Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A respirometric investigation of the activated sludge treatment of BKME during steady state and transient… Helle, Steve 1999

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

Item Metadata

Download

Media
831-ubc_1999-38893X.pdf [ 46.37MB ]
Metadata
JSON: 831-1.0058987.json
JSON-LD: 831-1.0058987-ld.json
RDF/XML (Pretty): 831-1.0058987-rdf.xml
RDF/JSON: 831-1.0058987-rdf.json
Turtle: 831-1.0058987-turtle.txt
N-Triples: 831-1.0058987-rdf-ntriples.txt
Original Record: 831-1.0058987-source.json
Full Text
831-1.0058987-fulltext.txt
Citation
831-1.0058987.ris

Full Text

A Respirometric Investigation of the Activated Sludge Treatment of BKME During Steady State and Transient Operating Conditions B Y S T E V E H E L L E B. Eng., McGi l l University, 1989 M. Eng., McGi l l University, 1990 A thesis submitted in partial fullf i l lment of the requirements for the degree of Doctor of Philosophy in The Faculty of Graduate Studies Department of Chemical Engineering We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A A P R I L 2 8 ™ , I 9 9 9 © Steve Helle, 1999 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of CAe^ T? i Ce ^ j r)g&n fl<j The University of British Columbia Vancouver, Canada Date /W/ DE-6 (2/88) 11 Abstract Activated sludge is commonly used to treat bleached kraft pulp mill effluent (BKME). Treatment performance during steady state operation is acceptable, but may be less than adequate during periods of unsteady state operation, such as process spills or temperature changes. Nineteen batches of B K M E were treated in two lab scale activated sludge bioreactors. The effects of the solids retention time (SRT) and the presence of an aerobic selector on the activated sludge treatment of B K M E during steady state and transient operating conditions were investigated. Respirometric methods for the study of activated sludge kinetics were investigated in detail, including the unsteady state nature of the assay, and the presence of multiple substrates. B K M E is composed of many different substrates, and the composition was found to be variable from batch to batch. The substrate fractions were classified into two main groups, the readily biodegradable (70%) and the slowly biodegradable (30%). The removal rates of the readily biodegradable substrates were approximately an order of magnitude greater than the removal rates of the slowly biodegradable substrates. The substrate removal rates and the stoichiometry were more dependent upon the wastewater characteristics than the solids retention time or the presence of an aerobic selector. The pH affects the removal rates of all of the substrates in B K M E approximately to the same extent. Below pH 6 and above pH 9, the metabolic activity of the activated sludge was low enough to result in incomplete removal of the slowly biodegradable substrates. Outside of the pH range 5.5 to 9.5, the decay rate of the bacteria increased. Operating at a high SRT (fifteen days) mitigated the negative impact of pH on the biomass. Lll The optimum temperature for activated sludge operation was related to the steady state operating temperature. For biomass acclimated to 35°C, temperature increases to 45°C resulted in poor treatment performance, due to the increased microbial decay rate. The biomass was capable of partially adapting to the higher operating temperature within a few days. Temperature increases above 42°C resulted in energy spilling during the metabolism of methanol and. formic acid, and very low yields on these substrates. iv Table of Contents Abstract • ii Table of Contents i v List of Tables viii List of Figures ix Dedication xviii Chapter 1 Introduction & Objectives 1 Introduction • 1 Objectives '• 3 Outline 4 Chapter 2 Literature Review - Activated Sludge Operation 7 2.1 B K M E Characteristics 7 2.2 Activated Sludge Treatment of B K M E 12 General B K M E Treatment 12 Effect of SRT and HRT on Treatment Performance 16 Selectors 20 Temperature 23 pH 25 2.3 Microbiological Aspects of Wastewater Treatment 26 Substrate Metabolism 27 Variability of Yields and Metabolism : 27 Uncoupled Metabolism 30 Endogenous metabolism 31 2.3 Effect of Transient Operating Conditions on Activated Sludge Performance '. 33 BOD 33 Temperature and pH Shocks 34 Toxic Compounds 34 Operating Conditions 38 2.4 Summary 39 Chapter 3 Literature Review - Activated Sludge Models 41 3.1 Introduction • • 41 3.2Monod 43 Alternatives to Monod 45 3.3 Multiple Substrates 52 Substrate Interactions 53 Multicomponent Kinetics 55 IAWQ Model 59 3.4 Mass Transfer 63 Adsorption ....64 External & Internal Diffusion 67 Oxygen Transfer 70 3.5 Temperature 71 3.6 pH 72 V 3.7 Summary 73 Chapter 4 Literature Review - Model Parameter Measurements 75 4.1 Introduction • 75 4.2 Continuous/Pilot Plant Setups 78 4.3 Fed Batch Test 79 4.4 Batch Tests 80 4.5 Respirometry •. 81 Introduction 81 Low F / M Respirometric Methods 86 AOUR and similar assay techniques..... 86 IAWQ and similar assays 92 Measurement of adsorbed substrate 96 Toxicity assessment 98 Online respirometry 100 High F / M Respirometric Methods 101 Biochemical oxygen demand 101 Maximum growth rate measurements 104 Correlation between OUR and viability 105 4.6 Yield 107 4.7 Decay '. 108 4.8 Summary 109 Chapter 5 Materials & Methods I l l 5.1 Lab Scale Activated Sludge Units I l l 5.2 Standard Tests 115 5.3 Batch Tests , 115 Wastewater Characteristics ....115 Infinite Dilution 116 Decay ....117 Temperature and pH : 117 5.4 Respirometric Method 118 AOUR Procedure 118 Data Analysis 120 Chapter 6 Respirometry 122 6.1 Respirometric Data Analysis 122 Typical Respirometric Data.... 122 Curve Fitting 135 Effect of Biomass Concentration 150 Verification of Models 163 Low Substrate Concentration 164 High Substrate Concentration 167 Summary 171 6.2 Effect of DO Concentration 172 6.3 Multi-Substrate Wastewaters 179 Methanol - Acetic Acid Mixtures 188 Formic acid - Acetic Acid Mixtures 197 Methanol - Formic Acid Mixtures 202 vi Methanol - Acetic Acid - Formic Acid Mixtures 210 Summary • 214 Chapter 7 B K M E Kinetics 216 7.1 Operating Data.... 216 7.2 Measurement of B K M E Kinetics 220 Discrepancies Between Kinetic Tests 220 Evidence of Multiple Substrates 227 Wastewater Fractions 231 Discussion of Significance of Findings 239 7.3 Variability of Kinetics and Wastewater Composition '. 247 Batch Tests 247 Caustic Extraction Effluent 264 Slowly Biodegradable Kinetics 267 7.4 Operating Conditions & Adaptation 268 AOUR Kinetics Variability 269 Variability of Single Substrate Kinetics. 273 Effect of SRT / Selector on Kinetics... 279 Adaptation to Different Batches of Wastewater 283 7.5 Yield & Decay 288 Microbial yield • 288 Decay 291 Chapter 8 Effect of pH and Temperature Shocks on B K M E Kinetics 305 8.1 Effect of pH on Kinetics 305 Readily Biodegradable Kinetics vs. pH 305 Slowly Biodegradable Kinetics vs. pH 313 Microbial Decay vs. pH : ,. 316 8.2 Effect of Temperature on Respirometric Kinetics 327 Effect of Temperature on Methanol Respirometry 327 Effect of Temperature on Formic Acid Respirometry 340 Effect of Temperature on Acetic Acid Respirometry 344 Summary 347 8.3 Adaptation to Shock Temperatures 352 Temperature Shock #1 352 Temperature Shock #2 356 Temperature Shock #3 364 Summary • 366 8.4 Incorporation of Temperature and pH into the Activated Sludge Model 368 Chapter 9 Conclusions 373 Respirometry 373 B K M E Treatment 374 Response to Transient Operating Conditions. 376 Chapter 10 Recommendations for Future Work 379 Activated Sludge Kinetics and Stoichiometry 379 Respirometry 380 Shock Loads 381 Chapter 11 Nomenclature 383 Chapter 12 Bibliography viii List of Tables Table 2.1 B K M E Composition and Treatment 11 Table 4.1 Effect of f/m Ratio 77 Table 4.2 Respirometric Methods 84 Table 5.1 Experimental Conditions 114 Table 6.1 AOUR Parameters 142 Table 6.3 Respirometric Analysis of Substrate Mixtures 215 Table 7.1 Reactor Performance 220 Table 7.2 Summary of Kinetic Constants of the Various Substrate Groups in B K M E -237 Table 8.1 Average Activation Energies 349 ix List of Figures Figure 3.1 Growth rate vs. substrate concentration, Monod, Blackman, and Powell relationships 46 Figure 3.2 Growth rate vs. substrate concentration, Monod and Moser relationships 46 Figure 3.3 Growth rate vs. substrate concentration, Monod, Tessier, and Konak relationships 49 Figure 3.4 Multisubstrate removal during a batch test 49 Figure 3.5 COD composition of domestic wastewater before and after activated sludge treatment 60 Figure 3.6 a) IAWQ death regeneration model; b) I A W Q endogenous decay model 60 Figure 3.7 Effect of external mass transfer on the growth rate vs. substrate concentration relationship 65 Figure 3.8 Effect of internal mass transfer resistance on the growth rate vs. substrate concentration relationship 65 Figure 4.1 Interpretation of respirometric data 88 Figure 4.2 IAWQ batch test 95 Figure 4.3 Growth test 95 Figure 6.1.1 Oxygen consumption following methanol injections 123 Figure 6.1.2 OUR following methanol injections 123 Figure 6.1.3 AOUR vs. methanol concentration 125 Figure 6.1.4 AOUR vs. methanol concentration 125 Figure 6.1.5 AOUR standard deviation vs. methanol concentration 127 Figure 6.1.6 Oxygen consumption vs. methanol concentration 127 Figure 6.1.7 Respirometric yield vs. methanol concentration 128 Figure 6.1.8 Respirometric yield vs. methanol concentration 128 Figure 6.1.9 OC standard deviation vs. average and respirometric yield standard deviation vs. average 130 Figure 6.1.10 OUR vs. methanol addition rate 130 Figure 6.1.11 Respirometric yield vs. SUR 132 Figure 6.1.12 SUR vs. methanol concentration 134 Figure 6.1.13 AOUR vs. formic acid, acetate, and methanol concentration ....134 Figure 6.1.14 OC vs. formic acid, acetate, and methanol concentration .....136 Figure 6.1.15 SUR vs. formic acid, acetate, and methanol concentration 136 Figure 6.1.16 AOUR vs. methanol concentration with Monod, Powell, and Blackman curve fits 137 Figure 6.1.17 AOUR vs. methanol concentration with Monod, Tessier, Konak, and Moser curve fits 137 Figure 6.1.18 AOUR residual vs. expected value for Monod, Powell, and Blackman curve fits 139 Figure 6.1.19 AOUR residual vs. expected value for Monod, Moser, Konak, and Tessier curve fits 139 Figure 6.1.20 AOUR residual vs. expected value for all curve fits 141 X Figure 6.1.21 Monod parameters 95% confidence intervals 143 Figure 6.1.22 Powell parameters 95% confidence intervals 144 Figure 6.1.23 AOUR vs. methanol concentration, with Monod and Powell curve fits : 146 Figure 6.1.24 Residuals from Monod and Powell curve fits vs. experiment # 146 Figure 6.1.25 AOUR vs. acetate concentration, with Monod and Powell curve fits 148 Figure 6.1.26 AOUR vs. formic acid concentration, with Monod and Powell curve fits 148 Figure 6.1.27 AOUR vs. substrate concentration for formic acid/methanol/acetate mixtures 149 Figure 6.1.28 AOUR vs. substrate concentration for B K M E 149 Figure 6.1.29 AOUR vs. methanol concentration, for various M L V S S concentrations, Powell and Monod curve fits 151 Figure 6.1.30a AOUR vs. methanol concentration, for various M L V S S concentrations, a) Powell curve fits; b) Monod curve fits 151 Figure 6.1.30b AOUR vs. methanol concentration, for various M L V S S concentrations, a) Powell curve fits; b) Monod curve fits 153 Figure 6.1.31 Half saturation constant, maximum A O U R vs. M L V S S concentration 153 Figure 6.1.32 Monod parameters 95% confidence intervals 154 Figure 6.1.33 Powell coefficients vs. MLVSS concentration 156 Figure 6.1.34 Powell parameters 95% confidence intervals 158 Figure 6.1.35 AOUR vs. methanol concentration, for various M L V S S concentrations, external mass transfer curve fits 160 Figure 6.1.36 AOUR vs. acetate concentration, for various M L V S S concentrations, Powell and Monod curve fits 162 Figure 6.1.37 AOUR vs. formic acid concentration, for various M L V S S concentrations, Powell and Monod curve fits 162 Figure 6.1.38 OUR following methanol injections 165 Figure 6.1.39 Methanol concentration and OUR vs. time for a batch test 168 Figure 6.1.40 Respirometric yield vs. time for a batch test 168 Figure 6.1.41 Methanol concentration, acetate concentration, and OUR vs. time during two batch tests 170 Figure 6.1.42 OUR vs. time during growth on methanol and acetate 170 Figure 6.2.1 OUR vs. DO in the presence of excess substrate 174 Figure 6.2.2 OUR vs. DO in the presence of excess methanol 174 Figure 6.2.3 AOUR vs. DO and substrate based on the Monod model 176 Figure 6.2.4 AOUR vs. DO and substrate based on the Powell model 176 Figure 6.2.5 DO vs. time under different methanol feed rates 177 Figure 6.2.6 OUR vs. flow rate, and OUR vs. DO slope vs. flow rate 177 Figure 6.2.7 OUR vs. DO for replicate methanol injections 178 . Figure 6.2.8 OUR vs. DO for replicate methanol injections 178 Figure 6.2.9 OUR vs. DO for replicate methanol injections 180 Figure 6.2.10 OUR vs. DO for one large methanol injection 180 xi Figure 6.3.1 Simulated OUR vs. time following an injection of two substrates 182 Figure 6.3.2 Simulated SUR and respirometric yield vs. substrate composition 184 Figure 6.3.3 Simulated SUR and respirometric yield vs. substrate composition 184 Figure 6.3.4 OUR vs. time for injections of methanol, acetate and methanol/acetate mixtures 187 Figure 6.3.5 OUR vs. time for injections of methanol, acetate and methanol/acetate mixtures 187 Figure 6.3.6 AOUR vs. substrate concentration for methanol/acetate mixtures 191 Figure 6.3.7 AOUR vs. methanol concentration for methanol/acetate mixtures......191 Figure 6.3.8 OUR vs. time for injections of methanol, acetate and methanol/acetate mixtures 193 Figure 6.3.9 OC vs. substrate concentration for methanol/acetate mixtures 193 Figure 6.3.10 Measured and expected respirometric yield vs. substrate composition for methanol/acetate mixtures 194 Figure 6.3.11 OUR vs. time for injections of methanol, acetate and methanol/acetate mixtures 194 Figure 6.3.12 SUR vs. substrate concentration for methanol/acetate mixtures 196 Figure 6.3.13 SUR, OUR, OC vs. substrate composition for methanol/acetate mixtures 196 Figure 6.3.14 OUR vs. time for injections of formate, acetate and formate/acetate mixtures 198 Figure 6.3.15 AOUR vs. substrate concentration for formate/acetate mixtures 198 Figure 6.3.16 OC vs. substrate concentration for formate/acetate mixtures 199 Figure 6.3.17 SUR vs. substrate concentration for formate/acetate mixtures 199 Figure 6.3.18 OUR vs. time for injections of formate, acetate and formate/acetate mixtures 201 Figure 6.3.19 AOUR vs. substrate concentration for formate/acetate mixtures 201 Figure 6.3.20 SUR vs. substrate concentration for formate/acetate mixtures 203 Figure 6.3.21 SUR, respirometric yield vs. substrate composition for formate/acetate mixtures 203 Figure 6.3.22 AOUR vs. substrate concentration for formate/methanol mixtures 204 Figure 6.3.23 OC vs. substrate concentration for formate/methanol mixtures 204 Figure 6.3.24 SUR vs. substrate concentration for formate/methanol mixtures 206 Figure 6.3.25 OUR vs. time for injections of formate, methanol and formate/methanol mixtures 206 Figure 6.3.26 OUR vs. time for injections of formate, methanol and formate/methanol mixtures : 207 Figure 6.3.27 OUR vs. time for injections of formate, methanol and formate/methanol mixtures 207 Figure 6.3.28 AOUR vs. substrate concentration, for formate/methanol mixtures 209 Figure 6.3.29 SUR, respirometric yield vs. substrate composition for formate/methanol mixtures 209 Figure 6.3.30 OUR vs. time for injections of formate, methanol, acetate, and formate/methanol/acetate mixtures 211 Xll Figure 6.3.31 AOUR vs. substrate concentration for formate/methanol/acetate mixtures 211 Figure 6.3.32 OC vs. substrate concentration for formate/methanol/acetate mixtures 212 Figure 6.3.33 AOUR vs. substrate concentration for formate/methanol/acetate mixtures 212 Figure 6.3.34 OC vs. substrate concentration for formate/methanol/acetate mixtures 213 Figure 7.1.1 M L V S S vs. time over the course of the project 217 Figure 7.1.2 Effluent VSS vs. time over the course of the project 217 Figure 7.1.3 SRT vs. time over the course of the project 218 Figure 7.2.1 AOUR and OC vs. BOD and COD for B K M E 221 Figure 7.2.2 OUR, BOD and COD vs. time during a batch biodegradation test of B K M E 221 Figure 7.2.3 SUR vs. BOD for B K M E measured by three different experimental methods 223 Figure 7.2.4 OUR vs. time for injections of samples withdrawn during a batch biodegradation test of B K M E 223 Figure 7.2.5 AOUR vs. BOD for B K M E samples withdrawn from a batch biodegradation test 229 Figure 7.2.6 OC vs. BOD for B K M E samples withdrawn from a batch biodegradation test 229 Figure 7.2.7 OUR and BOD vs. time during a batch biodegradation test of B K M E 232 Figure 7.2.8 Composition of B K M E 232 Figure 7.2.9 AOUR vs. the corresponding BOD fraction lor B K M E samples withdrawn from a batch biodegradation test 234 Figure 7.2.10 Maximum AOUR and Monod half saturation constant vs. sample time for B K M E samples withdrawn from a batch biodegradation test 234 Figure 7.2.11 Fractional AOUR vs. the corresponding BOD fraction for B K M E samples withdrawn from a batch biodegradation test 236 Figure 7.2.12 Fractional OC vs. the corresponding BOD fraction for B K M E samples withdrawn from a batch biodegradation test 236 Figure 7.2.13 SUR of wastewater fractions vs. the corresponding BOD fraction for B K M E 238 Figure 7.2.14 Treated B K M E BOD of the various substrate fractions vs. activated sludge loading 238 Figure 7.2.15 SUR of the various B K M E substrate fractions vs. BOD 240 Figure 7.2.16 SUR of the various B K M E substrate fractions vs. activated sludge loading 240 Figure 7.2.17 Treated B K M E BOD, assuming two substrate fractions, vs. activated sludge loading 243 Figure 7.2.18 SUR of B K M E , assuming two substrate fractions, vs. BOD 243 Figure 7.2.19 Effect of wastewater composition on treated B K M E BOD, assuming two substrate fractions, vs. activated sludge loading 245 Figure 7.3.1 OUR and BOD vs. time during a batch biodegradation test of B K M E 248 Figure 7.3.2 OUR vs. time for injections of samples withdrawn during a batch biodegradation test of B K M E 248 Figure 7.3.3 AOUR vs. the corresponding BOD fraction for B K M E samples withdrawn from a batch biodegradation test 249 Figure 7.3.4 OC vs. the corresponding BOD fraction for B K M E samples withdrawn from a batch biodegradation test 249 Figure 7.3.5 OUR and BOD vs. time during a batch biodegradation test of B K M E 251 Figure 7.3.6 OUR vs. time for injections of samples withdrawn during a batch biodegradation test of B K M E 251 Figure 7.3.7 AOUR vs. readily biodegradable BOD fraction for B K M E samples withdrawn from a batch biodegradation test 252 Figure 7.3.8 OUR and BOD vs. time during a batch biodegradation test of B K M E 252 Figure 7.3.9 OUR vs. time for injections of samples withdrawn during a batch biodegradation test of B K M E 254 Figure 7.3.10 AOUR and OC vs. readily biodegradable BOD fraction for B K M E samples withdrawn from a batch biodegradation test 254 Figure 7.3.11 OUR and BOD vs. time during a batch biodegradation test of B K M E 256 Figure 7.3.12 15 day BOD curves for samples withdrawn during a batch biodegradation test of B K M E 256 Figure 7.3.13 B O D s , BODv, OC, COD vs. time during a batch biodegradation test of B K M E 257 Figure 7.3.14 OUR vs. time for injections of samples withdrawn during a batch biodegradation test of B K M E 257 Figure 7.3.15 AOUR vs. BOD for B K M E samples withdrawn from a batch biodegradation test 258 Figure 7.3.16 OC vs. the corresponding BOD fraction for B K M E samples withdrawn from a batch biodegradation test 258 Figure 7.3.17 OUR vs. time for a number of batch B K M E biodegradation tests 260 Figure 7.3.18 OUR vs. time for a number of batch B K M E biodegradation tests 260 Figure 7.3.19 BOD vs. time for a number pf batch B K M E biodegradation tests 261 Figure 7.3.20 BOD vs. time for a number of batch B K M E biodegradation tests 261 Figure 7.3.21 OUR vs. time for a batch biodegradation test of caustic extraction effluent 265 Figure 7.3.22 Slowly biodegradable fraction removal rates vs. slowly biodegradable BOD 265 Figure 7.4.1 Maximum AOUR vs. time over the course of the project 270 Figure 7.4.2 OC vs. time over the course of the project 270 xiv Figure 7.4.3 Maximum AOUR and the Monod half saturation constant vs. wastewater batch over the course of the project 271 Figure 7.4.4 OC and BOD vs. wastewater batch over the course of the project 271 Figure 7.4.5 Respirometric OC vs. wastewater BOD 274 Figure 7.4.6 Maximum AOUR obtained with methanol, formate, and acetate, vs. time over the course of the project 274 Figure 7.4.7 Respirometric yield obtained with methanol, formate, and acetate, vs. time over the course of the project 275 Figure 7.4.8 Maximum methanol AOUR vs. wastewater batch over the course of the project 275 Figure 7.4.9 Maximum formate AOUR vs. wastewater batch over the course of the project 276 Figure 7.4.10 Maximum acetate AOUR vs. wastewater batch over the course of the project 276 Figure 7.4.11 Ratio of methanol AOUR to formate A O U R and methanol yield to formate yield vs. time over the course of the project 278 Figure 7.4.12 Maximum AOUR vs. SRT for all of the wastewater batches studied .....278 Figure 7.4.13 Maximum AOUR obtained with methanol, formate, and acetate, vs. SRT for all of the wastewater batches studied 282 Figure 7.4.14 OUR vs. time for injections of B K M E batch L 282 Figure 7.4.15 AOUR vs. BOD for B K M E batch L 285 Figure 7.4.16 OUR vs. time for injections of B K M E batch M 285 Figure 7.4.17 OUR vs. time for injections of B K M E batch N 286 Figure 7.4.18 OUR vs. time for injections of B K M E batch N 286 Figure 7.5.1 Apparent yield vs. SRT 289 Figure 7.5.2 Apparent yield vs. wastewater batch 289 Figure 7.5.3 Decay coefficient vs. wastewater batch 292 Figure 7.5.4 Endogenous OUR and decay coefficient vs. SRT 292 Figure 7.5.5 Endogenous OUR, AOUR obtained with methanol, formate, and acetate, vs. SRT 295 Figure 7.5.6 Endogenous OUR, AOUR obtained with methanol, formate, and acetate, vs. SRT , 295 Figure 7.5.7 OUR, AOUR obtained with methanol and formate, vs. time during a batch decay test, semilog graph 299 Figure 7.5.8 OUR, AOUR obtained with methanol and formate, and M L V S S vs. time during a batch decay test '. 299 Figure 7.5.9 OUR, AOUR obtained with methanol, formate, and acetate vs. time during a batch decay test, semilog graph 301 Figure 7.5.10 OUR, AOUR obtained with methanol, formate, acetate, and M L V S S vs. time during a batch decay test at 35°C 301 Figure 7.5.11 OUR, AOUR obtained with methanol, formate, acetate, and M L V S S vs. time during a batch decay test at 20°C 303 Figure 7.5.12 Apparent yield vs. SRT 303 Figure 8.1.1 AOUR vs. substrate concentration at different pH's 306 XV Figure 8.1.2 Maximum AOUR and Monod half saturation constant vs. pH 306 Figure 8.1.3 AOUR vs. substrate concentration at different pH's 308 Figure 8.1.4 OC vs. substrate concentration at different pH's 308 Figure 8.1.5 Maximum AOUR and Monod half saturation constant vs. pH 309 Figure 8.1.6 Maximum AOUR vs. pH 309 Figure 8.1.7 Maximum AOUR, obtained with methanol, formate, and acetate, vs. pH 310 Figure 8.1.8 Relative maximum AOUR, obtained with methanol, formate, and acetate, vs. pH ...310 Figure 8.1.9 OUR vs. time for injections of methanol at different pH values .312 Figure 8.1.10 Endogenous OUR vs. pH 312 Figure 8.1.11 OUR vs. time for batch B K M E biodegradation tests at pH 8 and at pH 9 314 Figure 8.1.12 BOD vs. time for batch B K M E biodegradation tests at pH 8 and a tpH9 314 Figure 8.1.13 SUR vs. BOD for batch B K M E biodegradation tests at pH 8 and at pH 9 315 Figure 8.1.14 SUR vs. BOD at pH 8 and at pH 9, measured using the fed batch assay 315 Figure 8.1.15 AOUR vs. substrate concentration obtained with biomass exposed to pH 10 .317 Figure 8.1.16 Maximum AOUR and Monod half saturation constant 10 vs. pH, measured using biomass exposed to pH 10 317 Figure 8.1.17 Relative maximum AOUR, obtained with methanol, vs. time exposed to various pH changes 319 Figure 8.1.18 Relative maximum AOUR, obtained with formate, vs. time exposed to various pH changes 319 Figure 8.1.19 Relative maximum AOUR, obtained with acetate, vs. time exposed to various pH changes 320 Figure 8.1.20 Fed batch test data, BOD vs. time after exposure to pH 10 320 Figure 8.1.21 First order AOUR decay coefficient and maximum AOUR vs. pH 322 Figure 8.1.22 Recovery of relative maximum AOUR vs. time after exposure topH 10 322 Figure 8.1.23 Recovery of relative maximum AOUR vs. time after exposure to pH 10 323 Figure 8.1.24 Relative maximum AOUR after exposure to pH 10 vs. biomass concentration 323 Figure 8.1.25 Recovery of relative maximum AOUR obtained with formate vs. time after exposure to pH 10 325 Figure 8.1.26 Recovery of relative maximum AOUR obtained with methanol vs. time after exposure to pH 10 325 Figure 8.1.27 Recovery of relative maximum AOUR obtained with acetate vs. time after exposure to pH 10 326 Figure 8.2.1 OUR vs. time for injections of methanol following a temperature increase from 34°C to 44.5°C 329 xvi Figure 8.2.2 OUR vs. time for injections of methanol following a temperature increase from 34°C to 44.5°C, and then a decrease to 34°C 329 Figure 8.2.3 Maximum AOUR, maximum SUR, and respirometric yield, obtained with methanol vs. time from temperature increase from 34°C to44.5°C 331 Figure 8.2.4 OUR vs. time for injections of methanol following a temperature decrease from 34°C to 27°C 331 Figure 8.2.5 OUR vs. time for injections of methanol following a temperature increase from 34°C to 42.4°C 332 Figure 8.2.6 OUR vs. time for injections of methanol following a temperature increase from 34°C to 49°C 332 Figure 8.2.7 Relative maximum AOUR, obtained with methanol, vs. time f rom temperature change for various temperature changes 334 Figure 8.2.8 Respirometric yield, obtained with methanol, vs. time from temperature change for various temperature changes 334 Figure 8.2.9 Maximum SUR, obtained with methanol, vs. time from temperature change for various temperature changes ....335 Figure 8.2.10 Maximum SUR, obtained with methanol, vs. time from temperature change for various temperature changes, on a semi-log graph 335 Figure 8.2.11 Methanol SUR decay coefficients vs. inverse temperature 337 Figure 8.2.12 Methanol maximum SUR vs. inverse temperature 337 Figure 8.2.13 Methanol maximum SUR vs. temperature 338 Figure 8.2.14 Active fraction vs. temperature ' 338 Figure 8.2.15 Methanol SUR decay rate and respirometric yield vs. temperature. 339 Figure 8.2.16 Effect of rapid temperature adjustment on maximum AOUR, maximum SUR, and respirometric yield 339 Figure 8.2.17 Maximum methanol AOUR vs. temperature 341 Figure 8.2.18 Relative maximum methanol AOUR vs. temperature 341 Figure 8.2.19 OUR vs. time for injections of formate following a temperature increase from 34°C to 54°C 342 Figure 8.2.20 Relative maximum AOUR, obtained with formate, vs. time from temperature change for various temperature changes 342 Figure 8.2.21 Respirometric yield, obtained with formate, vs. time from temperature change for various temperature changes 343 Figure 8.2.22 Maximum SUR, obtained with formate, vs. time from temperature change for various temperature changes 343 Figure 8.2.23 Formate SUR decay rate and respirometric yield vs. temperature 345 Figure 8.2.24 Formate maximum AOUR vs. temperature 345 Figure 8.2.25 OUR vs. time for injections of acetate following a temperature increase from 34°C to 45°C 346 Figure 8.2.26 Maximum AOUR, maximum SUR, and respirometric yield, obtained with acetate, vs. time from temperature increase from 34°C to 45°C 346 XVII Figure 8.2.27 Maximum acetate AOUR vs. temperature 348 Figure 8.2.28 Maximum methanol AOUR, maximum formate AOUR, and maximum acetate AOUR, vs. temperature : 348 Figure 8.2.29 Endogenous OUR vs. temperature 350 Figure 8.2.30 Endogenous OUR vs. time from temperature change for various temperature changes 350 Figure 8.2.31 Endogenous OUR, maximum methanol AOUR, and maximum formate AOUR vs. time from temperature change from 34°C to 49°C 351 Figure 8.2.32 Endogenous OUR, maximum methanol AOUR, and mass transfer limited methanol AOUR vs. time from temperature change from 34°Cto44.5°C 351 Figure 8.2.33 Methanol AOUR vs. substrate concentration at 25°C, 35°C, and 40°C 353 Figure 8.3.1 Maximum AOUR and endogenous OUR vs. time following a temperature increase from 34°C to 44°C 355 Figure 8.3.2 M L V S S and treated effluent BOD vs. time following a temperature increase from 34°C to 44°C 355 Figure 8.3.3 Maximum AOUR measured at 35°C, 4 5 ° C , and 25°C, vs. time following a temperature increase from 34°C to 44°C 357 Figure 8.3.4 Maximum AOUR vs. temperature before and after a temperature increase from 34°C to 44°C 357 Figure 8.3.5 M L V S S and apparent yield vs. time in a control reactor, and a temperature shocked reactor 359 Figure 8.3.6 Treated BOD and COD vs. time in a control reactor, and a temperature shocked reactor 359 Figure 8.3/7 Maximum AOUR vs. time in a control reactor, and a temperature shocked reactor : 361 Figure 8.3.8 Respirometric yield vs. time in a control reactor, and a temperature shocked reactor 361 Figure 8.3.9 Maximum AOUR vs. temperature using control biomass and biomass exposed to a 10°C temperature increase 363 Figure 8.3.10 Relative maximum AOUR vs. temperature using control biomass and biomass exposed to a 10°C temperature increase 363 Figure 8.3.11 AOUR vs. substrate concentration using control biomass and biomass exposed to a 10°C temperature increase 365 Figure 8.3.12 Maximum AOUR vs. temperature for biomass acclimated to 35°C and biomass acclimated to 27°C 365 Figure 8.3.13 Relative maximum AOUR vs. temperature for biomass acclimated to 35°C and biomass acclimated to 2 7 ° C 367 Figure 8.3.14 Respirometric coefficients vs. time during a temperature decrease to 27°C 367 Figure 8.3.15 Apparent yield vs. time during a temperature decrease to 27°C 369 Figure 8.3.16 Optimum temperature vs. operating temperature 369 Figure 8.4.1 AOUR vs. pH, at various temperatures 371 Figure 8.4.1a AOUR vs. pH, at various temperatures 371 I xviii Dedication This thesis is dedicated to Lucky - who understood the meaning of life. 1 Chapter 1 Introduction & Objectives Introduction Large amounts of water are used in the production of bleached pulp. Due to residual chlorine, it is difficult to recycle the effluent from the bleaching process, and this waste stream, along with others, is discharged to the environment. The combined bleached kraft mill effluent (BKME) contains a large number of organic compounds and toxic compounds. If the wastewater is not treated there is potential for lethal toxic effects on the ecosystem in the immediate vicinity of the outfall. If the wastewater is well mixed and diluted during discharge to be below the acute toxicity threshold in the receiving water, there may still be sublethal toxic effects. In addition, the oxygen demand of the wastewater may lead to eutrophication of the receiving water, especially i f nutrients are present. In order to minimise these problems, B K M E is treated prior to discharge to the environment. Many pulp mills have recently installed activated sludge wastewater treatment plants. In the activated sludge process, a consortium of microorganisms grow on the organic compounds in the wastewater. The biomass growth rate is controlled by wasting excess biomass, which must be disposed of. By changing the growth rate (or solids retention time (SRT)), or adding a selector, different microbial populations may be selected. Different populations may lead to different treatment characteristics. When both the pulp mill and the treatment plant are operating at steady state, the activated sludge process is adequate for removing the easily biodegradable organic compounds and the acute toxicity from B K M E . Pulp mills do not always operate at 2 steady state due to changes in the wood source, and the bleaching process. These changes will have effects on the wastewater composition and subsequently may impact the wastewater treatment. More severe impacts on the treatment system are possible if process spills occur during the pulping or bleaching operations. For example, i f there is a black liquor spill, a large amount of caustic material and organic matter will be sent to the treatment plant. If the spill is large, it may be beyond the buffering capacity of the pH control system, and the activated sludge will be exposed to a.pH shock. If the black liquor spill is neutralised before reaching the wastewater treatment system, the activated sludge will still be exposed to an organic shock load. B K M E is typically too hot for biological wastewater treatment, and must be cooled. There can occasionally be problems with a pulp mill's effluent cooling system, especially in the summer months, and the temperature in the activated sludge unit may rapidly increase beyond the optimal, or acceptable, temperature range for the activated sludge biomass. The goal of this project was to study the effects of rapid temperature and pH changes on the activated sludge treatment of B K M E . The tools chosen to quantify the activated sludge performance were the measurement of organic removal rates from the wastewater, measurement of the rate of decay of the biomass, and the measurement ofthe stoichiometry of the process. A survey of the literature revealed that activated sludge growth rates, and/or organic removal rates, were most often measured using respirometry, so this was the method chosen. As the project progressed, some shortcomings and 3 invalid assumptions of the respirometric method slowly became evident. This led to a full investigation of respirometry. Concurrently to the investigation of respirometry, the appropriate models to use to -characterise B K M E and the activated sludge treatment of B K M E were determined. The kinetics of B K M E treatment by activated sludge has received limited investigation. The majority of previous studies have used older activated sludge models, ignoring recent findings obtained from municipal wastewater treatment investigations. In this study, these newer models were adapted to apply to the specific case of B K M E , which is different from typical domestic sewage. The activated sludge model and the respirometric method were used to study the effects of pH and temperature on the treatment process. Experiments performed later in the study were done in more detail, as more was known about respirometric methods and the nature of B K M E . Objectives Hypothesis: The stoichiometry and rates of substrate metabolism for the activated sludge treatment of B K M E are a complex function of: effluent characteristics, environmental conditions (pH, temperature), and operating conditions (SRT, aerobic selector). In order to verify this hypothesis, the following objectives for this project were set: • To determine the effect of operating conditions and effluent characteristics on the activated sludge treatment of B K M E . In particular, to investigate the effects of varying the SRT from 5 to 20 days, adding an aerobic selector, and wastewater composition. 4 • To evaluate methods for measuring activated sludge organic removal rates, decay rates, and stoichiometry, focusing on respirometric methods. • To determine the appropriate model to use for the activated sludge treatment of B K M E . • To determine the effects of transient operating conditions on the activated sludge treatment of B K M E , such as changing wastewater composition, organic concentration, operating temperature, and operating pH. Outline Following is a brief outline of this thesis: chapters 2 through 4 give background information, which is relevant to the topic, and / or the results. Chapters 6 through 8 give the results of this project. Each builds upon the results of the preceding chapters. Chapter 2 discusses the characteristics of B K M E and B K M E treatment. In addition, the effect of transient operating conditions on activated sludge (organic concentration, temperature, and pH), and the variability of treatment results are discussed. Since only a small amount of research on the effects of transient operating conditions on B K M E treatment is available in the literature, this review will mainly discuss municipal treatment results. This discussion is relevant for chapters 7 (all of chapter 2) and 8 (sections 2.3 and 2.4). Chapter 3 discusses activated sludge models, the effects of multiple substrates on data interpretation, and the incorporation of mass transfer, pH, and temperature, into activated sludge models. This discussion is relevant for chapters 6 (3.2, 3.4), 7 (3.3), and 8 (3.5, 3.6). 5 Chapter 4 discusses various assays for measuring activated sludge growth rates and organic removal rates, focusing mainly on respirometric methods. Many different respirometric methods have been developed, however most are very similar, and most make the same assumptions. The discussion of respirometric methods has been divided into two sections - the methods which employ a low food to microorganism ratio, and those which employ a high food to microorganism ratio, (relevant for chapter 6 and 7). Chapter 5 presents the materials and methods used in obtaining the results presented in chapters 6 through 8. Chapter 6 is a detailed investigation of the respirometric method, focusing on the measurement of the removal rates of methanol, formic acid, and acetate (known components of B K M E ) . A variety of models are investigated, including ones which contain mass transfer effects. The assumption of pseudo-steady state is discussed. Also investigated were the effects of dissolved oxygen (DO) concentration on respirometry, and interactions among substrates when more than one substrate is present. This last result is important for the interpretation of B K M E respirometric data, since B K M E is a mixture of many substrates. Chapter 7 presents the substrate removal rate data obtained using B K M E . Major points presented in this chapter are the importance of multiple components in the wastewater for the interpretation of respirometric data, the variability of the removal rates, the variability of the wastewater composition, the effect of operating conditions on the treatment performance, and the adaptation of the biomass to different batches of wastewater. The significance of these findings on activated sludge modeling is discussed. Finally, since results from the measurement of activated sludge yield and 6 decay do not agree with standard interpretations, possible alternate interpretations of the data are presented. Chapter 8 discusses the effects of pH and temperature on the performance of the activated sludge process. The effects of sudden temperature changes and sudden pH changes on biomass acclimated to pH 8 and 35°C were investigated. Two week long temperature increases were also investigated to observe the adaptation ability of the biomass. Chapter 9 ties all of the results together, and summarises the project. 7 Chapter 2 Literature Review - Activated Sludge Operation In this chapter a review of the aspects of activated sludge relevant to this project will be presented. The first subject will be the nature and treatability of BKME. Next, factors, which affect the activated sludge treatment and stoichiometry will be discussed. Among these factors are the solids residence time (SRT), the hydraulic and environmental characteristics of the activated sludge process, shock loads, wastewater variability, and population dynamics. 2.1 BKME Characteristics Bleached kraft mill effluent (BKME) is a complex mixture, containing simple inorganic salts, over 300 known (with many more unknown) low molecular weight organic compounds, and many high molecular weight (>lkDa) organic constituents (Axegard et al 1993). The high molecular weight components of BKME are generally assumed to be chlorolignins (Sagfors and Starck 1988, Dahlman et al. 1993), which are the highly oxidized degradation products originating from the residual lignin in the unbleached pulp, and carbohydrates from hemicelluloses (Axegard et al. 1993). There are many toxic compounds present in BKME, mainly resin and fatty acids, and chlorinated phenols. There are also unidentified toxic compounds (Heimburger et al 1988a). Due to the vast amount of different compounds present, pulp mill effluent is usually characterised by general parameters, such as BOD (biochemical oxygen demand), COD (chemical oxygen demand), TOC (total organic carbon), colour, pH, TSS (total suspended solids), and AOX (adsorbable organic halide). 8 The wastewater from the manufacturing of pulp originates during various aspects of the process. During the kraft pulping process, the lignin is removed from the wood fibres by dissolving it in the cooking chemical solution (white liquor: sodium hydroxide and sodium sulfide). Ninety to ninety-five percent ofthe lignin is removed as well as some ofthe polysaccharides, especially the hemicelluloses. The cooking liquor is then concentrated by evaporation, and the cooking chemicals are recovered in the kraft recovery furnace. The main source of pollution from the pulping process is the condensates generated during concentration ofthe cooking liquors. The condensates are high in methanol (which comprises 80% of the total BOD ofthe condensates) (Kemeny and Baerjee 1997, Barton et al 1998), and also contains other alcohols (mainly ethanol), ketones, phenolic substances, sulfur compounds and terpenes. The condensates contain approximately one third of a pulp mill's BOD load (Springer 1993b). Cooked pulp is washed with water to remove the dissolved organic material and residual lignin from the pulp. This process is called brown stock washing and the wash water is sent to the evaporators. The washing results are improved as the amount of wash water increases, but so is the demand on the evaporators. Black liquor carry-over to the bleach plant will affect the bleach plant effluent quality (Tana and Lehtinen 1996). Older mills often have high washing losses to the sewer; this effluent contains phenolic compounds, organic acids, terpenes, and resin acids (Axegard et al 1993, Kemeny and Baerjee 1997). Another source of BOD from the pulping process comes from process spills, such as black liquor spills (Springer 1993). Black liquor is very rich in BOD (hydroxy acids, 9 formic acid, acetic acid), and contains high concentrations of wood extractives, which are toxic (Sjostrom 1993). To further reduce the organic load to the bleach plant, most mills use extended delignification or oxygen delignification. The organic matter removed from the pulp prior to bleaching is sent to the kraft recovery furnace and not to the treatment plant. The residual lignin in the pulp after the pulping stage is removed in the bleaching process. The bleaching process consists of a chlorination stage (usually with 100% chlorine dioxide substitution), followed by extraction ofthe chlorinated lignins in alkali, then more bleaching with chlorine dioxide and extraction. It is the first 2 stages ofthe bleaching process, the chlorination stage (Cl) and the extraction stage (El) which contain most ofthe pollutants originating in the bleach plant. The effluent from the extraction stage contains organic compounds with a higher molecular weight than the organic compounds in the chlorination stage effluent. The majority of the A O X in spent chlorination liquor has a molecular weight below 10 000 (80%), with 30% being below 1000. Fifty-five percent of the spent alkali liquor has a molecular weight greater than 25 000, with only 5% being below 1000 (Kringstad and Lindstrom 1984). Another study ofthe molecular weight distribution of the A O X in B K M E found 85% to have a molecular weight less than 1000 when measured with non-aqueous size exclusion chromatography (Jokela and Salkinoja-Salonenl992). H M W COD and A O X are mainly composed of strongly oxidised degradation products originating from the residual lignin in the unbleached pulp (Axegard et al.1993), much of which will degrade to L M W compounds in the environment (Wilson and Holeran 1992, Eriksson et al. 1985, Fitzsimmons and Eriksson 1990). 10 The majority of the BOD in the chlorination stage effluent is due to methanol, while formic acid is the major contributor to the extraction stage BOD. L M W acidic compounds identified in B K M E (when the pulp is bleached with chlorine) in quantity are acetic acid, glyceric acid, oxalic acid, malonic acid, succinic acid, and malic acid. Next to methanol, the greatest components of the L M W neutral compounds are the various hemicelluloses. Also present are a wide variety of phenolic compounds, many of which are chlorinated. Many of these compounds are biodegradable, but at slow rates (Kringstad and Lindstrom 1984). When modern bleaching sequences are used, (100% chlorine dioxide substitution, enhanced extraction) methanol is still a major component of the total mill effluent, and alkaline extraction effluent (Kemeny and Banerjee 1997), and both methanol and formic acid are present in significant quantities in the first chlorine dioxide stage effluent (Dahl et al 1998). Totally chlorine free bleach plant effluents also contain significant amounts of monocarboxylic acids, mainly formic acid, glycolic acid, and 3-hydroxypropanoic acid (Ristolainen and Alen 1998). These compounds are formed under both alkaline and acidic bleaching conditions. Reduction in the amount of pollution generated during bleaching is achieved by substituting CIO2 for CI2 in the first chlorination stage, as long as the level of substitution is greater than 50%. High chlorine dioxide substitution and oxygen delignification decrease COD, BOD, A O X , toxicity, colour, and makes the waste water easier to treat by biological methods. Modern bleaching sequences (extended delignification, oxygen delignification, and chlorine dioxide substitution) produce significantly less organic material (measured by BOD, COD, TOC, A O X , toxicity, and colour), and material of lower M W than the older bleaching processes (Axegard et al 1993, Graves et al. 1993, 11 Heimburger et al. 1988a). New bleaching and cooking methods produce less organic pollution, but the ratio of BOD to COD in the effluent remains approximately the same (Graves et al 1993, Heimburger 1988a, 1988b, Saunamaki 1995). Typical characteristics, and treatment efficiency for B K M E are shown in table 2.1. The content of B K M E depends on many factors, including the type of tree being pulped, the age of the wood chips and the specifics of the process (Graves et al. 1993, Servizi and Gordon 1973, Heimburger 1988a, 1988b). Effluent from production of hardwood pulp contains organic material of lower M W , and less chlorinated phenolics, than the effluent from production of the corresponding softwood pulps (Axegard et al. 1993, Heimburger et al. 1988a). More than twice as much A O X , BOD , and COD is generated during the bleaching of softwood pulp than hardwood pulp (Cook 1990, Heimburger et al 1988b). The spread in removal efficiencies is due to the differences in the content of B K M E , as well as the differences in the wastewater treatment plants used. Table 2.1 B K M E Composition and Treatment Parameter Amount per ton of wood pulped Typical Concentration Removal Efficiency BOD 10-30 80 - 95% 2 COD 3 0 - 120 3 30 - 70% TOC 25 - 50% 4 A O X 0 .2-7 1 3 15-70% 1 Colour 20-220 5 ~ 0 Acute toxicity 200-400 5 90- 100% 1 , 2 1 Wilson and Holloran 1992 2 Leach 1976 3 Axegard etal. 1993 4 Jokelaetal. 1993 5 Heimburger et al. 1988a 12 The characteristics of B K M E from any given mill are highly variable, with variations on the order of 100%, as can be seen in the literature (Simpura and Pakarinen 1993, Lo et al. 1994a). These variations were measured with global parameters. The variation of specific compounds in the wastewater is probably even greater. These variations may be due to changing operating conditions, different species of trees being pulped, or differences in the age ofthe wood chips used. In a study of a TMP (thermomechanical pulping) mill, resin acid discharge was greater in the winter months than in the summer months, presumably due to greater resin acid degradation during wood chip storage in the warmer summer months. Variation in B K M E BOD from -50 mg/l to greater than 400 mg/l was observed in an eight month study, with large differences observed from day to day. The large increases in BOD corresponded to black liquor spills. Black liquor spills, and the pulping of green chips, also resulted in effluent high in toxicity (Servizi and Gordon 1973). This variability in effluent quality and strength will be expected to influence the wastewater treatment performance. 