UNIVERSITE MONTPELLIER II
SCIENCES et TECHNIQUES du LANGUEDOC
UNIVERSITY of QUEENSLAND
SCHOOL of ENGINEERING
THESE
Réalisée en cotutelle entre l’Université Montpellier II et The University of Queensland
pour obtenir le grade de
DOCTEUR de L'Université Montpellier II
et
DOCTOR of PHILOSOPHY of The University of Queensland
Discipline : Génie des Procédés
Ecole Doctorale : Science des Procédés - Science des Aliments
Présentée et soutenue le 28 Septembre 2007 à The University of Queensland
par
Romain LEMAIRE
ETUDES FONDAMENTALES ET DEVELOPPEMENT DE PROCEDES
INNOVANTS POUR L’ELIMINATION BIOLOGIQUE DE L’AZOTE ET DU
PHOSPHORE DANS DES EFFLUENTS D’ABATTOIR.
Jury:
Dr. Philip Bond, Président, Australie
Dr. Nicolas Bernet, Directeur de Thèse, France
Prof. Zhiguo Yuan, co-Directeur de Thèse, Australie
Prof. Jürg Keller, Directeur adjoint de Thèse, Australie
Dr. Mathieu Sperandio, Rapporteur, France
Dr. David Barnes, Rapporteur, Australie
DEVELOPMENT AND FUNDAMENTAL INVESTIGATIONS
OF INNOVATIVE TECHNOLOGIES
FOR BIOLOGICAL NUTRIENT REMOVAL
FROM ABATTOIR WASTEWATER.
by
Romain LEMAIRE
Bachelor of Industrial and Environmental Processes Engineering
Masters of Environmental and Water Processes
Institut National des Sciences Appliquées de Toulouse, France
PhD conducted under a cotutelle de thèse agreement between
The University of Queensland in Australia and l’Université de Montpellier II in France.
Thesis submitted in partial fulfilment of the requirements
for the degree of Doctor of Philosophy.
Advanced Water Management Centre
School of Engineering
The University of Queensland, Australia
&
Laboratoire de Biotechnologie de l'Environnement
Ecole Doctorale : Science des Procédés – Science des Aliments
Université de Montpellier II, France
July 2007
Declaration
I declare that the work presented in this thesis is, to the best of my knowledge and believe,
original and my own work, except as acknowledged in the text, and that the material has not
been submitted, in whole or in part, for a degree at this or any other institution.
Romain Lemaire
Ph. D. Candidate
Prof. Zhiguo Yuan
Principal Advisor
i
Acknowledgements
I remembered coming across a PhD student when I was still an undergraduate student in
France. I was asking him some questions about the way his PhD was going and he told me:
“don’t worry too much about the subject of your PhD, what you really want is supervisors that
are easy to talk to, timely with their comments and interested in your work”. I did not fully
understand his advice at the time but I can ensure you that I do now. While I had the chance to
work on a subject I was passionate about, my excellent relation with both my principal
supervisors, Prof. Zhiguo Yuan and Dr. Nicolas Bernet, their strong commitments, their
availability and their insightful comments were instrumental in directing my research and
helping me achieve my goal. I would also like to thank my associate supervisor Prof. Jürg
Keller for his contribution and his valuable expertise in the field. I should also mention that,
as the director of AWMC, Jürg welcomed me in his centre for a year when I was still an
undergraduate student 6 years ago. I have not left since then as AWMC has become a vibrant
and renowned research centre a credit to Jürg’s effort and passion.
I would also like to thank the Environmental Biotechnology CRC, a Cooperative Research
Centre established and funded by the Australian Government together with industry and
university partners. They financed my scholarship and my PhD research through Project P5.
I had the chance to work with many people over the course of my PhD from whom I have
learned a lot and also contributed to my work as detailed on page III. I greatly appreciate the
opportunity to work with Dr. Rikke Meyer, Dr. Greg Crocetti, Prof. Linda Blackall, Rick
Webb, Dr. Gulsum Yilmaz, Dr. Maite Pijuan, Dr. Adrian Oehmen, Dr. Raymond Zeng,
Annelies Taske, Marcos Marcelino and all the other members of AWMC. My thanks also go
to Dr. Beatrice Keller-Lehmann, Jianguang Li and Dr. Sandra Hall for their technical support
and advice and to all the administration staff (Wendy, Jan and Chris) for their daily support
and their patience especially with us, the overseas students.
I would also like to specially thank all my family in France who have provided me with love
and support throughout these years far from home as well as all the friends that I have made
here in Australia during the last 6 years and those that I have back in France.
Last but not least, I would like to thank my beautiful girlfriend Ruth who involuntarily
became an expert on “granular sludge” and “biological nutrient removal” over the last three
years!
ii
Statement of the Contribution by Others
This thesis contains the reporting of some important contributions made by other researchers
that I have worked with throughout the course of my PhD. These contributions are
acknowledged as follow:
•
•
•
•
•
•
•
•
•
Dr. Gulsum Yilmaz, of the Advanced Water Management Centre (AWMC) worked
closely with me on the Biological Nutrient Removal (BNR) project P5, funded by the
Environmental Biotechnology CRC. The development of the strategy to maintain the
activity of BNR biomass during long term starvation periods, described in Appendix C,
was a joint effort. She also contributed to the bioreactor operation, maintenance, sample
collection and batch test experiments reported in Appendix F.
Dr. Gregory Crocetti of AWMC performed the FISH quantification and analysis reported
in Appendix D. He also contributed to the development of a novel microbial technique
used to determine the distribution of different microbial population inside aerobic granules
(described in Appendix E).
Dr. Rikke Meyer, formerly of AWMC, currently at the Department of Microbiology at
Aarhus University, Denmark, constructed the N2O microsensors used in this thesis and
assisted with the experimental work reported in Appendix D.
Ms Annelies Taske, formerly a student with the Department of Microbiology and
Molecular Sciences, assisted with the N2O microsensor measurements reported in
Appendix D.
Mr Marcos Marcelino, a visiting PhD student at AWMC for 4 months, developed and
calibrated a model based on ASM2d and ran the simulations reported in Appendix B.
Dr. Raymond Zeng of AWMC contributed to some of the design and construction of
bioreactors used in this thesis.
Dr. Adrian Oehmen, formerly of AWMC, currently at the Chemistry Department of the
Universidade Nova de Lisboa, Portugal, worked closely with me in the lab for the first 4
months of my PhD thesis and contributed to some of the bioreactor operation and
maintenance reported in Appendix A.
Dr. Beatrice Keller-Lehmann of AWMC operated the FIA and GC for the analysis of
phosphate, ammonia, nitrate/nitrite, VFA, PHA and glycogen. Jianguang Li of AWMC
assisted with operation of the FIA. Graham Kerven of the School of Land, Crop and Food
Sciences conducted the ICP-AES ion analysis.
Mr Rick Webb of the Centre of Microscopy and Microanalysis (CMM) assisted with the
sample preparation for the observation of aerobic granules with scanning electron
microscope and conducted the transmission electron microscopy analysis reported in
Appendix G.
iii
List of Publications and Presentations
Journal publications forming chapters of the thesis, with author’s contribution:
•
Lemaire R., Yuan Z., Bernet N., Yilmaz G. and Keller J. submitted. A Sequencing Batch
Reactor System for High-Level Biological Nitrogen and Phosphorus Removal from
Abattoir Wastewater. Chemosphere.
Author’s Contribution: All of the experimental study and analysis of the results. Writing of
the paper, with contributions from the other authors.
•
Lemaire R., Marcelino M. and Yuan Z. submitted. Achieving the Nitrite Pathway Using
Aeration Phase Length Control and Step-feed in a SBR Removing Nutrients from Abattoir
Wastewater. Biotechnology and Bioengineering.
Author’s Contribution: All of the experimental study and analysis of the results. Writing of
the paper, with contributions from the other authors.
•
Yilmaz G., Lemaire R., Keller J. and Yuan Z. (2007). Effectiveness of an Alternating
Aerobic, Anoxic/Anaerobic Strategy for Maintaining Biomass Activity of BNR Sludge
during Long-term Starvation. Water Research, 41(12) 2590-2598.
Author’s Contribution: Reactor-based experimental study (jointly with G. Yilmaz) and
analysis of the results (assisted by G. Yilmaz). Writing of the paper, with contributions from
the other authors.
•
Lemaire R., Meyer R., Taske A., Crocetti G., Keller J. and Yuan Z. (2006). Identifying
causes for N2O accumulation in a lab-scale sequencing batch reactor performing
simultaneous nitrification, denitrification and phosphorus removal. Journal of
Biotechnology, 122(1) 62-72.
Author’s Contribution: Reactor-based experimental study and analysis of the results (assisted
by R. Meyer). Writing of the paper, with contributions from the other authors.
•
Lemaire R., Yuan Z., Blackall L. L. and Crocetti G. in press. Microbial Distribution of
Accumulibacter spp. and Competibacter spp. in Aerobic Granules from a Lab-Scale
Biological Nutrient Removal System. Environmental Microbiology.
Author’s Contribution: Development of the method and analysis of the results (in
collaboration with G. Crocetti). Writing of the paper, with contributions from the other
authors.
•
Yilmaz G., Lemaire R., Keller J. and Yuan Z. in press. Simultaneous Nitrification,
Denitrification and Phosphorus Removal from Nutrient-Rich Industrial Wastewater using
Granular Sludge. Biotechnology and Bioengineering.
Author’s Contribution: Reactor-based experimental study and analysis of the results (jointly
with G. Yilmaz). Writing of the paper, with contributions from the other authors.
•
Lemaire R., Webb R. and Yuan Z. submitted. Micro-scale Observations of the Structure
of Aerobic Microbial Granules used for the Treatment of Nutrient-Rich Industrial
Wastewater. ISME Journal.
Author’s Contribution: Experimental study and analysis of the results (assisted by R. Webb).
Writing of the paper, with contributions from the other authors.
iv
Conference presentations:
•
•
•
•
Blackall L. L., Lemaire R. and Crocetti G. (2007). Studies into Granular Sludge in the
Activated Sludge Process. 4th ASM Conference on Biofilms, 25-29 March, Quebec City,
Canada.
Lemaire R., Yilmaz G., Keller J. and Yuan Z. (2007). Aerobic granular sludge: breaking
the sludge barrier. Ozwater 2007 AWA conference, 4-8 March, Sydney, Australia.
Lemaire R., Blackall L. L., Yuan Z. and Crocetti G. (2006). Microscale structure of the
microbial community within simultaneous nitrification denitrification and phosphorus
removal granules. 11th International Symposium on Microbial Ecology ISME-11, 20-25
August, Vienna, Austria.
Lemaire R., Meyer R., Crocetti G., Zeng R. J., Yuan Z., Blackall L. L. and Keller, J.
(2005). The microbial community dynamics in a simultaneous nitrification, denitrification
and phosphorus removal bioreactor using different carbon sources. 4th Activated Sludge
Population Dynamic IWA Specialist Conference, 17-20 July, Gold Coast, Australia.
Patent:
•
Lemaire R., Yuan Z. and Keller J. “Biological process for at least partial removal of
nitrogen, phosphorus and BOD from wastewaters which have very high levels of nitrogen
as well as significant phosphorus levels, such as abattoir wastewaters.” PCT filed on 16th
October 2007, IP Australia.
v
Abstract
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen (N) and
phosphorus (P). These nutrients must be removed to very low levels before any discharge into
local waterways to avoid causing eutrophication in receiving aquatic systems. Reliable
biological COD and nitrogen removal systems have been developed and applied for abattoir
wastewater treatment using continuous activated sludge systems. However, P removal is
achieved primarily through chemical precipitation with the addition of large amounts of iron
or alum salts. Biological P removal is often consider as being a cheaper and more sustainable
option as no chemical dosing is required and the quality of the excess sludge is usually
suitable for land fertilisation purposes. Compare to well established continuous wastewater
treatment systems, sequencing batch reactor (SBR) technology offers a great deal of
operational flexibility and appears to be a promising vehicle for achieving high levels of N
and P removal from abattoir wastewater.
The overall objective of this thesis is to develop a biological process that achieves high level
of COD, N and P removal from abattoir wastewater, producing effluent suitable for being
discharged into river systems (i.e. over 95% of the N and P contents removed). Achieving
stable and reliable biological P removal from a wastewater containing high level of nitrogen is
the main obstacle to overcome to achieve this objective. This study also investigate the
feasibility of using two innovative technologies, namely the simultaneous nitrification,
denitrification and phosphorus removal (SNDPR) process and the aerobic granular sludge
technology, to further enhance the performance of the SBR system designed.
In this thesis, three different lab-scale SBRs were operated to demonstrate the feasibility to
achieve high-levels of biological COD, N and P removal from abattoir wastewater. The first
SBR was employed to experiment novel strategies, which can easily be implemented to
current water treatment facilities used by the meat industry, to produce an effluent suitable for
river discharge. The second SBR was operated to provide a platform for an in-depth
investigation of the previously proposed SNDPR process, which has the potential for
application to the treatment of abattoir wastewater. The third SBR was operated for the
treatment of abattoir wastewater by combining together the aerobic granular sludge
technology and the SNDPR process. A wide range of techniques was employed in this thesis
which includes reactor process studies, microbial investigations (fluorescent in-situ
hybridisation combined with confocal laser scanning microscopy) and several micro-scale
techniques (e.g. microsensors, electron microscopy and light microscopy). The combination
of these multi-disciplinary techniques has helped deliver significant insightful information.
The main contributions of this thesis are as follow. An SBR system was demonstrated to
effectively remove 95%, 97% and 98% of the total COD, total N and total P present in
abattoir wastewater. It could provide a real alternative to chemical P removal, which is the
common practice in the meat industry. A multi-step feeding strategy was employed to prevent
the accumulation of nitrate or nitrite in the SBR providing the right condition for the
development of a stable biological P removal. The incorporation of a high-rate pre-fermentor
as an integrated component of the nutrient removal system was found to be important. This
stream, which contains a high-level of volatile fatty acids, provides supplementary carbon
sources critical for both P and N removal. Further, an aeration control strategy consisting of
stopping the aeration in the SBR immediately after ammonia is fully oxidised was effective in
vi
achieving stable N removal via the nitrite pathway. It benefited the nutrient removal
performance of the SBR by saving some valuable amount of COD.
The investigation of the SNDPR process revealed that the production of N2O, often observed
in lab-scale SNDPR bioreactors and source of concern due to the high global warming
potential of N2O, was found to be linked to the loss of diversity amongst the denitrifying
microbial community due to the use of synthetic wastewater containing only a single carbon
source. N2O accumulation is unlikely to be an issue in a SNDPR bioreactor treating real
wastewater which contains a large variety of different carbon sources.
The SNDPR process and the aerobic granular sludge technology were successfully combined
in a single SBR process. The size and the dense structure of aerobic granules positively
contributed to the oxygen mass transfer limitation required to achieve reliable SNDPR. It was
also demonstrated that the minimum hydraulic retention time (HRT) for a granular sludge
system is not governed by the sludge settleability, as is the case in a system with floccular
sludge. Mass transfer limitations in granules are an important factor to be considered in the
design of the HRT and the COD and nutrient loading rate in a granular sludge system.
The in-depth study of the structure and function of these aerobic granules fed with abattoir
wastewater revealed interesting structural features that have not been reported in synthetic-fed
granules. These observations initiated new hypotheses regarding the microbial structure and
the fate of mature granules, the effect of pH on the granule structure stability and the role
played by protozoa in the overall system performance. It also provided some new directions
and recommendations for further experimental studies on aerobic granules in relation to their
structure and behaviour in real systems.
vii
Résumé en français
L’industrie de la viande utilise de larges quantités d’eau lors de l’abattage, du découpage et du
nettoyage des équipements. Les effluents produits sont très chargés en DCO, azote et
phosphore. Afin d’éviter toute pollution des milieux aquatiques environnants, ces effluents
doivent subir des traitements poussés. Le but principal de cette thèse était de développer un
procédé de traitement par boues activées qui puisse éliminer plus de 95% de la DCO, de
l’azote et du phosphore dans les effluents d’abattoir permettant alors un rejet direct de
l’effluent traité en rivière. La forte teneur en azote des effluents d’abattoir est l’obstacle
majeur empêchant d’établir une élimination biologique du phosphore stable et efficace.
Durant cette thèse, un procédé biologique capable d’éliminer 98% du phosphore tout en
abattant 95% de la DCO et 97% de l’azote a été développé dans un réacteur batch à
alimentation séquentielle (SBR). Par rapport aux procédés chimiques classiques d’élimination
du phosphore, ce nouveau procédé biologique offre une vraie alternative financière et
environnementale pour l’industrie de la viande. Une stratégie d’alimentation séquentielle a
permis de réduire l’accumulation des nitrates dans le SBR rendant ainsi possible l’élimination
biologique du phosphore.
Cette thèse aborde aussi l’étude et l’utilisation de technologies innovantes pour améliorer les
performances du procédé SBR. Le procédé de nitrification, dénitrification et déphosphatation
simultanées (SNDPR) a été incorporé au procédé à boues granulaires aérobies. La taille, la
densité et l’activité microbienne des granules aérobies génèrent de forts gradients d’oxygène à
l’intérieur des granules, permettant alors d’obtenir un procédé SNDPR plus efficace. Le
volume du réacteur et la demande en DCO nécessaire pour éliminer l’azote et le phosphore
dans les effluents d’abattoir ont ainsi pu être fortement réduits. La structure interne et la
composition microbienne de ces granules ont également été étudiées.
viii
Table of Contents
Declaration .................................................................................................................................. i
Acknowledgements ....................................................................................................................ii
Statement of the Contribution by Others...................................................................................iii
List of Publications and Presentations....................................................................................... iv
Abstract ..................................................................................................................................... vi
Résumé en français..................................................................................................................viii
Table of Contents ...................................................................................................................... ix
List of Abbreviations................................................................................................................. xi
1.0
Introduction..................................................................................................................... 1
2.0
Literature Review ........................................................................................................... 2
2.1. Introduction to biological nutrient removal processes................................................... 2
2.1.1. Nitrogen removal process....................................................................................... 2
2.1.2. Biological phosphorus removal process ................................................................ 4
2.1.3. Advanced biological nutrient removal processes................................................... 5
2.1.4. Achieving BNR using sequencing batch reactor technology.................................. 6
2.1.5. SBR control strategy............................................................................................... 6
2.2. Abattoir wastewater treatment ....................................................................................... 7
2.2.1. Abattoir wastewater characteristics....................................................................... 7
2.2.2. Principal wastewater treatment processes used in the meat processing industry . 8
2.2.3. SBR technology for abattoir wastewater treatment................................................ 9
2.2.4. Challenges facing abattoir wastewater treatment................................................ 10
2.3. New possible BNR technologies ................................................................................. 11
2.3.1. Simultaneous Nitrification, Denitrification and Phosphorus Removal................ 11
2.3.2. Aerobic Granular Sludge technology................................................................... 12
3.0
Thesis Overview ............................................................................................................ 15
3.1. Research Objectives..................................................................................................... 15
3.2. Research Methods........................................................................................................ 17
3.2.1. Operation of lab-scale SBRs used in this thesis ................................................... 17
3.2.2. Analytical Methods............................................................................................... 19
3.3. Research Outcomes...................................................................................................... 21
4.0
Conclusions and Recommendations for Future Work.............................................. 28
4.1. Main Conclusions of the Thesis................................................................................... 28
4.2. Recommendation for Future Research......................................................................... 29
ix
References ................................................................................................................................31
Appendix A A Sequencing Batch Reactor System for High-Level Biological Nitrogen
and Phosphorus Removal from Abattoir Wastewater .......................................38
Appendix B Achieving the Nitrite Pathway Using Aeration Phase Length Control and
Step-feed in a SBR Removing Nutrients from Abattoir Wastewater................52
Appendix C Effectiveness of an Alternating Aerobic, Anoxic/Anaerobic Strategy for
Maintaining Biomass Activity of BNR Sludge during Long-term
Starvation...........................................................................................................73
Appendix D Identifying causes for N2O accumulation in a lab-scale sequencing batch
reactor performing simultaneous nitrification, denitrification and
phosphorus removal ..........................................................................................88
Appendix E Microbial Distribution of Accumulibacter spp. and Competibacter spp. in
Aerobic Granules from a Lab-Scale Biological Nutrient Removal System....103
Appendix F Simultaneous Nitrification, Denitrification and Phosphorus Removal
from Nutrient-Rich Industrial Wastewater using Granular Sludge.................119
Appendix G Micro-scale Observations of the Structure of Aerobic Microbial Granules
used for the Treatment of Nutrient-Rich Industrial Wastewater .....................142
Appendix H Résumé détaillé en français.............................................................................162
x
List of Abbreviations
AOB
AS
BNR
BOD5
CLSM
COD
DAF
DO
EBPR
EPS
FISH
FOG
GAO
HRT
MLSS
MLVSS
NOB
OCT
OLR
ORP
OUR
PAO
PCA
PHA
SBOD5
SBR
SCOD
SEM
SKN
SND
SNDPR
SRT
SVI
TCOD
TDN
TDP
TEM
TKN
TN
TP
UASB
VSS
VFA
WWTP
Ammonia oxidising bacteria
Activated sludge
Biological nutrient removal
Five-day biochemical oxygen demand
Confocal laser scanning microscope
Chemical oxygen demand
Dissolved air flotation
Dissolved oxygen
Enhanced biological phosphorus removal
Extracellular polymeric substances
Fluorescent in situ hybridisation
Fat, oil and grease
Glycogen accumulating organisms
Hydraulic retention time
Mixed liquor suspended solids
Mixed liquor volatile suspended solids
Nitrite oxidising bacteria
Optimum cutting temperature
Organic loading rate
Oxido-reduction potential
Oxygen uptake rate
Polyphosphate accumulating organisms
Perchloric acid
Polyhydroxylalkanoates
Soluble five-day biochemical oxygen demand
Sequencing batch reactor
Soluble COD
Scanning electron microscope
Soluble Kjeldahl nitrogen
Simultaneous nitrification and denitrification
Simultaneous nitrification, denitrification and phosphorus removal
Sludge retention time
Sludge volume index
Total COD
Total dissolved nitrogen
Total dissolved phosphorus
Transmission electron microscope
Total Kjeldahl nitrogen
Total nitrogen
Total phosphorus
Upflow anaerobic sludge-blanket
Volatile suspended solids
Volatile fatty acid
Wastewater treatment plant
xi
1.0
Introduction
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen (N) and
phosphorus (P). These nutrients must be removed from the wastewater to very low levels
before any discharge into local waterways to avoid causing eutrophication in receiving aquatic
systems. Over the past two decades, biological COD and N removal from abattoir wastewater
has received much greater attention than has the biological P removal. Reliable biological
COD and nitrogen removal systems have been developed and applied for abattoir wastewater
treatment using continuous activated sludge systems. Currently, P removal is achieved
primarily through chemical precipitation with the addition of large amounts of iron or alum
salts. Biological P removal is often consider as being a cheaper and more sustainable option as
no chemical dosing is required and the quality of the excess sludge is usually suitable for land
fertilisation purposes due to its low salt content and high phosphate recovery potential. The
high level of nitrogen in abattoir wastewater has proved to be an obstacle to the development
of a stable and reliable biological P removal process. Compare to well established continuous
wastewater treatment systems, sequencing batch reactor (SBR) technology offers a great deal
of operational flexibility and appears to be a promising vehicle for achieving high levels of N
and P removal from abattoir wastewater. The meat industry in Australia is indeed starting to
employ the SBR technology to treat their wastewater on site.
The overall objective of this thesis is to develop a biological process that achieves high level
of COD, N and P removal from abattoir wastewater, producing effluent suitable for being
discharged into river systems (i.e. over 95% of the nitrogen and phosphorus contents
removed). Achieving stable and reliable biological P removal from a wastewater containing
high level of nitrogen is the main obstacle to overcome to achieve this objective. This novel
SBR process will have to be developed in a way that its future application to current abattoir
wastewater treatment facilities can be achieved without large additional infrastructure and
operating costs making it a real alternative option for the meat industry.
This study will also investigate the feasibility of using two innovative technologies to further
enhance the performance of the SBR system designed. These technologies are still in their
conceptual stage and knowledge gaps have to be filled before any possible industrial
application. Firstly, the recently developed and demonstrated simultaneous nitrification,
denitrification and phosphorus removal (SNDPR) process will be applied to the treatment of
abattoir wastewater. This process promises savings in aeration costs and would also reduce
demand for volatile fatty acids, which is crucial for biological P removal but often limiting in
abattoir wastewater. Secondly, the feasibility of achieving high-levels of COD, N and P
removal from abattoir wastewater using aerobic granular sludge technology will be
investigated in this thesis. The excellent settleability of aerobic granular sludge allows for
more biomass to be maintained in a relatively small reactor volume, enhancing the ability of
the reactor to withstand high loading rates. This is of great interest for the treatment of high
nutrient containing industrial wastewater such as abattoir wastewater compared to
conventional floccular sludge systems. It is also believed that SNDPR could be more easily
implemented in a granular system than in a floccular system.
1
2.0
Literature Review
This chapter explains the principles of the biological nitrogen and phosphorus removal
processes, examines the characteristics of abattoir wastewater and identifies the main
challenges for its treatment. The advantages and knowledge gaps related to two novel
attractive technologies are also presented. A description of the achievements and limitations
of previous studies is provided and based on these limitations, the focus of this research is
established.
2.1.
Introduction to biological nutrient removal processes
The general deterioration of water quality in rivers and streams in urban areas has resulted in
an effort to prevent eutrophication by reducing the nutrient levels in wastewater - such as
nitrogen (N) and phosphorus (P) - being discharged into local waterways (Mainstone and Parr,
2002). In order to achieve this reduction the wastewater needs to receive tertiary treatment,
usually through biological treatment which is considered to be the easiest and most costeffective way to remove nutrients from wastewater streams. Biological nutrient removal
(BNR) relies on the activity of a diverse microbial community that transfers the nutrients from
the wastewater (liquid phase) to the atmosphere (gas phase) and/or into biosolids (solid
phase).
2.1.1. Nitrogen removal process
A conventional BNR process achieves nitrogen removal through a continuous two-stage
treatment process (Figure 1): aerobic nitrification and anoxic denitrification (Metcalf & Eddy,
1991). During the nitrification stage, ammonia oxidising bacteria (AOB) aerobically oxidise
ammonium, the major form of nitrogen in wastewater, to nitrite (i.e. nitritation process) which
is subsequently oxidised to nitrate (i.e. nitratation process) by nitrite oxidising bacteria
(NOB). Both AOB and NOB are autotrophic bacteria using oxygen as electron acceptor.
During the denitrification stage, heterotrophic denitrifiers reduce nitrate to nitrite and then
finally to dinitrogen gas under oxygen deficiency, using external organic substrates (chemical
oxygen demand - COD) as electron donor. Typically the COD present in the incoming
wastewater is used for denitrification in which case the denitrification reactor is located prior
to the nitrification reactor (pre-denitrification) (Figure 1). This design is necessary to ensure
sufficient amount of COD for complete denitrification as most of the COD present in the
wastewater would be oxidised under aerobic conditions if the nitrification stage was located
first. Nitrate produced by nitrification is then returned to the denitrification reactor via a
recycle flow.
2
Anoxic (O2 , NO3-)
Aerobic (O2 , NO3-)
Denitrification
influent
Nitrification
NH4+ Æ
NH4+
COD Æ CO2
Clarifier
NO3effluent
COD Æ CO2
NO3- Æ N2
recycle
Waste
(WAS)
recycle of settle biomass (RAS)
Figure 1. Conceptual design of a continuous nitrification and denitrification plant.
However, it has been observed that these two processes can occur concurrently in a singlesludge, single-stage process under low dissolved oxygen (DO) conditions called simultaneous
nitrification and denitrification (SND) (Munch et al., 1996; Bertanza, 1997; Keller et al.,
1997; Fuerhacker et al., 2000). SND relies on the formation of anoxic zones in the central part
of the microbial aggregates caused by the mass transfer limitation of oxygen (Figure 2). In the
aerobic zone on the edge of the aggregate, autotrophic bacteria can nitrify using oxygen,
whereas in the anoxic zone at the centre of the aggregate, heterotrophic bacteria can denitrify.
Therefore, factors that affect oxygen mass transfer limitation such as bulk liquid oxygen
concentration, the aggregate size, and the specific activity of the microbial aggregates (oxygen
uptake rate per volume of biomass) (Pochana and Keller, 1999; Meyer et al., 2005) also affect
SND. Performing N removal via SND in full-scale plants has several potential benefits.
Firstly, it will reduce the capital and operating cost through the elimination of the separate
denitrification tank and recycle flow depicted in Figure 1. Secondly, the application of low
DO for SND will save aeration costs.
N2
O2
Aerobic
N2
Anoxic
NOx-
NH4+
Figure 2. Illustration of oxygen profile inside a microbial aggregate under low oxygen
concentration.
Additionally, nitrogen removal can be achieved through nitrification and denitrification via
nitrite, by-passing the nitrate stage (Figure 3). This can be achieved by inhibiting or
eliminating the NOB population from the system as NOBs are responsible for the second step
of the nitrification (from NO2- to NO3-). This nitrification and denitrification via nitrite (also
called “nitrite pathway”) offers many advantages: (i) 40% lower requirement of COD in the
denitrification stage, (ii) 25% lower consumption of oxygen in the nitrification stage, (iii)
3
higher denitrification rates and smaller sludge production (Turk and Mavinic, 1986).
Implementing the nitrite pathway could have significant benefits for the operation of largescale wastewater treatment facilities both for domestic and industrial effluents.
O2
COD
NO3-
O2
NH4+
COD
NO2-
N2
Figure 3. Illustration of nitrification and denitrification via nitrite (i.e. nitrite pathway).
2.1.2. Biological phosphorus removal process
The removal of phosphorus from wastewater is typically achieved by either chemical
precipitation with the addition of iron and alum salts or through biological accumulation.
However, the biological alternative has a number of significant advantages such as lower
operating costs due to the non-requirement of chemicals addition, lower waste sludge
production and higher reuse potential of the sludge as soil fertiliser due to its high phosphate
recovery potential and its low salt content (Gaterell et al., 2000).
Wastewater treatment plants (WWTPs) performing biological P removal recirculate the
sludge through an anaerobic and an aerobic zone as shown in Figure 4. This process is called
enhanced biological phosphorus removal (EBPR) and is based on the ability of
polyphosphate-accumulating organisms (PAOs) to take up P and accumulate it intracellularly
as polyphosphate (poly-P) when exposed to alternating anaerobic (O2 and nitrite/nitrate (NOx-)
absent) and aerobic conditions (Comeau et al., 1986; Wentzel et al., 1988). PAOs are capable
of taking up volatile fatty acids (VFAs) anaerobically which gives them a selective advantage
over ordinary heterotrophic organisms, which are unable to take up VFAs anaerobically.
VFAs are therefore the essential carbon substrates for EBPR. However they are not generally
present to a sufficient concentration in wastewater. Many EBPR full-scale plants thus use
prefermentors to generate additional VFAs from the incoming wastewater that can be added to
the anaerobic zone.
Prefermenter
influent
O2
O2
Anaerobic
Aerobic
P-release
Clarifier
P-uptake
VFA Æ PHA
COD Æ CO2
PHA
effluent
recycle
recycle of settle biomass (RAS)
Figure 4. Conceptual design of an EBPR plant.
4
Waste
(WAS)
In the anaerobic zone, PAOs store VFAs as intracellular polyhydroxylalkanoates (PHAs)
using energy gained from hydrolysing intracellular poly-P and releasing it to the liquid phase
in the form of phosphate. In the subsequent aerobic zone, PAOs take up an amount of
phosphate that is in excess of what was released under the proceeding anaerobic conditions,
using their internally stored carbon (i.e. PHA) to provide the energy required. The
accumulated P is then removed from the system together with the excess sludge that is
produced in the process.
EBPR process has been successfully implemented in WWTPs and good P removal is regularly
achieved (van Loosdrecht et al., 1997). However the control of the process can be difficult
and EBPR failure has been frequently observed. Several reasons have been identified
including excessive aeration (Brdjanovic et al., 1998), high nitrate concentration in the
anaerobic zone (Pitman et al., 1983; Chang and Hao, 1996; Furumai et al., 1999) and the
proliferation of microorganisms known as glycogen-accumulating organisms (GAOs) able to
compete with PAOs for the carbon substrates in the anaerobic stage without contributing to
any P removal (Mino et al., 1995). To store PHA from VFAs in anaerobic phase, GAOs gain
energy exclusively from the glycolysis of intracellular glycogen, not from poly-P hydrolysing.
Under aerobic condition, GAOs oxidise PHA for cell growth and glycogen replenishment.
Therefore they can survive under EBPR conditions, but do not release/accumulate any
phosphorus. It has been demonstrated in a recent survey of full-scale EBPR plants that carbon
(i.e. VFA) requirements are increased considerably by the presence of GAOs (Saunders et al.,
2003).
2.1.3. Advanced biological nutrient removal processes
In the case of very sensitive receiving environment, wastewater streams can be subjected to
advanced tertiary treatment before being discharged to the local waterways. It usually consists
of a combination between a conventional two-stage treatment process for N removal and the
EBPR process for P removal. The conceptual design of this advanced BNR process is detailed
in Figure 5. Such continuous BNR systems can vary slightly in terms of design but they
generally include at least 3 separate mixing tanks, several recycling lines and a clarifier unit
which make them relatively complex to operate and quite expensive to build. For example, the
5-stages Bardenpho® process, often referred to as the “Rolls-Royce” of all BNR systems
because it can achieve really high-levels nutrient removal, consists in 5 separate tanks (i.e. a
post-anoxic and second aerobic tanks are added to the process described in Figure 5 as
polishing steps before the clarifier). In addition, an anaerobic prefermeter is often required at
the head of the plant to provide sufficient VFAs for stable bio-P removal.
Prefermenter
influent
(O2, NO3-)
(O2, NO3-)
(O2, NO3- )
Anaerobic
Anoxic
Aerobic
NH4+
NH4+ Æ NO3-
NH4+
VFA Æ PHA
COD Æ CO2
P-release
NO3- Æ N2
COD Æ CO2
PHA
Clarifier
effluent
P-uptake
recycle
recycle of settle biomass (RAS)
Waste
(WAS)
Figure 5. Conceptual design of a continuous BNR plant achieving both N and P removal.
5
2.1.4. Achieving BNR using sequencing batch reactor technology
Sequencing batch reactor (SBR) activated sludge processes are known to have several
advantages over conventional continuous flow systems (Irvine and Busch, 1979; Wilderer et
al., 2001). In recent years, the use of single tank SBRs for the biological treatment of
wastewater has been widely extended from lab-scale studies to full-scale WWTPs (Tilche et
al., 1999; Artan et al., 2001; Keller et al., 2001; Puig et al., 2004) since it offers a great deal of
operational flexibility by easy adjustment of aerobic, anoxic and anaerobic periods through
temporal control of aeration and filling in a cycle with no need for separate basins, recycling
lines or clarifiers. SBRs have been utilised extensively for the removal of COD and, in many
cases, nitrogen, from wastewater. It can be particularly useful in treating industrial effluent
often presenting a large composition and flow variability. A comprehensive review of the
SBR technology can be found in Wilderer et al. (2001).
More recently, a number of phosphorus removal processes using the SBR technology have
also been developed which involves the use of a number of innovative concepts. Keller et al.
(2001) demonstrated a novel filling process called “UniFed”. The unique feature of this
process is the uniform introduction of the influent into the settled sludge during the settling
and decant periods of the SBR operation. This creates true anaerobic conditions under the
sludge blanket by removing any remaining nitrate or nitrite using some of the freshly
introduced influent COD. As a result, PAOs can release large amount of phosphate
anaerobically which is critical to successful biological phosphorus removal. It also achieves a
“selector” effect, which helps in generating a compact, well settling biomass in the reactor.
Another innovative strategy is the use of a step-feeding scheme, characterised by several
aerobic and anoxic phases in a SBR cycle (Anderottola et al., 2000; Lin and Jing, 2001; Puig
et al., 2004). This strategy aims at improving the nitrification and denitrification performance
of the SBR. The feeding is carried out in several anoxic periods so that easily biodegradable
substrate would enhance the denitrification efficiencies and low nitrate levels may be
achieved even from wastewater highly concentrated in nitrogen (Puig et al., 2004).
2.1.5. SBR control strategy
SBRs usually operate with fixed lengths for the different phases of filling, mixing (anaerobic,
aerobic or anoxic), settling and decanting. Due to influent fluctuation and system state
variations, it is beneficial to operate a SBR process with varying phase lengths. Therefore,
higher levels of process control and automation are necessary to optimise the SBR operation.
Many researchers have suggested that for a nitrogen removal system, on-line measurements of
ORP, DO and pH contain some characteristic patterns that indicate the end of the
biodegradation processes (Al-Ghusain et al., 1994; Wareham et al., 1994; Al-Ghusain and
Hao, 1995; Hao and Huang, 1996). The control systems designed are inferential, due to the
fact that ORP and pH are indirect measures of the nitrification and denitrification processes.
Figure 6 shows typical ORP, pH and DO profiles in an alternating aerobic-anoxic nitrogen
removal bioreactor with excess aerobic and anoxic periods. NH4+, and NO3- profiles are also
shown in Figure 6. It is clear that the bending points detected on the ORP and pH curves
correspond to the ends of the nitrification and denitrification, which enable the design of a real
time control system for the process based on ORP, pH and DO signals. For example, Peng et
al. (2004) recently demonstrated that stable nitrification and denitrification via nitrite could be
obtained through the control of the aeration time in a sequencing batch reactor (SBR) treating
domestic wastewater. DO and pH signals were used to detect the end of the nitritation process
(i.e. complete oxidation of NH4+). Aeration was stopped as soon as nitritation finished, as
indicated by a bending point on the pH profile and a sharp increase of the DO level. An
6
external carbon source (glucose) was then added to enable denitrification in the following
anoxic period.
12
8.0
aeration off
ammonia valley
nitrate apex
7.5
6
7.0
3
6.5
pH
ORP (m V/30), nitrogen &
DO concentration (m g/L)
9
0
0
-3
-6
-9
1
2
3
ammonia break point
aeration
on
4
5
6
7 6.0
nitrate knee
ORP
DO
pH
5.5
NO3-N
NH4-N
aeration on
5.0
tim e (hour)
Figure 6. ORP, pH, DO, NH4+ and NO3- profiles in a nitrogen removal bioreactor (illustrative
from (Yuan et al., 2003)).
In recent years, aeration control using nutrient sensors has been studied by many researchers.
The continuous improvement of reliability, accuracy, ease of maintenance and in-situ location
of ammonia and nitrate sensors (Londong and Wachtl, 1996; Lynggaard-Jensen et al., 1996;
Ingildsen and Wendelboe, 2003) has resulted in some full-scale applications (Ingildsen and
Olsson, 2002). Compared to other types of sensors, nutrient sensors support the direct control
of the ammonia and nitrate nitrogen concentrations in the system. The control systems
designed based on these sensors therefore exhibit more flexibility between nitrification and
denitrification (Yuan et al., 2003).
2.2.
Abattoir wastewater treatment
2.2.1. Abattoir wastewater characteristics
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen and phosphorus.
In a recent review of the wastewater treatment in the meat industry, Mittal (2006) reported an
average water usage of 300-700 L per pig slaughtered and 1,000-2,500 per cattle. The daily
water usage for large plants processing up to 3,000 animals a day can therefore be quite
consequent. Bhamidimarri (1991) indicated that a typical New-Zealand meat processing plant
produces 10,000 m3.d-1 of wastewater with a pollution load equivalent to a city of 60,000100,000 inhabitants. However, the water consumption varies considerably depending on the
type of abattoir (e.g. integrated slaughterhouse, processing plant, rendering plant), the type of
animals slaughtered (e.g. pig, sheep, cattle), the local practices employed (e.g. number of
shifts per day and number of working days per week), the climate and also between private
and public abattoirs (Koenig and Yiu, 1999).
7
Abattoir wastewater is composed of a mixture of grease, fat, protein, blood, intestinal content,
manure, cleaning products as well as about 80% fresh water (Johns et al., 1995). The main
characteristics of primarily treated abattoir wastewater reported in different studies are
presented for comparison in Table 1. The strength of the wastewater varies among abattoirs
depending upon operating practices. It also varies from day to day depending on the number
and type of animals being processed. The main differences concern the COD and FOG
concentrations whereas nitrogen and phosphorus content are less variable. According to Johns
(1995), the typical concentrations of TKN and TP in slaughterhouse wastewater are 120200 mg.l-1 and 15-40 mg.l-1, respectively, which are considerably higher than in domestic
wastewater. Primarily treated wastewater also contains a high level of organic N that usually
breaks down to ammonia during the subsequent treatment steps. The potentially high
TCOD:TKN and TCOD:TP ratios in abattoir wastewater should be cautiously interpreted as
40 to 70% of the total COD is slowly biodegradable and not directly usable for N or P
removal.
Table 1. Some characteristics of primarily treated abattoir wastewater. Adapted from Mittal
(2004). Values in bracket are average value.
Reference
Country
TCOD
SCOD
FOG
TKN
N-NH4
TP
(mg.l-1)
(mg.l-1)
(mg.l-1) (mgN.l-1) (mgN.l-1) (mgP.l-1)
Borja et al. (1994)
Spain
(5,100)
--(310)
(95)
(30)
Caixeta et al. (2002)
Brazil
2,000-6,200
-40-600
-20-30
15-40
Li et al. (1986)
China
628-1,437
-97-452 44-126
25-105
10-16
Manjunath et al. (2000)
India
1,100-7,250
-125-400 90-150
-8-15
Martinez et al. (1995)
Spain
(6,700)
(2,400)
(1,200)
(268)
-(17)
Nunez and Martinez (1999)
Spain
1,440-4,200 720-2,100 45-280
---Russell et al. (1993)
New-Zeal.
(1,900)
--(115)
(30)
(15)
Sachon (1984)
France
(5,113)
-(897)
(248)
-(22)
Sayed et al. (1987)
Holland
1,500-2,200
--120-180
-12-20
Sayed et al. (1988)
Holland 1,925-11,118 780-10,090
-110-240
-13-22
Stebor at al. (1990)
US
4,200-8,500 1,100-1,600 100-200 114-148
65-87
20-30
Thayalakumaran (2003b)
New-Zeal. 490-2,050
400-1,010 250-990 105-170 26-116
25-47
2.2.2. Principal wastewater treatment processes used in the meat processing
industry
Similar primary treatment processes (settling, screening, flocculation, dissolved air flotation –
DAF) are used worldwide to partially remove the suspended solids and the FOG from the raw
abattoir wastewater. The main process variation concerns secondary treatments and depends
mostly on the location of the processing plant. Johns (1995) and more recently Mittal (2006)
highlighted the differences existing between abattoirs located in Europe, North and South
America, New-Zealand and Australia which are the main meat producing regions in the world
(Figure 7). In Europe, primarily treated wastewater is usually directly discharged into
municipal WWTPs for further treatment necessiting the payment of a surcharge. The treated
effluent is then discharged into local waterways. In contrast, countries where land availability
is less of an issue (North and South America, Australia and New-Zealand), large anaerobic
and aerobic ponds systems are often employed to remove COD and achieve some partial
nitrification before the effluent is used for land irrigation. These large pond systems achieve
good COD removal but limited N removal and almost no P removal (Keller et al., 1997).
Although this effluent contains the required nutrient for plant growth, the high nitrogen
(ammonium and nitrate) infiltration rate in the soil may result in groundwater pollution during
intensive irrigation (Russell et al., 1993). Activated-sludge systems are also employed in
8
North America to remove the COD and some of the nutrients before land application.
Recently, advanced tertiary treatments using biological and physicochemical methods have
been employed in the US and in Australia to achieve complete nitrification and partial or even
complete denitrification together with chemical phosphorus removal. However, their use is
very limited due to high-cost involved (Mittal, 2006).
Australia
South America
TSS
FOG
Europe
Primary treatment (DAF, screening, settling)
COD
Anaerobic Pond, UASB
COD
Partial Nitri.
Denit.
New-Zealand US, Canada
Aerobic Pond
BNR
AS
Irrigation
Municipal WWTP
Receiving Ecosystem
Figure 7. Principal wastewater treatment processes used in the meat industry worldwide. The
targeted pollutant of each level of treatment is indicated on the left margin. Adapted from
Johns (1995) and Mittal (2006).
2.2.3. SBR technology for abattoir wastewater treatment
Recently, several studies using the SBR technology to simultaneously remove COD, N and P
from piggery wastewater have been very successful (Tilche et al., 1999; Obaja et al., 2003;
Obaja et al., 2005). However, the characteristics of the piggery wastewater differ greatly from
typical abattoir wastewater. TKN and TP are typically 3 to 4 times higher in piggery
wastewater than in abattoir wastewater while the amount of TCOD is relatively similar. This
has a direct effect on the overall TCOD:TKN and TCOD:TP ratios which are much lower in
the case of piggery wastewater than in typical abattoir effluent. Therefore, an external source
of easy biodegradable COD (acetate, methanol, non-digested pig manure) has to be added to
achieve good N and P removal. In addition, the large amount of inorganic salts, minerals and
metal ions present in the piggery wastewater promote chemical P removal by precipitation as
only scarce P release occurred during the anaerobic stage of the process (Bortone et al., 1994).
The absence of P release combined with the lack of microbial analysis of the sludge
undermines the simultaneous “biological” removal of COD, N and P claimed in these studies.
Studies on nutrient removal have also been carried out with abattoir wastewater.
Subramaniam et al. (1994) and Keller et al. (1997) firstly reported that simultaneous
biological nutrient and COD removal was possible in SBR from anaerobic-treated abattoir
wastewater under anoxic/anaerobic and aerobic conditions. In both studies, N removal was
partially achieved through SND leaving enough readily biodegradable COD to remove 90%
of the total N and P present in the anaerobic-treated abattoir wastewater without any external
carbon addition. However, nutrient removal performance depended greatly on the degree of
anaerobic treatment of the sludge - the longer the treatment, the less readily biodegradable
COD is left for nutrient removal. The P removal was quite unstable due to some intermittently
9
high levels of NOx- being recycled to the anaerobic period reducing the amount of VFAs
available for PAOs. More recently, Thayalakumaran (2003a) fed primary treated abattoir
wastewater directly into a SBR under anoxic/anaerobic and aerobic conditions with low DO
(0.5 mgO2.l-1) to promote SND. As a result, 93% of the total N and 96% of the total P in the
influent were biologically removed. It should be noted that the wastewater used in that study
contained a particularly high fraction of soluble COD and readily biodegradable COD,
representing 60% and 17% of the total COD, respectively. Most primary treated abattoir
effluents have much lower soluble and readily biodegradable COD fractions which could not
sustain high level of biological nutrient removal. The high variability of primary treated
abattoir wastewater is also an obstacle to achieve stable N and P removal. Instead, using
anaerobic-treated abattoir wastewater as influent for the SBR process offers more stability in
term of influent composition due to the relatively long hydraulic retention time (HRT) of
common anaerobic treatment processes. For example, the average HRT of large anaerobic
pond systems widely used in the meat industry is around 10 days which is long enough to
level out most of the daily fluctuations of flow and composition of the raw abattoir
wastewater.
2.2.4. Challenges facing abattoir wastewater treatment
Effect of high nitrate level on Bio-P removal
The deterioration of Bio-P removal due to the presence of nitrate in the designated anaerobic
period has been the topic of several studies. Comeau et al. (1986) reported that, when the
nitrate concentration in the return sludge was less than 5 mg.l-1, good P-release was easily
achieved. Pitman et al. (1983) found that a nitrate concentration higher than 10 mg.l-1
inhibited P-release resulting in a failed EBPR. Chang and Hao (1996) observed that when
nitrate levels in effluent were reduced from 10.9 to 5.6 mg.l-1, P removal efficiency increases
from 80 to 98%. In SBR performing both N and P removal, nitrate concentration and the end
of cycle must be low enough to provide a substantial true anaerobic period at the start of the
next cycle, which is fundamental for Bio-P removal (Furumai et al., 1999).
In general, the remaining level of nitrate in SBR treating domestic wastewater varies from 2 to
5 mg.l-1, which should not greatly affect the P removal. However, when dealing with abattoir
wastewater containing a high level of nitrogen, the remaining nitrate levels are generally
much higher and P removal can be severely affected due to the increased competition for
organic substrate between denitrifiers and PAOs (Subramaniam et al., 1994; Keller et al.,
1997).
Limitation of readily biodegradable COD
Sufficient COD, in particular readily biodegradable COD (primarily VFA), is required to
achieve good biological nutrient removal. Therefore, a challenge for the abattoir wastewater
primary treatment is to reduce the carbon content through anaerobic pond systems and, at the
same time, have sufficient and suitable COD remaining to perform complete N and P removal
in the BNR system (Keller et al., 1997). In some cases, COD required for denitrification can
be provided by by-passing a fraction of the raw wastewater stream directly into the BNR
system (Metzner and Temper, 1990). However, the direct addition of raw wastewater
increases the fat content of the BNR mixed liquor and could lead to poor settling properties
(see next paragraph). Alternatively, the high level of VFA necessary for P removal, mainly in
the form of acetate or propionate (Pijuan et al., 2004; Oehmen et al., 2006), could be supplied
from a well-controlled pre-fermenter.
10
Settling and bulking problems caused by the high level of fat/oil/grease
The high concentration of FOG and particulate matter (i.e. colloidal) of the abattoir
wastewater (Table 1) is a concern as it generally produces a light, poor settling sludge
(Hopwood, 1977). Travers and Lovett (1984) found that at low DO (<0.5 mg.l-1), fat
degradation was inhibited, leading to poor settling properties (sludge volume index – SVI – of
250 ml.gMLSS-1) and excessive numbers of filamentous organisms. In a recent study of the
edible oil industry, Reddy et al. (2003) observed significantly high TSS in the treated effluent
as a result of sludge oil aggregation, pin point floc formation and high numbers of free
swimming bacteria. However, microscopic analysis showed low abundance of filamentous
bacteria. The high oil content of the effluent resulted in sludge bulking with SVI values as
high as 770 ml.gMLSS-1. Interestingly, Thayalakumaran et al. (2003a) had no major settling
problem in their SBR (SVI between 100-200 ml.gMLSS-1, effluent MLSS of 33 mg.l-1) when
directly using primary treated abattoir wastewater containing higher fat levels than
anaerobically treated wastewater.
Fluctuations in abattoir wastewater composition and flow
One of the challenges for abattoir WWTPs is to cope with the large fluctuations of the
wastewater flow and composition inherent to most industrial activities. Koenig and Yiu
(1999) cited the type of animal slaughtered, the methods used, the type of equipment
available, as well as the local mode of production as the main factors affecting the amount and
composition of abattoir wastewater. In some cases, low activity periods (e.g. annual
maintenance or seasonal production variations) would even result in complete interruptions of
wastewater flows to the WWTPs for weeks and even months affecting the stability of the
overall BNR process. It is crucially important to maintain the viability of biomass (activity
and integrity) during the long idle or starvation periods to ensure a rapid return to previous
level of treatment when normal operational condition are resumed.
2.3.
New possible BNR technologies
The following technologies may potentially address the challenges facing the treatment of
abattoir wastewater. The main advantages, drawbacks and knowledge gaps are presented.
2.3.1. Simultaneous Nitrification, Denitrification and Phosphorus Removal
Simultaneous nitrate/nitrite and phosphorus removal can be achieved in anaerobic-anoxic
EBPR systems using ability of PAOs to simultaneously reduce NOx- and take up P (Kuba et
al., 1993; Kerrn-Jespersen et al., 1994; Meinhold et al., 1999). This can be highly beneficial to
lower the COD demand (same carbon source used for N and P removal) and reduce the
aeration costs. In addition, PAOs exposed to such anaerobic-anoxic conditions are 40% less
efficient in generating energy compare to in normal anaerobic-aerobic EBPR systems and thus
have a 20-30% lower cell yield resulting in less sludge production (Kuba et al., 1994).
Combining this anaerobic-anoxic EBPR system with SND could deliver even more substantial
savings of COD especially if nitrification and denitrification via nitrite pathway is achieved.
Indeed, the process of simultaneous nitrification and denitrification and phosphorus removal
(SNDPR) has already been demonstrated in lab-scale SBRs treating mainly synthetic
wastewater under alternating anaerobic and low DO aerobic stages (Zeng et al., 2003). In this
process, nitrogen and phosphorus are removed simultaneously to very low levels. During the
11
anaerobic stage, COD (i.e. acetate) is taken up accompanied by phosphorus release. During
the subsequent aerobic stage, the conversion of ammonia to gaseous nitrogen products via
nitrate or nitrite and P uptake are achieved concomitantly with almost no nitrite or nitrate
accumulating in the bulk liquid.
However, it was observed that nitrous oxide (N2O) rather than N2 was the major
denitrification end-product. This is a significant environmental concern due to the high
greenhouse gas potential of N2O (Zeng et al., 2003; Meyer et al., 2005). Meyer et al. (2005)
identified the PHA-driven denitrification as a potential cause for this N2O accumulation in
lab-scale SNDPR processes. Furthermore, unstable nitrate/nitrite removal has been reported in
such SNDPR bioreactors using floccular biomass. Meyer et al. (2005) showed that incomplete
coupling between nitrification and denitrification would occur if the aerobic/anoxic zones in
the microbial aggregates were not formed as illustrated in Figure 2, leading to the
accumulation of NOx- in the bulk liquid. Based on the process data of these bioreactors, it was
also reported that GAOs, known to compete with PAOs for carbon sources but without
contributing to P removal, appear to be primarily responsible for the denitrification process
(Zeng et al., 2003; Lemaire et al., 2006). Without denitrification by PAOs, there is no true link
between SND and EBPR, and the two processes merely occur in the same sludge at the same
time compromising the carbon savings proposed to be obtained by SNDPR. These findings
undermine the stability and robustness of the SNDPR process which are crucial for any
possible industrial application of this novel BNR technology.
2.3.2. Aerobic Granular Sludge technology
What is a granule?
Microbial granules can be described as compact and dense microbial aggregates of different
bacterial species with an approximately spherical external appearance (Figure 8). The growth
of such granules is sometimes regarded as a special case of biofilm development (Grotenhuis
et al., 1991; El-Mamouni et al., 1998). While granules were first reported in an upflow
anaerobic sludge blanket (UASB) bioreactor two decades ago (Lettinga et al., 1980), recent
research efforts have been dedicated to the study of aerobic granules in SBR systems
(Morgenroth et al., 1997; Beun et al., 1999; Peng et al., 1999; Etterer and Wilderer, 2001; Tay
et al., 2001). The large microbial diversity found in both anaerobic and aerobic granules has
led researchers to hypothesize that granulation is not a function of specific microbiological
groups but of reactor operating conditions (Beun et al., 1999). Compared to conventional
flocs, anaerobic or aerobic granules have a wide range of beneficial properties, including: a
regular, dense and compact structure, high biomass retention in the bioreactor, good
settleability, and the ability to withstand high flows rate and high organic loading rates. These
properties explain why granular UASB process has been extensively applied to anaerobic
wastewater treatment and why the recent aerobic granular technology has been the subject of
extensive studies yielding more than 60 publications over the past 3 years (Liu and Tay,
2004).
12
(a)
(b)
Figure 9. Aerobic granule(s) visualized (a) by scanning electron microscopy and (b) light
microscopy. These granules were obtained in this thesis study for the treatment of abattoir
wastewater. See appendix F and G for details of these granules and the conditions under
which they were obtained.
Factors affecting aerobic granulation
The formation of aerobic granules is a gradual process from suspended sludge flocs to
compact aggregates, to granular sludge and finally to mature aerobic granules (Tay et al.,
2001). However, the exact mechanisms involved in the successive stages of the aerobic
granulation process have not yet been fully explained. Recently, aerobic granules cultivated in
synthetic wastewater in SBR have been reported to achieve COD and/or N removal (Tay et
al., 2002; Liu et al., 2003; Yang et al., 2003) and, in some cases, bio-P removal (Dulekgurgen
et al., 2003; Lin et al., 2003). Although nearly all studies on aerobic granulation have been
carried out in lab-scale SBR with only a few using real wastewater (de Villiers and Pretorius,
2001; Arrojo et al., 2004; de Bruin et al., 2004), aerobic granulation technology is moving
towards industrial use for biological wastewater treatment. Liu et al. (2005b) recently
published some guidelines for up-scaling aerobic granular SBR from lab-scale to full-scale.
At laboratory scale, many factors influencing the properties of aerobic granules in SBR have
been described (Liu and Tay, 2004). Four major selection pressures directly contributing to
the formation mechanism of granules in SBR have been identified in the literature:
•
•
•
•
Short settling and discharge time to select rapid settling granule (Arrojo et al., 2004;
McSwain et al., 2004; Qin et al., 2004).
High volume exchange ratio (defined as the volume of effluent discharged divided by the
working volume of the SBR) (Liu et al., 2005a).
Feast and famine regime to enhance the conversion of readily biodegradable substrates
into slowly biodegradable stored substrate (Beun et al., 2001; de Kreuk and van
Loosdrecht, 2004).
High hydrodynamic shear force and high air upflow velocity to induce a compact, dense
and round granule structure (Liu and Tay, 2004).
13
Benefits of aerobic granular sludge technology for industrial wastewater treatment
The potential of the aerobic granular sludge technology is very promising, in particular for the
biological removal of high nutrient containing industrial wastewater such as abattoir effluents.
First of all, the excellent settleability of granular sludge allows for more biomass to be
maintained in a relatively small reactor volume, enhancing the ability of the reactor to
withstand high loading rates and reducing the capital costs associated with the construction of
large bioreactors. Secondly, the large size of granules and their dense and compact structure
are expected to positively contribute to the oxygen mass transfer limitation required for stable
and reliable SNDPR systems. Therefore, combining SNDPR and granular sludge technologies
could indeed result in a very attractive novel BNR technology for industrial application.
14
3.0
Thesis Overview
This chapter offers an overview of the research work undertaken in this thesis. First, the
research objectives are formulated based on the primary objectives of this study (Chapter 1),
then the research methods employed to address theses objectives are briefly described and
finally, the major outcomes of this research are outlined.
3.1.
Research Objectives
i.
Development of a process that achieves high levels of COD, nitrogen and
phosphorus removal from abattoir wastewater.
As explained in the previous chapter, the biological removal of COD and nitrogen from
abattoir wastewater has been widely studied resulting in successful full-scale applications
whereas in comparison, biological phosphorus removal has not received great attention due to
the general incompatibility between high level of nitrate/nitrite accumulation and stable EBPR
processes. As a result, most full-scale WWTPs in the meat industry remove phosphorus
through chemical precipitation by dosing iron or alumina ions. This method adds considerable
costs to the overall abattoir effluent treatment and produces a sludge rich in metal ions
limiting possible further application of this sludge as soil enrichment alternative. The first
objective of this thesis is to investigate the feasibility of biologically removing COD, nitrogen
and phosphorus to high-levels (>95%) from abattoir wastewater to produce an effluent of river
discharge quality using an innovative SBR process aimed at limiting the nitrate/nitrite
accumulation.
ii.
Design of an automatic control system to consistently achieve nitrification and
denitrification via nitrite.
The accumulation of nitrate/nitrite is not the only challenge faced by the meat industry to
develop a reliable and cost effective biological phosphorus removal process. The limited
amount of easily biodegradable COD contained in abattoir effluent is another source of
concern for the removal of the high levels of N and P. It was established in the previous
chapter that the use of the nitrite pathway to remove the high level of N present in abattoir
wastewater could save significant amount of COD and improve the overall performance of the
BNR system. Automatic control systems based on pH and DO signals have been used in
several studies to promote nitrification and denitrification via nitrite in ammonia rich effluent.
However, these studies relied upon the dosage of an external carbon source rather than the
COD contained in the original wastewater to achieve a sufficient level of denitrification. The
use of external carbon sources considerably undermines the overall benefits of the nitrite
pathway. The second objective of this thesis is to implement an automatic control strategy to
the SBR process developed under the first objective in order to achieve nitrite pathway
without the need of external carbon source dosing. This automatic control strategy would
improve COD usage efficiency and thus improve the reliability and robustness of biological
phosphorus removal processes for the meat industry.
15
iii.
Development of an operational strategy to maintain the biomass activity during
long term starvation condition and during the subsequent resuscitation period.
