to the baseline cycle K L a , the more dramatic the N 2 0 generation response. Similar trends in SBR response were also demonstrated in the data presented in Section 4.7.2, when the SBR was subjected to wastewater that simulated increases in SDC utilization rate and where the baseline biomass N 2 0 generation rate was essentially constant during the period of conducted experiments. Therefore, the drawn conclusions with respect to the general trends in ammonia oxidation rate and N 2 0 generation response with increasing oxygen mass transfer rate, as discussed in this section, are considered valid. The mechanism responsible for the apparently contradicting observed phenomenon with respect to N 2 0 generation (i.e. increased oxygen supply/mixed liquor concentration and increased N 2 0 generation) may be related to the concentration of nitrite within the reactor, as suggested by the Anderson et al (1993) data: • As previously discussed and illustrated in Figure 4.53, increasing the rate of oxygen supply to the SBR caused a higher concentration of nitrite to accumulate earlier during the aerobic-phase of the cycle, due to the apparent competitive advantage that the AOB had over the NOB in scavenging available oxygen. Other wastewater studies (e.g. Hanaki et al, 1990), as well as mixed culture studies with Nitrosomonas europaea and Nitrobacter winogradskyi isolates (e.g. Laanbroek and Gerards, 1993), have also shown that AOB can out compete NOB for oxygen, in oxygen-limited environments, given the observed accumulation of nitrite in the systems. In addition, Philips et al (2002) reports literature AOB and NOB KQ values, for wastewater biomass, that range from 0.25 to 0.5 mg 0 2 /L and 0.34 to 2.5 mg 0 2/L, respectively. The lower AOB K 0 values, relative to the NOB values, indicate that AOB kinetics are less impacted by low DO levels, when compared to NOB. Therefore, low DO levels may allow some nitrite to accumulate in the system, due to the differing kinetic response of the two groups of organisms to low DO levels. • As will be shown in Section 4.8, the biomass can readily respond, in terms of an increased N 2 0 generation rate, to increased reactor nitrite levels. • Furthermore, the increased nitrification rates induced a more rapid reduction in SBR pH levels during the initial portion of the aerobic-phase. Figure 4.57 shows the pH profiles for the three sets of 189 6.0 4 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 15% K L a ( S B R - T S 9 a ) 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) + 3 4 % K L a ( S B R - T S 8 a ) 9.0 8.5 8.0 £ 7 . 5 7.0 6.5 6.0 Baseline — • — Perturbation r i ***** IT 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) + 6 8 % K L a ( S B R - T S 1 2 a ) Figure 4.57: SBR pH concentration-time profiles for a) + 15% K L a (SBR-TS9a), b) + 34% K L a (SBR-TS8a) and c) + 68% K L a (SBR-TS 12a) perturbation - baseline wastewater experiments 190 experiments. The data clearly show that the relative difference in the perturbation and baseline cycle pH levels, during the early portion of the aerobic-phase, increased with the increasing relative difference in the K La perturbations. This further shifted the nitrite-nitrous acid equilibrium in favour of nitrous acid and, if nitrous acid was the true alternate TEA rather than nitrite proper, would further favour increased N 2 0 generation. • The possible net effect, then, is the higher nitrite-nitrous acid concentrations earlier in the aerobic-phase might increase the AOB ability to use nitrite as an alternate TEA in an oxygen-limited environment, regardless of the DO concentration or the extent of the oxygen limitation. In other words, the increased availability of nitrite-nitrous acid (i.e. tending to increase the N 2 0 generation rate) offset the effect of additional available oxygen (i.e. tending to decrease the N 2 0 generation rate), resulting in the net effect of an increased N 2 0 generation rate. The data also show that the "net" effect of increased N 2 0 generation was limited to the early portion of the aerobic-phase, as the provision of additional oxygen clearly reduced the total mass of N 2 0 generated over the course of the aerobic-phase. Very limited specific information was found in the literature with respect to the described observation relating N 2 0 generation and nitrite levels in SBR systems: • Park et al (2001) presents data that shows a rapid increase and decrease in the N 2 0 emission rate (i.e. an indication of the off-gas concentration), subsequent to the start of the aerobic-phase of the cycle, that coincides with a rapid increase and decline in nitrite concentration. However, no evidence was provided that AOB were indeed the source of the N 20. Furthermore, the mixed liquor N 2 0 concentration-time profile showed that N 2 0 had accumulated in the mixed liquor during the anoxic-phase, with the concentration quickly reduced to a steady-state level once aeration began. Thus, it is possible that the measured off-gas N 2 0 originated from the anoxic-phase mixed liquor, and was simply stripped from the mixed liquor at the start of the aerobic-phase. • Okayasu et al (1997) presents data for an SBR system treating "night soil". In one figure, the aerobic-phase N 2 0 emission rate was shown to rapidly increase to about 160 mg N/hr, after several hours of steady-state at 100 mg N/hr, at the same point when significant nitrite began to accumulate in the reactor. The figure scale did not allow identification of the nitrite levels early in the aerobic-191 phase, but they were about 10 mg N/L at the time of the peak N 2 0 emission rate. The researchers acknowledged that either autotrophic or heterotrophic organisms could have been responsible for the N 2 0 generation. • The various parameter-time profdes presented by Zeng et al (2003) for their anaerobic/aerobic SBR system were similar to those obtained during this research, except that both nitrite and nitrate concentrations were essentially zero during both phases of the cycle. As noted by the authors, N 2 0 generation was significant even though the "... measurable nitrite concentration was very low (close to zero)". The scale of the figures did not allow identification of actual nitrite concentrations. Again, the researchers did not explicitly demonstrate the source of N 20, assuming it was generated heterotrophically in anoxic micro-zones with the floes; and they did not even consider the possibility that the N 2 0 was generated by AOB. Burgess et al (2002) provided data for a continuous-flow reactor system that was subjected, at different times, to ammonia spikes and reductions in aeration rate. Both types of perturbations reduced the reactor DO concentration, with simultaneous increases, and subsequent decreases, in off-gas N 2 0 and nitrite concentrations over a time scale of less than 100 min. Thus, their data also show a correlation between increased N 2 0 generation rate with increased nitrite concentration. Similar to the highlighted studies, the authors assumed that AOB nitrite reduction was the source of N 2 0. Process Monitoring and Control Implications From a process monitoring and control perspective, the data presented in this section show that the off-gas N 2 0 data can be used to identify a change in the oxygen-competition dynamic between AOB and NOB that affects the relative difference in the-ammonia and nitrite oxidation rates, induces subtle differences in transient nitrite levels, and ultimately provides an indication of how the ammonia oxidation rate has changed due to the altered oxygen competition. The data presented in this section were obtained from experiments that altered the AOB-NOB oxygen-competition dynamic by changing the reactor oxygen supply rate. The Section 4.5.3 experiments also changed the oxygen competition dynamic, but did so indirectly by changing the heterotrophic OUR via an increased SDC utilization rate. Section 4.7.2 combines changes in oxygen supply rate with SDC utilization rate to further investigate SBR response to altered oxygen competition. Section 4.8 presents and discusses the experiments that were conducted to specifically investigate the 192 significance of nitrite and pH (nitrous acid) levels on N 2 0 generation. Section 4.8.1 also provides a hypothesized theoretical framework for the relationship between N 2 0 generation and nitrite-nitrous acid concentration. 