© 2024 IWA Publishing

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 the libraries for

pledging to support the transition of this title to Open
Access through the 2024 KU Partner Package program.
Experimental Methods for
Membrane Applications
in Desalination and Water Treatment
Sergio G. Salinas-Rodríguez
Loreen O. Villacorte
Experimental Methods for Membrane Applications in

Desalination and Water Treatment

 Experimental Methods for
Membrane Applications in
Desalination and Water Treatment

SERGIO G. SALINAS-RODRÍGUEZ

LOREEN O. VILLACORTE
Published by: IWA Publishing
Unit 104 – 105, Export Building
1 Clove Crescent
London E14 2BA, UK
Telephone: +44 (0)20 7654 5500
Fax: +44 (0)20 7654 5555
Email: [email protected]
Web: www.iwapublishing.com

First published 2024

© 2024 IWA Publishing

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 licences
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 IWA Publishing and should not be acted upon without independent consideration and professional
advice. IWA and IWA Publishing 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

Library of Congress Cataloguing in Publication Data

A catalogue record for this book is available from the Library of Congress

Reference:
Salinas Rodriguez SG, Villacorte LO (2024) Experimental Methods for Membrane Applications in
Desalination and Water Treatment, 1st edn IWA Publishing, London. doi: 10.2166/9781789062977

Cover design: Hans Emeis

Graphic design: Hans Emeis

ISBN 9781789062960 (Hardback)

ISBN 9781789062977 (eBook)
ISBN 9781789062984 (ePub)

This is an Open Access book distributed under the terms of the Creative Commons Attribution Licence
(CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with
no derivatives, provided the original work is properly cited (https:// creativecommons.org/licenses/
by-nc-nd/4.0/). This does not affect the rights licensed or assigned from any third party in this book.
 Foreword

Experimental Methods for Membrane Applications in Desalination and Water Treatment

Water

Few other substances are so abundant on our beautiful planet that they would be able to
cover it in a layer more than three kilometers thick. Seen from space, our planet is blue and
white from water.

Still, water scarcity is a grave and global issue, because water is often not in the right place, at
the right time, and in the right quality. This book is dedicated to the last issue, water quality.
More specifically, the focus is on experimental membrane processes in water treatment
(this, of course, you will have picked up from the book title). Membrane processes in water
treatment are literally as old as life itself, but still a vibrant experimental field, as will be clear
when you enjoy the book.

As a technology, membrane filtration is highly effective, proven to be able to mitigate the

increasingly global challenges of water scarcity and limited access to clean water. Depending
on membrane type, the filtration process can remove a wide range of water contaminants,
making it uniquely suitable for purifying unconventional but abundant water sources such as
seawater, highly polluted surface or groundwater, and various types of wastewater. As water
scarcity impacts billions of people globally, thousands of membrane-based purification
plants have been planned or installed in both developed and developing regions. This
means that plant engineers and operators who have process and analytical knowledge of
membrane technology are urgently needed. Researchers are also needed to further improve
the sustainability and economic feasibility of the technology.

Unfortunately, knowledge on membrane

processes is currently fragmented in various
academic publications, most of which are
not freely available to operators, engineers
and researchers, particularly in developing
countries. This book aims to address this
critical issue by bringing it all together in
a series of chapters written by some of the
foremost experts in the field.

The Grundfos Foundation is proud to

co-sponsor this book.

Poul Toft Frederiksen

Head of Programme, Research and Learning,
The Grundfos Foundation

V
VI
 Contributors

Chapter(s)

Aamer Ali, PhD, MSc, Assistant Professor 5

Center for Membrane Technology, Department of Chemistry and Bioscience,
Aalborg University, Denmark

Adam C. Hambly, PhD, Senior Researcher Water Technology & Processes 12

Dep. of Environmental and Resource Engineering, Technical University of Denmark, Denmark

Alberto Tiraferri, PhD, MSc, Full Professor of Applied Environmental Engineering 4

Department of Environment, Land and Infrastructure Engineering, Politecnico di Torino, Italy