Abattoir wastewater treatment plants (WWTPs) have to cope with large fluctuations of the
wastewater flow and composition inherent to most industrial activities. In some cases, low
activity periods for the industry (e.g. annual maintenance or seasonal production variations)
would even result in complete interruptions of wastewater flows to the WWTPs for weeks and
even months. It is crucially important to maintain the viability of biomass during the long idle
or starvation periods and to ensure the process can fully recover when normal operations are
resumed. Based on the limited literature available on the subject, it appears that keeping the
sludge under alternating aerobic, anaerobic/anoxic conditions could be a suitable strategy for
maintaining the biomass activities during an extended starvation period. However, most
studies investigated the effect of different starvation conditions (aerobic, anoxic or anaerobic)
on nitrifiers species only (i.e. AOB and NOB). This thesis aims to develop an integrative
operational strategy easily applicable for the industry to (i) preserve the N and P removal
capability of the biomass during long starvation periods and (ii) to resuscitate the biomass
once the starvation period ceases.
iv.
Identification of the causes of N2O production in an SNDPR process enabling its
application to the treatment of abattoir wastewater.
The possible benefits of implementing the SNDPR technology to the treatment of abattoir
wastewater have been presented in the previous Chapter (i.e. lower COD demand, less
accumulation of nitrate/nitrite, lower aeration costs and smaller sludge production). However,
N2O rather than N2 has been observed in the SNDPR process as the end-product of
denitrification, which poses a significant environmental problem. Before looking into the
future application of SNDPR to abattoir wastewater treatment the causes of N2O
production/accumulation must be first identified and then eliminated. This forms another aim
of this thesis.
v.
Determination of the microscale microbial distribution of PAO and GAO in
aerobic granules
As defined in Chapter 2, the aerobic granular sludge technology could be beneficial for the
stability and reliability of the SNDPR process. Due to their large size and dense structure,
aerobic granules are expected to positively contribute to the oxygen mass transfer limitation
required for reliable SNDPR. As a result, different conditions (i.e. aerobic, anoxic or
anaerobic) are likely to co-exist at different depths in granules creating different ecological
micro-niches. It was also established in Chapter 2 that without denitrification by PAOs, there
is no true link between SND and EBPR with the two processes merely occurring in the same
sludge at the same time and thus compromising the carbon savings proposed to be obtained by
SNDPR. If aerobic granules are to positively contribute to SNDPR, denitrification should be
achieved by PAOs and not by GAOs. Therefore, PAOs should be the dominant population in
the central part of the granules where no oxygen is present (anoxic zones) and denitrification
occurs. To demonstrate ecological positions and roles for both populations in this complex
system, a novel microbial method is developed in this thesis to establish the spatial
distribution of the main PAO and GAO population in SNDPR granules, namely
Accumulibacter spp. and Competibacter spp., respectively.
16
vi.
Investigation of the feasibility of COD, N and P removal from abattoir
wastewater using granular sludge technology.
As discussed in Chapter 2, the excellent settleability of aerobic granular sludge allows for
more biomass to be maintained in a relatively small reactor volume, enhancing the ability of
the reactor to withstand high loading rates. This is of great interest for the treatment of high
nutrient containing industrial wastewater such as abattoir wastewater compared to
conventional floccular sludge systems. To date, aerobic granular sludge technology has
mainly been studied with synthetic wastewater, with the only real wastewater studies being
performed using domestic wastewater. The feasibility of achieving high-levels of COD, N and
P removal from abattoir wastewater using granular sludge technology is investigated in this
thesis. In addition, the microscale structure of these aerobic granules is examined using a
multi-disciplinary approach.
3.2.
Research Methods
In this thesis, three different lab-scale SBRs were operated to demonstrate the feasibility to
achieve high-levels of biological COD, N and P removal from abattoir wastewater. The first
SBR, referred to as RivD-SBR, was employed to experiment novel strategies, which can
easily be implemented to current water treatment facilities used by the meat industry, to
produce an effluent suitable for river discharge (>95% TCOD, TN and TP removal). RivDSBR was used to address research objectives 1, 2 and 3. The second SBR, referred to as
SNDPR-SBR, was operated to provide a platform for an in-depth investigation of the
previously proposed simultaneous nitrification, denitrification and phosphorus removal
(SNDPR) process, which has the potential for application to the treatment of abattoir
wastewater. SNDPR-SBR was first employed to identify the causes of N2O emission in labscale SNDPR bioreactors fed with synthetic wastewater (research objective 4). The operation
of this reactor was then modified to transform its floccular biomass into granules, which are
believed to reinforce the SNDPR process. The microbial community structure in this new
granular SNDPR-SBR was then examined via a newly developed method (research
objective 5). The third SBR, referred to as Granular-SBR, was operated for the treatment of
the high nutrient-containing abattoir effluent (research objective 6).
The operations of the three SBRs are briefly described below, which are followed by an
overview of the key analytical methods employed in this thesis to address all of the above
research objectives.
3.2.1. Operation of lab-scale SBRs used in this thesis
RivD-SBR
RivD-SBR had a working volume of 7 l (Figure 9) and was operated with a cycle time of 6 h
in a temperature-controlled room (18-22°C). Each cycle, 1 l of abattoir wastewater collected
weekly from a local abattoir (mixture of primarily treated and anaerobically treated
wastewater) and stored in a cold room at 4°C was pumped into the reactor over 3 feeding
periods. Each feeding period was followed by an anaerobic/anoxic period and an aerobic
period (Table 2). During the aerobic periods, air was provided intermittently using an on/off
control system to keep the DO level between 1.5 and 2 mgO2.l-1. The HRT and SRT in the
SBR were kept constant at 42 h and 15 days, respectively. Full description of the SBR
operation is given in Appendix A.
17
MIXER
WASTAGE
ToC
meter
7L
Working volume
6L
EFFLUENT
pH
meter
Solenoid
valve
Extra
VFA
DO
meter
ORP
meter
pumps
Air
flowmeter
Prefermented
RAW
effluent
Air
Anaerobic
POND
effluent
Figure 9. Design of RivD-SBR.
Table 2. Operating conditions of RivD-SBR (6 h cycle).
Duration
(min)
Fill no-mix 1
5
No-aerated mix 1*
30
Aerated mix 1
55
Fill mix 2
3
No-aerated mix 2*
70
Aerated mix 2
35
Fill mix 3
2
No-aerated mix 3*
60
Aerated mix 3
18
Wastage
2
Settle
70
Decant
10
*
anoxic or anaerobic, depending when nitrate and nitrite are depleted.
SBR sequences
SNDPR-SBR
A 5 l SBR performing SNDPR, which was previously designed and operated at the AWMC
lab, was used in this thesis (Figure 10) to identify the causes for N2O accumulation in labscale SNDPR processes (research objective 4). Initially, the SBR 6h cycle consisted of a 90
min anaerobic, 220 min aerobic, 40 min settling, and 10 min decanting period. In each cycle,
3 l of synthetic wastewater (230 mgCOD.l-1 as acetate, 23 mgN.l-1 as NH4+ and 18 mgP.l-1 as
PO43-) was pumped into the reactor resulting in an HRT of 10 h. The SRT was kept at 20 days.
Aeration was provided intermittently using an on/off control system to keep the DO level
relatively low between 0.35 and 0.5 mg.l-1. Full description of the initial SBR operation is
given in Appendix D.
18
WASTAGE
5L
Working volume
ToC
meter
pH
meter
DO
meter
2L
EFFLUENT
flowmeter
N2
Solenoid valve
Magnetic
stirrer
ORP
meter
pump
Synthetic
FEED
Air
Figure 10. Design of SNDPR-SBR.
The SBR operation was then modified to promote the formation of aerobic granules. The
settling time was gradually reduced from 40 min to only 5 min and the nutrient concentration
in the synthetic wastewater was increased to 350 mgCOD.l-1 as acetate, 35 mgN.l-1 as NH4+
and 23 mgP.l-1 as PO43-. The cycle time was reduced to 4 h and consisted in 55 min anaerobic
period followed by 170 min aeration, 5 min settling, and 10 min decant. The DO was kept
between 1.3-1.7 mgO2.l-1. Full description of the new SBR operation is given in Appendix E.
Granular-SBR
The design of granular-SBR was similar than SNDPR-SBR (Figure 10) with a working
volume of 5 l. Granular-SBR was also operated in a temperature-controlled room (18-22°C).
The reactor was seeded with the granular sludge obtained from the lab-scale SNDPR-SBR fed
with synthetic wastewater. During the first 4 months of operation, the SBR was fed with a
mixture of anaerobically treated abattoir effluent and synthetic wastewater and the cycle time
varied between 4-8h. After the initial 4 month adaptation period, only anaerobically treated
abattoir effluent was fed into the SBR. Once steady state was established, the cycle time was
fixed at 8 h and consisted of an 18 min feeding, 60 min anaerobic/anoxic, 315 min aerobic, 80
min post-anoxic, 2 min settling and 5 min decanting period. During the aerobic period the DO
level was kept between 3.0 and 3.5 mgO2.l-1. The HRT gradually increased from 6.7 h during
the adaptation period to 13.3 h once stable operation was reached (3 l fed each cycle). Full
description of the granular SBR operation is given in Appendix F.
3.2.2. Analytical Methods
A wide range of techniques was employed to address the research objectives defined in this
thesis. It includes reactor process studies, microbial investigations (fluorescent in-situ
hybridisation – FISH combined with confocal laser scanning microscopy – CLSM) and
several micro-scale techniques (e.g. micro-sensors, electron microscopy and light
microscopy). The combination of these multi-disciplinary techniques has helped deliver
significant outcomes.
19
Chemical analysis
The ammonia (NH4+ + NH3), nitrate (NO3-), nitrite (NO2-) and phosphate-P (PO43-P) were
analysed using a Lachat QuikChem8000 Flow Injection Analyser (Lachat Instrument,
Milwaukee). Dissolved nitrous oxide (N2O) was measured on-line with a N2O microsensor
constructed according to Andersen et al. (2001). Total and soluble COD (TCOD and SCOD),
soluble BOD5, total and soluble Kjeldahl nitrogen (TKN and SKN), total phosphorus and
total dissolved phosphorus (TP and TDP), mixed liquor suspended solid (MLSS) and volatile
MLSS (MLVSS) were analysed according to the standard methods (APHA, 1995). The major
ions present in the SBRs bulk liquid (Ca2+, Fe2+, K+, Mg2+, Na+ and HS-) were measured by
Inductively Coupled Plasma - Atomic Emission Spectrometry (ICP-AES Varian Vista-PRO,
Varian, Inc.). VFAs were measured by Perkin-Elmer gas chromatography with column DBFFAP 15 m x 0.53 mm x 1.0 µm (length x ID x film) at 140°C, while the injector and FID
detector were operated at 220°C and 250°C, respectively. High purity helium was used as
carrier gas at a flow rate of 17 ml.min-1. Polyhydroxylalkanoates (PHA=PHB+PHV+PH2MV)
and glycogen were determined using the method described in Oehmen et al. (2005). The
phosphorus fractionation in granules was determined using the cold perchloric acid (PCA)
extraction procedure developed by De Haas et al. (2000) as detailed in Appendix F.
Microbial analysis
Sludge samples were fixed and FISH probed as previously described (Amann, 1995).
Oligonucleotide probes used in this thesis were the combination of EUB338 i-iii (EUBmix)
for the detection of all bacteria (Daims et al., 1999), the combination of PAO462, PAO651
and PAO846 (PAOmix) for Accumulibacter spp. (Crocetti et al., 2000), the probe
combination (GAOmix) of GAOQ989 (Crocetti et al., 2002) and GB_G2 (Kong et al., 2002)
for Competibacter spp., NTSPA662 (Daims et al., 2001) for Nitrospira spp., NIT3 (Wagner et
al., 1996) for Nitrobacter spp. and NSO1225 (Mobarry et al., 1996) for most of the ammonia
oxidising bacteria (AOB) from the Betaproteobacteria. Additionally, probes for the proposed
Actinobacterial PAOs (Kong et al., 2005) and Defluviicoccus spp.-related GAOs (Meyer et
al., 2006) were also used. Fluorescently labelled oligonucleotides were purchased from
Thermo (Ulm, Germany) with fluorescein isothiocyanate (FITC) or one of the
sulfoindocyanine dyes indocarbocyanine (Cy3) or indodicarbocyanine (Cy5). FISH images
were collected with a Zeiss LSM 510 (Carl Zeiss, Jena, Germany) CLSM using an argon laser
(488 nm), a helium neon laser (543 nm) and a red diode laser (633 nm) fitted with 515-565
nm BP, 590 nm LP and 660-710 nm BP emission filters, respectively. Two different Zeiss oil
immersion objectives were used, a Plan-Neofluar 40x/1.3 and a Plan-Apochromat 63x/1.4.
FISH quantification was performed according to Crocetti et al. (2002) where the relative
abundance of each group was determined as mean percentage of all bacteria based on pixel
area counting.
Nile-Blue A staining (Ostle and Holt, 1982) was used to determine cells containing
intracellular PHA and visualised on the confocal laser scanning microscope (see above).
Fixed granule samples for FISH and Nile Blue A staining were embedded in optimum cutting
temperature (OCT) compound (TissueTek, Sakura, USA) for cryosectioning as previously
described (Meyer et al., 2003). Embedded granules were then frozen and sectioned into 10 µm
thick slices using a cryotome operated at -20°C (Kryo 1720, Leitz, Germany). The granule
sections were collected on SuperFrost Plus microscope slides (Menzel-Glaser, Germany).
Finally, the slides were dehydrated by sequential immersion for 3 min in 50%, 80% and 98%
ethanol, followed by air-drying.
20
Physical analysis
To monitor the granule structure and characteristics, granule size distribution and density were
measured. To determine the size distribution of the granules, 30 ml of well mixed granular
sludge was sampled from the SBR at the end of the aeration period and pumped through a
Malvern laser light scattering instrument, Mastersizer 2000 series (Malvern Instruments,
Worcestershire, UK). The granule density, defined as the quantity of dry mass per biomass
volume, was measured by the blue dextran method adapted from Di Iaconi et al. (2004) and
further described in Appendix E.
The gradient of oxygen in granules was measured with oxygen microsensors (tip diameter
<10 µm), which were constructed as described by Revsbech et al. (1989). Granules were
transferred to a flow-cell with an upward flow where replicate oxygen profiles were then
measured and averaged as described in Meyer et al. (2003). The pH gradient in granules was
also measured with pH microsensors using the same experimental set-up.
Whole granules were photographed using an Olympus SZH10 stereo microscope mounted
with a digital camera. To visualise the structure of aerobic granules at a microscale level,
scanning and transmission electron microscopy (SEM and TEM, respectively) were
employed. Prior to visualisation, granules were fixed in subsequent chemical solutions to
minimise any structural artefacts arising from the dehydratation process (see Appendix G for
more details). For TEM studies, dehydrated granules were embedded in Epon resin and
ultrathin sections were cut and mounted on copper grid for visualisation with a JEOL 1010
transmission electron microscope operated at 80 kV. For SEM studies, dehydrated granules
were mounted onto metal stubs and sputter coated with platinum to reduce charging. Viewing
of samples was conducted using a JEOL 6300F scanning electron microscope operated at 5 or
10 kV. To observe the internal structure, some dehydrated granules were frozen in liquid
nitrogen, fractured, sputter coated with platinium and visualised as described above.
3.3.
i.
Research Outcomes
Multi-step feeding strategy to prevent the accumulation of NOx-
Abattoir wastewater contains a high level of ammonia and organic nitrogen and the complete
nitrification of these nitrogen compounds produces a high level of nitrate and/or nitrite, which
has proved to be an obstacle to the development of a stable and reliable Bio-P removal
process. The detrimental effect of nitrate/nitrite accumulation for biological P removal
processes was discussed in Chapter 2. To address this challenge, a multi-step feeding strategy
was implemented to keep the nitrate/nitrite concentration low during the entire SBR cycle
providing suitable anaerobic conditions to perform reliable Bio-P removal. The SBR feeding
strategy was split into 3 separate feeding stages each followed by a non-aerobic and aerobic
periods. 50% of the total influent volume was fed during the first feeding, 30% during the
second and 20% during the third. The “UniFed” filling process described in Chapter 2 was
employed for the first feeding period. Fresh influent was introduced directly into the sludge
blanket from the bottom of the SBR once the settling period was finished.
21
anO2
18
O2
anO2
anO2
O2
O2
settle+decant
30
14
PO4-P
12
NH4-N
25
20
NOx-N
10
15
8
6
10
P (mg L-1)
N (mg L-1)
16
4
5
2
0
0
0
50
100
150
200
250
300
350
Time (min)
Figure 11. Nitrogen and phosphorus profiles during a cycle study of RivD-SBR. The vertical
arrows indicate the 3 feeding periods.
This multi-feed strategy successfully limited the level of NOx- recycled to the anaerobic
period. Figure 11 shows the nitrogen and phosphorus transformations in a typical SBR cycle
study during the steady state period of the RivD-SBR. At the end of each aerobic period,
NH4+ was fully oxidised, and the low level of NOx- accumulated was then removed in the
following anoxic period. Very low level of NOx- was carried over to the next cycle, and was
denitrified very quickly at the beginning of the first anaerobic period. Most of the anaerobic P
release by PAOs occurred in the first non-aerated period. This indicates that the first nonaerated period was crucially important for P removal as PAOs could freely utilise all the VFA
available without having to compete with heterotrophic denitrifiers. This multi-step strategy
allowed RivD-SBR to consistently achieve 95%, 97% and 98% of TCOD, TN and TP
removal, respectively. The full detail of this study is provided in Appendix A
ii.
Strategies to meet the carbon demand for biological nutrient removal
The performance of a biological nutrient removal system depends greatly on the availability
of easily biodegradable carbon sources in the wastewater, particularly VFAs. Therefore, it is
required to optimise the use of the limited available COD to remove the high levels of nutrient
in abattoir wastewater. Considering that it is difficult to control the VFAs content in large
anaerobic ponds, a more controllable VFA source is necessary for reliable biological nutrient
removal from abattoir wastewater. In this thesis, abattoir raw wastewater was subjected to a
one-day high-rate pre-fermentation step before being mixed with anaerobic ponds effluent to
increase the VFAs concentration of the wastewater fed into RivD-SBR. The pre-fermentation
was performed in a 50 l tank continuously mixed with the temperature kept at 37˚C. No
inoculum was introduced in the pre-fermentor, and hence the microbial population present in
the raw abattoir wastewater was used to carry out the fermentation. The overall VFA
concentration in the raw wastewater stream more than doubled as a direct result of this prefermentation. Acetate and propionate were the most abundant VFAs in the raw abattoir
wastewater before and after pre-fermentation with propionate having a slightly higher
production rate than acetate. However, it should be highlighted that the use of raw wastewater
should be minimised due to its high FOG content which would likely deteriorate the sludge
settleability.
22
An alternative approach to reduce the carbon demand is to achieve nitrogen removal via the
nitrite pathway as presented in Chapter 2. To achieve this nitrite pathway in RivD-SBR, an
on-line aeration phase length control system was integrated to the step-feed strategy described
previously. The control strategy employed was based on the slope of the pH signal and on the
oxygen uptake rate (OUR). The exact time of complete NH4+ oxidation (i.e. end of nitritation
process) in each aerobic period could be detected through the pH bending point and the sharp
OUR drop. The aeration could then be automatically switched off preventing the further
oxidation of nitrite into nitrate. Instead of using an external carbon source, the organic carbon
contained in wastewater was used for denitrification, thus delivering the true benefit of
implementing the nitrite pathway. Figure 12 presents the level of nitrite pathway achieved in
the SBR, measured as the average amount of NO2- produced per NOx- produced (mgN.l-1)
during the 3 aeration periods, and the relative abundance of NOBs in the SBR after
implementation of the aeration control strategy. RivD-SBR had been already running for 5
months performing high level of COD, N and P removal (see Appendix A) before the aeration
length control was implemented. During that time, no nitrite accumulation was observed
during the aerobic periods. The control of the length of each aeration period was first
implemented manually, which resulted in a rapid accumulation of nitrite reaching 95% of the
total amount of NOx- produced and a sharp decrease of the NOB population (Figure 12). Then,
the application of fixed aeration periods deteriorated the nitrite pathway previously
established and NOB population slightly recovered. Finally, the implementation of the
automatic aeration length control strategy re-established the nitrite pathway in the SBR. This
strategy was therefore successful in controlling the level of the nitrite pathway through the
elimination or the reduction of the NOB population in the system. Removing N via the nitrite
pathway benefited the nutrient removal performance of the SBR by reducing the demand for
COD. The full detail of this study is presented in Appendix B.
N-NO2 : N-NOX ratio (%)
80
70
60
50
1.8
Start
automatic
aeration
control
90
Start
manual
aeration
control
1.6
1.4
Starvation
period
1.2
1
Start overaeration for
NOB recovery
40
0.8
0.6
30
0.4
20
% NO2
NOB population
10
0
150
200
250
300
NOB : all bacteria ratio (%)
100
0.2
350
400
450
500
550
0
600
Days
Figure 12. Degree of nitrite accumulation and abundance of NOB Nitrospira (FISH probe
Nitspa-662) in RivD-SBR. The NOB quantification shown is an average (error bars=SE, n=3).
iii.
Strategy to maintain the biomass activity during long term starvation condition.
RivD-SBR was put twice into a so-called “sleeping mode” for a period of 5-6 weeks when the
abattoir from where the wastewater was sourced was closed for annual maintenance. The
“sleeping mode” operation consisted of 15 min mixed aeration and 345 settling in a 6 h cycle.
23
During the first starvation period, the nitrifying activity was closely monitored through
weekly pulse addition of ammonia and nitrite, while the activity of PAOs was only monitored
by measuring the phosphate-P concentration in the liquid phase. During the second starvation
period, the PAO activity was more comprehensively studied with weekly batch tests to
monitor the anaerobic and aerobic activities of PAOs. The recovery processes of these
organisms after the starvation were also investigated by applying a resuscitation strategy
consisting of gradually increasing the wastewater load in the first few days after normal
operation was resumed.
Batch tests monitoring the specific activity of nitrifiers and PAOs population in RivD-SBR
over the course of each starvation period demonstrate a clear activity drop for both
populations. However, after gradually resuming the normal reactor operation, the nitrification
rate, denitrification rate and the amount of P-release and P-uptake quickly improve and
reached their initial value within 4 days after both starvation periods (Table 3). Good nutrient
removal performances were consistently achieved for several months after each starvation
periods confirming that the strategies employed to maintain the biomass activity during long
term starvation condition and to resuscitate that same biomass were very successful.
Table 3. Nitrification rate (rNH4+), denitrification rate (rNOx-) and amounts of P released and
uptaken over a cycle, measured during cycle studies performed before the start of the
starvation period, immediately after the starvation period (50% of normal load), 2 days after
starvation (75% of normal load) and 4 days after the starvation (100% of normal load).
Parameter monitored
“Sleeping mode” I or II
Before starvation
After 1st cycle (50%)
After 2 days (75%)
After 4 days (100%)
rNH4+
(mgN.l-1.h-1)
I
II
18.2
25.5
8.2
7.4
12.9
20.8
17.6
29.1
rNOx(mgN.l-1.h-1)
I
II
4.8
12.3
1.9
1.8
4.5
9.6
5.7
11.7
P-release
(mgP.l-1)
I
II
18.8
36.9
2.5
4.4
9.6
31
19.6
47
P-uptake
(mgP.l-1)
I
II
16.1
34.2
2.3
4.8
8.1
28.9
17.4
44.3
iv.
Organisms responsible for denitrification in lab-scale floccular and granular
SNDPR bioreactors
FISH analysis performed on the SNDPR-SBR floccular sludge has demonstrated a very high
abundance of Accumulibacter spp. (PAOs) and Competibacter spp. (GAOs) accounting
together for 70% of the total bacteria present in the reactor. Figure 13 shows that
Accumulibacter spp. were always more abundant than Competibacter spp.. An interesting
trend appears in Figure 13 when comparing the N removal efficiency to the percentage of
GAOs in the system. Over the 5-months period, N removal efficiency decreased from 100%
to 53% and the Competibacter spp. population in the reactor simultaneously decreased from
19% to 8% of all bacteria. At the same time, the Accumulibacter spp. population increased
from 48% to 70%. These results strongly support the hypothesis that Competibacter spp. are
the microorganisms primarily responsible for the denitrification occurring in the SNDPR
reactor. It poses the question of whether there is a direct link between the enrichment of
Competibacter spp. and N2O accumulation from denitrification. More details of this study are
given in Appendix D.
24
100
0.75
0.70
80
0.65
60
0.60
0.55
40
P/C (mol/mol)
% of all bacteria or % N removed
Accumulibacter
Competibacter
N removal
P release/VFA uptake
0.50
20
0.45
0
0.40
Jun 7
Jul 12
Aug 23
Sep 20
Oct 25
Figure 13. Abundance of Accumulibacter and Competibacter correlated with the N removal
efficiency and the carbon-uptake to P-release ratio over the 5 month period.
In an attempt to demonstrate ecological positions and roles for each population in aerobic
granules, the spatial distribution of PAOs and GAOs in this complex system was correlated
with the dissolved oxygen profiles. The goal was to verify at a microscale level, which
microbial community between PAO and GAO was more likely responsible for the
denitrifcation in lab-scale granular SNDPR bioreactors fed with synthetic wastewater. A novel
method using fluorescence in situ hybridisation (FISH) and confocal laser scanning
microscopy (CLSM) was developed in this thesis to study the microscale distribution,
organisation, and community composition of the bacterial community in aerobic granules.
Oxygen profiles inside SNDPR granules were determined by microsensors. The exact method
is described in depth in Appendix E. The population distribution of Accumulibacter spp. (the
main PAO) and Competibacter spp. (the main GAO) within 24 different granules was
expressed as the relative abundance at different depth of Accumulibacter spp. divided by that
of Competibacter spp. (referred to as the PAO:GAO ratio). Using the ratio data, a mean
distribution was calculated and is presented in Figure 14 alongside the in-situ dissolved
oxygen profile. For each 50 µm zone of the granules, the PAO:GAO ratio strongly correlated
(significant at the 0.01 level) with the dissolved oxygen concentration (Pearson
correlation=0.86). Accumulibacter spp. was dominant (i.e. PAO:GAO ratio >1) in the aerobic
zones (0-200 µm) while Competibacter spp. dominated (i.e. PAO:GAO ratio <1) in the central
anoxic zones (200 µm inwards) of the studied granules. Therefore Competibacter spp. would
be mostly responsible for denitrification in SNDPR granules which confirms the results
obtained with floccular SNDPR systems also fed with synthetic wastewater. The low
microbial diversity observed in both floccular and granular SNDPR bioreactors was suggested
for this lack of denitrification by PAOs. The apparent role of GAOs in denitrification clearly
compromises the carbon savings proposed to be obtained by SNDPR. More information is
provided in Appendix E.
25
3.5
PAO/GAO ratio
3.0
PAO/GAO ratio
O2 profile
-1
0.5
O2 concentration (mg l )
0.6
2.5
0.4
2.0
0.3
1.5
0.2
1.0
0.1
0.5
0.0
0.0
0
50
100
150
200
250
300
350
400
Depth (µm)
Figure 14. Average profile of the PAO:GAO ratio within 24 granules (error bar=95%CL) and
mean O2 profiles in granules at the end of the aerobic period (error bars=S.D., n=6).
v.
Management of N2O accumulation in SNDPR
To elucidate the factors responsible for N2O accumulation in lab-scale SNDPR bioreactors
operated with synthetic wastewater and test how it can be prevented, several anoxic batch
tests were performed under different conditions. Sludge was sampled at the end of the
anaerobic period from the parent reactor (SNDPR-SBR) and transferred to two 14.75 ml vials
sealed with rubber stoppers to which a N2O microsensor was inserted for on-line monitoring
of the N2O concentration. The mixed liquor was stirred using a magnetic stirrer, and the vials
were filled completely to avoid any exchange of N2O between liquid and gas phases. The two
mini-reactors were operated in parallel with one acting as a negative control or a duplicate.
Substrates could be added in small amounts at any time during experimentation with a syringe
through the rubber stopper. Net N2O production and consumption rates could then be
measured with different electron acceptors (nitrate, nitrite and N2O) using different carbon
sources (intracellular PHA, acetate, propionate, methanol and abattoir wastewater). Both rates
were corrected for MLVSS variations in different batch tests.
It was hypothesised that if denitrification is carried out simultaneously by GAOs and other
denitrifiers in the SNDPR system, the N2O accumulated by GAOs could be reduced by other
denitrifiers, provided that carbon is available to these cells. To verify this hypothesis, anoxic
batch tests were also performed with a mixture of sludge from the SNDPR reactor and a labscale nitrifying-denitrifying reactor treating domestic wastewater containing additional carbon
in the form of methanol for denitrification. The addition of raw abattoir wastewater as carbon
source after 35 min incubation with nitrate led to an immediate reduction and depletion of
N2O (Figure 15). At the same time, the nitrate reduction rate increased, underlining the
capacity of non-PAO and non-GAO denitrifiers to reduce nitrate to N2 while removing N2O
accumulated by the SNDPR sludge simultaneously.
The accumulation of N2O in the SNDPR sludge is likely to be a result of the high enrichment
of PAOs and GAOs because of the application of synthetic wastewater containing a single
carbon source (i.e. acetate). If real wastewater is fed it is a realistic expectation that some
carbon would be available for denitrification by non-PAO or non-GAO organisms during the
low-DO aerobic period. The N2O accumulation is unlikely to be an issue in future application
26
of the SNDPR process to remove nutrients from abattoir wastewater. This was indeed verified
when abattoir wastewater was fed into Granular-SBR and SNDPR was achieved. Detailed
explanation of this study is given in Appendix D.
0.6
add high-strength WW
10
0.3
8
-1
N2O-N
NOx-N
0.4
6
0.2
4
0.1
2
0
N-NOx (mg l )
-1
N-N2O (mg l )
0.5
0
0
20
40
60
80
time (min)
Figure 15. N2O and NOx- concentration in the bulk liquid during an anoxic test in a 500 ml
reactor. Raw high-strength wastewater was added at T=35 min.
vi.
Nutrient removal from abattoir wastewater using aerobic granular sludge
technology
Good nitrogen and phosphorus removal performance was achieved once stable operation was
established. The organic, nitrogen and phosphorus loading rates applied were
2.7 gCOD.l-1.d-1, 0.43 gN.l-1.d-1 and 0.06 gP.l-1.d-1, respectively. The removal efficiency of
soluble COD, total dissolved nitrogen and total dissolved phosphorus were 89%, 93% and
88%, respectively. The remaining soluble COD measured in the effluent (162 mg.l-1) was
non-biodegradable as indicated by the very low soluble BOD5 value (<2 mg.l-1). The oxidised
nitrogen accumulating at the end of each cycle (i.e. around 10 mgN.l-1) was almost
exclusively nitrite (94.7%, SD=2.5%, n=63) suggesting that N was removed via the nitrite
pathway in this granular SBR. However, the high suspended solids in the effluent (around
0.3 g.l-1) limited the overall removal efficiency to 68%, 86% and 74% for total COD, total
nitrogen and total phosphorus, respectively. Interestingly, Competibacter spp. was no longer
detected in the granules via FISH analysis and Accumulibacter spp. were found to be involved
in the denitrification indicating that true SNDPR occurred in this granular SBR fed with real
wastewater. It was also found that the minimum hydraulic retention time in this aerobic
granular sludge system was not governed by the sludge settleability, as is the case in a
floccular sludge system, but likely by the limitations associated with the transfer of substrates
in granules. More details of this study can be found in Appendix F. The structure of these
aerobic granules fed with nutrient rich industrial wastewater was also studied at a microscale
level. Observations were made using a wide range of techniques including light microscopy,
scanning and transmission electron microscopy, fluorescent in-situ hybridisation (FISH)
combined with confocal laser scanning microscopy (CLSM) and oxygen and pH
microsensors. This multi-disciplinary approach allowed for interpretations to be made about
the general structure and fate of mature granules, the microbial community structure, the
effect of pH on the granule structure stability and the possible role played by protozoa in the
overall system performance. Images of the granule structure and their interpretations are
provided in Appendix G.
27
4.0
4.1.
Conclusions and Recommendations for Future Work
Main Conclusions of the Thesis
A sequencing batch reactor system was demonstrated to effectively remove nitrogen,
phosphorus and COD from abattoir wastewater. This provides a more cost-effective
alternative to chemical phosphorus removal, the current practice in the meat industry which
requires the addition of large amount of chemicals. The implementation of this novel SBR
technology to the existing water treatment facilities employed in the meat industry is believed
to be relatively simple and realistic. Indeed, this lab-scale RivD-SBR technology has been
recently scaled up to a 10 m3 on-site pilot plant for further trials. The following conclusions
can be drawn:
•
•
•
•
•
It is possible to achieve a high degree (>98%) of biological phosphorus removal from
abattoir wastewater in the presence of high levels of nitrogen (200 – 300 mgN.l-1) using a
SBR process.
The multi-step feeding strategy prevents high-level accumulation of nitrate or nitrite, and
hence facilitates the creation of anaerobic conditions in the SBR. This strategy is strongly
recommended for practical use in the biological treatment of abattoir wastewater.
It is important to incorporate a high-rate pre-fermentor as an integrated component of the
nutrient removal system in SBR. This stream, which contains a high-level of VFAs, is
effective in providing supplementary carbon sources for both phosphorus and nitrogen
removal.
The aeration control strategy consisting of stopping the aeration in the SBR immediately
after NH4+ is oxidised is effective in achieving stable N removal via the nitrite pathway. It
benefits the nutrient removal performance of the SBR by saving some valuable amount of
COD.
The intermittent aeration of 15 minutes in every 6 hours is effective in maintaining the
biomass activities of activated sludge performing biological nitrogen and phosphorus
removal. Sludge can be stored under such conditions for at least six weeks with its
nitrifying, denitrifying and phosphorus removal capabilities adequately maintained to
allow for a quick recovery when wastewater feed resumes. The resuscitation strategy of
gradually increasing the wastewater load in the first few days after a starvation period was
also demonstrated to be successful.
Two novel technologies were also identified to have strong potentials to remove nutrient from
abattoir wastewater, namely the simultaneous nitrification, denitrification and phosphorus
removal (SNDPR) process and the aerobic granular sludge technology. The main conclusions
drawn from the in-depth study of these two technologies are:
•
GAOs and not PAOs are primarily responsible for the denitrification in lab-scale SNDPR
bioreactor treating synthetic wastewater with floccular biomass. This compromises the
carbon savings proposed to be obtained when using SNDPR process to remove nutrient
from wastewater.
28
•
•
•
•
The production of N2O in lab-scale SNDPR bioreactors is likely linked to the loss of
diversity amongst the denitrifying microbial community due to the use of synthetic
wastewater containing only a single carbon source. However, N2O accumulation is
unlikely to be an issue in a SNDPR bioreactor treating real wastewater as such wastewater
contains a combination of different carbon sources. It will enable denitrifiers other than
PAO and GAO to participate to the denitrification and reduce the N2O presumably
produced by GAO.
The size and the dense structure of aerobic granules positively contributed to the oxygen
mass transfer limitation required to achieve reliable SNDPR process. Large and stable
anoxic zones were created in the centre part of the granule which provided a better
coupling between nitrification and denitrification. However, GAOs were still the main
denitrifier in lab-scale granular SNDPR bioreactor treating synthetic wastewater.
Granular sludge can be maintained in a bioreactor operated under alternating anaerobic
and aerobic conditions using anaerobically pre-treated abattoir wastewater as feed. Highlevels of COD, nitrogen and phosphorus removal can be achieved through true SNDPR
process with PAOs carrying out the denitrification. However, the effluent produced
requires post-treatment in order to remove the suspended solids before discharge into the
receiving environment.
The minimum HRT for a granular sludge system is not governed by the sludge
settleability and retention, as is the case in a system with floccular sludge. Mass transfer
limitations in granules are likely an important factor to be considered in the design of the
HRT and the COD and nutrient loading rate in a granular sludge system.
4.2.
Recommendation for Future Research
During the course of this PhD thesis, a number of other questions were raised that call for
further investigations. Recommendations for future research in this field are listed below:
•
•
•
The lab-scale SBR process developed in this PhD thesis for the treatment of abattoir
wastewater still needs to be scaled up to a pilot and/or full scale treatment plant in order to
validate the results obtain in this thesis and offer practical application for the industry. The
specific feed mixture used in this study (80-90% anaerobic pond effluent supplemented
with extra VFAs and 10-20% pre-fermented raw wastewater) may potentially limit the
wider validity of results. Further investigations using different mixture ratio and better
quality anaerobic pond effluent would therefore be beneficial.
The operation and design of anaerobic pre-treatment processes currently employed in most
abattoirs should be oriented towards maximising VFA production instead of solely
focussing on removing COD. This will facilitate the downstream biological nutrient
removal treatment of abattoir wastewater.
The SBR cycle operation could be improved by the use of on-line NH4+, NO3-, NO2- and
PO43- sensors which are getting more and more reliable. Efficient control strategies could
then be designed based on these on-line sensors as biological processes such as
nitrification, denitrification and aerobic P-uptake would be directly monitored during the
SBR cycle.
29
•
•
•
•
The optimal aeration frequency and duration of the alternating anoxic/anaerobic and
aerobic strategy to maintain the biomass activities of BNR process during long starvation
period are yet to be identified through further experimental studies.
Interesting structural properties of aerobic granules have been observed in this thesis using
a multi-disciplinary approach. However, more specific studies are needed to fully
understand the granule formation process and the exact role of EPS, mineral precipitation
and protozoa in the overall behaviour of aerobic granules. The impact of pH fluctuations
on the granule structure stability requires also further experimental work.
In depth microsensors studies of aerobic granules could provide opportunities to model the
complex diffusion processes occurring inside these granules and help to better understand
and possibly predict the overall performance of granular sludge systems.
Aerobic granular sludge technology seems to be particularly well suited for the treatment
of industrial wastewater. The stability of aerobic granules under starvation condition needs
to be investigated as industries are often subjected to temporal fluctuation of water usage
which directly affects the flow and/or concentration of the effluent entering the treatment
plant.
30
References
Al-Ghusain I. and Hao O.J. (1995). Use of pH as control parameter for aerobic/anoxic sludgedigestion. Journal of Environmental Engineering-Asce. 121(3): 225-235.
Al-Ghusain I.A., Huang J., Hao O.J. and Lim B.S. (1994). Using pH as a real-time control
parameter for waste-water treatment and sludge-digestion processes. Water Science
and Technology. 30(4): 159-168.
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
Anderottola G., Canziani R., Foladori P., Ragazzi M. and Tatano F. (2000). Laboratory scale
experimentation for RBCOD production from OFMSW for BNR systems: Results and
kinetics. Environmental Technology. 21(12): 1413-1419.
Andersen K., Kjaer T. and Revsbech N.P. (2001). An oxygen insensitive microsensor for
nitrous oxide. Sensors and Actuators B-Chemical. 81(1): 42-48.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Arrojo B., Mosquera-Corral A., Garrido J.M. and Mendez R. (2004). Aerobic granulation
with industrial wastewater in sequencing batch reactors. Water Research. 38(14-15):
3389-3399.
Artan N., Wilderer P., Orhon D., Morgenroth E. and Ozgur N. (2001). The mechanism and
design of sequencing batch reactor systems for nutrient removal - the state of the art.
Water Science and Technology. 43(3): 53-60.
Bertanza G. (1997). Simultaneous nitrification-denitrification process in extended aeration
plants: Pilot and real scale experiences. Water Science and Technology. 35(6): 53.
Beun J.J., Heijnen J.J. and van Loosdrecht M.C.M. (2001). N-removal in a granular sludge
sequencing batch airlift reactor. Biotechnology and Bioengineering. 75(1): 82-92.
Beun J.J., Hendriks A., Van Loosdrecht M.C.M., Morgenroth E., Wilderer P.A. and Heijnen
J.J. (1999). Aerobic granulation in a sequencing batch reactor. Water Research.
33(10): 2283-2290.
Bhamidimarri S.M.R. (1991). Appropriate industrial-waste management technologies - the
New-Zealand meat industry. Water Science and Technology. 24(1): 89-95.
Borja R., Banks C.J. and Wang Z. (1994). Performance and Kinetics of an Upflow Anaerobic
Sludge Blanket (Uasb) Reactor Treating Slaughterhouse Waste-Water. Journal of
Environmental Science and Health Part a-Environmental Science and Engineering &
Toxic and Hazardous Substance Control. 29(10): 2063-2085.
Bortone G., Malaspina F., Stante L. and Tilche A. (1994). Biological nitrogen and phosphorus
removal in an anaerobic/anoxic sequencing batch reactor with separated biofilm
nitrification. Water Science and Technology. 30(6): 303-313.
Brdjanovic D., Slamet A., van Loosdrecht M.C.M., Hooijmans C.M., Alaerts G.J. and Heijnen
J.J. (1998). Impact of excessive aeration on biological phosphorus removal from
wastewater. Water Research. 32(1): 200-208.
Caixeta C.E.T., Cammarota M.C. and Xavier A.M.F. (2002). Slaughterhouse wastewater
treatment: evaluation of a new three-phase separation system in a UASB reactor.
Bioresource Technology. 81(1): 61-69.
Chang C.H. and Hao O.J. (1996). Sequencing batch reactor system for nutrient removal: ORP
and pH profiles. Journal of Chemical Technology and Biotechnology. 67(1): 27-38.
31
Comeau Y., Hall K., Hancock R. and Oldham W. (1986). Biochemical model for enhanced
biological phosphorus removal. Water Research. 20: 1511-1521.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
de Bruin L.M.M., de Kreuk M.K., van der Roest H.F.R., Uijterlinde C. and van Loosdrecht
M.C.M. (2004). Aerobic granular sludge technology: an alternative to activated
sludge? Water Science and Technology. 49(11-12): 1-7.
de Haas D.W., Wentzel M.C. and Ekama G.A. (2000). The use of simultaneous chemical
precipitation in modified activated sludge systems exhibiting biological excess
phosphate removal Part 2: Method development for fractionation of phosphate
compounds in activated sludge. Water Sa. 26(4): 453-466.
de Kreuk M.K. and van Loosdrecht M.C.M. (2004). Selection of slow growing organisms as a
means for improving aerobic granular sludge stability. Water Science and Technology.
49(11-12): 9-17.
de Villiers G.H. and Pretorius W.A. (2001). Abattoir effluent treatment and protein
production. Water Science and Technology. 43(11): 243-250.
Di Iaconi C., Ramadori R., Lopez A. and Passino R. (2004). Preliminary biomass
characterization in a sequencing batch biofilm reactor. Annali di Chimica. 94(12): 889898.
Dulekgurgen E., Ovez S., Artan N. and Orhon D. (2003). Enhanced biological phosphate
removal by granular sludge in a sequencing batch reactor. Biotechnology Letters.
25(9): 687-693.
El-Mamouni R., Leduc R. and Guiot S.R. (1998). Influence of synthetic and natural polymers
on the anaerobic granulation process. Water Science and Technology. 38(8-9): 341347.
Etterer T. and Wilderer P.A. (2001). Generation and properties of aerobic granular sludge.
Water Science and Technology. 43(3): 19-26.
Fuerhacker M., Bauer H., Ellinger R., Sree U., Schmid H., Zibuschka F. and Puxbaum H.
(2000). Approach for a novel control strategy for simultaneous
nitrification/denitrification in activated sludge reactors. Water Research. 34(9): 24992506.
Furumai H., Kazmi A.A., Fujita M., Furuya Y. and Sasaki K. (1999). Modeling long term
nutrient removal in a sequencing batch reactor. Water Research. 33(11): 2708-2714.
Gaterell M.R., Gay R., Wilson R., Gochin R.J. and Lester J.N. (2000). An economic and
environmental evaluation of the opportunities for substituting phosphorus recovered
from wastewater treatment works in existing UK fertiliser markets. Environmental
Technology. 21(9): 1067-1084.
32
Grotenhuis J.T.C., Vanlier J.B., Plugge C.M., Stams A.J.M. and Zehnder A.J.B. (1991). Effect
of ethylene glycol-bis(beta-aminoethyl ether)-N,N-tetraacetic acid (egta) on stability
and activity of methanogenic granular sludge. Applied Microbiology and
Biotechnology. 36(1): 109-114.
Hao O.J. and Huang J. (1996). Alternating aerobic-anoxic process for nitrogen removal:
Process evaluation. Water Environment Research. 68(1): 83-93.
Hopwood D. (1977). Effluent treatment in meat and poultry-processing Industries. Process
Biochemistry. 12(8): 5-&.
Ingildsen P. and Olsson G. (2002). Exploiting online in-situ ammonium, nitrate and phosphate
sensors in full-scale wastewater plant operation. Water Science and Technology. 46(45): 139-147.
Ingildsen P. and Wendelboe H. (2003). Improved nutrient removal using in situ continuous
on-line sensors with short response time. Water Science and Technology. 48(1): 95102.
Irvine R.L. and Busch A.W. (1979). Sequencing batch biological reactors - overview. Journal
Water Pollution Control Federation. 51(2): 235-243.
Johns M.R. (1995). Developments in wastewater treatment in the meat processing industry: A
review. Bioresource Technology. 54(3): 203-216.
Johns M.R., Harrison M.L., Hutchinson P.H. and Beswick P. (1995). Sources of nutrients in
wastewater from integrated cattle slaughterhouses. Water Science and Technology.
32(12): 53-58.
Keller J., Subramaniam K., Gosswein J. and Greenfield P.F. (1997). Nutrient removal from
industrial wastewater using single tank sequencing batch reactors. Water Science and
Technology. 35(6): 137-144.
Keller J., Watts S., Battye-Smith W. and Chong R. (2001). Full-scale demonstration of
biological nutrient removal in a single tank SBR process. Water Science and
Technology. 43(3): 355-362.
Kerrn-Jespersen J.P., Henze M. and Strube R. (1994). Biological phosphorus release and
uptake under alternating anaerobic and anoxic conditions in a fixed-film reactor.
Water Research. 28(5): 1253-1255.
Koenig A. and Yiu W.C. (1999). Waste management in Hong Kong abattoirs. Water Science
and Technology. 40(1): 379-387.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kong Y.H., Nielsen J.L. and Nielsen P.H. (2005). Identity and ecophysiology of uncultured
actinobacterial polyphosphate-accumulating organisms in full-scale enhanced
biological phosphorus removal plants. Applied and Environmental Microbiology.
71(7): 4076-4085.
Kuba T., Smolders G.J.F., van Loosdrecht M.C.M. and and Heijnen J.J. (1993). Biological
phosphorus removal from wastewater by anaerobic and anoxic sequencing batch
reactor. Water Science and Technology. 27(5-6): 241-252.
Kuba T., Wachtmeister A., Vanloosdrecht M.C.M. and Heijnen J.J. (1994). Effect of nitrate
on phosphorus release in biological phosphorus removal systems. Water Science and
Technology. 30(6): 263-269.
Lemaire R., Meyer R., Taske A., Crocetti G.R., Keller J. and Yuan Z.G. (2006). Identifying
causes for N2O accumulation in a lab-scale sequencing batch reactor performing
simultaneous nitrification, denitrification and phosphorus removal. Journal of
Biotechnology. 122(1): 62-72.
33
Lettinga G., Vanvelsen A.F.M., Hobma S.W., Dezeeuw W. and Klapwijk A. (1980). Use of
the upflow sludge blanket (USB) reactor concept for biological wastewater-treatment,
especially for anaerobic treatment. Biotechnology and Bioengineering. 22(4): 699-734.
Li C.T., Shieh W.K., Wu C.S. and Huang J.S. (1986). Chemical Bio-Fluidized Bed Treatment
of Slaughterhouse Waste-Water. Journal of Environmental Engineering-Asce. 112(4):
718-728.
Lin Y.F. and Jing S.R. (2001). Characterization of denitrification and nitrification in a stepfeed alternating anoxic-oxic sequencing batch reactor. Water Environment Research.
73(5): 526-533.
Lin Y.M., Liu Y. and Tay J.H. (2003). Development and characteristics of phosphorusaccumulating microbial granules in sequencing batch reactors. Applied Microbiology
and Biotechnology. 62(4): 430-435.
Liu Y. and Tay J.-H. (2004). State of the art of biogranulation technology for wastewater
treatment. Biotechnology Advances. 22(7): 533-563.
Liu Y., Wang Z.W., Qin L., Liu Y.Q. and Tay J.H. (2005a). Selection pressure-driven aerobic
granulation in a sequencing batch reactor. Applied Microbiology and Biotechnology.
67(1): 26-32.
Liu Y., Wang Z.W. and Tay J.H. (2005b). A unified theory for upscaling aerobic granular
sludge sequencing batch reactors. Biotechnology Advances. 23(5): 335-344.
Liu Y., Yang S.F. and Tay J.H. (2003). Elemental compositions and characteristics of aerobic
granules cultivated at different substrate N/C ratios. Applied Microbiology and
Biotechnology. 61(5-6): 556-561.
Londong J. and Wachtl P. (1996). Six years of practical experience with the operation of online analysers. Water Science and Technology. 33(1): 159-164.
LynggaardJensen A., Eisum N.H., Rasmussen I., Jacobsen H.S. and Stenstrom T. (1996).
Description and test of a new generation of nutrient sensors. Water Science and
Technology. 33(1): 25-35.
Mainstone C.P. and Parr W. (2002). Phosphorus in rivers - ecology and management. Science
of the Total Environment. 282: 25-47.
Manjunath N.T., Mehrotra I. and Mathur R.P. (2000). Treatment of wastewater from
slaughterhouse by DAF-UASB system. Water Research. 34(6): 1930-1936.
Martinez J., Borzacconi L., Mallo M., Galisteo M. and Vinas M. (1995). Treatment of
slaughterhouse wastewater. Water Science and Technology. 32(12): 99-104.
McSwain B.S., Irvine R.L. and Wilderer P.A. (2004). The influence of-settling time on the
formation of aerobic granules. Water Science and Technology. 50(10): 195-202.
Meinhold J., Filipe C.D.M., Daigger G.T. and Isaacs S. (1999). Characterization of the
denitrifying fraction of phosphate accumulating organisms in biological phosphate
removal. Water Science and Technology. 39(1): 31-42.
Metcalf & Eddy (1991). Wastewater engineering: treatment, disposal and reuse. Singapore,
McGraw-Hill, Inc.
Metzner G. and Temper U. (1990). Operation and optimization of a full-scale fixed-bed
reactor for anaerobic-digestion of animal rendering waste-water. Water Science and
Technology. 22(1-2): 373-384.
Meyer R.L., Saunders A.M. and Blackall L.L. (2006). Putative glycogen-accumulating
organisms belonging to the Alphaproteobacteria identified through rRNA-based stable
isotope probing. Microbiology-Sgm. 152: 419-429.
Meyer R.L., Saunders A.M., Zeng R.J.X., Keller J. and Blackall L.L. (2003). Microscale
structure and function of anaerobic-aerobic granules containing glycogen
accumulating organisms. FEMS Microbiology Ecology. 45(3): 253-261.
34
Meyer R.L., Zeng R.J.X., Giugliano V. and Blackall L.L. (2005). Challenges for simultaneous
nitrification, denitrification, and phosphorus removal in microbial aggregates: mass
transfer limitation and nitrous oxide production. FEMS Microbiology Ecology. 52(3):
329-338.
Mino T., Liu W.T., Kurisu F. and Matsuo T. (1995). Modeling glycogen-storage and
denitrification capability of microorganisms in enhanced biological phosphate removal
processes. Water Science and Technology. 31(2): 25-34.
Mittal G.S. (2004). Characterization of the effluent wastewater from abattoirs for land
application. Food Reviews International. 20(3): 229-256.
Mittal G.S. (2006). Treatment of wastewater from abattoirs before land application - a review.
Bioresource Technology. 97(9): 1119-1135.
Mobarry B.K., Wagner M., Urbain V., Rittmann B.E. and Stahl D.A. (1996). Phylogenetic
probes for analyzing abundance and spatial organization of nitrifying bacteria. Applied
and Environmental Microbiology. 62(6): 2156-2162.
Morgenroth E., Sherden T., van Loosdrecht M.C.M., Heijnen J.J. and Wilderer P.A. (1997).
Aerobic granular sludge in a sequencing batch reactor. Water Research. 31(12): 31913194.
Munch E.V., Lant P. and Keller J. (1996). Simultaneous nitrification and denitrification in
bench-scale sequencing batch reactors. Water Research. 30(2): 277-284.
Nunez L.A. and Martinez B. (1999). Anaerobic treatment of slaughterhouse wastewater in an
Expanded Granular Sludge Bed (EGSB) reactor. Water Science and Technology.
40(8): 99-106.
Obaja D., Mace S., Costa J., Sans C. and Mata-Alvarez J. (2003). Nitrification, denitrification
and biological phosphorus removal in piggery wastewater using a sequencing batch
reactor. Bioresource Technology. 87(1): 103-111.
Obaja D., Mace S. and Mata-Alvarez J. (2005). Biological nutrient removal by a sequencing
batch reactor (SBR) using an internal organic carbon source in digested piggery
wastewater. Bioresource Technology. 96(1): 7-14.
Oehmen A., Keller-Lehmann B., Zeng R.J., Yuan Z.G. and Keller E. (2005). Optimisation of
poly-beta-hydroxyalkanoate analysis using gas chromatography for enhanced
biological phosphorus removal systems. Journal of Chromatography A. 1070(1-2):
131-136.
Oehmen A., Saunders A.M., Vives M.T., Yuan Z.G. and Keller H. (2006). Competition
between polyphosphate and glycogen accumulating organisms in enhanced biological
phosphorus removal systems with acetate and propionate as carbon sources. Journal of
Biotechnology. 123(1): 22-32.
Ostle A.G. and Holt J.G. (1982). Nile blue-A as a fluorescent stain for poly-betahydroxybutyrate. Applied and Environmental Microbiology. 44: 238-241.
Peng D.C., Bernet N., Delgenes J.P. and Moletta R. (1999). Aerobic granular sludge - A case
report. Water Research. 33(3): 890-893.
Peng Y.Z., Chen Y., Peng C.Y., Liu M., Wang S.Y., Song X.Q. and Cui Y.W. (2004). Nitrite
accumulation by aeration controlled in sequencing batch reactors treating domestic
wastewater. Water Science and Technology. 50(10): 35-43.
Pijuan M., Saunders A.M., Guisasola A., Baeza J.A., Casas C. and Blackall L.L. (2004).
Enhanced biological phosphorus removal in a sequencing batch reactor using
propionate as the sole carbon source. Biotechnology and Bioengineering. 85(1): 56-67.
Pitman A.R., Venter S.L.V. and Nicholls H.A. (1983). Practical experience with biological
phosphorus removal plants in Johannesburg. Water Science and Technology. 15(3-4):
233-259.
35
Pochana K. and Keller J. (1999). Study of factors affecting simultaneous nitrification and
denitrification (SND). Water Science and Technology. 39(6): 61-68.
Puig S., Vives M.T., Corominas L., Balaguer M.D. and Colprim J. (2004). Wastewater
nitrogen removal in SBRs, applying a step-feed strategy: from lab-scale to pilot-plant
operation. Water Science and Technology. 50(10): 89-96.
Qin L., Liu Y. and Tay J.-H. (2004). Effect of settling time on aerobic granulation in
sequencing batch reactor. Biochemical Engineering Journal. 21(1): 47-52.
Reddy K., Drysdale G.D. and Bux F. (2003). Evaluation of activated sludge treatment and
settleability in remediation of edible oil effluent. Water Sa. 29(3): 245-250.
Revsbech N.P. (1989). An oxygen microsensor with a guard cathode. Limnology and
Oceanography. 34(2): 474-478.
Russell J.M., Cooper R.N. and Lindsey S.B. (1993). Soil denitrification rates at waste-water
irrigation sites receiving primary-treated and anaerobically treated meat-processing
effluent. Bioresource Technology. 43(1): 41-46.
Sachon G. (1984). Traitment des eaux residuaires des slaughterhouses de betail. In: Food
Industries and the Environment. Proceedings of the International Symposium,
Budapest, Hungary, 9-11 Sep 1982. Hollo J. (eds). Amsterdam, Holland, Elsevier:
293-305. In French.
Saunders A.M., Oehmen A., Blackall L.L., Yuan Z. and Keller J. (2003). The effect of GAOs
(glycogen accumulating organisms) on anaerobic carbon requirements in full-scale
Australian EBPR (enhanced biological phosphorus removal) plants. Water Science
and Technology. 47(11): 37-43.
Sayed S. and Dezeeuw W. (1988). The performance of a continuously operated flocculant
sludge UASB reactor with slaughterhouse waste-water. Biological Wastes. 24(3): 199212.
Sayed S., Vancampen L. and Lettinga G. (1987). Anaerobic treatment of slaughterhouse waste
using a antigranulocytes sludge UASB reactor. Biological Wastes. 21(1): 11-28.
Stebor T.W., Berndt C.L., Marman S. and Gabriel R. (1990). Operating experience: anaerobic
treatment at packerland packing. Proceedings 44th Purdue Industrial Waste
Conference, Chelsea, MI, Lewis.
Subramaniam K., Greenfield P.F., Ho K.M., Johns M.R. and Keller J. (1994). Efficient
biological nutrient removal in high-strength waste-water using combined anaerobicsequencing batch reactor treatment. Water Science and Technology. 30(6): 315-321.
Tay J.H., Liu Q.S. and Liu Y. (2001). Microscopic observation of aerobic granulation in
sequential aerobic sludge blanket reactor. Journal of Applied Microbiology. 91(1):
168-175.
Tay J.H., Liu Q.S. and Liu Y. (2002). Characteristics of aerobic granules grown on glucose
and acetate in sequential aerobic sludge blanket reactors. Environmental Technology.
23(8): 931-936.
Thayalakumaran N., Bhamidimarri R. and Bickers P.O. (2003a). Biological nutrient removal
from meat processing wastewater using a sequencing batch reactor. Water Science and
Technology. 47(10): 101-108.
Thayalakumaran N., Bhamidimarri R. and Bickers P.O. (2003b). Characterisation of aerobic
bio treatment of meat plant effluent. Water Science and Technology. 48(8): 53-60.
Tilche A., Bacilieri E., Bortone G., Malaspina F., Piccinini S. and Stante L. (1999). Biological
phosphorus and nitrogen removal in a full scale sequencing batch reactor treating
piggery wastewater. Water Science and Technology. 40(1): 199-206.
Travers S.M. and Lovett D.A. (1984). Activated-sludge treatment of abattoir wastewater .2.
Influence of dissolved-oxygen concentration. Water Research. 18(4): 435-439.
36
Turk O. and Mavinic D. (1986). Preliminary assessment of a shortcut in nitrogen removal
from wastewater. Can Journal of Civil Engineering. 13: 600-605.
vanLoosdrecht M.C.M., Hooijmans C.M., Brdjanovic D. and Heijnen J.J. (1997). Biological
phosphate removal processes. Applied Microbiology and Biotechnology. 48(3): 289296.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Wareham D.G., Mavinic D.S. and Hall K.J. (1994). Sludge-digestion using ORP-regulated
aerobic-nnoxic cycles. Water Research. 28(2): 373-384.
Wentzel M.C., Loewenthal R.E., Ekama G.A. and Marais G.V. (1988). Enhanced
polyphosphate organism cultures in activated-sludge systems .1. Enhanced culture
development. Water Sa. 14(2): 81-92.
Wilderer P., Irvine R.L. and Goronszy M.C. (2001). Sequencing batch reactor technology.
London, IWA.
Yang S.F., Tay J.H. and Liu Y. (2003). A novel granular sludge sequencing batch reactor for
removal of organic and nitrogen from wastewater. Journal of Biotechnology. 106(1):
77-86.
Yuan Z., Keller J. and Lant P. (2003). Optimization and control of biological nitrogen
removal in activated sludge processes: a review of recent developments. In:
Biotechnology for the Environment. Agathos S.N. and Reineke W. (eds), Kluwer
Academic Publishers. 3C.
Zeng R.J., Lemaire R., Yuan Z. and Keller J. (2003). Simultaneous nitrification,
denitrification, and phosphorus removal in a lab-scale sequencing batch reactor.
Biotechnology and Bioengineering. 84(2): 170-178.