2.2 Activated Sludge Treatment of BKME General BKME Treatment The preferred method of B K M E treatment is primary clarification followed by secondary biological wastewater treatment. The most widely used form of biological wastewater treatment for pulp and paper effluents is the aerated stabilisation basin. The advantages of aerated stabilisation basins over activated sludge are much simpler operation, lower costs, resistance to shock loads, and minimal sludge production (Folke and Guerra 1992). The disadvantages are the large land area required and the high 13 suspended solids content in the treated effluent. The minimum detention time is 4 days, with most mills operating at 10. to 15 days. The high COD to BOD ratio of B K M E (table 2.1) implies that it will not be possible to remove all of the organic matter from the wastewater using biological wastewater treatment. Many ofthe new treatment plants are activated sludge designs, used to meet today's stricter environmental regulations. The activated sludge process can produce an effluent very low in BOD and SS, and the treatment efficiency is less variable than that obtained with aerated lagoons (Melcer et al. 1995). Detention times are around 3 to 8 hours, making activated sludge very attractive when there is minimal area available for a treatment plant. Operating conditions are chosen to select for floe forming microorganisms with good settling characteristics. This allows for mixed liquor suspended solids (MLSS) concentrations on the order of 2000 to 5000 mg/l and also results in the production of large amounts of excess sludge, 0.5 to 0.75 kg/kg BOD removed. If pure oxygen is used, the MLSS may be in the range of 5000 to 7000 mg/l, allowing for shorter hydraulic detention times. However there is little difference in performance between air and oxygen activated sludge plants (Springer 1993). One ofthe pollution parameters of concern in B K M E is A O X , although there is no clear evidence linking A O X to environmental effects (Heimburger et al. 1988b, Axegard et al. 1993). In lagoons, A O X removal is hypothesised to occur by sorption onto the biomass and subsequent dechlorination under anaerobic conditions in the benthic layer (Bryant et al. 1987, Bryant and Amy 1989, Chernysh et al. 1992). Activated sludge units are usually reported to achieve greater A O X removal than aerobic lagoons. In the activated sludge process the excess sludge is wasted, and any A O X adsorbed onto the 14 sludge will also be wasted. (Boman et al. 1988). A n alternate explanation is that the higher solids retention time of activated sludge compared to lagoons allows for the growth of biomass which can degrade some of the high molecular weight A O X and COD compounds (Bomann et al. 1988, and Wilson and Holloran 1992). A study of a full scale treatment plant found the main removal mechanism of chlorinated phenols to be sorption into the sludge (Leuenberger et al. 1985). Others have observed minimal adsorption of the A O X onto the biomass (Hall and Randle 1992, Nevalainen et al. 1991, Rempel et al. 1992). Most, i f not all, studies have found that conventional aerobic biological treatment cannot remove all of the A O X in the wastewater. To achieve zero A O X goals, chlorine must be eliminated from the bleaching sequence, or the bleach effluent must be recycled. The other pollution parameters of concern in B K M E are BOD, COD, and toxicity. Secondary treatment removes the majority of the BOD and the acute toxicity from B K M E . Approximately half of the COD is removed. The remaining COD is a measure of the recalcitrant organic compounds in the wastewater (this is demonstrated by the very low BOD to COD ratio of the treated effluent). [Recalcitrant, as used in this study, refers to those compounds which are not degraded during wastewater treatment, and does not mean that these compounds will not be degraded in the environment.] More information on the nature of the compounds that are in the wastewater may be obtained by measuring the molecular weight (MW) distribution. In several studies of the molecular weight distribution of the TOC before and after biological treatment, there was greater removal of the low M W organics, and no significant reduction of the high M W organics (Graves et al. 1993, Lindstrom and Mohamed 1988, Y u 1993). Studies of full scale activated sludge plants treating B K M E found removal of all size classes of 15 A O X from B K M E , with a slightly greater removal of the lower M W compounds (Jokela et al. 1993, Stuthridge et al. 1991). Other full scale studies have found limited removal of high M W A O X (Strang 1992) and high M W TOC (>300 daltons) (Sonnenberg et al. 1995). There is a slight increase in the high molecular weight organics during aerobic treatment, due to microbial product formation (Sonnenberg et al. 1995, Bryant and Amy 1989). Effluent colour is associated with the carbon which is not removed during biological treatment, and is probably composed of lignin residuals (Kemeny and Baerjee 1997). The biological treatment efficiency of B K M E depends upon the pulping and bleaching sequence used at the mill, as well as the wood species being pulped. Oxygen delignification appears to make the effluent more amenable to biological treatment, as well as reducing the total amount of pollutants. Nevalainen et al. (1991) found 40% COD removal, and 22% A O X removal, during activated sludge treatment of conventional effluent. This increased to 50% COD reduction, and 40% A O X reduction, during the treatment of oxygen bleached pulp. There was also a slight increase in BOD removal. In another study, COD removal in the waste treatment plant increased from 27% to 32% with CIO2 substitution, and to 45% with oxygen delignification. BOD removal decreased with C10 2 substitution, from 90 to 87%, and decreased to 80% with oxygen delignification, although the overall amount of BOD in the effluent greatly decreased (by 50%) (Graves etal. 1993). 16 Effect of SRT and HRT on Treatment Performance The ability to recycle the suspended solids in the activated sludge process allows the solids retention time (SRT) to be controlled independently from the hydraulic residence time (HRT). The SRT is a very important variable, for many reasons. The most obvious, and probably the most important, is that the mixed liquor volatile suspended solids (MLVSS) concentration is directly related to the SRT. Sludge settleability is a very important treatment parameter for the activated sludge process, and is influenced by the SRT and other factors. If the SRT is too low, the growth rate of the biomass will be above that required for flocculation, and the sludge will not settle. Operation at high SRT may result in sludge with poor settling qualities. Most activated sludge plants designed for the treatment of B K M E are operated at SRTs between 5 and 10 days. This allows for removal of some of the more slowly biodegradable compounds, while maintaining a sludge with good settling qualities. The growth rate (and consequently the SRT) may have an effect on the biomass characteristics, such as cell morphology and the enzymes expressed (Harder and Dijkhuizen 1983). For some bacteria, at high growth rates enzyme systems with low affinity but high capacity for the carbon source are expressed, while at low growth rates enzyme systems with high affinity, low capacity, are expressed. The microbial population present in the wastewater treatment plant depends on the SRT. As the SRT decreases, bacteria with relatively slow growth rates will wash out of the system, resulting in a lower microbial diversity than i f the SRT were greater. High SRTs are often necessary for the removal ofthe more recalcitrant compounds in the wastewater, which are thought to be removed by microorganisms with low growth rates. 17 The common reasoning for this assumption is that i f the more recalcitrant compounds degrade at low rates, then the bacteria which utilise these substrates grow at low rates. An additional explanation is offered by the fact that as the SRT increases, the specific loading of all ofthe organic compounds decreases. If the recalcitrant compounds are inhibitory, or i f they are only metabolised when other preferential substrates are already utilised, then increasing the SRT will increase the removal rates of these compounds. For difficult to degrade compounds, such as pentachlorophenol, degradation occurs best at high sludge ages (greater than eight days), suggesting catabolic degradation by slow growing specific degraders (Melcer and Bedford 1988, Nyholm et al. 1992, Ettala et al. 1992). Other compounds, such as lindane, have higher removal efficiencies at intermediate sludge ages and high loading rates, implying removal by co-oxidation (Nyholm et al. 1992). In a study of phenol biodegradation, at a range of SRTs (3 to 14 days), the specific phenol loading rate was found to be more important than the SRT in determining the phenol breakthrough (Nakhla et al. 1994). The specific loading and the SRT are related. The improved removal of compounds with slow biodegradation rates with increasing SRT may be offset by the increase in soluble microbial product (SMP) formation. SMP formation is directly related to sludge age (Rittman et al. 1987, Pribyl et al. 1997). SMPs are also directly related to the influent substrate concentration (Chudoba 1985). In the treatment of B K M E , increasing the SRT, and / or the HRT, is often observed to result in increased COD or A O X removal (Yu 1993, Rempel et al. 1992, Liu et al. 1993, Bomann et al. 1988, Wilson and Holloran 1992, Hall and Randle 1992, 18 Simpura and Pakarinen 1993, Cook 1990, Strehler and Welander 1994). For example, decreasing the HRT from 20 hours to 2 hours in lab scale activated sludge treatment of bleach plant effluent resulted in the COD removal decreasing from 50% to 40%, and A O X removal decreasing from 63% to 47%. A further reduction in HRT below 2 hours resulted in process deterioration. The mixed liquor solids level was kept constant during these experiments, so the SRT would have decreased with the HRT (Yu 1993). In another study of activated sludge treatment of B K M E where the SRT was kept constant, the soluble organic carbon and A O X removal decreased with increasing HRT. The explanation offered was that the increased biomass at lower HRTs resulted in greater adsorption of the organic material onto the floes. When the SRT was increased, greater A O X and soluble organic carbon removal was observed. The BOD was equally removed at all of the SRTs and HRTs studied (Oleszkiewicz et al. 1992). In a comparison of lagoon to activated sludge treatment systems, the long HRT of the lagoons was found to be more important than the long SRT of the activated sludge units for increasing A O X removal. Only very high SRTs (30 days) achieved equal A O X reduction compared to the long HRT lagoons (Hall and Randle 1992). As the SRT was increased from 10 days to 30 days, COD reduction was found to increase from 44% to 64%, with minimal improvement upon a further increase in SRT to 40 days (Simpura and Pakarinen 1993). Rempel et al. (1992) found a closer correlation between A O X reduction and SRT than between HRT and A O X reduction. A n SRT of 10 days or greater gave the best removal of resin acids and chlorinated phenolics (Rempel et al. 1992). Increasing the SRT from 2.5 to 5 days also results in better removal of chlorinated phenolics (Lindstrom and Mohamed 1988). BOD removal is not usually affected within the treatment parameters 19 tested (i.e. the BOD is completely removed at all test conditions which give good sludge settling characteristics). Unlike the BOD assay, A O X and COD measurements encompass compounds whether or not they are easily degraded. Increasing the HRT provides longer contact between the biomass and the wastewater resulting in increased removal ofthe more recalcitrant compounds across the treatment system. Conversely, decreasing the HRTs may not allow enough time for degradation of the various compounds in the wastewater. When the removal mechanism of a recalcitrant compound is adsorption onto the sludge, low SRTs wil l result in greater removal compared to higher SRTs, where biodegradation will be more important (Nyholm et al. 1992). In practical applications the HRT and the SRT are closely related. Increasing the SRT may require increasing the HRT to avoid excessive M L V S S concentrations, making it difficult to determine which parameter is more important. The effect of HRT on B K M E treatment becomes more evident when plug flow systems, or aeration tanks in series, are studied. There is a rapid removal of BOD at the start ofthe treatment unit, and a slow steady removal of COD across the whole treatment unit (Simpura and Pakarinen 1993, Servizi and Gordon 1973, Oleszkiewicz et al. 1992., Fein 1992, Saunamaki et al. 1991). For example, in an activated sludge unit with a selector, followed by two aeration tanks in series, 96% ofthe BOD was removed in the selector, with minimal removal in the following aeration tanks. In contrast, 56% of the COD removed was removed in the selector, 22% in the first aeration tank, and the remaining 20% in the second aeration tank (Simpura 1993). 20 Selectors Similar to the SRT and the HRT, the aeration tank hydraulic pattern may also have an important impact on the treatment process. Biomass in a plug flow system is exposed to different environmental conditions compared to biomass in a completely mixed system. These environmental differences may affect the microbial kinetics, the stoichiometry, and the microbial population. Often, biomass from completely mixed (CSTR) systems exhibit poor settling characteristics, such as sludge bulking, pin floes, or dispersed growth. Sludge bulking is one of the main operating problems of activated sludge. Some causes of sludge bulking are low nutrient levels, low dissolved oxygen levels, low pH, and a low food to microorganism ratio (f/m). The f/m in the aeration tank is the ratio of the substrate concentration in the aeration tank to the biomass concentration. A CSTR activated sludge unit will have a lower f/m than a plug flow activated sludge unit. In a plug flow reactor the f/m wil l be high initially, and decrease as substrate is removed. For a CSTR, the f/m may be increased by the addition of a selector to the activated sludge process, to provide conditions of relatively high substrate concentrations at the start of the treatment process. Filamentous microorganisms have a lower growth rate than floe forming bacteria at high substrate concentrations, but a higher growth rate at lower substrate concentrations. The selector insures that the substrate concentration is high, selecting for floe forming bacteria (Chudoba et al 1973). A n alternate explanation is that the selector selects for floe formers with higher growth rates than filaments at all substrate concentrations. The selector selects for fast growing organisms simply by providing a 21 high initial f/m (Ekama and Marais 1986). A number of studies have compared the substrate utilisation rates of biomass from CSTRs with the biomass from systems with selectors. The biomass from units with selectors was found to have faster substrate utilisation rates (Ekama and Marais 1986, Smets et al 1994, Cech et al 1984). A third explanation for the success of selectors is the greater storage capacity of floe formers compared to filaments, and their greater ability to survive starvation periods. Storage was found to be the main removal mechanism for acetate using biomass from an intermittently fed bioreactor. The storage polymers were then utilised in the subsequent starvation period. The formation of storage polymers resulted in increased growth yield, as the substrate was utilised more efficiently (Majone et al 1996, van Loosdrecht et al 1997). A respirometric study of an aerobic selector utilised for the treatment of pulp mill wastewater suggested that the substrate was assimilated in the selector and not stored (Wessberg etal 1993). In dealing with industrial wastewater, completely mixed activated sludge systems are often used in order to ensure maximum dilution and minimal impact of toxic compounds on the biomass. A synthetic wastewater containing phenol, a potentially toxic compound was treated in a selector, which resulted in inhibition of the bacteria. Better results were obtained using a compartmentalised system, which had better settling sludge than a CSTR treating the same wastewater. In the compartmentalised system, the phenol was removed in the first compartment, with the rest ofthe compartments serving for endogenous metabolism, which aided in the selection of floe forming bacteria. The maximum specific phenol removal rate was higher for the biomass from the compartmentalised system, as was the yield (obtained respirometrically) (Chudoba et al 22 1991). Nyholm et al (1992) recommend tanks in series for removal of difficult to degrade compounds, to ensure a loading suitable for acclimation of the biomass. Selectors do not have to be aerobic, and are often anoxic or anaerobic. A number of pulp and paper mills use an anaerobic or anoxic selector to prevent sludge bulking (Kang and Fitfield 1992, Flippin and Bellanca 1992, Biornstad et al 1993). All pilot studies found improved settling with the use of a selector. The use of an aerobic selector also worked in improving sludge settling when treating pulp mill wastewater (Wessberg et al 1993). The addition of an anaerobic stage may improve the degradation of certain compounds in addition to improving sludge settling. Anaerobic processes are efficient in degrading and dechlorinating low molecular mass molecules, although the rate of anaerobic dechlorination appears to be too slow to be of practical use in wastewater treatment (Fitzsimons and Eriksson 1990). In lagoons, both high and low MW AOX adsorb to aerobic biomass, which settles to the anaerobic zone, where anaerobic dehalogenation takes place (Wilson and Holloran 1992). One study has greater removal of AOX across an anaerobic/aerobic lagoon compared to activated sludge, but these were full scale treatment plants treating different wastes. The anaerobic aerobic system also removed a greater percentage of the higher MW compounds (Jokela et al 1993). Another study (lab-scale) found that the anaerobic zone does not increase AOX removal (Hall and Randle 1992). A study involving the profiling of the microbial community in activated sludge found that the community structure appears to be affected more by the nature of the wastewater than the presence of an anaerobic stage (Hiraishi 1998). 23 It has been shown that activated sludge operated with anaerobic stabilisation of the sludge can lead to synchronisation of the biomass. Synchronisation means that most of the cells are in the same physiological and metabolic state at the same time (Chudoba et al 1991). The same environment has been shown to lead to increased uncoupling between catabolism and anabolism, and hence lower yield (Chudoba et al 1992b). Temperature The microbial growth rate increases exponentially with temperature, when the temperature is below the optimum. As the temperature approaches the optimum, the increase in growth rate with temperature decreases. When the temperature passes the optimum, the microbial growth rate decreases rapidly with further increases in temperature. The optimum temperature for wastewater treatment plants is often reported to be around 35°C, with an acceptable range of 15 to 40°C for mesophilic bacteria (Bailey and Ollis 1986). The effluent from a pulp mill is usually warmer than this, so the effluent must be cooled prior to activated sludge treatment. If the wastewater is treated in a lagoon, then the large surface area and residence time may be enough to allow the effluent to cool to the desirable operating range. In the winter months, lagoon performance may decrease due to low temperatures. ASB performance also decreases at temperatures greater than 40°C (Springer 1993). In a study of aerated lagoons treating B K M E , reduced removal efficiencies of chlorinated phenolic compounds, A O X , and toxicity were observed at 10°C compared to 25°C (Melcer et al 1995). The optimal temperature for the treatment of CTMP (chemi-thermomechanical pulping) effluent was at 20°C. Below this temperature, treatment efficiency decreased. Temperature increases up to 40°C brought about a slight decrease 24 in removal efficiency, while treatment was poor at 50°C. The MLSS concentration was low at 50°C (Liu et al 1993). Increasing the temperature (to 40°C) resulted in lower resin and fatty acid removal due to sludge adsorption, which is an important treatment mechanism when the treatment time is short. Resin and fatty acid biodegradation rates increased with temperatures up to 40°C (Liu et al 1993b). Decreased removal of A O X at higher temperatures (35°C compared to 32°C) across an aerated lagoon was also attributed to decreased sorption onto the biomass at the higher temperatures (Bryant and Amy 1989). Under aerobic conditions, A O X removal was found to increase from 47% at 22°C to 70%) at 30°C, while TOC removal was not affected by the rise in temperature (Chernysh et al 1992). A study using an aerobic suspended biofilm found better A O X removal at 50°C compared to 37°C, which was attributed to increased abiotic removal at the elevated temperature (Strehler and Welander 1994). Activated sludge operation at 50°C resulted in increased COD removal compared to operation at 35°C (Barr et al 1996). The maximum growth rate of nitrifying bacteria was found to increase monotonically from 15°C to 25°C (Antoniou et al 1990). Over a temperature range from 10°C to 35°C, the Monod half saturation constant for nitrification was found to be lowest at 15°C. The nitrification was most efficient at . 15°C, which is different from the typical optimum temperature for carbon assimilation. (Charley 1980). The optimal temperature for sludge yield is not necessarily the optimum temperature for the growth rate, or substrate utilisation rate (Coultate and Sundaram 1975). Depending on the substrate, and the environmental conditions (aerobic / anaerobic), the Monod half-saturation coefficient has been found to both increase and decrease with increasing temperature (Esener et al 1983, Muck and Grady 1975). There is greater sludge production at lower operating temperatures, 0.22 kg SS/kg BOD at 29°C compared to 0.33 lb SS/lb BOD at 18°C, during the treatment of pulp mill wastewater (Springer 1993). The lower yield at higher temperatures may be due to the microbial decay. The decay rate of bacteria increases with increasing temperature. A general rule of thumb is that the "decay rate constant is influenced by temperature in a manner similar to the maximum specific growth rate constant" (Muck and Grady 1975). The relationship between microbial yield and temperature sometimes passes through a maximum at approximately 20°C. Specific oxygen demand is much greater at high (50°C) operating temperatures, compared to 35°C (Carter et al 1975). A further effect of temperature is to decrease the sludge settling velocity and increase the sludge volume index (SVI). The change in SVI is due to changes in the floe structure, and the nature of the exocellular material that the bacteria secrete to flocculate. A temperature range of 10°C to 35°C was studied (Cetin 1990). pH For most enzymatic systems, the enzymes are active within a certain pH range, usually 4 to 5 pH units. Inside this pH range, the activity versus pH has a bell shaped profile (often with a maximum around pH 7.4). Outside of this pH range, the proteins denature as they are protonated or deprotonated (Copeland 1996). Similar relationships between activity and pH are found for bacterial cultures as are found with pure enzymes (Rosso et al 1995, Lallai et al 1988). In a study of CTMP aerobic wastewater treatment, BOD and resin and fatty acid removal was the same at operating pHs from 5 to 8. COD removal was greatest at pH 7 (80%), decreasing to 70% at pH 5, and to 74% at pH 8 (Liu et al 1993). Glucose uptake 26 rates, using biomass from a waste stabilisation pond, containing algae, were optimal from pH 6 to 8. Below pH 6, there was a slight decrease in the glucose uptake rates, while above pH 9, there was a large decrease. The decrease in glucose consumption above pH 9 was attributed to the algae (Mayo and Noike 1996). For growth on phenol, the microbial yield was found to be maximal at pH 6, decreasing drastically below this pH, and decreasing gradually above this pH (Lallai et al 1988). An increase in the pH caused an increase in sludge settling velocity, and a decrease in the SVI (a pH range of 6 to 10 was studied). As the pH increased, larger floes were formed due to an increased number of reactive sites on the cell surfaces and exocellular polymers (Cetin 1990). The effects of pH and temperature on bacteria growth rate are usually found to be independent (Rosso et al 1995). 2.3 Microbiological Aspects of Wastewater Treatment In this section a brief review ofthe basic microbiological aspects pertinent to the data interpretation in subsequent chapters will be presented. The majority of bacteria present in activated sludge treatment systems treating B K M E are heterotrophs. For heterotrophic bacteria, the carbon source in the wastewater serves two important functions: energy generation and biomass formation. Among the carbon and energy sources for heterotrophic growth in B K M E are sugars, organic acids, (including acetate and formate), as well as alcohols, mainly methanol. Activated sludge systems may also contain autotrophic bacteria which obtain their energy from the oxidation of substrates in the wastewater (such as ammonium) and their carbon from carbon dioxide. 27 Substrate Metabolism The generation of energy from sugars may be demonstrated using glucose as an example. In many bacteria, the glucose is oxidised to two molecules of pyruvate via the Embden-Meyer-Oscar pathway. This results in the generation of ATP. In aerobic bacteria, pyruvate is oxidised to carbon dioxide via the citric acid cycle. This results in the reduction of N A D and F A D to N A D H 2 and F A D H 2 respectively. NADFf 2 and FADH2 transfer their electrons to the electron transfer chain in the cellular membrane. Under aerobic conditions, the terminal electron acceptor is oxygen. The electron transfer chain serves to pump hydrogen ions outside of the cellular membrane (which is impermeable to hydrogen and hydroxyl ions), creating a potential across the membrane. This potential is used for the generation of ATP. There are many variations of this pathway, depending on the bacterial species and environmental conditions. When glucose is the growth substrate, cellular building blocks are obtained from many points along the dissimilation pathway. The dissimilation of acetate may follow a pathway similar to that of glucose. Acetate is oxidised via the citric acid cycle. Electrons from acetate are transferred to N A D and F A D , and then to the electron transfer chain for the generation of membrane potential and ATP. When acetate is the carbon source, an additional pathway is required, which is the glyoxylate cycle. This allows for the synthesis of three and four carbon cellular building blocks. If the glyoxylate cycle was not present, the bacteria would not be able to regenerate the constituents of the citric acid cycle as they are removed to build cellular components. 28 The dissimilation of one-carbon substrates is different from the metabolism of sugars and acetate. Several different pathways of methanol metabolism have been found in bacteria. The first step involves oxidation of methanol to formaldehyde catalysed by methanol dehydrogenase. This reaction occurs outside the bacterial membrane, in the cytoplasm (in gram negative bacteria). The other substrates in B K M E must first be transported across the cell membrane before being dissimilated. Methanol dehydrogenase interacts directly with the electron transfer chain. Methanol is oxidised to formaldehyde. Formaldehyde is transported into the cell and further oxidised to formic acid and carbon dioxide by a number of different pathways. If methanol is to serve as the carbon source, pathways must be present to build cellular material from one-carbon compounds. The common building block for these pathways is formaldehyde. There are several such pathways in a class of bacteria known as methylotrophs. Variability of Yield and Metabolism The microbial yield is an important parameter for a waste water treatment process. The yield determines how much sludge must be wasted which may be a large operating expense. The microbial yield depends upon the relationship between anabolism and catabolism. Microbial cells do not have the ability to exert fine control over the uptake and catabolism of carbon substrate, and can readily dissociate catabolism from anabolism. The anabolic enzymes are adjusted to the growth rate, but the catabolic enzymes are maintained in excess. As a consequence, i f the carbon source is added as a pulse to a bacterial culture, catabolism will be higher than anabolism, and microbial yield will be lower than i f the carbon source had been added continuously. This has been 29 observed using bacteria during active growth, as well as washed suspensions of bacteria under conditions where biosynthesis is severely impeded (Tempest and Neijssel 1980, Brooks and Meyers 1973, Stouthamer 1979, Russell and Cook 1995). One of the main factors which determines the microbial yield is the nature of the substrate. The yields on organic acids are lower than the yields on carbohydrates. This is partly due to organic acids being more oxidised than carbohydrates, and therefore provide less energy (Narang 1997). Different substrates are utilised by different pathways, which has an effect on the amount of carbon assimilated, and hence the yield (Gommers et al . 1988). If multiple substrates are present, the nature of the substrate mixture will also influence the yield. Substrates such as formate and methanol may not be assimilated when other substrates are present; the enzymes required for methanol and formate assimilation may not even be present. Instead, the methanol or formate wil l be utilised solely as an energy source. This allows for a greater percentage of the other substrates to be assimilated, increasing the microbial yield. This depends on the particular substrates, the ratio of the substrates, the growth rate, and the bacterial species. Lower ratios of formate or methanol to the other carbon source, and higher growth rates, tend to suppress assimilation of the formate or methanol, and lead to higher yields (due to greater assimilation of the other carbon source) (Harder and Dijhuizen 1982, Gommers et al 1988). The true microbial yield depends on the substrate, the transport system, cell composition, and metabolic pathway (including the formation of storage polymers). The apparent yield, which is the measured yield, will be lower than the true yield due to 30 expenditure of energy for non-growth processes. The apparent yield depends upon the true yield, decay rate and endogenous metabolism. These processes are influenced by the f/m, growth rate, substrate composition, population dynamics (predation by protozoa), and feeding patterns (which may trigger energy spilling, select for floe formers over filaments, etc.). The yield also depends upon other environmental factors, such as the presence of metals (Gokcay and Yetis 1996), temperature, and the pH. Even i f the true yield was constant, the apparent yield for a wastewater treatment process is influenced by many factors, and will be quite variable (Gaudy and Ramanathan 1971). Uncoupled Metabolism Anabolism and catabolism are not always perfectly coupled, not all ofthe energy obtained from substrate oxidation is utilised by the cell. The uncoupling between anabolism and catabolism may be a response to nutrient limitation, and appears to be a common characteristic of growth with an excess of energy (Rusell and Cook 1995). Under nutrient limitation, bacteria will increase the rate of transport of substrate into the cell, and increase the initial metabolism of the substrate. This is in order to maintain the driving force for accumulation of substrate, and also to allow the cells to increase their processes of cell synthesis without being limited by a lack of available energy, i f the limiting nutrient were to suddenly increase. These goals may be achieved either through a greater amount of metabolic enzymes, or synthesis of high affinity enzymes (Harder and Dijkhuizen 1983, Neijssel and Tempest 1976, Tempest et al 1985). In chemostat studies at varying growth rates, the maximal carbon uptake rate was found to be constant, independent ofthe growth rate. The excess catabolic activity ofthe bacteria decreased as the growth rate increased. 31 Metabolic uncoupling is necessary to maintain growth potential. If bacteria are adapted for scavenging small molecules from dilute aqueous environments, a sudden increase in substrate would result in accumulation of too much substrate (Koch 1979). This in turn will result in an excess of energy, which must be spilled. Mechanisms for energy spilling include: production of storage compounds, excretion of catabolic products, deletion of sites of oxidative phosphorylation, branching of the respiratory chain, dissipation ofthe energised membrane, and wastage of ATP (Stouthamer 1979, Russell and Cook 1995). A l l of these mechanism result in a decreased yield. Endogenous Metabolism, Maintenance Energy and Microbial Decay As the SRT increases, the M L V S S concentration also increases. For a given HRT and wastewater, the biomass concentration in an activated sludge unit is directly proportional to the SRT times the net yield. It is commonly observed that the net yield decreases with increasing SRT. This decrease in net yield is attributed to microbial decay and / or endogenous metabolism. These processes become important in activated sludge operation due to the low microbial growth rates typical in wastewater treatment plants. In rapidly growing cultures there is much less of an impact of bacterial decay on the net growth rate. Maintenance energy is energy expended by the bacteria for maintenance functions. These functions include maintaining ion gradients across cell membranes, protein and R N A turnover, overflow metabolism, metabolic shift, futile cycles, modification ofthe respiratory chain, and metabolic uncoupling (Tempest and Neijssel 1984, Russell and Cook 1995). When no exogenous substrate is present, the bacteria will obtain the energy required for maintenance from endogenous substrates, and this is 32 referred to as endogenous metabolism. Maintenance energy and endogenous metabolism are often assumed to be the same, but this is not necessarily the case. When no substrate is present, the amount of maintenance energy required to keep the cell functioning may be lower than when exogenous substrate is present (Russell and Cook 1995). The maintenance energy is not a constant, but has been observed to be a function of the growth rate (Tempest and Neijssel 1980). Maintenance energy also depends upon the substrate, the limiting nutrients, the temperature, the dissolved oxygen concentration, the substrate addition pattern, the particular bacteria, and the physiological state ofthe bacteria. A l l of these factors can change with changing growth rate, causing an increase in maintenance requirements at lower growth rates (Stouthamer 1990). Under starvation conditions, during endogenous metabolism, bacteria can maintain catabolic enzymes, and immediately shift up metabolism after exposure to substrate. Prolonged starvation results in decreased yield after substrate addition (Koch 1979), presumably due to greater energy wasting. As the growth rate decreases, the fraction of viable biomass also decreases. A significant fraction of biomass from chemostats operated at low growth rates may be able to degrade substrate, but unable to multiply, and hence, are not viable. In the study of activated sludge it is difficult to count the numbers of viable biomass, but indirect results indicate that activated sludge viability decreases with decreasing growth rate (increasing SRT) (Huang 1982). For activated sludge operated at an SRT of 10 days, the viability of the M L V S S may be as low as 20%. The explanation for the apparent low percentage of viable microorganism in activated sludge is provided by the concept of microbial decay. 33 Microbial decay is usually attributed to the death of microorganisms and the subsequent utilisation of the cellular material by the remaining microbes. If some of the cellular material released by the decay process is recalcitrant, this material will stay in the system (due to the recycle of sludge in the activated sludge process), and increase with increasing SRT. The viable fraction of the MLVSS will decrease with increasing SRT. 2.4 Effect of Transient Operating Conditions on Activated Sludge Performance B O D Wastewater treatment plants may be exposed to organic shock loads - sudden increases in the amount of substrate concentration in the wastewater. The activated sludge response to organic shock loads is varied, and depends upon the prior operating conditions. Increased substrate concentration may result in increased microbial growth rate, changes in the substrate utilisation pattern, changes in the floe structure, and changes in stoichiometry. A common response to increased substrate is an immediate utilisation of the substrate at an increased rate, but a lag period prior to an increase in the growth rate of the biomass (Rozich and Gaudy 1985). Activated sludge systems are commonly observed to have the capability to absorb shock loads (up to three fold increases in substrate concentration (Krishnan and Gaudy 1976)). This is thought to be a property of biomass cultivated at low growth rates. The biomass maintains a capability to metabolise substrate concentrations greater than in the steady state environment. As a result the 34 yield during the transient period is usually lower than the yield during steady state. This may be explained by the concept of metabolic uncoupling described above. Increasing the SRT increases the biomass in the system, which increases the capability to handle shock loads. One study found increasing the SRT from 5 to 10 days resulted in no difference in activated sludge performance. This was hypothesised to be due to the higher percentage of viable cells at the low SRT, smaller floes, and less slime (reduced mass transfer resistance) (Selna and Schroeder 1978). Another study found biomass at high growth rate had a slow response to substrate increases. Biomass at low growth rates also had a slow response (in growth rate), and the growth rate and oxidation were uncoupled. Biomass at a medium growth rate had the most rapid response to substrate increases (Daigger and Grady 1982a). Response to changes in substrate concentration involves physiological shifts. During unsteady state operation, as during shock loads, the microbial kinetics depend upon the past culture history, and will change as the environmental conditions change (Mona et al 1979, Manickam and Gaudy 1985, Storer and Gaudy 1969, Selna and Schroeder 1978). Sludge characteristics may change following large organic shock loads. The increased growth rate of the biomass may result in the formation of pin floes, which have poor settling characteristics. Temperature and pH Shocks Biomass acclimated to 35°C could tolerate shock temperatures up to 45°C, but above this temperature the system was inhibited. Biomass acclimated to 50°C could tolerate temperature drops to 45°C, but not 35°C (Carter 1975). For biomass acclimated 35 to 25°C, temperature increases to 47°C and 57.5°C resulted in a loss of biomass, and deterioration of treatment. Within two to four days, the biomass started to recover (treatment performance recovered with the biomass), although the morphology was different than prior to the shock. A temperature decrease to 8°C (from 25°C) resulted in loss of biomass and poor treatment. Lower growth rates, and higher biomass concentrations, were found to improve the accommodation to the shock temperature (George and Gaudy 1973). Anaerobic biomass acclimated to 39°C could tolerate temperature increases to 45°C. Above this temperature, the microbial decay increased, and the activity of the population decreased (van Lier et al 1990). A 10°C temperature increase to 35°C resulted in increased substrate removal, and increased biomass growth rate. There was a lag period of a few hours after the increase in substrate uptake rate before the growth rate increased (Topiwala and Sinclair 1971). The response of CSTRs, with and without cell recycle, to pH shocks was studied. Sudden decreases in pH caused a sudden decrease in biomass concentration, but once the system had time to acclimate to the new pH, the biomass concentration increased to levels higher than the pre-shock value. Acclimation consisted of the growth of fungi, tolerant to low pH. In the system with no cell recycle, recovery of the system following a pH shock did not depend on the magnitude of the shock, but on the time required for a new population to develop (approximately five days). In the system with cell recycle, less change in population was observed; the original population adapted to the new conditions before the fungi took over (George and Gaudy 1973b). Uncoupled growth results not only from the presence of an excess energy source as described above, but also from inhibitory compounds, unfavorable temperatures, 36 minimal media, and transient periods. (Stouthamer 1979, Coultate and Sundaram 1975). When yeast was grown in chemostats at superoptimal temperatures, the yield was decreased due to increased maintenance by viable cells, and increased energy dissipation by non-viable cells (Van Uden and Madeira-Lopez 1976). For some mesophilic bacteria, decreases in growth yield (due to increased maintenance energy expenditures and uncoupled metabolism) occur at temperatures equal to or greater than the optimum temperature for growth rate (Farmer and Jones 1976, Forrest 1967, Esener et al 1983). Other bacteria exhibit complementary behaviour where uncoupling increases below the optimum temperature for growth (Forrest 1967). Toxic compounds Activated sludge biomass must acclimate to the compounds present in the wastewater before degradation can occur. For acclimation, the environmental conditions are important. The acclimation period for the biodegradation of nitrilotriacetic acid (NTA) is much shorter in laboratory studies than in full scale systems, presumably due to the prevailing optimum operating conditions in the laboratory (Stephenson et al 1984). For rapid acclimation, the inoculum should contain a large amount of active cells, and the initial substrate concentration should be low enough to avoid substrate inhibition effects, but not too low or acclimation may not occur (Kim and Maier 1986). Adaptation is affected by the presence of other substrates (Tokuz 1991), depends upon the specific compounds, and is influenced by the presence of predatory protozoa (Wiggins et al 1987). Lag periods before acclimation occurs are often on the order of days, but in some cases can be several months (Nyholm et al 1992). Acclimation periods have been explained as the time necessary for microbial populations to grow to sufficient size to 37 achieve detectable bioconversion rates, induce new enzymes, undergo genetic changes, and to exhaust preferential substrates (Linkfield et al 1989). For the case of phenol, adaptation of the activated sludge was found to involve a shift in the microbial population to a greater percentage of gram negative bacteria, and a less diverse population of protozoa (Lewandowski 1990). Once the compound of interest is removed from the wastewater feed to the activated sludge plant, the biomass may de-adapt. The course of de-adaptation appears to depend op the specific case. If special enzymes are required, and these enzymes are only produced when the compound of interest is present, then de-adaptation will be related to the SRT (Senthilnathan and Ganczarczyk 1989). In a study using parachlorophenol, the bacteria de-adapted faster than could be explained by the SRT (Arbuckle and Kennedy 1989). Similar results were obtained using 3,5-dichlorophenol, where the biomass de-adapted when 3,5-dichlorophenol was removed from the wastewater for four days (Broecker and Zahn 1977). Acclimation of the activated sludge process to a particular organic compound will improve the response ofthe system to a shock load by the same compound. Such acclimation does not necessarily prevent system failure (Tokuz 1991). Activated sludge units operated at longer SRTs appear to be more resistant to toxic shock loads than activated sludge units operated at shorter SRTs (Santiago and Grady 1990, Rozich and Gaudy 1985). Increasing the sludge age also increases the biomass concentration. This results in a decrease in the toxicant to microorganism ratio, which can help to mitigate the inhibitory effects of toxic compounds (Lange et al 1989). Fast growing cells have been 38 found to be more sensitive to toxic compounds, due to greater permeability into the cells (Bull and Brown 1979). When multiple phenol shocks were applied to activated sludge biomass, there was a decrease in the specific substrate removal rate with each consecutive pulse (Okaygun et al 1992). Other studies have found increased biodegradation after exposure to a toxic compound, and further increases in biodegradation of toxic compounds during shock loads once the cultures are acclimated (Lewandowski 1990). Traditionally, CSTRs are thought to provide the best protection against inhibitory shock loads, due to the large dilution factor. Simulation studies indicate that plug flow reactors are better at mitigating the effect of shock loads than CSTRs unless there is severe substrate inhibition (Santiago and Grady 1990). Plug flow reactors should provide a sludge with better settling properties than CSTRs. Operating Conditions The SRT affects the growth rate of the microorganism, which affects the physiological properties. SRT also exerts a selection pressure on the population, based on growth rate. Changes in the SRT will cause the biomass properties to change. Approximately three times the difference between the old SRT and the new SRT must pass before a new steady state is achieved (Zaloum 1992). Other factors which affect treatment are the temperature, pH, f/m ratio, wastewater composition and wastewater strength. A l l of these factors may be variable. For a mixed microbial population treating a mixture of substrates, steady state may never be achieved. The population dynamics between the various bacterial species, and the predator prey relationship between the protozoa and the bacteria wil l lead to oscillations 39 in the populations (Bazin 1990). Shifts in microbial populations were observed after changes in wastewater strength (Okaygun et al 1992). Low dissolved oxygen concentration, or low nutrient levels, may lead to predominance of filamentous bacteria. If the microbial population is changing following changes in environmental conditions, the treatment parameters may be changing as well. Due to these difficulties, a general rule of thumb is to wait 3 to 10 sludge ages after process changes for the biomass to adapt to the new conditions. Adaptation to temperature changes requires from 10 to 20 days to reach new equilibrium conditions (Chou and O'Brien 1987). 2.5 Summary B K M E is a complex mixture of many organic compounds. The composition of B K M E depends on the operating conditions of the pulp mill, and can be highly variable. Common constituents are methanol, formate, acetate, sugars, and many toxic compounds such as phenols and resin acids. B K M E treatment by biological processes is very effective, especially for BOD and acute toxicity removal. Treatment efficiency increases with SRT and HRT. This is due to the large amount of recalcitrant substrates in the wastewater which need long contact times (high HRT), or slow growing microorganisms (high SRT), for biodegradation. The activated sludge operating conditions affect the bacterial population. Decreasing the SRT, or adding a selector, has been observed to result in bacteria with greater growth rates. The population may also be affected by the composition of the wastewater. For example, when grown on methanol as a sole substrate bacteria wil l synthesise the necessary enzymes for growth on a C l substrate. If another substrate is 40 present as a carbon source, the bacteria may not synthesise the enzymes required to use methanol as a carbon source, and use methanol solely as an energy source. Many different bacterial populations and metabolic pathways are possible. The response of an activated sludge unit to an increase in substrate loading can be complex. The growth rate of activated sludge bacteria is usually low. A n excess of dissimilatory enzymes are maintained to scavenge low substrate concentration environments, and to accommodate sudden increases in substrate concentration. This allows the bacteria to handle large increases in substrate concentration, however the energy generated by substrate dissimilation is wasted until anabolic enzymes can be synthesised for an increase in the microbial growth rate. A lag period between the substrate concentration increase and increased growth rate is often observed. The increase in substrate removal rate following an increase in substrate concentration is often immediate. Bacteria at low growth rates are able to handle changing conditions (substrate, temperature, pH) better than bacteria at high growth rates. Increasing the SRT often results in a greater capacity to handle shock loads, due to the lower growth rate and increased biomass concentration. 41 Chapter 3 Literature Review - Activated Sludge Models 3.1 Introduction Many activated sludge models have been developed over the years. The majority of models are based upon the Monod equation. The Monod model is an empirical relationship which relates biomass growth of a pure culture on a single substrate to the substrate concentration, and is similar to the Michaelis-Menton model for enzyme kinetics. In this chapter the models based on the Monod equation, as well as some alternative models, wil l be discussed. The treatment of wastewater by the activated sludge process consists of a complicated series of reactions. The bacteria in activated sludge usually grow in floes, which are an amalgamation of live, dead, and non-viable bacteria held together by exocellular materials excreted by the biomass under the growth conditions typical in wastewater treatment plants. During wastewater treatment, the many different substrates in the wastewater (soluble, colloidal, and particulate), all with different degradation rates, must first be transported into the floe to come into contact with the mixed culture. The substrates are then degraded by the bacteria. Depending on the substrates, the environmental conditions, and the species of bacteria present, the substrates may be used for growth, converted to storage products for later use, or only partially metabolised and returned to the wastewater. During steady state operation of wastewater treatment plants, the growth rate of the biomass is balanced by the decay rate and sludge wasting. Due to the low substrate concentrations in completely mixed activated sludge reactors, and the corresponding low 42 growth rates, the decay rate will be a significant factor. The decay of bacteria produces soluble and particulate organics, some of which are biodegradable and will be metabolised by the remaining bacteria, and some of which are recalcitrant (during the residence time of an activated sludge plant). The Monod model, without modifications, rarely fits experimental data from activated sludge units, due to the complexities discussed above. Many different modifications have been made to improve the model. These include: removing the assumption that the biomass and substrate are homogenous, adding mass transfer steps, either adsorption into the floe and conversion to storage products, or diffusion through the floe before the bacteria can metabolise the substrates, or accounting for soluble microbial product formation. The two most popular modifications in the activated sludge literature seem to be: 1) to divide the substrate into various components with different kinetic coefficients, and 2) to divide the removal of the substrate into various processes, such as an adsorption step followed by a metabolism step. A l l of these modifications seem to add enough flexibility to allow the models to fit the data. Since almost all activated sludge units are operated at conditions which achieve complete removal of the influent substrate, it is very difficult to compare all of the different models to determine which one is the most appropriate (Orhon et al 1989). The most applicable model probably depends upon both the wastewater and biomass characteristics, as the relative importance of all of the different processes discussed above may change (Sheffer et al 1984). 