Almotasembellah Abushaban, PhD, MSc, Assistant Professor in Water Desalination 1, 7, 15

The Applied Chemistry and Engineering Research Center of Excellence,
Mohammed VI Polytechnic University, Morocco

Barun Lal Karna, ME (Research), Assistant Chief Engineer, Sofitel Sydney Darling Harbour 11
Water Research Centre, School of Civil and Environmental Engineering,
University of New South Wales, Australia

Cejna Anna Quist-Jensen, PhD, MSc, Associate Professor of Membrane Technology 5

Center for Membrane Technology, Department of Chemistry and Bioscience,
Aalborg University, Denmark

Claus Hélix-Nielsen, PhD, MSc, Professor DTU Sustain 4

& President Danish Natural Sciences Academy
Head of Dep. of Environmental and Resource Engineering, Technical University of Denmark, Denmark

Francisco Javier García Picazo, MEng, Assistant Chemist 19

Environmental Chemistry Services, City of San Diego, United States of America

Guillem Gilabert-Oriol, PhD, MSc, Research and Development Leader 2, 3

DuPont Water Solutions, Spain, Adjunct Professor in the Universitat Rovira i Virgili, Spain

Gustavo A. Fimbres Weihs, PhD, Lead Research Fellow 19

School of Chemical and Biomolecular Engineering, The University of Sydney, Australia

Helen Rutlidge, PhD, BSc, Lecturer 11

School of Chemical Engineering, University of New South Wales, Australia

Helga Calix Ponce, MSc, Researcher 13

Denmark

Irena Petrinic, PhD, MSc, Associate Professor 4

University of Maribor, Slovenia

Jan Frauholz, MSc, Process Engineer 4

Aquaporin A/S, Denmark & RWTH Aachen, Germany

Javier Rodriguez Gómez, Laboratory supervisor 18

Genesys and PWT brands, H2O innovation, Spain

VII
Jia Xin Tan, BEng 19
Faculty of Chemical and Process Engineering Technology,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

Johannes S. Vrouwenvelder, PhD, MSc, Professor of Environmental Science and Engineering, 17

Director of Water Desalination and Reuse Center (WDRC)
Biological & Environmental Science & Engineering Division (BESE), King Abdullah University
of Science and Technology (KAUST), Saudi Arabia
Delft University of Technology , Faculty of Applied Sciences, Department of Biotechnology ,
The Netherlands

Karima Bakkali, BSc, Research Engineer 15

Mohammed VI Polytechnic University, Morocco

Kathleen Foo, MSc, PhD Research Fellow 19

Faculty of Chemical and Process Engineering Technology,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

Léonie Le Bouille, MSc, Research Fellow 15

IHE Delft Institute for Water Education / CIRSEE Suez / Delft University, The Netherlands / France

Loreen O. Villacorte, PhD, MSc, Lead Water Treatment Specialist 1, 13

Global Technology and Innovation, Grundfos Holding A/S, Denmark

Luca Fortunato, PhD, MSc, Research Scientist 17

Water Desalination and Reuse Center (WDRC), Biological & Environmental Science & Engineering
Division (BESE), King Abdullah University of Science and Technology (KAUST), Saudi Arabia

Lucia Ruiz Haddad, MSc, PhD Candidate 14

Environmental Science and Engineering Program, Biological and Environmental Science and
Engineering (BESE) Division, King Abdullah University of Science and Technology (KAUST), Thuwal,
Kingdom of Saudi Arabia

Maria Salud Camilleri-Rumbau, PhD, MSc, Researcher, R&D Project Manager 4

Technology Centre of Catalonia - Fundació Eurecat, Spain / Aquaporin A/S, Denmark

Mohamed Chaker Necibi, PhD, Associate Professor in Circular Economy 15

International Water Research Institute (IWRI), Mohammed VI Polytechnic University, Morocco

Mohamed Fauzi Haroon, PhD, Associate Director Analytical Science & Technology 14
Moderna, United States of America