37
Appendix A
A Sequencing Batch Reactor System for High-Level Biological Nitrogen and
Phosphorus Removal from Abattoir Wastewater
Romain Lemaire1, Zhiguo Yuan1, Nicolas Bernet2, Gulsum Yilmaz1 and Jürg Keller1
1
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane QLD 4072, Australia
2
INRA, UR50, Laboratoire de Biotechnologie de l'Environnement, Avenue des Etangs,
Narbonne, F-11100, France
ABSTRACT
A sequencing batch reactor (SBR) system is demonstrated to biologically remove nitrogen,
phosphorus and COD to very low levels from abattoir wastewater. Each 6h cycle contained
three anoxic/anaerobic and aerobic sub-cycles with wastewater fed at the beginning of each
anoxic/anaerobic period. The step-feed strategy was applied to avoid high-level build-up of
nitrate or nitrite during nitrification, and therefore to facilitate the creation of anaerobic
conditions required for biological phosphorus removal. A high degree removal of total
phosphorus (>98%), total nitrogen (>97%) and total COD (>95%) was consistently and
reliably achieved after a three-month start-up period. The concentrations of total phosphate
and inorganic nitrogen in the effluent were consistently lower than 0.2 mgP.l-1 and 8 mgN.l-1
respectively. Fluorescence in-situ hybridization (FISH) revealed that the sludge was enriched
in Accumulibacter spp. (20-40%), a known polyphosphate accumulating organism (PAO),
whereas the known glycogen accumulating organisms (GAOs) were almost absent. The SBR
received two streams of abattoir wastewater, namely the effluent from a full-scale anaerobic
pond (75%) and the effluent from a lab-scale high-rate pre-fermenter (25%), both receiving
raw abattoir wastewater as feed. The pond effluent contained approximately 250 mgN.l-1 total
nitrogen and 40 mgP.l-1 of total phosphorus, but relatively low levels of soluble COD (around
500 mg.l-1). The high-rate lab-scale pre-fermentor, operated at a temperature of 37˚C and a
sludge retention time of 1 day, proved to be a cheap and effective method for providing
supplementary VFAs allowing for high-degree of biological nutrient removal from abattoir
wastewater.
Keywords: Abattoir wastewater, biological nutrient removal, pre-fermentation, SBR, stepfeed, PAO.
38
INTRODUCTION
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen (N) and
phosphorus (P). Over the past two decades, biological COD and N removal from abattoir
wastewater has received much greater attention than has the biological P removal. Reliable
biological COD and nitrogen removal systems have been successfully developed and applied
for abattoir wastewater treatment using continuous activated sludge systems (Beccari et al.,
1984; Frose and Kayser, 1985; Willers et al., 1993). However, P removal continues to be
achieved primarily through chemical precipitation, despite biological P removal being a much
cheaper and more environmentally sustainable option.
The main challenges for biological phosphorus removal from abattoir wastewater are:
•
•
The wastewater contains a high level of ammonia and organic nitrogen. The complete
nitrification of these nitrogen components produces a high level of nitrate, which has
proved to be an obstacle to the development of a stable and reliable Bio-P removal process
(Pitman et al., 1983; Comeau et al., 1986; Furumai et al., 1999). Phosphorus removal
requires alternating anaerobic and aerobic/anoxic conditions. The high level of nitrate (due
to the high influent nitrogen concentrations) makes the creation of true anaerobic
conditions in the system difficult;
Abattoir wastewater contains substantial amounts of fat, oil and grease (FOG), which
would deteriorate the sludge settleability when directly fed to activate sludge systems.
Primary treatment is typically required before abattoir wastewater is treated in biological
nutrient removal systems. In Australia, the raw abattoir wastewater is typically pre-treated
in anaerobic ponds with a hydraulic retention time ranging between 7 – 14 days. While
reducing the FOG content, this anaerobic treatment process also removes a large fraction
of the COD from the wastewater, resulting in COD limitations (particularly Volatile Fatty
Acids – VFAs) for N and P removal (Keller et al., 1997).
In this paper, we demonstrate the use of a sequencing batch reactor (SBR) system for
biological nitrogen and phosphorus removal from abattoir wastewater. In recent years, the use
of SBRs for the biological treatment of wastewater has been widely extended from lab-scale
studies to real WWTPs (Tilche et al., 1999; Artan et al., 2001; Keller et al., 2001; Puig et al.,
2004). SBRs offer a great deal of operational flexibility as it allows for easy adjustment of
aerobic, anoxic and anaerobic periods through temporal control of aeration and filling
(Wilderer et al., 2001). To address the first challenge described above, a step-feed scheme,
characterised by several aerobic and anoxic phases in a SBR cycle with the wastewater fed to
the reactor during the anoxic phases is employed (Anderottola et al., 2000; Lin and Jing,
2001; Puig et al., 2004). This operational strategy allows timely removal of nitrate so that,
when an adequate amount of COD is available, nitrate build-up is avoided (Puig et al., 2004).
To address the second challenge, the proposed system is equipped with a high-rate prefermentor, which is used to provide additional VFAs when the anaerobic pond effluent does
not contain a sufficient amount for the biological phosphorus and nitrogen removal.
Details of the design, operation and performance of the proposed system are presented.
Further potential optimisations are also highlighted.
39
MATERIAL AND METHODS
Reactor set-up and operation
A lab-scale SBR with a working volume of 7 l was used in this study. The SBR was seeded
with non-EBPR (enhanced biological phosphorus removal) sludge from a full-scale SBR
treating abattoir wastewater in Queensland, Australia. 1 l of EBPR sludge (MLSS around
4 g.l-1) enriched in a lab reactor (Lemaire et al., 2006) was added on Day 60 to initiate the
EBPR process in the reactor as there seemed to be no EBPR organisms present in the initial
seed sludge used. The SBR was operated with a cycle time of 6 h in a temperature-controlled
room (18-22˚C). In each cycle, 1 l of abattoir wastewater (more details given below) was
pumped into the reactor over the three filling periods with a volume distribution of 0.5 l, 0.3 l
and 0.2 l respectively. Each filling period was followed by non-aerated (either anoxic or
anaerobic depending on when the oxidised nitrogen was completely consumed) and aerated
periods (Table 1). During aerated periods, air was provided intermittently using an on/off
control system to keep the dissolved oxygen (DO) level between 1.5 and 2 mg.l-1. After the
settling period, 1 l supernatant was removed from the reactor resulting in a HRT of 42 h. 115
ml of mixed liquor was wasted every cycle resulting in a SRT of 15 days. The pH in the
system was recorded, ranging between 7.1-7.9, but not controlled. The ORP signal was also
recorded to give indications of the nitrate levels in the reactor during the anoxic periods. The
reactor was mixed with an overhead mixer except during the settling, decanting and first
filling periods.
Table 1. Operating conditions of lab-scale SBR (6h cycle)
HRT = 42 h
duration (min) DO (mgO2.l-1)
Fill no-mix 1
5
~0
No-aerated mix 1 (anoxic or anaerobic*)
30
~0
Aerated mix 1 (no aeration in the last 5 min)
55
1.5-2
Fill mix 2
3
~0
No-aerated mix 2 (anoxic or anaerobic*)
70
~0
Aerated mix 2 (no aeration in the last 5 min)
35
1.5-2
Fill mix 3
2
~0
No-aerated mix 3 (anoxic or anaerobic*)
60
~0
Aerated mix 3 (sludge wasted at the end)
20
1.5-2
Settle
70
~0
Decant
10
~0
*
when nitrate and nitrite depleted
SBR sequences
Wastewater
The wastewater used in this study was from a local abattoir in Queensland, Australia. At this
site, the raw effluent passes through four parallel anaerobic ponds before being treated in a
SBR for biological nitrogen and COD removal. The anaerobic ponds serve to reduce FOG and
COD, and also to produce easily biodegradable COD, particularly VFAs, to facilitate the
down-stream biological nitrogen removal. The four anaerobic ponds produced VFAs at
different concentrations due to different organic loading rates. Pond A, which was the only
easily accessible pond for wastewater collection and hence had to be used in the study, was
under-loaded leading to much lower COD and VFAs concentrations in comparison to other
40
ponds (see Table 2). Therefore, extra VFAs were added to pond A effluent to simulate the
higher VFA levels present in other ponds, as will be detailed in Table 3.
Raw wastewater and anaerobic pond effluent from the abattoir were collected on a weekly
basis and stored at 4˚C. The raw wastewater was subjected to one-day pre-fermentation before
being pumped into the SBR. The pre-fermentation was performed in a 50 l tank continuously
mixed with a submersible pump. No inoculum was introduced in the pre-fermentor, and hence
the microbial population present in the raw abattoir wastewater was used to carry out the
fermentation. The temperature inside the tank was kept at 37˚C by a heating probe and would
not require any special heating system in a full-scale installation due to the temperature of the
abattoir raw wastewater (typically around 40˚C). The aim of this pre-fermentation step was to
increase the level of easy biodegradable COD, in particular VFAs, which is critical for bio-P
removal. The characteristics of the pre-fermented raw wastewater and the anaerobic pond
effluent are compared in Table 2.
Table 2. Characteristics of the different types of wastewater used in this study. The intervals
represent the mid-95% range.
Parameter
Pre-fermented
raw wastewater
7460-9300
2360-2840
703-869
271-317
139-160
44-53
38-43
Anaerobic pond
A effluent b
430-720
205-245
24-32
218-262
207-224
33-37
32-34
Anaerobic pond
B effluent
740-950
440-531
272-358
240-262
220-226
37-40
33-36
TCOD (mg.l-1)
SCOD (mg.l-1)
VFAa (mgCOD.l-1)
TN (mg.l-1)
NH4-N (mg.l-1)
TP (mg.l-1)
PO4-P (mg.l-1)
a
Acetate and propionate only
b
Pond effluent used in this study; additional acetate and propionate was added (see Table 3) to
simulate Pond B effluent, which was non-accessible for wastewater collection on site.
Table 3. Characteristics of the SBR influent during its nine-month operation.
Influent Parameters
Day 0-30
Ratio VFA:TP
3.7
Ratio TCOD:TN
5.5
% pre-fermented raw
15%
wastewater in influent
VFAs added to Pond A to
No
simulate other ponds
a
-1
-1
250 mgCOD.l acetate, and 100 mgCOD.l propionate
Day 30-80 After day 80
12.2
15.1
8.7
12
15%
25%
Yesa
Yesa
The abattoir closed down for a month over Christmas due to annual maintenance. During this
period, the SBR cycle operation was modified in order to preserve the reactor biomass as no
wastewater was available. The sludge was aerated and mixed for 15 min in each 6h cycle and
was allowed to settle for the rest of the cycle. The performance during this starvation and
recovery period is the focus of another paper (Yilmaz et al., 2007)
41
Analyses
The ammonia (NH3 + NH4+), nitrate (NO3-), nitrite (NO2-) and total phosphate (PO43--P) were
analysed using a Lachat QuikChem8000 Flow Injection Analyser (Lachat Instrument,
Milwaukee). Total and soluble chemical oxygen demand (TCOD and SCOD, respectively),
total Kjeldahl nitrogen (TKN), total phosphorus (TP), mixed liquor suspended solid (MLSS)
and volatile MLSS (MLVSS) were analysed according to the standard methods (APHA,
1995). VFAs were measured by Perkin-Elmer gas chromatography with column DB-FFAP
15m x 0.53mm x 1.0µm (length x ID x film) at 140˚C, while the injector and FID detector
were operated at 220˚C and 250˚C, respectively. High purity helium was used as carrier gas at
a flow rate of 17 ml.min-1. 0.9 ml of the filtered sample was transferred into a GC vial to
which 0.1 ml of formic acid was added.
Fluorescence in situ hybridisation (FISH) was performed as specified in Amann (1995).
Oligonucleotide probes used in this study were the combination of EUB338 i-iii (EUBmix)
for the detection of all bacteria (Daims et al., 1999), the combination of PAO462, PAO651
and PAO846 (PAOmix) for Accumulibacter spp. (Crocetti et al., 2000), the probe
combination (GAOmix) of GAOQ989 (Crocetti et al., 2002) and GB_G2 (Kong et al., 2002)
for Competibacter spp., the probe combination (DF1mix) of TFO_DF218 and TFO_DF618
for Cluster 1 Defluviicoccus vanus-related spp. (Wong et al., 2004) and the probe combination
(DF2mix) of DF988, DF1020 and helper probes H966 and H1038 for Cluster 2
Defluviicoccus vanus-related spp. (Meyer et al., 2006). FISH quantification was performed as
described in Crocetti et al. (Crocetti et al., 2002).
RESULTS
Figure 1 presents the influent and effluent COD, N and P concentrations, along with the
MLSS concentration in the reactor and its volatile fraction, during the nine months operation
of the SBR. Also presented in Figure 1 is the fraction of Accumulibacter spp. (i.e. PAOs) in
the system. GAOs, namely Competibacter spp. and the putative Defluviicoccus vanus-related
spp. (Cluster 1 and 2), were negligible in this reactor (<1% of the total microbial population at
all time). According to the effluent and MLSS data (Figure 1c and 1d), the SBR reached a
steady state around day 170. The study can be divided into two periods: the start-up period
from day 0 to 170 and the steady state period from day 170 to 275.
42
starvation period
PO4-P
350
100
NH4-N
TN
TP
300
80
60
150
40
-1
200
P (mg L )
-1
N (mg L )
250
100
(a)
50
CODt
CODs
VFA
0
3.0
20
0
-1
COD (g L )
2.5
2.0
(b)
1.5
1.0
0.0
PO4-P
60
40
20
(c)
20
0
0
8
0.8
6
0.6
4
0.4
2
0.2
MLSS
VSS:MLSS
(d)
0
0
50
100
VSS:MLSS ratio
Accumulibacter
40
-1
NOx-N
P (mg L )
80
NH4-N
60
-1
MLSS (g L )
-1
N (mg L ) and % Accumulibacter
0.5
150
200
250
0.0
300
Days
Figure 1. Characteristics of the influent (a) and (b), effluent nutrient levels and the
Accumulibacter spp. population (c), and MLSS and the VSS:MLSS ratio in the reactor (d). No
wastewater was fed during the 33 days of starvation period.
43
Start up period (day 0 to 170)
Complete nitrification was achieved in the SBR after less than one week of operation as
shown by the absence of NH4+ in the effluent (Figure 1c). However, denitrification was
incomplete and NOx- (nitrate + nitrite) accumulated in the reactor reaching 60 mgN.l-1 in the
effluent during the first 30 days of operation (Figure 1c). Clearly, In order to improve the
denitrification, more COD was needed during anoxic periods. Therefore, extra VFAs (i.e.
acetate and propionate) were added to pond A effluent on day 30 in order to simulate the
concentration in the other ponds (typically 250 mgCOD.l-1 acetate and 100 mgCOD.l-1
propionate). These additional VFAs improved the denitrification and the level of NOx- in the
effluent dropped to 15 mgN.l-1 (Figure 1c). The similar levels of PO43- measured in the
influent (Figure 1a) and effluent (Figure 1c) clearly indicate that phosphorus removal was
negligible during the first 60 days. P removal was likely limited by the slow development of
PAOs, which were possibly inhibited by the level of nitrate present during most of the time
over a cycle as speculated by Keller et al. (1997). The fact that non-EBPR sludge was used to
seed the reactor could have also contributed to the slow development of PAOs.
After the introduction of 1 l lab-scale EBPR sludge enriched in Accumulibacter spp. (details
of the culture can be found in Lemaire et al., (2006)) on day 60, P removal improved
dramatically, and consistent high-level of P removal was achieved and maintained thereafter.
The process data clearly suggests that the enriched Accumulibacter spp. culture managed to
survive and develop in a very different environmental setting. This is confirmed by the FISH
quantitation results (Figure 1c) and the decrease of the organic fraction in the biomass due to
intracellular poly-P storage by Accumulibacter spp. (Figure 2d).
However, Figure 1c also shows that while P removal was improving, NOx- started to
accumulate again in the system. It was believed that a shortage of easily biodegradable COD
in the reactor trigger this NOx- accumulation as PAOs and denitrifiers were competing for the
same carbon sources. In order to further increase the amount of VFA available for P and N
removal, the amount of pre-fermented raw wastewater in the influent was increased from 15%
to 25% on day 80 resulting in a higher VFA:TP ratio and TCOD:TN ratio in the influent
(Table 3). Denitrification improved immediately and from day 100 onwards, less than
10 mgN.l-1 remained in the effluent. Good COD, nitrogen and phosphorus removal was
achieved after day 100. There was one interruption to the reactor operation between day 125
and 160, when the abattoir closed down and no wastewater could be supplied to the SBR
(Figure 1). A detailed report of the reactor performance in this period can be found in Yilmaz
et al. (2007). The reactor biomass concentration decreased by 30% during this long starvation
period.
Steady state period (from day 170 to 275)
Following the long starvation period, the reactor performance quickly recovered (within four
days) as clearly indicated by the low nutrient level in the effluent shortly after the normal SBR
operation was resumed (Figure 1c). The biomass concentration returned to its previous level
after 2 weeks and remained relatively constant around 5 g.l-1 with an organic fraction
fluctuating between 0.7 and 0.75 (Figure 1d). Table 4 details the SBR effluent quality after the
starvation period, between day 170 and 275. For comparison, the COD and nutrient levels in
the influent are also presented. The SBR process consistently achieved 95, 97 and 98% of
TCOD, TN and TP removal, respectively. The remaining COD in the effluent could be
regarded as non-biodegradable and represented about 5% of the total COD initially present in
44
the influent. It was observed that the sludge volume index (SVI) was relatively high
throughout the study period, between 180 and 250 ml.gMLSS-1. This could have partially
been caused by the remaining high fat/oil/grease content of the pre-fermented raw wastewater
as suggested by Johns (1995). However, the suspended solids concentration in the effluent
was lower than 25 mg.l-1 at all times (data not shown).
Table 4. Influent and effluent characteristics between day 170 and 275. (N represents the
number of samples analysed between day 170 and 275)
Parameter
(mg.l-1)
TCOD
SCOD
TKN
N-NH4
N-NOx
TP
P-PO4
Influent (N=13)
mid-95% range
2600-3120
1150-1320
236-277
196-215
not detected
38-41
35-38
Effluent (N=32)
mean
mid-95% range
mean
2870
1240
256
206
129-151
114-127
5.3-7.7
0.2-0.8
1.9-2.8
0.7-1.4
0.04-0.09
140
121
6.5
0.5
2.3
1.0
0.06
40
37
Removal of
TCOD, TN
and TP
95 %
97 %
98 %
Figure 2 shows the nitrogen and phosphorus transformations in a typical SBR cycle study
during the steady state period. At the end of each aerobic period, NH4+ was fully oxidised, and
the low level of NOx- accumulated was then removed in the following anoxic period. It can be
seen that a very low level of NOx- was carried over to the next cycle, and was denitrified very
quickly at the beginning of the first anaerobic period. PO43- level increased during each
anaerobic period due to both anaerobic P release by PAOs and wastewater feeding (containing
approximately 35 mgP.l-1), but most P release occurred in the first non-aerated period. PO43was then fully taken up during the subsequent aerobic periods.
anO2
18
O2
anO2
anO2
O2
O2
settle+decant
30
14
PO4-P
12
NH4-N
25
20
NOx-N
10
15
8
6
10
P (mg L-1)
N (mg L-1)
16
4
5
2
0
0
0
50
100
150
200
250
300
350
Time (min)
Figure 2. Nitrogen and phosphorus profiles during a SBR cycle study performed on day 243.
The vertical arrows indicate the 3 feeding periods.
45
Performance of the pre-fermentor
The impact of the one-day pre-fermentation performed on raw wastewater is depicted in
Figure 3. The overall VFA concentration more than doubled as a direct result of this prefermentation. Acetate and propionate were the most abundant VFAs in the raw abattoir
wastewater before and after pre-fermentation with propionate having a slightly higher
production rate than acetate. Also shown in Figure 3 is the impact of pre-fermentation on the
NH4+ and PO43- concentrations. While PO43- concentration stayed constant, NH4+
concentration doubled due to partial mineralisation of the organic nitrogen which represents
around 75% of the raw wastewater total nitrogen. The one week storage of the pre-fermented
raw wastewater in the cold room at a temperature of 4˚C affected VFAs levels more than
nutrient levels with a 20 % reduction of acetate and propionate concentration.
450
400
before pre-fermentation
after pre-fermentation
after 1 week storage at 4OC
350
mg L
-1
300
250
200
150
100
50
0
acetate
propionate
other VFAs
PO4-P
NH4-N
Figure 3. Concentration of the main VFAs, PO43- and NH4+ in the raw wastewater before and
after pre-fermentation, and after one week storage at 4°C. Other VFAs include iso-butyric,
butyric and iso-valeric acids. (Error bars=SD, n=11)
DISCUSSION
Multi-feed strategy to promote biological P removal
Biological phosphorus removal from wastewaters containing a high level of nitrogen, such as
abattoir wastewater, is challenging. Large accumulation of nitrate or nitrite must be avoided in
order to secure anaerobic conditions required by PAOs. Several studies using the SBR
technology to simultaneously remove COD, N and P from piggery wastewater have been
reported (Tilche et al., 1999; Obaja et al., 2003; Obaja et al., 2005). However, the
characteristics of piggery wastewater differ greatly from those of abattoir wastewater. The
large amount of inorganic salts, minerals and metal ions present in the piggery wastewater
promote chemical P removal by precipitation, as evidenced by the scarce P release observed
during the anaerobic stage of the process (Bortone et al., 1994). Subramaniam et al. (1994)
and Keller et al. (1997) attempted to achieve simultaneous COD, N and P removal
biologically from abattoir wastewater using SBR systems. However, P removal was quite
unstable due to intermittent recycling of high levels of NOx- to the anaerobic period.
The use of a multi-feed strategy in this study aimed to limit the level of NOx- recycled to the
anaerobic period. Figure 2 shows that the strategy was very successful. The NOx- level was
less than 8 mgN.l-1 throughout the cycle, despite of the high level of NH4+ and organic
nitrogen in the wastewater (over 250 mgN.l-1, see Table 4). Based on the amount of
46
wastewater fed over the three feeding periods in the SBR cycle (i.e. 0.5 l, 0.3 l and 0.2 l,
respectively) and the PO43- concentration in the influent (i.e. 35 mgP.l-1), the amounts of PO43introduced in the SBR bulk liquid during each feeding steps were estimated to be 2.7, 1.5 and
1.0 mgP.l-1, respectively. The true anaerobic P release by PAOs following those three feeding
periods was thus estimated to be 25.3, 4.5 and 1.0 mgP.l-1, respectively. The very large
discrepancy between the first anaerobic P release and the second and third is mostly due to the
very low level of NOx- in the bulk liquid prior to the first feeding (< 0.5 mgN.l-1) compared to
the second (8 mgN.l-1) and third (6 mgN.l-1) feeding (Figure 2). This indicates that the first
non-aerated period was crucially important for P removal as PAOs could freely utilise all the
VFA available without having to compete with heterotrophic denitrifiers. Without the multifeed strategy employed in this study, high level of NOx- is likely to have accumulated in the
SBR at the end of the cycle and to have been recycled into the first anaerobic period
preventing the stable high level of P removal reported.
Pre-fermentation of raw wastewater
The performance of a biological nutrient removal system depends greatly on the availability
of easily biodegradable carbon sources in the wastewater, particularly VFAs. Considering the
fact that it is difficult to control the VFAs content in large anaerobic ponds, a more
controllable VFA source is necessary for reliable biological nutrient removal from abattoir
wastewater. In this study, a high-rate pre-fermentation step was demonstrated to be a cheap
and effective option. Table 3 shows that the VFA:TP ratio increased from 12.2 to 15.1 when
the pre-fermented wastewater fraction in the SBR influent increased from 15% to 25% on
day 80. This caused an immediate reduction in the nitrate level, with a drastic improvement to
the reliability of P removal (Figure 1b and 1c). The results show that it is both necessary and
practically feasible to include a high-rate pre-fermentor to generate VFAs that may be
supplemented to the nutrient removal SBR when an inadequate amount of VFAs is present in
the pond effluent.
However, it should be highlighted that the use of raw wastewater should be minimised. There
is evidence suggesting that a high fraction of raw feed deteriorates the sludge settleability
(data not shown) likely due to its higher FOG content compared to pond effluent. An over
supply of carbon sources through this stream would also increase aeration costs and sludge
production in the SBR system. Controlled addition of this stream using an on-line control
system would be highly beneficial. However, the control of VFAs supplement to biological
phosphorus removal systems in accordance to the actual demand for VFAs (varying with
time) is still unresolved (Olsson et al., 2005).
An alternative solution that is being investigated is to reduce the demand for carbon sources
by achieving nitrogen removal via nitrite instead of nitrate. This strategy, if successful, would
reduce the carbon demand for denitrification by 40% (Turk and Mavinic, 1986). This would
therefore reduce the amount of additional carbon supply, which in turn will also reduce the
overall oxygen requirement. Such an improvement would have significant benefits for the
operation of large-scale wastewater treatment facilities. Peng et al. (2004) demonstrated that
stable nitrite accumulation during the nitrification process could be obtained through an
aeration control system based on pH and DO signals. On-line control systems based on simple
pH and DO signals are being developed to achieve this nitrite pathway in our lab-scale SBR.
A further opportunity to reduce the demand for carbon sources is to enhance the
denitrification by PAOs. It has been found that Accumulibacter spp. are capable of taking up
phosphorus under anoxic conditions (Kuba et al., 1993; Meinhold et al., 1999; Zeng et al.,
2003a). This is particularly attractive as the same carbon could be used for both denitrification
47
and P removal. However, the exact conditions necessary to promote this type of denitrification
are still unclear and further investigations are needed (Zeng et al., 2003a).
The low abundance of GAOs in the sludge
Competibacter spp. have been widely reported to be abundant in both lab-scale EBPR reactors
(Mino et al., 1995; Crocetti et al., 2002; Kong et al., 2002; Zeng et al., 2003b) and full-scale
EBPR plants (Crocetti et al., 2002; Saunders et al., 2003; Kong et al., 2006). Surprisingly, in
this study, Competibacter spp. were scarcely present in the reactor representing always less
than 1% of the total microbial population. Defluviicoccus vanus-related Alphaproteobacteria
organisms, a new putative GAO recently reported in literature (Wong et al., 2004; Meyer et
al., 2006), was also found to be in very low abundance in the reactor.
Several factors have been suggested in the literature that may influence on the competition
between PAOs and GAOs. Filipe (2001) and Oehmen et al, (2005) found that pH has a
significant impact on the PAO and GAO competition with Accumulibacter spp. possessing
advantages over Competibacter spp. for anaerobic carbon uptake at relatively high pH (>8).
The pH in the study fluctuated between 7.1 and 7.9 during a cycle (uncontrolled), which
should have unlikely provided any selective advantages to PAOs over GAOs. Saito (2004)
reported that the presence of nitrite in the anaerobic or aerobic period inhibited the PAO
activity and could therefore enhance the presence of GAOs in the system. The presence of
nitrite in the reactor during all three aerobic periods and during the second and third anaerobic
periods apparently did not promote the growth of GAOs, as suggested by Saito (2004). Some
studies also showed that better EBPR performance was achieved at relatively low temperature
(5-15˚C) due to a shift in the microbial community from GAOs to PAOs (Whang and Park,
2002; Erdal et al., 2003). The temperature used in this study, controlled between 18-22˚C, is
very similar to many reactor studies where GAOs appeared to be a problem, and is therefore
not believed to be a significant contributor to the low abundance of GAOs. A more likely
reason for the limited growth of GAOs in this reactor could be the large fraction of propionate
present in the influent (propionate to acetate COD ratio was 0.8). (Pijuan et al., 2004; Oehmen
et al., 2006) revealed that propionate as a carbon source may provide selective advantage to
PAOs. The pre-fermentor used in this study largely contributed to the increase of the
propionate fraction. If this hypothesis is true, the operation of the pre-fermentor should be
optimised to not only maximise the total amount of VFAs produced but also to control the
VFAs composition and particularly the acetate to propionate ratio.
CONCLUSION
A sequencing batch reactor system was demonstrated to effectively remove nitrogen,
phosphorus and COD from abattoir wastewater. This provides a more cost-effective and
environmentally friendly alternative to chemical phosphorus removal, which is a common
practice at present. The following conclusions are drawn:
•
•
•
It is possible to achieve a high degree (>98%) of biological phosphorus removal from
abattoir wastewater in the presence of high levels of nitrogen (200 – 300 mgN.l-1).
The multi-step feeding strategy prevents high-level accumulation of nitrate or nitrite, and
hence facilitates the creation of anaerobic conditions. The strategy is strongly
recommended for practical use in the biological treatment of abattoir wastewater.
It is important to incorporate a high-rate pre-fermentor as an integrated component of the
nutrient removal system. This stream, which contains a high-level of VFAs, is effective in
providing supplementary carbon sources for both phosphorus and nitrogen removal.
48
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology CRC, Australia. Dr Yilmaz
Gulsum thanks Istanbul University for fellowship support. FISH quantification was carried
out by Dr Gregory Crocetti from the Advanced Water Management Centre at The University
of Queensland, Australia.
REFERENCES
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
Anderottola G., Canziani R., Foladori P., Ragazzi M. and Tatano F. (2000). Laboratory scale
experimentation for RBCOD production from OFMSW for BNR systems: Results and
kinetics. Environmental Technology. 21(12): 1413-1419.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Artan N., Wilderer P., Orhon D., Morgenroth E. and Ozgur N. (2001). The mechanism and
design of sequencing batch reactor systems for nutrient removal - the state of the art.
Water Science and Technology. 43(3): 53-60.
Beccari M., Carrieri C., Misiti A. and Ramadori R. (1984). Design of single sludge systems
for the treatment of wastewaters with high ammonia content. La Tribune du Cebedeau.
37(491): 387-394.
Bortone G., Malaspina F., Stante L. and Tilche A. (1994). Biological nitrogen and phosphorus
removal in an anaerobic/anoxic sequencing batch reactor with separated biofilm
nitrification. Water Science and Technology. 30(6): 303-313.
Comeau Y., Hall K., Hancock R. and Oldham W. (1986). Biochemical model for enhanced
biological phosphorus removal. Water Research. 20: 1511-1521.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Erdal U.G., Erdal Z.K. and Randall C.W. (2003). The competition between PAOs
(phosphorus accumulating organisms) and GAOs (glycogen accumulating organisms)
in EBPR (enhanced biological phosphorus removal) systems at different temperatures
and the effects on system performance. Water Science and Technology. 47(11): 1-8.
Filipe C.D.M., Daigger G.T. and Grady C.P.L. (2001). pH as a key factor in the competition
between glycogen-accumulating organisms and phosphorus-accumulating organisms.
Water Environment Research. 73(2): 223-232.
Frose G. and Kayser R. (1985). Effective treatment of wastewater from rendering plants.
Proceedings 40th Annual Purdue Industrial Waste Conference, Indiana, Butterworths.
49
Furumai H., Kazmi A.A., Fujita M., Furuya Y. and Sasaki K. (1999). Modeling long term
nutrient removal in a sequencing batch reactor. Water Research. 33(11): 2708-2714.
Johns M.R. (1995). Developments in wastewater treatment in the meat processing industry: A
review. Bioresource Technology. 54(3): 203-216.
Keller J., Subramaniam K., Gosswein J. and Greenfield P.F. (1997). Nutrient removal from
industrial wastewater using single tank sequencing batch reactors. Water Science and
Technology. 35(6): 137-144.
Keller J., Watts S., Battye-Smith W. and Chong R. (2001). Full-scale demonstration of
biological nutrient removal in a single tank SBR process. Water Science and
Technology. 43(3): 355-362.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kong Y.H., Xia Y., Nielsen J.L. and Nielsen P.H. (2006). Ecophysiology of a group of
uncultured Gammaproteobacterial glycogen-accumulating organisms in full-scale
enhanced biological phosphorus removal wastewater treatment plants. Environmental
Microbiology. 8(3): 479-489.
Kuba T., Smolders G.J.F., van Loosdrecht M.C.M. and and Heijnen J.J. (1993). Biological
phosphorus removal from wastewater by anaerobic and anoxic sequencing batch
reactor. Water Science and Technology. 27(4-5): 241-252.
Lemaire R., Meyer R., Taske A., Crocetti G.R., Keller J. and Yuan Z.G. (2006). Identifying
causes for N2O accumulation in a lab-scale sequencing batch reactor performing
simultaneous nitrification, denitrification and phosphorus removal. Journal of
Biotechnology. 122(1): 62-72.
Lin Y.F. and Jing S.R. (2001). Characterization of denitrification and nitrification in a stepfeed alternating anoxic-oxic sequencing batch reactor. Water Environment Research.
73(5): 526-533.
Meinhold J., Filipe C.D.M., Daigger G.T. and Isaacs S. (1999). Characterization of the
denitrifying fraction of phosphate accumulating organisms in biological phosphate
removal. Water Science and Technology. 39(1): 31-42.
Meyer R.L., Saunders A.M. and Blackall L.L. (2006). Putative glycogen-accumulating
organisms belonging to the Alphaproteobacteria identified through rRNA-based stable
isotope probing. Microbiology-Sgm. 152: 419-429.
Mino T., Liu W.T., Kurisu F. and Matsuo T. (1995). Modeling glycogen-storage and
denitrification capability of microorganisms in enhanced biological phosphate removal
processes. Water Science and Technology. 31(2): 25-34.
Obaja D., Mace S., Costa J., Sans C. and Mata-Alvarez J. (2003). Nitrification, denitrification
and biological phosphorus removal in piggery wastewater using a sequencing batch
reactor. Bioresource Technology. 87(1): 103-111.
Obaja D., Mace S. and Mata-Alvarez J. (2005). Biological nutrient removal by a sequencing
batch reactor (SBR) using an internal organic carbon source in digested piggery
wastewater. Bioresource Technology. 96(1): 7-14.
Oehmen A., Saunders A.M., Vives M.T., Yuan Z.G. and Keller H. (2006). Competition
between polyphosphate and glycogen accumulating organisms in enhanced biological
phosphorus removal systems with acetate and propionate as carbon sources. Journal of
Biotechnology. 123(1): 22-32.
Oehmen A., Vives M.T., Lu H.B., Yuan Z.G. and Keller J. (2005). The effect of pH on the
competition between polyphosphate-accumulating organisms and glycogenaccumulating organisms. Water Research. 39(15): 3727-3737.
50
Olsson G., Nielsen M., Yuan Z., Jensen A. and Steyer J.-P. (2005). Instrumentation, control
and automation in wastewater systems, IWA Publishing.
Peng Y.Z., Chen Y., Peng C.Y., Liu M., Wang S.Y., Song X.Q. and Cui Y.W. (2004). Nitrite
accumulation by aeration controlled in sequencing batch reactors treating domestic
wastewater. Water Science and Technology. 50(10): 35-43.
Pijuan M., Saunders A.M., Guisasola A., Baeza J.A., Casas C. and Blackall L.L. (2004).
Enhanced biological phosphorus removal in a sequencing batch reactor using
propionate as the sole carbon source. Biotechnology and Bioengineering. 85(1): 56-67.
Pitman A.R., Venter S.L.V. and Nicholls H.A. (1983). Practical experience with biological
phosphorus removal plants in Johannesburg. Water Science and Technology. 15(3-4):
233-259.
Puig S., Vives M.T., Corominas L., Balaguer M.D. and Colprim J. (2004). Wastewater
nitrogen removal in SBRs, applying a step-feed strategy: from lab-scale to pilot-plant
operation. Water Science and Technology. 50(10): 89-96.
Saito T., Brdjanovic D. and van Loosdrecht M.C.M. (2004). Effect of nitrite on phosphate
uptake by phosphate accumulating organisms. Water Research. 38(17): 3760-3768.
Saunders A.M., Oehmen A., Blackall L.L., Yuan Z. and Keller J. (2003). The effect of GAOs
(glycogen accumulating organisms) on anaerobic carbon requirements in full-scale
Australian EBPR (enhanced biological phosphorus removal) plants. Water Science
and Technology. 47(11): 37-43.
Subramaniam K., Greenfield P.F., Ho K.M., Johns M.R. and Keller J. (1994). Efficient
biological nutrient removal in high-strength waste-water using combined anaerobicsequencing batch reactor treatment. Water Science and Technology. 30(6): 315-321.
Tilche A., Bacilieri E., Bortone G., Malaspina F., Piccinini S. and Stante L. (1999). Biological
phosphorus and nitrogen removal in a full scale sequencing batch reactor treating
piggery wastewater. Water Science and Technology. 40(1): 199-206.
Turk O. and Mavinic D. (1986). Preliminary assessment of a shortcut in nitrogen removal
from wastewater. Can Journal of Civil Engineering. 13: 600-605.
Whang L.M. and Park J.K. (2002). Competition between polyphosphate- and glycogenaccumulating organisms in biological phosphorus removal systems - effect of
temperature. Water Science and Technology. 46(1-2): 191-194.
Wilderer P., Irvine R.L. and Goronszy M.C. (2001). Sequencing batch reactor technology.
London, IWA.
Willers H.C., Tenhave P.J.W., Derikx P.J.L. and Arts M.W. (1993). Temperature-dependency
of nitrification and required anoxic volume for denitrification in the biological
treatment of veal calf manure. Bioresource Technology. 43(1): 47-52.
Wong M.T., Tan F.M., Ng W.J. and Liu W.T. (2004). Identification and occurrence of tetradforming Alphaproteobacteria in anaerobic-aerobic activated sludge processes.
Microbiology-Sgm. 150: 3741-3748.
Yilmaz G., Lemaire R., Keller J. and Yuan Z. (2007). Effectiveness of an alternating aerobic,
anoxic/anaerobic strategy for maintaining biomass activity of BNR sludge during
long-term starvation. Water Research. 41(12): 2590-2598.
Zeng R.J., Saunders A.M., Yuan Z.G., Blackall L.L. and Keller J. (2003a). Identification and
comparison of aerobic and denitrifying polyphosphate-accumulating organisms.
Biotechnology and Bioengineering. 83: 140-148.
Zeng R.J., van Loosdrecht M.C.M., Yuan Z.G. and Keller J. (2003b). Metabolic model for
glycogen-accumulating organisms in anaerobic/aerobic activated sludge systems.
Biotechnology and Bioengineering. 81(1): 92-105.
51
Appendix B
Achieving the Nitrite Pathway Using Aeration Phase Length Control and Stepfeed in a SBR Removing Nutrients from Abattoir Wastewater
Romain Lemaire, Marcos Marcelino and Zhiguo Yuan
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane 4072 QLD, Australia.
ABSTRACT
Aeration phase length control and step-feed of wastewater are used to achieve nitrogen
removal from wastewater via nitrite in sequencing batch reactors (SBR). Aeration is switched
off as soon as ammonia oxidation is completed, which is followed by the addition of a fraction
of the wastewater that the SBR receives over a cycle to facilitate denitrification. The end-point
of ammonia oxidation is detected from the on-line measured pH and oxygen uptake rate
(OUR). The method was implemented in a SBR achieving biological nitrogen and phosphorus
removal from anaerobically pre-treated abattoir wastewater. The degree of nitrite
accumulation during the aeration period was monitored along with the variation in the nitrite
oxidising bacteria (NOB) population using fluorescence in situ hybridisation (FISH)
techniques. It is demonstrated that the nitrite pathway could be repeatedly and reliably
achieved, which significantly reduced the carbon requirement for nutrient removal. Modelbased studies show that the establishment of the nitrite pathway was primarily the result of a
gradual reduction of the amount of nitrite that is available to provide energy for the growth of
NOB, eventually leading to the elimination of NOB from the system.
Keywords: aeration control, denitritation, industrial wastewater, nitritation, NOB, multi-step
feeding.
INTRODUCTION
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen (N) and
phosphorus (P) with relatively high C/N and C/P ratios. Primary treatment is usually required
before abattoir wastewater can be treated in biological nutrient removal (BNR) systems. In
Australia, the raw abattoir wastewater is typically pre-treated in anaerobic ponds with a
hydraulic retention time ranging between 7–14 days. While reducing the amounts of fat, oil
and grease (FOG), this anaerobic treatment process also removes a large fraction of the COD
from the wastewater, resulting in COD limitations (particularly easily biodegradable COD
such as volatile fatty acids – VFAs) for subsequent N and P removal (Keller et al., 1997).
It has been widely reported that N removal through nitrification and denitrification via nitrite
(i.e. nitrite pathway) not only reduces the oxygen consumption in the nitrification stage by
52
25% but also reduces the COD requirement in the denitrification stage by 40% (Turk and
Mavinic, 1986). The use of this nitrite pathway to remove the high level of N present in
abattoir wastewater could therefore save significant amount of COD and improve the overall
performance of the BNR system.
Several factors have been identified to promote the nitrite pathway by selectively inhibiting or
limiting the growth of nitrite oxidising bacteria (NOB) over ammonia oxidising bacteria
(AOB). High pH (Villaverde et al., 1997; Pambrun et al., 2004), high concentrations of free
ammonia (Balmelle et al., 1992; Villaverde et al., 2000) and free nitrous acid (Anthonisen et
al., 1976; Vadivelu et al., 2006), low dissolved oxygen (DO) (Munch et al., 1996), and high
temperature combined with short sludge retention time (SRT) (in the so-called SHARON –
Single reactor High activity Ammonium Removal Over Nitrite – process detailed in Hellinga
(1998)) all likely contribute, to various degrees, to the inhibition or elimination of NOB and
the accumulation of nitrite. These factors have mainly been applied to wastewaters containing
a high concentration of ammonium with low COD:N ratios (e.g. anaerobic digester
supernatant, landfill leachate), whereas denitrification was achieved in a post anoxic period
with the dosage of external carbon sources (Hellinga et al., 1998; Lai et al., 2004; Peng et al.,
2004a; Mace et al., 2006).
These strategies may not be directly applicable to other types of wastewaters requiring
simultaneous biological COD, nitrogen and phosphorus removal, due to the possible
inhibitory effect that these factors could have on other bacterial populations involved in a
more complex BNR system. For example, it has been reported recently that nitrite could
severely inhibit the aerobic/anoxic P-uptake and the anaerobic P-release by the polyphosphate
accumulating organisms (PAOs) in enhanced biological phosphorus removal (EBPR)
processes (Saito et al., 2004; Yoshida et al., 2006; Zhou et al., 2007). The accumulation of
nitrite has also been shown to trigger the emission of nitrous oxide (N2O) during the
denitrification step (Itokawa et al., 2001; Zeng et al., 2003).
Peng et al. (2004) demonstrated that stable N removal via nitrite could be obtained through
the control of the aeration time in a sequencing batch reactor (SBR) treating domestic
wastewater. Aeration was stopped as soon as nitritation finished (i.e. complete oxidation of
NH4+), as indicated by a bending point on the pH profile. An external carbon source (glucose)
was added to enable denitrification in the following anoxic period. Clearly, the use of an
external carbon source rather than the COD contained in wastewater for denitrification
considerably undermines the overall benefits of the nitrite pathway.
This study investigates the possibility of achieving nitrogen removal from abattoir wastewater
via the nitrite pathway by integrating aeration phase length control with step-feed of
wastewater. Instead of using an external carbon source, the organic carbon contained in
wastewater is used for denitrification, thus delivering the true benefit of implementing the
nitrite pathway. A lab-scale SBR, operated with three non-aerated and aerated sub-cycles with
wastewater added to the three non-aerated periods, was employed to remove COD, N and P
from abattoir wastewater. An on-line aeration phase length control system was implemented
based on pH and DO signals to switch between aerobic and anoxic periods. The reactor
performance in terms of the N and P removal efficiency and the degree of nitrite accumulation
was monitored, along with the variation in the NOB population. The mechanisms responsible
for the achievement of the nitrite pathway are explained using model-based analysis.
53
MATERIAL AND METHODS
Reactor set-up and operation
A lab-scale SBR with a working volume of 7 l was used in this study. The SBR was operated
with a cycle time of 6 h in a temperature-controlled room (18-22˚C). In each cycle, 1 l of
abattoir wastewater (more details given below) was pumped into the reactor over the three
filling periods with a volume distribution of 0.5 l, 0.3 l and 0.2 l respectively. Each filling
period was followed by non-aerated (either anoxic or anaerobic depending on when the
oxidised nitrogen was completely consumed) and aerated periods (more details can be found
in Lemaire et al. (submitted)). During aerated periods, air was provided intermittently using
an on/off control system to keep the DO level between 1.5 and 2 mgO2.l-1. After the settling
period, 1 l supernatant was removed from the reactor resulting in a HRT of 42 h. 115 ml of
mixed liquor was wasted every cycle to keep a constant SRT of 15 days. The pH in the system
was recorded, ranging between 7.1 and 7.9, but not controlled. The ORP signal was also
recorded to give indications of the nitrate levels in the reactor during the anoxic periods. The
reactor was mixed with an overhead mixer except during the settling, decanting and first
filling periods. The SBR cycle operation was controlled by a programmable logic controller
(PLC – Opto Control).
Wastewater
The wastewater used in this study was from a local abattoir in Queensland, Australia. At this
site, the raw effluent passes through four parallel anaerobic ponds before being treated in a
SBR for biological N and COD removal. The anaerobic ponds serve to reduce FOG and COD,
and also to produce some easily biodegradable COD, particularly VFAs, to facilitate the
down-stream biological nitrogen removal.
Raw wastewater and anaerobic pond effluent from the abattoir were collected on a weekly
basis and stored at 4˚C. The raw wastewater was subjected to one-day pre-fermentation before
being pumped into the SBR (more details can be found in Lemaire et al. (submitted)). The aim
of this pre-fermentation step was to further increase the level of easily biodegradable COD, in
particular VFAs, which is critical for bio-P removal. The characteristics of the pre-fermented
raw wastewater and the anaerobic pond effluent are compared in Table 1. The wastewater fed
to the lab-scale SBR consisted of a mixture of anaerobic pond effluent and pre-fermented raw
wastewater as later described in Table 2. The modification of the fraction of raw prefermented wastewater used in the influent did not modify the overall N and P content of the
influent due to the identical levels of N and P present in both type of wastewater.
54
Table 1. Characteristics of the different types of wastewater used in this study. The intervals
represent the mid-95% range.
Parameter
Pre-fermented
Anaerobic pond
(mid-95% range)
raw wastewater
effluent b
1,090-1,270
TCOD (mg.l-1)
6,400-8,320
769-909
2,110-2,550
SCOD (mg.l-1)
502-626
699-797
VFAa (mgCOD.l-1)
235-254
260-306
TN (mg.l-1)
223-229
141-157
NH4-N (mg.l-1)
36-39
44-50
TP (mg.l-1)
34-35
37-42
PO4-P (mg.l-1)
a
Acetate and propionate only
b
Additional acetate and propionate were added to the anaerobic pond effluent to mimic the effluent of
better operated, but physically inaccessible pond
Aerobic phase length control to promote nitrite pathway
The SBR was operated for approximately 18 months. Aerobic phase length control for
achieving N removal via nitrite was trialled in the last 13 months. The control strategy
employed was based on the slope of the pH signal and on the oxygen uptake rate (OUR).
Figure 1 shows that the exact time of complete NH4+ oxidation in each aeration periods could
be detected through the pH bending point and the sharp OUR drop. During each aeration
period, the pH slope was calculated once the maximum pH has been reached and started to
decrease. The slope of the pH was determined based on pH values in a 2 minute moving
window. Due to the on-off control system of the DO in the reactor, the OUR was calculated
during the time the oxygen valve was in an “off” state. Aeration is switched off when all the
following three criteria are met:
(i)
The pH slope is lower than a pre-specified value entered by the operator (typically
between 0.005 – 0.01 pH unit.min-1). This was used as the main criterion.
(ii)
The OUR is lower than a pre-specified value entered by the operator (typically 1.2
mgO2.l-1.min-1). This was only a safety condition to reduce the risk of switching
aeration too early should the pH algorithm incorrectly detected a bending point. As
such the set-point was not conservative giving more weight to the pH criterion.
(iii)
The aeration period length is greater than a minimum aeration time entered by the
operator (typically 15 min). The role of this criterion was to ensure a minimum
aeration period in the case that the reactor was seriously under-loaded due to for
example a very low nitrogen concentration in the feed, in which case both the pH and
OUR algorithms could detect a bending point soon after aeration is started.
55
8.0
14
10
8
1.5
6
1.0
4
0.5
7.8
7.6
pH
2.0
12
-1
NO3
DO
pH
OUR
NO2
NH4
2.5
NH4, NO2 and NO3 (Nmg l )
DO (mgO2 l-1) and OUR (mgO2 l-1 min-1)
3.0
7.4
7.2
2
0.0
0
30
50
70
90
110
130
150
170
190
7.0
210
Time (h)
Figure 1. Typical pH, DO, OUR and ammonium profiles during the first two aeration periods
of a SBR cycle, showing the correlation between the complete NH4+ oxidation (vertical dot
line), the pH bending point (horizontal arrows) and the OUR drop. The nitrate and nitrite
profiles are also presented.
The implementation and demonstration of the aeration length control consisted of three stages.
In the first stage, from day 160 to 340, the aeration control was performed manually by the
operator. Based on the pH slope and OUR calculated on-line, the length of each of the three
aeration period was adjusted on a daily basis to ensure that the aeration was stopped
immediately after complete oxidation of NH4+ in the SBR. Then, from day 340 to 410, the
manual aeration control was ceased and fixed aeration lengths of 60 min, 36 min and 24 min
for the three aeration periods, respectively, were applied, which were longer than the time
required for complete NH4+ oxidation. The purpose of Stage II was to demonstrate that the
nitrite pathway previously established under Stage I could be deteriorated by extending
aeration. In Stage III (from day 410 onwards), the automatic aeration length control was
implemented to re-establish the nitrite pathway. It should be mentioned that the abattoir
closed down from day 480 to 525 due to annual maintenance. During this period, the SBR
cycle operation was modified in order to preserve the reactor biomass as no wastewater was
available. The sludge was aerated and mixed for 15 min in each 6h cycle and was allowed to
settle for the rest of the cycle, more details of this operation can be found in Yilmaz et al.
(2007).
Analyses
The ammonia (NH3 + NH4+), nitrate (NO3-), nitrite (NO2-) and total phosphate (PO43--P) were
analysed using a Lachat QuikChem8000 Flow Injection Analyser (Lachat Instrument,
Milwaukee). Total and soluble chemical oxygen demand (TCOD and SCOD, respectively),
total Kjeldahl nitrogen (TKN), total phosphorus, mixed liquor suspended solid (MLSS) and
volatile MLSS (MLVSS) were analysed according to the standard methods (APHA, 1995).
VFAs were measured by Perkin-Elmer gas chromatography with column DB-FFAP 15m x
0.53mm x 1.0µm (length x ID x film) at 140°C, while the injector and FID detector were
56
operated at 220°C and 250°C, respectively. High purity helium was used as carrier gas at a
flow rate of 17 ml.min-1.
Fluorescence in situ hybridisation (FISH) was performed as specified in Amann (1995).
Oligonucleotide probes used in this study were EUBmix (Daims et al., 1999) for the detection
of all Bacteria, NTSPA662 (Daims et al., 2001) for Nitrospira and NIT3 (Wagner et al., 1996)
for Nitrobacter. FISH images were collected using a Zeiss LSM 510 confocal laser scanning
microscope with a 63x Plan-Apochromat oil immersion lens. FISH quantification was
performed according to Crocetti et al. (2002) where the relative abundance of each group was
determined in triplicate as mean percentage of all bacteria.
Model based simulations
To investigate the mechanisms responsible for the decrease of the NOB population in the SBR
with aeration phase length control and step-feed of wastewater, a model was devised based on
the IWA ASM2d (Henze et al., 1999). The main change to ASM2d involved describing
nitrification and denitrification as two-step processes, namely ammonia oxidation followed by
nitrite oxidation, and nitrate reduction followed by nitrite reduction. Consequently, the
nitrifier population in the original ASM2d model was replaced with two populations namely
AOB and NOB. However, the two-step denitrification process was assumed to be catalysed by
the same organisms as described in ASM2d, namely ordinary heterotrophic bacteria and
PAOs. Both groups of bacteria were assumed to be able to reduce both nitrate and nitrite. The
resulting model is presented in Table S1 – S6 in the Supplementary Materials.
In the simulation studies described below, default parameter values recommended in Henze et
al. (1999) were used for the ASM2d model parameters. For the new kinetic parameters related
to the two distinct nitrifier populations, identical values were selected for AOB and NOB.
Rigorous model calibration was not considered necessary given the purpose of the simulation
studies was purely theoretical. The model parameters along with the wastewater composition
used in the simulations are summarised in Table S1 – S6 in the Supplementary Materials.
Two different scenarios were tested, with and without aeration length control. In the case with
aeration length control, aeration was switched off as soon as the ammonium concentration in
the SBR reached 1 mgN.l-1. The lengths of the three aeration periods used in the second case
were 60, 36 and 24 min, respectively, identical to those used in the SBR without aeration
length control. These lengths are in average 10 to 15 min longer than those used in the first
case.
RESULTS AND DISCUSSION
Effect of the aeration control strategy on the nitrite pathway
Figure 2a presents the level of nitrite pathway achieved in the SBR, measured as the amount
of NO2- produced (mgN.l-1) per NOx- produced (mgN.l-1) during the 3 aeration periods, and
the relative abundance of NOBs in the SBR throughout Stage I, Stage II and Stage III. Tests to
identify the main NOB species present in the SBR using common FISH probes (i.e. NIT3 for
Nitrobacter and NTSPA662 for Nitrospira) showed that only Nitrospira was present in the
system. Therefore, only the Nitrospira population was quantified. The reactor used in this
study had been already running for 5 months performing high level of COD, N and P removal
(Lemaire et al., submitted) before the aeration length control was implemented. In that time,
no nitrite accumulation was observed during the aerobic periods as depicted in Figure 2a.
57
% NO2
(a)
NOB population
Starvation
period
80
1.4
70
1.2
60
1
50
0.8
40
0.6
30
0.4
20
0.2
10
Stage I
Stage II
Stage III
45
40
0
P-PO4
N-NH4
N-NOx
VFA influent
(b)
800
700
35
-1
600
30
500
25
400
20
A
15
B
C
D
200
10
100
5
0
150
300
-1
0
N and P (mg l )
1.6
VFA influent (mgCOD l )
N-NO2 : N-NOX ratio (%)
90
1.8
NOB : all bacteria ratio (%)
100
200
250
300
350
400
450
500
550
0
600
Days
Figure 2. (a) Degree of nitrite accumulation in the three stages and the abundance of
Nitrospira as the dominant NOB (FISH probe NTSPA662). The NOB quantification shown is
an average (error bars=SE, n=3). (b) Ammonium, oxidised nitrogen and phosphate in the
effluent and VFAs in the influent.
During Stage I, the manual control of the length of each aeration period resulted in a gradual
accumulation of nitrite reaching 95% of the total amount of NOx- produced on day 280
(Figure 2a). This high level of nitrite pathway was maintained until the start of Stage II.
During that second stage, the implementation of fixed aeration periods rapidly deteriorated the
nitrite pathway previously established, resulting in only 20% of NO2- accumulation 50 days
after the start of Stage II (Figure 2a). The implementation of the automatic aeration length
control strategy during Stage III resulted in the recovery of the nitrite pathway in the SBR. A
nitrite accumulation of 85% was reached 150 days (including 50 days of starvation period)
after the application of the automatic controller. The aeration phase length control and stepfeed strategy were therefore successful in establishing the nitrite pathway.
When comparing the level of nitrite pathway in the SBR and the Nitrospira population
dynamics it clearly appears that the nitrite pathway was achieved through the elimination or
the reduction of the NOB population. However, some delay was observed between the level
of nitrite pathway measured and the abundance of NOB. While the NO2- accumulation
decreased from 98% to 20% during Stage II, the Nitrospira population only increased from
0.3% to 0.5% but later increased to 1.2% of the total bacterial population 40 days into
58
Stage III (Figure 2a). The presence of this lag phase could be due to the complex dynamics
involved in the NOB growth processes when the availability of their main energy source
(i.e. NO2-) is modified. It should be said that in Stage III, no wastewater was fed to the SBR
between day 480 and 530 due to the annual closure of the abattoir and the reactor cycle was
modified in order to preserve the SBR biomass population (more detail given in Yilmaz et al.
(2007)). The slight increase of the Nitrospira population observed during this starvation
period was likely cause by the reduction of the total amount of bacteria present in the reactor
(MLVSS in the SBR decreased by 20% over that period, data not shown) which had a direct
impact on the relative amount of Nitrospira present in the reactor.
Effect of nitrite pathway on the overall SBR performance
In order to demonstrate the benefit of the nitrite pathway in COD savings, the COD
concentration in the SBR influent was adjusted several times during the experimental period
through changing the fraction of fermented raw feed and/or the VFA content in the pond
effluent. The resulting VFA concentration profile in the SBR influent along with nutrient
levels in the effluent are shown in Figure 2b.
From day 160 to 250, the level of nitrite pathway increased from 0 to 95% after the length of
the aeration periods was manually controlled. High levels of COD, N and P removal (95%,
97% and 98%, respectively) were consistently achieved. Figure b shows that as the level of
nitrite pathway increased the amount of NOx- in the effluent decreased, likely due to the
reduced COD requirement for N removal via the nitrite pathway. The implementation of the
aeration length control increased the anoxic time to aerobic time ratio for the SBR cycle which might
have also improved the denitrification performance. The period between day 250 and 280 with
stable high level of nitrite pathway is referred to as “Period A” in Figure 2b and the reactor
performance in this period is summarised in Table 2.
From day 280 to 310, the VFA concentration in the SBR feed was gradually decreased from
600 to 400 mgCOD.l-1 by first reducing the fraction of pre-fermented raw wastewater in the
influent from 25% to 15%, and then gradually reducing the VFA concentration in pond
effluent by 40% (mimicking a poorly performing anaerobic pond). The bio-P removal was
immediately affected due to this sudden VFA shortage but soon recovered (Figure 2b). With
the further reduction of VFA, NOx- started to accumulate in the effluent on day 300 due to
incomplete denitrification apparently caused by the inadequate COD supply. The
accumulation of NOx- was very detrimental to bio-P removal as it deteriorated anaerobic P
release (data not shown), resulting in the loss of P removal (the effluent PO43- concentration
was similar to that in the influent). The amount of VFA had to be slightly increased to
400 mgCOD.l-1, which led to the recovery of both N and P removal (Figure 2b). The stable
period from day 310 and 340 is referred to as “Period B” in Figure 2b and the reactor
performance is also summarised in Table 2.
With the implementation of fixed length aeration on day 340, the effluent NOx- and P levels
deteriorated considerably, likely due to the increased COD requirement for denitrification via
the gradually recovered nitrate pathway. Therefore, more VFAs (710 mgCOD.l-1) had to be
supplied to stabilise the N and P removal performance (Figure 2b). The stable period from day
380 to 420 is referred to as “Period C” in Figure 2b and Table 2.
Following the implementation of the automatic aeration control strategy, the VFA content in
the SBR feed was reduced on day 420 by decreasing both the fraction of pre-fermented raw
wastewater in the influent from 15% to 10% and the VFA concentration in pond effluent by
30%. Once again, NOx- and PO43- immediately accumulated in the effluent due to the sudden
VFA and COD shortage but promptly recovered (Figure 2b). Day 540 to 600, after the normal
59
wastewater load was resumed in the SBR following the long starvation period, is defined as
“Period D” in Figure 2b and Table 2.
Table 2. Summary of the SBR performance and the VFA requirements over four stable
periods, correlated with the degrees of nitrite pathway achieved.
Parameter
(mid-95% range)
NO2- accumulation (%)
NOx- effluent (mgN l-1)
PO43- effluent (mgP l-1)
% pre-fermented raw in
SBR influent
Total VFAs in SBR
influent (mgCOD l-1)
Period A
day 230-280
81-92
0.9-1.6
0.06-0.20
Period B
day 310-340
87-99
2.9-6.5
0.01-0.07
Period C
day 380-420
29-47
5.4-8.3
0.1-3.6
Period D
day 540-600
78-84
1.5-2.6
0.05-0.13
25%
15%
15%
10%
579-632
450-531
710-813
540-571
Table 2 clearly shows that during “Period C”, when the level of nitrite pathway was the
lowest, the worst N and P removal were achieved, despite the amount of VFA in the influent
being considerably higher than in other periods. This clearly demonstrates the benefits of the
nitrite pathway in saving COD and also in enhancing the nutrient removal performance.
