By Guest On 30 August 2023

©2021 The Author(s)
This is an Open Access book distributed under the terms of the Creative Commons
Attribution-Non Commercial-No Derivatives Licence (CC BY-NC-ND 4.0), which
permits copying and redistribution in the original format for non-commercial
purposes, provided the original work is properly cited.
(http://creativecommons.org/licenses/by-nc-nd/4.0/). This does not affect the rights
licensed or assigned from any third party in this book.
This title was made available Open Access through a

partnership with Knowledge Unlatched.
IWA Publishing would like to thank all of the libraries for

pledging to support the transition of this title to Open Access
through the 2020 KU Partner Package program.
Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

by guest
Tuning
B i o lo g i c a l
nuTrienT
r e m o va l P la n T s
K e n H a rt ley

by guest
Tuning Biological Nutrient
Removal Plants

by guest
Tuning Biological Nutrient
Removal Plants
Ken Hartley

by guest
Published by IWA Publishing
Alliance House
12 Caxton Street
London SW1H 0QS, UK
Telephone: +44 (0)20 7654 5500
Fax: +44 (0)20 7654 5555
Email: [email protected]
Web: www.iwapublishing.com
First published 2013

© 2013 IWA Publishing
Cover Illustration: One of two annular oxidation ditch-MBRs at the 126,000 EP Cleveland Bay Wastewater Treatment Facility
in Townsville, Queensland, photographed during startup in late 2007. Effluent discharges to the Great Barrier Reef Marine
Park and average effluent limits are TN 4 mg/L, TP 0.8 mg/L. Three existing secondary clarifiers retained from the old plant
were converted into a primary clarifier and two MBRs. A three-cell anaerobic reactor and four hollow fibre membrane trains
are located in a covered stainless steel centre section, surrounded by an annular oxidation ditch. More information about the
plant can be found in Section 6.4.3. Photograph courtesy of Townsville Water.
Apart from any fair dealing for the purposes of research or private study, or criticism or review, as permitted under the UK
Copyright, Designs and Patents Act (1998), no part of this publication may be reproduced, stored or transmitted in any
form or by any means, without the prior permission in writing of the publisher, or, in the case of photographic
reproduction, in accordance with the terms of licenses issued by the Copyright Licensing Agency in the UK, or in
accordance with the terms of licenses issued by the appropriate reproduction rights organization outside the UK.
Enquiries concerning reproduction outside the terms stated here should be sent to IWA Publishing at the address printed
above.
The publisher makes no representation, express or implied, with regard to the accuracy of the information contained in this
book and cannot accept any legal responsibility or liability for errors or omissions that may be made.
Disclaimer
The information provided and the opinions given in this publication are not necessarily those of IWA and should not be acted
upon without independent consideration and professional advice. IWA and the Author will not accept responsibility for any
loss or damage suffered by any person acting or refraining from acting upon any material contained in this publication.
British Library Cataloguing in Publication Data

A CIP catalogue record for this book is available from the British Library
ISBN 9781780404820 (Paperback)
ISBN 9781780404837 (eBook)

by guest
Contents
About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Chapter 1
What is tuning? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Chapter 2
Influent and effluent characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 The Catchment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Flow Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Mass Loading Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3.1 Connected population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.2 Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.3 COD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.4 SS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.5 pH and alkalinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.6 Dissolved sulfide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.7 TDS or conductivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3.8 Pathogens . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.9 Load ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.10 COD/BOD5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3.11 Nutrient ratios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

by guest
vi Tuning Biological Nutrient Removal Plants
2.3.12 Total COD fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.3.13 Soluble unbiodegradable organic N . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.3.14 Nitrifier maximum specific growth rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.15 Temperature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.3.16 Energy content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.17 Other sewage components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Sewer Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.5 Effect of Primary Treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Effluent Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Chapter 3
Biological nutrient removal – process fundamentals . . . . . . . . . . . . . . . . . . . . . . . . 27
3.1 The Basic Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Kinetics of Biological Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.1 Substrate utilisation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.2.2 Substrate concentration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 Solids Retention Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.1 Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.3.2 Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4 Carbon Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4.1 Effect of SRT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.4.2 Loading conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.3 F:M ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.4.4 Mixed liquor suspended solids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.4.5 Nutrient requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.4.6 Carbonaceous oxygen demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.5 Nitrogen Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5.1 The nitrogen cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5.2 Nitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5.3 Denitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3.5.4 Simultaneous nitrification and denitrification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.5.5 Total process oxygen demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.6 Phosphorus Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.6.1 Mechanisms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.6.2 Process characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.6.3 Adverse factors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.7 Supplementary Chemical Dosing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.7.1 Phosphorus removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.7.2 Nitrogen removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.7.3 Other effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.8 Alkalinity and pH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.8.1 Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.8.2 Process behaviour . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.9 Rules of Thumb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

by guest
Contents vii
Chapter 4
Sludge settleability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.1 Characterising Settleability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.2 Settling Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.3 Factors Affecting Settleability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.1 Non-BNR activated sludge processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.2 BNR processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.4 Effect of SVI on Effluent Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
4.5 Foaming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Chapter 5
BNR flowsheets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.1 Nitrogen Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.1.1 Modified Ludzack-Ettinger (MLE) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.1.2 Bardenpho . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.2 Phosphorus Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.2.1 Phoredox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3 Nitrogen and Phosphorus Removal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.1 Modified Bardenpho . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.2 Johannesburg (JHB) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.3 University of Cape Town (UCT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.4 Modified UCT (MUCT) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.5 Westbank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.3.6 Membrane Bioreactor (MBR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Chapter 6
Reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1 Mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.1 Mixing regime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.1.2 Mixing intensity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.2 Aeration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.2.1 Oxygen transfer rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
6.2.2 Oxygen transfer efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.2.3 Carbon dioxide stripping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6.3 Compartmentalised Reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.2 Process characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
6.3.3 Operating principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.4 Oxidation Ditches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.4.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6.4.3 Hydraulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
6.4.4 Mechanical aerator characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

by guest
viii Tuning Biological Nutrient Removal Plants
6.5 Membrane Bioreactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

6.5.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
6.6 Sequencing Batch Reactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
6.6.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Chapter 7
Secondary clarifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.1 Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
7.2 Mass Balances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.3 Operating Diagrams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
7.4 Effluent Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.4.1 Factors affecting effluent SS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.4.2 Other effects of clarifiers on effluent quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
7.5 Clarifier Stress Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Chapter 8
Sludge processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8.2 Aerobic Digestion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
8.2.1 Stabilisation performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.2 Operating characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.2.3 Operating examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
8.3 Sludge Dewatering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.3.1 Mass balances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
8.3.2 Belt filter press . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
8.3.3 Centrifuge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Chapter 9
Plant characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
9.1 Mass Balances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
9.2 Operational Process Capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
9.2.1 Bottlenecks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
9.2.2 Failure curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
9.3 Capacity Envelopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
9.3.1 Continuous flow process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
9.3.2 SBR process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162

by guest
Contents ix
9.4 Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

9.4.1 Water cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
9.4.2 Plant energy balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
9.4.3 Energy models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
9.4.4 Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Chapter 10
Process control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
10.1 The Role of Control and Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
10.2 Disturbances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
10.3 Control Priorities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
10.3.1 The system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
10.3.2 The importance of dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
10.3.3 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
10.4 Feedback Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
10.4.1 Open and closed loop control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
10.4.2 Low level control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
10.4.3 Integral action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
10.4.4 Derivative action . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
10.4.5 Computer realisation of the PID controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
10.4.6 When is PID control appropriate? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
10.4.7 Controller tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
10.4.8 Manual PI tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
10.4.9 Ziegler-Nichols frequency response method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
10.4.10 Autotuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
10.5 Instrumentation and Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
10.5.1 Instruments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183
10.5.2 Estimating the oxygen uptake rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
10.6 Pumping . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
10.6.1 Pump characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
10.6.2 Relationship between flow rate and power . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
10.7 Essential Control Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
10.7.1 Dissolved oxygen control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
10.7.2 Ammonia-Based DO control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
10.7.3 Recycle flow controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
10.7.4 Chemical precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
10.7.5 Anaerobic digestion process control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
10.8 Minimising Operating Cost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
10.8.1 Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
10.8.2 The human factor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
10.9 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 196

by guest
x Tuning Biological Nutrient Removal Plants
Chapter 11
Evolutionary operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
11.1 Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
11.2 Application to Sewage Treatment Plants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
11.2.1 Plant improvement strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198
11.2.2 Learning curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
11.2.3 Startup phase . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
11.3 Long-Term Improvement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
11.3.1 Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
11.3.2 Tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
11.3.3 Stress testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
11.3.4 Debottlenecking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
11.3.5 Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
11.4 EVOP Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
11.4.1 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
11.4.2 Trend plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
11.4.3 General principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
11.5 Faux EVOP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Appendix 1
A note on the statistics of variability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Average and Median . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Percentiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Appendix 2
Chemicals: useful properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Appendix 3
Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241

by guest
About the Authors
Ken Hartley
Ken Hartley is a water and wastewater process engineer with over forty years experience in ten countries. He
has worked for the South Australian Engineering and Water Supply Department (now the South Australian
Water Corporation), consulting engineers GHD and The University of Queensland. Since 1998 he has been
an independent consulting engineer and since 2005 an adjunct professor in chemical engineering at The
University of Queensland.
Ken has authored over fifty published papers and two previous books, Operating the Activated Sludge
Process and Chemistry of Chlorinated Swimming Pools.
Gustaf Olsson
Gustaf Olsson has contributed Chapter 10. Gustaf is professor in industrial automation and since 2006
professor emeritus at Lund University, Sweden. He has devoted his research to control and automation
in water systems, electrical power systems and industrial processes. His research has lately been focused
on the water-energy nexus.
Gustaf has spent extended periods as a guest professor and visiting researcher at other universities and
companies in Sweden, UK, USA, Australia, Japan, Malaysia and China. He has been invited as guest
lecturer and speaker in another thirteen countries. Gustaf has authored eight books published in various
languages and contributed chapters to another seventeen including this one. He has over 140 scientific
publications and has served in various positions within the International Water Association.

by guest
Acknowledgements
Over the years, the author has had the opportunity to mess about in a large number of interesting treatment
plants. Plant owners have generously allowed previously unpublished insights to be incorporated in this
book for the benefit of others. Their support is gratefully acknowledged as follows:
ACTEW (ACT Electricity and Water), Canberra. For allowing operating data from their Lower Molonglo
plant to be included in Chapter 7.
Byron Shire Council, northern NSW. For allowing operating data from their West Byron and Brunswick
Valley plants to be included in Chapters 3, 4, 7, 8, 9 and 11.
Logan Water, southeast Queensland. For allowing operating data from their Flagstone, Jimboomba and
Loganholme plants to be included in Chapters 4 and 7.
Melbourne Water Corporation. For allowing operating data from their Eastern plant to be included in
Chapter 7.
Queensland Urban Utilities, Brisbane. For allowing energy consumption data from their Gibson Island and
Wacol plants to be included in Chapter 9.
Redland Water, southeast Queensland. For allowing operating data from their Thorneside plant to be
included in Chapters 9 and 11.
South Australian Water Corporation. For allowing influent data from their Port Pirie plant to be included in
Chapter 2.
Townsville Water, north Queensland. For allowing operating data from their Cleveland Bay plant to be
included in Chapters 2, 3, 6 and 9 and Appendix 1; and for permission to use the cover photograph.
Unitywater, southeast Queensland. For allowing operating data from their Coolum, Maroochydore,
Nambour, Redcliffe and Woodford plants to be included in Chapters 2, 3, 4, 6, 8, 9 and 11.
Water Corporation, Perth. For allowing the results of a study of the effect of sample storage time on BOD
and COD results to be included in Chapter 2.

by guest
Preface
This is a book about understanding and tuning sewage treatment plants. I wrote it for two people – you and
me. For myself because I wanted to record my learnings and insights for future reference; for you because I
wanted to share my insights so you could use them and build on them. The original title of the book was The
Quintessential Operator because I believe tuning is the quintessence of plant operation. However IWA
Publishing pointed out that you would be less likely to find the book on the internet if the keywords
were not included in the title. So the sub-title became the title.
The intent is not to cover all processes but to focus on the key components of most BNR plants and
discuss the approach to or philosophy of tuning for maximum performance and capacity. The underlying
tuning principles – based on an understanding of the behavioural characteristics of processes and their
interactions – are universally applicable.
We are indeed fortunate that Gustaf Olsson has been willing to contribute Chapter 10 on automatic
process control, an area in which I am unknowledgeable. Now that he has joined forces he may feel
obliged to answer our further queries!1
We are also indebted to the many Australian plant owners who have generously permitted their plant data
to be incorporated – they are acknowledged individually in Acknowledgements.
The philosophy of this book is Evolutionary Operation (EVOP)2 applied to sewage treatment plants – a
process of ongoing improvement.3 The capacity and performance capabilities of a plant are not fixed; both
are amenable to ongoing enhancement through systematic and enthusiastic effort. I hope this book will
encourage sewage treatment enthusiasts to lift their sights and set new benchmarks.
Ken Hartley
1
You’ll enjoy Gustaf in: Olsson G. (2012). ICA and me - a subjective review. Water Research, 46(6), 1585–1624.
2
Box G. E. P. and Draper N. R. (1969). Evolutionary Operation: A Statistical Method for Process Improvement. John Wiley.
3
Goldratt E. M. and Cox J. (2004). The Goal: A Process of Ongoing Improvement. 3rd revised ed. North River Press, Great Barrington,
Massachusetts, USA.

by guest
Chapter 1
What is tuning?
Process tuning is the quintessence of plant operation. Tuning is the art of striving to achieve and maintain
peak plant performance, or to achieve maximum plant capacity, as defined by a set of operating goals.
Tuning increases both performance and capacity by testing and pushing out the operating capacity
envelope of the plant.
Sewage treatment plants1 have a special role in the nutrient cycle that sustains life on Earth. Sewage
treatment provides controlled decomposition of the concentrated human wastes produced in our towns
and cities, recycling the nutrients without overloading the local environment. This is illustrated in the
simplified nutrient cycle shown in Figure 1.1. Currently the main purpose of treatment is to purify the
sewage, however in the future direct nutrient recovery and recycling will become more important.
The core processes used in sewage treatment plants are adaptations of the natural nutrient cycle. Process
behaviour is subject to climatic influences and the natural performance variability of all biological
systems. The process loading (sewage flow and organic loading) is also variable because of seasonal,
diurnal, social and statistical influences. Treatment plants therefore need to be continuously optimised to
maintain best performance under variable loading conditions.
Conscious effort applied to any area of human activity results in improvement in performance. Such
improvements have been demonstrated repeatedly in construction, manufacturing and chemical
processing operations and the curves of gradually increasing production rate, improving performance and
decreasing cost are called learning curves. In the process industries this has been termed performance
and capacity creep (Brennan & Greenfield, 1994).
A valuable technique for maximising the rate of learning is a method of routine operation called
evolutionary operation, or EVOP (Box & Draper, 1969). This is a structured method for deliberate
learning based on the philosophy “that it is inefficient to run an industrial process in such a way that
only a product is produced, and that a process should be operated so as to produce not only a product
but also information on how to improve the product.” EVOP is a “tool in which a continuous
investigative routine becomes the basic mode of operation for the plant and replaces normal static
operation” (ibid, p5). A similar process of ongoing improvement in manufacturing plants is the subject
of an absorbing novel by Goldratt and Cox (2004).
1
I prefer the original name sewage to wastewater. Wastewater connotes uselessness whereas sewage is a resource containing pure water,
nutrients and energy which are amenable to recovery and reuse. From the multitude of plant names in use, I usually choose to use sewage
treatment plant (STP).

by guest
2 Tuning Biological Nutrient Removal Plants
F O2, CO2, N2 IN AIR &

IR O WAT
E RVO ER
RES
LIVING
ANIMAL
MATTER
2 Proteins
CO
W
as
Fats
te
w l
Pr
ro a
G nim
th
od
uc
People
A
tio
io
as
n
pt
te
2
&
m
O
Pr
D
su
ea
od
on
th
uc
C
tio
od
n
Fo
DEAD
LIVING PLANT
ORGANIC
MATTER Decay & Death
MATTER
Carbohydrates Food Sewage
C
Proteins
N
Fats
P
t
en
Fo
tm
od
ea
Ni
Pr
Tr
tro
2
od
O
e
M ,P
uc
ge
ag
n
an
tio
Pl
N2
tio
w
nF
an
uf
Se
si
Effluent
n
ac
po
tG
ixa
2
O
Biosolids
tu
om
ro
C
re
tio
ec
d
th
n
PRODUCTS OF
2
N
DECOMPOSITION
O
2 CO2
Ph CO NO3
oto 2 PO4
sy
nth
SOLAR ENERGY es
is
Figure 1.1 Simplified major nutrient cycle (after Fair et al., 1968).
The basic tuning strategy is illustrated in Figure 1.2. The steps are as follows:
(1) Define goals and set targets.
(2) Set the plant control variables.
(3) Measure the outcomes.
(4) Interpret the data and develop ideas for improvement.
(5) Re-adjust the process.
(6) Conduct trials and tests on specific processes or the whole plant.
These steps form the basis of an ongoing EVOP strategy which will produce long-term benefits. Tuning
transcends troubleshooting in its usual connotation of major unexpected problems. Close monitoring and
interpretation of plant behaviour gives early warning of undesirable trends and the need for corrective action.

by guest
What is tuning? 3
CONDUCT
TRIALS
ADJUST
PROCESS
OPERATING
GOALS
INTERPRET MEASURE
DATA PERFORMANCE
Figure 1.2 Plant operating cycle.
It is important to realise that the actual capacity and performance capabilities of an operating plant are
nearly always different from the original design values. This arises because of design and construction
margins, influent flow and quality characteristics differing from design, and the accumulation of specific
operational know-how on the plant which improves performance and enables bottlenecks to be identified
and overcome. The operator’s main task is to determine the real capabilities of the plant and to get the
very best out of it despite apparent deficiencies in the design. Design and operating perspectives are
different. Capital cost, completion time and contractual issues weigh heavily during design and
construction. However, savings in initial cost and construction time are soon forgotten once operation
begins.
In general terms, the plant operating goals can be expressed as follows:
• Meet effluent quality standards.
• Meet biosolids quality standards.
• Meet odour standards.
• Meet energy consumption or greenhouse emission limits.
• Maximise plant capacity.
• Meet an operating cost limit.

by guest
These goals will be quantified to produce specific targets which may be tightened as plant performance is
enhanced over time. Inevitably there will need to be trade-offs among the various goals.
The key to better performance is improved understanding of plant behaviour. This book is therefore
ordered as follows: influent and effluent characteristics; individual process characteristics; overall plant
characteristics; and the EVOP process.

by guest
Chapter 2
Influent and effluent characteristics
2.1 THE CATCHMENT

Sewage characteristics vary from plant to plant and operators need to develop a detailed understanding of
their own influent. Sewage has several components – domestic, commercial, industrial and institutional
effluents, stormwater inflow and groundwater infiltration. The pollutional components of most sewages
are predominantly domestic, modified to some degree by the other effluents. This chapter examines the
characteristics of domestic sewage conveyed in separate sewers with no deliberate stormwater conveyance.
Rain
Off-Gases
CATCHMENT Evaporation
Local
Food
Production
Off-Gases
Stormwater
Inflow
Fertilisers
HOUSE- Sewage Effluent
Food Retail STP
HOLDS
Solid Groundwater
Wastes Infiltration
Biosolids
Landfill
Processing
Water
Stormwater
Supply
Figure 2.1 The simple sewerage catchment.

by guest
Domestic sewage has many constituents derived from both the water supply and its household uses.
Here our attention is focused on the parameters of most importance in nutrient removal plants.
Discussion of other constituents can be found in a range of references (e.g. Metcalf and Eddy, 2003;
Ternes & Joss, 2006).
Sewage has two primary characteristics, flow and mass loading. For pollutants added during use of the
water, concentration is a secondary characteristic – (mass load)/flow. However, concentration is the
predominant characteristic for constituents such as alkalinity which are derived in significant part from
the water supply and modified by groundwater and stormwater ingress.
Figure 2.1 shows the behaviour of a simple sewerage catchment. Flow is derived predominantly from
household water use. Pollutants come mainly from household food and cleaning products. We look at a
mass balance on the human body later but it is interesting to reflect that the truckloads of food and
detergents arriving at the local supermarkets represent the main part of the load subsequently arriving at
the sewage treatment plant.
2.2 FLOW CHARACTERISTICS

Sewage flow generally exhibits seven distinct characteristics: (1) a base flow generally called average dry
weather flow (ADWF), (2) a diurnal flow pattern, (3) a (weak) weekly flow pattern resulting from the weekly
living cycle, (4) a seasonal flow pattern derived from rainfall-induced inflow and groundwater infiltration,
(5) occasional short-term variations associated with holiday periods and special community events, (6)
long-term increase caused by growth in the contributing population, and (7) statistical variability
overlying the whole picture. The rhythms of life. These characteristics are interwoven in the historical
record and it is often difficult to separate them. For practical purposes, it is more useful to use a
statistical approach to the whole record. Growth can be quantified from the long-term trend. For periods
of more or less constant load, use the median (50 percentile) flow as the baseline. This is the daily flow
exceeded on 50% of days and is usually within 10% of the average.1
Figure 2.2 shows the daily flow and rainfall at the Maroochydore STP for the two years 2005–06. The
influences of stormwater inflow and groundwater infiltration can both be seen. The simplest way to reduce
this record to a useable form and increase our insight is to produce a flow frequency distribution. Flow
frequency plots for Maroochydore and three other plants are shown in Figure 2.3. These diagrams are
produced by sorting and ranking the daily flows over several years. Three years of data are needed to
reach the 0.1% level.
60 120
Daily Rainfall (mm/d)
Daily Flow (ML/d)
40 80
20 40
0 0
1-Jan 1-Apr 1-Jul 1-Oct 1-Jan 1-Apr 1-Jul 1-Oct 1-Jan
Figure 2.2 Two years of daily flow and rainfall at the Maroochydore STP (Qld) (2005–06).
1
The simple statistics used in this book are explained in Appendix 1.

by guest
Influent and effluent characteristics 7
Woodford
Daily Flow (normalised to median)

5
Maroochydore
4
3 Cleveland Bay
Port Pirie
1
0
0.01 0.1 1 10 100
% of Days Equal or Greater
Figure 2.3 Flow frequency distributions.
Two orders of plant size are represented (0.3–24 ML/d). The Maroochydore and Cleveland Bay plants
are of similar size and the effect of the higher rainfall at Maroochydore is evident. The small plant at
Woodford has a lower rainfall than Maroochydore but leakier sewers. At the other end of the scale, the
medium size plant at Port Pirie is located in an arid region and suffers very little flow peaking. Because
of wet weather peaks the average flow always exceeds the median, with margins of 2–12% at the four
example plants as listed in the Figure.
For practical purposes, the following assumptions are generally made about this type of statistical flow
frequency diagram:
• The period of records used is assumed to provide a reasonable representation of future flow
characteristics. Rainfall for the historical period can be compared with the long-term rainfall record
but predictions are always uncertain.
• The diagram is constructed from the flows on random rather than consecutive days. Because wet days
generally occur in runs (see Figure 2.2), it is assumed that a diagram constructed using consecutive
days would not be much different. Therefore, for example, the flow shown on the diagram for 8%
exceedance is assumed to be a useful representation of the flow for the peak month in a year (30
consecutive days rather than 30 random high rainfall days).

by guest
• The diagram is based on daily flows but it is assumed that percentage of days in the data period can be
interpreted generally as percentage of time. For example, the 0.1 percentile can be interpreted as 0.1%
of a year (8.76 hours per year).
• To provide insight into low probability events, it is assumed the plot can be extrapolated to lower
frequencies. Of course, the maximum flow rate is limited by the maximum capacity of the
sewerage system to deliver flow to the plant.
With percentiles it is useful to think in approximate practical terms. For example:
% greater than % less than

Peak month in a year 10 90
Peak week in a year 2 98
Peak day in a year 0.3 99.7
Peak day in 5 years 0.05 99.95
Peak hour in a year 0.01 99.99
While the frequency plot shows the percentage of time the flow exceeds various values, areas under the
curve give percentage of volume discharged. Referring to the Woodford curve replotted in Figure 2.4 (using
a natural scale to provide a better perspective on the percentages involved), the total area under the curve
(from 0–100%) gives the ratio of average flow:median flow (1.12 in this case as listed in Figure 2.3).
For illustration, if the plant treatment capacity happened to be 1.8Q50 (which is lower than the real
figure) and bypassing occurred at higher flows, the percentage of total sewage flow bypassed would be
given by the area ABCA as a percentage of the total area under the curve.
C
Daily Flow (normalised to median)
2
A B
0
0 20 40 60 80 100
% Equal or Greater
Figure 2.4 Woodford flow frequency distribution.

by guest
Figure 2.5 illustrates diurnal flow patterns for the Redcliffe and Cleveland Bay STPs (both in Qld). For
Redcliffe the figure shows the average and range for five weekdays and the average for the two weekend
days. The morning peak on the weekend lags the weekdays by about an hour. This is a typical
observation, reflecting the change in lifestyle at the weekend. The amplitude and timing of the
diurnal curve varies with the size and nature of the catchment, and with the lifestyle of the contributing
population.
2.5
Hourly Flow (ratio to average)
1.5
0.5
Redcliffe-Weekday average & range Weekend ave CB ave

0
0:00 4:00 8:00 12:00 16:00 20:00 0:00
Time of Day
Figure 2.5 Diurnal flow patterns at the Redcliffe STP (Qld), median flow 14 ML/d; data for seven days,
8–14 Jun 97. For comparison an average weekday pattern at the Cleveland Bay plant is also shown
(2–6 Aug 04).
2.3 MASS LOADING CHARACTERISTICS

The constituents of domestic sewage come from three main sources: (1) mass inputs during household
use of the water (the most important), (2) the source water supply (inorganics), and (3) groundwater
infiltration to the sewers (especially salinity and organic colour). The sewage quality is also often
modified by biochemical transformations in the sewerage system, particularly in the rising mains. And
like the flow, the quality also exhibits substantial statistical variability.
The domestic mass load comprises human and other household wastes. Bodily wastes constitute
the major part of the load and an indicative steady state mass balance on the adult human body is shown
in Table 2.1 and Figure 2.6. These figures relate to an adult living a light activity lifestyle on a diet
typical in Europe pre-1981 and vary with age, weight, gender, activity level and diet. The total outputs
are supplemented in the sewer by other household outputs from the kitchen, bathroom and laundry.
The fundamental constituent is the organic content (organic carbon), quantified here using COD
(chemical oxygen demand). This is the residue from the organic food providing the basic energy
requirement of the human body. The table indicates that close to 90% of the chemical energy in food is
extracted by the human body. The levels of the primary nutrients, nitrogen and phosphorus, reflect both
diet and, in the case of phosphorus, the use of powdered laundry detergent. The table gives a TN:TP
ratio of 7.5 whereas detergent use reduces the ratio in a typical Australian sewage to about 5.

by guest
Table 2.1 Mass balance on a 65 kg adult (light activity)1.
Component Whole Intake Output

body
Diet Air Gases Sweat Urine Faeces Total
(%) (g// d) (g// d) (g//d) (g// d) (g//d) (g// d) (g// d)
Major Items
Water 63 3700 300 2000 1300 80 3700
O2 800
CO2 900 0.2 0.13 0.1 900
CH4 Trace Trace
COD 900 10 40 60 110
TS 37 630 15 454 35 95
VS 30 600 7 254 306 60
VFA 0.028 4 1 5
Total N 3.02 13 0.6 10 27 15
NH3-N 0.1 1.0 0.06 1.2
Urea [CO(NH2)2] 2 21 0 23
Total P 1.1 2.0 0.0005 1.55 0.55 2.0
Total S 1.3
Inorganic TDS 17
pH 5.8 + 8 6.2 7.1
Cations
Ca 2.00 1.0 0.06 0.24 0.67 1.0
Mg 0.04 0.26 0.006 0.13 0.12 0.26
K 0.24 3.8 0.7 2.7 0.44 3.8
Na 0.16 7.6 2.2 5.2 0.15 7.6
Anions
Cl 0.16 7.0 2.1 4.8 0.06 7.0
SO4-S 0.07 1.15
HCO3 38 0.05 0.2 3
Heavy Metals
(masses in mg// d)
Arsenic, As 0.1
Cadmium, Cd 0.16 0.002 0.16 0.2
Chromium, Cr 0.07 0.008 0.06 0.07
Copper, Cu 0.00015 2.0 0.04 2 2
Iron, Fe 0.0070 6.1 0.1 6 6
Lead, Pb 0.34 0.04 0.3 0.3
Manganese, Mn 4.0 0.02 4 4
Mercury, Hg 0.005
(Continued)

by guest
Table 2.1 Mass balance on a 65 kg adult (light activity)1 (Continued).
Component Whole Intake Output

body
Diet Air Gases Sweat Urine Faeces Total
(%) (g// d) (g// d) (g//d) (g// d) (g//d) (g// d) (g// d)
Nickel, Ni 0.26 0.003 0.26 0.3
Tin, Sn 4.0 4 4
Vanadium,V 2.0 2 2
Zinc, Zn 0.0025 7.0 0.5 5–10 7
Overall
Energy MJ/d 11 11
MJ/kgVS 25 25
COD:VSS g/g 2.0
VFA:COD gVFA/ 0.05
gCOD
TN:COD g/g 0.14
TP:COD g/g 0.018
TN:TP g/g 7.5
1
Adapted from Ciba-Geigy Ltd (1981). European diet.
2
16% of protein
3
Partial pressure 6.7 kPa
4
Dissolved
5
Dissolved P in urine is mainly inorganic; solid P in faeces is mainly calcium phosphate
6
Major components: amino acids 2.8 g/d, lactic acid 1.2 g/d
7
50% dissolved
8
Sweat contains lactic acid; bicarbonate increases from 0 to 6 as sweat production rate increases
FOOD
WATER
AIR
Gases
Sweat
Urin
e
Faec
es
Figure 2.6 Steady state material balance on the human body; see Table 2.1.

by guest
Variability in daily COD loading is illustrated in Figure 2.7 and load frequency distributions are
compared with the associated flow frequencies in Figure 2.8. For COD loads, a useful indicator of
variability is the 90%ile load (exceeded on only 10% of days).
40 60
Daily COD Load (t/d)
Daily Flow (ML/d)

30 45
20 30
10 15
COD
0 0
1-Jan 1-Apr 1-Jul 1-Oct 1-Jan 1-Apr 1-Jul 1-Oct 1-Jan
Figure 2.7 Maroochydore STP – daily COD load data added to the daily flow plot in Figure 2.2.
4.5
Mar flow
4
Mar COD mass
Daily Load (normalised to median)
CB flow
3.5
CB COD mass
3
2.5
1.5
0.5
0
0.1 1 10 100
% Equal or Greater
Figure 2.8 Mass load frequency distributions for the Maroochydore and Cleveland Bay plants. The flow lines
are repeated from Figure 2.3 for comparison with the COD mass load distributions. The data periods are listed
in Figure 2.3.
Examples of diurnal concentration and mass loading patterns are shown in Figure 2.9. These diagrams
show the same plants and data periods as the flow diagram in Figure 2.5 and it is clear the factors affecting
the diurnal flow pattern also affect the mass loading pattern.

by guest
(a)
Concentration (ratio to average)
1.5
0.5
Redcliffe-COD SS TKN TP VFA CB-COD

0
0:00 4:00 8:00 12:00 16:00 20:00 0:00
Time of Day
(b)
4
Mass load (ratio to average)
Redcliffe-COD TKN TP CB-COD
0
0:00 4:00 8:00 12:00 16:00 20:00 0:00
Time of Day
Figure 2.9 Diurnal curves of (a) concentration and (b) mass load for the Redcliffe and Cleveland Bay plants.
Data are for the same weekday periods as shown in Figure 2.5.
It is more difficult and expensive to collect quality data than flow data. Twenty four hour composite
sampling is required to obtain a reliable estimate of the daily loading, and flow-proportioned sampling
provides a more accurate result than time-based sampling. The two sampling methods are compared in
Figure 2.10.
The most common sewage characteristics are listed in Table 2.2 and discussed below in the order listed in
the table.
2.3.1 Connected population

Total plant loading is commonly expressed in terms of equivalent population or persons (EP). This is a
rubbery figure because industrial inputs generally have different EP values for flow and the various
pollutants. With purely domestic sewage the EP equals the actual connected population. There are about
1000 municipal STPs in Australia with capacities ranging from less than 100 EP to 4 million EP (COD
basis – Western TP, Melbourne). The overall average and median capacities are about 20,000 EP and
3000 EP respectively.

by guest
2
Hourly Value (ratio to average)
Flow 1.45
Concentration 1.15
1
0.7
Ci = average concentration for the ith time interval
Qi = average flow for the ith time interval
ti = duration of ith time interval, d
0.1
0
0:00 4:00 8:00 12:00 16:00 20:00 0:00
Time of Day
Figure 2.10 Comparison of time-based and flow-proportioned sampling results.

by guest
Table 2.2 Typical plant load characterisation.
Parameter (& associated Symbol Typical values

text section)
Connected population (2.3.1) EP Australian average 20,000 EP
Median flow (2.3.2) Q50 200–250 L/EP · d
Median Loads:
COD (2.3.3) 120–130 g/EP · d
SS (2.3.4) 60 g/EP · d
pH (2.3.5) 7.0–7.5
Total alkalinity (2.3.5) 200–250 mgCaCO3/L
Sulfide (dissolved) (2.3.6) 2–10 mg/L
TDS or conductivity (2.3.7) TDS 800–1500 mg/L
Conductivity TDS/0.64 mS/cm
Pathogens (indicators):1 (2.3.8)
E.coli 107–1012, typical 1011 cfu/EP · d
F-RNA bacteriophage 107–109 , typical 108 pfu/EP · d
Somatic coliphage 108–1011, typical 1010 pfu/EP · d
Clostridium perfringens 107–108, typical 108 cfu/EP · d
Load Ratios: (2.3.9) Flow Mass Load
Average/median 1.05–1.1
Median/ADWF 1.0–1.05
Diurnal minimum 0.2–0.3 0.2–0.3
hour/diurnal average
Diurnal maximum 1.5–2.0 1.5–2.5
hour/diurnal average
90%ile/50%ile 1.2–1.7 1.3–1.5
(peak month)
99.7%ile/50%ile 1.4–4 1.7–2.0
(peak day)
Peak flow/median flow 3–5
Median Characteristics:
VSS/SS (2.3.4) 0.85–0.90
Particulate COD, CODp/VSS (2.3.4) 1.5–2.0
COD/BOD5 (2.3.10) 2.4
Nutrient Ratios: (2.3.11)
TOC/COD 0.30
TKN/COD 0.10
TP/COD 0.020
NH3-N/TKN 0.70
PO4-P/TP 0.80
(Continued)

by guest
Table 2.2 Typical plant load characterisation (Continued).
Parameter (& associated Symbol Typical values

text section)
Particulate N/VSS 0.07
Particulate P/VSS 0.01
Total COD Fractions: (2.3.12)
Soluble readily fbs 0.05–0.25
biodegradable (RBCOD)
Particulate biodegradable fbp 0.5–0.7
Soluble unbiodegradable fus 0.05
Particulate fup 0.20
unbiodegradable
VFA (as COD)/RBCOD fbsa Variable, depending on
sewer transformations
Soluble unbiodegradable (2.3.13) fnus 0.02–0.03
organic N/TKN
Nitrifier maximum specific (2.3.14) μnm20 0.5–1.0 d−1
growth rate
Temperature2 (2.3.15) Latitude Seasonal Range
38S (Melbourne) 14–24°C
28S (Brisbane) 19–28
18S (Townsville) 24–32
Chemical energy content (2.3.16) 12.5 MJ/kgCOD
Other sewage components (2.3.17)
1
National Resource Management Ministerial Council et al., 2006.
2
These temperatures are typical bioreactor operating temperatures.
2.3.2 Flow
The main external factors influencing the median flow are groundwater infiltration rate and water supply
restrictions. For example, a severe drought in southeast Queensland led to the introduction of water
restrictions in 2006, reducing domestic sewage flows from 230 L/EP · d to less than 140 L/EP · d.
Sewage strength increased proportionately. Subsequently a flow of 200 L/EP · d became a common
figure for long-term planning.
2.3.3 COD
Domestic COD load generally runs at 120–130 g/EP · d in Australia. This is a little above the human
waste output of 110 g/EP · d listed in Table 2.1, the difference being other household inputs and
commercial inputs not accounted for separately. The value will vary with the age distribution of the
contributing population because of the varying bodily outputs. COD can also be lost through sewer
transformations, discussed below.
COD, rather than BOD, is adopted as the organic load basis in this text. COD accounts for the total
organic load (biodegradable and unbiodegradable), has the advantage of simple and rapid analysis, and is

by guest
less likely than BOD to decrease with sample storage time.2 However, in practice, COD results seem no less
susceptible to analytical variability than BOD and are subject to chloride interference (increasing the test
result) when TDS levels are elevated. The COD:BOD ratio for Australian sewages is typically about 2.4
although this can vary because of transformations in the sewerage system and in the samples collected.
COD fractions and sewer transformations are discussed in Sections 2.3.12 and 2.4 below.
2.3.4 SS
The typical median mass load is about 60 g/EP · d which is 85–90% volatile (organic). This corresponds well
to the human TS outputs of Table 2.1, allowing for a dissolved fraction. The empirical composition of the
volatile component of sewage settleable solids is C10H19O3N which has a COD of 2.0 kgCOD/kgVSS and
N and P contents of about 7% and 1% respectively (Parkin & Owen, 1986). The COD is calculated as follows:
C10 H19 O3 N + 12.5O2 − 10CO2 + NH3 + 8H2 O

MW 201 400
COD = 400/201 = 2.0 mgO2 /mgVSS
In some sewages subject to high levels of wet weather inflow and infiltration carrying silt, the inorganic
fraction of the SS increases substantially during the wet season.
2.3.5 pH and alkalinity

Sewage alkalinity is important for buffering of acidity produced or added during treatment. Alkalinity is
typically 200–250 mg/L as CaCO3, which comes from both the water supply and household products.
Saline groundwater infiltration can also increase alkalinity.
2.3.6 Dissolved sulfide

Hydrogen sulfide is produced by reduction of sulfate in the sewerage system, particularly the rising mains.
Some of this sulfide is precipitated by dissolved metals such as iron and copper. The remaining dissolved
sulfide is a potential source of odour and corrosion and may inhibit treatment processes such as biological
phosphorus removal. Sulfide is discussed further in Section 2.4 below.
2.3.7 TDS or conductivity

High levels of total dissolved solids (TDS) can result from infiltration of saline groundwater. For example,
dry weather diurnal variation of sewage TDS concentration in the coastal Port Pirie catchment is shown in
2
A laboratory study conducted by the Western Australia Water Corporation (Beenyup WWTP Laboratory, 2006) examined the effect of
sample storage time before testing on the results of BOD and COD analyses. Standard Methods (1998) recommends initiating the BOD
analysis within 6 hours of sampling to minimise reduction of the BOD result. COD results are apparently more stable and samples can be
preserved by acidification.
In the study, samples from the Beenyup and Subiaco catchments were stored at 4°C for up to 48 hours before analysis. It was found
that the BOD result decreased as sample storage time increased. However, the COD of samples stored in the same way did not change
significantly over that period. This may be due to changes in the form of the organics, perhaps RBCOD being taken up into bacterial
cells and becoming slowly biodegradable COD, so that a lower percentage of the total organics is oxidised in the 5-day BOD test period.
The percentage decline in BOD with time would then depend on the RBCOD fraction in the original sample.
In the tests conducted in the study, the BOD of Subiaco sewage decreased by an average of 8% with 24 hours storage, and 12% with
48 hours storage. The corresponding declines for Beenyup sewage were 9% and 14%.

by guest
Figure 2.11. High TDS is a concern with respect to the effluent receiving environment, reuse of treated
effluent and potential inhibition of treatment processes such as nitrification.
30,000
25,000
20,000
TDS (mg/L)
15,000
10,000
5,000
0
12:00 16:00 20:00 0:00 4:00 8:00 12:00
Time of Day (15/16-Aug-00)
Figure 2.11 Sewage diurnal salinity profile, Port Pirie STP. Salinity is higher when there is less sewage flow to
dilute the constant saline inflow.
2.3.8 Pathogens
E coli is the indicator organism for pathogenic3 bacteria. A typical concentration is 107–108 cfu/100 mL.
F-RNA and somatic coliphages 4 are indicators for pathogenic enteric viruses.
Clostridium perfringens is a spore-forming bacterium5 used as an indicator for pathogenic protozoa such
as Cryptosporidium and Giardia and for helminths (intestinal worms).
2.3.9 Load ratios

The load ratios listed have practical plant application:
Average/median: This facilitates conversion between the median daily flow (exceeded on 50% of days)
and the average daily flow (total annual flow divided by 365 days). The median is relevant to effluent
discharge licences expressed in percentile (percentage of time) terms; the average is relevant to licences
limiting the total annual mass discharged.
Diurnal minimum and maximum hours: These ratios affect two things: (a) the diurnal variation in load on
plant equipment such as recycle pumps and aeration blowers, and (b) the diurnal variability in effluent
quality as a result of the plant process response to the loading variation.
3
Pathogenic: disease-causing
4
Coliphages are viruses that infect bacteria.
5
A bacterial spore is a resting form of the organism which is much more resistant than the vegetative form to disinfectants.

by guest
Peak month: This load represents the effective maximum load to which the main plant processes respond.
For example, (a) if operating the activated sludge process at a constant solids retention time (SRT), the
maximum MLSS concentration will be about the annual average multiplied by the peak month load
ratio, and (b) the load on the sludge digesters will also vary by about the same amount.
Peak day: Tells us the likely maximum daily load on the plant over a typical year.
Peak flow: The maximum instantaneous flow rate the plant is designed to pass. Higher sewage flows will
overflow in the sewerage system or bypass plant processes. Different parts of the plant may have
different design peak flow values.
2.3.10 COD//BOD5
Typically around 2.4 in Australian sewages. The COD fractions discussed below then give a biodegradable
COD/BOD ratio of 1.8.
2.3.11 Nutrient ratios

The nutrient ratios listed in Table 2.2 are typical for Australia. While the carbon and nitrogen contents come
mainly from bodily wastes, about a third of the phosphorus comes from powdered detergents (which may
reduce over time due to change in detergent formulation). Sewage phosphorus characteristics are generally
the most stable of all the quality characteristics and can be a useful benchmark for developing a better
understanding of a sewage when the overall data are limited and variable.
2.3.12 Total COD fractions

To understand the behaviour of biological treatment processes, the total COD is divided into the four main
categories listed. This subdivision is illustrated in Figure 2.12. The significance of these fractions is as
follows:
Soluble readily biodegradable COD (RBCOD), fbs: This fraction comprises the truly soluble
biodegradable COD and is the most rapidly biodegradable component. It can be further subdivided
into volatile fatty acids (VFAs) and fermentable COD which can be converted to VFAs. It produces
rapid denitrification and it drives the biological phosphorus removal process. It is difficult to
measure because it relates to the biodegradable fraction of the “truly soluble” (low molecular
weight) component of the total COD. It can be measured in two indirect ways: (a) by bioassay
using a bench-scale biological reactor to measure and interpret the varying response of oxygen
utilisation rate during oxidation of a sample of the sewage (Ekama et al., 1986; Wentzel et al.,
1995), or (b) by measuring the COD of a flocculated and membrane filtered sewage sample (the
total “truly soluble” COD), and deducting the “truly soluble” (unbiodegradable) COD in the
effluent from an activated sludge process (full-or pilot-scale) treating the same sewage (representing
the unbiodegradable soluble COD) (Mamais et al., 1993).6 The second method is generally simplest
and has become the most popular choice for design and operational purposes.
The VFA fraction of the RBCOD is variable. Table 2.1 shows that in fresh sewage the VFA
concentration should be about 5% of the total COD concentration and this can increase to as much
6
Note that the pore size of filter paper used for routine SS measurements is coarser than that of membranes used for differentiating
between “soluble” and “particulate” material where molecular weight cut-offs are involved.

by guest
as 25% through fermentation in the sewerage system (see Section 2.4 below). The VFA and total
RBCOD fractions can be boosted by prefermentation processes at the treatment plant.
Unbiodegradable COD, fus & fup: These fractions are readily estimated from normal plant data. The
soluble unbiodegradable COD is generally assumed to equal the soluble COD in the effluent from
an activated sludge process with a reasonably long SRT (although some of this COD actually
comprises metabolic products of the biological treatment process). Infiltration of organically
coloured groundwater can sometimes contribute to fus and residual colour in the plant effluent can
then interfere with UV disinfection.
The particulate unbiodegradable COD has its biggest influence on the sludge mass in an activated
sludge process. The fup value can be estimated from plant operating data using a simple activated
sludge model to calibrate the MLSS concentration to the process load and SRT.
Particulate biodegradable COD, fbp: This is the residual COD fraction, (1-fbs-fus-fup).
Total COD
Biodegradable Unbiodegradable
Particulate Soluble Particulate Soluble

Slowly Readily Unbiodegradable Unbiodegradable
Biodegradable Biodegradable
fbp fbs fup fus
Fermentable VFA
fbp + fbs + fup + fus = 1
(1 – fbsa)fbs fbsa.fbs
Figure 2.12 Sewage COD components. A small fraction may also be associated with active biomass which
has grown during transport in the sewerage system but this can generally be ignored in practice.
2.3.13 Soluble unbiodegradable organic N

Sewage nitrogen can be classified in a similar fashion to COD, see Figure 2.13. Ammonia is commonly
about 70% of the sewage TKN.7 Of the organic N (the nitrogenous component of organic compounds),
the biodegradable fraction is converted to ammonia and the particulate unbiodegradable fraction is
incorporated in the process sludge mass. In purely biological processes, the soluble unbiodegradable
fraction, fnus, passes unchanged into the effluent and thereby constitutes an important part of the effluent
total nitrogen. The value of fnus is typically 0.02–0.03, representing 1–2 mgN/L.
7
Oxidised N is usually, but not always, negligible in sewage.

by guest
TKN
Organic N
Unbiodegradable
Ammonia N Biodegradable Soluble Particulate

Unbiodegradable Unbiodegradable
fnus
Figure 2.13 Sewage nitrogen components.
The fnus value is commonly ascribed to the total soluble organic N measured in the effluent from
an operating BNR process. This total concentration may actually have several components –
soluble unbiodegradable N from the influent, residual (undegraded) soluble biodegradable N from
the influent and the N component of soluble metabolites produced by the biomass. It is difficult to
measure the individual components and it is commonly assumed that the soluble organic N measured in
the effluent is the unbiodegradable fraction of the influent TKN and that the other components are
negligible.
2.3.14 Nitrifier maximum specific growth rate

The growth rate of the nitrifying organisms is inhibited by a range of unidentifiable compounds which can be
present in sewage, particularly if there is a significant industrial input. The nitrifier maximum specific
growth rate, μnm20 or μmax, 8 is therefore generally listed as a sewage characteristic.
In conventional activated sludge processes the maximum specific growth rate typically varies from 0.4–
0.6 d−1 (0.4–0.6 of the organism mass per day). In biological N and P removal processes the growth rate can
be higher – up to about 1.0 d−1.9 See Chapter 3 for more detail.
2.3.15 Temperature
Detailed temperature data are scarce (see Wanner et al., 2005). Average sewage temperature is in the order
of 3 degrees higher than the water supply temperature (depending on per capita flow) because of household
8
Maximum specific growth rate is the organism growth rate in a nitrifying process treating the sewage at high (non-inhibitory) ammonia
concentration (several mg/L), high dissolved oxygen concentration (several mg/L), non-inhibitory pH (7.2–8.0) and temperature 20°C.
9
These growth rates are associated with a nitrifier endogenous decay rate of 0.04 d-1. Some detailed microbiological models use other
(higher) combinations of growth and decay rates.

by guest
and industrial water heating, and varies diurnally with the heat input. The temperature may dip a few degrees
in wet weather due to rainwater infiltration.
Plant process temperatures are 0.5–1°C higher than the sewage temperature because of reaction heat and
are more stable because of the process thermal mass (Wanner et al., 2005). Table 2.2 lists typical seasonal
process temperatures for a range of latitudes.
2.3.16 Energy content

Sewage contains both heat energy and chemical energy. The heat energy is represented by the elevation of
sewage temperature above the water supply temperature. Heat energy equals the heat capacity of water
(4.19 kJ/kg.°C).
Chemical energy content can be measured as the heat of combustion of the sewage organics. Because
the energy content of sewage organics can be transformed into methane by anaerobic bacteria, an
alternative approach is to calculate the heat of combustion of an equivalent quantity of methane (Parkin
& Owen, 1986). The equivalent quantity can be determined from the equivalent oxygen demand or COD
of the sewage. The COD of methane is 4 kgCOD/kgCH4 as calculated from the following equation:
CH4 + 2O2 CO2 + 2H2 O
The net energy content of sewage is therefore 50.0 (MJ/kgCH4)/4 (kgCOD/kgCH4) = 12.5 MJ/
kgCOD, where 50 MJ/kg is the net heat of combustion of methane (25°C, constant pressure, excluding
latent heat from water vapour condensation).
The heat and chemical energy contents of sewage are summarised in Table 2.3. It can be seen that
the chemical energy in sewage could raise the sewage temperature by 1.6°C and in fact part of the
energy released in the plant biological processes does raise the effluent temperature as mentioned in
2.3.15 above.
Table 2.3 Sewage energy content.
Parameter Heat energy Chemical energy

Specific energy 4.19 kJ/kg.°C 12.5 MJ/kgCOD
Per unit plant load:
Energy 4,190 MJ/ML.°C 12,500 MJ/tCOD
Power 48 kW per ML/d per°C 140 kW per tCOD/d
Per EP:
Per capita load 230 L/EP · d 125 gCOD/EP · d
Energy 0.96 MJ/EP · d.°C 1.6 MJ/EP · d
Power 11 W/EP.°C 18 W/EP
2.3.17 Other sewage components

There are many other sewage components and characteristics which may be individually important in
specific circumstances. Examples include heavy metals; synthetic organics such as pharmaceuticals and
personal care products (PPCPs), endocrine disruptors and pesticides (Ternes & Joss, 2006); and illicit
drugs (Lai et al., 2011).

by guest
2.4 SEWER TRANSFORMATIONS

Biological transformations of sewage quality occur in the sewerage system, particularly the rising mains.
These are predominantly anaerobic processes occurring in biofilm growing on the pipe wall at long
SRTs. There are three main processes:
• Fermentation of soluble COD to VFAs and hydrogen.
• Metabolic uptake of VFAs by sulfate-reducing bacteria, producing sulfide.
• Metabolic uptake of VFAs and hydrogen by methane-forming bacteria, producing methane.
Sulfide and methane production in the sewerage system reduces the readily biodegradable COD content of
the sewage and is therefore detrimental to biological nutrient removal in the downstream treatment plant.
Guisasola et al. (2008, 2009) and Sharma et al. (2008) found in both field and laboratory tests that (a)
non-VFA soluble COD concentration decreased, (b) VFA concentration increased for about 2 hours,
then decreased, (c) sulfide production increased steadily, and (d) methane production also increased
steadily. Sulfide and methane production increase with sewage hydraulic residence time and also with
the surface area:volume ratio of the sewers. Up to 100 mg COD/L of methane was found in one of the
two full-scale sewer systems studied, while methane production represented up to about 70% of the total
soluble COD loss in a laboratory-scale sewer system. Some methane concentrations were supersaturated
as found in other studies (Hartley and Lant, 2006).
These sewer processes affect the sewage characteristics at the STP inlet and have exerted some average
influence on the typical values of the characteristics listed in Table 2.2. Other influences on plant influent
characteristics include:
• Chemicals dosed in the sewerage system for control of sulfide, including oxygen, nitrate, hydrogen
peroxide, iron, magnesium hydroxide and various proprietary formulations.
• Internal plant recycles from the sludge processing stream back to the plant inlet.
2.5 EFFECT OF PRIMARY TREATMENT

The sewage characteristics listed in Table 2.2 are modified by primary treatment when this is used in the
treatment train. Table 2.4 shows how these changes can be estimated.
2.6 EFFLUENT CHARACTERISTICS

Effluent quality standards can be expressed in two ways, percentile concentration limits and mass discharge
limits. Most concentration limits relate to the percentage of daily average concentrations which must not
exceed the stipulated limit over a specified time period, usually a month or a year. The mass discharge
approach limits the maximum mass of a pollutant that can be discharged over a specified time period.
The relationship between the two – percentage of time versus total mass – is similar to time-based versus
flow-weighted sampling (Figure 2.10 above).
For an annual mass discharge limit the average effluent concentration has to be limited to the annual mass
limit divided by the total flow for the year. The total flow is 365 times the average daily flow, which is
greater than the median daily flow (see Section 2.2). The annual average concentration required is less
than the median concentration. With a mass discharge limit the precise average effluent quality required
is not known in advance because neither the magnitude nor distribution of rainfall and flow to be
expected can be known. Close tracking of performance is essential to ensure the overall mass discharge
target will be met.

by guest
by guest
Table 2.4 Effect of primary treatment on sewage characteristics.
24
Parameter Sewage Primary effluent Typical values
Symbol Formula Symbol Formula Sewage (Table 2.2) Primary Effluent

Concentrations:
COD COD CODPE = COD(1-RCOD) 550 mg/L 300 mg/L
SS SS SSPE = SS(1-RSS) 260 mg/L 100 mg/L
TKN TKN TKNPE = TKN – fvfnvSS.RSS 55 mg/L 46 mg/L
TP TP TPPE = TP – fvfpvSS.RSS 11 mg/L 9.7 mg/L
Pathogens (indicators):
E.coli 108 cfu/100 mL 107 cfu/100 mL
F-RNA bacteriophage 105 pfu/100 mL 104 pfu/100 mL
Somatic coliphage 107 pfu/100 mL 106 pfu/100 mL
Clostridium perfringens 105 cfu/100 mL 104 cfu/100 mL
PST Performance:
Overflow rate VO 1.0 m/h

SS removal coefficients Rm, a, k 0.8, 0.5, 1.0 m/h
SS removal RSS = Rm – a.exp(-k/VO) 0.62
COD removal RCOD = RSS(fup + fbp) 0.46
BOD removal RBOD = RSSfbp 0.34
TKN removal RN = RSS(fv.fnv.SS/TKN) 0.17
TP removal RP = RSS(fv.fpv.SS/TP) 0.20
COD Fractions:
Soluble biodegradable fbs fbs,pe = fbs/(1-RCOD) 0.20 0.37
Particulate biodegradable fbp fbp,pe = fbp(1-RSS)/(1-RCOD) 0.55 0.40
Soluble unbiodegradable fus fus,pe = fus/(1-RCOD) 0.05 0.09
Tuning Biological Nutrient Removal Plants
Particulate unbiodegradable fup fup,pe = fup(1-RSS)/(1-RCOD) 0.20 0.14

TKN Fraction:
Soluble unbiodeg organic N fnus fnus,pe = fnus/(1-RN) 0.03 0.036
Concentration Ratios:
Particulate COD/VSS fcv 2.0 2.0
Particulate N/VSS fnv 0.07 0.07
Particulate P/VSS fpv 0.01 0.01
VSS/SS fv = (fup + fbp)COD/(fcvSS) 0.85 0.85
COD/BOD5 = 1.8/(1-fus-fup) = 1.8/(1-fus,pe-fup,pe) 2.4 2.4
TKN/COD = TKNPE/CODPE 0.10 0.15
TP/COD = TPPE/CODPE 0.02 0.032
Two other effluent characteristics of interest are interactions between the different parameters, and
variability. Common interactions are listed in Table 2.5 and it can be seen that effluent SS is the
parameter having the most influence on other parameters.
Table 2.5 Effluent quality interactions.
Effluent quality Main components Source &// or magnitude

parameter (mg//L UNO)
COD Soluble unbiodegradable COD Sewage soluble unbiodegradable COD
fus × sewage COD
Residual soluble ---
biodegradable COD
Particulate COD 1.5 × VSS
BOD5 (inhibited)1 Residual soluble BOD ---
Particulate BOD ∼ 0.6 × SS
SS VSS (volatile) 0.6–0.8 × SS
ISS (inorganic) ---
Turbidity, NTU SS/[1.5 to 2.5]
True colour, PCU Sewage true colour ---
DOC ---
Iron ---
UV transmissivity, % True colour2 170 × (True Colour)-0.3
SS ---
Total N NH3-N ---
NOX-N (NO2-N + NO3-N) ---
Soluble unbiodegradable Sewage soluble unbiodeg organic N
organic N fnus × sewage TKN
Particulate organic N 0.10 × VSS
Total P Soluble P (ortho-P) ---
Particulate P ∼ 0.03–0.05 × VSS
1
Carbonaceous BOD – nitrification inhibited. Nitrification in the BOD test adds an uncertain fraction of the total nitrification
demand of 4.6 mgO2/mgNH3-N, where NH3 includes the ammonium ion, NH+ 4.
2
Correlation of unfiltered UV transmissivity data from 3 trickling filter and 3 activated sludge effluents; data ranges: true
colour 10–127 PCU, UVT 27–85%; R2 0.66.
Typical variabilities are listed in Table 2.6, although it should be noted that variability tends to increase as
a plant nears its operating capacity. For BNR effluents, a clear rule of thumb is that for most parameters the
90 percentile is about twice the 50 percentile. The exception is ammonia for which the 90 percentile is four
times the 50 percentile. These rules of thumb provide a basis for determining the median effluent
concentrations at which the plant has to be operated to meet its percentile limits. Appendix 1 provides
more detail about effluent variability characteristics.

by guest
Table 2.6 Typical effluent quality variability.
BNR PLANTS1
Parameter Median Concentration 90:50%ile Ratio

(median & range, mg// L) (median & range)
BOD5 3 (2–10) 2.0 (1.3–3.6)
SS 6 (0.5–15) 2.4 (1.8–5.9)
Oil & grease 2 (1 plant) 1.5 (1 plant)
Ammonia N 0.5 (0.1–11) 4.1 (1.7–6.8)
Nitrate N 1.4 (0.2–15) 2.4 (2.1–7.3)
Total N 5.1 (1.9–18) 2.0 (1.2–2.6)
Total P 1.7 (0.3–4.6) 2.4 (2.1–3.6)
Disinfection of Secondary Effluent
Chlorination:
Coliphage count in disinfected effluent: 90:50%ile 3.0 (0.5 logs)2
UV Disinfection:3
Faecal coliform count in disinfected effluent, 90:50%iles:
Redcliffe STP: 4.0 (0.6 logs)
Maroochydore STP: 6.2 (0.8 logs)
Nambour STP: 20 (1.3 logs)
1
Hartley,1995, plus data from author’s files. Analysis of 12 months data from 9 BNR
plants of various types. Variability does not correlate with plant size or type, or with
median effluent quality.
2
Metcalf & Eddy, 2003, Table 15-1, p1638.
3
Author’s data for 3 Queensland plants. Redcliffe, contractual acceptance test Feb-Mar
2001, 255 results; Maroochydore, operating data 1997–2000, 550 results; Nambour,
operating data 1998–2001, 168 results.

by guest
Chapter 3
Biological nutrient removal – process
fundamentals
3.1 THE BASIC PROCESS

Here we focus on the common suspended growth form of the BNR process, leaving aside fixed film variants.
The modern BNR process is a development of the basic activated sludge process which was invented in
1914 by Edward Ardern and William Lockett who were chemists with the Manchester Corporation in
England (Ardern & Lockett, 1914a, 1914b).
Ardern and Lockett began their first historic paper with these words (1914a, p523):
It has long been known that if sewage be exposed to the air for a sufficient period of time, the organic contents are
gradually oxidised, with the formation of a deposit of so-called “humus” and the final production of nitrate from
the ammonium salts and the nitrogenous organic matter.
This purification change ... takes place, however, comparatively slowly, and even if aided by direct aeration, by
no means becomes a practical method of sewage purification.
Ardern and Lockett, working first with bottles and then with casks, solved the problem by aerating a
batch of sewage, allowing the humus formed to settle, decanting the supernatant, adding a fresh batch of
sewage and resuming aeration in the presence of the humus previously formed. Successive decanting,
addition of new batches of sewage and aeration eventually produced a high concentration of settleable
solids and ‘a percentage purification … quite equal to that yielded by efficient bacterial filters’ in a
matter of hours (1914a, p525).
The people attending the meeting at which Messrs Ardern and Lockett presented their first paper
recognised immediately the importance of the results. For example, ‘Mr F R O’Shaughnessy regarded
the paper as an epoch-making one, provided that the process experimentally established by the authors
could ultimately be applied on a large scale at a reasonable cost’ (1914a, p536). ‘Mr S E Melling
thought the process described was ideal inasmuch as it resolved itself into a single-tank treatment
(1914a, p537).’ And ‘Mr J T Thompson said that one striking feature of the new process was that it was
aerobic from beginning to end, and it was probable that the amount of smell would be negligible
(1914a, p537).’