Mohammad Mahdi A. Shirazi, PhD, MSc, Senior Postdoctoral Fellow 5

Center for Membrane Technology, Department of Chemistry and Bioscience,
Aalborg University, Denmark

Mohaned Sousi, PhD, MSc 16

Water Supply, Sanitation and Environmental Engineering Department,
IHE Delft Institute for Water Education, The Netherlands

Morten Lykkegaard Christensen, PhD, MSc, Associate Professor of Wastewater Treatment 2

and Membrane Technology Department of Chemistry and Bioscience,
Aalborg University, Denmark

VIII
Muhammad Ali, PhD, Martin Naughton Assistant Professor in Environmental Microbiology 14
Department of Civil, Structural & Environmental Engineering, Trinity College Dublin,
The University of Dublin, Ireland

Muhammad Nasir Mangal, PhD, MSc 10

Membrane specialist, Berghof Membranes, The Netherlands

Nuria Peña García, PhD, MSc, Research Director 9, 18

Director of Scientific Global Services for Genesys and PWT brands, H2O innovation, Spain

Pascal E. Saikaly, PhD, MSc, Professor of Environmental Science and Engineering, 14

Chair of Environmental Science and Engineering Program, Biological and Environmental Science
and Engineering (BESE) Division, King Abdullah University of Science and Technology (KAUST),
Thuwal, Kingdom of Saudi Arabia

Pierre Le-Clech, PhD, MSc, Associate Professor of Water and Wastewater treatment 11
School of Chemical Engineering, University of New South Wales, Australia

Poul Toft Frederiksen, PhD, MSc, Head of Programme Research and Learning F
The Grundfos Foundation, Denmark

Pouyan Mirzaei Vishkaei, MSc 1

Researcher, Water Supply, Sanitation and Environmental Engineering Department,
IHE Delft Institute for Water Education, The Netherlands

Rita Kay Henderson, PhD, MSc, Professor of Water Quality and Treatment 11
School of Chemical Engineering, University of New South Wales, Australia

Sergio G. Salinas-Rodriguez, PhD, MSc, Associate Professor of Water Supply Engineering 1, 7, 8, 15

Water Supply, Sanitation and Environmental Engineering Department,
IHE Delft Institute for Water Education, The Netherlands

Steven J. Duranceau, PhD, PE, Professor and Director 6

Environmental Systems Engineering Institute Department of Civil, Environmental & Construction
Engineering, College of Engineering and Computer Science,
University of Central Florida, United States of America

Urban J. Wünsch, PhD, Postdoctoral Researcher 12

National Institute of Aquatic Resources, Technical University of Denmark, Denmark

Vanida A. Salgado-Ismodes, MSc, PhD research fellow 7

Water Supply, Sanitation and Environmental Engineering Department,
IHE Delft Institute for Water Education, The Netherlands

Victor Augusto Yangali Quintanilla, PhD, MSc, Lead Water Treatment Specialist 4
Global Technology and Innovation, Grundfos Holding A/S, Denmark

Victoria Sanahuja-Embuena, PhD, MSc, Chemical Engineer, Scientist 4

Aquaporin A/S, Denmark

Wen Yew Lam, BEng 19

Faculty of Chemical and Process Engineering Technology,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

IX
Weng Fung Twong, BEng 19
Faculty of Chemical and Process Engineering Technology,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

Xuan Tung Nguyen, BSc, Project Manager 4

Aquaporin Asia, Singapore

Yie Kai Chong, BEng 19

Faculty of Chemical and Process Engineering Technology,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

Yuli Ekowati, PhD, MSc, Postdoctoral Researcher 1, 13

Global Technology and Innovation, Grundfos Holding A/S, Denmark

Yong Yeow Liang, PhD, Senior Lecturer 19

Faculty of Chemical and Process Engineering Technology, Centre for Research in Advanced Fluid
and Processes, Universiti Malaysia Pahang Al-Sultan Abdullah, Kuantan, Pahang, Malaysia

X
About the editors

Sergio G. Salinas-Rodriguez is Associate Professor and

desalination and water treatment technology professional
at IHE Delft Institute for Water Education. He has a PhD
in Desalination and Water Treatment from the Technical
University of Delft, an MSc in Water Supply Engineering
from UNESCO-IHE Institute for Water Education, a Master’s
in Irrigation and Drainage and a BSc in Civil Engineering
from San Simon Major University. He also obtained the
University Teaching Qualification in the Netherlands.

