High Rate Anaerobic Treatment of Complex Wastewater

7 downloads 89 Views 2MB Size Report
Greg Crocetti and Phil Hugenholtz were free with their ideas and time in the ..... be rate limiting compared to methanogenesis (Pavlostathis and Giraldo-Gomez,.
High Rate Anaerobic Treatment of Complex Wastewater

Damien J. Batstone B. Chemical Engineering (Hons)

A thesis submitted for the degree of Doctor of Philosophy in Chemical Engineering at The University of Queensland

October 1999

Cover: False colour SEM image of a cleaved granule from the Spearwood high rate anaerobic treatment system. Colour was based on presumptive identification.

The research in this thesis was financially supported by Australian Postraduate Awards (Industry) and Environmental Solutions International Ltd.

Batstone, D. (2000). High-Rate Anaerobic Treatment of Complex Wastewater ISBN: 1 86499 3863 Copyright  by Damien Batstone All rights reserved. No part of this thesis, apart from bibliographic data, may be reproduced in part or in full, in any form or by any means, or stored in a database or electronic retrieval system without the prior written permission of the author.

ii

DECLARATION I declare that the work presented in this thesis, to the best of my knowledge and belief, is original, except as acknowledged in the text. This material has not been submitted in whole or part, for a degree at this or any other university.

_____________________ Damien J. Batstone

iii

ABSTRACT High-rate anaerobic degradation of soluble organic pollutants is becoming very popular, particularly for carbohydrate-based industrial wastewaters. Despite the successes achieved, there are significant limitations in the application of this technology to more complex wastewaters. These are defined as containing other organic compounds such as particulate and soluble proteins and fats, and originate from abattoirs (slaughterhouses), meat and food processing and similar industries. Complex wastewater is often difficult to degrade and components such as solids and fats have slow degradation kinetics and can diminish process performance. Also, the growth of granular sludge, which is critical for optimal performance in upflow reactors, is slow and granule properties such as shear strength and settling velocity are poorer. This is reflected in a lower treatment efficiency of 50%-60% in systems treating complex wastewater compared with efficiencies of 85%-95% in carbohydrate fed treatment systems. This thesis examines specific aspects in the treatment of complex (proteinaceous) wastewater in high rate upflow anaerobic treatment plants and the influences of different conversion processes and microbial characteristics on design and operation. The research problem was approached in two ways: The macroscopic conversion processes were examined by investigating and modelling a two-stage full-scale high rate hybrid reactor in Spearwood, Western Australia, designed and operated by ESI Ltd. This allowed localisation of the key conversion process; specifically hydrolysis of solids, which was found to occur mainly within the methanogenic reactor. Degradation of soluble proteins was rapid and all proteins were fully acidified in the acidogenic (first) stage even at very low retention times. Because of the rapid protein degradation rates, partial acidification, which is often a strategy to improve granulation rates, is incompatible with pH, flow and concentration equalisation. The influence of a protein feed on granulation compared with a carbohydrate feed was examined by sampling granules from the above reactor, as well as two full scale brewery fed reactors and a full scale reactor fed fruit and vegetable cannery wastewater. The cannery fed granules had the highest shear strength and settling characteristics while the protein fed granules had low strength and density, low settling velocity and a comparatively wide size distribution. Both brewery fed granules had very similar and suitable properties. Molecular studies using fluorescent in-situ hybridisation (FISH) probing and microscopy indicated that the granules from the complex (protein) wastewater fed reactor had limited structural characteristics, possibly due to limited reaction rates (as opposed to diffusion rates). Granules from the cannery reactor and both brewery reactors had structures that appeared to be the result of diffusion limitations. Therefore, the critical operational constraints when treating complex wastewater are the particulate biomass and particulate substrate. Awareness of process status could be increased by monitoring of biological and substrate solid inventory in the methanogenic reactor. The model developed in this thesis can greatly assist this. Complications due to particulate substrate and poor granule properties may be intrinsic to complex feeds. These constraints are probably

iv

best addressed by design of a methanogenic reactor specifically for complex wastewater. The design should attempt to separate substrate hydrolysis, minimise shear on the granules and retain solids.

v

ACKNOWLEDGEMENTS A PhD thesis, while outwardly a sign of the candidate’s professional and personal growth is also a product of many other people’s hard work. I was lucky to have a principal supervisor in Dr Jürg Keller with whom I enjoy a very productive working relationship. Jürg allowed me wide academic freedom while maintaining a capable hand on overall research direction. My associate supervisors were Dr Bob Newell and Dr Linda Blackall, both consummate experts in their fields and very free with their time whenever I needed help. My industry supervisor Mark Newland at ESI ltd helped a great deal with all types of concerns and problems during my stay in Western Australia. A special thanks goes to Rick Webb from the Department of Microscopy and Microanalysis, whose expertise created the stunning EM and LM images in chapter 10. Paul Addisson from the Department of Physiology and Pharmacology applied his microtome to bacteria instead of histological samples. Marion Dunstan, Gary Asmussen, Peter Abeydeera and Beatrice Keller from the Process and Environmental Analytical Centre did much of the analytical work for my thesis. Greg Crocetti and Phil Hugenholtz were free with their ideas and time in the application of FISH probing and this complete novice would have had no hope without their assistance. I was lucky enough to start extraction of EPS soon after Catherine Biggs perfected the method. Abdul Harris and Ian Ramsay were my fellow anaerobic experts and my work benefited from our discussions. Matt Rossinski derivitised samples for me for amino acid analysis. Our department has a deserved reputation for free exchange of ideas and assistance and I thank Assoc. Prof. Ian Cameron, Dr Bill Clark, and Dr Pratap Pullammanappillil for assistance in their areas of expertise. Any major project is a nightmare without proper funding and I avoided this thanks to Australian Postgraduate Awards (Industry), a joint venture between the Australian Government and an Industry partner. The Australian Government provided my scholarship while ESI Ltd paid all my operating costs and probably exceeded the record for in-kind contributions and patience. I also appreciate letting me work on their full-scale systems and occasionally perturbing normal operation to get system responses. Thanks also go to Golden Circle Canneries, Queensland Breweries, and Matilda Bay Breweries for allowing me to study their reactors. Part of this work follows directly from the model developed in the PhD work of Des Costello, Mohammad Romli, and Ian Ramsay. I gratefully acknowledge the excellent foundation they have given me. Thanks also too all my fellow students at the Advanced Wastewater Management Centre; especially Matt Wood, Marc Steffans and Steve Pickering who maintained our computer systems under sometimes difficult situations. Our administration officer, Lesley Johnstone has provided great support and advice together with Sharon Psomadellis and Ray Johnson.

vi

I thank those who have been my friends during this period. You have supported me often during this period. My parents Druce and Bronwyn my brothers Martin and Seamus and my sister Aislinn have always been a source of strength. Thank you to my darling Anna, who always loved me and always was there for me.

vii

PUBLICATIONS Papers in refereed publications Batstone, D. J., Keller, J., Newell, R. B. and Newland, M. (1997). “Model Development and full scale validation for anaerobic treatment of protein and fat based wastewater.” Wat. Sci. Tech. 36(6-7): 423-431. Batstone, D.J., Keller, J., Newell, R.B. and Newland, M. (2000) “Modelling anaerobic degradation of complex wastewater I: Model development” Bioresource Tech. 75: 67-74 Batstone, D.J., Keller, J., Newell, R.B. and Newland, M. (2000) “Modelling anaerobic degradation of complex wastewater II: Parameter estimation and validation using slaughterhouse effluent” Bioresource Tech. 75: 75-85 Papers submitted to refereed publications Batstone, D.J., Keller, J., and Blackall, L.L. “Use of established RNA Probes for Community Analysis in Full Scale High Rate Anaerobic Reactors” Submitted to FEMS Microbiol. Ecol. Batstone, D.J. and Keller, J. "Variation of Bulk Properties of Anaerobic Granules With Wastewater Type", Submitted to Wat. Res. Conference posters and papers Batstone, D. J., Keller, J., Newell, R. B. and Newland, M. (1997). “Model Development and full scale validation for anaerobic treatment of protein and fat based wastewater.” 8th Int. Conf. on Anaerobic Digestion, Sendai, Japan: 2:158166 (Also published in Wat. Sci. Tech, see above). Batstone, D. J., Keller, J., Newell, R. B. and Newland, M. (1997). “Flow profiles in high rate upflow anaerobic reactors” EERE ’97, Noosa, Qld, Australia Batstone, D.J., Keller, J. and Blackall, L.L. (1999) “Structural and phylogenetic characterisation of anaerobic biofilms and granules” 4th Int. Conf. on Biofilm Reactors, New York, USA.

viii

TABLE OF CONTENTS Declaration Abstract Acknowledgements Publications Table of Contents 1. Introduction Research Objective Research Approach Thesis Outline Contributions 2. Anaerobic Digestion Processes 2.1 Overview of Anaerobic Digestion 2.2 Hydrolysis of Particulate Substrates 2.3 Acidogenesis/Fermentation 2.4 Obligate Hydrogen Producing Acetogenesis and Hydrogenotrophic Methanogenesis 2.5 Acetoclastic Methanogenesis 2.6 Inhibition in Anaerobic Systems 2.7 Modelling Anaerobic Degradation Processes 2.8 Motivation, Objectives and Approach References 3. Model Development 3.1 Model Background 3.2 Model Outline 3.3 Liquid and Mass Balances 3.4 Physico-chemical Reactions 3.5 Biochemical Reactions 3.6 Model Implementation 3.7 Summary References Nomenclature 4. Analytical Methods 4.1 Physical analysis 4.2 Chemical analysis 4.3 Hydraulic Retention time study References 5. System Characterisation and Data Collection Strategy 5.1 The Spearwood Plant 5.2 Initial Characterisation 5.3 Data Set Collection 5.4 Comparison of Treatment Efficiencies 5.5 Reactor Hydraulics 5.6 Conclusions References 6. Parameter Estimation and Validation 6.1 Parameter Estimation Procedure

ix

ii iii v vii viii 1 2 3 3 5 7 7 9 18 24 30 30 32 35 36 40 40 41 44 47 49 60 63 64 66 68 68 70 72 72 73 73 76 82 86 87 90 91 93 93

6.2 Parameter Estimation and Elimination 6.3 Estimation of Parameters from Measured Data 6.4 Validation of Parameters 6.5 Yields of Products from Proteins 6.6 Critical Analysis 6.7 Conclusions References 7. Anaerobic Granules 7.1 Desirable Granule Properties 7.2 Structure 7.3 Mechanism of Granule Formation 7.4 EPS Role and Factors Affecting Production 7.5 Other factors Affecting Granulation 7.6 Conclusions References 8. Experimental Methods for Analysis of Granules 8.1 Granule Sampling and Storage 8.2 Bulk Property Tests 8.3 Microscopic Methods References 9. Bulk Properties of Anaerobic Granules 9.1 Chemical Composition 9.2 Density 9.3 Size Distribution 9.4 Settling Distribution 9.5 Shear Strength 9.6 Extracellular Polymeric Substances (EPS) 9.7 General Discussion 9.8 Conclusions References 10. Structure and Ecology of Anaerobic Granules 10.1 Golden Circle Cannery Granules 10.2 Yatala Granules 10.3 Matilda Bay Granules 10.4 Spearwood Granules 10.5 Discussion 10.6 Conclusions References 11. Conclusions and Recommendations Conclusions Recommendations Appendices Appendix A: Treatment of Logged Data Appendix B: Data Collected for Parameter Estimation and Validation Appendix C: Kinetic Parameters Appendix D: Model Library Files Appendix E: Data from Bulk Property Tests Appendix F: Additional CLSM, TEM, LM and SEM Images

x

98 101 110 117 118 121 122 125 126 128 135 136 139 139 142 145 145 147 153 156 157 157 158 158 160 162 163 167 169 170 171 172 176 179 182 186 190 197 199 199 200

Chapter 1

Introduction This thesis investigates fundamental and practical aspects of high rate anaerobic treatment of complex wastewater in upflow anaerobic systems. Complex wastewater is defined here as largely proteinaceous and containing a large range of components, some of which complicate anaerobic treatment (such as particulates and fat). Anaerobic digestion is biologically catalysed transformation of soluble organics to carbon dioxide and methane. The other major method of biodegradation is aerobic treatment, which oxidises organics to carbon dioxide using oxygen. High rate anaerobic treatment has undeniable advantages compared to aerobic treatment when treating industrial wastewater (Speece, 1996). Some of the major advantages are: • No aeration costs; • Production of a by-product (biogas) which can be used for heating and power generation; • Much lower generation and better stabilisation of biomass (i.e., biomass generally does not decay further); • Lower footprint and higher space-time loading; and • Better process stability (when system is matched to loads). The major disadvantage is the lack of an established nutrient removal capability, though there are several developing technologies such as the Kluyver group’s Anammox process (Strous et al., 1997) or removal via external nitrification and recycle (Hendriksen and Ahring, 1996). The incapability of treatment to secondary standard and poor performance on low strength wastewater have been cited as disadvantages but both have been demonstrated to be possible by increasing the residence time (Seghezzo et al., 1998) or by treatment in anaerobic baffled reactors (ABR), (Barber and Stuckey, 1999). High rate upflow reactors were introduced by Lettinga et al. (1980) (see chapter 5 for application). These depend on the development of granular sludge, which consists of large (>1 mm), rapidly settling anaerobic agglomerates (see Chapter 7). Novel reactor designs have been developed for specific applications such as the internal circulation (IC) reactor (Paques BV, NL), which has an extremely low footprint and very high loading rate, or the compartmentalised ABR (Barber and Stuckey, 1999). However, upflow reactors remain the most common design for high rate anaerobic treatment as they are relatively low cost, compact, and reliable.