by guest
Messrs Ardern and Lockett had discovered the three key features of the activated sludge process:
• During aeration the sewage is mixed with a large mass of previously grown organisms.
• The solids are flocculent and can be removed from the liquid by settlement. In some modern forms of
the process the solids are separated using flotation or membranes but the majority of plants still
use settling.
• Separate control is exercised over the solid and liquid phases, so that the solids are retained in the
process for a much longer time that the liquid.
Ardern and Lockett wrote, ‘For reference purposes and failing a better term, the deposited solids resulting
from complete oxidation of sewage have been designated “activated sludge” ’ (1914a, p524). The sludge is
a complex population of microorganisms, mainly bacteria, with some protozoa and rotifers. The bacteria
effect treatment by utilising the polluting materials as food and converting much of it into bacterial cells.
Dissolved and colloidal materials are thereby transformed into easily removable solids.
Ardern and Lockett’s initial experiments were carried out on a batch basis but the first full-scale plants
were built on a continuous flow basis. Most plants are still of that type but the sequencing batch reactor
(SBR) process is identical in principle to Ardern and Lockett’s original experimental arrangement.
The format of the continuous flow process is shown in Figure 3.1. Its features are as follows:
• A bioreactor, often compartmentalised, with provisions for aeration and mixing of the activated
sludge and sewage mixture, which is called mixed liquor.
• A settling basin or clarifier to separate activated sludge from the treated effluent.
• A sludge recycling system to return settled activated sludge to the bioreactor. BNR processes also
generally have other recycles to transfer mixed liquor between the various bioreactor compartments.
• A sludge wasting system to remove sludge at the rate that it grows, thus maintaining the desired mass
of solids in the process. Waste sludge can be extracted from either the bioreactor or the return
activated sludge (RAS) flow.
• There may be chemical dosing provisions to enhance nitrogen and phosphorus removal.
Waste Activated
Sludge
Sewage Effluent
Bioreactor Clarifier
Return Activated Sludge

Oxygen
Figure 3.1 Basic format of the continuous flow activated sludge process.
The operator has four main controls:

(1) Sludge wasting rate, governing the solids retention time and mass of solids in the process.
(2) Aeration rate, governing the concentrations of both dissolved oxygen and dissolved carbon dioxide
in the bioreactor.

by guest
Biological nutrient removal – process fundamentals 29
(3) Sludge recycle rates, governing the solids concentration and mass of sludge in the clarifier and
bioreactor compartments.
(4) Chemical dosing rates, influencing nitrogen and phosphorus removal.
Sometimes the bioreactor compartment sizes and feed locations can also be adjusted.
The process can have one or more of the following functions:
• Removal of carbonaceous organic materials – this is the primary function of nearly all BNR plants.
• Oxidation of ammonia to nitrate – nitrification.
• Removal of total nitrogen by nitrification and the subsequent reduction of nitrate to nitrogen –
denitrification.
• Removal of phosphorus by biological and supplementary chemical means.
3.2 KINETICS OF BIOLOGICAL PROCESSES

The kinetics (rates) of BNR processes revolve around bacterial growth rates. This section provides an
overview of bacterial kinetics.
Bacteria need a supply of substrate or food, predominantly carbon, for building new cellular material plus
a supply of growth energy for carrying out that work. The bacteria utilising organic material, or COD, in
sewage are heterotrophs which obtain both their cell carbon and their growth energy from organic
substrates. A second group of bacteria called autotrophs obtain their cell carbon from inorganic carbon
dioxide and their growth energy from the oxidation of inorganic compounds such as ammonia.
3.2.1 Substrate utilisation

In order to grow, bacteria need substrates which can be synthesized into new cellular material, plus a supply
of energy with which to carry out this synthesis. In the case of the heterotrophs, part of the organic substrate
is oxidised to provide the energy needed to synthesize the rest into new cellular material.
Bacteria also need energy to maintain the existing cells. This maintenance energy is used for activities
such as the rebuilding of unstable cellular components and motility. In addition, a percentage of cells die
each day and become substrate for living cells. The energy used for cell maintenance and for oxidation
of dead cells is considered to be obtained by oxidation of previously formed cellular material and is
called endogenous respiration.
About 20% of cellular mass is relatively resistant to biodegradation and as a result of endogenous
respiration a residue of inert cellular material accumulates within the sludge mass.
The heterotrophic growth process is summarised in Figure 3.2. The total mass of substrate utilised with
time is dS/dt where S is the substrate composition (measured as COD). A proportion Y.dS/dt is converted
into bacterial cells, where Y is the yield coefficient (0.45 mgVSS/mgCOD for heterotrophs1). The rest of the
substrate, (1−Y)dS/dt, is oxidised to provide growth energy. At the same time, endogenous respiration
causes a decrease in the active bacterial mass. The rate of decrease is bX where X is the concentration of
active cells and b is the decay coefficient (0.24 d−1 for heterotrophs). The net growth rate of active
bacterial mass is then given by:
dX/dt = Y · dS/dt − bX (3.1)
1
For sewage organics, 33% of the COD utilised is oxidised and the resulting energy is used to convert the other 67% into cellular
material with a COD of 1.48 mgCOD/mgVSS. The yield coefficient is therefore 0.67 mgCOD/mgCOD or 0.67/1.48 = 0.45
mgVSS/mgCOD. The yield is different for other organic substrates.

by guest
End Products
CO2 + H2O +
NH3 + H3PO4
Oxidised for
growth energy,
Portion oxidised to provide (1-Y)dS/dt
energy for growth
Sewage Endogenous respiration

Organics to provide energy for
C10H19O3NP0.1 cell maintenance Reduction in mass
COD utilised, dS/dt

Oxidised of active cells due
to endogenous
Portion synthesised respiration,
into new cells (growth) bX
Residue
New Cells
C5H7O2NP0.1 Synthesised
into new cells,
Y.dS/dt
Residue of cellular material accumulating Mass of
as a result of endogenous respiration and active cells,
loss of viability dX/dt
Residue of
cellular material
Figure 3.2 Heterotrophic substrate utilisation.
3.2.2 Substrate concentration

Bacterial growth in a batch culture is illustrated in Figure 3.3. From A to B (the lag phase) the bacteria are
adapting their enzyme systems to the substrate. From B to C the bacterial growth rate is constant
(exponential growth), while the substrate concentration falls exponentially. However, at Point C the
substrate concentration becomes low enough to limit the rate of growth of the bacteria which then enter
the so-called declining growth phase. Slowing in the bacterial growth rate can also be caused by other
changes in solution chemistry such as accumulation of inhibitory end-products or changes in pH or
temperature but we are here concerned with the effect of substrate concentration.
Log Scale
D
C
Bacterial concentration, X
Substrate concentration, S
B
A
Natural Scale
Figure 3.3 Bacterial growth in a batch culture.

by guest
At Point D the synthesis rate has fallen to equal the decay rate and net growth is stationary. Beyond D the
bacterial mass declines because endogenous respiration exceeds the rate of synthesis of new cells.
The rate of change of active mass is described by the equation:
dX/dt = mX − bX
(3.2)
= (m − b)X
where
μ = growth rate of active bacteria, time−1
(μ − b) = net growth rate
A batch culture can be converted to a continuous system by withdrawing cells and oxidation products as they
are formed and adding food as it is used, to maintain uniform culture conditions. The continuous culture will
then operate indefinitely at some point on the batch culture curve in Figure 3.3. The aim of sewage treatment
is to reduce the substrate concentration, S, to a low value. Bacteria in treatment processes therefore normally
operate in the declining growth phase between Points C and D.
In 1942 the French microbiologist, Monod, investigated the dependence of the growth rate, μ, on the
substrate concentration, S (not published till 1949: Monod, 1949). He produced the now famous Monod
relationship (see Figure 3.4) expressing the dependence of bacterial growth rate on substrate concentration:
m = mm S/(KS + S) (3.3)
where
μm = maximum specific growth rate (substrate not limiting), time−1
KS = half velocity (or half saturation) coefficient, equal to the substrate concentration at which μ = μm/2
μm
Bacterial Growth Rate (1/time)
μm/2
KS
Substrate Concentration
Figure 3.4 The Monod curve.
The substrate which limits growth rate is the nutrient in shortest supply relative to bacterial needs, known
as the limiting nutrient. Nutrient requirements are discussed in Section 3.4.5.

by guest
An equation for the rate of change of substrate concentration can be derived from Equations 3.1–3.3:
dS/dt = kSX/(KS + S) (3.4)

where
k = rate coefficient for substrate utilisation
= μm/Y
Standard references (e.g. Metcalf & Eddy, 2003) can be consulted for more detail.
3.3 SOLIDS RETENTION TIME

3.3.1 Definition
Process solids retention time (SRT), also called mean cell residence time or sludge age, is the key operating
parameter as it governs or influences nearly all other parameters of interest. In the way that hydraulic
retention time (HRT) equals the volume of water in the process divided by the hydraulic flow rate, the
SRT equals the mass of solids in the process divided by the mass leaving the system each day, or the
mass wasting rate. SRT and HRT are independent.
The significance of the SRT is that under steady state operating conditions the net biological growth rate,
(μ − b), is the inverse of the SRT. Thus setting the wasting rate at, say, one tenth of the sludge mass per day
(SRT 10 days) sets the sludge growth rate at one tenth of its mass per day. Both the process sludge mass and
the effluent substrate concentration will attain constant values.
There is a minimum SRT below which no bacterial growth can occur. This is set by the maximum growth
rate of the bacteria and is given by Equation 3.5:
SRTmin = 1/(mm − b) (3.5)
3.3.2 Calculation
The basic equation for SRT is:
SRT = (MR + MC )/MW (3.6)
where
SRT = solids retention time, d

MR = mass of solids in the bioreactor, kg
= VRXM when the MLSS concentration is constant throughout the bioreactor
= ΣVRiXMi when various bioreactor compartments have different MLSS concentrations
XM = mixed liquor suspended solids (MLSS) concentration, mg/L
VR = bioreactor volume, ML
MC = mass of solids in the clarifier, kg
MW = mass of solids wasted, kg/d
= Q WX W
QW = volume wasted, ML/d
XW = solids concentration in the waste sludge, mg/L

by guest
As an example, if one tenth of the mass of solids in the process is wasted each day the SRT is 10 days. Except
at short SRT, or under conditions of critical clarifier loading (most commonly, wet weather), the mass of
solids in the clarifier is generally only a small fraction of the total and for operational purposes can be
ignored or assigned a nominal value such as 10% of the total solids.
In practice, a portion of the daily solids production is lost in the plant effluent and this must be taken into
account in calculating the SRT. Equation 3.6 should therefore be modified to read:
SRT = (MR + MC )/(MW + ME ) (3.7)
where
ME = mass of solids lost in the effluent, kg/d
= Q EX E
QE = effluent flow rate, ML/d
XE = effluent SS concentration, mg/L
If the waste sludge is directed to a thickener or dewatering process and the clarified return flow is directed to
the bioreactor, Equation 3.7 should be further modified to account for the solids returned to the process
(Figure 3.5):
SRT = (MR + MC )/(SC MW + ME ) (3.8)

where
SC = thickener solids capture
= QUXU / MW
= 1 – [(QW – QU)XO / MW]
QU = thickened sludge flow, ML/d
XU = thickened sludge SS concentration, mg/L
XO = thickener return flow SS concentration, mg/L
Bioreactor Effluent
Clarifier
VR, X M Q E, X E
Overflow
Waste Sludge
(Q W-Q U), X O
Q W, X W
Underflow
Thickener
Q U, X U
Figure 3.5 Solids flows.

by guest
3.4 CARBON REMOVAL

This section deals with the removal of carbonaceous organic materials. In the following Sections the
pioneering work of Professor Marais and his colleagues at the University of Cape Town has been
heavily drawn upon (Marais & Ekama, 1976; Ekama & Marais, 1977; UCT et al., 1984; Henze et al., 2008).
3.4.1 Effect of SRT

The main process parameters vary with SRT as shown in Figure 3.6.
• Mass of solids. The mass of solids in the bioreactor increases with SRT; at a particular SRT the mass
depends solely on the daily mass COD loading.
• Wasting mass. The daily mass to be wasted declines with increasing SRT because of the effect of
endogenous respiration (explained below).
• Oxygen demand. The oxygen demand increases with SRT; at a particular SRT the demand varies with
the COD and nitrogen loadings.
• Effluent soluble biodegradable COD (or BOD). For municipal sewage this is low and equals only a
few mg/L at SRTs greater than about 5 days. The effluent biodegradable COD does not depend on the
feed COD – higher feed COD is compensated for by increased biomass.
Sludge mass (kg)
Nitrogenous oxygen
demand (kg/d)
Carbonaceous oxygen
demand (kg/d)
Wasting mass (kg/d)
Effluent soluble biodegradable COD (mg/L)
0 10 20 30 40
SRT (d)
Figure 3.6 Variation of process parameters with SRT at a particular plant loading.
A typical classification of SRTs is as follows: high rate (COD removal only), 2–3 days; so-called
conventional (carbon removal plus nitrification), 5–15 days; and extended aeration, 15–40 days or more.
Most nitrogen and biological phosphorus removal processes operate in the high conventional to extended
aeration range.
There is a minimum SRT below which no bacterial growth can occur. This is given by the inverse of the
maximum net growth rate of the bacteria. At shorter SRTs the bacteria cannot grow as fast as they are wasted
and the solids are said to be washed out of the process. The minimum SRT for carbon removal is less than

by guest
one day but the high rate range of 2–3 days is the minimum producing sufficient flocculation for effective
solids settling and retention.
3.4.2 Loading conditions

In practice the loading on a plant is not steady but varies continuously. Instantaneous values of bacterial
growth rate and therefore SRT are correspondingly variable and in normal plant operation use of moving
average values calculated over several days is the most useful approach for monitoring purposes.
3.4.3 F:M ratio

As an alternative to SRT the process can be controlled using the food:microorganism (F:M) ratio. This is
also called the sludge loading rate and equals the daily mass COD (or BOD) loading per unit mass of
MLSS (kgCOD/kgMLSS.d).
F:M ratio is directly related to SRT and the approximate relationship for a typical municipal sewage is
shown in Figure 3.7.
2
F:M Ratio (kgCOD/kgMLSS.d)
1.5
0.5
Settled Sewage
Raw Sewage
0
0 10 20 30 40
SRT (d)
Figure 3.7 Relationship between F:M ratio and SRT for a typical municipal sewage.
Under normal operating conditions F:M ratio is inferior to SRT as a control parameter. F:M ratio has the
following disadvantages:
• It is an ambiguous term in that one process condition can be described by different F:M values,
depending on whether food loading is expressed as BOD, total or biodegradable COD or TOC,
and biomass as total, volatile or active mass (MLSS components are explained later). On
theoretical grounds the ratio is correctly expressed in terms of the mass of active bacteria but this
cannot be directly measured.
• The ratios of the active mass and mixed liquor volatile suspended solids (MLVSS) to total MLSS
depend on the amount of inert solids in the process feed, hence the F:M ratio describing a
particular process condition is affected by the feed characteristics.

by guest
• Frequent measurement of BOD, COD or TOC is necessary for process control. If BOD is used, values
are five days old when obtained.
• Choice of F:M ratio sets the approximate MLSS concentration to be maintained but gives no direct
indication of the solids wasting rate required.
In contrast, operation at a fixed SRT simply requires that the daily wasting equal (1/SRT) of the mass of
solids in the process. The appropriate MLSS concentration sets itself automatically. Day to day
fluctuations in COD loading result in automatic variations in MLSS concentration, but because of the
slow bacterial growth rates employed MLSS concentration varies only slowly. A change in operating
SRT can be made simply by adjusting the wasting rate; the MLSS concentration then gradually moves to
its new equilibrium value over a period of 2 to 3 SRTs.
A particular case where use of F:M ratio is preferred is process startup using a small quantity of seed
sludge. These startup conditions necessarily entail non-steady state operation with the biomass increasing
day by day. Where practical, a low initial loading can be applied and then increased daily in proportion
to the MLSS concentration to keep the F:M ratio constant.
3.4.4 Mixed liquor suspended solids

Components of MLSS
The solids mass in the mixed liquor of a carbon removal activated sludge process has five separate
components:
(1) Active mass. The mass of live organisms providing treatment.

(2) Endogenous residue. The accumulated inert residue of endogenous respiration.
(3) Inert SS from the feed. A portion of the volatile SS in the feed is unbiodegradable (fup × feed COD,
Table 2.2). These solids accumulate in the sludge mass.
(4) Stored substrate. Biodegradable organics in colloidal or suspended form are adsorbed onto the
bacterial cells and solubilised before being absorbed. Except at very short SRTs or during high
peak diurnal loadings the stored mass is a negligible fraction of the total MLSS.
(5) Inorganic mass. Components 1 to 4 are all organic (volatile), forming the MLVSS. There is also an
inert inorganic fraction, amounting in carbon and nitrogen removal processes to some 25% of the
MLSS with raw sewage feed and 17% with settled sewage feed.
In phosphorus removal processes there are two additional components of interest – inorganic polyphosphate
granules within the cells of the phosphorus accumulating organisms (PAOs), and inorganic precipitates
produced by chemical dosing for phosphorus removal. These are discussed later.
The various components cannot be readily measured in practice and for operational purposes the MLSS is
generally divided into the simple volatile and inorganic fractions.
Mass of solids
The effect of SRT on the process solids mass in a carbon removal process is shown in Figure 3.8. The values
shown are approximate only because the quantity of inert organics in the feed can vary considerably. The
solids mass is essentially the same for nitrifying and denitrifying processes because the mass of nitrifiers is
negligible compared with the heterotrophs. If the process incorporates enhanced phosphorus removal, the
sludge mass will be 10–20% higher.

by guest
14
Sludge Mass (kgSS/kgCOD per day)

12 MLSS
10 MLVSS
Raw Sewage MLSS
8 MLVSS
6
Settled Sewage
4
2
Active Mass
0
0 10 20 30 40
SRT (d)
Figure 3.8 Effect of SRTon the sludge mass in a carbon removal process treating a typical municipal sewage
at 20°C. With settled sewage there is less inert feed VSS in the sludge mass. A rise in temperature of 10°C
reduces the mass by about 9%.
It can be seen that the active mass is only a small part of the total. The active fraction declines as SRT
increases because of the increasing effect of endogenous respiration. Because the endogenous respiration
rate varies with temperature, the active fraction also declines with temperature rise. The active fraction
defines the stability of the waste sludge, or its tendency to produce odours and attract vectors during air
drying. Thus at higher temperatures a given degree of stability can be achieved at shorter SRT.
The daily sludge production, or wasting mass, equals the mass of sludge in the process divided by the
SRT. Approximate values for municipal sewage are shown in Figure 3.9.
0.6
Wasting Mass (kgSS/kgCOD)
0.5
Raw Sewage
0.4
0.3
Settled Sewage
0.2
0.1
0
0 10 20 30 40
SRT (d)
Figure 3.9 Effect of SRT on the mass of solids to be wasted at 20°C. The mass decreases by about 9% per
10°C rise in temperature.

by guest
3.4.5 Nutrient requirements

Bacterial growth requires a large number of different elements, most in trace quantities only (Wood &
Tchobanoglous, 1975).
Heterotrophic cell material contains, in addition to hydrogen and oxygen (which the organism can derive
from water), carbon, nitrogen, phosphorus and sulfur, in order of decreasing abundance (see Table 3.1).
These six elements account for about 95% of the cellular dry weight. Many other elements are present in
the remaining fraction, including potassium, sodium, magnesium, calcium, iron, manganese, cobalt,
copper, molybdenum and zinc (Stanier et al., 1976).
Table 3.1 Composition of the heterotrophic bacterial cell.
Element Approximate percentage

of dry weight
VSS 93
Carbon 49
Oxygen 26
Nitrogen 11
Hydrogen 6
Phosphorus 3
Sulfur 1
Potassium 1
Sodium 1
Calcium 0.5
Magnesium 0.5
Chlorine 0.5
Iron 0.2
All Others 0.3
The general chemical formula for organic cell material in a heterotrophic bacterium is C5H7O2N which
has a COD of 1.42 mgCOD/mgVSS, calculated as follows:
C5 H7 O2 N + 5O2 − 5CO2 + NH3 + 2H2 O

MW 113 160
COD = 160/113 = 1.42 mgO2 /mgVSS
An empirical figure of 1.48 mgCOD/mgVSS is often used for the total MLVSS including both biomass
and residual influent organics.
In municipal sewage organic carbon (or COD) is the nutrient limiting the growth of the biomass and other
nutrients are present in excess of minimum requirements. Because net sludge production varies with SRT,
nutrient requirements also vary with SRT. Nitrogen and phosphorus constitute about 10% and 3%
respectively of the overall MLVSS and approximate growth requirements are shown in Figure 3.10. The
percentage of phosphorus in the biomass is greater in enhanced P removal processes (discussed in
Section 3.6 below).

by guest
0.06
Nutrient Requirements (kg/kgCOD)

0.05
NITROGEN
0.04
Raw Sewage
0.03
0.02 Settled Sewage

PHOSPHORUS
Raw Sewage
0.01
Settled Sewage
0
0 10 20 30 40
SRT (d)
Figure 3.10 Approximate nitrogen and phosphorus requirements at 20°C. The differing requirements for raw
and settled sewage reflect the different percentages of inert feed VSS in the sludge mass.
3.4.6 Carbonaceous oxygen demand

In a carbon removal process, oxygen is used for two purposes:
• Oxidation of one third of the feed biodegradable COD to provide growth energy.
• Endogenous respiration.
The magnitude of the carbonaceous oxygen demand under steady state conditions is shown in Figure 3.11.
0.8
Carbonaceous Oxygen Demand
0.6
(kgO2/kgCOD)
Endogenous Respiration
0.4
0.2
COD Oxidation
0
0 10 20 30 40
SRT (d)
Figure 3.11 Carbonaceous oxygen demand at 20°C. A 10°C rise in temperature increases the oxygen
demand by about 5%.

by guest
3.5 NITROGEN REMOVAL

3.5.1 The nitrogen cycle
Figure 3.12 shows the essentials of the nitrogen cycle as it applies to sewage treatment processes. Gaseous
nitrogen is incorporated in organic compounds (nitrogen fixation) by both heterotrophic and autotrophic
microorganisms (bacteria, algae and fungi). As discussed in Chapter 2, sewage contains both organic
nitrogen and ammonia. In the treatment plant the organic nitrogen is released as ammonia
(ammonification) and some of this is reconverted to organic form when supplying the metabolic needs of
bacterial growth.
NITROGEN FIXATION
N2
Nitrogen gas Heterotrophs & autotrophs
DENITRIFICATION N2O Anammox

Organic N
Heterotrophs Nitrous oxide (anaerobic
ammonium
oxidisers)
AMMONIFICATION
Nitrosomonas
(micro-aerophilic conditions) Heterotrophs
NO Nitrosomonas
NH3
Nitric oxide (anoxic conditions) Ammonia
HNO2 NH2OH
Nitrous acid (nitrite) Hydroxylamine
s
na
omo
NOH ros
Nit
r
te
ac
ob
HNO3 NITRIFICATION
itr
Nitric acid (nitrate)

N
Autotrophs
Figure 3.12 The nitrogen cycle (after Henze et al., 2008).
The residual ammonia can then be oxidised progressively to nitrite and nitrate (nitrification). The
oxidised nitrogen can in turn be reduced to nitrogen gas (denitrification).
Intermediate nitrogenous compounds and pathways can be involved in these overall conversions as
illustrated in Figure 3.12.
3.5.2 Nitrification
Nitrifiers
Oxidation of ammonia to nitrate is carried out by two groups of bacteria (the nitrifiers), generally represented
as Nitrosomonas and Nitrobacter although other families are also involved. These organisms are aerobes,

by guest
growing under aerobic conditions and decaying via endogenous respiration under both aerobic and
non-aerobic conditions.2 They are also classified as chemical autotrophs because they obtain their
cellular carbon from dissolved carbon dioxide and their growth energy from oxidation of ammonia to
nitrite (Nitrosomonas) and nitrite to nitrate (Nitrobacter). Although Nitrosomonas and Nitrobacter are
generally used to represent the ammonium oxidising bacteria (AOBs) and nitrite oxidising bacteria
(NOBs), other bacteria are also known to be involved.
The overall nitrification process is described by these two chemical equations:
Nitrosomonas NH+ NO−
4 + 1.5O2 − 2 +H2 O + 2H
+
Nitrobacter NO− NO−

2 + 0.5O2 − 3
Overall NH+ NO−

4 + 2O2 − 3 + H2 O + 2H
+
These equations show that:

• The nitrogenous oxygen demand is 4.6 kgO2/kgN nitrified (3.43 for ammonia oxidation and 1.14 for
nitrite oxidation).
• Acid production reduces the total alkalinity by 7.1 mg/L as CaCO3 for each mg/L of ammonia-N
oxidised. Alkalinity and pH are discussed in Section 3.8.
The yield of nitrifiers is very low (0.1 mgVSS/mgN nitrified), which has three implications: (a) in practical
terms, the mass of nitrifiers makes a negligible contribution to the overall process sludge mass in combined
carbon and nitrogen removal processes, (b) nitrifier endogenous respiration is low and makes negligible
contribution to the overall oxygen demand, and (c) metabolic nitrogen uptake by the nitrifiers is small
and has negligible effect on the amount of ammonia nitrified.
Figure 3.13 shows the effect of SRT on the nitrifier active and total masses. The difference between the
two comprises the nitrifier endogenous and inorganic fractions. Comparison of Figure 3.13 with Figure
3.8 for the heterotrophs shows two things:
• The small nitrifier mass is evident. For example, assuming raw sewage with a TKN:COD ratio of 0.1,
an operating SRT of 20 days and a metabolic N uptake in accordance with Figure 3.10, the total
nitrifier mass is only 1.5% of the total heterotroph mass.
• As SRT varies, the two types of active mass behave differently. As SRT increases above about 10
days, the heterotroph active mass hardly increases while the nitrifier active mass increases
significantly. For example, the ratio of the 40d:10d active masses is 1.3 for the heterotrophs but
2.6 for the nitrifiers. This comes about because of the difference in the endogenous decay rates,
b – 0.24 d−1 (at 20°C) for the heterotrophs but only 0.04 d−1 for the nitrifiers. This is an important
difference: whereas nitrification can be improved by increasing the SRT, heterotrophic
denitrification performance hardly changes as SRT is varied. To increase the denitrifying sludge
mass the process anoxic mass fraction has to be increased (denitrification is discussed below).
Kinetics
SRT is particularly important for nitrification because the nitrifiers grow much more slowly than the
heterotrophs. The growth rates of Nitrosomonas and Nitrobacter are given by the Monod equation (3.3)
2
Conditions in which dissolved oxygen is absent but nitrite and/or nitrate are present are called anoxic. When oxygen, nitrite and nitrate
are all absent the conditions are called anaerobic.

by guest
where the substrate concentration is ammonia-N for Nitrosomonas and nitrite-N for Nitrobacter.
Nitrosomonas has a slower maximum specific growth rate than Nitrobacter and governs the overall
nitrification rate under most conditions. Nitrite is normally oxidised to nitrate as fast as it is formed so
little nitrite is generally found in nitrified effluents; however elevated nitrite concentrations may be found
under conditions of low dissolved oxygen, high pH or high temperature (Ma et al., 2009).
Nitrifier Mass (kgSS/kgN nitrified per day) 2.5
2
Total Mass
1.5
Active Mass
0.5
0
0 10 20 30 40
SRT (d)
Figure 3.13 Effect of SRT on the nitrifier mass in a fully aerobic process at 20°C.
Nitrifier growth rate is affected by several factors as discussed below.

Ammonia-N (substrate) concentration3
The relationship between growth rate and ammonia-N concentration is described by the Monod equation
(Equation 3.3). The fundamental nitrifier maximum specific growth rate, μnm, is sensitive to a wide range
of inhibitory substances, some of which are present in sewage but cannot be specifically identified. The
maximum specific growth rate is therefore sewage-specific and can vary over a range of 0.3–1.0 d−1 (at
20°C). It has been observed that alum dosing can enhance the nitrification rate, presumably by removing
inhibitors (Section 3.7.3). The half velocity coefficient in Equation 3.3, Kn, is about 1.0 mgN/L at 20°C.
Temperature
Nitrification is very sensitive to temperature. Both the maximum specific growth rate and the half velocity
coefficient increase by 12% per °C rise in temperature, or 3.2 times per 10°C. On the other hand, the
endogenous decay coefficient increases by only 3% per °C rise. A temperature of 20°C is used as the
reference for maximum specific growth rate, then termed μnm20.
Figure 3.14 illustrates the effects of SRT, maximum specific growth rate and temperature on effluent
ammonia concentration under steady state operating conditions.
3
At normal pH levels ammonia is present in two forms, free ammonia, NH3, and ammonium ion, NH+ 4 . The total concentration is
sometimes referred to as free and saline ammonia, FSA. In this book, unless specifically described as free ammonia, the
terminology mgNH3-N/L is used to represent the concentration of total free and saline ammonia expressed as nitrogen.

by guest
40
30 20 20 10°C
unm20 0.6 /d
35 unm20 0.4 /d
30
Effluent NH3-N (mg/L)

25
20
15
10
0
0 5 10 15
SRT (d)
Figure 3.14 Effect of SRT on effluent ammonia nitrogen concentration produced by a steady state complete
mix reactor at different temperatures and maximum specific growth rates at DO 2 mg/L and pH 7.2.
Dissolved oxygen concentration

The effect of dissolved oxygen concentration is also described by Monod kinetics, with a half velocity
coefficient, KO, around 0.3 mg DO/L. The effect of DO concentration on nitrifier growth rate is shown
in Figure 3.15.
0.8
(fraction of maximum)
Nitrifier Growth Rate
0.6
0.4
0.2
0
0 Ko 0.3 1 2 3
DO (mg/L)
Figure 3.15 Effect of dissolved oxygen concentration on nitrifier maximum specific growth rate.
pH
Growth rate is very sensitive to pH outside the range 7–8, as shown in Figure 3.16. The equations describing
the effect of pH are as follows (Henze et al., 2008):
pH ≤ 7.2:
mnmpH = mnm7.2 2.35(pH−7.2) (3.9)

by guest
where
μnmpH = maximum specific growth rate at pH of interest, d−1
pH ≥ 7.2:
mnmpH = mnm7.2 1.13(9.5 − pH)/(9.8 − pH) (3.10)
0.8
(fraction of maximum)
Nitrifier Growth Rate
0.6
0.4
0.2
0
4 5 6 7 8 9 10
pH
Figure 3.16 Effect of pH on nitrifier maximum specific growth rate.
Salinity
Elevated and variable salinity levels also reduce nitrifier growth rates, especially at TDS concentrations
above 10 g/L. Nitrite oxidation may be more affected than ammonia oxidation so that process nitrite
levels increase.
Concurrent biological phosphorus removal
Calibration of process models to the nitrification performance of full-scale nitrogen and phosphorus removal
plants indicates that the presence of an anaerobic zone for enhanced biological phosphorus removal at the
head of the process train may increase the value of μnm20 to about 1.0 d−1. The reason is uncertain but may
result from enhanced removal of nitrification-inhibiting compounds.
Alum dosing
Calibration of process models to the performance of full-scale plants with bioreactor alum dosing also shows
μnm20 values as high as 1.0 d−1. It is presumed this results from precipitation of inhibitors.
Load variation
Most sewage treatment plants do not operate under steady state conditions. The diurnal plant load varies
cyclically as shown in Chapter 2. The effect of this on nitrification is two-fold:
• Effluent ammonia concentration also varies cyclically.

• Because the peaks in effluent ammonia are greater than the troughs (geometric rather than arithmetic
variation), the average effluent ammonia concentration increases.
Figure 3.17 shows the approximate magnitudes of these effects.

by guest
Effluent NH3-N (ratio to steady state value)

Daily average Diurnal peak
6
5
Peak TN mass load 3xaverage
4
3
2x
2
3x 1.6x
2x
1.6x
1
0
2 3 4 5 6
SRT/SRTmin
Figure 3.17 Approximate magnitude of variation in effluent ammonia concentration under diurnal cyclic
loading conditions. The daily average and peak effluent ammonia concentrations are shown as ratios to
the steady state effluent concentration (Henze et al., 2008; Gloag, 2009). SRTmin is the minimum SRT for
nitrification as given by Equation 3.5.
Unaerated mass fraction

Nitrifiers can only grow under aerobic conditions. When a process contains both aerated and unaerated
(anoxic or anaerobic) zones, the nitrifiers grow in the aerated zones and decay via endogenous
respiration in both aerated and unaerated zones.
Figure 3.18 illustrates the effects of process unaerated mass fraction on nitrification. Two alternative
scenarios are shown, a total SRT of 10 days and an aerobic SRT of 10 days. The constant total SRT case
represents a process with constant sludge mass but varying aerated and unaerated fractions brought about
by different design, or, in an operating plant, change in reactor unaerated fraction or transfer of sludge
mass from reactor to clarifier. Under these conditions, as the unaerated mass fraction increases, both the
total active nitrifier mass and the active nitrifier mass in the aerobic zone decline and effluent ammonia
rises. At the operating conditions shown in the diagram, nitrification fails completely at an unaerated
fraction of 0.56.
The constant aerobic SRT case represents, for example, an operating plant in which sludge transfers from
reactor to clarifier but the reactor MLSS concentration is held constant by increasing the total SRT (reducing
the sludge wasting). The corresponding total SRT is shown in the figure. It can be seen that, as the unaerated
mass fraction and total SRT increase, the total active mass of nitrifiers in the process also increases; however
the active mass in the aerobic zone declines and the effluent ammonia concentration slowly deteriorates.
Nitrification fails completely when endogenous decay matches growth at an unaerated mass fraction of
0.82 (total SRT 55 days).
In the above discussion, note two important features of the unaerated mass fraction of the sludge.
Firstly, there is always an unaerated fraction in the clarifier and this is an integral part of the process
just as much as the bioreactor zones. Secondly, the unaerated mass fraction of the process may
differ from the unaerated volume fraction because of different concentrations of sludge in the different
zones.

by guest
Total SRT for aerobic SRT 10d Effluent NH3-N

Nitrifier active mass total Nitrifier active mass in aerobic zone
40 1.6
(kgVSS/kg NH3-N load per day)

30 1.2
Effluent NH3 -N (mg/L)
Nitrifier Active Mass

Aerobic SRT 10d
Total SRT (d)
20 0.8
Total SRT 10d
10 Total SRT 10d Aerobic SRT 10d 0.4
Total SRT 10d Aerobic SRT 10d
0 0
0 0.2 0.4 0.6 0.8 1
Unaerated Mass Fraction
Figure 3.18 Effect of process unaerated mass fraction on nitrifier active mass and effluent ammonia
concentration in a complete mix reactor under steady state conditions with temperature 20°C, DO 1 mg/L,
pH 6.8, μnm20 0.6 d−1. The nitrifier active mass is expressed in terms of the nitrification ammonia load,
equal to the effluent ammonia concentration with no nitrification.
Nitrification capacity
Nitrification capacity is a useful parameter which helps in understanding the overall
nitrification/denitrification behaviour of a process. Nitrification capacity is the nitrate concentration in
the effluent with full nitrification and no denitrification and is calculated using Equation 3.11:
Nc = Nti − Nte − Ns (3.11)
where
Nc = nitrification capacity, mgNO3-N/L in the effluent flow

Nti = process influent TKN, mgN/L
Nte = effluent soluble TKN, mgN/L
= fnus · Nti + Nae
fnus = process influent soluble unbiodegradable organic N fraction (Table 2.2)
Nae = effluent ammonia, mgN/L
Ns = nitrogen incorporated in sludge mass, mgN/L process influent
Figure 3.19 shows that as the influent COD:TKN ratio increases the nitrification capacity decreases because
of the greater fraction of influent nitrogen incorporated in the biomass.

by guest
Process Nitrification Capacity/Feed TKN, Nc/Nti

Markers show
typical feed ratios
0.9
Primary effluent feed

0.8
Raw sewage feed

0.7
0.6
0.5
4 6 8 10 12 14
Feed COD/FeedTKN, Sti/Nti
Figure 3.19 Effects of influent quality and primary treatment on process nitrification capacity with complete
nitrification (effluent ammonia zero). Based on typical sewage and primary effluent characteristics in
Table 2.4. SRT 20d, temperature 20°C.
3.5.3 Denitrification
Stoichiometry
Under anoxic conditions the oxygen required by the heterotrophic bacteria for oxidation of organics is
obtained from nitrate and nitrite (Figure 3.12). The facultative heterotrophs in the biomass can switch
between oxygen and nitrate respiration almost instantaneously.
Nitrate respiration is called denitrification because nitrate is reduced to nitrogen gas as follows:
5C6 H12 O6 + 24NO−

3 30CO2 + 18H2 O + 12N2 + 24OH
−
For comparison, the reaction with oxygen is:
5C6 H12 O6 + 30O2 30CO2 + 30H2 O
Here, glucose has been used as an example of an organic substrate but the stoichiometry is similar for
most organic materials.
From these equations it can be calculated that:
• Each mg of NO3-N denitrified is equivalent to 2.9 mg of oxygen.
• Alkali produced during denitrification increases the alkalinity by 3.6 mg/L as CaCO3 for each mg/L
of NO3-N denitrified.
If denitrification is included in a nitrifying process, oxygen use for nitrification can be reduced by up to 60%
and alkalinity depletion by up to 50%.
The total COD utilised in heterotrophic denitrification is 2.9/(1−YHD) where YHD is the bacterial yield
coefficient under anoxic operating conditions (see Section 3.2 above). Table 3.2 lists the denitrification
COD requirements reported for various substrates.

by guest
Table 3.2 Denitrification COD requirements.
Substrate Substrate Yield coefficient COD needed for Reference

COD (gCOD// gCOD) denitrification (table
(gCOD// g (gCOD// gNO3-N) footnotes)
substrate) [2.86// (1−YHD)]
Aerobic1 Anoxic
[YH] [YHD]
Sewage 0.67 0.53 6.1 2
Acetic Acid CH3COOH 1.07 0.54 0.45 5.2 3
Carbohydrate CH2O 1.07 0.72 0.61 7.3 4
Glucose C6H12O6 1.07 0.67 0.57 6.7 3
Lactic Acid CH3CHOHCOOH 1.07 0.65 0.52 6.0 4
Methanol CH3OH 1.5 0.67 0.36 4.5 4
Protein C16H24O5N4 1.5 0.64 0.51 5.8 4
1
Aerobic yield shown for comparison.
2
Muller et al. (2003)
3
Sperandio et al. (1999)
4
Orhon et al. (1996)
Kinetics
The fundamental process format used for nitrogen removal uses the Modified Ludzack-Ettinger (MLE)
flowsheet shown in Figure 3.20 (Ludzack & Ettinger, 1962; Barnard, 1973). Influent nitrogen passes
through the anoxic zone and is nitrified in the aerobic zone. Nitrate is then recycled to the anoxic zone
where it is denitrified using the influent readily biodegradable and slowly biodegradable COD. This
process format is sometimes extended by adding secondary anoxic and aerobic zones downstream of the
primary zones (see Chapter 5).
ML recycle, flow a.Q
Bioreactor
Influent, flow Q Effluent
Anoxic Aerobic Clarifier
Waste Activated
Sludge
RAS, flow s.Q
Figure 3.20 MLE process for nitrogen removal.
The denitrification rate is zero order with respect to the nitrate concentration, that is, the denitrification
rate is not affected by the nitrate concentration down to very low nitrate levels (,1 mg/L). There are four
different denitrification rates, known as K1, K2, K3 and K4, progressively decreasing in magnitude as listed
in Table 3.3. K1 applies for denitrification with readily biodegradable COD (RBCOD) and K2 with slowly
biodegradable COD (SBCOD). K3 is the denitrification rate in secondary anoxic reactors where nitrate is
used only for endogenous respiration while K4 is an even slower endogenous rate applying during
aerobic digestion of waste activated sludge.

by guest
Table 3.3 Denitrification rates.
Rate Value at 20°C Temperature

(mgNO3-N// mgVASS · d)1 coefficient, θ2
K1 (RBCOD) 0.720 1.200
K2 (SBCOD) 0.101 1.080
K3 (endogenous) 0.072 1.029
K4 (aerobic digestion) 0.048 1.029
1
VASS = biomass active VSS
2
KT = K20θ(T-20)
The maximum concentration of nitrate (relative to the influent flow) that an anoxic reactor can denitrify is
called the denitrification potential, Dp. This depends primarily on the influent RBCOD and SBCOD and the
process anoxic mass fraction. Whether the denitrification potential is achieved depends on the quantities of
nitrate and oxygen fed to the anoxic reactor.
The behaviour of a primary anoxic reactor is depicted in Figure 3.21. This shows how effluent nitrate
concentration (expressed as a fraction of maximum nitrification capacity – effluent nitrate concentration
with complete nitrification and no denitrification) varies with primary anoxic mass fraction for a range of
operating conditions. Several sets of lines are shown. Solid lines ABC, D, E relate to a process operating
at 20°C with raw sewage influent.
A
1
(K1 + K2) denitrification (RBCOD)
fbs = 0.05
Effluent NO3-N /Nitrification Capacity, Nne/Nc
0.8
fbs = 0.15
K2 denitrification (SBCOD) PE, fbs 0.27 (ex-sewage 0.15)
0.6
fbs = 0.25
Recycle limit: (a + s) = 1
B
D E
0.4
(a + s) = 2
F
0.2 (a + s) = 5
(a + s) = 10
25-20-15°C
C (a + s) = 100
0
0 0.2 0.4 0.6 0.8 1
Primary Anoxic Mass Fraction, fx1
Figure 3.21 Denitrification behaviour of a primary anoxic reactor. In this diagram, Nc is the nitrification
capacity with full nitrification (zero effluent ammonia); temperature is 20°C (except as noted), SRT is 20
days and sewage and primary effluent characteristics are typical values from Table 2.4. Secondary
denitrification in the clarifier is assumed to be negligible.
AB depicts the initial high rate (K1 + K2) denitrification. At B the influent RBCOD (for influent
RBCOD/total COD, fbs, of 0.25) is exhausted and denitrification slows to the K2 rate depicted by BC.
The dashed lines on either side of ABC show performance at 15 and 25°C.

by guest
Lines AD and AE show the denitrification performance with fbs values of 0.15 and 0.05.
Long-dashed line A-F shows the reduced denitrification performance with primary effluent as the
influent. This line is drawn for a primary effluent fbs value of 0.27, increased from a sewage value of
0.15 by primary sedimentation (Table 2.4).
These sloping denitrification lines depict the denitrification potential of the primary anoxic zone.
However, whether the full potential denitrification is achieved depends on the process recycle ratios, a
and s. Because only part of the aerobic reactor effluent can be recycled to the anoxic zone, the minimum
effluent nitrate achievable by the MLE process (ignoring any secondary denitrification in the clarifier) is
given by:
Nne /Nc = 1/(a + s + 1) (3.12)
where
Nne = effluent nitrate concentration
Nc = nitrification capacity, mgNO3-N/L in the effluent flow (Equn 3.11)
a = mixed liquor recycle ratio, recycle flow/influent flow
s = RAS recycle ratio, RAS flow/influent flow
In the diagram, the horizontal lines show the recycle limits for a range of recycle ratios.
Two further factors affect denitrification. Firstly, dissolved oxygen input to the anoxic reactor in the
influent or recycles reduces denitrification potential. Each mg of DO fed to the anoxic reactor reduces
denitrification potential by (1/2.9) mgNO3-N. A DO concentration of Oa mg/L in the mixed liquor
recycle increases effluent nitrate concentration by (a · Oa/2.9) mgNO3-N/L. A higher recycle ratio
returns more nitrate to the anoxic zone for removal but it also brings more oxygen which reduces
denitrification; therefore, when the recycles contain DO there is an optimum recycle ratio which gives
the lowest effluent nitrate concentration. The optimum recycle ratio is illustrated in Figure 3.22 and the
effect on effluent nitrate in Figure 3.23.
20
Optimum ML Recycle Ratio, ao
15
Dp/Nc = 1.1
10
1.0
5 0.9
0.8
0.7
0
0 1 2 3
DO in ML Recycle, Oa (mg/L)
Figure 3.22 Effect of dissolved oxygen in the mixed liquor recycle flow on the optimum recycle ratio giving
minimum effluent nitrate concentration. Dp/Nc is the ratio of the anoxic reactor denitrification potential, Dp,
to the process nitrification capacity, Nc. This diagram is drawn for a nitrification capacity of 40 mg/L, a RAS
recycle ratio, s, of 1 and zero DO in the RAS flow.

by guest
0.5
Effluent NO 3-N/Nitrification Capacity, Nne/Nc

0.4
0.3
Oa 2 mg/L
0.2
Oa 1 mg/L
0.1
Oa 0.5 mg/L
0
0 5 10 15 20
ML Recycle Ratio, a
Figure 3.23 Effect of dissolved oxygen in the mixed liquor recycle flow on the effluent nitrate concentration.
Corresponds to the Dp/Nc = 1.0 line in Figure 3.22.
The second factor is the provision of an anaerobic reactor at the head of the process train for enhancement
of biological phosphorus removal. The anaerobic reactor causes respective increases in the K2 and K3
denitrification rates in the anoxic reactors of about 2.5 and 1.5 times (Henze et al.,). The fundamental
reason for this is currently unknown. In the author’s full-scale plant experience, the K2 denitrification
rate increases from 0.101 mgNO3-N/mgVASS.d (Table 3.3) to about 0.21. The net result is that, with
the same total unaerated mass fraction, the process denitrification performance is unchanged by the
addition of an anaerobic reactor. For example, Figure 3.21 shows that, for a process without an
anaerobic reactor, at an fbs value of 0.15 and a recycle ratio of 100 the effluent nitrate concentration
reaches a minimum at a primary anoxic mass fraction of 0.46. If an anaerobic reactor with a mass
fraction of 0.10 is placed at the head of the process, the effluent nitrate concentration still reaches its
minimum at a total unaerated mass fraction of 0.46, made up of 0.10 anaerobic and 0.36 primary anoxic.
3.5.4 Simultaneous nitrification and denitrification

So far, the discussion has assumed that nitrification and denitrification occur in separate reactors or
locations. However, under conditions of low dissolved oxygen nitrification and denitrification can both
occur in the same volume. This is called simultaneous nitrification and denitrification (SND).
DO concentration varies within the bacterial flocs and both integrated biochemical conversions and local
spatial zonation can be involved. Integrated biochemical SND can release larger quantities of the active
greenhouse gas nitrous oxide, N2O.
For practical purposes, SND can be considered to function through spatial aerobic/anoxic zonation of the
flocs resulting from the consumption of oxygen diffusing inwards, as shown in Figure 3.24. The nitrifiers
inhabit the aerobic annulus while denitrification occurs in the anoxic core. The relative aerobic and anoxic
fractions depend on the operating bulk liquor DO concentration and the oxygen uptake rate (OUR) or
respiration rate.

by guest
Aerobic outer annulus

DO gradient and average DO in
aerobic zone depend on OUR
Anoxic core DO Bulk Liquor DO
Aerobic fraction of floc depends on floc size ,

bulk liquor DO and DO gradient
Figure 3.24 SND floc model.
Note two features of this process. Firstly, COD is utilised aerobically as well as anoxically so the extent of
denitrification for a given amount of COD is less than for compartmentalised formats. Secondly, the aerobic
and anoxic fractions are sensitive to variations in operating conditions, so effluent ammonia and nitrate
concentrations can be variable. The nitrifiers are motile and it has been observed under steady state
operating conditions that a reduction in bulk liquor DO set point can cause an immediate rise in effluent
ammonia concentration followed by a gradual recovery as nitrifiers move up-gradient to the smaller
aerobic zone (Hartley, 1997b).
3.5.5 Total process oxygen demand

Steady state conditions
The process oxygen demand consists of the total carbonaceous demand, plus the nitrogenous demand
for nitrification, less the oxygen equivalent of the nitrate denitrified for anoxic oxidation of organics.
When the process incorporates enhanced biological phosphorus removal, the carbonaceous demand is
shared between two groups of organisms, the ordinary heterotrophs and the phosphorus- and glycogen-
accumulating organisms (discussed in the next Section).
The carbonaceous and nitrogenous demands have been discussed in Section 3.4.6, 3.5.2 and 3.5.3 above.
The total picture is shown in Figure 3.25.
1.5
Oxygen Demand (kgO2/kgCOD)
Carbonaceous
0.5
Denitrification - settled
Nitrification - settled
Denitrification - raw
Nitrification - raw
0
0 10 20 30 40
SRT (d)
Figure 3.25 Total process oxygen demand under steady state conditions at 20°C, showing the effects of full
nitrification and denitrification.

by guest
The oxygen demand manifests itself as a respiration rate or oxygen uptake rate in the mixed liquor. The
respiration rate can be calculated by dividing the oxygen demand (Figure 3.25) by the sludge mass
(Figure 3.8) and is shown in Figure 3.26.
Respiration Rate (mgO2/gMLSS.h) 30
Raw sewage:
20 Carbonaceous + nitrogenous
Carbonaceous
Endogenous
Settled sewage:
Carbonaceous + nitrogenous
Carbonaceous
Endogenous
10
0
0 10 20 30 40
SRT (d)
Figure 3.26 Mixed liquor respiration rate under steady state conditions at 20°C. If the process incorporates
enhanced biological or chemical phosphorus removal, the respiration rates will be 10–20% lower than shown
due to the increased sludge mass. Values in mgO2/gMLSS · h can be converted to mgO2/L · h by multiplying
by MLSS concentration in g/L.
Dynamic conditions
The loading on sewage treatment plants generally follows a cyclic diurnal loading pattern (Chapter 2). Under
this type of loading the process oxygen demand also varies cyclically. A simple rule of thumb to estimate the
variation in oxygen demand over the day is that the amplitude of the total oxygen demand varies by about
one third of the amplitude of the COD mass load, where the amplitude is the maximum departure (up or
down) from the daily average (Ekama & Marais, 1977).
3.6 PHOSPHORUS REMOVAL

3.6.1 Mechanisms
So far two groups of bacteria have been discussed, the ordinary heterotrophic organisms (OHOs) which
grow on carbonaceous organic substrate under aerobic or anoxic conditions, and the autotrophic nitrifiers
which grow on inorganic carbon dioxide and oxidise ammonia or nitrite to obtain their energy
requirements. Enhanced biological phosphorus removal (EBPR) results from the activities of a third
group of bacteria, the phosphorus accumulating organisms (PAOs).
PAOs are special heterotrophs which are able to compete with the OHOs under appropriate conditions by
sequestering VFAs in the sewage for their own exclusive use. The basic process format required for PAO
growth is shown in Figure 3.27. It consists of an anaerobic reactor (no oxygen or nitrate) followed by an
aerobic reactor. In the anaerobic reactor the PAOs take up VFAs and store them within their cells; then
in the aerobic reactor the stored VFAs are used as substrate for their normal heterotrophic growth. The

by guest
OHOs cannot utilise VFAs in the anaerobic reactor but a proportion of them can ferment influent non-VFA
RBCOD to VFAs, benefiting the PAOs. Therefore the PAOs are able to sequester much of the influent
RBCOD for their own use.
ANAEROBIC REACTOR AEROBIC REACTOR

OHOs: Ferment non-VFA RBCOD to VFAs PAOs: Use stored PHA for growth
PAOs: Take up VFAs, store as PHA Take up P/cations, replenish polyP stores
Release P plus cations K, Mg, Ca Replenish glycogen stores
Deplete glycogen stores
RBCOD
(VFA/non-VFA) Anaerobic Aerobic
Clarifier
Reactor Reactor
Waste Activated
Sludge
Return Activated Sludge
Figure 3.27 Behaviour of the EBPR process.
The energy required by the PAOs to take up VFAs under anaerobic conditions is provided by the
degradation of internal reserves of both glycogen and polyphosphates. Glycogen is formed of linked
glucose units and is universally used by animals as a ready-energy store. However, the use of
polyphosphates for this purpose is confined to the small group of bacteria called PAOs. As VFAs are
taken up in the anaerobic reactor, phosphate is released from the cells into the bulk liquor. When the
cells pass into the aerobic reactor, phosphorus is taken up to replenish their polyphosphate reserves.
EBPR occurs by taking the normal wasting of daily sludge growth from the aerobic reactor where the
phosphorus content of the PAOs is at a maximum (P/VSS 0.38 for PAO cells, compared with the
normal 0.03 for the OHOs in the overall biomass).
The VFA molecules taken up in the anaerobic reactor are stored as complex long-chain linkages of fatty
acids called polyhydroxyalkanoates (PHAs). Animals commonly store food reserves as fatty acids which
have a high energy content. The stored PHA is then used in the aerobic reactor for heterotrophic energy
and growth, including replenishment of polyphosphate and glycogen reserves.
The counter-ions potassium, magnesium and calcium provide electrical neutrality in the polyphosphate
reserves. These elements are released in the anaerobic zone and taken up in the aerobic zone in parallel with
phosphorus. The approximate stoichiometry involved is as follows (Henze et al., 2008; Wentzel et al.,
1989):
P release: 0.5 gP/gCOD uptake

(for acetic acid, CH3COOH, COD = 1.07 mg/mg acid)
Cation release: P:K:Mg:Ca 1:0.33:0.33:0.05 molar
1:0.42:0.25:0.06 by mass
From Table 2.1, waste output from the human body contains P:K:Mg:Ca in the mass ratios of about
1:1.9:0.13:0.5. The only element potentially deficient with respect to the requirements of EBPR is
magnesium and most water supplies will easily make up the shortfall.

by guest
The process kinetics in the anaerobic reactor are dominated by the rate of fermentation of non-VFA
RBCOD to VFAs by the OHOs. This governs the anaerobic mass fraction required and, being a first
order reaction, makes a plug flow (or cells-in-series) configuration most efficient.
The minimum aerobic SRT required to fully utilise the stored PHA has been modeled and tested using
enriched culture laboratory systems (Henze et al., 2008). The value varies from 10 days at 5°C to 1.5 days at
25°C. A simple exponential fit to the results gives:
SRTAER,MIN = 92T−1.33 , where SRTAER,MIN = days and T = 8C (3.13)
3.6.2 Process characteristics

Process models have been developed from intensive laboratory studies using almost pure PAO cultures and
mixed cultures of PAOs and other organism groups (Henze et al., 2008; Wentzel et al., 1990). The overall
performance characteristics of the process when treating raw sewage are summarised in Figures 3.28 to 3.31.
These graphs show the following:
• Figure 3.28 shows that P removal is greatest at low SRT and declines as SRT increases because
of the decrease in net sludge production. P removal increases with anaerobic mass fraction because
of the increased fermentation of non-VFA RBCOD. Given that sewage typically has a TP:COD
ratio of about 0.02 (Chapter 2), the figure shows that EBPR can potentially provide full P removal.
24 0.04
fxa
0.25
P Removal (mgP/mg influent COD)

18 0.15 0.03
0.10
P Removal (mg/L)
0.05
12 0.02
6 0.01
0 0.00
0 5 10 15 20 25 30
SRT (d)
Figure 3.28 Predicted performance of an EBPR process treating raw sewage over a range of anaerobic mass
fractions, fxa. Process conditions: COD 600 mg/L, fus 0.05, fup 0.20, fbs 0.20, sewage VFA negligible,
temperature 20°C, 3 anaerobic cells in series, zero DO and NO3 input, effluent SS zero.
• Figure 3.29 shows the degree of performance improvement provided by subdividing the
anaerobic reactor into cells in series. This arises from the first order nature of the RBCOD
fermentation process.

by guest
24 0.04

P Removal (mg/L)
18 N 0.03
4
3
2
1
12 0.02
6 0.01
0 5 10 15 20 25 30
SRT (d)
Figure 3.29 Effect of the number of anaerobic cells in series, N, on P removal for fxa 0.10. Other process
conditions as for Figure 3.28.
• The effect of sewage fbs, the RBCOD fraction of total COD, is shown in Figure 3.30. P removal at zero
RBCOD is the normal metabolic uptake of the OHOs. As RBCOD increases, the PAO fraction of the
biomass increases at the expense of the OHOs and total P removal is almost tripled at an fbs value of
0.2. It can also be seen that the absolute COD concentration also affects performance because of the
impact on OHO population and RBCOD fermentation.
• Figure 3.30 also shows the P fraction of the MLSS and therefore of plant effluent SS. The P content
reaches 5% at an fbs value of 0.15.
• Figure 3.31 shows that as fbs and P uptake increase, the total sludge mass increases and the MLVSS:
MLSS ratio declines. These changes are caused by the increasing polyP content of the sludge. At an
fbs value of 0.2, the sludge mass increases by about 20% (compare with Figure 3.8); also, although not
shown in the figure, the carbonaceous oxygen demand decreases by about 5%.
The effect of temperature on EBPR is complex (Henze et al., 2008). Different temperature
coefficients apply to the anaerobic and aerobic components of the process. The anaerobic process
rate reaches a maximum at 20°C while the aerobic rate continues to increase up to 30°C. In full-
scale plants better P removal performance has sometimes been reported at lower temperatures. The
effect of temperature on PAO performance is confounded by the interaction of another group of
bacteria, the glycogen accumulating organisms or GAOs; this interaction is discussed in Section 3.6.3
below.
Increased P release and uptake have been observed as pH increases over the typical plant operating
range. This is attributed to the greater energy requirement for VFA uptake at higher pH levels. Like
temperature, the effect of pH is also confounded by interactions with GAOs (discussed below).

by guest
0.030 0.12
Influent COD
0.025 800 mg/L 0.1
600 mg/L
0.020 0.08
Particulate P/MLSS
P Removal
400 mg/L
0.015 0.06
P/MLSS
(COD 600 mg/L)
0.010 0.04
0.005 0.02
0.000 0
0 0.05 0.1 0.15 0.2 0.25 0.3
RBCOD/COD, fbs
Figure 3.30 Effect of RBCOD fraction of total COD, fbs, on P removal and on the P fraction of the MLSS. SRT
20d; except as shown, other process conditions as for Figure 3.28.
9 0.9
Sludge Mass (kgSS/kgCOD per d)
Sludge Mass
8 0.8 MLVSS/MLSS
7 0.7
MLVSS/MLSS
6 0.6
0 0.05 0.1 0.15 0.2 0.25 0.3
RBCOD/COD, fbs
Figure 3.31 Effect of RBCOD fraction of total COD, fbs, on the process total sludge mass and the MLSS
volatile fraction. SRT 20d; other process conditions as for Figure 3.28.
3.6.3 Adverse factors

Factors adversely affecting biological P removal are the feeding of dissolved oxygen, nitrate or sulfide to the
anaerobic reactor, and the competitive growth of glycogen accumulating organisms (GAOs).

by guest
Dissolved oxygen and nitrate

Dissolved oxygen and oxidised nitrogen (nitrite and nitrate) can enter the anaerobic reactor in the feed
and recycle flows. DO and nitrate are used by the OHOs for oxidation of RBCOD, thereby reducing the
RBCOD available to the PAOs. The resulting reductions in RBCOD are 3.0 mgCOD/mgDO and
8.6 mgCOD/mgNO3-N.
Figure 3.32 shows a gross correlation of P removal versus influent TKN for thirteen full-scale BNR
plants in Australia and the USA (Hartley, 1995; Morales et al., 1991). The data show a small decline in P
removal as the influent nitrogen increases, presumably due to increased recycling of nitrate to the
anaerobic reactor.
0.020
0.015
0.010
0.005
dP/COD = –0.0165(TKN/COD) + 0.0141