4.7.2 Slowly Degradable Carbon Utilization Rate Experiments The K La perturbation - SDC utilization rate experiments illustrate the general SBR nitrification response to increased aeration (i.e. oxygen mass transfer) rates with the simulated SDC wastewater component. Aeration System Operation and Average Ammonia Oxidation Rates This section presents a collective examination of the effect that K La perturbations had on ammonia oxidation kinetic rates when the SBR was operating under increasing SDC utilization rates. Examination of the experimental data (Table 4.6) revealed a somewhat different situation from that of the baseline wastewater experiments, as discussed below: • Increases in the relative ammonia oxidation rates (RS C) were positively correlated with relative increases in the oxygen mass transfer coefficients (RS A). However, unlike the situation for the baseline wastewater experiments, the increased oxygen mass transfer rate did not induce the same relative change in the ammonia oxidation rate. The low and medium SDC utilization rate experiment data are shown in Figure 4.58. For these experiments, the relative increases in the oxygen mass transfer coefficients induced a much larger relative change in the ammonia oxidation rate. • In addition, for the experimental conditions, the SDC utilization rate affected the relative change in the ammonia oxidation rate given the same relative change in the oxygen mass transfer coefficient. Consider the + 33% K La perturbation data as an example (Table 4.6). Here, the induced relative increases in the ammonia oxidation rate were 76%, 57% and 0% for the low, medium and high SDC utilization rates, respectively. For this level of K La perturbation, the extent of "removal" of the autotrophic "oxygen limitation" was controlled by the SDC utilization rate and related heterotrophic OUR. The following sections present and discuss the detailed data from the K La perturbation - SDC utilization rate experiments. Section 4.7.1 noted that the carbon loading period for the low and medium SDC utilization rate experiments was from elapsed time = 30 min to 150 min, and from elapsed time = 30 min 193 Table 4.6: Aeration system operation, average ammonia oxidation rate and relative statistics for slowly degradable carbon utilization aeration rate experiments Experiment Cycle Aerobic-Phase Aeration Oxygen Mean Average Ammonia Relative Statistics3 and SDC Rate K L a 1 Aerobic-Phase Oxidation Rate2 Grouping Utilization Rate DO Reactor Specific A B C D (mg N/L/hr (mL/min) (hr'1 @ 20°C) (mg/L) (mg N/L/hr) perg MLSS/L) SBR-TS18a baseline low 1,040 7.8 0.20 6.3 2.44 perturbation low 1,510 10.4 0.26 11.1 4.30 33% 30% 76% SBR-TS18D baseline low 1,040 7.8 0.23 6.8 2.72 11.4% perturbation low 1,248 9.0 0.25 8.8 3.52 15% 9% 29% SBR-TS17a baseline medium 1,040 7.8 0.24 6.8 2.78 perturbation medium 1,510 10.4 0.27 10.7 4.37 33% 13% 57% SBR-TS17D baseline medium 1,040 7.8 0.22 5.2 2.20 -20.6% perturbation medium 1,248 9.0 0.22 6.9 2.92 15% 0% 33% SBR-TS19a baseline high 1,040 7.8 0.15 0.0 0.00 perturbation high 1,510 10.4 0.18 0.0 0.00 33% 20% 0% SBR-TS19D baseline high 1,040 7.8 0.17 0.0 0.00 perturbation high 2,040 13.1 0.31 7.2 3.05 68% 82% -Notes: 1. "Clean water" tests conducted using synthetic wastewater (Section 4.3.2) 2. Perburbation cycle rates adjusted to account for ammonia assimilation: low carbon rate = 1 n mg N/L/hr medium carbon rate = 3 2 mg N/L/hr high carbon rate = 6 4 mg N/L/hr 3. A = (perturbation - baseline) / baseline x 100% for oxygen mass transfer coefficient data B = (perturbation - baseline) / baseline x 100% for mean DO concentration C = (perturbation - baseline) / baseline x 100% for specific ammonia oxidation rate D = (2nd baseline - 1st baseline) / 1st baseline x 100% for specific ammonia oxidation rate 194 .= 0.60 CD | 0.50 o » ° - 4 0 Q 0.30 • Low SDC • Mpr i ium snn 0.15 0.33 Relative Difference in K L a F i g u r e 4.58: Relat ive difference in ammonia oxidat ion rate versus relative difference in oxygen mass transfer coefficient for l ow ( S B R - T S 18a/b) and medium ( S B R -T S 17a/b) S D C ut i l izat ion rate experiments 195 to 120 min for the high SDC utilization rate experiments. Therefore, for the purposes of clarity, data for samples collected after the carbon loading period were not shown in the figures. Baseline Cycle Generated N20 Mass The SDC utilization rate experiments presented in this section were conducted in three sets over a two week period. The day after each set of experiments was conducted, a regular "baseline" SBR cycle was monitored for off-gas N 2 0 concentrations when the SBR was treating only the regular baseline wastewater (i.e. no additional carbon loading). This monitoring was done to provide information on the stability of N 2 0 generation, under baseline conditions, over this experimental period. During this period, the baseline cycle generated N 2 0 / oxidized ammonia mass ratio decreased from about 0.22 to 0.17. The relatively small change in this ratio indicates that baseline biomass N 2 0 generation was quite stable over the period that the SDC utilization rate experiments were conducted. +15% KLa Perturbations (baseline, low and medium SDC utilization rates) The first set of presented data are for the + 15% K La perturbations for the baseline wastewater (i.e. no "additional" SDC utilization, shown for comparative purposes), and low and medium SDC utilization rate experiments. As shown in Tables 4.5 and 4.6, the + 15% K La perturbation induced increases in perturbation cycle ammonia oxidation rates, relative to the baseline cycle rates, of 16%), 29%> and 33%> for the baseline wastewater, low and medium SDC utilization rate experiments, respectively. Figure 4.59 illustrates the off-gas N 2 0 profiles for these experiments, with the cumulative stripped N 2 0 mass profiles shown in Figure 4.60. After the time of appearance of measurable quantities of N 2 0 in the off-gas, the general trend for both parameters was that the difference in the baseline and perturbation cycle values, at any given elapsed time, markedly increased with increased SDC utilization rate. For both the low and medium SDC utilization rate experiments, the perturbation cycle DO profile was virtually indistinguishable from the baseline cycle profile (Figure 4.61). Similar to the DO profiles, the baseline and perturbation cycle pFl profiles were almost identical to each other for the low (Figure 4.62b) and medium (Figure 4.62c) SDC utilization rate experiments (Note: The pH