1

CHAPTER 1: INTRODUCTION

High rate upflow anaerobic reactors are best applied to soluble high strength carbohydrate streams (such as the cannery reactor examined in the later part of this thesis). Under these conditions, degradation rates are extremely rapid and granulation rate is good. Treatment is complicated by the presence of solids, inhibitory compounds and components that cause foaming or can potentially cause granule washout. Compounds in complex wastewater can cause the following problems: • Fats can be difficult to hydrolyse and can damage the system by coating the granules (decreasing mass transfer), attaching to the granule and causing washout, or inhibiting activity (Rinzema et al., 1994). • Granule growth and accumulation tends to be lower (Fang et al., 1994, Moosbrugger, 1994) and granule strength is lower. • Particulate material can disrupt sludge bed development by blocking liquid distribution systems, increasing wash-out, diluting the bed with inactive material and favouring growth on the particle surface rather than in granular biomass (Lettinga and Hulshoff Pol, 1991). Hydrolysis rates of fats vary depending on fatty acid chain length, state (solid or liquid) and specific surface area (Martinelle and Hult, 1994). Protein hydrolysis rates depend on whether the protein is globular or fibrous, surface area and solubility (McInerney, 1988). In many carbohydrate fed systems, a single particulate solid such as starch or cellulose is present, and hydrolysis is relatively predictable. Protein based complex wastewaters may have large numbers of different proteins and fats, some readily degradable, some undegradable. This may explain the relatively low number of anaerobic systems treating complex wastewaters. Lettinga and Hulshoff Pol (1991) found only 6 treatment plants (out of 200) that treated wastewater with a large component of particulate protein or fats. A comparison of different full-scale systems indicated that plants treating proteinaceous wastewaters with particulate material could only achieve 45-65% COD removal while carbohydrate fed systems could achieve above 90% (Chapter 5). Despite the above problems, anaerobic process technology remains a lucrative and under-utilised option for the treatment of complex wastewater. There are actually several advantages when treating complex wastewater compared to treating highly soluble carbohydrate wastewater: • Because the wastewater matrix is diverse, trace element and nutrient addition is rarely necessary. • Ammonium formed in the degradation of proteins allows for strong pH buffering and often removes requirements for alkalinity recycling or pH control (often the largest cost in plants fed with carbohydrate based wastewater). • Because proteins and fats are generally less oxidised than carbohydrates, the biogas has a higher methane content (approximately 70%-80%), which increases calorific value and ease of use. • Extent of degradation is often higher (Speece et al., 1996) allowing lower effluent soluble COD.

1.1 Research Objective The major hurdle to acceptance of anaerobic treatment of complex wastewater by process suppliers and clients is the lack of understanding of the conversion processes

2

CHAPTER 1: INTRODUCTION

and their influence on the process. This thesis has found that knowledge gained by studying anaerobic treatment of carbohydrate wastewater is not directly applicable to complex wastewater as the issues are different. This is especially true during hydrolysis and fermentation/acidogenesis. As described in Chapter 2, acidogenesis of soluble carbohydrates is strongly driven by two driving forces: maximising yield and finding internal electron acceptors. These produce a small number of products in varying amounts. In contrast, protein acidogenesis products are well defined but diverse, caused by the diversity in amino acids. The influence of these changed issues on plant design and operation is not well understood. Therefore, the following research objectives have been addressed in this thesis: • Identify and investigate the critical conversion processes in anaerobic digestion of complex wastewater. • Describe these conversion processes in the form of a mathematical model. • Identify and study in detail a main limitation for the application of anaerobic treatment to complex wastewater and compare it to carbohydrate based wastewater.

1.2 Research Approach In this thesis, these objectives have been approached in three main ways: Conversion processes as observed at a macroscopic scale are examined: This was done by characterisation of a full-scale two stage anaerobic system. This allowed localisation of different conversion processes within the system. The main objectives were analysis of acidogenesis and solids hydrolysis, and potential effects on the system. This system was simulated by a structural dynamic model: The parameters were estimated using a parameter estimation procedure and a single data set. This model was then validated using data sets with and without recycle. The results from parameter estimation and validation allowed further insights into the process. Function and characteristics of the granules are examined: Successful granulation is a major factor for stable and efficient operation of upflow anaerobic reactors. While granulation on complex wastewaters is known to be poor, quantification of properties and the reasons for poor granulation have not been examined. This thesis addresses fundamental and applied aspects of relevant anaerobic treatment processes and is therefore focused on full-scale reactors. Many of the issues affecting anaerobic degradation of complex wastewater cannot be addressed in laboratory scale plants, and formulation of synthetic wastewater with a similar influent matrix is difficult.

1.3 Thesis Outline The knowledge background and research motivation are outlined in Chapter 2. Conversion processes and potential inhibition mechanisms are reviewed. Two main

3

CHAPTER 1: INTRODUCTION

areas have been identified as having significant knowledge gaps within the scope of this thesis: (a) The behaviour of complex solids within two-stage systems has not been analysed. While models have been directed towards solids degradation, these have been largely oriented towards solids digesters rather than high rate systems. (b) Few models have been applied to full-scale systems and none have been applied to systems treating complex wastewater. Laboratory scale reactors are often operated with synthetic wastewater, which does not simulate complex feeds. Costello (1989) developed a mechanistic model based on the hydrogen regulation model for glucose of Mosey (1983). This was implemented and tuned by Romli (1993), and Ramsay (1997) added degradation of soluble proteins. In Chapter 3 particulate hydrolysis of fats, proteins and carbohydrates as well as long chain fatty acid (LCFA) degradation are introduced to this model and some modifications of hydraulic equations are made to better reflect full-scale systems. Chapter 4 describes the experimental methods used for parameter estimation and full scale validation of the developed model. Chapter 5 characterises a full-scale system treating pig abattoir (slaughterhouse) wastewater in Spearwood, Western Australia, designed and operated by ESI Ltd, Osborne Park, WA. The plant is a hybrid upflow anaerobic sludge blanket (UASB) reactor with filter packing in the upper half to retain solids and encourage attached biomass growth. In this chapter, the influent matrix and process behaviour are studied and the hydraulics are characterised and modelled. Sample analysis for parameter estimation and validation relied on the dynamic step caused by start-up of the abattoir on Monday mornings. Chapter 6 presents the parameter estimation and validation. Parameter estimation relies on decreasing the number of degrees of freedom by identifying those parameters applicable from the literature or directly measured prior to using a dynamic parameter estimation procedure. Chapter 7 describes current knowledge of anaerobic granule properties and mechanisms of formation and growth. Major areas that are not covered in the literature are the effect of wastewater type on granule bulk properties and comparison of this with microstructure, ecology and reactor performance. In chapters 8, 9, and 10 four granular sludges from four full-scale reactors are characterised. The Golden Circle fruit and vegetable cannery feeds three UASB reactors. This was classified as soluble carbohydrate wastewater. Yatala brewery feeds two similar rectangular UASB reactors while Matilda Bay brewery uses a hybrid anaerobic reactor similar to that at Spearwood, WA. The fourth reactor is the Spearwood system, treating slaughterhouse wastewater from the abattoir in a hybrid reactor. This comparison of granular characteristics between three wastewater types and two reactor types allows some interesting conclusions with respect to the effect of these factors.

4

CHAPTER 1: INTRODUCTION

Chapter 8 describes a number of bulk property, individual property and microscopic analysis techniques used for characterisation of granules. Bulk property tests include shear strength and EPS extraction via cation exchange resin. Microscopic tests include fracturing of whole granules prior to SEM examination and fluorescent in-situ hybridisation (FISH) probing with three probes simultaneously to indicate areas of the four main intracellular conversion processes (acidogenesis, acetogenesis, methanogenesis). In Chapter 9, the bulk properties of the anaerobic granules is described. The cannery fed granules had the best bulk properties, the brewery fed granules were adequate for high-rate applications while the complex fed granules had the poorest properties. In Chapter 10, microstructure and ecology of the granules is presented using a variety of microscopic techniques. The multiple probe FISH method provided largely unambiguous results. All four granules had dominant structural features that were not predicted by the classical growth models presented in Chapter 7. This indicates that the influences caused by the constraints of full-scale operation have a high influence on structure and ecology.

Contributions This thesis provides contributions to the knowledge and application of anaerobic treatment processes in the following areas: • Expansion of a structural model for soluble protein by addition of hydrolysis and long chain fatty acid oxidation. • Development of a stepwise parameter estimation procedure for full-scale systems. • Prediction of the behaviour of particulate solids in the system. • Comparative analysis of granule properties from different full-scale reactors allowing novel conclusions about granule suitability and application. • Investigations of granule microstructure and ecology leading to discussion of granule life cycle, environmental constraints, and influences of structure on desirable properties.

References Barber, D. J. and Stuckey, D. C. (1999). “The use of the anaerobic baffled reactor (ABR) for wastewater treatment: a review.” Wat. Res. 33(7): 1559-1578. Costello, D. J. (1989). Modelling, Optimisation and Control of High-Rate Anaerobic Reactors. PHD Thesis Chemical Engineering. Brisbane, Australia, The University of Queensland. Fang, H. H. P., Chui, H.K., Li, Y.Y. and Chen, T. (1994). “Performance and granule characteristics of UASB process treating wastewater with hydrolyzed proteins.” Wat. Sci. Tech 30(8): 55-63. Hendriksen, H. and Ahring, B. (1996). “Integrated removal of nitrate and carbon in an upflow anaerobic sludge.” Wat. Res. 30(6): 1451-1458. Lettinga, G. and Hulshoff Pol, L. W. (1991). “UASB-process design for various types of wastewaters.” Wat. Sci. Tech 24(8): 87-107. Lettinga, G., van Velsen, A. F., Hobma, S. W., de Zeeuw, W. and Klapwijk, A. (1980). “Use of the Upflow sludge blanket (USB) reactor concept for biological wastewater treatment, especiallly for anaerobic treatment.” Biotech. Bioeng. 22: 699-734. Martinelle, M., Hult, K. (1994). Kinetics of triglyceride lipases. Lipases. P. Woolley, Petersen, S. Cambridge, Cambridge University Press: 363.