He has over 75 publications in books, chapters, international

peer-reviewed journals and conference proceedings in the
areas of seawater and brackish water desalination, water
treatment, water reuse, and natural organic matter characterization.

Sergio is involved in teaching and curriculum development of the MSc Programme in Water
and Sustainable Development at IHE Delft. His projects comprise capacity building, research
and innovation (e.g., EU-MEDINA, EU-MIDES, EU-MAR2PROTECT). He has mentored more
than 50 MSc students, co-promoted 4 PhD students, and currently supervises 2 PhD students.
Sergio lectures and coordinates several courses on Desalination and membrane technology.

Loreen Ople Villacorte is Lead Water Treatment Specialist at Global Technology and
Innovation in Grundfos Denmark. He has broad experience in research, conceptualization,
development and validation of water treatment technologies including applications of
traditional and emerging membrane technologies.

For the last 18 years, he has held various roles in the academia
and the industry across three countries (Philippines,
Netherlands and Denmark), primarily driving research and
technology development projects to tackle water challenges
in drinking water production and transport, wastewater
treatment or reuse, oil-water separation and industrial
cooling systems. Most of these projects were implemented
through cross-functional collaborations and involves
understanding the physics, biology and chemistry of water
to enable development of effective treatment solutions. He
has published >25 scientific articles and filed numerous
patents in water treatment and desalination applications.

He is a civil engineer with a master’s degree in water supply engineering at IHE-Delft and a
doctoral degree in desalination and water treatment from the Technical University of Delft,
IHE-Delft and Wetsus.

XI
XII
 Contents

Foreword V
Contributors VII
About the editors XI

Chapter 1
Feedwater Quality Guidelines and Assessment Methods
for Membrane-based Desalination 1
1.1 Introduction 1
1.2 Particulate fouling potential 7
1.3 Inorganic fouling and scaling potential 9
1.4 Organic fouling potential 10
1.4.1 Organic carbon 11
1.4.2 UV absorbance and fluorescence 11
1.4.3 LC-OCD 11
1.4.4 TEP 12
1.4.5 Oil and grease 12
1.5 Biofouling potential 13
1.5.1 Bacterial growth potential 13
1.5.2 Assimilable organic carbon 14
1.5.3 Biodegradable dissolved organic carbon 15
1.5.4 Phosphate 15
1.6 Outlook and opportunities 16
1.7 Abbreviations and symbols 17
1.8 References 18

Part 1
Membrane processes 25

Chapter 2
Microfiltration and ultrafiltration 27
2.1 Introduction 27
2.1.1 Advantages of ultrafiltration compared to conventional treatment 28
2.2 Design and optimize membrane processes 29
2.3 Objective of the filtration process 30
2.4 Membrane types 32
2.5 Basic equations 33
2.6 Normalization 35
2.7 Membrane fouling 36
2.8 Sustainable flux 38
2.9 Membrane design and module 40
2.10 Pretreatment 41
2.11 Cleanings 41
2.11.1 Optimization of hydraulic cleaning 42