43 3.2 Monod Models of microbial growth relate biomass growth rate with either biomass concentration, substrate concentration, or both. Although biomass concentration can indirectly affect the growth rate through such mechanisms as changing the pH, changing the dissolved oxygen concentration, etc., the growth rate is intrinsically independent of population density (Powell 1972). The Monod equation (equation 3.1, figure 3.1) is the standard way of correlating the microbial growth rate with substrate concentration, and will be discussed before alternatives and modifications are presented. In the absence of endogenous metabolism: M- = V-UAX V ^ „ where p is the specific growth rate and is defined as follows: » = - , l c ( 3 ' 2 ) If the yield is defined as: d7dt * then, the rate of consumption of substrate is (assuming a constant yield): dSl = p^_^S_ = _ A _ ( 3 . 4 ) H dt X Y KM +S *MAXKM+S The apparent microbial yield is not constant. The apparent yield at the low growth rates common in activated sludge units is often lower than the apparent yield of biomass growing at higher rates. It is assumed that the true yield is constant, but the yield appears to decrease because of the increase in the percentage of energy going towards cell maintenance and other non-growth processes at low growth rates. This may 44 be accounted for in the Monod model by adding a term for microbial decay, usually modeled as a first order reaction (also purely empirical). When the substrate concentration is equal to 0, the change in biomass concentration is due solely to microbial decay as follows: ^- = -kdX (3.5) at The growth rate now becomes: S V=VMAX V ^ „-K (3-6) M This modification allows the Monod model to be applied to continuous activated sludge units. Using mass balances, and assuming that there is no biomass in the influent, and no separation of substrate in the clarifier, the following equations are obtained for substrate, biomass, and yield (Orhon and Artan 1994, pp207-211): (3.7) xY(SQ-S^x Y„ = — (3.9) obs According to equation 3.7, the substrate concentration leaving a continuous activated sludge treatment system does not depend on the influent substrate concentration, but only upon the SRT (Grady and Williams 1975). At low substrate concentrations (S « K M ) , the Monod equation simplifies to a first order equation: q = q M A 2 L S = K S ( 3 1 0 ) KM 45 At high substrate concentrations (S » K M ) , the Monod equation becomes zero order (q = q M A x ) -Alternatives to Monod j The Monod equation is purely empirical in nature, although there is a similarity to Michaelis-Menten kinetics. It has been widely observed in the literature that the Monod equation approaches its asymptotic value too slowly to accurately model experimental data (Powell 1972, Bader 1978) and a number of alternate models have been proposed (Powell 1972, Bailey and Ollis 1986 p391). The Blackman model (equation 3.11, figure 3.1) often fits experimental data better than Monod, but due to the discontinuity it is not as easy to use (Bader 1978). p = S / A w h e n S < A p M A x (3.11) P = PMAX when S > A p M A x According to the Blackman equation, the maximum growth rate is not just some fictional condition approached asymptotically at high substrate concentrations, but is attained at a finite substrate concentration (S = A PMAX)-Powell (1972) modified the Monod model by assuming that the substrate must first diffuse into the cell through a membrane according to a first order reaction (which takes place inside the cell membrane), then the substrate is metabolised inside the cell according to Monod kinetics. If steady state is assumed, then the following equation is derived: 46 1.2 0 1 2 3 4 5 6 7 8 9 10 Substrate Figure 3.1 Monod kinetics L/K = 0 ( ), Blackman kinetics L/K = oo ( ), Powell kinetics L/K = 2 ( ), Powell kinetics L/K = 8 ( ), Powell kinetics L/K = 38 ( ), Powell kinetics L/K = 398 ( ). 1.2 0 1 2 3 4 5 6 7 8 9 10 Substrate Figure 3.2 Monod kinetics n = 1 ( ), Moser kinetics n = 2 ( ), Moser kinetics n = 3 ( ), Moser kinetics n = 5 ( ), Moser kinetics n = 10 ( ). 47 where F is a constant related to the substrate diffusivity. In theory F is a function ofthe geometry and physical characteristics of the organisms and their environment, but in equation 3.12 it is used as an average parameter for the whole system, obtained empirically. Introducing L=PMAX/(Y F) and solving for p: Equation 3.13 is shown graphically in figure 3.1, for a number of values of L . Powell data, and found that equation 3.13 gave the best fit in the majority ofthe cases. If L is small, the Monod model may be used with an apparent half saturation constant Ff= KM + L . Koch and Coffman (1970) presented an independent derivation of equation 3.12. Diffusion into the entire cell was assumed instead of diffusion across a membrane (as in Powell's derivation). The diffusion was assumed to be rapid, resulting in a constant substrate concentration throughout the cell. In comparing equation 3.12 to Monod's, Koch (1979) found equation 3.12 to give a better fit to the growth vs. substrate (glucose) data. It was hypothesised that the cell's transport mechanisms are present in excess and capable of supporting the cell's need for glucose at an external glucose concentration where uptake is non-saturated and still in the first order phase. Dabes (Dabes et al 1973) derived equation 3.12 (the Powell equation) by assuming that cell metabolism could be simulated using a series of linked reversible enzymatic reactions. If one reaction is much slower than all of the other reactions, the Monod equation results. If there are two slow reactions which are separated by any number of fast reactions with a large equilibrium constant, the Blackman equation (3.11) (3.13) compared equation 3.13 with Monod's and Tessier's (equation 3.15) using published 48 results. Equation 3.11 implies that there are 2 regimes, one where the external substrate concentration limits the growth rate, and one where an internal "pacemaker" enzyme limits growth. If two slow reactions are separated by an equilibrium constant which is not large, then equation 3.12 results. Dabes compared equation 3.11 and 3.12 with the Monod equation for a number of published data sets, and found that either equation 3.11 or 3.12 gave the best fit. The Blackman and Powell equations are derived based upon the assumption of two or more reactions in series, at least one of which follows Monod kinetics. A number of other equations have been proposed to describe experimental microbial growth data. These equations are more complicated mathematically, are completely empirical, and do not appear to be commonly used. They all have the advantage over Monod's in that they saturate faster with increasing substrate concentration than Monod's equation does. Some examples of these are: Moser's (figure 3.2): S" ^  = ^ MAX~k~^¥  ( 3 1 4 ) Tessier's (figure 3.3): u - i w ( l - e * s ) (3-15) and Konak's (1974) (figure 3.3): ys=KkMAX-»J (3.16) When n=l, equation 3.16 reduces to Tessiers equation (3.15), i f n = 2, then Monod's equation results. If n 1 then equation 3.16 may be integrated to give: \&3 -(VMAX- I t / = (1 - n ) K S (3.17) 49 1.2 1 1 0 1 2 3 4 5 6 7 8 9 10 Substrate Figure 3.3 Monod kinetics n = 2 ( ), Tessier kinetics n = 1 ( ), Konak kinetics n = 3 ( k Konak kinetics n = 5 ( ), Konak kinetics n = 7 ( ), Konak kinetics n=10( ). 350 Figure 3.4 Example of multisubstrate removal during a batch test following equation 3.22. The overall substrate (thick solid line) removal rate is equal to the sum ofthe individual substrate removal rates (each individual substrate is represented by a seperate dotted line). 50 The advantage of equations 3.11 to 3.17 over Monod's equation is faster saturation of the growth rate with respect to the substrate concentration, which more often allows for a better fit of experimental data. Another source of difficulty with Monod's model is that K M does not appear to be a constant, but a function of either the biomass concentration or the initial substrate concentration. In continuous studies, the biomass concentration is related to the initial substrate concentration through the yield coefficient, making it difficult to determine which has a greater effect on K M . The majority of studies where K M was not found to be a constant were studies of continuous systems. During batch tests X is usually constant so the relationship between X and K M would not be noticed! Contois (1959) proposed that K M is directly proportional to biomass concentration, resulting in The explanation given was that at higher biomass concentrations there wil l be more inhibitory microbial products, resulting in a larger apparent half saturation constant. Fujimoto (1963) derived a similar equation by assuming that substrate must be adsorbed by the cell before it can be utilised. Equation 3.13 was derived using similar reasoning that is used to derive Monod's equation, but used S / X (adsorbed substrate) instead of S (bulk substrate). Roques et al (1982) also found that K M was.proportional to biomass concentration and proposed the following: KyX+S (3.18) S (3.14) M1 V-UAX S + M+b(SQ-S) 57 This equation can simplify to Monod's i f b = 0, to Powell's i f S « So (and L in Powell's equation is small), to Contois's i f M = 0 (and using (So-S)=X / Y , assuming Y is constant). Chen and Hashimoto (1978) and Elmaleh and Ben Aim (1976) found K M to be a function of initial substrate concentration. Grady and Williams (1975) used a first order approximation of Monod's equation and also found K M to be proportional to So- The main implication of equation 3.14 and similar (equation 3.13, Chen's, Elmaleh's, and Grady's) is the prediction that the substrate concentration in the treated wastewater of a continuous system is a function of the substrate concentration in the untreated wastewater, as was found in the above studies, and not independent of the untreated wastewater substrate concentration as predicted by equation 3.7. It is interesting to note that most of these studies dealt with mixed substrates measured using global parameters (this subject will be discussed in more detail in the next section). In a study dealing with pure compounds, Cech et al (1984) also found K M to be proportional to the biomass concentration, but the Monod model was still used to describe the data. Roques equation (3.14) reduces to Monod's during batch tests where initial rates are measured (because S=So), or the equation of Contois' may be used instead. Most of the equations discussed in this section may be simplified to Monod's under special conditions. Equations for continuous systems similar to equations 3.6 to 3.8 may be developed by adding the term for microbial decay and using appropriate mass balances. Roels (1983, pp237-239) has shown that under substrate limiting growth, the biomass concentration is not a strong function of the microbial model chosen,.but depends mainly on the yield coefficient. The substrate concentration under these conditions can not usually be determined accurately enough to choose the proper model. 52 The Monod model appears to be adequate when substrate is not limiting (zero order region), and for correlating steady state biomass concentration and growth rate in continuous cultures. Even in cases where reliable substrate data allows the use of alternate models, most researchers use Monod due to its resemblance to Michaelis-Menten enzyme kinetics, and it's ease of use (Bader 1978). 3.3 Multiple Substrates All ofthe preceding models are based upon the assumption that there is only one substrate used for growth, even though these models are often applied to wastewaters. Wastewaters are not usually composed of single substrates, but a variety of compounds all with different biodegradation rates. If wastewaters are characterised using global parameters, such as BOD, COD, or TOC, and one of the preceding models is used, complications may arise (Hao and L i 1987, El-Rehaili 1994, Jones 1973). The kinetics measured using one parameter will not be applicable to the other parameter, and i f COD is used, the non-biodegradable fraction must be accounted for. Jones (1973) pointed out that i f the kinetics of a multiple substrate mixture are measured using a global substrate measurement (such as COD), the measured kinetics will be dominated by the substrate component with the greatest degradation rate. There have been a number of attempts at dealing with the multicomponent nature of wastewaters. The approach taken by the IAWQ model involves separating the substrates into different fractions depending upon their kinetics and physical properties. Another approach assumes the presence of a large number of substrates in the wastewater and that the substrate removal rate decreases as individual substrate components are removed (Grau et al 1975). This second approach will be referred to as the multicomponent model. 53 Substrate Interactions When dealing with microbial growth on mixed substrates, a number of different substrate interactions are possible. The IAWQ model and the multicomponent model assume that the various substrates in the wastewater do not interact. This assumption is based upon the argument that although in batch cultures sequential substrate removal might occur, simultaneous substrate removal is more likely to be the case in a continuous steady state system. Furthermore, the substrate removal rates are assumed to be independent of the presence of other substrates. In the case of simultaneous substrate removal, the various substrates may still influence the other substrates removal rates. A few models for simultaneous use of multiple substrates have been proposed, such as the one by Gujer and Kappeler (1992), where the growth rates on the individual substrates are multiplied by a correction factor (based upon a simplification of competition kinetics). The correction factor is to temper an unrealistic increase in growth rate that would be obtained by simply adding the growth rate due to the two substrates. No experimental evidence was given for this sort of correction factor. Orhon has proposed that competitive inhibition enzyme kinetics be used to model multiple substrate utilisation, and presented the following expression (Orhon and Artan 1994, pp 138-140): (3.20) dX S7X (3.21) 54 This model is theoretical and not based upon experimental evidence. Both models (3.20 and 3.21) predict that the presence of a second substrate with a slower degradation rate will markedly decrease the degradation rate of the first substrate. There have been a few cases of substrate interactions using mixed cultures reported in the literature. Wuhrmann (Discussion of Tischler and Eckenfelder 1968) points out that substrate removal rates may not always be independent. In a study using glucose and galactose as substrates, it was found that the removal rates during a batch test were lower when both substrates were present. Another study (Zollinger et al 1964) found that galactose was only removed from solution once all of the glucose had been removed, and sequential substrate utilisation for these substrates was hypothesised. In a study using continuous reactors (Ghosh and Pohland 1972) it was found that at hydraulic residence times typical for activated sludge units (>4 hours), the presence of galactose had no effect on the glucose removal rate. It was also found that the yield was lower when both substrates were present, and the evidence suggested that the biomass used glucose as a carbon source and galactose as an energy source. In contrast to the studies with mixed cultures, there have been many studies of mixed substrate utilisation using pure cultures (see Harder and Dijkhuizen 1982, 1983, Bull and Brown 1979). Since most of these studies were performed using continuous cultures, the results should be more applicable to continuous wastewater treatment systems than the results obtained from batch tests. A general conclusion from these studies is that at low dilution rates mixed substrates are utilised simultaneously, while at higher dilution rates, one ofthe substrates is likely to be used preferentially before the other substrates are used. This behaviour may be due to the high levels of ATP generated during high growth conditions, which 55 leads to metabolic controls such as catabolite repression and inhibition (Philbrook and Grady 1985). In batch tests, i f carbohydrate mixtures are used, substrate removal is likely to be sequential. If one of the substrates is an organic acid, simultaneous removal can be expected. The pattern of dual substrate removal may also depend upon the previous culture history (Narang 1997). There are exceptions to these generalities; some mixtures are used simultaneously at all dilution rates, while others are used preferentially at all dilution rates. From the evidence given above, the most likely response of an acclimatised mixed, continuous culture operating at growth rates typical of wastewater treatment systems appears to be simultaneous utilisation ofthe substrates. The question which remains to be answered is whether or not the individual substrate removal rates are independent of the other substrates, or a function of the other substrates present. In contrast to the predictions of equations 3.20 and 3.21, the growth rate on a substrate mixture has been found to be greater than the growth rate on either ofthe individual substrates (Narang 1997). The prediction of possible substrate interactions is further complicated i f the bacterial culture is not acclimated to the wastewater, such as during a shock load to a wastewater treatment system. In addition to substrate removal rates and growth rates, biomass yield is also influenced by substrate mixtures. The yields during mixed substrate growth appear to be additive. Multicomponent Kinetics If each substrate is removed independently of the others, then the overall removal rate (as measured by BOD or COD) will be a summation of the individual substrate 56 removal rates (figure 3.4). Tischler and Eckenfelder (1968) showed theoretically that i f the individual components are removed according to zero order kinetics, the overall removal rate can be approximated using first order kinetics (this was found empirically -the authors expressed no preference for a pseudo first order reaction over any other pseudo order reaction). They then showed experimentally that glucose, analine, glycine, phenol, and acetate removal rates were constant during a batch test (zero order kinetics), and the overall removal rate was approximated as the summation of the individual removal rates and modeled using first order kinetics. This concept was also demonstrated using glucose, phenol, and sulfanilic acid as substrates (Siber and Eckenfelder 1980). Chudoba (Discussion of Tischler and Eckenfelder 1968) has also found zero order substrate removal rates for a range of substrates. Glucose and benzoate were removed from batch culture at constant rates, which were additive for the mixtures (Zollinger et al 1964). From these studies it appears that the metabolism of individual substrates by mixed microbial communities is not influenced by the metabolism ofthe other substrates present. A l l of these studies used batch tests which are more likely to exhibit substrate interactions than continuous cultures due to the high initial substrate concentrations. Grau et al (1975) developed an empirical equation which models the batch test removal of substrate from a multicomponent wastewater assuming that the overall removal rate is equal to the summation of individual zero-order removal rates: dt c J \ S (3.22) n is not limited to integers, but for ease of analysis, n is usually taken to be either 1 or 2. This equation is different from the limiting case ofthe Monod equation when S « K M (equation 3.10), in that the substrate removal rate depends on the ratio of remaining 57 substrate to original substrate as opposed to simply the substrate concentration. This modification is to account for the decreasing removal rate as components are eliminated from the wastewater. Adams and Eckenfelder (1975) compared equation 3.22 with the first order approximation of the Monod model. Equation 3.22 predicts that the effluent quality will vary with the influent strength and this was supported by the experimental data. The Monod model predicted much less effluent variability than was observed. Equation 3.22, with n=l, developed for multicomponent effluents, is similar to the equation 3.19 (which was also developed to bypass the limitation of the Monod model to relate effluent substrate concentration to influent substrate concentration), if So is very large compared to S and b and M. If n=l, equation 3.22 is equivalent to the model of Grady and Williams (1975). Equation 3.22 is a generalisation of the empirical first order approximation of the Monod equation, and appears to fit experimental data very well (Grau et al 1975). Unlike equation 3.19, and that of Elmaleh and Ben Aim (1976), equation 3.22 does not predict a maximum substrate uptake rate. Since most wastewater treatment plants operate under low loading conditions, this is unlikely to be a major concern except during shock loads. An alternate explanation for the success of equation 3.22 is the production of soluble microbial products (SMP) (Rittmann et al 1987). SMP have been found to be produced in proportion to the initial biodegradable organic concentration and also to the concentration of active biomass. When the feed is composed of only one or a few simple compounds, the concentration of organic compounds remaining in the effluent is often found to be proportional to the initial substrate concentration, however the chemical but of a different chemical composition than the original substrate. Chudoba (1985) found 58 that refractory products may amount to 1-10% of the substrate consumed. Artan et al (1989) predict that there will be no differences between the performances of completely mixed and plug flow activated sludge units (assuming substrate removal follows the Monod model) because they will produce the same amount of microbial products. If SMP formation is ignored, the Monod equation, as well as equation 3.22, predict that a plug-flow reactor should have superior treatment efficiency compared to a completely mixed system. Most studies have found no difference between the two process configurations. During plug flow operation, the different substrates may be removed sequentially instead of simultaneously as assumed by equation 3.22 (Braha and Hafher 1987). The biomass behavior during plug flow operation should be similar to the biomass behaviour during the batch tests used to measure the model coefficients. Different microbial populations wil l develop under different flow regimes, and the model parameters of plug flow, completely mixed, and batch reactors may not be comparable. Argaman (1991) offers a different explanation for the similarity of treatment efficiencies of plug flow and completely mixed units. If the removal rate is a summation of zero order reactions, then equation 3.22 should be integrated as a zero order reaction, even if n is 1 or 2. For a zero order reaction, there will be no difference between batch, plug flow, or completely mixed continuous reactors. The resulting equations are as follows. If n=l from batch test data, then (3.23) If n=2, then 59 S = 1 + (3.24) A common mistake is to use k obtained from a first order curve fit to batch test data (equation 3.23), and use equation 3.24 to design a continuous stirred tank reactor (which is the first order integration of equation 3.22 for a CSTR). This will result in a major oversizing of the activated sludge system. Equations 3.23 and 3.24 may be used instead of equation 3.7 in modeling the activated sludge process. IAWQ Model The multicomponent model accounts for the fact that there are many different substrates present, but still uses global substrate measurements (such as BOD or COD). Greater accuracy would be obtained by measuring each substrate individually. Often, the substrates in wastewaters can be broadly classified into only two different groups based upon removal rates: those rapidly biodegradable, and the slowly biodegradable. Dual substrate models, such as the IAWQ activated sludge model #1, were developed to more realistically model the activated sludge process (see Orhon and Artan 1994 for a thorough treatment of this model). These models divide the organic components of domestic wastewater into various fractions, as shown in figure 3.5 (Henze 1992). The biodegradable component of the wastewater is divided into readily and slowly biodegradable fractions. The division was at first based on the obvious physical property of the wastewater. The readily biodegradable fraction was assumed to be the water-soluble components, while the slowly biodegradable components were assumed to be the particulate matter. It is now realised that the slowly biodegradable component can 60 400 350-; 300-100 50-| 0 Soluble Suspended Before Treatment Soluble Suspended After Treatment -45 -40 -35 -30 -25 -20 CD Q o u 15 10 t-5 0 Figure 3.5 Breakdown of COD before and after activated sludge treatment for a typical domestic wastewater. From top to bottom: inert soluble • , readily biodegradable rapidly hydolysable • , slowly hydrolysable , heterotrophs , inert particulates Figure 3.6 a) IAWQ model with death - regeneration b) IAWQ model with endogenous decay. S - soluble substrate; Sp - soluble microbial products; Xg -particulate substrate; X J J - biomass; X p - particulate micobial products. 61 contain soluble and colloidal matter, as well as particulates. The removal rate of the readily biodegradable component is assumed to follow the Monod equation. The slowly biodegradable fraction is first adsorbed, or enmeshed, by the active biomass, then hydrolysed to readily biodegradable substrates. The hydrolysis step is considered to be the rate limiting step, so the adsorption step is not accounted for in the IAWQ model. Hydrolysis is modeled using a function similar to Monod's, but this is usually simplified to a first order reaction. The readily biodegradable organic substrates produced by hydrolysis may be removed independently of the original readily biodegradable components, as found by Spangers and Vanrolleghem (1995), or there may be substrate interactions as predicted by equation 3.20 (Gujer and Kappeler 1992) or 3.21 (Orhon and Artan 1994). The easiest interpretation is that the readily biodegradable organics produced by hydrolysis are the same as those originally present in the wastewater. The yields of the two fractions are usually considered to be equal, and commonly taken to be 0.67 gCOD/gCOD (Kapeller and Gujer 1992). The particulate matter in the wastewater is also divided into different components. These are the particulate degradable organics already discussed, the inert particulate matter, and the active biomass. Active biomass is converted to inert particulate matter through microbial decay. There are two ways to model microbial decay (figure 3.6); both give similar results for aerobic systems, and both offer advantages over just using equation 3.5 in that both predict the formation of inert particulate matter from the decay of active mass. This allows the decrease ofthe viable fraction of biomass with increasing SRT to be predicted, but it also makes it difficult to measure the amount of active 62 biomass present, since VSS measurements will measure the inert solid matter along with the active biomass. The first method of modeling microbial decay (figure 3.6b) is by assuming endogenous decay. The biomass decays according to a first order reaction, and a fraction fex is converted to inert particulate matter while the rest (l-f e x) is oxidised during the degradation process. The second method (figure 3.6a) is the death regeneration approach. In this model the biomass decays according to a first order reaction, with a fraction fp x being converted to inert endogenous residue, and the rest (l-f p x) converted into slowly biodegradable (particulate) substrate. This substrate is then hydrolysed to soluble substrate and converted into active biomass with an assumed yield of 0.67. In the measurement of the decay rates by respirometry, it is assumed that growth on the readily biodegradable substrate, the step which requires oxygen, is the rate limiting step. At low temperatures, the hydrolysis step will be slower than the death or growth step, and microbial decay will be underestimated (Lishman and Murphy 1994). The IAWQ model accounts for soluble microbial products by assuming that they are produced in relation to the growth and decay rates. Since activated sludge plants are operated at low growth rates, the inert organic compounds produced are usually assumed to be mainly due to microbial decay. The endogenous decay model can easily be modified so that some active biomass is converted into inert particulates, some into soluble products, and the rest is oxidised. 63 3.4 Mass Transfer Another complication that arises in the modeling ofthe activated sludge process results from the three phase nature ofthe mixed liquor. The substrate might be both soluble and particulate, the biomass tends to grow in floes, and the oxygen required for metabolism is usually added in gaseous form (air bubbles). A physical mass transfer step may be a rate limiting step in the overall substrate removal process. Mass transfer will probably not interfere with the measurement of PMAX, at high substrate concentrations, as the reactions will be kinetically limited. However, at low substrate or low oxygen concentrations (mass transfer may affect both substrate and oxygen availability), the reactions may be mass transfer limited. If this is the case, any measurement of KM will be affected, and K M will be overestimated (Shieh 1980, Horvath and Engasser 1974). Sometimes the K M calculated in the presence of mass transfer resistance is called the apparent K M , but this may lead to confusion when the results are applied to a reactor with different fluid mechanical conditions. There are several different methods for including mass transfer steps in the activated sludge models. The Powell equation (equation 3.13) accounts for diffusion of the substrate across the cell membrane into the cell with the assumption that the substrate flux across the cell membrane is equal to the substrate utilisation rate. Another approach assumes that the substrate must adsorb onto the floe (which has a finite capacity to hold substrate) before the biomass has access to the substrate. This approach assumes a heterogeneous system. If a homogenous system is assumed, the main mass transfer resistance may be in the bulk solution (external mass transfer resistance) or in the microbial floe (internal mass transfer resistance),' and both substrate and oxygen mass transfer may be rate limiting (Cussler 1997). In this section adsorption and external and internal diffusion will be discussed. The Powell equation was discussed in section 3.2. Adsorption The first step in soluble substrate utilisation may be adsorption onto the floe, or the bacteria (Tsezos and Bell 1989). Following adsorption, the substrate is transformed, possibly into storage products, before being metabolised for microbial growth. Biosorption of the substrate in the wastewater prior to metabolism can serve to dampen influent fluctuations of substrate concentration, and help to provide a constant effluent quality (Fujie et al 1997). Many models incorporating a mass transfer step have been developed (Fujie et al 1988, Sheffer et al 1984). The model of Andrews (Busby and Andrews 1975) seems to be the most well developed and accepted. The substrate is treated as one component, but must undergo two reaction steps in order to be utilised by the biomass. The removal of substrate from the wastewater is assumed to be by the pathway shown below. S • > X s > X A > X i Substrate Stored Active Inert The total M L V S S is composed of three fractions: stored mass, active mass, and inert mass. The attachment step is modeled as follows: dt ~ s S+Ks j XT-^--Xs (3.25) The synthesis step is assumed to follow Monod kinetics 65 Substrate Figure 3.7 Effect of external (to the floe) mass transfer on the Monod growth rate vs. substrate concentration relationship. 1.4 Substrate Figure 3.8 Effect of internal mass transfer resistance on the Monod growth rate vs. substrate concentration relationship. 66 The formation of inert mass due to microbial decay follows a first order reaction. ^T = kIY1XA (3.27) at This model predicts higher substrate removal rates when there is no initial stored mass. This is an alternate explanation for decreasing removal rates during a batch test, with the high initial rate of substrate removal being due to adsorption. As the floes become saturated, removal occurs by the slower metabolic processes. This predicts a higher ratio of SUR to OUR at the beginning of the batch test than during the rest of the test, which is often found experimentally. This model can explain the increased efficiency of the contact stabilisation activated sludge process. Stored mass is converted into active mass during the stabilisation step, allowing for maximum biosorption of substrate during the contact step. The amount of biosorption which occurs depends upon the operating conditions of the activated sludge process. Biosorption was found to increase with the SVI and the population of filamentous microorganisms (Pujol and Canler 1992). Adsorption also depends upon the operating temperature (Bell and Tsezos 1987), pH, cell age, and cell viability (Amy et al 1988). Substrate removal by biosorption has been found to increase as the SRT decreases (Jacobsen et al 1993). For the treatment of B K M E by activated sludge, biosorption probably plays a role in the removal of the more difficult to degrade compounds. Biosorption is often related to the octanol water partition coefficient of the compound (Steen and Karickhoff 1981). Increasing hydrophobicity increases biosorption. Adsorption is an important step in the removal of A O X during B K M E treatment in a lagoon (Amy et al 1988). 67 External & Internal Diffusion Mass transfer effects may be accounted for by using one of two standard chemical engineering mass transfer models. The first assumes that mass transfer is proportional to the concentration difference between the substrate in the bulk fluid and the floe phase. The substrate concentration in the floe is assumed to vary linearly from the center ofthe floe to the floc-liquid interface. This approach is termed the external diffusion model. In the second method, the solute flux is proportional to the solute gradient. The substrate concentration in the floe wil l be a minimum in the center of the floe, and close to the bulk concentration at the interface ofthe floe and bulk solution. This method is termed internal diffusion. For external diffusion, the following equation may be used: i ? = h (- s° - v - ^ xi f r s (3- 28) Equation 3.28 is usually solved by assuming steady state (Fforvath and Engasser 1974). This results in an equation similar to Powell's (3.12). The effect of external mass transfer is shown in figure 3.7. When the external mass transfer resistance is large, the reaction rate is independent of the microbial kinetic coefficients. The external mass transfer coefficient, h, is proportional to floe surface area. For slurries, there is relatively little movement between the suspended particles and the fluid, which may make external mass transfer important. However, it has been shown, both for oxygen and glucose, that external mass transfer has much less of an effect on the overall kinetics than internal mass transfer (Baillod and Boyle 1970, Mueller et al 1966). External mass transfer may be accounted for in the internal mass transfer model by using the appropriate boundary conditions (Ramachandran 1975). The 68 continuity equation, assuming constant density, constant diffusivity and Monod kinetics, is: f =-«vs+ DV2S-iW^T, ( 3 - 2 9 ) at r i M + b This equation is solved assuming steady state, and the assumption that the advective term is insignificant (u=0), i.e. the mass transfer is solely due to molecular diffusion. The floes are usually assumed to be spherical with a constant size and even distribution of microorganisms. The non-linearity of the Monod expression means that this equation must be solved using numerical methods (Ramachandran 1975). With these assumptions, equation 3.29 becomes: d2S 2D3S . S £ T T + T- = VMAXPT——Q (3-30) or r or KM + S The effect of internal mass transfer on kinetics is shown in figure 3.8. Unlike the situation for external mass transfer, the reaction rate is always a function of the microbial kinetic parameters (Horvath and Engasser 1974). According to Benefield and Molz (1983) the assumption of pseudo steady state is valid. They calculated that the time required to reach equilibrium in the floe is approximately 2.2 minutes, compared with hydraulic retention times in the order of 4-12 hours for a typical activated sludge plant. Equation 3.30 predicts that small floes are better for reducing mass transfer effects. This is in contrast with the large bioflocs required for good settling in the secondary clarifier. Equation 3.30 also predicts that as floe size increases, there will be zero nutrient concentration in the middle ofthe floe, perhaps leading to loss of viability at the floe centers. If both substrate and oxygen are limiting, more of the floe wil l be viable than if just one were limiting due to greater nutrient penetration (Benefield and Molz 69 1983). Benefield and Molz (1983) showed that mass transfer effects can be quite important in determining the response to transient operating conditions. In particular, the microbial kinetics can change from substrate limited to oxygen limited and back to substrate limited during the course of one retention time during a shock load. A more recent study ofthe effect of internal diffusion on the response to shock loads used the IAWQ model (Tyagi et al 1996). Their model accounted for the natural variability of floe size in a typical activated sludge process. It was found that i f mass transfer was ignored, the effect of the shock load would be underestimated. The same kinetic coefficients from the literature were used for both cases. If batch test data are interpreted according to the IAWQ model with mass transfer effects, the coefficients obtained, in particular K M , wil l be different than i f the same data were interpreted according to the IAWQ model without mass transfer effects. Using coefficients obtained from one model in another model in order to simulate shock loads will lead to erroneous results. Logan and Hunt (1988) have taken the observation that bacteria form floes at low nutrient concentrations, and proposed that there is greater substrate availability for bacteria in floes than for free bacteria due to advective mass transfer. If mass transfer is modeled using advection instead of diffusion, similar results are obtained, implying that all of the previous results may also be obtained assuming advective mass transfer is limiting. Comparing free bacteria with bacteria growing in floes, assuming both are mass transfer limited, it was calculated that the mass transfer to the floe could be greater than the mass transfer to the free bacteria i f advective flow was assumed. If the floes interact with rising air bubbles, this will further increase the mass transfer rate. These results give a competitive advantage to bacteria growing in floes in low nutrient environments (mass 70 transfer limited), such as the activated sludge process, and explain why there is no change in cell viability throughout the floe (Wagner et al 1994). Similar to adsorption, diffusion depends upon the activated sludge operating conditions. It has been found that the effective diffusivity changes with SVI, suggesting that changes in the diffusivity are due to floe porosity (Eliosov et al 1996). The importance of floe size on diffusion has been demonstrated by measuring activated sludge before and after disruption of the floes by ^lending (Mueller et al 1966, Baillod and Boyle 1970). The overall reaction rates increased as the floe size decreased. The diffusivity was found to increase with floe diameter. Oxygen Transfer At dissolved oxygen concentrations greater than 2 mg/1, oxygen should not be rate limiting. Below 2 mg/1, the microbial kinetics may be limited by oxygen. If substrate is already limiting microbial growth, the growth rate will be limited by both substrate and oxygen. Dual limitation may be modeled using the double Monod model or the non-interacting model. The double Monod model (equation 3.31) is most often used, and is supported by experimental evidence (Bae and Rittmann 1996). r * ) { a ) KK, +sj V K2 +S2J (3.31) If S 2 » K2 (as is usually the case for oxygen, K02 < 0.2 mg/1), then the term for S 2 (DO) may be ignored, which results in the Monod model. If oxygen is utilised faster than it can diffuse into the floe, the oxygen concentration in the floe will be lower than the dissolved oxygen concentration in the bulk solution, and oxygen may be a limiting substrate even when the bulk concentration 71 is greater than 2 mg/1 (Beccari et al 1992, Siegrist and Gujer 1987, Mueller et al 1966). The diffusion of oxygen into the floe is assumed to follow the same rules as the diffusion of substrate into the floe (equation 3.30, with the reaction term replaced by equation 3.31). The oxygen diffusivity depends upon many factors, such as the type of wastewater, and the properties of the floe. Dissolved oxygen diffusivity through water is temperature dependent, increasing with increasing temperature (Lin et al 1998). This is offset the decreased oxygen solubility at elevated temperatures. 3.5 Temperature A common rule of thumb is the microbial growth rate doubles for every 10°C increase in temperature. The maximum microbial growth rate can be described as a function of temperature in accordance with the Arrhenius expression For most wastewater treatment plants, where the temperature varies in the range of 10°C to no more than 30°C, this equation is usually simplified to \^  = n 2 0 e ( r - 2 0 ) (3.33) As the temperature increases above 30°C, the decay rate increases, resulting in a decrease in the net growth rate as the fraction of active enzymes decreases. Enzyme denaturation also follows an exponential relationship with temperature. kd=Ae-{EIRT) (3.34) If enzyme inactivation is assumed to be irreversible, the following equation is obtained 72 Ae -(£, / RT) ^MAX ~ j K e ' ^ E l ( 3 . 3 5 ) This equation accounts for both the increase in reaction rate as the temperature increases, and the subsequent decrease in reaction rate when the temperature rises too high. Equations ofthe form of 3.35 were developed for analysing enzyme kinetics, and become empirical when applied to microbial growth rates, but nevertheless fit the data. The assumption is that there is one rate controlling reaction, and this reaction is rate controlling at all temperatures. Many other empirical correlations between growth rate and temperature have been developed. Some of these fit the data better than model 3.34, but at the expense of introducing more parameters. A l l of the correlations are empirical in nature. 3.6 pH If one enzymatic reaction is considered to be the rate controlling reaction, then the effect of pH on this reaction will also be the effect of pH on the overall growth rate. If the following equilibrium is assumed OH~ H* ^inactive ^active ^ *S ^ inactive H* • OH' the resulting effect of pH on growth rate is K, [H>] Alternately, the enzyme substrate complex may be assumed to ionised instead ofthe enzyme. This results in the following KH+ 73 Both of these equations predict a bell shaped curve of growth rate vs. pH, which is commonly observed. A number of empirical equations have been proposed as well, but are used less frequently. Denaturation by pH changes occurs by the enzyme being protonated or deprotonated. The relationship between ka and the pH may be modeled with the following equation (Laidler and Buntins 1973): K\+ K2*pH+ K3*pH2 K4+K5* pH+.K6*pH2 a v A . vc - „ 7r . vc _ „ url v J where K I , K2, K3, K4, K5, arid K6 are related to equilibrium constants for the various sites of pH denaturation, and rate constants for the rate of denaturation. 3.7 Summary Microbial kinetics are usually modeled with Monod's equation. For a given biomass and a given substrate, the Monod model often saturates too slowly to fit the data. This observation may be accounted for by modeling two reactions in series, with the first reaction usually being a mass transfer step such as diffusion into the floe. When mixed substrates are used (and measured with a global parameter), the Monod half saturation constant appears to be a function of the initial substrate concentration, and not a constant. This observation may be accounted for by either the production of soluble microbial products, or by the multiple component nature of the wastewater, with each substrate having unique model coefficients. A common observation of municipal wastewater treatment has been that there are two main processes occurring, one much faster than the other. The model of Busby and 74 Andrews (1975) accounts for this by dividing the substrate utilisation process into two steps (adsorption and biodegradation), while the IAWQ model divides the substrate into two components (readily and slowly biodegradable). Both of these models help to overcome some major failings of assuming that there is only one substrate, or only one reaction, occurring in an activated sludge unit. The models may be further refined by accounting for microbial decay, which is responsible for the decrease in yield with increasing SRT. Decay may proceed by endogenous metabolism, or cell lysis, with the resulting liberated organic compounds utilised by the remaining viable bacteria. If microbial decay is assumed to produce non-biodegradable particulate matter, then the active fraction of the sludge may be modeled as a function of SRT. Most ofthe models fit the experimental data fairly well. Rarely are the models tested with a secondary set of experiments. For example, i f coefficients are obtained in a batch test, they will not necessarily be applied to a continuous system which is treating the same wastewater for verification. There are undoubtedly many different reactions occurring during the activated sludge process, and it is difficult to know which ones are the most important. Most ofthe models have many parameters and can fit most sets of data, making the choice sometimes seem arbitrary. Assumptions made in interpreting batch test data may not be valid, resulting in errors when the findings are applied to continuous treatment plants. 75 Chapter 4 Literature Review - Model Parameter Measurements 4.11ntroduction Microbial model parameters depend upon many factors, including the culture history and the nature of the assay used for their measurement. Two very important aspects of microbial assays, especially when dealing with mixed cultures, are the initial food to microorganism ratio (f/m), and the length of the assay. Results from tests with a short duration and a low initial f/m are highly dependent upon previous culture conditions. For example, pure culture studies, comparing the growth rates of biomass removed from continuous reactors operating at different SRTs, found that K M decreased and PMAX increased with increasing dilution rate (Templeton and Grady 1988). The Monod coefficients were higher when measured by a batch test with a high initial f/m compared to when measured by a fed-batch test with a low initial f/m (Templeton and Grady 1988, Chudoba et al 1992a). For batch tests, as the initial f/m increases, the growth yield is often observed to decrease. The initial response to a greater substrate concentration than that to which the bacteria are acclimated will be: metabolic uncoupling and a resultant increase in energy spilling (Liu 1996), microbial product formation, or storage. A l l of these response will result in a lower yield. Once the biomass has adapted to the new conditions by increasing the growth rate, more energy will be expended towards growth than towards non-growth functions, and greater amounts of energy are required for cell growth resulting in a lower yield (Chudoba et al 1992a). 76 The fact that the measured Monod coefficients and microbial yield depend upon the initial f/m implies that the bacteria are adapting to the new environmental conditions of the batch test. The higher that the initial f/m is, and the longer the assay, the greater the chance that the biomass will adapt. This adaptation may take many forms, such as the selection for species with high growth rates, expression of enzyme systems with higher capacity, or production of a greater amount of anabolic enzymes for reduction of metabolic uncoupling (see section 2.3). Assay methodologies may be categorised by the initial f/m based, upon the response of the biomass, into those that employ low (f/m < 0.04), intermediate (0.04 < f7m < 2), and high (f/m > 2) f/m ratios (table 4.1). Values of the f/m ratio greater than 0.04 (mg COD/mg VSS) may result in changes in microbial physiology (Grady et al 1996). If the f/m is greater than 2, the microorganisms will multiply, enriching the culture for organisms with the greatest growth rates on the available substrates (Chudoba et al 1992a). Grady et al (1996) have proposed the use of the term extant (and the subscript e) to describe coefficients which are measured without giving the biomass time to adapt to the assay conditions, and the term intrinsic when the biomass is allowed to fully adapt and reach its maximum potential (the growth rate is unrestricted). Extant kinetics represent the capabilities of the culture at the time of sampling. This section describes various methods for determining activated sludge kinetics, most of which are for measuring extant kinetics. For this project, the actual performance of the bioreactors under real conditions (low substrate, high biomass concentrations), is of more interest than the hypothetical performance under unrestricted growth conditions. 