5

CHAPTER 1: INTRODUCTION

McInerney, M. J. (1988). Anaerobic hydrolysis and fermentation of fats and proteins. Biology of Anaerobic Microorganisms. Zehnder. New York, John Wiley & Sons: 373-416. Moosbrugger, R. E., Loewenthal, R.E. and Marais, GvR. (1990). “Pelletisation in a UASB system with protein (casein) as substrate.” Water SA 16(3): 171-178. Mosey, F. E. (1983). “Mathematical modelling of the anaerobic digestion process: Regulatory mechanisms for the formation of short-chain volatile acids from glucose.” Wat. Sci. Tech. 15: 209232. Ramsay, I. R. (1997). Modelling and Control of High-Rate Anaerobic Wastewater Treatment Systems. PhD Thesis Chemical Engineering. Brisbane, University of Queensland: 270. Rinzema, A. B., M., van-Knippenberg, K., Lettinga, G. (1994). “Bactericidal effect of long chain fatty acids in anaerobic digestion.” Wat. Environ. Res. 66(1): 40-49. Romli, M. (1993). Modelling and verification of a two-stage high-rate anaerobic wastewater treatment system. PhD Thesis, Dept Chemical Engineering. St Lucia, The University of Queensland: 206. Seghezzo, L., Zeeman, G., vanLier, J. B., Hamelers, H. V. M. and Lettinga, G. (1998). “A review: The anaerobic treatment of sewage in UASB and EGSB reactors.” Biores. Tech. 65(3): 175-190. Speece, R. E. (1996). Anaerobic Biotechnology. Nashville, Tennessee, Archae Press. Strous, M., VanGerven, E., Zheng, P., Kuenen, J. and Jetten, M. (1997). “Ammonium removal from concentrated waste streams with the anaerobic ammonium oxidation (anammox) process in different reactor configurations.” Wat. Res 31(8): 1955-1962.

6

Chapter 2

Anaerobic Digestion Processes This chapter describes anaerobic digestion and the implications of treating complex wastewater from a macroscopic point of view. This requires that the substrate types and major extracellular intermediates and products be described together with the digestion mechanisms. This chapter is divided into seven main sections. The first is an overview of anaerobic digestion pathways and processes. The next four describe the four main steps through which anaerobic digestion proceeds. The sixth section describes regulatory and inhibitory mechanisms common in anaerobic systems and the final section reviews current methods of modelling anaerobic digestion processes and outlines the objectives for modelling of complex systems.

2.1Overview of Anaerobic Digestion Anaerobic digestion is the biological degradation by a complex microbial ecosystem of organic and occasionally inorganic substrates in the absence of an oxygen source. During the process, organic material is converted to mainly methane, carbon dioxide and biomass. The nitrogen not utilised in growth is generally released as or reduced to ammonia. As both a chemical and biological process anaerobic treatment is completely different from aerobic treatment. Aerobic organisms have a ready electron acceptor in the form of oxygen (O2) while anaerobic digestion is strongly regulated by finding thermodynamically suitable electron acceptors. In many cases, ionic hydrogen (protons) or bicarbonate act as an electron acceptor to produce hydrogen gas or formate as product. Anaerobic treatment has advantages over aerobic treatment in that there are no power requirements for air supply, the methane can be used for energy production and there is a much lower sludge production. Aerobic degradation of organics yields much more energy than anaerobic degradation. Comparisons are deceptive, as the endproducts, and correspondingly free energy yields are different, but complete aerobic digestion of glucose to carbon-dioxide yields up to 38 mole ATP/mole glucose while anaerobic fermentation to mixed organic acids yields 2-4 mole ATP/mole glucose (Brock et al., 1994). Subsequent anaerobic digestion to methane and carbon-dioxide by different organisms yields much lower amounts of ATP.

7

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

However, anaerobic treatment may be more difficult to operate under stable conditions and is more susceptible to various shocks. It also requires higher substrate levels as a driving force and therefore may require higher retention times or different process configurations to achieve discharge quality effluent. Because there are no proven commercial nitrogen or phosphorous removal processes, high rate anaerobic treatment cannot remove these components except through biomass growth. This is often inadequate, especially where high protein levels in the substrate cause high product nitrogen or phosphorous levels in the effluent. Anaerobic digestion proceeds through a series of parallel and sequential processes by a variety of consortia. In contrast to aerobic digestion, where oxygen is an external electron acceptor, the gaseous and dissolved products (largely methane and carbon dioxide) have the same combined carbon oxidation state as the primary substrates. Thus anaerobic digestion is largely constrained by the need to find appropriate internal electron acceptors. When this is impossible, hydrogen ions or bicarbonate must be used and the thermodynamics of these reactions require that the products, elemental hydrogen or formate, be maintained at a low concentration (approximately 0.1 mbar and 0.19mM respectively, for syntrophic organic acid oxidation, Section 2.5.1). Components in anaerobic digestion can be grouped into primary substrates, which are in the feed wastewater, intermediates, and products. Primary substrates can be grouped into oil and grease, particulate carbohydrates and particulate proteins (particulate substrate) or long chain fatty acids (LCFA), soluble sugars and amino acids (soluble substrate). While intermediates can be a wide range of gaseous and soluble compounds, final products are normally methane and carbon dioxide. Anaerobic digestion processes consist of four main steps (Figure 2.1): • Hydrolysis is an enzyme mediated extracellular step which solublises particulates and substrates that cannot be directly utilised by the anaerobic organisms. • Acidogenesis or fermentation is the degradation of soluble substrates such as amino acids and sugars, which can be degraded largely without an external electron acceptor. The products are largely organic acids and alcohols. • Syntrophic acetogenesis is the degradation of fermentation products to acetate using hydrogen ions or bicarbonate as an external electron acceptor. This process is coupled with hydrogen or formate utilising methanogenesis, which maintains a low hydrogen or formate concentration. In many studies, the electron acceptor product (hydrogen or formate) is referred to as hydrogen as the hydrogen/bicarbonate couple is thermodynamically and stoichiometrically similar to formate. Therefore, in this thesis, the production of hydrogen during oxidative acetate producing reactions is referred to as obligate hydrogen producing acetogenesis (OPHA) and uptake of hydrogen or formate to produce methane is referred to as hydrogenotrophic methanogenesis. • Acetoclastic methanogenesis is the cleavage of acetate to methane and carbon dioxide via highly specialised organisms.

8

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Components in Feed

Particulate Proteins and Carbohydrates, Oil and Grease

Intracellular

Soluble Proteins, Sugars Fermentation/Acidogenesis Organic Acids, Alcohols

Intracellular

Syntrophic Acetogenesis/Methanogenesis Methane from Hydrogen Utilisers

Intracellular

Acetate

Increasing variety of catalyst (biomass)

LCFA1

Increasing organic substrate complexity

Hydrolysis Increasing succeptability ro inhibition

Extracellular

Methanogenesis

CH4, CO2

Figure 2.1: Basic digestion pathways of anaerobic digestion showing main substrates. The major intermediates carbon dioxide and water are not shown.

2.2 Hydrolysis of Particulate Substrates Particulate substrates are generally a conglomerate of a large number of different materials. To assist the reaction engineering approach of modelling, they have been grouped into carbohydrates, proteins and lipids. Organisms cannot take up non-soluble and particulate substrates that are too large to pass through the cell membrane. Therefore, extra-cellular enzymes are released that cleave polymers into smaller substrate molecules. This process is commonly referred to as hydrolysis. The somewhat confusing Oxford Concise Dictionary definition of hydrolysis is: “Any reaction in which a compound is decomposed and the hydrogen and hydroxyl (ions) of the water molecule become attached to separate products.” (Brown, 1993). That is, a substrate molecule is split in a reaction and water taken up. Hydrolysis here and in much biochemical engineering literature is used to describe the degradation of substrates with large or non-soluble molecules into fermentable substrate (eg Pavlostathis and Giraldo Gomez, 1991). Mainly enzymatic hydrolysis is addressed in this thesis.

9

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

This section of the review discusses each particulate substrate (carbohydrates, proteins and oil and grease, O&G), indicates special problems with specific hydrolases and then discusses general mechanisms of hydrolysis. 2.2.1 Particulate Carbohydrates Carbohydrates are named for their elemental components, as their empirical formula is generally of the form Cn(H2O)m. Correspondingly, the only contribution to chemical oxygen demand from carbohydrates is the carbon molecules, as the hydrogen can be fully oxidised by carbohydrate oxygen. Carbohydrates mainly originate directly or indirectly from plants. Generally, plant material is a mixture of cellulose (25%-60%), hemicellulose (15%-30%) and lignin (15%-20%) (Tong and McCarty, 1991). The remainder is tannins, soluble sugars and ash. The first two components are very similar and are digested anaerobically via similar mechanisms. Tong and McCarty (1991) list typical chemical compositions of lignocellulosic materials. Cellulose is a regular linear polymer of D-glucose with β-1,4 glucosidic bonds as shown in Figure 2.2. HOCH2 O

HOCH2 O OH

OH

O

HOCH2 O

O

OH OH

OH OH

OH n

D-Glucose

Cellulose

Figure 2.2: Glucose and Cellulose (Hart, 1992) Wood has a standard cellulose chain length of 5000-10000 glucose units (Tong and McCarty, 1991). Individual chains are bound into fibres by hydrogen bonds and Van der Wall’s forces. Although the individual bonds are rather weak, there are such a large number they provide enormous strength to the fibres. Crystalline cellulose has regular, strong hydrogen bonds whereas amorphous cellulose has irregular bonds. Amorphous cellulose has faster hydrolysis rates. Hemicellulose represents a group of linear and branched polymers of most natural minor sugars as well as uronic acids. Hemicellulose normally consists of less than 200 monomer units and is in an amorphous state. It is not well structured and is generally more easily hydrolysed. It can be solublised in weak acid or alkali conditions. Lignin is a dense three-dimensional polymer of aromatic molecules. It is hydrophobic and is linked by carbon as well as ether bonds. It is much more difficult to hydrolyse than either of the other polymers.

10

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Hydrolysis of particulate carbohydrates Depending on operating conditions degradation of carbohydrate based particulates may be rate limiting compared to methanogenesis (Pavlostathis and Giraldo-Gomez, 1991). Anaerobic organisms are not capable of hydrolysing un-degraded lignin (Chennowyth, 1991, Gujer and Zehnder, 1983). Aerobic bacteria degrade lignin by oxidatively cleaving with an extracellular enzyme, ligninase, with H2O2 as an oxidant. The presence of untreated lignin usually interferes with the hydrolysis of cellulose and hemicellulose. Cellulose is taken as an example substrate here. Degradation of cellulose from the fiber to glucose requires four major enzyme assisted steps as shown in Figure 2.3 (as many as 26 individual hydrolysis steps). Cellulose

Cello-oligosaccharides

Cellobiose

Glucose

Figure 2.3: Hydrolysis of cellulose fibres The cellulose fibre is reduced to the one-dimensional polymers, then the polymer cellobiose, and then glucose. Different enzymes catalyse each step. The enzymes are not isolated in their steps and synergism is actually necessary for degradation (Woodward, 1991). For example, Clostridium Thermocellum, one of the major cellulolytic anaerobic bacteria in thermophilic environments produce a cellulase matrix that catalyses the complete hydrolysis (Mayer, 1987). Many researchers (Gonzales, 1988, Humphrey, 1979, Tong and McCarty, 1991) regard the process as having two main steps: solublisation of the cellulose to cellobiose and formation of glucose. Hemicellulose is hydrolysed more easily than cellulose and most microorganisms that produce cellulases also produce hemi-cellulases (Tong and McCarty, 1991). Cellulases have a pH optimum of about 4-6 but since the optimal pH of each of the steps may be different, it is difficult to determine a single optimum. Tong and McCarty (1991) and Saha et al. (1985) have reviewed enzyme structure and mechanics. 2.2.2 Proteins Proteins are natural polymers of different amino acids joined by peptide (amide) bonds. The backbone of a protein is a repeating sequence of one nitrogen and two carbon atoms (Figure 2.4). H R O H R O H R O N C C N C C N C C H

H

H

Figure 2.4: Protein chain with amino acids linked by amide groups

11

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

20 amino acids are commonly found in nature. These are differentiated by the R group, which defines the function of the amino acid. A protein has three structural components: • Amino acid composition and sequence (primary structure). • The three-dimensional shape as set by bond angles and hydrogen bonds forms a helical shape in complex proteins. This is the secondary structure • The tertiary structure defines the macro-molecular shape as set by bonding between di-sulfide groups and to a lesser extent, other inter-R bonding. There are two major areas of importance for hydrolysis processes. Amino acid composition (primary structure) affects the products. The tertiary structure defines the proteins as either fibrous or globular. Fibrous proteins are structural materials such as keratin, which is protective and collagen, which is generally connective. Globular proteins are often chemically functional and act as enzymes, hormones, transport proteins or storage proteins. Hydrolysis of particulate proteins Hydrolysis of proteins is regarded as the rate-limiting step of protein degradation (Pavlostathis and Giraldo-Gomez, 1991). Protein structure is one of the main factors in rate of hydrolysis. Globular proteins are rapidly hydrolysable while fibrous proteins are difficult (McInerney, 1988). In general, all proteins apart from the most rigid of keratin (such as the outer layer of hair and fingernails) are hydrolysable (Figure 2.5).