Mean 0.0129
0.000
0.00 0.05 0.10 0.15
Influent TKN/COD
Figure 3.32 Effect of influent TKN on EBPR performance in full-scale BNR plants. Five Australian plants and
eight US plants of various configurations; SRT variable up to 25d.
The ordinate value is 0.014 mg P removed/mgCOD and the overall mean is 0.013. These values are at the
low end of the performance range shown in Figure 3.30 and presumably reflect the action of adverse factors
in these full-scale plants.
Sulfide
Sulfide fed to the anaerobic reactor causes inhibition of EBPR. Figure 3.33 shows P removal performance
plotted against sulfide concentration in the combined anaerobic reactor feed after sewage was mixed with
return activated sludge (Hartley et al., 1999; Yamamoto et al., 1991). The Yamamoto data are from a pilot
plant study while the Thorneside data are from a full-scale BNR plant incorporating a prefermenter
generating considerable sulfide. The Thorneside data points are the averages for three operating periods
when the BNR process was nominally operating at steady state. Sulfide concentration measured within
the anaerobic reactor was zero.

by guest
0.03
P Removal = –0.0014(Sulfide) + 0.0258
P Removal/Feed Biodegradable COD

0.02
0.01
Thorneside Yamamoto et al
0
0 10 20 30 40
Sulfide in Anaerobic Reactor Feed (mg/L)
Figure 3.33 Effect of sulfide on biological P removal. Feed sulfide is the effective combined concentration
after the influent mixes with process recycles.
Note that P removal at zero sulfide was 0.026 mgP/mg biodegradable COD, equivalent to 0.019
mgP/mg total COD at a CODb:CODt ratio of 0.75. This reflects a better performance than Figure 3.32,
possibly because of the presence of a prefermenter at Thorneside.
PAO-GAO competition
Glycogen accumulating organisms (GAOs) have a similar metabolism to PAOs except that they rely solely
on glycogen stores for energy to take up VFAs under anaerobic conditions (Henze et al., 2008). They
compete with PAOs for the available RBCOD and therefore reduce the biological P removal
performance. The important factors governing the relative activities of PAOs and GAOs in operating
EBPR processes are type of carbon source (acetate or propionate), pH and temperature.
Lopez-Vasquez et al. (2009) conducted a detailed modeling study to determine the relative
PAO-GAO populations over a range of temperature, pH and VFA speciation conditions. The main
results are condensed into the contour diagram in Figure 3.34a. This diagram shows isopleths
(contours) of PAO mass as a percentage of the total (PAO + GAO) mass. They therefore represent
the percentage of RBCOD sequestered by the PAOs and used for P removal. The axes show the
two parameters which are potentially controllable by the plant operator, VFA speciation (acetate
concentration as a percentage of total VFA concentration) and mixed liquor pH. There are three
sets of isopleths on the diagram relating to three (uncontrollable) operating temperatures, 10, 20
and 30°C.
Because of the different growth characteristics of the PAOs and GAOs, each group of organisms is
favoured under different operating conditions. To highlight the area of the diagram giving good PAO
growth, Figure 3.34b shows the PAO 80% (or 87%) isopleths for the three temperatures. In the areas
above these isopleths, PAOs make up more than 80% of the total PAO-GAO biomass meaning the same
percentage of the total RBCOD sequestered goes to the PAOs. At 10°C virtually the whole biomass is
PAOs over the whole area of the diagram (the 87% isopleth is the lowest appearing in this operating
window). At 20°C the operating region favouring PAOs has shrunk considerably, while at 30°C good
PAO growth occurs only within a very narrow range of conditions.

by guest
(a) (b)
10°C 20°C 30°C
7.5 7.5
99 95 99 80% - 30C
95 90
40 60 80 40
20 80 90 80% - 20C 80% - 20C
10 60 60
Mixed Liquor pH
Mixed Liquor pH
7.0 8060 7.0
50
6.5 6.5
10 87% - 10C
40 87
6.0 6.0
0 25 50 75 100 0 25 50 75 100
Substrate Composition: Substrate Composition:

Acetic Acid (as C) Mass % (HAc/[HAc+HPr]) Acetic Acid (as C) Mass % (HAc/[HAc+HPr])
Figure 3.34 (a) PAO-GAO diagram showing isopleths of PAO mass as a percentage of total PAO + GAO
mass. The triangular marker shows the Cleveland Bay STP operating point (refer to text).(b) 80 (or 87)
percentile lines from (a). Zones of PAO dominance lie above the 80 percentile lines.
While the composition of the VFAs is known if they are fed to the anaerobic reactor from an external
source, this is not the case for an EBPR process relying predominantly on VFA production by internal
fermentation.
The author has found the diagram reliable in the course of a review of the EBPR performance of the
Cleveland Bay Wastewater Treatment Facility in the tropical Australian city of Townsville (Hartley,
2009). The sewage was well fermented in-sewer and during the review period the anaerobic zones of
two parallel independent bioreactors were fed primary effluent with a mean total COD of 292 mg/L and
a total VFA concentration of 59 mg/L (VFA-COD/total COD 0.22). Acetic acid was 85% of total
VFAs. Mixed liquor temperature and pH were 30°C and 6.85 respectively. The operating point is shown
by the triangular marker in Figure 3.34(a). By linear interpolation, the PAO percentage of PAOs +
GAOs given by the diagram is 49%. PAO and GAO enumeration was conducted on mixed liquor
samples from both bioreactors, giving PAO percentages of 43% and 51% for a mean of 47%, very close
to the diagram figure.
3.7 SUPPLEMENTARY CHEMICAL DOSING

Chemical dosing is commonly used to supplement biological nutrient removal, with the aims of enhancing
phosphorus and/or nitrogen removal, and increasing the operating pH. Chemicals can be dosed upstream or
downstream of the biological process or direct to the process itself. In this chapter the focus is on integral
dosing direct to the biological process.
3.7.1 Phosphorus removal

Phosphorus removal can be enhanced by dosing aluminium or iron salts to the mixed liquor. The use of alum
(aluminium sulfate) is discussed here but iron salts behave in a similar fashion.

by guest
When alum is dosed to the mixed liquor the useful reactions which occur are as follows (Letterman
et al., 1999):
• Aluminium precipitates as aluminium hydroxide, Al(OH)3.4
• The higher molecular weight fraction of dissolved organics is adsorbed onto the hydroxide
precipitate.
• Phosphate is incorporated to form a mixed aluminium hydroxide phosphate precipitate (Boisvert
et al., 1997; Galarneau & Gehr, 1997).
The extent to which these reactions occur depends mainly on the concentrations of the reactants and the
pH. The solid phase acts as a sorbent for phosphorus and an equilibrium is established between the
phosphorus concentrations in the liquid and solid phases; as process conditions change, phosphorus is
released or taken up to maintain the equilibrium.5 Phosphate is also sorbed by preformed aluminium
hydroxide flocs (in water treatment sludge, e.g.) but the degree of sorption decreases with the age of
the flocs.
Typical performance is shown by the alum dosing jar test results plotted in Figure 3.35 (GHD, 1994).
Twenty four tests were performed on an effluent from an activated sludge process exhibiting a degree
of EBPR (filtered P 2.2 mg/L). Depending on alum dose and pH, the final filtered P varied from 0.01–
0.9 mg/L as shown in Figure 3.35a. Optimum pH was 5.5–6.0. Removal of true colour (used as an
indicator of dissolved organics) was best at pH 5.5 (Figure 3.35b) while removal of SS was best at pH
6.0–6.5 (Figure 3.35c).
Phosphorus removal efficiency can be expressed in terms of aluminium adsorption capacity, mgP/mgAl.
The stoichiometric maximum for aluminium phosphate, AlPO4, is the atomic weight ratio, 31/27 or 1.15
mgP/mgAl. The jar test results in Figure 3.35a can be displayed in the form of Langmuir adsorption
isotherms by taking vertical slices at chosen pH values.
The Langmuir adsorption isotherm (Metcalf & Eddy, 2003) has the same mathematical form as the
Monod bacterial growth equation:
q = qm PS /(KP + PS ) (3.14)
where
q = aluminium adsorption capacity, mgP/mgAl

qm = maximum (saturation) aluminium adsorption capacity, mgP/mgAl
PS = soluble phosphate concentration, mgP/L
KP = half saturation coefficient, equal to the soluble P concentration at
which q = qm/2, mgP/L
Table 3.4 lists the calculated qm and KP values for the pH range of interest in biological processes and
Figure 3.36 shows the resulting isotherms.
4
Nielsen et al. (1973) found that after drying alum sludge at 103°C, 1.25H2O molecules per molecule of Al(OH)3 remained.
5
This means that under steady state operation, the alum dosing location within the bioreactor is not critical; no saving of alum accrues
from dosing in the anaerobic reactor where PO4 is higher as the P is released again downstream where PO4 concentrations are lower. A
common dosing location is the reactor outlet to the secondary clarifiers which has the potential advantage of providing a more rapid
effluent response to changes in dose rate but a sufficient contact time is required (which could be determined by jar testing).

by guest
(a) 350
Filtered P Isopleths (mg/L) Data Points
Alum Dose (mg/L as solid alum)

300 0.92 0.01 0.01 0.03
0.01
250 0.86 0.01 0.01 0.02
0.02
200 0.63 0.01 0.01 0.04
150 0.75 0.02 0.02 0.06

0.05
100 0.66 0.06 0.03 0.11

0.10
50 0.84 0.07 0.10 0.38

0.50
0
3 4 5 6 7 8
pH
(b) 350
True Colour Isopleths (PCU) Data Points
300 16 10 12 18
10
250 16 10 13 16
11
200 16 11 13 18
13
150 17 13 14 15 19
100 17 14 15 18 21
20
50 17 18 18 22
0
3 4 5 6 7 8
pH
(c) 350
Settled Turbidity Isopleths (NTU) Data Points
300 5.8 0.6 0.5 0.7
250 12.5 0.7 0.4 0.5
200 7.2 1.2 0.4 0.5
150 6.3 7.5 0.5

0.5 0.8
1
100 4.8 8.0 0.3 1.3
2
50 1.0 3.3 0.5 1.8
0
3 4 5 6 7 8
pH
Figure 3.35 Jar test results for alum dosing of a biological process effluent with the following quality: BOD 10
mg/L, SS 7 mg/L, turbidity 4.5 NTU, true colour 30 PCU, total P 2.9 mg/L, filtered P 2.2 mg/L. Test procedure:
dose predetermined quantity of acid or alkali and mix, dose alum and flash mix, flocculate at a velocity gradient
(G) of 40 s−1 for 15 minutes, settle for 30 minutes.

by guest
Table 3.4 Aluminium adsorption constants.1
pH qm mgP//mgAl KP mgP// L
6.0 1.10 0.10
6.5 0.96 0.13
7.0 0.83 0.18
7.5 0.75 0.27
1
Empirical correlations are:
qm = 24.9(pH)−1.74, r2 = 0.998
KP = 0.00183exp(0.661pH), r2 = 0.991
1.2
1 pH 6.0
Adsorption Capacity, q (mgP/mgAl)
pH 6.5
0.8
pH 7.0
0.6 pH 7.5
0.4
0.2
0
0 0.2 0.4 0.6 0.8 1
Soluble P, Ps (m g/L)
Figure 3.36 Adsorption isotherms for the removal of phosphate with alum.
Chemical phosphorus removal characteristics can be derived using Equation 3.14 in conjunction with the
following chemical equations:
Distribution of ortho-phosphate species in pH range of interest
H2 PO−
4 H+ + HPO2−
4 pK2 = 7.206
Phosphate precipitation
Al3+ + H2 PO−
4 AlPO4 + 2H+
Al3+ + HPO2−
4 AlPO4 + H+
6
Second dissociation constant for phosphoric acid, K2 = [H+][HPO2− −
4 ] / [H2PO4 ]; pK2 = negative logarithm of K2; [HPO4 ] =
2−
[H2PO−4 ] at pH = pK2 . pK values vary slightly with temperature and are normally expressed as the 25°C values.

by guest
Aluminium hydroxide precipitation
Al3+ + 4.25H2 O Al(OH)3 1.25H2 O + 3H+
Solids production is the sum of aluminium phosphate and aluminium hydroxide precipitates. Alkalinity
depletion is the total hydrogen ion production. The formulas used are listed in Table 3.5. The derived process
characteristics are shown in Table 3.6 and Figure 3.37.
Table 3.5 Alum dosing relationships.
Alum Dose
dA = 11.1(KP + PS)/(qmPS) (3.15)
where dA = alum dose as solid alum, Al2(SO4)314.3H2O, mg alum/mgΔP
and other symbols are defined in the text
Solids Production
XA = 0.334 + 0.064qmPS/(KP + PS) (3.16)
where XA = solids production, mgSS/mg solid alum dosed
Alkalinity Depletion
ΔTA = {qmPS/[11.1(KP + PS)]}{2/[1 + 10∧(pH-pK2)] + 1/[1 + 10∧(pK2-pH)]−4 · 84} + 0.501 (3.17)
where ΔTA = decrease in total alkalinity, mgCaCO3/mg solid alum dosed
Estimate of Biological P Removal
ΔPB = PTI – PS – DA/dA (3.18)
where ΔPB = P removed biologically, mg/L
PTI = process influent total P, mg/L
PS = process effluent soluble phosphate concentration, mgP/L
DA = process alum dose as solid alum, Al2(SO4)314.3H2O, mg alum/L of influent
and dA is defined above (Equation 3.15)
Table 3.6 Alum dosing stoichiometry.
Final pH Final soluble P Alum dose required

(mgPO4-P// L) (mg solid alum// mgΔP)
7.0 2 14.6
1 15.8
0.5 18.2
0.2 25.4
0.1 37.4
6.5 2 12.3
1 13.1
0.5 14.6
0.2 19.1
0.1 26.6

by guest
(a) Alum dose required

50
40
(mg solid alum / mg dP)

pH 7.5
30
Alum Dose 7.0

6.5
20
6.0
10
0
0 0.2 0.4 0.6 0.8 1
Soluble P, Ps (mg/L)
(b) SS production
0.4
0.39
pH 6.0
(mg SS/mg solid alum dosed)
0.38
6.5
Solids Production
0.37
7.0
7.5
0.36
0.35
0.34
0.33
0 0.2 0.4 0.6 0.8 1
(c) Alkalinity depletion

0.50
0.45
(mg CaCO3/mg solid alum dosed)
0.40
Alkalinity Depletion
pH 7.5
7.0
0.35
6.5
0.30
6.0
0.25
0.20
0 0.2 0.4 0.6 0.8 1
Figure 3.37 Alum dosing characteristics.

by guest
Equation 3.18 in Table 3.5 enables an estimate to be made of the magnitude of the biological phosphorus
removal occurring in a bioreactor operating with supplementary alum dosing.
3.7.2 Nitrogen removal

Aluminium hydroxide sorption of dissolved organics is shown by the removal of true (dissolved) colour in
Figure 3.35b. The organics sorbed contain nitrogen. Alum dosing therefore reduces soluble
unbiodegradable organic N. Figure 3.38 shows the effect of alum dosing to the mixed liquor of two
BNR plants over 18–22 month periods. The effluent concentrations exhibit a wide scatter, partly because
they have been calculated indirectly from several analyses:
West Byron: Nue = Nte – Nae – Nxe
Maroochydore: Nue = Ne – Nae – Nne – Nxe
where
Nue = effluent soluble unbiodegradable organic N, mg/L
Nte = effluent TKN, mg/L
Ne = effluent TN, mg/L
Nae = effluent ammonia N, mg/L
Nne = effluent nitrate N, mg/L
Nxe = effluent particulate organic N, mg/L
= 0.07Xe
Xe = effluent SS, mg/L
2.5
West Byron Jan 06-Oct 07
Effluent Soluble Organic N, Nue (calc, mg/L)
Maroochydore Jan 06-Jun 07

2 Medians: Sewage ML
COD BOD TN pH
West Byron --- 196 52 6.70
Maroochydore 451 --- 51 6.80
1.5
Nue = Nti(0.005 + 0.028exp(-0.04Da)

where Nti = sewage TKN
1
0.5
0
0 20 40 60 80 100 120
Alum Dose, Da (mg/L solid alum)
Figure 3.38 Effect of biological process alum dose on effluent soluble organic N at the West Byron and
Maroochydore STPs.
3.7.3 Other effects

In the author’s experience, the removal of soluble organics by alum dosing to the bioreactor can have two
other beneficial effects:
(1) Increasing the nitrification rate, presumably by the removal of inhibitory substances; this can
counterbalance any decrease in nitrification rate associated with an alum-caused decline in pH.
(2) Reducing the chlorine demand in a tertiary disinfection or oxidation process.

by guest
3.8 ALKALINITY AND pH

3.8.1 Chemistry
The alkalinity of a water is its acid-neutralising capacity.7 In sewage treatment processes, pH is
controlled mainly by the carbonate system, although other acid-base systems such as ammonia and
phosphate contribute when their concentrations are high (in digesters, e.g.) or the pH is particularly high
or low.
The carbonate system is formed from the dissolution of carbon dioxide (Loewenthal & Marais, 1976). A
small proportion of the dissolved carbon dioxide hydrolyses to carbonic acid which then dissociates into
bicarbonate and carbonate ions. These carbonic species together with the hydrogen and hydroxyl ions of
the water exist in a state of dynamic equilibrium described by the following equations:
CO2 + H2 O H2 CO3 Carbonic acid

+ −
H2 CO3 H + HCO3 Bicarbonate pK1 = 6.378
HCO3 − H+ + CO3 2− Carbonate pK2 = 10.33
+ −
H2 O H + OH Water pKw = 14.0
The relative concentrations of the different species vary with pH as shown in Figure 3.39. This chemical
system provides buffering against pH change when acid or base is dosed to the process or produced by
reaction. The degree of buffering available in the water is measured by total alkalinity. In simple terms,
total alkalinity is the mass of strong acid required to convert the CO3 2− and HCO3− ions to H2CO3 and
is given by:

[TA] = 2 CO3 2− + [HCO3 − ] + OH− ] − [H+
where [ ] represents molar concentration, moles/L.
Fraction of Total Species Concentration
1
H2CO3* HCO3 CO3
0.8
0.6
0.4
0.2
0 pK1 pK2
2 4 6 8 10 12 14
pH
Figure 3.39 Variation of carbonic species concentration with pH.
7
Conversely, the acidity of a water is its alkali-neutralising capacity.
8
Only a small fraction of the CO2 is hydrolysed to H2CO3 and the equilibrium is virtually unaffected by temperature or pH. K1 is
therefore calculated as [H+][HCO−3 ]/[H2CO3*] where [H2CO3*] = [CO2] + [H2CO3]. In common usage the star symbol is omitted.

by guest
Total alkalinity is measured by the quantity of strong acid required to reduce the sample pH to about 4.5
(Standard Methods, 1998). It is normally expressed in terms of equivalent calcium carbonate, CaCO3,
calculated by multiplying the mass of acid by the ratio of equivalent weights, {CaCO3}/{acid}.9
3.8.2 Process behaviour

The pH value in a nutrient removal process depends mainly on the following factors:
• The alkalinity and pH of the sewage.
• Carbon dioxide produced by oxidation of organics.
• The acid produced by nitrification.
• The alkali produced by denitrification.
• The acidity produced by dosing of chemicals such as alum or iron salts.
• Chemical alkali dosed to control the pH.
10
9 CO2 0.5 mg/L
2
8 5
10
Mixed Liquor pH
20
7 50
100
6
3 CO2 saturation ~ 0.5 mg/L
2
-50 0 50 100 150 200 250
Total Alkalinity (mgCaCO3/L)
Figure 3.40 Variation of mixed liquor pH with alkalinity and dissolved CO2 concentration.
About one third of the influent biodegradable COD is converted to carbon dioxide. This is dissolved in the
water and reduces the pH. The concentration of dissolved carbon dioxide in equilibrium with the small
quantity in the atmosphere is only about 0.5 mg/L, hence the water becomes supersaturated. Carbon
dioxide therefore transfers from the water to the air phase, primarily in the aeration zones. Thus the
aerators fulfil two functions – they transfer oxygen from the air to the water and they also transfer carbon
dioxide from the water to the air. The aerators are the lungs of the process.
The carbon dioxide concentration in the water is governed by the balance between the rate of production
and the rate of mass transfer to atmosphere. The carbon dioxide does not reduce the alkalinity, but does
reduce the pH at a given alkalinity. It will be observed that the pH of a clarified effluent is higher than
9
Equivalent weights of CaCO3, H2SO4 and HCl are 50, 49 and 36.5 respectively.

by guest
that of the mixed liquor because additional carbon dioxide is stripped from the surface of the clarifier,
particularly at the weirs and launders. However, alkalinity is unchanged.
The effect of mixed liquor alkalinity and carbon dioxide concentration on mixed liquor pH is shown in
Figure 3.40 (UCT et al., 1984). It can be seen that regardless of the actual concentration of carbon dioxide,
once the alkalinity falls below about 50 mgCaCO3/L the pH becomes unstable and falls rapidly. The effect
of aeration efficiency on pH is discussed in more detail in Chapter 6.
3.9 RULES OF THUMB

Useful rules of thumb are listed in Table 3.7.
Table 3.7 Nutrient removal rules of thumb.
A. Nitrification 1. Oxygen demand = 4.6 times NH3-N oxidised

2. TA lost = 7.1 times NH3-N oxidised
B. Denitrification 1. Oxygen recovered = 2.9 times NO3-N denitrified
2. TA recovered = 3.6 times NO3-N denitrified
3. COD used = 4.5–6.1 times NO3-N denitrified (Table 3.2)
C. Nitrogen Removal 1. Nitrification is governed by aerobic SRT
2. Denitrification is governed by anoxic fraction
3. DO discharged to the anoxic zone reduces denitrification by
0.3 mgN/mgDO
4. N removal capacity = 0.1 times COD
D. Biological Phosphorus Removal 1. RBCOD needed = 10 mgRBCOD/mgP removed
2. DO added to the sewage reduces RBCOD by
3 mgRBCOD/mgDO
3. DO added to the anaerobic zone reduces P removal by
0.3 mgP/mgDO
4. NO3-N added to the anaerobic zone reduces P removal by
1 mgP/mgNO3-N
5. Biological P removal capacity = 0.015 times COD
6. P release in anaerobic reactor = 0.5 mgP/mgCOD uptake =
5 mgP/mgP removed
E. Alum Dosing 1. Alum dose (as solid alum) required for P removal = 20 times
P removed
2. Alkali needed to counteract alum acidity = 0.35
mgCaCO3/mg solid alum

by guest
Chapter 4
Sludge settleability
4.1 CHARACTERISING SETTLEABILITY

Sludge settleability is one of the most important design and operating characteristics of the traditional
forms of activated sludge process utilising settling for biomass retention. Both the operating capacity of
the process and the effluent quality produced are dependent on the sludge settling and thickening
characteristics. Sludge quality is characterised using the simple Sludge Volume Index (SVI) test, which
came into use during the early years of activated sludge plant operation and was formalised by
Mohlman (1934).
SVI is the volume in mL occupied by one gram of MLSS after 30 minutes settling, commonly in a
standard one litre measuring cylinder. This represents the sludge volume after all of the flocs have been
brought into contact as closely as possible without being compacted. Values range from 50–100 mL/g
for dense, well-compacting sludges to several hundred for light, poorly compacting sludges. The latter
are known as bulking sludges and an SVI of 150 mL/g has often been used as a cut-off for this
description. The SVI value is the inverse of the settled sludge density and Mohlman chose this mode of
expression because increasing values reflect the increase in settled sludge blanket depth as settleability
deteriorates.
Activated sludge flocs are built around a backbone of filamentous bacteria which give the flocs their
strength. High SVI occurs when the filament length extending outside of the flocs becomes excessive,
interfering with settling and compaction. SVI does not become excessive until filament lengths exceed
10 km/gMLSS and highly filamentous sludges can have filament lengths of 100 km/gMLSS (Sezgin
et al., 1978; Sezgin, 1982)!
When measured in a one litre cylinder, SVI is calculated from:
SVI (mL/g) = settled volume (mL)/initial concentration (g/L) (4.1)
When the sludge is so filamentous that no settlement takes place in the cylinder at all, the SVI is a
maximum. Equation 4.1 shows that the maximum SVI value declines as the MLSS concentration increases.
When the settled volume is greater than 300–400 mL the small diameter of the cylinder tends to impede
agglomeration and settlement and increase the measured SVI value (Dick & Vesilind, 1969). This is
commonly known as the wall effect. If samples of mixed liquor are diluted or concentrated and SVI
measured at different initial concentrations, the test values vary as shown in Figure 4.1. This change in
the test value occurs even though there is no change in the nature of the sludge itself.

by guest
600
500 Plant C
NO SETTLING
400
SETTLED VOLUME 350 mL
SVI (mL/g)
300
Plant D
Plant A1
200
Plant A2
Plant E
100
0
0 5 10 15
Initial SS Concentration (g/L)
Figure 4.1 Effect of initial concentration in the test cylinder on the SVI value measured (after Dick & Vesilind,
1969).
To overcome this problem, three alternative test procedures have come into use:
• Diluted SVI (DSVI): Serial dilutions of the sludge are used to reduce the settled sludge volume to less
than 350 mL/g (Stobbe, 1964; Lee et al., 1983).
• Stirred SVI (sSVI): A slow speed stirrer is fitted to the standard one litre test cylinder to overcome the
wall effect (Standard Methods, 1998).
• Stirred Specific Volume Index (SSVI3.5 or simply SSVI): A larger cylinder fitted with a slow speed
stirrer is used (approximately 100 mm diameter × 500 mm water depth) and the sludge is initially
diluted or concentrated to a standard initial concentration of 3.5 g/L. This test procedure was
developed by White (1975, 1976) who originally designated it the Stirred Specific Volume (SSV).
The standard unstirred SVI test has enjoyed widespread acceptance as an aid to operation because it is
simple and requires no special apparatus. Stirring is not essential for normal plant control where
comparative values are of more significance than absolute values, but the limitations of the unstirred test
should be kept in mind:
• At settled volumes above 300–400 mL, a change in MLSS concentration may produce a change in the
measured value of SVI when no change in the nature of the sludge has occurred.
• When settlement during the test is small, the measured SVI is rather insensitive to actual changes in
sludge settling characteristics.
• The test result is sensitive to temperature, methods of sampling and agitation, diameter of the settling
column and time between sampling and the start of testing so useful results can only be obtained by
following exactly the same procedure each time the test is performed.
SSVI produces the most reliable measure of the actual settling characteristics of the sludge and is generally
accepted as the standard. In practice this test is often performed at the sample SS concentration rather than
3.5 g/L; this usually gives reliable results because of the large cylinder and the stirring employed. The sSVI
test is identical except for the smaller cylinder and for practical purposes is usually considered equivalent to
the SSVI.

by guest
Sludge settleability 73
4.2 SETTLING RATES

At the usual MLSS concentrations, sludge settles in the hindered settling mode. The flocs displace water as
they settle and because they are close together the resultant upward flow of water has sufficient velocity to
produce a measurable reduction in the floc settling rate relative to a fixed plane. The sludge settling rate
therefore varies with the SS concentration and the relationship is well represented by the exponential
equation:
VS = VO e−nX (4.2)
where
VS = zone (sludge interface) settling velocity, m/h

VO = constant, nominally equal to the settling velocity at X = 0, m/h
n = constant, L/gSS
X = initial sludge SS concentration, gSS/L
Table 4.1 Ozinsky & Ekama settleability relationships.

Settling Parameters:
VO = α · exp(−β · SSP) m/h (4.3)

n = γ + δ · SSP L/g (4.4)
Where SSP = sludge settleability parameter (SVI, DSVI or SSVI, mL/g) and values of constants α, β, γ and δ are
listed below.
Settling Rate:
VS = VOe-nX (4.2)
or, substituting for VO and n above:
ln VS = ln α − β · SSP − γ · X − δ · SSP · X (4.5)
Conversions:
From Equation 4.5, conversions between SSPs can be made for the same sludge settling rate:
SSP2 = [ln αSSP2 − γSSP2X − (ln αSSP1 − βSSP1SSP1 − γSSP1X − δSSP1X · SSP1)]/[βSSP2 + δSSP2X] (4.6)
Constants:
Process Type NDEBPR ND
SSP SSVI SSVI DSVI SVI

ln α 2.70065 2.45095 2.30854 2.14370
α 14.88941 11.59936 10.05973 8.53094
β 0.00808 0.00636 0.00297 0.00165
γ 0.22632 0.15128 0.29721 0.20036
δ 0.00264 0.00287 0.00095 0.00091
Correlation coeff, r2 0.916 0.849 0.776 0.842

by guest
Several studies have shown that the values of the constants VO and n can be related to the SVI of the sludge.
The most comprehensive was undertaken by Ozinsky and Ekama (1995), who examined the data from nine
previous studies, rejecting the data from four as being statistically inadequate. The data from the other five
were analysed to produce statistically significant and useful relationships between the sludge settling
constants and the associated SVI, DSVI and SSVI values. The main outcomes of their study are set out
in Table 4.1.
8
X=1 g/L
6
Settling Rate, Vs (m/h)
X=2
5
4 X=3
3 X=4
2
X=6
1 X=8
0
0 20 40 60 80 100 120 140 160
SSVI (NDEBPR) (mL/g)
Figure 4.2 Effect of SSVI on sludge settling rate.
(a) 160 (b) 160

140 140
120 120
SSVI (ND) (mL/g)
SSVI (ND) (mL/g)
100 100
80 80
X=10 g/L
60 X=1 g/L X=4
60
X=2 X=2
40 X=4 40 X=1
X=10
20 20
0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300
SVI (ND) (mL/g) DSVI (ND) (mL/g)
(c) 160
140
SSVI (NDEBPR) (mL/g)
120
100
80
60 X=1 g/L
X=2
40
X=4
X=10
20
0
0 20 40 60 80 100 120 140 160
SSVI (ND) (mL/g)
Figure 4.3 Conversions between sludge settleability parameters.

by guest
Figure 4.2 shows the relationship between sludge settling rate and SSVI. Figure 4.3 shows conversions
from the other settleability parameters prepared using Equation 4.6 in Table 4.1. Note from Figure 4.3c that
there is a small difference between the SSVI data for nitrification denitrification (ND) versus nitrification
denitrification excess biological phosphorus removal (NDEBPR) processes, however, given the variable
nature of sludge settleability, this is small enough to be ignored for practical purposes within the normal
range of MLSS concentration.
Bear in mind that these settling relationships are purely empirical and are not precise (as reflected in the
correlation coefficients listed in Table 4.1). Individual plants may have sludge characteristics differing
somewhat from the overall mean. One factor which can reduce the settling rates of denitrifying sludges
is the formation of gaseous microbubbles within the flocs.
4.3 FACTORS AFFECTING SETTLEABILITY

Nearly one hundred years after the activated sludge process was invented (Ardern & Lockett, 1914a, b),
positive prediction and control of SVI remain elusive. SVI is a measure of the relative proportions of
zoogleal and filamentous bacteria in the sludge. This depends on their relative growth characteristics
which in turn depend on the process operating environment and the concentrations and form (soluble or
particulate) of macro- and micronutrients.
There has been considerable research into the types of filamentous organisms and the associated filament
lengths affecting settleability. However this has not proven particularly useful from a practical perspective.
SVI changes at the same time as the filament length so microbiological monitoring does not provide useful
early warning of changes in SVI.
4.3.1 Non-BNR activated sludge processes

A range of factors has been reported to affect settleability in activated sludge processes generally, including
feed quality, SRT, DO concentration, pH, reactor mixing characteristics, anoxic fraction and chemical
dosing (Jenkins et al., 2004; Henze et al., 2008).
Feed quality
Feed parameters reported to affect SVI include nutrient availability, the level of soluble readily
biodegradable COD (discussed below), sulfide level and the presence of industrial wastes; some
industrial wastes, carbohydrates for example, are notorious for producing bulking sludge.
Operating parameters
In short SRT conventional processes DO concentration affects SVI as shown in Figure 4.4. The effect of DO
concentration varies with SRT because of the variation of respiration rate with SRT (Figure 3.26). Bacteria
within the flocs are supplied with oxygen by diffusion from the bulk liquor. The resulting DO gradient is
steeper at higher respiration rates, requiring higher bulk liquor DO concentrations to maintain a
particular DO concentration at the centre of the floc. When the interior of the floc is under oxygen stress
the filamentous organisms growing from the floc into the bulk liquor have an advantage and become
more dominant.
At extremely short SRTs (very high growth rates) bacterial growth becomes more dispersed and floc sizes
smaller. SVI therefore becomes less dependent on DO concentration.

by guest
DO Concentration (mg/L)
SVI > 150 mL/g
4
SVI > 150 mL/g

2
0 2 4 6 8 10 12
SRT (d)
Figure 4.4 DO concentration needed to maintain SVI below 150 mL/g in a complete mix aeration basin
treating settled municipal sewage (after Palm et al., 1980).
Reactor mixing characteristics

Continuous flow systems having a low degree of longitudinal mixing (plug flow or a large number of cells in
series) are less prone to bulking than complete mix systems, reportedly because the higher substrate
concentration at the inlet end of the reactor gives the floc-forming bacteria a competitive advantage over
the filamentous bacteria. The use of small compartments (selectors) at the head of the reactor can provide
the same effect. Anaerobic, anoxic and aerobic selectors have all been used but are not always successful
in providing a low SVI.
4.3.2 BNR processes

Sludge settleability in BNR processes seems to be governed predominantly by process anoxic fraction,
influent RBCOD and chemical dosing.
Anoxic fraction
In a study of seven full-scale BNR plants in South Africa, Casey and Alexander (2001) found a significant
inverse correlation between average DSVI and process aerobic mass fraction. In these compartmentalised
plants the DSVI increased from 61 mL/g at an aerobic mass fraction of 76% to 162 mL/g at an aerobic
fraction of 44%. Figure 4.5 shows the results.
Hartley and Lant (2010) reported another study of sludge settleability in eight full-scale Australian BNR
plants. Process formats covered a wide range including ND, EBPR and NDEBPR in continuous or
intermittently decanted formats, with or without primary treatment and prefermentation. None of the
plants had chemical dosing to the liquid stream processes. The first part of the study involved analysis of
one or two years of routine operating data from six continuous flow, compartmentalised BNR plants. In
the second part, the operating regime of two continuous inflow, intermittently aerated and decanted
plants was systematically varied over fifteen and eleven month periods and process behaviour monitored.
Settleability data were standardised to SSVI where necessary using Ozinsky & Ekama’s relationships.
For all eight plants it was found that SSVI increased with primary anoxic mass fraction in a consistent
manner. Correlation of SSVI with other process parameters was poor.

by guest
200
Daspoort
160
DSVI (mL/g) Rynfield

120
Vlakplaats
Goudkoppies
80 Northern
Olifantsvlei
Bushkoppie
40
0 0.2 0.4 0.6 0.8 1
Aerobic Fraction
Figure 4.5 Relationship between DSVI and process aerobic mass fraction (after Casey & Alexander, 2001).
Primary anoxic referred to the anoxic zone or period in which the biomass was mixed with process
influent or influent which had passed through an anaerobic zone only. Secondary anoxic referred to an
anoxic zone or period fed only with mixed liquor which had already passed through an aerobic zone or
period. It was demonstrated in the intermittent plants that secondary anoxic mass fraction (settle and
decant phase) had no effect on SSVI.
As an example of the data, Figure 4.6 shows the raw MLSS and SVI trends from one of the six
compartmentalised plants. Figure 4.7 shows the excellent correlation between SSVI and primary anoxic
fraction for all eight plants. Compartmentalised process formats included ND (one plant), EBPR (one
plant, primary anoxic fraction zero), and NDEBPR (four plants). The relationship shown is generally
consistent with Casey and Alexander’s (Figure 4.5). For each of the compartmentalised plants, variability
of the SSVI over the data period, expressed as the ratio of 90:50 percentiles, varied from 1.16 to 1.45
with a median ratio of 1.3.
700 7
MLSS
600 6
500 5
MLSS (g/L)
SVI (mL/g)
400 4
300 3
SVI
200 2
100 1
0 0
01-Jan 01-Feb 01-Mar 01-Apr 01-May 01-Jun 01-Jul 01-Aug 01-Sep 01-Oct 01-Nov 01-Dec 01-Jan
2002
Figure 4.6 Data trends for the Loganholme Stage 5–6 NDEBPR plant. Median values for MLSS, SVI,
equivalent SSVI: 4.4 g/L, 195 mL/g, 77 mL/g.

by guest
160
1.81x
140 y = 59.33e
120
SSVI (mL/g) 100
80
60
40 Continuous Flow
Intermittent-Jimboomba
20
Intermittent-Flagstone
0
0 0.1 0.2 0.3 0.4 0.5
Primary Anoxic Fraction
Figure 4.7 Overall correlation of compartmentalised and intermittent plant results.
In a study of full-scale Australian oxidation ditch plants, Hartley (2008) used the effluent ammonia:
nitrate ratio as a surrogate for the operating anoxic mass fraction. He found that SSVI was a minimum at
an optimum ammonia:nitrate ratio and increased at higher and lower ratios, as shown in Figure 4.8. The
optimum ratio varied between plants within the range 0.1–10 and SSVI could be controlled by
appropriate aeration control.
200
Bucasia Ditch 1(SRT 17d)
Bucasia Ditch 2(10d)
Coolum(30d)-APT Online
Coolum(30d)-APT Offline
150 West Byron(20d)

SRT 20d
SSVI or sSVI (mL/g)
SRT 10d
100
50
0
0.01 0.1 1 10 100
Effluent NH3-N:NO3-N Ratio (3SRT average, Nrave)
Figure 4.8 Relationship between SSVI and effluent ammonia:nitrate ratio in four oxidation ditches (Hartley,
2008).

by guest
The various data sets have somewhat different correlations because (a) SSVI is also affected by the
influent RBCOD fraction (see below), and (b) ammonia:nitrate ratio is affected not just by anoxic
fraction, but also by several factors affecting the ammonia concentration; these include SRT, DO
concentration in the aerobic zone, pH, temperature and nitrifier maximum specific growth rate (affected
by sewage composition).
All of these results show that sludge settleability in the BNR plants studied (compartmentalised,
intermittent and oxidation ditch process formats) was governed largely by primary anoxic fraction. This
is consistent with the UCT hypothesis, described below, that settleability is related to the biochemistry of
denitrification. The oxidation ditch, with a very large recirculation between the aerobic and anoxic zones,
appears to differ from the other process formats in that lowest SSVI occurs at some positive anoxic
fraction rather than zero.
Influent RBCOD
For ND and NDEBPR processes, the Water Research Group at the University of Cape Town (UCT)
conducted a comprehensive settleability study which they published in a lengthy series of papers and
summarised in Ekama et al. (1996) and Casey et al. (1999). They found that the low F/M group of
filaments (such as Microthrix parvicella and 0092) caused practically all bulking problems in long SRT
nutrient removal processes and that they could not be controlled by anaerobic, anoxic or aerobic
selectors. Whether or not the process included biological phosphorus removal, it was the operating
nitrogen regime which governed sludge settleability.
According to the model they developed, in a two-stage anoxic-aerobic system utilising mainly slowly
biodegradable COD, if denitrification in the primary anoxic zone was incomplete and nitrite was present,
the intracellular intermediate nitric oxide persisted and inhibited the denitrifying floc-formers in the
aerobic zone. The filamentous organisms, which did not compete well in the anoxic zone because they
could only denitrify nitrate to nitrite, were not inhibited in the aerobic zone where they proliferated.
Intracellular nitric oxide did not accumulate when the COD was readily biodegradable (RBCOD) so a
significant fraction of RBCOD relieved the competitive pressure on the floc-formers.
Data from the Coolum oxidation ditch BNR plant (Figure 4.8) are consistent with this model (Hartley,
2008). The ditch was preceded by an anaerobic reactor (process mass fraction 0.10) for biological
phosphorus removal. Because sewage RBCOD was low (RBCOD:total COD about 0.05), an in-line
activated primary tank (APT) was provided for prefermentation of the influent. Operating data for the
three years 2001–03 are shown in Figure 4.9. For the 20 months 3-Jul-01 to 3-Mar-03 the APT was
off-line for maintenance. The sSVI increased when the APT was taken off-line and decreased again
when it was returned to service. With the APT on-line and off-line the median sSVIs were 100 and
158 mL/g respectively.
300
sSVI (mL/g)
200
100
APT Off-Line
0
01-Jan-01 30-Jun-01 27-Dec-01 25-Jun-02 22-Dec-02 20-Jun-03 17-Dec-03
Figure 4.9 Effect of prefermentation on sSVI in the Coolum NDEBPR oxidation ditch process.

by guest
Prefermented influent RBCOD was not measured at Coolum but the RBCOD:total COD ratio monitored
in APT effluent at a sister plant at Thorneside (Brisbane) averaged about 0.25 (Hartley et al., 1999; alluded
to in Section 3.6.3, Sulfide).
Chemical dosing
When biological phosphorus removal is supplemented by alum or iron dosing to the mixed liquor, the MLSS
concentration increases because of the accumulation of chemical precipitate and simultaneously the SSVI
decreases. A useful rule of thumb is that the sludge settling rate is unchanged by chemical dosing.
Figure 4.10 shows the changes in MLSS and sSVI when alum dosing was introduced to the NDEBPR
Maroochydore B Plant. After initiation, the dose rate was eased back to achieve the desired effluent
phosphorus concentration (not shown). Over a period of about three SRTs the MLSS increased and the
sSVI declined. Median data for the period before dosing, and for the eventual period of more or less
steady state performance after dosing, are tabulated below the trend plot. The MLSS increased from 4.3
to 5.7 g/L while the sSVI fell from 91 to 64 mL/g. The associated sludge settling rates calculated from
the Ozinsky & Ekama equations are listed, showing that the settling rate was essentially unchanged after
the introduction of alum dosing.
120 7
MLSS
Alum Dose (mg solid/L)
100 6
5
sSVI (mL/g)
80
MLSS (g/L)
4
60
Start Alum Dosing
sSVI 3
40
2
20 1
Alum Dose
0 0
1-Nov-03 1-Dec-03 1-Jan-04 1-Feb-04 1-Mar-04 1-Apr-04 1-May-04 1-Jun-04
Figure 4.10 Change in MLSS and sSVI with the introduction of alum dosing to the Maroochydore B Plant.
Alum is dosed to the mixed liquor, with dose rate expressed relative to the sewage flow. Process SRT was
25 days. The time required for the full effect to occur (29-Jan to 4-Apr) was about 2.7SRTs, consistent with
the 3 residence time rule for biomass changes in complete mix systems (see Section 6.1).
4.4 EFFECT OF SVI ON EFFLUENT QUALITY

There is a strong relationship between effluent suspended solids concentration and SVI. Flocculation
improves as filamentous growth increases, reducing the loss of dispersed solids in the effluent. This is
illustrated for three different process types in Figure 4.11.

by guest
60
Pipes (1979) - conventional AS
Jimboomba - ND intermittent
50 West Byron - NDEBPR OD
Effluent SS (mg/L)
40
30
20
10
0
0 50 100 150 200 250 300 350 400
SVI (mL/g)
Figure 4.11 Relationship between effluent SS and SVI for three different process types, conventional
activated sludge, ND continuous feed intermittent decant (Hartley & Lant, 2010) and NDEBPR oxidation
ditch (Hartley 2008). The oxidation ditch reactor had a median chemical precipitant dosage of 51 mg/L as
solid alum which tended to increase the effluent SS concentration (see Chapter 7, Section 7.4.1).
4.5 FOAMING
Foaming is not directly related to settleability but is another sludge characteristic prone to causing operator
concern. There is an extensive literature on this subject (for example, Jenkins et al, 2004) but no universal
approach to control.
Foam or scum is commonly formed by the actinomycete group of filamentous bacteria of
which Microthrix and Nocardia are two of the most common. These organisms are slow growing
heterotrophs mainly restricted to using long chain fatty acids (LCFAs) and lipids (fats) in treatment
plants (Andreasen & Nielsen, 2000). They flourish in BNR plants because they are able to take up and
store sparingly soluble LCFAs under anaerobic conditions and are able to grow both anoxically and
aerobically. They increase their competitive advantage at lower temperatures when (a) the solubility of
LCFAs and lipids decreases further, and (b) predation by rotifers is reduced (Pajdak-Stos & Fialkowska,
2012).
From a practical perspective:
(1) The concentration of LCFAs and lipids in the process influent varies depending on catchment inputs
and the degree of removal by prefermentation upstream of the biological process.
(2) These filamentous organisms have a foam-forming propensity in aerated reactors and there is a
dynamic equilibrium between the proportion of foam-formers in the liquid phase and in the
surface foam.
(3) This equilibrium varies from time to time, presumably because of changes in bioreactor chemistry
caused by changes in process influent quality. As a result the extent of surface foam can vary rapidly
as illustrated by the example monitoring results depicted in Figure 4.12.

by guest
(4) The most positive method of foam control is mechanical removal from the surface of both
bioreactors and clarifiers. In the reactor, this reduces foam-former numbers in both the surface
and liquid phases because of the dynamic equilibrium between the two phases. The SRT of the
foam-formers is reduced, enabling their numbers to be controlled.
OD Scum (% of straight)
240
160
80
0
1-Oct-11 1-Dec-11 1-Feb-12 1-Apr-12 1-Jun-12 1-Aug-12 1-Oct-12 1-Dec-12 1-Feb-13
Figure 4.12 Extent of oxidation ditch scum coverage at the Brunswick Valley STP. The chain and flight scum
harvester ran for preset periods each day, generally totalling about 40 minutes. Length of channel scum
coverage upstream of the harvester was quantified relative to the length of straight channel between the
harvester and the first upstream bend (see Figure 6.11). Process SRT averaged approximately 30 days.
The behaviour is consistent with the premise of greater scum former advantage in winter.

by guest
Chapter 5
BNR flowsheets
The development of biological nutrient removal technology has been marked by a succession of different
process flowsheets aimed variously at overcoming limitations, improving performance, making operation
simpler or more flexible, minimising cost or differentiating products. The main configurations developed
are shown in Figure 5.1 and the logic is explained below.
In these flowsheets the recycle ratios are expressed as:
recycle ratio = recycle flow/influent flow

where
s = recycle ratio for RAS recycle (or s-recycle)

a = recycle ratio for mixed liquor recycle (or a-recycle)
r = recycle ratio for anaerobic recycle (or r-recycle)
Although the flowsheets are drawn for continuous flow processes with separate spatial zones and secondary
clarifiers, most of the basic process configurations can be implemented, in whole or in part, as sequential
time steps in sequencing batch reactors.
5.1 NITROGEN REMOVAL

5.1.1 Modified Ludzack-Ettinger (MLE)
See Figure 5.1a. This is the basic nitrogen removal process described in Chapter 3 and shown in Figure 3.20
(Ludzack & Ettinger, 1962; Barnard, 1973). Given a sufficient anoxic fraction, the mixed liquor recycle ratio
limits the degree of nitrogen removal which can be achieved because it sets the proportion of nitrified liquor
recycled for denitrification.
5.1.2 Bardenpho
See Figure 5.1b. The nitrogen removal performance of the MLE process can be increased by adding
secondary anoxic and aerobic zones (Barnard, 1973). COD may be dosed to the secondary anoxic zone.
The secondary aerobic or reaeration zone is intended to strip nitrogen bubbles from the floc ahead of the
clarifier and oxidise ammonia released by endogenous respiration in the secondary anoxic zone.

by guest
5.2 PHOSPHORUS REMOVAL

5.2.1 Phoredox
See Figure 5.1c. This is the basic phosphorus removal process described in Chapter 3 and shown in
Figure 3.27 (Barnard 1976a). It is also known as the A/O process (for anaerobic/oxic). This process is
only viable at low temperatures because it has to be operated in a non-nitrifying mode to protect the
anaerobic zone from recycled nitrate.
5.3 NITROGEN AND PHOSPHORUS REMOVAL

5.3.1 Modified Bardenpho
See Figures 5.1d and e. N and P removal can be integrated by adding an anaerobic reactor ahead of the
primary anoxic reactor in the MLE or Bardenpho processes (Barnard, 1976b). These configurations have
become known as 3-stage and 5-stage Modified Bardenpho respectively. The 3-stage process is also
known as A2O (for anaerobic/anoxic/oxic).
A disadvantage of the 3-stage process is a propensity for nitrate recycle in the RAS, reducing the
effectiveness of the anaerobic zone. This is overcome in the 5-stage process by removal of the residual
nitrate in the secondary anoxic zone (to which COD may be dosed).
5.3.2 Johannesburg (JHB)

See Figure 5.1f. When full N removal is not required, an alternative method of protecting the anaerobic
zone of a 3-stage process from recycled nitrate is to add an anoxic reactor to the RAS recycle (which
supplements the sludge mass in the secondary clarifier). This approach was developed at the
Johannesburg treatment plant (Osborn & Nicholls, 1978). The RAS anoxic reactor is smaller than the
secondary anoxic reactor in the 5-stage Modified Bardenpho process because (a) only part of the residual
nitrate, that in the RAS stream, has to be denitrified, and (b) the RAS SS concentration is higher than the
MLSS concentration.
As shown in Figure 5.1f, the RAS reactor can be implemented as a pre-anoxic zone to which part
of the sewage can be fed to enhance RAS denitrification and ensure the anaerobic zone is fully
anaerobic.
5.3.3 University of Cape Town (UCT)

See Figure 5.1g. The UCT format was developed to enable the anaerobic zone to be positively protected
from recycled nitrate over a wide range of loading and operating conditions (Rabinowitz & Marais,
1980). In this process the RAS is denitrified in the primary anoxic zone and a third r-recycle is fed back
to the anaerobic zone. However, the a-recycle requires close control to ensure the anoxic zone is not
overloaded with recycled nitrate which will then be recycled to the anaerobic zone.
5.3.4 Modified UCT (MUCT)

See Figure 5.1h. The operating flexibility of the UCT process was improved by splitting the primary anoxic
zone into two as shown in Figure 5.1h (Siebritz et al., 1982; UCT et al., 1984). This completely separates the
nitrogen and phosphorus removal cycles of the process, allowing them to be optimised separately. When
operated in a short SRT mode this process has sometimes been called VIP (for Virginia Initiative Plant,
Daigger et al., 1987).

by guest
BNR flowsheets 85
5.3.5 Westbank
See Figure 5.1i. A variety of other formats have been used to suit specific circumstances. The Westbank
format is an example which was developed in Canada and first applied in the small town of Westbank
(USEPA, 2009). It provides a high level of operational flexibility to adjust the process to best suit the
actual influent characteristics. It includes prefermentation of primary sludge to enhance the level of
RBCOD in the influent. Westbank is basically a Johannesburg process with step feed of influent to the
pre-anoxic, anaerobic and primary anoxic zones. The fermented primary sludge stream can be fed to the
anaerobic zone either directly or after separation of the solids for separate processing. The step feeds can
be adjusted to optimise nitrogen and/or phosphorus removal and to protect the anaerobic zone from
storm flows.
The primary clarifier and sludge fermenter can be replaced by an activated primary tank (APT)
combining both clarification and sludge fermentation (Barnard, 1984; Hartley et al., 1999).
5.3.6 Membrane Bioreactor (MBR)

See Figure 5.1j. In all of the preceding formats there is a sludge fraction in the clarifier which acts like
the RAS anoxic zone in the Johannesburg process. In the MBR the membranes replace the clarifier and
most commonly they are submerged in the sludge. The effluent is referred to as permeate. For high
levels of nitrogen removal the membranes need to be placed in a separate membrane zone which is
usually highly aerobic due to aeration used for membrane cleaning. The membrane zone therefore
forms a secondary aerobic zone and the membrane recycle (equivalent to the RAS in the clarifier
formats) is usually directed to the primary aerobic zone to protect the primary anoxic zone from the
high DO recycle.
The membrane recycle ratio also tends to be quite high, 4 say, to minimise the overall process volume by
minimising the difference in MLSS concentration between the membrane zone and the upstream zones.
In a similar way, membranes can be incorporated in any of the BNR formats. In all formats which include
a secondary anoxic zone the primary aerobic zone upstream can incorporate a de-aeration zone to reduce the
dissolved oxygen concentration in the anoxic zone feed.
(a) MLE
Primary Aerobic
anoxic zone
zone
Mixed liquor recycle, a
Influent Effluent
Clarifier
WAS
RAS recycle, s
Figure 5.1 The major BNR flowsheets. Other combinations and variants are also in use.

by guest
(b) Bardenpho
Primary Aerobic Secondary Reaeration
anoxic zone anoxic zone
zone zone
Influent Effluent
Clarifier
WAS
RAS recycle, s
PHOSPHORUS REMOVAL
(c) Phoredox
Anaerobic Aerobic
zone zone
Influent Effluent
Clarifier
WAS
RAS recycle, s
NITROGEN & PHOSPHORUS REMOVAL

(d) 3-Stage Modified Bardenpho
Anaerobic Anoxic Aerobic
zone zone zone
Influent Effluent
Clarifier
WAS
RAS recycle, s
Figure 5.1 The major BNR flowsheets. Other combinations and variants are also in use (Continued).

by guest
BNR flowsheets 87
(e) 5-Stage Modified Bardenpho

Anaerobic Primary Aerobic Secondary Reaeration
zone anoxic zone anoxic zone
zone zone
Influent Effluent
Clarifier
WAS
RAS recycle, s
(f) Johannesburg
Pre-anoxic Anaerobic Primary Aerobic
zone zone anoxic zone
zone
Influent Mixed liquor recycle, a
Effluent
Clarifier
WAS
RAS recycle, s
(g) UCT
zone zone zone
Anaerobic Mixed liquor
recycle, r recycle, a
Influent Effluent
Clarifier
WAS
RAS recycle, s

by guest
(h) Modified UCT
Anaerobic Anoxic Anoxic Aerobic

zone zone 1 zone 2 zone

recycle, r recycle, a
Influent Effluent
Clarifier
WAS
RAS recycle, s
(i) Westbank
zone
Mixed liquor
Influent Primary effluent recycle, a
PST
Effluent
Fermenter Clarifier
Fermented
primary sludge WAS
RAS recycle, s
(j) MBR-BNR
Anaerobic Primary Primary Secondary Membrane
zone anoxic aerobic anoxic zone
zone zone zone (secondary
aerobic)
Mixed liquor
recycle, a
Influent Permeate
WAS
Membrane
Anaerobic recycle, s
recycle, r

by guest
Chapter 6
Reactors
The biological reactor (or bioreactor) is the heart of the process, where treatment takes place. This chapter
looks firstly at reactor mixing and aeration characteristics, and then discusses the basic behavioural
characteristics of four major categories of bioreactor – compartmentalised, oxidation ditch, sequencing
batch and membrane.
6.1 MIXING
Different types of reactors are characterised by their mixing patterns. There are two separate aspects to
mixing: mixing regime and mixing intensity. These are discussed in turn.
6.1.1 Mixing regime

Mixing regime relates to the residence time distribution (RTD) of water passing through the reactor. There
are three different classifications of mixing regime – complete mix, plug flow and arbitrary flow
(Levenspiel, 1972).
Complete mix
In the ideal complete mix reactor the inflow is instantaneously mixed uniformly throughout the whole
volume. This type of reactor is commonly depicted as shown in Figure 6.1a. A complete mix reactor
therefore has the following properties:
• The reactor contents are of uniform quality throughout.
• The quality of the contents and the quality of the effluent leaving the reactor are identical.
(a) (b)
Figure 6.1 (a) Complete mix and (b) plug flow reactors.

by guest
Because the inflow is mixed through the whole volume, some of the influent leaves the reactor with the
effluent immediately. On the other hand, because the inflow remaining in the reactor at any time is mixed
uniformly, other parts of the inflow remain in the reactor for a very long time. The RTD can be determined
by injecting a slug of inert chemical tracer into the inflow, representing a parcel of inflow, and measuring its
concentration in the outflow over time. The complete mix RTD is shown in Figure 6.2a.
(a)
Effluent Tracer Concentration

39% of the influent has left
1.0 within half the MRT
0.5 13% of the influent remains longer

than twice the MRT
0 0.5 1.0 1.5 2.0

Mean Residence Time, Time
MRT (V/Q) (fraction of MRT)
(b)
All parcels of influent

leave after the same
residence time
0 1.0 2.0
MRT Time
(c)
2.0
Large
Dispersion
(N = 5)
1.5 Plug Flow
Complete
Mix
Small
(N = 1)
Dispersion
1.0
Intermediate
Dispersion
(N = 20)
0.5
0 0.5 1.0 1.5 2.0

MRT Time
Figure 6.2 Residence time distributions for (a) complete mix, (b) plug flow, and (c) arbitrary flow reactors. N is
the equivalent number of complete mix reactors in series.

by guest
Reactors 91
The rates of many process reactions approximate first order kinetics, in which the rate of consumption of
a reactant is proportional to its concentration:
dC/dt = kC (6.1)
where
C = reactant concentration
k = rate constant, time−1
For a first order removal reaction in a complete mix reactor, the steady state effluent concentration of the
reactant is given by:
CE = CI /(1 + kt) (6.2)

where
CE = effluent concentration
CI = influent concentration
t = reactor mean hydraulic residence time, V/Q
V = reactor volume
Q = flow rate
Plug flow
The plug flow reactor is the opposite of complete mix and is depicted in Figure 6.1b. There is no mixing of
influent with the reactor contents and all parcels of inflow take the same time to pass through. The RTD for
the plug flow reactor is shown in Figure 6.2b.
For a first order removal reaction in a plug flow reactor, the steady state effluent concentration of the
reactant is given by:
CE = CI e−kt (6.3)
where the nomenclature is the same as in Equations 6.1 and 6.2.
A batch reactor has a similar performance characteristic to a plug flow reactor but reaction conditions
vary with time rather than space. The batch reactor is filled, reactions take place over time and the
reactor is then emptied. As for the plug flow reactor, all parcels of inflow remain within the reactor for
the same time.
Arbitrary flow
The complete mix and plug flow regimes represent the extremes of a range. In between is arbitrary flow,
where an intermediate degree of mixing takes place. Typical RTDs are shown in Figure 6.2c. Most real
reactors fit into this intermediate category.
Within a continuous flow reactor, the internal mixing which governs the shape of the RTD is referred to
as longitudinal dispersion, or backmixing. Similar types of RTDs are produced by placing complete mix
reactors in series or by recycling from the outlet to the inlet of a plug flow reactor.
For first order reactions, the effluent concentration from a plug flow reactor is significantly lower than
from a complete mix reactor. If a complete mix reactor is subdivided into N complete mix cells in series

by guest
the effluent concentration, as given by Equation 6.4, approaches plug flow quality for large N:
CE = CI /(1 + kt/N)N (6.4)
where
t = mean hydraulic residence time of the total reactor
In nutrient removal plants, the two processes most sensitive to the mixing regime are the fermentation
reactions within the anaerobic reactor and nitrification within the aerobic reactor. The effect of
subdividing the anaerobic reactor is illustrated in Chapter 3, Figure 3.29, while the effect of the aerobic
reactor mixing regime on nitrification is illustrated in Figure 6.3. In contrast, denitrification is a zero
order reaction over the usual nitrate operating range, meaning that denitrification rate is not affected by
nitrate concentration and complete mix conditions are as effective as plug flow. However reactor
subdivision is often used to prevent short circuiting under non-ideal mixing conditions.
Relative Effluent Ammonia Concentration
0.8
0.6
0.4
0.2
0
0 1 2 3 4 5
No. of CM Cells in Series
Figure 6.3 Typical effect of subdividing a nitrifying reactor on effluent ammonia concentration.
The complete mix regime has a particular advantage when the plant load is highly variable, due to
industrial inputs for example: complete mix reactors provide resilience to shock loads because the
biomass is exposed only to effluent-level, rather than influent-level, concentrations. In addition, spatial
oxygen demand is uniform in complete mix reactors but varies in plug flow reactors.
Poor mixing occurs in various ways, the most common being dead (inactive) volume, partial short
circuiting (bypassing) around the active volume and internal recirculation. Tracer testing can be used to
determine the actual mixing characteristics of a plant reactor (Levenspiel, 1972; Monteith and
Stephenson, 1981). Practical operational targets for a complete mix reactor could be set at, for example:
active volume fraction .90%, bypass flow ,5%. Guidelines for the conduct of a tracer test are set out
in Table 6.1.