77 This requires that batch tests employ a low initial f7m, to minimize bacterial growth and metabolic changes (i.e. enzyme induction). Table 4.1 Effect of f/m ratio F / M / Description Assays F / M < 0.04 Measures extant kinetics i f short term . AOUR, infinite dilution, pilot plants 0.04 > F / M < 2 Possible changes in microbial physiology IAWQ test, adsorption tests F / M > 2 Measures intrinsic kinetics, growth will occur, possible changes in population Growth tests The analysis of batch test data usually involves many assumptions, which may not always be valid depending on the circumstances. One assumption is simultaneous removal of all of the substrates present. The tests start with a high substrate concentration (compared to a continuous unit), so there is a possibility that the substrates (the readily biodegradable and slowly biodegradable) may be removed sequentially instead of simultaneously as assumed. Microbial growth is assumed to be negligible during the test (for low initial f/m). At the start of the test, the bacteria are taken from a low growth environment (continuous system) and exposed to an excess of substrate (batch test). As the bacteria adapt to the new conditions their growth rate may increase. The yield is assumed to be constant. As the bacteria adapt to their new environment, it is possible that the yield will change. It is commonly observed that the yield is related to the growth rate, even after accounting for endogenous metabolism. The growth rate increases when the bacteria are exposed to an excess of substrate, so the yield may increase as well. 78 4.2 Continuous/Pilot Plant Setups The traditional approach for determining activated sludge model parameters is to operate a series of continuous activated sludge units at different SRTs, measuring the treated effluent substrate concentration at steady state. Unlike the other tests described in this section, these are long term measurements. It is generally assumed that 3 to 10 SRTs are required for a continuous unit to reach steady state (Zaloum 1992), which for a SRT of fifteen days may require 150 days to reach steady state. The data are interpreted according to the Monod model with microbial decay (equations 3.7 to 3.9). Y and ka may be calculated from equation 4.1, and a graph of (So-S)/X versus 0. P M A X and K M may be calculated from equation 4.2 and a graph of 1/S versus 0/(1+kd0). Many other linearisations ofthe Monod model are available (Ong 1990), although these two were found to give the best estimates when the error was normally distributed. Sazl = i^ +1 (41) X Y x Y L-i^MAx. §J£ _ _ L (4 2) S KM \+kdQx KM The data may also be interpreted assuming a first order reaction (equation 3.10), or using equation 3.22. A drawback to continuous tests (other than the time involved) is that the biomass will adapt to the different operating conditions, so that each measurement used to calculate the stoichiometric and rate constants will come from a different microbial population. The parameters will be an average ofthe parameters from each ofthe different populations. The results will be acceptable for modeling steady state operation, 79 but extrapolation outside ofthe test conditions or prediction of microbial behaviour during shock loads and transient operating conditions may not be reliable. 4.3 Fed Batch Test In order to avoid the excessive time required for operating continuous systems at different SRTs, and also the complications arising from changing populations over the long tests periods, the infinite dilution test was developed. The infinite dilution test (Williamson and McCarty 1975) involves continuous addition of concentrated feed to a bioreactor and withdrawal of samples for analysis. Due to the conditions in the bioreactor it is assumed that steady state with respect to substrate removal is rapidly reached (less than one hour), that the biomass does not have time to adapt to the batch test conditions, and no shifts in population will occur. The substrate to be tested is fed to the bioreactor at different feed rates, from which the substrate uptake rates are calculated. The steady state substrate concentrations in the bioreactor are measured for each feed rate, and a curve of substrate uptake versus substrate concentration may be obtained. The assumption that the biomass does not have time to adapt to the new operating conditions is validated by work on shock loads to continuous units, where lags in the biomass growth rate upon an increase of substrate loading were observed (Selna and Schroeder 1978, Mona et al 1979). This is probably due to the low microbial growth rates and the physiological conditions ofthe bacteria from activated sludge units (Daigger and Grady 1982b, Manickam and Gaudy 1985). The infinite dilution test has been modified in order to increase the statistical accuracy of the assay while minimizing the number of experiments (Templeton and Grady 1988, Philbrook and Grady 1985). The modified test assumes that the Monod 80 model will fit the data a priori, and involves measurement at only the maximum substrate uptake rate, and one half the maximum substrate uptake rate (which corresponds to K M for the Monod model). The modified fed batch test will not give any indication i f an alternative model to Monod is preferable, or whether there are mass transfer effects. 4.4 Batch Tests Batch tests consist of taking a sample of wastewater, and adding biomass. The substrate concentration, OUR, and biomass concentration are measured to determine the substrate removal rate or the biomass growth rate. If many different starting substrate concentrations are used, the initial rate (growth or substrate uptake) at each substrate concentration may be measured to obtain the relationship between rate and substrate concentration. A more efficient method involves obtaining the model parameters from a single batch test. If substrate concentration is measured over the duration of the batch test, the conefficients may be calculated by curve fitting the integrated Monod equation to the data (Robinson and Tiedje 1983). The starting conditions are very important for this sort of test, as they wil l determine i f unique parameter estimates may be obtained If the data follow the Monod model, unique estimates for all of the parameters wil l not be attainable i f the initial substrate concentration is less than K M , or very much greater than K M Another possibility for measuring the coefficients from a batch test is to measure the OUR and substrate concentration throughout the batch test, until all of the substrate has been removed. The OUR is used as a measure of the substrate removal rate. A plot of the OUR vs. substrate concentration can be used to obtain the Monod constants (Huang et al 1985). 81 When substrate is measured by global parameters (such as BOD, COD or TOC) during a batch test involving a multicomponent wastewater, the removal rates will probably not follow the Monod model, but will follow a first or second order model (equation 3.22, Grau et al 1975). If the substrate removal goes through two distinct phases in a batch test (a initial high substrate removal rate followed by a sustained slower removal rate), the IAWQ model, or the model of Busby and Andrews (1975), may be more applicable to the data than the unmodified Monod model. 4.5 Respirometry Introduction As applied to wastewater treatment, respirometry is the study of bacterial metabolic rates through the measurement of bacterial respiration rates. In order to measure the biodegradation rates of the organic compounds in the wastewater, an accurate assay must be available to measure the various compounds, especially when low substrate concentrations are used. Analyses for substrate tend to be time consuming, inaccurate, or non-existent, so respirometric methods are often used as a surrogate. The dissolved oxygen concentration can be measured accurately, immediately, and continuously. If biomass (X), substrate (S), and dissolved oxygen (O) are given in the same basis (i.e. COD), a simple mass balance may be derived (equation 4.3). The change in biomass concentration is equal to the amount of substrate which is removed less the substrate which is oxidised. AX = AS + AO (4.3) 82 Based upon equation 4.3 and the yield coefficient (equation 3.3), the OUR is proportional to the substrate degradation rate and the biomass growth rate. The equations from section 3.2 can be used to correlate the OUR to the substrate concentration. The relationship based upon the Monod model is shown in equation 4.4. dO _ Y S 1 S S - -—»™*Y^rl= T ^ q M A x X Y ~ T s = 0URMAXXT^~S ( 4 4 ) Only a limited amount of information may be provided by respirometry. The yield and the initial substrate concentration can not be calculated independently, but require a separate test, either the measurement of the substrate or the yield. Similarly, the active biomass concentration and maximum growth rate are interdependent, and one can not be calculated without prior knowledge of the other (Dochain et al 1995). Most of the respirometric tests are very similar, but different assumptions have been made, allowing for the calculation of different parameters with the same basic data (see appendix for examples). In most respirometric studies, product formation and energy spilling are assumed to be negligible. The respiration of protozoa and nitrifiers are also assumed to be negligible, as well as their influence on the heterotrophic respiration and growth rate (Mahendraker and Viraraghavan 1995). The OUR due to endogenous metabolism may be accounted for and subtracted from the overall OUR. It is important that the biomass and substrate are measured in consistent units (which accounts for the change in oxidation state as substrate is converted to biomass, as in the measurement of COD). The growth rate and the substrate biodegradation rate are proportional to the OUR (equation 4.4), and the amount of growth and substrate degraded are proportional to the amount of oxygen consumed (equation 4.3), assuming a constant yield coefficient. These 83 relationships, combined with the ease of use and automation, make respirometry ideal for many different purposes. Oxygen uptake rates may be measured a number of different ways, including batch tests and continuous methods (table 4.2). The original methods (Warburg) used closed vessels with air in the headspace. The air in the headspace serves to replenish the dissolved oxygen in the liquid phase as it is consumed by the bacteria. The microbial consumption of oxygen is followed by measuring the oxygen decrease in the headspace, often using pressure (the CO2 produced by the bacteria is removed using a caustic solution). Air or oxygen may be continuously bubbled through the mixed liquor in a respirometer. In this case, the OUR is measured by following the dissolved oxygen in the liquid phase, or the decrease in oxygen between the inlet and outlet gas streams. The mass transfer of oxygen from the gas phase to the liquid phase must be accounted for in the data analysis. Mass transfer limitations from the bulk liquid to the bacteria are usually assumed to be negligible. In order to avoid the mass transfer problems and simplify the data analysis, the respirometer may be completely filled with sample and mixed liquor so there is no headspace, with the DO measured using a DO probe. Since oxygen is not added during the test, the oxygen supply for each OUR measurement will be limited to what is initially present in the mixed liquor and sample. Due to the low solubility of oxygen in water this oxygen supply is relatively small. When it is depleted, the OUR measurement must be interrupted and the mixed liquor re-aerated. The OUR is very simply calculated as the slope of the dissolved oxygen concentration versus time. 84 A l l respirometric methods require sufficient agitation for mass transfer, regulation of the temperature and pH, and the presence of the necessary nutrients for the microorganisms. Table 4.2 Respirometric Methods Method Measurements Comments Headspace Pressure drop, DO in liquid phase. Gas to liquid mass transfer. Continuous O2 supply O2 in gas inlet/outlet, DO in liquid phase. Gas to liquid mass transfer. Filled DO in liquid phase. Limited oxygen supply. Respirometric methods may be broadly classified into two categories, based upon the initial f/m ratio. The first category of tests employs a small microbial inoculum and a large f/m ratio, and the respiration is followed over a long time period, usually at least one day. These tests are used to: • determine microbial growth rates, • measure the biodegradation potential of various substrates, • measure the inhibition of the growth rate by toxic substances, • determine the ability ofthe biomass to adapt to different substrates, • measure the active fraction of the M L V S S , • measure the BOD, and • determine the required treatment time for a given waste. Significant growth occurs during the high f/m tests, so the microbial composition at the end ofthe test may not be the same as the composition at the start of test, as discussed in section 4.1. Low initial biomass concentrations are used in an attempt to ensure that the oxygen uptake is a true measure of biomass growth, rather than oxidative assimilation, storage, or maintenance (Gaudy et al 1988). The high f/m also ensures that the biomass 85 has time to adapt to the new environmental conditions, allowing the bacteria to reach their maximum growth rate regardless of the culture history. If mixed populations are used, as in activated sludge, these kinds of tests will select for the fast growing microorganism, which will have different growth rates and stoichiometry than the slow growers. The second category of respirometric methods uses large biomass concentrations and smaller f/m ratios, and generally take less than six hours to complete. The f/m ratio ranges from low to medium, and the biomass and substrate concentrations are similar to those found in wastewater treatment plants. The validity ofthe assumptions made will depend in large part on the initial f/m, as discussed in section 4.1. The greater the initial f/m, the greater the possibility of changes in biomass properties. Theoretically, the biomass is not given sufficient time to adapt to the new conditions. Therefore, the results of these tests will depend strongly on the culture history. It has been argued that these conditions wil l give growth rate parameters and stoichiometric values closer to those found in an actual wastewater treatment plant than the tests from the first category (Chudoba et al 1992a). These tests have been used to: • determine the biomass concentration, • measure the biodegradation kinetics (with and without inhibition), • measure the hydrolysis rate of the slowly biodegradable.organics, • measure the substrate adsorption onto the biomass, • determine the short term BOD, • determine the readily biodegradable fractions of wastewater, and • determine the required treatment time. 86 Under the conditions used in these tests, endogenous respiration and/or maintenance respiration may be a significant fraction of the total OUR and must be taken into account. The energy expended for cell maintenance may either come from available substrate, or from endogenous mass if no substrate is available. In this literature review, the main focus will be on the methods which employ a low f/m ratio. A few examples will be given of tests using a high initial f/m, specifically those used to determine the maximum growth rate and the active fractions of activated sludge M L V S S . A recent review of respirometric methods, with the emphasis on growth tests, is available (Mahendraker and Viraraghavan 1995). Low F/M Respirometric Methods A O U R and Similar Assay Techniques Over the years, many respirometric methods have been developed. Most are very similar, and make the same assumptions. The majority of respirometric tests were developed independently rather than by expanding on previously published methods. The most thoroughly developed method for determining the kinetics of concentrated biodegradable compounds which has been used by a number of experimenters is the method of Cech and Chudoba (Cech et al 1984, Chudoba et al 1985). This is the method used in this project, and it shall be referred to as the AOUR method and described here briefly. The AOUR method consists of a number of batch respirometric tests. The mixed liquor is transferred from a continuous wastewater treatment unit to a respirometer and aerated. When any substrate initially present has been utilised (a constant baseline OUR is attained), the aeration is stopped and a DO probe is inserted into the respirometer. The 87 endogenous respiration rate is measured. Then, a small amount of substrate is added. This causes a temporary increase in the OUR until all of the added substrate has been metabolised, at which point the OUR returns to the endogenous rate. A typical dissolved oxygen trace for one batch test is shown in figure 4.la. The substrate biodegradation rate is assumed to be proportional to the exogenous OUR (or the AOUR), which is calculated by subtracting the endogenous OUR from the total OUR after substrate addition. The proportionality constant is 1 - Y. SUR=AOUR = AOUR l-Y OCIS The yield is calculated according to equation 4.6, Y = ($-0C)IS=\-(0C1S) (4.6) which is based on the assumption that the substrate which is not oxidised is used for growth, and that all of the added substrate is utilised. OC represents the oxygen consumed due to the added substrate, and does not include the oxygen used for maintenance or endogenous metabolism. The AOUR and the amount of oxygen consumed are calculated by drawing best fit lines through the DO trace. This procedure is repeated using a number of different initial substrate concentrations. Whenever the dissolved oxygen level in the respirometer drops below 2 mg/1, the mixed liquor is reaerated. For data analysis, pseudo steady state is assumed. The bacteria are assumed to immediately reach the metabolic state corresponding to the added substrate concentration. The maximum AOUR achieved during the batch test is assumed to be the AOUR which corresponds to the initial substrate concentration of that test. If these assumptions are valid, the AOUR may be converted to a substrate uptake rate and graphed versus the added substrate concentration 7.0 6 .0H c n>5.0-X o . -a 4.0-> O 8 3.0-Q 2.0-1.0-1.5-T J 1 2 -a> E § 0.9-o U 10.6->-o 0.3-0-0.25-0.2-O 0 . 1 -0 .05-0 O2 "consumed Endogenous OUR Maximum OUR OUR due to substrate metabolism B O2 consumed Maximum AOUR Shaded area is equal to the oxygen consumed due to the added substrate. Endogenous OUR 1 1 1 r Time Figure 4.1 Interpretation of respirometric data. 89 to obtain the relationship between the substrate uptake rate and the substrate concentration. The data is almost universally assumed to follow the Monod model, although any ofthe equations discussed in chapter 3 may be applied to the data. If there is substrate inhibition, then an appropriate model must be used. The curve in figure 4.1a may be transformed into an oxygen consumed (due to substrate) versus time curve (figure 4.1b) (Smets et al 1996b). This is calculated by subtracting the DO value at a given time from the initial DO, correcting for the amount of oxygen consumed due to endogenous metabolism. OC = D00-DO-tOURendogenous (4.7) This allows for easier visual inspection of the data, and is similar to the representation of BOD curves. Another method is to graph the OUR versus time, as is done with the IAWQ method (figure 4.1c). The AOUR test, or variations thereof, has been used to measure various aspects of the activated sludge process. The biodegradation rates of wastewater (all substrates lumped together), and various single substrates, including ammonia and nitrate, have been measured. The inhibitory effects of selected toxic compounds have been determined. The substrate removal rates using biomass from different activated sludge processes were compared on several occasions to determine the effects of different operating conditions or process configurations. Equation 4.6 has been used both to measure the microbial yield of single substrates and also to calculate the amount of biodegradable organics (either as BOD or COD) in unknown samples using an assumed yield. 90 One of the variations on the AOUR method described above includes using two respirometers. The substrate is added to only one of the respirometers. The DO profile from the respirometer which did not receive substrate is subtracted from the one which did receive substrate, after first correcting for differences between the two respirometers (Smets et al 1994, Ellis et al 1996). The OC and AOUR are calculated from the resulting curve as shown in figure 4.1b. The second respirometer is used to correct for changes in the endogenous respiration rate over the course of the test. This will be valid i f the change in endogenous OUR is a function of time, but i f it is a function of the DO then there will be a slight error introduced since the DO in the respirometer which receives substrate wil l be lower than the DO in the control respirometer. Other variations use open respirometers with the mixed liquor aerated throughout the test (Ros and Dular 1992, and Spanjers 1995). The advantage of these methods is the ability to handle large amounts of substrate since there is continuous aeration. In contrast, the AOUR method may only be properly used with substrate concentrations that have an oxygen demand of less than 8 mg/l (at 20°C) before reaeration is required. In open respirometers, the gas to liquid mass transfer step must be accounted for in the data analysis. The main resistance to oxygen mass transfer on the liquid side is assumed to be in the liquid phase, and is represented by the liquid side mass transfer coefficient, K^a. The K L a is measured by turning the air flow off and letting the DO level drop (due to endogenous respiration), then following the rate of increase of DO once the air is turned back on. Once the K^a is known, the exogenous oxygen uptake rate may be measured by adding known amounts of sample to the respirometer. Before sample addition, the DO should be constant, as a steady state is achieved during which the endogenous OUR will 91 be equal to the oxygen transferred from the gas to the liquid phase. Upon addition of substrate, the DO will drop as the OUR increases above the endogenous rate. The exogenous oxygen uptake rate is equal to the K L a times the difference between the DO level before addition of substrate and the minimum DO level after addition of substrate. The oxygen demand, OC, of the added substrate may be calculated by integrating between the DO trace and the endogenous DO. Once the exogenous OUR (or AOUR) and the OC are measured, calculations may proceed as with the AOUR method. One criticism of short term batch respirometric methods is the possibility of the loss of enzyme systems during the endogenous respiration period before the start of the test (Grady and Philbrook 1984). This is unlikely to be significant at the SRT's, and corresponding low growth rates, used in most activated sludge units (Cech et al 1984 also see section 2.3). In a further study of the AOUR method (Chudoba et al 1986), it was found that the measured Monod constants of activated sludge from a completely mixed unit were approximately constant for the first 10 hours of endogenous respiration, although the endogenous respiration rate decreased during this period. A similar experiment on activated sludge from a contact stabilisation process found that the Monod constants increased during the first 18 hours of endogenous respiration. It was argued that this would be a good method to optimise the contact stabilisation process. Similar results were found in a separate study (Drtil et al 1993a), where it was also found that the yield increased with the duration of the endogenous period before the AOUR was measured. The decrease in the endogenous respiration and increase in yield may be due to a decrease in cell maintenance (including energy wasting and spilling) as the bacteria adapt to a higher growth rate caused by the substrate pulses. 92 The AOUR method has been compared to the infinite dilution test for single substrates (Cech et al 1984). It was found that the respirometric method gave higher K M values, perhaps due to substrate adsorption (adsorbed substrate would not be measured during the infinite dilution test). A n apparent dependence of K M on the biomass concentration was also noted, although no explanation was offered. Visual inspection of short term batch respirometric data in the literature reveals that the rate data often saturate faster than allowable by the Monod formula, although the Monod model is used regardless. A possible explanation is that mass transfer effects, which i f ignored, may cause the rates to appear to saturate faster than they actually do. IAWQ and Similar Assays The AOUR test is generally used to determine substrate removal rates, although it may also be used to determine the yield, or the amount of substrate in wastewater. The IAWQ method is used to determine the amount of substrate in wastewater, but it has also been used to determine substrate removal and growth rates. The IAWQ method may be thought of as a AOUR test with only a single large addition of substrate, usually wastewater, with the main goal of measuring the substrate concentration. There are four main differences between the two tests: 1) the endogenous respiration is not accounted for in the IAWQ test, 2) the yield is not measured during the test, but is assumed to be a constant, usually 0.66, 3) the Monod model is assumed, but there is no direct measurement of the half saturation constant, and 4) the test is only performed at one initial substrate concentration. Because of the large amount of substrate present, the mixed liquor requires reaeration many times throughout the test. In order to avoid the 93 reaeration steps, air may be bubbled continuously through the respirometer and the data analysed as described for the open respirometers in the AOUR section. The standard procedure (Ekama et al 1986) involves adding mixed liquor to wastewater and following the OUR. A typical data set is shown in figure 4.2. The maximum substrate uptake rate for readily biodegradable substrate is calculated from the initial OUR, using an assumed yield, and the amount of readily biodegradable substrate is calculated from the area under the curve, using the same assumed yield. The second OUR plateau is said to be due to hydrolysis of the slowly biodegradable substrate. This portion of the curve cannot easily be used to calculate the maximum hydrolysis rate, since, under the conditions of the batch test the hydrolysis rate will be 60 to 80% of the maximum hydrolysis rate. The entire OUR vs. time curve may be modeled to calculate the readily biodegradable substrate biodegradation rate coefficients, and the slowly biodegradable substrate hydrolysis rate coefficients. Hydrolysis may be modeled as a first order reaction (Kappeler and Gujer 1992), or by using a relationship similar to the Monod equation (Orhon et al 1995). A more sophisticated approach was used by Spanjers and Vanrolleghem (1995). The IAWQ model was compared to a modified model which assumed that the readily biodegradable component produced by the hydrolysis of particulate matter has its own set of biodegradation coefficients distinct from the original readily biodegradable component. The modified model fitted the batch test data better (this is not surprising since more parameters were introduced). The same study also demonstrated the importance of the initial f/m ratio in obtaining reliable parameter estimates. 94 Many respirometric tests for measuring substrate removal rates or stoichiometry have been reported on in the literature. Many of these assays do not fit into the classification of the AOUR or IAWQ methods, but share many similarities. One modification of the IAWQ test involves only one OUR measurement so that no re-aeration ofthe mixed liquor is required (Xu and Hasselblad 1996). This is achieved by diluting the mixed liquor and substrate prior to the start of the test. The breakpoint in the OUR curve is the point at which the readily biodegradable component has been removed. The readily biodegradable substrate is then calculated from the amount of oxygen utilised at this point, using a yield obtained with acetate as the substrate in a separate test. Another variation involves using a small sludge inoculum, and a sufficient assay time to allow measurable microbial growth (Wentzel et al 1995). The OUR increases as the biomass grows at its maximum rate on the readily biodegradable substrates, and on the.hydrolysis products ofthe slowly biodegradable substrates (figure 4.3). In order to calculate the amount of readily biodegradable organic matter initially present in the wastewater, the amount of oxygen used due to the slowly biodegradable organic matter must be calculated. This procedure involves more manipulation of the data than the procedure of Ekama et al (1986), but does not require activated sludge as the test relies on the biomass initially present in the wastewater. This method employs a high initial f/m and therefore will probably result in changing biomass characteristics throughout the assay, but it is assumed that the stoichiometry will remain constant. The method used for a detailed investigation of nitrification combined features from the AOUR method and the IAWQ test (Ossenbruggen, et al 1996). As in the IAWQ 95 Shaded area proportional to readily biodegradable substrate Initial OUR proportional to maximum growth rate ~ ^ OUR due to slowly biodegradable substrate Time Figure 4.2 IAWQ batch test for measurement of maximum growth rate and readily biodegradable substrate. ID o Shaded area proportional to readily biodegradable substrate ^ g J ^ ^ ' ^ O U R due to slowly biodegradable substrate cm ZD o 05 O u - b Mmax i r Time Figure 4.3 Growth test, growth measured using OUR. 96 test, enough substrate was added to ensure high microbial activity for over an hour. As in the AOUR test, the endogenous respiration rate was measured. The OUR, ammonium, and nitrite concentrations were followed throughout the test. The specific OUR was used as the response variable. The kinetics were calculated from the OUR versus ammonium and nitrite concentrations throughout the batch test, not just using the initial OUR and initial substrate concentration as in the AOUR test. The test was repeated with five different starting concentrations. The data were modeled using a two step nitrification model. A combination of the IAWQ method and the AOUR method was used to determine the kinetic coefficients of a kraft mill wastewater (Slade and Dare 1993, Slade et al 1991). The amount of readily biodegradable substrate was calculated from the IAWQ method, then the Monod coefficients for readily biodegradable substrate utilisation were measured using the AOUR technique. Measurement of Adsorbed Substrate Respirometric techniques have been used to determine the amount of substrate adsorbed to biomass during batch tests, which is very difficult to measure directly. Using a simple mass balance, the difference between the SUR calculated from the OUR and the measured SUR may be interpreted to be due to substrate adsorption. Major assumptions required are constant rate coefficients and yield throughout the batch test. In one study, measurement of accumulated substrate during a batch test was made using mass and energy balances (Guiot and Nyns 1986). B K M E was used as the substrate. A batch test was performed similar to the IAWQ method, but soluble COD 97 was measured as well. The OUR profile was similar to the one shown in figure 4.2, and the COD removal appeared to follow the multicomponent rate equation (equation 3.22). It was assumed that the high initial substrate uptake rate (which was greater than the substrate uptake rate calculated from the OUR based on an assumed and constant yield) was due to substrate adsorption. As long as there was adsorbed substrate, the metabolic rate operated at its maximum. The drop in the OUR was interpreted to be the point at which all of the adsorbed substrate had been metabolised, driving the bacteria to metabolise the remaining soluble substrate. Torrijos et al (1994) also found that the substrate removal rate was greater than the oxygen uptake rate during the first five minutes of a batch test. Assuming a constant yield, this led to the interpretation that a small amount of substrate was adsorbed at the start of the batch test. Other evidence for substrate adsorption comes from following the change in the yield measured by the AOUR method during sludge regeneration. The increase in yield was attributed to the ongoing oxidation of accumulated substrates (Drtil et al 1993a). At the start of sludge regeneration, there was an excess of accumulated substrate, so all of the new substrate added during the respirometric assay was used for growth. The ability of the sludge to adsorb substrate was partially restored after a period of regeneration, thereby increasing the apparent yield (the stored substrate would not be oxidised during the course of the AOUR test). There was no attempt at quantifying the amount of adsorbed substrate. Some studies have indicated that the nature of the wastewater component limits the amount of adsorbed substrate, with only a fraction of all of the organics in the 98 wastewater being adsorbed. Other studies have concluded (assumed) that the amount of substrate adsorbed is based solely upon the capacity of the biomass, which is greatest at the start of a batch test. Toxicity Assessment A n important use of respirometry is for the determination of wastewater toxicity. Possible applications of toxicity assessment are to track down toxic substreams which may require pretreatment, to determine the maximum concentration of these substreams that a treatment system can tolerate, and to determine the toxicity of different additives used in industrial processes. Respirometric toxicity assays are limited in that they cannot determine the effect of the influent on floe stability and settleability, compaction and effluent clarity, all of which are important treatment parameters of the activated sludge process. Respirometric toxicity assays are complicated i f the toxic compound to be tested is also a food source for the bacteria. If this is the case, at low concentrations the test compound will cause an increase in the OUR, due to increased metabolism. As the concentration is increased, the OUR will start to drop as the toxic effects of the compound are exerted on the bacteria. If just one concentration is tested, it is difficult to tell i f the sample is toxic or not. A number of different procedures have been developed to overcome this difficulty. The first group of methods involve measuring the OUR at many different concentrations of the test compound. If the sample is not a food source, the endogenous OUR will be measured, and any decrease will be an indication of toxicity. If the sample is also a food source for the microorganisms, the OUR in the presence of the sample may 99 be greater than the endogenous OUR. In this case, the OUR must be measured using various sample concentrations, i f the OUR decreases with increasing sample concentration, the sample is toxic (Arthur 1984). In order to avoid these complications, an excess of substrate may be added to the respirometer, as well as the sample to be tested. This will ensure that the OUR is at it's maximum rate regardless of whether the sample being tested is a food source or not. An example of this type of assay is the OECD activated sludge respiration inhibition test, which involves the incubation of activated sludge with synthetic sewage and the test compound. After three hours the OUR is measured and compared to a control sample with synthetic sewage but no test compound. Since the synthetic sewage contains many different substrates in high concentrations, the biomass metabolic rate wil l be at it's maximum and the sample should not cause an increase in the OUR (Klecka et al 1985). Toxicity tests are best done using the activated sludge of interest, in order to achieve the most pertinent results. In order to reduce variability and simplify tests, different cultures have been investigated, including E. coli, Vibrio fischeri (Microtox ©), and Polytox © . I t has been found that the toxicity response depends upon the test culture used. Activated sludge toxicity tests are less sensitive than the Microtox assay and a test using glucose consumption by E. coli. This may be due to the large concentration of sludge adsorbing the toxicants and reducing the effective concentration (Dutka and Kwan 1984). The toxicity of 3,5 dichlorophenol was greater when measured by respirometry than when tested in a continuous activated sludge system (Broecker and Zahn 1977). 100 Growth tests are more sensitive than respiration tests (King 1984). The effect of toxic compounds on microbial growth rates may be measured by using a high f/m respirometric assay. OnLine Respirometry Respirometry may be used for continuous monitoring and/or control of activated sludge plants. The microbial kinetics, the wastewater strength, the treated wastewater BOD, possible wastewater toxicity, the f/m, and the total or viable biomass may be monitored. These measurements may be used in order to control the activated sludge process (Spanjers et al 1996). On-line respirometers may be batch or continuous. In batch processes, the respirometer is charged, the OUR is measured, then the cycle is repeated. Alternately, flow through respirometers may be used where the DO is measured at the inlet and the outlet of the respirometer and the OUR is calculated from the difference in DO and the flow rates. Open or closed respirometers may be used, and the oxygen may be measured in the gas or liquid phase. The principles of on-line respirometry are the same as for the respirometric procedures described above. The OUR of samples taken from the start of the treatment process will be proportional to the wastewater strength and biomass viability. The OUR on return sludge samples will give an indication of viable biomass. On-line respirometry to measure the f/m has been described (Arthur and Arthur 1994). This technique involves two automated manometric respirometers. The first respirometer takes samples from near the inlet to the aeration tank to get a measure of the strength of the wastewater (and check for toxicity), while the other takes samples from the return sludge flow to measure the amount of biomass. In the absence of substrate, the 101 OUR will be directly proportional to the amount of viable biomass, although the proportionality constant will be different for different sets of biomass. The resulting information may be used to control the recycle rate and the sludge wasting. A number of respirometers for determination of influent STBOD and toxicity have been developed. Many of these methods use more sophisticated activated sludge models which account for the different fractions of wastewater and biomass. If biomass in the respirometer is limiting the respiration rate (excess substrate), the OUR will be directly proportional to the biomass concentration, and can be used to estimate the viable biomass. If biomass is not limiting, the OUR is proportional to the substrate concentration, and may be used to obtain a rapid measurement (30 minutes) of the BOD5 (Therien 1983). High F/M Respirometric Methods Biochemical Oxygen Demand Measurements One common use of batch tests with small initial inocula is to measure the wastewater BOD. In the standard method, diluted wastewater is incubated with a small amount of seed at 20°C, for five days, after which time the dissolved oxygen depletion is measured. The BOD test was devised in 1898 by the British Royal Commision on Sewage Disposal to provide a direct measure of the amount of oxygen depletion in receiving waters (Bailey 1986). The five day period is arbitrary, originally chosen as the residence time of a typical British river. It is often argued that the meaning of the BOD is suspect (Orhon & Artan 1994). Since the substrate degradation often does not go to completion during five days, the BOD5 value will be a function of not just the wastewater 102 strength, but also the wastewater biodegradability, and the acclimation of the bacterial seed to the sample. The oxygen demand is modeled as a first order process, asymptotically approaching the ultimate BOD value (the value at the completion ofthe biochemical reactions). The faster the BOD reaction, the greater the percentage of substrate which will be oxidised during the five day test. Usually, wastewater leaving a treatment plant is less biodegradable (smaller first order constant) than the untreated wastewater, so a lower percentage of substrates will be measured in the effluent than in the influent. The ultimate BOD, which is not a function ofthe oxidation rate, is more meaningful than the intermediate five day value. The ratio of the B O D 5 to the ultimate BOD is 0.65 to 0.7 for domestic sewage, and anywhere from 0.1 to 0.9 for industrial wastewaters (Orhon & Artan 1994). This ratio depends upon the BOD rate constants. The BOD 2o value is often assumed to be equal to the ultimate BOD. The correlation of the BOD5 with the COD gives an indication of the ratio of biodegradable substrate concentration to total substrate concentration in the sample. If the sample contains only a readily biodegradable substrate, such as glucose, the B O D 5 will be approximately 80% of the COD. The ultimate BOD is usually close to the COD value i f the sample contains only readily biodegradable substrates. Due to a lack of clear meaning ofthe BOD5 value, and also to the lengthy time required for measurement, substitute tests to measure wastewater strength are often used. In order to speed up the reactions, many variations of the BOD test use respirometers with headspace, or with the continual addition of air. The greater availability of oxygen means the sample will not have to be diluted as much and a greater microbial seed 103 concentration may be used than is used in the standard B O D 5 test. This results in faster biochemical reactions and the BOD may be measured in a day or less. One example is the HBOD (headspace BOD) test (Logan and Wagenseller 1993), which involves the measurement of the dissolved oxygen after one day. Other examples are the many manometric tests. These variations of the standard BOD test are easier to automate than the standard method, so the oxygen demand may be recorded throughout the test, in addition to the BOD5 the BOD coefficients and the ultimate BOD may be calculated from the data. The tests described in the previous section (low f/m respirometric methods) for determining short term oxygen demand or readily biodegradable COD may also be used to determine the wastewater strength. These tests use environmental conditions closer to that of the wastewater treatment plant and thus give a more meaningful indication of the wastewater strength. Due to the much higher amount of biomass present, endogenous metabolism must be accounted for or very high oxygen demand values wil l be measured. Therefore only the oxygen demand due to the initial metabolism of the substrate will be measured. In the BOD test, the oxygen demand due to the growth on the substrate, and also the oxygen demand due to the endogenous decay of the bacteria which have grown on the substrate are measured. Consequently, short term BOD is usually less than the B O D 5 . If the BOD5 values are required for regulatory purposes, a correlation may be found between a given respirometric method and a given wastewater, so the respirometric method may be used for more efficient monitoring. The BOD5 has been successfully correlated with the respiration rate, one hour oxygen demand, and one half hour oxygen 104 demand (Arthur 1984). Each wastewater, including treated and untreated, requires a separate correlation. Maximum Growth Rate Measurements Procedures similar to the manometric BOD tests have been used for many purposes, including measuring biodegradability and microbial adaptation to various compounds or wastewaters. However, these results are not directly applicable to activated sludge treatment due to the greatly different environments between the two processes. In this review only a few examples of these tests which have been used to determine activated sludge kinetics shall be described. Growth tests are performed under different conditions from low f/m tests. A high f/m is used, and the tests usually last longer allowing the bacteria to grow during the course ofthe respirometric assay. The initial OUR is low due to the small inoculum. As the biomass grows, the OUR increases in direct proportion to the increase in biomass. The endogenous metabolism and decay are assumed to be negligible, and the OUR will increase exponentially until some factor becomes limiting. Using different starting concentrations, the growth rate versus substrate curve can be obtained (Gaudy et al 1988). The Monod constants may also be calculated from just one respirometric growth test using an appropriate initial f/m (Smets et al 1996a). The estimate of the initial biomass concentration will affect parameter determination. Also, i f there are insufficient amounts of nutrients present, the calculated yield will be higher than expected. It was recommended that nutrients be present in quantities far in excess of their stoichiometrically required dose in order not to affect the measured kinetic coefficients. 105 A more efficient way of measuring the maximum growth rate is to measure the change in OUR in the presence of an excess of substrate (Kappeler and Gujer 1992). It can be shown that the slope of the logarithm of the OUR vs. time curve is equal to the maximum growth rate minus the endogenous decay rate (using the IAWQ model). The half saturation constant may be calculated at the point where the OUR drops, under the assumption that the substrate concentration is very low, by applying the IAWQ model to the entire OUR vs. time curve. However, since there is likely to be slowly biodegradable matter still present in the wastewater once the readily biodegradable matter is removed (Wentzel et al 1995), this method will give erroneous results for K M . A common critique of methods which involve excessive biomass growth is the changing biomass composition during the tests. At the end of the test, the biomass may not be the same as at the start of the test, and may have different kinetic parameters. In particular, the batch test tends to select organisms with high growth rates (Wentzel et al 1995, Bull and Brown 1979). This has been demonstrated experimentally (Novak et al 1994). The organisms selected based on the maximum growth rate in batch tests do not necessarily play a significant role in the nutrient-limited environment, such as a wastewater treatment plant (Harder and Dijkuizen 1982). Correlation Between OUR and Viability There have been many attempts to determine the percentage of active biomass in activated sludge using respirometry. One method (Huang et al 1985) involves measuring the maximum OUR using a batch test. The microbial viability is then assumed to be proportional to the OUR. Using cell counts from another paper, it was found that a portion ofthe OUR was due to non-viable cells. Another method involving OUR 106 measurements (although it was not clear whether they were endogenous or exogenous rates) also found that a majority of respiration was due to non-viable cells (Walker and Davies 1977). These calculations were based on the assumption that the respiration rate per cell is constant. The OUR was correlated to viable biomass using data obtained from a batch test (Jorgensen et al 1992). Viable biomass in a continuous unit was then calculated by measuring the OUR and using the conversion factor (assuming that the specific OUR of biomass from a continuous activated sludge unit is the same as the specific OUR of a rapidly growing culture). Another technique involves measuring the maximum specific growth rate using a high f7m batch test, and measuring the volumetric growth rate using a procedure similar to the AOUR procedure, then dividing the two numbers to obtain'the viable biomass concentration (Blok 1976). Alternately, the initial OUR at the start of the growth test may be used as the measure of the specific growth rate (Orhon et al 1995, Kappelar and Gujer 1992). According to the IAWQ model, under the conditions of a growth test the biomass concentration is proportional (through the yield coefficient) to the OUR times the growth rate. The maximum growth rate is measured during the test, so the initial biomass concentration may be calculated if the yield is known. This method makes the further assumption that, at the start of the growth test, the endogenous metabolism is minimal compared to the overall OUR. If the biomass changes during the growth test (selection for fast growing organisms over bacteria with lower growth rates), the maximum growth rate obtained will be for the fast growing organisms, and not the complete mixture of organisms present at the start of the test. This wil l result in an overestimation of the initial biomass concentration. 