Figure 2.5: Cow hair from anaerobic reactor showing intact keratin (A) compared with degradation of collagen centre via anaerobic organisms (B) (Photo by Author). There are three main groups of proteases: serine, metallo and acid proteases which have alkaline (8-11), neutral (6-8) and acidic (4-6) pH optimums respectively. The triggers for enzyme production vary widely. Some clostridia secrete in growth phase, stationary phase, and under stress. Enzyme production may be suppressed when readably biodegradable substrates such as glucose or amino-acids are supplied (Ramsay, 1997). 2.2.3 Fats and Oils Lipids are glycerol bonded to long chain fatty acids (LCFA), alcohols and other groups by an ester or ether linkage (Brock et al., 1994). Fats and oils have all the alcohol groups esterified with fatty acids as shown in Figure 2.6 and these form the

12

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

bulk of glyceridic material in mixed oils and fat with other glyceridic compounds usually a result of processing. CH 2 -O- Fatty Acid

CH 2 OH CH OH

CH-O- Fatty Acid

CH 2 OH

CH 2 -O-Fatty Acid

Glycerol

Triglyceride

Figure 2.6: Glycerol and Triglycerides While there are many other factors, the melting point of the mixed esters is largely affected by the number of double bonds and the length of the fatty acid chain. As saturation of the fatty acids in the lipid increases, the melting point increases. Unsaturated oils such as soybean, rapeseed and cottonseed are commercially hydrogenated to produce margarine and other spreadable vegetable fats (Austin, 1984). The only differentiation between fats and oils is the melting point. Hydrolysis of oils is normally more rapid than fats due to the higher level of emulsification and hence higher specific surface area. Table 2.1 shows the relative fatty acid concentration of several common fats and oils. Long chain fatty acids (LCFA) Fatty acids are carboxylic acids with a hydrophilic acid group at one end and a hydrophobic alkyl group at the other end. Because of this, they normally act as cell boundaries for simple organisms. In larger organisms they are normally found as long term energy storage or thermal insulation as extra-muscular fat cells (animals) or seeds (plants). This is the largest source of environmental lipids. The number of inter-carbon double bonds in the chain differentiates fatty acids. Those with no double bonds are unsaturated, single double bonds mono-unsaturated and those with two or more, poly-unsaturated fatty acids. Gunstone (1996) made four main generalisations about naturally occurring fatty acids. (i) Natural fatty acids are mostly straight-chained molecules with an even number of carbon atoms. Chain lengths vary from 2 to 80 carbon atoms although the norm is 12 to 22. (ii) Monounsaturated fatty acids are normally olefinic compounds and have the double bond in the cis (Z) configuration in one of a number of preferred positions. This is most commonly in the ∆9 (9 carbon atoms from the carboxyl group) or n-9 (9 carbon atoms from the methyl group). (iii) Polyunsaturated fatty acids are mainly polyolefinic with methylene interrupted arrangement of double bonds having cis (Z) configuration as in Figure 2.7. COOH Figure 2.7: Polyolefinic fatty acid with methylene interrupted cis double bonds.

13

Butyric Valerate Caprylic Capric Lauric Myristic Myrstoleic Palmitic Palmitoleic Stearic Oleic Linoleic Linolenic Arachidic Arachidonic Clupanadonic Lignoceric

4:0 5:0 8:0 10:0 12:0 14:0 14:1 16:0 16:1 18:0 18:1 18:2 18:3 20:0 20:2 22:4 24:5

C3H7COOH C4H11COOH C7H15COOH C9H19COOH C11H23COOH C13H27COOH C13H25COOH C15H31COOH C15H29COOH C17H35COOH C17H33COOH C17H31COOH C17H29COOH C19H39COOH C19H35COOH C21H35COOH C23H37COOH

Formula 10 2 78 7 1 -

Olive 4.0 2.0 1.0 2.5 3.0 10.0 25.0 11.0 28.5 2.5 -

Butter 0.6 22.9 2.2 24.7 49.7 -

Cotton 8.3 5.4 24.9 52.7 7.9 0.9 -

Soybean 7.5 3.5 46.3 42.0 0.5 0.2

Corn 0.2 1.1 44.0 0.1 4.5 39.1 10.1 0.4 0.4 -

Palm 7.0 16.0 17.0 1.0 27.0 20.0 12.0 -

Menhaden 8.0 1.5 12.1 15.0 2.3 33.4 9.0 8.2 10.5 -

Whale 6.5 4.5 20.9 17.4 50.6 0.1 -

Linseed 8.0 7.0 48.0 17.5 8.8 2.0 6.0 2.5 -

Coconut

Beef Tallow 3.0 28.0 24.0 42.0 2.0 -

1.0 0.2 28.0 3.0 13.0 46.0 6.0 0.7 2.0 -

Lard

14

1. Source: Austin (1984), Hasenhuettl (1991). These references (and many others) did not give the basis for composition. However, comparison with some data in Hui et al. (1996) indicated that the basis was %w/w. 2. Number of carbon atoms:number of double bonds.

Acid

Structure2

Table 2.1: Fatty acid composition of various common oils and fats (%w/w)1

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

(iv) Although fatty acids rarely have distinctive structural features apart from the carboxyl group and double bonds, acids are known which contain hydroxy, keto or epoxy groups. Free fatty acids often disappear from solution within 24 hours and adsorb onto the solid phase (Rinzema et al., 1994). Inhibition of anaerobic organisms by LCFA Rinzema et al. (1994) indicated that because fatty acids adsorb onto organic solids, the inhibition mechanism is to block transfer of substrate into the cell. This particularly will inhibit organisms utilising active transport. It would then follow that the inhibition is a function of the LCFA:biomass or LCFA:surface area ratio. Hwu et al. (1996) found that the inhibition was strongly surface area related on a variety of sludges with an oleate IC50 (50% loss of relative activity) of 0.26-3.34 mM at a pH of 7.0. The most resistant type was granular sludge developed on milk fat, which indicated acclimatisation was possible. Acetoclastic methanogens were the most strongly inhibited, followed by hydrogenotrophic methanogens, OHPA, and acidogenic organisms. In contrast, Rinzema et al. (1994) found that LCFA concentration was more important than LCFA:Biomass ratio for oleic acid, having a lethal effect on methanogenic organisms at 6.7-9.0 mM. After toxic inhibition, the reactor required 150-800 hours recovery. Given that Hwu et al. (1996) had extensive analysis and further developed experimental evidence of biosorption (Hwu 1997), surface related inhibition is probably the key mechanism. Hydrolysis of triglycerides Hydrolysis is catalysed by long-chain fatty acid ester hydrolases, called lipases. These act at the lipid-water interface in enzymatic hydrolysis to degrade the insoluble reactant to soluble products. There is little work on degradation of lipids in anaerobic environments compared with carbohydrate and protein substrates. Most of this has been focused on the rumen, reviewed by McInerney (1988). There are three main products from the hydrolysis of fats. These are non-fatty acid products (mainly glycerol), unsaturated fatty acids, and saturated fatty acids. In the rumen, the non-fatty acid components are degraded and unsaturated fatty acids hydrogenated but saturated fatty acids will not be oxidised (Ratledge, 1994), probably due to the high partial pressure of hydrogen. Lipase production can be stimulated by the presence of both triglycerides and by fatty acids (Finnerty, 1988). It is not known whether the main source of the lipase is the fatty acid utilising organisms (Brockerhoff and Jensen, 1974, Ratledge, 1994) or organisms which can utilise glycerol and hydrogenate unsaturated fatty acids (McInerney, 1988). One particular characteristic of lipases is increased activity with insoluble rather than soluble lipids. Martinelle and Hult (1994) indicated that the activity of lipases increases greatly when the concentration of triglycerides reaches saturation and forms a second phase. The lipases are adsorbed at the interface. Because there is an

15

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

adsorption mechanism, combined reaction and adsorption rate may be dependent on surface area of the insoluble triglycerides. Some proteins that affect surface tension can inhibit adsorption. Verger et al. (1991) found lipase adsorption was inhibited by 50% when an inhibitory protein covered 5%10% of the surface. This is especially of interest when dealing with complex mixed wastewater where there are large amounts of varied proteins. Bacterial lipases can be divided into three main types; Non-specific lipases, 1,3specific lipases, and fatty acid specific lipases (Finnerty, 1988). Non-specific lipases can hydrolyse any fatty acid triglyceride regardless of structure, acting at any of the fatty acids. These can completely hydrolyse the ester bonds acting equally at all alcyl sites. 1,3-specific lipases can only act at the outside bonds of the triglycerides, yielding 1,2diacylglycerols and 2-monoacylglycerols. These glyceride esters are unstable and undergo acyl migration to 1,3-diacylglycerol and 1-monoacylglycerol. Subsequently, these can be degraded further by the 1,3-specific lipase to glycerol and free fatty acids. Fatty acid specific lipases catalyse the removal of a specific fatty acid, preferentially removing cis-∆9-monounsaturated fatty acids. Other fatty acids are degraded very slowly, especially those containing an additional double bond between ∆1 and ∆9. Lipid hydrolysis is generally regarded as being faster than carbohydrate or protein hydrolysis (Pavlostathis and Giraldo Gomez, 1991) and since fatty acids are not degraded in the rumen, it is often assumed that fermentation is the rate limiting step in degradation of lipids (McInerney, 1988, above reference). Others (Ratledge, 1994, Martinelle and Hult, 1994) have indicated that the growth of the bacteria is dependent on the pH optimum of the lipase (usually 6-7), stirring speed and particle size, indicating that hydrolysis may be rate limiting. Also, many of the original references comparing kinetic values assumed ruminant conditions. Under these conditions, the partial pressure of hydrogen prevents degradation of LCFA. 2.2.4 General Mechanics of Hydrolysis There are three main mechanical pathways for release of enzymes and hydrolysis. (a) The organisms secrete enzymes to the bulk liquid where it adsorbs onto a particle or reacts with a soluble substrate (Jain et al., 1992). (b) The organism attaches to the particle, secretes enzymes into the vicinity of the particle. The organism benefits from the soluble substrates being released (Vavilin et al., 1996a). (c) The organism has an attached enzyme which may double as a transport receptor to the interior of the cell. This method requires the organism to adsorb onto the surface of the particle.