by guest
Reactors 93
Table 6.1 Tracer testing guidelines.
1. These guidelines relate to the testing of a simple flow-through reactor. More complex flow patterns require special
consideration.
2. Select an inert tracer which has a low background concentration. Fluorescent tracers used in clean water are not
generally suitable in operating sewage treatment plants. Lithium (in the form of LiOH) is a common choice but other
common elements such as fluoride have also been used. Check that the tracer is not chemically reactive in the test
situation, for example, fluoride reacts with aluminium and is not suitable where alum sludge is present.
3. Over the test period, operate the test reactor as close to steady state as possible. Record the flow rate over the full test
period.
4. Slug dose the reactor feed with sufficient tracer solution to allow accurate measurement of the tracer at 10% of its initial
concentration when uniformly distributed through the reactor volume.
5. Run the test for about 3HRTs (hydraulic residence times). Ideally, collect outflow grab samples at about the following
times (percentage of nominal HRT): just before dosing (for background tracer concentration), then 0.5%, 2%, 7% then
7% increments to 0.5HRT, then 13% increments to 1.5HRT, then 20% increments to 3HRT, making 30 samples
altogether. Record the sample collection times.
6. Analyse the results using a formula applicable to the hydraulic regime. A useful model for a simple reactor with a
complete mix active zone is shown below (Monteith & Stephenson, 1981). The complete mix portion of the results can be
analysed using:
C/CO = [(Q1 V/Q2 )/tA ]EXP(−t/tA )
where C, CO = tracer concentration at time t & time zero, and tA = VA/Q1
REACTOR
Tracer Bypass, Q2
Samples
Injection
Q Q1 Active Volume
VA
Dead Volume
VD
Total Volume, V = VA + VD
Example test results for an anaerobic digester with a hydraulic jet mixing system are plotted below (20 samples): analysis
gives active volume fraction 98.8%, bypass flow fraction −3.2%, HRT 15.2 days.
Digester 1
1.2 1200
Corrected for Co = 16.5 kg/2.5 ML
background Li = 11.55 mg/L Li
(0.012 mg/L)
1 1000
Excluded from regression

0.8 800
Flow (kL/d)
C/Co
y = 1.04544e-0.06656x
0.6 600
R2 = 0.971
0.4 400
Flow
0.2 200
0 0
0 5 10 15 20 25 30 35 40 45 50
Days Since Dosing

by guest
Time constant
A useful rule of thumb for the response of a complete mix process to a step change in input is illustrated in
Figure 6.4. If the process has a conservative steady state input, an alkali dose for example, and a step change
in input dose rate occurs, the output responds as shown. At a time equal to the process mean residence time,
τ, the change in output reaches 63% of the ultimate change. At three mean residence times it reaches 95% of
the ultimate. This is the basis for the rule of thumb that when a process change is made, it will take about
three residence times for the full effect to become evident.
C
Input
Output
ΔC 0.95ΔC dC = 1 – e-t/τ
0.63ΔC
Initial Concentration
0 τ 3τ t
Figure 6.4 Response of complete mix process output to a step change in input.
Although this rule is illustrated here in relation to flow, it will be seen later that it applies in other disparate
areas such as the process solids phase and the rate of performance improvement achieved in plant
optimisation programs. In non-residence time applications the mean residence time, τ, is referred to as
the time constant, equal to the time to achieve 63% of the ultimate change.
6.1.2 Mixing intensity

Mixing intensity is the level of turbulence. There are two separate aspects to mixing intensity in reactors: (a)
ensuring the mixed volume is fully active and the biomass (solids phase) is uniformly distributed, and (b) the
effect of the associated shear level on flocculation of the biomass.
Active mixing
Commonly accepted inputs required for active mixing are listed in Table 6.2.
Flocculation
Activated sludge flocs are complex structures made up of bacteria, organic particles adsorbed from the
sewage and inorganic components, all embedded in a cloud of exopolymeric substances (EPS) produced
by the bacteria (Biggs, 2000; Wilen et al., 2010). They are considered to have three structural levels
(Biggs & Lant, 2000; Snidaro et al., 1997): individual bacterial cells (size in the order of 1 µm), bacteria
tightly bound together by a gel matrix to form microcolonies (median diameter around 13 µm), and

by guest
Reactors 95
polymer linkages of microcolonies to form flocs. As described at the start of Chapter 4, the flocs are
generally built around a backbone of filamentous bacteria which give them additional strength. The floc
particle size distribution is bimodal, with a small unflocculated cell and microcolony group in the 0.5–
5 µm range and a large particle size group in the 25–2000 µm range (Parker et al., 1971, 1972; Ekama
et al., 1997). Sludge settling rate and effluent SS concentration both relate to the particle size distribution
of the solids.
Table 6.2 Minimum power inputs for active mixing.
Type of system Minimum Equivalent Equivalent

mixing power velocity
gradient (G)
Unaerated reactors, mechanical mixers — 10 W/m3 100 s−1
Fine bubble diffused air, uniformly distributed 0.6 L/s per m2 5 W/m3 70 s−1
tank surface
Coarse bubble diffused air, spiral flow 0.3 L/s per m3 10 W/m3 in a 100 s−1
tank volume 4 m deep
tank
Mechanical aeration — 15 W/m3 120 s−1
Oxidation ditch
Mechanical aeration 0.2 m/s 5–10 W/m3 70–100 s−1
circulation
velocity
Diffused air; low speed propellers 0.2 m/s Propellers 50 s−1
circulation 2 W/m3
velocity
The size distribution of the flocs is determined by the turbulence level of their environment. Flocs grow
by turbulent aggregation but they also suffer breakage and surface erosion through turbulent shear; for a
particular turbulence level a given sludge has a steady state floc size distribution. Using sludge from the
Wacol nitrifying-denitrifying plant in Brisbane, Biggs & Lant (2000) showed that, over a G range of 19–
346 s−1, the steady state volume-weighted median floc diameter, d (in μm), varied inversely with the
velocity gradient, G (s−1), according to:
d = 450G−0.49 (6.5)
The 95:50 percentile diameter ratio was about 3. When diluted sludge samples were first flocculated
at various G values (19–113 s−1), then deflocculated completely to microcolonies (median diameter
13 µm) by sonication and finally reflocculated at the original G values, the median floc sizes increased
again to their original values, reaching 90% of their steady state diameters in 25–30 minutes. Repeated
sonication and reflocculation gave the same result. Therefore, as the sludge within a plant
encounters varying shear conditions, the degree of flocculation will vary; at normal MLSS
concentrations (an order higher than the test value) reflocculation after a high-G zone will normally take
only a few minutes.
The sludge flocculation characteristics also depend on a range of other conditions which are discussed in
Chapter 7, Secondary Clarifiers.

by guest
6.2 AERATION
The reactor aerators are the lungs of the treatment process – they transfer oxygen to the water and strip the
carbon dioxide produced; the aerators also provide mixing for full reactor utilisation and are one of the
plant’s major energy consumers.
6.2.1 Oxygen transfer rate

Oxygen transfer occurs predominantly by diffusion of oxygen from air bubbles mixed with the water.
Aeration performance is measured using two main parameters – oxygen transfer rate (OTR, kgO2/h) and
oxygen transfer efficiency (OTE, kgO2/kWh). Oxygen transfer rate is proportional to the dissolved
oxygen saturation deficit and, in clean water, can be described by the following equation (Metcalf &
Eddy, 2003):
rC = dC/dt = KL a(CS − CW ) (6.6)
where
rC = rate of change in concentration, mg/L · h
K La = overall oxygen mass transfer coefficient, h−1
CS = saturation concentration, mg/L
CW = dissolved oxygen concentration, mg/L
In an activated sludge system where the biomass exerts an oxygen demand:
rC = KL a(CS − CW ) − rM (6.7)
where
rM = rate of oxygen use by the microorganisms, mg/L · h
Under steady state conditions the rate of oxygen transfer equals the rate of oxygen utilisation and the DO
concentration remains constant. Then rC = 0 and:
rM = KL a(CS − CW ) (6.8)
Under operating conditions the transfer rate is modified by several factors as follows:
• KLa varies with temperature according to:
KL aT = KL a20 uT−20 (6.9)

where
KLaT and KLa20 = KLa values at T°C and 20°C
θ = temperature correction factor, 1.024
In practice, the transfer rate does not change significantly with temperature because the change in
mass transfer coefficient is almost counterbalanced by an opposite change in saturation
concentration. Tables of oxygen solubility can be found in Standard Methods, 1998.
• KLa varies with the type of aeration device, reactor geometry, mixing intensity and wastewater
characteristics. These aspects are accounted for by the alpha factor:
a = KL a (wastewater)/KL a (clean water) (6.10)

by guest
Reactors 97
Values of alpha vary with all the factors noted above and can range from 0.3 to 1.2. Typical values are
0.4 to 0.8 for diffused air and 0.6 to 1.2 for mechanical aeration.
• Fine pore diffusers, both ceramic and membrane, generally suffer from some degree of external
fouling by biological slimes and chemical precipitates. Internal fouling can also occur from
excessive dust in the airflow or from sludge which enters the piping through leaks during air
outages. The reduction in transfer rate resulting from external fouling is accounted for by a fouling
factor, F (USEPA, 1989; Hartley, 1990; Hartley & Gobbie, 1991). The fouling factor is generally
applied in conjunction with alpha to give an overall factor, αF.
Fouling increases with time and can vary widely depending on industrial components of the sewage,
chemical composition of the mixed liquor, the process SRT, the uniformity of conditions through the
reactor and the diffuser characteristics and history. Biological fouling is the most common. Fouling is
slower at longer SRT (Rosso & Stenstrom, 2006). Hartley & Gobbie (1991) found for three plants that
the decline in performance caused by fouling followed a semi-logarithmic relationship. Taking the
value of F as 1 at a service life of one month, they found that each tenfold increase in service time
decreased F by 3–13%. In addition, each doubling of service time increased diffuser pressure loss at
constant airflow by 10–38%.
For diffuser pressure loss:
P = AtM (6.11)
where
P = diffuser element pressure loss, kPa
t = diffuser service time, months
A = constant equal to pressure loss at a service time of one month, kPa
M = pressure increase coefficient, kPa month units
An unpublished study (Hartley, 2002) using data from 32 treatment plants found that the M value
varied from 0.041 to 0.46 as shown in Figure 6.5. A range of ceramic and membrane diffuser types
were included but the fouling rate did not correlate with the diffuser material. Fouling rates were
highest when the sewage contained a significant organic industrial component (such as abattoir,
dairy or poultry wastewaters).
0.5
Fouling Coefficient, M
0.4
(kPa, month units)
0.3
0.2
0.1
0
1 10 100
% Equal or Greater
Figure 6.5 Range of diffuser fouling rates.

by guest
• Oxygen saturation concentration varies with the chemical composition of the wastewater. This is
accounted for by the beta factor:
b = CS (wastewater)/CS (clean water) (6.12)
The value of beta varies from 0.7 to 0.98 and is commonly assumed to be 0.95.
• Oxygen saturation concentration also varies with atmospheric pressure. This is of special
significance for plants located at high altitude and is accounted for by using an appropriate value
of CS.
Aerator performance is rated at so-called standard conditions, defined as zero DO concentration,
20°C (10°C in Europe), standard atmospheric pressure (101.3 kPa) and clean water.
The field oxygen transfer rate, OTR, is related to the standard oxygen transfer rate, SOTR, by:
OTR = SOTR aF uT−20 (bCS − CW )/CS20 (6.13)

where
CS = oxygen saturation concentration in clean water at operating temperature and pressure

CS20 = oxygen saturation concentration in clean water at 20°C and one atmosphere pressure
For surface aeration the operating pressure is taken as atmospheric pressure. However, for diffused air
where the diffusers are submerged at greater than 90% of the reactor water depth, the tank average
saturation concentration corresponds to a water depth of 20–45% (commonly 40%) of the full depth and
the operating pressure is taken as the absolute pressure at that depth (US EPA, 1989).
Oxygen transfer rate is controlled by varying the power input to keep the DO concentration constant at a
set point value. In mechanical aeration systems the impeller immersion, speed or number of impellers
operating can be varied while in diffused air systems the air flow is varied. A typical SOTR
characteristic for fine pore ceramic dome diffusers is shown in Figure 6.6.
1400 20
SOTR (gO2/d/m depth/dome)
SOTE (gO2/m3 air/m depth)
1200 18
Min. Recommended
Min. Recommended
Peak Transfer
1000 16
800 14
600 12
400 10
0 2 4 6 8 0 1 2 3 4
Standard Air Flow (m /h/dome) 3
Standard Air Flow (m3/h/dome)
Figure 6.6 Typical performance characteristics for fine pore ceramic dome diffusers (Hartley, 1990). Range of
air flow is 6.5:1 for a range in transfer rate of 3.4:1. SOTE (%) = (gO2/m3 air)/2.77.

by guest
Reactors 99
6.2.2 Oxygen transfer efficiency

Two parameters are used to describe oxygen transfer efficiency:
• Standard oxygen transfer efficiency (SOTE), used for diffused air systems to describe the percentage
of oxygen which dissolves in clean water under standard conditions. Under field conditions this is
simply called oxygen transfer efficiency (OTE). OTE is greater at higher SRT as oxygen transfer
inhibitors such as surfactants are more completely degraded (Rosso & Stenstrom, 2006). SOTE
depends on water depth, air flow rate per diffuser pore (which governs bubble size) and diffuser
layout (which governs bubble travel time to the surface). Typical values are 5–8% per metre depth
for fine bubble systems and 2–3% per metre depth for coarse bubble systems (Metcalf & Eddy, 2003).
• Standard aeration efficiency (SAE), used for both mechanical surface aeration and diffused air
to describe energy efficiency (kgO2/kWh) at standard conditions (ASCE, 2007). Typical values
(wire to water) are 2 kg/kWh for mechanical aeration and up to 4 kg/kWh for fine pore diffused air.
Typical performance characteristics for fine pore ceramic dome diffusers are shown in Figure 6.6. Note that
air at 20°C and standard atmospheric pressure has a density of 1.20 kg/m3 and its oxygen content is 23% by
mass (276 gO2/m3 air).
Fine pore diffusers need periodic cleaning to restore aeration capacity and reduce energy consumption. A
study of three plants using ceramic diffusers (Hartley, 1990; Hartley & Gobbie, 1991) concluded that
refiring was the most effective cleaning technique, and provided sufficient blower capacity was
available, the most economic cleaning interval was 5–10 years. For plants using membrane or sintered
plastic diffusers which could be cleaned satisfactorily by hosing, Rosso & Stenstrom (2006) concluded
that annual cleaning was the most economic. Blower power and diffuser back-pressure should be
monitored to track the degree of fouling.
6.2.3 Carbon dioxide stripping

The mixed liquor pH is governed by the concentration of dissolved carbon dioxide produced by COD
oxidation as explained in Section 3.8. Carbon dioxide is stripped by the aeration air bubbles. However,
the stripping rate and therefore the mixed liquor pH varies with the airflow rate which is governed by the
oxygen transfer efficiency. The net effect on pH is illustrated in Figure 6.7. The figure shows the effect
of diffused air oxygen transfer efficiency on pH for a range of influent total alkalinity:COD ratios for
both sewage and primary effluent. The sewage lines are drawn assuming full nitrification and
denitrification; however for primary effluent it is assumed only 60% of the nitrate is denitrified due to
the higher TKN:COD ratio (see Figure 3.21). Observations are as follows:
• Reactor pH falls as oxygen transfer efficiency increases and as feed alkalinity decreases.
• Reactor pH is lower with primary treatment because of the decrease in alkalinity recovery associated
with denitrification.
• Alkalinity dosing required to increase pH can be determined from the graph.
Mixed liquor leaving the reactor is still supersaturated with carbon dioxide and pH rises through the
clarifiers due to further stripping from the surface and effluent weirs. At loads below plant capacity it
may be possible to increase mixed liquor pH by reducing the number of diffusers operating. This
increases the air flow per diffuser, reducing transfer efficiency and increasing total air flow required
(Figure 6.6).
Mechanical aerators have lower oxygenation efficiencies (higher effective air flows) than fine pore
diffused air and will generally operate towards the left hand side of the pH graph.

by guest
7.5
Raw: full DN
PE: 60% DN
7.0 INFLUENT
TA:COD
Mixed Liquor pH
0.6
1.11
0.5
6.5
0.93
0.4
6.0
0.74
5.5
5% 15% 25% 35%
SOTE
Figure 6.7 Effect of oxygen transfer efficiency on reactor pH. Lines are shown for reactors treating raw
sewage or primary effluent. For raw sewage, nitrification and denitrification are complete; for primary
effluent, nitrification is complete but denitrification is only 60% complete. For each influent there are three
lines representing different influent alkalinities (native alkalinity plus any alkali dosing); the primary effluent
alkalinities are the same as the raw sewage values and the TA:COD ratios shown represent 46% COD
removal by primary treatment. The typical sewage and primary effluent qualities listed in Table 2.4 have
been used in the preparation of this figure.
6.3 COMPARTMENTALISED REACTORS

6.3.1 Description
Compartmentalised reactors separate the various process steps into individual compartments generally as
reflected in the BNR flowsheets in Chapter 5. Compartments may range from complete mix to plug flow
and are commonly subdivided for control of mixing characteristics. Swing zones and de-aeration zones
may be provided. A swing zone is located at the head of the main aerobic zone and can be operated
either aerobically or anoxically, providing flexibility to adjust mass fractions. A de-aeration zone is
located at the end of the main aerobic zone and operates at low DO concentration to minimise oxygen
transfer to both primary and secondary anoxic zones.
Equipment is provided for aeration, mixing, mixed liquor recycling, sludge wasting, scum harvesting and
chemical dosing. Instruments may be provided for monitoring and/or control of liquid and air flows, and
dissolved oxygen, pH, ammonia, nitrate, phosphate and MLSS concentrations.
This type of reactor can be the most straightforward to operate process-wise but provision of increased
operating flexibility generally brings increased process and equipment complexity.

The main operating characteristics of a compartmentalised process are set by the relative volumes of the various
zones and the arrangement and magnitudes of the recycles. The recycles are generally arranged to protect the
anoxic zones from recycled oxygen and the anaerobic zone from recycled oxygen and nitrate. Where mixed
liquor is recycled upstream within the reactor, MLSS concentration will decrease in the same direction.
Tables 6.3 and 6.4 list the equations for calculating MLSS concentrations from recycle ratios and for
calculating the actual operating recycle ratios from measured MLSS concentrations. These are needed for
effective SRT control. The tables also provide equations for sludge mass fractions which can be varied
by adjusting the recycle ratios. Sludge settleability in compartmentalised processes is affected by the
primary anoxic fraction (see Chapter 4).

by guest
Reactors 101
Table 6.3 UCT & Modified UCT process balance.

zone zone zone

recycle, rQ recycle, aQ
fVO, XO fXC
fVA, XA fVX, XO
Influent, Q Effluent
Clarifier
WAS
RAS recycle, sQ, XS
Mixed Liquor Suspended Solids (MLSS) concentrations

XS = XO [(1 + s)/s] (6.14a)
XA = XO [r/(1 + r)] (6.14b)
Recycle ratios
s = XO /(XS − XO ) (6.14c)
r = XA /(XO − XA ) (6.14d)
Sludge mass fractions
f XA = f VA (1 − f XC )[r/(1 + r)]/{f VA [r/(1 + r)] + f VX + f VO } (6.14e)
f XX = f VX (1 − f XC )/{f VA [r/(1 + r)] + f VX + f VO } (6.14f)
f XO = f VO (1 − f XC )/{f VA [r/(1 + r)] + f VX + f VO } (6.14g)
f XA + f XX + f XO + f XC = 1 (6.14h)
Volume fractions
f VA + f VX + f VO = 1 (6.14i)
where
a = mixed liquor recycle ratio relative to Q

r = anaerobic recycle ratio relative to Q
s = RAS recycle ratio relative to Q
fVA = anaerobic zone volume fraction
fVO = aerobic zone volume fraction
fVX = total anoxic zone volume fraction
fXA = anaerobic zone mass fraction
fXC = clarifier sludge mass fraction, typically in the order of 0.10
fXO = aerobic zone mass fraction
fXX = total anoxic zone mass fraction
Q = influent flow rate
XA = MLSS concentration in anaerobic zone
XO = MLSS concentration in aerobic zone
XS = RAS SS concentration
XX = MLSS concentration in anoxic zone

by guest
Table 6.4 Westbank & Johannesburg process balance.

zone
Mixed liquor
Influent, Q Primary effluent recycle, aQ
PST
f1Q f2Q f3Q fVO, XO fXC
fVP, XP fVA, XA fVX, XO
Effluent
Fermenter Clarifier
Fermented
primary sludge f4Q
WAS
RAS recycle, sQ, XS
Flow balance
f1 + f2 + f3 + f4 = 1 (6.15a)
Mixed Liquor Suspended Solids (MLSS) concentrations
XS = XO [(1 + s)/s] (6.15b)
XP = XO [(1 + s)/(f 1 + s)] (6.15c)
XA = XO [(1 + s)/(f 1 + f 2 + f 4 + s)] (6.15d)
Recycle ratio
s = XO /(XS − XO ) (6.15e)

F = f VP (1 + s)/(f 1 + s) + f VA (1 + s)/(f 1 + f 2 + f 4 + s) + f VX + f VO (6.15f)
f XP = f VP (1 + s)(1 − f XC )/[F(f 1 + s)] (6.15g)
f XA = f VA (1 + s)(1 − f XC )/[F(f 1 + f 2 + f 4 + s)] (6.15h)
f XX = f VX (1 − f XC )/F (6.15i)
f XO = f VO (1 − f XC )/F (6.15j)
f XP + f XA + f XX + f XO + f XC = 1 (6.15k)
Volume fractions
f VP + f VA + f VX + f VO = 1 (6.15l)
where
a = mixed liquor recycle ratio fXA = anaerobic zone mass fraction

s = RAS recycle ratio fXC = clarifier sludge mass fraction, typically in the order of 0.10
f1 = fraction of influent fed to pre-anoxic zone fXO = aerobic zone mass fraction
f2 = fraction of influent fed to anaerobic zone fXP = pre-anoxic zone mass fraction
f3 = fraction of influent fed to primary anoxic zone fXX = primary anoxic zone mass fraction
f4 = fraction of influent fermented and fed Q = influent flow rate
to anaerobic zone XA = MLSS concentration in anaerobic zone
fVA = anaerobic zone volume fraction XO = MLSS concentration in primary anoxic and aerobic zones
fVO = aerobic zone volume fraction XP = MLSS concentration in pre-anoxic zone
fVP = pre-anoxic zone volume fraction XS = RAS SS concentration
fVX = primary anoxic zone volume fraction

by guest
Reactors 103
6.3.3 Operating principles

The basic operating principles are as follows:
(1) Set the daily sludge wasting to maintain the desired SRT. The solids distribution within the reactor
and the associated WAS concentration need to be known for this purpose.
(2) Set appropriate DO concentrations in the aerobic and de-aeration zones.
(3) Set appropriate mixed liquor recycle ratios.
(4) For a Westbank type process, set an appropriate feed distribution.
(5) Set appropriate chemical dose rates for secondary denitrification, supplementary phosphorus
removal and pH control.
(6) Monitor performance and adjust process settings as required.
6.4 OXIDATION DITCHES

6.4.1 Description
The oxidation ditch (OD) is a bioreactor configured like a race track with a recirculating flow pattern. It was
invented by Pasveer in the Netherlands in 1954 (Pasveer, 1959; Baars, 1962) as a continuous feed,
intermittent decant (CFID) process. The initial plants for small communities were constructed as ditches
fitted with floating horizontal shaft brush type aerators and in the original format the reactors are still
commonly called Pasveer channels.
The OD reactor format has since developed into four main types:
Pasveer channel: CFID format, also commonly known as IDEA (intermittently decanted extended
aeration), used for small communities; no clarifiers.
Horizontal shaft OD: Continuous flow OD fitted with horizontal shaft mechanical aerators for aeration
and circulation. Usually has multiple aerobic and anoxic zones. Horizontal shaft aerators are not
generally available in powers above 45 kW, which limits the capacity of individual reactors.
Vertical shaft OD: Continuous flow OD fitted with vertical shaft mechanical aerators for aeration and
circulation. Vertical shaft aerators are available in powers of up to about 132 kW, allowing larger
reactors to be used. These reactors are commonly referred to as Carrousels after the trade name used
when the system was originally patented (since expired) by the Dutch inventors, DHV consultants,
in the 1960’s.
Diffused air OD: Continuous flow OD aerated with diffused air and circulated using large diameter
propeller mixers. Generally has one aerobic and one anoxic zone and is more energy efficient. The
aeration and circulation rates are independent. There are no equipment constraints on reactor size.

Figure 6.8 illustrates the basic mechanically aerated OD format. Mixed liquor is impelled around the
circuit by one or more mechanical aerators. To maintain the biomass in suspension the circulating
velocity needs to be a minimum of about 0.2 m/s at typical SSVI values. This circulation typically
provides a mixed liquor:feed flow ratio in the order of 100. The DO concentration jumps up at each
aerator and declines downstream due to the sludge respiration. With zero DO upstream, the DO
concentration at the aerator equals the oxygen transfer rate divided by the circulating flow rate. The
slope of the DO profile varies with the respiration rate and the anoxic fraction is controlled by varying
the aeration rate to maintain a DO set point at a probe located towards the downstream end of the
aerobic zone. The anoxic fraction can be varied by adjusting the set point.

by guest
Anaerobic reactor
Feed & RAS
Aerator Circulation
DO Probe
Mixed liquor to clarifier
Aerator Outlet Feed Aerator
DO
DO Probe
Aerobic Anoxic
Distance around circuit

Figure 6.8 Basic oxidation ditch reactor format.
The basic BNR OD process has the 3-stage Modified Bardenpho format as shown in Figures 5.1 and
6.9. The OD circulating flow is the mixed liquor recycle, with a ratio in the order of 100. Because of the
large recycle and the tapering DO profile, the OD can achieve full N removal without the need for a
secondary anoxic reactor. Provided the anoxic fraction is large enough and the influent COD:TKN is
sufficient, the minimum effluent nitrate N achievable is governed by the recycle ratio and is in the
order of 0.4 mgN/L at a recycle ratio of 100 (refer to Figure 3.21; Equation 3.12; Table 2.4). With
additional denitrification in the clarifier sludge layer, there is essentially no nitrate recycle to the
anaerobic reactor. In addition, because the mixed liquor recycle ratio is so high, the ammonia and
nitrate concentrations are almost constant, and equal to the effluent concentrations, throughout the
OD circuit.
The anoxic fraction of the OD governs both nitrogen removal and sludge settleability (see Chapter 4).
The process nitrogen removal characteristic is illustrated in Figure 6.10 (Hartley, 1997a). There is an
anoxic fraction operating window which gives lowest TN in the effluent. As anoxic fraction increases
nitrate decreases while ammonia increases. As a rule of thumb, lowest TN occurs where ammonia and
nitrate are equal.
Nitrate is governed predominantly by anoxic fraction and influent TKN:COD ratio with the minimum
achievable governed by the mixed liquor recirculation ratio. Ammonia is governed by a range of process
factors, primarily aerobic SRT, DO concentration, pH and temperature.

by guest
Reactors 105

zone zone zone
OD circulation, aQ
Inflow, Q
Clarifier
OXIDATION DITCH
WAS
RAS
Figure 6.9 OD 3-stage Modified Bardenpho format.
(a) 8
Total N
N Concentration (mg/L)
6 NOx-N
OPERATING
WINDOW
4
2
NH3-N
0
0 0.2 0.4 0.6 0.8 1
Anoxic Fraction
(b) 8
TK N
TKN:C
,18C
25d,7.0,20C
15d,7.0,18
:COD
15d,6.6
TN Concentration (mg/L)
OD 0.1
6
0.13
0
2
Nitrification parameters:
SRT, ML pH, temperature
0
0 0.2 0.4 0.6 0.8 1
Anoxic Fraction
Figure 6.10 (a) OD anoxic fraction operating window concept for producing minimum effluent TN. (b) Effects
of influent TKN:COD ratio and nitrification parameters on the width of the operating window and the minimum
achievable TN (typical).

by guest
The operating window shifts as operating conditions change. Figure 6.10 illustrates how changes in the
ammonia and nitrate limbs affect the window – changing its width, shifting the optimum anoxic fraction,
varying the effluent ammonia:nitrate ratio and changing the minimum effluent TN achievable.
Figure 6.11 shows how the anoxic fraction is controlled using the DO set point (SP). The OD shown in
this figure is aerated by a uniform diffused air zone. The DO rises through the aeration zone and then
declines downstream. The maximum DO at the end of the aeration zone equals the process oxygen
demand minus the oxygen consumption within the aeration zone, divided by the circulation flow around
the circuit. A short, high-intensity aeration zone and a low circulation rate are desirable to maximise the
average DO concentration. It can be seen that with a constant circulation velocity the DO concentration
increases with load, enhancing the nitrification rate and reducing effluent ammonia variability. The
average DO is lowest at plant startup when the connected population is lowest but nitrification
performance can be maintained by increasing the operating SRT.
(a)
Anaerobic zone Influent
Scum to
digester RAS
Scum DO Probe
To clarifiers
harvester
Mixer Aeration zone
(b)
DO
(mg/L) Outflow to Feed from
clarifiers anaerobic zone
4
Full load - max demand

3
DO probe
Circulation
2 - ave v = 0.25 m/s
- min DO SPs at startup & full load

1
Startup - ave demand

0
0 Circuit length, Lc 139m Distance
Aeration zone
Aerobic zone Anoxic zone
Figure 6.11 (a) 18,000 EP diffused air oxidation ditch (Brunswick Valley STP): OD 6m × 4m WD × 139m
circuit length, 3.3 ML volume; anaerobic reactor 0.37 ML, (b) DO profiles over the full range of
operating conditions.

by guest
Reactors 107
The effect of DO concentration on nitrification rate is non-linear and the effect of the tapering DO profile
on nitrification rate is given by:
mn = mnm {1 − (KO /DOmax )LN[1 + (DOmax /KO )]} (6.16)
where
μn = nitrifier average specific growth rate in tapering DO zone, d−1

μnm = nitrifier maximum specific growth rate, DO non-limiting, d−1
KO = nitrifier half velocity coefficient for DO, around 0.3 mg/L
DOmax = maximum DO concentration at the downstream end of a diffuser zone or at a mechanical
aerator, mg/L
The anoxic zone is nominally the circuit length devoid of oxygen downstream of the aerobic zone. The
length of the aerobic zone is controlled using a DO probe located towards its downstream end. The DO
SP controls the aeration blowers. In operation it is not necessary to know the anoxic fraction; effluent
ammonia and nitrate are monitored and the DO SP adjusted from time to time as necessary to minimise
effluent TN or sludge SSVI (Chapter 4).
To achieve satisfactory DO concentrations, very long diffused air ODs can use non-uniform diffused
air zones. A short high intensity section to raise the DO is followed by a uniform low intensity section
to maintain the DO to the end of the zone. Generally, two DO probes are provided for separate
aeration control of the two sections. Varying the anoxic fraction for tuning purposes may then
involve turning air headers on or off at the downstream end of the aeration zone and shifting the DO
control location.
The appropriate locations for the ditch inflow and outflow are shown in Figure 6.11 – inflow near the start
of the main anoxic zone where COD is needed, outflow to the clarifiers near the downstream end of the
aerobic zone where COD and ammonia are lowest and the sludge has had time for flocculation
downstream of the higher energy aeration zone. This applies to both diffused air and mechanically
aerated ditches.
6.4.3 Hydraulics
The key feature of an oxidation ditch is the large flow of mixed liquor around an open channel circuit. The
low velocity provides good flocculation, resulting in very low effluent SS (see Section 6.1.2 and
Figure 4.11). It also provides effective conditions for scum separation and removal by mechanical
harvesting.
To prevent solids settlement and allow effective control, a uniform velocity distribution has to be
achieved in the channel cross-section. Channel bends contribute to vertical distribution by inducing
spiral flow (Chow, 1959). However at sharp bends without aerators turning vanes are generally used to
control horizontal distribution (as shown in Figure 6.11). Diffused air and vertical shaft ditches have
good vertical mixing and no other hydraulic controls are needed.
With horizontal shaft aerators, however, baffles are needed to achieve good vertical flow distribution in
the channel and maximise aerator performance (Hartley, 1987). In general, an inclined baffle is used
downstream of each aerator to distribute the surface discharge downwards while a vertical baffle is
provided upstream where necessary to reduce the surface inflow velocity to the aerator, maximising its
oxygen transfer rate.

by guest
Annular ditches have spiral flows which affect their velocity distributions and dissolved oxygen
profiles. Figure 6.12 shows the DO contours measured in an annular oxidation ditch-MBR plant with
diffused air; on the basis of this survey, location H was chosen for the DO probe used for aeration
control. Velocity tests conducted in the same plant showed that air flow rate had no effect on channel
velocity – see Figure 6.13.
0.91/0.75 1.29/1.06
1.45/1.0
F
E
0.69/0.50 1.07/0.80
0.5
G Y 1.0
0.49/0.44
1.0 1.63/1.30
H 0.2 0.63/0.12 X 1.09/0.40
0.5
D
0.2
0.20/0.15 W
K V
0.29/0.20
I
0.10/0.08 U
L 0.07/0.06
0.08/0.06
1.0/0.14 1.5
M T
2.2/2.0
N
0.16/0.07 J O
0.13/--
--/0.13
1.0
S C 0.97/1.01
P 0.10/--
0.68/0.85
0.50/0.50 R
0.5
B 0.52/0.18
0.22/0.10 Q
0.2
A 0.26/0.15
TION
ULA
CIRC
Figure 6.12 Dissolved oxygen profile in an annular oxidation ditch-MBR at the Cleveland Bay plant
(Townsville, Queensland; see cover photograph). The membrane trains form the rectangular centre
section; anaerobic cells surround the membrane compartment and feed into the oxidation ditch anoxic
zone. The figure shows DO and velocity test locations, DO readings at surface/at 1–1.5 m depth, and DO
contours at the surface (solid lines) and at 1–1.5 m depth (dashed lines).

by guest
Reactors 109
0.4
OD Circulating Velocity (m/s)

0.3
0.2
7000 m3/h
3700 m3/h
0.1 2000 m3/h
y = 0.00020x + 0.06111
R2 = 0.97426
0
0 500 1000 1500 2000
Circulator Motor Speed (rpm)
Figure 6.13 Velocity test results for the diffused air OD in Figure 6.12 measured at three different air flows.
Liquid flow was circulated with large diameter banana blade mixers. Velocity was independent of aeration
air flow.
6.4.4 Mechanical aerator characteristics

Horizontal shaft aerators
For a horizontal shaft oxidation ditch, the oxygen transfer rate is approximately proportional to the aeration
power over the design operating range. The aeration power is related to the aerator operating parameters
approximately as follows (Hartley, 1987):
P / (N.I) 0.9 2.4

n (6.17)
where
P = total aerator operating power

N = number of aerators operating
I = aerator immersion
n = aerator speed
The ditch circulating velocity, v, varies with power and aerator speed approximately as follows:
v / P0.5 n−0.67 (6.18)
Vertical shaft aerators

Vertical shaft aerators are located at 180 degree bends. The aerator acts as a radial flow pump with a
hydraulic jump at the impeller discharge which entrains air bubbles for aeration. The flow around the
channel is the combined primary pumping capacity of the operating impellers.
At very low speed (and normal immersion settings) the impeller is drowned and the pumping rate is
proportional to the impeller speed. As the speed increases the water level at the impeller discharge is

by guest
depressed by the velocity head and the operating immersion decreases. When the water level reaches the top
of the impeller blades (full immersion), the pumping rate reaches its maximum. With further increase in
speed the impeller becomes partly immersed, however the pumping rate (and velocity around the ditch)
remain approximately constant because the decrease in operating immersion is counterbalanced by the
increase in speed.
Oxygen transfer rate is approximately proportional to the aeration power over the design operating range.

The oxidation ditch is a simple reactor with significant operating flexibility. Its particular operating
advantages are as follows:
• Effective nitrogen control: Effluent nitrogen concentration can be continuously optimised by control
of the operating anoxic fraction. Any desired nitrogen concentration can be maintained – very low
values for discharge to the environment or higher values for recycle to agriculture.
• Control of sludge settleability: Appropriate control of the anoxic fraction allows the SSVI to be
maintained at a low value to maximise plant capacity.
• Low effluent SS: Effective flocculation in the circulating channel produces very low effluent
SS concentration.
• Simple scum control: The circulating channel geometry facilitates effective scum harvesting.
The basic operating principles are as follows:
(1) Set the daily sludge wasting to maintain the desired SRT.
(2) Set the oxidation ditch circulators to maintain an average velocity of 0.2–0.25 m/s; this maximises
DO concentration and flocculation effectiveness while preventing sludge settlement.
(3) Monitor effluent ammonia and nitrate continuously or daily.
(4) Choose a DO set point to maintain the desired ammonia:nitrate ratio or total N concentration in the
effluent and an acceptable SSVI. Adjust the DO set point periodically as necessary to compensate
for variations in loading and operating conditions – feed TKN:COD ratio and mixed liquor
temperature, for example. Aeration control could be cascaded from the channel ammonia:nitrate
ratio using continuous metering.
(5) Set appropriate chemical dose rates for supplementary phosphorus removal and pH control.
6.5 MEMBRANE BIOREACTORS

6.5.1 Description
The use of membranes in a BNR process increases its process and hardware complexity. The particular
benefits provided by the (UF) membranes are:
• Very low effluent SS: SS concentration below 1 mg/L and turbidity below 0.1 NTU.
• Very low particulate nutrients: Particulate N and P concentrations are respectively about 10% and 3–
5% of effluent VSS concentration (Chapter 3).
• Pathogen removal: Typical log removals of 6–7 for faecal coliforms (producing effluent ,1
cfu/100 mL) and 4 for viruses; complete removal of protozoan cysts and helminths. Removals
vary with the degree of biomass caking on the membranes.
In the MBR-BNR flowsheet the secondary clarifier is replaced by membrane solids separation. Pressure
membranes can be used but more commonly the membranes are submerged in the mixed liquor and the

by guest
Reactors 111
permeate is extracted by pump. Figure 6.14 shows the flowsheet for a typical five-stage MBR-BNR process.
In small plants, flat plate membranes are sometimes incorporated in the main aerobic zone. However, in the
larger plants hollow fibre membranes are arranged in a separate membrane zone containing parallel trains
which can be taken off-line individually for automated chemical cleaning. This zone can be highly aerobic
because the membranes are continually cleaned by aeration.

zone Anoxic aerobic Anoxic zone
aerobic)
Mixed liquor
recycle, a
Influent Permeate
WAS
Anaerobic Membrane
recycle, r recycle, s
Figure 6.14 Typical MBR-BNR flowsheet, from Figure 5.1j.
Membranes can substitute for clarifiers in any of the BNR flowsheets but adaptation of the UCT process
is used here as an example. In the UCT process the RAS recycle is directed to the anoxic zone to protect the
anaerobic zone from recycled nitrate (Chapter 5). In the MBR version shown in Figure 6.14 the RAS
(membrane recycle) discharge is shifted downstream to the aerobic zone to protect the anoxic zone from
the high level of recycled oxygen (both high DO concentration and high recycle rate).

Bioreactor
Incorporation of a membrane zone in the flowsheet as shown in Figure 6.14 produces five
notable characteristics. Firstly, because of the recycle arrangement the MLSS concentration steps up
through the zones with the highest concentration in the membrane zone. The MLSS relationships
calculated by mass balance are summarised in Table 6.5. The membranes limit the maximum
operating MLSS concentration, typically to a value of about 12 g/L. From a design perspective, the
recycle ratios are economic variables – the higher the recycle ratios the less the difference in MLSS
between the zones, therefore the smaller the zone volumes required for a given process sludge mass.
From an operating perspective, varying the recycle ratios can change both the total operating sludge
mass (and associated SRT) and the mass fractions of the various zones – see Figure 6.15. The operating
recycle ratios can be calculated from the measured MLSS concentrations using the equations given in
Table 6.5.
Secondly, the membrane zone contains a significant fraction of the total sludge mass, typically about
20%. This reduces the mass fractions available to the other zones.

by guest
Table 6.5 MBR-BNR process balance.

zone anoxic aerobic anoxic zone
aerobic)
Mixed liquor
recycle, aQ
fVO, XO
fVA, XA fVX1, XX1 fVX2, XO fVM, XM
Influent, Q Permeate
WAS
Anaerobic Membrane
recycle, rQ recycle, sQ
MLSS concentrations
XO = XM [s/(1 + s)] (6.19a)
XX1 = XO [a/(1 + a)] (6.19b)
XA = XX1 [r/(1 + r)] (6.19c)
Recycle ratios
s = 1/[(XM /XO ) − 1] (6.19d)
a = 1/[(XO /XX1 ) − 1] (6.19e)
r = 1/[(XX1 /XA ) − 1] (6.19f)

FX = f VA XA + f VX1 XX1 + (f VO + f VX2 )XO + f VM XM (6.19g)
f XM = f VM XM /FX (6.19h)
f XX2 = f VX2 XO /FX (6.19i)
f XO = f VO XO /FX (6.19j)
f XX1 = f VX1 XX1 /FX (6.19k)
f XA = f VA XA /FX (6.19l)
f XM + f XX2 + f XO + f XX1 + f XA = 1 (6.19m)
where
a = mixed liquor recycle ratio fXM = membrane zone mass fraction

r = anaerobic recycle ratio fXO = primary aerobic zone mass fraction
s = membrane recycle ratio fXX1 = primary anoxic zone mass fraction
fVA = anaerobic zone volume fraction fXX2 = secondary anoxic zone mass fraction
fVM = membrane zone volume fraction FX = zone mass function
fVO = primary aerobic zone volume fraction XA = MLSS concentration in anaerobic zone
fVX1 = primary anoxic zone volume fraction XM = MLSS concentration in membrane zone
fVX2 = secondary anoxic zone volume XO = MLSS concentration in primary aerobic &
fraction secondary anoxic zones
fXA = anaerobic zone mass fraction XX1 = MLSS concentration in primary anoxic zone

by guest
Reactors 113
25%
Membrane Zone
Change in Maximum Sludge Mass
& Process Mass Fractions
Aerobic/Sec Anox
0%
Prim Anox
-25%
Anaerobic
Sludge Mass
Mass Fractions
-50%
-50% -25% 0% 25% 50%
Change in Recycle Ratios from Design Values
Figure 6.15 Effects of changes in the operating recycle ratios on maximum allowable sludge mass and
reactor mass fractions. MLSS concentration in the membrane zone is constant; design r, a and s recycle
ratios are 1, 6 and 4 and are varied in concert; process volume fractions/design mass fractions are –
anaerobic 0.15/0.07, anoxic 0.30/0.28, aerobic 0.36/0.40, secondary anoxic 0.05/0.05, membrane
0.14/0.20.
Thirdly, nitrogen removal is mediated only by the anoxic and primary aerobic zones. The membrane zone
further nitrifies residual ammonia from the upstream zones, meaning the process produces low effluent
ammonia. However, tuning of nitrogen removal performance requires effluent data from the upstream
zones rather than the final discharge from the membrane zone.
Fourthly, MBR processes are reported to have somewhat lower nitrification rates than conventional
processes (Monti & Hall, 2008), and to be more prone to foaming (Monti et al., 2006). Prevention of
effluent solids discharge by replacement of clarifiers with membranes leads to shifts in the composition
of the biomass (Hall et al., 2010).
Fifthly, because MBRs operate at average MLSS concentrations 2–3 times higher than conventional
processes, the reactor hydraulic residence time is 2–3 times shorter (for the same SRT). This means the
effluent quality is more responsive to load variability; effluent concentrations are more variable which
increases average values (see Chapter 2 and Appendix 1).
Membranes
This section discusses hollow fibre ultrafiltration membranes with water flow from outside to inside. These
membranes are generally arranged in parallel trains with a number of modules in each train. Trains can be
isolated for chemical cleaning. Feed to the biological process is fine screened (typically 1mm apertures) to
protect the modules from debris accumulation.

by guest
This type of membrane has a nominal pore size in the ultrafiltration range of 0.01–0.1 µm and is capable
of producing an effluent turbidity of less than 0.1 NTU. Faecal coliform counts can be reduced to zero.
MLSS concentration in the membrane zone is limited to about 12,000 mg/L to enable the required flux
rates to be maintained.
General process principles are as follows (Trussell et al., 2006; Trussell et al., 2007):
• As permeate flows into the membranes, biomass cakes on the membrane surface, increasing the
transmembrane pressure (TMP). In addition, microbial growth occurs on the membranes and
soluble microbial products (SMP – colloidal organic material) accumulate in the pores. High
intensity aeration is applied intermittently to scour the membrane surfaces and limit the degree of
caking. Aeration frequency varies with permeate flow and cleaning may be assisted by resting and
backflow phases. Permeability gradually decreases and chemical cleans (using hypochlorite for
organics and citric acid for any inorganic precipitants) are used to restore permeability at intervals
governed by the system’s TMP limit. Nevertheless, the permeability will suffer a long-term
decline and membrane life expectancy (based on required hydraulic capacity) may be about 10 years.
• As MLSS concentration increases, the associated increase in ML viscosity requires increasing
aeration intensity to maintain permeability. At constant aeration intensity, an increase in MLSS
concentration reduces membrane permeability and increases the ongoing fouling rate. The MLSS
limit is set to suit the design aeration intensity and the expected duration of flow peaks.
• Operating permeability also varies with SRT because of the variation in mixed liquor SMP level. As
SRT increases from 2 to 10 days the membrane fouling rate decreases 20-fold. Fouling rate continues
to decline up to an SRT of around 20 days.
• A study on the effects of carbon dosing to enhance nutrient removal (Ahmed et al., 2008) found
that synthetic wastewaters of different compositions caused increasing deterioration in membrane
permeability in the following dominant substrate order: glucose (least) , acetate , propionate
, methanol (worst). This contrasts with a reported improvement in permeability characteristics
produced by alum dosing which coagulates mixed liquor SMP (Holbrook et al., 2004); the author
has also observed this alum permeability enhancement at the Cleveland Bay (Townsville) plant.
• Membrane permeability can be calculated and monitored using the following formula:
LP20 = J.e−0.0239(T−20) /TMP (6.20)
where
LP20 = membrane permeability standardised at 20°C, L/m2h · kPa

J = membrane operating flux, L/m2h
T = membrane operating temperature, °C
TMP = transmembrane pressure, kPa
Permeability values are typically about 1.5 L/m2h · kPa at 20°C.

• MBRs have a well defined hydraulic capacity which declines as the duration of the peak flow
increases. Controls may be provided to automatically reduce throughput after a preset peak flow
duration. Because membrane systems are expensive there is generally little reserve capacity and
excess flow has to be bypassed or stored. This contrasts with clarifier systems where capacity
varies with MLSS concentration, SVI, RAS flow rate and internal sludge storage capacity and
failure is more gradual and more easily managed.

by guest
Reactors 115

Refer to Sections 6.3 or 6.4 for operating principles related to the main process format used in the MBR.
Principles specific to the MBR variant are as follows:
(1) To maximise SRT, operate at maximum MLSS concentration in the membrane zone and with
maximum recycle ratios.
(2) Adjust recycle ratios to change the mass fractions in the various zones.
(3) To tune nitrogen removal performance, use data from the membrane zone feed rather than effluent.
The membrane zone further nitrifies ammonia to nitrate but does not remove nitrogen.
(4) Manage membrane cleaning to maximise membrane availability during the peak flow season.
(5) Because of the reliance on significant quantities of mechanical, electrical and control equipment, a
sound maintenance management plan is essential.
6.6 SEQUENCING BATCH REACTORS

6.6.1 Description
A sequencing batch reactor (SBR) or intermittent process in its most basic form is a simple box reactor
operated in a sequential fashion. In a repetitive cycle, the reactor is filled with sewage, the batch of
sewage is reacted, the treated effluent is discharged. As described at the start of Chapter 3, Ardern and
Lockett used this batch treatment mode in their original development of the activated sludge process.
Physically it is the simplest of processes, however process-wise it is one of the most complex.
Development of the full-scale SBR was begun in earnest by Pasveer with his CFID (continuous flow
intermittent decant) oxidation ditch as mentioned in the introduction to Section 6.4 above. A variety of
process formats is now in use with a range of cycle designs, control methods and hardware
(Figure 6.16). The process is controlled by combinations of time, water level and on-line analytical
instruments. Some forms of the process incorporate mixed liquor recycles to inlet compartments
designed as selectors for SVI control or to promote biological N and P removal. The main advantage of
the SBR is low capital cost; its disadvantages are difficulty in reliably achieving high levels of biological
N and P removal and process complexity which makes operational tuning difficult.
(a)
Continuous Intermittent
Feed Decant
SBR
(b)
Parallel Modules
Feed Balance Balance Effluent

SBR
Tank Tank
Recycle
Figure 6.16 SBR process formats: (a) the simplest, a single tank CFID process, and (b) a generic format.

by guest
The generic name SBR was coined by Irvine in 1967 (Wilderer et al., 2001). Irvine provided a
classic summary of this process as summarised in Wilderer et al. (ibid, p70): “the SBR’s greatest
strength was that an infinite number of positive operating strategies could be employed to meet effluent
limits and …… its greatest weakness was that an infinite number of negative operating strategies were
also possible.”
Because of its commercial competitiveness and black box nature, a variety of proprietary formats has
developed. Process variants are described by a range of names – IDEA (intermittently decanted extended
aeration), IDAL (intermittently decanted aerated lagoon), ICEAS (intermittent cycle extended aeration
system), CASS (cyclic activated sludge system).

Figure 6.17 shows an example process format used here to explain SBR process characteristics. Figure 6.18
shows the tank geometry and Figure 6.19 the process operating cycle for each SBR.
WAS Effluent
Influent PS Alum
Mixer
SBR 1 Decanter
DO
Blowers
Mixer
DO
SBR 2 Decanter
Alum
Figure 6.17 Example SBR flowsheet.
Two parallel SBRs are fitted with mechanical mixers and diffused air aeration; blowers are controlled
from DO probes. Effluent is discharged using decanting troughs. There is no upstream balancing tank.
EBPR is not included and alum dosing is used for P removal. The two SBRs are operated 180 degrees
out of phase, each being fed separately for 50% of the time. Normal control is by time but the rate of
increase in water level is monitored and high flow events trigger a switch to a shorter high flow cycle
which increases the decant time from 13% to 27% of the total cycle time. If necessary, to prevent loss
of biomass under extreme flow conditions treatment can be interrupted and continuous decanting
implemented.

by guest
Reactors 117
12m decanter
DO probes
Tank 17 sq x 5.0 max WD

Volume 1445 kL
5.5
5.0
TWL
3.5
LWL
00
Figure 6.18 SBR geometry.
NORMAL T0 T1 T2 T3 T4 T5 T6 T7
CYCLE Feed
Mix
0.4 mg/L 2 mg/L
Aerate (with DO SPs)
Settle
Decant
Decanter Ascend
Waste Sludge
0 1 2 3 4
Hours
HIGH FLOW
CYCLE Feed
Mix
Aerate
Settle
Decant
Decanter Ascend
Waste Sludge
0 1 2 3 4
Hours
Figure 6.19 Normal and high flow cycles.

by guest
The normal cycle time is 4.36 hours, giving 5.5 cycles per day. The half cycle alternates the times the two
SBRs are fed each day so that the daily sewage loads are shared equally. The treatment phases are as follows:
Mix: An anoxic phase in which feed is used for denitrification.
Aerate: Split into two phases – aerobic feed, then aerobic batch after feed ceases. As shown, different
dissolved oxygen set points are used for these two phases.
Settle: Aeration ceases and sludge settling occurs.
Decant: About halfway through the settling period the decanter descends to draw off effluent. At the end
of decant the decanter starts its rise a few minutes before the next cycle starts so that the trough is above
the water level when feeding and aeration start.
Waste: To control SRT, sludge is wasted during each aeration phase after feed ceases so that the reactor is
well mixed and the volume to be extracted can be calculated accurately from the water level.
Figure 6.20 shows the process decanting behaviour. Plot (a) shows the water and sludge interface levels
while (b) shows the weir approach velocity, which is maintained below a typical design limit of 40 m/h
to prevent scour of the sludge blanket and carryover of solids in the effluent. Weir approach velocity is
calculated from the cross-sectional area defined by the width of the weir and the depth to the top of the
settling sludge layer. Ideally, decanting should start at the appropriate time to maintain the clear water
depth constant over the decanting time, producing a constant weir approach velocity and minimising the
risk of solids carryover; however, sludge settling rate varies from time to time as MLSS and SVI vary.
The decant needs to finish before sludge compaction starts to reduce the clear water depth as depicted in
the post-decant extension of Plot (a).
(a) 6
Settle Phase Decant Phase
Cycle End
5
Water Level
Depth (m)
4
LWL
3
Sludge Interface
2
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Hours from Settle Start
(b) 50
Cycle End
Settle Phase Decant Phase

Weir Approach Velocity (m/h)
40 Design Limit
30
20
10
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Hours from Settle Start
Figure 6.20 Process behaviour during the settle and decant phases for the high flow cycle.

by guest
Reactors 119
Sludge settleability in the SBR process appears to be governed predominantly by the primary anoxic
fraction – see Chapter 4.
Figure 6.21 shows typical cycle profiles for ammonia, oxidised nitrogen and biodegradable influent
COD. Decanted effluent quality is represented by the concentrations at the start of settle. Processes
occurring during the various phases are as follows:
Anoxic Feed Phase: COD increases only slowly as part of the feed mass is consumed by denitrification
of the residual nitrate from the previous cycle. Feed ammonia and ammonification of feed organic
N cause the ammonia concentration to increase. Ideally this phase ends when nitrate is exhausted as
shown in the figure.
Aerobic Feed Phase: Operates at a low DO (say 0.4 mg/L) for simultaneous nitrification and
denitrification (SND, see Chapter 3, Section 3.5.4). Ammonia is fed and nitrified.
Aerobic Batch Phase: Feeding ceases. Residual biodegradable COD is oxidised and ammonia
nitrified.
Settle and Decant Phase: The sludge settles and respires endogenously using residual DO and nitrate.
There is little change to ammonia and COD concentrations.
8 40
Biodegradable COD (mg/L)

Anoxic Feed Aerobic Feed Aerobic Batch Settle & Decant
Quality
6 30
Nitrogen (mg/L)
NOx-N
4 20
NH3-N
Effluent
2 10
CODb
0 0
0 1 2 3 4
Hours
Figure 6.21 Typical cycle profiles for ammonia, oxidised N and biodegradable COD. Effluent nitrate varies
with influent TKN:biodegradable COD ratio (0.15 in this example).
The SBR process is subject to the same limitations on minimum achievable nitrate as the continuous flow
process. In the continuous process the mixed liquor recycle ratio sets the proportion of nitrified effluent
recycled for denitrification and limits the minimum nitrate concentration (Chapter 3, Section 3.5.3). The
SBR process behaves the same way. After nitrification in the aerobic phases a proportion of the reactor
contents is decanted as effluent. The retained proportion is then mixed with influent for denitrification
during the following anoxic phase.
SBR volumetric exchange can be related to the continuous flow recycle ratio as follows:
a = recycled flow/discharged flow

= retained volume/decanted volume
= (total volume − decant volume)/decant volume
= (1/a) − 1 (6.21a)

by guest
or
a = 1/(a + 1) (6.21b)
where
a = continuous flow recycle ratio

α = SBR volumetric exchange ratio (Wilderer et al., 2001)
= decant volume/total volume
The smaller the fraction of reactor volume decanted and refilled during each cycle, the lower the minimum
nitrate achievable. Achieving low nitrate in the SBR process therefore requires a large reactor volume; this
has a relatively greater cost impact than the larger recycle required in the continuous flow process. In the
SBR process, therefore, greater reliance tends to be placed on supplementary SND (simultaneous
nitrification and denitrification) during the aerated feed phase to increase the extent of denitrification.