107 Similar logic has been applied to the determination of the active biomass concentration using the decay test. During a decay test the OUR is proportional to (1-f e x )bHX H . The term bn is measured during the decay test, fex is assumed, therefore using the initial OUR, the initial biomass concentration may be calculated. However the OUR data during a decay test often do not fit the first order model, possibly due to the presence of slowly biodegradable hydrolysable material at the start ofthe test (Sollfrank et al 1992). If this is the case the active biomass concentration wil l be overestimated. The AOUR technique in combination with continuous culture techniques (steady state operation at different SRT's) was used to determine the kinetic constants for the microorganisms responsible for degradation of xenobiotic compounds, rather than to determine the constants for the whole sludge (Chudoba et al 1989a,b). The concentration of responsible microorganisms was calculated from the ratio ofthe maximum volumetric removal rate calculated from the AOUR technique, and the specific removal rate calculated from the continuous reactors. Different populations of bacteria may be present at each of the continuous culture steady states. 4.6 Yield The accepted method for measuring the yield coefficient is to seed wastewater with a small amount of biomass and measure total and soluble COD. The COD due to biomass is calculated as the difference between the total and soluble COD. In order to calculate the true yield, samples can be taken over a number of days and the apparent yield calculated for each data point. The true yield can be calculated by extrapolating to time = 0 on a graph of 1/Y versus time (Slade et al 1991). This test is very sensitive to the initial values of substrate and biomass due to analytical errors (Orhon and Artan 108 1992). The yield may also depend upon the initial conditions (f/m) as discussed earlier., so a more appropriate assay would involve using f/m values closer to f/m values found in continuous systems. The AOUR procedure may be used to calculate bacterial yield i f a known amount of substrate is added (equation 4.6). Liebeskind et al (1996) added known amounts of acetate to a respirometer and measured the oxygen consumption. The yield was found to increase from 0.59 to 0.68 as the SRT increased from 1.88 to 24.0 days. This method has also been used by Cech et al (1984), who found a range of yields on various substrates. Another method of calculating the true yield is to run continuous activated sludge units at different SRT's, and use equation 4.3 to calculate both Y and kd. 4.7 Decay Microbial decay is measured by removing a sample of mixed liquor and placing it in a batch reactor and measuring the OUR or the biomass concentration for at least 10 days. The method involving OUR measurement was found to be easier and more reliable (Marais and Ekama 1976). The data may be interpreted assuming either the endogenous decay model or the death regeneration model. If the death regeneration model is used, the regeneration step may not be the rate limiting step, especially at low temperatures (Lishman and Murphy 1994). At low temperature hydrolysis may be slower than either death or regeneration. Since the regeneration step is the step which requires oxygen, this may lead to underestimating the decay rate. This argument does not apply to the endogenous decay model, i f the assumptions made are correct (oxygen utilisation is by. endogenous metabolism and not growth). 109 When bacteria are exposed to starvation conditions, the metabolism may be slowed and the membrane made more impermeable. This will result in a lower decay rate for starving bacteria than for actively growing bacteria. If this is the case, the decay coefficient measured by batch tests will not be applicable to continuous activated sludge units. Decay decreases with temperature (Sollfrank et al 1992) and has been found to follow a simplified Arrhenius relationship (Marais and Ekama 1976). 4.8 Summary Many different respirometric assays for measuring activated sludge substrate removal rates and stoichiometry have been developed. A l l of these assays may be classified into two broad groups based upon the initial f/m, low initial f/m tests and high initial f/m tests. The results from respirometric batch assays are a function ofthe assay. Tests which employ a low initial f/m simulate the environmental conditions of the treatment plant to a much greater degree. These tests give an accurate indication ofthe state and capability ofthe biomass at the time of sampling. Respirometry may be used to measure the substrate removal rate, the yield on the substrate, or the substrate concentration i f one of these is known a priori. For wastewaters, the substrate concentration and substrate removal rate are usually unknowns. If the yield is known from a separate assay, or assumed, the substrate concentration and removal rate may be calculated using respirometric techniques. Respirometry may also be used as an indication of the biomass concentration. The maximum OUR is directly proportional to the active fraction of biomass. The proportionality constant is the maximum specific growth rate, which is difficult to 110 measure under environmental conditions similar to activated sludge (low growth rates). Changes in the OUR are related to growth or decay of biomass, and may be used to calculate growth rates or decay rates. Ul Chapter 5 Materials & Methods 5.1 Lab Scale Activated Sludge Units Two continuous lab-scale activated sludge reactors were operated. Each reactor consisted of a 5-litre cylindrical jacketed Plexiglas aeration tank and a 1.5 litre conical glass clarifier. The working volume of the aeration tank was 4 litres, and the working volume ofthe clarifier was 1 litre. The aeration tank and the clarifier were connected by a short 1" diameter tube. This tube had to be cleaned periodically to prevent clogging. Reactor temperature was maintained at 35 C by circulating water from a constant temperature bath (Fisher Scientific) through the annular Plexiglas jacket surrounding the aeration tank. The clarifier did not have a water jacket, but was insulated. The temperature ofthe clarifier varied between 25°C and 30°C depending upon the room temperature. For one experimental run (run #3), an aerated selector was used ahead ofthe aeration tank. The selector consisted of a 600 ml jacked glass vessel. Temperature was maintained using the same circulating water bath as the main aeration tank, thereby maintaining the selector temperature very similar to the main aeration tank temperature. The selector volume was 300 ml, 7.5% of the aeration tank volume (which was lowered to 3.7 litres during the operation of the selector). Oxygen was added using an aquarium air pump. Air was supplied to the aeration tanks using both aquarium air pumps and building air. The redundancy was to ensure a constant supply of air. When the power went down, the building air would (sometimes) stay on providing mixing and air, but •112 often the building air would be turned off, so a second source was required. The reactor dissolved oxygen content (DO) was usually between 3 to 5 mg/L, and never fell below 2 mg/l. Mixing was provided by a magnetic stir bar and a stir plate. The sides ofthe aeration tank were cleaned daily in order to prevent wall growth from forming. In order to reduce biomass attaching to the sides of the clarifier, a pump was set up to mix the contents ofthe clarifier every 30 minutes. This did not completely work, so the sides of the clarifier were wiped clean daily. Feed, waste sludge, and sludge recycle were pumped using Masterflex peristaltic pumps (Cole-Parmer, Chicago, IL). Size 14 tubing, either silicone or C-flex, was used for both the feed line and the recycle line and an influent flow rate of 8 litres/day was maintained. This gave a hydraulic retention time of 12 hours. The tubing was cleaned of attached biomass every day, and the flow rate was adjusted as the tubing stretched. Although biomass grew in the feed line, this had no measurable effect on the effluent characteristics, COD and BOD. One pump with two pump heads was used for the recycle of both activated sludge units, but two separate pumps were used for the feed rates in order to have more precise control. The recycle line clogged repeatedly, but at the low flowrates desired, it would have been impossible to control using a larger size of tubing. The recycle ratio was maintained at approximately 1. Sludge waste rate was regulated using a Chrontrol (San Diego, CA) timer controlling Masterflex peristaltic pump (size 16 tubing). Approximately 10 millilitres was removed every waste cycle, which varied from every 20 minutes to every 2 hours depending on the M L V S S levels 113 and the desired SRT. The sludge wastage rate was measured daily and varied as necessary. The SRT was calculated according to the following formula: XV Qx = (5.1) X XQW+X(Q0-QW) Untreated B K M E was obtained from Harmac Pacific Ltd. (Nanaimo, BC). Harmac Pacific produces 1135 tonnes per day of bleached kraft pulp from softwood -douglas fir (17%), western hemlock/balsam (55%), and western red cedar (28%). The pulp is bleached using 100% chlorine dioxide substitution (Rodden 1994, Rodden 1998). The effluent was sampled from a sample port before the cooling towers (Harmac Pacific does not have primary effluent clarification). Approximately 50 20 litre buckets were collected on each effluent run. The time to fill all of the buckets was approximately 30 minutes, during which time any variation in the' effluent would result in variation in the composition of the individual buckets. The variation between buckets (< 5%) was small compared to the variation between effluent runs (up to 100%). The wastewater was warm when sampled (50°C). Transportation to the fridge took approximately three hours. The effluent was then stored at 4°C until use. Just before use, the pH was adjusted to 7-7.5, (initial pH varied from 4 to 11 depending on the effluent batch) and nutrients were added in the ratio of BOD:N:P = 100:5:1. Nutrients were added in the form of concentrated N H 4 O H (Fisher Scientific, Nepean, ON) and H 3 P O 4 (BDH, Toronto, ON). Approximately every six weeks a new batch of effluent was obtained from Harmac Pacific. Over the course of this study, 19 different batches of effluent were obtained from Harmac Pacific Ltd. This effluent was treated over six different runs. Each run was started with fresh sludge from Harmac's activated sludge unit. The first run involved lab 114 set up, learning how to operate activated sludge units, and developing experimental methods. The second run investigated the effects of SRT, the third run continued the study of SRT and also looked at the effect of an aerobic selector on the treatment of B K M E . The fourth run studied the biomass from Harmac's activated sludge treatment system, before adaptation to the lab scale conditions occurred. During this run, biomass decay rates were measured. During the fifth run, supporting evidence for what was discovered during the third run was sought, however due to low biomass activity, this was only partially successful. During the sixth run, the respirometric method was studied in great detail, and the decay experiment was repeated. Temperature and pH shocks were performed randomly throughout the various runs. The schedule of the effluent batches and operating conditions is presented in table 5.1. Table 5.1 Experimental Conditions Run Batch Dates Reactor #1 Reactor #2 1 A April 21 1994 SRT = 5 SRT =17 B June 21994 SRT = 5 SRT = 23 C July 14 1994 SRT = 7 SRT = 26 D September 2 1994 SRT = 5 SRT =10 2 E November 24 1994 SRT = 5 SRT = 10 F January 5 1995 SRT = 5 SRT =11 G March 1 1995 SRT = 5 SRT = 19 H April 5 1995 S R T - 5 SRT = 25 I May 17 1995 SRT = 5 SRT =17 J June 29 1995 SRT = 5 SRT =15 K August 10 1995 SRT = 7 SRT =12 3 L November 24 1995 SRT = 11 SRT =11, Aerobic selector M January 10 1996 SRT = 10 SRT =11, Aerobic selector N March 8 1996 SRT = 10 SRT =11, Aerobic selector 4 0 June 17 1996 Decay experiment 5 P August 22 1996 SRT = 5 SRT = 19 6 0 November 27 1996 SRT = 5 SRT = 23 R January 18 1997 SRT = 11 SRT = 24 S March 27 1997 Decay experiment SRT =15, Synthetic effluent 115 5.2 Standard Tests Mixed liquor suspended solids, B O D 5 and COD were determined using standard methods (Greenberg et al., 1992). COD was measured in H A C H COD vials, on 2 ml of sample. COD reagents were added to the sample vials using dispensers. The samples were digested in a H A C H COD reactor, and absorbance was measured in a H A C H 2000 Spectrophotometer at both 600nm and 620 nm. Calibration curves were done for each batch of reagents prepared. BOD was measured in 60 ml Wheaton BOD bottles. The BOD bottles were incubated in a Fisher Scientific incubator at 20°C. Dissolved oxygen was measured using a YSI model 59 DO meter and probe. Seed for the BOD test was obtained from the lab scale reactors. Seed control and samples were done in triplicate when there was enough sample. Acute toxicity was measured using a Microtox 500 analyser. 5.3 Batch Tests Wastewater Characteristics from Batch Tests A typical wastewater batch biodegradation test was as follows: 200 ml of concentrated mixed liquor removed from the continuous lab scale unit was added to 800 ml of untreated B K M E to provide an initial BOD of 200 mg/1, and an initial M L V S S of 1000 mg/1. A constant temperature was maintained by using glass vessels with water jackets and an external circulating water bath. The mixed liquor was continuously aerated and well mixed using a magnetic stir bar. The OUR was measured by transferring a well mixed sample to a water jacketed BOD vessel maintained at the same temperature as the batch test vessel. The DO concentration in the BOD vessel was 116 monitored until below 2 mg/l, then the sample was returned to the batch test vessel, and a new sample was transferred to the BOD vessel. Mixing in the BOD vessel was provided by the DO probe and a. magnetic stir bar. The OUR was measured for a minimum of six hours. Samples were periodically taken from the main batch test vessel for measuring solids, BOD, COD, and respirometric kinetics (see the AOUR procedure). Samples for use in the BOD, COD, and respirometric tests were centrifuged for 15 minutes to remove the solids immediately upon sampling. The solids were discarded (or returned to the continuous unit). BOD and COD tests were either started immediately upon sampling, or the samples were stored at 4°C for a maximum of six hours. Samples for respirometric measurements were either utilised immediately, or frozen overnight to be used the next day. Samples for measuring M L V S S were immediately filtered as per the M L V S S standard method. Infinite Dilution or Fed-Batch Test This procedure was based upon the method of Williamson and McCarty (1975). A large volume of mixed liquor (usually one litre) was placed in a water jacketed vessel and fed untreated B K M E at a constant rate using a syringe pump. The mixed liquor was aerated and well mixed. The M L V S S concentration was usually adjusted to around 1000 mg/l. Samples were withdrawn for BOD analysis periodically (every 15-30 minutes), centrifuged for 5 minutes to remove the solids, then stored at 4°C until the start of the BOD test. The solids removed from the BOD samples were returned to the test vessel. This ensured that the volume of mixed liquor didn't vary by more than 5%. The test was run for 2-4 hours. At this point the test was repeated at a new feed rate using fresh sludge. The M L V S S levels were measured at the start and end of the test.. The BOD test 117 on the collected samples was started at the end of the day when all of the samples had been collected. Batch Decay Test Decay rates were measured by removing a sample of mixed liquor from the continuous unit aeration tank, placing in a water jacketed vessel, and adjusting the conditions as required (pH and temperature). The sample of mixed liquor was then mixed and aerated at a constant temperature, with no feed, for a number of days or until the biomass activity was very low. At specified time intervals, samples were removed from the decay reactor and placed in the respirometer. Biomass activity was measured as outlined below (the AOUR respirometric procedure) after which the sample was discarded. If decay due to extreme temperature or pH was being studied, the sample taken for the measurement of the respirometric activity was adjusted to the temperature or pH the biomass was acclimated to prior to the start of the decay experiment. For example, if the activated sludge was acclimated to a pH of 8, and the decay experiment was carried out at pH 10, then the samples from the decay experiment were adjusted to pH 8 for the respirometric measurements. Temperature and pH studies For experiments in which the pH was adjusted, the pH of both the mixed liquor and substrate samples were adjusted to the experimental pH using 10M H2SO4 or 10M NaOH. pH adjustment was rapid. Minor adjustments were made with IM solutions as required. For all experiments, constant temperature was maintained using water jacketed vessels attached to heated water baths. The response time for small temperature changes was very fast. For large temperature changes, ice water or hot water was added to the water bath to achieve quick temperature adjustment. For experiments at temperatures below room temperature, ice water or cold tap water was added to the water bath as required. For decay experiments, measurements were taken immediately following the temperature or pH adjustment, until the biomass activity was too low to measure. For steady state experiments, measurements were taken immediately following the pH or temperature change, and continued until the experimental values reached a steady state. For all experiments where pH and temperature are not specified, the pH and temperature were at the values for which the biomass was acclimated. This is a temperature of 34 +/-1°C and a pH of 8 +/- 0.5. 5.4 Respirometric Method AOUR Procedure Activated sludge model parameters were determined through respirometric methods (Cech et al, 1984). Three respirometers were made by Canadian Scientific Glass blowing. Their working volumes were 230, 210, and 205 ml. Temperature was controlled using an external circulating water bath connected to the water jacket on the respirometer. Mixing was provided by both the mixer on the DO probe and a magnetic stir bar. At the start of a test, mixed liquor samples were transferred from the aeration tank ofthe continuous lab scale system to the respirometer. If a different MLVSS 119 concentration from the concentration in the continuous unit was desired during the respirometric test, the mixed liquor was diluted with treated effluent, or concentrated by settling and centrifuging. M L V S S was measured either at the start of the test, on a separate sample of mixed liquor from the aeration tank of the continuous lab scale system, or at the end of the test, on the respirometer contents. Upon addition to the respirometer, the mixed liquor was aerated for 15 to 60 minutes at a constant temperature. The duration of the aeration time did not have any noticeable effect on the measured kinetics. After the mixed liquor had been aerated, the air flow was stopped and a DO probe was inserted into the respirometer. Stirring was provided both with a stir plate and from the DO probe itself. The data from the DO meter (YSI model 59, or Orion model) were imported directly into a spreadsheet using Collect. Collect is a software program which takes the data from a RS 232 port. The data are parsed, then sent to the keyboard buffer so any program "thinks" that the data are being typed in from the keyboard. This allowed for real time monitoring of OUR instead of just the DO. The DO and temperature were recorded every second. The probe was allowed time to equilibrate with the sample temperature, and the endogenous respiration rate was measured for at least 2 minutes. Then, a known amount of substrate was added through an injection port using a Becton-Dickson syringe and the change in respiration rate was followed. When the respiration rate returned to the endogenous respiration rate, another sample was added. When the DO level fell below 2 mg/1, the DO probe was removed from the respirometer, and the air flow was started again. Once the mixed liquor sample in the respirometer was saturated with dissolved oxygen, the air flow was stopped and the DO probe was reinserted. More substrate samples were added as before. This procedure was repeated 120 until a full set of OUR versus substrate concentration data was obtained. When B K M E was used as the substrate, varying initial concentrations of B K M E in the respirometer were achieved by injecting varying volumes of B K M E into the respirometer. The order of injection was arbitrary. When other substrates were investigated, the initial concentration in the respirometer was varied either by injecting varying volumes of a stock solution, or by injecting the same volume of a series of dilutions. The two methods gave similar results. The range of substrate concentrations was chosen to cover the complete range of AOUR response, from the initial first order region of the AOUR vs. substrate curve, to the zero order region where the biomass' are at maximum activity. Data Analysis Typical respirometric data obtained using a range of methanol concentrations are shown in figure 4.1a. There are many ways to analyse the data. In the method developed by Cech et al (1984), the DO versus time data is used as is, and has already been described (chapter 4). Alternately, the data may be transformed to oxygen consumption for substrate metabolism versus time as shown in figure 6.1.1. The data may also be transformed by calculating the slope of the curves shown in figure 4.2c by using a smoothing interpolating formula, as was done to obtain the curves shown in figure 6.1.2. In this method of presenting the data, the OUR due to the addition of substrate is equal to the maximum point on the curve minus the OUR before addition of substrate (baseline), this value is referred to as the AOUR. The amount of oxygen consumed to metabolise the given substrate is equal to the area between the OUR curve and the baseline extrapolated until it intersects the OUR curve once the substrate has been all metabolised. For fast data analysis in the lab, this was the method which was used with the slope being 121 calculated at each second using linear regression over 10 data points. When this method was used it became easy to change how many points the slope was calculated over to adjust for the respiration rate. The three methods of interpreting the data described above all gave the same results for a given data set, with the main difference being in visual presentation. The method used for this project is very similar to the one developed by Cech, et al (1984), but the use of a computer to analyse the data allows for more accuracy (in the original method a strip chart recorder was used). The method used for this project will also be more accurate than the method discussed by Smets et al (1994), which uses two respirometers, as it will be very difficult to get identical conditions (temperature, pH, biomass concentration, etc.) in each respirometer and slight differences can have large effects on data interpretation. 122 Chapter 6 Respirometry 6.1 Respirometric Data Analysis Typical Respirometric Data Even though respirometry has been used extensively in the study of activated sludge kinetics, the majority of the assumptions made in interpreting respirometric data obtained from low f/m batch tests have not been verified. Some of the influences which have not been investigated are the effects of mass transfer, dissolved oxygen concentration, and the assumption of pseudo steady state. Measuring the kinetics of B K M E biodegradation is central to the goals of this project. Since methanol is the main readily biodegradable substrate in B K M E , and in order to verify the respirometric test using a single substrate before applying the method to a more complex wastewater, the data on methanol kinetics as measured using respirometry wil l be presented first. Data on formic acid and acetic acid kinetics will be presented as well, since these compounds are also major contributors to B K M E B O D . Typical activated sludge respirometric data for a range of added methanol concentrations are shown in figure 6.1.1 and 6.1.2. In figure 6.1.1, the oxygen consumed due to the addition of methanol has been plotted versus time for many different methanol additions. The OUR increased immediately (< 2 seconds) after the addition of methanol. The more methanol which was added, the more oxygen the biomass consumed. Also, i f the region right after the addition of methanol is closely examined, it can be seen that the greater the addition of methanol, the greater the slope of the oxygen consumption versus time curve. This is more evident in the OUR vs. time curves as shown in figure 6.1.2 for 123 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5 Time (minutes) Figure 6.1.1 Oxygen consumption for 8 different methanol injections of varying strength, superimposed for comparison. Initial methanol concentration listed on chart. MLVSS: 3750 mg/l 4.5 0—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—r-1—i—i—i—|—i—i—i—i—|—i—1—i—i—j 1 1.5 2 2.5 3 3.5 4 Time (minutes) Figure 6.1.2 Methanol injections, same data as figure 6.1.1, converted to OUR. MLVSS: 3750 mg/l 124 the same set of data. The AOUR is calculated by subtracting the OUR before substrate addition (the endogenous or baseline OUR) from the maximum OUR after substrate addition. The AOUR is a measure ofthe substrate uptake rate and may be plotted versus the amount of methanol added (figure 6.1.3). The AOUR followed saturation kinetics, that is, at low substrate concentrations the AOUR increased with increasing substrate concentration, until a maximum AOUR was reached. The methanol was not inhibitory to the activated sludge at the concentrations used (since the AOUR did not decrease at high methanol concentrations). Another set of data, at a lower biomass concentration, is shown in figure 6.1.4. The AOUR again followed saturation kinetics but was lower due to iess biomass being present. At steady state, the endogenous OUR and the OUR due to available substrate, and consequently the AOUR, are directly proportional to the active biomass concentration. Figure 6.1.5 shows the standard deviation of the AOUR data versus the average AOUR for each level of methanol addition with more than three replicates. The standard deviation versus methanol concentration is shown in figures 6.1.3 and 6.1.4. The standard deviation does not appear to be a function ofthe AOUR, methanol concentration, or the solids concentration. Besides the metabolic rates, the yields may also be calculated by respirometry. It is necessary to know the yield in order to convert the oxygen uptake rate into a substrate uptake rate. The oxygen consumption per substrate added is shown in figure 6.1.6 for the 2 data sets already discussed. For each data set the amount of oxygen consumed appeared to be directly proportional to the amount of added substrate, which is expected if the yield is a constant with respect to substrate concentration. If a constant yield is 125 - i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i — i — i — i — | — i 1 — i — r 0 1 2 3 4 5 6 Methanol (mg COD/I) Figure 6.1.3 AOUR ( • ) , average AOUR (where more than three replicates) (•), and standard deviation ( A ) VS methanol concentration. MLVSS 3755 mg/1. C o CO E 1.4-1.2-1-0.8-0 . 6 -O 0.4-< 0 . 2 -0 -A I A 4r I -1 1 1 1 1 1 1 [ 0.5 - i — i — i — | — i — i — i — i — r ~ 1 1.5 2 Methanol (mg COD/I) i — i — i — r p 0 . 1 r0 :09 r0 :08 r'0.07 c _0 -0 .06 -0 .05 ird De\ -0 .04 r0 .03 Standa r0 .02 -0 .01 - 0 2.5 Figure 6.1.4 AOUR ( • ) , average AOUR ( • ) , and standard deviation ( A ) VS methanol concentration. MLVSS 1370 mg/1. 126 assumed, the slope of the oxygen consumed versus methanol concentration is equal to the substrate oxidation coefficient (OC/S). The yields thus obtained (1 - OC/S) from the two sets of data were 0.23 ± 0.01 and 0.36 ± 0.01 (mg COD biomass / mg COD methanol) respectively. Although these two sets of data were collected only a few days apart, the yields were quite different and the 95% confidence intervals did not overlap. The variability of the kinetic and stoichiometric parameters wil l be discussed in chapter 7. In the original AOUR method (Cech et al 1984), the yield is calculated for each data point independently (Y = 1 - OC / S). When this was done with the data in figure 6.1.6, it was seen that the yield was not a constant, but increased at low substrate concentrations as shown in figures 6.1.7 and 6.1.8. This was due to the slight curvature of the lines in figure 6.1.6 at low substrate concentration. The AOUR data are shown in figures 6.1.7 and 6.1.8 for comparative purposes. The yield appeared to follow an inverse relationship with the metabolic rate at low substrate concentrations, then when the AOUR was constant, the yield was also constant. There are three possible explanations for this phenomenon. The standard deviation of the oxygen consumption data is graphed in figure 6.1.9: The standard deviation increased with increasing oxygen consumption, but the error in oxygen consumption at low methanol concentrations will have a greater impact on the error ofthe calculated yield. If the error in the OC data is ± OC, then the error in the calculated yield will be +/- OC / COD. At low methanol concentrations, ± OC is approximately 0.001. When the COD is close to this value, the error in the calculated yield will be large. Therefore, the apparent trend in the yield may be just an artifact of the data analysis. 127 0.1-0.09-0.08-c o 0.07-'->- 3 D "> 0.06-(D o 0.05-D -r— 0.04-a tn 0.03-0.02-0.01-0-Average OUR (mg 0 2 / l minute) Figure 6.1.5 AOUR Standard deviation vs. methanol concentration. MLVSS 3755 mg/1 ( o ) , and MLVSS 1370 mg/1 (•). 0 1 2 3 4 5 6 Methanol (mg COD/I) Figure 6.1.6 Oxygen consumption (•), and average oxygen consumption (•), vs methanol concentration, MLVSS 3755 mg/1. Oxygen consumption (A), and average oxygen consumption (•), vs methanol concentration, MLVSS 1370 mg/1. 128 I r—^, 1 I I I I 0 1 2 3 Methanol (mg COD/I) Figure 6.1.7 Respirometric yield (•), average yield (•), and AOUR (A) vs methanol concentration. MLVSS 3755 mg/1. 1.4 1.2 - 1 ^UIUU -0.8 0 -0.6 CD -0.4 AOUR -0.2 2 3 4 Methanol (mg COD/I) Figure 6.1.8 Respirometric yield ( • ), average yield (•), and AOUR (A) vs methanol concentration. MLVSS 1370 mg/1. 129 A second explanation for greater yield at low substrate concentration may be due to energy spilling as discussed in section 2.4. As the catabolic activity increases, the anabolic capabilities of the bacteria will be surpassed, and the yield wil l decrease as energy is spilled to avoid the buildup of toxic compounds or ATP. The activated sludge used in the respirometric assay was obtained from continuous activated sludge units. The growth rate ofthe biomass was low, allowing for energy spilling when the methanol concentration was suddenly increased. If energy spilling is the explanation for the decrease in yield with increasing methanol, the yield may be expected to keep decreasing, instead of leveling off as observed. Also, a greater effect on yield would be expected using the biomass from the activated sludge unit operated at an SRT of 15 days (compared to the biomass from the unit operated at an SRT of 5 days). This was not observed. A third explanation for the observed behaviour of the yield comes from the non-steady state nature of the method. If at the start of the batch test there is an oxygen reserve in the floe that is not measured by the dissolved oxygen probe, and this reserve gets used up during the test, then more oxygen will be consumed during the metabolism ofthe substrate than will be measured by the decrease in dissolved oxygen in the bulk mixed liquor. Consequently, the yield will appear to increase at small substrate additions. At larger substrate additions, the oxygen reserve in the floe becomes insignificant compared to the amount utilised from the bulk liquid. Assuming that the reserve of oxygen used in the floes is a constant, the variation of yield with substrate addition may be modeled (lines in figures 6.1.7 and 6.1.8). The dissolved oxygen reserve in the floe is BO Average Y (mg COD/mg COD) 0.3 0.35 0.4 0.45 0.5 i i ' U O 0.02 -0.2 -0.18 -0.16 c -0.14 O "5 -0.12 > a; ~u -0.1 "S - D -0.08 ~D C D -0.06 1/1 > --0.04 -0.02 Average OC (mg O /I) Figure 6.1.9 OC standard deviation vs. average OC. MLVSS 1370 mg/l ( • ) , MLVSS 3755 mg/l (•). Respirometric yield standard deviation vs. average yield. MLVSS 1370 mg/l (A), MLVSS 3755 mg/l (•). 0.9-0.6 0.8 1 1.2 SUR (mg C O D / I minute) 1.8 Figure 6.1.10 OUR vs. methanol addition rate. MLVSS: 1500 mg/l 131 calculated to be 0.005 mgCVl mixed liquor, which corresponds to approximately 1.7 mgCVl floe (using a M L V S S concentration of 3755 mg/l, and assuming a floe density of 1.25 • 106 mg floc/1 floe, (Benefield and Molz 1983)). This value of 1.7 mg/l floe will correspond to the steady state dissolved oxygen concentration in the floe at which the rate of diffusion of oxygen into the floe is equal to the endogenous OUR. Due to the large error in the calculated yields at low substrate concentrations, it is difficult to determine i f the increase in yield is real or not. The yield has been assumed to be constant for a given set of experiments, unless otherwise noted, and has been calculated by linear regression ofthe oxygen consumption versus substrate addition curves. This method of calculation will put more emphasis on the points at higher substrate concentrations (which agrees with the first and third possibilities outlined above). A n alternate method of calculating the yield is shown in figure 6.1.10. Methanol was added continuously to the respirometer, and the OUR was measured. For calculating the substrate uptake rate, the residual methanol was assumed to be negligible, so the substrate uptake rate was equal to the substrate addition rate. The yield is equal to 1 -OUR / SUR, and was 0.41 mg COD/mg COD. The AOUR yield calculated using the same biomass was 0.32 to 0.39 mg COD/mg COD, slightly lower than that measured using continuous methanol addition. With the continuous method, the yield decreased with increasing methanol addition rate (figure 6.1.11). As the methanol addition increased, the metabolism increased, which may result in a decrease in the yield as discussed previously. Using the calculated yield, the substrate uptake rate may be calculated using equation 4.5. The SURs for the two sets of data discussed are presented in figure 6.1.12. 0.6 0 . 1 -0—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i—|—i—i—i— 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 SUR (mg C O D / I minute) Figure 6.1.11 Respirometric yield vs SUR, continuous addition of methanol to a respirometer. MLVSS: 1500 mg/l 133 The two sets of data are similar at high substrate concentrations, but quite different at low substrate concentrations. The explanation for the discrepancy at low methanol concentrations will be discussed in the next section. These two data sets were collected just a few days apart, the specific maximum SURs are similar even though the yields are quite different. The amount of enzymes does not appear to have changed, but the amount of the substrate used for energy has. One possible explanation is the presence of slowly biodegradable components present in the mixed liquor during one of the experiments. Another explanation is the presence of compounds or conditions causing stress to the microorganisms, resulting in more substrate being oxidised for maintenance purposes. Yet another possibility is that the substrate utilisation enzyme levels (catabolism) have not changed, but the growth enzymes (anabolism) have changed, resulting in more or less substrate being wasted. The yield on methanol appeared to be more variable than the substrate removal rate. Different substrates can be expected to have different oxidation kinetics, and different yields. Acetic acid and formic acid are other readily biodegradable substrates present in B K M E . A set of AOUR data and oxygen consumption data for acetic acid and formic acid is shown in figures 6.1.13 and 6.1.14. For this set of biomass, the AOUR due to acetic acid was slightly lower than that due to formic acid, and both were lower than the AOUR due to methanol. The oxygen consumption per mg COD formic acid added was similar to the oxygen consumption per mg COD methanol added, implying that the yield on formic acid was similar to the yield on methanol. This is expected since both compounds are one carbon compounds and are probably oxidised by similar pathways. The yield on acetic acid (0.7 mg COD biomass/ mg COD substrate) was much greater 134 0.0014 0.0012-D C ; E 0.001 -oo > -0.0008-cn Q 0.0006-o U cn 0.0004-_E O O 0.0002-2 3 4 Methanol (mg COD / I ) Figure 6.1.12 SUR vs. methanol concentration. MLVSS 1370 mg/1 (•), and MLVSS 3755 mg/l<». 2.5-_ 2-m "5 c "E — 1.5-o cn X 1-ZD o < 0.5- A * i T— i—|—i—i—i—i—|—rn— i—i—|—r 4 5 6 Substrate (mg COD / I ) 3 Figure 6.1.13 AOUR vs substrate, formic acid (A), acetate (•), and methanol ( • ) . MLVSS 1810 mg/1 135 than the yield on methanol and formic acid. The yield on a two-carbon compound is expected to be greater than yield on a one-carbon compound. A n important observation from this set of data is that when the OURs are converted to SURs (figure 6.1.15), the SUR of acetic acid was greater than that of formic acid, even though formic acid gave the higher AOUR. This was due to the differences in the yields of the two compounds and demonstrates that caution must be exercised when comparing OURs due to the oxidation of different compounds. A higher OUR does not necessarily mean a higher SUR as commonly assumed. Curve Fitting The data set of methanol AOURs obtained during the treatment of batch S was chosen for identification of the best model since it is the most complete. For the purposes of curve fitting, all data points were treated as equals. More error is expected at low substrate concentrations, where the OUR has to be calculated over a smaller number of data points, but an examination of the standard deviations of the OUR data did not show any trends (figure 6.1.5). Certain linear transformations of nonlinear models, such as the Monod model, may result in unequal weights being applied to the data. In this study, all coefficients have been calculated using non-linear regression. The Monod, Powell, Moser, Konak, Tessier, and Blackman models were fitted to the experimental data using nonlinear regression techniques (figures 6.1.16 and 6.1.17) and the calculated coefficients are listed in table 6.1. Of the models tested, Monod is the most commonly used in the respirometry literature for non-inhibitory microbial kinetics. Visual inspection of figure 6.1.16 shows that the Monod model does not saturate quickly enough with increasing substrate concentration to adequately describe the data. 136 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Substrate (mg COD/I) Figure 6.1.14 Oxygen consumption vs. substrate, formic acid (A), acetate (•), and methanol (•). MLVSS 1810 mg/1 Ul CO Q O U CO .£ ZD ul 0.0018-0.0016-0.0014-0.0012-0.001 -0.0008-0.0006 0.0004-0.0002-A A 0 0 0.5 2 T 1.5 2.5 3 Substrate (mg COD/I) I ' ' 3.5 4.5 Figure 6.1.15 SUR vs substrate, formic acid (A), acetate (•), and methanol (•). MLVSS 1810 mg/1 Methanol (mg COD/I) Figure 6.1.16 AOUR vs. methanol concentration (o), with Monod ( ), Powell ( and Blackman ( ), curve fits. M L V S S 3755 mg/l. Methanol (mg COD/I) Figure 6.1.17 AOUR vs. methanol concentration (o), with Monod ( Konak ( ), and Moser ( ), curve fits. M L V S S 3755 mg/l. 15 30 -), Tessier (-138 Examination of kinetic data in the literature reveals that this phenomenon is not limited to this study, though the Monod model has been assumed in all of these other cases. A l l of the equations discussed share a common parameter, the maximum AOUR, which is the theoretical AOUR at infinite substrate concentration. It would be nice to be able to measure this theoretical value, but it is impractical to measure the JOUR at infinite substrate concentration. In figure 6.1.16, AOUR measurements at various high substrate concentrations (S » K) all had approximately the same value. This value, 3.7 mg O2 /1 minute (0.00098 mg 0 2 / mg M L V S S minute), may be called the maximum experimental AOUR and be used to compare with the maximum AOUR parameter calculated using the various models. For example, the Monod model greatly overpredicts the AOUR at high methanol concentration. At 50 mg COD /1, the AOUR for the Monod equation (0.00122 mg 02 / mg M L V S S minute) is 25% greater than the experimental maximum AOUR. The experimental AOUR did not reach the theoretical maximum AOUR as predicted by Monod. Powell's, Moser's, Konak's, and Tessier's equations all predict similar maximum AOUR's, all of which are slightly higher than the experimental maximum OUR (from 4 to 7% higher), but these are more reasonable values than predicted by Monod. The use of Blackman's equation results in a maximum AOUR coefficient which is slightly (7%) lower than the experimental maximum. At first glance all of the other equations seem to fit the data better than Monod's model. A n examination of the residuals (expected AOUR - actual AOUR), shown in figure 6.1.18 and 6.1.19, will help to identify which models are the most appropriate. Any systematic departures from the estimated regression equation (because the model is not adequate), or non constant variance can be detected by studying the residuals 139 - 1 —r-i—i—i—i—|—i—i—i— | i i—i—r-|—i—i—I—r-p-i—i—I I | i—r-r-i—| i —i—[-—r—i—i—i—pi—<~'—i—I-'-!—n-| 0 0 . 5 1 1 . 5 2 2 . 5 3 3 . 5 4 4 . 5 5 Expected Value Figure 6.1.19 Residual vs. expected value for the curve fits presented in figure 6.1.17. Monod (•), Moser (•), Konak (A), and Tessier ( ). MLVSS 3755 mg/l. 140 (Himmelblau 1980). In order to more easily compare the residuals obtained from the different curve fits, they were fitted with polynomial equations, and all of the resulting curves are shown in figure 6.1.20. These curves have no theoretical value, they just make it easier to look visually inspect the residuals. Examination ofthe residuals (figure 6.1.20) shows that the Monod equation, which has only 2 parameters, has the worst fit with an obvious trend in the residuals. The Blackman equation (2 parameters) also has a poor fit, and residuals seem to follow an opposite trend to the residuals obtained from the Monod curve fit. Both are limiting cases ofthe Powell equation, which appears to give the best fit to the data (no obvious trends in the residuals). If L = 0 in Powell's equation, Monod's equation results, and i f K= 0 in Powell's equation, Blackman's equation results. Both L and K were found to be significant in Powell's equation ( 0) using the t test. Comparison of the models shows that the third coefficient in Powell's equation is significant. Tessier's equation, which has only 2 parameters, fits better than Monod's and Blackman's, which have only 2 parameters, and Konak's equation, which has 3 parameters. Moser's equation (3 parameters) appears to fit the data fairly well, and the third coefficient, n, was found to be significant. Another way to compare regression equations is to compare the pooled variances (Himmelblau 1980). According to this method, the Powell equation gives the best fit to the data, followed by Moser's equation. -0.4--0.5—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—\—i—i—i—i—\—i—i—i—i—|—i—i—i—i-0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Expected value Figure 6.1.20 Residual vs. expected values from figures 6.1.18 and 6.1.19. Monod ( Powell ( ), Moser ( ), Konak ( ), Tessier ( ), and Blackman ( MLVSS 3755 mg/l. 142 Table 6.1 A O U R Parameters Model Maximum OUR K Other parameter Monod (3.1) 0.00124 0.774 -Powell (3.11) 0.00103 0.0917 L = 0.884 Moser (3.15) 0.00105 0.358 n= 1.595 Konak(3.17) 0.00105649 0.990044 n= 1.09735 Tessier (3.16) 0.00104 1.264 -Blackman(3.12) 0.000919 - A = 3.6159 Experimental 0.000985? - -In order to investigate parameter correlation, the 95% confidence regions for the Monod and Powell parameters were calculated (figures 6.1.21 and 6.1.22). For the Monod equation, O U R M A X and K M appear to be correlated, which is indicated by the non-circular nature of the confidence region. Since the Powell equation has three parameters, the confidence regions for 2 parameters are shown on the plane of the third parameter held constant at the value obtained during regression. The confidence regions for the Powell parameters are much smaller than the confidence regions for the O U R M A X and K M obtained using the Monod equation. The confidence regions for the Powell coefficients may be further reduced if equation 6.1 is used for regression instead of equation 3.13 (figure 6.1.22). Equation 6.1 is obtained by solving equation 3.12 for u. before substituting L = U M A X / ( Y F) for F. ' OUR = ±[K.F+ OURMAX + F.S-4{K-F+ OURUAX + F-SJ - WURMAX -F-s) (6.1) All future calculations of the Powell coefficients were done using equation 6.1, and F was converted to L after nonlinear regression. The Powell parameters are also correlated, in particular the parameters L and K , which is expected since L is a modifier of K . The Powell half saturation value is equal to K + L / 2 . K still has some dependence on the O U R M A X , but not as much as with Monod's equation. 143 0.8 0 . 7 V K, M (mg COD/1) 0.6 0.5 0.4 0 . 0 0 1 0 . 0 0 1 0 5 0 . 0 0 1 1 0 . 0 0 1 1 5 0 . 0 0 1 2 0 . 0 0 1 2 5 Maximum AOUR (mg 02/mg MLVSS minute) Figure 6.1.21 Monod parameters 95% confidence intervals. 144 Figure 6.1.22 Powell parameters 95% confidence intervals. Outer ellipse (and dashed ellipse in figure 6.1.22c) calculated with equation 3.11; inner ellipse calculated with equation 6.1. 145 The confidence intervals for the Monod and Powell equations are shown in figure 6.1.23. As expected from the above discussion, the Powell equation has much smaller confidence intervals and approximated the data much better than the Monod equation. At high substrate concentrations, the experimental data do not even fall between the Monod 95% confidence intervals. The standardised residuals may be used to investigate i f the microbial activity is shifting with time. The kinetic assays usually take a few hours per batch of biomass, and may take up to 8 hours if many data points are obtained (as in the data shown in figures 6.1.3 and 6.1.4). During this time period, the biomass is exposed to different environmental conditions than the conditions in the continuous activated sludge unit, from which the biomass was obtained. In particular, the only BOD source is methanol, and the BOD is added discontinuously (once per data point), so that the biomass experiences short bursts of high substrate concentrations, followed by short periods of starvation. There is a possibility that the biomass will adapt to these new environmental conditions, and the OUR response will change over the course of the experiment. Figure 6.1.24 shows the standardised residuals for both the Monod and Powell correlations versus the time that the data point was collected for the experiment shown in figure 6.1.3. The Monod data shows runs in the residuals, probably due to the correlation among the Monod residuals. There is no noticeable trend with time in the residuals from the Powell correlation. The assumption that the biomass does not adapt over the course of the experiment is justified when the Powell equation is used. The Monod and Powell equations were fitted to acetate oxidation data (figure 6.1.25) and formate oxidation data (figure 6.1.26). For acetate, the Powell equation does 146 6 5H 0 1 2 3 10203040 Methanol (mg COD/I) Figure 6.1.23 Methanol AOUR (o) vs substrate, Monod ( ), and Powell ( ), curve fits with 95% confidence intervals (dashed lines). MLVSS 3755 mg/l. 0.6-0.4H 0.2-o ^ 0-0) -0.2H -0.4 -0.6-• fl- flfi— • • • « • • • • • • • • • • • • • • a • • eft] •• °a •• • • a D A . • i • • • • • _ • I a i 1 1 1 1 i 1 1 1 1 i 1 1 1 1 i 1 10 20 30 40 Experiment # 50 60 70 Figure 6.1.24 Residual from Monod curve fit (•), and from Powell curve fit (•) vs. experiment # (each experiment was approximately 2 minutes apart). MLVSS 3755 mg/l. 147 not give a better fit than the Monod equation (for the data set shown, although the Powell equation resulted in better fits when other data sets were analysed). For formate, the Powell equation gives the best fit. For other formate respirometric data sets, the Monod equation gave the best fit to the data. B K M E is a mixture of methanol, formic acid, acetate, and other compounds, Both the Monod and Powell equations have been developed for application to growth on single substrates. AOUR tests were done on a mixture of these compounds to test the applicability of the Powell and Monod models to respirometric data obtained using multicomponent substrate mixtures. AOUR curves for methanol - acetate - formic acid mixtures obtained using different biomass samples are presented in figure 6.1.27. For three of the curves, the Monod model gave a better fit than the Powell model. A possible explanation is that these curves were obtained at intermediate M L V S S concentrations (see next section). Another explanation is the presence of formic acid, which often fits the Monod model better than the Powell model. It has been empirically observed that i f the respirometric kinetics of one substrate in the mixture are best fit with the Monod model, then the respirometric kinetics of the mixture will also be best fit with the Monod model. AOUR curves for two different samples of B K M E also are more closely approximated by the Monod equation (figure 6.1.28). It is very difficult to get a complete AOUR curve for untreated wastewater, as the concentrations of the various substrates in effluent are fairly low. Very large injection volumes (up to one half of the volume of the respirometer) are required to approach the maximum OUR. Injections of this size into the respirometer result in a large loss of biomass, which affects the respirometric results. 148 Acetate (mg COD/I) Figure 6.1.25 AOUR vs. acetate concentration (•), with Monod (solid), and Powell (dotted) curve fits. MLVSS 1810 mg/1. 0.7 0-f—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—|—i—i—i—i—I 0 0.5 1 1.5 2 2.5 Formate (mg COD/I) Figure 6.1.26 AOUR vs. formic acid concentration (A), with Monod (solid), and Powell (dotted) curve fits. MLVSS 1810 mg/1. 