Comparing Period B with Periods A and D, during all of which over 80% nitrite pathway was
achieved, the P removal performance was almost identical and the N removal deteriorated
only slightly, despite of a significantly lower VFA content in the feed (10-20%). Noting the
immediate deterioration in the reactor performance each time the VFA content in the feed was
reduced (Figure 2b), it appears that the biomass can slowly adapt to the feed variations,
provided that a critical amount of VFA is supplied. However, abrupt changes in the feed
should be avoided in reactor operation. The process data also showed that it is important to
consider both N and P removal when assessing the possible COD and/or VFA savings via the
nitrite pathway, as P removal depends strongly on the level of N removal.
The aeration control strategy was successful in achieving stable N removal via the nitrite
pathway with reduced requirement for COD and VFAs. As a direct result, the fraction of prefermented raw wastewater in the influent was reduced from an initial 25% to 10% without
affecting the performance of the SBR. The reduction of the raw wastewater fraction lowered
the amount of FOG and colloidal matters, and is expected to improve the sludge settleability
(Hopwood, 1977; Travers and Lovett, 1984) although this aspect was not investigated in this
study.
The steep-feed strategy employed in this SBR ensured that no external carbon addition was
needed to carry out the post denitrification or denitritation making the overall BNR process
more attractive. However, compared to external carbon sources (e.g. methanol, ethanol,
acetate and glucose), which contain no nitrogen, wastewater contains nitrogenous compounds
and therefore the step-feed of wastewater increases the concentration of ammonium and
organic nitrogen while reducing that of oxidised nitrogen compounds. This problem was
resolved in this study by applying three feeding periods, with a gradually reduced amount of
wastewater fed in each period (50%, 30% and 20%, respectively). The overall high-level
removal of N, P and COD (>95%) demonstrates that this multiple step-feed strategy was
successful.
60
Integration of aeration length control with step-feed of wastewater
In recent years, the use of SBR to remove nutrients from domestic or industrial wastewater
has been extended from lab-scale studies to full-scale applications (Tilche et al., 1999; Artan
et al., 2001; Keller et al., 2001; Puig et al., 2004). SBR usually operates with fixed lengths for
the different phases of operation including filling, mixing (anaerobic, aerobic or anoxic),
settling and decanting. Due to influent fluctuations and system state variations, it is beneficial
to operate a SBR process with varying phase lengths. Therefore, higher levels of process
control and automation are necessary to optimise the SBR operation. Many researchers have
suggested that for a nitrogen removal system, on-line measurements of ORP, DO and pH are a
cheap and easy way to detect the end of the nitrification and denitrification processes (AlGhusain et al., 1994; Wareham et al., 1994; Al-Ghusain and Hao, 1995; Hao and Huang,
1996).
More recently, Peng et al. (2004) used a control strategy based on the pH bending point to
initiate the anoxic phase (by stopping aeration) and on the addition of an external carbon
source (glucose) to achieve denitritation. In this study, we applied the control strategy to a
more complex system achieving COD, N and P removal, and achieved denitritation through a
step-feed strategy suppressing the need of external carbon dosage.
7.9
DO
-1
-1
3
OUR
7.8
pH
7.7
2.5
7.5
2
7.4
1.5
-1
7.3
7.2
1
7.1
0.5
7
0
6.9
15
N-NH4
N-NO2
12
40
N-NO3
30
9
20
-1
P-PO4
-1
N (mg l )
pH
7.6
P (mg l )
DO (mg l ) and OUR (mg l min )
3.5
6
10
3
0
0
0
50
100
Time (min)
150
200
Figure 3. Example of pH, DO, OUR, nitrogen and phosphorous profiles during a SBR cycle
performed on day 550 after the automatic aeration control was implemented. Vertical dot lines
indicate when the aeration was automatically stopped and black arrows represent the
additional anoxic time gained by aeration control. White arrows indicate feeding time.
61
Figure 1 already demonstrated the simultaneity between the depletion of NH4+, the bending
points on pH and OUR drop during a SBR cycle where the aeration lengths were not
controlled. The aeration length control strategy previously described could therefore be
implemented in this complex BNR system. Figure 3 presents the pH, DO, OUR, nitrogen and
phosphorus profiles in a SBR cycle after this automatic aeration length control strategy was
implemented. This strategy was reliable in detecting the end of the nitritation process and
stopping the aeration as indicated by the dot lines on Figure 3. The success of this control
strategy was further confirmed by the good long-term performances of the SBR presented
previously.
However, some technical issues and possible improvement of the control algorithm were
identified. The pH and OUR profiles in each of the three aeration periods in a cycle were quite
different (Figure 3) making it difficult to select universally applicable threshold values for the
control algorithms. The pH profile even changed over time as shown by the difference
between the profiles depicted in Figure 1 and in Figure 3. This was mainly due to the large
difference of the initial pH value observed at the start of each aeration period but also to the
different PAO activities in each aeration period, with most of the activity occurring in the first
aeration period as indicated by the highest P uptake in Figure 3. The pH and OUR profiles
evolution over time of in the SBR suggests that it would have been preferable to design an
algorithm where the aeration was controlled based on relative rather than absolute threshold
values. For example, the aeration could be turned off when a certain percentage (e.g. 5-10%)
of the maximum pH slope and OUR value obtained in the on-going aerobic period is reached.
Such an algorithm would have been easy to implement but the PLC software used in our study
did not provide enough flexibility to enable the implementation of this algorithm.
Mechanisms leading to NOB elimination: model-based investigation
Many factors have been identified in literature to selectively inhibit or limit the growth of
NOB, leading to their elimination from BNR systems. These included high pH, high
ammonium/ammonia concentration and high nitrite/free nitrous acid concentration
(Anthonisen et al., 1976; Balmelle et al., 1992; Villaverde et al., 1997; Villaverde et al., 2000;
Pambrun et al., 2004; Vadivelu et al., 2006). However, these factors unlikely played a key role
in the elimination/reduction of the NOB population in our system. Comparing the cycle study
data with and without aeration length control (Figure 3 vs. Figure 1), both pH and ammonium
varied within a similar range. While the nitrite profiles were different in the two cycles, nitrite
accumulation could not have been initiated by itself. The lower affinity of NOB with oxygen
in comparison to AOB has also been found to facilitate the washout of NOB (Blackburne et
al., in press) under low DO conditions. However, the DO concentration applied in our study
was relatively high, and was not lowered during the implementation of the nitrite pathway.
In the model used this study, none of the above factors were considered. As shown in the
Supplementary Materials, the growth parameters including the maximum specific growth rate,
the decay rate and the affinity constants with respect to oxygen and with respect to nitrogen
sources were assumed identical for AOB and NOB. Yet, the simulation results successfully
predicted the onset and full establishment of the nitrite pathway with aeration length control
and step-feed, which was not the case when the aeration lengths were fixed (Figure 4). The
results demonstrate that the nitrite pathway could be established without assuming that NOB
have poorer growth kinetics than AOB.
Figure 4 shows a quick onset of nitrite accumulation after the implementation of the aeration
length control. This is not unexpected. Nitrite, the energy source for NOB, is the product of
ammonia oxidation. This implies that nitrite oxidation should always lag behind ammonia
62
oxidation. In other words, certain levels of nitrite must still be present in the liquid phase at
the time when ammonia oxidation is completed. If the aeration is continued, the remaining
nitrite would be converted to nitrate by NOB (Figure 1). However, if aeration is switched off
at the exact time that ammonia oxidation finishes, as implemented in both the experimental
and simulation studies, the residual nitrite would be carried into the following anoxic phase
and reduced by denitrifiers and is thus no longer available for nitrite oxidisers (Figure 3). This
means that NOB gained less energy for growth in a cycle with aeration length control in
comparison to a cycle without aeration length control, resulting in comparatively less NOB
growth. The reduced NOB growth would lead to a slightly lower nitrite oxidation rate in the
following cycle, which is proportional to the NOB population size. As a result, the nitrite
accumulation when ammonia oxidation finishes would be slightly higher than that in the
previous cycle albeit the difference could be very small, further reducing the growth of NOB.
Over many cycles the NOB population would decrease gradually, which should be
accompanied by increased nitrite accumulation, a situation confirmed by both the
experimental and simulation results.
1.0
120
-1
0.8
100
Without aeration
length control
80
With aeration
length control
60
0.6
0.4
NOB
NO2 / NOX
40
AOB
20
0
0
50
100
150
200
250
NO2 / NOX
AOB and NOB (mgCOD l )
140
0.2
0.0
300
Days
Figure 2. Simulated variation of the NOB and AOB population sizes and NO2:NOx ratio at
the end of the aerobic periods without aeration length control (fixed aeration lengths) and with
aeration length control.
Both the experimental and modelling results demonstrate that the elimination of NOB is a
relatively slow process. This is likely because of the similar growth kinetics possessed by
these groups of nitrifiers, as assumed in the model. Factors such as high free nitrous acid
concentration that lower the NOB growth rate would help to speed up the elimination process.
However, from the discussions above, we hypothesise that the primary reason for NOB
elimination using aeration length control and the step-feed of wastewater, is the gradual
reduction in energy supply to NOB. Indeed, further simulation studies showed that the nitrite
pathway could be established even if NOB possess faster growth kinetics than AOB (data not
shown). However, the time required for achieving the nitrite pathway becomes considerably
longer in the latter case. In practice, it could be very difficult to establish the nitrite pathway in
such conditions as there is typically a delay between the actual depletion of ammonium and
the time of its detection. Such a delay could be sufficient for NOB to convert all the residual
nitrite to nitrate and therefore not losing energy source for growth.
63
CONCLUSION
Nitrogen removal via the nitrite pathway can be achieved by integrating step-feed of
wastewater with an on-line aeration phase length control system that switches off aeration as
soon as ammonium oxidation is completed. The implementation of the nitrite pathway
significantly reduces the carbon demand for biological nutrient removal.
In such systems, nitrite oxidising bacteria are eliminated likely due to the gradual reduction of
their energy source through the use of denitrification, rather than due to the inhibition of their
growth kinetics. However, any reduction of the NOB growth kinetics would help to speed-up
the onset and establishment of the nitrite pathway.
The end-point of ammonium oxidation can be reliably detected from the on-line pH and DO
signals. Therefore, the nitrite pathway can be implemented with relatively simple and cheap
on-line sensors.
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology CRC, a Cooperative Research
Centre established and funded by the Australian Government together with industry and
university partners.
REFERENCES
Al-Ghusain I. and Hao O.J. (1995). Use of pH as control parameter for aerobic/anoxic sludgedigestion. Journal of Environmental Engineering-Asce. 121(3): 225-235.
Al-Ghusain I.A., Huang J., Hao O.J. and Lim B.S. (1994). Using pH as a real-time control
parameter for waste-water treatment and sludge-digestion processes. Water Science
and Technology. 30(4): 159-168.
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
Anthonisen A.C., Loehr R.C., Prakasam T.B.S. and Srinath E.G. (1976). Inhibition of
nitrification by ammonia and nitrous-acid. Journal Water Pollution Control
Federation. 48(5): 835-852.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Artan N., Wilderer P., Orhon D., Morgenroth E. and Ozgur N. (2001). The mechanism and
design of sequencing batch reactor systems for nutrient removal - the state of the art.
Water Science and Technology. 43(3): 53-60.
Balmelle B., Nguyen K.M., Capdeville B., Cornier J.C. and Deguin A. (1992). Study of
factors controlling nitrite buildup in biological processes for water nitrification. Water
Science and Technology. 26(5-6): 1017-1025.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
64
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
Hao O.J. and Huang J. (1996). Alternating aerobic-anoxic process for nitrogen removal:
Process evaluation. Water Environment Research. 68(1): 83-93.
Hellinga C., Schellen A., Mulder J.W., van Loosdrecht M.C.M. and Heijnen J.J. (1998). The
SHARON process: An innovative method for nitrogen removal from ammonium-rich
waste water. Water Science and Technology. 37(9): 135-142.
Henze M., Gujer W., Mino T., Matsuo T., Wentzel M.C., Marais G.V.R. and Van Loosdrecht
M.C.M. (1999). Activated Sludge Model No.2d, ASM2d. Water Science and
Technology. 39(1): 165-182.
Hopwood D. (1977). Effluent treatment in meat and poultry-processing Industries. Process
Biochemistry. 12(8): 5-&.
Itokawa H., Hanaki K. and Matsuo T. (2001). Nitrous oxide production in high-loading
biological nitrogen removal process under low COD/N ratio condition. Water
Research. 35(3): 657-664.
Keller J., Subramaniam K., Gosswein J. and Greenfield P.F. (1997). Nutrient removal from
industrial wastewater using single tank sequencing batch reactors. Water Science and
Technology. 35(6): 137-144.
Keller J., Watts S., Battye-Smith W. and Chong R. (2001). Full-scale demonstration of
biological nutrient removal in a single tank SBR process. Water Science and
Technology. 43(3): 355-362.
Lai E., Senkpiel S., Solley D. and Keller J. (2004). Nitrogen removal of high strength
wastewater via nitritation/denitritation using a sequencing batch reactor. Water
Science and Technology. 50(10): 27-33.
Lemaire R., Yuan Z.G., Yilmaz G. and Keller J. (submitted). A sequencing batch reactor
system for high-level biological nitrogen and phosphorus removal from abattoir
wastewater. Chemosphere.
Mace S., Dosta J., Gali A. and Mata-Alvarez J. (2006). Optimization of biological nitrogen
removal via nitrite in a SBR treating supernatant from the anaerobic digestion of
municipal solid wastes. Industrial & Engineering Chemistry Research. 45(8): 27872792.
Munch E.V., Lant P. and Keller J. (1996). Simultaneous nitrification and denitrification in
bench-scale sequencing batch reactors. Water Research. 30(2): 277-284.
Pambrun V., Paul E. and Sperandio M. (2004). Treatment of nitrogen and phosphorus in
highly concentrated effluent in SBR and SBBR processes. Water Science and
Technology. 50(6): 269-276.
Peng Y.Z., Chen Y., Peng C.Y., Liu M., Wang S.Y., Song X.Q. and Cui Y.W. (2004). Nitrite
accumulation by aeration controlled in sequencing batch reactors treating domestic
wastewater. Water Science and Technology. 50(10): 35-43.
Peng Y.Z., Li Y.Z., Peng C.Y. and Wang S.Y. (2004a). Nitrogen removal from
pharmaceutical manufacturing wastewater with high concentration of ammonia and
free ammonia via partial nitrification and denitrification. Water Science and
Technology. 50(6): 31-36.
65
Puig S., Vives M.T., Corominas L., Balaguer M.D. and Colprim J. (2004). Wastewater
nitrogen removal in SBRs, applying a step-feed strategy: from lab-scale to pilot-plant
operation. Water Science and Technology. 50(10): 89-96.
Saito T., Brdjanovic D. and van Loosdrecht M.C.M. (2004). Effect of nitrite on phosphate
uptake by phosphate accumulating organisms. Water Research. 38(17): 3760-3768.
Tilche A., Bacilieri E., Bortone G., Malaspina F., Piccinini S. and Stante L. (1999). Biological
phosphorus and nitrogen removal in a full scale sequencing batch reactor treating
piggery wastewater. Water Science and Technology. 40(1): 199-206.
Travers S.M. and Lovett D.A. (1984). Activated-sludge treatment of abattoir wastewater .2.
Influence of dissolved-oxygen concentration. Water Research. 18(4): 435-439.
Turk O. and Mavinic D. (1986). Preliminary assessment of a shortcut in nitrogen removal
from wastewater. Can Journal of Civil Engineering. 13: 600-605.
Vadivelu V.M., Yuan Z.G., Fux C. and Keller J. (2006). The inhibitory effects of free nitrous
acid on the energy generation and growth processes of an enriched Nitrobacter culture.
Environmental Science & Technology. 40(14): 4442-4448.
Villaverde S., Fdz-Polanco F. and Garcia P.A. (2000). Nitrifying biofilm acclimation to free
ammonia in submerged biofilters. Start-up influence. Water Research. 34(2): 602-610.
Villaverde S., GarciaEncina P.A. and FdzPolanco F. (1997). Influence of pH over nitrifying
biofilm activity in submerged biofilters. Water Research. 31(5): 1180-1186.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Wareham D.G., Mavinic D.S. and Hall K.J. (1994). Sludge-digestion using ORP-regulated
aerobic-nnoxic cycles. Water Research. 28(2): 373-384.
Yilmaz G., Lemaire R., Keller J. and Yuan Z. (2007). Effectiveness of an alternating aerobic,
anoxic/anaerobic strategy for maintaining biomass activity of BNR sludge during
long-term starvation. Water Research. 41(12): 2590-2598.
Yoshida Y., Takahashi K., Saito T. and Tanaka K. (2006). The effect of nitrite on aerobic
phosphate uptake and denitrifying activity of phosphate-accumulating organisms.
Water Science and Technology. 53(6): 21-27.
Zeng R., Yuan Z. and Keller J. (2003). Enrichment of denitrifying glycogen-accumulating
organisms in anaerobic/anoxic activated sludge systems. Biotechnology and
Bioengineering. 81(4): 397-404.
Zhou Y., Pijuan M. and Yuan Z. (in press). Free nitrous acid inhibition on anoxic phosphorus
uptake and denitrification by poly-phosphate accumulating organisms. Biotechnology
and Bioengineering.
66
Process
concentration of
5. Aerobic growth on SA (XH)
6. Anoxic growth on SF (XH)
NO3 → NO2
7. Anoxic growth on SF (XH)
NO2 → N2
8. Anoxic growth on SA (XH)
NO3 → NO2
9. Anoxic growth on SA (XH)
NO2 → N2
10. Fermentation
11. Lysis
12. Storage of XPHA
13. Aerobic storage of XPP
14. Anoxic storage of XPP
NO3 → NO2
15. Anoxic storage of XPP
NO2 → N2
16. Aerobic growth of XPAO
17. Anoxic growth of XPAO
NO3 → NO2
oxygen
SF
SA
-1
-1
mgCOD l mgCOD l
Fermentable
COD
VFA COD
1- fSI
ν 3, NH
1- fSI
ν 4, NH
-1/YH
1-1/YH
-1/YH
22. Aerobic growth of XAOB
23. Aerobic growth of XNOB
24. Lysis of XAOB
25. Lysis of XNOB
26. Precipitation
27. Redissolution
SNO2
-1
mgN l
SN2
-1
mgN l
SP
-1
mgP l
Nitrate nitrogen
Nitrite nitrogen
Dissolved N2
Phosphate
4
ν 4,PO
4
4
4
4
-iNBM
4
-1/YH
ν 7 , NH
4
-1/YH
-iNBM
-1/YH
-iNBM
1
iNSF
ν 11, NH
−
(1 − YH )
(8 / 7 ⋅ YH )
(1 − YH )
−
(8 / 7 ⋅ YH )
-iPBM
(1 − YH )
(8 / 7 ⋅ YH )
(1 − YH )
−
(1.72 ⋅ YH )
(1 − YH )
(1.72 ⋅ YH )
(1 − YH )
−
(1.72 ⋅ YH )
(1 − YH )
(1.72 ⋅ YH )
(1 − YH )
(1.72 ⋅ YH )
4
ν 6,PO
4
-iNBM
-iNBM
(1 − 1 / YPAO )
(8 / 7)
-iNBM
ν 19, NH
YPHA
(8 / 7)
Y
− PHA
1.72
−
(1 − 1 / YPAO )
(8 / 7)
(1 − 1 / YPAO )
(1.72)
YPHA
1.72
(1 − 1 / YPAO )
(1.72)
ν 24, NH
ν 25, NH
4
4
1/YNOB
ν 1, ALK
ν 1, ALK
4
4
4
4
ν 1, ALK
4
-iPBM
ν 1, ALK
4
iPSF
ν 11,PO
ν 1, ALK
ν 1, ALK
ν 1, ALK
ν 1, ALK
-1
ν 1, ALK
-1
ν 1, ALK
ν 1, ALK
-iPBM
ν 1, ALK
-iPBM
ν 1, ALK
4
4
4
4
4
4
4
4
4
ν 1, ALK
1/YAOB
-iPBM
ν 1, ALK
4
-1/YNOB
-iPBM
ν 1, ALK
ν 19,PO
1
-iNBM
ν 1, ALK
4
4
-iNBM -1/YAOB
ν 1, ALK
4
4
-iPBM
1
− (3.43 − YAOB )
YAOB
− (1.14 − YNOB )
YNOB
fSI
ν 1, ALK
4
-iPBM
−
fSI
ν 1, ALK
-1
Y
− PHA
(8 / 7)
Inert soluble
Alkalinity
COD
ν 1, ALK
fSI
4
YPO4
-YPHA
SI
SALK
-1
-1
mgCOD l mmol l
ν 7,PO
4
-1
1-1/YPAO
ν 1, PO
ν 2,PO
ν 3,PO
4
4
ν 6 , NH
-1
SNO3
-1
mgN l
4
-1/YH
18. Anoxic growth of XPAO
NO2 → N2
19. Lysis of XPAO
20. Lysis of XPP
21. Lysis of XPHA
Ammonium +
ammonia nitrogen
ν 1, NH
ν 2, NH
1- fSI
1-1/YH
SNH4
-1
mg COD l
ν 24,PO
ν 25,PO
4
4
-1
1
ν 1, ALK
ν 1, ALK
ν 1, ALK
ν 1, ALK
ν 1, ALK
ν 1, ALK
4
4
4
4
4
4
4
4
Table S1. Enhanced ASM2d - reaction stoichiometry for soluble components
1. Aerobic hydrolysis
2. Anoxic hydrolysis
3. Anaerobic hydrolysis
4. Aerobic growth on SF (XH)
SO2
-1
mgO2 l
Process
XI
-1
mgCOD l
1. Aerobic hydrolysis
2. Anoxic hydrolysis
3. Anaerobic hydrolysis
4. Aerobic growth on SF (XH)
5. Aerobic growth on SA (XH)
6. Anoxic growth on SF (XH)
NO3 → NO2
7. Anoxic growth on SF (XH)
NO2 → N2
8. Anoxic growth on SA (XH)
NO3 → NO2
9. Anoxic growth on SA (XH)
NO2 → N2
10. Fermentation
11. Lysis
12. Storage of XPHA
13. Aerobic storage of XPP
14. Anoxic storage of XPP
NO3 → NO2
15. Anoxic storage of XPP
NO2 → N2
16. Aerobic growth of XPAO
17. Anoxic growth of XPAO
NO3 → NO2
18. Anoxic growth of XPAO
NO2 → N2
19. Lysis of XPAO
20. Lysis of XPP
21. Lysis of XPHA
22. Aerobic growth of XAOB
23. Aerobic growth of XNOB
24. Lysis of XAOB
25. Lysis of XNOB
26. Precipitation
27. Redissolution
XH
XPAO
XPP
XPHA
XAOB
XNOB
XTSS
XMeOH
XMeP
-1
-1
-1
-1
-1
-1
-1
-1
-1
mgCOD l
mgCOD l mgP l mgCOD l mgCOD l mgCOD l mgTSS l mgTSS l mgTSS l
Slowly
Ordinary
biodegradable
heterotrophs
particulate COD
Poly-P
accumulating
organisms
Poly-P
PHA
Ammonia
oxidising
bacteria
Nitrite
oxidising
bacteria
-1
-1
-1
1
1
iTSSBM
iTSSBM
iTSSBM
iTSSBM
iTSSBM
1
iTSSBM
1
iTSSBM
1
iTSSBM
1
fXIBM
fXIBM
1- fXIBM
1- fXIBM
-YPO4
1
1
-YPHA
1
-YPHA
ν 14,TSS
1
-YPHA
ν 15,TSS
1
-1/YPAO
ν 16,TSS
1
-1/YPAO
iTSSBM
1
-1/YPAO
iTSSBM
ν 19,TSS
-1
1
1
fXIBM
1- fXIBM
Metal
phosphate
ν 13,TSS
-1
1- fXIBM
Metal
hydroxide
ν 10,TSS
ν 11,TSS
ν 12,TSS
-1
-1
fXIBM
Total
suspended
solids
-1
-1
-3.23
-0.60
iTSSBM
iTSSBM
ν 24,TSS
ν 25,TSS
1.42
-1.42
-3.45
3.45
4.87
-4.87
Table S2. Enhanced ASM2d - reaction stoichiometry for particulate components
Inert
concentration of particulate
COD
XS
-1
mgCOD l
Table S3. Enhanced ASM2d - reaction kinetics.
Process
Kinetic rate
kH ⋅
1. Aerobic hydrolysis
XS
SO
XH
⋅
K O + SO K + X S
X
⋅ XH
XH
2. Anoxic hydrolysis
ηFE ⋅ k H ⋅
XS
S NOX
KO
XH
⋅
⋅
⋅X H
K O + S O K NOX + S NOX K + X S
X
XH
3. Anaerobic hydrolysis
ηFE ⋅ k H ⋅
XS
K NOX
KO
XH
⋅
⋅
K O + S O K NOX + S NOX K + X S
X
µH ⋅
4. Aerobic growth on SF (XH)
µH ⋅
5. Aerobic growth on SA (XH)
6. Anoxic growth on SF (XH)
NO3 → NO2
7. Anoxic growth on SF (XH)
NO2 → N2
8. Anoxic growth on SA (XH)
NO3 → NO2
9. Anoxic growth on SA (XH)
NO2 → N2
µ H ⋅ η NO3 ⋅
S NH 4
S PO4
SO
SF
SF
S ALK
⋅
⋅
⋅
⋅
⋅
⋅ XH
K O + S O K F + S F S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
S NH 4
S PO4
SO
SA
SA
S ALK
⋅
⋅
⋅
⋅
⋅
⋅ XH
K O + S O K A + S A S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
S NO X
S NO3
S NH 4
S PO4
KO
SF
SF
S ALK
⋅
⋅
⋅
⋅
⋅
⋅
⋅
⋅XH
K O + S O K NO X + S NO X S NO2 + S NO3 K F + S F S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
µ H ⋅ ηNO2 ⋅
S NOX
K NOX + S NO X
µ H ⋅ ηNO3 ⋅
⋅
S NO2 + S NO3
µ H ⋅ ηNO2 ⋅
16. Aerobic growth of XPAO
17. Anoxic growth of XPAO
NO3 → NO2
18. Anoxic growth of XPAO
NO2 → N2
19. Lysis of XPAO
20. Lysis of XPP
21. Lysis of XPHA
22. Aerobic growth of XAOB
q PP ⋅
S NH 4
S PO4
SF
SF
S ALK
⋅
⋅
⋅
⋅
⋅ XH
K F + S F S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
S NH 4
S PO4
S NO2
KO
SA
SA
S ALK
⋅
⋅
⋅
⋅
⋅
⋅
⋅ XH
K O + S O K NO2 + S NO2 K A + S A S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
q HFe ⋅
12. Storage of XPHA
⋅
S NO3
S NH 4
S PO4
SA
SA
S ALK
⋅
⋅
⋅
⋅
⋅
⋅
⋅ XH
K NOX + S NOX S NO2 + S NO3 K A + S A S A + S F K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK
K NOX
KO
SF
S ALK
⋅
⋅
⋅
⋅ XH
K O + S O K NOX + S NOX K F + S F K ALK + S ALK
⎛
⎞
S NOX
KO
⎜ bAerH ⋅ S O
⎟ ⋅X H
+ bAnoxH ⋅
⋅
⎜
K O + SO
K O + S O K NOX + S NOX ⎟⎠
⎝
X PP X PAO
SA
S ALK
qPHA ⋅
⋅
⋅
⋅ X PAO
K A + S A K ALK + S ALK K PP + X PP X PAO
11. Lysis
14. Anoxic storage of XPP
NO3 → NO2
15. Anoxic storage of XPP
NO2 → N2
S NO2
S NOX
10. Fermentation
13. Aerobic storage of XPP
S PO4
SO
X PHA X PAO
K MAX ⋅ X PP X PAO
S ALK
⋅
⋅
⋅
⋅
⋅ X PAO
K O + S O K PS + S PO4 K ALK + S ALK K PHA + X PHA X PAO K iPP + K MAX − X PP X PAO
ρ13 ⋅ ηNO3 ⋅
ρ13 ⋅ ηNO2 ⋅
µ PAO ⋅
S NO3
S NOX
KO
⋅
⋅
K O + S O K NOX + S NOX S NO3 + S NO2
S NOX
S NO2
KO
⋅
⋅
K O + S O K NOX + S NOX S NO3 + S NO2
S NH 4
S PO4
SO
X PHA X PAO
S ALK
⋅
⋅
⋅
⋅
⋅ X PAO
K O + S O K NH 4 + S NH 4 K PO4 + S PO4 K ALK + S ALK K PHA + X PHA X PAO
ρ16 ⋅ ηNO3 ⋅
ρ16 ⋅ ηNO2 ⋅
S NO3
S NOX
KO
⋅
⋅
K O + SO K NOX + S NOX S NO3 + S NO2
S NOX
S NO2
KO
⋅
⋅
K O + S O K NOX + S NOX S NO3 + S NO2
⎛
S NOX
KO
⎜ bAerPAO ⋅ S O
+ bAnoxPAO ⋅
⋅
⎜
K
S
K
K
S
+
+
O
O
O
O
NO X + S NO X
⎝
⎛
S NOX
KO
⎜ bAerPP ⋅ SO
+ bAnoxPP ⋅
⋅
⎜
+
+
+ S NOX
K
S
K
K
S
O
O
NO
O
O
⎝
X
24. Lysis of XAOB
25. Lysis of XNOB
⎞
⎟ ⋅ S ALK (K ALK + S ALK ) ⋅ X PAO
⎟
⎠
⎞
⎟ ⋅ S ALK (K ALK + S ALK ) ⋅ X PAO
⎟
⎠
⎛
⎞
S NOX
S
K
O
O
⎜ bAerPHA ⋅
⎟ ⋅ S ALK (K ALK + S ALK ) ⋅ X PAO
+ bAnoxPHA ⋅
⋅
⎜
K O + SO
K O + S O K NOX + S NOX ⎟⎠
⎝
S NH
S PO
SO
S ALK
⋅
⋅
⋅
⋅ X AOB
µ AOB ⋅
KO + SO K NH + S NH K PO + S PO K ALK + S ALK
4
4
23. Aerobic growth of XNOB
⋅X H
XH
µ NOB
4
4
4
4
S PO4
S NO2
SO
S ALK
⋅
⋅
⋅
⋅
⋅ X NOB
K O + S O K NO2 + S NO2 K PO4 + S PO4 K ALK + S ALK
bAerAOB ⋅
S NOX
SO
KO
+ bAnoxAOB ⋅
⋅
⋅ X AOB
KO + SO
K O + S O K NOX + S NOX
⎛
S NOX
KO
⎜ bAerNOB ⋅ S O
+ bAnoxNOB ⋅
⋅
⎜
K O + SO
K O + S O K NOX + S NOX
⎝
K PRE ⋅ S PO4 ⋅ X MeOH
26. Precipitation
27. Redissolution
K RED ⋅ X MeP ⋅ S ALK (K ALK + S ALK )
69
⎞
⎟ ⋅ X NOB
⎟
⎠
Table S4: Definition of the stoichiometric coefficients and their ASM2d default values or
selected values for the new coefficients (in bold).
Coefficient
Definition
Value
Unit
iNSI
iNSF
iNXI
iNXS
iNBM
iPSI
iPSF
iPXI
iPXS
iPBM
iTSSXI
iTSSXS
iTSSBM
fSI
YH
YPAO
YPO4
YPHA
YAOB
YNOB
fXI
N content of inert soluble COD SI
N content of fermentable soluble substrates SF
N content of inert particulate COD XI
N content of slowly biodegradable particulate substrate XI
N content of biomass XH, XPAO, XAOB, XNOB
P content of inert soluble COD SI
P content of fermentable soluble substrates SF
P content of inert particulate COD XI
P content of slowly biodegradable particulate substrate XI
P content of biomass XH, XPAO, XAOB, XNOB
TSS to COD ratio for XI
TSS to COD ratio for XS
TSS to COD ratio for biomass XH, XPAO, XAOB, XNOB
Production of SI in hydrolysis
Yield coefficient heterotrophs
Yield coefficient PAO (biomass/PHA)
Poly-P requirement (PO4 release) per PHA stored
PHA requirement for Poly-P storage
Yield coefficient AOB per NO2Yield coefficient NOB per NO3Fraction of inert COD generated in biomass lysis XH, XPAO, XAOB, XNOB
0.01
0.03
0.02
0.04
0.07
0
0.01
0.01
0.01
0.02
0.75
0.75
0.90
0
0.625
0.625
0.40
0.20
0.17
0.07
0.10
gN gCOD-1
gN gCOD-1
gN gCOD-1
gN gCOD-1
gN gCOD-1
gP gCOD-1
gP gCOD-1
gP gCOD-1
gP gCOD-1
gP gCOD-1
gTSS gCOD-1
gTSS gCOD-1
gTSS gCOD-1
gCOD gCOD-1
gCOD gCOD-1
gCOD gCOD-1
gP gCOD-1
gCOD gP-1
gCOD gN-1
gCOD gN-1
gCOD gCOD-1
70
Table S5: Definition of the kinetic parameters and their ASM2d default values at 20oC or
selected values for the new parameters (in bold).
Parameters
Description
Default (20oC) or
selected (in bold)
ηNO3
ηFe
Hydrolysis rate constant
Anoxic hydrolysis reduction factor
Anaerobic hydrolysis reduction factor
Saturation/inhibition coefficient for oxygen
Saturation/inhibition coefficient for nitrate
Saturation coefficient for particulate COD
Maximum growth rate on substrate
Maximum rate for fermentation
Rate constant for lysis and decay
Saturation/inhibition coefficient for oxygen
Saturation coefficient for growth on SF
Saturation coefficient for fermentation of SF
Saturation coefficient for growth on acetate/propionic SA
Saturation/inhibition coefficient for nitrate
Saturation/inhibition coefficient for nitrite
Saturation/inhibition coefficient for ammonium (Nutrient)
Saturation/inhibition coefficient for phosphate (Nutrient)
Saturation/inhibition coefficient for alkalinity (HCO3-)
Anoxic(NO2-) reduction factor for denitrification
Anoxic(NO3-) reduction factor for denitrification
Maximum growth rate of AOB
Maximum growth rate of NOB
Rate constant for lysis of XAOB
Rate constant for lysis of XNOB
Saturation/inhibition coefficient for oxygen
Saturation coefficient for ammonium (AOB activity)
Saturation coefficient for nitrite (NOB activity)
Saturation/inhibition coefficient for alkalinity (HCO3-)
Saturation/inhibition coefficient for phosphate (Nutrient)
Maximum growth rate of PAO
Rate constant for storage of XPP
Rate constant for storage of XPHA (base XPP)
Rate constant for lysis of XPAO
Rate constant for lysis of XPP
Rate constant for lysis of XPP
Saturation/inhibition coefficient for oxygen
Saturation/inhibition coefficient for nitrate
Saturation/inhibition coefficient for nitrite
Saturation coefficient for growth on acetate/propionic SA
Saturation/inhibition coefficient for ammonium (Nutrient)
Saturation coefficient for phosphorous in storage of PP
Saturation/inhibition coefficient for phosphate (Nutrient)
Saturation/inhibition coefficient for alkalinity (HCO3-)
Maximum ratio of XPP/XPAO
Inhibition coefficient for PP storage
Saturation coefficient for PHA
Anoxic(NO2-) reduction factor for denitrification
Anoxic(NO3-)reduction factor for denitrification
Rate constant for P precipitation
Rate constant for redissolution
Saturation coefficient for alkalinity
3.00
0.60
0.40
0.20
0.50
0.10
6.00
3.00
0.40
0.20
4.00
4.00
4.00
0.50
0.50
0.05
0.01
0.10
0.80
0.80
1.0
1.0
0.15
0.15
0.20
1.0
1.0
0.10
0.01
1.00
1.50
3.00
0.20
0.20
0.20
0.20
0.50
0.50
4.00
0.05
0.20
0.01
0.10
0.34
0.02
0.01
0.60
0.60
1.00
0.60
0.50
kH
KO2
KNO3
KX
µH
qFe
bH
KO2
KF
Kfe
KA
KNO3
KNO2
KNH4
KP
KALK
ηNO2
ηNO3
µAOB
µNOB
bAOB
bNOB
KO2
KAOBNH4
KNOBNO2
KALK
KP
µPAO
qPP
qPHA
bPAO
bPP
bPHA
KO2
KNO3
KNO2
KA
KNH4
KPS
KP
KALK
KMAX
KIPP
KPHA
ηNO2
ηNO3
kPRE
KRED
KALK
71
Units
d-1
gO2 m-3
gN m-3
gXs gXH-1
d-1
gSF gXH-1 d-1
d-1
gO2 m-3
gCOD m-3
gCOD m-3
gCOD m-3
gN m-3
gN m-3
gN m-3
gP m-3
gHCO3- m-3
d-1
d-1
d-1
d-1
gO2 m-3
gN m-3
gN m-3
gHCO3- m-3
gP m-3
d-1
gXPP gXPAO-1 d-1
gXPHA gXPAO-1 d-1
d-1
d-1
d-1
gO2 m-3
gN m-3
gN m-3
gCOD m-3
gN m-3
gP m-3
gP m-3
gHCO3- m-3
gXPP gXPAO
gXPP gXPAO
gXPP gXPAO
m3(gFe(OH)3)-1 d-1
d-1
mole HCO3-m-3
Table S6: Wastewater composition used in simulation.
Parameter Value
SO2
SA
SF
SNO3
SNO2
SN2
SNH4
SP
SHCO
SI
XI
XS
XH
XPAO
XPP
XPHA
XA
XMeOH
XMeP
Xash
XAOB
XNOB
0
625
450
0
0
15
160
40
73
225
100
200
10
0
0
0
0
0
0
10
0
0
72
Unit
mg l-1
mgCOD l-1
mgCOD l-1
mgN l-1
mgN l-1
mgN l-1
mgN l-1
mgP l-1
mmol l-1
mgCOD l-1
mgCOD l-1
mgCOD l-1
mgCOD l-1
mgCOD l-1
mgP l-1
mgCOD l-1
mgCOD l-1
mgSS l-1
mgSS l-1
mgSS l-1
mgCOD l-1
mgCOD l-1
Appendix C
Effectiveness of an Alternating Aerobic, Anoxic/Anaerobic Strategy for
Maintaining Biomass Activity of BNR Sludge during Long-term Starvation
Gulsum Yilmaz, Romain Lemaire, Jurg Keller and Zhiguo Yuan
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane 4072 QLD, Australia.
Water Research (2007) 41(12): 2590-2598
ABSTRACT
The effectiveness of an aerobic, anoxic/anaerobic strategy for maintaining the activity of
activated sludge performing biological nitrogen and phosphorus removal during long-term
starvation is investigated. A lab-scale sequencing batch reactor (SBR) treating abattoir
wastewater and achieving high-levels (>95%) of nitrogen, phosphorus and COD removal was
used. The reactor was put twice into a so-called “sleeping mode” for a period of 5-6 weeks
when the abattoir, where the wastewater was sourced, was closed down for annual
maintenance. The “sleeping mode” operation consisted of 15 minutes aeration in a 6 hour
SBR cycle. The sludge was allowed to settle in the remaining time of the cycle. The decay
rates for ammonia oxidising bacteria (AOB) and nitrite oxidising bacteria (NOB) were
determined to be 0.017 d-1 and 0.004 d-1, respectively. These decay rates correlated well with
AOB and NOB population quantified using molecular techniques (FISH). There was
negligible phosphate accumulation in the reactor during the first 1-2 weeks of starvation,
which was followed by a linear net release of phosphate in the remaining 4-5 weeks at a very
slow rate of 1-2 mgP.gVSS-1.d-1. A sudden decrease in the aerobic activities of polyphosphate
accumulating organisms (PAOs), observed via anaerobic/aerobic batch tests, occurred after
two weeks of starvation. This correlated with a dramatic increase of several metal ions in the
liquid phase. The underlying reasons are not clear. A resuscitation period with a gradual
increase of the wastewater load was applied during the re-startup of the reactor after both
“sleeping mode” periods. Each time, the performance of the reactor in terms of nitrogen and
phosphorus removal fully recovered in four days.
Keywords: Activated sludge, decay, EBPR, nitrifier, PAO, recovery, starvation
73
INTRODUCTION
Several industry sectors such as the dairy and food processing industries, paper mills and
abattoirs are known to be great consumers of water. As a result, large volumes of wastewater
are produced and need to be treated before being released into the municipal sewage or the
natural ecosystems. One of the challenges for such wastewater treatment plants (WWTPs) is
to cope with the large fluctuations of the wastewater flow and composition inherent to the
industrial activities. In some cases, low activity periods for the industry (e.g. annual
maintenance or seasonal production variations) would even result in complete interruptions of
wastewater flows to the WWTPs for weeks and even months. It is crucially important to
maintain the viability of biomass during the long idle or starvation periods.
Many researchers have investigated the effects of starvation on the bacterial population and
activities. The main discoveries to date are summarized below:
The availability of electron acceptor has a major impact on the bacterial decay rates. This has
been well-demonstrated for nitrifiers (Nowak et al., 1994; Roslev and King, 1995; Siegrist et
al., 1999; Morgenroth et al., 2000; Lee and Oleszkiewicz, 2003; Salem et al., 2006). It is now
well established that nitrifiers decay at a higher rate under aerobic conditions than under
anoxic or anaerobic conditions.
There is no or little loss (<5%) of nitrifying activities when activated sludge is starved under
alternating aerobic, anoxic/anaerobic conditions for a period of up to one week (Morgenroth
et al., 2000; Yuan et al., 2000; Lee and Oleszkiewicz, 2003). The growth of nitrifiers during
aerobic period utilising the ammonia released through the decay of heterotrophic bacteria can
largely compensate for the reduction of nitrifying activity caused by the decay of nitrifiers
(Morgenroth et al., 2000). This may explain, at least partially, the observation made in Lee
and Oleszkiewicz (2003) that nitrifiers decayed under alternating aerobic, anoxic conditions at
a rate that is 40% lower than under purely anoxic conditions.
Lopez et al. (2006) investigated the activity of polyphosphate accumulating organisms
(PAOs) during long term aerobic or anaerobic storage (up to 28 days). It was found that,
similar to nitrifiers, PAOs also displayed a considerably lower decay rate under anaerobic
condition than under aerobic condition. It confirmed the results reported in an earlier study
(Brdjanovic et al., 1998) that excessive aeration of enhanced biological phosphorus removal
(EBPR) processes under starvation condition was detrimental to PAOs.
These findings suggest that keeping the sludge under alternating aerobic, anaerobic/anoxic
conditions would be a feasible strategy for maintaining the biomass activities during an
extended idle period.
This paper evaluates the effectiveness of a specific operating strategy, which creates
alternating anaerobic, anoxic and aerobic conditions, in maintaining the nutrient removal
capacity of activated sludge. The simple, easily implementable strategy proposed and
evaluated consists of providing 15 minutes aeration in every six hours with biomass allowed
to settle in the remaining period. The activities of ammonia-oxidising bacteria, nitriteoxidising bacteria as well as PAOs were closely monitored through weekly batch tests over
two five-week starvation periods. The recovery processes of these organisms after the
starvation were also investigated.
74
MATERIAL AND METHODS
Reactor set-up and operation
A lab-scale sequencing batch reactor (SBR) treating abattoir wastewater with consistently
high level (>95%) removal of COD, nitrogen and phosphorus was used in this study. The
SBR with a working volume of 7 l was seeded with sludge from a full-scale SBR treating
abattoir wastewater. The SBR was operated with a cycle time of 6 h in a temperaturecontrolled room (18-22°C). In each cycle, 1 l of abattoir wastewater (composition summarised
in Table 1) was pumped into the reactor under anoxic or anaerobic (when oxidised nitrogen
was completely consumed) conditions, allowing for denitrification and phosphorus release.
During the subsequent aerobic periods, air was provided intermittently using an on/off control
system to keep the dissolved oxygen (DO) level between 1.5 and 2 mgO2.l-1. In these periods,
nitrification and phosphorus uptake occurred. The hydraulic retention time (HRT) and sludge
retention time (SRT) applied to the reactor were 42 h, and 15 days, respectively. The pH in
the system was recorded but not controlled and fluctuated between 7.0 and 7.9 during a
typical cycle.
Table 1. Characteristics of the SBR influent.
Mid-95% range
(n=67)
1,792-3,126
992-1,288
552-712
238-267
193-220
37-40
33-36
Parameter
TCOD (mg.l-1)
SCOD (mg.l-1)
VFA (mgCOD.l-1)
TN (mg.l-1)
NH4-N (mg.l-1)
TP (mg.l-1)
PO4-P (mg.l-1)
Due to annual maintenance, the abattoir closes down for up to six consecutive weeks each
year. During that time, the supply of wastewater to the SBR was interrupted and the SBR
cycle operation was modified according to the alternating aerobic, anoxic (or anaerobic when
nitrate depleted) strategy. The new 6h cycle consisted of 15 min mixed aeration and 345 min
settling periods. This operation is referred to as the “sleeping mode” operation in the sequel
with distinction made between the first and the second “sleeping mode” operations (i.e.
“sleeping mode I” and “sleeping mode II”, respectively) which were applied to the SBR a
year apart. This starvation operation lasted for 30 consecutive days during “sleeping mode I”
and for 44 days during “sleeping mode II”. During the first starvation period, the nitrifying
activity was closely monitored through weekly pulse addition of ammonia and nitrite, while
the activity of PAOs was only monitored by measuring the ortho-phosphate concentration in
the liquid phase. During the second starvation period, the PAO activity was more
comprehensively studied with weekly batch tests to monitor the anaerobic and aerobic
activities of PAOs.
75
Experiments performed
Before each “sleeping mode”
The performance of the reactor was monitored through measuring the NH4+, NO3-, NO2-,
PO43-, COD and TSS concentrations in the effluent, and the MLSS and MLVSS
concentrations in the reactor 3 times a week. The detailed performance was further monitored
by weekly cycle studies, during which mixed liquor samples were taken from the reactor
every 10 – 35 minutes over the 6h cycle, and filtered for NH4+, NO3-, NO2- and PO43-,
analysis.
During “sleeping mode I”
Short experiments (2h) were carried out once a week directly into the parent SBR to monitor
the nitrification activity of the biomass by measuring the ammonium and nitrite oxidation
rates after a pulse addition of ammonia or nitrite. During each of these experiments, the SBR
was continuously aerated. 15 mgN.l-1 of ammonium and 10 mgN.l-1 of nitrite were added at
the start and after 1h of aeration, respectively. Mixed liquor samples were taken and filtered
every 10 min for analysis of NH4+, NO3- and NO2-. The pH was manually controlled in the
range 7.4-7.6. The ammonia oxidation rate and nitrite oxidation rate were determined as the
slopes of the ammonia and nitrite profiles, respectively. The decay rates of ammonia oxidising
bacteria (AOB) and nitrite oxidising bacteria (NOB) were determined using Excel through
fitting an exponential function to the measured profiles of ammonia and nitrite oxidation rates
based on the least square optimisation algorithm. Between each nitrification experiment,
liquid samples were regularly taken from the parent reactor to monitor the evolution of NH4+,
NOx- and PO43- concentrations.
During “sleeping mode II”
During the second starvation period, only the PAO activity was monitored through weekly
anaerobic/aerobic batch tests carried out in a 200 ml reactor. At the start of the experiment,
200 ml of sludge was taken out from the parent SBR and acetate was added to reach a
concentration of 100 mgCOD.l-1. After 1h of anaerobic period, the 200 ml reactor was sparged
with air for 1h. Samples were taken every 10 min for analysis of acetate and PO43-. The pH
was manually controlled in the range 7.4-7.6. The sludge was then returned to the parent SBR
at the completion of each batch test. Liquid samples were regularly taken from the parent
reactor to monitor PO43- and cations concentrations.
After each “sleeping mode”
After each starvation period, the SBR normal operation was gradually resumed. On the first
day, the reactor received only 50% of the normal wastewater feed, which was further
increased to 75% on the second day and 100% on the fourth day. Cycle studies were
performed during the first cycle on Day 1 (50% feed), the second cycle on Day 2 (75%) and
the second cycle on Day 4 (100% feed).
Physico-chemical analyses
The ammonia (NH3 + NH4+), nitrate (NO3-), nitrite (NO2-) and total phosphate (PO43--P) were
analysed using a Lachat QuikChem8000 Flow Injection Analyser (Lachat Instrument,
Milwaukee). Total and soluble chemical oxygen demand (TCOD and SCOD, respectively),
total Kjeldahl nitrogen (TKN), total phosphorus (TP), mixed liquor suspended solid (MLSS)
and volatile MLSS (MLVSS) were analysed according to the standard methods (APHA,
1995). VFAs were measured by Perkin-Elmer gas chromatography with column DB-FFAP
76
15m x 0.53mm x 1.0µm (length x ID x film) at 140°C, while the injector and FID detector
were operated at 220°C and 250°C, respectively. High purity helium was used as carrier gas at
a flow rate of 17 ml.min-1. 0.9 ml of the filtered sample was transferred into a GC vial to
which 0.1 ml of formic acid was added. Calcium, potassium, magnesium and sodium cations
were measured by Inductively Coupled Plasma - Atomic Emission Spectrometry (ICP-AES
Varian Vista-PRO, Varian, Inc.).
Microbial analyses
Sludge samples were fixed and FISH probed as previously described (Amann, 1995).
Oligonucleotide probes used in this study were EUBmix (Daims et al., 1999) for all Bacteria,
PAOmix (Crocetti et al., 2000) for Accumulibacter spp., NTSPA662 (Daims et al., 2001) for
Nitrospira spp., NIT3 (Wagner et al., 1996) for Nitrobacter spp. and NSO1225 (Mobarry et
al., 1996) for most of the β-proteobacterial ammonia oxidising bacteria (AOB). FISH images
were collected using a Zeiss LSM 510 Meta Confocal microscope with a 63x PlanApochromat oil immersion lens. FISH quantification was performed according to (Crocetti et
al., 2002) where the relative abundance of each group was determined as mean percentage of
all bacteria.
RESULTS
Steady state operation before both starvation periods
The SBR was operated at steady state conditions for over one month before the first starvation
period started and for more than a year before the second started. This SBR consistently
achieved over 95% COD and N removal and 98% of P removal after reaching steady state
conditions. Table 2 shows the concentration of NH4+, NOx- (= NO2- + NO3-) and PO43- in the
effluent during the month preceding each starvation period. NH4+ and PO43- in the effluent
were almost undetected, indicating that nitrification and phosphorus removal were complete.
The average MLSS concentration and MLVSS/MLSS ratio were 4.5 g.l-1 and 0.75,
respectively before the “sleeping mode I” and 4.8 g.l-1 and 0.73 before “sleeping mode II”.
Table 2. Nutrient concentration in the effluent during the month of steady state operation
before and after each “sleeping mode” study.
Nutrient
PO43- (mgP.l-1)
NH4+ (mgN.l-1)
NOx- (mgN.l-1)
“sleeping mode” I
(Mid-95% range)
Before
After
0.00-0.04 0.01-0.04
0.0-0.5
0.02-0.08
7.2-12.2
2.9-4.6
77
“sleeping mode” II
(Mid-95% range)
Before
After
0.00-0.17 0.04-0.08
0.07-0.43 0.04-0.30
1.6-3.3
1.8-3.7
Nitrifiers monitoring during “sleeping mode I”
Figure 1 shows the profiles of NH4+ and NO2- oxidation rates measured weekly during the
first starvation period. These rates were calculated from the NH4+ and NO2- profiles measured
during batch tests as explained previously. Initially, the NH4+ oxidation rate was slightly
higher than the NO2- oxidation rate (19.3 mgN.l-1.h-1 compare to 17.8 mgN.l-1.h-1). However,
after 33 days of “sleeping mode I” operation, the NH4+ oxidation rate decreased by 40% to
11.9 mgN.l-1.h-1 while the NO2- oxidation rate barely changed (16.1 mgN.l-1.h-1). The NH4+
oxidation rate therefore decreased 4 times quicker than the NO2- oxidation rate. The decay
rates calculated from the NH4+ and NO2- oxidation rates (Figure 1) were 0.017 d-1 for AOBs
and 0.004 d-1 for NOBs. Using FISH quantification techniques, the NOB population
(Nitrospira spp. and Nitrobacter spp.) was estimated to be 3.1% of the total microbial
population before the starvation period and 2.7% after that. The AOB population (most of the
β-proteobacterial AOB) was estimated at 5.8% and 3.8% before and after the starvation
period, respectively.
rNH4 and rNO2 (mgN.l-1.h-1)
20
-0.0037x
y = 17.879e
R2 = 0.9881
18
16
rNH4
rNO2
Expon. (rNH4)
Expon. (rNO2)
14
12
y = 19.302e-0.0165x
2
R = 0.9745
10
0
5
10
15
20
25
30
Days
Figure 1. Variation of NH4+ and NO2- oxidation rates over the first starvation period and the
best fits produced by the first-order decay model.
The measured NO3- concentration in the reactor during “sleeping mode I” is shown in
Figure 2. NH4+ and NO2- concentrations were also monitored but almost undetectable in
periods other than during the nitrification experiments and hence were not given in Figure 2.
NO3- was present for most of the time so that the condition in the reactor was predominantly
anoxic with short aerobic periods. The accumulation of NO3- seemed to have been primarily
caused by the weekly nitrification experiments carried out with the addition of NH4+ and
NO2-. It should be noted, however, that the total amount of NOx- produced during the 2 h
oxidation experiment was 3.5 to 7 mgN.l-1 higher than the total amount of NH4+ and NO2added (i.e. 25 mgN.l-1 in total, data not shown). It was likely caused by the oxidation of extra
NH4+ coming from bacterial decay and from the slow breakdown of organic nitrogen under
anoxic conditions. The denitrification rates over the period are also presented in Figure 2. The
denitrification rate in the first week was significantly higher than that in the following weeks,
suggesting that the biomass quickly ran out of carbon sources for denitrification.
78
10
60
N-NO3
8
N-NO3 (mg.l-1)
Denit. rate
40
6
30
4
20
Denitrification rate
(mgN.l-1.d-1)
50
2
10
0
0
0
5
10
15
20
25
30
35
Days
Figure 2. Nitrate profile during the starvation period and denitrification rates in periods
between each nitrification experiment (shown by black arrows).
PAO activities during “sleeping mode II”
Figure 3 presents the anaerobic P-release rate, the aerobic P-uptake rate and the P-release:Cuptake ratio measured during each batch test throughout the starvation period. The values
depicted are relative to their initial value at the start of the starvation on Day 0. Over the 43
days starvation period, the anaerobic P-release rate and aerobic P-uptake rate decreased by
70% and 60% respectively while the P-release:C-uptake ratio decreased by only 15%. In
addition, Figure 3 shows the amounts of P released anaerobically and taken up aerobically
during each anaerobic/aerobic batch test. It appears that the amount of P released in each
batch test remained relatively constant (some slight increase after week 4). In contrast, the
total amount of P taken up aerobically started to decrease sharply after 15 days of starvation
and by the end of the starvation period, the amount of P taken up aerobically was only 30% of
that released anaerobically.
P release
P uptake
relative P-rel. rate
relative P-upt. rate
relative P-rel./C-upt. ratio
70
60
1.0
50
0.8
40
0.6
30
0.4
20
0.2
10
0.0
Prelease and Puptake (mgP/L)
r(t) : r(t=0) in mgP/gVSS.h
P/C (t) : P/C (t=0) in Pmol/Cmol
1.2
0
0
5
10
15
20
25
30
35
40
45
Time (Days)
Figure 3. Anaerobic P-release rate, aerobic P-uptake rate and P-release/C-uptake ratio during
anaerobic/aerobic batch tests, expressed relatively to their initial values at the start of the
starvation. The total amounts of P released anaerobically and taken up aerobically during the
batch tests are also depicted (black and grey bars).
79
The concentration of several cations (Ca2+, K+, Mg2+ and Na+) in the liquid phase during the
entire starvation period is presented in Figure 4. The very sharp release of those cations
observed between Day 17 and Day 20 correlates well with the sharp decrease in the aerobic
activities of PAOs reported in Figure 3. This release of cations occurred at the time PO43started to be released in the reactor (later depicted in Figure 5).
70
400
350
Ca
K
Mg
Na
50
40
300
250
200
30
150
20
Na (mg.l-1)
Ca, K and Mg (mg.l-1)
60
100
10
50
0
0
0
5
10
15
20
25
30
35
40
Days
Figure 4. Profiles of some common cations (Ca++, K+, Mg++ and Na+) in the SBR during the
second starvation period.
Figure 5 compares the evolution of the MLSS, the volatile fraction of the biomass (i.e.
VSS:MLSS ratio) and the PO43- concentration in the SBR during both “sleeping mode”
periods. The volatile fraction of the biomass was relatively constant during “sleeping mode I”
and increased slightly during “sleeping mode II” by about 10% (Figure 5a and 5b). The
evolution of the MLSS in the SBR was similar in the two “sleeping mode” periods with a final
decrease of approximately 20% of their initial values (Figure 5a and 5b). In “sleeping
mode I”, the MLSS decreased sharply in the first week while in “sleeping mode II” it stayed
relatively constant for the first 2 weeks and then started to decrease. This difference could be
due to the different starvation conditions applied in the two periods (i.e. anoxic/aerobic for
“sleeping mode I” and “anaerobic/aerobic for sleeping mode II”). The PO43- profiles in the
SBR during the two “sleeping mode” periods present similar trend, with an initial period of 7
days for “sleeping mode I” and 15 days for “sleeping mode II” with no accumulation of PO43followed by a period with a relatively linear accumulation of PO43- (Figure 5c and 5d).
80
MLSS/MLSSo and Fv/Fvo
1.2
1
0.8
0.6
MLSS
0.4
Fraction volatile (Fv)
0.2
(a)
(b)
(c)
(d)
0
-1
P-PO4 (mg.l )
100
80
60
40
20
0
0
5
10
15
20
25
30
Days
0
5
10
15
20
25
30
35
40
45
Days
Figure 5. MLSS and volatile fraction (Fv) profiles relative to their initial values during (a)
“sleeping mode I” and (b) “sleeping mode II”. PO43- profile in the SBR during (c) “sleeping
mode I” and (d) “sleeping mode II”.
Recovery after both “sleeping mode”
Table 3 compares the nitrification rate, the denitrification rate and the amount of P released
and P uptake measured during SBR cycles shortly before the starvation period started,
immediately after the completion of the starvation period (with 50% of normal wastewater
load), two days after (with 75% of normal wastewater load) and four days after resuming
wastewater feeding (with 100% of normal wastewater load) for both “sleeping mode” studies.
The steady-state operations of the SBR prior to the two starvation experiments were quite
different as a number of operating parameters had been changed for other research purposes.
This resulted in a considerably higher nutrient removal performance of the process prior to
“sleeping mode II” experiments. This is of no further relevance to this investigation as the
behaviour in each starvation test is only related to the performance before the start of the test.
After gradually resuming the normal reactor operation, the nitrification rate and denitrification
rate quickly improved and reached their initial value within four days for both starvation
periods (Table 3). The recovery of the P removal activity followed the same trend with the
amount of P-release and P-uptake over a cycle returning to their initial level within four days
after resuming wastewater feeding. The P removal activity was negligible in the first cycle
after both starvation periods.
The good nutrient removal performances were consistently achieved for several months after
both starvation periods. Table 2 summarizes the nutrient levels in the effluent during the
months immediately after each starvation period. Notwithstanding the quick recovery of the
reactor performance, it took around 30 days (2 sludge ages) for the MLSS to return to the
original level in both starvation studies (data not shown).
81
Table 3. Nitrification rate (rNH4+), denitrification rate (rNOx-) and amounts of P released and
uptaken over a cycle, measured during cycle studies performed before the start of the
starvation period, immediately after the starvation period (50% of normal load), 2 days after
starvation (75% of normal load) and 4 days after the starvation (100% of normal load).