The SBR is physically simple but operationally complex. The basic operating principles, based on the
example process format described above, are as follows:
(1) Set the basic cycle as described above.
(2) Set the sludge wasting to maintain the desired SRT.
(3) Monitor effluent ammonia and nitrate
(4) To vary effluent ammonia concentration, adjust the cycle fractions and DO set points for the aerobic
feed and aerobic batch phases, and the process SRT.
(5) To vary effluent nitrate concentration, adjust the anoxic feed fraction and the DO concentration in
the aerobic feed phase. The cycle time can also be shortened to reduce the volumetric exchange ratio
if this is a limiting factor.
(6) To minimise effluent SS concentration, increase the duration of the settle and decant phase and
adjust the start of the decant phase so that the water level falls at the same rate as the sludge
interface (assuming that the decant rate can be varied). The SRT can also be shortened to reduce
MLSS concentration and increase settling rate if this is a limiting factor.
(7) Set appropriate chemical dose rates for supplementary phosphorus removal and pH control.

by guest
Chapter 7
Secondary clarifiers
This chapter explains the behavioural characteristics of secondary clarifiers and provides some useful
operating diagrams. It focuses on the popular circular centre-feed clarifier however the principles are
similar for other forms.
7.1 DESCRIPTION
The primary function of the secondary clarifier is to clarify – to produce a clear effluent by separating the
solids from the mixed liquor for return to the bioreactor. Thickening of the return sludge is a secondary
function and should not be allowed to compromise the primary clarification function (avoid the tail
wagging the dog).
Secondary clarifiers may be circular or rectangular; circular clarifiers can have centre or peripheral feed.
The circular centre feed clarifier is the most common because of its simplicity and good performance.
Diameters above 45 m are uncommon, even in large plants, because of equipment and operational
limitations. Side water depths are typically at least 10% of the diameter and vary from 3–6 m. Greater
depths tend to produce lower effluent SS concentrations because of greater resistance to hydraulic and
wind disturbances (Ekama et al., 1997). They also provide more sludge storage capacity for handling wet
weather flow events.
Feed enters radially near the surface within a circular feed well. Flocculator feed wells are sometimes
used to reduce effluent SS concentration. This is unnecessary if the feed is already well flocculated, for
example, mixed liquor direct from an oxidation ditch (see Figure 4.11, Chapter 4).
Effluent is collected by peripheral or inboard launders. Effluent SS concentration is lower if the launders
are arranged to minimise the effect of the typical upwelling current at the side wall; peripheral launders
may be hung on the inside of the wall or a peripheral Stamford baffle may be used to deflect the current.
Sludge may be scraped to a centre well for drawoff or collected by suction (hydraulic) collectors hung
from a rotating bridge. Scrapers require a conical floor, typically at a 1:12 slope, and solids flow to the
centre well, partly as a hydraulic density current and partly under the impetus of the scraper. Log spiral
scrapers (making a constant angle to any radius) provide the best performance; echelon scrapers having a
number of parallel blades provide poorer performance because several revolutions of the scraper are
required to mechanically move a particle to the centre well.
Suction sludge collectors operate on a flat floor and remove sludge from the whole radius or diameter at
once. The underflow is drawn from a moving location hence the hydraulic behaviour of the clarifier is more

by guest
complex. One drawback of the suction collector is its fixed maximum hydraulic capacity – it is not amenable
to increase in the RAS flow should that ever be desired.
Both scrapers and suction collectors are usually also fitted with scum removal systems.
7.2 MASS BALANCES

Secondary clarifiers are mass balance driven. Under steady state conditions, what goes in must come out.
Diurnal and wet weather variations in flow rate produce increases and decreases in the clarifier solids
inventory. Table 7.1 summarises the steady state mass balance relationships. Although the flow patterns
in a clarifier are more complex than shown, the idealised approach depicted provides a good
representation of clarifier behaviour and capacity.
Table 7.1 Idealised clarifier behaviour.
Feed Q (1+s)Q, XM Effluent Q, XE ~ 0

VR, XM
Solids profile
AC Water Level XE
UO
Overflow
Zone
Feed Level
Underflow Zone
VCU, XU
VS UU
0 XU XS
RAS sQ, XS
Nomenclature:
AC Clarifier surface area
MXT Process total sludge mass
n Settling constant (see Table 4.1)
Q Plant flow
s RAS recycle ratio
sNL Non-limiting RAS recycle ratio
UO Overflow rate, Q/AC
UU Underflow rate, sQ/AC = sUO
VCU Clarifier underflow zone volume
(Continued )

by guest
Secondary clarifiers 123
Table 7.1 Idealised clarifier behaviour (Continued).
VO Settling constant (see Table 4.1)

VR Reactor volume
VS Solids settling rate in the underflow zone
XE Effluent SS concentration
XM MLSS concentration
XMC MLSS concentration at maximum capacity with equal SS concentrations in reactor and clarifier
underflow zone
XS RAS SS concentration
XU Underflow SS concentration
Relationships:
MXT = VRXM + VCUXU (7.1)
VS = VO · exp(−nXU) (4.2)
XMC = MXT/(VR + VCU) (7.2)
XS = XM(1 + s)/s (7.3)
XU = XM(1 + s)/[(VS/UO) + s] (7.4)
sNL = exp(nXM − 2) (7.5)
There are three process zones: the reactor, the clarifier overflow zone and the clarifier underflow zone.
Under normal operating conditions the solids mass in the clarifier overflow zone is negligible. The process
biomass is distributed between the reactor and the clarifier underflow zone but under dry weather flow
conditions the solids mass in the underflow zone is also generally negligible. As plant flow increases,
solids transfer from the reactor to the underflow zone.
In the underflow zone solids are transported downwards by two mechanisms – the bulk downward RAS
flow and settlement through the water. The solids concentration, XU, is constant for most of the depth
(Equation 7.4) and then increases to the RAS concentration, XS (Equation 7.3), near the floor where the
settlement mechanism ceases.
As the plant feed flow increases, the underflow concentration XU also increases because the overflow
rate, UO, is increasing (Equation 7.4). However, the RAS concentration remains constant if the recycle
ratio, s, is held constant. Because XU increases the solids mass in the underflow zone increases, reducing
the reactor MLSS concentration, XM.
The maximum capacity is reached when the increasing overflow rate, UO, reaches the settling rate of the
incoming mixed liquor. The underflow concentration is then at the MLSS concentration (Equation 7.2). To
ensure the RAS flow does not impose a lower capacity limit due to thickening limitations, the recycle ratio at
the limiting overflow rate should be not less than the value given by Equation 7.5 (see Section 7.3 below).
Thickening failure can be overcome by increasing the recycle ratio, however recovery from clarification
failure requires reduction in overflow rate, reduction in MLSS concentration or decrease in SSVI.
Clarifier behaviour is well illustrated by the clarifier sludge profile shown in Figure 7.1. The data were
measured at the Lower Molonglo Water Quality Control Centre in Canberra during a 1997 clarifier study.
This plant incorporated chemically enhanced primary treatment and, because of the low primary effluent
strength, had a very high clarifier:reactor volume ratio of 2:1. At the time the solids profile was measured
the sludge blanket was stable and extended well above the feed level, with a concentration equal to the
MLSS value. The large clarifier volume allowed for large transfers of solids between the reactors and

by guest
clarifiers. When plant flow increased the blanket rose but the associated transfer of solids from the reactors
reduced the MLSS concentration until a new equilibrium blanket level was attained. In effect the plant was
operating with a variable volume reactor – the main reactor plus the clarifier sludge volume.
Water level
Height Above Floor (m) 6
3
Bottom of
2 feedwell
0
0 2 4 6 8
SS Concentration (g/L)
Figure 7.1 Clarifier SS profile measured at the Lower Molonglo WQCC. Clarifier dimensions 36 m dia ×
6.26 m side water depth, flat floor; feed well extends 4 m below WL. MLSS concentration 3.1 g/L, SSVI
74 mL/g, overflow rate 0.86 m/h, RAS recycle ratio 0.75, measured RAS concentration 7.3 g/L.
Significant endogenous denitrification occurred within the sludge blanket. Nitrogen micro-bubbles
trapped within the blanket may have contributed to the low sludge settling rate and the apparent decrease
in solids concentration in the upper levels.
Note particularly that, in accordance with the theory, (a) in stable operation, the blanket concentration
equalled the MLSS concentration, (b) the recycle concentration was (1 + s)/s times the MLSS
concentration, and (c) with a safe recycle ratio (Section 7.3), the depth of the sludge thickening zone
was minimal.
7.3 OPERATING DIAGRAMS

The operating condition of a clarifier can be evaluated using the operating diagrams shown in Figure 7.2
(UCT et al., 1984; Ekama et al., 1997). The diagrams show whether clarifier failure can be expected at
any particular operating condition but give no information on the quantity of solids in the tank.
The diagrams have been drawn for an NDEBPR process and separate diagrams are provided for SSVI
values of 50, 70, 90 and 120 mL/g. Interpolation can be used for other values. Table 7.2 describes
diagram construction to facilitate production of further diagrams to suit specific plant circumstances.
The diagrams can be explained by reference to Table 7.2. Each diagram has three sets of lines. Lines A are
clarification failure boundaries. These are drawn at the sludge settling rate applicable to the particular SSVI
and MLSS values. Regardless of the recycle ratio, if the overflow rate at the operating MLSS concentration
lies above the appropriate failure boundary, the clarifier will fail in clarification with gross loss of solids. The
safe and failure operating regions are noted on the diagram.

by guest
(a) NDEBPR Process

SSVI 50 mL/g
4
2 g/L MLSS 3 g/L
3
Overflow Rate (m/h)
4 g/L
2 5 g/L
6 g/L
1
Underflow Rate
1.35 m/h
0
0 1 2
Recycle Ratio
NDEBPR Process
SSVI 70 mL/g
4
Underflow Rate MLSS 2 g/L

1.15 m/h
3
Overflow Rate (m/h)
3 g/L
2 4 g/L
5 g/L
1 6 g/L
0
0 1 2
Recycle Ratio
NDEBPR Process
SSVI 90 mL/g
4
Underflow Rate
0.97 m/h MLSS 2 g/L
Overflow Rate (m/h)
2 3 g/L
4 g/L
1 5 g/L
6 g/L
0
0 1 2
Recycle Ratio
Figure 7.2a Clarifier operating diagrams for SSVIs of 50, 70 and 90 mL/g.

by guest
(b) NDEBPR Process

SSVI 120 mL/g
4
Underflow Rate
0.76 m/h
Overflow Rate (m/h)

3
MLSS 2 g/L
2
3 g/L
1 4 g/L
5 g/L
0
0 1 2
Recycle Ratio
Figure 7.2b Clarifier operating diagram for SSVI of 120 mL/g.
Table 7.2 Clarifier operating diagram.
C: Boundary of region in which only the

clarification criterion needs to be met
Overflow Rate, Uo (m/h)
3 A: Clarification failure boundary

Fail Constant MLSS
Safe
2 B: Thickening failure boundary
Constant MLSS
Safe
Fail
0
0 1 2
Recycle Ratio, s
Nomenclature:
AC Clarifier surface area
n Settling constant (see Table 4.1)
Q Plant flow
s RAS recycle ratio
UO Overflow rate, Q/AC
UU Underflow rate, sQ/AC = sUO
VO Settling constant, m/h (see Table 4.1)
XM MLSS (clarifier feed) concentration
α Function in equation for thickening failure boundary
Relationships:
A. Clarification failure boundary
UO = VO.exp(−nXM) (7.6)
(Continued )

by guest
Table 7.2 Clarifier operating diagram (Continued).
B. Thickening failure boundary

UO = (VO/s)[(1 + α)/(1 − α)]exp[−n(1 + s)XM(1 + α)/(2s)] (7.7)
Where α = {1 − 4s/[n(1 + s)XM]} 0.5
(7.8)
C. Boundary of region in which only the clarification criterion has to be met
UO = VO/[s · exp(2)] (7.9)
or
s = exp(nXM − 2) (7.10)
Lines B are thickening failure boundaries. These show the minimum RAS recycle ratio required to avoid
thickening failure at particular values of overflow rate, SSVI and MLSS concentration. The safe and failure
operating regions are marked.
Line C divides the operating region in which both clarification and thickening failure can occur from the
region in which only clarification failure can occur. If the operating overflow rate is above the clarification
boundary for the operating MLSS, clarification failure will occur regardless of the recycle ratio. If the
operating recycle ratio is to the left of the appropriate failure boundary, thickening failure will occur. The
safe operating region lies below and to the right of the appropriate failure boundaries; failure will occur
by clarification and/or thickening if the operating point lies anywhere above or to the left of the
failure boundary.
Clarifiers have two safety mechanisms providing robustness against failure. If the operating point moves
outside of the failure boundaries, gross solids loss will not occur until sufficient sludge has accumulated in
the clarifier to be entrained in the effluent. Short-term (diurnal) overloads can be tolerated if the failure
condition is reversed before solids loss begins. Further, the clarifier has a self-compensating mechanism –
as solids transfer to the clarifier the feed MLSS concentration declines, expanding the safe operating
region. The MLSS concentrations shown on the diagrams relate to the values at the peak flow condition,
which will be somewhat lower than the preceding values.
7.4 EFFLUENT QUALITY

7.4.1 Factors affecting effluent SS
This section discusses the factors affecting the effluent SS concentration produced by the clarifiers. The
effect of the reactor mixing regime on sludge flocculation is discussed in Chapter 6, Reactors. Other
factors are discussed here.
In essence, under average operating conditions (no gross solids loss), the clarifier effluent SS
concentration is governed by four factors:
(1) The degree of flocculation, which governs the floc size distribution.
(2) The clarifier overflow rate, governing the percentage of flocs whose settling rate is less than the
overflow rate (dispersed solids) and which are therefore carried over in the effluent.
(3) Rising sludge due to denitrification in the clarifier.
(4) Excessive scraper speed which can produce a trailing cloud of SS overflowing the tank. Effluent SS
may increase measurably at peripheral scraper speeds above 2–3 m/minute, depending on tank
depth, blade height, SSVI and sludge accumulation.

by guest
Factors 1–3 are discussed below.
Degree of flocculation
The degree of flocculation is dependent on the following:
• Shear history
• Environmental chemistry
• SRT
• SSVI
• MLSS concentration
• Chemical precipitant dosing
Shear history is discussed in Chapter 6; the other drivers are discussed in turn here.
Environmental chemistry
Flocculation is affected by the ionic strength, inorganic composition and pH of the aqueous environment. In
particular, elevated cation concentrations enhance the binding capacity of the EPS. In laboratory
flocculation studies, Biggs (2000) found that adding 160 mg/L of calcium to the ND sludge increased
the median steady state floc size by 40%. Calcium additions of 80 mg/L or less had no effect on steady
state floc size but reflocculation after shear was faster.
SRT
Flocculation is better at lower growth rates (longer SRTs) because EPS production is greater (Wilen et al.,
2010). Figure 7.3 shows the effect of SRT on the percentage of dispersed solids in the mixed liquor and the
resultant performance of a full-scale plant. At very short SRT the bacterial growth rate is high and
flocculation poor. As SRT increases the bacteria excrete greater quantities of natural polymer and
flocculation becomes more complete.
(a) 40 (b) 100

Percent Dispersion
Effluent SS (mg/L)
30 75
20 50
10 25
0 2 4 6 8 10 12 0 2 4 6 8 10 12
SRT (d) SRT (d)
Figure 7.3 (a) Percentage of MLSS which are dispersed, measured on laboratory activated sludge units.
• Chao and Keinath (1979), SVI 100-725; ○ Bisogni & Lawrence (1971), SVI 80-600.
(b) Effect of SRT on effluent SS at the 25,000 EP, mechanically aerated, conventional activated
sludge plant at Redcliffe, Queensland (Hartley, 1985). Monthly average data for Jan-81 to
Mar-82.
• SVI 77-112; ○ SVI 142-430.

by guest
SSVI
The impacts of both settleability (as SVI) and shear history are illustrated in Figure 4.11. For a given plant
type, the higher the SVI the lower the percentage of very small flocs and the lower the effluent SS. There is a
fair scatter due to the effects of other parameters but the general SVI trends are clear.
The three plant types shown represent a gradation in mixed liquor shear history ahead of clarification.
The conventional activated sludge plants had shorter SRTs and higher mixing intensities in both reactors
and sludge transfer systems. The intermittently decanted plant (Jimboomba) had no primary treatment,
long SRT (greater than 20d), lower reactor mixing intensity and no sludge transfer requirement prior
to settling.
The oxidation ditch plant (West Byron) also had long SRT, plus the low mixing intensity characteristic of
oxidation ditches, enhanced in this case by the use of fine pore diffused air rather than mechanical aeration.
Flow along an oxidation ditch channel provides uniform mixing at the lowest practical turbulence level; if
the mixed liquor offtake is located well downstream of the aeration zone, reflocculation can take place
before transfer to the clarifier. Flow at a typical velocity of 0.25 m/s in the 6 m wide × 4 m deep channel
at West Byron had a mean G value of only 3 s−1.
MLSS concentration
Increasing MLSS concentration improves flocculation and reduces effluent SS concentration. Figure 7.4
illustrates this using data from a pilot plant settleability study and from the parallel full-scale plant
(Lower Molonglo, Canberra) which is an ND plant with chemically enhanced primary treatment.
50
Full-Scale
SBR 1
40 SBR 2
SBR 3
Effluent SS (mg/L)
SBR 4
30
SBR 5
SBR 6
SBR 7
20
10
0
0 1 2 3 4 5
MLSS (g/L)
Figure 7.4 Effect of MLSS concentration on the supernatant SS concentration at the end of 30 minute mixed
liquor settling tests (Wastewater Futures, 2007). Data were collected during a pilot sludge settleability study
conducted using seven SBRs operated in parallel with the full-scale Lower Molonglo plant in Canberra. The
SBRs all operated with different cycle regimes and different SSVIs. Representative data from the full-scale
plant are also shown.

by guest
Chemical precipitant dosing

Dosing of alum or iron to the biological reactor causes hydroxide and phosphate precipitates to be
incorporated in the floc structure (see Chapter 3). This weakens the floc, reducing the average floc size
and increasing the percentage of smaller flocs discharged in the effluent.
Figure 7.5 illustrates this using operating trends from the West Byron NDEBPR oxidation ditch plant in
Byron Bay, NSW. Ferric sulfate was dosed at the plant inlet for sulfide precipitation and alum was dosed to
the bioreactor for supplementation of bio-P removal. Ferric sulfide entering the reactor would have been
oxidised to ferric sulfate which formed co-precipitates with the alum.
Alum to OD Alum+Equiv Ferric 60d MA Equiv Dose Eff SS (mg/L)

150 15
Equiv Alum Dose
120 12
(mg solid alum/L)
Effluent SS (mg/L)
90 9
60 6
30 3
0 0
1-Nov 1-Jan 1-Mar 1-May 1-Jul 1-Sep 1-Nov 1-Jan 1-Mar 1-May 1-Jul 1-Sep 1-Nov
Figure 7.5 West Byron STP: two-year trends from plant startup in Nov-05. Correlation between effluent SS
and 3SRT moving average precipitant dose. Median MLSS and unstirred SVI were 3.2 g/L and
80 mL/g respectively.
The figure shows two years of trends for precipitant dose (alum plus equivalent ferric) and effluent SS.
Both daily and three-SRT (60 day) moving average chemical dose rates are shown because it takes about
three SRTs for changes in reactor dose rate to have full effect. By using appropriate scales the chemical
dose rate and SS trends have been aligned. It can be seen that the 3SRT moving average dose rate aligns
well with the effluent SS trend, indicating that precipitant dosing had a significant effect on effluent SS
behaviour which varied in proportion to the dose rate.
Further insight is provided by the behaviour of an SBR treating the discharge from anaerobic lagoons
receiving beef abattoir wastewater. Figure 7.6 shows operating trends for a five month period at the end
of 2003. The SBR was operated using a 6-hour cycle which achieved near-complete nitrogen removal –
anoxic feed 0.5 h, aerobic feed 2.5 h, aeration with no feed 0.5 h, settle and decant 2.5 h. On 5-Aug a
trial period of alum dosing began in order to reduce the phosphorus discharge in the effluent. Table 7.3
summarises the average performance for the two weeks before alum dosing began and for the two
months following a 60 day (three SRT) adjustment period after dosing began.
It can be seen that alum dosing caused the MLSS to increase and the SVI to fall, consistent with the
Maroochydore example discussed in Chapter 4. Effluent SS also increased as seen in the West Byron
example above. Further, ammonia became less stable because of the reduction in mixed liquor pH
associated with the alum dosing. An unexpected result was an increase in effluent nitrate concentration.
Part of the nitrate removal was occurring by simultaneous nitrification and denitrification (SND) during
the aerated feed phase (as discussed in Chapters 3 and 6). The observed reduction in nitrate removal was
consistent with a reduction in average floc size, causing a diminution of the volume percentage occupied
by the anoxic core. The required nitrate removal was subsequently reinstated by reducing the operating
DO concentration during the aerobic feed phase.

by guest
(a)
Alum Dosing Alum Dose Effluent PO4-P Effluent SS
600 60
Effluent Conc (mg/L)

Dose Rate (mg/L)
50
400 40
30
200 20
10
0 0
20-Jul-03 20-Aug-03 20-Sep-03 20-Oct-03 20-Nov-03 20-Dec-03
(b)
MLSS & SVI MLSS SVI
12 200
MLSS (mg/L)
10 150
SVI (mL/g)
8 100
Start Alum
6 50
4 0
(c)
Effluent N NH3-N NOx-N
60
Start Alum
Effluent N (mg/L)
40
20
0
Figure 7.6 Trend plots from an SBR treating abattoir wastewater, showing the behaviour when alum dosing
for P removal was initiated.
Table 7.3 Abattoir SBR – operating data.
Parameter Average Operating Value
Before Alum With Alum

Dosing 20-Jul to Dosing1 5-Oct to
4-Aug-03 16-Dec-03
Influent
COD biodegradable mg/L 1350 1350
Total N mg/L 190 206
Total P mg/L 40 43
(Continued)

by guest
Table 7.3 Abattoir SBR – operating data (Continued).
Parameter Average Operating Value
Before Alum With Alum

Dosing 20-Jul to Dosing1 5-Oct to
4-Aug-03 16-Dec-03
Effluent
SS mg/L 12 20
NH3-N mg/L 1.6 7.0
NOX-N mg/L 5.5 18
PO4-P mg/L 29 10
Process
Alum Dose mg/L as Al2(SO4)3 0 360
SRT d 20 20
MLSS g/L 6.0 6.8
SVI mL/g 158 74
1
Period beginning three SRTs (60d) after dosing began.
This experience supports the premise that alum or iron dosing weakens the floc structure, producing a
reduction in average floc size and an increase in effluent SS.
Overflow and settling rates

Figure 7.7 shows two years of SSVI and clarifier effluent SS data from a large conventional activated
sludge plant with primary treatment and step feeding. Median SSVI and effluent SS values for this
period were 90 mL/g and 19 mg/L respectively, which are generally consistent with the conventional
activated sludge data in Figure 4.11. However, in Figure 7.7 it can be seen that judicious choice of scales
highlights a strong correlation between the effluent SS trend and the ratio of overflow rate (Uo) to sludge
settling velocity (Vs) at the feed MLSS concentration. The sludge settling velocity, governed by both
MLSS concentration and SSVI, was measured as initial settling velocity in the routine SSVI tests, and is
generally consistent with expectations from Equation 4.2.
Uo / Vs SSVI SSe
3 120
Effluent SS (mg/L)
SSVI (mL/g)
2 80
Uo / Vs
1 40
0 0
01-Nov-08 01-Feb-09 01-May-09 01-Aug-09 01-Nov-09 01-Feb-10 01-May-10 01-Aug-10 01-Nov-10
Figure 7.7 Two years of daily data from the circular secondary clarifiers at Melbourne’s 350 ML/d Eastern
Treatment Plant. SSe is the clarifier effluent SS concentration, Uo the clarifier overflow rate and Vs the
sludge settling rate at the MLSS concentration in the clarifier feed.

by guest
It can also be seen in Figure 7.7 that over the last few months, when the operating Uo:Vs ratio increased to
1 or more, the effluent SS concentration became more variable and excessive solids losses occurred, as
would be expected.
Rising sludge
Rising sludge occurs in the clarifier of a nitrifying plant when denitrification in the clarifier underflow zone
produces sufficient volume of precipitated gas bubbles to float solids to the surface. Incipient rising sludge
can be seen as a few isolated lumps of sludge, particularly just behind the scraper, while a full blown problem
produces a thick scum layer covering the whole tank.
When mixed liquor enters the clarifier any residual dissolved oxygen is exhausted within a few minutes.
The biomass then switches to nitrate respiration, producing nitrogen. When the water becomes
supersaturated with nitrogen, bubbles form and attach themselves to the solids. Other gases,
predominantly nitrogen and carbon dioxide, also precipitate so that the composition of the gas bubbles
reflects the partial pressures of all the gases in solution. When the solids density has been reduced
sufficiently, clumps of sludge break free and rise to the surface.
Experience shows that rising sludge will generally not occur if the nitrate concentration in the clarifier
feed is less than 15–20 mgN/L.
7.4.2 Other effects of clarifiers on effluent quality

Impacts of clarification on effluent SS and denitrification have been discussed above. Clarifiers can also
affect effluent quality in other ways:
• Depending on the relative volumes of reactors and clarifiers, large feed wells can contain up to 10% of
the total sludge mass in the process. This volume acts as a secondary zone in which endogenous
ammonia release and further nitrification/denitrification may occur. These processes may have a
measurable effect on very low effluent ammonia and total N concentrations.
• Chemical precipitant dosing reduces effluent soluble organic N (Chapter 3). Given that chemical
precipitant dosing increases effluent SS and associated particulate N, there may therefore be an
optimum precipitant dose producing minimum total N concentration.
• Similarly, precipitant dosing will increase effluent particulate P. There may therefore be an optimum
precipitant dose producing minimum total P concentration.
7.5 CLARIFIER STRESS TESTING

Plant clarifier capacity can be assessed by stress testing – taking clarifiers off-line to increase the load on the
remaining operating units. The procedure can be illustrated using a stress test conducted at the Loganholme
oxidation ditch plant in Brisbane (Hartley, 2012). The facility comprised parallel A and B Plants with design
peak wet weather flow capacities of 90 and 110 ML/d respectively (200 ML/d total).
The A Plant had four secondary clarifiers, two of 32 m diameter and two of 36 m diameter, all with
3 m side water depths. The B Plant had four 40 m diameter clarifiers with 5 m side water depths.
Operating SRT was 10 days. The test was intended to provide data on (a) the relationship between
overflow rate and effluent SS, and (b) the clarifier gross failure capacity. The test was carried out by
progressively reducing the number of clarifiers on-line, at constant RAS flow per clarifier, to increase the
clarifier overflow rates and reduce the RAS recycle ratios.

by guest
The impact on effluent SS is shown in Figure 7.8. The effluent SS concentration was lower in the B Plant
than in the A Plant, possibly because of the B Plant’s greater side water depth.
15
A Plant Trend:
SS = 1.47Ln(Uo/Vs) + 8.2
12
B Plant Trend:
Effluent SS (mg/L)
SS = 0.7Ln(Uo/Vs) + 3.7
9
A Plant
6
3 B Plant
0
0 0.2 0.4 0.6 0.8 1
Uo/Vs
Figure 7.8 Effluent SS correlated with clarifier loading. Uo is clarifier average overflow rate based on the
average plant inflow rate for the hour preceding the sampling time; Vs is the sludge settling rate at the
MLSS concentration, calculated from the MLSS and SSVI using the relationships of Ozinsky & Ekama
(Table 4.1).
Figure 7.9a shows the results of the A Plant test with only one 32 m diameter clarifier on-line. In this test
the clarifier loading was high enough to demonstrate thickening failure. The following can be observed:
(1) The ratio of overflow rate to theoretical settling rate at the MLSS concentration, UO/VS, increased
steadily over the morning from 0.1 to 0.9.
(2) At the same time the RAS recycle ratio (at a constant 95 L/s, governed by the RAS bellmouth
setting – full down) decreased from greater than 1 to 0.2.
(3) As a result the sludge blanket level (measured with hand-held instrument) increased from near-zero
to 50% of the side water depth, above the bottom of the feed well located at 32% (2 m below the
surface).
(4) Effluent SS was 7 mg/L at 07:30 and 08:30 when the blanket was below the feedwell; effluent SS
was consistent with previous tests (Figure 7.8). At 09:30 the blanket level at the monitoring point
was above the bottom of the feedwell and the effluent SS was measured at only 5 mg/L, presumably
because the blanket was providing a degree of additional flocculation. However, at 10:30 when
the blanket level was nearly 0.6 m above the bottom of the feedwell and 1.4 m below the
surface, the effluent SS had increased to 120 mg/L, suggesting that disturbance of the blanket
was causing carryover.
The clarifier operating condition during the test is shown in the operating diagram in Figure 7.9b, drawn
for the operating SSVI of about 80 mL/g. The lines rising from the origin are failure boundaries for the
MLSS concentrations shown. Above the curved 1.15 m/h underflow rate line, clarification (overflow
rate) governs capacity whereas below the underflow rate line thickening (RAS recycle ratio) governs.
The failure boundary for MLSS 2.8 g/L is the failure line for the test condition. The operating conditions
during the test are shown as the 0.43 m/h underflow rate line. It can be seen that the operating line lies well

by guest
within the thickening zone, meaning that the RAS recycle ratio (thickening) governs capacity. Failure occurs
when the underflow line reaches the failure boundary at point F, shown in both (a) and (b). This means that at
an SSVI of 80 mL/g, an MLSS concentration of 2.8 g/L, and a recycle flow of 95 L/s, when the RAS
recycle ratio falls below about 0.3 the sludge cannot thicken sufficiently for the solids outflow in the
RAS to match the solids inflow to the clarifier. Solids will therefore accumulate in the clarifier until
solids loss in the effluent makes up the difference. It can be seen in Figure (a) that just after Point F was
reached the effluent SS had increased to 120 mg/L.
(a) 1 2
MLSS 2.8 g/L, sSVI 81 mL/g
Uo = clarifier overflow rate based on 30 min MA flow
Vs = ML settling rate
2. Blanket Level (depth fraction)
0.8 1.6
4.Underflow Zone SS (g/L)

5.Effluent SS (0.01*mg/L)
1.RAS ratio (at 95 L/s) 3.Uo/Vs
1. RAS Ratio
0.6 1.2
3. Uo / Vs
2.Blanket Level
0.4 Bottom of feed well 0.8

(fract'n of SWD -LH axis)
0.2 4.Underflow Zone SS F 0.4
5.Effluent SS
0 0
6:30 7:30 8:30 9:30 10:30 11:30
Time of Day
ND Process: SSVI 80 mL/g

(b) 4
Underflow Rate
1.15 m/h 2 g/L
3
Overflow Rate (m/h)
MLSS 2.8 g/L
2
4 g/L
F
5 g/L
1
6 g/L
Underflow rate 0.43 m/h

0
0 1 2
Recycle Ratio
Figure 7.9 (a) A Plant clarifier behaviour with only one clarifier online. When effluent SS were seen to have
risen sharply at 10:30 am the stress test was abandoned and other clarifiers brought back online. RAS flow
95 L/s (maximum flow from one bellmouth). Parameters 1, 2 & 5 were measured, other parameters were
calculated. (b) Clarifier operating diagram showing the operating conditions during the stress test
(underflow rate 0.43 m/h). As the test progressed and the recycle ratio declined the operating condition
moved up the 0.43 m/h line until thickening failure occurred at point F, marked in both (a) and (b).

by guest
The A Plant gross clarifier capacity at SSVI 80 mL/g can be estimated from these test results as follows:
• Point F in Figure 7.9b represents the A Plant test capacity (with minimal sludge transfer to the
clarifiers as only 20% of the total clarifier area was on-line).
• During the test period the plant COD load was only 60% of the design load. At the full design load the
A Plant MLSS concentration would increase to 4.7 g/L, however sludge transfer to the clarifiers
under peak conditions would reduce that to 3.6 g/L.
• From Figure 7.9b, at an MLSS concentration of 3.6 g/L the maximum overflow rate would be
1.2 m/h (limited by the RAS capacity), giving an A Plant clarifier peak flow capacity of 105 ML/d.
The clarifier testing produced results consistent with theoretical expectations and provided insight into the
clarifier behaviour under high load conditions.

by guest
Chapter 8
Sludge processes
8.1 OVERVIEW
This chapter describes the characteristics of the sludge stabilisation and dewatering processes common to
BNR plants. Primary treatment is rare and the waste activated sludge is usually stabilised by
aerobic digestion.
A typical solids stream process format is shown in Figure 8.1. The aerobic digester may be decanted
periodically to thicken the sludge and increase the solids digestion time. Sludge is usually dewatered by
belt filter press (BFP) or centrifuge. Belt washwater is required for the BFP.
Influent Effluent
LIQUID STREAM PROCESSES
Filtrate Decant
or WAS
Centrate Washwater
SLUDGE AEROBIC
DEWATERING DIGESTION
Sludge Cake
Figure 8.1 Interaction between the liquid stream and solids stream processes.
The following sections discuss the characteristics of aerobic digesters, BFPs and centrifuges.
8.2 AEROBIC DIGESTION

The usual goal of aerobic digestion is to stabilise the WAS to a standard suitable for agricultural reuse.
A standard commonly applied is a reduction of the biosolids specific oxygen uptake rate at 20°C
(SOUR20) to less than 1.5 mgO2/h/gTS (Grade A solids stabilisation for vector attraction reduction,
EPA NSW, 1997).

by guest
Figure 8.2 shows the basic single stage digester which has the following features:
• Single tank with variable volume to allow wasting and dewatering to be operated at different times
and rates. For example, sludge may be wasted to the digester on seven days per week but sludge
dewatering may operate on only five days.
• Possible supernatant decanting to increase the SRT over the HRT and increase VS destruction.
• Intermittent aeration for nitrification and denitrification of released nitrogen. Aerating intermittently
reduces alkalinity loss and reduces nitrogen recycle to the bioreactor in the dewatering return stream.
• Alum (or iron) dosing may also be provided to precipitate P released by VSS destruction. This reduces
phosphorus recycle to the bioreactor. When a plant requires alum for supplementary P removal,
dosing of the digester reduces overall alum consumption, and therefore the alum content of the
digested sludge, if the digester is operated at a phosphate concentration higher than the reactor
level (because alum P uptake is greater at higher P level, see Figure 3.36). In addition, alum
precipitate in the digester feed can take up additional P at the higher digester phosphate level.
• Alkali dosing to counteract the alkalinity loss due to nitrification/denitrification and alum dosing and
maintain an acceptable digester pH.
Alkali Al or Fe WAS, QW
AEROBIC
Supernatant
Digested Sludge, QS DIGESTER
Decant, QD
VD
Intermittent
Aeration
Figure 8.2 Basic single stage aerobic digester with alum or ferric dosing for precipitation of released P and
alkali dosing to counterbalance alkalinity loss due to nitrification/denitrification and alum dosing.
The hydraulic and solids residence times can be calculated as follows:

HRT = VD /QW (8.1)
SRT = VD /QS (8.2)
where
HRT = hydraulic retention time, d

SRT = solids retention time, d
QS = sludge flow to dewatering, kL/d
QW = WAS flow into digester, kL/d
VD = digester operating volume, kL
8.2.1 Stabilisation performance

Figure 8.3 shows the sludge stabilisation performance of a single stage complete mix digester. Multi-stage
digesters can achieve the same degree of stabilisation with a shorter total residence time.

by guest
Sludge Processes 139
(a) Process Temperature 10°C

Single Stage
4
SOUR at 20°C
SOUR at 10°C
SOUR (mgO2 /h/gTS)

2 SRT10 WAS SRT10
SRT15 SRT 15
SRT30 SRT25 SRT20
SRT 20
1.5 SRT30 SRT25
0
0 2 4 6 8 10
Digester SRT
(b) Process Temperature 20°C

Single Stage
4
SOUR at 20°C
3
SOUR (mgO2 /h/gTS)
1.5 WAS SRT10

SRT 15
SRT 20
SRT30 SRT25
1
0
0 2 4 6 8 10
Digester SRT
(c) Process Temperature 30°C

Single Stage
4
SOUR at 20°C
SOUR at 30°C
3
SOUR (mgO2 /h/gTS)
2 SRT10
SRT30 SRT15
SRT25 SRT20
1.5
1 WAS SRT10
SRT 15
SRT25 SRT 20
SRT30
0
0 2 4 6 8 10
Digester SRT
Figure 8.3 Sludge stabilisation performance of single stage aerobic digesters fed WAS with MLVSS/MLSS
0.7. Figures (a) to (c) relate to process operating temperatures (WAS source and digester) of 10–30°C and
show lines for WAS source SRTs of 10–30d. Dotted lines represent the SOUR within the digester and solid
lines the SOUR measured on samples adjusted to 20°C.

by guest
Table 8.1 summarises the single stage digester residence time needed to achieve an SOUR20 of 1.5
mgO2/h/gTS.
Table 8.1 Digester SRT to achieve grade A solids stabilisation1.
WAS Source SRT d 10 20 30
Temperature °C 10 20 30 10 20 30 10 20 30
Digester SRT d .10 9 5 8 4 2 5 2 ,1
1
WAS MLVSS/MLSS 0.7.
8.2.2 Operating characteristics

Table 8.2 lists typical characteristics for an aerobic digester fed WAS from a bioreactor operating at 20°C
and 20d SRT. A simple test procedure for measuring SOUR is set out in Table 8.3.
Table 8.2 Typical digester operating characteristics.
SRT Oxygen Demand Nutrient Chemical Dose Rate (when used)

(d) (mg// g feed SS) Release
(mg// g feed SS)
N P Alum1 (mg solid Alkali2
alum// g feed SS) (mgCaCO3/ g feed SS)
N// DN N// DN plus alum
5 1.4 9 6 50 30 40
1
Example WAS P/MLSS 0.05 (see Figure 3.30). Digester PO4-P ∼ 1 mg/L.
2
Replacement of alkalinity loss due to nitrification/denitrification and alum dose. With full digester nitrification/denitrification
and an operating pH close to 6.0, alkali dosing may not be needed.
Table 8.3 SOUR test procedure.
Procedure
1. Sample the digester during the period sludge is being fed to the dewatering system.
2. Mix the sample and divide into two. Measure the SS concentration of one portion.
3. Use the other portion to measure the oxygen uptake rate. Conduct the test at a temperature as near as
possible to 20°C. Aerate the sample to raise its DO concentration to several mg/L. Record the sample
temperature. Stop aerating and maintain uniform SS by gentle stirring (to minimise surface aeration) with
a magnetic stirrer or similar. Monitor the rate at which DO falls – time and DO.
4. Plot the falling DO profile and calculate the slope of the straight part of the line (mg/L · min). To adjust to
20°C, multiply the result by 1.029(20-T), where T is the test temperature.
5. Calculate the specific oxygen uptake rate in mgO2/h/gTSS from:
SOUR20 = rate of decrease in DO (mg/L · min) × 60/SS (g/L)
(Continued)

by guest
Table 8.3 SOUR test procedure (Continued).
Example Results
9
Regress open squares
8 y = –0.060x + 6.857
7 R2 = 0.999
6
DO (mg/L)
1 TSS 6.08 g/L, temperature 23.7°C

SOUR = 0.534 mgO2/h/gTSS at 20°C
0
0 10 20 30 40 50 60 70
Time (minutes)
8.2.3 Operating examples

West Byron STP
The various features of aerobic digester operation are illustrated in Figure 8.4. These trends are taken from
the first two years operation of the new West Byron plant, with a plant load of about half the design load. The
plant comprised an oxidation ditch with anaerobic prereactor and an aerobic digester; biological P removal
was supplemented by dosing of alum and magnesium hydroxide to the oxidation ditch.
For the first six months the digester operation was quite variable while the behaviour and interaction of
the various plant operating characteristics were explored and a routine monitoring schedule developed.
Performance became more settled as understanding grew.
Plot 1 shows that, once operation was stabilised, the WAS feed was taken from a bioprocess operated at
an SRT around 20 days. The digester HRT and SRT were both around 15 days (no supernatant was
decanted). The digester was fed 7 days per week and digested sludge was dewatered 5 days per week,
causing the digester water level to vary. WAS was taken from the RAS line and SS averaged 7 g/L,
while the SS of the digested sludge averaged about 5 g/L.
Plot 2 shows that the digester was aerated for about 40% of the time and the DO concentration during
aeration averaged about 7 mg/L. With this aeration regime the effluent ammonia and nitrate were both
low. The optimum aeration regime giving low N and maximum alkalinity recovery depends on the
oxygen demand and aeration DO concentration and can be established by trial. Figure 8.5 shows a useful
aeration profile but the daily percentage of aeration time giving lowest N will vary with the operating
DO and the associated decay time when aeration ceases.
Plot 2 also shows that for most of the time digester phosphate was maintained at a low level by digester
alum dosing. Plot 3 shows that the average alum dose was high compared with the 50 mg/g feed SS listed in
Table 8.2 for full P removal. This is because the digester pH was maintained at high level by alkali dosing
(Plots 3 and 4); alum P uptake is lower at higher pH (see Figure 3.36). Digester P rose over the period

by guest
Nov-06 to Mar-07 when the alum dose was reduced. Note the lag of about six weeks between change in alum
dose and the full P response, roughly equal to three HRTs (see Chapter 6). The most economical alum dose
would have maintained the P level at about 1 mg/L where alum P uptake is near its maximum.
1. Loading Dig SS WAS SS Design HRT Dig HRT WAS SRT WL

30 100
WL
SRT, HRT (d), SS (g/L)
24 80
Digester WL (%)
18 WAS SRT 60
WAS SS Dig HRT
12 40
6 20
Dig SS
0 0
2. Nutrients Dig NH3-N Dig NO3-N Dig PO4-P DO during aeration Aeration time
12 0.6
NO3-N
Aeration Time (fraction)

DO Aeration time
10 0.5
N, P, DO (mg/L)
8 PO4-P 0.4
6 0.3
4 0.2
2 0.1
Dig NH3-N
0 0
3. Chemical Dosing Alkali as mgCaCO3/g WAS SS Alum as mg solid/gWAS SS

150
125
Chemical Dose
100
75
50
25
0
4. Alkalinity & pH TA pH
150 8.0
125 7.5
TA (mgCaCO3/L)
100 7.0
pH
75 6.5
50 6.0
25 5.5
0 5.0
Figure 8.4 Operating trends for the aerobic digester at the West Byron STP after plant startup, Nov-05.
Air Air Air

On Off On
07:00 15:00 Nominal Time of Day 07:00
Figure 8.5 A practical digester aeration regime allowing for continuous aeration and mixing during the day
while sludge is being fed to the dewatering system and intermittent aeration at night to provide an
adjustable daily anoxic mass fraction for ammonia and nitrate control.

by guest
Brunswick Valley STP

Figure 8.6 shows operating trends from the aerobic digester at the new Brunswick Valley STP after startup in
Dec-10 at about half the design load. Relationships between operating characteristics are highlighted by
appropriate choice of scales.
1. Digester: HRT, SRT & TSS HRT-7d MA SRT-7d MA TSS

25 15
20 12
HRT, SRT (d)
TSS
15 9
TSS (g/L)
SRT
10 6
HRT
5 3
0 0
1-Dec 1-Feb 1-Apr 1-Jun 1-Aug 1-Oct 1-Dec 1-Feb 1-Apr 1-Jun 1-Aug 1-Oct 1-Dec
2. Digester: Aeration & Nutrients Dig NH3-N Dig NO3-N Dig PO4-P Aerated Fraction
12 0.5
Daily Aerated Fraction

9 0.4
N, P (mg/L)
Aerated Fraction
6 0.3
PO4-P
3 0.2
NO3-N
NH3-N
0 0.1
3. Digester: Chemical Dosing Alum Caustic Soda

150 400
(mgCaCO3/g feed SS)

120 320
(mg solid/g feed SS)
Caustic Soda Dose

Alum Dose
90 240
60 160
30 80
0 0
4. Digester: pH & Alkalinity pH TA

8.0 150
7.6 120
TA (mgCaCO3/L)
7.2 90
pH
6.8 60
6.4 30
6.0 0
5. SVI & Alum Dose OD-SVI Digester SVI Digester Alum Dose
300
Alum Dose (mg/g feed SS)
Unstirred SVI (mL/g)
225
150
75
0
Figure 8.6 Operating trends for the aerobic digester at the Brunswick Valley STP after plant startup, Dec-10.
The bioreactor was initially seeded with dewatered sludge cake from sister plant West Byron, producing a very
low initial SVI. Coloured versions of Plots 1, 2, 3, 4 & 5 can be seen in Figure 11.13, Plots 19, 22, 23, 24 &
21 respectively.

by guest
The digester was fed with WAS taken from the RAS system. To supplement bio-P removal, the bioreactor
was dosed with alum at an initial rate of about 100 mg solid alum/L of plant flow, gradually declining to
about 30 mg/L; like West Byron, the WAS therefore contained alum precipitate. Over the nineteen
month data period plotted, the bioreactor SRT was gradually decreased from about 40 days to 27 days.
The trend plots in Figure 8.6 show the following digester behaviour:
(1) Digester SRT was gradually increased from 5 to 15 days (Plot 1) by increasing the extent of
supernatant decanting (and reduced feed rate in the last month).
(2) Virtually complete N removal was maintained using a daily aerated fraction of 0.34 (Plot 2).
(3) Alum and caustic soda doses were maintained to the digester over the data period to precipitate P
released and maintain pH above 6.0 (Plot 3). Plot 2 shows the resulting P levels. Plot 4 shows the
relationship between digester pH and alkalinity. An alkalinity of 10–20 mg/L was sufficient to
maintain the pH above 6.0. Over Dec-11/Jan-12 the alum dose decreased (Plot 3) and PO4-P
rose (Plot 2). Alum dose was therefore increased at the start of Feb-12 to reduce P recycle from
the digester; this reduced pH (Plot 4). Note that reduction of alum dose has a two-fold effect on
P uptake because the associated pH increase reduces the adsorption capacity of the remaining
alum (Section 3.7.1).
(4) Trend Plot 5 shows that the digester SVI lies below the bioreactor (oxidation ditch) trend,
presumably resulting primarily from the additional alum dosed to the digester. It can be seen that
the step increase in alum dose at Feb-12 widened the gap between the bioreactor and digester
SVIs while the previous step decreases (Jun-11 and Oct-11) had the opposite effect.
Figure 8.7 shows the belt filter press dewatering characteristics of the sludge. Cake TS concentration varies
inversely with digested sludge SVI (Plot 1) while solids capture varies directly with SVI (Plot 2). Further,
comparison of Figure 8.7, Plot 2 with Figure 8.6, Plot 4 shows that solids capture is higher at lower pH –
filtrate SS concentration is lowest around Apr-11 and from Feb-12 onwards when pH is lowest; this is
presumably because the coagulation properties of the alum sludge component are best at low pH values
(see Chapter 3, Section 3.7).
1. BFP: Cake TS Cake TS Digester SVI

17 50
16
Digester SVI (mL/g)

15
Cake TS (%)
14 150
13
12
11 250
2. BFP: Solids Capture Filtrate SS Digester SVI

700 50
600
Digester SVI (mL/g)
Filtrate SS (mg/L)
500
400
150
300
200
100
0 250
Figure 8.7 Dewatering characteristics of aerobically digested sludge on the belt filter press at the Brunswick
Valley STP, Dec-10 to Jul-12. Coloured versions of Plots 1 & 2 can be seen in Figure 11.13, Plots 25 &
26 respectively.

by guest
Bane (1997) shows that cake TS also varies inversely with the VS:TS ratio of the sludge.
Theoretically, there may be an economic optimum alum dose giving the lowest total sludge disposal cost.
Increasing the digester alum dose reduces SVI and increases cake TS. In rough terms, Plot 5 of Figure 8.6
suggests that an increase in digester alum dose from about 50 to 70 mg/g feed SS as occurred at Feb-12
would decrease the digester SVI from 150 to 130 mL/g. This increase in alum dose would theoretically
generate an increase in sludge SS mass of about 5 mg/g feed SS or 0.5% (Section 3.7). Plot 1 of
Figure 8.7 suggests the BFP cake TS may increase from 14% to about 14.5%, giving a net decrease in
cake tonnage of about 3% (fairly insignificant).

The basic operating principles for an aerobic digester are as follows:
(1) Aerate intermittently to nitrify and denitrify. This reduces N recycle to the bioreactor and
minimises loss of alkalinity. The DO concentration is not critical. The optimum aerobic fraction
is about 50% and the aeration time may have to be set somewhat shorter to allow for the DO
decay time.
(2) If practical, periodically allow the sludge to settle and decant supernatant to maximise the SRT and
digestion performance (VS destruction).
(3) Dose alum (or Fe) to precipitate the P released and prevent recycle to the bioreactor. A digester
PO4-P concentration of about 1 mg/L will utilise close to the maximum alum P uptake
capacity (Figure 3.36) and give lowest overall plant chemical consumption. Because the
phosphorus sorption capacity of alum decreases as pH increases, upward spikes in digester
pH can cause significant release of chemical P as can be seen in Figure 8.6 (Plots 2 and 4,
Jan-12).
(4) Dose alkali as necessary to maintain the pH just above 6.0. This is a low pH that can be
maintained without too much instability, is not too corrosive, allows effective nitrification and
minimises digester alum and alkali consumptions. Alkali dosed to the digester increases the
alkalinity of the return flow to the liquid stream process and contributes to the overall process
alkalinity balance.
(5) Routinely monitor operating volume, NH3-N, NO3-N, PO4-P, pH and total alkalinity. Other
parameters worth tracking are HRT, SRT, TSS, VSS and SOUR.
8.3 SLUDGE DEWATERING

This section describes the operating characteristics of the two most common sludge dewatering devices – the
belt filter press and the decanter centrifuge.
8.3.1 Mass balances

Sludge thickening and dewatering are mass balance processes – they are usually operated under more or
less steady state conditions and the flow and solids outputs have to equal the inputs. The two main
performance parameters are thickened or dewatered sludge concentration and solids capture.
Table 8.4 sets out the useful mass balance equations which can be used to calculate and monitor these
parameters.

by guest
Table 8.4 Mass balance for thickening and dewatering processes.
Washwater
QW, CW
Polymer
QP [CP = 0]
Feed Solids Output

QF, CF QS, CS
THICKENING or DEWATERING
PROCESS
Liquid Output
QL, CL
Aim:
1. To determine the process solids capture and solids output flow, knowing:
(a) All input flows and SS concentrations
(b) SS concentrations in solids and liquid outputs
2. Alternatively (for a thickener with a thickened sludge flow meter), to determine the solids output
(thickened sludge) concentration when the solids output flow is known.
Nomenclature:
Q Flow rate
C SS concentration
SC Solids capture (output solids mass/input solids mass)
Relationships:
By mass balance:
SC = 1 − [QLCL/(QFCF + QWCW)] (8.3)
QL = QF + QP + QW (cake flow negligible) (8.4)
QS = [QFCF + QWCW − (QF + QP + QW)CL]/(CS − CL) (8.5)
For a thickener with a thickened sludge flow meter:
CS = CL + {[QF(CF − CL) − QPCL]/QS} (8.6)
8.3.2 Belt filter press

In the belt filter press (BFP), sludge is dewatered firstly by drainage through a filter belt and then by
compression between the filter belt and a press belt. The two belts are guided by rollers and run at the
same speed. Polymer is used for conditioning of the feed.
To effectively optimise the performance of any process there must be some understanding of its
performance characteristics. For the BFP, the performance characteristics of most interest are cake solids
concentration (dryness), solids capture and polymer dosing rate. Cake solids and solids capture are
affected by several operating variables:
(1) Machine Settings:
1.1 Belt tension
1.2 Washwater flow rate and pressure

by guest
(2) Machine Operating Parameters:

2.1 Belt speed
2.2 Feed rate
2.3 Polymer type and dosing rate
(3) Feed Sludge Characteristics
3.1 Solids concentration
3.2 Dewaterability
The machine settings, belt speed and polymer type can usually be left once set up. Day-to-day operation
requires periodic adjustment of the feed rate and polymer dose to match variations in the feed sludge
dewaterability. Sludge dewaterability is affected particularly by chemical dosing, SVI and pH as shown
above. Operating guidelines are set out in Table 8.5 and a generalised test procedure in Table 8.6.
Table 8.5 BFP operating guidelines.
To achieve maximum cake solids concentration at acceptable solids capture (.90%) with minimum polymer
dose:
1. Set standard washwater flow rate and belt tension.

2. Operate at slow belt speed.
3. Operate at high feed rate.
4. Adjust the polymer dose to achieve full drainage within 300 mm of the end of the drainage zone.
5. Where the BFP is fed aerobically (rather than anaerobically) digested sludge, allow for the BFP solids
capture in the calculation of the bioprocess sludge wasting rate to achieve the desired SRT.
6. Conduct a planned test program to determine the process characteristics of the specific machine/sludge
combination.
7. Conduct ongoing monitoring of loading and performance to facilitate tuning.
Table 8.6 BFP test procedure.
Systematically vary feed rate, polymer dose and belt speed while holding belt tension, washwater flow rate
and feed sludge characteristics constant. Each of the variables should be varied over the full practical range to
enable the optimum operating region to be defined. Steps in the testing procedure are as follows (assuming
the BFP is fed from an aerobic digester):
1. Decide the test values for each of the variables.

2. The filtrate flow rate has to be determined for each test run. It can be calculated from Equation 8.4,
Table 8.4:
QL = QF + QP + QW
where QL = liquid (filtrate) flow rate
QF = feed (sludge) flow rate
QP = polymer flow rate, including any dilution water
QW = washwater flow rate
(Continued)

by guest
Table 8.6 BFP test procedure (Continued).
The polymer and washwater flow rates can be measured manually prior to the test series. If washwater
flow cannot be measured directly, one method is to disconnect the washwater piping from the spray
bars and direct the flow into a drum. Throttle the washwater line to adjust the pressure to the normal
operating value.
3. Polymer dose rate has to be adjusted for the various test runs. A method of measuring the dose rate set, or
the total consumption over a test period, is required. Any dilution water flow also needs to be measured.
4. To maintain constant feed quality, ensure no WAS is fed to the digester during the test series. This may
require that the digester be filled before testing begins. Aerate the digester continuously while tests are in
progress to maintain a uniform feed solids concentration. Sample the digester at the start, middle and end
of the tests to determine an average feed SS concentration. Measure the digester SVI and pH.
5. Conduct BFP test runs. Continue each run until steady state performance is judged to have been
achieved. Record the operating conditions and sample the cake and filtrate for TS and SS analysis.
During the first test it may be worthwhile continuing the run for a further 10 minutes and resampling
twice at 5 minute intervals to assess whether steady state has been achieved and the degree of
variability in the results.
6. Analyse the samples and record all test conditions and results.
7. As far as possible, review the test results progressively and adjust the test program if necessary. Analyse
the overall test results and extend the tests if further performance improvements appear possible.
Figure 8.8 shows the results of a BFP test conducted on the aerobically digested sludge at the
Brunswick Valley STP. Twenty four test runs were conducted over a range of belt speeds, feed rates and
polymer doses.
The figure shows both the test results and the derived performance contours. With a limited number of
test runs it is difficult to draw accurate curves, however the trends are clear. The significant features of these
characteristics are as follows:
(1) For the sludge tested, there is an optimum solids feed rate of about 200 kgDS/h · m belt width
(equivalent to a sludge feed rate of 10 L/s at a sludge SS concentration of 6.7 g/L). This feed
rate gives the highest cake solids concentration and solids capture.
(2) There is a minimum polymer dose of about 3 kg powder/tDS which gives an acceptable cake TS
concentration (about 14%). Above this dose the TS increases slightly. Below this dose the cake
solids decreases to about 12% TS and then increases again, however the solids capture
deteriorates sharply.
(3) Belt speed has little effect on performance over the tested range of 70%–90% (3.68–4.64 m/min).
It also became apparent that the routine operating point (shown) as determined intuitively by the
operator was close to optimum – acceptable cake solids concentration and solids capture at low
polymer dose.
8.3.3 Centrifuge
The centrifuge used most commonly for sludge dewatering is the solid bowl decanter. A relatively high dose
of polymer is required for conditioning of the feed because of the high shear forces caused by turbulence at
the inlet to the bowl and by the high settling rates occurring in the clarification zone.

by guest
(a) Cake TS
12 12
Belt Speed 3.68 m/min Belt Speed 4.64 m/min
14.7 13.4
9 9
15.7
Polymer Dose (kg/tDS)

13.6
14.9 15.0%
14.2% 14.8%
14.2 14.8
Current oper'n
6 6 14.5
14.0% 13.6
13.9
14.2 14.7 14.5%
12.0% 12.3 12.0%
13.7 14.5
13.5%
11.5 13.9 13.4 13.0%
14.2 13.0% 14.3 14.0%
3 14.0 3
12.7
14.5% 14.5%
14.5 14.5
Shows cake TS test data & idealised performance contours Shows cake TS test data & idealised performance contours
0 0
0 100 200 300 0 100 200 300
Solids Feed Rate (kg/h.m belt width) Solids Feed Rate (kg/h.m belt width)
(b) Solids capture
12 12
Belt speed 3.68 m/min Belt speed 4.64 m/min
95.4 96.5
9 9
90.1
92.2 93.7
95.5
operation
97.6
96.1
86.4 92.1
Current op'n
6 93.3 6
98%
95.6 96.2
96.8 95% 95.6
93.7 96%
90.6
92.7 90.7 92.2 92.2
3 89.4 90% 3 92%
86%
40%
37.2 40%
43.0
Shows solids capture test data & idealised performance contours Shows solids capture test data & idealised performance contours
0 0
0 100 200 300 0 100 200 300
Solids Feed Rate (kg/h.m belt width) Solids Feed Rate (kg/h.m belt width)
Figure 8.8 Dewatering performance of a 1.2 m effective belt width BFP fed from the aerobic digester at the
Brunswick Valley plant in Aug-12. The then current operating point is also shown. Over the test period, the
bioreactor was being alum dosed at an average of 52 mg solid alum/L of sewage flow and the digester
was being further dosed at 97 mg/g feed SS. Digested sludge concentration was 6.9 g/L, unstirred SVI
124 mL/g and pH 6.5. Digester characteristics are further discussed in the text.
A cross-section of a typical solid bowl centrifuge is shown in Figure 8.9. This is a counter-current
machine in which the liquid and solids streams flow in opposite directions. Co-current machines in
which both streams discharge at the same end are also available.
The bowl rotates at high speed causing the solids to settle to the bottom of the pool whose depth is
controlled by adjustable centrate outlets at the wide end of the bowl. The bowl acts simply as a settling
basin and gravity thickener operating under greatly increased acceleration. The conveyor or scroll rotates
at a slightly different speed to the bowl and moves the solids up the conical section at the end. As the
solids move up the inclined beach and out of the end of the bowl they are subjected to additional
dewatering by drainage on the dry section of beach. As the solids emerge from the pool their weight is
increased because of the loss of buoyancy, and at the point of emergence the centrifugal acceleration is
greater than further up the beach due to the greater diameter. A component of the solids weight acts
along the sloping beach, hence only solids at a concentration high enough to withstand this slippage
force can be conveyed up the beach and out of the machine.

by guest
Bowl
Scroll
Feed
Pool
Beach
Centrate Cake
Figure 8.9 Cross-section of a typical counter-current solid bowl centrifuge.
The operating variables affecting centrifuge performance are as follows:
(1) Machine Settings:

1.1 Bowl speed (centrifugal force in units of g)
1.2 Pool depth
(2) Machine Operating Parameters:
2.1 Scroll differential speed (sometimes controlled by a scroll torque setting)
2.2 Feed rate
2.3 Polymer type and dosing rate, including dilution water
(3) Feed Sludge Characteristics
3.1 Solids concentration
3.2 Dewaterability
In this list the machine settings are the parameters which do not need frequent adjustment. The operating
parameters are adjusted as necessary to match variations in the feed sludge characteristics. Feed rate and
pool depth can be combined into hydraulic residence time (HRT; see Table 8.7). First order variables are
then g value, HRT, scroll differential and polymer dose. The scroll differential governs the depth of the
sludge layer in the pool and the corresponding solids thickening time. Note that the polymer and
associated dilution water flows contribute to the total feed rate and residence time. Because of the high g
forces, a high molecular weight polymer with a high cationic charge is necessary.
The relationship between centrifuge g value and bowl speed is as follows:
Centrifugal acceleration ( g value) = (pDn/60)2 /[(D/2)g] (8.7)
where
D = bowl inside diameter, m
n = bowl rotational speed, rpm
g = gravitational acceleration, 9.8 m/s2

by guest
Table 8.7 Centrifuge pool volume and HRT.
P α
Lc
Nomenclature:
D Bowl internal diameter, m
LC Bowl cylindrical length, m
P Pool depth (weir height), m
Q Total feed flow, sludge + polymer + polymer dilution water, L/s
RH Hydraulic residence time, s
VP Pool volume, L
α Beach angle, degrees
Relationships:
VP = (1000π/4)[4LCDP − (4LC − D/tanα)P2 − 5P3/(3tanα)] (8.8)
RH = VP/Q (8.9)
Idealised performance characteristics are shown in Figure 8.10 (Vesilind, 1974). Cake solids concentration
and solids capture depend on residence time generally as shown but actual dewatering characteristics vary
with the nature of the sludge. At very low residence time (Zone 1), only the heavier solids are removed as
cake, yielding a dry cake and poor capture. As the residence time is increased (Zone 2), the finer solids,
which are more difficult to dewater, are also captured, resulting in higher solids recovery but lower cake
solids concentration. Once all the solids are captured, further increase in the residence time may increase
the cake solids concentration (Zone 3).
Solids
Recovery
Cake Solids Concentration
100%
ZONE ZONE ZONE

1 2 3 4
Hydraulic Residence Time
Figure 8.10 Idealised decanter performance characteristics.

by guest
If the centrifuge is operating in Zone 2, increasing the centrifugal force (bowl speed) increases the solids
recovery but not the cake solids concentration. Once the recovery is approaching 100% increased centrifugal
force can increase the solids concentration, but usually only marginally. As low a speed as possible should
be used as this reduces operating costs and increases the life of the machine.
A centrifuge test procedure is set out in Table 8.8. This is the original set of procedures used to collect the
test results shown in Figure 8.11. The centrifuge was fed anaerobically digested sludge and operated at a g
value of 2460. It is apparent that centrifuge performance with this type of feed is somewhat different from the
idealised pattern shown in Figure 8.10: Zone 2 behaviour does not occur. The dominant operating
parameters are HRT and scroll torque (differential speed). Polymer dose has a second order effect on
cake TS but this can be inconsistent and periodic adjustments may be required to maintain the desired
trade-off between polymer consumption and cake TS. Minimum polymer dose is that required to
produce acceptable solids capture.
Table 8.8 Decanter centrifuge test procedure.
The general test procedure, showing the expected outputs, is as follows:

1. Select bowl speed, pool depth and conveyor differential:
a. Vary feed rate
b. At each feed rate, vary polymer dose
Output:
Poly Dose
Cake TS
Solids Capture
Residence Time
2. At selected residence time and conveyor differential:

a. Vary bowl speed
b. At each bowl speed, vary polymer dose
Output:
Cake TS
Poly Dose
Solids Capture
Bowl Speed (g-value)
(Continued)

by guest
Table 8.8 Decanter centrifuge test procedure (Continued).
3. At selected residence time, bowl speed and polymer dose:

a. Vary pool depth
b. At each pool depth, vary conveyor differential
Ouput:
Scroll Differential
Cake TS
Solids Capture
Pool Depth
25 1.0
20 0.8
Solids Capture
Cake TS (%)
15 0.6
Cake TS, Torque 55%

Torque 40%
Solids Capture
10 0.4
0 10 20 30 40 50
HRT (s)
Figure 8.11 Test characteristics for a high speed solid bowl decanter centrifuge dewatering anaerobically
digested sludge. Digester feed solids were 25% fermented primary sludge, 75% WAS from a BNR
process; the digested sludge averaged TS 2.7%, VS/TS 63%. Bowl speed was 3200 rpm, g value 2460.
Scroll differential was controlled by the torque settings shown. Polymer dose rates were 8–15 kg/tDS for
the 55% torque setting, 9–22 kg/tDS for the 40% torque setting; within these ranges, variation in polymer
dose had no consistent effect on performance. Reduction of the g value to 1390 at an HRT of 26 seconds
and a scroll torque setting of 55% reduced the cake TS concentration from 20.4% to 19.3%.

by guest
Operating guidelines are set out in Table 8.9.
Table 8.9 Decanter centrifuge operating guidelines.
To achieve maximum cake solids concentration at acceptable solids capture (.90%) with minimum polymer
dose:
1. Set appropriate bowl speed and pool depth.