149 0 5 10 15 20 25 COD (mg/l) Figure 6.1.27 AOUR vs substrate for formic acid/methanol/acetate mixtures. MLVSS values as shown on chart. 3.5 Volume injected (ml) Figure 6.1.28 AOUR vs substrate for BKME. MLVSS concentrations. 1644 mg/1 (•), 1858 mg/1 (•). 150 On a few occasions, sludge was concentrated by centrifugation, then diluted to the. original M L V S S concentration with a large volume of fresh effluent to obtain data points at high effluent concentrations. The results obtained with this method are presented in figure 6.1.28. Similar to the methanol - acetate - formic acid mixtures, the oxidation kinetics of effluent were found to follow Monod kinetics. This may be due to the relatively low M L V S S concentrations used for these experiments, as well as to the large number of dilute substrates present in B K M E . In summary, the Powell equation was found to give the best fit to the individual substrate data. As discussed previously, the Powell equation is derived by assuming two or more reactions in series. If just one reaction is dominating, then the Monod equation will result, which was found for certain acetate and formic acid data sets. With substrate mixtures, the Monod equation has been found to give the best representation of the data, and will be considered for future data analysis due to its widespread acceptance in the literature. Effect of Biomass Concentration Respirometric data collected using various biomass concentrations with methanol as the substrate are shown in figures 6.1.29 and 6.1.30. Theoretically, the biomass concentration wil l not have an effect on the Monod parameters, which may be calculated using equation 6.2 i f q is expressed on a volumetric basis. S where q is the total AOUR, and not the specific AOUR. 151 5-T 4.5-Methanol (mg COD/I) Figure 6.1.29 Methanol AOUR vs. substrate, using various MLVSS concentrations. 4440 mg/1 (•), 2965 mg/1 (A), 1535 mg/1 (•), 755 mg/1 (•). The solid curve fits are made with Powell's equation, the dotted curve fits are made with Monod's equation. Methanol (mg COD/I) Figure 6.1.30a Methanol AOUR vs. substrate, using various MLVSS concentrations. 4150 mg/1 (A), 2735 mg/1 (•), 1505 mg/1 (•), 730 mg/1 (•). The curve fits are made with Powell's equation. 152 The Monod parameters were calculated for the data shown in figure 6.1.29, using all ofthe data points (curves shown), and separately for each M L V S S concentration (curves not shown). When the parameters were calculated for each biomass level, the specific maximum AOUR was constant, regardless of the biomass concentration (figure; 6.1.31). However, the half saturation constant was directly proportional to the solids concentration (figure 6.1.31). This is verified by examining the 95% confidence regions for the Monod parameters which show a distinct trend in the half saturation constant (figure 6.1.32). Ideally, the 95% confidence regions for the four data sets would be superimposed, since the same activated sludge was used for the collection of all four sets of data. The 95% confidence regions for the entire data set (dashed ellipse) should be within the regions for the four data subsets, which is clearly not the case. A consequence of mass transfer effects is an apparent increase in the half saturation constant. To see i f the increase in K M was due to mass transfer effects, the ability of the Powell equation to fit the same data set was investigated. The Powell equation was derived by assuming two reactions in series. At high substrate concentrations, the reaction rate is limited by the number of available enzymes, and the volumetric rate should be directly proportional to the active biomass concentration. At low substrate concentrations, the rate limiting reaction may be either a metabolic reaction or a mass transfer step. If this second rate limiting reaction is a metabolic reaction, then the volumetric rate will be proportional to the biomass concentration (assuming a constant substrate concentration). If the second rate limiting reaction is a mass transfer step, the volumetric rate will also be proportional to the biomass concentration, assuming that changes in biomass concentration are brought about by changes in the number of 153 5 4.5-3 Methanol (mg COD/I) Figure 6.1.30b Methanol AOUR vs. substrate, using various MLVSS concentrations. 4150 mg/l (A), 2735 mg/l (•), 1505 mg/l (•), 730 mg/l (•). The curve fits are made with Monod's equation. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 MLVSS (mg/l) Figure 6.1.31 Half saturation constant (Monod) vs. MLVSS concentration. Data from figure 6.1.28 (•), and figure 6.1.27 (•). Maximum specific AOUR, from figure 6.1.28 (•) and figure 6.1.27 (A). 154 0 r , i • i i 1 1 - — i — • • • ' 1 — ' ' 1 1 — 1 — 0.0009 0.001 0.0011 0.0012 Maximum AOUR (mg 0 2/mg M L V S S minute) Figure 6.1.32 Monod parameters 95% confidence intervals, calculated for each M L V S S level (solid circles), and using all ofthe data (dashed circle). 155 floes, and that biomass properties are independent of concentration. In either case, all three kinetic parameters of the Powell equation should be independent ofthe biomass concentration, when the specific rate is related to the substrate concentration. Fitting the Powell equation (6.1) to the complete data set, and separately to the four M L V S S subsets, the O U R M A X was found to be a constant regardless ofthe biomass concentration (figures 6.1.29 and 6.1.30a). Similar to the Monod equation, L , and K , were found to be directly proportional to the M L V S S . (figure 6.1.33). The Powell coefficients, K and L, are closely related to the Monod half saturation constant K M , and. appear to follow the same relationship with M L V S S as K M . The Powell equation, similar to the Monod equation, is not suitable for determining the parameters at one M L V S S level, and applying those parameters to predict the results obtained at a different M L V S S concentration. The Powell equation may be modified to account for the dependence ofthe coefficients on the biomass concentration if the rate limiting mass transfer step is independent of the biomass concentration (the rate is independent, not the parameter). This situation will arise when the change in biomass is due to a change in floe size and not a change in floe concentration, or i f the rate limiting mass transfer step is inside the D O probe and not in the floe or bulk solution. As the biomass changes, the metabolic reaction will change in direct proportion, but the mass transfer rate will remain constant (for a constant substrate concentration). In an equation relating the specific metabolic rate to the substrate concentration, this will have the same effect as making the mass transfer coefficient a function of the biomass concentration. For this scenario, equations 3.11 and 6.1 become: 156 0.25-0.2-Q 0 .15^ O U co Ji o.H 0.05-1 1 1 I 1 1 1 1 i 1 1 1 1 i 1 1 1 1 i 1 1 1 ! I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 0 500 1000 1500 2000 2500 3000 3500 4000 4500 MLVSS (mg/l) Figure 6.1.33 Powell kinetic coefficients vs. biomass concentration L (•), K (•) , and maximum AOUR (A). 157 yMAX(K + LxX+S) ix = 21 ' 1-4LXXS 1 x (K + L X X + S ) (6.3) li = MKFx + OURMAXX + FXS - 4{KFX +OURMAXX+ F^)2 - WUR^XF.s) (6.4) where L and F have been replaced with L x and F x , to demonstrate the dependence of these parameters on the biomass concentration. Using the modified Powell equation, O U R M A X and L x were found to be constant regardless of biomass concentration. This equation gave a better fit to the data than the Monod equation. This was verified by looking at the confidence regions of the parameters for the data set shown in figure 6.1.29. The confidence regions are much smaller for the Powell parameters (figure 6.1.34) than for the Monod parameters (figure 6.1.32). The confidence regions of the modified Powell parameters for the four data subsets are not superimposed, but are scattered about the confidence region for the complete data set with no obvious trends in the parameters. It must be emphasised that the modified Powell equation is an empirical one, the good fit of the equation to the data does not mean that the assumptions behind the equation are valid. The assumptions made do not make sense for the circumstances used to collect this set of data, but may explain observations in the literature where K M was found to a function of biomass (see section 3.2, alternatives to Monod). If the change in biomass concentration is due to changing operating conditions, such as the SRT, then the biomass properties wil l be changing as well, making the mass transfer coefficient a function of biomass concentration. In this study, the biomass properties can be expected to remain constant since all of the data were collected using the same set of sludge. An 158 Figure 6.1.34 Modified Powell parameters 95% confidence intervals, calculated for each MLVSS level (solid circles), and using all of the data (dashed circles). 159 alternate explanation for the observed dependence ofthe kinetic coefficients on the biomass concentration, which does not require the biomass properties to be a function of biomass concentration, will be outlined in the following sub-section (verification of AOUR model). Figure 6.1.35 shows curves calculated using external mass transfer and the Monod equation (equation 3.28), and external/internal diffusion with the Monod equation (equation 3.30). The curves were calculated using 2 point collocation techniques. When diffusion through the bulk solution was ignored, the internal diffusion model did not approximate the data as well as when external diffusion was included in the model (not shown). This may be due to the extra parameter when external diffusion is added to the model. In both cases, the internal diffusion mass transfer coefficient was found to be a linear function of the biomass concentration. The mass transfer coefficients, similar to the Monod and Powell coefficients, are supposed to be independent of the biomass concentration. The dependence of the mass transfer coefficient on the biomass concentration indicates that equations 3.28 and 3.30 are unsuitable for modeling respirometric data. Convective mass transfer into the floe is assumed to be negligible in equations 3.28 and 3.30. This is a common assumption in modeling mass transfer effects in the activated sludge process. Recent observations indicate that the floes are more porous than previously thought, and that the floes do not contain dead centers as predicted by the diffusion hypothesis. Assuming that the main mass transfer resistance is due to convective mass transfer, instead of diffusion, results in the same overall mass transfer effects as i f diffusion is assumed to be the rate limiting step. From the data obtained, it is 160 • y Methanol (mg COD/I) Figure 6.1.35 Methanol AOUR vs. substrate, using various MLVSS concentrations: 4440 mg/l (•), 2965 mg/l (A), 1535 mg/l (•), 755 mg/l (•). The solid curve fits are made with Monod's model assuming external mass transfer resistance, the dotted curve fits are made assuming internal and external mass transfer resistance. 161 difficult to differentiate between the two processes, and both are likely to be important. The situation is further complicated by the heterogeneous nature of the activated sludge floes, which wil l result in variable diffusivity. The modified Powell equation fits the data better than either of the diffusion scenarios. A l l equations default to the Monod equation when mass transfer becomes non-rate limiting. In the modified Powell equation, i f the mass transfer resistance is small, then the equation defaults to the Monod equation with a half saturation constant equal to K+ LX/2 . This is a function of the biomass concentration, and for the activated sludge used in this study, M L V S S levels below approximately 1000 mg/1 resulted in the Monod model giving as good a fit as the Powell model. Above 1000 mg MLVSS/1, the Powell equation usually gave a better fit. Even at low biomass concentrations where the data are fit by the Monod model, K will still appear to be a function of the biomass concentration. Respirometric data using acetate as the substrate were collected at several biomass concentrations (figure 6.1.36). The modified Powell equation gave a slightly better fit to the data than the Monod equation. The Monod half saturation constant was found to be a function of the biomass concentration, as when methanol was used as the substrate. The modified Powell coefficients were found to be constant regardless of the biomass concentration. Respirometric data using formic acid as the substrate, at various biomass concentrations, is presented in figure 6.1.37. For this data, the Powell equation did not give a better fit than the Monod equation, and neither seemed to really fit the data. Neither model accounted for the large increase in AOUR observed at high substrate concentrations. 162 2 - j 1 -8-1.6--"5 c 1-4-£ — 1-2-CN -0 1 -CD E, 0.8-an no 0.6-< ] 0.4-0 .2-0 J 4 10 20 30 40 Acetate (mg COD/I) Figure 6.1.36 Acetate AOUR vs. substrate, using various MLVSS concentrations: 5060 mg/l (A), 3560 mg/l (•), 1920 mg/l (•). The solid curve fits are made with Powell's equation, the dotted curve fits are made with Monod's equation. 2-T 1.8-1.6-"5 C 1.4-£ 1-2-0 1 -CD JE 0.8-no 0.6-< 0.4-0.2-j 0-6 8 Formate (mg COD/I) Figure 6.1.37 Formic acid AOUR vs. substrate, using various MLVSS concentrations: 5060 mg/l (A), 3560 mg/l (•), 1920 mg/l (•). The curve fits made with Powell's equation and with Monod's equation fell on the same line. 163 Verification of Models The Powell equation fit the respirometric data, but the values of L and K were functions of the biomass concentration. One of the major assumptions made in the analysis of respirometric data is that of pseudo steady state. Pseudo steady state implies that as soon as the substrate is added to the respirometer, the bacteria start metabolising the added substrate at a rate equivalent to the steady state rate corresponding to the initial substrate concentration. This immediate change in metabolism causes an immediate change in the oxygen uptake rate which is measured by the DO probe. Under these assumptions, the maximum AOUR achieved corresponds to the metabolic rate which corresponds to the added substrate concentration. This assumption may seem less reasonable if the actual process is investigated in more detail. Upon addition of substrate, the substrate mixes throughout the respirometer and diffuses through the bulk solution to the floe. It is possible that the equilibrium constant favors substrate adsorption onto the floe. The substrate must then diffuse into the floe to the bacteria. The next step is transport of the substrate across the cell membrane (except when methanol is the substrate, which is oxidised in the cytoplasm). The substrate is metabolised, a process which involves the transfer of electrons from the substrate to the electron transfer chain. In the case of aerated activated sludge, the terminal electron acceptor is oxygen. The oxygen utilisation rate in the bacteria will increase above the rate during endogenous metabolism. This will cause oxygen from the floe to enter into the bacteria at a greater rate, lowering the oxygen concentration in the floe, causing the oxygen from the bulk solution to diffuse into the floe at a greater rate. Finally, it is the decrease in the bulk oxygen concentration which is measured by the DO probe. In summary, the OUR will 1.64 only increase once substrate dissolved oxygen concentration in the floe starts to decrease, which happens after the substrate has diffused into the floe and has been metabolised. Low Substrate Concentrations The unsteady state nature of the AOUR assay can explain the dependence of the parameters K M (Monod equation) and L and K (Powell equation) on the biomass concentration, i f oxygen diffusion is rate limiting. The following discussion assumes that the substrate concentrations is such that the AOUR is below the maximum, in the first order region of the AOUR vs. substrate curve. If the number of floes is doubled, the substrate will disappear from the bulk solution twice as fast but each floe will only see half as much substrate (a finite amount of substrate added, evenly distributed to all of the floes). Consequently, the amount of oxygen consumed per floe wil l decrease by half, and the driving force for oxygen diffusion from the bulk liquid to the floe will decrease by approximately half. This will result in a decrease in the specific oxygen consumption rate. Since the number of floes is double, the overall volumetric rate will be very similar to the overall volumetric rate before the number of floes was doubled. Different results would be obtained i f the substrate concentration in the bulk solution was kept constant. Then, once steady state was attained, the overall rate would be directly proportional to the biomass concentration. In the AOUR assay, substrate is added as a pulse, and is removed from the bulk solution before steady state is attained. This hypothesis is supported by examination of the OUR profiles during the AOUR test (figure 6.1.38). The initial OUR following methanol addition was the same regardless of the biomass concentration. As the biomass concentration increased, the maximum volumetric OUR increased, but the rate at which the OUR increased following 165 Time (minutes) Figure 6.1.38 Methanol injections, various methanol concentration. MLVSS 1535 mg/1 ( ), and MLVSS 4440 mg/1 ( ). 166 substrate addition was independent of biomass concentration. The response time of the AOUR assay is mass transfer limited. This provides insight into the success of the modified Powell equation in fitting batch test respirometric data. The mass transfer parameter, L x or F x , is probably related to the oxygen diffusivity, and not the substrate diffusivity, when methanol is used as a substrate. This observation may not hold for other, larger, substrates. It is the unsteady state nature of the respirometric method that results in L x being a function of the biomass concentration. A study using a fluorometer found that the rate limiting step of glucose utilisation by bakers' yeast was the substrate permease. The mixing time for the glucose to enter the cells was about four and a half seconds (Einsele et al 1978). This result agrees with the hypothesis that substrate metabolism will not reach it's maximum rate immediately upon substrate addition into the bulk liquid. Given the above discussion, the modified Powell equation may be used to model respirometric data. The Monod equation may also be modified to account for the increase in the half-saturation constant with biomass. q = q ^ x i d h (6-5) If the unmodified Monod equation is used, K will be overestimated at high biomass concentrations. If the equation is applied to predict steady state behaviour of activated sludge plants, the SUR will be underpredicted. Many other factors affect the K value (discussed in chapter 7), so the assumption of pseudo-steady state wil l not result in too much error when dealing with complex wastewaters. 167 This discussion supports the hypothesis made earlier that the yield appears to increase at low substrate concentrations due to a small reserve of dissolved oxygen in the floe. The mass of oxygen in the floe appears to depend mainly upon the bulk dissolved oxygen concentration, the oxygen diffusivity, and the floe size. High Substrate Concentrations When the substrate concentration is high enough to result in the maximum metabolic rate (zero order region) the assumption of pseudo steady state becomes valid. Upon addition of substrate, the OUR will increase until the maximum rate is reached. The OUR will stay at the maximum rate as long as there is enough substrate. Since the rate follows the zero order model at this point (therefore not a function of substrate concentration), the initial concentration is not of critical importance. If the initial concentration were kept constant throughout the test, the rate would be the same as measured with the AOUR test. The maximum SUR calculated from respirometry may be verified using a batch test experiment with the measurement of substrate and OUR. A n example of such a batch test is presented in figure 6.1.39. The measured SUR is 0.42 mg BOD/1 minute, the SUR calculated form AOUR tests performed with the same set of biomass is 0.40 mg BOD/1 minute. The AOUR method gives a reasonable approximation of the actual SUR, but is much simpler and takes significantly less time. The yield increased as the batch test progressed (figure 6.1.40). Initially, the yield may have been low due to the sudden shock the biomass experienced when the methanol was added. This may have resulted in energy spilling. As the biomass adapted to the new conditions, less energy was spilled, and the yield increased. 168 -50 0 50 100 150 200 250 Time (hours) Figure 6.1.39 Batch test, methanol BOD (•) and OUR (o) vs time. Methanol was added at timeO. MLVSS: 2000 mg/1 0.5 CO I E 0.3-1 O 0.25-U cn 0.2-2 0.15-> 0.1-0.05-0-j I I I | ' I ' ! | ! I I I | | ' ! I I | ' I I I -50 0 50 100 150 200 250 Time (hours) Figure 6.1.40 Yield (1-OUR/SUR) calculated for batch test data in figure 6.4.39. 169 Using a different set of biomass, the batch test was repeated, once using methanol and once using acetate as substrates (figure 6.1.41). For methanol, the measured SUR • was 2.2 mg COD/1 minute, compared to 2.5 mg COD/1 minute calculated by the AOUR method. The yield on methanol increased by the end of the batch test, similar to the previous example. For acetate, the measure SUR was 1.3 mg COD/1 minute, compared to 1.3 mg COD/1 minute calculated by the AOUR method. The AOUR method gives comparable results to the batch test method. The slight discrepancies are due to the error involved in measuring substrate (the BOD and COD test), and the higher f/m ofthe batch test. If the SUR were calculated from the OUR over the course of the batch test, the SUR of methanol would appear to decrease. The COD and BOD measurements indicate that the SUR was constant, and that the yield must have been changing. A standard method for measuring the maximum growth rate is to use a batch test with a high f/m ratio. The maximum growth rate is directly related to the exponential increase in the OUR with time. The maximum growth rate ofthe activated sludge used in this project was measured on methanol and acetate. There was a long lag phase before the biomass started to grow, and then the growth rate was much higher than predicted by respirometry (figure 6.1.42). The growth calculated for methanol was 0.5 /hour,, compared to 0.1 /hour predicted by the AOUR assay. The growth rate calculated for acetate was 0.8 /hour, compared to 0.06 predicted by the AOUR assay. The probable explanation for the large growth rate is the selection of fast growing microorganisms by the batch test. These organisms were initially present in very small quantities, which explains the long lag phase. In a separate experiment, the maximum growth rate as measured, by the AOUR assay was estimated before and after a high f/m batch test. The 170 Time (minutes) Figure 6.1.41 COD removal and OUR during two batch tests. Substrate added at time 0. Methanol COD (•) and OUR ( ). Acetate COD (A) and OUR ( - - ) . MLVSS: 3100 mg/1. 0 T 8 1 0 Time (hours) 1 8 Figure 6.1.42 OUR vs. time during a batch growth test. Methanol (•), acetate (o, •), acetate and methanol mixture (A , •). 1.71 growth rate of the bacteria at the end ofthe batch test was an order of magnitude greater than the growth rate before the batch test. The high f/m batch test does not appear to be appropriate for the measurement of activated sludge kinetics for the treatment of B K M E . Summary Respirometric data for methanol, formic acid, and acetate were presented. These substrates were chosen as representative substrates of B K M E . By comparing the AOUR profiles of these substrates against that of B K M E , it is evident that these substrates compose the bulk of the readily biodegradable substrate in B K M E , although many other substrates are present as well. Most of the other substrates are other carboxylic acids. When respirometry was performed with sugars (various sugars were used, including xylose), no response was obtained, which implies that these were not readily utilisable substrates. See chapter 7 for further details on the composition of B K M E . The relationship between the OUR and the SUR is dependent on the yield, which is different for each substrate. It is important to note that there is a higher OUR during formic acid metabolism than during acetate metabolism despite the greater removal rate for acetate. This is due to the larger yield on acetate. The respirometric method appears to be a valid surrogate for measuring the actual SUR. The modified Powell equation was found to give a better fit to the data than the Monod equation. This is hypothesised to be due to the unsteady state nature ofthe assay. If this is true, care must be taken in applying the batch kinetic constants to continuous systems operating at steady state, particularly K and / or L . The yield, OUR, and SUR calculated at high substrate concentrations are close to the steady state values. This will allow the use of batch test data to calculate the zero order kinetic coefficients. Given the 172 small saturation coefficients found for methanol, formic acid, and acetate, an assumption of zero order kinetics in dealing with the continuous system should not introduce too much error. 6.2 Effect of DO Concentration The dissolved oxygen concentration may be expected to have a large effect on the oxygen uptake rates. Conventional wisdom states that as long as the DO is above 2 mg/l, it will not affect the metabolic rate of the bacteria, but i f the DO falls below this value, then the kinetics will start to become oxygen limited. For the majority of kinetic tests in this study, the DO concentration was kept above 2 mg/l. For a few tests, the DO concentration was allowed to fall to 0 mg/l while the substrate was present in excess (as in figure 6,2.1). For this set of data, it appears that as long as the DO is above 0.5 mg/l, there is little effect on the OUR. Below 0.5 mg/l, the OUR decreases rapidly. The dependence of OUR on the DO concentration appears to be similar to the dependence of the AOUR on the substrate concentration. If Monod model is assumed for the dependence of the OUR on DO, the following equation is obtained by using equations 4.2, and 3.31: f \ OUR O U R ^ . „ _ + AOUR S baseline ' ^ "MAX V K M + SJ ^DO (6.6) K ^ + D O O U R = {0URbasdine + ± ( K F + O U R ^ + F S - J(KF + O U R M A X + F S ) 2 - 4 O U R M A X • FS Similarly, i f Powell kinetics are assumed, equation 6.7 results r. ^ oaseime *~ \ MAA 1 V MAA / MAA J J (6.7) In figure 6.2.2, the results of two different injections are plotted. Both sets of data were obtained at excess substrate concentration, and the OUR has been divided by 173 (OURbaseime+AOURMAx) for easier comparison (at high substrate concentrations, the teim in brackets in equations 6.6 and 6.7 is equal to (OURbaseiine+AOURMAx))- For one of the data sets, the OUR appeared to follow Powell kinetics with respect to the DO, while for the other set, the OUR followed Monod kinetics with respect to the DO. The main experimental difference between the two data sets is the M L V S S concentration. These results agree with those discussed in section 6.1, where it was found that the AOUR followed Monod kinetics at low M L V S S , and Powell kinetics at high M L V S S . It was hypothesised in section 6.1 that the mass transfer resistance was due to oxygen diffusion and not substrate diffusion (when methanol is the substrate). The coefficient L in the Powell equation is related to the mass transfer resistance. Typical values of L for methanol respirometric data were in the range 0.16 to 0.2 mg/1. Typical values of L for acetate respirometric data were in the range 0.12 to 0.2 mg/1. Values of L for the curves shown in figure 6.2.1 (the relationship between OUR and DO obtained using both methanol and acetate) range from 0.14 to 0.22. The similarity of these three sets of numbers implies that in all three cases the mass transfer resistances are similar. This implies that the main mass transfer resistance is related to the oxygen diffusion into the floe, and not the substrate diffusion. This situation may change when larger substrates are used, but the biomass in this study did not readily metabolise larger substrates (such as carbohydrates), and this was not investigated. The model constants for DO obtained from the curves in figure 6.2.2, along with the AOUR model constants for the substrate (methanol), were used to generate figures 6.2.3 and 6.2.4. For simplicity, OURbaseime was assumed to be 0. The standard method for measuring the OUR is to place a sample of mixed liquor into a BOD bottle, insert the 174 5-i 4.5-4-"5 3.5-C £ 3-CN 0 2.5-CO J . 2-ZD 1.5-o 1-0.5-o-0.6 0.8 Dissolved oxygen (mg/l) Figure 6.2.1 OUR vs DO measured with an excess of substrate. Methanol, MLVSS 2965 mg/1 (•); methanol, MLVSS 4440 mg/1 (•); acetate, MLVSS 3105 mg/1 (A); acetate, MLVSS 5380 mg/1 (A); acetate, MLVSS 1920 mg/1 (•); acetate, MLVSS 3560 mg/1 (•); acetate, MLVSS 5060 mg/1 (A). 0.9-0.8-"5 0.7-c ' £ 0.6-0 0.5-CO £ , 0.4-an ZD 0.3-o 0.2-0.1-o-1.5 2 2.5 Dissolved oxygen (mg/l) Figure 6.2.2 OUR vs DO measured with an excess of methanol. MLVSS 4440 mg/l (•), MLVSS 1505 mg/l(o). 175 DO probe and measure the decrease in DO for 15 minutes. Assuming the Powell model for both substrate and oxygen, and assuming that the substrate concentration remains constant, as the DO decreases, the OUR will remain approximately constant until the DO drops below 0.5 mg/l (figure 6.2.4). If the Monod model is assumed, and the substrate concentration is low, the OUR will also be approximately constant as the DO decreases. However, i f the substrate concentration is high, the OUR will decrease steadily as the DO drops from 6 to 2 mg/l, and drop even faster below 2 mg/l (figure 6.2.3). If the OUR is a function of the DO, it becomes difficult to measure. In order to verify this hypothesis, the OUR was measured-while feeding methanol continuously to the respirometer at low solids concentration (figure 6.2.5). Under these conditions, the methanol concentration should be approximately constant in the respirometer. The slope of the OUR vs. DO curve was calculated for each methanol concentration and is shown in figure 6.2.6. At low methanol concentrations, the OUR was approximately constant with DO, the slope was 0.01 / minute, compared to 0.07 at high methanol concentrations. As the catabolic rate increased, the dependence of OUR on the DO increased as predicted by figure 6.2.3. Further verification of the dependence ofthe OUR on DO was obtained from replicate AOUR experiments at different DO concentrations. For small injections, there was no noticeable trend in the AOUR as the DO decreased (figure 6.2.7). For larger injections, the AOUR decreased slightly as the DO decreased, as expected (figure 6.2.8). The OUR decreased during the injection, and then was slightly lower for the second injection (at a lowed initial DO) during which it also decreased. The only difference between the two injections was the DO concentration. A similar experiment, with a Figure 6.2.4 AOUR vs. DO and substrate based on Powell kinetics for both substrate and DO. 177 Time (minutes) Figure 6.2.5 DO vs. time under different continuous methanol feed rates. •0.07 T —i 1 r 5 10 15 20 Flow Rate (mg methanol COD/hour) hO.OS "5 ho.05 ~ a. o -0.04 P O a -0.03 ^ 0* r-0.02 O o haw §-OO 25 Figure 6.2.6 Based on the data in figure 6.2.5. OUR calculated at 2 minutes (•), OUR calculated at 5 minutes (•), Slope of OUR vs. DO plot (A). MLVSS 1370 mg/l. 178 0.9-0.8-D 0.7-C E 0.6-0 0.5-CD -J , 0.4-cm ZD 0.3-0 0.2-0.1-o-4.5 3.5 2.5 Dissolved oxygen (mg/l) Figure 6.2.7 OUR vs. DO for replicate methanol injections, average maximum OUR is graphed with 95% CI. MLVSS 1400 mg/l. 1.6 1.4-1.2-"5 c E 1-o 0.8-CD E 0.6-cm ZD 0 0.4-0.2 0 4.5 3.5 2.5 Dissolved oxygen (mg/l) Figure 6.2.8 OUR vs. DO for replicate methanol injections. MLVSS 1400 mg/l. 179 larger M L V S S concentration is shown in figure 6.2.9. There was no noticeable effect as the DO decreased. This is expected since the kinetics are now described by Powell's equation. A final example is presented in figure 6.2.10, A very large injection was added to the respirometer and when the DO dropped below 1 mg/l, the DO probe was removed, the mixed liquor aerated, and then the DO probe was replaced. This was repeated until the OUR dropped back to its pre-injection value. The OUR decreased with DO, and the pattern was the same each time the mixed liquor was aerated. This verifies that the decrease in OUR is due to the DO level and not the substrate concentration, as the substrate concentration would be continuously decreasing throughout the test. (Each line in figure 6.2.10 would be lower than the previous one if the decrease in OUR was due to substrate instead of DO). It appears from the data presented in this section that the dependence of activated sludge kinetics on the DO follows the Powell model. As the diffusional resistance becomes negligible compared to the reaction resistance, which happens at low solids concentrations, the Powell equation simplifies to the Monod equation. Under these conditions, the OUR becomes a function of the DO even at DO values greater than 2 mg/l, which contradicts most assumptions found in the literature. The dependence ofthe OUR on substrate and oxygen will follow the same form (low M L V S S - both can be modeled with Monod, high M L V S S - both can be modeled with Powell).)) 6.3 Multi-Substrate Wastewaters When dealing with wastewaters, the actual composition, sometimes even the major organic source, is often unknown. The biodegradable organic compounds are often treated as a single substrate. In order to apply respirometry to the determination ofthe 180 0 I I I I I | I I M | I I I I | I I I I | I I I I | I I I I | I 1 I I | I I I I | I I I I | I I I I | 6 5.5 5 4.5 4 3.5 3 2.5 2 1.5 1 Dissolved oxygen (mg/l) Figure 6.2.9 OUR vs. DO for replicate methanol injections. MLVSS 3750 mg/l. Dissolved oxygen (mg/l) Figure 6.2.10 OUR vs. DO for one large methanol injection, with re-aeration when the DO reached 1 mg/L After initial injection ( ), after 1st re-aeration ( ), after 2nd re-aeration ( ), after 3rd re-aeration ( ), and after 4th re-aeration ( ). MLVSS 2800 mg/l. 181 kinetics of B K M E biodegradation, it is necessary to know the effects that unknown substrate mixtures have on data interpretation. During the utilisation of multiple substrates, the metabolism of each substrate will exert an oxygen demand. Since oxygen is a global parameter, it is difficult to differentiate the oxygen uptake due to the individual compounds. It is also difficult to differentiate the amount of oxygen consumed to metabolise each of the individual compounds, and hence difficult to calculate the individual yields. These effects conspire to make the interpretation of respirometric data obtained using substrate mixtures difficult. The discussion in this section will focus on determining the SUR when all compounds of the mixture are present, i.e. at the start of the batch test. For any given mixture of substrates, it is likely that the time for complete removal of each substrate will be different, depending on the initial concentrations and the individual removal rates. The composition of the mixture will change as the batch test progresses. If the substrates are removed simultaneously, then the SUR will be high when all of the substrates are present. As substrates are removed from the mixture, the SUR wil l decrease as discussed in the section on n-order kinetics, section 3 .3 . For a two substrate mixture, initially the SUR will be equal to the sum of the SUR for the two compounds. When one substrate is completely removed, the SUR will be equal to the SUR of the remaining substrate. The best way to deal with this problem is to model the process as a mixture of independent substrates and calculate the model parameters and stoichiometry for each substrate. This will be addressed in chapter 7. In this section, the possible error in estimating the removal rates based upon respirometry and the assumption that the mixture behaves as a 182 4 3.5 3 2.5H cm 8 2 1.5H 1 0.5 0 OUR, OUR 1 ji'OC 2 | i i; OC 1 0 0^5 1 1.5 2 2.5 3 3.5 4 4.5 5 Time Figure 6.3.1 OUR vs time profile following an injection of two substrates which are used simultaneously. 183 . single substrate is examined. In either case, the possible substrate interactions must still be studied experimentally. The calculation of the SUR depends on the OUR and the yield. If the yields of the various substrates in a mixture are the same, the calculated yield (equation 4.6) will be the same as the actual yield, and the SUR calculated from the OUR will be correct. If the yields of the various compounds differ, then the yield will appear to change throughout the batch test as substrates are removed. The appropriate yield must be used to convert the OUR to a SUR at each point in the batch test. A further complication is that when more than one substrate is present, the calculated yield will be a weighted average of the yields of the individual substrates, and the calculated SURs may be in error. Consider the case demonstrated in figure 6.3.1. If the two compounds are used simultaneously, the actual SUR is: SURactual = SURt + SUR2 = ° U R l + ° U R l (6.8) a c , u a l 1 2 OQ/Sy OC2/S2 V Since the total OUR and the total oxygen consumption are measured, and not OURi, O U R 2 , OCi or O C 2 , the measured SUR will be (assuming that the individual OUR's and OC's are additive): S U R - 0UR^i OUBi+OUS? ( 6 9 ) — e d OCmalIS (oc{ + oc2)/(s]+s2) V . SURmeasured will only be equal to SURactuai if the yields (and thus OC/S) of the two compounds are equal. Figure 6.3.2 further demonstrates the error introduced by the average yield. We shall assume that there are two substrates, A and B. The yield on substrate A is 0.2 and the. yield on substrate B is 0.8. This is an extreme difference, but could be illustrative of 184 0.2 0.3 0.4 0.5 0.6 0.7 0.8 SI or( 1-S2) with SI + S2 = 1 Figure 6.3.2 Actual SUR (horizontal lines), measured Y (weighted average of actual yields, ), and SUR calculated using measured yield ( ) for different ratios of a two substrate mixture (Yl = 0.2, Y2 = 0.8, and OUR total = 1). 0.2 0.3 0.4 0.5 0.6 0.7 0.8 SI or( 1-S2) with SI + S2 = 1 Figure 6.3.3 Actual SUR (horizontal lines), measured Y (weighted average of actual yields, ), and SUR calculated using measured yield ( ) for different ratios of a two substrate mixture (Yl = 0.4, Y2 = 0.6, and OUR total = 1). 185 a methanol and acetic acid mixture. When both substrates are present in excess, the O U R (and S U R ) due to each compound will be at it's maximum, and the total O U R (and total S U R ) is not a function ofthe mixture composition. If simultaneous substrate metabolism is assumed, the total O U R will be equal to the sum of the O U R M A X ' S of the individual substrates. The same is true for the actual S U R . However, if the data is analysed assuming there is only one substrate present, then the calculated yield will depend on the wastewater composition, as shown in figure 6.3.2. This is due to the fact that the calculated yield is a weighted average ofthe individual substrate yields. Therefore, the calculated S U R (which is calculated using the calculated yield) wil l also vary as the composition of the mixture varies, whereas the actual S U R is independent ofthe mixture composition. Figure 6.3.2 shows that when there is more of substrate A (Y = 0.2), the S U R is likely to be underpredicted, and when there is more of substrate B (Y = 0.8), the S U R is likely to be overpredicted. The difference between the calculated S U R and the actual S U R also depends upon the value of the O U R M A X for the individual substrates. In the example shown in figure 6.3.2, the total O U R is 1. If the majority of this O U R is due to substrate B (Y=0.8), then the actual S U R will be 4.25. If the majority ofthe O U R is due to the metabolism of substrate A (Y=0.2), then the actual S U R will be 2. In contrast, the calculated S U R is independent ofthe O U R M A X for the individual substrates, in the example shown. The S U R calculated from assuming that the mixture is a single substrate is based on a constant O U R (=1) and a variable yield, and ranges from 1.5 to > 5. The example shown in figure 6.3.2 is an extreme case. When the yields ofthe individual substrates are closer together, the errors in calculating the S U R are not as great 186 (figure 6.3.3). If the individual yields are the same, there is no error in calculating the SUR. In this case, the calculated yield is independent of the mixture composition, and the actual SUR is independent of the fraction of the OUR due to the metabolism of the various substrates. The OUR may be converted to SUR at any point during the batch test by dividing by the oxidation coefficient. When dealing with the respirometric analysis of substrate mixtures, it appears to be necessary to determine the yields of the individual compounds independently. If the yields are different, then a method of determining the OUR due to each of the substrates is required. Otherwise, i f the mixture is analysed as i f it were a single substrate, the calculated SUR will be incorrect. In order to better understand what is occurring during respirometry and to test the assumptions made in this discussion, substrate mixtures typically found in B K M E were investigated. The major assumptions are simultaneous and additive substrate removal, OUR, and OC. It is possible that substrates may be removed sequentially, or that the removal of substrate mixtures may be simultaneous, but follow competition kinetics as proposed by Orhon (Orhon and Artan 1994?). This will further complicate data interpretation. Verification of the difficulties associated with substrates which have different yields on respirometric data interpretation, as predicted by figure 6.3.2 and 6.3.3, was also sought. The mixtures chosen were: methanol - acetic acid, methanol -formic acid, acetic acid - formic acid, and methanol - acetic acid - formic acid. The data obtained using these mixtures was compared to the data obtained using methanol, acetic acid, and formic acid on their own. 187 0 1 ' 1 1 1 I • 1 1 1 I 1 1 1 1 I 1 ' 1 1 I 1 1 ' 1 I ' 1 1 1 I 1 1 1 1 1 3 4 5 6 7 8 9 10 Time (minutes) Figure 6.3.4 OUR vs time for injections of methanol (2.6 mg COD/1) ( ), acetate (1.9 mg/l) ( ), and methanol (2.6 mg/l)/acetate (1.9 mg/l) mixture ( ). MLVSS: 1550 mg/l. 2.5 0—] r—i 1 1 1 1 1 1 1 1 — i i — i 1 — | 1 1 1 1 1 I—I r— i — | 1 1 f—i [ 2 2.5 3 3.5 4 4.5 5 Time (minutes) Figure 6.3.5 OUR vs time for injections of methanol (0.5 mg COD/1) ( ), acetate (0.4mg/l) ( ), methanol (0.5 mg/l)/acetate (0.4 mg/l) ( ), and expected curve for the methanol/acetate mixture assuming AOUR due to methanol and acetate are additive ( ). MLVSS: 1800 mg/l. 188 Methanol - Acetic Acid Mixtures Figure 6.3.4 shows an example of simultaneous substrate utilisation with methanol and acetic acid as the two substrates. The total AOUR is equal to the sum of the individual AOURs, and the area under the curve is equal to the sum of the area under the individual curves. The acetate is being metabolised at the same rate and with the same yield as when it is the sole substrate. The methanol is also being metabolised at the same rate and with the same yield as when it is the sole substrate. This data is at the maximum uptake rate of both methanol and acetate. If any deviation from simultaneous utilisation is expected, it is most likely to occur when the bacteria are metabolising the substrates at their maximum rates. These results imply that there are two separate biomass components for each of these substrates, as would be expected in biomass taken from a continuous low rate activated sludge unit receiving wastewater that contains both of these substrates. Over four years of activated sludge operation on B K M E , this was the response most commonly observed during the oxidation of methanol - acetic acid mixtures. Although methanol and acetate are removed simultaneously, the difference in the yields of the two compounds will cause difficulties during the interpretation of the respirometric data. For methanol, the AOUR is 0.7 mg/l minute, Y is 0.53, and the SUR. is 1.5 mg/l minute. For acetate, the AOUR is 0.1 mg/l minute, Y is 0.82, and the SUR is 0.56 mg/l minute. For the mixture, the AOUR is 0.8 mg/l minute (which is equal to AOURmethanoi + AOURaCetate)- The measured yield is 0.67 (which is the average of the yield on methanol and acetate). The actual SUR is equal to SURmethanoi + SURaCetate-189 1 The additive nature of the individual SURs is verified by examining the OUR profile. The OUR due to methanol oxidation has the same profile whether or not acetate is present.. If the methanol SUR were changing in the presence of other substrates, the elevated OUR due to methanol metabolism would either last longer than, or less than, the OUR profile obtained from the oxidation of methanol as the sole substrate. The fact that exactly the same amount of time is required for complete removal from the wastewater whether or not acetate is present implies that the rate of methanol utilisation is not affected by acetate. The same is true for the OUR profile due to acetate oxidation. The calculated SUR (2.4 mg/l minute) is different from the actual SUR (2 mg/l minute due to the calculated yield. If the yields of the two compounds are known beforehand, the SUR may be calculated from the OUR profile of the mixture, using the following procedure. The OUR profile is used to calculate the AOUR due to methanol oxidation and the AOUR due to acetate oxidation. These values are divided by the corresponding oxidation coefficients (obtained from the known yields), and added together to calculate the SUR. This method is only possible with certain ratios of substrates. If both substrates are present in quantities which result in their removal from the respirometer at the same time, it will be difficult to determine each substrate's contribution to the AOUR. In this case the SUR of the mixture must be calculated by measuring the SURs of the compounds individually, and then adding them together. The common response to methanol - acetate mixtures was simultaneous removal, with the OUR, OC, and SUR being additive. On at least one occasion a different response to methanol - acetate mixtures was observed. An example of this response is 190 shown in figure 6.3.5. The OUR due to the metabolism of the methanol - acetic acid mixture was very close to, but slightly lower than, the OUR due to the metabolism of methanol alone. The two substrates were still used simultaneously, as the time required to metabolise the mixture was the same as the time required to metabolise the individual substrates. The OUR was not additive, the OC was not additive, but the SUR was additive. The data for a number of methanol - acetic acid mixtures, as well as for methanol and acetic acid, are presented in figure 6.3.6. Regardless of the amount of acetic acid, the maximum AOUR on the methanol - acetate mixtures was approximately constant. As the amount of acetic acid increased, the half saturation constant appeared to increase. This was due to the increase in the COD of the solution from the added acetic acid. In figure 6.3.7 the AOURs are plotted versus methanol concentration; there is no noticeable difference between any of the curves. This implies that the OUR response was due solely to the methanol. The acetate was being utilised, but was not exerting an oxygen demand when methanol was present. This conclusion is also evident from the OUR profiles. In figure 6.3.5, there was just enough acetic acid to be utilised in the same amount of time as the added methanol, but the OUR profile was similar to the OUR profile of methanol alone. When an excess of acetic acid was added, figure 6.3.8, the acetic acid did not exert an OUR when the methanol was being oxidised. As soon as the methanol was depleted, and the OUR due to methanol oxidation dropped, the OUR then increased to the level it would be at if acetic acid were the sole substrate. The acetic acid was still removed in the same amount of time as if it were the sole substrate even though no oxygen demand was exerted for the metabolism of acetic acid when methanol was 191 2.5-„ 2H "5 c E — 1.5-CN o CD J - 1-at ZD o < 0.5-o o ° l o o o O AO • o o T ' : 1 i T i 1 1 1 1 1 1 r 4 6 8 10 Substrate (mg COD/I) 12 14 Figure 6.3.6 AOUR vs substrate for a number of methanol/acetate mixtures. Methanol (•), acetate (•), methanol/acetate : 1/2.8 mg COD/mg COD (A), 1/1.4 (o), 1/0.7 (o), 1/0.35 (•). MLVSS: 1800 mg/l. 2.5-2H 0) D C " E ^ 1.5-O CD at ZD o < H 0.5-A O Oo o o A 0 o E -i 1 1 1 r i i r -i 1 1 1 1 r 0 1 2 3 4 5 6 Methanol (mg COD/I) Figure 6.3.7 AOUR vs methanol for a number of methanol/acetate mixtures. Methanol (•), methanol/acetate : 1/2.8 mg COD/mg COD (A), 1/1.4 (o), 1/0.7 (o), 1/0.35 (•). MLVSS: 1800 mg/l. 192 present. It appears that regardless of how the cell processed the acetate (partially oxidised or not oxidised), the removal rate was constant. The dip in the OUR curve as the catabolism switched from methanol oxidation to acetate oxidation was repeatable (the experiment was repeated four times on two different days), but occurred only for a certain ratio of methanol to acetate, and only for certain strengths of injections. The yield of acetic acid metabolism must have been changing during the experiment, with the yield being approximately 1 when methanol was present (no acetate being oxidised), and decreasing to 0.76 when no methanol was present, with only a few seconds required for the bacteria to switch between the two modes. The oxygen consumption data for the various mixtures are shown in figure 6.3.9. As discussed previously, if the yields of the two substrates are constant, the measured yield will vary predictably with the mixture composition. For this set of experimental data, the experimental values for oxygen consumption were consistently lower than the predicted values based upon the yield of methanol and acetate. Consequently, the measured yield did not match the predicted values (green curve, figure 6.3.10) which were obtained assuming the actual yields were constant. With the assumption that the acetic acid yield is equal to 1 as long as methanol is present, and equal to 0.76 after the methanol has been completely oxidised, the predicted yield does match the measured yield (red curve, figure 6.3.10). The change in slope of the predicted yield corresponds to the point when the acetic acid and methanol are reduced to zero at the same time. When there was a greater amount of acetic acid added, the methanol was used up before all of the acetic acid was used up, and the OUR profile had a shoulder, which corresponded to the oxidation of the remaining acetate (figure 6.3.11). 