Parameter monitored
“Sleeping mode” I or II
Before starvation
After 1st cycle (50%)
After 2 days (75%)
After 4 days (100%)
rNH4+
(mgN.l-1.h-1)
I
II
18.2 25.5
8.2
7.4
12.9 20.8
17.6 29.1
rNOx(mgN.l-1.h-1)
I
II
4.8
12.3
1.9
1.8
4.5
9.6
5.7
11.7
P-release
(mgP.l-1)
I
II
18.8 36.9
2.5
4.4
9.6
31
19.6
47
P-uptake
(mgP.l-1)
I
II
16.1 34.2
2.3
4.8
8.1
28.9
17.4 44.3
DISCUSSION
Nitrifiers decay rate under anoxic conditions with intermittent aeration
During the five batch tests, a sum of 75 mg NH4-N and 50 mg NO2-N were added to each litre
of the mixed liquor. The AOB and NOB growth as a result of these additions is believed to be
insignificant relative to the size of their populations, considering the fact that the SBR had a
volumetric N load of approximately 150 mgN.l-1.d-1 during its normal operation. On the other
hand, these ammonia and nitrite additions could have affected the maintenance metabolism of
the nitrifiers as these bacteria were not truly maintained under starvation conditions, which
could have impacted on the cell decay rate. However, one aim of aerating for 15 min in a 6 h
cycle was to provide nitrifiers the opportunity to oxidise the nitrogen released from cell lysis
and therefore avoid extended starvation conditions. The most significant impact caused by
these additions was a change of the electron acceptor condition in the reactor. As shown in
Figure 2, NO3- accumulated in the reactor during most of the time of the starvation period,
which was caused by the addition of ammonia and nitrite during batch tests. It should be
noted, however, that the NO3- measurements were carried out during periods when the reactor
was mixed (15 min every 6h). Therefore, the measured NO3- concentration may not be
representative to its level under the sludge blanket during settling periods, which constituted
over 95% of the time over a cycle. It is likely that denitrification under the sludge blanket
caused depletion of nitrate, resulting in localised anaerobic conditions. Unfortunately, the
nutrient level under the sludge blanket was not measured directly. The decay rates determined
during “sleeping mode I”, 0.017 d-1 for ammonia oxidisers and 0.004 d-1 for nitrite oxidisers
(Figure 1), were likely those of AOB and NOB under mainly alternating anoxic and anaerobic
conditions with short intermittent aeration periods.
These decay rates are significantly lower than the decay rates generally reported in literature,
including those obtained under anoxic or anaerobic conditions. Salem et al. (2006) presented a
detailed comparison of nitrifier decay rates reported in various studies. Most studies only
determined the AOB decay rate, which ranged between 0.15-0.21 d-1, 0.025-0.06 d-1 and 0.050.2 d-1, respectively, for activated sludge systems at 20°C (similar to the temperature used in
this study) under aerobic, anaerobic and anoxic conditions. Salem et al. (2006) found that the
NOB decay rates in an activated sludge system at 20°C were 0.21 d-1, 0.06 d-1 and 0.12 d-1
under aerobic, anaerobic and anoxic conditions respectively.
The low decay rates found in this study, which were supported by the very quick recovery of
nitrifying activity after the normal operation of the reactor was resumed (Table 3), were likely
a result of the intermittent aeration strategy employed. Lee and Oleszkiewicz (2003) found
82
that the nitrifier decay rate under alternating aerobic and anoxic conditions was 62% and 40%
lower than those under pure aerobic and anoxic conditions, respectively. Yuan et al. (2000)
reported that the nitrifier decay rate was reduced from 0.08 d-1 to 0.03 d-1 in a storage tank
when the aeration frequency was changed from 15 min in every 30 min to 15 min in every 45
min. The importance of short aeration periods during sludge storage was further described by
Morgenroth et al. (2000). They suggested that intermittent aeration during anoxic storage of
biomass would result in the removal of the potential toxic compounds such as H2S that may
be produced under anaerobic or anoxic conditions. However, Salem et al. (2006) recently
studied the effect of H2S on the nitrification activity of AOB and NOB and found that H2S did
not inhibit their activity and therefore was not considered to be toxic for nitrifiers. More
recently, Mansar et al. (2006) reported an AOB and NOB decay rate of 0.015 d-1 and less than
0.001 d-1, respectively, in a conventional activated sludge system under anoxic conditions for
a period of 7 days. However, the starvation conditions used by Manser et al. (2006) were not
completely anoxic as the reactor was aerated once a day for 5 min with the hypothesis that a
complete lack of ATP for the nitrifiers would be prevented. The starvation conditions were
indeed mainly anoxic with only limited aeration and were therefore relatively similar to that
employed in this study, which could explain the similar low decay rates determined.
The relatively faster decay rate of AOB in comparison to NOB revealed by the process data in
this study is not in agreement with Salem et al. (2006), who observed that the AOB and NOB
decay rate is the same at the same corresponding conditions. Nevertheless, the results found in
our study are supported by the FISH quantification results. Over the starvation period, the
AOB population targeted by FISH probe NSO1225 decreased by some 40% which correlates
well with the 40% decrease in the measured AOB activity. The NOB population targeted by
probes NTSPA662 and NIT3 decreased by 13% only which also agrees well with the
observed 10% decrease in the measured NOB activity. It should be noted, however, that the
FISH quantification results should be interpreted qualitatively, as they are given as
percentages of the size of the total bacteria population (targeted by the EUBmix probe) which
is also subject to reduction during starvation. Also, FISH probes target the ribosome RNA,
and consequently the signal intensity is proportional to the quantity of ribosome RNA present
in cells rather than the number of cells.
PAOs activity under long term starvation condition
The behaviour of PAOs under long term starvation conditions has received limited attention to
date. Lopez et al. (2006) investigated the decay rate and the activity of PAOs under
maintained aerobic and anaerobic conditions for 4 weeks. The decay rate and activity of PAOs
were both found to be considerably smaller under anaerobic conditions as compared to
aerobic starvation conditions.
In this study, the monitoring of the PAO activity during “sleeping mode II” through
anaerobic/aerobic batch tests revealed a sudden activity drop after about two weeks of mainly
anaerobic starvation conditions with only brief aerobic periods. This sudden drop was
simultaneously observed for (i) the P-uptake rate and amount of P taken-up (Figure 3), (ii) the
release of cations in the parent SBR (Figure 4) and (iii) the beginning of PO43- accumulation
and decrease of biomass concentration in the SBR (Figure 5d and 5b). The exact reasons
behind this sudden loss of activity are not clear.
The PO43- profiles measured during the two starvation periods (Figure 5c and 5d) are very
different from those reported in Lopez et al. (2006) under either aerobic or anaerobic
conditions. During a four weeks anaerobic storage of an EBPR sludge, Lopez et al. (2006)
observed that all the intracellular poly-phosphate was released in the first few days, and then P
83
release ceased for the remaining period of the study. In contrast, under aerobic condition, P
was released at almost a constant rate over a three weeks period. The primary difference
between the storage conditions used in this study and those used in Lopez et al. (2006) is that
alternating aerobic, anoxic and anaerobic conditions rather than pure aerobic or anaerobic
conditions were applied. An advantage of this strategy is that PAOs were provided conditions
to perform their normal anaerobic and aerobic (or anoxic) metabolism. During anaerobic
periods, which likely happened under the sludge blanket during settling periods, PAOs should
be able to take up the VFAs produced by sludge hydrolysis and fermentation through the
release of internal poly-phosphate. The formation of intracellular polyhydroxyalkanoates
(PHA) should allow them to take up some of the phosphate released during the subsequent
aerobic or anoxic period.
Two different rates of P release were observed during each “sleeping mode” (Figure 5c and
5d). The specific P release rate in the first week of “sleeping mode I” and in the first two
weeks of “sleeping mode II” were calculated as 0.4 mgP.gVSS-1.d-1 and less than 0.1
mgP.gVSS-1.d-1, respectively. If we consider that Accumulibacter spp. (i.e. the only
established organisms performing the PAO phenotype detected in the reactor using FISH
techniques) constituted approximately 30% of the bacterial population prior to each starvation
period (data not shown), the P release rates specific to the PAO biomass are roughly estimated
to be 1.3 mgP.gVSS-1.d-1 and less than 0.3 mgP.gVSS-1.d-1 for “sleeping mode I” and
“sleeping mode II”, respectively. These estimations are considerably lower than the anaerobic
P release rate required for maintenance reported in the literature (45 to 70 mgP.gVSS-1.d-1, see
e.g. (Smolders et al., 1995; Filipe et al., 2001; Oehmen et al., 2005; Lopez et al., 2006). As
discussed above, these low P release rates were likely due to the anaerobic and aerobic/anoxic
recycling of phosphorus. Figure 5a shows that, during “sleeping mode I”, the MLSS
concentration decreased by 20% in the first week, implying that a considerable amount of
COD was released in this period. This availability of COD supported denitrification at a
relatively high rate as depicted in Figure 2, creating anaerobic conditions particularly under
the sludge blanket during settling periods. Part of the COD released was likely recycled by
PAOs, enabling these organisms to effectively keep their pools of poly-phosphate.
The relatively constant P release rates observed after one week of “sleeping mode I” and two
weeks of “sleeping mode II” (2.1 mgP.gVSS-1.d-1 and 1 mgP.gVSS-1.d-1, respectively) were
probably a consequence of maintenance processes and the lysis of PAO cells, both leading to
release of PO43-. Unfortunately, there are to our knowledge, no techniques at present to
distinguish between the two processes. The relatively constant VSS:MLSS ratio suggests that
the lysis of PAO cells could be the dominant process as it should lead to a proportional
decrease of MLSS and VSS. When maintenance was the primary cause for P release, Lopez et
al. (2006) reported that the volatile fraction of the biomass increased by 40% in two days of
starvation. However, it is not possible to be conclusive here due to the complex nature of the
VSS in the sludge used in this study. In addition to PAO biomass, it contains ordinary
heterotrophs, nitrifiers, as well as other biodegradable or non-biodegradable organics
originating from the wastewater or cell lysis.
Recovery of the process performance after long term starvation
When studying the impact of starvation on biomass activities, the most important questions to
be answered from a practical point of view are (i) can the process fully recover when the
starvation period ceases? (ii) how quickly can a full recovery occur?
The transient phase from a starved to a fully functional system is critical, and needs to be
managed carefully. Even when an adequate amount of microorganisms is maintained in the
84
system through applying appropriate storage strategies, the level of intracellular materials
such as enzymes and RNAs are likely to be very low, if not exhausted, due to maintenance
mechanisms and stressful conditions, and need to be re-synthesised. Some cells may need to
be ‘waken up’ through resuscitation strategies (Van Loosdrecht and Henze, 1999). Tappe et
al. (1999) reported that five-day resuscitation allowed a Nitrobacter culture starved for 35
days to resume 50% of its original activity. A resuscitation period is believed to be even more
important for PAOs, as their metabolism requires the cycling of phosphorus and carbon
compounds under alternating anaerobic, aerobic or anoxic conditions. After an extended
period of starvation, the pools of polyphosphate, PHA and glycogen, which are essential for
the PAO metabolism (Smolders et al., 1994) are likely to be very low and need to be redeveloped slowly.
The stepwise strategy adopted in this study was very successful as shown by the different
activity rates depicted in Table 3. The nitrification rates in the first cycle after “sleeping mode
I” and “sleeping mode II” were 50-70% lower than that obtained before each starvation
started, but increased quickly to reach their previous level in just 4 days (Table 3). The
complete nitrification observed throughout the whole month following the end of both
starvation periods (Table 2) confirms that the recovery was permanent. The denitrification
performance even improved after “sleeping mode I”, as indicated by the lower level of NOxin the effluent (Table 2). The reasons for this N removal improvement are not known.
The quick recovery of P removal after a long starvation period has not been widely reported in
literature. The very low amount of P release and P uptake in the first cycle after the starvation
(Table 3) is not unexpected. It was most likely due to the low level of polyphosphate and PHA
in the PAO cells. The high level of NOx- in the reactor after “sleeping mode I” would have
also inhibited the anaerobic VFA uptake by PAOs (Furumai et al., 1999). However, the
amount of P-release and P-uptake measured 4 days after both long-term starvations are almost
identical to those measured before starvations started (Table 3), suggesting PAOs had a full
recovery. This recovery is, like for the nitrifiers, a genuine long term recovery as indicated by
the excellent long term P removal performance observed after each starvation (Table 2).
CONCLUSION
Alternating anoxic/anaerobic and aerobic strategy is effective in maintaining the biomass
activities of activated sludge performing biological nitrogen and phosphorus removal. Sludge
can be stored under such conditions for at least six weeks with its nitrifying, denitrifying and
phosphorus removal capabilities adequately maintained to allow for a quick recovery when
wastewater feed resumes. While an intermittent aeration of 15 minutes in every 6 hours was
used in this study and proved to be effective, the optimal aeration frequency and duration are
yet to be identified through further experimental studies.
The resuscitation strategy of gradually increasing the wastewater load in the first few days
after a starvation period was also demonstrated to be successful.
ACKNOWLEDGEMENT
The project was funded by the Environmental Biotechnology CRC through Project P5. The
first author thanks Istanbul University for fellowship support. FISH quantification was carried
out by Dr Gregory Crocetti from the Advanced Wastewater Management Centre at The
University of Queensland, Australia.
85
REFERENCES
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Brdjanovic D., Slamet A., van Loosdrecht M.C.M., Hooijmans C.M., Alaerts G.J. and Heijnen
J.J. (1998). Impact of excessive aeration on biological phosphorus removal from
wastewater. Water Research. 32(1): 200-208.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
Filipe C.D.M., Daigger G.T. and Grady C.P.L. (2001). pH as a key factor in the competition
between glycogen-accumulating organisms and phosphorus-accumulating organisms.
Water Environment Research. 73(2): 223-232.
Furumai H., Kazmi A.A., Fujita M., Furuya Y. and Sasaki K. (1999). Modeling long term
nutrient removal in a sequencing batch reactor. Water Research. 33(11): 2708-2714.
Lee Y. and Oleszkiewicz J.A. (2003). Effects of predation and ORP conditions on the
performance of nitrifiers in activated sludge systems. Water Research. 37(17): 42024210.
Lopez C., Pons M.N. and Morgenroth E. (2006). Endogenous processes during long-term
starvation in activated sludge performing enhanced biological phosphorus removal.
Water Research. 40(8): 1519-1530.
Manser R., Gujer W. and Siegrist H. (2006). Decay processes of nitrifying bacteria in
biological wastewater treatment systems. Water Research. 40(12): 2416-2426.
Mobarry B.K., Wagner M., Urbain V., Rittmann B.E. and Stahl D.A. (1996). Phylogenetic
probes for analyzing abundance and spatial organization of nitrifying bacteria. Applied
and Environmental Microbiology. 62(6): 2156-2162.
Morgenroth E., Obermayer A., Arnold E., Bruhl A., Wagner M. and Wilderer P.A. (2000).
Effect of long-term idle periods on the performance of sequencing batch reactors.
Water Science and Technology. 41(1): 105-113.
Nowak O., Schweighofer P. and Svardal K. (1994). Nitrification inhibition - a method for the
estimation of actual maximum autotrophic growth-rates in activated-sludge systems.
Water Science and Technology. 30(6): 9-19.
86
Oehmen A., Zeng R.J., Yuan Z.G. and Keller J. (2005). Anaerobic metabolism of propionate
by polyphosphate-accumulating organisms in enhanced biological phosphorus removal
systems. Biotechnology and Bioengineering. 91(1): 43-53.
Roslev P. and King G.M. (1995). Aerobic and anaerobic starvation metabolism in
methanotrophic bacteria. Applied and Environmental Microbiology. 61(4): 1563-1570.
Salem S., Moussa M.S. and van Loosdrecht M.C.M. (2006). Determination of the decay rate
of nitrifying bacteria. Biotechnology and Bioengineering. 94(2): 252-262.
Siegrist H., Brunner I., Koch G., Phan L.C. and Le V.C. (1999). Reduction of biomass decay
rate under anoxic and anaerobic conditions. Water Science and Technology. 39(1):
129-137.
Smolders G., Vandermeij J., Vanloosdrecht M. and Heijnen J. (1994). Model of the anaerobic
metabolism of the biological phosphorus removal process - stoichiometry and pH
influence. Biotechnology and Bioengineering. 43(6): 461-470.
Smolders G.J.F., Klop J.M., van Loosdrecht M.C.M. and Heijnen J.J. (1995). A metabolic
model of the biological phosphorus removal process I. Effect of sludge retention time.
Biotechnology and Bioengineering. 48(3): 222-233.
Tappe W., Laverman A., Bohland M., Braster M., Rittershaus S., Groeneweg J. and van
Verseveld H.W. (1999). Maintenance energy demand and starvation recovery
dynamics of Nitrosomonas europaea and Nitrobacter winogradskyi cultivated in a
retentostat with complete biomass retention. Applied and Environmental
Microbiology. 65(6): 2471-2477.
Van Loosdrecht M.C.M. and Henze M. (1999). Maintenance, endogeneous respiration, lysis,
decay and predation. Water Science and Technology. 39(1): 107-117.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Yuan Z.G., Bogaert H., Leten J. and Verstraete W. (2000). Reducing the size of a nitrogen
removal activated sludge plant by shortening the retention time of inert solids via
sludge storage. Water Research. 34(2): 539-549.
87
Appendix D
Identifying Causes for N2O Accumulation in a Lab-scale Sequencing Batch
Reactor performing Simultaneous Nitrification, Denitrification and Phosphorus
Removal
Romain Lemaire1, Rikke Meyer1, Annelies Taske1, Gregory R. Crocetti1,2, Jürg Keller1 and
Zhiguo Yuan1
1
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane 4072 QLD, Australia.
2
School of Biotechnology and Biomolecular Sciences, The University of New South Wales,
Sydney 2052 NSW, Australia.
Journal of Biotechnology. (2006) 122(1): 62-72
ABSTRACT
The recently described process of simultaneous nitrification, denitrification and phosphorus
removal (SNDPR) has a great potential to save capital and operating costs for wastewater
treatment plants. However, the presence of glycogen accumulating organisms (GAOs) and the
accumulation of nitrous oxide (N2O) can severely compromise the advantages of this process.
In this study, these two issues were investigated using a lab-scale sequencing batch reactor
performing SNDPR over a five month period. The reactor was highly enriched in
polyphosphate-accumulating organisms (PAOs) and GAOs representing around 70% of the
total microbial community. PAOs were the dominant population at all times and their
abundance increased while GAOs population decreased over the study period. Anoxic batch
tests demonstrated that GAOs rather than denitrifying PAOs, were responsible for
denitrification. N2O accumulated from denitrification and more than half of the nitrogen
supplied in a reactor cycle was released into the atmosphere as N2O. After mixing SNDPR
sludge with other denitrifying sludge, N2O present in the bulk liquid was reduced immediately
if external carbon was added. We therefore suggest that the N2O accumulation observed in the
SNDPR reactor is an artefact of the low microbial diversity facilitated by the use of synthetic
wastewater with only a single carbon source.
Keywords: GAO, microbial diversity, N2O, PAO, SBR, SNDPR.
88
INTRODUCTION
The general degradation of water quality in rivers and streams in both rural and urban areas
has resulted in stronger actions taken towards preventing eutrophication by reducing nutrient
levels in wastewater discharged into local waterways. On a global scale, pollution with
greenhouse gases contributing to global warming is the focus of a large field of research, and
the link between eutrophication and greenhouse gas (N2O, CO2, CH4) emissions from natural
systems (Brink et al., 2001) reinforces the motivation for removing nutrients from wastewater
in an efficient and environmentally friendly way. However, biological wastewater treatment
plants contribute to the emission of greenhouse gas. While CO2 is an inevitable product from
the breakdown of organic carbon, the much stronger greenhouse gas nitrous oxide (N2O), an
intermediate product of both nitrification and denitrification (Beline et al., 2001), may also be
emitted from wastewater treatment processes. In the present study, we investigate and discuss
the cause and prevention of N2O emission from a novel biological nutrient removal process,
from which the accumulation of N2O had previously been identified as a major concern.
In conventional biological nutrient removal systems, ammonium (NH4+) is first oxidised to
nitrite or nitrate (NOx-), which is then reduced to di-nitrogen gas in a two-stage aerobic and
anoxic (O2 absent, NOx- present) process (Metcalf & Eddy, 1991), while phosphorus (P) can
be removed through enhanced biological phosphorus removal (EBPR). EBPR is based on the
ability of polyphosphate-accumulating organisms (PAOs) to take up P and accumulate it
intracellularly as polyphosphate when exposed to alternating anaerobic (O2 and NOx- absent)
and aerobic conditions (Comeau et al., 1986; Wentzel et al., 1988). An essential requirement
for successful EBPR is to only provide carbon under anaerobic conditions in order to provide
PAOs with a selective advantage, as other heterotrophic organisms cannot take up carbon in
the absence of an electron acceptor. Unfortunately, these conditions are not compatible with
the conditions required for N removal, and EBPR failure is regularly observed when NOx- is
present in the designated anaerobic zone due to competition for carbon between denitrifying
organisms and PAOs (Pitman et al., 1983).
It has been found that PAOs capable of denitrification (DPAOs) can perform P uptake with
NOx- as electron acceptor (Kuba et al., 1993; Meinhold et al., 1999). This finding offers a
possibility for removing N and P from wastewater, even with low COD:N and COD:P ratios,
as the same carbon could be used for both denitrification and P removal. More recently, Zeng
et al. (2003b) successfully demonstrated a process in an anaerobic-aerobic reactor where
nitrification and denitrification occurs simultaneously with P uptake under low-oxygen
concentration. This novel integrated process was called simultaneous nitrification,
denitrification and phosphorus removal (SNDPR). In SNDPR, acetate was taken up under
anaerobic conditions, accompanied by phosphorus release. During the following aerobic
period, phosphorus was fully taken up, while NH4+ was converted through simultaneous
nitrification and denitrification to gaseous nitrogen without accumulation of nitrite or nitrate
intermediates. However, detailed off-gas analysis in several studies of the SNDPR process
have shown that N2O rather than N2 was the major denitrification end-product (Zeng et al.,
2003b; Meyer et al., 2005), which is of significant environmental concern due to the high
global warming potential of N2O. This finding significantly diminishes the overall benefits of
the SNDPR process and limits the prospect of implementing this process in wastewater
treatment plants.
This study seeks to characterise the SNDPR process with the aim to elucidate factors
responsible for N2O accumulation, and test how it can be eliminated. Emission of N2O from
SNDPR is the main obstacle for pursuing the full potential of this process, and it is therefore
our goal to address how this issue can be managed.
89
MATERIAL AND METHODS
Reactor setup and operation
The biomass was enriched in a 5 l sequencing batch reactor (SBR) seeded with sludge from
the Caboolture Sewage Treatment Plant, Queensland, Australia. The SBR was operated at
room temperature (20-22°C) with a cycle time of 6 h, consisting of a 90 min anaerobic period,
followed by 220 min aeration, 40 min settling, and 10 min decant. Three litres of synthetic
wastewater was pumped into the reactor in the first 7 min of the anaerobic period, and 3 l
supernatant was removed after settling, resulting in a hydraulic retention time (HRT) of 10 h.
Before settling, 62.5 ml mixed liquor was wasted to keep the solids retention time (SRT) at
approximately 20 days. Aeration was provided at a flow of 0.5 l.min-1, using an on/off control
system to keep the dissolved oxygen level at between 0.35-0.5 mg.l-1. The pH in the system
was recorded but not controlled, and fluctuated between 7.0 and 7.5. The synthetic wastewater
fed in each cycle was prepared as described in Zeng et al. (2003b) with the adjustment of
carbon, nitrogen and phosphorus to the level of 230 mg.l-1 COD as acetate, 23 mg.l-1 NH4-N
and 18 mg.l-1 PO4-P. The performance of the reactor was monitored by weekly cycle studies,
during which samples for NH4+, PO43-, and NOx- were taken regularly during the cycle and
N2O was measured on-line with a N2O microsensor (see below). The N2O concentration was
measured with high temporal resolution, and these data were used for calculating the amount
of N2O stripped from the sludge during the cycle. As described above, the aeration of the
reactor was controlled in an on/off manner. Assuming the N2O transfer from the liquid to the
gas phase is negligible during aeration off periods (due to the limited surface area in this
reactor and the relatively high solubility of N2O, 26.2 mmol.l-1 at 22°C), the slopes of the N2O
data in these periods represent the net biological production (when slopes are positive) or
consumption (when slopes are negative) rates of N2O. Further assuming that these rates
remain constant during each aeration on/off cycle as each cycle is of a period of 1-3 min only,
the integration of the net biological production/consumption rates of N2O over the entire
aerobic period of the SBR operation yields a rough estimate of the amount of N2O that was
stripped during the entire aerobic period.
Physico-chemical analyses
Ammonia nitrogen (NH4-N), nitrate nitrogen (NO3-N), nitrite nitrogen (NO2-N) and
orthophosphate (PO4-P) were analysed using a Lachat QuikChem8000 Flow Injection
Analyser (Lachat Instrument, Milwaukee). Mixed liquor suspended solid (MLSS) and volatile
MLSS (MLVSS) were analysed according to the standard methods of APHA (APHA, 1995).
Acetate was measured by Perkin-Elmer gas chromatography with column DB-FFAP 15 m x
0.53 mm x 1.0 µm (length x ID x film) at 140°C while the injector and FID detector were
operated at 220°C and 250°C respectively. Polyhydroxybutyrate (PHB), polyhydroxyvalerate
(PHV) and glycogen were determined using the method described in Zeng et al. (2003b).
Oxygen mass transfer limitation
The gradient of oxygen in granules present in the reactor was measured with oxygen
microsensors (tip diameter <10 µm), which were constructed as described by Revsbech et al.
(1989). Granules were sampled from the parent reactor 1h into the aerobic period, at which
time NH4+ was usually depleted, and transferred to flow-cell with an upward flow as
described in Meyer et al. (2003). Replicate oxygen profiles were then measured over the
90
following hour. Profile measurements were analysed in the absence of NH4+ to simulate the
conditions in the parent reactor after NH4+ depletion when oxygen was more likely to fully
penetrate the flocs or granules.
The granules analysed with oxygen microsensors were larger than 500 µm in diameter, as the
floccular biomass was not amenable to microsensor analysis. The sensor was moved stepwise
into the granules from above, and movement of the sensor and data acquisition was obtained
with the software Profix (Unisense A/S, Aarhus, Denmark). The composition of medium in
the flowcell was identical to that in the parent reactor at the end of the anaerobic period except
for NH4+. The oxygen concentration was adjusted within the nutrient medium by controlling
the air/nitrogen ratio in the gas inlet.
Microbial analyses
Sample fixation and FISH experiments were performed as previously described in Amann
(1995). Oligonucleotide probes used in this study were EUBmix for the detection of all
bacteria (Daims et al., 1999), PAOmix for “Candidatus Accumulibacter phosphatis” (referred
to as Accumulibacter in the following) (Crocetti et al., 2000) and GAOQ989 (Crocetti et al.,
2002) and GB_G2 (Kong et al., 2002) for “Candidatus Competibacter phosphatis” (referred to
as Competibacter in the following). FISH quantification was performed using previously
published method in Crocetti et al. (2002).
Batch tests
Measurement of net N2O production and consumption by denitrification in SNDPR sludge
with nitrate or nitrite as electron acceptors
Sludge was sampled at the end of the anaerobic period from the parent reactor and amended
with KNO3 or NaNO2 to reach a concentration of 7.5 mgN.l-1 of NO3- or NO2- before being
transferred to two 14.75 ml vials sealed with rubber stoppers to which a N2O microsensor was
inserted for on-line monitoring of the N2O concentration. The mixed liquor was stirred using a
magnetic stirrer, and the vials were filled completely to avoid any exchange of N2O between
liquid and gas phases. The two mini-reactors were operated in parallel with one acting as a
negative control or a duplicate. A sample was taken at the end of the experiment for NOxanalysis to confirm the depletion of nitrite or nitrate from denitrification. Substrates could be
added in small amounts at any time during experimentation with a syringe through the rubber
stopper. The N2O microsensors used in these experiments were constructed according to
Andersen et al. (2001). The maximum N2O net production rate was calculated from the very
first part of the graph when N2O accumulated almost linearly, whereas the gross consumption
rate was calculated from the last part of the graph when nitrate or nitrite was depleted and the
N2O concentration started to decrease. Both rates were corrected for MLVSS variations in
different batch tests.
Measurement of net N2O production and consumption by denitrification in mixed SNDPR and
denitrifying sludges in presence of external carbon sources.
To investigate how the presence of denitrifying bacteria other than those found in the SNDPR
sludge affect N2O accumulation, batch tests were performed in duplicate with a mixture of
sludge from the SNDPR reactor and a lab-scale nitrifying-denitrifying reactor treating
domestic wastewater containing additional carbon in the form of methanol for denitrification.
The SNDPR sludge was sampled at the end of the anaerobic period, mixed with the
denitrifying sludge in a 1:2 ratio, and amended with KNO3 to a concentration of 7.5 mgN.l-1
91
of NO3- at time=zero, before being transferred to the two sealed vials for monitoring of the
N2O concentration. After 10 min, methanol was injected into one of the vials to a
concentration of 30 mgCOD.l-1. The addition of external carbon was delayed to allow
observation of N2O accumulation when only bacteria with intracellular PHA were active,
before studying the effect of other denitrifying bacteria on N2O accumulation after external
carbon was added. As a negative control, this experiment was repeated with SNDPR sludge
only.
It could be argued that adding methanol to a methanol-adapted sludge would bias the results
from the experiment detailed above, and therefore the effect of adding real wastewater to
SNDPR sludge mixed with denitrifying sludge that had not been adapted to a particular
carbon source was also tested. This was done in a 500 ml reactor from which NOx- samples
could be taken during the experiment. SNDPR sludge was sampled at the end of the anaerobic
period, mixed in a 1:2 ratio with sludge from a denitrifying SBR treating domestic
wastewater, and transferred to the 500 ml reactor. The reactor was sealed at the top and a N2O
microsensor was mounted through the lid for on-line measurements, but it should be noted
that the headspace in the reactor would cause some stripping of N2O to the gas phase during
the experiment. Prior to the experiment, helium was sparged for 10 min to obtain anaerobic
conditions before nitrate was added to a concentration of 7.5 mgN.l-1 of NO3- at Time=zero.
After 35 min, 50 ml of high-strength wastewater from meat processing industry (approx. 6000
mgCOD.l-1) was added to the mixed liquor.
Denitrification test
To determine whether PAOs were contributing to denitrification in the reactor, batch tests
were performed to measure if denitrification was accompanied by P uptake under strictly
anoxic conditions. Sludge (500 ml) was taken from the SNDPR reactor at the end of the
anaerobic period and placed in the same 500 ml reactor described before mounted with a N2O
microsensor. After 10 min of helium sparging, 5 ml of NaNO3 or NaNO2 solution was added
to obtain an initial concentration of 8 mgN.l-1 of NO3- or NO2- in the reactor. Liquid samples
were taken at regular intervals to monitor the PO43- , NH4+ and NOx- concentration.
RESULTS and DISCUSSION
Process and microbial characterisation
Over the 5 months operation period, MLSS varied between 3.94 and 4.64 g.l-1 with a standard
deviation (S.D) of the mean of 0.23, and MLVSS varied between 2.18 and 2.7 g.l-1 with a S.D.
of 0.21. The microbial community consisted mostly of Accumulibacter PAOs and
Competibacter GAOs, where the sum of these two populations accounted for 63 to 78.5% of
the biomass. Accumulibacter increased in abundance from 45 to 70% over the 5 months
operational period while the abundance of Competibacter decreased from 17.6% to only 8.6%
of the total population.
Weekly cycle studies showed little variation in the performance of the reactor in June- July
(the first 2 months), but the performance gradually changed in the months of August to
October towards increasing NOx- level present at the end of the reactor cycle. Figure 1 shows
two cycle studies representing the stable period from June to July (Figure 1a), and the period
after a gradual increase in the effluent NOx- concentration from August to October
(Figure 1b). During the anaerobic period of the reactor cycle, acetate was rapidly consumed
and stored as PHA (PHB and PHV), which was accompanied by release of phosphorus and
92
the consumption of glycogen (see Figure 1b). During the aerobic period, phosphorus was
completely taken up, PHA was oxidised and the glycogen pool was replenished (Figure 2b).
Ammonium was removed without accumulation of NOx- in the June-July period (Figure 1a),
but in the August-October period, NOx- accumulated in the liquid (Figure 1b).
A transient accumulation of N2O was observed in the aerobic period throughout the study
(Figure 1a and b). The N2O concentration peaked at the same time as NH4+ was depleted in
the June-July test (Figure1a) and slightly earlier in the August-October test (Figure 1b). The
accumulated N2O was eventually removed before the end of the cycle for both periods.
Stripping of N2O occurred during the aerobic period was estimated to be 27.5 mgN,
corresponding to 51% of the N removed in each cycle. Although SNDPR has been proposed
as an economically feasible process (Zeng et al., 2003b), the apparent emission of N2O poses
an environmental problem which hinders its implementation in wastewater treatment.
anaerobic
aerobic
160
4
12
120
3
8
80
2
4
40
16
0
0
NH4-N
NO2-N
NO3-N
Glycogen
PHB+PHV
PO4-P
acetate
N-N2O
(b)
12
8
120
4
0
0
1
2
3
4
5
0
-1
1
N-N2O (mg l )
-1
P-PO4 and Acetate (mg l )
-1
Glycogen and PHA (C-mmol l )
-1
N-NH 4 , N-NO 3 and N-NO 2 (mg l )
(a)
3
80
2
40
1
0
0
6
Time (h)
Figure 1. Typical weekly cycle study of the parent SBR (a) for the first 2 months (June-July)
when complete N removal was achieved and (b) in September when the reactor showed only
partial removal of N. The dashed line indicates the transition from the anaerobic to the aerobic
period.
Factors affecting N2O accumulation
Nitrous oxide can potentially accumulate from both nitrification and denitrification (Beline et
al., 2001), however, accumulation of N2O under strictly anoxic conditions (Figure 2) indicates
that denitrification was the main source of N2O in the present study. Di-nitrogen gas is most
often the predominant product of denitrification and N2O is rarely monitored in wastewater
treatment systems.
93
7
nitrite addition
nitrate addition
5
-1
N-N2O (mg l )
6
4
3
2
1
0
0
20
40
60
80
100
120
time (min)
Figure 2. N2O concentration during anoxic batch tests (14.75 ml sealed reactors) amended
with either nitrate or nitrite (7.5 mgN.l-1) at T=0.
A number of factors have been suggested to cause N2O accumulation from denitrification in
activated sludge. Hanaki et al. (1992) reported N2O production during denitrification in the
presence of low COD:N ratio (3.5), low pH (6.5), and short solids retention time (<1 day).
Kishida et al. (2004) also reported N2O production under low COD:N ratio (2.6). The SNDPR
reactor in our study was operated at a COD:N ratio of 10, pH of 7-7.5, and a solids retention
time of 20 days. As such, these factors are therefore not believed to affect the N2O production
in our study.
The N2O reductase is very sensitive to oxidative stress (Otte et al., 1996; Noda et al., 2003),
and denitrification occuring at low oxygen concentrations may therefore lead to N2O
accumulation, as it has been shown in pure culture studies (Otte et al., 1996). As SND relies
on an oxygen gradient into the biomass, it is possible that denitrification takes place under
microaerobic conditions. However, the N2O accumulation observed under completely anoxic
conditions (Figure 2) excludes oxygen as the single causing factor.
Another possible cause for N2O production is the presence of nitrite. Zeng et al. (2003b)
suggested that nitrite accumulation at even at very low levels triggered the N2O production
observed in their study. Itokawa et al. (2001) also reported nitrite accumulation as a possible
cause of N2O production in a denitrifying sludge, yet hypothesised that the accumulation may
be in combination with denitrification based on intracellular carbon. The potential rates of net
N2O production and consumption from denitrification investigated in anoxic, sealed batch
reactors with on-line monitoring of N2O, showed that the maximum net production rates of
N2O varied greatly with the electron acceptor provided (Table 1). Addition of nitrite to the
sludge led to a 5 times higher net N2O production rate, as compared to nitrate addition.
Figure 2 shows, as an example, one of 5 replicate batch tests. The peak concentration of N2O
corresponded to 77% of the N added in the incubations with nitrite, whereas only 26% of the
N added as nitrate accumulated as N2O. When accumulation of N2O stopped, there was no
significant difference between the N2O reduction rates measured in the different incubations
(Table 1), and these rates were similar to N2O reduction rates obtained when only N2O was
added as electron donor (data not shown). These results could suggest an effect of nitrite on
N2O production. However, a difference in the rate of nitrate and nitrite reduction could also
affect the accumulation of N2O if nitrate reduction is the rate limiting step. The two anoxic
batch tests performed with nitrate or nitrite as electron acceptors in the non-sealed 500 ml
reactor allowed NOx- samples to be taken simultaneously to N2O measurements. Figure 3a
94
and 3b show that the nitrite reduction rate was twice as high as the nitrate reduction rate and
nitrite did not accumulate in the batch test amended with nitrate. Nitrate reduction being the
rate limiting step may therefore explain the differences in N2O accumulation rate observed in
Figure 2. The profile showing the sum of nitrogen oxides (NOx- plus N2O) in Figure 3b
suggests that high concentrations of nitrite (> approx. 5 mgN.l-1) may inhibit N2O reduction as
the concentration of nitrogen oxides did not decrease in the first 10 min, indicating that nitrite
was only reduced to N2O in this period. After the 10 min, an inflection point in the N2O
concentration curve was observed, indicating a lower net N2O production rate. At the same
time, the nitrogen oxide concentration began to decrease, and the inflection point therefore
reflected the onset of N2O reduction. It should be noted that N2O accumulated even in the
absence of nitrite (Figure 3a), and the presence of nitrite cannot be the sole factor responsible
for the accumulation of N2O.
Table 1. Production and consumption rates of N2O (µmol.h-1.gVSS-1) by the SNDPR sludge.
Electron
acceptor
NO2NO3-
Mean N2O net production rate
± S.E. (n=5)
201.5 ± 14.2
37.7 ± 4.2
Mean N2O gross consumption rate
± S.E. (n=5)
90.8 ± 4.7
86.7 ± 5.5
In addition to the physico-chemical factors that may trigger N2O accumulation, the faster
reduction of nitrite as compared to N2O may simply be related to the enzyme kinetics of the
organism(s) carrying out denitrification in the reactor. Nitrous oxide accumulation has
previously been observed in SNDPR reactors (Zeng et al., 2003b; Meyer et al., 2005) and in
reactors enriched in denitrifying glycogen-accumulating organisms (DGAO) (Zeng et al.,
2003a). These reactors share the feature of performing denitrification using intracellular
carbon, which was taken up under anaerobic conditions where NOx- was absent. The DGAO
and the SNDPR reactors could therefore harbour similar denitrifying organisms, and the
observed N2O accumulation could be linked either to the phenotype of denitrification from
stored carbon, or to the particular denitrifying organism enriched in these systems. Several
attempts were made in the present study to investigate N2O accumulation from denitrification
in the presence of an external carbon source by adding acetate or propionate to one of two
parallel anoxic batch tests that were otherwise identical to the tests shown in Figure 2.
Addition of carbon led to a substantial increase in N2O accumulation, and it was found that
acetate and propionate was first accumulated intracellularly as PHA before being used as
electron donor in denitrification (data not shown). A true test of denitrification using
extracellular versus intracellular carbon could therefore not be performed.
95
14
140
(a)
100
8
80
6
60
4
40
2
20
0
0
NH4-N
NO2-N
NO3-N
N2O-N
N-NOx+N-N2O
PO4-P
10
-1
8
120
100
80
6
60
4
-1
12
N (mg l )
-1
10
P (mg l )
120
P (mg l )
-1
N (mg l )
12
40
(b)
2
20
0
0
0
0.5
1
time (h)
1.5
2
2.5
Figure 3. Measurements of NH4+, NO3-, NO2-, N2O, and PO43- during anoxic batch tests (500
ml reactor) with either nitrate (a) or nitrite (b) addition. It should be noted that some stripping
of N2O occurred during these batch tests.
Organisms responsible for denitrification
The role of PAOs versus GAOs in the denitrification process was specifically addressed by
investigating the ability of the SNDPR sludge to simultaneously reduce NOx- and take up
PO43- under strictly anoxic conditions. Batch tests showed that PO43- was not taken up by the
biomass under conditions where NOx- reduction occurred, using both nitrite and nitrate as
electron acceptors (Figure 3a and 3b). Although PAOs have the potential to denitrify (Kuba et
al., 1993; Meinhold et al., 1999), they did not appear to play a role in denitrification in this
reactor. As carbon was only available to organisms capable of anaerobic acetate uptake,
GAOs were therefore assumed to be responsible for denitrification.
The role of GAOs in denitrification was further supported by the developments in
denitrification performance and the abundance of PAOs and GAOs over the study period.
Figure 4 shows the abundance of Accumulibacter and Competibacter at five sampling times
during the 5 months period, the N removal efficiency and the P release to acetate uptake ratio
over the same period. While Accumulibacter increased from 46 to 70% of the total biomass,
Competibacter decreased from 18 to 9%. It is evident that Accumulibacter became more
abundant and more active, as the P release to acetate uptake ratio increased over the period to
reach 0.69 Pmol.Cmol-1 on day 140. This ratio reflects the fraction of carbon being taken by
PAOs relative to GAOs, and it was in this case higher than what was reported in previous
studies (0.5 Pmol.Cmol-1) of lab-scale reactors enriched in PAOs using acetate as the sole
carbon source (Smolders et al., 1994; Oehmen et al., 2004). The increase in PAO abundance
and PAO activity was accompanied by a decrease in the abundance of Competibacter, which
96
Accumulibacter
Competibacter
N removal
P release/VFA uptake
100
0.75
0.70
80
0.65
60
0.60
0.55
40
P/C (mol/mol)
% of all bacteria or % N removed
halved its population over the 5 months period. If Competibacter was primarily responsible
for denitrification, the decline in their population can explain the deterioration of the N
removal efficiency from 100% to 53% over the same period.
0.50
20
0.45
0
0.40
Jun 7
Jul 12
Aug 23 Sep 20
Oct 25
Figure 4. Abundance of Accumulibacter and Competibacter correlated with the N removal
efficiency and the carbon-uptake to P-release ratio over the 5 month period.
It should be noted that denitrification in the SNDPR process relies on the formation of anoxic
zones in the central part of the microbial aggregates caused by the mass transfer limitation of
oxygen. Lack of anoxic zones caused by a change in floc size or oxygen uptake rate of the
aggregates could therefore also cause the observed effect on denitrification. Measurement of
the oxygen penetration into aggregates during the June-July and the August-October periods
did, however, confirm that anoxic microzones were present in the aggregates throughout the
study (Figure 5).
-500
-400
-300
Depth (µ m)
-200
-100
liquid
0
granule
100
200
1h into aeration Oct.
300
1h into aeration Jun.
400
0
0.2
0.4
0.6
0.8
1
-1
Dissolve oxygen (mg l )
Figure 5. O2 profiles measured in aggregates at 1 h into the aeration period (i.e. after
nitrification activity ceased) in June and October. Profiles shown are averages (error
bars=S.E., n=3).
97
The apparent role of GAOs in denitrification has two implications to the SNDPR process.
Firstly, it compromises the carbon savings proposed to be obtained by SNDPR. Without
denitrification by PAOs, there is no true link between SND and EBPR, and the two processes
merely occur in the same sludge at the same time. Secondly, it poses the question of whether
there is a direct link between the enrichment of Competibacter and N2O accumulation from
denitrification. The enrichment of Competibacter in previous SNDPR and DGAO reactors
where N2O accumulation was observed (Zeng et al., 2003a; Zeng et al., 2003b) suggests this,
however, the question cannot be addressed fully without studying Competibacter in pure
culture.
Management of N2O accumulation in SNDPR
Whether N2O accumulation is linked to the particular organisms performing denitrification in
the SNDPR reactor or to the phenotype of PHA-driven denitrification, the high enrichment of
particular organisms in lab-scale reactors fed with synthetic wastewater containing a single
carbon source may exacerbate the effect of that organism or phenotype on N2O accumulation.
The abundance of Accumulibacter and Competibacter in full-scale treatment plants is usually
below 15% (Saunders et al., 2003), and the diversity of denitrifying bacteria in sludge treating
real wastewater may impact greatly on the potential for accumulating denitrification
intermediates. We therefore hypothesise that if denitrification is carried out by GAOs and
other denitrifiers simultaneously, the N2O accumulated by GAOs could be reduced by other
denitrifiers, provided that carbon is available to these cells.
The hypothesis was first tested by adding methanol during an anoxic batch test containing a
mixture of sludge from the SNDPR reactor, and a reactor treating domestic wastewater
amended with methanol (Figure 6a). Methanol was added to one of two parallel batch reactors
after 10 min of incubation with nitrate, and while the control reactor continued to accumulate
N2O, the N2O concentration immediately decreased in the test reactor after methanol addition.
As a negative control, the same experiment was repeated with SNDPR sludge alone,
demonstrating that methanol had no effect on N2O accumulation in this sludge (Figure 6b).
Hence, the denitrifying bacteria in the sludge mixed with the SNDPR sludge did not only
denitrify without accumulating N2O, they also acted as scavengers for the N2O accumulated
by GAOs in the SNDPR sludge.
98
1.5
add methanol
(a)
1
-1
N-N2O (mg l )
1.25
nitrate only
0.75
nitrate + methanol
0.5
0.25
0
0
10
20
30
40
50
-0.25
(b)
2.5
-1
N-N2O (mg l )
add methanol
2
1.5
1
nitrate only
nitrate + methanol
0.5
0
0
10
20
30
40
50
60
70
-0.5
time (min)
Figure 6. N2O profiles in the bulk liquid during anoxic batch tests in 14.75-ml sealed reactors
with and without methanol addition as carbon source using (a) a 2:1 mixture of SNDPR
sludge and denitrifying sludge, or (b) SNDPR sludge alone as a negative control. The initial
nitrate concentration at T=0 was 7.5 mgN.l-1 of NO3-.
The denitrifying sludge used in the experiment above was adapted to methanol, which may
have biased the results. The experiment was therefore repeated using raw, high-strength
wastewater as the carbon source and mixing the SNDPR sludge with sludge from a
denitrifying reactor which had not been adapted to a particular carbon compound. This batch
test was carried out in a 500 ml reactor, allowing sampling of NOx- during the incubation.
Addition of raw wastewater after 35 min of incubation with nitrate lead to an immediate
reduction and depletion of N2O (Figure 7). At the same time, the nitrate reduction rate
increased, underlining the capacity of the non-GAO denitrifiers to reduce nitrate to N2 while
simultaneously removing N2O accumulated by GAOs. The somewhat abrupt decrease in the
nitrate and N2O concentration upon addition of the raw wastewater was caused by the dilution
effect imposed by the added volume of liquid.
0.6
add high-strength WW
10
0.3
8
-1
N2O-N
NOx-N
0.4
6
4
0.2
N-NOx (mg l )
N-N2O (mg l-1)
0.5
2
0.1
0
0
0
20
40
60
80
time (min)
Figure 7. N2O and NOx- concentration in the bulk liquid during an anoxic test in a 500 ml
reactor. Raw high-strength wastewater was added at T=35 min.
99
The net production of N2O from denitrification is most likely linked to the organisms
(Competibacter and possibly other GAOs) responsible for denitrification in the present study,
while the enrichment of these organisms and loss of diversity amongst the denitrifying
microbial community enhanced the effect of these organisms on N2O accumulation in the
SNDPR reactor. This loss of diversity was primarily a result of feeding the reactor synthetic
wastewater containing only a single carbon source. Real wastewater contains a combination of
different carbon sources with different degradability, and because GAOs take up mainly
volatile fatty acids, many carbon compounds are likely to still be present when the aerobic
period sets in. It is therefore a realistic expectation that some carbon would be available for
denitrification by non-GAO organisms during the aerobic period. We therefore conclude that
N2O accumulation is unlikely to be an issue in a SNDPR reactor treating real wastewater.
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology Cooperative Research Centre
(EBCRC), Australia and the Danish National Science Research Council. The authors would
like to thank Xiuheng Wang for participating in some of the experimental studies.
REFERENCES
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
Andersen K., Kjaer T. and Revsbech N.P. (2001). An oxygen insensitive microsensor for
nitrous oxide. Sensors and Actuators B-Chemical. 81(1): 42-48.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Beline F., Martinez J., Marol C. and Guiraud G. (2001). Application of the N-15 technique to
determine the contributions of nitrification and denitrification to the flux of nitrous
oxide from aerated pig slurry. Water Research. 35(11): 2774-2778.
Brink C., Kroeze C. and Klimont Z. (2001). Ammonia abatement and its impact on emissions
of nitrous oxide and methane in Europeb - Part 1: method. Atmospheric Environment.
35(36): 6299-6312.
Comeau Y., Hall K., Hancock R. and Oldham W. (1986). Biochemical model for enhanced
biological phosphorus removal. Water Research. 20: 1511-1521.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
100
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Hanaki K., Hong Z. and Matsuo T. (1992). Production of nitrous oxide gas during
denitrification of wastewater. Water Science and Technology. 26(5-6): 1027-1036.
Itokawa H., Hanaki K. and Matsuo T. (2001). Nitrous oxide production in high-loading
biological nitrogen removal process under low COD/N ratio condition. Water
Research. 35(3): 657-664.
Kishida N., Kim J.H., Kimochi Y., Nishimura O., Sasaki H. and Sudo R. (2004). Effect of
C/N ratio on nitrous oxide emission from swine wastewater treatment process. Water
Science and Technology. 49(5-6): 359-365.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kuba T., Smolders G.J.F., van Loosdrecht M.C.M. and and Heijnen J.J. (1993). Biological
phosphorus removal from wastewater by anaerobic and anoxic sequencing batch
reactor. Water Science and Technology. 27: 241-252.
Meinhold J., Filipe C.D.M., Daigger G.T. and Isaacs S. (1999). Characterization of the
denitrifying fraction of phosphate accumulating organisms in biological phosphate
removal. Water Science and Technology. 39(1): 31-42.
Metcalf & Eddy (1991). Wastewater engineering: treatment, disposal and reuse. Singapore,
McGraw-Hill, Inc.
Meyer R.L., Saunders A.M., Zeng R.J.X., Keller J. and Blackall L.L. (2003). Microscale
structure and function of anaerobic-aerobic granules containing glycogen
accumulating organisms. FEMS Microbiology Ecology. 45(3): 253-261.
Meyer R.L., Zeng R.J.X., Giugliano V. and Blackall L.L. (2005). Challenges for simultaneous
nitrification, denitrification, and phosphorus removal in microbial aggregates: mass
transfer limitation and nitrous oxide production. FEMS Microbiology Ecology. 52(3):
329-338.
Noda N., Kaneko N., Milkami M., Kimochi Y., Tsuneda S., Hirata A., Mizuochi M. and
Inamori Y. (2003). Effects of SRT and DO on N2O reductase activity in an anoxicoxic activated sludge system. Water Science and Technology. 48(11-12): 363-370.
Oehmen A. (2004). The competition between polyphosphate accumulating organisms and
glycogen accumulating organisms in the enhanced biological phosphorus removal
process. PhD Thesis. University of Queensland, Brisbane, Australia.
Otte S., Grobben N.G., Robertson L.A., Jetten M.S.M. and Kuenen J.G. (1996). Nitrous oxide
production by Alcaligenes faecalis under transient and dynamic aerobic and anaerobic
conditions. Applied and Environmental Microbiology. 62(7): 2421-2426.
Pitman A.R., Venter S.L.V. and Nicholls H.A. (1983). Practical experience with biological
phosphorus removal plants in Johannesburg. Water Science and Technology. 15(3-4):
233-259.
Revsbech N.P. (1989). An oxygen microsensor with a guard cathode. Limnology and
Oceanography. 34(2): 474-478.
Saunders A.M., Oehmen A., Blackall L.L., Yuan Z. and Keller J. (2003). The effect of GAOs
(glycogen accumulating organisms) on anaerobic carbon requirements in full-scale
Australian EBPR (enhanced biological phosphorus removal) plants. Water Science
and Technology. 47(11): 37-43.
Smolders G., Vandermeij J., Vanloosdrecht M. and Heijnen J. (1994). Model of the anaerobic
metabolism of the biological phosphorus removal process - stoichiometry and pH
influence. Biotechnology and Bioengineering. 43(6): 461-470.
101
Wentzel M.C., Loewenthal R.E., Ekama G.A. and Marais G.V. (1988). Enhanced
polyphosphate organism cultures in activated-sludge systems .1. Enhanced culture
development. Water SA. 14(2): 81-92.
Zeng R., Yuan Z. and Keller J. (2003a). Enrichment of denitrifying glycogen-accumulating
organisms in anaerobic/anoxic activated sludge systems. Biotechnology and
Bioengineering. 81(4): 397-404.
Zeng R.J., Lemaire R., Yuan Z. and Keller J. (2003b). Simultaneous nitrification,
denitrification, and phosphorus removal in a lab-scale sequencing batch reactor.
Biotechnology and Bioengineering. 84(2): 170-178.
102
Appendix E
Microbial Distribution of Accumulibacter spp. and Competibacter spp. in
Aerobic Granules from a Lab-Scale Biological Nutrient Removal System
Romain Lemaire, Zhiguo Yuan, Linda L. Blackall and Gregory R. Crocetti
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane 4072 QLD, Australia.
ABSTRACT
Granular sludge for simultaneous nitrification, denitrification and phosphorus removal
(SNDPR) was generated and studied in a laboratory-scale sequencing batch reactor (SBR).
The SBR was monitored for 450 days during which the biomass was transformed from flocs
to granules, which persisted for the last 130 days of operation. Short sludge settling time was
employed to successfully generate the granules, with the 10th and 90th percentiles of diameter
being 0.7 and 1.6 mm, respectively. Good phosphorus removal and nitrification occurred
throughout the SBR operation but only when granules were generated was denitrification and
full nutrient removal complete. Fluorescence in situ hybridisation and oxygen microsensors
were used to study the granules at a microscale. Accumulibacter spp. (a polyphosphate
accumulating organism, PAO) and Competibacter spp. (a glycogen non-polyphosphate
accumulating organism, GAO) were the most abundant microbial community members
(together 74% of all Bacteria) and both are capable of denitrification. In the aerobic period of
the SBR operation, the oxygen penetrated 250 µm into the granules leaving large anoxic
zones in the centre part where denitrification can occur. In granules > 500 µm in diameter,
Accumulibacter spp. was dominant in the outermost 200 µm region of the granule while
Competibacter spp. dominated in the granule central zone. The stratification of these two
populations between the outer aerobic and inner anoxic part of the granule was highly
significant (P<0.003). We concluded that GAO Competibacter spp., and not the PAO
Accumulibacter spp., was responsible for denitrification in this SBR. This is undesirable for
SNDPR as savings in carbon demand cannot be fulfilled with P removal and denitrification
being achieved by different groups of bacteria.
Keywords: Aerobic granule, biological nutrient removal, GAO, microbial communities,
PAO, SNDPR.
103
INTRODUCTION
The removal of nutrients, mainly nitrogen (N) and phosphorus (P), from domestic and
industrial wastewater is required in order to prevent eutrophication of receiving water
systems. Biological nutrient removal (BNR) systems are the most cost effective way to reduce
the nutrient load from large volumes of wastewater. In conventional BNR systems, N is
removed via a two-stage process (Metcalf & Eddy, 1991) featuring aerobic nitrification and
anoxic denitrification. However, it has been observed that these two processes can occur
concurrently in a single-sludge, single-stage process under low dissolved oxygen conditions
called simultaneous nitrification and denitrification (SND) (Munch et al., 1996; Bertanza,
1997; Keller et al., 1997; Fuerhacker et al., 2000). SND relies on the formation of anoxic
zones in the central part of microbial aggregates caused by the mass transfer limitation of
oxygen. In the aerobic zone on the edge of the aggregate, autotrophic bacteria can nitrify
using oxygen, whereas in the anoxic zone in the centre of the aggregate, heterotrophic bacteria
can denitrify. Factors that affect oxygen mass transfer limitation such as bulk liquid oxygen
concentration, the aggregate size, and the specific activity of the microbial aggregates (oxygen
uptake rate per volume of biomass) (Pochana and Keller, 1999; Meyer et al., 2005) also affect
SND.
The removal of P from wastewater is typically achieved by either chemical precipitation or
through the biological process called enhanced biological phosphorus removal (EBPR). EBPR
is based on the ability of polyphosphate-accumulating organisms (PAOs) to take up P and
accumulate it intracellularly as polyphosphate when exposed to alternating anaerobic (O2 and
nitrite/nitrate (NOx-) absent) and aerobic conditions (Comeau et al., 1986; Wentzel et al.,
1988). Simultaneous NOx- and P removal was achieved in anaerobic-anoxic EBPR systems
using denitrifying PAOs (Kuba et al., 1993; Kerrn-Jespersen et al., 1994; Meinhold et al.,
1999). The use of denitrifying PAOs in BNR systems is highly beneficial in terms of lower
COD requirement (i.e. same carbon source used for both N and P removal) and reduced
aeration cost. EBPR and SND were amalgamated into one process called simultaneous
nitrification, denitrification and phosphorus removal (SNDPR) by Zeng et al. (2003). In the
anaerobic period of SNDPR, COD was taken up and P was released. During the following
aerobic period, there were concomitant P uptake and NH4+ removal through SND. No
accumulation of nitrite or nitrate was observed. However, incomplete nitrate removal has been
reported in lab-scale SNDPR bioreactors. Meyer et al. (2005) showed that if the
aerobic/anoxic zones in the microbial aggregates were not formed, incomplete coupling
between nitrification and denitrification could occur. Based on the process data of these
bioreactors, it was also reported that glycogen-accumulating organisms (GAOs), known to
compete with PAOs for carbon sources but without contributing to P removal, appear to be
primarily responsible for the denitrification process in many lab-scale SNDPR systems (Zeng
et al., 2003; Lemaire et al., 2006). The presence of GAOs in such systems reduces the
potential for SNDPR to be used by the wastewater treatment industry.
Recent research efforts exploring the formation and use of aerobic granular biomass for
nutrient removal in sequencing batch reactor (SBR) systems (Tay et al., 2002; Lin et al., 2003;
Liu et al., 2003a; Yang et al., 2003; Cassidy and Belia, 2005; de Kreuk et al., 2005), suggest
that granules could be beneficial for SNDPR. Compared to conventional flocs, granules are
relatively large, compact, dense microbial aggregates of different bacterial species with an
approximately spherical external appearance. The granular structure might positively
contribute to the oxygen mass transfer limitation that is important in facilitating
aerobic/anoxic zones required for SNDPR. Several studies investigated the structure and
distribution of bacteria in nitrifying biofilms (Schramm et al., 2000; Gieseke et al., 2002;
104
Gieseke et al., 2003; Lydmark et al., 2006) and anaerobic granules (Sekiguchi et al., 1999; Liu
et al., 2003b; Diaz et al., 2006) using molecular and microsensor techniques, but few
attempted to comprehensively describe the relationship between the oxygen mass transfer
limitation and the spatial organisation of the microbial community in aerobic granules (Ivanov
et al., 2005; Kishida et al., 2006).
This study reports on a lab-scale SBR operated for SNDPR with selection for granular
biomass. The microscale distribution, organization, and community composition of the
bacterial community in granules were studied using fluorescence in situ hybridisation (FISH)
and confocal laser scanning microscopy (CLSM). Oxygen profiles inside SNDPR granules
were determined by microsensors. The spatial distribution of PAOs and GAOs in granules
was correlated with the dissolved oxygen profiles in an attempt to demonstrate “ecological
positions” and roles for each population in this complex system.
RESULTS
Reactor performance
Figure 1 shows various process data collected from the SBR from day 600 (i.e. 9 months
before initiation of granule selection) until day 1000. The SBR effluent quality is depicted in
Figure 1a. The low concentration of PO43- and NH4+ in the effluent indicates that good P
removal and nitrification were consistently achieved in the SBR. However, incomplete
denitrification occurred occasionally as shown by the accumulation of NOx- (mostly nitrate) in
the effluent between day 700 and day 900. Before granulation started, the mixed liquor
suspended solid (MLSS) and volatile suspended solid (VSS) concentration in the SBR were
relatively stable with means of 4.2 g.l-1 (SD=0.53, n=32) and 2.4 g.l-1 (SD=0.36, n=25),
respectively, and a mean VSS:MLSS ratio of 0.59 (SD=0.04, n=25) (Figure 1b). During the
same period, the aggregates in the SBR were relatively small as indicated by the 10th, 50th and
90th percentile of the aggregates size distribution illustrated in Figure 1c.