2. Operate at slow scroll differential speed/high scroll torque (high solids residence time)
3. Operate at low feed rate (high hydraulic residence time).
4. Adjust the polymer dose to give acceptable solids capture.
5. When the centrifuge is fed aerobically (not anaerobically) digested sludge, allow for the centrifuge solids
capture in the calculation of the bioprocess sludge wasting rate to achieve the desired SRT.
6. Conduct a planned test program to determine the process characteristics of the specific machine/sludge
combination.
7. Conduct ongoing monitoring of loading and performance to facilitate tuning.

by guest
Chapter 9
Plant characteristics
This chapter discusses some operating characteristics of the total integrated plant – mass balances,
operational capacity and energy consumption.
9.1 MASS BALANCES

The various processes within the plant interact and simple mass balances help in understanding plant
behaviour. Steady state balances based on average operating conditions are generally sufficient to
achieve the following:
(1) Understanding what governs the flow rates and strengths of the various streams.
(2) Understanding the interactions of the various processes.
(3) Checking the accuracy of sampling results for various streams. Alternatively, calculating the
constituent concentrations in various streams without the need for sampling, especially where
accurate sampling is difficult.
(4) Determining the performance of processes when not all the performance data can be measured.
(5) Reducing the number of flow meters needed, or checking the accuracy of flow meters. Redundant
flow meters can cause heartache when the metered flows do not balance accurately.
The principles involved can be illustrated using the simple single-train flowsheet shown in Figure 9.1. The
simple steady state mass balance is set out in Table 9.1. Note that mass balances can be calculated in different
ways depending on the data available but the version shown is practical.
Examples of the application of mass balance principles are as follows:
RAS concentration
As shown in Chapter 7 (Table 7.1), the RAS SS concentration as calculated by mass balance is directly
related to the recycle ratio. At constant recycle ratio the RAS SS concentration is a constant multiple of
the MLSS concentration. Therefore when sludge wasted for SRT control is taken from the RAS line,
there is no need for RAS SS sampling if the RAS is flow-paced to the plant throughput.

by guest
Sludge thickeners
The concentrations of thickened sludge samples are notoriously variable. However, if the feed and thickened
sludge volumes, and the feed and return stream SS, are monitored, the average thickened sludge SS
concentration can be reliably calculated by mass balance. See Chapter 8, Table 8.4, Equation 8.6.
1 Pre
Influent Treatment
2 3 Secondary 4 5
Anaerobic Anoxic Aerobic Clarifier
Effluent
WAS 7
6
RAS
9
Thickener 13 Washwater
8 Thickened
Sludge
Aerobic
Digester
11 10
Polymer 12
BFP
Filtrate 14 15 Sludge
Cake
Figure 9.1 Simple BNR plant flowsheet.
Table 9.1 Mass balance for flowsheet shown in Figure 9.1.
Stream 1 2 3 4 5 6 7 8
Volumetric Flow, Q Q1 Q1 Q4 Q5 Q1 Q6 Q7 Q8
+ Q6 + Q6 + Q13 − Q15
Constituent Mass Flow, Q1C1 Q2C2 Q3C3 Q4C5 Q 5 C5 Q6C6 Q7C3 M7
M=Q·C −M9
(C = concentration)
Stream 9 10 11 12 13 14 15
Volumetric Flow, Q Q7 Q8 Q11 Q8 Q13 Q12 M15/C15

− Q8 + Q11 + Q13 (∼0)
− Q15
Constituent Mass Flow, Q9C9 Q8C10 Q11C11 M10 Q13C5 Q14C14 M12
M=Q·C (∼0) + M11 + M13
−M14

by guest
Plant characteristics 157
Sludge dewatering
The same principle applies to sludge dewatering. However, in this case the measurements of cake TS
concentration are more reliable than estimates of cake volume. Solids capture and cake production are
readily calculated from Equations 9.1 and 9.2, assuming the cake volume (at 10–20% TS) is negligible
compared with the total feed flow:
Solids capture (fraction) = 1 − (filtrate or centrate SS/feed SS) (9.1)

Cake DS mass (kg/d) = feed SS(g/L) × feed flow (kL/d) × solids capture (9.2)
Cake total mass (t cake/d) = cake DS mass/cake TS concentration (fraction) (9.3)
Dewatered sludge P content

A plant mass balance shows that all the phosphorus removed from the sewage reports to the dewatered
sludge cake. The cake P content is therefore:
Cake P content (fraction of DS) = Q(PTI − PTE )/[cake DS mass](kg/d) (9.4)
where
Q = sewage flow, ML/d
PTI = sewage total P, mg/L
PTE = effluent total P, mg/L
Biological P removal
When biological P removal is supplemented by chemical dosing the degree of biological P removal can be
simply estimated by deducting the calculated chemical P removal from the total removal:
DPB = PTI − PSE − DPAL (9.5)
where
ΔPB = biological P removal, mg/L of effluent flow
PSE = effluent soluble P, mg/L
ΔPAL = P removed by alum, mg/L of effluent flow
See Equation 3.18 in Table 3.5 for more detail.
9.2 OPERATIONAL PROCESS CAPACITY

The process capacity of a plant can be defined as the load at which the effluent quality will just meet (or will
just start to fail) its licensed performance standards, taking into account the inherent variability of the
influent and the process performance. This is different to the hydraulic capacity which is the maximum
flow which the plant can safely pass. It is also possible for the sludge stabilisation standards to limit capacity.
It is convenient to express the process capacity in terms of the median flow rate at the maximum load. At
higher loads one or more effluent quality parameters will exceed their license limits on an unacceptable
percentage of days. The operational capacity is inevitably different from the original design capacity
because of design and construction margins, differences between the design and actual sewage qualities
(strength and constituent ratios), and operating skill levels.

by guest
9.2.1 Bottlenecks
Process failure is caused by the single component of the plant, referred to as the bottleneck, which first
reaches its capacity. The capacity of the bottleneck governs the capacity of the whole plant. Potential
process bottlenecks and other factors affecting the liquid stream capacity include:
• Primary clarifier area when provided
• Flow distribution between parallel trains or units
• Reactor volume (governing MLSS concentration for a desired SRT)
• Secondary clarifier area
• RAS pumping capacity
• Aeration capacity
• Process control settings and tuning (responsiveness, variability): SRT, DO, mass fractions, recycle
ratios, chemical dose rates
• Biomass settleability (SVI)
• Recycles from the solids processing stream
• Degree of influent flow and quality variability resulting from diurnal and seasonal influences, storm
events, industrial wastewater inputs, special community events and the like.
Optimum flow distribution between parallel trains is important. Maximum plant capacity is achieved when
all trains fail at the same time; one train failing before the others wastes unused capacity in the non-failing
trains. This is explained in Table 9.2. As a rule of thumb, if one train in a plant receives 10% more than its
optimum share of the load, the total plant capacity is reduced by 9% (or putting it another way, the plant
capacity can be increased by 10% simply by redistributing the load).
Table 9.2 Effect of plant flow distribution on capacity.
Suppose a plant has two parallel trains, A and B. Each train fails when its load exceeds the capacity of its
bottleneck (the train component with the lowest capacity). As an example, suppose secondary clarification is
the bottleneck in each train and failure occurs by gross solids loss in the clarifier effluent. In this discussion
Train A is the failing train; Train B represents the rest of the plant and may be one or multiple parallel trains.
Suppose that the capacity of Train B is 50% greater
than Train A. If the trains are loaded optimally, Train A Train B capacity 0.6LO
receives 40% of the total (optimum) plant load, LO, Train B

Train A capacity 0.4LO
60%, and the clarifiers on both trains fail at the same time.
The train loads at failure equal the train capacities as
shown in Figure (a).
Train A B
Figure (a) Optimal loading at plant failure.

Suppose now that the trains are loaded non-optimally
and that Train A receives 10% too much load, that is,
Train A receives 44% of the total plant load instead of
40%. Train B therefore receives 56% instead of 60% as
depicted in Figure (b).
(Continued)

by guest
Table 9.2 Effect of plant flow distribution on capacity (Continued).
The train loads when Train A fails are now 0.4 LO for Train B 0.60 optimal
Train A (from Fig (a)) but only (0.56/0.44)0.4 LO = 0.51 0.56 actual
LO for Train B. The total plant capacity is reduced by 9% 0.44 actual Train A 0.40 optimal
to 0.91 LO.
Train A B
Figure (b) Flow distribution to one train is

10% higher than optimal.
Generalising, if several parallel trains with a total capacity of CO are loaded such that one train receives
(1 + x) times its optimum share, the total plant capacity, C, is reduced to CO/(1 + x). The result is as follows:
x: 0 0.1 0.2 0.3 0.4 0.5
C/CO: 1.0 0.91 0.83 0.77 0.71 0.67
9.2.2 Failure curves

Failure frequency analysis can be used to identify bottlenecks and determine theoretical plant capacity
(Gloag & Hartley, 2005). The technique involves using a dynamic process model to replicate several
years of historical loading and performance data. Once the model is calibrated the average plant load can
be progressively increased to simulate catchment growth while maintaining the historical variability of
influent flow and quality. By comparing the predicted effluent quality with the plant’s licence standards
a failure frequency diagram can be constructed that links the likelihood of failure to load growth. The
bottlenecks causing failure can also be identified.
Example failure curves for total N generated by this method are shown in Figure 9.2. Each of the curves
shows the percentage of days on which a particular licence limit would be breached at the associated plant
median loading and with the historical load variability. These curves illustrate typical plant failure responses
to increasing load.
40
Failure Frequency (% of days)
30
Long Term 80%ile Limit
20
10
Short Term 80%ile Limit
Maximum Limit
0 1 2 3 4 5
Median Flow Rate (ML/d)
Figure 9.2 Example of plant failure curves for effluent TN.
9.3 CAPACITY ENVELOPES

A capacity envelope is a useful aid to understanding the operating limitations of a process. A capacity
envelope is a graph showing the process failure boundaries as they are currently understood. It shows

by guest
what combinations of circumstances will cause a steady state process failure, but not the frequency with
which failure will occur. Provided the plant operating point lies within the failure boundaries gross
process failure will not occur.
The initial envelope can be drawn using the design parameters for the process. After a period of operating
experience with the plant, the diagram can be adjusted to reflect its actual operational capacity. With time,
improved understanding of the process behaviour allows the boundaries to be pushed out and additional
capacity or performance to be realised.
The capacity envelopes discussed here relate to the flow-COD capabilities which define the gross
capacity of the process. Similar diagrams can be drawn for N and P removal performance.
9.3.1 Continuous flow process

An example of a flow-COD capacity envelope for a BNR process is shown in Figure 9.3. This diagram is
based on the plant design data and shows flow rate (influent or RAS) on the abscissa and organic loading on
the ordinate. It is drawn for one SRT (the design value of 25 days) and separate diagrams can be drawn for
other SRTs.
12
RAS Capacity
Peak Flow 5 x 50%ile

SRT 25d
Clarifier Failure
10
(RAS non-limiting)
Aeration Capacity
SVI 100 mL/g
8 Min'm RAS Flows
COD Mass Load (t/d)
g/L
6 g/L MLSS SVI 150

0m
55
6 5 g/L
D
CO
Q50, 150 RAS at 3Q50, SVI 150 5Q50, 150 5Q50, 250 SVI 250
Inf
4 g/L
4 50%ile Load
3 g/L mg/L
D 110
Inf CO
2 g/L
2
1 g/L
0
0 10 20 30 40
Flow Rate (ML/d)
Figure 9.3 Flow-COD capacity envelope for a continuous flow process.
The diagram depicts seven sets of information:

(1) Firstly, the two sloping lines drawn through the origin are operating lines drawn for (a) dry weather
flow with a median influent COD strength of 550 mg/L (equal to the slope of the line), and (b) wet
weather flow of five times dry weather flow, with a COD strength of 110 mg/L. The diamond
marker shows the median dry weather operating point at the full design load (4.1 tCOD/d).
(2) The horizontal dotted lines show the MLSS concentrations produced by the associated COD loads.
(3) The dashed vertical line at far right shows the plant design peak hydraulic capacity, 37.5 ML/d.
(4) The horizontal line at 9 t/d shows the aeration capacity as the maximum sustained COD load that
can be handled.

by guest
(5) The curved lines in the top right hand corner show the clarifier overflow limits for three different
SVI values. Ignoring any reduction in MLSS concentration through solids accumulation in the
clarifiers, operation outside of these lines would produce significant loss of solids in the effluent.
(6) The solid vertical line shows the installed maximum RAS flow rate. Note that the abscissa here
represents RAS flow, not plant flow.
(7) The four curved lines through the origin are also represented as RAS flow on the abscissa. These
lines show the minimum RAS flows required at given COD mass loads to ensure RAS flow does not
limit clarifier capacity to less than the overflow limit. Four lines are shown for the four sets of plant
flow and SVI value noted in the figure – 1, 3 and 5Q50 and 150 or 250 mL/g. These lines are the
equivalent of Line C in a clarifier operating diagram (see Table 7.2).
The plant can operate safely within the envelope (operating space) defined by the various failure boundaries.
Table 9.3 explains how to draw this type of capacity diagram.
Table 9.3 Failure boundaries for a continuous flow process.
Peak Flow Limit

Design value as may be modified by testing
Aeration Limit
MSI = 24.OTRF/MOS (9.6)
Clarifier Limits
MLSS concentration:
MSI = VRX / MXS (9.7)
Clarifier overflow limit:
VS = VOe−nX (4.2)
Q = 24VSAC (9.8)
Minimum RAS flow:
QR = sQ (9.9)
s = exp(nX−2) (9.10)
where
MOS = specific oxygen demand, kgO2/kgCOD (Fig 3.25)

MSI = influent COD load, kg/d
MXS = specific sludge mass, kgMLSS/(kgCOD/d) (Fig 3.8)
n = sludge settling constant, L/g (Table 4.1)
OTRF = aeration system maximum field oxygen transfer rate, kg/h (6.13)
Q = influent flow rate, m3/d
QR = RAS flow, m3/d
s = RAS recycle ratio
VO = sludge settling constant, m/h (Table 4.1)
VR = reactor volume, m3
VS = sludge settling rate, m/h
X = MLSS concentration, g/L

by guest
9.3.2 SBR process

The SBR process has a high degree of operating flexibility. To maximise its capacity and performance it is
important to understand how the process behaves.
A capacity envelope for an SBR process (treating an abattoir wastewater – discussed in Section 7.4) is
shown in Figure 9.4. Construction of the diagram is explained in Table 9.4. The example diagram relates
to an SRT of 20 days and separate diagrams are needed for other SRTs. Four sets of information are shown:
(1) Four vertical lines (decant volumetric limits) show the maximum throughput over a cycle time or
greater based on the physical decanter constraint. The shorter the cycle time the greater the
maximum throughput because the maximum decant volume can be discharged more frequently.
(2) Two horizontal lines show the maximum COD load that the aeration system can handle. Greater
aeration capacity can be realised by increasing the aerated fraction of the cycle.
(3) The curved lines in the upper right hand sector show the decant scour limits, that is, the combination
of flow and COD load at which the settled sludge blanket will start to scour into the effluent during
decant. These boundaries are the equivalent of the clarifier failure boundaries for the continuous flow
process. Scour limits are shown for several operating conditions: (a) A, B and C show the limits for
cycle times of 4, 6 and 8 hours (with react fraction 0.5 and SVI 150 mL/g), (b) D and E show the
extremes for an SVI of 100 at a cycle time of 8 hours, and an SVI of 250 for a cycle time of
4 hours, (c) F shows what happens to the 8 hour line if the react fraction is increased from 0.5 to 0.6.
(4) Horizontal lines show the MLSS concentrations expected at reactor TWL for various COD loads.
SRT 20d
Decant Volumetric Limits
TC = 12h 8h 6h 4h
8
TA/TC 0.6 Aeration Capacity
7 8 g/L
(Sustained Load)
7 g/L TA/TC 0.5 D 8 0.5 100

6
COD Mass Load (t/d)
6 g/L
5 C 8 0.5 150
5 g/L B 6 0.5 150 Decant Scour

4 Limits
F 8 0.6 150
4 g/L
A 4 0.5 150
3
3 g/L
E 4 0.5 250
2
2 g/L TC TR/TC SVI (mL/g)
1 1 g/L MLSS TC = Total cycle time

TA = Aeration time
0 TR = React time
0 1 2 3 4 5
Flow (ML/d)
Figure 9.4 Capacity envelope for an arbitrary SBR process.
This diagram shows clearly the trade-offs inherent in maximising the operating capacity of an SBR
process. To achieve greater throughput the cycle has to be shortened. With a shorter cycle time the scour
limit decreases because the sludge has less time to settle. Aeration capacity (and aerobic sludge age for
nitrification) can be enhanced by increasing the aerated fraction of the cycle. But then the scour limit
decreases because of the shorter settling time or denitrification is reduced because of the shorter anoxic
mix time. The decrease in scour limit can be compensated for by increasing the cycle time (e.g. the
diagram shows that if the process was operating at a 4 hour cycle time with 0.5 aerated fraction, an
increase to 0.6 aerated fraction would be safe if the cycle time was also increased to 8 hours).

by guest
Table 9.4 Failure boundaries for an SBR process.
Decant Volumetric Limit

Q = 24QD / TC (9.11)
Aeration Limit
MSI = (24.OTR / MOS)(TA/TC) (9.12)
Decant SCOUR limit (Refer to Figure 6.20)
VS = VOe−nX (4.2)
Q = 6 LWVWVSTC[1−(TR/TC)]2 (derivation in footnote1) (9.13)
MLSS Concentration
MSI = VRX / MXS (9.14)
where
DC = supernatant depth, (DW – DS), m QF = decant flow rate, m3/h

DO = reactor depth at TWL, m t = time during cycle, h
DS = sludge depth, m TA = aeration time (T4-T1 in Fig 6.19), h
DW = water depth, m TC = cycle time (T7 in Fig 6.19), h
LW = decanting weir length, m TD = decant time (T7-T5 in Fig 6.19), h
MOS = specific oxygen demand, kgO2/kgCOD TR = react time (T4 in Fig 6.19), h
(Figure 3.25) TS = settle time to start of decant (T5-T4 in Fig 6.19), h
MSI = influent COD load, kg/d VD = weir descent velocity, m/h
MXS = specific sludge mass, kgMLSS/ VO = sludge settling constant, m/h (Table 4.1)
(kgCOD/d) (Figure 3.8) VR = reactor volume at TWL, m3
MXT = MLSS mass, kg VS = sludge settling rate at MLSS concentration X, m/h
n = sludge settling constant, L/g (Table 4.1) VW = limiting weir approach velocity during decant, m/h
OTR = aeration system maximum field oxygen (typically about 40 m/h based on the area
transfer rate, kg/h (Equn 6.13) bounded by the weir width and the water depth
Q = daily feed volume, m3/d above the settled sludge layer, see footnote2)
QD = maximum decant volume, m3/decant X = MLSS concentration at TWL, g/L
1
The clarification limit (Equn 9.13) is based on the maximum safe weir approach velocity (no scour of settled solids). The
equation is based on the settling & decant behaviour depicted in Figure 6.20, assuming the effluent is decanted optimally
such that the water level decrease parallels the sludge interface settling rate (by adjustment of T5 in Figure 6.19). The
derivation is as follows (symbols are defined in Table 9.4). From Equn 9.11, QD = Q.TC/24 (Equn (i)). The decant flow rate,
QF = QD/TD (ii). The reactor sludge depth, DS = DO – VSt (iii). The water level during decant is DW = DO – VD(t-TS) (iv), and
the supernatant depth is DC = DW – DS (v). Now, the weir approach velocity, VW, is given by: VW = QF/(LWDC). Substituting
Equns (i) – (v), setting TS = TD (where minimum VW generally occurs) and simplifying gives Q = 6 LWVWVSTC[1-(TR/TC)]2
(Equn 9.13 above).
2
Fair, Geyer & Okun (1968) recommend a maximum horizontal velocity to prevent scour of settled sludge in sedimentation
tanks of 10 times the particle settling velocity. For activated sludge, the settling constant VO represents the particle settling
velocity (i.e. the sludge settling velocity at zero MLSS concentration – Equn 4.2). A maximum weir approach velocity of 40
m/h should then be safe for VO values above 4 m/h, equivalent to SSVI values for ND processes of less than 160 mL/g
(Table 4.1).

by guest
9.4 ENERGY CONSUMPTION

To provide some perspective on sewage treatment plant energy consumption it is useful to start with the
overall water cycle energy balance.
9.4.1 Water cycle

The water cycle with typical energy inputs is shown in Figure 9.5. Evaporation of a nominal 300 L/person.d
total community water consumption from the ocean requires a solar energy input of about 680 MJ/p.d or 8
kW/person. The water supply and sewerage systems consume about 0.9 MJ/p.d of which the STP uses
about 60%. In comparison, the average household consumes electricity and food at the rates of 36 and
10 MJ/p.d respectively.
SOLAR ENERGY
HOUSEHOLD ENERGY
36 MJ/p.d
(400 W/p)
FOOD
10 MJ/p.d
(adult)
SEWAGE CONVEYANCE
200 L/p.d ELECTRICAL ENERGY
300 L/p.d 130 gCOD/p/d 0.08 MJ/p.d
+ 3°C (100 kWh/ML)
EVAPORATION
680 MJ/p.d (8 kW/p)
100 L/p.d (300 L/p.d, 2.26 MJ/kg)
ELECTRICAL ENERGY Losses (industry, irrigation)
0.3 MJ/p.d SEWAGE ENERGY CONTENT
(270 kWh/ML) 1.6 MJ/p.d Chemical (12.5 MJ/kgCOD)
(inc 2% at WTP) 2.5 MJ/p.d Heat (4.2 kJ/kg.°C)
ELECTRICAL ENERGY USE
0.5 MJ/p.d
(700 kWh/ML)
WTP BNR STP
1 kWh = 3.6 MJ
CO2 emission from black coal-fired electricity generation = 1.0 kgCO 2 / kWh
Figure 9.5 Typical water cycle energy balance for nominal per capita water consumptions.
The sewage contains wastes with a chemical energy content of 1.6 MJ/p.d and waste heat in the order of
2.5 MJ/p.d (Chapter 2).
Figure 9.6 shows the overall 1998 energy consumptions of the fifteen largest water and sewerage
businesses in Australia. The median sewerage energy consumption was 330 kWh/ML, less than half the
figure of 800 shown in Figure 9.5 for systems with BNR plants. BNR plants are more energy intensive
than conventional non-BNR plants and in 1998 totalled less than 10% of all plants in Australia. Further,
the energy consumption figures in Figure 9.6 relate to energy purchased, not energy used. Energy
purchased is lower at the larger plants where power is generated using digester gas.

by guest
1400
Waste water Businesses
Energy Consumption (kWh/ML)

1200 Water Businesses
1000
800
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100
Percent Greater Than
Figure 9.6 The 1998 purchased energy consumptions for the fifteen largest Australian water and wastewater
businesses – water collection, treatment and distribution; sewage collection, treatment and disposal (WSAA,
1999).
The total wastewater business energy consumptions are also affected by topography. The sewage
pumping component is greater when the topography is flat or undulating and where the sewage has to be
pumped a substantial distance for treatment and disposal.
9.4.2 Plant energy balance

A simple energy balance for a typical BNR plant with no primary treatment is shown in Figure 9.7. Specific
energy consumption in Australian plants using secondary clarifiers is in the order of 6 W/person (1.1
kWh/kgCOD) at full load, increasing at lower loads (see Section 9.4.3 below). This figure excludes
non-BNR demands such as influent and effluent pump stations and UV disinfection. The influent
chemical energy is about three times the plant electrical demand but this is not recovered in most
BNR plants.
Heat
21 110%
Aerobically Digested
BNR STP Biosolids
Influent Organics 3 15%
19 100% Electrical Demand

Effluent Organics
6 31%
1 6%
LEGEND
6 31%
% of Influent Electrical
W/p average Chemical Energy
Energy
Figure 9.7 Average energy balance for a BNR plant operating at full design load.

by guest
9.4.3 Energy models

General
Plant power can be formulated generally as:
PT = PI + PCOD + PQ (9.15)
where
PT = average total plant power, kW

PI = average load-invariant power3, kW
= kILD/24
PCOD = average COD mass load-dependent power, kW
= kLLO/24
PQ = average flow-dependent power, kW
= kQQO/24
kI = load-invariant energy constant, kWh/kgCOD
kL = load-dependent energy constant, kWh/kgCOD
kQ = flow-dependent energy constant, kWh/ML
LD = plant design average load, kg total COD/d
LO = plant average operating load, kg total COD/d
QO = plant average operating flow, ML/d
ET = specific total energy consumption at operating load, kWh/kg
total COD operating load
SO = plant average influent COD concentration, mg/L
= LO/QO
and specific energy consumption is:
ET = 24PT /LO
= (kI LD + kL LO + kQ QO )/LO
= kI LD /LO + (kL + kQ /SO ) (9.16)
The specific energy could alternatively be related to flow rather than COD load:
ET,Q = ET,COD .SO (9.17)
The average energy data for sixteen conventional Australian BNR plants are listed in Table 9.5 and
plotted in Figure 9.8. The overall correlation gives:
ET = 0.36(LD /LO ) + 0.75 = 1.1 kWh/kgCOD when operating at design load (9.18)
(or 550 kWh/ML for SO = 500 mgCOD/L)

Published figures for twelve Austrian BNR plants, all incorporating anaerobic sludge digestion and
on-site power generation, give: average 0.74 kWh/kgCOD, range 0.36–1.67 (Nowak, 2002).
3
Load invariant terminology is used in lieu of fixed to differentiate from the fixed rate component of electricity tariffs.

by guest
Table 9.5 Design and operating data.1
Plant Design ave LD / LO Specific energy

flow (ML// d) consumption kWh// kgCODT
With PST, Diffused Air
Maroochydore2 27 1.29 0.89
Nambour3 8 1.23 1.81
St Marys4 24 1.23 1.80
No PST, Diffused Air
Banora Point4 11.5 0.93 0.74
Brendale4 7.5 1.21 1.05
Brunswick Valley6 3.8 2.28 1.65
Gibson Island5 34 1.00 0.59
Noosa Coastal4 12 1.50 1.62
Penrith4 20 1.17 1.41
South Windsor4 3.2 1.26 1.59
Wacol5 6.9 1.00 1.50
West Wodonga4 5.9 1.72 0.96
No PST, Mechanical Air
Albury4 12 1.00 0.88
Bendigo4 27 1.67 1.01
Coolum6 8.3 1.72 1.33
Thorneside6 7.5 1.23 1.38
1
Ratio of average design to operating flows used for load ratio in Equn 9.18.
2
Hartley (2000)
3
JWP (2001)
4
AWWA (1997)
5
Tarong Energy (2000)
6
Plant data
1.5
kWh/kgCODt (Et)
M
With PST, Diffused Air
0.5 No PST, Diffused Air
Conventional BNR: No PST, Mechanical Air
Et = 0.357(Ld/Lo) + 0.751 MBR, With PST, Diff Air
0
0.8 1.2 1.6 2 2.4
Design Load / Operating Load (Ld/Lo)
Figure 9.8 Average specific energy consumptions for sixteen Australian continuous flow BNR plants, with
and without primary treatment. Non-core demands such as influent and effluent pumping and effluent UV
disinfection have been excluded. M is the Maroochydore STP discussed in the text. The MBR plant is not
included in the correlation and is discussed in the text.

by guest
Plant-specific energy characteristics

It is evident that energy consumptions can vary widely. Operators are encouraged to examine the energy
characteristics of their own plants to find ways to reduce consumption.
As an example, the energy characteristics of the Maroochydore STP have been explored in more detail.
This is a plant with primary treatment, Westbank BNR format and anaerobic sludge digestion (but no on-site
power generation). The monthly average plant powers for the 23 months Jan-99 to Nov-00 are plotted
against the monthly average flows in Figure 9.9. Over that period the connected population increased by
only 6% and the plant COD load can be regarded as essentially constant. There was a good spread of
monthly flows because 1999 was a wet year while 2000 was dry.
800
Pt = 11.6Q + 328
700
Monthly Average Power, Pt (kW)
600
500
400
300
COD-Dependent
200 238 kW
100
Invariant 90 kW
0
0 5 10 15 20 25 30 35
Monthly Average Flow, Q (ML/d)
Figure 9.9 Monthly average electrical energy consumptions at the Maroochydore STP Jan-99 to Nov-00.
Average COD load was 9.9 t/d compared with the design average of 13 t/d.
The electric drives in the plant can be categorised as load-invariant (screening, odour control, scrapers,
mixers, lighting, building services), COD-dependent (aeration blowers, sludge pumps, sludge thickening
and dewatering), or flow-dependent (influent pumps, grit handling, recycle pumps, UV disinfection,
effluent pumps). The estimated average operating power for each category is shown in the figure.
Maroochydore has a more significant flow-dependent power component than many STPs because of
influent and effluent pumping and UV disinfection. Without these demands, the average flow-dependent
power falls by two thirds. They are not included in the specific energy shown in Figure 9.8 (as noted in
the caption).
Note that invariant power will increase at the lower flows if the flow- and COD-dependent drives have
insufficient turndown and become invariant at the lower plant loads. In some cases it may be possible to
reduce invariant power by implementing intermittent operation of fixed speed drives.
MBRs
Membrane bioreactors generally have a somewhat higher energy consumption than conventional BNR
plants for several reasons:
• Membrane aeration for cleaning purposes
• Increased pumping associated with (a) fine screening of the influent, (b) higher recycle rate from the
membrane reactor compared with a clarifier, (c) additional recycle system to protect the anoxic zone
from dissolved oxygen, and (d) effluent head loss through the membranes.

by guest
• Greater switchroom complexity and cooling requirements associated with an increased number of
drives and controls.
The Cleveland Bay STP (Townsville) is used here as an example of MBR energy consumption
characteristics (with UF membranes). The BNR flowsheet is as shown in Figure 5.1j, with an oxidation
ditch for the nitrifying/denitrifying reactor. One of the two reactors is shown in Figure 6.12 and in the
cover photograph. The plant has primary treatment and anaerobic sludge digestion but no on-site power
generation. There is no on-site influent pumping or UV disinfection and effluent pumping is minimal.
The plant power characteristics are summarised in Figure 9.10. Power varies with flow rate, Figure (a),
but is insensitive to COD load, Figure (b), possibly because the COD-dependent power is a small component
of total power. Overall, the average specific energy consumption was 1.68 kWh/kgCOD which is 40%
above the conventional BNR trend as shown in Figure 9.8.
(a) 1000
800
600
400
200
Pt = 3.41Q + 654
0
0 20 40 60 80 100
Monthly Average Flow, Q (ML/d)
(b) 1000
800
600
400 Pt = 0L + 728
Black markers
200 Constant Flow Data are pre-PST
(20-30 ML/d, 19 months) commissioning
0
8,000 9,000 10,000 11,000 12,000 13,000
Monthly Average Load, L (kgCOD/d)
Figure 9.10 Monthly average electrical energy consumption at the Cleveland Bay MBR-BNR plant Feb-08 to
Jun-10. Design ADWF 29 ML/d, design average load 14.5 tCOD/d. Operating COD load 10.9 t/d average.
Monthly flow varied widely because of the impact of the tropical wet season. Plot (b) shows only dry
season data.

by guest
9.4.4 Tuning
It can be seen that plant energy consumptions vary over a wide range. There is obviously scope for reduction
in many plants. Operators should explore their own plants’ energy consumption characteristics in detail and
pursue practical reduction methods.

by guest
Chapter 10
Process control
Gustaf Olsson
Instrumentation, control and automation (ICA) attracted the attention of the water and wastewater industry
in the 1970s. Still, however, dynamic systems and process control are seldom part of the general civil
engineering or environmental engineering curricula. Consequently many water and wastewater system
designers are unaware of the potential of ICA. It has been demonstrated that ICA may increase the
capacity of BNR plants by 10–30% today. The advanced knowledge of the mechanisms involved
in biological nutrient removal that is being gained today is producing an increased understanding of the
processes and the possibilities for control. There is a sophisticated relationship between the operational
parameters in a treatment system and its microbial population and biochemical reactions, and hence its
performance. With further understanding and exploitation of these relationships the improvements due to
ICA may reach 20–50% of the total system investments within the next 10–20 years. Various case
studies of advanced control in water and wastewater treatment systems have shown significant savings in
operating costs and remarkably short payback times (Olsson et al., 2005).
10.1 THE ROLE OF CONTROL AND AUTOMATION

ICA in wastewater treatment systems has come a long way and is now an established and recognized area
of technology in the profession (Olsson & Newell, 1999). A number of factors have contributed to make
this progress possible:
• Instrumentation technology is today much more mature. Complex instruments like on-line nutrient
sensors and respirometers are now regularly used in the field.
• Actuators have improved over the years. Today variable speed drives in pumps and compressors are
proven technologies and are commonly used to allow a better controllability of the plant.
• Instrumentation technology is today much more mature. Complex instruments like on-line nutrient
sensors and respirometers are now regularly used in the field.
• Computing power can be considered almost “free”.
• Data collection is no longer a great obstacle. Software packages and SCADA systems are available
for data acquisition and plant supervision.

by guest
• Control theory and automation technology offer powerful tools. Benchmarking and various tools for
evaluating control strategy performance have been developed.
• Advanced dynamic models of many unit processes have been developed. Commercial simulators are
available to package the knowledge of plant dynamics.
• Operators and process engineers are often educated in instrumentation, computers and control ideas.
Still, however, there is a great need for better education in these areas.
• There are obvious incentives for ICA, not the least from an economic point of view. Plants are also
becoming increasingly complex which necessitates automation and control.
Today the main obstacle for more ICA is the lack of process flexibility. Plant design and operation still have
to be integrated in a systematic way.
10.2 DISTURBANCES
A major incentive for control is the presence of disturbances, and their impact has to be
compensated. Liquid and mass flow disturbances related to the influent have been described in
Chapter 2. Compared to most other process industries the disturbances to a wastewater treatment
plant are significant. The wastewater influent typically varies substantially both in its concentration,
composition and flow rate, with time scales ranging from fraction of hours to months. Discrete
events such as rainstorms, toxic spills and peak loads may also occur from time to time. As a result,
the plant is hardly ever in steady state, but is subject to transient behavior all the time (Olsson &
Newell, 1999).
Consistent performance must be maintained despite the disturbances. The traditional way of
dampening the disturbances has been to design plants with large volumes to attenuate large load
disturbances. This solution incurs large capital costs. On-line control systems, which have been
demonstrated to cope well with most of these variations, are a much more cost-effective and thus
attractive alternative. Disturbance rejection is indeed one of the major incentives for introducing on-line
process control.
Too often unnecessary disturbances are created within the plant itself. Often this depends on a lack of
understanding how the various parts of the plant interact. One example may illustrate the problem: if the
influent flow rate cannot be varied continuously but the pumps are operated in an on/off mode the
consequence is that the plant will be subject to sudden flow rate changes. In particular, the clarifier
operation will suffer from such sudden flow rate changes (see further Section 7.4 and in particular
Figure 7.8).
Recycling of water and sludge in a wastewater treatment plant creates apparent couplings between
various unit processes. If these interactions are not considered, then the plant operation will suffer. For
example, if anaerobically digested sludge supernatant is recycled to the plant influent during a high load,
then the nitrogen load to the plant may be very large and can be measured as a significant increase in the
oxygen uptake rate.
It is crucial to identify the sources of disturbances in order to obtain a high performance operation of
a plant. Then the control system can be structured so that disturbances are attenuated or even avoided
(Olsson et al., 2005). Further internal disturbances may be generated due to inadequate or inappropriate
operations including human errors, unsuitable or malfunctioning actuators and/or sensor breakdowns.
These may potentially cause major operational problems. Many of the internal disturbances may be
avoided (or their impacts minimized) through introducing on-line control systems, including early
warning systems.

by guest
Process control 173
10.3 CONTROL PRIORITIES

10.3.1 The system
It is quite apparent that a good operation and control depends on a whole chain of components that have to
work together. Online sensors and analyzers have to be maintained and calibrated. Actuators, such as
pumps, valves, motors and blowers, have to satisfy the requirements of good operation. In other words:
they have to deliver the control actions that are demanded by the control algorithms. The communication
system for signals between sensors, actuators and computers has to produce reliable information with
a minimum of signal disturbances. The software includes not only control algorithms, but also a data
acquisition system, methods for data screening, signal filtering, data bases, and human friendly
operator communication.
Most important of all: people. No control system can be presented to operators who have not been able to
influence the design of it. It is all built on trust. Any well-intended and functioning control system can be a
total failure if the operating people do not trust it. Therefore people involvement and education is a crucial
part of a successful system. So, what are the priorities?
(1) Keep the plant running. Section 10.4.
(2) Satisfy the effluent requirements by controlling key concentrations. Section 10.7.
(3) Minimize the cost by advanced control. This includes more advanced control of single variables. By
coordinating the control of various plant parts – such as the sludge treatment and the wastewater
treatment – the plant operation can be further improved. Finally a system-wide operation is
obtained if the sewer operation is integrated with the influent flow pumping control to the
wastewater treatment plant. This can make it possible to optimally use the available volumes and
the sludge for the best operation. See Section 10.8.
10.3.2 The importance of dynamics

Every operator knows that the plant is never in steady state. Instead most flow rates and concentrations are
varying with time. The fact that the plant is a dynamic system also means that the result of a corrective action
will take some time; it will never appear immediately. Therefore the time scales of the process changes are so
important. The dynamics of a wastewater treatment system involve a wide range of time scales, from
seconds to months. Typical time scales in a wastewater treatment plant are:
(1) Fast: minutes – hours
(2) Medium fast: hours – several hours
(3) Slow: days – months
The time scale influences the design of the control strategies. One way to express the control task is – for
example – to supply the right amount of air, or add the correct amount of chemicals, or move the sludge
to the right place to match the substrate load, at the right time.
In a sequential batch reactor the system is purposefully in a transient state (see Section 6.6). An oxidation
phase will continue until the oxidation is completed, and then a reduction phase (such as denitrification) will
take over and will finish when the reduction has been completed. A sequential batch system is therefore very
suitable for dynamic process control.
We can see that there is a wide difference between the fast and the slow time scales. Often it is possible to
separate the various control actions into different time domains. This means that in the fast time scale the
variables that change very slowly will be considered constant. For example, the dissolved oxygen
(DO) controller can act as if the biomass concentration is constant. In the slow time scale, for example

by guest
for the control of the total sludge inventory, then the DO concentration can be considered to change
instantaneously.
Sometimes the sensor time lag has to be taken into consideration. To get a DO reading takes a number
of seconds. This delay is small compared to the typical time for a DO change, a fraction of an hour.
A respirometer reading will take a longer time, typically half an hour. It is obvious that such a
measurement can be used only for slower corrective actions, in the order of hours.
It is always important to consider the dynamics when closing the loop. Sometimes the controllers are
tuned to be too “ambitious”. For example, a DO sensor may show a new DO concentration value every
10 seconds. This does not mean that the airflow rate should be changed so often, since the typical
response time in a full-scale aerator is 15–30 minutes. A change of the airflow more often than every
minute will only produce meaningless control actions and wear out the actuators. Instead a control action
every 5–12 minutes is more adequate. Then the DO measurement fed to the controller should be an
averaged value (and possibly further filtered) over the same time interval of 5–12 minutes.
10.3.3 Modelling
Modelling for control is not the same as modelling for understanding basic kinetic mechanisms.
Consequently models like the Activated Sludge Models 1,2,3 (Henze et al., 2000) or the Anaerobic
Digestion Model (Batstone et al., 2002) are not meant to be the basis for controller design. Instead, they
represent detailed descriptions of the way we understand the mechanisms of the biological processes. In
control, on the other hand, one has to identify certain key parameters that are crucial for the operation of
the plant. Such parameters can be the oxygen uptake rate (OUR), respiration rate, and reaction rates for
BOD removal, nitrification or denitrification. Redox (ORP) can reflect the progress of the reactions, in
particular in the oxygen-free denitrification process, where nitrate is reduced to nitrogen gas.
Key parameters can be calculated from simpler measurements. For example the DO concentration can be
used as a basis for the estimation of oxygen uptake rates, see further Section 10.5. On-line measurements of
ammonia nitrogen or nitrate can be further elaborated to find the adequate reaction rates. Consequently
estimation of dynamic parameters is an important part of the modelling that can form the basis for more
advanced control.
10.4 FEEDBACK CONTROL

The automation of a wastewater treatment plant is based on feedback control (also called closed loop
control). The automation has to be implemented from the bottom up, starting with all the equipment that
keeps the plant running. Pumps, valves and motors have to operate properly, the sensors must be
calibrated and maintained and the signals have to be communicated correctly to the control system.
10.4.1 Open and closed loop control

To get a proper understanding of control we have to distinguish between open loop and closed loop
(feedback) control. Many processes are manipulated without any measurements made. For example:
• Sludge scrapers in a clarifier may be operated based only on timers.
• Screens are cleaned at regular intervals.
• Air blowers are turned on and off based on timers.
These control actions are examples of open loop control. There is no measurement made.
Instead the actuator is working at regular intervals. In other words: after having made the control action

by guest
Process control 175
there is no confirmation about the result of the action. For better operation it is necessary to make
measurements.
The fundamental principle of control is feedback, Figure 10.1. The process (for example, an aerator, a
chemical dosage system, or an anaerobic reactor) is all the time subject to disturbances. This is why the
control action can never be the same. In our daily life we experience feedback all the time. Driving a car
is built on feedback: the eyes and the ears are the sensors that measure and monitor the environment,
which is changing all the time. Based on this information we make a decision in the brain (our
computer). The brain signal has to be transferred to the muscles, the hands and arms (the actuators) and
further on to the steering wheel and the speed control of the car.
Process
Actuator Measurement
Decision
Figure 10.1 Illustration of the feedback principle. The process is influenced by external disturbances,
illustrated by the arrow from above to the process.
So, a control system has to include:

• Measurements – a sensor system
• Signal treatment – take away all non-essential data (noise)
• Decision – calculation of a control action
• Actuators – transferring the computer system to “muscles”, to motors, valves, pumps etc.
• A goal for the control – this may be, for example, “to keep the flow rate constant”, “to maintain a
constant pressure”, “to keep the DO constant”, or “to ramp up the pump speed from zero to full
speed”. This goal is often expressed as the setpoint (or reference value).
The open loop examples mentioned above can be converted to feedback control:
• Sludge scrapers in a primary clarifier: if the sludge depth in the clarifier is monitored, then the
scrapers will start only when the sludge depth reaches a preset value.
• Screen cleaning: the differential pressure over the screen is measured and cleaning initiated when the
pressure reaches a preset value.
• Aeration blowers: the DO in the reactor is monitored; based on this measurement a correction of the
air flow is made.
The reason for control is the appearance of disturbances. If there were no disturbances we would not need
any control! If the effect of the disturbance is measured within the plant, such as a change in the DO
concentration, a rising sludge blanket, or a varying suspended solids concentration, the measured
information is fed back to a controller that will activate a pump, a valve, or a compressor, so that the
influence on the plant behavior is minimized. Having made the decision it has to be implemented via an
actuator, which is typically a motor, a pump, a valve or a compressor. In other words: control is about
how to operate the plant or process towards a defined goal, despite disturbances (Olsson & Newell,
1999; Olsson et al., 2005).

by guest
In control engineering feedback control is represented by a block diagram that describes the signals of
the control system, Figure 10.2. This kind of a simple control loop appears in all the local control of levels,
pressures, temperatures and flow rates. The controller has two inputs, the measurement (actual) value y and
the reference (setpoint) value uc and one output, the control signal u. In this simple case the controller uses
only the difference between the two inputs. The controller tries to make the error e as small as possible in the
best possible way. Here is the challenge: what do we mean by “best”? This is why controller tuning becomes
important. This is discussed later in this Section.
External
disturbances
uc u y
e Control Technical
Σ
law process
–1
Controller
Figure 10.2 Simple feedback control structure.
Sometimes the disturbances can be measured upstream, before they hit the plant, for example the influent
flow rate, or influent COD or ammonia-nitrogen concentrations. Then the information can be forwarded
from such a sensor to partly compensate ahead for the disturbance. Such control action is called
feedforward. For example, the aeration can be increased before a load increase hits the plant. Another
example is when the return sludge pumping can be increased to lower the sludge blanket as a preparation
of the settler for an expected increase in the hydraulic load.
10.4.2 Low level control

The “low level control” includes all control actions that are made to keep the plant running, such as the
control of local flow rates, liquid levels, air pressures or various concentrations that are not immediately
connected to the effluent quality. Most of these control actions are traditional process control loops, as
illustrated in Figure 10.2. In most of these cases the reference value (setpoint) is kept constant. The
process variable is subject to disturbances and the regulator is used to make the process variable
approach the setpoint in the best possible way.
The dominant type of controller in wastewater treatment (and in most process industries) is the PID
(proportional-integral-derivative) controller. Referring to Figure 10.2, the control error is first calculated as:
e = uc − y (10.1)
where uc is the setpoint value and y the measurement (the sensor signal). The controller consists of four
terms:

1 de
u=K· e+ e · d t + Td + u0 (10.2)
Ti dt
where K is the controller gain, Ti the integral time and Td the derivative time. The first term, the proportional
part, reacts at the present control error. The second term, the integral part, sums up all the previous control

by guest
Process control 177
errors. The third term, the derivative part, predicts future control errors by using the derivative of the control
error, or the rate of change of the error. The last term u0 is used to define the controller output for zero
control error.
It is worth noting that in commercial products the proportional gain K is often replaced by the
proportional band PB, that is, the range of error that causes a full range change of the control variable:
100
PB = (10.3)
K
Usually the derivative term is calculated using only the measured variable y instead of the error e:

1 dy
u=K· e+ e · d t − Td + u0 (10.4)
Ti dt
The reason is that the control signal u will get an unnecessary “kick” if the setpoint is suddenly changing,
which will make de/dt very large.
The controller can be used with all its parts or with only P, PI or PID control. In a large majority of cases
the controller is a PI (proportional-integral) controller (TD = 0). Figure 10.3 shows the principal behavior of
P, PI and PID controllers for a step change in the setpoint value. In particular, note that P control cannot
eliminate the steady-state offset.
1.4
1.2
PID control
1
Sensor signal
0.8 PI control
P control
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
Time
Figure 10.3 The setpoint has been changed 1 unit. The resulting sensor signal is shown for P, PI and PID
control respectively. The I-term of the PI controller will eliminate the steady state off-set. The D-term
increases the damping and decreases the overshoot.

by guest
10.4.3 Integral action

The purpose of the integral part of the PID controller is only to eliminate the stationary error (the steady state
off-set), in other words to make e = 0 after some time, see Figure 10.3. The integral action can be understood
intuitively in the following way. In steady-state all variables are constant. For the argument, assume that e
would be constant and nonzero. Then the integral in Equations 10.2 or 10.4 would increase and the u term
would increase, meaning that it is not in steady-state. So, the only way to keep u constant after a long time is
to make e = 0.
One of the most well-known possible sources of degradation of performance is the so-called integrator
windup phenomenon. This occurs when the controller output saturates (for example, a valve becomes fully
open, or a motor runs at full speed). This problem is of particular concern at the process start-up. In this case
the system operates as in the open-loop case, since the actuator is at its maximum (or minimum) limit,
independently of the process output. The control error e decreases more slowly as in the ideal case
(when there are no saturation limits) and therefore the integral term becomes large (it winds up). Thus,
even when the value of the process variable attains that of the reference signal, the controller still
saturates due to the integral term and this generally leads to large overshoots and controller settling
times. For more details, see Olsson & Newell (1999).
Most commercial computer based PID controllers today are supplied with anti-windup features. They
solve the problem either by putting an upper limit on the integral term or by turning off the integral
action when the controller output saturates. A problem with the latter approach is that the actuator or the
manipulated variable itself may saturate before the controller output.
10.4.4 Derivative action

Derivative action adds a kind of predictive capability by reacting to the rate of change of the error. In the
absence of noise, this adds a stabilising effect, which can counteract the destabilising effect of integral
action. Most of the processes in a wastewater treatment plant are relatively slow and do not have any
oscillatory behavior. This means that the measured variable will change in only one direction, when the
control actions are changed. For example, if the air flow is changed, then the DO concentration will
change in one direction, up or down. If a pump speed increases, then the flow rate will increase and a
level in a tank will increase. The motivation to use the D term is to speed up the correction. For
example, consider two cases of DO concentration changes:
(1) The DO concentration is 1 mg/l below the setpoint and is decreasing.
(2) The DO concentration is 1 mg/l below the setpoint but is increasing.
It is intuitively obvious that the air flow increase should be larger in case (1). This is why the D term can help.
The term de/dt (or −dy/dt) is positive in case (1) and negative in case (2). This will adjust the size of the air
flow rate (u).
Still it is not common that the D term is used, and the PI controllers are much more common. The reason is
that the signal y is often corrupted by noise. To take a derivative of a noisy signal cannot be recommended,
since the dy/dt value will become even more noisy. This will result in a control signal u that is not smooth at
all, and will wear out the equipment and provide a poor control instead of an improved one. So, the use of the
D term requires more elaborate controller tuning.
There are two kinds of solutions to the D term challenge. One is to first smooth the signal y in a filter. In
the simplest case the filtered signal can be obtained simply by taking an average. For example: the DO
concentration can be measured every 10 seconds, while the air flow is changed only every 10 minutes.
In this case the averaged DO concentration every 10 minutes is used by the controller.

by guest
Process control 179
The other solution is by sampling in the computer realization. The derivate is replaced by a difference
over the sampling interval. This is discussed below.
10.4.5 Computer realisation of the PID controller

When realised in a computer (which is the normal case) the continuous controller (Equation 10.2 or 10.4) is
approximated so that the integral term is replaced by a summation and the derivative term by a difference.
The process output error (Equation 10.1) is calculated at each sampling instant (k = 1,2,3,….):
e(kh) = uc (kh) − y(kh) (10.5)
It is assumed that the sampling period h is constant. Any signal variations that may take place during the
sampling interval are neglected. Alternatively, the signal is calculated as an average value, as noted above.
The time-discrete form of the PID controller (Equation 10.4) is:
u(kh) = uP (kh) + uI (kh) + uD (kh) + u0 (10.6)
where uP is the proportional part, uI the integral part and uD the derivative part. The offset signal u0 is used to
define the controller output for zero control error. According to Equation 10.4 the proportional part of the
controller is:
uP (kh) = K · e(kh) (10.7)
The integral in Equation 10.4 is approximated by finite differences and is equal to:
h
uI (kh) = uI [kh − h] + K · e(kh) (10.8)
Ti
The integral part of the digital controller forms a recursive expression, which means that its value is
updated at every sampling instance. The last term may become very small for a small h and a large Ti.
In the derivative part the derivative in (10.4) is replaced by a difference approximation:
y(kh) − y(kh − h)
uD (kh) = −K · Td (10.9)
h
It is very important how to select the sampling time h. In Section 10.3 we noted that the sampling time for
DO control should not be too short.
In a computer implementation there are additional features that have to be included in the PID controller.
An important part is anti-windup that is easily implemented in the digital form.
10.4.6 When is PID control appropriate?

It can be observed that a majority of controllers in the process industry are PI controllers. Actually PI control
is adequate for all processes where the dynamics are represented by a first order, such as level control in
single well-mixed reactors. Often the process has been designed so that it does not require tight control.
This means that even if the process has somewhat more complex dynamics, proportional control is
adequate if it is complemented with integral action to provide a zero steady-state offset.
PID control is appropriate for processes where the dominant dynamics are of second order. In a system
characterized by two response times that differ in magnitude then a D action can improve the operation.
When there is a dominant time delay a PID controller is typically not adequate, and a dead time
compensator is usually needed. With such a compensator the loop gain can be increased and the

by guest
response improved. Also, systems with oscillatory modes usually require more sophisticated control than a
PID can offer. Such systems, however, are not common in wastewater treatment systems.
10.4.7 Controller tuning

To find the optimum values of K, Ti and Td is called controller tuning. Good control performance can be
achieved with a proper choice of the controller parameters, but poor performance and even instability can
result from a poor choice of values. Controller tuning methods based on the dynamic performance have
been used for many decades and here we will give some guidance for the tuning. Controller tuning for
wastewater treatment systems is discussed in Section 17.3 of Olsson & Newell (1999) and more general
tuning methods are described in many control engineering books, for example Marlin (2000), Chapter 10.
Generally speaking we have the following qualitative considerations to make concerning the settings of
the controller parameters:
• If the gain K is too small, then the control action will become too slow and the controller is not acting
as quickly as we wish.
• If K is too large, then the controller is too sensitive to disturbances and may overreact. In some cases
the controller will cause the system to be unstable, that is the process variable will oscillate with
increasing amplitudes. This is illustrated in Figure 10.4 (left figure).
• If the integral time Ti is made short, then the regulator may overreact. It is too “ambitious” to make the
error equal to zero, and the control system may be unstable. On the other hand, if Ti is too large, then
the error may approach zero too slowly. See Figure 10.4 (right figure).
• Ti should be longer than the typical time constant of the individual process unit, so that the error e
gradually approaches zero.
• Increasing the derivative time constant Td gives a damping effect (Figure 10.3). However, much care
should be taken to avoid increasing it too much. Otherwise the controller may be too sensitive to the
rate of change and may overreact.
1.5 1.4
P control
1.2
Ti=3.5
1
1 Ti=5
Sensor signal
Sensor signal
Ti=10
0.8
0.6 Only P control
0.5
0.4
PI control
0.2
0 0
0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20
Time Time
Figure 10.4 A setpoint change of 1 unit and the resulting sensor signal. Left figure: P control with varying K. As
K increases the response time gets shorter, but the oscillatory behavior will increase. In all cases there is a
steady-state offset. Right figure: Tuning of the integral time Ti for a fixed K. With Ti = ∞ (only P control)
there is a steady state offset. As Ti decreases the steady state is reached earlier, but at a price of more
oscillatory behavior.

by guest
Process control 181
The very first task in tuning a control loop is to be clear on the objectives. What is the purpose of the
control loop? There are three main types of loops:
• Servo loops: in which the setpoint varies and the controller needs to be tuned to follow frequent
and quite rapid changes in the desired output or setpoint as quickly and as accurately as
possible. Examples are the inner loops of cascade controllers and valve positioners. Cascade
control consists of two or more feedback controllers with the outputs of the higher-level
controllers cascading to the setpoints of those below. This is discussed in more detail in Section
10.7, DO control.
• Regulator loops: in which the setpoint is usually constant and the controller is used for disturbance
rejection. The controller should try and keep the sensor signal as close as possible to a desired
output which seldom changes. This is the majority of control loops in a process plant.
• Averaging loops: these are used to damp out disturbances, usually controlling the level or pressure of
buffer tanks, which try to maintain a constant outflow at the average of the quite variable inflow.
Examples are equalization tank level control in two stage anaerobic digesters, and level control in
tanks for buffering filter backwash.
The second task is now to find the controller parameters. The general rule is to use as few parameters as
possible, for example:
• Use only P action:
for liquid levels in vessels, unless it is an averaging loop or the actual level is important,
○
○ on the inner loop of cascade loops.
• Use only I action:

○ for averaging loops,
○ where adherence to the exact desired output setting is important.
• Use D action:
○ for high order process dynamics, where the initial reaction is slow,
○ but: use extra care with D action when the measurement is noisy.
No matter what tuning technique is chosen, and there are nearly as many techniques as there are control
engineers, the general procedure involves three steps:
• Identify a simple model of the process (the form of the model).
• Estimate parameters for the model form chosen, usually by some form of stimulus-response
experiment on the process.
• Choose the controller parameters according to some rule.
The tuning of servo controllers differs slightly from that for regulatory controllers. The former are generally
tuned a little more tightly (higher gains) with the emphasis on tracking changes in setpoint as quickly as
possible. Slave controllers in cascade loops are often tuned for a servo response.
10.4.8 Manual PI tuning

Manual tuning of PID control is surprisingly common. The tuning procedure relies on the specific properties
of the three terms described above. We restrict our discussion to P and PI control because D control is a little
specialized. The systematic tuning goes as follows:
(1) Determine whether the priority for the closed-loop system is reference tracking (i.e. a setpoint that is
changing) or load disturbance (when the setpoint is constant).

by guest
(2) Determine whether steady state accuracy is essential to the control system performance.
(3) P control tuning: Introduce proportional action by increasing the value of the proportional gain K
until the speed of the response is acceptable. This is illustrated in Figure 10.4, left figure.
(4) I control tuning: If steady state accuracy is considered important, then introduce integral action into
the controller by slowly decreasing the integral time from Ti = ∞ (when no integral control takes
place) to smaller values. The Ti should be decreased so that an acceptable settle time is achieved. See
Figure 10.4.
(5) Balancing the controller terms: Increasing K may increase the overshoot. To compensate for this K
has to be decreased. A little fine tuning between K and Ti will be necessary to achieve acceptable
time responses.
The manual procedure is a trial and error process. It is hoped that steps (4) and (5) will eventually converge
to an acceptable solution. Basically, while K changes the speed of response, changing Ti alters the settling
time, with a tendency to introduce overshoot. Excessive overshoot usually needs to be avoided, but the settle
time must be reasonably short so that the desired output level is reached.
Naturally the manual tuning method is somewhat laborious, time-consuming and an ineffective use of
resources. We can do better by using some system knowledge.
10.4.9 Ziegler-Nichols frequency response method

Two methods for controller tuning were presented by Ziegler and Nichols as long ago as 1942. The
methods are still widely used, either in their original form or in some modification. One method is
based on the open loop step response and the other is based on the frequency response. It should
be emphasized that the methods usually provide a first approximation of an acceptable control
performance but most often not the best tuning. A large number of tuning rules have been devised in the
last 70 years (see Åström & Hägglund, 1995, 2006). The rules are generally based on a simple model of
the plant.
The frequency tuning method works as follows:
• Set the controller as a P controller (with Ti = ∞ and Td = 0). Increase the gain K until the process
output oscillates, being at the stability margin. Note the value of the gain Ku.
• Note the period time Pu of the oscillations.
The parameters of the P, PI, and PID controllers are calculated according to Table 10.1 and comparative
responses are shown in Figure 10.5.
Table 10.1 Ziegler-Nichols tuning of PID parameters.
Controller K Ti Td
P 0.5Ku – –
PI 0.45Ku 0.83Pu –
PID 0.6Ku 0.5Pu 0.125Pu
There are some apparent drawbacks with this Z-N method. First, not all processes can be allowed to
oscillate close to the stability limit. Second, many processes in wastewater treatment are relatively slow,
so the tuning experiment will take too long.

by guest
Process control 183
1.6
1.4
1.2
1
Sensor signal
0.8
0.6
0.4
Ziegler-Nichols tuning
0.2
0
0 2 4 6 8 10 12 14 16 18 20
Time
Figure 10.5 The sensor signal for a setpoint change of 1 unit. The controller tunings are set according to the
Ziegler-Nichols rules in Table 10.1. The P controller (broken line) has a steady-state offset. The PI controller
(dotted line) has become quite oscillatory because of the I-term, but will finally obtain a zero steady-state
offset. The PID controller (solid line) has a better damping and reaches steady state more quickly.
10.4.10 Autotuning
Today most commercial computer control systems are delivered with PID controllers including automatic
tuning, or autotuning. The identification of the process model and the tuning of the controller are made
automatically. An identification experiment is automatically performed after a specific request by the
operator and the values of the PID parameters are updated at the end of it. For this reason the overall
procedure is also called one-shot automatic tuning or tuning-on-demand.
The design of an automatic tuning procedure involves many critical issues. The choice of the
identification procedure is usually based on an open-loop step response or on a relay feedback
experiment. In the latter case the controller is simply (automatically) replaced by an on-off controller that
will cause the process output to oscillate slightly. The process model parameters are found from this
identification experiment. Finally a tuning rule is executed. At the end the tuned PID controller will
automatically take over the control.
The autotuning methods have been developed over the last two decades and show very good results in the
process industry, see Åström- & Hägglund (2006).
10.5 INSTRUMENTATION AND MONITORING

10.5.1 Instruments
To measure is to know. Due to developments over the last two decades, instrumentation is no longer the
main obstacle for ICA (Olsson & Newell, 1998, 1999; Olsson et al., 2005). The increased confidence in

by guest
instrumentation is now driven by the fact that clear definitions of performance characteristics and
standardized tests for instrumentation have become available (ISO 15839:2003). The most common
online instrumentation is shown in Table 10.2.
Table 10.2 Commonly used online instrumentation.
Physical variables Primary wastewater properties Advanced online sensors

Flow rate Dissolved oxygen (DO) Organic content
Level Turbidity (COD, TOC)
Pressure Suspended solids Ammonium
Temperature Sludge concentration Nitrite/nitrate
Conductivity Sludge blanket level Phosphate
pH
Redox (ORP)
Biogas flow rate
It is always important to screen the sensor data for high-frequency noise, missing data, values out of range
or trends from drifting sensors. The screened measurement is then a basis for monitoring or directly for
control.
To track the process operational state via the instrumentation is called monitoring. Even reliable
instrumentation can fail during operation, which can have serious consequences if the instrumentation is
used in closed loop control. Therefore real time data validation is needed before using measurements for
control purposes. If confidence in a measurement decreases, it might be possible (on a short-term basis)
to use an estimated value, but eventually control must be set to a default scheme until confidence in the
measurement has been restored.
10.5.2 Estimating the oxygen uptake rate