193 1 1 1 1 1 I ! 1 1 1 I 1 1 1 1 I 1 1 1 1 1 1 1 1 1 I ! 1 I 1 I 1 1 1 1 I 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (minutes) Figure 6.3.8 OUR vs time for injections of methanol (0.5 mg COD/1) ( ), acetate (1.9 mg/l) ( ), acetate (0.75 mg/l) ( ), and a methanol (0.5 mg/l)/acetate (1.5 mg/l) mixture ( •)• MLVSS: 1800 mg/l. Substrate (mg COD/I) Figure 6.3.9 OC vs substrate for a number of methanol/acetate mixtures. Methanol (•), methanol/acetate : 1/2.8 mg COD/mg COD (A), 1/1.4 (o), 1/0.7 (•), 1/0.35 (•). Curves for the mixtures were calculated by adding together the OC of methanol and acetate. MLVSS. 1800 mg/l. 194 % Acetate 100 90 80 70 60 50 40 30 20 10 0 Q Q I I | > > I I I I I I I I I I I I ' I I I I ' I I I I I ' I I I I I I I I ' ' ' I ' I ' 1 I 1 1 1 1 I o . H 0 10 20 30 40 50 60 70 80 90 100 % Methanol Figure 6.3.10 Measured respirometric yield vs substrate (•), expected yield if the yield of methanol and acetate are constant ( ), expected yield if the yield of methanol is not constant ( ). MLVSS: 1800 mg/l. 3 -°-5 I 1 1 ' 1 I 1 ' ' 1 I 1 1 1 ! I 1 ' 1 1 1 1 1 ' 1 I ' 1 ' ' I 2 3 4 5 6 7 8 Time (minutes) Figure 6.3.11 OUR vs time for injections of methanol ( ), acetate ( ), methanol/acetate mixture 1/2.8 mg COD/mg COD (- ), 1/1.4 ( ), 1/0.7 ( ), 1/0.35 ( ). MLVSS: 1800 mg/l. 195 The AOUR may be converted to SUR, but as pointed out earlier, due to the difference in yields of the two compounds (Ym e thanoi = 0.2, Y a c e t i C acid = 0.76), i f the measured yields are used, the SUR calculated for the mixtures will be in error (figure 6.3.12). In this particular case, further difficulty is added by the non-constant nature of the yield. A figure similar to 6.3.2 may be constructed (figure 6.3.13). At higher proportions of methanol, the SUR is underpredicted, associated with the lack of OUR due to acetic acid.- At lower proportions of methanol, the SUR is overpredicted due to the differences in the yields of the 2 compounds. In this case, it is impossible to calculate the actual SUR of the mixture without measuring the SUR of the compounds individually. . The yield was not constant for methanol - acetate mixtures, the two compounds appeared to be used simultaneously when present as a mixture, and they appeared to be removed at approximately the same rate as they were when present as single substrates. Unlike the first example of the microbial response to methanol - acetate mixtures, in this case the methanol and acetate appeared to be metabolised by the same biomass component. Continuous culture studies have demonstrated that when bacteria are supplied with two carbon (or greater) compounds, such as acetate, and a one carbon compound, such as methanol or formic acid, they will use the two carbon compound for growth and the one carbon compound for energy. The results presented here show that when methanol is present as an energy source, the bacteria are capable of utilising all of the acetate present without oxidising any of it. There also appears to be a very slight increase in the efficiency of the utilisation of the methanol. Although the bacteria prefer to obtain their energy from the oxidation of the methanol, i f there is none present, they are very 196 O 6 8 10 Substrate (mg COD/I) 12 14 Figure 6.3.12 SUR vs substrate for a number of methanol/acetate mixtures. Methanol (•), acetate (•), methanol/acetate : 1/2.8 mg COD/mg COD (A), 1/1.4 (o), 1/0.7 (•), 1/0.35 (•). SUR curves for the mixtures were calculated by adding together the SUR of methanol and acetate (dashed^ir^es^MLVSS 1800 mg/l. 100 90 80 70 60 50 40 30 20 10 0 •j I < i i—i L _L L ~i—i—|—i—n 20 30 1—II I I—|— i — i— I— r 70 80 90 100 -i—n—i—|—r 40 50 60 % Methanol Figure 6.3.13 Measured SUR (A), OUR (•), and OC (•) vs substrate for a number of methanol/acetate mixtures. The actual SUR is represented by the dotted line. MLVSS: 1800 mg/l. 197 quick to oxidise a portion of the acetate for energy. Methanol provides excess energy because growth on methanol is carbon limited. Growth on acetate is energy limited, and some acetate must be oxidised for energy. If both substrates are present, the bacteria can maximise their yield by using the excess energy obtained from the methanol to metabolise the acetate. This has been found previously in pure culture studies grown on methanol - acetate mixtures. Formic acid - Acetic Acid Mixtures Formic acid and acetic acid were used simultaneously, similar to methanol -acetic acid mixtures. In figure 6.3.14, the OUR profiles of formic acid, acetic acid, and their mixture, are plotted. At high concentrations, the OUR of the mixture was greater than the OUR of either substrate alone, but less than would be expected i f the OURs were additive. At low concentrations, the OUR of the mixture was similar to the OUR due to the formic acid alone, similar to the second methanol - acetate example presented in the previous section. The AOUR for a range of substrate concentrations is shown in figure 6.3.15. At all concentrations, the AOUR was less than would be predicted by adding together the AOUR for the two individual substrates. Since the OUR profiles suggest that the SUR's are additive, the OC must not be additive. The yield ofthe mixture was greater than would be expected by assuming the yields of the individual compounds are constant. This is verified by the oxygen consumption data, figure 6.3.16, which demonstrates that oxygen consumption during the oxidation of the mixture was less than expected, hence the yield was greater than expected. The SUR calculated assuming that the mixture is a single substrate was 2.8 mg/l minute. This is slightly higher than the SUR calculated using the single substrate 198 1.2-2.5 T 3 ' I 1 1 3.5 4 4.5 5 Time (minutes) 5.5 6 6.5 Figure 6.3.14 OUR vs time for injections of formic acid (2.2 mg COD/1, 0.4 mg/l) ( , ), acetate (0.75 mg COD/1, 3.8 mg/l) ( , ), formic acid/acetate mixture (2.2 mg/l / 3.8 mg/l) ( ), (0.4 mg/l / 0.75 mg/l) (- ). MLVSS 1800 mg/l. 0 1 2 3 4 Substrate (mg COD/I) Figure 6.3.15 AOUR vs substrate for a formic acid/acetate mixture. Formic acid (A), acetate (•), formic acid/acetate (0.57/1 mg COD/mgCOD) (•). OUR curve for the mixture was calculated by adding together the OUR of formic acid and acetate (dashed line). MLVSS: 1800 mg/l. 199 O co E, U O « 1.5H 0 1 2 3 4 Substrate (mg COD/I) Figure 6.3.16 OC vs substrate for a formic acid/acetate mixture. Formic acid (A), acetate (•), formic acid/acetate (0.57/1 mg COD/ mgCOD) (•). OC curve for the mixture was calculated by adding together the OC of formic acid and acetate (dashed line). MLVSS: 1800 mg/l. 0 1 2 3 4 Substrate (mg COD/I) Figure 6.3.17 SUR vs substrate for a formic acid/acetate mixture. Formic acid (A), acetate (•), formic acid/acetate (0.57/1 mg COD/mg COD) (•). SUR curves for the mixture was calculated by adding together the SUR of formic acid and acetate (dashed line). MLVSS: 1800 mg/l. 200 respirometric data and the assumption of simultaneous substrate utilisation (2.6 mg/l minute, figure 6.3.17). A greater discrepancy between these values would be expected on the basis of the large difference between the yield of acetic acid and that of formic acid (figure 6.3.2). The large difference in yield is partially offset by the increase in the yield ofthe mixture. If a number of different mixtures had been tested, a relationship similar to the one shown in figures 6.3.10 and 6.3.13 would be expected. In the case of formic acid - acetic acid mixtures, it was not possible to tell which of the substrates was being less oxidised when both are present (compared to when it was the sole substrate), but it was probably the acetic acid. If the formic acid yield is considered to be a constant, then the acetic acid yield changed from 0.76 as a sole substrate to 0.9 in the presence of formic acid. This was not as large a change as with the methanol - acetic acid example, but still significant. On a separate occasion, with different biomass, three different mixtures of formic acid and acetic acid were studied. Sample data is shown in figure 6.3.18. For the sample data shown, the OUR of the mixture was close to the expected OUR, implying that the yields were constant and the OUR was additive, unlike the previous example of formic acid - acetate mixtures. The data also suggests that the SUR's of the individual substrates were not influenced by each other, i.e. the SUR was additive as with all of the other examples discussed. The AOUR data is plotted in figure 6.3.19. As with the previous example, the AOUR of the mixtures was less than expected by adding together the AOURs ofthe individual substrates, probably due to the yields changing slightly when the two substrates were both present. The calculated yields were slightly greater than expected: 201 1-8-1.6-"5 1.4-CN O 1-e, o.8-| -> 0.6-o 0.4-0.2-0 1 2 3 4 5 6 Time (minutes) Figure 6.3.18 OUR vs time for injections of formic acid (4.3 mg COD/1) ( ), acetate (3.8 mg COD/1) ( ), formic acid (4.3 mg COD/1) /acetate (3.8 mg COD/1) mixture ( ), expected curve ( ) assuming AOUR due to formic acid and acetate are additive. MLVSS 2800 mg/l. • A • • 6 8 10 Substrate (mg COD/I) 12 14 Figure 6.3.19 AOUR vs substrate for a number of formic acid/acetate mixtures. Formic acid (A), acetate (•), formic acid/acetate : 1/0.43 mg COD/mg COD (•), 1/0.9 (A), 1/1.73 (#). MLVSS: 2800 mg/l. 202 based on the oxidation ofthe compounds individually (figure 6.3.21). Once again, this is probably due to the increased efficiency of acetate utilisation in the presence of a one carbon energy source (formic acid). As with the previous examples, the SUR calculated by treating the mixture as a single substrate is incorrect (figure 6.3.20) due to the difference in yields of the two compounds. The actual SUR is obtained by adding the SURs of formic and acetic acid together. The relationship between the calculated yield, the calculated SUR, and the actual SUR is shown in figure 6.3.21. The data follows the trends predicted previously. The overprediction of the SUR is somewhat mitigated by the increased yield on the mixture. Methanol - Formic Acid Mixtures The calculation of SURs of methanol - formic acid mixtures should not be difficult since methanol and formic acid have similar yields. The yield on the mixture will be the same as the yields on the individual substrates. It also has been predicted that the yield on a mixture of two one carbon compounds should not be greater than predicted based upon the yield ofthe two compounds individually . Therefore, unlike the mixtures involving acetate, the yield on methanol and on formic acid should not change due to the presence of the other substrate. AOUR data is shown in figure 6.3.22, and oxygen consumption data in figure 6.3.23, for a typical response of activated sludge to methanol - formic acid mixtures. The two substrates were utilised simultaneously, and the AOUR and oxygen consumption data of the mixture was very close to what we would predict by studying the single substrates. Both the AOUR and OC data were additive. As expected, the calculated SUR is very close to the SUR obtained assuming the substrates were removed simultaneously (figure 203 8 10 Substrate (mg COD/I) 14 Figure 6.3.20 SUR vs substrate for a number of formic acid/acetate mixtures. Formic acid (A), acetate (•), formic acid/acetate : 1/0.43 mg COD/mg COD (•), 1/0.9 (A), 1/1.73 (•). SUR curves for the mixture were calculated by adding together the SUR of formic acid and acetate (dashed lines).. MLVSS: 2800 mg/l. % Acet ic ac id 100 | I i—i—r—r-i—i—i i | i—i—i—i—|—i—i—i—i—]—i—i—i i | i — i —nr 0 10 20 30 40 50 60 70 80 90 100 % Formic acid Figure 6.3.21 Measured SUR (A), measured yield (•), expected yield if yields are constant ( ), actual SUR ( ). MLVSS: 2800 mg/l. 204 R 8 T 1.6-s 1.4-"5 c 1.2-£ 1-0 CO ^E 0.8-ZD 0.6-0 < 0.4-0.2-I I I I I I I I 1.5 2 2.5 3 3.5 Substrate (mg COD/I) Figure 6.3.22 AOUR vs substrate for a formic acioVmethanol mixture. Formic acid (A), methanol (•), formic acid/methanol (0.8/1 mg COD/mg COD) (o). OUR curve for the mixture was calculated by adding together the OUR of formic acid and methanol (dashed line). MLVSS: 2100 mg/l. 2.5-2H O O) i 5 . U O r 0.5-O " 0 . 0 : 'Tn—n—rr—r 0.5 1.5 2 2.5 3 3.5 Substrate (mg COD/I) 4.5 Figure 6.3.23 OC vs substrate for a formic acid/methanol mixture. Formic acid (A), methanol (•), formic acid/methanol (0.8/1 mg COD/mg COD) (o). OC curve for the mixture was calculated by adding together the OC of formic acid and methanol (dashed line). MLVSS: 2100 mg/l. 205 6.3.24). In this case the mixture behaved as expected based on the analysis of the single substrates. If the mixture were treated as a single substrate, the error in calculating the SUR and the yield would be minimal. On one occasion, a different response of the activated sludge was observed to methanol - formic acid mixtures. This response was different from all ofthe other examples discussed in that the SUR's were not additive. Figures 6.3.25, 6.3.26, 6.3.27 show typical OUR profiles from this data set. The oxygen uptake rate of the methanol -formic acid mixture was no greater than the oxygen uptake rate of methanol alone. Unlike the methanol - acetic acid scenario, the amount of oxygen consumed was greater than the amount of oxygen consumed when there was just methanol present. This was especially noticeable at the higher substrate concentrations (figure 6.3.27). The OUR during the oxidation of the mixture remained elevated for longer than the OUR during the oxidation of either methanol or formic acid implying that the SUR was not additive. The substrates were removed at a slower rate when present as a mixture than would be predicted by the data obtained when they were the sole substrates. This data implies that at high substrate concentrations, the substrates were being used sequentially, first the methanol was oxidised, then the formic acid. AOUR data for a number of methanol - formic acid mixtures is presented in figure 6.3.28. The AOUR ofthe mixtures did not increase above that of the AOUR due to methanol. If the amount of formic acid present is ignored^ and the AOUR is plotted versus just the methanol concentration, the AOUR increased at low methanol concentrations when there was a large amount of formic acid present. This implies that: at low concentrations the methanol and formic acid were used at the same time, but 206 Substrate (mg COD/I) Figure 6.3.24 SUR vs substrate for a formic acid/methanol mixture. Formic acid (A), methanol (•), formic acid/methanol (0.8/1 mg COD/mg COD) (o). SUR curve for the mixture were calculated by adding together the SUR of formic acid and methanol (dashed line). MLVSS: 2100 mg/l. Time (minutes) Figure 6.3.25 OUR vs time for injections of formic acid (0.43 mg COD/1) ( ), methanol (0.53 mg COD/1) ( ), formic acid (0.43 mg COD/1) / methanol (0.53 mg COD/1) mixture ( ). MLVSS: 1800 mg/l. 207 0—|—i—i—i—i—|—!—i—n~l—|—i—i—i—i—|—i—i—i—i—|—I— 1—i— 1—|— 1— 1— 1— 1—|— 1— 1— 1— 1—I 2 2.5 3 3.5 4 4.5 5 5.5 Time (minutes) Figure 6.3.26 OUR vs time for injections of formic acid (0.86 mg COD/1) ( ), methanol (1. lmg COD/1) ( ), formic acid (0.86 mg COD/1) /methanol (1.1 mg COD/1) mixture ( ). MLVSS: 1800 mg/l. 3-i 1 0 —(—i—r-i—r-[—i—i—I i | —r-i—r—pn—r— i—i—| i i i—i | i — 1 1 I 1 1 1 ! I 1 1 1 1 I 2 2.5 3 3.5 4 4.5 5 5.5 6 Time (minutes) Figure 6.3.27 OUR vs time for injections of formic acid (2.2 mg COD/1) ( ), methanol (2.6 mg COD/1) ( ), formic acid (2.2 mg COD/1) /methanol (2.6 mg COD/1) mixture ( — ). MLVSS: 1800 mg/l. 208 methanol was used preferentially at high concentrations. The same enzyme system must be responsible for the oxidation of formic acid and methanol. This enzyme system will preferentially oxidise the methanol, and then the formic acid. When methanol is oxidised, it is first turned into formic acid, then carbon dioxide. The data can be modeled based upon this sequential oxidation, and the curves in figure 6.3.28 are obtained. They do not match the data perfectly, but are reasonably approximate. Although the substrates were not used simultaneously, the yields of the methanol - formic acid mixtures can be predicted accurately (figure 6.3.29), implying that the yield of methanol and formic acid were constant even in the presence of each other. This is the expected response of mixtures of two one carbon compounds (yield not changed by the presence of other one-carbon compounds). The calculated SUR ofthe mixture was very close to the SUR of methanol alone (dashed line, figure 6.3.29). If methanol and formic acid were assumed to be used simultaneously, the SUR would be overpredicted by approximately 25%. From just the respirometric data ofthe mixture it is not possible to determine i f the two compounds were being used simultaneously or sequentially. Respirometric data of the compounds on their own must be available for comparison. Evidence for sequential utilisation was an increase in K M but not O U R M A X for the mixture over methanol alone. More evidence was provided by the OUR profiles, where it was shown that the oxidation of the mixture took longer than the oxidation of either of the substrates on their own. 209 2 3 4 Substrate (mg COD/I) Figure 6.3.28 AOUR vs substrate for a number of formic acioVmethanol mixtures. Formic acid (o), methanol (•), formic acid/methanol: 1/5 mg COD/mg COD (•), 1/2.5 (•), 1/1.25 (A), 1/0.63 ( ). MLVSS: 1800 mg/l. 90 100 I 1 < 1 1 I 1 1 • • i 1 • ' ' i 40 50 60 70 % Methanol Figure 6.3.29 Measured yield (•), expected yield if yields are constant ( ). Measured SUR (•), actual SUR ( ). MLVSS: 1800 mg/l. 210 Methanol - Acetic Acid - Formic Acid Mixtures When three substrates are present, interpretation of respirometric data becomes even more complicated. Two substrates may be utilised simultaneously and the third sequentially, or vice versa. Also, the yields on the substrates may be different when they are in a mixture than when they are on their own. In the first example of a methanol - acetate - formic acid mixture (figures 6.3.30, 6.3.31, and 6.3.32), the yields on the various substrates did not change when they were present in a mixture. This is evident from figure 6.3.32 where the predicted oxygen consumption ofthe mixture matches the experimental oxygen consumption of the mixture. If the yield does not change, the AOUR can be expected to be additive. This was not the case (figure 6.3.31). The AOUR of the three substrates were additive at low substrate concentrations (below saturation of the enzymes), but not at higher concentrations. These results imply that two of the substrates are oxidised by the same biomass component, and as long as they are present below saturating conditions, they wil l both be oxidised. When the substrates are present in saturating quantities, the biomass will preferentially use one of the substrates first, and only when that substrate has been utilised (or falls below the critical K M value) will the remaining substrate be utilised. The metabolism of the third substrate is not affected by the presence of the other substrates. Figure 6.3.30 further demonstrates this point. Methanol and acetate, or methanol and formic acid are oxidised first (simultaneously), followed by the remaining substrate (acetate or formic acid). It is difficult to tell i f the formic acid or the acetate was oxidised first, as the A O U R M A X of the two substrates were similar. Without knowing which substrates are utilised first, the SURs cannot be calculated. 211 U I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 1 1 1 1 I 2 3 4 5 6 7 8 Time (minutes) Figure 6.3.30 OUR vs time for injections of formic acid (0.9 mg COD/1) ( ), methanol (2.6 mg COD/1) ( ), acetate (1.9 mg COD/1) ( ), formic acid (0.9 mg COD/1) / methanol(2.6 mg COD/1)/acetate (1.9 mg COD/l) mixture ( ). MLVSS: 1500 mg/l. 3 - T 2.5-Substrate (mg COD/I) Figure 6.3.31 AOUR vs substrate for a formic acid/methanol/acetate mixture (0.33 /1 / 0.75 mg COD/mg COD/mg COD). Formic acid (A), methanol (•) acetate (•), formic acid/methanol/acetate (o). OUR curve for the mixture was calculated by adding together the OUR of formic acid, methanol, and acetate (dashed line). MLVSS: 1500 mg/l. 212 0 1 2 3 4 5 6 7 8 Substrate (mg COD/I) Figure 6.3.34 OC vs substrate for a formic acid/methanol/acetate mixture (0.33 /1 / 0.75 mg COD/mg COD/mg COD). Formic acid (A), methanol (•) acetate (•), formic acid/methanol/acetate (o). OC curve for the mixture was calculated by adding together the OC of formic acid, methanol, and acetate (dashed line). MLVSS: 1800 mg/l. 2 1 3 3.5-3 -^ 2-5-O D) O -O 1 5 H H 0.5H T T i —i r 2 3 4 Substrate (mg COD/I) 6 Figure 6.3.32 OC vs substrate for a formic acid/methanol/acetate mixture (0.33 /1 / 0.75 mg COD/mg COD/mg COD). Formic acid (A), methanol (•) acetate (•), formic acid/methanol/acetate ( o ) . OC curve for the mixture was calculated by adding together the OC of formic acid, methanol, and acetate (dashed line). MLVSS: 1500 4mg/L •m rr o / o i i >—r 2 3 4 5 6 7 8 Substrate (mg COD/I) Figure 6.3.33 AOUR vs substrate for a formic acid/methanol/acetate mixture (0.4 /1 / 0.75 mg COD/mg COD/mg COD). Formic acid (A), methanol (•) acetate (•), formic acid/methanol/acetate (o). OUR curve for the mixture was calculated by adding together the OUR of formic acid, methanol, and acetate (dashed line). MLVSS: 1800 mg/l. 214 Another example of the utilisation of the three substrates is presented in figures 6.3.33 and 6.3.34. This experiment used the same biomass as a few of the examples discussed above. In particular, the yield on the acetate increased to 1 while methanol was present; and formic acid was used sequentially when methanol was present in saturating concentrations. Here, the yields of the individual substrates changed when they were present in a mixture (figure 6.3.34). The yield of the mixture was greater than expected by calculating from the yields of the individual compounds. This was due to the presence of acetate and two energy sources, methanol and formic acid. The bacteria were able to satisfy their energy requirements from the methanol and formic acid present, and utilised the acetate for growth. Less ofthe acetate was oxidised than i f acetate had been the sole substrate. This resulted in an increase in the yield. These results imply that this sample of activated sludge is carbon limited, and not energy limited. Examination ofthe AOURs showed that the AOURs of the individual compounds were additive at low r concentrations, but not when the concentration increased. There are two explanations for this. The first is the OUR is lower than expected due to the increased yield on acetate. The second is the formic acid was not oxidised when methanol was saturating. Summary If substrate mixtures are composed of substrates with different yields, then the respirometric yield will be a weighted average of the yields of the individual substrates. This makes estimation of the SUR using respirometry very difficult, i f not impossible, unless the composition ofthe mixture is known. In addition to the composition, the individual yields must be known, and the individual OUR's. For unknown mixtures, such as B K M E , these problems will present great difficulties. 215 Data have been presented in this section on the response of activated sludge to various substrate mixtures. The general findings are presented in table 6.3. Sometimes the response was straightforward simultaneous removal of the substrates in the mixture, and sometimes some of the substrates were utilised sequentially. In no case did the removal follow competition kinetics as proposed by various authors. Adding to the difficulty of analysing mixtures where the two compounds have different yields, was the fact that, on occasion, the yields of the substrates were found to be different when other substrates were present. This was the case when acetic acid was present. The easiest mixture to analyse was the methanol - formic acid mixture. The yields of the two compounds are similar, and on most occasions the OUR and OC were additive. Table 6.3 Respirometric Analysis of Substrate Mixtures Mixture O U R additive O C Additive S U R additive Methanol - acetic acid No No Yes Formic acid - acetic acid No No Yes Methanol - formic acid Yes Yes Yes Methanol - formic acid No Yes no 216 Chapter 7 BKME Kinetics 7.1 Operating Data Mixed liquor volatile suspended solids data for the two lab scale reactors are shown in figure 7.1.1. The vertical dashed lines represent changes in wastewater batches. The vertical solid lines represent different experimental runs. Each run was started with fresh activated sludge seed from Harmac Pacific's wastewater treatment plant. The SRT (figure 7.1.3) in each activated sludge unit was controlled by wasting directly from the aeration tank based on the M L V S S , the VSS in the treated wastewater leaving the clarifier (figure 7.1.2), and the HRT. Unit one was operated at an SRT as close to 5 days as possible, except during the treatment of batches L , M , and N when the SRT was 10 days. The SRT in unit two was maintained as close to 15 to 20 days as possible, except for during the treatment of batches L , M , and N when the SRT was 10 days and an . aerobic selector was employed. The average effluent VSS in reactor one was 32 mg VSS/1 and the SRT was not difficult to control. There was day to day variation in the effluent VSS, and also day to day variation in the M L V S S . The M L V S S never achieved a true steady state value even though the SRT was fairly constant for most of unit one's operation. The biomass concentration seemed to go through cycles where every two weeks it would suddenly drop, then rise again. It is probably close to impossible to achieve true steady state conditions in a biological reactor with mixed populations and mixed substrates, which changes from batch to batch. Predation from protozoa will affect the M L V S S levels, and the populations of the different bacterial species present will probably oscillate. The data 217 0 100 200 300 400 500 600 Time (days) Figure 7.1.1 MLVSS vs. time over the course of the project. Reactor 1 (•), reactor 2 (o). Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. Figure 7.1.2 Effluent VSS vs. time over the course of the project. Reactor 1 (•), reactor 2 (o). Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 218 Figure 7.1.3 SRT vs. time over the course of the project. Reactor 1 (•), reactor 2 (o ) . Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 219 in figure 7.1.1 suggests that a pseudo steady state was achieved with respect to M L V S S in unit one. The average effluent VSS concentration leaving reactor two was 44 mg/l. Operating at SRTs greater than 15 days often results in sludge with poor settling qualities. This value, combined with an HRT of twelve hours made it difficult to control the SRT as the amount of mixed liquor to waste every day was very small and the fluctuations in the effluent VSS had a large impact on the SRT. One of the causes for the large variation in effluent VSS was the clarifier design. The underflow from the clarifier would occasionally become clogged, allowing the solids to accumulate in the clarifier and overflow with the effluent. Also, the biomass would stick to the clarifier walls, requiring occasional stirring, which then resulted in some biomass overflowing with the effluent before it had a chance to settle. The biomass from unit two usually settled much faster than the biomass from unit one, but stuck to the walls more frequently (either due to a greater amount of polymers present under starvation conditions or to the greater amount of solids present at the higher SRT). Consequently, more biomass from unit two was lost with the treated effluent due to stirring than from unit one. The high levels of effluent VSS made the SRT difficult to control. Figures 7.1.1 and 7.1.3 show that the M L V S S and SRT were highly variable for unit two (except when the SRT was controlled at 10 days). Steady state (with respect to M L V S S concentration) was never attained, but the SRT was significantly higher than the SRT from unit one at all times (except when they were both at 10 days). The BOD removal, COD removal, and toxicity removal (table 7.1) were the same for both reactors, and did not vary with time. Both units were operated at conditions in 220 excess of those required for acceptable B K M E treatment. Wastewater treatment plants are designed to operate at conditions that give good sludge settling characteristics, these operating conditions help ensure that all of the readily biodegradable organics are removed. The readily biodegradable organics are the source of BOD and acute toxicity in B K M E . Table 7.1 Reactor performance Reactor #1 Reactor #2 BOD Removal 95% 95% COD Removal 50% 55% Toxicity Removal 100% 100% 7.2 Measurement of BKME Kinetics In this section, various methods for measuring the activated sludge kinetics of B K M E treatment are investigated. In order to properly model the kinetics, the multiple substrate nature of B K M E must be accounted for. The implications of this finding will be discussed. If the multiple substrate nature is ignored, extrapolation of batch test data to continuous systems wil l be incorrect. Discrepancies between kinetic tests The kinetics of B K M E biodegradation by activated sludge may bemeasured using the respirometric method discussed in chapter 6 (figure 7.2.1). Most studies measuring B K M E biodegradation kinetics have ignored the multi-component nature of B K M E , and used the BOD or the COD to represent the degradable organics, which was done in this section as well. (To see i f this simple approach can be used to adequately model B K M E biodegradation kinetics). Various methods of calculating activated sludge 221 Total COD (mg/l) 0 10 20 30 40 50 0 2 4 6 8 10 12 Total BOD (mg/l) Figure 7.2.1 AOUR (•) vs. BOD and COD. OC (o) vs. BOD and COD. Batch L, SRT 5 days. 0 50 100 150 200 250 300 350 400 Time (minutes) Figure 7.2.2 OUR (•), BOD (o), and COD (A) VS. time during a BKME batch biodegradation test. Batch L, SRT 5 days. 222 kinetics were compared. These methods were the AOUR test, the batch test, and the infinite dilution fed batch test. The AOUR and oxygen consumption data in figure 7.2.1 is plotted versus the BOD and the COD. The BOD is probably a more accurate description ofthe biodegradable organics than the COD due to the large amounts of non-biodegradable components in the wastewater. The AOUR data follows Monod kinetics, and not Powell kinetics as found for methanol and acetate. This may be due to the multiple substrates present in low concentrations, as well as the difficulty in measuring the kinetics at large concentrations of effluent (due to biomass washout from the respirometer). The oxygen consumption data increases linearly with BOD concentration. Using the substrate oxidation coefficient from figure 7.2.1, the SUR may be calculated from the AOUR data (figure 7.2.3). The kinetic coefficients obtained depend on whether BOD or COD is taken as the substrate concentration. The A O U R M A X was similar to the values obtained on methanol and formic acid (see chapter 6, and section 7.3). This is as expected since these compounds are a major source of BOD in B K M E . The half saturation constant for the metabolism of effluent was much greater than the corresponding values for methanol and formic acid. Another method of measuring activated sludge kinetics is the batch test (figure 7.2.2). The BOD removal and COD removal were found to follow multicomponent kinetics (equation 3.22). There were three distinct regions in the OUR profile during the batch test. Each region corresponds to the oxidation of a different group of substrates, which will be discussed later. The kinetics of a continuous system may be calculated with batch test data using the method described by Argaman (1991) (figure 7.2.3). This 223 0.004 0.0035H Total BOD (mg/l) Figure 7.2.3 Kinetics calculated with: AOUR data (single substrate - Monod) (*), batch test data (n-order kinetics) (*), and fed-batch test (*). 3 1.5 2 2.5 3 3.5 4 4.5 5 5.5 Time (minutes) Figure 7.2.4 OUR profiles for injections of samples withdrawn during the batch test (figure 7.2.2) Samples from top to bottom (sample time in brackets): sample #1 (0 minutes), 2 (10), 3 (20), 6 (50), 8 (75), 9 (90), 11 (120). Batch L, SRT 5 days. 224 method assumes that each ofthe individual compounds is removed following zero order kinetics. This assumption is validated by the small half saturation constants for the typical compounds in B K M E found in Chapter 6. The calculated rates were slower than those calculated by the AOUR method, which relies on measuring the reaction rate at the start of many separate batch tests. The maximum SUR during the batch test (figure 7.2.2), which is the rate at the start ofthe test, was 0.0025 mg BOD/mg MLVSSnninute. This corresponds to the value obtained by respirometry (top curve, figure 7.2.3). If the Monod model is applied to the batch test data, and the maximum SUR is assumed to be 0.0025 mg BOD/mg M L V S S minute, the half saturation constant (the concentration at which the SUR =1/2 S U R M A X ) is approximately 120 mg/l. These coefficients result in a curve which does not fit the data. If the Monod model is fit to the batch test data (figure 7.2.2) using non-linear regression, ridiculously large, and meaningless, coefficients are obtained. It can be concluded that the Monod equation does not adequately describe the experimental batch test data. The data is best represented with the multicomponent model. A third way of measuring B K M E kinetics is the fed batch / infinite dilution method (middle curve, figure 7.2.3). The BOD measurement was used for this experiment since the large amount of non biodegradable COD makes COD measurements impractical. To compare all three methods, the predicted SUR vs. BOD for the AOUR method and batch test (modeled with the multicomponent model) are also plotted in figure 7.2.3. Under the flow rates used in the experiment, the AOUR results predict BOD values in the fed batch test to be less than 1 mg/l, when they went as high as 25 mg/l. The AOUR method greatly overpredicted the total substrate removal rate. On 225 the other hand, the multicomponent model underestimated the removal rate, and predicted treated BOD values approximately twice the actual values. Neither method (AOUR or batch test) accurately predicted the treated BOD values measured during the fed-batch test. The discrepancy between the three kinetic assays is likely due to the multicomponent nature of the wastewater. The BOD removal rates observed during the fed batch test may be modeled using the Monod model. The maximum substrate uptake rate was 1 mg/l minute. This is much lower than the maximum SUR predicted by the AOUR method (3 mg BOD/1 minute), and as found in the batch test method (0.5 mg BOD/1 minute). Had the flow rates been higher, the data would probably not have followed the Monod model, and the SURs measured would have been higher. Biomass washout prevented higher flow rates from being tested. The half saturation constant calculated based on the fed-batch data was 15 mg/l. This is larger than the half saturation constant obtained by the AOUR method, and much larger than the half saturation constants of the pure compounds tested in chapter 6. Although these values of K M are much larger than the values obtained for single substrates, they are much smaller than most values reported in the literature for wastewater treatment, and for the value obtained by fitting the Monod equation to the batch test data. This high value of K M is another indication of the multicomponent nature of the wastewater (Jones 1973). Others have noticed differences between the substrate concentration in the effluent and the values predicted by respirometry, and have attributed the discrepancy to soluble microbial product formation. SMPs would not be expected to form in such large quantities as indicated by the measured BOD, and are not usually measurable as BOD. 226 Fed-batch tests using methanol, acetate, and formic acid as the substrate, instead of BKME, resulted in negligible BOD in the reactor, thus no SMPs were formed from these compounds. The difference between the fed-batch and AOUR methods is also too great to be explained by the phenomena discussed in section 6.3. For example, if sequential substrate utilisation occurred in one of the assays and not the other, the maximum rate where it occurred would be lower, but not sufficiently low to account for the observed differences. Also, sequential utilisation would be more likely under the conditions in the respirometric test due to the higher substrate concentrations employed. The difficulty that the presence of compounds with different yields presents in converting the A O U R M A X into a S U R M A X is also unlikely to be great enough to account for the difference in the two assays. (This is supported by the similarity between the S U R M A X obtained from the respirometric test and from the batch test.) The yield obtained by the AOUR method was 0.6 mg COD / mg COD if based on the effluent BOD, and 0.3 mg COD / mg COD if based on the effluent COD. These values are greater than the yield on methanol or formic acid (0.2 to 0.5 mg COD / mg COD). This means that effluent organic compounds with a greater yield than methanol and formic acid are also being oxidised. One possible explanation is the presence of a large quantity of substrates such as acetate with very high yields. As was explained in section 6.3, the yield as measured by respirometry will be an average of the yields ofthe substrates present. Another explanation, which is certainly valid for the yield calculated based on the effluent COD, is that only a fraction of the organics are being oxidised during the AOUR tests, with the other fraction degrading at a slower rate or not at all. 227 Evidence of Multiple Substrates The discrepancy between the three methods of measuring the kinetics lies in the multicomponent nature of the wastewater. This is best demonstrated with the batch test data (figure 7.2.2). The BOD and COD removal during the batch test followed the multicomponent model as predicted by Grau et al (1975) for multicomponent wastewaters, with n = 1.4. If the different organic components in the wastewater all had the same removal rates, the BOD (or COD) removal would follow the Monod model. If the organics could be divided into two fractions with distinct removal rates, the total biodegradable organics, measured by COD, and the rapidly biodegradable organics, measured by BOD, then the BOD removal would follow the Monod model. Since neither the BOD nor the COD followed the Monod model, the situation is more complicated and there are more than two wastewater components (or the division between components is not the same as the division between BOD and COD). The OUR profile over the batch test (figure 7.2.2) also suggests the presence of multiple substrates. The OUR starts off high, drops to a plateau after the first 20 minutes, then drops to another plateau after 120 minutes. The standard interpretation of this type of OUR profile is that the initial high OUR is due to the readily biodegradable substrate. When the readily biodegradale substrate is gone, hydrolysis of the slowly biodegradable substrate becomes the rate limiting step. The second OUR plateau is due to the oxidation of the hydrolysis products (the hydrolysis products are metabolised at the same rate as the initial substrate, but the hydrolysis step is too slow for the OUR to be at its maximum). Another interpretation is the adsorption of substrate by the floe. The initial OUR plateau is due to the oxidation of the adsorbed substrate. When all of the adsorbed substrate has 228 been utilised, the OUR drops to the second plateau while the substrate in the bulk solutionis oxidised. Of course, i f the remaining substrate has different adsorption characteristics, it will probably also have different kinetics as well. A simpler explanation is suggested by comparison to figure 6.3.30, the OUR profile ofthe degradation of a three substrate mixture. The first OUR plateau is due to the biodegradation ofthe first readily biodegradable substrate, the second OUR plateau corresponds to the second substrate, and so on. There are many different biodegradable i components in B K M E , each with its own removal rate and stoichiometry. To determine i f the OUR and BOD removal behaviour during the batch test is due to the multicomponent nature of the wastewater, samples were withdrawn from the batch test at specified times, and the substrate removal rates of these samples were determined by the respirometric method. The OUR profiles of different batch test samples are shown in figure 7.2.4. The OUR profiles of samples one and two were quite different from samples three to nine, which were different from sample eleven. Note that the BOD of sample eleven is 61.8 mg/l, not much less than sample nine (86.6 mg/l). The AOUR vs. substrate curve for each sample of partially degraded wastewater are presented in figure 7.2.5, and the oxygen consumption data is presented in figure 7.2.6. For each sample the Monod model fit the data. As the wastewater was degraded, the removal rates ofthe samples decreased (the maximum SUR decreased, and the half saturation constant increased), implying that the composition ofthe wastewater changed throughout the batch test. If the composition did not change throughout the batch test, all ofthe curves in figure 7.2.5 and 7.2.6 should be superimposed, which is clearly not the case. 229 2.5 0 2 4 6 8 10 12 Total BOD (mg/l) Figure 7.2.5 AOUR vs. BOD curves of samples withdrawn during the batch test (figure 7.2.2) Batch L, SRT 5 days. 0 2 4 6 8 10 12 BOD (mg/l) Figure 7.2.6 Respirometric OC of samples withdrawn during the batch test (figure 7.2.2). Sample # 1(B), Sample 2 (•), Sample 3 (A), Sample 4 (•), Sample 6(D), Sample (o). Batch L, SRT 5 days. 230 The respirometric data of the complete wastewater and all of the samples follow the Monod model (figure 7.2.5) even though the batch kinetics are one and a half order (figure 7.2.2) because the respirometric method relies on the AOUR at the start of many short batch tests. The initial AOUR is due to the summation ofthe metabolism of all the components present in the particular sample being tested. The mixture of substrates present is always in the same ratio at the start of the respirometric test for a given sample, so it appears to behave as a single compound. As the batch test progresses, substrates are being removed from the wastewater. The first major drop in OUR corresponds to when the first major substrate component has been removed from the wastewater. Samples withdrawn after this time have lower removal rates than the original B K M E since they are missing the more readily biodegradable substrates. The second major drop in OUR corresponds to when the second major substrate component has been removed from the wastewater. Samples withdrawn after this time have even lower removal rates than the original B K M E since they are missing even more biodegradable substrates. These results are similar to those reported in continuous systems with tanks in series. In continuous systems, where the substrate concentration is usually low, substrate removal rates are often modeled as following first order kinetics. The first order substrate uptake rate coefficient was found to decrease from the first to the second tank: due to the more readily biodegradable substrates being removed in the first tank, leaving a more recalcitrant wastewater to be treated in the second tank (Huang et al 1985). The differences in the curves in figure 7.2.6 implies that the different components of B K M E will have different yields. This is expected from the results in chapter 6. 231 Wastewater Fractions The amount of readily biodegradable substrate in wastewater is usually determined by measuring the OUR during a batch test, and measuring the area under the OUR profile, as discussed in Chapter 4. The area under the curve is converted into readily biodegradable substrate using the yield. Usually, a yield of - 0.66 is assumed. The assumption of a constant yield is probably incorrect for BKME. The dependence of the yield on wastewater composition was demonstrated in chapter 6. In this chapter it was shown that the wastewater composition changes throughout the batch test, therefore the yield is probably changing throughout the batch test. Initially, the respirometric yield will be a weighted average of the yields of all ofthe substrates present (probably with an increased yield on the two-carbon and greater compounds due to the presence of one-carbon compounds). As the readily biodegradable substrates are removed one by one, the yield will change to be more representative of the yield ofthe slowly biodegradable substrates. A variable yield makes it difficult to convert the OUR into a SUR, and also to convert the area under the OUR curve into a substrate concentration. A more accurate determination of the readily biodegradable component is made from a direct measurement ofthe substrate. Straight lines may be drawn through the BOD values while the OUR is constant, as through points 3 to 10 (line 2, figure 7.2.7). r The first substrate group, represented by the first two BOD points is either made up of a single substrate with a high K M , or more likely a mixture of various substrates (lines la and lb). By point 3, this substrate group is gone, and the bacteria are working on the second substrate group, which is probably methanol. After the methanol has been all used up, acetate is probably the main substrate left (line 3a). When this is gone, the 232 250 I 0 50 100 150 200 Time (minutes) 400 Figure 7.2.7 BOD (o) and OUR ( ) vs time during a batch biodegradation of BKME. Lines drawn represent removal of different wastewater fractions. Batch L, SRT 5 days. Readily Figure 7.2.8 Composition of BKME. Outer ring represents BOD, inner circle represents COD. Batch L, SRT 5 days. 233 remaining substrate is a mixture of various compounds with relatively low biodegradation rates (line 3b). If all of the substrates are assumed to be utilised simultaneously, then these lines may be extended to time zero and the initial concentration of each substrate group can be determined. This result is presented in figure 7.2.8 for both BOD and COD fractions. For calculation of the COD fractions, the non-biodegradable COD was assumed to be the amount leaving the continuous reactor at a twelve hour hydraulic retention time. 1 When estimates are available for the wastewater fractions, the AOUR kinetics of each sample may be graphed versus the BOD of the corresponding fraction (figure 7.2.9). For example, the AOURs obtained for sample 1 are graphed versus the first readily biodegradable substrate (la), while the AOURs from samples 3 to 10 are graphed versus substrate 2. The results clearly show three distinct AOUR vs. substrate curves which correspond to the different OUR segments ofthe original batch test (figure 7.2.2). The Monod constants for each sample were calculated and are graphed versus the time the corresponding sample was taken during the batch test (figure 7.2.10). This figure confirms the earlier observation that the removal rates were decreasing throughout the batch test. The main reason for this decrease is the disappearance of substrates as the test continues (i.e. the sample taken at 50 minutes (#5) has fewer substrates than the original wastewater (sample #1)). The AOURs of sample 1 (original wastewater) is the summation of the AOUR of all of the substrates present. The AOUR due to substrate l a may be estimated by subtracting the AOUR due to the rest of the substrates. This procedure starts with the AOUR of substrate 3 a. Substrate group 3b does not exert a noticeable AOUR during the 234 0 0.5 1 1.5 2 2.5 3 3.5 Substrate (mg BOD/I) Figure 7.2.9 AOUR of samples withdrawn during the batch test (figure 7.2.2) vs. the corresponding BOD fraction. Sample # 1(B), Sample 2 (•), Sample 3 (A), Sample 4 (•), Sample 6 (•), Sample 8 (o), Sample 11 (&). Batch L , SRT 5 days. 0—| 1 1 1 1—I 1 1 [ — I 1—^ 1 I I 1 1 1 1 1 1 1 1 ! 