was manually adjusted during the SBR-TS 17b perturbation cycle (Figure 4.62c) between elapsed times = 20 and 30 min to raise the pH to the same initial aerobic-phase level as the baseline cycle. This action was needed because the 196 300 — 250 £ 200 O Z 150 in 3 100 It O 50 0 Baseline • Perturbation w£X-20 40 60 80 100 120 Elapsed Time (min) 160 180 a) Baseline Wastewater (SBR-TS9a) 500 400 300 200 100 0 Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) Low SDC utilization rate (SBR-TS18b) 500 300 8 200 o O 100 I I I Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 c) Medium SDC utilization rate (SBR-TS17b) Figure 4.59: SBR off-gas N 2 0 concentration-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K L a perturbation experiments 197 q = CO 3 S 25 20 15 10 5 I I I Baseline — • — Perturbation 20 60 80 100 120 140 160 180 Elapsed Time (min) a) Basel ine Wastewater ( S B R - T S 9 a ) 35 q z 30 •D 0> CL Z 25 trip O) E. 20 to to> IA (A 15 ra S 10 i i Baseline • Perturbation / y f / r / 20 60 80 100 120 Elapsed Time (min) 140 160 180 b) Low S D C utilization rate (SBR-TS18b ) J5 2 3 E 3 o I I Baseline — • — Perturbation t j J / 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 c) Medium S D C utilization rate (SBR-TS17b ) Figure 4.60: SBR cumulative stripped N 2 0 mass-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K L a perturbation experiments 198 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Baseline • Perturbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) Low S D C utilization rate ( S B R - T S 1 8 b ) 2.0 -j 1.8 1.6 1.4 • _J D> 1.2 • E, 1.0 • O 0.8 Q 0.6 -0.4 • 0.2 0.0 Baseline • Perturbation 60 80 100 120 Elapsed Time (min) 180 c) Medium S D C utilization rate ( S B R - T S 1 7 b ) Figure 4.61: SBR DO concentration-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K L a perturbation experiments 199 9.0 8.5 8.0 £ 7 . 5 7.0 6.5 6.0 — Baseline • Perturbation f 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) Basel ine Wastewater ( S B R - T S 9 a ) 9.0 8.5 8.0 7.5 7.0 6.5 6.0 Baseline — • — Perturbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) Low S D C utilization rate ( S B R - T S 1 8 b ) 9.0 8.5 8.0 7.5 7.0 6.5 6.0 Baseline — • — Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) Medium S D C utilization rate ( S B R - T S 1 7 b ) Figure 4.62: SBR pH-time profiles for a) baseline wastewater (SBR-TS9a), b) low SDC utilization rate (SBR-TS 18b) and c) medium SDC utilization rate (SBR-TS 17b) + 15% K L a perturbation experiments 200 low AOR during the baseline cycle resulted in completion of nitrification very close to the end of the aerobic-phase. Thus, there was little time for the aeration system to strip excess CO2 from the mixed liquor and increase the pH level to typical levels). In comparison, the perturbation cycle of the baseline wastewater experiment (Figure 4.62a) did generate a notably different pH profde, relative to the baseline cycle, reflecting the higher perturbation cycle nitrification rate. Comparison of'+15% KLa and + 33% KLa Perturbations (low and medium SDC utilization rates) The low SDC utilization rate experiments (SBR-TS18a and SBR-TS18b) were conducted on adjacent days, allowing direct comparison of the effect the + 15% KLa and + 33% K La perturbations had on SBR response: • Significant net or observed N 2 0 generation, as reflected in the cumulative stripped mass profile, began approximately 40 min earlier during the perturbation cycle than the baseline cycle for + 33% K La perturbation experiment (Figure 4.64b). Alternately, during the + 15%> KLa perturbation cycle (Figures 4.63a and 4.64a), the onset of N 2 0 generation began only about 20 min earlier than during the baseline cycle. • The + 15% K La perturbation provided a 1.6 times increase in the ratio of the measurable generated N 2 0 mass (i.e. cumulative stripped + mixed liquor mass) / oxidized ammonia mass (0.19 mg N/mg N), relative to the baseline cycle (0.12 mg N/mg/N), during the carbon loading period. Alternately, the + 33% K La perturbation induced a 2.4 times increase for this ratio (0.22 versus 0.09 mg N/mg N)over the same period. Thus the + 33%> K La perturbation increased the N 2 0 / ammonia mass ratio by about 50% over the + 15% KLa perturbation. • Linear models were fit, using least squares regression analysis, to the baseline (elapsed time = 120 min to 150 min) and perturbation cycle (elapsed time = 120 min to 150 min) cumulative stripped N 2 0 mass data for the + 15% K La perturbation experiment. The + 15% K La perturbation increased the estimated N 2 0 generation rate by 88% (i.e. 0.64 versus 0.34 mg N/min) over the baseline cycle. As shown in Table 4.6, the + 15% K La perturbation increased the ammonia oxidation rate by 29%. Similarly, linear models were fit to the baseline (elapsed time = 120 min to 150 min) and perturbation cycle (elapsed time = 100 min to 140 min) cumulative stripped N 2 0 mass data for the + 33% K La perturbation experiment. The + 33% K La perturbation increased the estimated N 2 0 201 — 500 H 400 o Z 300 | 200 it O 100 Baseline >—Perti jrbation 60 80 100 120 Elapsed Time (min) a) + 15% K L a - low S D C utilization rate (SBR-TS18b) 600 — 500 a 400 o Z 300 3 200 It O 100 Baseline •—Pert jrbation / *\ y T 1 \ \ / • ft«MM m a t f 60 80 100 120 Elapsed Time (min) 160 b) + 3 3 % K L a - low S D C utilization rate (SBR-TS18a) 350 300 E Q. 250 Q. q 200 z in 150 n O 100 3= O 50 0 Baseline • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 c) + 15% K L a - medium S D C utilization rate (SBR-TS17b) 350 300 250 200 150 100 50 0 I I I Baseline —•— Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) d) + 3 3 % K L a - medium S D C utilization rate (SBR-TS17a) Figure 4.63: SBR off-gas N 2 0 concentration-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS 18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K La - medium SDC utilization rate (SBR-TS 17a) experiments 202 60 O 2 50 •o a .— I 2 40 ~ O) co £ 30 o> in > w 1 I 2 ° 1 10 o I I I Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 a) + 15% K u a - low S D C utilization rate (SBR-TS18D) 60 80 100 120 Elapsed Time (min) 160 180 b) + 3 3 % K L a - low S D C utilization rate (SBR-TS18a) 30 25 g -z 20 ~ O) w £ 15 o> o > w I I 1 ° 3 E 5 3 ( J 0 I I I Baseline — • — Perturbation 20 60: 80 100 120 Elapsed Time (min) 140 160 180 c) + 15% K L a - medium S D C utilization rate (SBR-TS17b) 30 25 20 15 10 5 0 I I I Baseline • •—Pert urbatior / / y 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 d) + 3 3 % K L a - medium S D C utilization rate (SBR-TS17a) Figure 4.64: SBR cumulative stripped N 2 0 mass-time profiles for a) + 15% K La -low SDC utilization rate (SBR-TS18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33%) K La - medium SDC utilization rate (SBR-TS 17a) experiments 203 generation rate by 207% (i.e. 0.86 versus 0.28 mg N/min) over the baseline cycle. The + 33% K La perturbation increased the ammonia oxidation rate by 76%. " Examination of the induced increases in ammonia oxidation rate and N 2 0 generation rate illustrates two main points. First, the + 33% K La perturbation increased the ammonia oxidation rate in a fairly proportional amount, relative to the +15% K La perturbation. The same can be said of the estimated N 2 0 generation rates. Second, both aeration rate perturbations induced much larger increases in the N 2 0 generation rate, relative to the baseline cycles, when compared to the relative