16

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Tong and McCarty (1991) indicated that cellulolytic enzymes could take both forms ‘a’ and ‘c’ but I have not seen references to attached proteolytic or lipolytic enzymes. Fats are hydrophobic while the fatty acid and glycerol degrading bacteria are hydrophilic (Thaveesri et al., 1995) so it is unlikely they can attach unless they have a hydrophobic receptor enzyme. The optimal hydrolysis method depends on reactor configuration, liquid velocities and solids concentration and retention time. For example in a high solids digester with long retention times, ‘b’ or ‘c’ are preferred as attached growth will outcompete those organisms in the bulk liquid. Where solids retention time is slow (due to higher liquid velocities) and the solids concentration is low, ‘a’ would be preferred. A high rate reactor is in the second category. The number of processes required for bulk liquid enzymatic degradation are as follows: (Figure 2.8)

2. Transport to bulk

4. Adsorbtion of enzyme onto surface 5. Reaction

6. Transport of product to bulk

1. Production of enzyme

7. Deactivation of enzyme 3. Diffusion from bulk to particle

Figure 2.8: Hydrolysis of particulate substrates by bulk enzymes. Any of these steps may be rate determining under various conditions. Type ‘b’ and ‘c’ hydrolysis have similar issues as production, adsorption, reaction and deactivation all still occur. The organisms avoid diffusion by growing on the particle. 1. Production of enzyme from organisms Enzymes are normally assumed to be produced at a rate proportional to the production of biomass (Humphrey, 1979, Jain et al., 1991). This rate, however, is dependent on a wide range of environmental factors. Organisms may increase enzyme production during conditions of growth, peak or when substrate levels are low. Soluble substrate can cause the organisms to stop producing enzymes (Ramsay, 1997). Enzyme production may also be stimulated by addition of insoluble substrate or an increase in substrate surface area (Martinelle and Hult, 1994). 2 and 3: Transport of enzyme from organisms to bulk and enzyme to particle. This depends on the size of the biological floc or particle (since it may need to diffuse through the floc), liquid velocities, temperature and bulk enzyme concentration. In low solids systems with reasonable liquid velocities the film will be reasonably small

17

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

and diffusion should not control rate. In dense stagnant systems however, transport may be rate limiting. 4. Adsorption onto Particle. Adsorption is dependent on number of adsorption sites (area), enzyme concentration and enzyme adsorption characteristics. Jain et al. (1991) proposed there were many sites and adsorption is unlikely to be rate controlling. However, researchers have shown that surface area has a strong effect on the rate of enzymatic hydrolysis of cellulose (Walker and Wilson, 1991), lipids (Martinelle and Hult, 1994) and proteins (Ramsay, 1997). Sanders et al. (1999) compared hydrolysis rates on three different sizes fractions of starch and found a strong dependence on size. However, the reaction of starch to glucose is rapid and this study found the bacteria colonised the entire surface of the particles. 5. Reaction Degradation of pure substrates is usually a linear progression of reactions (as seen with cellulose and lipids above) and each of these steps needs to be treated differently. Gonzalez et al. (1989) investigated the stepwise degradation of cellulose. Each degradation step for pure substrate needs to be modelled using competitive product inhibition Michaelis-Menten kinetics. The most accurate model was the one that considered all three hydrolysis reactions as well as competitive bypasses (though this may have been because there were so many parameters to fit the data). This was in a system where there was no degradation of product. In an anaerobic reactor, product concentration is almost zero and there should be little inhibition. This reduces it to first order. For degradation of complex substrates such as aerobic sludge, most have used first order kinetics with good results (Shimizu et al., 1992, Gujer and Zehnder, 1983, Pavlostathis and Giraldo-Gomez, 1991, Pavlostathis and Gossett, 1986). 6. Transport of product to bulk A healthy reactor normally has minimal amounts of product in the bulk liquid (Jain et al., 1992) and as the substrate particle size is small, diffusion should be rapid. Before substrate diffusion rate becomes rate limiting, the concentration of product will probably inhibit enzymatic degradation (Gonzalez et al., 1989). 7. Deactivation of enzyme This occurs either because the enzyme complex is damaged from substrate not detaching or it is degraded by proteases.

2.3 Acidogenesis/Fermentation Acidogenesis if the first energy yielding step. Because LCFA require an external electron acceptor for oxidation, degradation of LCFA is covered in OHPA

18

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

(Acetogenesis). The main substrates for acidogenesis are soluble saccharides and proteins. 2.3.1 Soluble Carbohydrates The hydrolysis of soluble sugars to sugar is very rapid compared with hydrolysis of particulate carbohydrates (Tong and McCarty, 1991). Therefore, from a reaction engineering point of view, the degradation and regulation of soluble polysaccharides can be taken to be the same as that of the monomers. Fermentation processes of C5 and other C6 monosaccharides are very similar to glucose and this is taken as model substrate. C5 monosaccharides produce C3 and C2 alkyl acids with no carbon dioxide. For example, ethanol is produced instead of lactate (Table 2.2). Conversion of glucose to fructose is one of the first steps of the EMP pathway (Brock et al., 1994). The Embden-Meyerhof-Parnas (EMP) pathway is of central importance for generation of ATP in anaerobic organisms (Brock et al., 1994). There are two main steps during the EMP pathway: oxidation of glucose to pyruvate (an electron producing step) and reaction to products (which is normally electron consuming). From pyruvate, there are a number of possible products. Despite recent reviews predicting a large variety of possible products (eg Grasius et al., 1997: 14 products), products in mixed culture systems are largely constrained by environmental conditions. For example, production of propionate only (equation (d) below) is not normally observed as it is coupled with an oxidation reaction such as (f) to give (a). References to production of propionate only, even in pure cultures have not been seen though the key intermediate to propionate, succinate is produced by the organism Fibrobacter Succinogenes (Matheron et al., 1997). Table 2.2: Possible products from acidogenesis of glucose. Products

Reaction

Acetate, Propionate 3C6H12O6 4CH3CH2COO-+2CH3COO-+2HCO3++8H+ Lactate C6H12O6 2CH3CHOHCOO-+2H+ Ethanol C6H12O6+2H2O 2CH3CH2OH+2HCO3-+2H+ Propionate C6H12O6+2H2 2CH3CH2COO-+2H++2H2O2 Butyrate C6H12O6+2H2O CH3CH2CH2COO-+3H++2HCO3-+2H23 Acetate C6H12O6+4H2O 2CH3COO-+4H++2HCO3-+4H23 ATP yields from Brock et al. (1994) and Costello (1989) Ethanol production in yeasts yields energy via pyruvate. Heterofermentative lactic acid bacteria produce acetyl phosphate as an intermediate to ethanol which is non-energy yielding (Brock et al., 1994). May be able to degrade via acetyl-CoA which will be energy yielding.

(a) (b) (c) (d) (e) (f) 1. 2.

ATP/ mole G1 4/3 2 02 ? 3 4

Production and acidogenesis of lactate Although lactate largely has not appeared in the literature as a product of mixed culture fermentation, Romli (1993) showed lactate was a very important extracellular intermediate during rapid dynamics of glucose overload. Lactate appeared in almost stoichiometric levels as a function of glucose overload concentration, indicating that

19

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

all glucose was fermented via lactate. Lyberatos, (personal communication 1999) had found the same results. The products of lactate fermentation are the similar to that of glucose (though with adjusted energy yields) and therefore regulation and relative concentrations of products may be similar. Regulation of products by environmental conditions The two main conditions regulating products are pH and H2 levels (Table 2.3): Table 2.3: The effect of environmental conditions on products of lactate and glucose acidogenesis. Condition Products

high pH/high H2 Propionate/acetate

high pH/low H2 Butyrate/acetate

Low pH/low H2 Ethanol/acetate

Low pH/High H2 more ethanol/acetate

At high hydrogen concentrations in mixed culture systems, experimentally observed products of glucose acidogenesis are largely acetate, propionate and lactate (Romli, 1993, Ren et al., 1996). This is the case in a heavily loaded acidification reactor as there are no biological hydrogen consumers (growth rate is low) and production is faster than gas stripping. As hydrogen levels are decreased, reactions (e) and (f) become more favourable respectively. This is simplified as the mechanisms are closely related to intracellular:extracellular H2 ratios and cellular NAD+/NADH turnover. This is explained in more detail in Mosey (1983) and Ruzicka (1996). In some pure cultures this is not true and of production of butyrate only with several bars over-pressure of hydrogen is possible (Clostridium butyricum, Cord-Ruwisch, personal communication, 2000). Ren et al. (1996) found that a pH of 4.6-4.8 caused equimolar production of ethanol and acetate with no propionate production, even at high hydrogen levels. Ethanol production was probably favoured as it reduced terminal acid levels. Production of hydrogen gas would be slightly more favourable because of higher H+ levels. 2.3.2 Amino Acids Ramsay (1997) reviewed amino acid fermentation and much of this is summarised here. There are 20 common amino acids found in proteins. Products are dependent on amino acid content, which is available in FAO, UN 1970. Amino acids can be degraded in two main ways: (a) As a Stickland oxidation-reduction paired fermentation. (b) As a single amino acid with an external electron acceptor. Stickland reactions Stickland reactions require one amino acid to act as an electron donor (oxidation) and the other to act as an electron acceptor (reduction). The products of the oxidation step are always NH3, CO2 and a carboxylic acid with one carbon less than the original chain as well as ATP. The reduction step results in a carboxylic acid with the same number of total carbon atoms as the original amino acid and NH3. This makes it relatively easy to predict products and Ramsay (1997) produced a spreadsheet that

20

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

calculated relative product yields from an amino acid content. This has been used in Chapter 6. The coupled oxidation/reduction reaction for alanine and glycine in Clostridium sporogenes is shown below (Figure 2.9).

Reduction

Oxidation H H3C C- COO-

Glycine (Acceptor)

Alanine (Donor) NAD

2

NH2

H3C- COONH2

NADH

H3C C

-

COO

-

O

Pyruvate, NH3

CoA

CO2 Acetyl CoA

NAD NADH

Pi CoA

Acetyl Phosphate ADP ATP H3C- COOAlanine + ADP + Pi

Acetic Acid H3C- COO-

Acetic Acid Acetate + ATP + CO2 + NH3 + 4H

2 Glycine + 4H

2Acetate + 2NH3

Figure 2.9: Coupled Stickland digestion of Alanine and Glycine (From Brock et al., 1994) Table 2.4: Digestion of Amino Acids Amino Acid Form of R Group Glycine Hydrogen Alanine Alkyl Valine Alkyl Leucine Alkyl Isoleucine Alkyl Serine Alcohol Threonine Alcohol Cysteine Sulphur containing Methionine Sulphur containing Proline Forms ring with amino Phenylalanine Aromatic Tyrosine Aromatic Tryptophan Aromatic Aspartic acid Carboxyl Glutamic acid Carboxyl Lysine Nitrogen containing Arginine Nitrogen containing Histidine Nitrogen containing From: Ramsay (1997); Hart (1991).

21

Donor/Acceptor/Uncoupled Acceptor Donor Donor Donor/Acceptor Donor Donor Donor/Acceptor Donor Donor Acceptor Donor/Acceptor Donor/Acceptor Donor/Acceptor Donor Donor Donor Donor Uncoupled

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Amino acids can act as electron donor, electron acceptor or both (Table 2.4). Histidine is the only amino acid not able to be digested via Stickland reactions. The ATP yield varies according to amino acid content (Ramsay, 1997) but usually yield is approximately 0.5 mole ATP per mole amino acid (Andreesen et al., 1989). Degradation via Stickland reactions occurs rapidly compared to uncoupled degradation (Barker, 1981). Uncoupled reactions Uncoupled degradation of amino acids only occurs when there is a shortage in electron acceptors. For common mixed proteins and large pure proteins, these only account for 10% of amino acids fermented (Nagase and Matsuo, 1982). Products of amino acid fermentations The full spreadsheet of products is shown in Chapter 6. The main products from amino acid degradation depend upon whether pathways follow Stickland or uncoupled degradation. Stickland reactions normally produce volatile fatty acids up to valerate (C5) from non-aromatic amino acids. Aromatic amino acids produce aromatic intermediates such as phenol, cresol and indole derivatives which Ramsay (1997) found accumulated significantly as when hydrogen concentrations were high. Aromatic amino acids are low in abattoir wastewaters (See Chapter 6). Uncoupled degradation produces a range of products, as the pathways are less restricted. Sulphur containing amino acids will also produce methyl sulfide (Barker, 1961) which may allow an independent hydrogen sink via H2S. Carbon dioxide and hydrogen (from uncoupled degradation) are common gas products. 2.3.3

Other Fermentation Processes

This section is mainly intended to cover the conversion processes of other intermediates (eg minor fat and lipid hydrolysis products) that are not hydrogen producing. These include glycerol, unsaturated fatty acids and lactate. Fermentation of lactate was described with glucose in section 2.4.1. Fermentation of glycerol Zeng et al. (1996) reviewed possible degradation pathways of glycerol in Klebsiella pneumoniae. Glycerol is partially oxidised at each site. Zeng et al. (1996) assumed that the EMP pathway would be used after formation of pyruvate (as for glucose or lactate) while a coupled reaction would be used prior to this to allow for electron balancing (Figure 2.10). The major products found by Zeng et al. (1996) were acetate, lactate and 1,3propanediol. Succinic acid was found in small amounts but this will be rapidly fermented to propionate or acetate (Brock, 1994). Other pyruvic acid degradation products were also found (formate, 2,3-butanediol, ethanol; see Brock, 1994 for more details).