XIII
2.12 Membrane cascades 43
2.13 Summary 44
2.14 References 45

Chapter 3
Reverse Osmosis and Nanofiltration 47
3.1 The rise of reverse osmosis 47
3.2 Sustainaiblity of reverse osmosis 48
3.3 Understanding the osmosis process 49
3.3.1 Semi-permeable membranes 49
3.3.2 The reverse osmosis process 50
3.4 Equations 52
3.4.1 Fundamental equations 53
3.4.1.1 Osmotic pressure 53
3.4.1.2 Water flux 54
3.4.1.3 Salt transport 55
3.4.1.4 The difference between convective and concentration driven flows 55
3.4.2 System equations 56
3.4.3 Factors affecting membrane performance 58
3.4.3.1 Feed pressure 58
3.4.3.2 Feed concentration 58
3.4.3.3 Feed temperature 59
3.4.3.4 Concentration polarization 59
3.5 Reverse osmosis membranes 59
3.5.1 The significance of desalination 62
3.6 Performance monitoring 62
3.7 Normalization 63
3.7.1 Why normalization matters 64
3.7.2 Equations 64
3.7.2.1 Normalized permeate flow 64
3.7.2.2 Normalized salt rejection 66
3.7.2.3 Normalized pressure drop 66
3.8 Fouling 66
3.8.1 Biofouling 66
3.8.2 Organic fouling 67
3.8.3 Particulate fouling 67
3.8.4 Scaling 68
3.8.5 Integrity failure 68
3.9 References 70

Chapter 4
Forward Osmosis 71
4.1 Introduction: principles of forward osmosis 71
4.2 Materials and experimental set-up 73
4.2.1 Membrane configurations 73
4.2.2 Experimental modes 73
4.2.3 Draw solutions: properties, regeneration, types and selection criteria 76

XIV
4.3 Experimental methods 80
4.3.1 Typical parameters and phenomena 80
4.3.2 FO process design constraints and considerations 82
4.3.3 Best practices 86
4.4 Data analysis: Basic FO process design 87
4.4.1 FO fundamental equations 87
4.4.2 FO module mass balance 89
4.4.3 FO design considerations 91
4.5 Application examples 92
4.6 Outlook 95
4.7 References 96

Chapter 5
Membrane Distillation 97
5.1 Introduction 97
5.2 Materials, experimental set-up 100
5.2.1 MD membranes 100
5.2.1.1 Membrane properties 100
5.2.1.2 Membrane materials 101
5.2.2 Experimental set-up 104
5.2.2.1 MD configurations 104
5.2.3 Process 106
5.2.3.1 MD system 106
5.2.3.2 Operating parameters 107
5.2.4 MD modules 107
5.3 Methods 112
5.3.1 Process measurements and calculations 112
5.3.1.1 Permeate flux 112
5.3.1.2 Solute rejection 112
5.3.1.3 Logarithmic temperature difference 112
5.3.2 Membrane characterization 112
5.3.2.1 Physical and morphology properties 112
5.3.2.2 Chemical properties 120
5.3.2.3 Thermal properties 121
5.4 Applications and examples 122
5.5 Outlook 123
5.6 References 125

Part 2
Particulate fouling 137

Chapter 6
Silt Density Index 139
6.0 Abstract 139
6.1 Development of the fouling index 140
6.2 Silt as a component of membrane fouling 140
6.3 Standardizaton of the silt density index 141

XV
6.4 Methods and procedures 142
6.5 Limitations of the SDI 146
6.6 Alternatives to the SDI 149
6.7 Summary 151
6.8 References 152

Chapter 7
Modified Fouling Index (MFI-0.45) 155
7.1 Introduction 155
7.2 Theory particulate fouling 156
7.3 Measuring MFI-0.45 159
7.3.1 Filtration set-up and materials 159
7.3.1.1 Membrane filters 160
7.3.1.2 Filter holder 160
7.3.1.3 Feedwater reservoir 161
7.3.1.4 Electronic mass balance 161
7.3.1.5 Software for data acquisition 161
7.3.1.6 Pressure regulator and gauge 162
7.3.1.7 Pressure transducer 162
7.3.1.8 Non-plugging water 162
7.3.2 MFI-0.45 testing procedure 163
7.3.3 MFI-0.45 calculation procedure 163
7.4 Membrane properties of commercial membranes 165
7.5 Effect of filter material on MFI-0.45 values 166
7.5.1 Effect of membrane support holder in MFI-0.45 167
7.6 Application: water quality monitoring of North Sea water 168
7.7 Monitoring of MFI-0.45 in a full-scale desalination plant 169
7.8 References 171