increases in the ammonia oxidation rates. Also, DO levels (Figure 4.65a, b) were very similar for both the baseline and perturbation cycles. The medium SDC utilization rate experiments (SBR-TS 17a and SBR-TS 17b) were also conducted on adjacent days, allowing direct comparison of the effect the + 15% K La and + 33%> K La perturbations had on SBR response under a higher SDC utilization rate: • Similar to the low SDC utilization rate experiments, Table 4.6 shows that the + 15% K La and + 33% K La perturbations increased the perturbation cycle ammonia oxidation rate under the medium SDC utilization rate, relative to the baseline rate, by 33% and 57%, respectively. Again, the + 33% KLa perturbation increased the nitrification rate in a fairly proportional amount relative to the + 15% KLa perturbation. Due to the extremely low baseline cycle N 2 0 generation rates, as inferred from the cumulative stripped N 2 0 mass curves (Figure 4.64c, d), the relative changes in the N 2 0 generation rates between cycles were not determined. • Significant net N 2 0 generation, as reflected in the cumulative stripped mass profile, began approximately 70 min earlier during the perturbation cycle for the + 33%) K La perturbation experiment (Figure 4.64d) compared to a 30 min difference for the + 15% K La perturbation experiment (Figure 4.64c). Both of these perturbation cycle N 2 0 generation "onset times" occurred earlier in the cycle, relative to the baseline cycle onset times, compared to the same data for the low SDC utilization rate experiments. • The + 33% K La perturbation provided an 11 times increase in the ratio of measurable generated N 2 0 204 2.0 1.8 1.6 _ 1.4 * 1-2 £ i-o O 0.8 Q 0.6 0.4 0.2 0.0 I I Baseline —4— Perturbation 0 20 40 60 . 80 . 100 120 140 160 180 Elapsed Time (min) a) + 15% Ka) + 15% K L a , low carbon loading rate (SBR-TS18b) 2.0 - i 1.8 1.6 -_ 1.4 -_ J O l 1.2 E, 1.0 -O 0.8 -Q 0.6 -0.4 0.2 0.0 • i I seline B E -—•— Perturbation > -20 60 80 100 120 Elapsed Time (min) b) + 3 3 % K L a - low S D C utilization rate (SBR-TS18a) 2.0 1.8 1.6 _ 1.4 E. 1.0 O 0.8 Q 0.6 bjaselme < • Perturbation < • —f- 1 60 80 100 120 140 160 180 Elapsed Time (min) c) + 15% K L a - medium S D C utilization rate (SBR-TS17b) 2. E, 1. 0 0. D o . I I Baseline —•— Perturbation r < • < 9mmfi 20 60 80 100 120 Elapsed Time (min) 160 d) + 3 3 % K L a - medium S D C utilization rate (SBR-TS17a) Figure 4.65: SBR DO concentration-time profiles for a) + 15% K L a - low SDC utilization rate (SBR-TS 18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% K L a - medium SDC utilization rate (SBR-TS17b) and d) + 33% K L a - medium SDC utilization rate (SBR-TS 17a) experiments 205 mass / oxidized ammonia mass (0.11 mg N/mg N), relative to the baseline cycle (0.01 mg N/mg N), during the carbon loading period. Alternately, the + 15% K La perturbation induced only a 3 times increase for this ratio (0.03 versus 0.01 mg N/mg N) over the same period. In addition, the + 33% KLa perturbation increased the N 2 0 / ammonia mass ratio by about 270%> over the + 15% KLa perturbation. Clearly, under the medium SDC utilization rate, both the + 15% and + 33% K La perturbations provided a greater increase in the observed N 2 0 generation rate when compared to the same perturbations for the low SDC utilization rate experiments. Furthermore, under the medium SDC utilization rate, the + 33% K La perturbation proved more much significant in reducing the constraint on N 2 0 generation when compared to the + 33% K La perturbation applied under the low SDC utilization rate condition. Similar to the low SDC utilization rate experiments, the DO levels were again insensitive to the changes in oxygen supply rate (Figure 4.65c, d). Under the low and medium SDC utilization rates, both the + 15%> K La and + 33% K La perturbations did not significantly affect the mean aerobic-phase DO concentrations, but did induce much larger relative increases in ammonia oxidation rates (i.e. x 2) compared to the baseline wastewater experiments (Section 4.7.1) The baseline and perturbation cycle pH profiles were again very similar for the various experiments (Figure 4.66) (Note: As previously discussed, the pH was manually adjusted for all the shown perturbation cycles in Figure 4.66). Comparison of + 33% KLa and + 68% KLa Perturbations (high SDC utilization rate) As with the previous experiments, the high SDC utilization rate experiments (SBR-TS 19a and SBR-TS 19b) were conducted on adjacent days, allowing direct comparison of the effect the + 33% KLa and + 68%) K La perturbations had on SBR response. Figure 4.67 illustrates the dramatic effect that the + 68% K La perturbation had on the ammonia oxidation rate relative to the baseline cycle rate, as well as the + 33% K La perturbation cycle rate. Unlike the + 68% K La perturbation experiment, measurable levels of nitrite, nitrate and off-gas N 2 0 were not detected in samples collected during either the baseline or perturbation cycles for the + 33% K La 206 9.0 8.5 8.0 ^7.5 7.0 6.5 6.0 Baseline —•— Perturbation 20 40 60 80 100 120 Elapsed Time (min) a) + 15% K L a - low S D C utilization rate (SBR-TS18b) 9.0 8.5 8.0 7.5 7.0 6.5 6.0 60 80 100 120 Elapsed Time (min) b) + 3 3 % K L a - low S D C utilization rate (SBR-TS18a) 9.0 8.5 8.0 7.5 7.0 6.5 6.0 60 80 100 120 Elapsed Time (min) c) + 15% K L a - medium S D C utilization rate (SBR-TS17b) 9.0 8.5 8.0 £ 7 . 5 7.0 6.5 6.0 20 60 80 100 120 Elapsed Time (min) 140 160 180 Baseline —•— Perturbation Baseline —•— Perturbation Baseline —•— Perturbation 160 180 d) + 3 3 % K L a - medium S D C utilization rate (SBR-TS17a) Figure 4.66: SBR pH-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS18b), b) + 33% K L a - low SDC utilization rate (SBR-TS18a), c) + 15% K La - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K L a - medium SDC utilization rate (SBR-TS 17a) experiments 207 30 3" 25 z ro 20 E. x- 1 5 z + 10 i s 0 - A— Baseline - • — Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 3 3 % K L a ( S B R - T S 1 9 a ) 30 --J 25 z ro 20 • E. • ^ 15 • X z + 10 • X z 5 0 -. i A Baseline -' -A-—•—Perturba tion < — A — , iN 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) + 6 8 % K L a ( S B R - T S 1 9 b ) Figure 4.67: SBR ammonia concentration-time profiles for a) + 33% K L a (SBR-TS 19a) and b) + 68% K L a (SBR-TS 19b) high SDC utilization rate experiments 208 perturbation experiment. However, the + 33% K La - high SDC utilization rate experiment provided insight into the "denitrification" potential of the SBR under the operating condition. The perturbation cycle had an ammonia removal rate of 6.4 mg N/L/hr, compared to 4.2 mg N/L/hr for the baseline cycle. Assuming the same ammonia assimilation rates for both cycles, the difference in ammonia oxidation rates between the cycles was about 2.2 mg N/L/hr. The absence of oxidized nitrogen species in the mixed liquor and off-gas suggests that the oxidized ammonia was subsequently reduced to N 2 at a rate of 2.2 mg N/L/hr. This value compares favourably with the calculated 1.2 mg N/L/hr value for nitrate reduction under the medium carbon loading rate (Section 4.6.1), which was one-half the high carbon loading rate relevant to this discussion. Figure 4.68 shows the nitrite, off-gas N 20, cumulative stripped N 2 0 mass and DO profiles for the + 68% K La perturbation. Nitrite accumulated in the reactor, with the appearance of N 2 0 in the off-gas at about elapsed time = 70 min. The