22

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Simplified equations are as follows: Glycerol Acetate CH2OHCHOHCH2OH + H2O→ CH3COOH + 2 ATP + 3H2 + CO2 Butyrate CH2OHCHOHCH2OH → ½CH3CH2CH2COOH + 1.5ATP + 2H2 + CO2 Lactate CH2OHCHOHCH2OH → CH3CHOHCOOH + 1 ATP + H2 1,3-propanediol CH2OHCHOHCH2OH + H2 → CH2OHCH3CH2OH + H2O Glycerol H2O 3-Hydroxypropionaldehyde NADH NAD

1-3-Propanediol

Dihydroxyacetone ATP ADP

Dihydroxyacetonephosphate ADP ATP

NAD 2NADH NADH

NAD

Phosphoenolpyruvate ADP ATP

CO 2

Succinic acid ADP

ATP

Pryuvate Acetic

Lactic

Butyric

Others

Figure 2.10: Degradation pathways of glycerol by K. Pneumoniae (Modified from Zeng et al., 1996) Therefore, apart from the initial reactions (i.e., those leading to pyruvate), the glycerol degradation pathway is very similar to that of glucose. Zeng et al. (1996) indicated that 1,3-propandiol will be produced during high hydrogen and utilised via glycerol as hydrogen drops. Hydrogenation of unsaturated fatty acids. Hydrogenation of unsaturated fatty acid bonds is known to occur in the rumen (McInerney, 1988) but has not been studied in anaerobic reactors. It is known that there is a large variation of species capable of hydrogenating unsaturated compounds in the rumen and it can be assumed it will probably occur in other environments. Hydrogenation involves transferring to a trans isomer from the normal cis isomer and subsequently hydrogenating remaining cis bonds until the trans bond is also removed. There may be two hydrogenating mechanisms. The first hydrogenates to a monounsaturate and the second hydrogenates the remaining bond. Figure 2.11

23

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

illustrates the major hydrogenation pathway of the linolenic (cis-cis-cis octadeca9,12,15-trienoic) acid (Dawson and Kemp, 1970). cis-cis-cis octadeca-9,12,15-trienoic Isomerase

cis-trans-cis octadeca-9,11,15-trienoic trans-cis octadeca-11,15-trienoic trans octadeca-11-trienoic

Stearic Acid Figure 2.11: bio-hydrogenation pathways for linolenic acid It is not known what the hydrogen source is but the FAD-FADH2 transporter could be used as this is used for hydrogenation in other pathways. Yield would be 2/3 ATP per double bond. Hydrogenation can also be coupled to hydrogen producing reactions in fatty acid oxidising organisms. Saturation of bonds may also reduce toxicity (McInerney, 1988).

2.4 Obligate Hydrogen Producing Hydrogenotrophic Methanogenesis

Acetogenesis

(OHPA)

and

Obligate hydrogen producing acetogenesis is acetate producing reactions that can only oxidise the substrate while reducing hydrogen ions to hydrogen or bicarbonate to formate. Thermodynamics of the reactions require that acceptor concentrations be maintained at a low level. All organic fatty acids and ethanol are degraded by OPHA. Substrates have been grouped into C4+ fatty acids and propionate by differences in pathways. Ethanol conversion is not covered but information can be found in Dolfing (1988). As formate and hydrogen/CO2 are stoichiometrically and thermodynamically similar (see section 2.5.4), the electron acceptor product from OHPA is referred to here as hydrogen. 2.4.1 Thermodynamics of Reactions Simultaneous production of hydrogen during acetogenesis and removal via methanogenesis is only thermodynamically possible in a narrow range of hydrogen concentrations (expressed as partial pressure). Table 2.5 and Figure 2.12 show this. The ∆G’ shown in Figure 2.12 is slightly misleading, since it shows a very high yield for palmitate. However, the molecular weight, COD and number of carbon atoms per molecule is also high and a more relevant ∆G’ (per carbon-mole) is shown in table 2.5.

24

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

-2 0 0 .0 0 Operating Levels -1 8 0 .0 0

G' at pH 7.0, 25C (kj/mol substrate)

-1 6 0 .0 0 -1 4 0 .0 0 Palmitate

-1 2 0 .0 0 -1 0 0 .0 0 -8 0 .00

H 2->CH4

-6 0 .00 -4 0 .00



-2 0 .00 Butyrate

0 .0 0 Propionate

2 0 .0 0 4 0 .0 0 6 0 .0 0 2

3

4

5

6

-log pH2 (atm)

Figure 2.12: ∆G’ for the reactions shown in Table 2.5 at different hydrogen partial pressures. Apart from hydrogen, concentrations are 0.1M HCO3-, and 1mM organic acids at pH 7. The shaded region shows the operating region for syntrophic acetogenesis of butyrate. Valerate is equivalent thermodynamically to butyrate. Formate would have a similar graph at H2 concentration x 1.93. ∆G0 values taken from Brock (1994) and ∆G’ calculated from ∆G’=∆G0-5.71 ln  [C] [D]  in the  [A]a [B]b    reaction: a A + b B ⇔ c C + d D. c

d

Table 2.5: Thermodynamics of reactions for fatty acid oxidising organisms. Substrate H2, HCO3Propionate Butyrate Palmitate

Reaction

∆G0

∆G’

∆G’

(kj/mole)

(kj/mole)

(kj/cmole)

-12.0 -14.9 -25.9 -117

-12.0 -5.0 -6.5 -7.3

4H2+HCO3-+H+ CH4+ 3H2O -135.6 CH3CH2COO-+3H2O CH3COO-+3H2+HCO3-+H+ 76.5 CH3CH2CH2COO- + 2H2O 2CH3COO-+2H2+H+ 48.32 CH3(CH2)14COO- + 14H2O 8CH3COO-+14H2+7H+ 402.4 ’ ∆G Calculated for pH of 7, pH2 of 1e-5bar, pCH4 of 0.7, 0.1M HCO3- and concentrations. 







25

1mM organic acid

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

2.4.2 Oxidation of C4+ fatty acids. There are three methods of fatty acid degradation. β-oxidation is the main method observed for anaerobic organisms (McInerney, 1988, Brock et al., 1994). Another method available to anaerobic organisms is ω-oxidation, which oxidises by carboxylation of the far methyl carbon. α-oxidation is aerobic degradation by sequential oxidative decarboxylation (Finnerty, 1988). During β-oxidation of fatty acids, pairs of carbon atoms are removed from the carboxyl group end in a cycle until the chain is reduced to acetyl-CoA. This is then oxidised to acetate. There are two preliminary steps and four main steps in the process as indicated in Figure 2.13. Each loop of the cycle yields 1/3 ATP. The initial ATP invested to activate the fatty acid is regained when the final acetyl-CoA is de-activated. Therefore a complete oxidation of palmitate (A large constituent of beef tallow) to acetate would require 7 cycles as follow to give the products shown in table 2.5 to give 7/3 ATP. A generalised reaction for C4+ is as follows: CH3(CH2)nCOO

(n/2+1)CH3COO-+n/2H+ + nH2 + n/6 ATP R-(CH2)n-CH2-CH2-COOH

FLP Receptor Assisted Transfer ATP Acyl-CoA Synthetase

β-Oxidation Cycle

CoASH

AMP

O R-(CH2)n-CH2-CH2-C-SCoA 2/3 ATP

Acyl-CoA Dehydrogenase AMP

H2 O

R-(CH2)n-CH-CH-C-SCoA H2O

Enoyl-CoA Hydratase (L)OH

O

R-(CH2)n-CH-CH2-C-SCoA NAD+ 3-Hydroxyacyl-CoA Dehydrogenase

NADH O

O

R-(CH2)n-C-CH2-C-SCoA CoASH 3-Ketoacyl-CoA Thiolase O

O

R-(CH 2)n-C-SCoA + CH3-C-SCoA

AMP ATP CoASH CH3COOH Acetic Acid

Figure 2.13: β-oxidation cycle in anaerobic organisms (From Finnerty, 1988, Ratledge, 1994 and Hamilton, 1988).

26

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Each unsaturated bond adds 2/3 ATP and removes 1H2 from products. Odd chained organic acids such as valerate will produce a propionate from the final cycle. The partial ATP yields compare favourably with the ∆G’ given in table 2.5 assuming that free energy of 40-50 kj/mole substrate is required per mole of ATP (Thauer and Morris, 1984, Brock, 1994). LCFA’s are slightly more energy rich compared to VFA’s on a per-carbon basis but because of difficulties involved with adsorbing the fatty acid and production of hydrogen, it is often slower to degrade. Since β-oxidation requires a large amount of molecular reduction of perhaps FADH2 and certainly NADH + H+, it will be inhibited thermodynamically by hydrogen. Fermentation of fatty acids has long been regarded as very slow (McInerney, 1988). However, these studies have often concentrated on the rumen where hydrogen partial pressure is high and degradation therefore inhibited. The kinetic data used in this thesis (See Appendix C) indicates that LCFA oxidation is a slow process on a percarbon basis. Significant steps: (i) Transport and Activation: Fatty acids up to C11 may be able to diffuse through the cell membrane but those above that require adsorption on to a cell wall transport protein. Before β-oxidation commences activation by CoA is required. This has the added effect of detoxifying the fatty acid (Ratledge, 1994). (ii) Acyl-CoA Dehydrogenation: In aerobic organisms this is via FAD-FADH2 but the mechanism for anaerobic organisms is not known. Hamilton (1988) noted that hydrogenation of the alkyl bond is energy requiring even at 0.1 µM H2 (approximate levels during syntrophic association) and therefore cannot proceed via NAD+-NADH. Therefore, 2/3 ATP is hydrolysed to assist electron transport. (iii) Utilisation: the final cyclic reaction yields acetyl-CoA and an alkyl-CoA that is two carbon atoms shorter than the original fatty acid. The final reaction then yields either two acetyl-CoA molecules (acetate as an end product) or a propionyl-CoA and acetyl-CoA (acetate and propionate are end products). As noted previously, most natural LCFA’s are even chained and therefore acetate should be the major product. Acetate accumulation may also produce non-competitive inhibition or competitive accumulation with butyrate as the alternative product via acetyl-CoA. β-Oxidation of unsaturated fatty acids. Most monounsaturated fatty acids have the double bond in the ∆9 position (9 carbon atoms from the carboxyl group. This means that after 3 cycles, the double bond will be in the ∆3 position in the cis configuration (Figure 2.14). For the cycle to continue, the double bond needs to be in the trans configuration at ∆2 configuration. To overcome this a cis-∆3-trans-∆2-enoyl-CoA isomerase switches the bonds. The

27

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

double bond removes the acyl-CoA dehydrogenase and therefore 2/3 mole ATP and one hydrogen mole less is produced. This is reviewed in Ratledge (1994).

CH3(CH2)n-CH=CH-CH2-CH2-(CH2)5-CO-S-CoA 3 Turns of β-Oxidation cycle CH3(CH2)n-CH=CH-CH2-CO-S-CoA 3-cis-enoyl-CoA cis-∆3-trans-∆2-enoyl-CoA isomerase CH3(CH2)n-CH2-CH=CH-CO-S-CoA 2-trans-enoyl-CoA β-Oxidation resumes with hydratase Figure 2.14: Variation of β-Oxidation for double bonds (Ratledge, 1994) β-Oxidising organisms Syntrophomonas wolfei has been nominated as the main syntrophic organism degrading akyl acids up to C7 (McInerney et al., 1981). Syntrophomonas sapovorans (Roy et al., 1986) and Syntrophomanas wolfei subsp (Lorowitz et al., 1989) have been identified as mesophilic LCFA oxidising organisms while Thermosyntropha lipolytica has been identified as a thermophilic LCFA oxidising organism (Svetlitshnyi et al., 1996) 2.4.3

Propionate Oxidation

The digestion pathway of propionate is thought to reverse the digestion pathway to pyruvate and hence to acetyl-CoA and acetate via the methylmalonyl-CoA pathway (Stams and Plugge, 1995). Radioactivity studies using 14C markers found the carboxyl group was invariably converted to carbonate while the methyl and carboxyl group of the resulting acetate had equal radioactivity regardless of whether [2-14C] or [3-14C] propionate was used (Mackie and Bryant, 1995). This is consistent with the randomising step from malate to oxaloacetate in the above pathway. The free energy of reaction is so low that even at pH2 of 10-5 bar ATP production and hence yield would be very low. Based on free energy yield, ATP production per propionate oxidised could range from 1/3 ATP to 1/10 ATP. The most commonly reported syntrophic propionate oxidising organism is syntrophobacter wolinii (Stams and Plugge, 1995, Dubourguier et al., 1987, Harmsen et al., 1996)