Let us consider one apparent example of monitoring, to estimate the oxygen uptake rate (OUR) directly from
DO concentration measurements. In order to see the relationship been the DO and the OUR we set up the DO
mass balance for an aerator zone. Using the nomenclature of Section 6.2.1 and the additional parameters
defined below, the DO mass balance can be formulated as follows:
dC Qin Qout
= Cin − Cw + KL a(Cs − Cw ) − rM (10.10)
dt V V
where
Cin = DO concentration in reactor inflow
Cw = DO concentration in reactor outflow
Qin = reactor inflow
Qout = reactor outflow rate
V = reactor volume
and other parameters are defined in Section 6.2.1.
The left hand side is the rate of change of the DO concentration in a zone that is considered to be well
mixed. The first term on the right hand side denotes the mass of DO that will enter the zone via the

by guest
Process control 185
water flow, while the second term similarly denotes the amount of DO leaving the zone. Here we have
assumed that the concentration of DO is the same in the whole zone. The third term represents the
oxygen transfer from gaseous oxygen to dissolved oxygen (see Equation 6.6). Finally, the oxygen
consumption, called rM (see Equation 6.7), is the biomass respiration rate.
Looking more closely at the size of the terms the first two terms are much smaller than the last two. Let us
now assume that the DO is controlled and constant. This means that the term dC/dt is zero. Neglecting the
first two terms we now have:
KL a(Cs − Cw ) ≈ rM (6.8)
For diffused air aeration, Figure 6.6 shows that the oxygen transfer efficiency (gO2/m3 air/m depth)
typically varies by only about 15% over the normal operating range. In other words: the OUR (or rM)
varies approximately proportionally to the air flow rate. This means that the air flow can be used as an
important indicator of the biological activity of the organisms in the aerator.
In a sophisticated treatment plant there is a huge data flow from the process. More instrumentation will
provide even more data. Unlike humans, computers are infinitely attentive and can detect abnormal patterns
in plant data. The capability of computers to extract patterns (useful information) is rarely utilized beyond
simple graphing. Information technology is not commonly used to encapsulate process knowledge, that
is knowledge about how the process works and how to best operate it. If process knowledge can be
encapsulated, then not only is it retained but the computer can also assist decision-making in plant
operation (Rosen et al., 2004). The potential of substantial operator support for diagnosis and for
corrective actions exists and has been demonstrated.
10.6 PUMPING
10.6.1 Pump characteristics
Pumping and aeration are the two most energy intensive operations in a wastewater treatment plant. It is
quite logical to consider the pumping operations from an energy point of view. Furthermore, the
consequences of pump control for the behavior of the wastewater treatment plant also need to be
considered. In particular, the clarifiers are sensitive not only to the flow rate but also to the rate of
change of the flow rate. As a consequence the influent pumping ought to be as smooth as possible in
order not to upset the clarifiers. As a consequence we look at two basic properties of pumps:
• Pump efficiency
• Variable speed control
The operating point (or duty point) of a pump is determined by the intersection of the pump (QH curve) and
system characteristics, as shown in Figure 10.6. The QH curve defines the pressure (usually measured as
“head” in meters) that the pump can produce as a function of the flow rate. The system curve (the load)
defines the pressure needed to produce a certain flow rate. The intersection between the two curves is the
duty point A that defines the operating point of the pump.
Control by “throttling” is like trying to control your car’s speed by braking with one foot while continuing
to accelerate with the other. Of course there is a waste of energy, and furthermore it causes excessive wear
and tear on the equipment.
When selecting a pump it is important to choose a pump where the duty point is located within the
high efficiency area, Figure 10.6. Here the pump losses have a minimum. In a system with variable flow

by guest
it is of course impossible to operate the pump at the maximum efficiency point all the time. However, the
pump has to be designed so that the most common flow rates are around the most efficient operating point of
the pump.
Figure 10.6 The duty point A of a pump is determined by the intersection of the QH curve and the system
curve. The efficiency for a typical centrifugal pump is also shown. If the pump speed is changed then the
QH curve will be raised for higher speeds and lowered for smaller speeds, thus changing the duty point.
Similarly, if a throttle valve is used to decrease the flow rate it means that the pressure of the system will
increase. This will raise the system curve – while the QH curve remains the same – so that the duty point is
moved upwards and to the left. This can simply illustrate why it is not energy efficient to use a throttle valve
to change the flow rate. Consequently the energy to pump has to be increased. On the other hand, if a
variable speed pump is used to decrease the flow rate, then the QH curve is lowered, while the system
curve is not changed. Consequently the pressure can be decreased and the necessary pumping energy is
not so high.
The true pump duty point will almost always differ from the theoretical one. There are inaccuracies in the
calculation of main losses, as well as error margins in the published pump performance curves. Furthermore,
the pump characteristics will change with use due to wear and the dynamic pressure will increase with age
due to corrosion or sedimentation.
Variable speed control by means of a frequency converter is the most efficient way of adjusting pump
performance exposed to variable flow requirements. This simple approach can significantly reduce
the amount of electricity that a motor-pump system uses, and also lengthen the life of equipment that is
no longer subject to the jolting on/off braking that results from throttling. Variable speed control is
also desirable from a process operation point of view. An on/off pump creates sudden hydraulic
shocks. Firstly it causes pressure shocks that will wear the equipment. Secondly, and more important,
hydraulic shocks are extremely detrimental to sedimentation processes and will cause a decreasing
process performance.

by guest
Process control 187
10.6.2 Relationship between flow rate and power

The relationship between flow rate, pump speed and power requirement for a pump are described
by the affinity laws and illustrated in Figure 10.7. The pump flow rate Q is proportional to the pump
speed n:
Q=a·n (10.11)
So if the speed is halved then the flow rate will be halved. The head H is proportional to the square of the
speed n:
H = b · n2 (10.12)
The required pump power is very sensitive to the speed:
P = g · n3 (10.13)
Equation 10.13 shows that half the flow rate requires only 1/8 of the power. In reality, the pump
efficiency is different at different speeds and flow rates, so the practical relationship between P and n is
not exactly given by the equation. Still, this relationship explains why variable speed control is superior
to throttle valve control from an energy point of view. To calculate more precisely how much power can
be saved by variable speed control one also has to take the efficiency of the frequency converter and the
motor into consideration.
Figure 10.7 The relationship between flow rate and power requirement vs. speed, according to the
affinity laws.

by guest
There are various ways to increase pumping efficiency, and variable speed control is not the only way.
For example, in one treatment plant a variable speed pump was operating satisfactorily but had too large a
capacity and operated at only 60% of its maximum flow rate. At this flow rate the efficiency was only 65%
for the motor plus pump unit. It was considered profitable to install a new pump with an efficiency of 80% at
the most common flow rate.
10.7 ESSENTIAL CONTROL LOOPS

It is not sufficient to control the physical parameters only. Other variables that are directly related to the
effluent quality also have to be controlled. Some key control loops are (Olsson et al., 2005):
• DO control with a constant or variable DO setpoint.
• Aeration phase-length control in alternating plants based on nutrient sensors.
• Nitrate recirculation control in a pre-denitrification plant (see Figure 3.21) based on nitrate and DO
measurements in the aerator and in the anoxic zone (Ingildsen, 2002).
• Advanced solids retention time control based on local measurements of effluent ammonia
concentration and estimates of nitrification capacity.
• Return sludge control based on sludge blanket measurements in the clarifiers.
• Aeration tank settling (ATS) as one way to temporarily increase the plant capacity at storm conditions
(Nielsen et al., 1996, Gernaey et al., 2004).
• The control of anaerobic processes to regulate the biogas flow, while stabilizing the process and
maximizing its productivity (Steyer, 2006).
• Chemical precipitation control based on local measurements of phosphate concentration (Ingildsen,
2002).
10.7.1 Dissolved oxygen control

DO control is of primary importance in the activated sludge process, in both continuous flow plants and in
alternating or intermittent systems. The control of aeration has attracted considerable research since the
1970s, when the DO sensors reached a level of robustness and precision suitable for feedback control.
Today, the control of the DO to a setpoint is a mature technology. However, even a correct DO control
cannot compensate for insufficient air production at extremely high loads.
With the development of nutrient sensors in recent years the primary focus of aeration control has been
the on-line adjustment of the level of oxygen supply. For a continuous system, the problem is primarily the
on-line determination of appropriate DO setpoints (Ingildsen, 2002).
The importance of the DO profile in space and aeration control in plug-flow type reactors was
recognized early (Olsson-Andrews, 1978). For intermittently aerated systems, the control of the aerobic
phase-length has been studied intensively (Nielsen, 1983; Thornberg et al., 1993; Sorensen et al., 1994;
Nielsen-Ønnerth, 1995 and Potter et al., 1996).
Adequate design of the aeration system is of course the pre-requisite for energy efficient aeration. There
are two principally important parts, the compressor and the diffuser system. The compressor has to allow for
variable air flow rates, which is crucial for any control of the DO.
Many activated sludge systems are either of a plug flow type or can be described as a series of well mixed
tanks (see Section 6.1). The oxygen demand is much higher at the head end of the aerator than towards the
outlet. This means that there has to be a higher airflow early in the aerator than in the outlet area, where the
biological growth reaction should have been completed. This will require an aeration system that is tapered
with a higher capacity towards the head end.

by guest
Process control 189
The aeration system ideally should supply diffused air so that the DO is constant not only along the
aerator length but also constant in time. This requires two features:
• There have to be separate air valves along the aerator to supply the appropriate amount of air along
the aerator.
• The air valves have to be controlled so that the air flow can be varied with time.
Normally the microorganisms require around 1–3 mg/l of DO. Too many systems are still designed so that
there is insufficient air supply in the head end and an excess air supply towards the outlet. This results in
energy waste. In a nitrifying aerator there are two types of microorganisms that compete for the air (see
Section 3.5). The organic matter is consumed mainly by the heterotrophic organisms while the
ammonium nitrogen is oxidized to nitrite and nitrate via autotrophic organisms. The DO requirement is
larger for the autotrophic organisms (Figure 3.15). They are slower growing than the heterotrophic ones,
which means that ammonium is removed more slowly than the organic matter. This is illustrated in
Figure 10.8. Just before the middle of the reactor the DO concentration rises quickly. This is an indicator
that the organic matter has been consumed. The ammonium is still being oxidized and continues to
consume oxygen along the reactor. At the outlet all ammonium has been oxidized to nitrate.
Figure 10.8 The DO profile in a plug-flow reactor for organic removal combined with nitrification. The airflow
distribution along the aerator is uniform. The upper curve shows the DO profile (mg/l) and the lower curve
shows how the ammonium concentration decreases along the aerator.
We now assume that the DO is measured in one point in the aerator, typically in one of the aerated zones.
The air flow to this zone can be controlled independently of the other zones. Therefore the DO controller
should be able to maintain the DO concentration at a given setpoint value in the zone.
The DO concentration is compared with the DO setpoint and the DO controller (the master) will calculate
the necessary airflow change required to change the DO concentration towards the desired value,
Figure 10.9.

by guest
DO setpoint
DO
controller
DO sensor
Air flow
setpoint
Air flow
controller DO
conc
Aerator
Air
flow
Air supply
Figure 10.9 Structure of a standard cascade DO control. The master controller delivers the setpoint (air flow
setpoint) to the slave controller.
Note that the DO controller does not directly manipulate the air valve. Instead the desired airflow is
provided as a setpoint to an inner controller (the slave) controlling the air flow rate. This controller
receives the airflow rate measurement and compares it with the desired airflow. This difference will then
make the actuator (a compressor or a valve) change the airflow to the correct value. The loop is called a
cascaded control loop and is the standard configuration in this kind of system.
There are two fundamental reasons why the DO controller (called the master) is not coupled directly to
the valve:
(1) The first reason has to do with the valve characteristics. Usually the valves are nonlinear, as in a
butterfly valve. A 10% change of the valve signal will produce significantly different responses
if the valve is almost closed, in the mid-range or almost fully open. This means that a desired
airflow change has to result in widely different valve movements if the valve is almost closed or
if it is almost fully opened. If the airflow rate is measured, then the flow controller is able to
produce just the flow rate that is demanded. Having the closed-loop slave controller in place
means that the master controller will see a linear behavior of the airflow system. Without a
cascade structure the DO controller would need to have a different tuning for small loads and for
high loads because of the nonlinearity of the valve.
(2) The second reason has to do with the commissioning of the control system. The slave controller is
always tuned first with the master controller open loop (set in manual). The slave controller is
usually a simply P controller, since the steady state off-set is not critical for the slave controller.
Then one can ensure that the response of the airflow system is adequate. Now the master
controller will see a more ideal (and linear) behavior of the air flow system. The master
controller is then put into automatic mode and subsequently tuned. Having the slave controller
makes tuning of the master controller, usually a PI controller, much easier. The integral part
ensures that the DO concentration has no steady-state offset.

by guest
Process control 191
Just as in the equivalent human situation, the slave must be capable of reacting faster than the master to
ensure stability. Unless the master controller has a very slow response, the slave controller will generally
be a simple P controller. The master controller is usually of a PI type, Equation (10.4), and will remove
any offset which may occur.
Certainly, more complex controllers have been tried and reported in the literature. In 1984 experiments
were performed on the use of a more complex self-tuning regulator of higher order in a full-scale process
(Olsson et al., 1985). Surprisingly the controller parameters converged towards a PI-controller
performance. Despite this long-standing finding, attempts are still published to control the DO
concentration with advanced control algorithms, providing little additional benefit in the actual
implementation (Olsson, 2012a).
Figure 10.10 demonstrates DO control for a medium sized municipal plant.
Figure 10.10 Example of DO control during 20 days of operation of a municipal plant. The DO is controlled
around its setpoint value of 3 mg/l (middle curve). The upper curve shows the air valve opening and is an
indicator of the oxygen uptake rate (Section 10.5.2). The lower curve shows the influent flow rate to the plant.
The real challenge today is to decide what oxygen concentration is required for the process, given
disturbances and uncertainties. There are several aspects of this, one being the legislative framework
which the plant has to comply with. Internationally, there are different levels of total nitrogen limits. If

by guest
there is a limit of only the total nitrogen but not on the ammonia concentration, then the best strategy – from
an energy point of view – is to maximize the ammonia concentration while satisfying the total
nitrogen concentration.
There are also different time perspectives on the effluent limit compliance. For example, if the effluent
criteria are defined as averages over the day or over the week, then it is not crucial to keep the DO very tight
around the setpoint. On the other hand, if the effluent criteria have to be met at any time, then the control has
to be much more strict.
In some countries there are effluent fees connected to the plant performance. With such a criterion it is
apparent that operational costs can be compared with effluent costs, and the best strategy can be sought.
10.7.2 Ammonia-Based DO control

In a nitrifying aerator the organic removal is completed much before the ammonia-N removal, Figure 10.8.
Therefore it is of interest to verify that the ammonia-N concentration is sufficiently low towards the outlet.
It is most often profitable to measure the ammonia-N online towards the outlet, as indicated in Figure 10.11.
If the measured ammonia-N is smaller than the setpoint value of the ammonia-N then the DO setpoint in
the last zone of the aerator can be decreased. Consequently, if the ammonia-N is too large, then the DO
setpoint should be increased (up to a certain limit). Still this may not be sufficient, if the load is
extremely large. The DO setpoint correction should not be made more often than on an hourly basis.
Usually a simple P controller is sufficient. Significant energy savings can be achieved. Compared to a
fixed DO setpoint savings of 10–15% have been reported (e.g. Ingildsen, 2002).
Nitrification
40
35
30
25
Concentrations
20
15
10
-5
0 1 2 3 4 5 6 7 8 9 10
time (hr)
Figure 10.11 Ammonia-N concentration along the length of the aerator (simulated). The upper curve
indicates that the nitrification has to been completed. The DO setpoint should be increased. The middle
curve is an ideal ammonia-N removal and reaches a low value just at the outlet. The lower curve is too
“ambitious” and in the last part of the aerator no more nitrification takes place. Then the DO setpoint can be
lowered to slow down the nitrification rate and save energy.

by guest
Process control 193
10.7.3 Recycle flow controls

A number of flows in the biological step of a wastewater treatment plant need to be controlled. These include
the nitrate recirculation flow control, external carbon dosage control for enhancing denitrification or
phosphorus removal, SRT (sludge retention time) or sludge wasting control, sludge recycling flow
control. Other flows include the control of the wastewater flow through step feed, wastewater storage,
and aeration tank settling. Overviews of all these control actions are presented in Chapter 6 of Olsson
et al. (2005). Most of these variables may be operated relatively independently of each other. However,
external carbon dosage control should preferably be controlled in combination with nitrate recirculation.
10.7.4 Chemical precipitation

Phosphorus removal using chemicals has been discussed in Section 3.7. For tertiary chemical phosphorus
trimming downstream of a biological phosphorus removal process, the traditional control is a flow
proportional controller, where the dosage is proportional to the influent flow rate. It is often adjusted for
rain events and the charges of the particles and the performance is checked by grab samples or 24-hour
samples. Figure 10.12 illustrates a typical phosphate concentration variation in the influent to a tertiary
chemical process. It is not a trivial exercise to obtain an average effluent phosphate concentration of
0.3–0.5 mg/L PO4-P with such a plain strategy; usually an excess dosage is used in order to satisfy the
effluent requirement.
Figure 10.12 Example of variation in PO4 concentration into a tertiary process (Ingildsen, 2002).
However, using a feedback control strategy means that the dosage is controlled based on online
measurements of the phosphate concentration in the flocculation chamber. The controller can be a simple
PI controller. The feedback control makes it possible to reduce the dosage considerably. At the same
time there is a confidence that the effluent concentration is acceptable regardless of events taking place
upstream in the process. Figure 10.13 demonstrates that it is possible to control the dosage to yield a
quite constant effluent phosphate concentration, close to the target of 0.5 mg/L, in spite of variations in

by guest
the influent phosphate concentration. The standard deviation of the phosphate concentration from the
flocculation chamber in Figure 10.13 is 0.03 mg/L PO4-P. Furthermore, by simply adjusting the setpoint
of the controller to 0.3 mg/L the controller could readily adjust the dosage with a similar performance.
Figure 10.13 Effluent phosphate from the flocculation chamber using sensor feedback (Ingildsen, 2002).
10.7.5 Anaerobic digestion process control

While anaerobic digestion has not been discussed elsewhere in this book, a note is included here to provide a
broader perspective on the automatic control of biological processes.
Anaerobic digestion (AD) processes are usually more sensitive to process disturbances and subject to
more failures than aerobic processes. The control of biogas production can be quite unstable if the
controllers are poorly tuned. Indeed, one of the main reasons for imbalances that will eventually lead to
process failure is due to the inhibitory effects of the volatile fatty acids (VFA) at high concentrations.
Since VFA accumulation induces a pH drop and since pH is readily available information, it is
widely used for monitoring and control of AD processes. In AD operation the gas production,
gas quality, volatile acids content, temperature, sludge feed rate and alkalinity could be measured in the
early 1970s. It was also recognized early that a measure of organism activity was required. The
volatile-acids-to-alkalinity ratio related to gas production was considered such a measure. A
comprehensive review of instrumentation in anaerobic treatment can be found in Spanjers & van Lier
(2006).
The key message for better control is that it is not sufficient to base the feed rate control on only the
resulting biogas production. The gas measurements are simply too late for a good control. Therefore the
state of the reactor has to be monitored, such as pH, alkalinity, and temperature. Earlier information will
provide more reliable information for a better feed rate control.
In recent years there has been a lot of research and development on monitoring and control of the AD
process. J.P. Steyer has written an excellent overview of control of AD processes in Chapter 7 of Olsson
et al. (2005) and in Steyer et al. (1999, 2002, 2006). In particular Steyer describes an innovative work
on interval-based control. In our research (Liu et al., 2004, 2006) we found that the controller itself and

by guest
Process control 195
the control structure can be quite unsophisticated in the sense that a variable gain controller can achieve great
improvement of the operation. The system efficiency is maintained while process stability is ensured.
10.8 MINIMISING OPERATING COST

10.8.1 Approaches
In each one of the unit processes the control scheme may be more elaborate. All of this depends on suitable
sensors, instruments and actuators. The ultimate goal at this level is to optimize the unit process operation.
Control and automation can reduce the electrical energy requirement. A comprehensive overview of the
water-energy nexus is given in the book Olsson (2012b). Here we emphasise the importance of DO control
and proper operation of pumps.
The energy cost for aeration can be reduced, first by DO control with a fixed DO setpoint. By allowing the
setpoint to vary, depending on the ammonia measurement, the cost can be further decreased.
A further possibility for energy saving is to allow the air pressure setpoint to be variable. This means that
there is room for some energy savings by minimizing the air pressure at all times. This is accomplished by
the so called “most open valve” strategy. Assume that there are two or more parallel aerators. The DO
control uses the air flow valves to supply the aerator zones with the adequate air flow. The valves are
most probably not fully opened. By gradually reducing the air pressure (via the compressor) all the air
valves have to open up more. This can continue until the most open valve is almost fully opened. In this
way the pressure drop over the air valves is minimized and some energy can be saved.
In Section 10.7 we have also seen how the cost for chemicals can be decreased. In the example shown the
chemical costs were reduced by more than 35%, so the phosphate analyser could be paid for within a
few months.
The cost is also related to the personnel. The ICA development has meant that many plants are today
satisfactorily operated un-manned during evenings, nights and weekends.
10.8.2 The human factor

Often it is noted that the wastewater treatment performance may be poorer than plant design performance.
Sometimes the plant operators are constrained by treatment technologies. Sometimes they do not apply
sufficient control because they have inadequate training. Also, many times there are insufficient
incentives for the operator. These issues are addressed in Rieger & Olsson (2012). Olsson & Newell
(1999), argue that it is the “human factor” in successful operations that is often neglected, though it
could possibly create more problems than the technology.
The lack of incentives or misleading incentives for plant operating people may decrease the efficiency
and cause suboptimal effluent quality. Just one example is mentioned here: since the effluent
concentrations for ammonia and phosphorus are taken into account, but not total nitrogen concentration
or energy and chemical consumption, the operator is often over-aerating and overdosing P-precipitates to
prevent violation of effluent limits.
As stated in Rieger & Olsson (2012), “if the implementation of the new control system is successful, the
plant manager will earn praise for reduced costs. But the operators will have more work to do regarding
sensor maintenance and increased demands from the new and more complex actuators. If the system
fails, the chief operator will be held accountable because he is responsible for the effluent limits. This
‘blame culture’ is a common reason that WWTP operation is kept below possible best performance.”
The lesson is that technology alone is not sufficient. Without effective incentives, the sustained
application of a control concept or any other optimization measure might be jeopardized.

by guest
10.9 CONCLUDING REMARKS

Disturbances are everywhere and are the main reason for control. Uncertainty in the process or in its
environment makes automation both an opportunity and a great challenge. Application of automation in
water operations can be said to have two primary functions: information acquisition and process control.
For the former function, the level of automation is relatively high. Often many thousands of variables are
gathered on-line in the SCADA systems of treatment plants and more or less sophisticated data analyses
are standard components of the treatment operation. The latter function, process control, is less
developed and often limited to a few unit process control loops. Future development will be exploiting
the enormous capacity of data distribution that is possible today. Many SCADA systems are also
applying the technology from the Internet, which gives an almost unlimited potential for remote data
evaluation and decision. The distributed control room is already here. There is a limit to how much
expertise a treatment plant can afford. However, given that plant data can be made available anywhere it
is possible to utilize specialist competencies wherever they are located.
The increasing incorporation of ICA in water treatment operation is not only driven by the development
of automation technology, including impressive technical development of instrumentation and computer
technology, modelling and control algorithms. It is motivated by economy and environmental obligations
and turns out to be a necessary and worthwhile investment. It is already proven in several installations
that ICA investments have paid off quickly and we will see that ICA will become an increasing part of
the total investments.

by guest
Chapter 11
Evolutionary operation
This chapter completes the cycle begun with the introduction to EVOP in Chapter 1. Chapter 2 covered plant
influent and effluent characteristics, Chapter 3 basic process fundamentals, Chapters 4–8 the characteristics
of the main individual processes found in BNR plants, Chapter 9 the main characteristics of the overall plant
and Chapter 10 the automatic control systems. Chapter 11 discusses the non-automatic control system – the
plant operator’s tuning effort based on the philosophy of ongoing learning and improvement.
11.1 LEARNING
Conscious effort applied to any area of human activity results in improvement in performance.1 This goes
without saying in fields such as sport and music. However, it is often not objectively pursued in other fields
such as sewage treatment. Performance improvements have been demonstrated in construction,
manufacturing and chemical processing operations and the curves of gradually increasing production
rate, improving performance or decreasing cost are called learning curves or experience curves. Two
disparate examples are shown in Figure 11.1.
Peck (1984) provides examples from the construction industry and highlights three key factors emerging
from his experiences:
(1) Deliberate effort is needed. Where complex activities are involved, those close to the activity
are frequently unable to appreciate or remove the blocks which exist in the operation.
Budgetary over-runs are met by increasingly hard work, by way of long hours and physical
effort doing more or less the same thing. However, relatively minor modifications can
produce dramatic benefits. Monitoring emphasis should be shifted from the absolute progress
figures to rate of improvement. Ongoing improvement should be a normal part of everyday
operation.
(2) You have to believe that improvement is truly possible before you will seriously look for it. The
potential for improvement is obvious to football teams not at the top of the table, but is not so
obvious in everyday work activities. However, improvement potential always exists.
1
“There is no such thing as failure, only feedback.” (Michael J Gelb)
“Practice makes perfect.” (Anon)

by guest
(3) Delays or interruptions to the improvement process result in “forgetting”. That is, progress is
inhibited or regresses in the face of external delays, interruptions, unnecessary changes to work
practices, changes of personnel, reorganisations, industrial disputes and so on.
(a) Two teams in the Australian Football League competition
LEARNING CURVES - 5 YEARS OF AFL HISTORY Essendon Rank Brisbane Rank

T = idealised time constant (time for 63% of ultimate improvement)
1
Essendon
T =15 rounds
4
Team Rank
10
Brisbane
13
T =25 rounds
16
1997 1998 1999 2000 2001
(b) Cost of limestone production in the USA (after Hirschmann, 1964)

10
Price per Ton ($ constant)
1938
1945
1952
1929
1971
1
1 10 100
Total Industry Accumulated Experience (billion tons)
Figure 11.1 Two disparate learning curve examples. The time constant referred to in (a) is discussed in
the text.
11.2 APPLICATION TO SEWAGE TREATMENT PLANTS

11.2.1 Plant improvement strategies
Historically, more effort has been applied to understanding and applying plant improvement strategies in the
process industries than in the water industry. Figure 11.2(a) shows the traditional water industry approach
where major plant expansions are implemented at long intervals when the plant load reaches the nominal
(design) capacity. Between expansions, plant capabilities with respect to both capacity and performance
are assumed to remain constant.
Figure 11.2(b) shows the approach more common to the chemical process industries (Brennan &
Stephens, 1985a, 1985b). Major expansions occur at longer time intervals because ongoing increases in

by guest
Evolutionary operation 199
actual plant performance capability and capacity occur through a process sometimes called performance and
capacity creep. Performance and capacity enhancement generally occurs in two phases: a short-term startup
phase immediately following the commissioning of new facilities and a long-term performance and capacity
creep phase. Both phases can be modelled using the learning curve approach in which performance and
capacity are correlated with cumulative production experience.
(a) (b)
Plant Plant
Load Load
Design
Capacity Capacity
Creep
Expansions
Major
Initial Load Load
Initial
Capacity
Expansions
Major
Capacity
1
2
Time Time
Figure 11.2 Expansion of plant capacity: (a) traditional water industry perspective, (b) process
industry perspective.
11.2.2 Learning curves

For process plants, two logarithmic learning curve equations are commonly applied (Brennan & Stephens,
1985a).
For capacity increase:
QU, ultimate capacity

Plant Capacity
QI, initial capacity
Time
(Q − QI )/(QU − QI ) = [1 − exp ( − t/t)] = fraction of ultimate increase achieved (11.1)

or
Q = QU − (QU − QI ) exp ( − t/t) (11.2)
where
Q = capacity at time, t
QI = initial capacity

by guest
QU = ultimate capacity
t = time
τ = time constant
= time to achieve 63% of ultimate increase in capacity (see Figure 6.4)
The time required to achieve 95% of the ultimate increase is 3τ.
For performance improvement:
CI, initial quality

Effluent Quality
CU, ultimate quality
Time
(CI − C)/(CI − CU ) = [1 − exp (−t/t)] = fraction of ultimate improvement achieved (11.3)

or
C = CU + (CI − CU ) exp (−t/t) (11.4)
where
C = effluent concentration at time, t

CI = initial effluent concentration
CU = ultimate effluent concentration
t = time
τ = time constant
= time to achieve 63% of ultimate increase in capacity (see Figure 6.4)
11.2.3 Startup phase

Experience
Experience demonstrates that the design capacity and performance of a new process plant are only achieved
through a period of intensive effort during which equipment problems are resolved and process
characteristics explored.
Figures 11.3–11.5 show examples from the author’s experiences with the startup of new or upgraded
wastewater treatment plants incorporating relatively new technology. It can be seen that the desired
results are not necessarily achieved smoothly and that learning curves are idealisations of the real world.
However, the idealised learning curves depicted do reflect the intensive improvement efforts of
the operators.

by guest
(a) Effluent Nitrogen

NH3-N+NOx-N Learning Model
10
CI = 10 mg/L
8 CU = 1.3 mg/L
Effluent N (mg/L)
τ = 45 days
6
50%ile TN Limit
4
0
23-Apr 22-Jun 21-Aug 20-Oct 19-Dec 17-Feb 18-Apr 17-Jun
30d Periods, 1997-98
(b) Effluent Phosphorus PO4-P Learning Model

15
CI = 12 mg/L
Effluent PO4-P (mg/L)
CU = 2 mg/L
10 τ = 120 days
50%ile TP Limit
0
23-Apr 22-Jun 21-Aug 20-Oct 19-Dec 17-Feb 18-Apr 17-Jun
Figure 11.3 Startup phase of the new 30 kEP Thorneside BNR plant – an oxidation ditch with online activated
primary tank (APT) prefermenter and anaerobic pre-reactor. The rate of improvement in bio-P removal was
governed mainly by tuning of the APT (Hartley et al., 1999).
TN TN Learning Curve
Effluent Concentration (mg/L)
25
CI = 20 mg/L
TN CU = 3 mg/L
20
τ = 90 days
15
10
5
TP
0
21-Jun 21-Jul 20-Aug 19-Sep 19-Oct 18-Nov 18-Dec 17-Jan 16-Feb
Figure 11.4 Startup phase of an expansion and BNR upgrade to the 56 kEP B Plant at Maroochydore STP –
Westbank process format with primary treatment, side-stream static prefermentation and anaerobic sludge
digestion. Particular issues were BNR influent nutrient ratios, nitrification rate (SRT, pH), process COD
distribution and foaming/wasting.

by guest
(a) Effluent Nitrogen TN Learning Model

150
CI = 100 mg/L
Effluent Total N (mg/L)

CU = 25 mg/L
τ = 110 days
100
50
Goal
0
1-Sep-99 31-Oct-99 30-Dec-99 28-Feb-00 28-Apr-00 27-Jun-00 26-Aug-00
30d Periods, 1999-2000
(b) Effluent Suspended Solids SS Learning Model

150
CI = 80 mg/L
CU = 10 mg/L
Effluent SS (mg/L)
τ = 140 days
100
50
Goal
0
1-Sep-99 31-Oct-99 30-Dec-99 28-Feb-00 28-Apr-00 27-Jun-00 26-Aug-00
30d Periods, 1999-2000
Figure 11.5 Startup phase of a BNR upgrade to a 300 kEP abattoir WWTP with median TN/SS goals of
25/10 mg/L – a combined anaerobic pond/SBR process format. Particular problems were encountered
with aeration capacity, wasting and sludge dewatering capacity, control of SBR feed COD/N ratio and
SBR decanter solids carryover. Enhancements implemented over the startup period included increased
aeration & sludge dewatering capacity, improved load balancing, improved control interfaces, decanter
modifications and operating cycle modifications.
For the plant startup phase, data from the process industries give a τ value of 8 weeks for small simple
plants up to 1 year for large complex plants (Brennan & Stephens, 1985a). The startup data for the
wastewater treatment plant experience depicted in Figures 11.3 to 11.5 are summarised in Table 11.1 and
give τ values ranging from 20d (for simple alum dosing) to 140d for SS and N removal from an abattoir
wastewater. These values are generally consistent with the reported process industry experience.
Table 11.1 Learning curve parameters for startup phase.
Plant Ave Parameter Ci/ Cu Time Time to

SRT (d) Constant Achieve 95%ile
(τ) Performance (3τ)
Days x SRT Days x SRT

Thorneside 25 TN 7.7 45 1.8 135 5.4
TP (biological) 6.0 120 4.8 360 14
Maroochydore 17 TN 6.7 90 5.3 270 16
(Continued )

by guest
Table 11.1 Learning curve parameters for startup phase (Continued).
Plant Ave Parameter Ci/ Cu Time Time to

SRT (d) Constant Achieve 95%ile
(τ) Performance (3τ)
Days x SRT Days x SRT

Abattoir 18 SS 12 140 7.8 420 23
TN 4.4 140 7.8 420 23
TP (alum dosing)1 3.8 20 1.1 60 3.3
1
See Chapter 7, Fig. 7.7.
The data in Table 11.1 suggest the following rules of thumb:

Time constant (time to achieve 63% of ultimate startup performance): 5 × SRT
Startup optimisation time (time to achieve 95% of full startup performance): 3 × time constant 15 × SRT
Contractual startups
Plant design and construction contracts involving performance proving at startup should provide for the
extended time commonly required to shake down both equipment and process reliability in new sewage
treatment facilities. The theoretical effects of the available startup time on the required design effluent
quality and the resulting plant cost are illustrated in Figure 11.6. Plant cost has been assumed inversely
proportional to effluent nitrogen concentration raised to the power 0.59 (Hartley, 1998).
(a)
6
Contract Deadline
Effluent Concentration, C
CI, initial quality

C = Cu + (Ci-Cu)exp(-t/τ)
5
CI/CU = 6; τ = 2.5 months
(normalised)
3 Moving monthly performance range with good

operation (10-90%iles, geom SD 1.3)
2 CC, contract quality
1
CU, ultimate quality
0
0 2 4 6 8 10 12
Months from Startup, t
(b) 3
CI/CU = 6; τ = 2.5 months
Normalised Values
Plant Capital Cost
CU/CC
0
0 2 4 6 8 10 12
Duration of Contractual Startup Phase
Figure 11.6 Effect of startup duration on design effluent quality and cost, arbitrarily based on effluent TN cost
impact and using representative startup parameters from Table 11.1. CC is the contractual effluent quality
required (drawn in (a) for an arbitrary six month contractual startup period), CI is the initial effluent quality
achieved and CU is the ultimately achievable effluent quality for which the plant has to be designed to
enable CC to be met by the end of the contractual startup phase. Figure (a) also shows the potential
variability of the moving monthly effluent quality.

by guest
It is clear that contractual performance proving requirements can potentially have a significant effect on
project cost because of the design conservatism required to meet performance guarantees within a limited
time frame. Alternatively, the project completion time may have to be extended when it becomes evident
during the optimisation phase that the design performance will not be met by the deadline.
Additional investment is sometimes required during the startup phase for new sewage treatment plants
to enable full design capacity and performance to be demonstrated, especially when relatively new
technology is involved. The author analysed eight plant projects in this category which were
implemented between 1984 and 1994 to assess the magnitude of this additional startup investment.
Types of projects included greenfield and brownfield, municipal and industrial, full-scale and pilot-scale.
The magnitudes of the additional investment ranged from 1–31% of the initial plant construction cost.
The geometric mean was 4% with a geometric standard deviation of 3.7 (i.e. the 16–84 percentiles were
4/3.7–4 × 3.7 or 1–15%).
11.3 LONG-TERM IMPROVEMENT

11.3.1 Potential
To illustrate the degree of long term improvement potentially achievable in capacity or performance,
Australian plants in the chemical process industry are reported to have achieved typical capacity
increases between major expansions of around 50% over periods of 5–10 years (QU/QI = 1.5, τ = 2–3
years). An example from Brennan and Stephens (1985a) is shown in Figure 11.7.
20
Stage 3 cell bank
15
Relative Capacity
10
Stage 2 cell bank
Stage 1 cell bank
5
Process intensification
& learning phases
0
0 10 20 30 40 50
Years
Figure 11.7 History of maximum daily capacity of chlorine manufacturing plant (after Brennan & Stephens,
1985a).
This approach has not commonly been applied to sewage treatment plants but some degree of both
performance and capacity creep does occur. It is certain, however, that significant potential gains are
waiting to be exploited and there has been increasing interest in the process (Parker et al., 1999).
The overall capacity and performance creep process is illustrated in Figure 11.8. Capacity enhancement
draws on two main factors: (a) design and construction margins (deliberate or otherwise), which result in a

by guest
plant in which the various components do not all have the same capacity but a distribution of capacities, and
(b) advances in technology and process understanding, facilitating process intensification. A particular
cause of capacity distribution in sewage treatment plants arises from uncertain and variable quality of the
sewage. It is inevitable that the actual sewage received by a new plant will differ in some respects from
the design quality, resulting in the performance characteristics of various plant components differing
from design expectations.
Experience
DESIGN Own research
& CONSTRUCTION LEARNING Technological
MARGINS advances
Components
Whole plant Plant capacity distribution
(Different component margins)
I = kiQ0.6
Accumulating operational
Series of
know-how TUNING DEBOTTLENECKING
Minor Investments
EVOP program
Plant modifications
Equipment additions/
replacements
Process intensification
2.5
2
Process improvement from normal CAPACITY &
technical effort plus EVOP
PERFORMANCE LEARNING ACTUAL
Capacity
Capacity
CURVE CAPACITY
1.5
CREEP
Process improvement from 1
normal technical effort DESIGN
o
Design Life
Deferral of capital investment for major
Actual Life
LOAD CAPACITY
expansion or upgrade 0.5

o Enhanced effluent quality
o Savings in O&M per unit load - energy, 0
0 1 2 3 4 5 0 5 10 15 20
consumables, personnel
Years Years
Figure 11.8 Capacity and performance creep. Minor investment cost is discussed later in this chapter.
Capacity and performance are improved through a combination of ongoing operational tuning and
periodic minor investments in debottlenecking. The extent and rate of improvement achieved will
depend on the effort expended.
11.3.2 Tuning
Tuning improves performance and enhances plant capacity by testing and pushing out the actual plant
capacity envelope. This provides ongoing improvement through application of accumulating operational
know-how on the plant.
Tuning can have several goals:
(1) Improve performance – effluent quality, biosolids quality, odour emission, greenhouse emissions
(2) Increase capacity – liquid stream, solids stream

by guest
(3) Reduce consumption – energy, chemicals

(4) Reduce operating cost
A valuable technique for maximising the rate of learning is a method of routine operation called evolutionary
operation, or EVOP for short (Box & Draper, 1969). This is a structured method for deliberate learning
devised for the process industries. A continuous investigative routine becomes the basic mode of
operation and replaces normal static operation. EVOP is based on the principle that it is inefficient to run
a plant in such a way that only a product is produced, rather, a process should be operated so as to
produce not only a product but also information on how to improve the product or increase the rate of
production (plant capacity).
This method of operation involves the operators in an ongoing process of continuous improvement
through planned variation of operating parameters and evaluation of results. Understanding of the plant’s
behaviour improves and the plant is continually moved towards the optimal operating regime.
Formal statistical approaches can be applied in chemical process plants where close control of
throughputs and reaction conditions can be exercised. However, the EVOP technique has to be adapted
for use with sewage treatment processes, which have special characteristics:
• Uncontrolled & variable loading
• Uncontrolled & variable temperature
• Variable outputs
• Long process response times
• Usually operating at a load significantly less than capacity
Optimisation of a sewage treatment plant is perhaps more akin to that of a manufacturing plant as described
in an absorbing novel by Goldratt and Cox (2004). The implications for application of EVOP to sewage
treatment plants include the following:
• Use EVOP as an operating philosophy rather than a statistically rigorous procedure.
• Start EVOP early in the life of a plant, ideally as soon as the plant is commissioned.
• Document procedures and results carefully to minimise memory loss with time and
personnel changes.
• Substitute test run time (multiple of process response time) for number of replicate cycles.
• Include load as one of the variables.
• If using a rigorous statistical approach, use a three variable scheme to limit the overall time required.
• Recognise that variability of output is as important as the average.
• Provide close professional support to the operating personnel.
The statistical sophistication of the operating trials may vary. The fundamental philosophy of EVOP,
however, is that the plant should always be operated and monitored so as to deliberately improve the
operators’ understanding of its behavioural characteristics and allow its performance and capacity to be
continually improved or optimised – a structured process of ongoing improvement. Examples of the
author’s approach in sewage treatment plants are discussed throughout this book.
11.3.3 Stress testing

The capacities of various plant components can be assessed while the plant is operating below its full
capacity by conducting stress tests (Parker et al., 1999). In a stress test the load on a particular plant
component is gradually increased, or the operating parameters gradually adjusted, to quantify the process
response. For example, secondary clarifiers can be progressively taken off-line to assess the response of

by guest
the remaining units to increasing load (see Section 7.5), or the process SRT can be deliberately shortened to
assess the minimum practicable value.
11.3.4 Debottlenecking
As discussed in Chapter 9, the bottleneck principle states that the plant component with the least capacity
sets the capacity of the whole plant. Debottlenecking therefore increases plant capacity by increasing the
capacity of the limiting component. The capacity of the bottleneck is increased by equipment
addition/replacement/modification, process intensification or other plant modification.
Although tuning is usually thought of in terms of performance and debottlenecking in terms of capacity,
both can also be thought of in the opposite terms. An improvement in performance at a given capacity can
alternatively yield an increase in capacity for the same performance, and vice-versa.
For debottlenecking to be applied, the distribution of capacities for the plant components must be
determined. These capacities are not absolute quantities but apply only in the context of the overall
plant design, the sewage quality feeding the plant and the effluent quality standards which the plant
must meet. Process units generally do not have fixed individual process capacities but contribute to
the overall integral plant capacity. In principle, the procedure for identifying and quantifying
bottlenecks is to gradually increase the plant load (on paper or in practice) until a failure occurs – that
is, one of the plant’s performance objectives is breached. The cause of this failure is identified. This
bottleneck is then assumed to be overcome and the increase in load is continued until a further failure
occurs. And so on.
This is easily understandable in the case of a hydraulic limitation, for example. However, sometimes
there is no single bottleneck and the failure could be addressed in alternative ways (of varying cost
and practicality). Consider a conventional activated sludge plant. As the load is increased, the COD of
the primary effluent increases, passing more load to the secondary system. The MLSS concentration in
the biological reactors therefore increases, passing more load to the secondary clarifiers. Eventually the
clarifiers will fail with gross loss of biomass in the effluent. What is the bottleneck? If the return
activated sludge rate is the limitation, this would generally be regarded as the bottleneck. However, an
increase in overall process capacity could alternatively be achieved by upgrading the primary clarifiers,
or the bioreactors, or the secondary clarifiers.
Therefore, as each failure condition is reached, alternative methods of overcoming it are identified and
compared. This will be an iterative process because alternative debottlenecking actions may affect the next
bottleneck in different ways. Plant trials can be run to define actual capacities compared with theoretical
values. In this way a debottlenecking program can be developed which correlates capacity increase
with cost.
Plant capacity can be displayed using capacity and performance envelopes or diagrams (operating space
diagrams) as discussed in Chapter 9. These diagrams provide insight into limiting factors, and the potential
capacity gain from debottlenecking proposals can be readily seen. One typical characteristic of sewage
treatment plants is immediately obvious from such capacity envelopes – there is a large investment in
hydraulic (wet weather) capacity which is used for only a small percentage of the time and appears ripe
for exploitation.
11.3.5 Costs
In studies of Australian chemical process plants expanded in the 1960’s and 1970’s (Brennan and
Greenfield, 1994) it was found that the cumulative real fixed capital spent over successive expansions of

by guest
an existing plant could be related to plant capacity by the relationship:
SI = kI QB (11.5)
where
ΣI = cumulative fixed capital (investment) adjusted to a real basis

Q = capacity
kI, B = coefficients
The exponent B was approximately 0.8 for a parallel plant expansion but only 0.6 for an integral expansion
(involving debottlenecking investment). This means that additional capacity can be provided more cheaply
through capacity creep-type investment than through parallel plant expansions. Both the capital investment
and associated ongoing maintenance costs are lower.
Additional economic benefit accrues through the deferral of capital investment until it is needed to match
increasing load. Figure 11.9 illustrates the reduction achieved in the Net Present Value of a capital
investment if a given expansion in plant capacity is provided in an increasing number of increments.
100
(% of Single Stage Cost)
NPV of Total Capital
90
80
70
60
1 2 3 4 5 6
No. of Equal Expansion Stages
Figure 11.9 NPV reduction for incremental expansion over a 10 year period compared with single expansion
at time zero; total expansion 30%, discount rate 5%.
Capacity creep has also been found to provide savings in unit operating costs, particularly with respect to
personnel (Brennan & Greenfield, 1994). In the process plants studied, the total number of personnel
employed frequently remained unchanged despite capacity (production) growth; reductions in the
personnel requirement per annual tonne of product averaged 4.5% per year. With parallel plant
expansion, however, which involved significant additions of new equipment and increased materials
handling, personnel numbers increased.
At any stage in the development of a plant, the economic benefit of an integral expansion can be
compared with the cost of an alternative parallel expansion using the form of cost diagram shown in
Figure 11.10. Total annual cost (or net present value) is used in this diagram to reflect both capital
investment and annual operating and maintenance cost. Both curves, but particularly the integral
expansion curve, will vary with time as experience and understanding are developed. The plant strategy
therefore needs to be revisited periodically.

by guest
Parallel plant expansion
($ / unit capacity increase)

Total Unit Cost
Integral plant expansion
Increase in Plant Capacity

Figure 11.10 Relative costs for integral and parallel expansion.
11.4 EVOP PROCEDURES

As described above, the fundamental philosophy of EVOP is that the plant should always be operated and
monitored so as to deliberately improve the operators’ understanding of its behavioural characteristics and
allow its performance and capacity to be continually improved – a structured process of ongoing
improvement. This section describes methods used to enable operators to better understand and improve
their plants.
11.4.1 Method
The EVOP program should be integrated with the regular plant monitoring and control program. It is best
run by the process superintendent with the support of one or more process specialists who serve in an
advisory and supportive capacity. Plant data are collated, analysed and reported at regular intervals
(typically monthly) and when specific test programs have been undertaken. The operational group meet
at regular intervals to review the process data, provide helpful analysis and stimulate a flow of new ideas
to be incorporated in the investigation. Current progress is openly shared to facilitate input and feedback
from all plant personnel.
The basic strategy (introduced in Chapter 1) is illustrated in Figure 11.11. The actions are as
follows:
(1) Define goals and set targets.
(2) Set the plant control variables.
(3) Measure the outcomes – data collection program.
(4) Interpret the data and develop ideas for improvement – regular (monthly) process reviews, technical
memoranda and EVOP meetings.
(5) Re-adjust the process.
(6) Conduct trials and tests on specific processes or the whole plant.
In simple terms there are three basic steps: measure, interpret, act. These steps form the basis of an
ongoing EVOP strategy which will produce long-term benefits. The key to better performance is
improved understanding of plant behaviour.

by guest
Conduct special tests

CONDUCT
TRIALS
Adjust process settings
ADJUST
PROCESS
OPERATING
GOALS
INTERPRET MEASURE
DATA PERFORMANCE
Review daily KPIs Collect data – routine

Maintain operating statistics, trend - special programs
plots - site data, daily log,
Conduct monthly in-depth review laboratory support
Identify trends & implications Maintain data bases
Decide & record process actions – on-site
Improve understanding of plant - centralised
behaviour, record insights Maintain plant documentation
Develop ideas for improvement - design, licence,
agreements,
management plans,
manuals
Figure 11.11 The EVOP process operating cycle.
Step 1: Data collection

Careful observation and interpretation of plant behaviour are essential if the plant is to be tuned for
maximum performance. There are four distinct aspects to this monitoring effort:
• Visual observation on a daily basis.
• Operating measurements obtained both instrumentally and manually.
• Laboratory measurements of physical, chemical and biological parameters.
• Maintenance of good records and files of operating information (discussed in Data Interpretation
below).
An example routine monitoring schedule is set out in Table 11.2. This is the initial schedule used for the
two-year intensive performance proving phase after the new Brunswick Valley STP was commissioned
at the end of 2010. The plant had the flowsheet shown in Figure 9.1 except that WAS was wasted
directly to the digester.

by guest
by guest
Table 11.2 Brunswick Valley STP – routine data collection schedule.
ID Data Source// Type of Frequency DATA ANALYSES (no. per week; M = monthly)
Sampling Location Sample
WEEKLY ANALYSES
Total Alkalinity
Total Sulfide
TS
Quantity
VSS
Temperature
VFA
TP
Control Setting
Availability
TN
Oil & Grease
SS
COD
BOD
Hours Run
Storage Level
NOx-N
H2S
Flow
Deliveries
Faecal coliforms
PO4-P
SVI
NH3-N
pH
Dissolved Sulfide
M&E, I&C equip. SCADA Daily ✓ ✓

Flow meters SCADA Daily ✓
Process controls SCADA When ✓
changed
Dosing pumps Pump When ✓
setting changed

Electricity meters – Monthly ✓
Screenings – Weekly ✓
Grit – Weekly ✓
Sewage 24 h comp. 1 1 1 1 1 1 1 1 1 9
Sewage Grab 1 1 2
Grit tank exit Grab 1 1 2
Foul air In-situ 1 1
Evolutionary operation
Anaerobic reactor Grab Weekly 1 1

Oxidation ditch Grab per week 2 M 1 1 2 6
Clarifier effluent Grab per week 2 2 2 2 2 1 1 12
UV effluent Grab Weekly 1 1
Reuse effluent 24 h comp. Weekly 1 1 1 1 1 1 1 7
River discharge 24 h comp. Weekly 1 1 1 1 1 1 1 7
Digester Grab Weekly 1 M 1 1 1 1 5
BFP filtrate Grab Weekly 1 1
Sludge cake:
To storage Grab Weekly ✓ 1 1
Trucked – Weekly ✓ 1 1
Chemicals:
Ferric – Weekly ✓ ✓
(Continued)
211
by guest
212
Table 11.2 Brunswick Valley STP – routine data collection schedule (Continued).
ID Data Source// Type of Frequency DATA ANALYSES (no. per week; M = monthly)
Sampling Location Sample

WEEKLY ANALYSES
Total Alkalinity
Total Sulfide
TS
Quantity
VFA
VSS
Temperature
TP
Control Setting
Oil & Grease
Availability
SS
TN
Hours Run
Storage Level
COD
BOD
NOx-N
H2S
Flow
Deliveries
Faecal coliforms
PO4-P
SVI
NH3-N
pH
Dissolved Sulfide
Alum – Weekly ✓ ✓
Caustic soda – Weekly ✓ ✓
Hypochlorite – Weekly ✓ ✓
Polymer – Weekly ✓ ✓
WEEKLY ANALYSES 1 3 1 2 9 M 1 1 5 3 3 4 3 8 2 4 1 2 2 1 56
NOTES:
1. Shows data collection frequencies during normal operation.
2. Collect grab samples at same time each sampling day according to appropriate sampling protocol.
From time to time, the schedule will be adjusted or additional data collection programs implemented for
specific trials or investigations. This comprehensive effort is easily justified when the overall costs are fully
considered:
• The cost of the data collection program.
• The environmental cost associated with periodic discharges of inadequately treated effluent
and biosolids.
• The supplementary cost of investigating and remedying upsets in plant performance.
• The additional operating cost associated with greater than necessary consumptions of energy
and chemicals.
• The savings accruing from deferment of plant expansion achieved through increase in operational
capacity.
Step 2: Data interpretation

The data collected are useless if they are not examined and used for operational decision-making. The two
critical steps are recording and interpretation.
All data, however obtained, should be recorded in a single plant file to facilitate interpretation. The
routine use of trend plots of all important parameters is by far the best way to manage and study the
data. Tables camouflage trends; graphs illuminate them. The use of Excel facilitates this procedure.
Trend plots are discussed below.
Useful frequencies for process reviews are daily, weekly and monthly. Because STP data are highly
variable, decisions need to be made on the basis of trends. Daily data are useful only for picking up
large and unexpected changes in parameters. Because of the effort involved in data analysis and
trending, a few critical parameters can be reviewed weekly and the full range of plant parameters
monthly.
Decisions on process adjustments must then be made. Specific issues may need to be investigated.
Setpoints may need adjustment to counterbalance changes in loading or influent characteristics or
season. Structured trials can be planned to evaluate the effects of operational settings on process
parameters and define optimum operating regions. Capacity bottlenecks can be identified and methods
for overcoming them pursued. Effort to further improve performance should be ongoing.
Specific test programs should be reported regularly in technical memoranda or reports. These documents
can be structured simply as Aims, Procedure, Results, Interpretation, Conclusions, Recommendations and
Appendices.
Step 3: Process adjustment

The decisions on process changes made in Step 2 are acted upon. Specific tests are conducted. The outcomes
are measured (Step 1) and the cycle repeats. Improvement comes with continuing effort.
Earlier chapters have alluded to or described a sample of potential investigations and specific data
gathering aimed at better understanding specific plant operating characteristics: reactor or digester tracer
testing (Table 6.1), bioprocess minimum SRT testing (Chapter 6), aeration diffuser monitoring and
testing (Chapter 6), bioreactor SVI optimisation (Chapter 6), clarifier stress testing (Chapter 7), sludge
stability testing (Table 8.3), BFP and centrifuge testing (Tables 8.6 & 8.8), and plant load distribution
(Table 9.2).
Most importantly, the process control systems need to be optimally tuned to reliably maintain the
associated process variables at their set points (Chapter 10).

by guest
11.4.2 Trend plots

The most useful way to understand plant behaviour is to maintain trend plots of all relevant operating
parameters. Trend plots enable true trends to be differentiated from noise, highlight shifts in behaviour,
facilitate comparison of related trends and display months or years of data for multiple parameters on a
few understandable pages.
Figure 11.12 shows the trend plots used over the first two years of operation of the Brunswick Valley
STP. The plant flowsheet corresponded with Figure 9.1, with the exception that WAS was wasted from
the RAS line direct to the digester. Plots were added progressively as found useful, eventually
amounting to twenty eight plots displaying eighty four parameter trends. These trends and associated
tables of key operating parameters were updated monthly and used as the basis for monthly performance
reviews. Effects of previous adjustments to process parameters were examined. New insights and
resulting refinements to operating procedures were systematically recorded and implemented.
Useful aspects of this trending technique are as follows:
Production: Use an Excel file. Link the trends to the data sheets so they are automatically extended when
new data are added. For long-term use, a time-scale of 12 or 24 months is most useful, displaying
trends and relationships without requiring frequent adjustment of the plots.
Format: The proportions shown have proven their worth, allowing five plots containing multiple
parameters to be shown on each A4 page at a readable size; this facilitates comparison of trends.
Colours are essential for multiple trends, improving readability significantly and increasing the
number of trends that can usefully be shown in each plot. A grey background can improve
readability with multiple parameters. Line thickness or style and use or not of symbols are
additional differentiators for ease of interpretation. Moving averages can be added to clarify trends
(e.g. Plots 2 and 28). Design or target levels can also be shown.
Order: Trends can usefully be grouped as shown in the example: plant loading, effluent quality, energy
consumption, chemical consumption, liquid stream operating parameters, solids stream operating
parameters. Operating cost trends can also be added.
Lab vs site data: Laboratory data generally lag site data for the same or similar parameters by days or
weeks, so both can be plotted for timeliness and accuracy (e.g. Plots 5–7).
Relationships: Plots can be configured to examine and highlight potential relationships, and adjusted from
month to month as relationships emerge. For example, in Plot 24 the scales have been chosen to
highlight the close relationship between digester alkalinity and pH. In Plot 25 the digester SVI scale
has been reversed to highlight the close relationship between BFP cake TS and the feed sludge SVI.
Similarly, Plot 17 was set up to track the relationship between plant caustic soda dose, effluent
total alkalinity and mixed liquor pH. Anaerobic reactor pH was added to the plot when monitoring
was initiated. Plot 16 is a companion plot which trends the biological P removal calculated from
the plant alum dose (influent – effluent – chemical removal)2 and also shows the correlation
between bio-P removal and anaerobic zone PO4-P.
Plots 14 and 15 were configured to track the close relationship between DO set point, effluent
ammonia:nitrate ratio and SVI in the oxidation ditch bioreactor.
And Plot 18 was set up to determine whether there was a relationship between effluent oil and
grease level and oxidation ditch scum inventory (with constant daily operating time for the scum
harvester). For the data so far available no relationship is evident.
2
Alum P removal was calculated using the relationships given in Chapter 3, Section 3.7.

by guest
Trend Plots 1–5

BRUNSWICK VALLEY STP Dec-10 to Mar-13
A. PLANT LOADING
Rainfall Mullumbimby Inf Total Inf Mullum Lagoon
1. Influent Flows
15 100
Design flows:
Easter Sunday
DL start 26-Feb-11
Startup 3-Dec-10
Peak day 22 ML/d

Influent Flow (ML/d)
12 80
Rainfall (mm/d)
Peak rate 27 ML/d
9 60
6 40
Design ADWF 3.8 ML/d
3 20
0 0
1-Dec 1-Feb 1-Apr 1-Jun 1-Aug 1-Oct 1-Dec 1-Feb 1-Apr 1-Jun 1-Aug 1-Oct 1-Dec 1-Feb 1-Apr
2. Influent Strength COD conc BOD conc Alkalinity as CaCO3 COD mass COD Mass-45d MA
900 3
Concentration (mg/L)
COD Mass (t/d)

600 Design COD50 Mass 2
Design COD50 Conc
300 1
0 0
3. Influent Nutrients TKN VFA TP Sulfide-sewage sol -grit tank tot -grit tank sol
100 15
80 12
TKN, VFA (mg/L)
Sulfide (mg/L)
Design TP50
TP (mg/L)
60 9
Design TKN50
40 6
20 3
0 0
B. EFFLUENT
BOD SS O&G
4. Organics
15 Licence O&G Max 10
12 8
Licence BOD90
O&G (mg/L)
BOD (mg/L)
SS (mg/L)
9 6
Licence O&G90
6 4
Design SS50
3 2
0 0
5. Effluent Nitrogen (Lab Composites) TN NH3-N

15
12
Nitrogen (mg/L)
Licence TN90
9
3
Licence NH3-N90
0
Figure 11.12 Brunswick Valley STP Process Trend Plots 1–28, being updated and reviewed monthly during
the two year performance proving period.

by guest
Trend Plots 6–10

6. Clarifier Effluent Nitrogen (Site Grabs) TIN NH3-N
15
12
Nitrogen (mg/L)
Licence TN90
9
3
Licence NH3-N90
0
7. Effluent P & pH TP (lab composites) PO4-P Clarifier (site grabs) pH-Clarifier Effluent pH-River Discharge
3 8
Phosphorus (mg/L)
2 7
pH
Licence pHmin
1 6
Licence TP90
0 5
8. Faecal Coliforms Site Reuse Eff UV Eff River Discharge

1000
RiverLimitMax
F coliforms (cfu/100mL)
RiverLimit90
100
UVLimitMax
UVLimit90
10
1
BDL plotted as 0.1
> value plotted as value
0.1
C. ELECTRICITY
9. Plant Power Plant Meter (daily rdgs) Country Energy Invoices (mthly)
100
80
Power (kW)
60
40
20
0
10. Energy Consumption - Monthly Averages Country Energy Invoices Plant Meter
1200
Specific Energy (kWh/ML)
900
600
Targets shown:
300
Peak mth at 2 ML/d - 890 kWh/ML
Average at 2 ML/d - 711 kWh/ML
0
the two year performance proving period (Continued).

by guest
Trend Plots 11–15

D. CHEMICAL CONSUMPTIONS
11. Plant Alum & Ferric Consumption Total Plant Alum Dose OD Alum Dose Ferric Sulfate Dose Ferric as Equiv Alum
180
Alum, Ferric Dose (mg solid/L)
150
120
90
AlumLimitMax FerricLimitMax
60
AlumLimitAve
30 FerricLimitAve
0
12. Plant Caustic Soda Consumption Total Plant Dose OD Dose

150
120
(mg/L as NaOH)
Caustic Dose
90
CausticLimitMax
60
CausticLimitAve
30
0
E. OPERATION - LIQUID STREAM