large K La perturbation induced an 82% increase in mean DO concentration between the perturbation (0.31 mg/L) and baseline (0.17 mg/L) cycles. Finally, the pH profile for the + 68% K La perturbation cycle (Figure 4.69b) was significantly different than the baseline cycle profile, unlike the case for the + 33% KLa perturbation cycle (Figure 4.69a). In comparison to the + 33% KLa perturbation, the + 68% K La perturbation induced a greater nitrification rate, resulting in enhanced alkalinity destruction and thus a more significant reduction in pH levels. Nitrite Levels The explanation for the large changes in the N 2 0 concentration and cumulative stripped mass profiles, induced by the positive K La perturbations, may again be related to the reactor nitrite concentrations. Similar to the observations for the baseline wastewater K La perturbation experiments described in Section 4.7.1, the positive K La perturbations, for the low, medium and high SDC utilization rate experiments, induced a higher rate of aerobic-phase nitrite accumulation during the perturbation cycles relative to the baseline cycles. First consider the nitrite data for the low and medium SDC utilization rate experiments (Figure 4.70): • The nitrite data were fairly linear between elapsed time = 45 min and 135 or 150 min; therefore, linear models were fit to the nitrite data using least squares regression analysis. For both the low and medium SDC utilization rates, the + 15% K La perturbation increased the nitrite accumulation rate by 209 2.0 ~ 1.5 _i z cn d z 0.5 0.0 I l —A—Baseline # Ppr t i i rhat inn > - 4k t\ A L > 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 100 80 60 40 20 0 i i Baseline — • — Perturbation J r J —i 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 3.0 q z 2.5 •D O tripp Z2 .0 cn E „ to — 1.5 t> u> re g 1.0 3 E 0.5 o 0.0 I I Baseline — • — Perturbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) 2.0 1.8 1.6 _ 1.4 $™ £ LO O 0.8 ° 0.6 0.4 0.2 0.0 -Baseline - Perturbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) Figure 4 .68: SBR nitrite, off-gas N 2 0 , cumulative stripped N 2 0 mass and DO profdes for + 68% K L a - high SDC utilization rate experiment (SBR-TS 19b) 210 Baseline — • — Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 3 3 % K u a ( S B R - T S 1 9 a ) Baseline — • — Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) + 6 8 % K L a ( S B R - T S 1 9 b ) Figure 4.69: SBR pH-time profiles for a) + 33% K L a (SBR-TS 19a) and b) + 68% K L a (SBR-TS 19b) high SDC utilization rate experiments 211 2.0 _ 1.5 _i z o E 1 0 d z 0.5 1 1 —A—Base ine rbation • Pertu J 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) + 15% K L a - low S D C utilization rate (SBR-TS18b) -A—Baseline — Perturbation 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) + 3 3 % K L a - low S D C utilization rate (SBR-TS18a) -A — Baseline -4— Perturbation 20 40 60 80 100 120 140 ,160 180 Elapsed Time (min) c) + 15% K L a - medium S D C utilization rate (SBR-TS17b) -A—Baseline • - • — Perturbation 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) d) + 3 3 % K L a - medium S D C utilization rate (SBR-TS17a) Figure 4.70: SBR nitrite concentration-time profiles for a) + 15% K La - low SDC utilization rate (SBR-TS 18b), b) + 33% K La - low SDC utilization rate (SBR-TS 18a), c) + 15% KLa - medium SDC utilization rate (SBR-TS 17b) and d) + 33% K L a - medium SDC utilization rate (SBR-TS17a) experiments 212 about 50%, relative to the baseline cycles. Under the medium SDC utilization rate, the + 33% KLa perturbation increased the nitrite accumulation rate, relative to the baseline rate, by 105%. Thus the + 33% KLa perturbation increased the nitrite accumulation rate in a fairly proportional amount relative to the + 15%> K La perturbation. However, for the low SDC utilization rate experiment, the + 33%> K La perturbation increased the nitrite accumulation rate by 220% relative to the baseline cycle. This disproportionally large increase in the nitrite accumulation rate was unexpected given the nitrite accumulation response of the + 15% K La perturbation, and also considering that both the + 33% KLa and + 15%) K La perturbations increased the ammonia oxidation rate by quite similar amounts (i.e. 57%> versus 76% relative to the baseline cycles). • The increased nitrite accumulation rate becomes of interest when looking at the point in time that measurable net ~N20 generation began. For example, consider the + 15% K La perturbation - low SDC utilization rate experiment. Figure 4.64a shows that measurable N 2 0 generation began at approximately elapsed time = 100 min during the baseline cycle. At this same time, the nitrite concentration was approximately 0.5 mg N/L (Figure 4.70a). During the perturbation cycle, N 2 0 generation started at about elapsed time = 80 min (Figure 4.64a) when the nitrite concentration was also about 0.5 mg N/L. The other experiments demonstrated the same trend, although the absolute value of the "critical" nitrite concentration was slightly variable. • Finally, consider the + 15% and + 33%> K La perturbation cycles for the medium SDC utilization rate experiments. As shown in Figure 4.64c, N 2 0 appeared in the off-gas at approximately elapsed time = 110 min during the + 15% K La perturbation cycle. The nitrite concentration at this time was just over 0.6 mg N/L (Figure 4.70c).. Alternately, during the + 33% K La perturbation cycle, off-gas N 2 0 appeared at about elapsed time = 70 min (Figure 4.64d) when the nitrite concentration was around 0.7 mg N/L (Figure 4.70d). These data suggest that, for the same biomass operating under the same SDC utilization rate, measurable N 2 0 generation began at a point in time that corresponded to the same nitrite concentration even though the biomass was functioning under different aeration rate and DO conditions. The + 15% and + 33%> KLa perturbation cycles for the low SDC utilization rate experiments demonstrated a similar response. 213 The nitrite data (Figure 4.68) for the + 68% K La perturbation - high SDC utilization rate experiment show a distinct step increase between elapsed times = 60 and 75 min, from 0.5 to 0.8 mg N/L. At the same point in the aerobic-phase, measurable N 2 0 appeared in the off-gas (Figure 4.68). Discussion, and Process Monitoring and Control Implications The results of aeration rate experiments conducted with the simulated wastewater SDC component generally reflected the findings obtained using the baseline wastewater (Section 4.7.1). However, compared to the baseline wastewater situation, the presence of additional-higher SDC utilization further impacted N 2 0 generation, and its relation to ammonia oxidation, in several ways because of the reactors enhanced "denitrification potential" (Section 4.6) when operating under the additional SDC load. These impacts are significant from a process monitoring and control perspective: • First, the faster heterotrophic nitrite reduction rate, during treatment of wastewater that provided a higher SDC utilization rate, would reduce the nitrite accumulation rate during the early portion of the aerobic-phase, thus shifting back the point in time when the; "threshold" nitrite concentration related to significant (i.e. observed) N 2 0 generation was reached. '* • . • Second, the faster denitrification rate provided by the higher SDC utilization rate would also provide a faster N 2 0 reduction rate. N 2 0 would not be detected in the