28

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

2.4.4 Hydrogen Utilising Organisms Form of electron carrier Earlier in this chapter, it was noted that the major electron transfer mechanism could be either via formate or hydrogen. Although thermodynamically and stoichiometrically, these are very similar, there are three major differences (Boone et al., 1989): • Hydrogen has a higher diffusivity. • Formate is more soluble. • Formic acid is a stronger acid than carbon-dioxide (H2). Therefore, when interspecies distances are short, hydrogen transfer will be faster and when distances are long, the greater solubility of formate allows a greater concentration gradient and therefore better transfer. Boone et al. (1989) indicated that when organisms are further than 10µm apart, formate was the main mechanism for electron transfer. As formic acid is a relatively strong acid, formate production may be decreased at low pH. Harmsen et al. (1996) showed that in granular sludge syntrophic organisms were located at distances of below 5µm. Therefore where organisms can be immobilised closely, such as in high rate systems, syntrophic colonies allow hydrogen transfer. This means bulk hydrogen concentration will have little relationship with intra-colony hydrogen concentration as most hydrogen transfer occurs within the colony. Any measurements of hydrogen or formate concentrations will not necessarily directly reflect the state of the reaction vessel (syntrophic culture) but may be used to indicate responses to disturbances (eg Cord-Ruwisch et al., 1997). Long-term increase of bulk hydrogen concentrations could cause accumulation of VFA’s due to a decrease in diffusion of hydrogen to the bulk liquid from the area of reaction. Apart from this, treatment of the syntrophic culture as a reaction system is largely unaffected as stoichiometry and thermodynamics are unaffected and hydrogen/formate may be in enzyme assisted equilibrium (Thiele and Zeikus, 1988). Also acetogens may waste electrons as either hydrogen or formate and hydrogen utilising methanogens can accept either (Boone et al., 1993). Hydrogen utilising organisms The major pathway of hydrogen or formate removal in mesophilic high rate reactors is methanogenesis. This occurs by activation of a carbon-dioxide molecule or formate molecule and successive hydrogenation of this complex. As a final step, methyl-CoM is formed and this is reduced to methane with a yield of 1 ATP/mole methane formed. None of the methanogenic archaea can utilise energy from substrate level phosphorylation and ATP is probably generated from a proton motive force (Boone et al., 1993). There are large numbers of Methanogenic hydrogen/formate utilising organisms within 5 orders in the archaea (Boone et al., 1993).

29

CHAPTER 2: ANAEROBIC DIGESTION PROCESSES

Another potential sink for hydrogen/formate is homoacetogenic bacteria (Diekert and Wohlfarth, 1994), which mediate parallel oxidation/reduction of carboxyl groups to form a methyl and a carbonyl group. These are complexed with CoA to form acetyl CoA, which is then reduced to acetate with a yield of 1 ATP. These may be important in acidogenic reactors, where a higher growth rate allows homoacetogens to outcompete hydrogenotrophic methanogens and in a heavily loaded acidogenic reactor Zhang and Noike (1994) found these organisms contributed 20-30% of acetate production. However, fixed biofilms and granules allow the growth of hydrogenotrophic methanogens, which have a threshold of 0.04 mbar hydrogen gas (Cord-Ruwisch et al., 1988). The threshold value of 0.43 to 0.95 mbar hydrogen gas (Cord-Ruwisch et al., 1988) for homoacetogens is also incompatible with propionate oxidation (0.1 mbar). Therefore the contribution of homoacetogens as a hydrogen sink in fixed film reactor and lightly loaded acidogenic reactors is probably negligible. Acetate production from hydrogen/CO2 and formate becomes more favourable psychrophilic conditions ( 300µm Perhaps outer layer of bacteria 100-150µm layer as above Centre dead

>3mm Outer layer may shrink due to reduced specific surface area. Granule will grow and eventually oversettle or disentegrate.

Figure 10.14: Life line of Yatala and Matilda Bay Granules

(c) Breakup or Oversettling As the granule increases in size, the specific surface area drops and the active annulus decreases. The centre of inactive organisms has a lower strength and becomes more porous as the granule becomes larger. Eventually the granule breaks up and pieces wash out. Alternatively, the granule becomes too large and settles to the bottom of the reactor where there is little substrate flux and disintegrates. Another possibility is the core disintegrating in granules with a very high residence. This then fills with gas and the granule floats. Spearwood (a) Nucleation Because of the sparse structure and the low degradation rate of proteins, the nuclei are likely to be flocs. This has been seen in the effluent of the reactors during startup. Because these flocs have a low settling velocity, many are washed out and therefore development of a granular bed is slow.

Nuclei forms as a floc

1.5mm Granule grows in irregular shape Most are disk shaped.

Growth is largely unregulated but from the size analysis (Chapter 9), the granules are likely to be spherical up to 1.5mm. After this, the granule can grow in many Figure 10.15: Life cycle of Spearwood granules shapes but largely forms disks, perhaps because this maximises presented surface

193

CHAPTER 10: STRUCTURE AND ECOLOGY OF ANAEROBIC GRANULES

area (to flow) and minimises shear moments around the thin granule. Also as granules grow larger, they accumulate the dense packets of bacteria seen in Figure 10.8, perhaps because these require a long residence time. (c) Removal Granules may be removed from the reactor or altered in several ways. As the granule becomes larger (as a disk), the moment due to shear around the axis will become large and the granule may split. Because of the larger void space, EPS extrusion may trap gas in the granule and cause it to float. A large irregular shape is more likely to wash out. 10.5.5 Desirable Properties The results of this section are summarised in Table 10.3 Table 10.3: Possible influence of structure and morphology on desirable properties Property Golden Yatala Matilda Bay Activity + Strength ++ + + Size Distribution and Settling Velocity ++ Resistance to Shocks and Toxins ++ + + As Seed Sludge -

Spearwood + +/-

Activity All the granules improve activity over distributed organisms or flocs by interspecies transfer, especially syntrophic hydrogen/formate transfer. However, the structure and growth pattern of the brewery granules leave much of the granule unused, especially in the larger granules, which make up a significant part of the volume. The Spearwood granules have an advantage over the other granules because of their sparse structure (which increases diffusion and convection) but the low cell density causes a lower specific activity/volume. Golden Circle has approximately 40%-50% of the granule utilised (by image pixel analysis) and the structure is very dense. I would estimate that there are more organisms/volume than in Spearwood. However, because the high density and shape of the Golden Circle granules, there is probably little convection and diffusion is slow. Strength This does not seem to be a function of morphology as Yatala granules lack an outer layer of bacteria but have similar strength to Matilda granules. Estimated from the granule density as seen in SEM, Golden Circle granules should have the highest shear strength followed by Yatala and Matilda Bay granules while Spearwood granules will have a low strength. The larger granules of Yatala and Matilda Bay will have a very low compressive strength as most of the granule consists of dead material. This compares well with the shear strength and density measurements made in Chapter 9 where it was found that Golden Circle had high density and strength followed by Yatala and Matilda Bay granules and Spearwood granules had the lowest density and strength.

194

CHAPTER 10: STRUCTURE AND ECOLOGY OF ANAEROBIC GRANULES

Size distribution and settling velocity An optimum size distribution should be narrow, to define clear upper operational upflow velocities and large, to allow a high velocity. In the last chapter, it was shown that the densities were high enough to have less influence than size and shape. Here, Golden Circle has a clear advantage, as the size is limited to a maximum size, which is quite large and spherical, resulting in a high allowable settling velocity. However, the residence time needs to be low enough to achieve the required removal and avoid reactor overload. In the Golden Circle reactor, sludge settles too fast and causes gas surging through the packed bed. This is not a fault of the sludge but rather the reactor. It should have a lower cross-sectional area and an increased height to maintain the reactor volume with a higher upflow velocity. The expected washout sludge is also easier to handle, as it should consist of more hollow granules than fines. These are easier to remove in a DAF or settling chamber. The life cycle of Matilda Bay and Yatala produce a wide size distribution which results in a variety of settling velocities. However, the higher density and reasonably spherical shape produces relatively high velocities. Spearwood has a clear disadvantage here as there is a low driving force towards a spherical granule. This causes a low settling velocity. Resistance to shocks and toxins Golden Circle granules have a clear advantage here. The buffer zone protects the more delicate archaea from pH shocks and toxins and there is even a possibility that the bacteria on the exterior can metabolise the toxins. Matilda Bay and Yatala lack the buffer zone but their high density reduces diffusion into the granule. Spearwood has a disadvantage as the organisms are distributed and density is low, maximising diffusion and convection. Application as seed sludge Desirability of each of these as seed depends largely on the application. For example, Golden Circle sludge could be transported to a soluble polysaccharide plant (such as a sugar refinery) and be used at low pH with low levels of acclimatisation. In a complex wastewater plant, the lack of microbial diversity and specialised structure would make the sludge inappropriate. Spearwood has a slight advantage as general purpose sludge as it has a high diversity and the flexible structure allows slightly easier acclimatisation. Alpenhaar (1994) showed that protein based sludge should not be exposed to polysaccharides for fear of fluffy growths. Also, the low strength will cause some disintegration in shipping. Yatala and Matilda Bay would be adequate for any plant with a pH operating level of above 6.8 though the possible lack of molybdenum in the feed may cause a change in community.

195

CHAPTER 10: STRUCTURE AND ECOLOGY OF ANAEROBIC GRANULES

10.5.6 Implications for Design and Operation Pre-acidification and reactor pH The results from Yatala indicate that in-granule acidification is not necessary for development of effective granules in carbohydrate reactors with normal pH’s. However, in the case of Golden Circle, the growth of the external acidifying bacteria is essential for development of the buffer zone. Indeed, by a combination of low preacidification and low pH, better granules have been developed than would be seen in a reactor with conventional operating parameters. The main structural component in the Spearwood granules appears to be Methanosaeta in filamentous configuration and this indicates that pre-acidification in not essential. However, the only dense (microbe/volume) regions in the granules are the coccoid bacteria in more mature regions. These are only 20µm in diameter an probably do not have an overall effect on structural strength. A lower level of preacidification may result in a surface layer of bacteria growing on readily biodegradable proteins and polysaccharides. The lack of filamentous bacteria indicates that this shouldn’t cause fluffy growth as seen in Alpenhaar (1994). The results of Golden Circle indicate it may be advantageous to operate at a lower pH (with low pre-acidification). Further work with laboratory scale reactors should be done to optimise the setpoints. Reactor design This study has shown that the feed characteristics and operating set points strongly affect granule structure and consequently, settling velocity and activity. Reactor design is currently based on a fixed upflow velocity and operating volume, largely uninfluenced by feed characteristics. Reactor volume can be estimated by degradation rates of that substrate (there are precedents for most wastewater types) while upflow velocity (and therefore cross-sectional area) can be optimised for expected settling velocity. Thus a reactor treating Golden Circle’s wastewater would be tall with a small cross-sectional area (and consequent footprint) while Spearwood would have a large cross-sectional area. In the actual systems, plants, this is reversed. The proposed design would increase granule washout but granules can be settled and returned to the reactor. This also needs further work using laboratory or pilot scale equipment.