The reduction of the SBR settling time after day 870 resulted in an immediate increase in the
suspended solids concentration in the effluent due to the washout of slowly settling aggregates
from the SBR (Figure 1b). After a slight decrease of the biomass concentration in the SBR,
the MLSS and VSS concentrations reached 9 g.l-1 and 6 g.l-1 respectively while the size of the
microbial aggregates (now called granules) increased sharply to stabilise between 0.7 and 1.6
mm (10th and 90th percentile respectively) (Figure 1b and 1c). During the granule selection
process, the nutrient levels in the SBR effluent stayed relatively low with only a slight
increase of the PO43- concentration. The denitrification improved rapidly and was complete by
day 900. Figure 2 shows the SBR cycle study carried out on day 970. The profiles of PO43-,
NH4+ and NOx- in the SBR bulk liquid depicted in Figure 2 illustrated typical SNDPR patterns
with simultaneous removal of N and P during the aerobic stage without significant
accumulation of NOx- in the bulk liquid.
105
Settling time reduced
Loading rate increased
15
30
NH4-N (mg/l)
NOx-N (mg/l)
PO4-P (mg/l)
(a)
12
25
-1
15
6
P mg l
N mg l
-1
20
9
10
3
5
(b)
10
8
-1
MLSS
VSS
VSS:MLSS
SS effluent
0.8
0.6
6
0.4
4
0.2
2
0
0
1.6
Aggregate size (mm)
VSS:MLSS ratio, SSeff (g l )
0
-1
MLSS and VSS (g l )
0
d (0.1)
d (0.5)
d (0.9)
(c)
1.4
1.2
1
0.8
0.6
0.4
0.2
0
600
650
700
750
800
850
900
950
1000
Days
Figure 1. Process data since SBR was started. (a) Nutrient concentration in the SBR effluent.
(b) MLSS and VSS concentration and their ratio in the SBR and the effluent suspended solids
concentration. (c) 10th, 50th and 90th percentile of the volumetric size distribution of the
microbial aggregates in the SBR. The vertical dotted-line indicates when the SBR settling
time was reduced to promote the formation of granules.
25
-1
N (mg l )
20
15
180
NOx-N
160
PO4-P
140
acetate
120
100
10
80
60
5
40
-1
200
NH4-N
-1
Aerobic
P (mg l ), acetate (mgCOD l )
220
Anaerobic
20
0
0
0
0.5
1
1.5
2
2.5
3
3.5
4
time (h)
Figure 2. N, P and Acetate profile during a cycle study performed on day 970 when granule
samples were taken to visualize microbial population distribution within the granule structure.
106
Oxygen profiles
After 100 days of implementing operational conditions to promote the formation of granules
(i.e. reducing the settling time and increasing the organic and nutrient loading rates, see
Experimental Procedures), the SBR biomass was mainly constituted of large (diameter > 500
µm) and dense granules (biomass density = 125 gSS.l-1 biomass, SD=9, n=5) with only few
loose floccular aggregates observed. To determine the oxygen penetration boundary into the
>500 µm granules, oxygen gradient measurements were determined on day 970. Figure 3
shows that the oxygen profiles at the start and at the end of the aeration period were very
similar with a boundary between the aerobic and anoxic part of the granule occurring at
approximately 250 µm from the granule surface.
800
600
Depth (µm)
400
200
Liquid
0
Granule
-200
End aeration
-400
Start aeration
-600
-800
0
0.5
1
1.5
2
-1
O2 concentration (mg l )
Figure 3. Oxygen profiles in granules at the start and the end of the aerobic period performed
on day 970. Profile shown are average (error bars=S.D., n=6).
Microbial diversity and distribution of PAO and GAO in granules
A preliminary investigation of the diversity of PAO and GAO populations present in
homogenised granular sludge samples was carried out by FISH with probes targeting
organisms previously shown to exhibit the PAO phenotype (Accumulibacter spp.,
Actinobacter-related spp.) or GAO phenotype (Competibacter spp. and Defluviicoccus spp.).
Accumulibacter spp. (48% of all bacteria, SD=18, n=24) and Competibacter spp. (26% of all
bacteria, SD=8, n=24) were the only PAO and GAO (together 74% of all bacteria, SD=16,
n=24) detected in these granules. The abundance of the nitrifier populations was also
estimated using FISH with probes targeting most ammonia oxidising bacteria (AOB) from the
Betaproteobacteria as well as some of the known nitrite oxidising bacteria (NOB) such as
Nitrospira spp. and Nitrobacter spp. These are the most common AOB and NOB in activated
sludge. Nitrobacter spp. was not detected while few Nitrospira spp. clusters were observed
(under the detection limit of the quantification method employed). AOB comprised 1.9% of
all bacteria (SD=0.3%, 3 replicates) in homogenised granular sludge samples. Given the fact
that nitrate rather than nitrite accumulated at the end of aerobic periods during days 700 – 900,
NOB were likely present in the sludge in that period, but declined during the development of
the granular sludge.
107
The mean distribution of Accumulibacter spp. and Competibacter spp. obtained from 24
granules (15 granules of >500 µm diameter and 9 granules of <500 µm diameter) and
expressed as the PAO (Accumulibacter spp.) to GAO (Competibacter spp.) ratio calculated
for each concentric 50 µm zone (see description in Experimental Procedures) is depicted in
Figure 4 along with the mean dissolved oxygen concentration at each zone measured in
several granules at the end of the aerobic period. For each 50 µm zone of the granules, the
PAO:GAO ratio correlated strongly (significant at the 0.01 level) with the dissolved oxygen
concentration (Pearson correlation=0.86). Accumulibacter spp. was dominant from 0-200 µm
from the granule surface (i.e. PAO:GAO ratio > 1) while Competibacter spp. dominated from
200 µm inwards (i.e. PAO:GAO ratio < 1). The statistical validity of this observation was
tested by comparing the difference of the mean PAO:GAO ratio in the outer part of the
granule (0-100 µm, likely aerobic) and in an inner part (250-350 µm, likely anoxic). The
stratification of the two populations between the aerobic and anoxic part of the granules was
highly significant (P<0.003). The FISH micrographs of median sections of granules shown in
Figure 5a and 5e illustrate this stratification. However, this trend was not always observed in
smaller granules with diameters less than 500 µm as shown in the FISH micrograph depicted
in Figure 5c.
3.5
PAO/GAO ratio
3.0
PAO/GAO ratio
O2 profile
-1
0.5
O2 concentration (mg l )
0.6
2.5
0.4
2.0
0.3
1.5
0.2
1.0
0.1
0.5
0.0
0.0
0
50
100
150
200
250
300
350
400
Depth (µm)
Figure 4. Average profile of the PAO:GAO ratio within 24 granules (error bar=95%CL) and
mean O2 profiles in granules at the end of the aerobic period performed on day 970 (error
bars=S.D., n=6).
108
Figure 5. Reconstructed CLSM images of FISH micrographs (a, c and e) and Nile-Blue-stain
micrographs (b, d and f) of entire granule sections. In (a), (c) and (e) Accumulibacter spp.
cells are cyan (overlay of blue PAOmix and green EUBmix), Competibacter spp. cells are
yellow (overlay of red GAOmix and green EUBmix) and other Bacteria are green (green
EUBmix). In (b), (d) and (f), overlays were of transmitted light images (black and white) and
Nile-Blue-stained PHAs in red. Subsequent granule sections (7 µm apart) of two different
granules are presented in (a)-(b) and (e)-(f). (c) and (d) are not images of the same granules.
The granules were sampled at the end of the anaerobic (a-b and c) and at the end of the
aerobic (d and e-f) periods. Scale bar = 100 µm.
109
Intracellular poly-hydroxyalkanoates (PHA) distribution in granules section
PAOs and GAOs take up acetate and convert it to PHA under anaerobic conditions and then
oxidise their stored PHA in either aerobic or anoxic (NOx- as electron acceptor) conditions.
Nile Blue staining of PHA was carried out to observe the location in granules of cells
containing PHA. Figure 5b shows that at the end of the anaerobic period, cells containing
PHA could be detected from the edge of the granule to the centre. This was consistently
observed in all the granules sampled at the end of the anaerobic period. At the end of the
aerobic period, PHA was still present, however only in the centre of the granule as shown in
Figure 5d and 5f. In all the granules sampled at the end of the aerobic period, the depletion
boundary of PHA correlated relatively well with the penetration boundary of oxygen, around
250 µm inside the granule.
DISCUSSION
Role of granules for SNDPR process reliability
The denitrification performance of the SBR operating for SNDPR was occasionally less
efficient leading to the accumulation of NO3- in the effluent (Figure 1a). Previous research
showed the causes to be a lack of anoxic zones in the inner part of the microbial aggregates
(flocs or granules) (Meyer et al., 2005) and a decrease in the denitrifier population (Lemaire et
al., 2006). We hypothesised that biomass comprised of a high proportion of larger granules
would be better at sustaining stable gradations of dissolved oxygen which are required for the
establishment of a robust SNDPR process. The strategy employed on day 870 proved
successful in generating biomass in mostly large and dense granules, which rapidly achieved
complete denitrification (Figure 1a) due to good coupling between nitrification and
denitrification. Additionally, NOx- was barely detectable in the bulk liquid during the aeration
period of the SBR cycle (Figure 2), indicating that SND was occurring. However, in order to
benefit fully from the advantages of SNDPR, it is essential that PAOs carry out most of the
denitrification and not GAOs as previously found (Zeng et al., 2003; Lemaire et al., 2006).
Microscale distribution of Accumulibacter spp. and Competibacter spp.
The 15 large granules (diameter > 500 µm) were comprised of Accumulibacter spp (the PAO
in the SBR) mostly in the outer part of the granule, and Competibacter spp. (the GAO in the
SBR) mostly in the inner zone (P<0.003). This PAO:GAO stratification was strongly
correlated with the dissolved oxygen concentration gradient (Figure 3, Pearson coefficient =
0.86). In the 9 small granules (diameter < 500 µm) studied, the central anoxic zone was likely
absent and there was no clear stratification of PAOs and GAOs. A totally random distribution
was observed in 4 of the 9 small granules (data not shown). Overall, our results are in contrast
to those of Kishida et al. (2006) who reported that Accumulibacter spp. were located in
granules past the point where dissolved oxygen was not measurable, Competibacter spp. were
mainly located in the aerobic parts of the granules. This discrepancy may be due to aerobic
granules being relatively heterogeneous and having many different local microenvironments
(Ivanov et al., 2005) including channels and voids which allow transport of nutrients from the
bulk liquid to the granule interior. To address the heterogeneity, a sufficient number of
granules should be studied to allow statistical analysis of the data. Therefore, true
comparisons with Kishida et al. (2006) are difficult due to the few granules they studied. We
110
analysed 24 granules of various shapes and sizes so as to provide a representative sample of
granules in order to establish a microscale distribution profile with statistical support.
The irregular spherical shape of the granules prevented the use of basic software tools to
automatically generate the different zones of microorganism distribution, as have been used in
biofilm studies (Schramm et al., 2000; Gieseke et al., 2003; Lydmark et al., 2006). Our
developed methods were highly reproducible and the PAO:GAO ratio highlighted the
population distribution. Errors that might have been introduced by drawing concentric zones
following the specific contour of each granule were statistically evaluated by replication and
the SD was less than 3% of the mean (data not shown).
Several biofilm studies have focused on the microscale distribution of nitrifiers in aerobic
biofilms (Okabe et al., 1999; Schramm et al., 2000; Gieseke et al., 2003). Lydmark et al.
(2006) suggested that the availability of different specific substrates (ammonium or nitrite)
was the main factor responsible for the stratification of ammonia oxidisers and nitrite
oxidisers. Accumulibacter spp. and Competibacter spp. use the same substrates (acetate)
under anaerobic conditions. The presence of PHA in Accumulibacter spp. and Competibacter
spp. cells located in the centre of large granules indicated that acetate diffused fully into
granules (Figure 5b). Phosphorus transformations distinguish Accumulibacter spp. from
Competibacter spp., but there are no in situ methods that suitably allow the study of inorganic
phosphate (Pi) diffusion into granules. Through experimental and modelling results,
Falkentoft et al. (2001) reported that Pi diffusion limitation could be significant in continuousflow EBPR biofilm systems with constant phosphate concentration in the bulk liquid of
28 mgP.l-1. The batch process used in our study produced a concentration of Pi in the bulk
liquid of up to 100 mgP.l-1 (Figure 2) at the start of the aerobic period which should provide
sufficient concentration gradient for a full penetration of Pi inside the granules. In contrast,
the relatively low concentration of DO (in comparison to acetate and Pi) in the bulk liquid
phase has been found to create zones in granules devoid of oxygen or with low oxygen
concentrations. We hypothesise that the DO gradient in granules was most likely responsible
for the stratification of Accumulibacter spp. and Competibacter spp. in SNDPR granules. This
could either be caused by different affinities of Accumulibacter spp. and Competibacter spp.
with respect to oxygen, or directly by the availability of oxygen in different zones of the
granules.
Role of microbial diversity in denitrification process
An important feature of SNDPR is that most of the carbon in the bulk liquid (i.e. acetate) is
fully taken up and stored intracellularly during the anaerobic period by specific organisms (i.e.
PAOs and GAOs). Thus, most microorganisms present in the biomass will have no access to
an energy source which would allow them to carry out nitrate or nitrite reduction during the
oxidative period (oxygen and/or NOx- present), with the exception of denitrifiers able to use
exudates or extracellular polymeric substances as carbon sources. However, these alternate
carbon sources are limited and also slowly biodegradable. Consequentially, they are unlikely
to contribute in any substantial fashion to denitrification in SNDPR. The determination of the
organisms (whether they are PAOs or GAOs) primarily responsible for denitrification in
SNDPR is important, because carbon savings are a major advantage of SNDPR compared to
conventional BNR systems. PAOs are desirable because they use the same carbon source for
both N and P removal. Instead, if GAOs are the main denitrifiers, then there is no link
between SND and EBPR and no saving of carbon.
Using the PAO:GAO ratio, Accumulibacter spp. dominated in the aerobic zones and
Competibacter spp. in the central anoxic zones of the studied granules. Nile Blue staining on
111
several granule sections demonstrated intracellular PHA inclusions in cells throughout the
granules when sampled at the end of the anaerobic period. After the aerobic period, only cells
in the central anoxic zone (mostly Competibacter spp.) still contained PHA. We suggest that
these cells only oxidised a portion of their stored PHA while denitrifying the NOx- produced.
However, since Nile Blue staining is not quantitative, we could not confirm this hypothesis.
We conclude that Competibacter spp. would be mostly responsible for denitrification in
SNDPR granules. Previous research also found Competibacter spp. to be the main
denitrifying population in lab-scale SNDPR systems (Zeng et al., 2003; Lemaire et al., 2006).
Nitrate rather than nitrite was the primary end product of nitrification in the process leading to
the formation of granular sludge. The genome sequence of Accumulibacter spp. (Martin et al.,
2006) does not contain respiratory nitrate reductase genes but does have the genes for
enzymes in the denitrification pathway from nitrite reduction onwards. Several studies
reported that Accumulibacter spp. could reduce nitrate to N2 under anoxic conditions (Kuba et
al., 1993; Meinhold et al., 1999; Shoji et al., 2003). It is possible that strains of
Accumulibacter spp. other than those sequenced by Martin et al. (2006) may indeed be able to
reduce nitrate to N2. Since complex mixed cultures were studied, organisms other than
Accumulibacter spp. might have carried out nitrate reduction to nitrite. Accumulibacter spp.
then would have access to this formed nitrite to survive in anoxia. If Competibacter spp. has
the capacity to reduce nitrate, this organism would have a competitive edge over
Accumulibacter spp. in anoxic conditions which may explain the spatial organisation of the
organisms in our granules. The abundance of Competibacter spp. in the community shows this
organism has limited competitors for nitrate. A more complex community would provide this
competition and likely limit the dominance of Competibacter spp. in the centre part of the
granules. Therefore, maintaining a threshold of microbial biodiversity above that in the
present SBR with a synthetic feed by using more complex wastewater might facilitate
efficient SNDPR by allowing PAOs to carry out denitrification and P removal. Alternatively,
the long term inhibition or elimination of the nitrite oxidising bacterial population in a
SNDPR system would prevent nitrite being oxidised to nitrate allowing PAOs to perform
denitrification of nitrite to N2 without relying on GAOs or any other denitrifiers to carry out
the first denitrification step. The process data and FISH quantification results indicate that
NOB were indeed gradually eliminated during the granulation process. However, the impact
of this change on the interactions between PAOs and GAOs and on their spatial locations in
granules was not assessed in this study. Further investigations are warranted.
EXPERIMENTAL PROCEDURES
Reactor Setup and Operation
The biomass was enriched in a 5 l SBR operated at room temperature (20-22°C) and seeded
with sludge from the Caboolture Sewage Treatment Plant, Queensland, Australia. The SBR
was initially operated with a cycle time of 6 h, consisting of a 90 min anaerobic period,
followed by 220 min aeration, 40 min settling, and 10 min decanting periods (Lemaire et al.,
2006). After 870 days of operation, the SBR cycle was modified to promote the formation of
granules. The cycle time was shortened to 4 h and the time allowed for settling was gradually
reduced to reach 5 min by day 900. From day 900 onward, the new 4 h cycle operation
consisted of 55 min anaerobic period, followed by 170 min aeration, 5 min settling, and 10
min decant. Every cycle, three litres of synthetic wastewater was pumped into the reactor in
the first 7 min of the anaerobic period, and 3 l supernatant was removed after settling,
resulting in a hydraulic retention time (HRT) of 6.7 h. The synthetic wastewater was prepared
112
as described in Zeng et al. (2003) with 350 mg.l-1 COD as acetate, 35 mg.l-1 NH4-N and
23 mg.l-1 PO4-P. To provide adequate shear force in the SBR during the aerobic period, air
was provided at a flow of 5 l.min-1 throughout an air diffuser producing coarse bubbles
resulting in an upflow air velocity of 0.9 cm.s-1. Dissolved oxygen was kept between 1.3-1.7
mg.l-1 using an on/off control system. The pH in the system was recorded but not controlled,
and fluctuated between 7.0 and 7.5.
To monitor the reactor performance, the SBR effluent was sampled daily for analysis of PO43-,
NH4+ and NOx- using a Lachat QuikChem8000 Flow Injection Analyser (Lachat Instrument,
Milwaukee). Cycle studies were performed once a week during which liquid samples were
taken and filtered every 20-30 min for analysis of acetate, PO43-, NH4+ and NOx-. Acetate was
measured on a Perkin-Elmer gas chromatograph with column DB-FFAP 15 m x 0.53 mm x
1.0 µm (length x ID x film) at 140°C while the injector and flame ionisation detector were
operated at 220°C and 250°C, respectively. The reactor MLSS, VSS and effluent suspended
solids concentrations were measured weekly according to standard methods (APHA, 1995).
Oxygen Microsensors
The gradient of oxygen in granules was measured with oxygen microsensors (tip diameter
<10 µm), which were constructed as described by Revsbech et al. (1989). Granules were
sampled from the SBR on day 970 at the start of the aerobic period when NH4+ and PO43were present at high concentrations and at the end of the aerobic period, at which time NH4+
and PO43- were usually depleted. Granules with diameter > 500 µm were transferred to a flowcell with an upward flow as described in Meyer et al. (2003). Six replicate oxygen profiles
were then measured over the following hour and averaged. The sensor was moved stepwise
into the granules from above, and movement of the sensor and data acquisition was obtained
with the software Profix (Unisense A/S, Aarhus, Denmark). The composition of medium and
the dissolved oxygen concentration in the flow-cell was identical to that in the SBR at either
the start or the end of the aerobic period. The oxygen concentration was adjusted within the
nutrient medium by controlling the air/nitrogen ratio in the gas inlet.
Biomass Composition and Structure
The granule size distribution was determined with a Malvern laser light scattering Mastersizer
2000 series instrument (Malvern Instruments, Worcestershire, UK) on granules in 30 ml of
well mixed granular sludge from the SBR at the end of the aeration period. The granule
density, defined as the quantity of dry mass per biomass volume, was determined using the
procedure described below. The biomass volume was measured using the blue dextran method
adapted from Di Iaconi et al. (2004). Dextran blue is not absorbed by biomass. Briefly, 5 ml
granular sludge was gently mixed with 5 ml of dextran blue solution (1 g.l-1), the mixture was
centrifuged for 3 minutes at 10,000 rpm and the absorbance of the supernatant at 620 nm was
measured. The same procedure was carried out with distilled water (blank).
Biovolume (ml) = 5 x (1 – Blank A620/Sample A620)
The biomass density (mg.l-1) was calculated by dividing the dry weight of the biomass in the
10 ml suspension by the volume that the granules occupied in the initial 5 ml sample
determined above. The reported value was determined from the mean of five replicates.
Granule samples for FISH and chemical staining were fixed in 3% paraformaldehyde
(Amann, 1995) on day 970. The fixed granules were embedded in optimum cutting
temperature (OCT) compound (TissueTek, Sakura, USA) for cryosectioning as previously
described (Meyer et al., 2003) or they were homogenised in 1.5 ml tube using a pestle (Astral
113
Scientific, Australia). Embedded granules were then frozen and sectioned into 10 µm thick
slices using a cryotome operated at -20°C (Kryo 1720, Leitz, Germany). The granule sections
were collected on SuperFrost Plus microscope slides (Menzel-Glaser, Germany). Finally, the
slides were dehydrated by sequential immersion for 3 min in 50%, 80% and 98% ethanol,
followed by air-drying.
FISH was then carried out on the granule sections (PAO and GAO) or on homogenised
granule samples (nitrifiers) as described by Amann (1995). To ensure that only the median
part of each cryosectioned granule was visualised by FISH, microscopic slides with the largest
sections of subsequent granule sections (as determined by eye) were first selected for each
cryosectioned granule. Oligonucleotide probes used in this study were the combination of
EUB338 i-iii (EUBmix) for the detection of all bacteria (Daims et al., 1999), the combination
of PAO462, PAO651 and PAO846 (PAOmix) for Accumulibacter spp. (Crocetti et al., 2000),
the probe combination (GAOmix) of GAOQ989 (Crocetti et al., 2002) and GB_G2 (Kong et
al., 2002) for Competibacter spp, NIT3 (Wagner et al., 1996) for Nitrobacter spp., NSPA662
(Daims et al., 2001) for Nitrospira spp. and NSO1225 (Mobarry et al., 1996) for most of the
ammonia oxidizing bacteria (AOB) from the Betaproteobacteria. Additionally, probes for
proposed Actinobacteria PAOs (Kong et al., 2005) and Defluviicoccus spp. GAOs (Meyer et
al., 2006) were used. Fluorescently labelled oligonucleotides were purchased from Thermo
(Ulm, Germany) with fluorescein isothiocyanate (FITC) or one of the sulfoindocyanine dyes
indocarbocyanine (Cy3) or indodicarbocyanine (Cy5).
Hybridised granule sections were visualised with a Zeiss LSM 510 (Carl Zeiss, Jena,
Germany) CLSM using an argon laser for FITC excitation (488 nm), a helium neon laser for
Cy3 (543 nm) and a red diode laser for Cy5 (633 nm) fitted with 515-565 nm BP, 590 nm LP
and 660-710 nm BP emission filters, respectively. To obtain images of entire granule sections,
between 10 and 50 overlapping, consecutive images of 1024x1024 pixels were collected
(depending on the size of the granule) using a Zeiss Neofluar 40x/1.3 oil objective. The final
composite image of the granule section was then reconstructed from all the single images
collected using Adobe Photoshop 7.0 (Adobe Systems, USA). For preliminary investigation
of the microbial composition and for the quantification of nitrifier populations in
homogenised granule samples, images were taken using a Zeiss Apochromat 63x/1.4 oil
objective.
Consecutive granule sections following those used for FISH were stained with Nile Blue A
(Ostle and Holt, 1982) and visualised on the CLSM using the Cy3 settings (above) to
determine cells containing intracellular PHA. Images of sections of 15 granules sampled from
the SBR at the end of the anaerobic period and 15 from the end of the aerobic period were
reconstructed.
Image analysis and statistical analysis
Using the brush tool in Adobe Photoshop 7.0, sequential 50 µm wide concentric zones,
following the external contour of each granule, were manually drawn. The creation of these
zones was continued stepwise towards the centre of the granule, following the contour of the
outer boundary from each previous zone, until the centre of the granule was reached. The
relative abundance of each microbial population (Accumulibacter spp. (PAOmix-targeted) or
Competibacter spp. (GAOmix-targeted)) was determined in each concentric 50 µm granule
zone of 24 individual granules with diameters ranging from 400 to 1000 µm using the pixel
Measure/Count tool in Image Pro 4.0 (Media Cybernetics, USA). Briefly, the area of pixels
contributed by PAOmix or GAOmix probes above a manually determined threshold was
divided by the area of pixels contributed by the EUBmix probes (also applying threshold) for
114
each granule zone. Individual pixels counted as ‘positive’ PAOmix or GAOmix were also
required to have a ‘positive’ pixel signal from EUBmix, above given thresholds. The
population distribution of Accumulibacter spp. and Competibacter spp. within each granule
was expressed as the relative abundance of Accumulibacter spp. divided by that of
Competibacter spp. (referred as the PAO:GAO ratio). Using the ratio data, a mean distribution
was calculated. The statistical validity of the differences between the PAO:GAO ratio in the
outer layer of the granule (0-100 µm, likely aerobic) and the PAO:GAO ratio in an inner layer
of the granule (250-350 µm, likely anaerobic) was analysed using a Student's paired t-Test,
with a two-tailed distribution. The correlation between the PAO:GAO ratio and the oxygen
profile in the granule was established using the Pearson correlation coefficient.
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology CRC, a Cooperative Research
Centre established and funded by the Australian Government together with industry and
university partners.
REFERENCES
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Bertanza G. (1997). Simultaneous nitrification-denitrification process in extended aeration
plants: Pilot and real scale experiences. Water Science and Technology. 35(6): 53.
Cassidy D.P. and Belia E. (2005). Nitrogen and phosphorus removal from an abattoir
wastewater in a SBR with aerobic granular sludge. Water Research. 39(19): 48174823.
Comeau Y., Hall K., Hancock R. and Oldham W. (1986). Biochemical model for enhanced
biological phosphorus removal. Water Research. 20: 1511-1521.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
115
de Kreuk M., Heijnen J.J. and van Loosdrecht M.C.M. (2005). Simultaneous COD, nitrogen,
and phosphate removal by aerobic granular sludge. Biotechnology and Bioengineering.
90(6): 761-769.
Di Iaconi C., Ramadori R., Lopez A. and Passino R. (2004). Preliminary biomass
characterization in a sequencing batch biofilm reactor. Annali di Chimica. 94(12): 889898.
Diaz E.E., Stams A.J.A., Amils R. and Sanz J.L. (2006). Phenotypic properties and microbial
diversity of methanogenic granules from a full-scale upflow anaerobic sludge bed
reactor treating brewery wastewater. Applied and Environmental Microbiology. 72(7):
4942-4949.
Falkentoft C.M., Arnz P., Henze M., Mosbaek H., Muller E., Wilderer P.A. and Harremoes P.
(2001). Possible complication regarding phosphorus removal with a continuous flow
biofilm system: Diffusion limitation. Biotechnology and Bioengineering. 76(1): 77-85.
Fuerhacker M., Bauer H., Ellinger R., Sree U., Schmid H., Zibuschka F. and Puxbaum H.
(2000). Approach for a novel control strategy for simultaneous
nitrification/denitrification in activated sludge reactors. Water Research. 34(9): 24992506.
Gieseke A., Arnz P., Amann R. and Schramm A. (2002). Simultaneous P and N removal in a
sequencing batch biofilm reactor: insights from reactor- and microscale investigations.
Water Research. 36(2): 501-509.
Gieseke A., Bjerrum L., Wagner M. and Amann R. (2003). Structure and activity of multiple
nitrifying bacterial populations co-existing in a biofilm. Environmental Microbiology.
5(5): 355-369.
Ivanov V., Tay S.T.L., Liu Q.S., Wang X.H., Wang Z.W. and Tay J.H. (2005). Formation and
structure of granulated microbial aggregates used in aerobic wastewater treatment.
Water Science and Technology. 52(7): 13-19.
Keller J., Subramaniam K., Gosswein J. and Greenfield P.F. (1997). Nutrient removal from
industrial wastewater using single tank sequencing batch reactors. Water Science and
Technology. 35(6): 137-144.
Kerrn-Jespersen J.P., Henze M. and Strube R. (1994). Biological phosphorus release and
uptake under alternating anaerobic and anoxic conditions in a fixed-film reactor.
Water Research. 28(5): 1253-1255.
Kishida N., Kim J., Tsuneda S. and Sudo R. (2006). Anaerobic/oxic/anoxic granular sludge
process as an effective nutrient removal process utilizing denitrifying polyphosphateaccumulating organisms. Water Research. 40(12): 2303-2310.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kong Y.H., Nielsen J.L. and Nielsen P.H. (2005). Identity and ecophysiology of uncultured
actinobacterial polyphosphate-accumulating organisms in full-scale enhanced
biological phosphorus removal plants. Applied and Environmental Microbiology.
71(7): 4076-4085.
Kuba T., Smolders G.J.F., van Loosdrecht M.C.M. and and Heijnen J.J. (1993). Biological
phosphorus removal from wastewater by anaerobic and anoxic sequencing batch
reactor. Water Science and Technology. 27: 241-252.
Lemaire R., Meyer R., Taske A., Crocetti G.R., Keller J. and Yuan Z.G. (2006). Identifying
causes for N2O accumulation in a lab-scale sequencing batch reactor performing
simultaneous nitrification, denitrification and phosphorus removal. Journal of
Biotechnology. 122(1): 62-72.
116
Lin Y.M., Liu Y. and Tay J.H. (2003). Development and characteristics of phosphorusaccumulating microbial granules in sequencing batch reactors. Applied Microbiology
and Biotechnology. 62(4): 430-435.
Liu Y., Xu H.L., Yang S.F. and Tay J.H. (2003a). Mechanisms and models for anaerobic
granulation in upflow anaerobic sludge blanket reactor. Water Research. 37(3): 661673.
Liu Y., Yang S.F. and Tay J.H. (2003b). Elemental compositions and characteristics of
aerobic granules cultivated at different substrate N/C ratios. Applied Microbiology and
Biotechnology. 61(5-6): 556-561.
Lydmark P., Lind M., Sorensson F. and Hermansson M. (2006). Vertical distribution of
nitrifying populations in bacterial biofilms from a full-scale nitrifying trickling filter.
Environmental Microbiology. 8(11): 2036-2049.
Martin H.G., Ivanova N., Kunin V., Warnecke F., Barry K.W., McHardy A.C., Yeates C., He
S.M., Salamov A.A., Szeto E., Dalin E., Putnam N.H., Shapiro H.J., Pangilinan J.L.,
Rigoutsos I., Kyrpides N.C., Blackall L.L., McMahon K.D. and Hugenholtz P. (2006).
Metagenomic analysis of two enhanced biological phosphorus removal (EBPR) sludge
communities. Nature Biotechnology. 24(10): 1263-1269.
Meinhold J., Filipe C.D.M., Daigger G.T. and Isaacs S. (1999). Characterization of the
denitrifying fraction of phosphate accumulating organisms in biological phosphate
removal. Water Science and Technology. 39(1): 31-42.
Metcalf & Eddy (1991). Wastewater engineering: treatment, disposal and reuse. Singapore,
McGraw-Hill, Inc.
Meyer R.L., Saunders A.M. and Blackall L.L. (2006). Putative glycogen-accumulating
organisms belonging to the Alphaproteobacteria identified through rRNA-based stable
isotope probing. Microbiology-Sgm. 152: 419-429.
Meyer R.L., Saunders A.M., Zeng R.J.X., Keller J. and Blackall L.L. (2003). Microscale
structure and function of anaerobic-aerobic granules containing glycogen
accumulating organisms. FEMS Microbiology Ecology. 45(3): 253-261.
Meyer R.L., Zeng R.J.X., Giugliano V. and Blackall L.L. (2005). Challenges for simultaneous
nitrification, denitrification, and phosphorus removal in microbial aggregates: mass
transfer limitation and nitrous oxide production. FEMS Microbiology Ecology. 52(3):
329-338.
Mobarry B.K., Wagner M., Urbain V., Rittmann B.E. and Stahl D.A. (1996). Phylogenetic
probes for analyzing abundance and spatial organization of nitrifying bacteria. Applied
and Environmental Microbiology. 62(6): 2156-2162.
Munch E.V., Lant P. and Keller J. (1996). Simultaneous nitrification and denitrification in
bench-scale sequencing batch reactors. Water Research. 30(2): 277-284.
Okabe S., Satoh H. and Watanabe Y. (1999). In situ analysis of nitrifying biofilms as
determined by in situ hybridization and the use of microelectrodes. Applied and
Environmental Microbiology. 65(7): 3182-3191.
Ostle A.G. and Holt J.G. (1982). Nile blue-A as a fluorescent stain for poly-betahydroxybutyrate. Applied and Environmental Microbiology. 44: 238-241.
Pochana K. and Keller J. (1999). Study of factors affecting simultaneous nitrification and
denitrification (SND). Water Science and Technology. 39(6): 61-68.
Revsbech N.P. (1989). An oxygen microsensor with a guard cathode. Limnology and
Oceanography. 34(2): 474-478.
Schramm A., De Beer D., Gieseke A. and Amann R. (2000). Microenvironments and
distribution of nitrifying bacteria in a membrane-bound biofilm. Environmental
Microbiology. 2(6): 680-686.
117
Sekiguchi Y., Kamagata Y., Nakamura K., Ohashi A. and Harada H. (1999). Fluorescence in
situ hybridization using 16S rRNA-targeted oligonucleotides reveals localization of
methanogens and selected uncultured bacteria in mesophilic and thermophilic sludge
granules. Applied and Environmental Microbiology. 65(3): 1280-1288.
Shoji T., Satoh H. and Mino T. (2003). Quantitative estimation of the role of denitrifying
phosphate accumulating organisms in nutrient removal. Water Science and
Technology. 47(11): 23-29.
Tay J.H., Liu Q.S. and Liu Y. (2002). Characteristics of aerobic granules grown on glucose
and acetate in sequential aerobic sludge blanket reactors. Environmental Technology.
23(8): 931-936.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Wentzel M.C., Loewenthal R.E., Ekama G.A. and Marais G.V. (1988). Enhanced
polyphosphate organism cultures in activated-sludge systems .1. Enhanced culture
development. Water SA. 14(2): 81-92.
Yang S.F., Tay J.H. and Liu Y. (2003). A novel granular sludge sequencing batch reactor for
removal of organic and nitrogen from wastewater. Journal of Biotechnology. 106(1):
77-86.
Zeng R.J., Lemaire R., Yuan Z. and Keller J. (2003). Simultaneous nitrification,
denitrification, and phosphorus removal in a lab-scale sequencing batch reactor.
Biotechnology and Bioengineering. 84(2): 170-178.
118
Appendix F
Simultaneous Nitrification, Denitrification and Phosphorus Removal from High
Nutrient Containing Industrial Wastewater using Granular Sludge
Gulsum Yilmaz, Romain Lemaire, Jurg Keller and Zhiguo Yuan
Advanced Water Management Centre (AWMC), The University of Queensland, St Lucia,
Brisbane QLD 4072, Australia.
ABSTRACT
The biological removal of nitrogen and phosphorus from nutrient-rich abattoir wastewater
using granular sludge has been investigated. A lab-scale sequencing batch reactor, seeded
with granular sludge developed using synthetic wastewater, was operated for 13 months under
alternating anaerobic and aerobic conditions. It is demonstrated that the granules could be
sustained and indeed further developed with the use of abattoir wastewater. The organic,
nitrogen and phosphorus loading rates applied were 2.7 gCOD.l-1.d-1, 0.43 gN.l-1.d-1 and 0.06
gP.l-1.d-1, respectively. The removal efficiency of soluble COD, soluble nitrogen and soluble
phosphorus were 85%, 93% and 89%, respectively. However, the high suspended solids in the
effluent limited the overall removal efficiency to 68%, 86% and 74% for total COD, TN and
TP, respectively. This good nutrient removal was achieved through the process known as
simultaneous nitrification, denitrification and phosphorus removal, likely facilitated by the
presence of large anoxic zones in the centre of the granules. The removal of nitrogen was
likely via nitrite optimising the use of the limited COD available in the wastewater.
Accumulibacter spp. was found to be responsible for most of the denitrification, further
reducing the COD requirement for nitrogen and phosphorus removal. Mineral precipitation
was evaluated and was not found to significantly contribute to the overall nutrient removal. It
is also shown that the minimum HRT in a granular sludge system is not governed by the
sludge settleability, as is the case with floccular sludge systems, but likely by the limitations
associated with the transfer of substrates in granules.
Keywords: abattoir wastewater, BNR, granular sludge, HRT, SBR, SNDPR.
INTRODUCTION
The meat processing industry requires large quantities of water, much of which is discharged
as wastewater containing high levels of COD and nutrients such as nitrogen (N) and
phosphorus (P). Over the past two decades, biological COD and N removal from abattoir
wastewater has received much greater attention than has the biological P removal. Reliable
biological COD and nitrogen removal systems have been successfully developed and applied
for abattoir wastewater treatment using continuous activated sludge systems (Beccari et al.,
1984; Frose and Kayser, 1985; Willers et al., 1993). However, P removal continues to be
119
achieved primarily through chemical precipitation, despite biological P removal being a much
cheaper and more environmentally sustainable option.
There are two main challenges for biological phosphorus removal from abattoir wastewater.
First of all, the wastewater contains a high level of ammonia and organic nitrogen, the
oxidation of which results in a high level of nitrate accumulation. Nitrate accumulation has
proved to be an obstacle to the development of a stable and reliable enhanced biological
phosphorus removal (EBPR) process (Comeau et al., 1986; Furumai et al., 1999; Pitman et al.,
1983). Secondly, abattoir wastewater contains substantial amounts of fat, oil and grease
(FOG) which negatively impact on sludge settleability. As a solution, the raw abattoir
wastewater is often pre-treated in anaerobic ponds to reduce the FOG content, which also
removes a large fraction of COD, resulting in COD limitations (particularly Volatile Fatty
Acids – VFAs) for N and P removal (Keller et al., 1997).
Simultaneous nitrification and denitrification (SND) potentially offers a solution to the first
problem as it prevents accumulation of NOx- (NO3- + NO2-) in the system. SND relies upon
the formation of anoxic zones in the central part of microbial aggregates caused by the mass
transfer limitation of oxygen (Bertanza, 1997; Fuerhacker et al., 2000; Keller et al., 1997;
Munch et al., 1996). The NOx- formed in the outer aerobic layer of the aggregates due to
nitrification can be reduced in the central anoxic zone. The accumulation of NOx- in the bulk
liquid phase is therefore minimised.
To address the second challenge, optimised use of COD for N and P removal could be
achieved by combining EBPR with SND. Polyphosphate-accumulating organisms (PAOs)
have the ability to simultaneously reduce NOx- and take up P using the same carbon source
(Kerrn-Jespersen et al., 1994; Kuba et al., 1993; Meinhold et al., 1999). Indeed, the process of
simultaneous nitrification and denitrification and phosphorus removal (SNDPR) has already
been demonstrated in lab-scale SBRs treating mainly synthetic wastewater (Zeng et al.,
2003a). Such a process requires alternating anaerobic and aerobic conditions. In the aerobic
period, the conversion of ammonia to gaseous nitrogen products and P uptake were achieved
concomitantly. However, unstable nitrate/nitrite removal has been reported in lab-scale
SNDPR bioreactors using floccular biomass. Meyer et al. (2005) showed that incomplete
coupling between nitrification and denitrification would occur if the aerobic/anoxic zones in
the microbial aggregates were not formed, leading to the accumulation of NOx-.
Recent research efforts exploring the formation and use of aerobic granular biomass in
sequencing batch reactor (SBR) systems (de Kreuk et al., 2005; Mosquera-Corral et al., 2005),
suggest that granules could indeed be beneficial for SNDPR. While granules were firstly
reported in an up-flow anaerobic sludge blanket (UASB) bioreactor over two decades ago
(Lettinga et al., 1980), they have been recently investigated in aerobic SBR systems for
nutrient removal (Beun et al., 1999; Cassidy and Belia, 2005; de Kreuk et al., 2005; Lin et al.,
2003; Peng et al., 1999; Yang et al., 2003). Compared to conventional flocs, granules are
relatively large, compact, dense microbial aggregates of different bacterial species with an
approximately spherical external appearance. The size and the dense, compact structure of
granules are expected to positively contribute to the oxygen mass transfer limitation required
for reliable SNDPR systems. In addition, the excellent settleability of granular sludge allows
for more biomass to be maintained in a relatively small reactor volume, enhancing the ability
of the reactor to withstand high loading rates. This is of great interest for the treatment of
nutrient-rich industrial wastewater.
The work presented investigates the feasibility of using granular sludge to sustain a stable and
robust SNDPR system allowing high-level of nutrient removal from nutrient-rich industrial
wastewater such as abattoir effluents. A lab-scale SBR, seeded with granular sludge
120
developed with synthetic wastewater, was operated under alternating anaerobic and aerobic
conditions. The feed was changed gradually from synthetic wastewater to real abattoir
wastewater. The size distribution of the granules and the long term performance of the reactor
in terms of COD, nitrogen and phosphorus removal were monitored. The effect of chemical
precipitation on the overall nutrient removal performance and the impact of the SBR cycle
time on the reactor nitrification capability were also investigated.
MATERIAL AND METHODS
Abattoir wastewater
The wastewater used in this study was from a local abattoir in Queensland, Australia. At this
site, the raw effluent passes through four parallel anaerobic ponds before being treated in a
SBR for biological COD and N removal. Anaerobic pond effluent from the abattoir was
collected on a weekly basis and stored at 4˚C. The characteristics of the anaerobic pond
effluent are detailed in Table 1. The change in the wastewater composition during the one
week storage period is also shown in Table 1.
Table 1. Characteristics of the anaerobic pond effluent used in this study. The intervals
represent the mid-95% range.
Parameter
TCOD (mg.l-1)
SCOD (mg.l-1)
VFA* (mgCOD.l-1)
TN (mg.l-1)
NH4-N (mg.l-1)
TP (mg.l-1)
PO4-P (mg.l-1)
TSS (mg.l-1)
*
Containing acetate and propionate only
Anaerobic
Variation after one
pond effluent
week storage at 4°C
-12%
600-783
+5%
265-384
-15%
58-116
N.A.
225-277
+2%
215-240
N.A.
35-42
-5%
32-38
N.A.
185-217
N.A.: Not Analysed
SBR operation
The SBR had a working volume of 5 l and was operated in a temperature-controlled room
(18-22°C). The SBR had an height:diameter ratio of 5.5 and its mixing was carried out via a
combination of magnetic stirring (200 rpm) and intermittent sparging of either nitrogen gas
(anaerobic/anoxic periods) or oxygen (aerobic period). The reactor was seeded with aerobic
granules obtained from another lab-scale SBR fed with synthetic wastewater (Lemaire et al.,
in press). During the first 40 days (i.e. start-up period), the SBR was fed only with synthetic
wastewater. From day 40 to 133 (i.e. transition period), the SBR was fed with a mixture of
abattoir and synthetic wastewater. The percentage of abattoir wastewater in the influent was
gradually increased from 0 to 100% to allow the granular biomass to acclimate to the high
levels of nutrients present in the anaerobically pre-treated abattoir wastewater. During this
transition period, the average COD, ammonia and phosphate concentrations in the SBR feed
increased from 350 mgCOD.l-1, 40 mgN.l-1 and 20 mgP.l-1 to 1420 mgCOD.l-1, 240 mgN.l-1
and 40 mgP.l-1, respectively. From day 133 onwards the reactor was fed only with
121
anaerobically pre-treated abattoir wastewater. Additional acetate had to be supplemented to
the SBR as the amount of easily biodegradable COD (i.e. VFAs) available in the particular
abattoir wastewater used in this study was insufficient (due to highly effective operation of the
anaerobic pond) for the removal of the high levels of N and P present. The N, P and VFA
concentrations in the SBR feed during the entire period of study is depicted in Figure 1a.
Detailed operations of the SBR over the 13 months period are summarised in Table 2. In each
cycle, 3 l of anaerobically pre-treated abattoir wastewater supplemented with additional
acetate was pumped into the bottom of the reactor without mixing. The feeding period was
followed by a mixed anaerobic/anoxic period (Table 2) with intermittent sparging of nitrogen
gas (10 seconds every minute) at 2.5 l.min-1 to ensure good mixing and shear force in the
reactor. In the first two phases, the dissolved oxygen (DO) concentration during the
subsequent aerobic period was not controlled, and constant aeration was provided. The DO
level in the SBR was stable around 2.0-2.5 mgO2.l-1 during most of the aerobic period,
reaching 6 mgO2.l-1 only when ammonia and phosphate were completely depleted, typically in
the last 15 minutes of the aerobic period. After day 290, the DO level was controlled between
3.0-3.5 mgO2.l-1 during the entire aerobic period using an on/off aeration control system. To
provide adequate shear force during the aerobic period an air diffuser producing coarse
bubbles was used for aeration and the air flow rate was maintained at 4 l.min-1 (resulting in an
upflow superficial air velocity of 0.6 cm.s-1). On day 333 (phase IV), a post-anoxic period was
introduced after the aerobic period to further enhance denitrification. At the end of the cycle,
sludge was rapidly settled before 3 l of supernatant was decanted. During the initial four
months (phase I), a hydraulic retention time (HRT) of 6.7 h was applied. Once the percentage
of abattoir wastewater in the SBR feed reached 100%, the aerobic period was gradually
extended resulting in a HRT of 13.3 h. Wastage was manipulated to maintain MLSS at
approximately 20 g.l-1, resulting in an SRT of 15-20 days. The pH in the system, which was
recorded but not controlled, typically fluctuated between 7.0 and 8.6 over the cycle.
Table 2. Cycle composition during four phases of the SBR operation over the 13 month
period.
Phase I
Phase II
Phase III
Phase IV
Day 0-133
Day 133-290
Day 290-333
Day 333-390
Settle (min)
2
2
2
2
Decant (min)
5
5
5
5
Feed (min)
18
18
18
18
Anaerobic (min)
50
60
60
60
Aerobic (min)
160
from 160 to 400
400
315
Post anoxic (min)
0
0
0
80
Total cycle (min)
240
from 240 to 480
480
480
HRT (h)
6.7
from 6.7 to 13.3
13.3
13.3
DO (mgO2.l-1)
no control
no control
control (3.0-3.5) control (3.0-3.5)
Cycle operation
Reactor and granules monitoring procedures
To monitor the reactor performance, the SBR effluent was sampled daily for analysis of
phosphate-P (PO4-P), ammonia (NH4+ + NH3), nitrate and nitrite. Cycle studies were
performed once a week during which liquid samples were taken and filtered every 20-30
minutes for analysis of phosphate-P, ammonia, nitrate and nitrite. The reactor mixed liquor
suspended solids (MLSS) and volatile MLSS (MLVSS) concentrations were also measured
weekly as well as the effluent total suspended solids (TSS).
122
To monitor the granule structure and characteristics, granule size distribution and density were
measured once a week. To determine the size distribution of the granules, 30 ml of well mixed
granular sludge was sampled from the SBR at the end of the aeration period and pumped
through a Malvern laser light scattering instrument, Mastersizer 2000 series (Malvern
Instruments, Worcestershire, UK). The granule density, defined as the quantity of dry mass
(TSS) per biomass volume, was measured every week by the blue dextran method adapted
from Di Iaconi et al. (2004). The reported value was determined from the mean of five
replicates.
Chemical analyses
The ammonia (NH4+ + NH3), nitrate (NO3-), nitrite (NO2-) and phosphate-P (PO4-P)
concentrations were measured using a Lachat QuikChem8000 Flow Injection Analyser
(Lachat Instrument, Milwaukee). Total and soluble chemical oxygen demand (TCOD and
SCOD, respectively), soluble five-day biological oxygen demand (SBOD5), total and soluble
Kjeldahl nitrogen (TKN and SKN), total phosphorus and total dissolved phosphorus (TP and
TDP), MLSS and MLVSS were analysed according to the standard methods (APHA, 1995).
The major ions present in the SBR (Ca2+, K+, Mg2+, Na+ and HS-) were measured by
Inductively Coupled Plasma - Atomic Emission Spectrometry (ICP-AES Varian Vista-PRO,
Varian, Inc.). VFAs were measured by Perkin-Elmer gas chromatography with column DBFFAP 15m x 0.53mm x 1.0µm (length x ID x film) at 140°C, while the injector and FID
detector were operated at 220°C and 250°C, respectively.
The fractions of various types of phosphorus-containing compounds in granules were
determined using the cold perchloric acid (PCA) extraction procedure developed by de Haas
et al. (2000). The initial centrifugation steps were used to separate the supernatant (containing
the phosphate species initially present in the sample bulk liquid (PO43-, HPO42-, H2PO4- and
H3PO4) plus the “interstitial” loosely bound phosphates) from the solid fraction or “pellet”.
The subsequent steps were aimed at distinguishing the biologically and chemically stored
phosphorus in the “pellet” (i.e. granules) by measuring the total P and ortho-P of cold PCA
extracts. The slow hydrolysis of biologically stored phosphorus (poly-P + nucleic acids +
phospholipids + minor organics P = “complex P”) in cold PCA implies that, provided the
ortho-P is measured immediately after the PCA extraction, the ortho-P content of the PCA
extracts may be assumed to originate from chemically bound P (mineral complexes) while
non-ortho-P (total P – ortho-P) is assumed to be of biological origin (i.e. “complex P”) (de
Haas et al., 2000). The residue of the cold PCA extraction can be analysed for total P and
contains mostly poly-P associated in some manner with proteins, or other macromolecules
which are not extractable in cold PCA. However, these non-PCA soluble phosphate
compounds can be almost fully extracted with NaOH (de Haas et al., 2000). To determine the
origin (i.e. chemical or biological) of the residue P content of the cold PCA extraction, this
additional NaOH fractionation step was performed on 3 different residues.
Microbial analysis
Granule samples for FISH were fixed in 3% paraformaldehyde (Amann, 1995). The fixed
granules were embedded in OCT (TissueTek, Sakura, USA) for cryosectioning as previously
described (Meyer et al., 2003). FISH was then carried out on the granule sections as described
by Amann (1995). Oligonucleotide probes used in this study were the combination of
EUB338 i-iii (EUBmix) for the detection of all bacteria (Daims et al., 1999), the combination
of PAO462, PAO651 and PAO846 (PAOmix) for Accumulibacter spp. (Crocetti et al., 2000),
the probe combination (GAOmix) of GAOQ989 (Crocetti et al., 2002) and GB_G2 (Kong et
123
al., 2002) for Competibacter spp., Actino-658 (Kong et al., 2005) for proposed Actinobacteria
PAO, NTSPA662 (Daims et al., 2001) for Nitrospira and NIT3 (Wagner et al., 1996) for
Nitrobacter.
Hybridised granule sections were visualised with a confocal laser scanning microscope
(CLSM) Zeiss LSM 510 (Carl Zeiss, Jena, Germany). To obtain images of entire granule
sections, between 10 and 40 overlapping, consecutive images of 1024x1024 pixels were
collected (depending on the size of the granule) using a Zeiss Neofluar 40x/1.3 oil objective.
The final composite image of the granule section was then reconstructed from all the single
images collected using Adobe Photoshop 7.0 (Adobe Systems, USA). FISH quantification
was performed on reconstructed granule images as described in Crocetti et al. (2002).
Oxygen profiles
The gradient of oxygen in granules was measured with oxygen microsensors (tip diameter
<10µm), which were constructed as described by Revsbech et al. (1989). Granules were
sampled from the SBR at the start of the aerobic period when ammonia and phosphates were
present at high concentrations and at the end of the aerobic period, at which time ammonia
and phosphates were usually depleted. Profile measurements were performed as described in
Lemaire et al. (2006). The composition of medium and the dissolved oxygen concentration in
the measuring flowcell was identical to that in the SBR at the start or the end of the aerobic
period.
Calculation of the degree of struvite saturation during a SBR cycle
To assess the contribution of struvite precipitation (Mg NH4 PO4·6H2O) on nutrient removal
in a SBR cycle, the degree of saturation of the relevant ions in the reactor bulk liquid was
calculated over the course of a SBR cycle. Liquid samples were taken for phosphate-P,
ammonia and Mg analysis and the pH was recorded on-line. The degree of struvite saturation
was expressed by the critical supersaturation ratio (Sc), which relates to the conditional
solubility product (Ps) and the product of the concentrations of phosphate-P, ammonia and Mg
(Pso) as shown in Eq. (1)
Sc = (Pso/Ps)1/3
(1)
Ps depends on the pH, the ionic strength and the ion activity and was calculated as described
by Ohlinger et al. (1998). The chemical equilibria for phosphate species
(PO43-/HPO42-/H2PO4-/H3PO4), magnesium complexes (Mg2+/MgOH+; Mg2+/MgPO4-;
Mg2+/MgHPO4 and Mg2+/MgH2PO4+), ammonia (NH3/NH4+) and water were solved to
determine the speciation of the ions PO43-, NH4+ and Mg2+ from the measured concentration
of phosphate-P, ammonia and Mg and measured pH value. For each equilibrium equation, the
stability constant values applied by Ohlinger et al. (1998) were used. The ions activity
coefficients were calculated using the Davies approximation of the Debye-Huckel equations at
20°C. The ionic strength of the SBR bulk liquid was estimated based on the measured
concentrations of the major ions present. The solubility product of struvite (Ksp=10-13.26)
proposed by Ohlinger et al. (1998) was employed in this study as it includes the effect of ionic
strength and magnesium phosphate complex formation.
Anoxic batch tests
To determine whether PAOs were contributing to denitrification in the reactor, anoxic batch
tests were performed to measure if denitrification was accompanied by P uptake under strictly
anoxic conditions. Granular mixed liquor (200 ml) was taken from the SBR at the end of the
124
anaerobic period and placed in two 100 ml vessels. After 10 min of helium sparging, 5 ml of
NaNO3 solution was added to one vessel while 5 ml of NaNO2 solution was added to the other
vessel resulting in an initial concentration of 55 mgN.l-1 of N-NO3 or N-NO2 in the two
vessels, respectively. Liquid samples were taken every 10 min to monitor the concentration of
phosphate, ammonia, nitrate and nitrite.
RESULTS AND DISCUSSION
Granule structure and characteristics
Following the inoculation of the SBR, an initial start-up period of 40 days was used to
gradually increase the COD and nutrient concentrations in the synthetic feed (Figure 1a). As
shown in Figure 1b, the granule size increased from an initial diameter of 0.1-0.8 mm
(10th-90th percentiles) to 0.9-1.6 mm over that period. The introduction of anaerobically pretreated abattoir wastewater to the reactor influent on day 40 resulted in a temporary decrease
of the granule size by about 25%, which fully recovered within a few weeks. From thereon, a
relatively stable granule size of 0.7-1.6 mm was maintained in the reactor. The fluctuations of
the granule size observed after day 133 are hypothesized to be a consequence of the dynamic
processes regulating the size of the granules; big granules are expected to fracture into smaller
aggregates that can then grow into bigger granules again.
Figure 1c shows the MLSS and MLVSS concentrations and also the granule density over the
experimental period. MLSS linearly increased from 2.8 to 20.2 g.l-1 in the first 160 days,
largely as a result of the increased COD load but also partially due to the fact that no sludge
was deliberately wasted from the reactor in this period. Sludge was regularly wasted after
day 160 to maintain the biomass concentration at approximately 20 g.l-1. The MLVSS
concentration displayed a similar trend to that of MLSS, but the MLVSS:MLSS ratio
decreased with time. The granule density gradually decreased from 290 gTSS.lbiomass-1 with
synthetic feed to around 150 gTSS.lbiomass-1 with real wastewater feed, with reasons yet to be
identified. The density measured in this study is considerably higher than the density values
reported in literature. For example, de Kreuk et al. (2005) reported a granule density of
78-89 gTSS.lbiomass-1, while Cassidy and Belia (2005) reported a density of 12-32
gTSS.lbiomass-1 in a granular SBR using abattoir wastewater.
125
Start-up Transition
period
100% abattoir wastewater
300
900
800
250
600
NH4-N
PO4-P
VFA (HAc)
150
100
500
400
300
(a)
Acetate (mg l-1)
N and P (mg l-1)
700
200
200
50
100
0
0
1600
1400
1200
Granule size (µm)
1000
800
600
d(0.1)
d(0.5)
d(0.9)
400
(b)
0
350
20
300
16
250
12
200
150
8
MLSS
MLVSS
Density
4
100
(c)
0
Density (g lbiomass -1)
MLSS and MLVSS (g l-1)
200
50
0
0
50
100
150
200
250
300
350
400
Days
Figure 1. (a) N, P and VFA concentrations in the SBR feed. During the start-up phase only
synthetic wastewater was used, which was then stepwise changed to real wastewater during
the transition period and after day 133 only anaerobically pre-treated abattoir wastewater was
fed to the reactor. (b) Volumetric distribution of granule size in the reactor during the entire
study. d(0.1), d(0.5), d(0.9) are the 10th, 50th and 90th percentiles of the distribution. (c) MLSS,
MLVSS and granule density during the study.
Nitrogen and Phosphorus Removal Performance
The SBR performance in terms of nitrogen and phosphorus removal varied considerably
during the process of gradually increasing the fraction of abattoir wastewater in the feed (data
not shown). Indeed, the unstable performance continued for 3-4 months after 100% of abattoir
wastewater was applied, as shown in Figure 2. As the nitrification gradually improved, more
carbon was needed to denitrify the increasing amount of NOx- produced and therefore the
extra acetate was gradually increased from 330 to 750 mgCOD.l-1 (Figure 1a). The acetate
concentration in the influent was stable after day 280 (750 mgCOD.l-1) which corresponds to
the time when complete nitrification was achieved in the reactor.
126
-1
NH4-N, NOx-N and PO4-P (mg l )
120
100
6h
7h
PO4-P
NH4-N
NOx-N
6h
5h
80
8h
short anoxic
period added
before settle
60
40
20
0
130
160
190
220
250
280
310
340
370
400
Days
Figure 2. Long-term effluent NH4-N, NOx-N and PO4-P concentrations showing the nitrogen
and phosphorus removal performance of the SBR, and the impact of cycle time changes
(indicated by vertical arrows) on the nitrification performance. The NOx- data point around
day 360, enclosed with {}, was an outlier caused by a hardware failure of the control system.
Good nitrogen and phosphorus removal performance was achieved once stable operation was
established. Table 3 shows the average reactor performance during the last 2 months of
operation under an organic, nitrogen and phosphorus loading rate of 2.7 gCOD.l-1.d-1,
0.43 gN.l-1.d-1 and 0.06 gP.l-1.d-1, respectively. Both the ammonia and phosphate removal
efficiencies were over 98%. The removal efficiency for the soluble COD was 85%. The
remaining soluble COD measured in the effluent (162 mg.l-1) was non-biodegradable as
indicated by the very low soluble BOD5 value (<2 mg.l-1) (Table 3). It indicates that the
abattoir pond wastewater used contained a relatively high fraction of non-biodegradable COD
estimated to be around 40% of its soluble COD. During the last two months of stable
operation, the biomass yield in the granular SBR was estimated to be 0.3 gVSS.gCOD-1.
Considering a biomass composition of CH1.8O0.55N0.2P0.015, the biomass growth accounted for
approximately 16% and 22% of the total N and P removal in the SBR, respectively.
Table 3. Summary of reactor performance in the last 2 months (day 333 to day 390).
Parameter
TKN (mgN l-1)
SKN (mgN l-1)
NH4 (mgN l-1)
NO3 (mgN l-1)
NO2 (mgN l-1)
NOx (mgN l-1)
TP (mgP l-1)
TDP (mgP l-1)
PO4 (mgP l-1)
TCOD (mg l-1)
SCOD (mg l-1)
SBOD5 (mg l-1)
TSS (mg l-1)
Influent
Effluent
Nutrient removal
average (SD; NS) average (SD; NS)
efficiency
237.3 (6.6; 8)
25.3 (5.2; 8)
85.7 % (TN)
220.1 (4.1; 7)
7.6 (0.7; 7)
92.7 % (TDN)
221.7 (6.7; 7)
0.8 (0.5; 10)
99.7 %
N.D.
5.2 (1; 9)
--N.D.
3.3 (4.1; 9)
--N.D.