Chapter 8
Modified Fouling Index Ultrafiltration (MFI-UF) Constant Flux 173
8.1 Introduction 173
8.2 Theory Particulate fouling 176
8.2.1 Deposition factor 180
8.2.2 The particulate fouling prediction model 181
8.3 Measuring MFI-UF constant flux 181
8.3.1 Filtration set-up and materials 181
8.3.1.1 Membrane filters 182
8.3.1.2 Constant flow pump 183
8.3.1.3 Pressure transducer 184
8.3.1.4 Membrane filter holder 185
8.3.1.5 Syringe 185
8.3.1.6 Ultra-pure water 185
8.3.1.7 Tubing 186
8.3.1.8 Software 186
8.3.2 Membrane cleaning and conditioning 186
8.3.3 MFI-UF testing procedure 187

XVI
 8.3.3.1 Selection of filtration flux rate 187
8.3.4 Calculation procedure 188
8.3.4.1 Example of membrane resistance calculation of UPW 189
8.3.4.2 Example of MFI-UF calculation 190
8.3.5 Reproducibility 191
8.3.6 Blank and limit of detection 191
8.3.7 Sample storage 192
8.3.8 Concentration of particles 193
8.3.9 Membrane material 194
8.4 Variables and applications of the MFI-UF 195
8.4.1 Plant profiling and water quality monitoring 195
8.4.2 Flux rate 196
8.4.3 Predicting rate of fouling of seawater RO systems 197
8.4.4 Comparing fouling indices 198
8.5 References 200

Part 3
Inorganic fouling and scaling 205

Chapter 9
Inorganic Fouling Characterization Tools and Mitigation 207
9.1 Introduction 207
9.2 Main components of inorganic fouling 210
9.2.1 Colloidal matter/particulate 210
9.2.2 Metals 214
9.2.3 Scaling 218
9.2.4 Other components 221
9.3 Methods for inorganic fouling identification 222
9.4 Methods for inorganic fouling removal 225
9.5 References 228

Chapter 10
Assessing Scaling Potential with Induction Time and a
Once-through Laboratory Scale RO System 229
10.1 Introduction 229
10.2 Induction time measurements 231
10.2.1 Experimental setup 232
10.2.1.1 Glass reactor 232
10.2.1.2 Stirrer device 233
10.2.1.3 pH meter 233
10.2.1.4 Peristaltic pump 233
10.2.1.5 Thermostat 233
10.2.2 Experimental procedure 233
10.2.2.1 Preparation of artificial brackish water 233
10.2.2.2 Induction time measurement 234
10.2.3 Calculation of induction time 234
10.2.4 Cleaning of the reactor 235

XVII
 10.2.5 Example of application of induction time 235
10.3 Once through lab-scale RO system 238
10.3.1 Experimental set-up 238
10.3.2 Experimental protocol 240
10.3.3 Example of application 240
10.4 Outlook and final comments 243
10.5 References 245

Part 4
Organic fouling 247

Chapter 11
Practical Considerations of Using LC-OCD for Organic Matter 249
Analysis in Seawater
11.1 Introduction 249
11.2 LC-OCD analysis 252
11.2.1 Instrumentation and chromatogram integration 252
11.2.2 Effect of salinity on organic characterization and calibration 254
11.2.3 Level of detection 257
11.2.4 Reproducibility of LC-OCD 257
11.2.5 Characterisation of organic mixtures 259
11.2.6 Applications 259
11.2.6.1 OM composition in seawater 260
11.2.6.2 Fouling behaviour of organic matter 261
11.2.6.3 Effectiveness of pretreatment methods 261
11.3 Conclusions 261
11.4 References 262

Chapter 12
Fluorescence Excitation Emission Matrix (EEM) Spectroscopy 265
12.1 Introduction 265
12.2 Sampling & storage 267
12.3 Benchtop instrumentation 268
12.4 Quality assurance 270
12.5 Interferences 271
12.6 Data processing 272
12.7 Data analysis 273
12.7.1 PARAFAC 275
12.8 Application in membrane systems 275
12.9 References 281