10.6 Conclusions Once again while no statistical or causative link could be established between feed type and granule structure, the cannery fed granule, brewery fed granules and protein fed granules had distinctively different structures. Both brewery fed granules had very similar structures. The use of three probes for FISH was excellent for overall community analysis and provided largely unambiguous results as follows: The cannery fed granules had an outer layer of densely packed presumptively acidogenic bacteria followed by a largely inactive buffer region, with a core of Methanosaeta and syntrophic bacteria in association with hydrogen utilising archaea. 196

CHAPTER 10: STRUCTURE AND ECOLOGY OF ANAEROBIC GRANULES

Driving forces for this granule type were probably the low external pH (which caused the buffer region) and a readily biodegradable primary substrate (which caused the external layer of bacteria). Methanosarcina like archaea were scattered through the buffer region and in the outer part of the centre. The structure and conditions that caused it ensured a strong granule with a narrow size distribution and therefore narrow settling distributions as large granules fill with gas and washout. The constant depth buffer zone also helps ensure a spherical shape. Yatala and Matilda Bay fed granules had similar structures except that Matilda Bay fed granules showed evidence of an outer layer of acidogenic bacteria. The granules were active in the outer 100µm-150µm and were probably limited by diffusion. The rest consisted of cellular debris and inorganic material, which could be regarded as support material. Because gas exhaust into the bulk probably encounters less resistance than into the centre, the granules probably have no upper size limit and grow until the granule disintegrates from high centre porosity or the granule oversettles to the base of the reactor. The ratio of Methanosaeta to other archaea was much lower than in the other granules. This perhaps indicates that the reduced carbon flow is passing through another major intermediate (other than acetate). A candidate is formate. One possible explanation is a much larger amount of molybdenum in the granules, which is necessary for and stimulates formate dehydrogenase (FDH) in syntrophic archaea. The Spearwood fed granules had a dispersed structure with the main structural component appearing to be clumps of Methanosaeta. Most of the granule appeared active but much less dense than the others. Presumptively acidogenic bacteria were much more diverse, probably because of the diversity in primary substrate. Because the structure may be unregulated due to the loose structure and lack of a rate limiting step, non-spherical shapes can form. Calculation of the Thiele modulus indicated that diffusion was rate limiting but the structure of this granule indicated that reaction may be rate limiting. This may mean that the reaction rates used (which were taken from the model) may be much lower than the in-granule rates or that diffusion is higher than estimated. This may indicate that convection is occurring through the granule or that the soluble primary substrate (soluble protein) is close to threshold values. The dispersed structure and non-spherical shape of the granules would assist increased diffusion rates. The structure affects the desirable properties of these granules in a number of ways and simple changes such a adjustment of pH (via decreased caustic dosing) or preacidification may alter system performance. Also, knowledge of how a wastewater type affects desirable properties and granule structure allows modified reactor design via change in reactor cross-sectional area and height to achieve an optimum balance between reactor volume and up-flow velocity.

References Alphenaar, A. (1994). Anaerobic granular Sludge: Characterisation and factors effecting its functioning. Agricultural College. Wageningen Agricultural College, Wageningen: 222. Blaut, M. (1994). “Metabolism of methanogens.” Antonie van Leewenhoek 66, 187-208. Boone, D. R., Johnson, R. L. and Liu, Y. (1989). “Diffusion of interspecies electron carriers H2 and formate in methanogenic ecosystems and its implications in the measurement of Km for H2 or formate uptake.” Appl. Environ. Microbiol. 55: 1735-1741. 197

CHAPTER 10: STRUCTURE AND ECOLOGY OF ANAEROBIC GRANULES

Boone, D.R., Whitman, W.B, and Rouviere, P. (1993). Diversity and Taxonomy of Methanogens. in Methanogenesis. Ecology, Physiology, Biochemistry and Genetics. J. G. Ferry. Chapman and Hall, New York. Cord-Ruwisch, R., Seitz, H. J. and Conrad, R. (1988). “The capacity of hydrogenotrophic anaerobic bacteria to compete for traces of hydrogen depends on the redox potential of the terminal electron acceptor.” Arch. Microbiol. 149: 350-357. Fang, H. H. P., Chui, H.K., Li, Y.Y. and Chen, T. (1995a). “Performance and granule characteristics of UASB process treating wastewater with hydrolyzed proteins.” Wat. Sci. Tech 30(8) , 55-63. Fang, H. P., Chui, H. and Li, Y. (1995b). “Effect of Degredation Kinetics on the Microstructure of Anaerobic Biogranules.” Wat. Sci. Tech. 32(8) , 165-172. Guiot, S. R., Pauss, A. and Costerton, J.W. (1992). “A structured model of the anaerobic granule consortium.” Wat. Sci. Tech 25(7) , 1-10. Harmsen, H. J. M., , Kengen, H.M.P., Akkermans, A.D.L., Stams, A.J.M. and de Vos, W.M. (1996a). “Detection and localization of syntrophic propionate-oxidizing bacteria in granular sludge by InSitu Hybridization using 16S rRNA-based Oligonucleotide Probes.” Appl. Environ. Microbiol 62(5) , 1656-1663. Harmsen, H. J. M., Akkermans, A.D.L., Stams, A.J.M. and de Vos, W.M. (1996b). “Population dynamics of propionate-oxidizing bacteria under methanogenic and sulfidogenic conditions in anaerobic granular sludge.” Appl. Environ. Microbiol 62(6) , 2163-2168. Matheron, C., Delort, A.M., Gaudet, G., and Forano, E. (1997). “Re-investigation of glucose metabolism in Fibrobacter succinogenes, using NMR spectroscopy and enzymatic assays. Evidence for pentose phosphates phosphoketolase and pyruvate formate lyase activities.” BBA Mol Cell Res 1355(1) , 50-60. Ramsay, I. R. (1997). Modelling and Control of High-Rate Anaerobic Wastewater Treatment Systems. Department of Chemical Engineering. University of Queensland, Brisbane: 270. Richardson, J. F. and Peacock, D.G. (1994). Coulson and Richardson’s Chemical Engineering Vol III: Chemical and Biochemical Reactors and Process Control. Elsevier Science, Oxford, UK. Sekiguchi, Y., Kamagata, Y., Nakamura, K., Ohashi, A. and Harada, H. (1999). “Flourescence in situ hybridization using 16S rRNA-Targeted Oligonucleotides Reveals Localization of Methanogens and Selected Uncultured Bacteria in Mesophilic and Thermophilic Granules.” Appl. Environ. Microbiol. 65(3) , 1280-1288. Schnürer, A., Zellner, G. and Svensson, B. H. (1999). “Mesophilic syntrophic acetate oxidation during methane formation in biogas reactors.” Fems Microbiol. Ecology 29: 249-261. Thiele, J. H. and Zeikus, J.G. (1988). Interactions between hydrogen and formate producing bacteria and methanogens during anaerobic digestion. in Handbook on Anerobic Fermentations. C. E. Erickson and Daniel-Yee-Chak-Fung. Marcel Dekker Inc., NY. White, W. B. and Ferry, J.G. (1992). “Identification of formate dehydrogenase specific mRNA species and nucleotide sequence of the fdhC gene of Methanobacterium formicum.” J, Bacteriol. 174, 4987-5004.

198

Chapter 11

Conclusions and Recommendations The research objectives of this thesis are: • Identify and investigate the critical conversion processes in anaerobic digestion of complex wastewater. • Describe these conversion processes in the form of a mathematical model. • Identify and study in detail a main limitation for the application of anaerobic treatment to complex wastewater and compare it to carbohydrate based wastewater. Based on the research undertaken the following conclusions may be made: 1. Degradation of solids in complex wastewaters is rate-limiting and most degradation occurs in the methanogenic reactor. Therefore all negative effects of solids and fats on granulation are enhanced and decreases in flow may not have an immediate influence on reactor load, as the system will degrade residual solids in the reactor. 2. Kinetic rates of soluble proteins are rapid. This means a controlled level of preacidification is probably not possible when influent stream also contains soluble proteins as the ammonium release will buffer the system sufficiently to avoid low pH inhibition of fermentation processes. Removal of the acidification stage is not compatible with pH, concentration and flow equalisation. 3. Conversion of organics in proteinaceous wastewater is different from carbohydrates in that stoichiometric rather than regulated yields of VFA's from fermentation could be used but is not fully predicted by the theoretical Stickland degradation pathway. Comparison of the stoichiometric yields of organic acids, hydrogen and carbondioxide with those expected from the amino acid content of the wastewater indicated a shift towards propionate from acetate and butyrate. This may be due to a change in pathway from Stickland reactions or may be the cumulative influence of the nonamino acid constituents of organics. It is probably not due to a larger amount of carbohydrate, as an increase in acetate production would also be expected.

199

CHAPTER 11: CONCLUSIONS AND RECOMMENDATIONS

4. The least desirable granules in terms of shear strength and settling velocity were sampled from a reactor fed proteinaceous wastewater. Granules fed with complex wastewater had a lower settling velocity, broader settling distribution and much lower shear strength than other granules. This means net granulation rate is slower, and the fines in the effluent are higher. Also, the shape of the granules is irregular which may cause greater break-up due to shear moment about the axis of the granule. 4. Granules sampled from a reactor fed proteinaceous wastewater had limited structural characteristics but a very diverse bacterial ecology. While the carbohydrate (cannery) fed and brewery fed granules had structures that appeared strongly influenced by diffusional limitations (and other environmental constraints), the protein granules consisted of large, loosely packed presumptively syntrophic colonies and bundles of filamentous Methanosaeta. There were relatively large dense colonies of (probably acidogenic) bacteria throughout the granule rather than in an outer layer as seen in the other granules. Therefore, reaction rather than diffusion may be rate limiting for all protein fermentation. As it was found that the kinetic rates of soluble protein as observed at a macroscopic scale was rapid (during modelling), the in-granule processes may be slower, or the soluble protein concentration at the lower threshold. 5. Both microstructure and ecology of the carbohydrate fed granules was strongly affected by wastewater type and environmental conditions: (a) Soluble carbohydrate (cannery) fed granules had an acidogenic outer layer followed by an inactive buffer region. The centre mainly consisted of syntrophic colonies and clumps of Methanosaeta. This is probably caused by the low inreactor pH. This driving force produces a granule with strongly defined size distribution in mature granules and the rapid degradation rates probably cause high density and settling rates. (b) Brewery fed granules had an active region in the outer 200-300 µm. The centre was inactive. Therefore, diffusion seems to be the main driving force. There were also much lower numbers of acetoclastic organisms, which may indicate another major methanogenic precursor instead of acetate.

Recommendations 1. Operational The flow of solids and the solids inventory within a methanogenic reactor has a major influence on process response and process stability. To maintain awareness of the process status and particulate biomass/substrate within the reactor, the following steps can be taken: (a) Monitor solids flow around the reactor in terms of particulate COD and volatile solids.

200

CHAPTER 11: CONCLUSIONS AND RECOMMENDATIONS

(b) Estimate solids inventory in the reactor by estimating first order decay and total COD of gas flow during periods of no hydraulic flow. Combined with volatile solids measurement, this will also indicate biological inventory. (c) With the above information, estimate particulate biomass and substrate washout from the reactor during normal load and high solids load. The model developed during this thesis is ideal for this application and an implementation of the full model or a simplified model could address these objectives very well and would supply further information such as approximate biological population distribution. 2. Design and Implementation Retention of solids in the methanogenic reactor and the low granulation rates indicate that development of a methanogenic reactor design specifically for complex wastewater is merited. Some of the features such a reactor should include are: (a) Biomass and particulate solids retention by an upper filter or gas separation mechanism. (b) Large surface area to minimise upflow velocity within the reactor. This may alleviate both solids washout and the effects of slow settling granular sludge. (c) Possibly, multiple compartments in series to decrease effluent solids and avoid blocking of distribution systems by mineral solids. Such a reactor concept would need extensive development and optimisation in laboratory or pilot scale units. 3. Research The literature review has shown there is a large number of factors that affect hydrolysis rates in practice such as surface area, soluble substrate inhibition, substrate type and surface tension. However, this is a big task and defining such a project well is critical. The EPS extractions described in Chapter 9 were largely inconclusive because it is difficult to locate the EPS or define its function. Further work here should concentrate on locating and analysis of functional EPS via molecular probes, micromanipulators, and EPS staining. The results from Chapter 10 indicate several areas for further research: (a) The carbohydrate grown granules may have a well-defined acidogenic community. Acidogenic organisms have received the least attention amongst anaerobic organisms as the diversity is so high. This offers an opportunity for better examination via a clone library and possibly, extraction and purification. (b) The two brewery fed granules may produce methane from another major methanogic precursor than acetate. This may have a number of strengths such as increased resistance to saline inhibition and pH. There may also be weaknesses such as the dependence on molybdenum and inapplicability to other reactors.

201

CHAPTER 11: CONCLUSIONS AND RECOMMENDATIONS

Molecular and laboratory scale studies of these sludges should be done to further define these opportunities. (c) The protein based sludge has a structure derived from very different driving forces to the others. In-situ studies in flow cells or laminar tubes with microsensors could offer insights into driving forces for structure and lead to further improvements in granulation and sludge performance.

202

Appendices on Accompanying CD-ROM Appendix A: Treatment of Logged Data Appendix B: Data Collected for Parameter Estimation and Validation Appendix C: Kinetic Parameters Appendix D: Model Library Files Appendix E: Data from Bulk Property Tests Appendix F: Additional CLSM and SEM Images