8.5 (3.7; 9)
--34.3 (1.2; 8)
9.0 (2.8; 7)
73.8 %
32.4 (1.4; 7)
3.6 (1.2; 7)
89.5 %
33.6 (0.8; 7)
0.6 (0.3; 10)
98.3 %
*
1480 (138; 6)
467 (83; 5)
68.4 %
1072* (53; 6)
162 (33; 7)
84.8 %
N.A.
<2 (--; 4)
--205 (17; 8)
306 (88; 8)
---
*
including 750 mgCOD.l-1 of acetate added
SD: Standard Deviation; NS: Number of Samples; N.D.: Not Detectable, N.A.: Not Analysed
127
It is known that granular sludge systems produce effluent containing higher levels of
suspended solids compared to floccular sludge systems due to their specific design (i.e. short
settling time and sludge wasting with the effluent). In this study, the average effluent TSS
concentration was 306 mg.l-1, which is in the higher range of effluent TSS values reported in
literature for granular systems (McSwain et al., 2004; Mosquera-Corral et al., 2005; Pan et al.,
2004). However, the influent of those granular systems did not contain any solids (i.e.
synthetic wastewater) whereas in this study, the average influent TSS of 205 mg.l-1 would
have certainly contributed to the high level of TSS in the effluent given the short HRT of the
SBR. This high level of suspended solids in the effluent significantly reduced the TN, TP and
TCOD removal efficiencies of the SBR (Table 3). Cassidy and Belia (2005) reported higher
TN, TP and TCOD removal efficiencies in their granular SBR treating abattoir wastewater
under very similar organic, nitrogen and phosphorus loading rates and similar SRT. The
reason for this might be the much longer HRT they employed compared to this study (72 h vs.
13.3 h), resulting in only 8% of the SBR working volume being discharged in each cycle.
Considering that the same settling time of 2 min was applied, the suspended solids in their
effluent (55 mg.l-1) were therefore much lower than in our study which could explain the
higher nutrient removal efficiencies reported. More biomass was mechanically wasted in their
SBR each day in order to keep the SRT constant around 20 days. In either case though, an
additional filtration or floatation system would be necessary to remove the remaining solids
from the SBR effluent to meet environmental discharge limits. The additional costs associated
with such a solids removal step are likely limited at least for the meat industry since posttreatment using dissolved air flotation (DAF) is already required in many abattoirs to remove
remaining FOG and solids from the activated sludge process effluent before release to the
receiving waterways. On the other hand, the HRT applied in the last 2 months of the reactor
operation (i.e. 13.3 h) was less than 20% of the HRT currently applied in the full-scale
floccular SBR system used in the studied abattoir. This implies that the reactor volume could
theoretically be reduced by more than 80% if a granular SBR system was used, which would
result in considerable savings that could more than offset the extra costs for a post-treatment
step. It may therefore be concluded that granular sludge is a very attractive option for the
treatment of nutrient-rich industrial wastewater such as abattoir wastewater. However, the
operation of the anaerobic pre-treatment processes of the raw abattoir wastewater might have
to be modified (i.e. high rate prefermentor, anaerobic lagoons with shorter HRT) to ensure
that sufficient amount of VFAs are available for the subsequent N and P removal in the
granular sludge system.
Contribution of chemical precipitation in the overall nutrient removal efficiency
Nutrient-rich wastewaters often provide the right conditions for the formation of mineral
complexes such as apatites (Ca2HPO4(OH)2, hydroxydicalcium phosphate or HDP;
Ca5(PO4)3OH, hydroxyapatite or HAP; Ca3(PO4)2.xH2O, amorphous calcium phosphate or
ACP), newberyite (MgHPO4.3H2O) or struvite (MgNH4PO4.6H2O) (Maurer et al., 1999;
Musvoto et al., 2000). The precipitation of these minerals is controlled by pH, degree of
supersaturation, temperature and the presence of other ions in the wastewater and can occur
when the concentrations of Ca2+, Mg2+, NH4+ and PO43- ions exceed the solubility products of
the minerals (HDP: Ksp=10-22.6, Maurer et al. (1999); newberyite: Ksp=10-5.51, Abbona et al.
(1982); struvite: Ksp=10-13.26, Ohlinger et al. (1998)). It is well established that the large
release of PO43- and Mg2+ during the anaerobic phase of the EBPR process (ratio P:Mg of 3:1,
mole based) favours the precipitation of phosphates into diverse minerals, which is referred to
as “biologically induced precipitation” (Maurer et al., 1999).
128
In our granular SBR system, the combination of nutrient-rich industrial wastewater, high
loading rate and strong EBPR activity could have induced chemical precipitation, contributing
to the overall nutrient removal described previously. Figure 3 shows the concentrations of Ca,
Mg, ammonia and phosphate-P in the bulk liquid during a cycle study performed on day 270
along with the pH recorded. The important decrease of ammonia during the last 30 minutes of
the anaerobic period (around 30 mgN.l-1) was consistently observed in all cycles and, to our
knowledge, cannot be explained by any anaerobic biological processes. The relatively long
lag phase (15 min) observed between the end of feeding and the start of the sharp ammonium
decrease (Figure 3) indicates that the adsorption of ammonium to the granular sludge surface
unlikely contributed to this phenomenon as biosorption is a relatively fast process reaching
equilibrium within minutes (Nielsen, 1996). This means that chemical precipitation of
ammonium-containing minerals such as struvite likely occurred in this granular SBR during
the anaerobic phase, which is not surprising given the very high concentration of phosphate-P
(250 mgP.l-1), ammonia (120 mgN.l-1) and Mg (70 mg.l-1) at that time. To further demonstrate
the possibility for struvite formation in this SBR, the degree of struvite saturation, or critical
supersaturation ratio (Sc), is also depicted in Figure 3. When Sc>1, the SBR bulk liquid is
supersaturated and struvite precipitation could likely occur (represented by the grey period).
The initial formation of struvite (i.e. nucleation) can be facilitated by the presence of suitable
nuclei such as solid impurities (Doyle and Parsons, 2002). In our granular SBR, the
biopolymers matrix (i.e. external polymeric substances - EPS) of the granules may play this
role. When Sc<1, the bulk liquid is undersaturated and struvite would dissolve releasing PO43-,
NH4+, and Mg2+ in the bulk liquid. However, no corresponding increase of their respective
concentration was observed in the later part of the cycle (Figure 3). Any PO43- released under
aerobic condition would be immediately taken up by PAOs (alongside with Mg2+) to form
intracellular poly-P and any NH4+ released would be oxidised to nitrite/nitrate by nitrifiers,
therefore masking the possible effect of struvite dissolution. Nevertheless, indirect evidence of
this dissolution can be glanced when considering that 35 mgN.l-1 of NOx- accumulated in the
bulk liquid in the last 150 min of the cycle (data not shown) while only 20 mgN.l-1 of NH4+
was oxidised (Figure 3). The production of NOx- in excess of NH4+ consumption was
consistently observed in the last 1-2 h of the aerobic period (see also Figure 4) and was likely
due to the instantaneous oxidation of NH4+ released through the dissolution of struvite. It is
also interesting to notice that the calcium concentration in the bulk liquid more than doubled
in the last 200 min of the aerobic period (Figure 3). It could be that apatites formed during the
anaerobic period when the PO43- concentration was very high, later re-dissolved during the
aerobic period when PO43- level is very low and pH is decreasing (from 8.3 to 7.8). Unlike
PO43-, Ca2+ released from apatites dissolution (mostly HDP and ACP) was not subsequently
removed from the bulk liquid via biological processes and therefore accumulated. This agrees
well with the dynamic model proposed by Maurer et al. (1999) to describe the pH-sensitive
and partly reversible precipitation of calcium phosphates observed in EBPR systems.
129
0.6
0.4
0.2
0.0
8.2
200
80
150
60
100
8.0
7.8
pH
-1
N-NH4
Mg
250
-1
0.8
8.4
Struvite Sc
100
8.6
300
P-PO4 (mg l )
120
N-NH4, Mg and Ca (mg l )
struvite critical supersaturation (Sc)
1.0
Ca
pH
P-PO4
140
1.2
7.6
7.4
40
7.2
50
20
7.0
0
0
0
50
100
150
200
250
300
350
6.8
400
Time (min)
Figure 3. Cycle study performed on day 270 illustrating the role of chemical precipitation in
the overall nutrient removal. The vertical dotted line delimits the anaerobic and aerobic
periods. The profiles of total phosphate, ammonia, magnesium and calcium along with the pH
and the ratio of struvite critical supersaturation (Sc) are depicted. In the calculation of Sc, the
magnesium concentration during the anaerobic phase was estimated using the measured
concentrations at the beginning and the end of the anaerobic period and the phosphate profile,
assuming a Mg-released:P-released ratio of 0.33 mol.mol-1 (Smolders et al., 1994). The grey
zone indicates when struvite is likely to precipitate (Sc>1) in the cycle based on the
thermodynamic equilibrium. The rest of the time (Sc<1), struvite should dissolve.
To assess the exact contribution of chemical mineral precipitation to the overall nutrient
removal, the possible interactions between the main mineral formed and the different ions
species present in wastewater, as well as nucleation thermodynamics, crystal growth kinetics
and dissolution rates would have to be considered, which can rapidly become very complex.
A simpler method to evaluate the contribution of phosphate precipitation during the SBR
cycle is to apply the phosphorus fractionation procedure based on cold perchloric acid (PCA)
developed by de Haas et al. (2000) to mixed liquor samples taken at the end of the anaerobic
period and at the end of the cycle. This PCA fractionation procedure is capable of
distinguishing between biologically-stored forms of phosphate (mainly poly-P in EBPR
systems) and chemically precipitated forms of phosphate (ortho-P) in activated sludge
samples.
Table 4 presents the phosphorus fractionation of mixed liquor samples taken at the end of the
anaerobic and aerobic periods in different SBR cycles. First of all, it should be noted that the
typical EBPR pattern is well illustrated by the P fractionation results. By the end of the
anaerobic period, 30.5% of the total amount of P contained in the sample was released into the
bulk liquid (or supernatant) as ortho-P. This ortho-P was subsequently taken-up and stored
intracellularly as poly-P (i.e. “complex P”) in the aerobic period as shown by the higher
fraction of “complex P” in the granules at the end of the cycle compared to the end of the
anaerobic period (76.8% and 35.7%, respectively - see Table 4). The important information
given in Table 4 concerns the fraction of P held in mineral complexes formed in an SBR
cycle, which is represented by the ortho-P fraction in the PCA extract. As previously
suggested, this fraction was significant at the end of the anaerobic period (10.6%) but was
very small at the end of the SBR cycle (1.9%). This confirms that most of the mineral
complexes formed during the anaerobic period (due to higher bulk liquid phosphate-P
130
concentrations) were subsequently re-dissolved during the aerobic period. Therefore mineral
complexes (or precipitates) did not appear to contribute significantly to the overall nutrient
removal performance of the SBR. The P left over in the residue of the cold PCA extract was
found to be at least 95% “complex P” (i.e. most likely of biological origin) when employing a
NaOH based extraction procedure on samples taken at the end of a cycle and at the end of the
anaerobic period; less than 2% was of chemical origin (i.e. ortho-P) and around 3% were
unaccounted for (data not shown).
Table 4. Phosphorus fractionation determined by the cold PCA procedure in five mixed liquor
samples taken at the end of the anaerobic and at the end of the aerobic period during five
different cycles.
Sample/Extract
Supernatant
PCA
Residue (not extracted)
Recovery
Supernatant
PCA
Residue (not extracted)
Recovery
Ortho P
(% of sample TP)
Total P
(% of sample TP)
End of anaerobic period (n=5)
30.5% ; SD=1.8
30% ; SD=4.0
10.6% ; SD=1.8
46.3% ; SD=5.8
--29.5% ; SD=3.9
--105.7% ; SD=2.8
End of cycle (n=5)
0.2% ; SD=0.2
2.6% ; SD=0.9
1.9% ; SD=0.5
78.7% ; SD=4.3
--18.9% ; SD=2.3
--99.4% ; SD=5.0
“Complex P”
(Total P – Ortho P)
(% of sample TP)
(-0.5%) ; SD=2.1
35.7% ; SD=4.0
----2.4% ; SD=0.7
76.8% ; SD=3.8
-----
Simultaneous Nitrification, Denitrification and Phosphorus Removal (SNDPR) using
Granular Sludge
SNDPR via nitrite was previously demonstrated using floccular sludge and synthetic
wastewater (Zeng et al., 2003a). However, denitrification was not attributed to PAOs but to
glycogen-accumulating organisms (GAOs), their known competitors. Without denitrification
by PAOs, there is no true link between SND and EBPR, the two processes simply occurring in
the same sludge at the same time. Figure 4 shows a cycle study on the SBR performed on day
369. During the first 3 h of the aerobic period, before phosphates were completed taken up, 70
mgN.l-1 of NH4+ was removed with only the accumulation of less than 10 mgN.l-1 of NO2-,
demonstrating that simultaneous nitrification and denitrification (SND) occurred. The
oxidised nitrogen present at the end of the aerobic period was almost exclusively nitrite
(NO2- : NOx- ratio=0.95, SD=2.5%, n=63) as depicted in Figure 4, suggesting that N was
removed via the nitrite pathway in this granular SBR. At a microbial level, common nitrite
oxidising bacteria population (i.e. Nitrobacter and Nitrospira) were not detected in the SBR
(data not shown) providing further supporting evidence that SND via nitrite occurred in this
SBR.
131
-1
P-PO4 and acetate (mg l )
-1
400
N-NH4, N-NO2 and N-NO3 (mg l )
120
450
100
350
N-NO3
300
Acetate
P-PO4
250
N-NH4
200
N-NO2
80
60
40
150
100
20
50
0
0
0
50
100
150
200
250
300
350
400
450
Time (min)
Figure 4. Cycle study performed on the SBR on day 369 demonstrating the occurrence of
simultaneous nitrification, denitrification and phosphorus removal. The two dotted lines
delimit the anaerobic period (0-78 min), the aerobic period (78-393 min) and the post anoxic
period (380-470 min).
Most of the nitrite accumulation in the SBR (i.e. up to 45 mgN-NO2.l-1) occurred during the
last 2 h of the aerobic period, when phosphates was completely taken up, before being later
reduced to less than 5 mgN.l-1 in the post-anoxic period (Figure 4). The concomitance of the
nitrite accumulation and the depletion of phosphate suggested that PAOs might be responsible
for most of the denitrification and that SNDPR occurred in this granular SBR. This was
further confirmed by the anoxic batch test presented in Figure 5a. Phosphate was taken up in
the absence of oxygen while NO2- was being reduced. The P-uptake:N-denitrification ratio
was 0.84 mgP.mgN-1 which is comparable to the value of 1.18 mgP.mgN-1 reported by Zeng
et al. (2003b) for an enriched culture of denitrifying PAOs. When nitrate instead of nitrite was
added to the granular sludge under otherwise identical anoxic condition (Figure 5b), the
denitrification rate decreased by more than 80% (from 50.5 to 8.5 mgN.l-1.h-1) and no
simultaneous P-uptake was observed.
132
250
70
(a)
60
240
50
230
40
220
30
210
N-NH4
N (mg l-1)
N-NO2
10
200
P-PO4
0
190
60
290
50
P (mg l-1)
20
280
N-NH4
N-NO2
N-NO3
P-PO4
40
30
270
260
20
10
250
(b)
0
240
0
10
20
30
40
50
60
Time (min)
Figure 5. Measurement of phosphate, nitrate, nitrite and ammonia during anoxic batch tests
with nitrite (a) and nitrate (b) addition.
This study shows that SNDPR via nitrite can be achieved with nutrient-rich abattoir
wastewater using granular sludge and that SND and EBPR are truly linked, optimising the use
of the limited available COD for nutrient removal.
The presence of granules probably facilitated the SNDPR process. As shown in Figure 6,
oxygen was barely penetrating into the granules 1 h after the aerobic period started, when the
granule biomass was highly active. At the end of the aerobic period (i.e. the last 15 min) when
the microbial activity inside the granule was low and the DO level in the liquid phase reached
5-6 mgO2.l-1, oxygen was only penetrating the granules to a limited depth (i.e. 300-400 µm).
Anoxic zones in the inner part of the granules were therefore present during most of the
aerobic period.
133
800
600
400
Depth (µm)
200
Liquid
0
Granule
-200
-400
1h into aeration
-600
End of aeration
-800
-1000
0
1
2
3
4
5
6
7
O2 concentration (mg l-1)
Figure 6. O2 profiles in granules measured on day 240 at 1 h into the aeration period and at
the end of the aeration period (i.e. after DO level reached 5-6 mgO2.l-1 due to low microbial
activity in the granules). Profiles shown are averages (error bars=S.E., n=4).
Looking at a microbial community level, Accumulibacter spp. was the dominant PAO
detected in 10 different cryosectioned granules by fluorescent in situ hybridisation (41% of all
bacteria, SD=9, n=10, Supplementary Figures S1a and S1b). Actinobacteria PAOs were also
present but not dominant (4.1% of all bacteria, SD=1.2, n=7, Supplementary Figures S1c and
S1d). Very few clusters of Competibacter spp. cells, the only GAO detected in the system,
could be observed in each granule. Their low abundance was under the detection limit of the
FISH quantification method employed. Given the microbial composition of this granular
SNDPR system, the above stated proposition that PAOs (mainly Accumulibacter spp.) are
responsible for the denitrification in this reactor is strongly supported. The discrepancy
between the nitrite and nitrate reduction ability may also be explained by the information
contained in the genome sequence of Accumulibacter spp. recently published (Martin et al.,
2006). This genome does not appear to contain the genetic information for the known
respiratory nitrate reductase enzymes whereas genes for enzymes in the denitrification
pathway from nitrite reduction onwards were present. Since granular sludge is a complex
mixed cultures, organisms other than Accumulibacter spp. could have first reduced nitrate to
nitrite, before Accumulibacter spp. could finish the denitrification from nitrite to nitrogen gas.
It is very interesting to highlight that before being exposed to real wastewater conditions, the
original granules cultured on synthetic wastewater with acetate as the sole carbon source,
contained a high fraction of Competibacter spp. (around 25% of all bacteria) which was found
to be responsible for denitrification in that SNDPR system (Lemaire et al., in press). In this
study, the disappearance of Competibacter spp. from the SNDPR system and the
denitrification likely by Accumulibacter spp. coincide with the use of real abattoir wastewater.
In addition, the inhibition or elimination of the nitrite oxidiser bacteria population in this
SNDPR system prevented nitrite to be oxidised to nitrate which might have allowed PAOs to
perform full denitrification to N2 without relying on GAOs or any other denitrifiers to carry
out the first denitrification step.
134
Supplementary Figure S1. Reconstructed CLSM images of FISH micrographs of entire
granule sections (a, b and c) and part of the section (d). In (a) and (b) Accumulibacter spp.
cells are magenta (overlay of red PAOmix and blue EUBmix) and other Bacteria are blue
(blue EUBmix). In (c) and (d) Actinobacteria PAO cells are magenta (overlay of red Actino658 and blue EUBmix) and other Bacteria are blue (blue EUBmix). Scale bar = 100 µm for
(a), (b) and (c) and 10 µm for (d).
135
Hydraulic retention time (HRT) determination
Several factors were identified to have contributed to the variation of reactor performance.
These included the adaptation of the biomass to the increased N and P load caused by the
stepwise increase of the fraction of abattoir wastewater in the feed (see Figure 1a), and also
the variation of the VFA level in the feed. For example, the effluent P peak observed around
day 230 (Figure 2) correlated well with the low VFA concentration in the feed in this period
(Figure 1a). Of most significance to the reactor operation is the finding that the cycle time (or
equivalently the HRT when the amount of wastewater fed in each cycle remained constant)
has a major impact on the nitrification performance. Figure 2 shows that a cycle time of 8 h is
required in order to achieve complete nitrification. Between day 133 and day 250, when a
cycle time of 5 h or 6 h was applied, the degree of nitrification varied between 60 – 90%. This
is in clear contrast to the results obtained in periods when a cycle time of 8 h was applied (day
250 to day 280 and from day 300 onwards), during which over 99% of the influent ammonia
was removed (see also Table 3). It is interesting to observe the ammonia peak after day 280,
which further confirms that a cycle time of 6 h was too short.
Theoretically, the degree of nitrification is primarily determined by the SRT, with the HRT
playing little or no role. The fundamental reason for this prediction is that, in a steady state
operation of a reactor, the nitrifier concentration in the reactor is proportional to 1/HRT (see
Appendix). Consequently, the nitrification capability of the sludge is proportional to the
nitrogen load resulting in an effluent quality that is independent of HRT (see Appendix).
Based on the oxygen profiles inside the granules measured on day 240 (Figure 6), we
hypothesize that the contradiction between the theoretical and experimental results in this
study was caused by the mass transfer limitation of oxygen in granules. The limitation in
oxygen transfer implies that only a fraction of nitrifiers are able to contribute to the oxidation
of nitrogen. Consequently, the increase in the nitrifier population (XA) as the result of a
reduced HRT (Eq. (A3)) does not necessarily lead to a proportional increase in the
nitrification rate of the sludge, leading to incomplete nitrification when a too short HRT (or
cycle time) is applied. The shortening of the HRT leads to an increase in the COD and
nutrient loading of the process, and it is likely that this increased loading rate is the
performance-limiting factor in this case. However, no independent experimental study of the
effect of the loading rate was undertaken in this case as actual abattoir wastewater with a
given concentration range was used. The experimental results obtained in this study show that
a minimum cycle time of 8h is required for the wastewater and operational conditions
(SRT = 15–20 days, DO = 2.5–3.0 mgO2.l-1) used in this study.
In the design of a conventional wastewater treatment plant, the SRT is typically chosen to
ensure complete nitrification, while the HRT is designed such that the resulting solids
concentration is below a certain limit (e.g. 4 g.l-1) determined by the sludge settleability.
When granular sludge is used, a much higher solids concentration can be used (e.g. 20 g.l-1 in
this study). In this case, sludge settleability is no longer a limiting factor, and the minimum
HRT (or maximum loading rate) applicable is likely determined by other factors, as indicated
by the results obtained in this study. Mass transfer limitation could be one of these factors.
The design criteria for the HRT and loading rates in granular sludge systems require further
investigation.
136
CONCLUSION
Nitrogen and phosphorus removal using granular sludge from anaerobically pre-treated
abattoir wastewater was investigated. The following conclusions are drawn:
• Good nitrogen and phosphorus removal from nutrient-rich industrial wastewater can be
achieved biologically using granular sludge operated under alternating anaerobic and
aerobic conditions. The use of granules likely facilitated the simultaneous nitrification,
denitrification and phosphorus removal process observed;
• Unlike in granular SNDPR systems using synthetic wastewater, Accumulibacter spp. were
found to be responsible for both denitrification and P removal resulting in lower carbon
demand. Increased microbial diversity due to the use of more complex wastewater and
utilisation of the nitrite pathway are believed to be the main reasons for this denitrification
by PAOs.
• Mineral precipitation was observed in the reactor mostly during the anaerobic period.
However, the overall contribution of this mineral precipitation in the nutrient removal was
only minor due to subsequent dissolution in the aerobic period.
• The minimum HRT for a granular sludge system is not governed by the sludge
settleability and retention, as is the case in a system with floccular sludge. Mass transfer
limitations in granules are likely an important factor to be considered in the design of the
HRT and the COD and nutrient loading rate in a granular sludge system.
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology CRC, a Cooperative Research
Centre established and funded by the Australian Government together with industry and
university partners.
The authors also gratefully acknowledge the valuable input made by David de Haas from the
Advanced Water Management Centre, in Brisbane, Australia, on the phosphorus fractionation
procedure and Claudio Di Iaconi from the National Research Council, Water Research
Institute, in Bari, Italy, on granular sludge processes in general.
APPENDIX
The following mass balance equations can be made for an aerobic nitrifying system operated
with hydraulic and sludge retention times of HRT and SRT, respectively:
S NH
dX A
1
X A − bA X A −
XA
= µ A,max
dt
K NH + S NH
SRT
dS NH
S NH
XA
1
( S NN ,inf − S NH )
= µ A,max
+
dt
K NH + S NH Y A HRT
(A1)
where XA and SNH are, respectively, the nitrifier and ammonium (including ammonia)
concentrations in the system; SNN,,inf is the nitrifiable nitrogen concentration in the feed; µA,max ,
bA, KNH and YA are the maximum specific growth rate (d-1), the decay rate (d-1), the affinity
137
constant with respect to ammonia (mgN.l-1) of nitrifiers and the biomass yield
(mgCOD.mgN-1), respectively.
Solving Eq. (A1) in steady state for XA and SNH gives:
S NH =
XA =
K NH (b A + 1 / SRT )
µ A,max − b A − 1 / SRT
(A2)
Y A ( S NN ,inf − S NH )
HRT (b A + 1 / SRT )
(A3)
-1
-1
The nitrification capacity of the sludge (rN,max, mgN.l .d ) is:
rN ,max =
µ A,max
YA
XA =
µ A,max ( S NN ,inf − S NH )
HRT (b A + 1 / SRT )
(A4)
Eq. (A4) implies that, with a given HRT, the nitrifiers are able to oxidise the incoming
nitrogen load to the same level (SNH determined by Eq. (A2)), which is independent of the
HRT (and indeed the influent nitrifiable nitrogen concentration as well).
While the above equations were derived for continuously fed reactors, the conclusion that the
effluent ammonia concentration should be independent of HRT is applicable to a fed-batch
reactor such as the SBR used in this study.
REFERENCES
Abbona F., Madsen H.E.L. and Boistelle R. (1982). Crystallization of 2 Magnesium
Phosphates, Struvite and Newberyite - Effect of Ph and Concentration. Journal of
Crystal Growth. 57(1): 6-14.
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
APHA (1995). Standard methods for the examination of water and wastewater. Washington,
DC, American Public Health Association.
Beccari M., Carrieri C., Misiti A. and Ramadori R. (1984). Design of single sludge systems
for the treatment of wastewaters with high ammonia content. La Tribune du Cebedeau.
37(491): 387-394.
Bertanza G. (1997). Simultaneous nitrification-denitrification process in extended aeration
plants: Pilot and real scale experiences. Water Science and Technology. 35(6): 53.
Beun J.J., Hendriks A., Van Loosdrecht M.C.M., Morgenroth E., Wilderer P.A. and Heijnen
J.J. (1999). Aerobic granulation in a sequencing batch reactor. Water Research.
33(10): 2283-2290.
Cassidy D.P. and Belia E. (2005). Nitrogen and phosphorus removal from an abattoir
wastewater in a SBR with aerobic granular sludge. Water Research. 39(19): 48174823.
Comeau Y., Hall K., Hancock R. and Oldham W. (1986). Biochemical model for enhanced
biological phosphorus removal. Water Research. 20: 1511-1521.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
138
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
de Haas D.W., Wentzel M.C. and Ekama G.A. (2000). The use of simultaneous chemical
precipitation in modified activated sludge systems exhibiting biological excess
phosphate removal Part 2: Method development for fractionation of phosphate
compounds in activated sludge. Water SA. 26(4): 453-466.
de Kreuk M., Heijnen J.J. and van Loosdrecht M.C.M. (2005). Simultaneous COD, nitrogen,
and phosphate removal by aerobic granular sludge. Biotechnology and Bioengineering.
90(6): 761-769.
Di Iaconi C., Ramadori R., Lopez A. and Passino R. (2004). Preliminary biomass
characterization in a sequencing batch biofilm reactor. Annali di Chimica. 94(12): 889898.
Doyle J.D. and Parsons S.A. (2002). Struvite formation, control and recovery. Water
Research. 36(16): 3925-3940.
Frose G. and Kayser R. (1985). Effective treatment of wastewater from rendering plants.
Proceedings 40th Annual Purdue Industrial Waste Conference, Indiana, Butterworths.
Fuerhacker M., Bauer H., Ellinger R., Sree U., Schmid H., Zibuschka F. and Puxbaum H.
(2000). Approach for a novel control strategy for simultaneous
nitrification/denitrification in activated sludge reactors. Water Research. 34(9): 24992506.
Furumai H., Kazmi A.A., Fujita M., Furuya Y. and Sasaki K. (1999). Modeling long term
nutrient removal in a sequencing batch reactor. Water Research. 33(11): 2708-2714.
Keller J., Subramaniam K., Gosswein J. and Greenfield P.F. (1997). Nutrient removal from
industrial wastewater using single tank sequencing batch reactors. Water Science and
Technology. 35(6): 137-144.
Kerrn-Jespersen J.P., Henze M. and Strube R. (1994). Biological phosphorus release and
uptake under alternating anaerobic and anoxic conditions in a fixed-film reactor.
Water Research. 28(5): 1253-1255.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kong Y.H., Nielsen J.L. and Nielsen P.H. (2005). Identity and ecophysiology of uncultured
actinobacterial polyphosphate-accumulating organisms in full-scale enhanced
biological phosphorus removal plants. Applied and Environmental Microbiology.
71(7): 4076-4085.
Kuba T., Smolders G.J.F., van Loosdrecht M.C.M. and and Heijnen J.J. (1993). Biological
phosphorus removal from wastewater by anaerobic and anoxic sequencing batch
reactor. Water Science and Technology. 27: 241-252.
Lemaire R., Meyer R., Taske A., Crocetti G.R., Keller J. and Yuan Z.G. (2006). Identifying
causes for N2O accumulation in a lab-scale sequencing batch reactor performing
139
simultaneous nitrification, denitrification and phosphorus removal. Journal of
Biotechnology. 122(1): 62-72.
Lemaire R., Yuan Z.G., Blackall L.L. and Crocetti G.R. (in press). Microbial distribution of
Accumulibacter spp. and Competibacter spp. in aerobic granules from a lab-scale
biological nutrient removal system. Environmental Microbiology.
Lettinga G., Vanvelsen A.F.M., Hobma S.W., Dezeeuw W. and Klapwijk A. (1980). Use of
the upflow sludge blanket (USB) reactor concept for biological wastewater-treatment,
especially for anaerobic treatment. Biotechnology and Bioengineering. 22(4): 699-734.
Lin Y.M., Liu Y. and Tay J.H. (2003). Development and characteristics of phosphorusaccumulating microbial granules in sequencing batch reactors. Applied Microbiology
and Biotechnology. 62(4): 430-435.
Martin H.G., Ivanova N., Kunin V., Warnecke F., Barry K.W., McHardy A.C., Yeates C., He
S.M., Salamov A.A., Szeto E., Dalin E., Putnam N.H., Shapiro H.J., Pangilinan J.L.,
Rigoutsos I., Kyrpides N.C., Blackall L.L., McMahon K.D. and Hugenholtz P. (2006).
Metagenomic analysis of two enhanced biological phosphorus removal (EBPR) sludge
communities. Nature Biotechnology. 24(10): 1263-1269.
Maurer M., Abramovich D., Siegrist H. and Gujer W. (1999). Kinetics of biologically induced
phosphorus precipitation in waste-water treatment. Water Research. 33(2): 484-493.
McSwain B.S., Irvine R.L. and Wilderer P.A. (2004). The influence of-settling time on the
formation of aerobic granules. Water Science and Technology. 50(10): 195-202.
Meinhold J., Filipe C.D.M., Daigger G.T. and Isaacs S. (1999). Characterization of the
denitrifying fraction of phosphate accumulating organisms in biological phosphate
removal. Water Science and Technology. 39(1): 31-42.
Meyer R.L., Saunders A.M., Zeng R.J.X., Keller J. and Blackall L.L. (2003). Microscale
structure and function of anaerobic-aerobic granules containing glycogen
accumulating organisms. FEMS Microbiology Ecology. 45(3): 253-261.
Meyer R.L., Zeng R.J.X., Giugliano V. and Blackall L.L. (2005). Challenges for simultaneous
nitrification, denitrification, and phosphorus removal in microbial aggregates: mass
transfer limitation and nitrous oxide production. FEMS Microbiology Ecology. 52(3):
329-338.
Mosquera-Corral A., de Kreuk M.K., Heijnen J.J. and van Loosdrecht M.C.M. (2005). Effects
of oxygen concentration on N-removal in an aerobic granular sludge reactor. Water
Research. 39(12): 2676-2686.
Munch E.V., Lant P. and Keller J. (1996). Simultaneous nitrification and denitrification in
bench-scale sequencing batch reactors. Water Research. 30(2): 277-284.
Musvoto E.V., Wentzel M.C. and Ekama G.A. (2000). Integrated chemical-physical processes
modelling - II. Simulating aeration treatment of anaerobic digester supernatants. Water
Research. 34(6): 1868-1880.
Nielsen P.H. (1996). Adsorption of ammonium to activated sludge. Water Research. 30(3):
762-764.
Ohlinger K.N., Young T.M. and Schroeder E.D. (1998). Predicting struvite formation in
digestion. Water Research. 32(12): 3607-3614.
Pan S., Tay J.H., He Y.X. and Tay S.T.L. (2004). The effect of hydraulic retention time on the
stability of aerobically grown microbial granules. Letters in Applied Microbiology.
38(2): 158-163.
Peng D.C., Bernet N., Delgenes J.P. and Moletta R. (1999). Aerobic granular sludge - A case
report. Water Research. 33(3): 890-893.
Pitman A.R., Venter S.L.V. and Nicholls H.A. (1983). Practical experience with biological
phosphorus removal plants in Johannesburg. Water Science and Technology. 15(3-4):
233-259.
140
Revsbech N.P. (1989). An oxygen microsensor with a guard cathode. Limnology and
Oceanography. 34(2): 474-478.
Smolders G.J.F., Vandermeij J., Vanloosdrecht M.C.M. and Heijnen J.J. (1994).
Stoichiometric model of the aerobic metabolism of the biological phosphorus removal
process. Biotechnology and Bioengineering. 44(7): 837-848.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Willers H.C., Tenhave P.J.W., Derikx P.J.L. and Arts M.W. (1993). Temperature-dependency
of nitrification and required anoxic volume for denitrification in the biological
treatment of veal calf manure. Bioresource Technology. 43(1): 47-52.
Yang S.F., Tay J.H. and Liu Y. (2003). A novel granular sludge sequencing batch reactor for
removal of organic and nitrogen from wastewater. Journal of Biotechnology. 106(1):
77-86.
Zeng R.J., Lemaire R., Yuan Z. and Keller J. (2003a). Simultaneous nitrification,
denitrification, and phosphorus removal in a lab-scale sequencing batch reactor.
Biotechnology and Bioengineering. 84(2): 170-178.
Zeng R.J., Saunders A.M., Yuan Z.G., Blackall L.L. and Keller J. (2003b). Identification and
comparison of aerobic and denitrifying polyphosphate-accumulating organisms.
Biotechnology and Bioengineering. 83: 140-148.
141
Appendix G
Micro-scale Observations of the Structure of Aerobic Microbial Granules used
for the Treatment of Nutrient-Rich Industrial Wastewater
Romain Lemaire1, Richard I. Webb2 and Zhiguo Yuan1
1
Advanced Water Management Centre (AWMC), 2 Centre for Microscopy and Microanalysis
(CMM), The University of Queensland, St Lucia, Brisbane QLD 4072, Australia.
ABSTRACT
The structure and function of aerobic microbial granules from a lab-scale sequencing batch
reactor (SBR) treating nutrient-rich abattoir wastewater were investigated. These wastewaterfed granules were examined using a wide range of micro-scale techniques including light
microscopy, scanning and transmission electron microscopy, fluorescent in situ hybridisation
(FISH) combined with confocal laser scanning microscopy (CLSM) and oxygen and pH
microsensors, in conjunction with a range of measurements in the bulk liquid phase.
Interesting structural features were observed in these granules that have not been reported in
synthetic-fed granules. The complex nature of abattoir wastewater was suggested to be
responsible for accelerating the breaking process of large mature granules due to a rapid
clogging of the granules pores and channels and for the very diverse microbial community
observed displaying specific spatial distribution throughout the granules. More importantly,
the dissolution at lower pH of mineral complexes associated to the granule matrix of
extracellular polymeric substances (EPS) might have caused the structural damages observed
on the granules even though some pH buffer capacity were observed inside these granules.
Ciliate protozoa were found to be very abundant on the surface of these wastewater-fed
granules which could potentially assist with reducing the high levels of suspended solids
usually present in aerobic granular sludge effluent. All these observations provide support to
future studies on aerobic granular sludge treating real wastewater especially with regard to the
granule structure and the mechanisms involved in their formation.
Keywords: aerobic granule, electron microscopy, EPS structure, FISH, industrial wastewater,
micro-scale.
INTRODUCTION
The deterioration of water quality in rivers, lakes and other fresh water streams is often related
to the depletion of dissolved oxygen. When the level of organic compounds and nutrient
discharged into local waterways as a result of human activity is too high, the natural oxidation
processes of organic compounds and the proliferation of aquatic plant such as algae in
nutrient-rich waters (process known as eutrophication) will consume most of the dissolved
oxygen available in the water system. To prevent happening, the level of organic compounds
142
(often measured as chemical oxygen demand – COD) and nutrient – such as nitrogen (N) and
phosphorus (P) – in discharged wastewater must be significantly reduced by improving the
treatment processes. Tertiary wastewater treatment, usually based on biological treatment
known as the activated sludge system, is considered to be the easiest and most cost-effective
way to remove nutrients from wastewater streams. Biological nutrient removal relies on the
activity of a diverse microbial community that transfers organic matter and nutrients from the
wastewater (liquid phase) to the atmosphere (gas phase) and/or into biosolids (solid phase).
In conventional activated floccular sludge systems, microorganisms and small particles in
suspension in the wastewater shape into small aggregates or flocs (50 to 300 µm in size).
Under special condition, these aggregates can become much bigger and compact forming
granules (0.3 to 5 mm). While granules were first reported in an upflow anaerobic sludge
blanket (UASB) bioreactor two decades ago (Lettinga et al., 1980), recent research efforts
have been dedicated to the study of aerobic granules (Morgenroth et al., 1997; Beun et al.,
1999; Etterer and Wilderer, 2001). Aerobic granules can be described as compact and dense
aggregates of microbial origin with an approximately spherical external appearance which do
not coagulate under reduced hydrodynamic shear and settle significantly faster than
conventional activated sludge flocs. The growth of such granules is sometimes regarded as a
special case of biofilm development without the presence of any substratum for attachment
(Grotenhuis et al., 1991; El-Mamouni et al., 1998). It appears that aerobic granules do not
form naturally and must be cultivated in bioreactors under specific operating conditions
providing strong selective pressures (Liu and Tay, 2004). The large microbial diversity found
in aerobic granules has also led researchers to hypothesize that granulation is not a function of
specific microbiological groups (Beun et al., 1999). However, the exact mechanisms involved
in the successive stages of the aerobic granulation process have not yet been fully explained.
More recently, aerobic granules cultivated in synthetic wastewater bioreactors have been
reported to achieve COD and/or N removal (Tay et al., 2002; Liu et al., 2003; Yang et al.,
2003) and, in some cases, P removal (Dulekgurgen et al., 2003; Lin et al., 2003). Sequencing
batch reactors (SBR) have been identified as the most suitable bioreactor design to harbour
this novel aerobic granular sludge technology. The main advantages of the aerobic granule
technology are high biomass retention in the bioreactor, good settling properties, and the
capacity to withstand high organic loading rates which all contribute to the very small
footprint of this technology in comparison to conventional floccular activated sludge systems.
It should be noted that all these advantages originate from the unique compact and dense
structure of aerobic granules.
So far, most of the research have been primarily focussing on macro-scale characterisation of
aerobic granular sludge systems whether there were designed for COD removal only or for
both COD and nutrient removal. The effects of several key operating parameters (e.g.
dissolved oxygen (DO) concentration, shear force, settling time, feast/famine regime and
organic loading rate) on granule size and reactor performance were comprehensively
investigated with several different carbon substrates (acetate, glucose and phenol) (Liu and
Tay, 2004). Only few recent studies explored the micro-scale structure of aerobic granules
(Ivanov et al., 2005; Chen et al., 2007a; Liu and Tay, 2007; Wang et al., 2007; Zheng and Yu,
2007). The majority of these studies focussed on extracellular polymeric substances (EPS) and
its distribution within aerobic granules since non-biodegradable EPS appear to provide an
architectural structure and mechanical stability for such granules (Wang et al., 2007).
Unfortunately, to date, all these micro-scale studies were performed on granular systems fed
with synthetic wastewater containing a single carbon source (i.e. mainly acetate) and operated
for COD removal only. The ultimate goal of aerobic granular sludge technology is to treat real
wastewater, either from domestic or industrial origin, which often contains diverse carbon
143
sources along with a multitude of organic and inorganic compounds and some particulate
matters. This complex nature of real wastewater and the different substrates degradation rates
are likely to have an effect on the structure of aerobic granules and their ability to remove
COD and/or nutrients due to a change in the granule microbial communities affecting both the
type of EPS produced and the mass transfer of substrate within the granule (Schwarzenbeck et
al., 2005).
The aim of this study is to make micro-scale observations of the structure and function of
aerobic granules fed with real wastewater, providing support to future studies on granule
structure and the mechanisms involved in their formation. The granules from a lab-scale SBR
fed with nutrient-rich industrial wastewater (Yilmaz et al., submitted) were employed in the
study. The structure and function of granules were examined using a wide range of microscale techniques including light microscopy, scanning and transmission electron microscopy,
fluorescent in-situ hybridisation (FISH) combined with confocal laser scanning microscopy
(CLSM) and oxygen and pH microsensors, in conjunction with a range of measurements in
the bulk liquid phase. The work is to our knowledge the most comprehensive micro-scale
study on wastewater-fed aerobic granules using a variety of multi-disciplinary tools. Some of
the structural features observed provided support to the hypotheses made previously by other
researchers from aerobic granules obtained with synthetic feed. Others initiated new
hypotheses regarding the general and microbial structure and the fate of mature granules, the
effect of pH on the granule structure stability and the possible role played by protozoa in the
overall system performance. The work also provided some new directions and
recommendations for further experimental studies on aerobic granules in relation to their
structure and behaviour in real systems.
MATERIAL AND METHODS
Nutrient-rich industrial wastewater
The wastewater used in this study was from a local abattoir in Queensland, Australia. At this
site, the raw effluent passes through four parallel anaerobic ponds before being treated in a
SBR for biological COD and N removal. Anaerobic pond effluent from the abattoir was
collected on a weekly basis and stored at 4˚C. The average concentration of COD, ammonia
and phosphate were 1420 mgCOD.l-1, 240 mgN.l-1 and 40 mgP.l-1, respectively.
Reactor operation and sampling of granules
The aerobic granule SBR had a working volume of 5 l and was operated in a temperaturecontrolled room (18-22°C). It was operated on an 8 h cycle consisting of 18 min non-mixed
feeding (3 l each cycle), 60 min mixed anaerobic/anoxic, 315 min mixed aerobic, 80 min
mixed anoxic, 2 min settling and 5 min decanting periods. The DO level was controlled
between 3.0-3.5 mgO2.l-1 during the entire aerobic period using an on/off aeration control
system. The hydraulic retention time (HRT) was 13.3 h and the sludge retention time (SRT)
was kept around 15-20 days through sludge wastage. The pH in the system, which was
recorded but not controlled, typically fluctuated between 7.0 and 8.6 over the cycle.
All the granules analysed in this paper were sampled during steady state operation, after the
SBR had been operated for more than a year achieving stable COD, N and P removal.
Granules were either sampled at the end of the anaerobic period or at the end of the SBR
cycle.
144
Due to presence of mineral complexes inside these wastewater-fed granules over the course of
the SBR cycle (Yilmaz et al., submitted) and the dissolution of these complexes at lower pH,
the effect of pH fluctuation on the granule structure stability was investigated in several
anaerobic batch tests. For each batch test, 200 ml of a mixture of granule and liquor was
sampled from the parent SBR at the end of the anaerobic period and transferred into a pHcontrolled vessel kept under anaerobic condition. The vessel was consistently mixed and
granules were sampled after 1 h for examination under light and electron microscopy to
monitor their structure. Bulk liquid pH of 7.5 (pH in the parent SBR at the end of the
anaerobic period) and 6.5 were tested.
Physico-chemical analysis
To monitor the granule structure and characteristics, granule size distribution and density were
measured. To determine the volumetric size distribution of the granules, 30 ml of well mixed
granular sludge was sampled from the SBR at the end of the aeration period and pumped
through a Malvern laser light scattering instrument, Mastersizer 2000 series (Malvern
Instruments, Worcestershire, UK). The granule density, defined as the quantity of dry mass
per biomass volume, was measured by the blue dextran method described in Lemaire et al.
(in press), which was adapted from Di Iaconi et al. (2004).
The gradient of oxygen in granules was measured with oxygen microsensors (tip diameter
<10 µm), which were constructed as described by Revsbech et al. (1989). Granules were
sampled from the SBR at the start of the aerobic period when ammonia and phosphates were
present at high concentrations and at the end of the aerobic period, at which time ammonia
and phosphates were usually depleted. They were then transferred to a flow-cell with an
horizontal flow where replicate oxygen profiles were measured and averaged as described in
Meyer et al. (2003). The composition of medium and the dissolved oxygen concentration in
the measuring flow-cell was identical to that in the SBR at the start or the end of the aerobic
period. The pH gradient in granules was measured with pH microsensors using the same
experimental set-up than for oxygen profile measurement. Granules were sampled from the
SBR at the end of the anaerobic period. The medium used in the flow-cell was sampled from
the SBR (liquid phase) at the same time to keep the substrate concentrations identical.
Microbial analysis
Granule samples were fixed and FISH probed as previously described (Amann 1995). Prior to
FISH probing, fixed granule samples were embedded in optimum cutting temperature (OCT)
compound (TissueTek, Sakura, USA) for cryosectioning as previously described (Meyer et al.
2003). Embedded granules were then frozen and sectioned into 10 µm thick slices using a
cryotome operated at -20°C (Kryo 1720, Leitz, Germany). The granule sections were
collected on SuperFrost Plus microscope slides (Menzel-Glaser, Germany). Finally, the slides
were dehydrated by sequential immersion for 3 min in 50%, 80% and 98% ethanol and airdried.
Oligonucleotide probes applied on the granule sections were the combination of EUB338 i-iii
(EUBmix) for the detection of all bacteria (Daims et al. 1999), the combination of PAO462,
PAO651 and PAO846 (PAOmix) for Accumulibacter spp. (Crocetti et al. 2000), the probe
combination (GAOmix) of GAOQ989 (Crocetti et al. 2002) and GB_G2 (Kong et al. 2002)
for Competibacter spp., NTSPA662 for Nitrospira spp. (Daims et al. 2001), NIT3 for
Nitrobacter spp. (Wagner et al. 1996) and NSO1225 for most of the ammonia oxidising
bacteria (AOB) from the Betaproteobacteria (Mobarry et al. 1996). Additionally, probes for
the newly proposed Actinobacterial PAOs (Kong et al. 2005) and Defluviicoccus spp.-related
145
GAOs (Meyer et al. 2006) were also used. Fluorescently labelled oligonucleotides were
purchased from Thermo (Ulm, Germany) with fluorescein isothiocyanate (FITC) or one of the
sulfoindocyanine dyes indocarbocyanine (Cy3) or indodicarbocyanine (Cy5).
Microscopy Images
Whole fresh granules were photographed using an Olympus SZH10 stereo microscope with a
DP70 digital camera.
FISH images were collected with a confocal laser scanning microscope (CLSM) Zeiss 510
(Carl Zeiss, Jena, Germany) using an argon laser (488 nm), a helium neon laser (543 nm) and
a red diode laser (633 nm) fitted with 515-565 nm BP, 590 nm LP and 660-710 nm BP
emission filters, respectively. To obtain images of entire granule sections, between 10 and 50
overlapping, consecutive images of 1024x1024 pixels were collected (depending on the size
of the granule) using a Zeiss Neofluar 40x/1.3 oil objective. The final composite image of the
granule section was then reconstructed from all the single images collected using Adobe
Photoshop 7.0 (Adobe Systems, USA). For single images of specific part of the granule
section a Zeiss Apochromat 63x/1.4 oil objective was used. FISH quantification was
performed according to (Crocetti et al., 2002) where the relative abundance of each group was
determined as mean percentage of all bacteria based on pixel area counting.
To visualise the structure of aerobic granules at a micro-scale level, scanning and transmission
electron microscopy (SEM and TEM, respectively) were employed. Prior to visualisation,
granules EPS material was first stabilised using 2.5% glutaraldehyde and 75 mM lysine in 0.1
M cacodylate buffer for 10 minutes to minimise any structural damages arising from the
dehydration procedure (Jacques and Graham, 1989). All subsequent processing was
performed in a Pelco Biowave microwave oven. Granules were fixed in 3% glutaraldehyde in
0.1 M cacodylate buffer and after washing in 0.1 M cacodylate buffer, were postfixed in 1%
osmium tetroxide.
For transmission electron microscopy (TEM) studies, granules were dehydrated in a graded
acetone series and embedded in Epon resin. Semi-thin sections of 500 nm were stained with
1% Toluidine Blue and 1% borax and viewed with an Olympus BX61 stereo microscope.
Ultra-thin sections of 60 nm thickness were cut using a Leica Ultracut UC6 ultramicrotome
and mounted on the Formvar coated copper grids, stained with 5% uranyl acetate in 5%
methanol and Reynolds lead citrate, and viewed using a JEOL 1010 transmission electron
microscope operated at 80 kV. To further reduce any possible artefacts coming from the
fixation steps, some fresh granules were also frozen in a Leica EMPACT 2 high pressure
freezer and cryosubstituted at -850C over 2 days in 2% osmium tetroxide 0.5% uranyl acetate
in acetone. Specimens were then warmed to room temperature over 13 hours, washed in
acetone and embedded in Epon resin. Ultra-thin sections were cut and viewed by TEM as
described above.
For scanning electron microscopy (SEM) studies, granules were dehydrated in a graded
ethanol graded series and infiltrated with the drying agent hexamethyldisilazane and left
overnight to dry before being sputter coated with platinum. Viewing of samples was
conducted using a JEOL 6300F scanning electron microscope operated at 5-10 kV. To
observe the internal structure, some dehydrated granules were taken from 100% ethanol,
frozen in liquid nitrogen and fractured. These fractured granules were then thawed in ethanol,
dried as described above and coated with platinum. The internal structure of granules were
visualised by SEM as described above.
146
RESULTS AND DISCUSSION
General macro characteristics of mature granules
Over the course of this study, the granular sludge SBR was in steady state and consistently
removed 85%, 93% and 89% of the soluble COD, soluble nitrogen and soluble phosphorus
present in the abattoir wastewater, respectively. This good nutrient removal was achieved
through the process known as simultaneous nitrification, denitrification and phosphorus
removal (SNDPR), facilitated by the presence of large and stable anoxic zones in the inner
part of the granules (Yilmaz et al., submitted).
Figure 1 presents an overview of the macro-characteristics of the mature granules developed
in this SBR fed with industrial wastewater. The volumetric size distribution of the granules
presented in Figure 1a indicates that more than 80% of the biomass volume is made of
granules with a size larger than 600 µm while the volume percentage of biomass with a size
smaller than 300 µm (likely flocs) is less than 5% which demonstrates that granules were the
dominant form of bacterial aggregates in this SBR. The average density of these granules was
measured at 150 g.lbiomass-1, which is considerably higher than what is generally reported for
synthetic wastewater-fed granules (30-80 g.lbiomass-1). Figure 1b and 1c show that most
granules have a round shape with a clear outline; however, smaller granules with concave
shapes were also observed suggesting they might have been part of a bigger granule that
disintegrated, which will be further discussed later in this paper.
14
Volume (%)
12
(a)
10
8
6
4
2
0
10
100
Particle Size (µm)
1000 2000
Figure 1. (a) Volumetric size distribution of the SBR mature granules. Pictures of granules
obtained (b) with a scanner and (c) with a light microscope. Scale bars = 2 mm.
147
Micro-scale structure of granules and illustration of the role played by EPS
To explore the micro-scale structure of these granules especially in regards to the EPS
structural matrix, electron microscopy (both SEM and TEM) was employed (Figure 2). The
overall round shape of aerobic granules is actually made up of large cauliflower-like
outgrowths (Figure 2a). These outgrowths have also been reported in acetate-fed aerobic
granules (Tay et al., 2004; Liu and Tay, 2007), but were not as pronounced as observed in this
study. On the surface of the granule (Figure 2c) some glue-like substances (i.e. EPS), provide
the cohesive material to maintain the bacteria bound to each other. To demonstrate the
importance of the lysine based pre-fixation step before visualising granules with SEM, a SEM
image of a granule which has not been pre-fixed in lysine solution is shown in Figure 2e. The
EPS matrix has obviously shrunk during the dehydration process revealing more rod-shaped
bacteria underneath it. Extreme caution should be exercised when preparing biological
samples for SEM or TEM observation to avoid any structural artefacts. The internal EPS
structure was observed on granules fractured in liquid nitrogen to prevent any structural
damages coming from the use of a cutting tool. The SEM picture of the inner part of a
fractured granule is presented in Figure 2b. Figure 2d presents a close up image of the field
delimited in black in Figure 2b, which is located near the surface of the granule. All bacteria
cells visible on this image are embedded in the EPS matrix. The cellular origin of this EPS is
clearly demonstrated in the high magnification SEM image of a fractured granule displayed in
Figure 2f. Each individual cell is enclosed in a thick EPS capsule. An ostensible difference
can be made between the EPS of this cell capsule and the main EPS matrix cementing the
overall granule. The real distance between bacteria in granules is very difficult to assess due
to the thickness of the sections generally observed by light microscopy or CLSM where lots of
bacteria are piled up on the top of each other. The TEM image of a granule ultra-thin section
depicted in Figure 2g reveals the exact distance between each cell in three different bacteria
clusters defined by their size, shape, grey scale intensity and general texture at high
magnification. The cell to cell distance varied substantially depending on the type of bacteria
resulting in sparse or dense cell clusters. The space between cells is filled by some EPS
material, which means that the cells from the sparse cluster might indeed produce more EPS
than those from the dense cluster. The type of EPS produce by different clusters of bacteria is
also likely to be different throughout the granule. To illustrate that point, Figure 2h shows a
TEM image of two different clusters of bacteria, one cocci-shaped and one rod-shaped,
embedded in what appear to be different types of EPS matrix based on the grey scale intensity
and texture.
148
Figure 2. SEM images of (a) an entire granule, scale bar = 200 µm; (b) a fractured granule,
scale bar =200 µm; (c) granule outer surface delimited in black in (a), scale bar = 1 µm; (d)
granule inner surface delimited in black in (b) , scale bar = 1 µm; (e) outer surface of a granule
without any pre-fixation step, scale bar =1 µm; (f) granule inner surface at high magnification,
scale bar = 1 µm; TEM image of (g) three bacteria clusters in the outer part of a granule, scale
bar = 5 µm; (h) two bacteria clusters showing different types of EPS, scale bar = 5 µm.
149
Voids, channels and fate of large mature granules
Voids and channels have been reported in some acetate-fed granules (Ivanov et al., 2005;
Zheng and Yu, 2007) and were found to play a key role in the transport of substrate and
metabolites in and out of the granules. Such voids or cavities were observed in almost every
granule sections examined. One of these large voids can be seen in the centre of the granule
section depicted in Figure 3a. These voids were most of the time connected to the outside of
the granules via channels-like structures as illustrated in the light microscope images of Figure
3b and 3c and the TEM image of Figure 3d. The channels depicted in Figure 3c and 3d are
filled with some material, again likely EPS, and could be compared to a ground water system
where liquid can circulate in and out of the granules through some porous material. Very few
clusters of bacteria are present in these channels probably due to the constant flow of liquid
which prevent them from attaching firmly on to the granule and thus being washed away. In
comparison, the channel presented in Figure 3b is not filled with any material and appears to
be more like an open river system where liquid can circulate more freely. These types of
channels or interstices are usually located on the boundary line between two cauliflower-like
outgrowths. The two large blue shapes at the entrance of this channel are two protozoa
(ciliates) and their presence will be discussed later in this paper.
Figure 3. Light microscope images of semi-thin sections (500 µm) embedded in resin and
stained with Toluidine Blue of (a) an entire granule, scale bar=100 µm; (b) and (c) channels
on the granule surface, scale bar=10µm; (d) TEM image of an inner channel, scale bar=10µm.
As mentioned earlier, not all granules in our SBR were big and round shaped. A large number
of granules were of concave shape with smooth, dense cauliflower-like outgrowths on one
side and loose and fluffy organisation on the other. Some examples of such granules are
presented in Figure 4a and 4b. A close up image of the smooth and dense side of such a
granule is shown in the Figure 4c while the loose and fluffy side is shown in Figure 4d. These
highly heterogenous granule structures seem to originate from the disintegration of bigger
granules. The newly formed granules can then grow to become themself big mature granules
that would break into smaller aggregates. However, the properties of these “recycled” mature
granules would likely differ from that of the original large mature granules due to the
150
heterogeneity of the structure from which they had to re-develop. A recent research study by
Zheng and Yu (2007) found a positive correlation between the bioactivity of acetate-fed
granules and their porosity and also reported that the granule porosity decreased as the granule
size increased. They suggested that the pores of large granules were more readily plugged by
EPS leading to a decrease of the biological activity because of a lack of nutrient or
metabolites transport. In our system, the particular and colloidal matter and the high level of
fat, oil and grease (FOG) present in abattoir wastewater might have speeded up the clogging
process of the granule channels and pores and increased the mass transfer limitation of
nutrient and substrates. This would have weakened the inner structure of the granule and
explain why our granules only grew to a maximum size of 1.5-2 mm before breaking up into
smaller and less homogenous granules. Therefore, not only the well studied operating
parameters such as settling time, shear force and organic loading rate can have an impact on
the size of the granules but also the characteristics of the wastewater to be treated.
Figure 4. (a) SEM image of a broken granule, scale bar = 100 µm; light microscope images of
semi-thin sections (500 µm) embedded in resin and stained with Toluidine Blue of (b) a
broken granule, scale bar = 100 µm; (c) part of the granule edge delimited in white in (b),
scale bar = 10 µm; (d) part of the granule edge delimited in black in (b), scale bar = 10 µm.
Microbial population and distribution in granules
The mass transfer limitation of nutrient, substrates and metabolites inside large granule was
proposed to cause the breaking up of mature granules into smaller aggregates. This theory
implies that the core part of mature granules must be deprived of any substantial microbial
activity due to substrate diffusion limitation. Some researchers have used staining methods to
establish the distribution profile of total (SYTO 63) and dead cells (SITOX Blue, BacLight
Live-Dead staining kit) in acetate-fed aerobic granules (McSwain et al., 2005; Chen et al.,
2007b; Chiu et al., 2007). The problem with these methods is that no standard procedure is
available in the literature for the staining of entire granules and different incubation times,
ranging from 5 to 30 min have been employed by different groups. Due to the diffusion
151
limitation likely to occur in large granules, the time of incubation should probably be defined
based on the size of the granules if anything. Due to the lack of consistency of these staining
procedures, the cell distribution in our wastewater-fed granules was investigated qualitatively
using SEM and TEM images. A SEM image of the central part of a fractured granule is
illustrated in Figure 5a. This image can be compared to that of Figure 2f taken near the
granule surface. The same kind of EPS capsule is observed in both images but the shape of the
cells enclosed in these capsules are very different. Cells in the central part of the granules
(Figure 5a) are mal-formed compared to the smooth round shaped cells of Figure 2f. It is very
likely that these amorphous cells are indeed dead cells. To confirm that, several juxtaposing
TEM images were taken from the edge to the centre of the granule. Figure 5b and 5c are two
examples of what could be observed in the centre and on the edge of every granule,
respectively. Mostly cell walls embedded in the EPS matrix are left in the central part of the
granules whereas lots of compact cells are located on the edge of the granule. This cell
distribution in wastewater-fed granules supports the nutrient and substrate diffusion limitation
theory, which may be the main cause for the breaking up of large mature granules.
Figure 5. (a) SEM image of the central part of a fractured granule, scale bar = 1 µm; TEM
images of (b) the central part of a granule section, scale bar = 5 µm; (c) the edge of the same
granule section, scale bar = 5 µm.