Chapter 13
Transparent Exopolymer Particles 287
13.1 Introduction 287
13.2 Quantification methods 290
13.2.1 Alcian blue dye preparation 292
13.2.2 TEP0.4μm measurement 293

XVIII
 13.2.3 TEP10kDa measurement 297
13.2.4 Method calibration 300
13.2.4.1 Xanthan gum standard preparation 300
13.2.4.2 TEP0.4μm calibration 1 301
13.2.4.3 TEP0.4µm calibration 2 301
13.2.4.4 TEP10kDa calibration 302
13.2.5 Other considerations 302
13.2.5.1 Limit of detection 302
13.2.5.2 Impact of storage on TEP concentration 304
13.2.6 Application and interpretation 304
13.3 Summary and outlook 309
13.4 References 310

Part 5
Biological fouling 313

Chapter 14
Genomics Tools to Study Membrane-Based Systems 315
14.1 Introduction 315
14.2 Experimental design and sample preparation 318
14.2.1 Experimental design in a metagenomics 318
14.2.2 Sample collection and preservation 319
14.2.3 DNA extraction 319
14.2.4 Library preparation 320
14.2.5 Sequencing platforms 320
14.3 Bioinformatics analysis 321
14.3.1 Data pre-treatment 321
14.3.2 Amplicon-based approach 321
14.3.3 Metagenomics, read-based approach 322
14.3.4 Metagenomics, assembly-based approach 322
14.3.5 Metagenome-assembled genome (MAG) binning 322
14.3.6 Supervised and unsupervised binning 323
14.3.7 Functional annotation 323
14.3.8 Genome-resolved metatranscriptomics 323
14.4 Data sharing and storage 324
14.5 Bioinformatics analysis workflow examples 324
14.5.1 Amplicon sequences processing workflow 324
14.5.2 Genome-resolved metagenomics 325
14.5.3 Genome-resolved metatranscriptomics 327
14.6 Applications of genomics in membrane filtration research 329
14.7 Outlook 330
14.8 Data availability 331
14.9 References 332

Chapter 15
Biofouling Potentia Measurement using Bacterial Growth Potential (BGP) 337
15.1 Introduction 337

XIX
15.2 Materials 338
15.2.1 Laboratory equipment 338
15.2.2 Chemicals 339
15.2.3 Instrumental equipment 339
15.3 Methods and experimental procedure 340
15.3.1 Sample collection and storage 340
15.3.2 Cleaning glassware 341
15.3.3 Preparation of artificial seawater 341
15.3.4 Intact cell count by flow cytometry 342
15.3.5 Measurement of bacterial growth potential 342
15.3.6 Bacterial yield and calibration line 343
15.4 Applications 345
15.4.1 Example A: BGP monitoring of an SWRO pre-treatment 345
15.4.2 Example B: BGP in the intake and SWRO feed water 345
15.5 Data discussion and interpretation 346
15.6 ATP measurement 347
15.6.1 Introduction 347
15.6.2 Material and methods 348
15.7 References 353

Chapter 16
Assessing Biological Stability of Ultra-low Nutrient Water by Measuring Bacterial
Growth Potential 355
16.1 Introduction 355
16.2 Materials and experimental set-up 357
16.2.1 Equipment 357
16.2.2 Materials and methods 361
16.2.3 Method 361
16.3 Examples of application 364
16.4 Additional considerations 370
16.5 References 372

Chapter 17
Optical Coherence Tomography (OCT) as a Tool for (Bio)-fouling Assessment
in Desalination Systems 375
17.1 Introduction 375
17.2 Materials, experimental set-up 377
17.3 Methods 377
17.3.1 Imaging with optical coherence tomography 377
17.4 Data Analysis 378
17.4.1 Biovolume calculation 378
17.4.2 Image processing 381
17.5 Data discussion and interpretation 382
17.5.1 Biomass quantification 382
17.5.2 Membrane performance 383
17.6 Applications, examples 384
17.6.1 Biomass distribution 384