off-gas until the point in time that the N 2 0 generation rate exceeded the N 2 0 reduction rate. Again, referring to the cumulative generated N 2 0 mass profiles shown in Figure 4.17, the data show that, after the onset of N 2 0 generation, the N 2 0 generation rate increases over time until it reaches a fairly constant rate. The cumulative stripped N 2 0 mass profiles also demonstrate this phenomenon. Therefore, a higher N 2 0 reduction rate would further shift back the point in time when measurable N 2 0 would be detected in the off-gas, relative to SBR operation under a lower SDC utilization rate. Furthermore, if the induced heterotrophic N 2 0 reduction rate exceeded the maximum autotrophic N 2 0 generation rate, measurable concentrations of N 2 0 would not be detected in the off-gas at any point during the cycle. The combined effect of these two processes further exaggerates differences in the observed N 2 0 generation onset time, the N 2 0 generation rate, and ultimately the off-gas N 2 0 profiles, when compared to SBR operation under the baseline wastewater conditions. • In addition, for SBR operation under moderate (i.e. low and medium) SDC utilization rates, the KLa 214 perturbations provided much larger relative increases in the N 2 0 generation rates, when compared to the relative increases in the ammonia oxidation rates. These results contrast with the baseline wastewater aeration rate experiment findings (Section 4.7.1), where the relative increases in estimated N 2 0 generation rates were fairly well correlated to the relative increases in ammonia oxidation rates. Thus, the additional oxygen provided during the perturbation cycles of the moderate SDC utilization rate experiments not only increased the ammonia oxidation rate, but also removed some of the "constraint" on N 2 0 generation under these SBR operating conditions. Furthermore, the extent of removal of this "constraint" was much larger for SBR operation under the medium SDC utilization rate, compared to operation under the low SDC utilization rate. This result was unexpected given the similarity in the changes in ammonia oxidation rates, relative to the baseline cycles, provided by the K La perturbations for both SDC utilization rates. One factor that could have caused the observed responses, when the SBR was operating under the simulated SDC loads, would be a larger induced reduction in the heterotrophic N 2 0 reduction-consumption rate. The DO data do not suggest that such a change was likely, assuming that the heterotrophic N 2 0 reduction rate was solely dependent on the DO concentration. However, the higher rate of oxygen supply, regardless of DO concentration, may have shifted the heterotrophs energy generation mechanism more in favour of using oxygen rather than the various nitrogen compounds as electron acceptors, thus reducing the N 2 0 reduction rate. Increased inhibition of the N 2 0 reductase enzyme may have also occurred as a result of the increased oxygen supply rate. The observed SBR N 2 0 generation response to changes in aeration rate is useful from a process monitoring and control perspective. Specifically, the magnitude of change in N 2 0 generation rate for a purposely made change in aeration rate provides an indication of wastewater composition with respect to the SDC component and its heterotrophic utilization rate. It also provides an indication of the relative change in the ammonia oxidation rate. • Finally, unlike the N 2 0 profiles, regardless of low or medium SDC utilization rate or K La perturbation, the baseline cycle and perturbation cycle DO and pH profiles were practically identical to each other. Therefore, from a process monitoring perspective, these parameters would not provide any indication of the effect that a change in aeration rate had on ammonia oxidation rate for the given magnitude of K La perturbations and SDC utilization rates. Large K La perturbations (i.e. + 68%), such as that provided when the SBR was operated under the high SDC utilization rate, can induce 215 such a dramatic change in the ammonia oxidation rate that the rate change is reflected in the pH profde as well as the N 2 0 profdes. For this case, a notable change was also observed in the DO profile. 216 4.8 NITRITE-NITROUS ACID EXPERIMENTS Section 4.7 discussed the apparent correlation between the onset of N 2 0 generation and the presence of a suitably high "threshold" nitrite concentration. This phenomenon was observed when the SBR was operated with the baseline wastewater, as well as wastewater with simulated increases in the slowly degradable carbon component. Section 4.8 further examines the issue of bulk solution nitrite-nitrous acid concentration and its impact on N 2 0 generation. Section 4.8.1 discusses the influence of nitrite proper on N 2 0 generation, with Section 4.8.2 examining the effect of nitrous acid, through pH manipulation, on N 2 0 generation. Section 4.8.3 discusses the relationship between N 2 0 generation and nitrite levels as observed during long-term SBR operation. 4.8.1 Nitrite Spikes This section presents and discusses the results of experiments used to investigate the effect that reactor nitrite concentrations had on N 2 0 generation. Experimental Design The nitrite spike experiments were conducted using two back-to-back SBR cycles, with the SBR fed the same baseline synthetic wastewater during the baseline and perturbation cycles. Two nitrite spike experiments were conducted. The first experiment (NOME-3) involved the addition of a 3 mg N/L nitrite spike (i.e. concentrated sodium nitrite solution) to the SBR early in the aerobic-phase of the perturbation cycle, at elapsed time = 44 min. The addition time was, chosen such that little N 2 0 generation was occurring at that time. The SBR was operated with a 1,248 mL/min aerobic-phase aeration rate during both cycles. During the second experiment (NOME-3b), a 3 mg N/L nitrite spike was added to the SBR at elapsed time = 44 min, with an additional 2.5 mg N/L spike added to the SBR at elapsed time =119 min. The second spike was purposely added during the time when reactor off-gas N 2 0 concentration had reached a fairly constant level. The SBR was operated with a 1,040 mL/min aerobic-phase aeration rate during both cycles. The NOME-3 and NOME3b experiments were conducted approximately 6 months apart, and over this time, small differences in SBR performance necessitated a change in the baseline aeration 217 rate to maintain comparable ammonia oxidation rates. DO and pH Profiles Figure 4.71 illustrates the baseline and perturbation cycle DO and pH profdes for the NOME-3 and NOME-3b experiments. As expected, given the small variance in average aerobic-phase ammonia oxidation rates (i.e. 4% between the NOME-3 cycles; 2% between the NOME-3b cycles), the baseline and perturbation cycle profdes for both parameters were virtually identical to each other. Therefore, neither DO or pH were expected to influence the SBR response to the perturbation cycle nitrite spikes. Nitrite, N20 and Nitrous Acid Profiles Nitrite spike at elapsed time = 44 min: • Figure 4.72 illustrates the dramatic SBR response, with respect to N 2 0 generation, subsequent to the 3 mg N/L nitrite spikes added to the SBR at elapsed time = 44 min. In both experiments, the off-gas N 2 0 concentration reached its peak value within 25 to 35 minutes of the nitrite spike, then rapidly declined to baseline cycle levels. The off-gas N 2 0 profdes