1 0 20 40 60 80 100 120 Time (minutes) Figure 7.2.10 Maximum AOUR (•) and Monod half saturation constant (A) of samples withdrawn during the batch test, calculated from the data in figure 7.2.9. Batch L , SRT 5 days. 235 respirometric test. The AOUR due to substrate 2 is calculated by subtracting the AOUR due to substrate 3a from the AOUR measured on samples 3 to 10. This procedure is repeated for substrate lb, then la. In figure 7.2.11 the estimated AOURs of the substrate groups are presented. Group 2 has the greatest removal rate, followed by group 1, and then group 3. The substrate group with the greatest removal rate is not removed first during the batch test because it is present in a large quantity, while there is relatively little of substrate group 1. This observation goes against the common assumption that the compound with the greatest removal rate always disappears first during a batch test. Similarly, the oxygen consumption for each substrate group-may be calculated (figure 7.2.12). Unfortunately, the oxygen consumption due to substrate group #3 could not be measured. The oxygen consumption for groups 1 and 2 are similar. The corresponding yield is 0.44 mg COD/mg COD This is typical for methanol and formic acid, which are thought to be two of the main constituents of substrate groups 1 and 2. Group 3 is composed of acetate and other 2+ carbon compounds, and the yield will be higher than the yield on the one carbon compounds (chapter 6). A yield of 0.6 was assumed for group 3. This is supported by the yield from the continuous activated sludge units, which is approximately 0.5 (see section 7.5). The yield of the readily biodegradable component is less than the overall yield implying that the yield of the slowly biodegradable component (group 3) must be larger. It would seem that the bacteria preferentially grow on the slowly biodegradable fraction and use the readily biodegradable fraction as an energy source. This has been found in other cases when both one carbon and two carbon compounds are available as substrates. The one carbon compound is used for energy, and the two carbon compound is used for growth. In the 236 1.6 Substrate (mg BOD/I) Figure 7.2.11 Fractional AOUR of samples withdrawn during the batch test (figure 7.2.2) vs. the corresponding BOD fraction. Sample #1 (•), Sample 2 (•), Sample 3 (A), Sample 4 (•), Sample 6 (•), Sample 8 (o), Sample 11 (A). Batch L, SRT 5 days. Substrate (mg BOD/I) Figure 7.2.12 Fractional OC of samples withdrawn during the batch test (figure 7.2.2) vs. the corresponding BOD fraction. Sample #1 (•), Sample 2 (•), Sample 3 (A), Sample 4 (•), Sample 6(a), Sample 8 (o). Batch L, SRT 5 days. 237 case of activated sludge treating BKME, the readily biodegradable fraction is assumed to be composed of one carbon compounds, mainly methanol and formic acid. The SURs of the various components of BKME may now be calculated (figure 7.2.13). The AOURs are converted to SURs using the yield factor. Since the AOUR of substrate group 3b could not be measured, an alternate method of calculation is required. Based on the SURs calculated for groups 1 and 2, the amount of BOD due to these groups measured during the fed batch test should be negligible. The removal rate of these groups of compounds is great enough for complete removal at the relatively low loadings attained in the fed-batch test. By process of elimination, the substrate being measured during the fed-batch test must belong to group 3, in particular group 3b. If the BOD in the fed-batch test is assumed to be from group 3b, the removal rate of group 3b may be calculated (the feed rate of group 3b is known from figure 7.2.8 and the flow rate) and is presented in figure 7.2.13. This fraction has a large K M value and is probably composed of many miscellaneous substrates. The Monod constants calculated from the batch test, and from the respirometric tests on the batch test samples are summarised in table 7.2. Table 7.2 Summary of kinetic constants of the various substrate groups in BKME Overall substrate Group 1 Group 2 Group 3 a Group 3b B O D 235 46 97 46 46 A O U R M A X batch test 0.0014 0.00067 0.00053 0.0013 . -A O U R M A X respirometry 0.0013 0.00039 0.00094 0.00013 -Y batch test 0.66 0.86 0.6 0.56 -Y respirometry 0.66 0.44 0.44 0.6# -S U R M A X batch test 0.0038 0.0030 .0054 0.00015 -S U R M A X respirometry ( = A O U R / ( l - Y ) ) 0.0038 0.00078 0.002 0.00022 0.00013* # assumed. * this value was calculated from the fed-batch test. 238 0.002 -JJ- 0.001 B-\ I 0.0016H £ <£ 0.0014 > § 0.0012 co E 0.001 H § 0.0008 C Q 0.0006 o£ 0.0004 ZD m 0.0002 10 15 Substrate (mg BOD/I) 25 Figure 7.2.13 SUR of wastewater fractions vs. the corresponding BOD. Group la (•), Group lb (•), Group 3a (A), Group 3b (•), Group 2 (•). Batch L, SRT 5 days. Figure 7.2.14 BOD vs. activated sludge loading for all of the wastewater fractions. Total BOD ( ), Group la ( ), Group lb ( ), Group 3a ( ), Group 3b ( ), Group 2 ( ), based on OUR ( ), measured BOD values (O) 239 The overall kinetics calculated from the batch test data agree with the overall kinetics calculated from respirometry. The kinetics of substrate groups 1 and 2 do not agree. Group one was removed at a greater rate than group 2 during the batch test. Opposite results were obtained by the respirometric method on the partially degraded samples from the batch test. There are two possible explanations for this observation. The first is that the removal rates were actually different between the two tests as they were done on separate days and the removal rates were highly variable (see section 7.3). The second explanation is that at the very high substrate concentrations at the start of the batch test the bacteria behave quite differently than they do at the lower substrate concentrations common in the respirometric method. It is possible that there is a relatively small quantity of substrate group 1 components present and high enough concentrations to saturate the enzyme systems are not attained in the respirometer. The very high yield calculated for group 1 during the batch test is puzzling and implies that not all of the substrate taken up at the start of the test is metabolised, but that some is stored. Discussion of Significance of Findings Using the data from figures 7.2.13 (SUR kinetics) and 7.2.8 ( B K M E composition), the BODs of the treated effluent and the SURs of the various fractions during the fed-batch test may be calculated (figures 7.2.14, 7.2.15, and 7.2.16). At the loadings employed during the fed-batch test, all of the BOD is due to the substrate from group 3 (figure 7.2.14). This will be the case until loadings greater than 4 mg BOD/mg M L V S S day are reached, when incomplete removal of substrates from groups 1 and 2 will occur. This loading is approximately 10 times the typical loading of the continuous 240 100 150 BOD (mg/l) 250 Figure 7.2.15 SUR vs. BOD for all ofthe wastewater fractions. Total SUR ( ), Group la ( ), Group lb ( ), Group 3a ( ), Group 3b ( ), Group 2 ( ), based on OUR ( ), measured values (O). Figure 7.2.16 SUR vs. loading for all of the wastewater fractions. Total SUR ( ), Group la (——), Group lb ( ), Group 3a ( ), Group 3b ( ), Group 2 ( ), based on OUR ( ), measured values (O). 241 lab scale activated sludge unit. In figures 7.2.15 and 7.2.16 the corresponding SUR's are plotted. The major contribution to the SUR comes from group 2. This is due to the fact that group 2 is the largest contributor to the overall effluent BOD, and has the fastest degradation kinetics. If the organic compounds in B K M E are assumed to be all the same substrate, and the removal rate is obtained from respirometry on the whole effluent (figure 7.2.1), the dashed line in figure 7.2.14 is obtained. No increase in treated effluent BOD is predicted until the loading reaches 4 mg BOD/mg M L V S S day. The BOD in the treated effluent wil l be underpredicted until very high loadings are reached, and what is coming out ofthe treatment plant is basically what is going in. Analogously, the SUR wil l be overpredicted due to the overprediction ofthe yield {SUR = OUR / (1 - Y)} (figures 7.2.15 and 7.2.16). The yield is overpredicted during respirometry because substrate group 3 exerts very little oxygen demand even though it comprises a fair portion of the BOD (Y = 1 - OC / S). If the BOD is treated as one compound, the BOD of substrate 3 is used in calculating the yield, and the predicted reaction rates are faster than the actual rates. This explains why the BOD in the treated effluent is underpredicted in figure 7.2.14 i f the substrates in the effluent are assumed to have the same removal rates and stoichiometry. The predicted SUR in figure 7.2.15 (dashed line) follows the Monod form since the calculations were based on a single substrate. However, for the other substrate groups, the SUR vs. total BOD in the treated effluent does not follow the Monod form. If the SURs were plotted against their corresponding substrates, the kinetics would follow the Monod form. The discrepancy is due to the changing nature ofthe BOD in the treated effluent. As the loading increases, substrate groups 1 and 2 are efficiently 242 removed, and the concentration of substrate group 3 increases. When the loading increases beyond 4 mg BOD/mg M L V S S day, the concentrations of substrate group 1 and 2 start to increase in the treated effluent. So, initially the BOD is predominately composed of substrate group 3, then ever increasing proportions of groups 1 and 2 are added. For the same reasons the total SUR vs. BOD also does not follow the Monod model. The half saturation constant calculated using the modified infinite dilution test would be approximately 30 mg BOD/1, which is significantly greater than the half saturation constants ofthe different compounds which make up the BOD. Noting that since the removal rates of substrate groups 1 and 2 are similar, and greatly different from the removal rate of group 3, the kinetic analysis may be simplified by assuming that the wastewater is composed of just two fractions, the readily biodegradable fraction, groups 1 and 2, and the slowly biodegradable fraction, group 3. The readily biodegradable substrates may be considered those which are completely removed in the continuous activated sludge unit under normal loading conditions. Similarly, the slowly biodegradable ones are the substrates partially removed under the same conditions. There is approximately an order of magnitude difference between the removal rates ofthe readily and slowly biodegradable fractions. In order to estimate the removal rates, the following procedure is recommended. First, the readily and slowly biodegradable components of the wastewater may be estimated using a batch test where both OUR and substrate are measured. Second, AOUR experiments may be used to measure the removal rate of the readily biodegradable fraction. Due to the strong influence of this fraction on respirometric determinations, this wil l be a reasonably accurate measurement. Third, the removal rate of the slowly 243 200 180 160 140 "120H Continuous feed experiments are at a loading < 2 Overall BOD. Slowly biodegradable BOD Readily biodegradable BOD 1 1 1 '"I 2 4 - | — i — i — i — | — i — i — i — | — i — i — i — | — i — i — i — | — i — i — i — | — I — i — i — | — i — i — r 6 8 10 12 14 16 18 20 Loading (mg BOD/mg biomass day) Figure 7.2.17 BOD vs. loading assuming two wastewater fractions. Arrow marks the loading of the lab scale units. 3.5--- Overall SUR - / / I / / / / / / / / / Readily biodegradable SUR X / ^ s s • Slowly biodegradable SUR i i i 1 i 1 1 1 1 1 1 1 1 I 1 - l 1 1 1 1 1 r p i ""P i— j— r—• T 2.5-E o C D C O at ZD OO 20 40 60 80 100 BOD (mg/l) 120 140 Figure 7.2.18 SUR vs. loading assuming two wastewater fractions. biodegradable fraction may be calculated using the fed-batch test and assuming that all of the measured BOD belongs to the slowly biodegradable fraction. At the flow rates used in this study, the readily biodegradable substrates make up less than 5% of the BOD coming out of the fed-batch test (this was verified by measuring the AOUR). The third step will not be as accurate as the AOUR measurements in the determination of model parameters, but these numbers are important for a more accurate determination of effluent quality. The resulting kinetics are shown in figures 7.2.17 and 7.2.18. The simplification of B K M E to two substrates will not result in too much error due to the large difference in removal rates between the two groups. As shown in figure 7.2.17, the slowly biodegradable fraction dominates the BOD in the treated effluent under normal loading conditions. Figure 7.2.18 shows that the metabolic activity of the biomass measured during the AOUR test will be due almost exclusively to the readily biodegradable substrate. This is especially true in light of the fact that the yield on the slowly biodegradable fraction is greater than the yield on the readily biodegradable fraction. Depending on the particular situation (wastewater composition and removal rates), some intermediate biodegradable organics may behave as readily biodegradable under low loading conditions, but will start to be incompletely removed at intermediate loadings. As the loading increases, these substrates will switch from the readily biodegradable fraction to the slowly biodegradable fraction. The importance of the ratio of readily biodegradable organic matter to slowly biodegradable organic matter of a batch of wastewater is shown in figure 7.2.19. The data in this figure was calculated assuming the maximum substrate uptake rate ofthe readily biodegradable component is 0.0018 mg BOD/lmg M L V S S minute, with a half 245 200 0 1 2 3 4 5 6 7 8 9 10 Loading (mg BOD/ mg biomass day) Figure 7.2.19 BOD vs. loading assuming two wastewater fractions, effect of wastewater composition. 246 saturation constant of 0.9 mg BOD/1, the maximum substrate uptake rate for the slowly biodegradable component is 0.0003 mg BOD/mg MLVSS minute with a half saturation constant of 7.5 mg BOD/1, and a wastewater BOD of 180 mg/l. If there is no slowly biodegradable fraction present, there will have to be a significant increase in the loading (to near 4 mg BOD/mg MLVSS) to the activated sludge unit before an increase in effluent BOD will be noticed. If the wastewater is composed solely of slowly biodegradable material, an increase in effluent BOD will be evident as soon as the loading is increased. For the wastewater studied, with a typical batch being composed of 60 to 80% readily biodegradable matter, an increase in loading will only result in slight increases in effluent BOD, and this BOD will belong mostly (>95%) to the slowly biodegradable component of the organic matter in the wastewater. Assuming one wastewater fraction when there are two or more with greatly different kinetics will result in error in predicting the performance of continuous activated sludge units based on batch test and AOUR kinetic data. However, if the yields ofthe various substrates are equal, then the SUR and BOD in the continuous system may be accurately determined by measuring the OUR in the continuous system. The SUR is equal to the OUR divided by (1-Y). If the SUR is known, the BOD in the treated effluent may be calculated using a simple mass balance (S = So - Oh * SUR). Unfortunately, if the yields of the various compounds are different as with BKME, then the simple relationship between OUR and SUR, and hence between OUR and BOD, is no longer valid. Under these conditions the OUR data become difficult to interpret, and it is best to use the fecl-batch data for modeling the wastewater as one substrate. This approach will give adequate predictions of the treated wastewater BOD as long as the loading does not 247 surpass the removal rate of the readily biodegradable components (i.e. for the data presented here, loading < 4 mg BOD/mg MLVSS) . If the batch data is analysed using the multicomponent model according to the method of Grau et al (1975) and this equation is applied to a continuous system using the method of Argaman (1991), the BOD in the treated wastewater is overpredicted (figure 7.2.3). For B K M E , near the end of the batch test, at low substrate concentrations, the remaining substrates have low removal rates. This is reflected in the multicomponent model coefficients. At low loadings the substrate concentration coming out of the continuous system is low, so the batch test predicts low substrate removal rates. This is not the case. A very small portion of the substrate will belong to the readily biodegradable group. This group has a small KM-value and even at low concentrations has a high removal rate. Consequently, the overall removal rate will be greater than that predicted by the multicomponent model. The predictive power of the multicomponent model decreases when there is a large difference in the removal rates of the compounds removed at the start of the batch test and those removed at the end of the batch test. The multicomponent model is dependent on the ratios of the various substrates. 7.3 Variability of Kinetic Coefficients and Wastewater Composition Batch Tests Comments The wastewater batch used to obtain the results discussed in section 7.2 was batch L. Variability in the degradation rates and stoichiometry of different wastewater batches was observed over the course of this project. In this section, respirometric data and batch 248 1.4-1.2-1 1 c 0.8-O ro 0.6H at o 0 .4H 0.2H 60 •160 1-140 120 h-100 ^ CD -80 — Q O -60 -40 20 I 1 1 1 120 180 240 Time (minutes) 1 1 1 r 300 360 0 Figure 7.3.1 OUR (•) and BOD (•) vs. time during a batch B K M E biodegradation test. Batch M , SRT 5 days. 1.5 3.5 5.5 7.5 9.5 Time (minutes) 11.5 13.5 15.5 Figure 7.3.2 OUR profiles for injections of samples withdrawn during the batch test (figure 7.3.1). Samples from top to bottom (sample time in brackets): #1 (0 minutes), 2 (10), 4 (30), 6 (50), 10 (120), 14 (300). Batch M , SRT 5 days. 249 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Substrate (mg BOD) Figure 7.3.4 Batch test samples OC vs. fraction 1 BOD (•), and vs. fraction 2 BOD (•) (figure 7.3.1). Batch M, SRT 5 days. 250 test data for a number of other wastewater batches will be presented in order to demonstrate the variability in the wastewater composition and the microbial kinetics. Data from wastewater batch M is shown in figures 7.3.1, 7.3.2, 7.3.3, and 7.3.4. The OUR profile in figure 7.3.1 has the same general shape as the one in figure 7.2.2, but the second OUR plateau was lower and lasted longer. The BOD removal during the batch test followed the multicomponent model. The B K M E batch appeared to be composed of at least three different groups of substrates. The first substrate was removed during the first 50 minutes of the batch test. As the OUR steadily dropped during this time, the first substrate either had a very large half saturation constant, or was actually composed of many different organic substrates. The second OUR plateau was most likely due to the oxidation of just one substrate, probably methanol. Samples were withdrawn during this batch test and the removal rates ofthe samples were measured using the AOUR method. Examples of OUR profiles of various samples are presented in figure 7.3.2. From these curves, it is evident that the first substrate was removed by the time the sixth sample was taken. This corresponds to the time at which the OUR dropped during the batch test (figure 7.3.1). The AOUR data (figure 7.3.3), and the oxygen consumption data (figure 7.3.4), for the two substrates were calculated and support the hypothesis that the first substrate was formic acid and the second was methanol. At the time of this batch test, the yield on formic acid was lower than the yield on methanol, so greater oxygen consumption on formic acid oxidation, as was found for substrate one, is expected. Also, for the set of biomass used in the batch test, the formic acid oxidation rate was significantly greater than the methanol oxidation rate as determined in a separate experiment using the pure substrates. This batch of wastewater had a similar composition 251 0 60 120 180 240 300 360 420 Time (minutes) Figure 7.3.5 OUR (•) and BOD (•) vs. time for a batch BKME biodegradation test. Batch N, SRT 15 days. 13.5 Time (minutes) Figure 7.3.6 OUR profiles for injections of samples withdrawn during the batch test (figure 7.3.5). Samples from top to bottom (sample time in brackets): #1 (0 minutes), 2 (9), 3 (20), 5 (40), 7 (60), 8 (75). Batch N, SRT 15 days. 252 0.9-r 0.8-0.7-"5 _c 0.6-E 0.5-0 CO 0.4-cm ZD 0.3-0 < 0.2-0 . 1 -4 5 Readily Biodegradable BOD (mg/l) Figure 7.3.7 Batch test samples AOUR vs. BOD (sample time in minutes). Sample #1 (0 minutes) (•), sample 2 (9) (•), sample 3 (20) (A), sample 4 (30) (•), sample 5 (40) (•), sample 6 (50) (o). figure 7.3.5, batch N , SRT 15 days. Time (hours) Figure 7.3.8 OUR (•) and BOD (•) vs. time for a batch BKME biodegradation test. Batch O. 253 to the batch discussed in section 7.2, but the kinetics were considerably different. Methanol, which is normally removed at a high rate, was removed a relatively slow rate. For this batch, and this set of biomass, methanol behaved as a slowly biodegradable substrate. Consequently, batch M had a large slowly biodegradable fraction, but this was due more to the biomass composition than the wastewater. The next batch investigated in detail was batch N (figures 7.3.5, 7.3.6, 7.3.7). The OUR during the batch test dropped suddenly four times, meaning that this batch was composed of at least five different substrates (figure 7.3.5). This batch had enough of the different readily biodegradable substrates to see distinct drops in the OUR profile rather than a steady decline as for the first substrate in the previous two batches discussed. The BOD removal followed the multicomponent model as expected for multiple substrate wastewaters. Samples removed during the batch test were analysed using the respirometric method. Example OUR profiles on the samples (figure 7.3.6) show that it is very difficult to differentiate between the different substrates. The starting concentration in the AOUR tests on the whole wastewater (sample 1) was much less than the starting concentration during the batch test so the sudden drops in the OUR were not noticeable, and the OUR gradually declined to the baseline. Due to the complexity of this batch, the entire readily biodegradable substrate was treated as one compound with the removal rates as shown in figure 7.3.7. This will be adequate to model the continuous system until very high loadings are attained. The slowly biodegradable fraction of this batch was smaller than the others investigated. The composition of batch O (figures 7.3.8, 7.3.9, 7.3.10) also appeared to be different from the other batches investigated. All of the readily biodegradable organics 254 3.5 3H H 2.5-c E — 2-CN o jf 1.5H o 0.5 o-r-1.5 2.5 3 3.5 4.5 Time (minutes) Figure 7.3.9 OUR profiles for injections of samples withdrawn during the batch test (figure 7.3.8). Samples from top to bottom (sample time in brackets): 2 (10 minutes), 3 (20), 4(30). Batch O. — 3 5 o o Buu) o — OC 2.5-o 2 -C E 1.5-^ C N o 1 -CD AOUR 0.5-0 -3 4 5 Readily Biodegradable B O D (mg/l) Figure 7.3.10 Batch test samples AOUR (•), and OC (•)vs.BOD. Figure 7.3.8, batch O. 255 were removed during the first hour of the batch test, and there was only one major drop in the OUR, implying that there was only one major readily biodegradable substrate. The readily biodegradable substrate was probably composed of methanol, formic acid, and other compounds (Kringstad and Lindstrom 1984), but these compounds were present in just the right ratio to compensate for the different removal rates and therefore to disappear from the batch test at more or less the same time. With the given data it is impossible to separate the readily biodegradable substrate into various fractions. The OUR profiles (figure 7.3.9) from several AOUR experiments on different samples from the batch test had a slight shoulder indicating the presence of several substrates which appeared to be removed simultaneously. The readily biodegradable component may be treated as one substrate, with removal rates as presented in figure 7.3.10. In this instance, the AOUR was better approximated with the Powell equation than the Monod equation. The last batch investigated in detail was batch Q (figures 7.3.11, 7.3.12, 7.3.13, . 7.3.14, 7.316, 7.3.16). Once again, the substrate removal during the batch test could be approximated with the multicomponent model, and the OUR profile suggests there were at least three different substrates, with a large slowly biodegradable fraction (figure 7.3.11). In order to verify the use of the BOD test as an adequate measure of the substrates during a batch test, the BOD kinetics were measured on a few samples (figure . 7.3.12). If the wastewater composition was changing throughout the batch test, BOD may not be an adequate measure of the wastewater strength. It is possible that the BOD test will measure a greater percentage of the substrates in the first sample than in subsequent samples. Figure 7.3.13 demonstrates that the BOD5, B O D u , COD, and OC (during the AOUR test) all behaved identically. The BOD test measured the same 256 1.2-c 0.8-CN O 0.6-CD j * 0.4-O 0.2-0.5 I 1 1 1 1 I 1 1.5 2 Time (hours) 2.5 180 160 - 1 4 0 - 1 2 0 -100 co H 8 0 § CD 60 40 h 2 0 ho 3.5 Figure 7.3.11 OUR (•) and BOD (•) vs. time for a batch BKME biodegradation test. Batch Q, SRT 15 days. 300 8 10 12 Time (days) 20 Figure 7.3.12 15 day BOD data for batch test samples (sample time in brackets). Sample 1 (0 minutes) (•), sample 2 (10) (•), sample 6 (50) (A), sample 10 (105) (•), sample 11 (135) (•), sample 14 (o), BOD out (A), (figure 7.3.11), batch Q, SRT 15 days. 257 200 180H _ 160-"co U o 140H 120-Q 3 1 00- l 2 80H tn Q 60-O m 40-20-• A A A A A 0.5 2 1.5 Time (hours) 2.5 3 • h750 -700 -800 E -650 o U -600 -550 3.5 Figure 7.3.13 BKME biodegradation batch test (figure 7.3.11). BOD(5) (•), BOD(u) (•), OC (A), and COD (•) vs. time. Batch Q, SRT 15 days. Figure 7.3.14 OUR profiles for injections of samples withdrawn during the batch test (figure 7.3.11). Samples from top to bottom (sample time in brackets): # 1 (0 minutes), 3 (20), 5 (40), 7 (60), 9 (90). Batch Q, SRT 15 days. 258 1.8 1.6H •i 1 . 2 H CN H o I5 0.8-^ 0.6-O < 0.4-0.2-O f o Or, A O A • n—i—i—i—r 0 0.5 ' I ' 1 1 1 I 1 1 1 1 I ! 1.5 2 2.5 Substrate (mg BOD/I) 3.5 Figure 7.3.15 Batch test samples, AOUR vs. BOD (sample time in brackets). Sample 1 (0 minutes) (•), sample 2(10) (•), sample 3 (20) (A), sample 4 (30) (•), sample 5 (40) ( ), sample 6 (50) (o), sample 7 (60) (A), sample 8 (75) (o), sample 9 (90) (*), sample 10 (105) (<). (figure 7.3.11), batch Q, SRT 15 days. 0.5 r 1.5 2 BOD (mg/l) Figure 7.3.16 Batch test samples OC vs. fraction 1 BOD (•), and vs. fraction 2 BOD (•) (figure 7.3.11). Batch Q, SRT 15 days. 259 fraction of substrates for samples throughout the batch test, within the error of the measurements. The B O D u was not measured during the other batch tests, but the COD always correlated very closely with the BOD. Sample OUR profiles from the AOUR assay are difficult to interpret (figure 7.3.14), but demonstrate that all the substrates were removed simultaneously. The AOUR (figure 7.3.15) reveals the presence of two main substrate groups, with removal rates very similar to what would be expected from examining the OUR profile of the batch test. Once again, the first substrate appeared to be composed of formic acid and a few other compounds, and the second substrate was methanol. This was verified by the yields (figure 7.3.16). Note that for this set of biomass the yield on formic acid was greater than the yield on methanol which is the opposite to the biomass used to test batch M . This was confirmed with independent AOUR tests using methanol and formic acid, and may be explained by differences in metabolic pathways. For growth on formic acid and methanol, several different metabolic pathways are possible, and the efficiencies of the different pathways are quite different. In section 7.2, the removal rates obtained for the wastewater fractions from the batch test, and the removal rates determined by respirometry on the samples from the batch test were not the same (table 7.2). This may have been due to changes in the biomass between the two sets of measurements, or the different initial substrate concentrations. For the wastewater batches discussed in this section, the removal rates obtained from the batch tests and the respirometric method were in much better agreement 260 Time (hours) Figure 7.3.17 OUR vs. time for BKME biodegradation tests. Batch A ( ), batch B ( ), batch J ( ), batch K ( ), batch L ( ). 0 | i i — i i i i i i i—i i i I i I I — r - — \ — | i i i — i | — i — i i | — i - 1 — i - 1 — | 0 1 2 3 4 5 6 7 Time (hours) Figure 7.3.18 OUR vs. time for BKME biodegradation tests. Batch L ( ), batch M ( ), batch N ( ), batch O ( ), batch Q (—) . 1 6 0 * 0 1 2 3 4 5 6 7 Time (hours) Figure 7.3.19 BOD vs. time for BKME biodegradation tests. Batch A (•), batch B (•), batch J (•), batch K (•), batch F (A). 0—| 1 1 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 1 1 ! 1 1 ! 1 1 1 ! 1 1 ! 1 1 r—j 0 1 2 3 4 5 6 7 Time (hours) Figure 7.3.20 BOD vs. time for BKME biodegradation tests. Batch L (•), batch M (•), batch N (A) , batch O (•), batch Q (•). 262 Other batches were not investigated in as much detail, but serve to illustrate the variability of the wastewater composition and microbial kinetics. The OUR profiles for a number of different batch tests are presented in figures 7.3.17 and 7.3.18 for comparison. The OUR profiles give an indication of the composition of the wastewater, and also of the removal rates of the various substrate fractions. The relative amounts of the various fractions can be estimated by the areas under the different sections of the OUR profile and appeared to vary from batch to batch. Some batches of wastewater behaved as i f they contained three distinct fractions (B, N), others as if there were two (A, J, L , M , Q), and others as if there was just one (K ,0) readily biodegradable fraction. The BOD measurements corresponding to the OURs are plotted in figures 7.3.19 and 7.3.20, and serve as a verification of the OUR data. The OUR and BOD profiles for the different batch tests were quite different. This is due to both the changing composition of the wastewater from batch to batch, as well as changes in the biomass. The relative oxidation rates of the different fractions can be compared by comparing the OURs at different points during the batch tests. The initial OUR varied from 1 to 3.8 mg/l minute. This variation is probably due more to the differences in biomass than the differences between the wastewater batches, although at least one batch (K) did not seem to have this fraction of the wastewater. The OUR profile of the first fraction of wastewater sometimes declined gradually to the OUR plateau due to the second substrate group (M). This implies a large K M value, or more likely, that this fraction is composed of many different substrates. In many of the batches the first fraction to be removed was small, and was removed from the batch test within the first twenty minutes. In these cases, this fraction will dominate the measurement of the 263 removal rates by batch tests where the initial OUR is measured, and result in erroneously high rate measurements. This first wastewater fraction is probably due to formic acid oxidation. In cases where there appear to be many substrates in this fraction, the other substrates are probably other small molecular weight organic acids. The highest OUR does not necessarily correspond to the largest fraction or the one with the greatest removal rate. The second OUR plateau in the batch tests usually lasted longer than the first plateau and remained constant until a sudden drop at the end. This second substrate group appears to be the dominant substrate in BKME, in terms of quantity. In most cases this OUR is probably due to methanol oxidation. For one batch (O), there was no second OUR plateau. The organic acids and methanol were present in the right proportion to be removed at the same time during the batch test. The OUR of the second plateau varied greatly from batch to batch, implying that the rate and yield of methanol oxidation varied from batch to batch. Comparison of the BOD and OUR curves reveals that generally the lower the OUR, the slower the rate of BOD removal, as expected. The batch with the slowest BOD removal had an initial OUR that was comparable to that of the other batches, but the second OUR plateau was very low. All of the BOD removal curves followed the multicomponent model, with n ~ 1.5 for most of the batches. For the batch with the slowest BOD removal rate, n=l. The BOD remaining after the OUR dropped close to the baseline value gives an indication of the amount of slowly biodegradable substrate in the wastewater. The variability of the multicomponent model coefficients implies that the wastewater composition was changing from batch to batch. 264 Caustic Extraction Effluent A batch test with the measurement of OUR was performed using the first caustic extraction stage effluent (figure 7.3.21). This process stream is high in BOD, and one of the major sources of readily biodegradable organics in B K M E . The high strength of this wastewater made it possible to determine the various substrate groups and to test the assumption that the readily biodegradable substrates are removed simultaneously during a batch test. In order to avoid adding substrates during the batch test, large samples were removed at two points during the batch test. These samples were then used to resuspend fresh biomass, and large amounts of methanol and formic acid were added. After the drop in the OUR at 120 minutes, the addition of methanol caused no increase in the OUR, implying that the substrate which was being oxidised at that point was methanol. However, the addition of a large amount of formic acid at this point caused the OUR to increase from 0.8 to 1.2 mg/l minute. Formic acid caused an increase of 0.4 mg/l minute in the OUR in a separate test when no effluent was present. These results imply that the first substrate was formic acid, and was used simultaneously with the second substrate, methanol. The first substrate was mainly formic acid, but formic acid can not account for an OUR as high as 1.9 mg/l minute, so other substrates must be present. These other substrates are probably small molecular weight organic acids. Because the OUR was much higher than can be explained by the oxidation of formic acid, these compounds must be utilised simultaneously. The slow decrease in OUR to the level that is due to formic oxidation implies that there were many different substrates being removed. 265 2 0—I—i—i—i—i—|—i—i—i—i—|—i—i—i—i—I—i—i—i—'—|—i—i—1—i—I—1—1—1—1—| 0 50 100 150 200 250 300 Time (minutes) Figure 7.3.21 Batch test, OUR vs. time. Caustic extraction effluent. 0.0004 "aT 0.00035 C 0.0003H >, 0.00025 Is 0.0002 Q O 0.00015 cn CD JE. 0.0001 5? 0.00005 0 30 BOD (mg/ l ) 60 Figure 7.3.22 Slowly biodegradable BOD removal rates measured by the fed-batch test vs. slowly biodegradable BOD. Batch L (•), batch M (•), batch N (A), batch O (•). 266 The OUR drop at -260 minutes is due to the disappearance of methanol. Adding large amounts of methanol after this point caused the OUR to increase back to 0.8 mg/l minute (an increase of 0.4 mg/l minute). However, methanol caused an increase of 0.6 mg/l minute in the OUR in a separate test when no effluent was present. The fact that the addition of methanol to the batch test sample from -260 minutes did not increase the OUR by 0.6 mg/l minute means that either the yield of the remaining substrate and methanol mixture was different than expected (see section 6.3), or the methanol and the remaining substrate were utilised sequentially. If the yield was changing, the remaining substrate was probably acetate. A more likely explanation is that the remaining substrate was being utilised sequentially, since methanol alone causes the OUR to increase from 0.2 to 0.8 mg/l minute. Formaldehyde may be present in the effluent, or it may be a byproduct of the metabolism of other substrates. Methanol and formaldehyde may share a common biodegradation path in the activated sludge used for these tests, so sequential oxidation would be expected, as found for the special case of methanol and formic acid in section 6.3. In summary, the first major substrate in the batch test may have been formic acid (and related compounds), and the oxidation of these substrates is independent of the other substrates present. The second major substrate may have been methanol, and its oxidation appears to delay the oxidation of the third substrate group, probably formaldehyde. These results demonstrate that the assumption of simultaneous removal is appropriate for most of the readily biodegradable compounds, even when the initial BOD is very high. However, a few substrates will be removed sequentially. 267 Slowly Biodegradable Kinetics The kinetics ofthe slowly biodegradable fraction were determined for four of the wastewater batches using the infinite dilution / fed batch test (figure 7.3.22). Due to the difficulty in obtaining the data, and the scatter in the results, the Monod model was assumed for the purposes of data fitting. There is not enough data to attempt to fit more detailed models. For three ofthe batches tested (L ,N ,0 ) , the removal rates of this fraction were very similar. The slowly biodegradable removal rates were approximately an order of magnitude less than the readily biodegradable removal rates. These batches had similar wastewater compositions, dominated by the readily biodegradable fraction, but as pointed out earlier, the organics in the treated effluent (under normal loading) wil l belong to the slowly biodegradable fraction. For the fourth batch, M , the removal rate ofthe slowly biodegradable fraction was twice the removal rate found for the other wastewater batches. For this set of biomass, the removal rate of methanol was low enough for it to show up during the infinite dilution / fed batch test and to be classified as a slowly biodegradable substrate. The low removal rate of methanol during the treatment of this wastewater batch was verified by independent batch assays using methanol as the sole substrate. However, the removal rate of methanol was still larger than the removal rate of the other slowly biodegradable compounds, and thus dominated the measurement of the slowly biodegradable removal rate. If the other slowly biodegradable substrates could be separated from the methanol, the removal rates would probably be similar to the other three batches studied. As a consequence of grouping the methanol into the slowly biodegradable fraction, this 268 fraction was larger than the readily biodegradable fraction. The readily biodegradable fraction contained only formic acid and similar compounds. The biodegradation rates of this fraction were an order of magnitude greater than the slowly biodegradable removal rates. Figure 7.3.22 indicates that an increase in loading during the treatment of batch M will result in much larger BOD in the treated wastewater than during the treatment of the other batches, emphasizing the importance of the ratios of wastewater fractions (figure 7.2.17). The slowly biodegradable fraction of B K M E is undoubtedly composed of many substrates; the kinetics shown in figure 7.3.22 will be dominated by the substrate with the greatest removal rate. Since the removal rates are low, these substrates may be grouped into one fraction without too much error. 7.4 Operating Conditions & Adaptation In the previous section, the results of batch tests and fed-batch tests were presented. These results showed the variability of the wastewater composition from batch to batch, as well as the variability of the microbial kinetic coefficients. In this section the results of the AOUR tests performed over the course of this project will be presented. The AOUR test is comparatively easy to perform, and consequently much more data was collected than from the batch tests. The data also demonstrates the variability of the wastewater, and the variability of the kinetic coefficients. In addition, the adaptation of the bacteria to the different wastewater batches, and the effect of different operating conditions on the microbial kinetic coefficients were investigated. 269 AOUR Coefficient Variability A surnmary of the AOUR kinetic coefficients measured on the different wastewater batches over the course of this project will be presented. The AOUR vs. day of operation is presented in figure 7.4.1; the oxygen consumption data is presented in figure 7.4.2. The maximum rates and stoichiometry ofthe two lab scale activated sludge reactors followed the same trends over time. The data is easier to examine when graphed vs. wastewater batch instead of time (figure 7.4.3). The maximum AOUR for the biomass from both reactors followed the same trends. The rates were low during the treatment of batch E, and high during the treatment of batch L. The maximum substrate uptake rate of biomass grown on and treating batch L was 1.6 times greater than the biomass grown on and treating batch M . This is a statistically significant difference. The operating conditions of the activated sludge unit were identical throughout the treatment of these two batches (SRT of 10 days, HRT of 12 hours). Inspection of figure 7.4.3 reveals that the wastewater batch being treated seems to have a greater effect on the maximum AOUR than the operating conditions (SRT / aerobic selector). The two activated sludge units were fed the same batches of wastewater at the same time, and the maximum AOUR of the two different units followed the same trends as the batches of wastewater changed regardless of their different operating conditions. This is due to both the different characteristics of the wastewater batches and to the different populations of microorganisms that are selected by these different batches. The reason for the high maximum AOUR during the treatment of batch L is unknown. For the rest ofthe batches the maximum AOUR was approximately the same. It is interesting to note that the AOUR test did not indicate the 270 3T 0.0035-4— D C E 0.003-oo oo > ^ 0.0025-0.002 H CD E CN o ^E 0.0015-ZD 0 0.001 -< 1 0.0005H "x D 2 c d o T c. • Q m cm o T 1 Q • • T Q o 0 o T O 0 100 200 300 400 500 600 700 800 Day Figure 7.4.1 Maximum AOUR vs. time over the course of the project. Reactor 1 ( • ) , reactor 2 (o). Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 2 5 0 Figure 7.4.2 OC vs. time over the course of the project. Reactor 1 (•), reactor 2 (o). Red line represents BOD. Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 271 COCO-' S 0 . 0 0 2 5 -t/0 £ 0 . 0 0 2 -E 0 . 0 0 1 5 -o g 0 . 0 0 1 -ZD O 0 . 0 0 0 5 -< 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B C E F G H I J K L M N Q R S O Batch - 2 . 5 T h 2 _ o - 1 . 5 m co £ , h-1 V -0.5 Figure 7.4.3 Maximum AOUR and K vs. wastewater batch over the course of the project. Reactor 1 maximum AOUR ( • ) , K ( • ) , reactor 2 maximum AOUR ( • ) , K ( o ) . 300-250H 200H "co ^ - 1 5 0 Q O CD 1 0 0 H 5 0 H hioo i i i i i i i i i i r A B C D E F G H I J K L M N P Q R O Batch 120 -80 - 6 0 - 4 0 h 2 0 co £ . U o Figure 7.4.4 OC (A) and BOD (bars) vs. wastewater batch over the course of the project. Shaded area represents slowly biodegradable BOD (where measured). 272 possibility of poor performance during the treatment of batch M i f the loading were to increase (large slowly biodegradable component). The wastewater composition varied from batch to batch (figure 7.4.4). This variation partially explains the variation in the oxygen consumption data. As the BOD of the wastewater increased, the oxygen consumption during respirometric tests also increased, as expected. Linear regression of the short term AOUR oxygen consumption vs. the wastewater BOD (long term oxygen consumption) yields a correlation coefficient of 0.79 (figure 7.4.5). The slope is 0.46. For every milligram of BOD added to the respirometer, 0.46 mg was oxidised. The slowly biodegradable fraction, approximately 30% of the B O D wil l not be oxidised during the AOUR test. 0.24 mg of the readily biodegradable fraction (1 mg - 0.3 mg - 0.46 mg) were not oxidised during the AOUR test, which implies an average yield of 0.34 on this fraction (0.24/(0.46+0.24)=0.34). A quick way to estimate the untreated wastewater BOD is to do a AOUR test and divide the OC by 0.46. This method is valid for wastewaters which have a similar ratio of slowly to readily biodegradable organics as the wastewater used in this study, and for wastewaters with similar yields. The strength of the different wastewater batches varied greatly, from 110 to 280 mg BOD/1, but the overall composition did not vary as much. The slowly biodegradable fraction usually made up 25%) of the BOD. The large apparent difference in composition of batch M compared to the other batches is due to the methanol component being grouped into the slowly biodegradable fraction for this particular wastewater batch. The actual composition of this batch was similar to the composition of the other batches. The readily biodegradable fraction usually comprised 75% of the wastewater BOD. The data 2 73 presented here does not distinguish between the various fractions making up the readily biodegradable component (i.e. methanol, formic acid, and other undefined low molecular weight components), but as discussed in the previous section, the composition appeared to vary from batch to batch. Variability of Single Substrate Kinetic Coefficients For the wastewater batches treated in the second half of the study, respirometric tests were performed using methanol, formic acid, and acetic acid as substrates. These are the main components of B K M E BOD. The AOUR data is presented in figure 7.4.6; the yield data is presented in figure 7.4.7. Unlike the situation when measuring the AOUR using B K M E , the respirometric yield on methanol, formic acid, or acetic acid, is easily calculated. By providing a constant substrate over time (methanol, formate, or acetate as opposed to B K M E which is variable), the variability of the biomass from day to day may be followed. For clarity, the AOUR data is plotted versus wastewater batch for methanol "(figure 7.4.8), formic acid (figure 7.4.9), and acetate (figure 7.4.10). The acetate data show very little variation during the treatment of batches L to S. The yield on acetate is also constant during this period (figure 7.4.7). The situation is different for the methanol and formic acid oxidation rates. The removal rate of both substrates increased during the treatment of batch L . For the other batches the removal rates varied, but not as much. An interesting observation is that although the overall AOUR remained constant for most of the wastewater batches, the AOUR due to methanol.and formic acid fluctuated. For example, during the treatment of batches K , M , and N , the overall AOUR was the same. During the treatment of batch K , the majority of this AOUR came from 274 0 50 100 150 200 250 300 BOD (mg/l) F i g u r e 7.4.5 Respirometric OC vs. wastewater BOD for all of the wastewater batches studied. aT 0.0016 D C £ 0.0014 0.0012H o O O O > 5 g 0.001 H ° 0.0008 co E a: 0.0006H ZD § 0.0004-0.0002-x D • • 290 • cm • • 5 u A A -I R 390 • B M • A • ml Q • A A A D OH _ E J a EJ D DA OO Oi 0 490 590 690 790 Day F i g u r e 7.4.6 Maximum AOUR vs. time over the course of the project. Methanol (•,•), formic acid (A,A), acetate (»,o). Reactor 1 filled points, reactor 2 open points. Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 275 0.9-Q 0.8-o u 0.7-CO 0.6-Q ... o 0.5-u CO 0.4-JE 0.3-j > 0.2-0.1-0-• • • A a M • y i • A 6^  • • Q a o o o o. o • A A' ft JB • cc a D • A O C o c i • 290 390 490 590 690 790 Day Figure 7.4.7 Respirometric yield vs. time over the course of the project. Methanol (»,•), formic acid (A,A), acetate 0,o). Reactor 1 filled points, reactor 2 open points. Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 0.003 3 0.0025 c i oo £ 0.002 co E 0.0015 o co ^E cm ZD o < 0.001 0.0005H M N Batch Figure 7.4.8 Maximum methanol AOUR vs. wastewater batch over the course of the project. Reactor 1 (•), reactor 2 (•). 276 0.003 1 0.0025-c E S £ 0.002-£ 0.0015-j ro 0.001 -ZD O 0.0005-< ] H I J K L M N P Q R S O Batch Figure 7.4.9 Maximum formic acid AOUR vs. wastewater batch over the course of the project. Reactor 1 (•), reactor 2 (•). 0.003 nute) 0.0025-I A/SS 0.002--CO 0.0015-CN 0 ? 0 . 0 0 1 A at O 0.0005-H I J K L M N P Q R S O Batch Figure 7.4.10 Maximum acetate AOUR vs. wastewater batch over the course of the project. Reactor 1 (•), reactor 2 (•). 277 methanol oxidation. During the treatment of batch M , the majority of the AOUR came from formic acid oxidation (although there was still plenty of methanol in the wastewater). Finally, during the treatment of batch N , the contribution to the overall AOUR was equal for methanol and formic acid. This observation becomes evident when the A O U R M A X due to methanol is compared to the A O U R M A X due to formic acid (figure 7.4.11). During the treatment of batch K, this ratio is greater than one, during the treatment of batch M it is less than one, and is approximately one during the treatment of batch N . These results clearly demonstrate the changing microbial populations with time in the lab-scale activated sludge units. The microbial population appeared to be dependent on the wastewater composition, as it appeared to vary with the wastewater batch, and often followed the same trend in the two units. The yield on methanol and formic acid also varied from day to day. The ratio of the yield on methanol to the yield on formic acid is presented in figure 7.4.11. The yield on these compounds can change quite rapidly. The ratio of the yields generally followed the ratio of the AOURs. Low yields on one-carbon compounds are probably due to energy spilling. On days where there are fewer enzymes available for utilising methanol for growth, it can be expected that more of the methanol will be wasted than on days when there are more enzymes available for utilising the methanol for growth. This will result in a lower yield. These results present further evidence for the variability of the microbial population with time. 278 o at ZD o < o _o o at 4.5-1 4 3.5-3-2.5-2-1.5-1 0.5 O D • 6 o M 8 © D c* 1 ® o Q • 380 430 480 530 580 Day 630 680 730 780 Figure 7.4.11 Ratio of methanol AOUR to formic acid AOUR: Reactor 1 (•), reactor 2 (•). Ratio of methanol yield to formic acid yield: Reactor 1 (o), reactor 2 (•). Letters represent wastewater batches, solid vertical lines mark boundaries between experimental runs. 0.0025-] "5 -c 0.002-I OO OO ... >j -0.0015-CO £ CN -0 0.001 -CO jE at -ZD 0.0005-0 < o-15 20 SRT (days) 30 Figure 7.4.12 Maximum AOUR vs. SRT for all of the wastewater batches studied. Reactor 1 (•), reactor 2 (•). 279 Effect of SRT / Selector on Kinetic Coefficients Figure 7.4.12 shows the relationship between the removal rates ofthe readily biodegradable component of the wastewater as measured with the AOUR assay and the activated, sludge operating conditions. These values are averages of the respirometric assays, previously presented in figures 7.4.1 and 7.4.3, for each of the 18 batches treated over the course of this project. As can be seen, the removal rates did not change very much with changing operating conditions. The exceptionally high values at an SRT of 10 days corresponded to the treatment of batch L. For activated sludge operated at SRT's between 5 and 20 days the readily biodegradable kinetics can be considered constant at qMAx = 0.0025 ± 0.0005 mg BOD/mg biomass minute, and K M = 1.2 ± 0.3 mg BOD/1. The removal rates measured on the unit with a selector, at an SRT of 10 days, were also within this range. The removal rates ofthe biomass taken directly from the full scale activated sludge plant