13. SRT, MLSS, RAS, ML Temperature SRT ML Temperature MLSS RAS ratio
50 8
40 6
SRT (7d MA, d)
ML Temp (°C)
MLSS (g/L)
RAS Ratio
30 Design MLSS50 4
20 2
10 0
14. DO SP, Effluent N Eff NH3-N Eff NO3-N DO SP

15 1
12 0.8
Clarifier N (mg/L)
DO SP (mg/L)
9 0.6
6 0.4
3 0.2
0 0
15. SVI & Effluent N Ratio OD-SVI -SSVI Digester SVI Effluent NH3:NO3
300 10
NH3-N:NO3-N Ratio
225 1
(90d MA)
SVI (mL/g)
150 0.1
Design SVI90
75 0.01
0 0.001
Figure 11.12 (Continued).

by guest
Trend Plots 16–20

16. Bio-P Removal An PO4-P Estimated Bio-P Removal Clarifier PO4-P Sewage VFA
12 80
70
10
Sewage VFA (mg/L)

60
8
50
P (mg/L)
6 40
30
4
20
2
10
0 0
17. ML Alkalinity, pH Plant Caustic Dose (mgCaCO3/L) Effluent TA (mgCaCO3/L) ML pH Anaerobic pH

200 8.5
Caustic Dose, Effluent TA
160 8.0
120 7.5
pH
80 7.0
40 6.5
0 6.0
18. OD Scum vs Effluent O&G OD Scum Effluent O&G

240 0
OD Scum (% of straight)
200 1
Effluent O&G (mg/L)

160 2
120 3
80 4
40 5
0 6
F. OPERATION - SOLIDS STREAM

19. Digester HRT, SRT & TSS HRT-7d MA SRT-7d MA TSS
25 15
20 12
HRT, SRT (d)
15 9
TSS (g/L)
10 6
5 3
0 0
20. Digester VSS/TSS OD Digester

0.8
VSS/TSS
0.7
0.6
0.5

by guest
Trend Plots 21–25

21. SVI & Alum Dose OD-SVI Digester SVI Digester Alum Dose
300
Alum Dose (mg/g feed SS)
Unstirred SVI (mL/g)
225
150
Design SVI90
75
0
22. Digester Aeration & Nutrients Dig NH3-N Dig NO3-N Dig PO4-P Aerated Fraction
12 0.5
Daily Aerated Fraction

9 0.4
N, P, DO (mg/L)
6 0.3
3 0.2
0 0.1
23. Digester: Chemical Dosing Alum Caustic Soda

150 400
(mgCaCO3/g feed SS)

120 320
(mg solid/g feed SS)
Caustic Soda Dose

Alum Dose
90 240
60 160
30 80
0 0
24. Digester Alkalinity & pH pH TA

8.0 150
7.6 120
TA (mgCaCO3/L)
7.2 90
pH
6.8 60
6.4 30
6.0 0
25. BFP Performance 1 Cake TS Digester SVI

17 50
16
Digester SVI (mL/g)
15
Cake TS (%)
14 TS Limit Ave 150
13
12
11 250
Figure 11.12 (Continued).

by guest
Trend Plots 26–28

26. BFP Performance 2 Filtrate SS Digester SVI
700 50
600
Digester SVI (mL/g)

Filtrate SS (mg/L)
500
400
150
300
200
100
0 250
27. BFP Polymer Dose Liquid Dose Equivalent Powder Dose

3 6
Dose as Powder (kg/tDS)

2.5 5
Liquid Polymer Dose
(kg liquid /ML)
2 Liquid Limit Max 4
1.5 Liquid Limit Ave 3
1 2
0.5 1
0 0
28. Sludge Production kgDS/ML kgDS/kgCOD

400 0.8
300 0.6
Sludge Production
kgDS/kgCOD
30d MA for kgDS/ML
(kgDS/ML)
200 kgDS/ML Limit Ave 0.4
100 0.2
0 0
Trends vs variability: Plot 16 is a good example of the utility of plots for discerning trends when
day-to-day variability is high. Plot 11 is another example – until November, alum was dosed to the
oxidation ditch at a constant daily mass rate giving a variable mg/L rate. Despite the highly
variable mg/L value the overall downwards trend as the dosing rate was decreased is clearly evident.
11.4.3 General principles

Tuning a BNR plant can be complex, with many interactions between process variables. To achieve good
performance it is essential to collect operating data and review performance on a regular basis. Some general
principles are as follows:
(1) Biological processes are susceptible to rapid upset but are very slow to recover. It is therefore
essential to monitor plant performance closely in order to identify incipient upsets and initiate
recovery action as early as possible.
(2) It is difficult to have too much data. Do not skimp on analytical effort and ensure that plant
instruments are regularly calibrated and maintained.

by guest
(3) Maintain trend plots of key parameters and examine all results regularly to catch problems early.
(4) Look for interactions between and within processes so that understanding of the plant’s operating
characteristics can be actively enhanced. When opportunities present, conduct special tests to
improve understanding and quantify operating characteristics.
(5) Be aware of interactions between processes and the likely effects that changes to one process may
have on others.
(6) Best results are achieved if process operating conditions are maintained as constant as possible.
Keep process operating adjustments to the necessary minimum. Ensure process equipment
is reliable.
(7) Be patient. The full effects of many operating adjustments are not evident for 2–3 solids residence
times, even for changes seemingly governed by hydraulic residence time; this is because associated
changes in biomass composition are often involved.
(8) With statistical licence limits a good operating strategy is to operate routinely as far below the limits
as is practicable or economic; this provides a statistical buffer against periods of upset.
11.5 FAUX EVOP

EVOP as presented in this book would be regarded by its originators, George Box and Norman Draper,
eminent statisticians both, as rather elementary. The aim here is to promote the approach rather than the
mathematical detail, to encourage an interest in basic tuning to improve the performance of sewage
treatment plants without the need for statistical sophistication. Both environmental and economic
benefits can accrue – in terms of improved performance and deferral of plant expansions – in an industry
where tuning and learning curves are not part of the vernacular. This approach to operation is also a lot
of fun!

by guest
Appendix 1
A note on the statistics of variability
For operational purposes, the main statistical issue is variability – of influent, effluent and plant operating
parameters. To meet the operating goals, operators need to know how much all of these parameters vary and
they should aim to reduce variability.
For simplicity this discussion is based on sets of daily data collected over lengthy periods, usually
a month or a year to meet tuning, reporting and licensing requirements. To provide a reliable
representation of overall performance, these data should ideally be collected at uniform time intervals.
AVERAGE AND MEDIAN

Distributions of STP data are usually positively skewed, with a long tail of infrequent, high values. Values
do not vary in terms of plus and minus but in multiply and divide. The arithmetic mean (or average) is then
not the same as the median or 50 percentile (the value exceeded on half of the days) because a few very high
values pull the average up. Daily values are less than the average on more than 50% of days.
Most of these distributions can be normalised by taking the logarithms of the values; that is, the
logarithms are normally distributed and vary in terms of plus and minus log values. For a normal
distribution the mean and the median are equal. The mean of an ideal log-normal distribution is called
the geometric mean and is calculated by averaging the logs of the data. The geometric mean equals the
median and is therefore the value exceeded by 50% of the results, or the value exceeded 50% of
the time.
In general STP use, average relates to total quantities over a period while median relates to percentage
of time.
PERCENTILES
For a normal distribution 68% of the values lie within plus or minus one standard deviation of the mean. The
standard deviation adds to and subtracts from the mean. For a log-normal distribution this applies to the logs
of the data. Usually the data are expressed in their natural values and the geometric standard deviation
multiplies and divides the geometric mean (median). An exception is microbiological data for which the
variability is high; both the median and the variability are then normally expressed in logarithmic terms
(plus or minus a number of logs).

by guest
APPLICATION
Because most performance standards are expressed as upper limits, percentiles are usually expressed as
percentage less than; for example, the 90 percentile is exceeded by only 10% of the values and 90% are
less, or conform with the standard.
The variability of data known or assumed to be log-normally distributed is often expressed in terms of its
geometric standard deviation, or alternatively by the ratio of the 90 percentile to the 50 percentile (simpler to
understand). If required, other percentiles can then be estimated using the normal distribution ratios.
Selected ratios are listed in the Table A1.1 below.
Table A1.1 Log-normal distribution.
% less than % greater than No. of SDs from 50%ile Example of practical significance
99.99 0.01 3.736 Peak hour in a year
99.95 0.05 3.307 Peak day in 5 years
99.9 0.1 3.106
99.7 0.3 2.780 Peak day in a year
99 1 2.327
98 2 2.053 Peak week in a year
95 5 1.643
90 10 1.282 Peak month in a year
84.1 15.9 1.000 Peak 2 months in a year
80 20 0.842
50 50 0.000 Median
20 80 0.842
15.9 84.1 1.000 Minimum 2 months in a year
10 90 1.282 Minimum month in a year
5 95 1.643
2 98 2.053 Minimum week in a year
1 99 2.327
0.3 99.7 2.780 Minimum day in a year
0.1 99.9 3.106
0.05 99.95 3.307 Minimum day in 5 years
0.01 99.99 3.736 Minimum hour in a year
The simplest procedure is as follows:

(1) Determine the 50 and 90 percentiles of the data. Calculate the 90:50 percentile ratio. From
Table A1.1, this is 1.282 geometric standard deviations.
(2) Estimate the values for other percentiles lying outside the data range by ratio from the number of
standard deviations. For example, to estimate the value exceeded 0.3% of the time (the 99.7
percentile or peak day in Table A1.1):
log(99.7 percentile) = log(50 percentile) + (2.780/1.282)log(90:50 percentile ratio)

by guest
A note on the statistics of variability 225
The ratio of the average to the geometric mean is given by:
average/geometric mean = (1 + v2 )0.5

where
v = arithmetic coefficient of variation

= standard deviation/average
This has application when the total annual mass discharge of an effluent constituent must meet a limit. It is
then the average daily discharge which is important, not the median discharge. Figure A1.1 shows the fit to
this relationship of daily effluent N and P data from the Cleveland Bay MBR-BNR plant in Townsville
which has mass discharge limits. Each data point relates to a 4–6 month seasonal period from Mar-08
to Mar-10.
1.8
Theoretical relationship:
Ave/Geomean = (1 + v2)0.5
Average/Geometric Mean
1.6
Theoretical
CB - TN
1.4 CB - TP
1.2
1
0 0.2 0.4 0.6 0.8 1 1.2 1.4
Coefficient of Variation, v
(standard deviation/average)
Figure A1.1 Cleveland Bay effluent data conform with the theoretical relationship between the normal and
log-normal distributions.

by guest
Appendix 2
Chemicals: useful properties

by guest
by guest
228
Chemical Common Formula Mass % of Useful Commercial Strength SG (liquid)

Name Name Component Form (by mass) Bulk Density
(t// m3, solid)
Aluminium Liquid alum Al2(SO4)314.3H2O Al, 9.02% of Solution 46% solution 1.30
sulfate solid
Calcium Hydrated lime Ca(OH)2 Alkalinity, Solid 100% 0.5–0.6
hydroxide 1.35 kg CaCO3/kg

lime
Ethanol – C2H5OH COD, 2.09 Liquid (95 95% (5% water) 0.81
kg/kg ethanol Standard
Grade)
Ferric chloride – FeCl3 Fe, 34% of Solution 42% solution 1.45
solid
Ferrous – FeCl2 Fe, 44% of Solution 27% solution 1.29
chloride solid
Ferric sulfate – Fe2(SO4)3 Fe, 28% of Solution 42% solution 1.50
solid
Lithium – LiOH · H2O Li, 16.5% Solid 100% –
hydroxide
(monohydrate)
Magnesium – Mg(OH)2 Alkalinity, Slurry 60% slurry 1.50
Mg(OH)2
Methanol – CH3OH COD, 1.5 kg/kg Liquid 100% 0.79
methanol
by guest
Molasses – Typical composition, %:1 Sucrose 35% Liquid Total organics 1.35–1.45
Water 23.5 COD 87% as sucrose-
Sucrose, C12H22O11 34.6 BOD 43% equivalent 75%
Reducing sugars 13.0
Other organics 15.7
Ash 13.2
Sodium Soda ash Na2CO3 Alkalinity, Solid 100% 0.5–0.9 light
carbonate 0.94 kg CaCO3/kg 0.9–1.1 dense
Na2CO3
Sodium Caustic soda NaOH Alkalinity, Solution 46% solution 1.50
NaOH
Sodium – NaOCl 125 gCl2/L Solution 10.4% Cl2 1.20

hypochlorite
Sucrose (a) Liquid C12H22O11 COD 1.12 kg COD/kg Solution 41% solution 1.18
sugar sucrose
(b) D.Nitro 67% solution 1.33
1
From GHD, 1995. COD of sucrose is 1.12 mgCOD/mg sucrose. Molasses COD & BOD determined by laboratory analysis.
Chemicals: useful properties
229
Appendix 3
Abbreviations
(excludes standard laboratory terminology)
ACT Australian Capital Territory

ADWF average dry weather flow
A/O anaerobic oxic
A2O anaerobic anoxic oxic
APT activated primary tank
BFP belt filter press
BNR biological nutrient removal
CASS cyclic activated sludge system
CFID continuous feed intermittent decant
EBPR enhanced biological phosphorus removal
EP equivalent persons
EPS bacterial exopolymeric substances
EVOP evolutionary operation
F:M ratio food:microorganism ratio
GAO glycogen accumulating organism
HRT hydraulic residence time
ICEAS intermittent cycle extended aeration system
IDAL intermittently decanted aerated lagoon
IDEA intermittently decanted extended aeration
JHB Johannesburg
MLE Modified Ludzack-Ettinger
MLSS mixed liquor suspended solids
MLVSS mixed liquor volatile suspended solids
MBR membrane bioreactor
MRT mean residence time
MUCT Modified University of Cape Town
ND nitrification denitrification
NDEBPR nitrification denitrification enhanced biological phosphorus removal

by guest
NSW New South Wales (Australian State)

OD oxidation ditch
OHO ordinary heterotrophic organisms
OTE oxygen transfer efficiency
OTR oxygen transfer rate
OUR oxygen uptake rate
PAO phosphorus accumulating organism
Qld Queensland (Australian State)
RAS return activated sludge
RBCOD readily biodegradable COD
RTD residence time distribution
SA South Australia (Australian State)
SBCOD slowly biodegradable COD
SBR sequencing batch reactor
SC secondary clarifier
SMP soluble microbial products
SND simultaneous nitrification and denitrification
SOTE standard oxygen transfer efficiency
SOTR standard oxygen transfer rate
SOUR specific oxygen uptake rate
SP set point
SRT solids residence time
sSVI stirred sludge volume index
SSVI stirred specific volume index
STP sewage treatment plant
SVI sludge volume index
TMP transmembrane pressure
TWL top water level
UCT University of Cape Town
VASS volatile active suspended solids
Vic Victoria (Australian State)
WA Western Australia (Australian State)
WAS waste activated sludge
WD water depth
WQCC water quality control centre
WWTP wastewater treatment plant

by guest
References
Ahmed Z., Lim B.-R., Cho J., Song K.-G., Kim K.-P. and Ahn K.-H. (2008). Biological nitrogen and phosphorus
removal and changes in microbial community structure in a membrane bioreactor: effect of different carbon
sources. Water Research, 42(1–2), 198–210.
Andreasen K. and Nielsen P. H. (2000). Growth of Microthrix parvicella in nutrient removal activated sludge plants:
studies of in situ physiology. Water Research, 34(5), 1559–1569.
Ardern E. and Lockett W. T. (1914a). Experiments on the oxidation of sewage without the aid of filters. Society of
Chemical Industry Journal, 33(10), 523–539.
Ardern E. and Lockett W. T. (1914b). The oxidation of sewage without the aid of filters: Part 2. Society of Chemical
Industry Journal, 33(23), 1122–1124.
ASCE Standard ASCE/SEI 2-06 (2007). Measurement of Oxygen Transfer in Clean Water. American Society of Civil
Engineers, Reston, VA, USA.
Åström K. J. and Hägglund T. (1995). PID Controllers: Theory, Design and Tuning. ISA Press, Research Triangle Park,
USA.
Åström K. J. and Hägglund T. (2006). Advanced PID Control. ISA–The Instrumentation, Systems and Automation
Society, USA.
AWWA (1997). BNR Plants in Australia. Australian Water & Wastewater Association, Sydney, Australia.
Baars J. K. (1962). The use of oxidation ditches for treatment of sewage from small communities. Bulletin of the World
Health Organization, 26, 465.
Bane I. H. (1997). Dewaterability of activated biosolids. Water, 24(3), 56–57.
Barnard J. L. (1973). Biological denitrification. Water Pollution Control, 72(6), 705–720.
Barnard J. L. (1976a). A review of biological phosphorus removal in the activated sludge process. Water SA, 2(3),
136–144.
Barnard J. L. (1976b). Nutrient removal in biological systems. Water Pollution Control, 74, 143–154.
Barnard J. L. (1984). Activated primary tanks for phosphate removal. Water SA, 10(3), 121–126.
Batstone D. J., Keller J., Angelidaki R. I., Kalyuzhnyi S. V., Pavlostathis S. G., Rozzi A., Sanders W. T. M., Siegrist H.
and Vavilin V. A. (2002). Anaerobic Digestion Model No.1. Scientific and Technical Report No. 13, IWA
Publishing, London, UK.
Beenyup WWTP Laboratory (2006). Biochemical Oxygen Demand – Reduction in the BOD5 Test Upon Storage at
,4°C. Water Corporation, Perth, Australia.
Biggs C. A. (2000). Activated Sludge Flocculation: Investigating the Effect of Shear Rate and Cation Concentration on
Flocculation Dynamics. PhD thesis, Dept of Chemical Engineering, University of Queensland, Brisbane, Australia.
Biggs C. A. and Lant P. A. (2000). Activated sludge flocculation: on-line determination of floc size and the effect of
shear. Water Research, 34(9), 2542–2550.

by guest
Bisogni J. J. and Lawrence A. W. (1971). Relationships between biological solids retention time and settling
characteristics of activated sludge. Water Research, 5, 753–763.
Boisvert J.-P., To T. C., Berrak A. and Jolicoeur C. (1997). Phosphate adsorption in flocculation processes of aluminium
sulphate and poly-aluminium-silicate-sulphate. Water Research, 31(8), 1939–1946.
Box G. E. P. and Draper N. R. (1969). Evolutionary Operation: A Statistical Method for Process Improvement. John
Wiley, New York.
Brennan D. J. and Greenfield P. F. (1994). Competitive analysis: watch out for capacity creep. Chemical Engineering in
Australia, CHE19(4), 7–8.
Brennan D. J. and Stephens G. K. (1985a). The extent and rate of capacity increase in process plants. Process Economics
International, V(3 & 4), 15–25.
Brennan D. J. and Stephens G. K. (1985b). Changes in cost and performance resulting from plant expansion. Process
Economics International, VI(1), 16–26.
Casey T. and Alexander W. V. (2001). Design and Operating Strategies to Minimize Bulking by Anoxic-Aerobic
Filamentous Organisms in Nutrient Removal Activated Sludge Plants, WRC Report No. 775/1/01, Water
Research Commission, Pretoria.
Casey T. G., Wentzel M. C. and Ekama G. A. (1999). Filamentous organism bulking in nutrient removal activated
sludge systems. Paper 11: a biochemical/microbiological model for proliferation of anoxic-aerobic (AA)
filamentous organisms. Water SA, 25(4), 443–451.
Chao A. C. and Keinath T. M. (1979). Influence of process loading intensity on sludge clarification and thickening
characteristics. Water Research, 13, 1213–1223.
Chow V. T. (1959). Open-Channel Hydraulics. McGraw–Hill, New York.
Ciba-Geigy Ltd (1981). Geigy Scientific Tables, Vol 1, 8th edn., Summit, New Jersey, USA.
Daigger G. T., Randall C. W., Waltrip G. D., Romm E. D. and Morales L. M. (1987). Factors affecting biological
phosphorus removal for the VIP process, a high-rate University of Cape Town process. IAWPRC Conf on
Biological Phosphate Removal, Rome.
Dick R. I. and Vesilind P. A. (1969). The sludge volume index – what is it? Journal of WPCF, 41(7), 1285–1291.
Ekama G. A., Barnard J. L., Gunthert F. W., Krebs P., McCorquodale J. A., Parker D. S. and Wahlberg E. J. (1997).
Secondary Settling Tanks: Theory, Modelling, Design and Operation. IAWQ Sci & Tech Rpt No. 6.
Ekama G. A. and Marais G. v. R. (1977). The activated sludge process, Part 2 – Dynamic behaviour. Water SA, 3(1),
18–50.
Ekama G. A., Dold P. L. and Marais G. v. R. (1986). Procedures for determining influent COD fractions and the
maximum specific growth rate of heterotrophs in activated sludge systems. Water Science and Technology, 18,
91–114.
Ekama G. A., Wentzel M. C., Casey T. G. and Marais G. v. R. (1996). Filamentous organism bulking in nutrient
removal activated sludge systems. Paper 6: review, evaluation and consolidation of results. Water SA, 22(2),
147–152.
EPA NSW (1997). Environmental Guidelines: Use and Disposal of Biosolids Products. NSW Environment Protection
Authority, Sydney.
Fair G. M., Geyer J. C. and Okun D. A. (1968). Water and Wastewater Engineering: Volume 2. Water Purification and
Wastewater Treatment and Disposal. John Wiley, New York.
Galarneau E. and Gehr R. (1997). Phosphorus removal from wastewaters: experimental and theoretical support for
alternative mechanisms. Water Research, 31(2), 328–338.
Gernaey K. V., Nielsen M. K., Thornberg D., Höök B., Munk-Nielsen T., Ingildsen P. and Jørgensen S. B. (2004). First
principle suspended solids distribution modelling to support ATS introduction on a recirculating WWTP. 2nd Int.
IWA Conf. on Automation in Water Quality Monitoring, Vienna, Austria.
GHD (1994). Thorneside WPCW Augmentation, Technical Memo No. 7 – Alum Dosing Jar Tests. Redland Shire
Council.
GHD (1995). Thorneside WPCW Augmentation, Planning Report. Redland Shire Council.
Gloag G. (2009). Dynamic modeling study for design of Pimpama WWTP, Gold Coast, Australia. Personal
communication.

by guest
References 235
Gloag G. and Hartley K. (2005). Failure frequency analysis – quantifying the operational process capacity of wastewater
treatment plants. Proceedings Australian Water Association 21st Biennial Conv, Brisbane, Australia.
Goldratt E. M. and Cox J. (2004). The Goal: A Process of Ongoing Improvement. 3rd revised edn, North River Press,
Great Barrington, Massachusetts, USA.
Guisasola A., de Haas D., Keller J. and Yuan Z. (2008). Methane formation in sewer systems. Water Research, 42(6–7),
1421–1430.
Guisasola A., Sharma K. R., Keller J. and Yuan Z. (2009). Development of a model for assessing methane formation in
rising main sewers. Water Research, 43(11), 2874–2884.
Hall E. R., Monti A. and Mohn W. W. (2010). A comparison of bacterial populations in enhanced biological phosphorus
removal processes using membrane filtration or gravity sedimentation for solids-liquid separation. Water Research,
44(9), 2703–2714.
Hartley K. J. (1985). Operating the Activated Sludge Process. GHD, Brisbane, Australia.
Hartley K. J. (1987). Hydraulics of horizontal shaft oxidation ditches. Journal of WPCF, 59(7), 686–694.
Hartley K. J. (1990). Fouling & Cleaning of Fine Bubble Ceramic Dome Diffusers, Research Report No. 14, Urban
Water Research Association of Australia.
Hartley K. J. (1995). Biological Nutrient Removal Plants: Review of Full-Scale Operation, Research Report No. 94,
Urban Water Research Association of Australia.
Hartley K. J. (1997a). Use of the oxidation ditch reactor in BNR plants, Proc 3rd Australian Conf on BNR. Australian
Water & Wastewater Ass.
Hartley K. J. (1997b). SBR BNR Process R&D Project – Final Report, GHD, Brisbane, Australia.
Hartley K. J. (1998). The cost of Australian BNR plants. Water, Jnl Australian Water & Wastewater Ass, 25(1), 20.
Hartley K. J. (2000). Maroochydore STP: Energy KPI Analysis. Maroochy Water, Maroochydore, Qld, Australia.
Hartley K. J. (2002). Analysis of aeration diffuser fouling. Unpublished.
Hartley K. J. (2008). Controlling sludge settleability in the oxidation ditch process. Water Research, 42(6–7),
1459–1466.
Hartley K. J. (2009). Cleveland Bay WTF: Review of Biological P Removal Performance. Townsville Water,
Townsville, Qld, Australia.
Hartley Ken (2012). Loganholme WWTP: Stage 7 Process Commissioning. Logan Water Alliance, Brisbane, Australia.
Hartley K. J. and Gobbie M. R. (1991). Fouling and cleaning of fine bubble ceramic dome diffusers. Proc Australian
Water & Wastewater Ass 14th Fed Conv, Perth, Australia.
Hartley K. J. and Lant P. A. (2006). Eliminating non-renewable CO2 emissions from sewage treatment: an anaerobic
migrating bed reactor pilot plant study. Biotechnology & Bioengineering, 95(3), 384–398.
Hartley K. and Lant P. (2010). Sludge settleability in BNR processes. Water, Jnl Australian Water Ass, 37(3), 64–70.
Hartley K., Hertle C., Komarowski S. and Walpole R. (1999). Operating experience with two activated primary
tanks for sewage prefermentation. Proc Australian Water & Wastewater Ass 18th Fed Conv, Adelaide,
Australia.
Henze M., Gujer W., Mino T. and van Loosdrecht M. (2000). Activated Sludge Models ASM1, ASM2, ASM2d and
ASM3. IWA Publishing, London.
Henze M., van Loosdrecht M. C. M., Ekama G. A. and Brdjanovic D. (eds) (2008). Biological Wastewater Treatment:
Principles, Modelling and Design. IWA Publishing, London.
Hirschmann W. B. (1964). Profit from the learning curve. Harvard Business Review, 42(1), 125–139.
Holbrook R. D., Higgins M. J., Murthy S. N., Fonseca A. D., Fleischer E. J., Daigger G. T., Grizzard T. J., Love N. G.
and Novak J. T. (2004). Effect of alum addition on the performance of submerged membranes for wastewater
treatment. Water Environment Research, 76(7), 2699–2702.
Ingildsen P. (2002). Realising Full-Scale Control in Wastewater Treatment Systems Using In-Situ Nutrient Sensors. PhD
thesis, Dept. of Ind. Electrical Engineering and Automation (IEA), Lund University, Lund, Sweden.
ISO15839 2003. Water Quality – On-line Sensors/Analysing Equipment for Water – Specifications and Performance
Tests. 1st edn, 2003-10-15, SAI Global.
Jenkins D., Richard M. G. and Daigger G. T. (2004). Manual on the Causes and Control of Activated Sludge Bulking,
Foaming, and Other Solids Separation Problems. 3rd edn, CRC Press, Boca Raton & IWA Publishing, London.

by guest
John Wilson & Partners (2001). Nambour STP, Loads & Impacts Study. Maroochy Water, Maroochydore, Qld,
Australia.
Lai F. Y., Ort C., Gartner C., Carter S., Prichard J., Kirkbride P., Bruno R., Hall W., Eaglesham G. and Mueller J. F.
(2011). Refining the estimation of illicit drug consumption from wastewater analysis: co-analysis of
prescription pharmaceuticals and uncertainty assessment. Water Research, 45(15), 4437–4448.
Lee S.-E., Koopman B., Bode H. and Jenkins D. (1983). Water Research, 17(10), 1421–1426.
Letterman R. D., Amirtharajah A. and O’Melia C. R. (1999). Coagulation and flocculation. In: Water Quality and
Treatment: A Handbook of Community Water Supplies, R. D. Letterman (ed.), 5th edn, McGraw–Hill,
New York, Chapter 6.
Levenspiel O. (1972). Chemical Reaction Engineering. 2nd edn, John Wiley, New York.
Liu J., Olsson G. and Mattiasson B. (2004). Control of an anaerobic reactor towards maximum biogas production. Water
Science and Technology, 50(11), 189–198.
Liu J., Olsson G. and Mattiasson B. (2006). Extremum-seeking with variable gain control for intensifying biogas
production in anaerobic fermentation. Water Science and Technology, 53(4–5), 35–44.
Loewenthal R. E. and Marais G. v. R. (1976). Carbonate Chemistry of Aquatic Systems: Theory and Application. Vol 1.
Ann Arbor Science, Ann Arbor, Michigan, USA.
Lopez-Vazquez C. M., Oehmen A., Hooijmans C. M., Brdjanovic D., Gijzens H. J., Yuan Z. and van Loosdrecht
M. C. M. (2009). Modeling the PAO-GAO competition: effects of carbon source, pH and temperature. Water
Research, 43(2), 450–462.
Ludzack F. J. and Ettinger M. B. (1962). Controlling operation to minimize activated sludge effluent nitrogen. Journal of
WPCF, 34, 920–931.
Ma Y., Peng Y., Wang S., Yuan Z. and Wang X. (2009). Achieving nitrogen removal via nitrite in a pilot-scale
continuous pre-denitrification plant. Water Research, 43(3), 563–572.
Mamais D., Jenkins D. and Pitt P. (1993). A rapid physical-chemical method for the determination of readily
biodegradable soluble COD in municipal wastewater. Water Research, 27(1), 195–197.
Marais G. v. R. and Ekama G. A. (1976). The activated sludge process part 1 – steady state behaviour. Water SA, 2(4),
164–200.
Marlin T. E. (2000). Process Control. Designing Processes and Control Systems for Dynamic Performance. 2nd edn,
McGraw Hill, New York.
Metcalf & Eddy (2003). Wastewater Engineering: Treatment and Reuse. 4th edn, McGraw–Hill, New York.
Mohlman F. W. (1934). The sludge index. Sewage Works Journal, 6(1), 119–122.
Monod J. (1949). The growth of bacterial cultures. Annual Review of Microbiology, 3, 371.
Monteith H. D. and Stephenson J. P. (1981). Mixing efficiencies in full-scale anaerobic digesters by tracer methods.
Journal of WPCF, 53(1), 78–84.
Monti A. and Hall E. R. (2008). Comparison of nitrification rates in conventional and membrane-assisted biological
nutrient removal processes. Water Environment Research, 80(6), 497–505.
Monti A., Hall E. R., Dawson R. N., Husain H. and Kelly H. G. (2006). Comparative study of biological nutrient
removal (BNR) processes with sedimentation and membrane-based separation. Biotechnology and
Bioengineering, 94(4), 740–752.
Morales L. M., Daigger G. T. and Borberg J. R. (1991). Capability assessment of biological nutrient removal facilities.
Research Journal of WPCF, 63(6), 900–909.
Muller A., Wentzel M. C., Loewenthal R. E. and Ekama G. A. (2003). Heterotroph anoxic yield in anoxic aerobic
activated sludge systems treating municipal wastewater. Water Research, 37(10), 2435–2441.
Natural Resource Management Ministerial Council, Environment Protection and Heritage Council and Australian
Health Ministers’ Conference (2006). Australian Guidelines for Water Recycling: Managing Health and
Environmental Risks.
Nielsen H. L., Carns K. E. and DeBoice J. N. (1973). Alum sludge thickening and disposal. Journal of AWWA,
65, 385.
Nielsen M. K. (1983). Computer Control of Nitrogen Removal in a Single Sludge Sewage Treatment Plant. PhD thesis,
Inst. for Kemiteknik Danish Technical Univ., Denmark.

by guest
References 237
Nielsen M. K. and Ønnerth T. (1995). Improvement of a recirculating plant by introducing STAR control. Water Science
and Technology, 31(2), 171–180.
Nielsen M. K., Carstensen J. and Harremöes P. (1996). Combined control of sewer and treatment plant during rainstorm.
Water Science and Technology, 34(3–4), 181–187.
Nowak O. (2002). Benchmarks for the energy demand of nutrient removal plants. Proc IWA World Congress,
Melbourne.
Olsson G. (2012a). ICA and me – a subjective review. Water Research, 46(6), 1585–1624.
Olsson G. (2012b). Water and Energy – Threats and Opportunities. IWA Publishing, London, UK.
Olsson G. and Andrews J. F. (1978). The dissolved oxygen profile – a valuable tool for the control of the activated sludge
process. Water Research, 12, 985–1004.
Olsson G. and Newell R. B. (1998). Talking of RAS – reviewing, assessing and speculating. Final address at the 7th
IAWQ Symp. on ICA, Brighton, UK, July, 1997. Water Science and Technology, 37(12), 397–401.
Olsson G. and Newell R. B. (1999). Wastewater Treatment Systems. Modelling, Diagnosis and Control. IWA
Publishing, London, UK.
Olsson G., Nielsen M. K., Yuan Z., Lynggaard-Jensen A. and Steyer J. P. (2005). Instrumentation, Control and
Automation in Wastewater Treatment Systems. Scientific and Technical Report No.15. IWA Publishing,
London, UK.
Olsson G., Rundqwist L., Eriksson L. and Hall L. (1985). Self-tuning control of the dissolved oxygen concentration in
activated sludge systems. In: Advances in Water Pollution Control, R. A. R. Drake (ed.), Int. Assoc. on Water
Pollution Res. and Control (IAWPRC), London, pp. 473–480.
Orhon D., Sozen S. and Artan N. (1996). The effect of heterotrophic yield on the assessment of the correction factor for
anoxic growth. Water Science and Technology, 34(5–6), 67–74.
Osborn D. W. and Nicholls H. A. (1978). Optimisation of the activated sludge process for the biological removal of
phosphorus. Progress in Water Technology, 10(1–2), 261–277.
Pajdak-Stos A. and Fialkowska E. (2012). The influence of temperature on the effectiveness of filamentous bacteria
removal from activated sludge by rotifers. Water Environment Research, 84(8), 619–625.
Palm J. C., Jenkins D. and Parker D. S. (1980). Relationship between organic loading, dissolved oxygen
concentration and sludge settleability in the completely-mixed activated sludge process. Journal of WPCF, 52
(10), 2484–2506.
Parker D. S. (1983). Assessment of secondary clarification design concepts. Journal of WPCF, 55(4), 349–359.
Parker D. S., Kaufman W. J. and Jenkins D. (1971). Physical conditioning of activated sludge floc. Journal of WPCF, 43
(9), 1817–1833.
Parker D. S., Kaufman W. and Jenkins D. (1972). Floc breakup in turbulent processes. Jnl Sanitary Engineering Div,
Proc ASCE, 98(SA1), 79–99.
Parker D., Wahlberg E., Buttz J., Daigger G., Dentel S., Farrell J., Stephenson J. and Sutton P. (1999). Research
Priorities for Debottlenecking, Optimizing and Rerating Wastewater Treatment Plants. Project 99-WWF-1.
Water Environment Research Foundation, Alexandria, VA, USA.
Parkin G. F. and Owen W. F. (1986). Fundamentals of anaerobic digestion of wastewater sludges. Journal of
Environmental Engineering, ASCE, 112(5), 867–920.
Pasveer A. (1959). Contribution to the development in activated sludge treatment. Journal and Proceedings of the
Institute of Sewage Purification, 4, 436.
Peck G. M. (1984). The principles of learning and forgetting. Inst. of Engrs Aust, Muli-Disciplinary Engg Trans,
pp. 95–106.
Pipes W. O. (1979). Bulking, deflocculation and pinpoint floc. Journal of WPCF, 51(1), 62–70.
Potter T., Koopman B. and Svoronos A. (1996). Optimisation of a periodic biological process for nitrogen removal from
wastewater. Water Research, 30(1), 142–152.
Rabinowitz B. and Marais G. v. R. (1980). Chemical and Biological Phosphorus Removal in the Activated Sludge
Process. Research Report No. W32, Civil Engineering Department, University of Cape Town, South Africa.
Rieger L. and Olsson G. (2012). Why many control systems fail. Water Environment & Technology, Water Environment
Federation, June 2012, 24(6), 43–45.

by guest
Rosen C., Ingildsen P., Guildal T., Nielsen M. K., Jacobsen B. N. and Ønnerth T. (2004). On-line estimation of
nitrification and denitrification capacity at Avedøre wastewater treatment plant. Aquatech, Int. Conf. Upgrading
of Wastewater Treatment Plants, 30 September–1 October 2004, Amsterdam, The Netherlands.
Rosso D. and Stenstrom M. K. (2006). Economic implications of fine-pore diffuser aging. Water Environment Research,
78(8), 810–815.
Sezgin M. (1982). Variation of sludge volume index with activated sludge characteristics. Water Research, 16, 83–88.
Sezgin M., Jenkins D. and Parker D. S. (1978). A unified theory of filamentous activated sludge bulking. Journal of
WPCF, 50(2), 362–381.
Sharma K. R., Yuan Z., de Haas D., Hamilton G., Corrie S. and Keller J. (2008). Dynamics and dynamic modelling of
H2S production in sewer systems. Water Research, 42(10–11), 2527–2538.
Shipin O. V., Meiring P. G. J., Phaswana R. and Kluever H. (1999). Integrating ponds and activated sludge process in the
PETRO process. Water Research, 33(8), 1767–1774.
Siebritz I. D., Ekama G. A. and Marais G. v. R. (1982). A parametric model for biological excess phosphorus removal.
Snidaro D., Zartarian F., Jorand F., Bottero J. Y., Block J. C. and Manem J. (1997). Characterisation of activated sludge
flocs structure. Water Science and Technology, 36(4), 313–320.
Sørensen J., Thornberg D. and Nielsen M. K. (1994). Optimisation of a nitrogen-removing biological wastewater
treatment plant using on-line measurements. Water Environment Research, 66(3), 236–242.
Spanjers H. and van Lier J. B. (2006). Instrumentation in anaerobic treatment – research and practice. Water Science and
Technology, 53(4–5), 63–76.
Sperandio M., Urbain V., Audic J. M. and Paul E. (1999). Use of carbon dioxide evolution rate for determining
heterotrophic yield and characterising denitrifying biomass. Water Science and Technology, 39(1), 139–146.
Standard Methods for the Examination of Water and Wastewater (1998). 20th edn, American Public Health
Association/American Water Works Association/Water Environment Federation, Washington, DC, USA.
Stanier R. Y., Adelberg E. A. and Ingraham J. L. (1976). General Microbiology. 4th edn, Prentice–Hall, Upper Saddle
River, NJ, USA.
Steyer J. P., Buffiére P., Rolland D. and Moletta R. (1999). Advanced control of anaerobic digestion processes through
disturbances monitoring. Water Research, 33(9), 2059–2068.
Steyer J. P., Bouvier J. C., Conte T., Gras P. and Sousbie P. (2002). Evaluation of a four year experience with a fully
instrumented anaerobic digestion process. Water Science and Technology, 45(4–5), 495–502.
Steyer J. P., Bernard O., Batstone D. J. and Angelidaki I. (2006). Lessons learnt from 15 years of ICA in anaerobic
digesters. Water Science and Technology 53(4–5), 25–33.
Stobbe C. T. (1964). Uber das Verhaltern von Belebten Schlammes in Aufsteigender Wasserbewegung,
Veroffentlichungen, des Institut fur Siedlungswasserwirtschaft der Technischen Hochschule Hannover. (Study
of the Behaviour of Activated Sludge in Effluent Clarification. Institute for Urban Water Management,
Technical Institute of Hannover, Germany.)
Tarong Energy (2000). Macro Energy Audit, Water & Wastewater Infrastructure. Brisbane Water, Brisbane, Australia.
Ternes T. A. and Joss A. (eds) (2006). Human Pharmaceuticals, Hormones and Fragrances. IWA Publishing, London.
Thornberg D., Nielsen M. K. and Andersen K. (1993). Nutrient removal: on-line measurements and control strategies.
Trussell R. S., Merlo R. P., Hermanowicz S. W. and Jenkins D. (2006). The effect of organic loading on process
performance and membrane fouling in a submerged membrane bioreactor treating municipal wastewater. Water
Research, 40(14), 2675–2683; Corrigendum 40(18), 3479.
Trussell R. S., Merlo R. P., Hermanowicz S. W. and Jenkins D. (2007). Influence of mixed liquor properties and aeration
intensity on membrane fouling in a submerged membrane bioreactor at high mixed liquor suspended solids
concentrations. Water Research, 41(5), 947–958.
University of Cape Town, City Council of Cape Town & NIWR. (1984). Theory, Design and Operation of Nutrient
Removal Activated Sludge Processes. Water Research Commission, Pretoria.
US EPA. (1989). Design Manual: Fine Pore Aeration Systems. EPA/625/1-89/023, EPA Office of Research and
Development, Cincinnati.

by guest
References 239
US EPA. (2009). Nutrient Control Design Manual. EPA/600/R-09/012, EPA Office of Research and Development,
Cincinnati.
Vesilind P. A. (1974). Treatment and Disposal of Wastewater Sludges. Ann Arbor Science, Ann Arbor.
Wanner O., Panagiotidis V., Clavadetscher P. and Siegrist H. (2005). Effect of heat recovery from raw wastewater on
nitrification and nitrogen removal in activated sludge plants. Water Research, 39(19), 4725–4734.
Wastewater Futures. (2007). Secondary Treatment Augmentation Project, ActewAGL Project No. WC-20045, Sludge
Settleability Study Stage 2. Actew AGL.
Wentzel M. C., Ekama G. A., Dold P. L. and Marais G. v. R. (1990). Biological excess phosphorus removal – steady
state process design. Water SA, 16(1), 29–48.
Wentzel M. C., Ekama G. A., Loewenthal R. E., Dold P. L. and Marais G. v. R. (1989). Enhanced polyphosphate
organism cultures in activated sludge systems. Part 2: Experimental behaviour. Water SA, 15(2), 71–88.
Wentzel M. C., Mbewe A. and Ekama G. A. (1995). Batch test for measurement of readily biodegradable COD and
active organism concentrations in municipal waste waters. Water SA, 21(2), 117–124.
White M. J. D. (1975). Settling of Activated Sludge. Tech Report TR11, Water Research Centre.
White M. J. D. (1976). Design and control of secondary settlement tanks. Water Pollution Control, 75, 459–467.
Wilderer P. A., Irvine R. L. and Goronszy M. C. (eds.) (2001). Sequencing Batch Reactor Technology, Sc & Tech Rpt
No. 10, IWA Publishing, London.
Wilen B.-M., Lumley D., Mattsson A. and Mino T. (2010). Dynamics in flocculation and settling properties studied at a
full-scale activated sludge plant. Water Environment Research, 82(2), 155–168.
Wood D. K. and Tchobanoglous G. (1975). Trace elements in biological waste treatment. Journal of WPCF, 47(7),
1933–1945.
WSAA. (1999). The Australian Urban Water Industry: WSAAfacts’99. Water Services Association of Australia,
Melbourne.
Yamamoto R. I., Komori T. and Matsui S. (1991). Filamentous bulking and hindrance of phosphate removal due to
sulfate reduction in activated sludge. Water Science and Technology, 23, 927–935.

by guest
Index
A sludge dewaterability, 144

Abattoir wastewater treatment, 130, 202–203 sludge settleability, 144
Activated sludge history, 27 sludge stabilisation (see specific oxygen uptake rate)
Aeration (See also Reactors, oxidation ditch, aerator specific oxygen uptake rate (SOUR):
characteristics) performance, 138
diffusers: test procedure, 140
aeration characteristics, 98 Alkalinity & pH, 67–69
cleaning, 99 aerobic digester:
fouling, 97 alkali dosing, 140, 145
pressure loss, 97 bioreactor:
oxygen transfer: alkali dosing, 99–100
effect on mixed liquor pH, 99–100 effect of aeration efficiency on mixed liquor
efficiency, 99 pH, 99–100
transfer rate, 96, 184 effects of alkalinity and dissolved CO2 on mixed
Aerobic digesters: liquor pH, 68
aeration, 139–140, 142, 145 Alum dosing (See also Chemicals):
alkali dosing, 140–145 aerobic digesters (see Aerobic digesters, alum dosing)
alum dosing, 140–145 bioreactors:
denitrification rate, 48 effects on:
description, 137 alkalinity and pH, 64, 65
flowsheet, 137 effluent chlorine demand, 66
interaction with liquid stream effluent SS, 61, 130
processes, 137 effluent true colour, 61
nutrient release, 140 floc size, 130
operating principles, 145 nitrification, 44, 66
pH, 144 organic N removal, 66
residence time: phosphorus removal, 60–65
hydraulic, 138 simultaneous nitrification and denitrification, 130
solids, 138 sludge settleability, 80, 130

by guest
242 Index
Alum dosing (Continued) Molasses, 229

estimation of biological phosphorus Sucrose, 229
removal, 64, 66, 157 Soda ash, 229
process response time, 80 Sodium hypochlorite, 229
solids production, 64–65 Clarifiers:
rules of thumb, 69 primary:
effect on influent characteristics, 23
B performance, 23
Bacterial composition, 38 secondary (See also Effluent characteristics;
Belt filter press, 137, 144, 146–149 Nitrification, effects of clarifier sludge mass;
operating and performance characteristics, 146–149 Sludge settleability)
alum, effect of, 144–145 description, 121
test procedure, 147 effluent SS concentration, effects of:
operating guidelines, 147 alum dosing, 130
Biological phosphorus removal (see Phosphorus removal, environmental chemistry, 128
biological) flocculation, 128
Bioreactors, (see Reactors) MLSS concentration, 129
BNR process (see Process, BNR) overflow rate, 132
BNR process formats (see Process formats, BNR) rising sludge, 133
BOD, effect of sample storage, 17 scraper speed, 127
Bottlenecks (see Plants, capacity, bottlenecks) SRT, 128
SSVI, 129, 132
C effluent quality, other effects, 133
Capacity (see Plants, capacity) failure (see Clarifiers, secondary, operating diagrams)
Capacity creep, 199, 204–205, 208 (See also Learning feed wells, effect of, 133
curves) mass balance, 122
Capacity expansion: operating diagrams, 124
integral expansion, 208 rising sludge, 127
parallel expansion, 208–209 solids profile, 122–124
Carbon dioxide stripping, 99 stress testing, 133
Catchment: Contractual startups (see also Learning curves)
nutrient balance, 6 Control, see Process control
Centrifuge, solid bowl, 148–154
operating and performance characteristics, 153 D
operation: Debottlenecking (see Tuning)
guidelines, 154 Denitrification, 47 (See also Simultaneous nitrification
variables, 150 and denitrification)
pool volume, 151 alkalinity, 47
test procedure, 152–153 anoxic mass fraction, 49
Chemicals: carbon dosing, 48
Dosing, see Alum dosing; Alkalinity and pH COD requirement, 47
Properties of following chemicals, Appendix 2 denitrification rate
Alum, 228 aerobic digester, 48
Caustic soda, 229 liquid stream, 48
Ethanol, 228 effects of:
Ferric, ferrous salts, 228 anaerobic reactor, 51
Lime, 228 dissolved oxygen, 50
Lithium hydroxide, 228 mixing regime, 92
Magnesium hydroxide, 228 recycle ratio, 50
Methanol, 228 rules of thumb, 69

by guest
Index 243
Diffusers (see Aeration) L

Digester, aerobic (see Aerobic digesters) Learning curves, 197 (See also Evolutionary operation;
DSVI (see Sludge settleability) Operation; Startup; Tuning)
capacity creep, 199
E contractual startups, 203–204
Effluent characteristics, 23, 25 costs, effect on, 207
disinfected effluent, 26 equations, 199
interactions, 25 football, 198
suspended solids concentration (see Alum dosing, limestone production, 198
bioreactors, effects on effluent SS; Clarifiers, performance creep, 200–203, 205
secondary, effluent SS concentration) phases:
variability, 26, Appendix 1 startup, 200–204
Energy consumption (See also Plants, energy long-term, 204–205
consumption) time constant (see Time constants)
water businesses, 164–165 Lime (see Chemicals)
water cycle, 164 Lithium hydroxide (see Chemicals)
Ethanol (see Chemicals)
Equivalent persons (see Human body; Sewage) M
Evolutionary operation (EVOP), 1, 197–221 Magnesium hydroxide (see Chemicals)
(See also Tuning) Mass balances, 155 (See also Reactors)
application: biological P removal, 157
chemical process plants, 204–206 plant, 155
manufacturing plants, 1, 206 reactors:
STPs, 197–221 Johannesburg, 102
costs, effect on, 207–209 MBR, 112
data collection and interpretation, 210–213 Modified UCT, 101
monitoring schedules, 211–212 UCT, 101
principles, 206, 220 Westbank, 102
procedures, 209 secondary clarifiers, 122, 155
trend plots, 214 sludge dewatering, 145, 157
EVOP (see Evolutionary operation) sludge thickeners, 145, 156
MBR (see Reactors, membrane bioreactor)
F Membrane bioreactors (see Reactors, membrane
Failure curves (see Plant capacity, bioreactor)
failure curves) Membranes (See also Reactors)
Ferric dosing (see Alum dosing; Chemicals) permeability, 114
First order reactions (see Reactions, first order) chemical dosing, effect of, 114
Flocculation (see Mixing, flocculation) monitoring, 114
Foam, 81 Methanol (see Chemicals)
Formats (see Process formats) Mixing (See also Tracer testing)
arbitrary flow, 91–92
H cells or reactors in series, 91
History (see Activated sludge history) complete mix, 89–91
Human body, mass balance, 9 first order reaction, 91
Human factors, 195 flocculation, 94
intensity, 94
I nitrification, effect on (see Nitrification, effects of,
Influent (see Sewage) mixing regime)
Instrumentation (see Process control, plug flow, 89–91
instrumentation) power (see Mixing, intensity)

by guest
244 Index
Mixing (Continued) dissolved oxygen, 58

regime, 89 influent TKN:COD ratio, 58
residence time distribution (RTD), 89–90 nitrate, 58
response time (see Time constants) PAO-GAO competition, 59
series, cells or reactors (see Mixing, cells or reactors pH, 59
in series) RBCOD, 56
time constant (see Time constants) SRT, 55
velocity gradient (see Mixing, flocculation) substrate composition, 59
Molasses (see Chemicals) sulfide, 58
rules of thumb, 69
N Phosphorus removal, chemical (see Alum dosing)
Nitrification, 40 Plants:
effects of: capacity:
alum dosing, see Alum dosing bottlenecks, 158
ammonia concentration, 42 capacity envelopes:
biological phosphorus removal, 44 continuous flow plants, 160
clarifier sludge mass, 45 SBRs, 162
dissolved oxygen concentration, 43 clarifier stress testing, 133
load variability, 44 debottlenecking, 207
mixing regime, 92 failure curves, 159
pH, 43 failure frequency analysis, 159
salinity, 44 flow distribution, 158
SRT, 41 stress testing, 133, 206
temperature, 42 energy consumption (See also Energy consumption)
unaerated mass fraction, 45 energy balance, 165
rules of thumb, 69 energy models, 166
Nitrification capacity, 46 historical data, 167
Nitrogen cycle, 40 effect of plant load, 167–168
Nitrogen removal, rules of thumb, 69 MBRs, 167–169
Nutrient cycle, 1 Tuning, 170
Nutrient requirements (see Process, BNR, nutrient loading (see Sewage)
requirements) mass balances, 155–157
named in this book:
O Albury, NSW, 167
Operation (See also Evolutionary operation; Learning Banora Point, NSW, 167
curves; Tuning) Beenyup, WA, 17
operating goals, 3 Bendigo, Vic, 167
operating cycle, 3 Brendale, Qld, 167
Oxidation ditch (see Reactors) Brunswick Valley, NSW, 82, 106, 143–144,
Oxygen demand (See also Process, BNR, oxygen demand) 148–149, 167
dynamic loading, effect of, 53 Bucasia, Qld, 78
Bushkoppie, South Africa, 77
P Cleveland Bay, Qld, cover, 7, 9, 60, 108–109, 114,
Performance creep (see Learning curves) 169, Appendix 1
Person (see Human body) Coolum, Qld, 78–80, 167
pH (see Alum dosing; Alkalinity & pH) Daspoort, South Africa, 77
Phosphorus removal, biological, 53 Eastern, Vic, 132
effects of: Flagstone, Qld, 78
anaerobic mass fraction, 55 Gibson Island, Qld, 167
cells in series, 55 Goudkoppies, South Africa, 77

by guest
Index 245
Jimboomba, Qld, 78, 81, 129 PID control, 179

Johannesburg, South Africa, 84 Recycle control, 193
Loganholme, Qld, 77, 133 Process formats, BNR:
Lower Molonglo, ACT, 123–124, 129 Bardenpho, 83
Maroochydore, Qld, 6–7, 12, 26, 66, 80, 130, basic, 28
167–168, 201–202 Johannesburg, 84
Nambour, Qld, 26, 167 MBR, 85
Noosa Coastal, Qld, 167 MLE, 48, 83
Northern, South Africa, 77 Modified Bardenpho, 84
Olifantsvlei, South Africa, 77 Modified UCT, 84
Penrith, NSW, 167 Phoredox, 84
Port Pirie, SA, 7, 17, 18 UCT, 84
Redcliffe, Qld, 9, 13, 26, 128 VIP (see Modified UCT)
Rynfield, South Africa, 77 Westbank, 85
South Windsor, NSW, 167
St Marys, NSW, 167 R
Subiaco, WA, 17 Rainfall (see Catchment; Sewage, flow characteristics)
Thorneside, Qld, 58, 167, 201–202 Return activated sludge (RAS) (see Clarifiers, secondary)
Vlakplaats, South Africa, 77 Reactions, first order, 91
Wacol, Qld, 95, 167 Reactors (See also Mass balances; Mixing)
West Byron, NSW, 66, 78, 81, compartmentalized, 100
129–130, 141–144 operating principles, 103
West Wodonga, Vic, 167 process characteristics, 100
Westbank, Canada, 85 membrane bioreactor, 110 (See also Membranes)
Woodford, Qld, 7, 8 operating principles, 115
Primary clarifiers (see Clarifiers, primary) process characteristics, 111
Primary treatment (see Clarifiers, primary) oxidation ditch, 103
Process, BNR: aerator characteristics
biological phosphorus removal (see Biological horizontal shaft, 109
phosphorus removal) vertical shaft, 109
biological process fundamentals, 27 diffused air, 106
denitrification (see Denitrification) operating principles, 110
nitrification (see Nitrification) process characteristics, 103
nutrient requirements, 38 sequencing batch reactor, 28, 115 (See also
oxygen demand, 39, 52 Simultaneous nitrification and denitrification)
diurnal variation, 53 operating principles, 120
sludge mass, 36, 57 process characteristics, 116
kinetics, 29 series (see Mixing, cells or reactors in series)
F:M ratio, 35 Residence time distribution (see Mixing, cells or
SRT, 32 reactors in series)
Process balances (see Mass balances) Response time (see Time constants)
Process control, 171 Rules of thumb
Anaerobic digester control, 194 nutrient removal, 69
Chemical control, 193 oxygen demand, diurnal variation, 53
Control priorities, 173
Controller tuning, 180 S
Costs, operation, minimizing, 195 Salinity (see Nitrification, effects of, salinity;
DO control, 188 Sewage, salinity)
Human factors, 195 Sampling (see Sewage, sampling)
Instrumentation, 183 SBR (see Reactors, sequencing batch reactor)

by guest
246 Index
Scum (see Foam) DO concentration, 75

Secondary clarifiers (see Clarifiers, secondary) feed quality, 75, 79
Settleability (see Sludge settleability) microbubbles, 124
Sequencing batch reactors (see Reactors, sequencing prefermentation, 79
batch reactor) reactor mixing characteristics, 76
Sewage (See also Catchment; Human body) selectors, 76
COD fractionation, 19, 24 solids retention time, 75
energy content, 22 test procedures, 72
flow characteristics, 6 SND (see Simultaneous nitrification &
load characterization, 15 denitrification)
pathogens, 15 Soda ash (see Chemicals)
per capita loading, 9, 22 Sodium hypochlorite (see Chemicals)
quality characteristics, 9–22 Solids retention time (see Aerobic digesters, SRT control;
effect of primary treatment, 23 Process, BNR, SRT)
sampling, 13, 14, 17 Specific oxygen uptake rate (see Aerobic digesters,
salinity, 9, 18 specific oxygen uptake rate)
sewer transformations, 23 Startup, 36 (See also Learning curves)
variability, 12, 18, Appendix 1 Statistics, Appendix 1
Sewage treatment plants (see Plants) Sucrose (see Chemicals)
Sewer transformations (see Sewage, sewer Sugar (see Chemicals)
transformations) SVI (see Sludge settleability)
Simultaneous nitrification & denitrification
(SND), 51, 130 T
effect of alum dosing, 130 Time constants, 94, 199–200
Sludge dewatering (see Belt filter press; Centrifuge) Tracer testing, 92–93 (See also Mixing)
Sludge digestion (see Aerobic digesters) Trend plots, 214–220
Sludge flocculation (see Clarifiers, secondary, effluent Tuning (See also Evolutionary operation; Learning
SS concentration, effects of, flocculation) curves; Operation)
Sludge settleability (See also Aerobic digesters, sludge debottlenecking, 207
settleability) goals, 1, 3
effect on effluent SS, 81 principles, 220
relationships between settleability parameters
(DSVI, SSVI, sSVI, SVI), 74
settling rates, 73 V
factors affecting: Variability:
alum/ferric dosing, 80 (See also alum dosing) Effluent (see Effluent characteristics, variability)
ammonia:nitrate ratio, 78 Influent (see Sewage, variability)
anoxic fraction, 76
denitrification (see Sludge settleability, settling W
rates, factors affecting, microbubbles) Wastewater (see Sewage)

by guest
Tuning Biological Nutrient Removal Plants increases interest in tuning to enhance both
performance and capacity, to provide insight into typical plant operating characteristics, and to
stimulate operators’ interest in studying the behaviour of their own plants. the book focuses
on understanding of plant behavioural characteristics so that optimum performance can be
achieved and maintained.
Tuning Biological Nutrient Removal Plants is carefully organized to cover:

• influent and effluent characteristics;
• process fundamentals;
• individual process characteristics;
• overall plant characteristics;
• the evolutionary operation approach to tuning.
the approach is practical and the use of mathematics is kept to a minimum and information
is supplied in graphical and tabular form. real operating data from a wide range of plant
experiences is included. the book draws on the generosity of many australian plant owners in
permitting their plant data to be incorporated. not all process types are covered but the tuning
principles expounded are universally applicable.
the capacity and performance capabilities of a plant are not fixed; both are amenable to
on-going enhancement through systematic and enthusiastic effort. the book will help to set
new benchmarks in plant operation.
Tuning Biological Nutrient Removal Plants is a valuable resource for sewage treatment operations
and operations support personnel, sewage process design engineers - operating authorities,
consultants, contractors, operators of industrial wastewater treatment plants and sewage
treatment lecturers in chemical engineering departments and other training organisations.
iwapublishing.com
@IWAPublishing
ISBN: 9781780404820 (Paperback)
9 781780 404820
ISBN: 9781780404837 (eBook)

by guest

By Guest On 30 August 2023

Uploaded by

Copyright:

Available Formats

By Guest On 30 August 2023

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

By Guest On 30 August 2023

Uploaded by

Copyright:

Available Formats

©2021 The Author(s)

This title was made available Open Access through a

IWA Publishing would like to thank all of the libraries for

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

First published 2013

British Library Cataloguing in Publication Data

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

About the Authors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

2.3.12 Total COD fractions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

6.5 Membrane Bioreactors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

9.4 Energy Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

F O2, CO2, N2 IN AIR &

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Figure 1.2 Plant operating cycle.

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

2.1 THE CATCHMENT

Figure 2.1 The simple sewerage catchment.

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

2.2 FLOW CHARACTERISTICS

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Daily Flow (normalised to median)

Figure 2.3 Flow frequency distributions.

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

% greater than % less than

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Redcliffe-Weekday average & range Weekend ave CB ave

2.3 MASS LOADING CHARACTERISTICS

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Table 2.1 Mass balance on a 65 kg adult (light activity)1.

Component Whole Intake Output

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Table 2.1 Mass balance on a 65 kg adult (light activity)1 (Continued).

Component Whole Intake Output

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Daily Flow (ML/d)

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Redcliffe-COD SS TKN TP VFA CB-COD

Redcliffe-COD TKN TP CB-COD

2.3.1 Connected population

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Figure 2.10 Comparison of time-based and flow-proportioned sampling results.

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf

Table 2.2 Typical plant load characterisation.

Parameter (& associated Symbol Typical values

Downloaded from http://iwaponline.com/ebooks/book-pdf/869444/wio9781780404837.pdf