followed a similar pattern as the nitrite profdes, but offset by the approximate 30 min time period. • Figure 4.73a shows the calculated nitrous acid concentrations for the NOME-3 experiment. Subsequent to the nitrite spike, the nitrous acid concentration remained fairly constant for the next 80 min of the aerobic-phase even though the nitrite concentration, and then the off-gas N 2 0 concentration, rapidly decreased. In NOME-3b, the nitrous acid concentration did decrease following the nitrite spike (Figure 4.73b). However, the relative change in the nitrous acid concentration, following the nitrite spike, was much smaller than the relative change in the nitrite concentration. Nitrite spike at elapsed time = 119 min: • The nitrite spike at elapsed time =119 min of the NOME-3b perturbation cycle (Figure 4.72c) induced a similar N 2 0 generation response (Figure 4.72d) as the earlier spike. However, the peak N 2 0 concentration following the second nitrite spike was about 20% higher (i.e. 850 ppm versus 700 218 _ 1.4 0.2 0.0 I I J Baseline — • — Perturbation f f j t • I 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) a) D O (NOME-3 ) ft X iseline srturbation Be -•—p< 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) b) pH ( N O M E - 3 ) O 0.8 n -i seline f 1 — • — Perturbation 4 • 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) c) D O (NOME-3b) Baseline — • — Perturbation — I 0 20 40 60 80 100 120 140 160 180 Elapsed Time (min) d) pH ( N 0 M E - 3 b ) Figure 4.71: SBR DO and pH-time profdes for NOME-3 and NOME-3b nitrite spike experiments; perturbation = nitrite spikes 219 3.0 2.5 ? 2 . 0 cn E.1.5 O 1.0 z 0.5 0.0 K 1 Baseline f\ Perturbs tion I ' \ I i 4 i 4r 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 nitrite spike a) Nitrite (NOME-3 ) 250 f? 200 a n 150 100 50 0 I I Baseline — • — Perturbation 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 b) Off-gas N 2 0 (NOME-3 ) nitrite spikes c) Nitrite (N0ME-3D) 900 800 700 600 500 400 300 200 100 0 60 80 100 120 Elapsed Time (min) 180 d) Off-gas N 2 0 ( N 0 M E - 3 b ) Figure 4.72: SBR nitrite and off-gas N 2 0 concentration-time profiles for NOME-3 and NOME-3 b nitrite spike experiments; perturbation = nitrite spikes 220 9.E-05 • 8.E-05 • 7.E-05 • z 6.E-05 at 5.E-05 4.E-05 -o z 3.E-05 X 2.E-05 1.E-05 -O.E+00 r / k—Bas eline — • — Perturbation I I I a) N O M E - 3 6.E-04 5.E-04 2 4.E-04 £ 3.E-04 i 2.E-04 x 1.E-04 O.E+00 20 40 60 80 100 120 Elapsed Time (min) 140 160 180 —A—Baseline — • — Perturbation ; / \ / / X > \ / 60 80 100 120 140 160 180 Elapsed Time (min) b) N O M E - 3 b Figure 4.73: SBR nitrous acid concentration-time profdes for a) NOME-3 and b) NOME-3b nitrite spike experiments; perturbation = nitrite spikes 221 ppm) than the peak concentration following the first spike, even though the peak nitrite concentrations induced by the nitrite spikes were very similar in magnitude (Figure 4.72c). The nitrite "conversion" (i.e. removal) rate was clearly slower following the second spike, compared to the period after the first spike. The resulting higher nitrite concentrations, extended over time, may explain the increased peak N 2 0 concentration. • Figure 4.73b shows that the peak nitrous acid concentration following the second nitrite spike was more than 3 times larger than the peak nitrous acid concentration that followed the first nitrite spike. Nitrite Loss/N20 Recovery The NOME-3 nitrite spike increased the mass of generated N 20, relative to the baseline cycle, by about 130%. Approximately 30% of the added nitrite mass was recovered as N 20, with an additional 40% recovered as nitrate. The two NOME-3b nitrite spikes increased the generated N 2 0 mass by about 85%, relative to the baseline cycle. About 69% of the added nitrite was recovered as N 20, with another 13% recovered as nitrate. Discussion The nitrite spike experiments clearly highlighted the ability of the AOB to utilize exogenously supplied nitrite and reduce it to N 20. As discussed in Section 4.4, this type of response was almost certainly due to the action of AOB, as opposed to simply the chemical decomposition of nitrite to N 20. Furthermore, the AOB were shown to effectively compete with the NOB for the exogenous nitrite. At the time the NOME-3b experiment was conducted, the biomass was generating considerably more N 2 0 per baseline cycle than at the time when the NOME-3 experiment was conducted (i.e. 47 mg N versus 7.4 mg N). Therefore, for the given biomass populations, the AOB had an increased competitive advantage for utilizing nitrite relative to the NOB during the NOME-3b experiment. The larger fraction of exogenous nitrite mass recovered as N 2 0 in the NOME-3b experiment, compared to the NOME-3 experiment, confirmed this phenomenon. The biomass response to the exogenous nitrite was similar to some, but not all, pure culture research results. Anderson et al (1993) found that increasing the culture media nitrite concentration from 1 mM 222 (i.e. 14 mg N/L) to 20 mM (i.e. 280 mg N/L) significantly increased the Nitrosomonas europaea N 2 0 generation rate, when cultured under low (i.e. 0.5% 02) and higher (i.e. 5% 02) oxygen conditions. Alternately, Hynes and Knowles (1984) stated that their experiments with pure culture Nitrosomonas europaea cell suspensions found that the "same amount of N 20" was formed when the cells were aerobically cultured with no exogenous nitrite or with 2 mM of nitrite (i.e. 28 mg N/L). Similarly, Ritchie and Nicholas (1972) found that an exogenous nitrite concentration of 20 mM (i.e. 280 mg N/L) did not increase N 2 0 generation in pure culture Nitrosomonas europaea cell suspensions, relative to the case where no exogenous nitrite was added to the culture. The different AOB response to exogenous nitrite, with respect to N 2 0 generation, between the pure culture studies and those conducted for this research may be related to the influence of a mixed microbial population in the SBR biomass. In the SBR, NOB were competing with AOB for available nitrite, with the resultant effect of determining the bulk solution nitrite concentration. Considering a simple case where the system is in steady-state (i.e. constant nitrite concentration), the bulk solution nitrite concentration establishes the nitrite gradient across the AOB cell wall (i.e. bulk solution concentration < periplasmic space concentration) and the NOB cell wall (i.e. bulk solution concentration > periplasmic space concentration). Supply of exogenous nitrite to the SBR increases the bulk solution nitrite concentration, inducing several hypothesized, and related, effects: • The nitrite gradient across the AOB cell wall decreases, slowing the rate of passive nitrite transport from the cell into the bulk solution. • The nitrite concentration inside the AOB periplasmic space increases, due to exogenous nitrite diffusing into the cell and/or the accumulation of HAO-generated nitrite due to the aforementioned reduced passive transport rate. • The active enzyme sites are not saturated with their substrates under normal cellular conditions (Tortora et al, 1989), and the enzymatic reaction rate increases hyperbolically, to a maximum rate, with increasing substrate concentration (White et al, 1978). Therefore, the increase in internal cellular nitrite concentration could increase the relative saturation of the the nitrite reductase (NiR) enzyme, thus increasing the N 2 0 generation rate. Examination of mixed population