The microbial diversity within the granule structure and spatial distributions of various groups
of bacteria of importance to biological nutrient removal was investigated using a wide range
of FISH probes designed to identify the most common microbial communities found in
152
activated sludge performing nutrient removal. Accumulibacter spp., the main polyphosphate
accumulating organism (PAO) found in biological P removal system, were dominant in these
wastewater-fed granules (41% of all bacteria). This domination is illustrated in Figure 6a and
6b where FISH images of entire granule sections are reconstructed (Accumulibacter spp. in
magenta, ammonia oxidising bacteria (AOB) in cyan and other bacteria in blue). Oxygen
micro-profiles were measured in these granules during the first hour of the aeration period
when the microbial activity was the highest (i.e. substrates in excess) and at the end of the
aeration period, 4 h later, when the microbial activity was lower (i.e. most substrates
depleted). Oxygen was found to penetrate only as far as 50 µm inside the granules in the first
hour of the aeration period and around 400 µm deep at the end of the aeration (Yilmaz et al.,
submitted). Accumulibacter spp. were located on the outer part of these wastewater-fed
granules where the dissolved oxygen concentration was higher. This preferred location has
been already reported and quantified in acetate-fed granules performing N and P removal
(Lemaire et al., in press). Other P removal microorganisms recently speculated,
Actinobacteria-PAO, were also present but in much lower abundance (4.1% of all bacteria) in
comparison to Accumulibacter spp. as depicted in Figure 6c. Very few clusters of
Competibacter spp. cells, the main glycogen accumulating organisms (GAO) usually found in
large numbers in biological P removal processes, could be detected in the granules. This is
considered desirable as GAOs compete with Accumulibacter spp. for the same carbon source
but without performing any P removal. One of the few Competibacter spp. cluster present in
these wastewater-fed granules is shown in Figure 6e in cyan.
The nitrifying organisms (i.e. ammonium and nitrite oxidising bacteria, AOB and NOB
respectively) are also important microbial populations in nutrient removal processes due to
their ability to oxidise ammonium/nitrite to nitrite/nitrate, which can then be reduced to dinitrogen gas by other organisms. AOBs were present in these granules as expected from the
good nitrification performance of this SBR. A typical dense AOB cluster is depicted in Figure
6d in magenta. According to their strict aerobic metabolism, they should also be located on
the outer part of the granule where oxygen is always available. Fig 6a and 6b shows that most
AOB clusters were indeed situated in the first 200 µm from the granule surface but rarely in
the most outer part of the granule (i.e. 0-50 µm layer) where oxygen availability is high.
Instead, AOBs appear to grow just behind the thick layer of Accumulibacter spp. surrounding
the granule edge. Strangely, some AOB clusters were even found right in the centre of the
granule (indicated by the white circles on Figure 6a and 6b) where, according to the oxygen
micro-profiles, oxygen should not be usually present. However, these AOB clusters were
always located along the edge of large internal voids where small “pockets” of oxygen might
have been present after diffusing through some of the channels described earlier in this paper.
This could explain why some oxygen micro-profiles presented small surges of oxygen
concentration deep inside the granule. It clearly highlights the heterogenous nature of aerobic
granules and the need to study a sufficient number of granules when investigating their microscale structure.
No NOB targeted by the FISH probes applied was found in these wastewater-fed granules.
The oxidised nitrogen accumulating in the liquid phase of the SBR at the end of each cycle
was almost exclusively nitrite (data not shown) confirming the absence of NOB in the system
and that N was likely removed through the nitrite pathway. The limitation of oxygen transfer
inside the granules could have advantaged the nitrite reducing organisms over NOB by
providing large anoxic zones where denitrification could occur using the soluble COD present
in the wastewater. In addition, the domination of Accumulibacter spp. on the outer part of the
granules where oxygen was abundant might have prevented NOB to adequately perform their
aerobic metabolism due to insufficient oxygen availability.
153
Figure 6. Reconstructed CLSM images of FISH micrographs of entire granule sections (a, b
and c) and part of the section (d and e). In (a) and (b) Accumulibacter spp. cells are magenta
(overlay of red PAOmix and blue EUBmix) most of the ammonia oxidising bacteria (AOB)
from the Betaproteobacteria are in cyan (overlay of green NSO1225 and blue EUBmix) and
other Bacteria are blue (blue EUBmix). The white circle highlights the presence of AOB in
the centre of the granule. In (c) Actinobacteria PAO cells are magenta (overlay of red Actino658 and blue EUBmix) and other Bacteria are blue (blue EUBmix). In (d) AOB from the
Betaproteobacteria are in magenta (overlay of red NSO1225 and blue EUBmix) and other
Bacteria are blue (blue EUBmix). In (e) Accumulibacter spp. cells are magenta (overlay of
red PAOmix and blue EUBmix) Competibacter spp. are in cyan (overlay of green GAOmix
and blue EUBmix) and other Bacteria are blue (blue EUBmix). Scale bars = 100 µm for (a),
(b) and (c) and 10 µm for (d) and (e).
154
Impact of bulk liquid pH on the granule structure
Aerobic granular sludge technology is particularly well suited to treat industrial wastewater
due to its small footprint and capacity to withhold high loading rates. Most intensive water
user industries are subject to inherent production variability often resulting in changes of the
wastewater composition which could affect its pH. Yilmaz et al. (submitted) reported that
mineral complexes such as struvite and apatites could precipitate in these wastewater-fed
granules during the anaerobic phase of the SBR cycle when the concentration of Ca2+, Mg2+,
NH4+ and PO43- ions is the highest. Due to the dissolution of these complexes at lower pH and
the possible effect that could have on the overall granule structure, the influence of pH
variation on the macro-structure of wastewater-fed granules was investigated. Series of batch
tests were performed in a small pH-controlled vessel to investigate the effect of pH on the
granule structure. Figure 7a and 7c show light microscopy and SEM images of the granules
after a 1h batch test at pH 7.5, simulating the bulk liquid pH of the parent SBR at the end of
the anaerobic period, while example images obtained in the case of pH 6.5 are shown in
Figure 7b and 7d. After 1h batch test at pH 6.5, granules started to lose their smooth and
compact external appearance (Figure 7d) and most of the smaller granules even completely
disintegrate (Figure 7b). This disintegration led to a significant decrease of the volumetric size
distribution of granules with the 10th, 50th and 90th percentiles dropping by an average of 80%,
50% and 20%, respectively.
Figure 7. Light microscope images of granules after 1h anaerobic batch test performed (a) at
pH 7.5 and (b) at pH 6.5, scale bars = 1 mm. SEM images of the same granules after batch
test (c) at pH 7.5 and (d) at pH 6.5, scale bars = 200 µm.
155
To better understand the effects that lower bulk liquid pH can have on the granule structure,
microsensors were employed to measure the in situ pH profiles in these wastewater-fed
granules after immersion in a pH-controlled flow cell. One of several pH profile time series
measured is presented in Figure 8. The first pH profile (t=0) was measured as soon as the
granule was transferred from the parent SBR bulk liquid (pH=7.6) into the measuring flowcell where the pH was controlled at 6.5. The subsequent profiles were measured on the same
granule (different location) after 10, 20 and 60 minutes. The presence of a clear pH gradient
indicates that these granules have a pH buffer capacity due to diffusion limitation and/or in
situ biological activities. This buffering effect decreased overtime but a slight pH gradient
remained visible even after 60 min. If the dissolution of mineral complexes likely associated
with the granules biopolymers (i.e. EPS matrix) is indeed the main reason for the structural
damage observed at lower pH, the presence of sharp pH gradient inside the granules would
protect them from total disintegration. It could explain why during the batch tests at pH 6.5
small granules were more affected than large granules. The relatively limited pH buffer
capacity of smaller granules would result in similar pH levels in the bulk liquid and in the
granules and therefore increase their vulnerability under lower pH condition. Although these
are only preliminary results, the high probability of having mineral complexes associated with
the EPS matrix in granules treating nutrient-rich industrial wastewater calls for more
extensive research to be done on the effect of dynamic pH fluctuations on the granule
structural stability.
-800
-600
-400
Liquid
Depth ( µm)
-200
0
200
Granule
400
600
800
t=0
1000
t=10min
1200
t=20min
1400
t=60min
1600
6.3
6.5
6.7
6.9
7.1
7.3
7.5
7.7
pH
Figure 8. Typical pH profile time series measured in one granule using a microsensor in a
flow-cell controlled at pH 6.5.
Presence and role of ciliates on real wastewater-fed granules
Protozoa from the ciliate group (likely from the Vorticella genus based on the observed
morphology) were present in “bouquets” on the surface of almost every granule examined
(Figure 9a). These ciliates are attached to the granule via their retractable stalk as depicted in
Figure 9d. They are always located in the concave part of the granule (Figure 9b and Figure
3a) or in interstices between large cauliflower-like outgrowths (Figure 9c). Due to the high
shear force applied in the SBR, the constant collision between granules would have prevented
156
ciliates to stay attached on exposed part of the granules explaining why they are mostly
located in sheltered areas.
The presence of these ciliates on the surface of the granules raises the question of their
specific role and possible impact on the overall granular process. During measurement of
oxygen profiles in granules it was observed that when the microsensor tip had to progress
through a ciliate bouquet before reaching the surface of the granule (Fig 9e), the measured
oxygen concentration immediately dropped to zero. It indicates that ciliates were indeed
interfering with the oxygen diffusion process in the granule by creating some localised oxygen
depleted zones at the surface of the granule. The high abundance of ciliates on each granule
could have had an impact on the overall oxygen diffusion limitation in the granules but also
on the diffusion of other substrates.
However, the main role of these ciliates in the system was probably related to their predatory
behaviour. Like most protozoa, ciliates feed on small organic particulates including bacterial
cells. They can sweep into their mouth free floating bacteria by creating a vortex through the
rotation of their oral cilia. The mouth of these ciliates can be observed in Figure 9f although
their oral cilia are hidden inside. The TEM image presented in Figure 9g shows a cross section
of a ciliate mouth with a group of free floating bacteria close by. These bacteria are then
digested by the ciliate via the formation of internal food vacuoles that are depicted in Figure
9h (black arrows). High level of suspended solids in granular sludge effluent is a well known
drawback of the technology and post-treatments for solids removal are often required
(Schwarzenbeck et al., 2005). It is directly linked to the process operation (i.e. short settling
time) where slowly settling biomass has to be washed out from the system continuously. The
predation by these ciliates of the bacteria suspended in the SBR bulk liquid could help reduce
the level of suspended solids discharged in the effluent at the end of each cycle and reduce the
cost of post-treatments. More experimental work has to be done to estimate the fraction of
suspended solids removed from the bulk liquid by these ciliates.
157
Figure 9. (a)-(d) and (f) SEM images of ciliates attached to the granule surface, scale
bars = 1 mm (a), 200 µm (b), 100 µm (c) and 10 µm (d) and (f); (e) light microscope image of
a bunch of ciliates, scale bar =100 µm; TEM images of (g) oral cilia attracting free floating
bacteria in the ciliate mouth, scale bar =2 µm; (h) a ciliate cross section with internal food
vacuole indicated by the black arrows, scale bar =10 µm.
158
CONCLUDING REMARKS
The structure of aerobic granules treating nutrient-rich wastewater in a SBR was investigated.
Some interesting structural features were observed in these granules that have not been
reported in synthetic-fed granules. The particulate and colloidal matter along with the fat, oil
and grease present in abattoir wastewater appeared to have enhanced the breaking process of
large mature granules due to a rapid clogging of the granules pores and channels. The various
and complex types of substrates available also resulted in a very diverse microbial community
with specific spatial distribution throughout the granules. This diverse community is likely to
produce different sorts of EPS with different function. More importantly, the dissolution of
mineral complexes associated to the granule EPS matrix at lower pH could indeed affect the
structural stability of the granules. Further experimental studies are needed to understand the
real impact of pH on the granule stability especially in regard to dynamic pH fluctuations as
granules exhibit some in-situ pH buffer capacity. Finally the abundance of ciliates on the
surface of these wastewater-fed granules raises the question of their real contribution in
removing small particulate matter from the bulk liquid. More targeted studies on that aspect
could be beneficial for the overall aerobic granular sludge technology which is known to
produce an effluent with high levels of suspended solids.
ACKNOWLEDGEMENTS
This work was funded by the Environmental Biotechnology CRC, a Cooperative Research
Centre established and funded by the Australian Government together with industry and
university partners.
REFERENCES
Amann R.I. (1995). In situ identification of microorganisms by whole cell hybridization with
rRNA-targeted nucleic acid probes. In: Molecular Microbial Ecology Manual.
Akkermans A.D.L., van Elsas J.D. and de Bruijn F.J. (eds). Dordrecht, Holland,
Kluwer Academic Publications. 3.3.6: 1-15.
Beun J.J., Hendriks A., Van Loosdrecht M.C.M., Morgenroth E., Wilderer P.A. and Heijnen
J.J. (1999). Aerobic granulation in a sequencing batch reactor. Water Research.
33(10): 2283-2290.
Chen M.Y., Lee D.J. and Tay J.H. (2007a). Distribution of extracellular polymeric substances
in aerobic granules. Applied Microbiology and Biotechnology. 73(6): 1463-1469.
Chen M.Y., Lee D.J., Tay J.H. and Show K.Y. (2007b). Staining of extracellular polymeric
substances and cells in bioaggregates. Applied Microbiology and Biotechnology.
75(2): 467-474.
Chiu Z.C., Chen M.Y., Lee D.J., Wang C.H. and Lai J.Y. (2007). Oxygen diffusion in active
layer of aerobic granule with step change in surrounding oxygen levels. Water
Research. 41(4): 884-892.
Crocetti G.R., Banfield J.F., Keller J., Bond P.L. and Blackall L.L. (2002). Glycogen
accumulating organisms in laboratory-scale and full-scale activated sludge processes.
Microbiology. 148(11): 3353-3364.
Crocetti G.R., Hugenholtz P., Bond P.L., Schuler A., Keller J., Jenkins D. and Blackall L.L.
(2000). Identification of polyphosphate-accumulating organisms and design of 16S
159
rRNA-directed probes for their detection and quantitation. Applied and Environmental
Microbiology. 66(3): 1175-1182.
Daims H., Bruhl A., Amann R., Schleifer K.H. and Wagner M. (1999). The domain-specific
probe EUB338 is insufficient for the detection of all Bacteria: Development and
evaluation of a more comprehensive probe set. Systematic and Applied Microbiology.
22(3): 434-444.
Daims H., Nielsen J.L., Nielsen P.H., Schleifer K.H. and Wagner M. (2001). In situ
characterization of Nitrospira-like nitrite oxidizing bacteria active in wastewater
treatment plants. Applied and Environmental Microbiology. 67(11): 5273-5284.
Di Iaconi C., Ramadori R., Lopez A. and Passino R. (2004). Preliminary biomass
characterization in a sequencing batch biofilm reactor. Annali di Chimica. 94(12): 889898.
Dulekgurgen E., Ovez S., Artan N. and Orhon D. (2003). Enhanced biological phosphate
removal by granular sludge in a sequencing batch reactor. Biotechnology Letters.
25(9): 687-693.
El-Mamouni R., Leduc R. and Guiot S.R. (1998). Influence of synthetic and natural polymers
on the anaerobic granulation process. Water Science and Technology. 38(8-9): 341347.
Etterer T. and Wilderer P.A. (2001). Generation and properties of aerobic granular sludge.
Water Science and Technology. 43(3): 19-26.
Grotenhuis J.T.C., Vanlier J.B., Plugge C.M., Stams A.J.M. and Zehnder A.J.B. (1991). Effect
of ethylene glycol-bis(beta-aminoethyl ether)-N,N-tetraacetic acid (egta) on stability
and activity of methanogenic granular sludge. Applied Microbiology and
Biotechnology. 36(1): 109-114.
Ivanov V., Tay S.T.L., Liu Q.S., Wang X.H., Wang Z.W. and Tay J.H. (2005). Formation and
structure of granulated microbial aggregates used in aerobic wastewater treatment.
Water Science and Technology. 52(7): 13-19.
Jacques M. and Graham L. (1989). Improved preservation of bacterial capsule for electronmicroscopy. Journal of Electron Microscopy Technique. 11(2): 167-169.
Kong Y., Ong S.L., Ng W.J. and Liu W.-T. (2002). Diversity and distribution of a deeply
branched novel proteobacterial group found in anaerobic-aerobic activated sludge
processes. Environmental Microbiology. 4(11): 753-757.
Kong Y.H., Nielsen J.L. and Nielsen P.H. (2005). Identity and ecophysiology of uncultured
actinobacterial polyphosphate-accumulating organisms in full-scale enhanced
biological phosphorus removal plants. Applied and Environmental Microbiology.
71(7): 4076-4085.
Lemaire R., Yuan Z.G., Blackall L.L. and Crocetti G.R. (in press). Microbial distribution of
Accumulibacter spp. and Competibacter spp. in aerobic granules from a lab-scale
biological nutrient removal system. Environmental Microbiology.
Lettinga G., Vanvelsen A.F.M., Hobma S.W., Dezeeuw W. and Klapwijk A. (1980). Use of
the upflow sludge blanket (USB) reactor concept for biological wastewater-treatment,
especially for anaerobic treatment. Biotechnology and Bioengineering. 22(4): 699-734.
Lin Y.M., Liu Y. and Tay J.H. (2003). Development and characteristics of phosphorusaccumulating microbial granules in sequencing batch reactors. Applied Microbiology
and Biotechnology. 62(4): 430-435.
Liu Y. and Tay J.-H. (2004). State of the art of biogranulation technology for wastewater
treatment. Biotechnology Advances. 22(7): 533-563.
Liu Y., Yang S.F. and Tay J.H. (2003). Elemental compositions and characteristics of aerobic
granules cultivated at different substrate N/C ratios. Applied Microbiology and
Biotechnology. 61(5-6): 556-561.
160
Liu Y.Q. and Tay J.H. (2007). Characteristics and stability of aerobic granules cultivated with
different starvation time. Applied Microbiology and Biotechnology. 75(1): 205-210.
McSwain B.S., Irvine R.L., Hausner M. and Wilderer P.A. (2005). Composition and
distribution of extracellular polymeric substances in aerobic flocs and granular sludge.
Applied and Environmental Microbiology. 71(2): 1051-1057.
Meyer R.L., Saunders A.M. and Blackall L.L. (2006). Putative glycogen-accumulating
organisms belonging to the Alphaproteobacteria identified through rRNA-based stable
isotope probing. Microbiology-Sgm. 152: 419-429.
Meyer R.L., Saunders A.M., Zeng R.J.X., Keller J. and Blackall L.L. (2003). Microscale
structure and function of anaerobic-aerobic granules containing glycogen
accumulating organisms. FEMS Microbiology Ecology. 45(3): 253-261.
Mobarry B.K., Wagner M., Urbain V., Rittmann B.E. and Stahl D.A. (1996). Phylogenetic
probes for analyzing abundance and spatial organization of nitrifying bacteria. Applied
and Environmental Microbiology. 62(6): 2156-2162.
Morgenroth E., Sherden T., van Loosdrecht M.C.M., Heijnen J.J. and Wilderer P.A. (1997).
Aerobic granular sludge in a sequencing batch reactor. Water Research. 31(12): 31913194.
Revsbech N.P. (1989). An oxygen microsensor with a guard cathode. Limnology and
Oceanography. 34(2): 474-478.
Schwarzenbeck N., Borges J.M. and Wilderer P.A. (2005). Treatment of dairy effluents in an
aerobic granular sludge sequencing batch reactor. Applied Microbiology and
Biotechnology. 66(6): 711-718.
Tay J.H., Liu Q.S. and Liu Y. (2002). Characteristics of aerobic granules grown on glucose
and acetate in sequential aerobic sludge blanket reactors. Environmental Technology.
23(8): 931-936.
Tay J.H., Liu Q.S. and Liu Y. (2004). The effect of upflow air velocity on the structure of
aerobic granules cultivated in a sequencing batch reactor. Water Science and
Technology. 49(11-12): 35-40.
Wagner M., Rath G., Koops H.-P., Flood J. and Amann R. (1996). In situ analysis of
nitrifying bacteria in sewage treatment plants. Water Science and Technology. 34(1-2):
237-244.
Wang Z.W., Liu Y. and Tay J.H. (2007). Biodegradability of extracellular polymeric
substances produced by aerobic granules. Applied Microbiology and Biotechnology.
74(2): 462-466.
Yang S.F., Tay J.H. and Liu Y. (2003). A novel granular sludge sequencing batch reactor for
removal of organic and nitrogen from wastewater. Journal of Biotechnology. 106(1):
77-86.
Yilmaz G., Lemaire R., Keller J. and Yuan Z.G. (submitted). Simultaneous nitrification,
denitrification and phosphorus removal from nutrient-rich industrial wastewater using
granular sludge. Biotechnology and Bioengineering.
Zheng Y.M. and Yu H.Q. (2007). Determination of the pore size distribution and porosity of
aerobic granules using size-exclusion chromatography. Water Research. 41(1): 39-46.
161
Appendix H
Résumé détaillé en français
INTRODUCTION
L’industrie de la viande utilise de très grandes quantités d’eau lors de l’abattage et le
découpage des bêtes ainsi que pour le nettoyage des équipements. Les effluents produits sont
très chargés en DCO, azote et phosphore. Afin d’éviter toute pollution des cours d’eau
environnants, les effluents doivent être traités pour diminuer considérablement leurs teneurs
en DCO, azote et phosphore (> 95%) avant de pouvoir être rejetés dans le milieu naturel.
Durant ces vingt dernières années, l’élimination biologique de la DCO et de l’azote dans les
effluents d’abattoir a reçu beaucoup plus d’attention que l’élimination biologique du
phosphore. Cela a abouti au développement de procédés à boues activées fiables pour le
traitement en continu de la DCO et de l’azote dans ce type d’effluents. Cependant, le
phosphore continue à être principalement éliminé par l’intermédiaire de procédés chimiques
de précipitation, même si l’élimination biologique du phosphore est généralement moins
coûteuse et meilleure pour l’environnement. La haute teneur en azote des effluents d’abattoir
c’est avérée être un obstacle au développement d’un procédé fiable et résistant d’élimination
biologique du phosphore. Contrairement aux procédés de traitements des effluents en continu,
les procédés à réacteurs séquentiels discontinus, sequencing batch reactors en anglais (SBR)
permettent d’avoir une plus grande flexibilité de fonctionnement, ce qui peut être un atout
majeur pour le traitement des effluents d’abattoirs à fortes teneurs azotées et phosphorées.
D’ailleurs, en Australie, de plus en plus d’abattoirs ont opté pour le procédé SBR afin de
traiter leurs effluents sur site.
L’objectif principal de cette thèse est de développer un procédé à boues activées purement
biologique qui parvienne à réduire fortement les teneurs en DCO, azote et phosphore des
effluents d’abattoir (> 95%) afin de répondre aux contraintes de rejet en rivière de plus en plus
sévères et financièrement pénalisantes pour l’industrie. Pour ce faire, le principal obstacle à
surmonter est de parvenir à éliminer durablement le phosphore par suraccumulation
biologique dans un milieu fortement azoté. Pour pouvoir être considéré comme une réelle
alternative pour l’industrie de la viande, ce nouveau procédé SBR devra être facilement
transférable aux installations de traitement des eaux déjà présentes sur site afin d’éviter des
coûts de construction trop élevés et une opération trop complexe.
Ce projet de thèse s’intéressera aussi à l’utilisation de technologies innovantes pour améliorer
les performances du procédé SBR. Ces nouvelles technologies ou concepts sont encore au
stade de développement avec beaucoup de questions fondamentales à résoudre avant de
pouvoir être utilisées à l’échelle industrielle. Tout d’abord, le procédé de nitrification,
dénitrification et déphosphatation simultanée (SNDPR), récemment développé et validé pour
le traitement d’effluents synthétiques, sera utilisé pour traiter des effluents d’abattoir. Ce
procédé permet de réduire les coûts d’aération mais surtout de diminuer la demande en DCO
(surtout en acides gras volatils – AGV) qui est un paramètre très important pour l’élimination
du phosphore par suraccumulation biologique, souvent déficient dans les effluents d’abattoir
(faible teneur en AGV). Dans un deuxième temps, la possibilité d’éliminer la DCO, l’azote et
le phosphore en utilisant un procédé à boues activées granulaires sera étudiée. A cause de leur
162
structure dense et compacte, les boues granulaires décantent très rapidement, ce qui permet
d’avoir une concentration de boues plus élevée dans le réacteur et par conséquent de réduire le
volume du réacteur. Cette concentration élevée en boues granulaires permet au procédé de
pouvoir répondre à de fortes charges, ce qui est un grand avantage par rapport aux procédés
classiques à boues activées floculantes pour traiter des effluents industriels généralement très
concentrés. Il est également reconnu que le procédé SNDPR pourrait être plus facilement mis
en œuvre dans un système granulaire à cause d’un gradient en oxygène plus important dans
les granules que dans les flocs.
OBJECTIFS de RECHERCHE
1.
Développement d’un procédé pouvant éliminer les concentrations en DCO,
azote et phosphore contenues dans les effluents d’abattoir.
L’élimination biologique de la DCO et de l’azote dans des effluents d’abattoir a déjà été
largement étudiée, aboutissant au développement de procédés d’épuration de grande échelle.
En revanche, l’élimination biologique du phosphore dans ces mêmes effluents n’a été que peu
étudiée en raison de problèmes d’instabilité causés par une accumulation de nitrate trop élevée
lors de l’étape de nitrification. En effet, l’élimination du phosphore par suraccumulation
biologique nécessite la présence de périodes d’anaérobies (sans oxygène ni nitrate ou nitrite)
pendant lesquelles les bactéries déphosphatantes (PAO) peuvent stocker des AGV pour s’en
servir ensuite comme source d’énergie durant les périodes d’aérobies. Par conséquent, la
plupart des stations d’épuration traitant des effluents d’abattoir éliminent le phosphore par
précipitation chimique en ajoutant des ions métalliques (fer ou aluminium). Cette technique
alourdit considérablement les coûts de traitement et produit des boues à fortes teneurs
métalliques ce qui peut limiter leur valorisation agricole. Le premier objectif de cette thèse est
de développer un nouveau procédé de traitement en SBR limitant l’accumulation des nitrates
et/ou nitrites et permettant du même coup d’éliminer par voie biologique plus de 95% de la
DCO, de l’azote et du phosphore des effluents d’abattoir.
2.
Création d’un système de contrôle automatique permettant d’obtenir une
nitrification et une dénitrification via les nitrites.
La trop forte accumulation de nitrates et/ou nitrites n’est pas le seul problème limitant la
stabilité et l’efficacité du procédé d’élimination du phosphore par suraccumulation biologique
dans les stations de traitement d’effluents d’abattoir. La faible teneur en DCO facilement
biodégradable de ces effluents est aussi un problème pour arriver à éliminer les fortes
concentrations en azote et en phosphore. Le fait d’arrêter le processus de nitrification au stade
nitrite permet d’économiser de la DCO lors de la dénitrification (en théorie 40%) et
d’améliorer les performances du procédé. Des systèmes de contrôle automatique du pH et de
la concentration en oxygène dissous ont déjà été utilisés pour favoriser la nitrification et
dénitrification via les nitrites pour des effluents très concentrés en ammonium. Cependant,
l’ajout de composés carbonés facilement biodégradables est à chaque fois nécessaire durant
l’étape de dénitrification car la teneur en DCO des effluents n’est généralement pas suffisante
pour obtenir une dénitrification satisfaisante. L’utilisation de sources externes de carbone ne
permet donc pas de bénéficier pleinement des avantages du shunt des nitrates. Le second
objectif de cette thèse est de mettre en place sur le procédé SBR développé précédemment un
système de contrôle automatique pour établir un shunt des nitrates sans avoir recours à l’ajout
de composés carbonés facilement dégradables. Ce système aura pour but d’améliorer
163
l’utilisation de la DCO présente dans les effluents d’abattoir et rendra le procédé d’élimination
du phosphore par suraccumulation biologique plus efficace et plus résistant.
3.
Expérimentation d’une stratégie pour maintenir l’activité des boues durant de
longues périodes de famine ainsi que durant la phase de redémarrage du procédé.
Les stations de traitement d’effluents d’abattoir doivent s’adapter aux larges fluctuations de
débit et de composition de l’effluent d’entrée en fonction du niveau de production. Durant
certaines périodes de faible production (production saisonnière, maintenance annuelle des
équipements) aucun effluent n’arrivera en tête de station pendant des semaines voire même
des mois. Il est donc très important de pouvoir maintenir l’intégrité des boues biologiques
durant ces longues périodes d’interruption afin de s’assurer que le procédé de traitement soit
opérationnel lors du retour à une production normale. D’après le peu de recherches faites sur
ce sujet, il semble que la stratégie consistant à alterner les conditions aérobies et anaérobies
soit la plus efficace pour maintenir l’activité des boues durant de longues périodes de famine.
Cependant, ces travaux de recherche se sont seulement intéressés aux boues nitrifiantes. Cette
thèse a donc pour but de développer une stratégie de stockage des boues facilement applicable
par l’industrie qui puisse préserver les capacités nitrifiantes, dénitrifiantes et déphosphatantes
des boues lors de longues périodes d’interruption de la production et qui permette leur rapide
réactivation lors de la reprise d’une production normale.
4.
Identification des causes de la production de N2O dans le procédé SNDPR afin
de pouvoir utiliser ce procédé pour le traitement d’effluents d’abattoir.
Le procédé SNDPR présente de nombreux avantages qui pourraient être très bénéfiques pour
le traitement d’effluents d’abattoir comme une plus faible demande en DCO, moins
d’accumulation de nitrates ou nitrites durant l’étape aérobie, des coûts d’aération plus faible et
une production de boue limitée. Cependant, la production de N2O au lieu de N2 comme
produit final de la dénitrification pose un problème environnemental à cause du fort pouvoir à
effet de serre de ce gaz (300 fois plus élevé que celui du CO2). Avant de pouvoir utiliser cette
technologie pour le traitement d’effluents d’abattoir, les causes exactes de cette accumulation
et cette émission de N2O doivent être identifiées et si possible éliminées. Ceci constitue un
autre objectif de cette thèse.
5.
Etude de la distribution spatiale des populations microbiennes responsables de
la dénitrification à l’intérieur de boues granulaires.
L’utilisation de boues granulaires a la capacité à améliorer la stabilité et l’efficacité du
procédé SNDPR. La taille et la densité de ces granules aérobies sont supposées générer de
plus forts gradients d’oxygène qui seront bénéfiques pour la stabilité du procédé SNDPR en
créant des zones anaérobies au centre de chaque granule. La présence simultanée de zones
aérobies et anaérobies à l’intérieur d’un même granule entraînera vraisemblablement une
organisation des populations microbiennes en fonction du gradient d’oxygène existant dans
cette structure granulaire. Par exemple, les populations dénitrifiantes seront sans doute
localisées plutôt vers le centre du granule dépourvu d’oxygène. La réduction de la demande en
DCO grâce au procédé SNDPR implique que la dénitrification soit principalement assurée par
les populations déphosphatantes (PAO) qui utiliseront la même source de carbone pour
éliminer l’azote et le phosphore de façon simultanée. Malheureusement, il a souvent été
observé que d’autres populations ne participant pas à l’élimination du phosphore (glycogen
accumulating organisms – GAO) soient en fait responsables de la dénitrification dans le
procédé SNDPR rendant ce procédé moins attractif. Dans un procédé SNDPR à boues
164
granulaires idéal, les PAO seraient en charge de la dénitrification ce qui impliquerait qu’ils
soient plutôt localisés au centre du granule là où l’oxygène est absent. Pour vérifier les rôles
écologiques respectifs des populations de PAO et GAO dans le système SNDPR à boues
granulaires, une méthode a été développée afin d’établir les positions de chaque population à
l’intérieur des granules. Le développement de cette méthode ainsi que le rôle joué par ces
deux importantes populations dans le procédé SNDPR est un autre objectif de cette thèse.
6.
Possibilité d’éliminer la DCO, l’azote et le phosphore présents dans les
effluents d’abattoir par l’intermédiaire d’un procédé SBR aérobie à boues granulaires.
L’excellente décantation des boues granulaires et leur densité élevée permettent d’avoir une
forte concentration de boues dans un réacteur de petite taille. L’utilisation d’un tel procédé
peut donc permettre de traiter des effluents très chargés en DCO, azote et phosphore plus
efficacement par rapport aux procédés à boues floculantes conventionnels. Aujourd’hui, les
procédés à boues granulaires ont été principalement étudiés avec des effluents synthétiques.
Les quelques études utilisant des effluents réels se sont limitées à des eaux résiduaires
urbaines. La possibilité d’utiliser cette technologie pour traiter des effluents d’abattoir très
chargés en DCO, azote et phosphore est étudiée dans cette thèse. De plus, la structure des
granules ainsi développés est examinée en utilisant une grande variété de méthodes
microscopiques.
PRINCIPAUX RESULTATS
1.
Utilisation d’une stratégie d’alimentation fractionnée pour limiter l’accumulation
de nitrates et/ou nitrites dans le SBR
Pour réduire l’effet néfaste des nitrates sur la stabilité et l’efficacité du procédé d’élimination
du phosphore par suraccumulation biologique, l’alimentation du SBR avec l’effluent
d’abattoir a été répartie en 3 périodes pour chaque cycle. Après la décantation et la décharge
de l’effluent traité, le premier remplissage ne consiste qu’en 50% du volume total d’effluent à
traiter dans le cycle, et est suivi d’une période sans aération (anaérobie) puis avec aération
(aérobie). Lors du deuxième remplissage, 30% du volume total d’effluent à traiter sont
ajoutés, suivi là aussi d’une période anaérobie puis aérobie. Le reste du volume à traiter est
ajouté lors du troisième remplissage, suivi également d’une période anaérobie et aérobie.
Grâce à cette stratégie le niveau de nitrates à la fin de chaque période aérobie reste
suffisamment faible pour permettre une suraccumulation biologique du phosphore efficace et
stable. La figure 11 montre l’évolution des concentrations d’ammonium, de NOx- (nitrate +
nitrite) et de phosphates durant un cycle SBR typique. A la fin de chaque période aérobie,
l’ammonium était totalement nitrifié et les NOx- produits étaient rapidement consommés
durant la période anoxique qui suivait. Le peu de NOx- restant dans le SBR à la fin du cycle
était immédiatement consommé une fois le premier remplissage du cycle suivant achevé.
Grâce au faible niveau de NOx- durant tout le cycle du SBR, les populations déphosphatantes
(PAO) étaient très actives comme le montrent les forts relargages de phosphates durant les
périodes anaérobies et leur rapide consommation pendant les périodes aérobies. Cette stratégie
de remplissages successifs a permis de maintenir une élimination de plus de 95% de la DCO
totale en entrée, 97% de l’azote total et 98% du phosphore total.
165
anO2
18
O2
anO2
anO2
O2
O2
settle+decant
30
14
PO4-P
12
NH4-N
25
20
NOx-N
10
15
8
6
10
P (mg L-1)
N (mg L-1)
16
4
5
2
0
0
0
50
100
150
200
250
300
350
Time (min)
Figure 11. Concentrations en azote et phosphore pendant un cycle du SBR. Les flèches
indiquent les 3 périodes d’alimentation.
2.
Stratégies pour satisfaire la demande en DCO nécessaire pour éliminer les
fortes concentrations d’azote et de phosphore des effluents d’abattoir.
Les performances d’un procédé de traitement biologique pour l’élimination de l’azote et du
phosphore dépendent énormément de la DCO facilement biodégradable présente dans
l’effluent à traiter, en particulier la teneur en AGV. Pour éliminer les fortes concentrations
d’azote et de phosphore des effluents d’abattoir, il est donc primordial d’optimiser l’utilisation
de la DCO disponible. En pratique, les effluents bruts d’abattoir sont envoyés dans des
grandes lagunes anaérobies afin de réduire les teneurs en graisses ainsi que d’hydrolyser la
DCO particulaire. Malheureusement, beaucoup de DCO est abattue durant ce prétraitement,
ce qui rend l’élimination de l’azote et du phosphore difficile. La taille de ces lagunes ne
permet pas de contrôler avec précision les quantités d’AGV produites à cause de la
compétition entre les processus d’acidogénèse et de méthanogénèse. Afin de mieux contrôler
les niveaux d’AGV présents dans l’effluent d’entrée du SBR, une fraction de l’effluent brut a
été soumise à une pré-fermentation d’un jour à 37°C dans un réacteur batch de 50 L avant
d’être mélangée avec l’effluent prétraité en sortie de lagune anaérobie et introduite dans le
SBR. Aucun inoculum n’était utilisé dans notre pré-fermenteur, seules les bactéries présentes
dans l’effluent brut d’abattoir effectuèrent cette pré-fermentation où la teneur en AGV a plus
que doublé. Les acides acétique et propionique étaient les AGV les plus abondamment
produits lors de cette pré-fermentation. Cependant, l’utilisation d’une trop grande fraction
d’effluent pré-fermenté pendant 24h dans l’alimentation d’entrée du procédé SBR doit être
évitée à cause de la forte teneur en graisses de cet effluent et du risque que cela pose pour
avoir une bonne décantation des boues.
L’autre stratégie pour réduire la demande en DCO du procédé est d’utiliser le shunt des
nitrates lors de l’étape de nitrification et de dénitrification. Pour ce faire, un système de
contrôle automatique de la durée de chacune des 3 périodes aérobies du cycle SBR a été mis
en place. Ce système de contrôle est basé sur la pente du signal pH ainsi que sur la vitesse de
consommation d’oxygène (OUR). Durant chaque aération, l’instant exact où tout l’azote a été
nitrifié était déterminé grâce à la rupture de pente du profil pH et à la soudaine diminution de
l’OUR. L’aération dans le SBR pouvait donc être arrêtée automatiquement, évitant ainsi
l’oxydation des nitrites en nitrates. Au lieu d’apporter une source de carbone externe, la DCO
166
présente dans l’effluent d’abattoir était utilisée pour accomplir la dénitrification, ce qui a
permis de profiter pleinement des atouts du shunt des nitrates. La figure 12 montre la relation
entre le niveau moyen d’accumulation des nitrites dans le SBR à la fin de chacune des 3
aérations (indicateur du niveau de shunt des nitrates réalisé) et l’abondance relative des
populations microbiennes responsables de l’oxydation des nitrites en nitrates (NOB) dans les
boues du SBR. Pendant les 5 premiers mois de fonctionnement du SBR, la durée de chaque
phase d’aération n’était pas strictement contrôlée et tous les nitrites étaient oxydés en nitrates.
Après avoir d’abord ajusté manuellement la durée de chacune des 3 périodes d’aération afin
que l’aération s’arrête dès que tout l’azote était oxydé, la proportion de nitrites augmenta très
rapidement de 0% à 95% des NOx- produits et les populations NOB diminuèrent en
conséquence. Après avoir rétabli un régime d’aération plus longue que le temps nécessaire
pour avoir une totale oxydation de l’azote, le pourcentage d’accumulation des nitrites chuta et
les populations NOB augmentèrent très légèrement. Enfin, après 400 jours d’opération, le
contrôle automatique de la durée d’aération fut mis en place ce qui entraîna un retour a un
shunt des nitrates d’environ 70%. Après une période d’interruption de 6 semaines suite à la
fermeture annuelle de l’abattoir, le shunt des nitrates atteignit 85% et les populations NOB
disparurent presque complètement du SBR. Cette stratégie du contrôle de la durée d’aération a
donc permis d’atteindre un fort shunt des nitrates ce qui a fortement réduit la demande en
DCO du procédé durant l’étape de dénitrification.
N-NO2 : N-NOX ratio (%)
80
70
60
50
1.8
Start
automatic
aeration
control
90
Start
manual
aeration
control
1.6
1.4
Starvation
period
1.2
1
Start overaeration for
NOB recovery
40
0.8
0.6
30
0.4
20
% NO2
NOB population
10
0
150
200
250
300
NOB : all bacteria ratio (%)
100
0.2
350
400
450
500
550
0
600
Days
Figure 12. Accumulation de nitrite et abondance des bactéries nitratantes (NOB) du genre
Nitrospira (FISH probe Nitspa-662) dans le SBR. La quantification des NOB indiquée est une
moyenne (barres d’erreur=SE, n=3).
3.
Stratégie pour maintenir l’activité biologique des boues de traitement
d’effluents d’abattoir durant de longues périodes sans alimentation.
Le fonctionnement du SBR a du être modifié durant deux périodes de 5 à 6 semaines lorsque
l’abattoir a interrompu sa production pour cause de maintenance annuelle de ses équipements,
ce qui nous a obligé à arrêter l’alimentation du SBR. Le nouveau mode opératoire du SBR,
appelé « mode veille », consistait en seulement 15 min d’aération pour chaque cycle de 6h
(décantation le reste du temps). Durant le premier « mode veille » de 5 semaines, l’activité des
populations nitrifiantes était mesurée chaque semaine en injectant de l’ammonium puis des
nitrites et en suivant leur cinétique de consommation respective alors que le SBR était aéré
pendant 2 h. L’activité générale des populations déphosphatantes était suivie en mesurant la
167
concentration des phosphates dans le SBR plusieurs fois par semaine. Durant le second
« mode veille », de 6 semaines cette fois-ci, l’activité aérobie et anaérobie des PAO était plus
rigoureusement suivie à l’aide de tests en batch réalisés chaque semaine avec 200 ml de boues
prélevés dans le SBR. Après la réouverture de l’abattoir et le retour à une production
d’effluent normale, une stratégie de réalimentation progressive du SBR étalée sur 4 jours a été
expérimentée pour permettre aux boues de se ré-acclimater aux fortes concentrations de DCO,
d’azote et de phosphore.
Pendant ces deux « mode veille », l’activité des populations nitrifiantes a diminué de 40%
pour les AOB et de 10% pour les NOB, alors que pour les populations déphosphatantes
(PAO) elle a chuté de plus de 60%. Malgré cela, après avoir progressivement réalimenté le
SBR, les cinétiques de nitrification, de dénitrification, et de déphosphatation ont rapidement
retrouvé leur valeur initiale (voir Tableau 3). La nouvelle stratégie opératoire du SBR pendant
ces longues périodes d’interruption et lors de la réalimentation progressive, a donc permis de
maintenir les capacités nitrifiantes, dénitrifiantes et déphosphatantes des boues ainsi que de
rapidement retrouver les cinétiques d’élimination initiales.
Tableau 3. Vitesses de nitrification (rNH4+), dénitrification (rNOx-) et quantités de P relargué
et accumulé par cycle, mesurées durant des cycles réalisés avant la période de famine, juste
après le début de la période de famine (50% de la charge normale), 2 jours après (75% de la
charge normale) et 4 jours après la famine (100% de la charge normale).
Paramètres
“mode veille” I or II
Avant famine
Après 1st cycle (50%)
Après 2 days (75%)
Après 4 days (100%)
rNH4+
(mgN.l-1.h-1)
I
II
18.2 25.5
8.2
7.4
12.9 20.8
17.6 29.1
rNOx(mgN.l-1.h-1)
I
II
4.8
12.3
1.9
1.8
4.5
9.6
5.7
11.7
Relargage de P
(mgP.l-1)
I
II
18.8
36.9
2.5
4.4
9.6
31
19.6
47
Accumulation de P
(mgP.l-1)
I
II
16.1
34.2
2.3
4.8
8.1
28.9
17.4
44.3
4.
Identification des principaux microorganismes dénitrifiants dans des
bioréateurs SNDPR à boues floculantes et granulaires traitant des effluents
synthétiques
Les analyses microbiennes (FISH) faites sur les boues floculantes du bioréacteur SNDPR
alimenté avec un effluent synthétique ont montré une forte abondance de bactéries
Accumulibacter (PAO) et Competibacter (GAO) représentant environ 70% de la totalité des
bactéries dans les boues. La figure 13 indique que les PAO étaient toujours plus nombreux
que les GAO dans le SBR. Il est intéressant de noter que le taux d’élimination de l’azote a
diminué de 100% à 53% pendant que le pourcentage de GAO dans les boues est passé de 19%
à 8%. Durant cette même période, le pourcentage de PAO a lui augmenté de 48% à 70%. Ces
résultats semblent suggérer que les GAO, et non pas les PAO, étaient responsables de la
dénitrification dans le SBR. Cela pose donc la question de savoir s’il y a un lien entre cette
apparente dénitrification par les GAO et la production de N2O souvent constatée dans les
procédés SNDPR à boues floculantes traitant des effluents synthétiques.
168
100
0.75
0.70
80
0.65
60
0.60
0.55
40
P/C (mol/mol)
% of all bacteria or % N removed
Accumulibacter
Competibacter
N removal
P release/VFA uptake
0.50
20
0.45
0
0.40
Jun 7
Jul 12
Aug 23
Sep 20
Oct 25
Figure 13. Abondance de Accumulibacter et Competibacter corrélée à l’élimination de l’azote
et au rapport carbone accumulé:phosphate relargué pendant les 5 mois d’expérience.
Pour essayer de déterminer à l’échelle microbienne quelle était, entre les PAO et les GAO, la
population la plus apte à dénitrifier dans un procédé SNDPR traitant des effluents
synthétiques, on a comparé, mais cette fois à l’aide de boues granulaires, la distribution
spatiale des PAO et GAO à l’intérieur des granules ainsi que le gradient d’oxygène existant
dans ces mêmes granules. Pour pouvoir établir avec précision la distribution de ces deux
populations dans la structure granulaire, une nouvelle méthode utilisant la technique FISH et
la microscopie laser confocal (CLSM) a donc été développée durant cette thèse. Les gradients
d’oxygène à l’intérieur des granules ont eux été mesurés à l’aide de microsondes à oxygène.
La distribution des PAO et GAO a été déterminée en analysant une fine section médiane de 24
granules différents. Pour chaque granule étudié, le pourcentage relatif de PAO était divisé par
celui des GAO pour chaque niveau de profondeur (tous les 50 µm) à partir de la surface du
granule. En utilisant ce ratio PAO/GAO obtenu pour chaque zone, une distribution moyenne a
été calculée et est présentée dans la figure 14 ainsi que la concentration en oxygène mesurée
tous les 50 µm en partant de la surface du granule. Une forte corrélation a été établie entre le
pourcentage de PAO et la concentration en oxygène. Les PAO étaient beaucoup plus
nombreux que les GAO dans la zone aérobie des granules (0-200 µm) alors que les GAO
étaient dominateurs dans la zone centrale anaérobie (à partir de 200 µm de profondeur). A
cause de cette prédominance des GAO dans la partie du granule dépourvu d’oxygène, il
apparaît donc comme très probable que les GAO soient les bactéries dénitrifiantes dans ce
procédé SNDPR granulaire, ce qui va dans le même sens que les résultats obtenus avec des
boues SNDPR floculantes. Alors que cette dénitrification par les GAO compromet les
économies de DCO du procédé SNDPR, il est possible que l’utilisation d’un effluent réel à la
place d’un effluent synthétique puisse atténuer le rôle dénitrifiant des GAO.
169
PAO/GAO ratio
3.0
PAO/GAO ratio
O2 profile
-1
0.5
O2 concentration (mg l )
0.6
3.5
2.5
0.4
2.0
0.3
1.5
0.2
1.0
0.1
0.5
0.0
0.0
0
50
100
150
200
250
300
350
400
Depth (µm)
Figure 14. Profil moyen du rapport PAO:GAO dans 24 granules (bar d’erreur=95%CL) et
profil moyen de la concentration en O2 dans les granules à la fin de la période aérobie (barre
d’erreur=S.D., n=6).
5.
Gestion de l’accumulation de N2O dans le procédé SNDPR
Afin d’identifier les raisons de l’accumulation de N2O observée dans le procédé SNDPR
traitant des effluents synthétiques et essayer d’y remédier, plusieurs expériences en batch ont
été réalisées sous différentes conditions. Pour chaque expérience, des boues étaient prélevées
du SBR et transvasées dans deux petits flacons de 15 ml fermés hermétiquement avec un
bouchon en caoutchouc à travers lequel une microsonde à N2O était introduite pour suivre la
concentration en N2O dissous en continu. Les flacons étaient mélangés continuellement avec
un barreau magnétique et ne contenaient aucune poche d’air afin d’éviter tout transfert de N2O
de la phase liquide vers la phase gazeuse. Chaque expérience était faite en parallèle dans les
deux flacons avec un des flacons utilisé comme témoin. Durant chaque expérience, de faibles
volumes de substrat pouvaient être injectés dans chaque flacon par l’intermédiaire d’une
seringue. Les vitesses de production et de consommation du N2O lors de l’étape de
dénitrification pouvaient ainsi être mesurées en utilisant différents accepteurs d’électrons
(nitrate, nitrite et N2O) et plusieurs sources de carbone (réserves de carbone intracellulaire,
acétate, propionate, méthanol et des effluents d’abattoir).
Une des hypothèses proposées pour éviter l’accumulation de N2O est que si la dénitrification
dans le procédé SNDPR n’était pas uniquement supportée par les GAO comme vu
précédemment mais aussi par d’autres populations dénitrifiantes, l’excès de N2O produit par
ces GAO lors de la dénitrification pourrait donc être consommé par d’autres organismes
dénitrifiants, pourvu que suffisamment de DCO soit encore disponible pendant la période
aérobie du procédé SNDPR. Pour vérifier cette hypothèse, des expériences batch
supplémentaires ont été réalisées en mélangeant cette fois les boues du procédé SNDPR avec
celles d’un autre procédé éliminant l’azote d’effluents urbains par nitrification et
dénitrification avec ajout de méthanol. La figure 15 montre que l’addition d’effluent d’abattoir
après 35 min d’incubation avec des nitrates a permis aux boues du réacteur traitant un effluent
urbain de consommer immédiatement tout le N2O accumulé par les boues provenant du
procédé SNDPR.
170
L’accumulation de N2O observée dans les boues du procédé SNDPR est sûrement le résultat
de la très forte abondance des PAO et GAO dans le SBR due à l’utilisation d’un effluent
synthétique contenant leur source de carbone préférée (acide acétique) comme unique
composé organique. Si ce procédé SNDPR était alimenté avec un effluent d’origine
domestique ou industrielle, il est très probable que des organismes dénitrifiants autres que les
PAO ou GAO pourraient participer à la dénitrification en utilisant la multitude de composés
organiques présents dans cet effluent. Il est donc vraisemblable que l’accumulation de N2O
observée lors du traitement d’effluents synthétiques à une seule source de carbone par le
procédé SNDPR ne soit plus observée lors du traitement d’effluents réels contenant une large
diversité de composés organiques.
0.6
add high-strength WW
10
N2O-N
NOx-N
0.4
0.3
8
6
4
0.2
N-NOx (mg l-1)
N-N2O (mg l-1)
0.5
2
0.1
0
0
0
20
40
60
80
time (min)
Figure 15. Concentrations en N2O et NOx- dans la phase liquide pendant un test anoxique
dans un réacteur de 500 ml. L’eau usée brute chargée a été ajoutée à T=35 min.
6.
Elimination de la DCO, de l’azote et du phosphore des effluents d’abattoir à
l’aide de boues granulaires
Les charges en DCO, azote et phosphore appliquées dans le SBR granulaire étaient de
2.7 gDCO.l-1.j-1, 0.43 gN.l-1.j-1 et 0.06 gP.l-1.j-1. Les rendements d’épuration obtenus après
avoir atteint un régime stationnaire étaient de 89% pour la DCO soluble, 93% pour l’azote
dissous total et 88% pour le phosphore dissous total. Le reste de la DCO soluble dans
l’effluent de sortie (162 mg.l-1) était considérée non biodégradable à cause de la très faible
valeur de la DBO5 soluble mesurée (<2 mg.l-1). A la fin de chaque cycle SBR environ
10 mgN.l-1 de NOx- restait dans l’effluent principalement sous la forme de nitrites (95%)
signifiant que l’azote était vraisemblablement éliminé via le shunt des nitrates. Cependant, la
forte concentration des matières en suspension de l’effluent en sortie du SBR (autour de
0.3 g.l-1) limita les taux d’épuration de la DCO totale, de l’azote total et du phosphore total à
68%, 86% et 74% respectivement. Il est intéressant de noter que les GAO n’étaient pas
présents dans ces boues granulaires et que les PAO étaient responsables de la dénitrification.
Cela montre que le véritable procédé SNDPR a pu être établi dans ce SBR à boues granulaires
traitant des effluents d’abattoir. La structure interne de ces boues granulaires a également été
étudiée à l’aide de microscopes à balayage et à transmission électronique, de microsondes à
oxygène et à pH et de techniques de microbiologie (FISH).
171
CONCLUSION
Un procédé SBR (réacteur discontinu à alimentation séquentielle) a été développé pour
éliminer efficacement les fortes teneurs en DCO, azote et phosphore des effluents d’abattoir.
Ce procédé biologique offre une vraie alternative par rapport aux procédés chimiques
conventionnels utilisés pour éliminer le phosphore dans ces effluents, que ce soit au niveau du
coût ou de l’impact sur l’environnement. L’utilisation de ce nouveau procédé par l’industrie
de la viande est donc rapidement envisageable et relativement facile à mettre place. Ce
procédé est d’ailleurs en train d’être testé à l’échelle pilote (10 m3) dans un abattoir australien.
Les principales conclusions de la recherche effectuée sur ce procédé durant cette thèse sont :
•
•
•
•
•
Ce procédé SBR peut traiter de façon biologique des effluents d’abattoir à forte teneur en
azote (200 – 300 mgN.l-1) et d’obtenir des taux épuration supérieurs à 95% pour la DCO
totale, l’azote total et le phosphore total.
La stratégie d’alimentation fractionnée a permis de réduire l’accumulation des nitrates et
nitrites dans le SBR rendant possible l’élimination du phosphore par suraccumulation
biologique.
Il est important de faire passer une fraction de l’effluent d’abattoir brute dans un préfermenteur en amont du procédé SBR afin d’augmenter sa teneur en AGV. Cela
améliorera la stabilité du procédé SBR ainsi que ses taux d’épuration.
L’arrêt automatique du système d’aération du SBR en fonction du pH dès que tout
l’ammonium était oxydé a permis d’éviter la production de nitrates. Grâce à ce shunt des
nitrates, l’utilisation de la DCO présente dans l’effluent d’abattoir a été optimisée pour
obtenir une élimination de l’azote et du phosphore plus stable et plus poussée.
Pour maintenir l’activité biologique des boues lors de longues périodes d’interruption et
faciliter le redémarrage du procédé, la stratégie consistant à aérer les boues pendant 15
minutes toutes les 6h a été très efficace. Après avoir utilisé cette stratégie pendant plus de
5 semaines alors que l’alimentation du SBR avait été arrêtée, le retour aux cinétiques
d’épuration initiales a pris seulement 4 jours.
Deux autres nouveaux procédés très prometteurs pour le traitement biologique des effluents
industriels chargés ont été étudiés. Le procédé de nitrification, dénitrification et
déphosphatation simultanées (SNDPR) et le procédé à boues granulaires aérobies. Les
principales conclusions de ces deux études sont les suivantes :
•
•
•
Les GAO sont principalement responsables de la dénitrification dans le procédé SNDPR à
boues floculantes traitant des effluents synthétiques. L’absence de dénitrification par les
PAO compromet sérieusement la réduction de la demande en DCO supposée être réalisée
avec ce procédé lors de l’élimination de l’azote et du phosphore.
La production de N2O dans des SBR traitant des effluents synthétiques par le procédé
SNDPR est très probablement liée à la présence d’une seule source de carbone (acétate)
dans cet effluent. L’émission de gaz N2O ne sera vraisemblablement plus un problème lors
du traitement d’effluents d’abattoir à cause de la diversité des sources de carbone
présentes.
La taille, la densité et l’activité microbienne des granules aérobies contribuent à la
présence d’un fort gradient d’oxygène à l’intérieur des granules. L’existence de larges
zones anaérobies au centre de chaque granule permet d’obtenir un procédé SNDPR plus
stable grâce à un meilleur couplage entre les zones aérobies pour la nitrification et celles
172
•
anaérobies pour la dénitrification. Cependant, les GAO étaient toujours responsables de la
dénitrification dans ce procédé SNDPR à boues granulaires traitant des effluents
synthétiques.
Des taux d’épuration élevés peuvent être obtenus lorsque des effluents d’abattoir sont
traités avec des boues granulaires. Le procédé SNDPR a pu être employé avec cette fois
les PAO responsables de la dénitrification permettant enfin au procédé SNDPR de
développer tout son potentiel.
173
RESUME
L’industrie de la viande utilise de larges quantités d’eau lors de l’abattage, du découpage et du nettoyage des
équipements. Les effluents produits sont très chargés en DCO, azote et phosphore. Afin d’éviter toute pollution
des milieux aquatiques environnants, ces effluents doivent subir des traitements poussés. Le but principal de
cette thèse était de développer un procédé de traitement par boues activées qui puisse éliminer plus de 95% de la
DCO, de l’azote et du phosphore dans les effluents d’abattoir permettant alors un rejet direct de l’effluent traité
en rivière. La forte teneur en azote des effluents d’abattoir est l’obstacle majeur empêchant d’établir une
élimination biologique du phosphore stable et efficace. Durant cette thèse, un procédé biologique capable
d’éliminer 98% du phosphore tout en abattant 95% de la DCO et 97% de l’azote a été développé dans un réacteur
batch à alimentation séquentielle (SBR). Par rapport aux procédés chimiques classiques d’élimination du
phosphore, ce nouveau procédé biologique offre une vraie alternative financière et environnementale pour
l’industrie de la viande. Une stratégie d’alimentation séquentielle a permis de réduire l’accumulation des nitrates
dans le SBR rendant ainsi possible l’élimination biologique du phosphore. Cette thèse aborde aussi l’étude et
l’utilisation de technologies innovantes pour améliorer les performances du procédé SBR. Le procédé de
nitrification, dénitrification et déphosphatation simultanées (SNDPR) a été incorporé au procédé à boues
granulaires aérobies. La taille, la densité et l’activité microbienne des granules aérobies génèrent de forts
gradients d’oxygène à l’intérieur des granules, permettant alors d’obtenir un procédé SNDPR plus efficace. Le
volume du réacteur et la demande en DCO nécessaire pour éliminer l’azote et le phosphore dans les effluents
d’abattoir ont ainsi pu être fortement réduits. La structure interne et la composition microbienne de ces granules
ont également été étudiées.
Development and Fundamental Investigations of Innovative Technologies for
Biological Nutrient Removal from Abattoir Wastewater.
ABSTRACT
The meat processing industry requires large quantities of water, much of which is discharged as wastewater
containing high levels of COD and nutrients such as nitrogen (N) and phosphorus (P). These nutrients must be
removed to very low levels before the wastewater can be discharged into local waterways to avoid causing
eutrophication. The aim of this thesis was to develop a biological process that could achieve more than 95% of
COD, N and P removal from abattoir wastewater, producing an effluent suitable for direct discharge into river
systems. The main challenge is to achieve stable and reliable biological P removal in nitrogen-rich wastewater. A
sequencing batch reactor (SBR) system was demonstrated to effectively remove 95%, 97% and 98% of the total
COD, total N and total P present in abattoir wastewater. It could provide a more cost-effective and
environmentally friendly alternative to chemical P removal, which is the common practice in the meat industry at
present. A multi-step feeding strategy was employed to prevent the accumulation of nitrate or nitrite in the SBR
providing the right condition for the development of a stable biological P removal. An automatic aeration length
control strategy was developed and demonstrated to remove N via the nitrite pathway which benefited the
nutrient removal performance of the SBR by reducing the amount of readily biodegradable COD required. This
study also investigated the feasibility of using two innovative technologies to further enhance the performance of
the SBR system. The simultaneous nitrification, denitrification and phosphorus removal (SNDPR) process and
the aerobic granular sludge technology were successfully combined in a single SBR process. The size and the
dense structure of aerobic granules positively contributed to the oxygen mass transfer limitation required to
achieve reliable SNDPR. The structure and function of these granules fed with nutrient-rich wastewater were
closely investigated using a wide range of microbial and micro-scale techniques, which yielded insightful
information about aerobic granules and provided support for future in-depth studies on the mechanisms involved
in aerobic biogranulation.
DISCIPLINE: Génie des Procédés – Process Engineering
MOTS-CLES: Boues granulaires aérobies, Dénitrification, Effluent d’abattoir, Elimination biologique de
l’azote et du phosphore, Nitrification, Populations microbiennes, Réacteur séquentiel discontinu, Shunt des
nitrates.
KEYWORDS: Aerobic granular sludge, Denitrification, Abattoir wastewater, Biological nutrient removal,
Nitrification, Microbial communities, Sequencing batch reactor, Nitrite pathway.
INTITULE ET ADRESSE DES LABORATOIRES:
Laboratoire de Biotechnologie de l’Environnement (LBE), Avenue des Etangs, Narbonne, 11100, France.
Advanced Water Management Centre (AWMC), The University of Queensland, 4072 QLD, Australie.