studies also provides some contradictory results regarding the 223 hypothesized mechanisms. The HAO enzyme, responsible for catalyzing the oxidation of hydroxylamine to nitrite, and the NiR enzyme, that allows the reduction of nitrite to N 20, are believed to be located in the AOB periplasmic space between the cell wall and cytoplasmic membrane (Bock and Wagner, 2001). This close physical association of the AOB enzymes prompted Hynes and Knowles (1984) to doubt that NOB, with their cytoplasmic membrane-bound nitrite oxidoreductase (NoR) enzyme (Bock and Wagner, 2001), could out compete the AOB NiR for AOB HAO-generated nitrite. This hypothesis was supported by their experimental results, where they investigated Nitrosomonas europaea N 2 0 generation when grown in the presence of Nitrobacter winogradskyi, in a mixed population limited to these organisms. They found no difference in the Nitrosomonas europaea N 2 0 generation rate in the presence or absence of Nitrobacter winogradskyi, over a wide range of oxygen concentrations for their experimental conditions. Alternately, Kester et al (1997) found that co-culturing Nitrosomonas europaea and Nitrobacter winogradskyi resulted in "strongly reduced ... N 2 0 emissions" compared to the case where Nitrosomonas europaea was independently cultured under the same conditions. The researchers concluded that "the low nitrite concentration in the mixed culture", resulting from NOB oxidation to nitrate, probably reduced the N 2 0 generation rate. The Kester et al (1997) results demonstrated the potential for the AOB to generate more N 2 0 in the presence of presumably higher intracellular nitrite concentrations. By extension, the addition of exogenous nitrite should also increase the N 2 0 generation rate up to the point where the nitrite limitation has been removed. Fluorescent in-situ hybridization (FlSH) analyses of mixed liquor samples obtained from reactor systems, similar to those used in the current research, showed that AOB and NOB colonies grew in close physical proximity to one another (e.g Simm et al, 2004c). However, even in the extreme case where there was no measurable bulk solution nitrite due to the "direct" movement of nitrite from AOB to NOB through their immediately adjacent cell walls, the NOB "nitrite demand" would still influence the AOB intracellular nitrite concentration and the N 2 0 generation potential. The current research findings, and the hypothesized theoretical framework, are consistent with the Kester et al (1997) and Anderson et al (1993) results. Furthermore, the findings of the nitrite spike experiments corroborate those of the aeration rate experiments (Section 4.7). Specifically, a situation (i.e. increased aeration rate) that provides a competitive kinetic advantage of the AOB over NOB, and thus results in a higher concentration of nitrite to accumulate earlier in the aerobic-phase, will cause N 2 0 to be generated 224 earlier in the aerobic-phase and at a higher rate. The mechanism providing this effect would be the same as that described for the exogenous supply of nitrite, but the increased bulk solution and AOB intracellular nitrite concentrations would be the result of additional HAO-generated nitrite. The NOME-3 data indicated that the N 2 0 generation response was more obviously correlated with nitrite, rather than nitrous acid, concentrations. Alternately, the similarity of the perturbation cycle nitrite, nitrous acid and N 2 0 profdes for the NOME-3b experiment precluded the identification of the relative importance of nitrite versus nitrous acid in the N 2 0 generation response. Section 4.8.2 further investigates the issue of nitrite proper versus nitrous acid. 4.8.2 pH (Nitrous Acid) This section presents and discusses the results of experiments used to investigate the effect that reactor nitrous acid levels had on N 2 0 generation. Experimental Design The pH (nitrous acid) experiments were conducted using two back-to-back SBR cycles, with the SBR fed the same baseline synthetic wastewater during the baseline and perturbation cycles. The SBR was operated with a 1,040 mL/min aerobic-phase aeration rate during both cycles. Two pH (nitrous acid) experiments were conducted for this research. During the perturbation cycle of the first experiment (NOME5-b), the pH was maintained at approximately 7.2 from the start of the aerobic-phase of the cycle. This pH level was chosen as it was the typical level reached at the end of the baseline cycle prior to DO breakthrough. The pH was manually controlled using discreet additions of 0.1 N HC1, supplied to the SBR at elapsed times = 12 min and 31 min. During the second experiment (NOME-5c), the perturbation cycle pH was maintained at approximately 7.9 from the start of the aerobic-phase until elapsed time = 120 min. This pH level was chosen at it was close to the typical level, at the time, of the reactor pH at the end of the anoxic-phase of the cycle. The pH was manually controlled using discreet additions of 0.1 N HC1, supplied to the SBR at typically 5 min intervals up until elapsed time = 120 min. Unfortunately, the electrical signal received by the data logging system from the in-situ pH probe was erratic on the day NOME-5c was conducted. Therefore, 225 mixed liquor samples were manually withdrawn from the SBR for off-line pH measurement, with the mixed liquor then immediately returned to the SBR. General SBR Response - Low pH (nitrous acid) Experiment Figure 4.74 illustrates the pH, DO, ammonia and nitrate profdes for the NOME-5b experiment. The data clearly indicate that the SBR response to the reduced aerobic-phase pH was negligible in terms of DO, ammonia and nitrate levels. The baseline cycle average ammonia oxidation rate was 11.9 mg N/L/hr, almost identical to the 11.7 mg N/L/hr rate for the perturbation cycle. Nitrite, Nitrous Acid and Off-Gas N20 Profiles - Low pH (nitrous acid) Experiment As shown in Figure 4.75, the perturbation cycle aerobic-phase nitrite levels were slightly lower than the baseline cycle levels. The difference in nitrite concentrations between the cycles was consistently around 0.1 mg N/L between elapsed times = 45 and 120 min. The reverse situation was evident for the calculated nitrous acid concentrations (Figure 4.75). Up until about elapsed time = 105 min, the lower perturbation cycle pH shifted the nitrite-nitrous acid equilibrium in favour of nitrous acid, such that the perturbation cycle nitrous acid concentrations exceeded those of the baseline cycle. The difference in nitrous acid concentrations was most significant during the early portion of the aerobic-phase. The off-gas N 2 0 data (Figure 4.75) show a subtle, but distinct, difference in the cycle profiles. Although not shown, the headspace N 2 0 data illustrate the same trend. Close examination of the off-gas and headspace N 2 0 data shows that measurable N 2 0 levels were recorded at around elapsed time = 60 min during the baseline cycle, with comparable levels measured at elapsed time = 45 min during the perturbation cycle. At these times, the baseline and perturbation cycle nitrite concentrations were 0.5 mg N/L and 0.3 mg N/L, with the perturbation cycle concentration 40% lower than the baseline cycle concentration. However, the nitrous acid concentrations at the these times were 4.3 x 10"5 mg N/L (baseline cycle) and 4.4 x 10"5 mg N/L (perturbation cycle), within 2% of each other. General SBR Response - High pH (nitrous