XX
 17.6.2 Biomass and performance decline 385
17.6.3 Biomass thickness map 386
17.7 Additional considerations 387
17.7.1 OCT image analysis 387
17.7.2 Biomass accumulation and membrane performance 388
17.7.3 Biomass location in the flow channel 389
17.7.4 Use of OCT in biofouling studies 390
17.7.5 Mapping the biofouling 390
17.8 Summary 391
17.9 References 392

Part 6
General applications 397

Chapter 18
Membrane Autopsy 399
18.1 Introduction 399
18.2 Materials, experimental set-up 401
18.3 Membrane autopsy protocol 402
18.4 Methods 403
18.4.1 Visual inspection 403
18.4.2 Analytical methods for foulant and damage identification 408
18.4.3 Membrane performance 416
18.4.4 Cleaning tests 418
18.5 References 420

Chapter 19
CFD as a Tool for Modelling Membrane Systems 421
19.1 Introduction 421
19.1.1 What is not modelled 423
19.1.2 How modelling can assist membrane systems 423
19.2 Methods 424
19.2.1 Geometry 425
19.2.2 Flow types 427
19.2.2.1 1D, 2D and 3D 427
19.2.2.2 Laminar, transient, turbulent 427
19.2.2.3 Single phase 428
19.2.2.4 Multiphase 428
19.2.3 Boundary conditions 431
19.2.3.1 Steady-state and transient-state 432
19.2.4 Initial conditions 432
19.2.5 Meshing and algorithms 433
19.2.6 Convergence 434
19.3 Data Analysis 434
19.3.1 Verification 434
19.3.2 Validation 436
19.4 Data discussion and interpretation 437

XXI
 19.4.1.1 Data processing and assessment 437
19.5 Applications, examples 438
19.5.1 Flow stability 438
19.5.1.1 Laminar steady 438
19.5.1.2 Laminar unsteady – oscillating vs. vortex shedding 438
19.5.1.3 Quasiperiodic flow 439
19.5.1.4 Turbulent flow 441
19.5.2 Mass transfer and vortex shedding 441
19.5.3 Spacer design 442
19.5.3.1 Two-dimensional feed spacer 442
19.5.3.2 Three-dimensional feed spacer 442
19.5.4 Flow perturbation 443
19.5.4.1 Electro-osmosis 444
19.5.4.2 Modelling electro-osmosis in CFD 445
19.5.4.3 Significant learnings of EOF slip velocity in CFD studies 446
19.5.4.4 Oscillating flow 447
19.5.4.5 Vibrations 449
19.5.5 Fouling modelling 449
19.5.5.1 Particulate fouling 449
19.5.5.2 Tracer test 452
19.5.5.3 Biofouling 452
19.6 Additional considerations 455
19.6.1 Multi-scale modelling 455
19.6.1.1 Techno-economics 456
19.7 Outlook 456
19.8 References 457

XXII
 doi: 10.2166/9781789062977_0001

Chapter 1

Feedwater Quality Guidelines

and Assessment Methods for
Membrane-based Desalination
Loreen O. Villacorte, Grundfos, Denmark

Yuli Ekowati, Grundfos, Denmark

Almotasembellah Abushaban, UM6P, Morocco

Pouyan Mirzaei Vishkaei, IHE Delft, The Netherlands

Sergio G. Salinas-Rodriguez, IHE Delft, The Netherlands

The learning objectives of this chapter are the following:

• To review the existing feedwater quality guidelines for membrane-based

desalination

• To present and discuss the existing and proposed methods for assessing fouling and
scaling potential of feedwater.

1.1 INTRODUCTION

Amidst the global problem of dwindling freshwater water resources, desalination of

unconventional but abundant water resources such as seawater and brackish water has
grown rapidly over the last three decades. From a global operational capacity of ~7.5 million
m3/day in 1990 to ~115 million m3/day in 2023, water desalination technologies have
been the leading solution to address the growing municipal, agricultural and industrial
demand for clean freshwater (Figure 1a and 1b). Furthermore, desalination technologies
are generally applied for triple barrier wastewater reuse applications, which currently has a
global installed capacity of >60 million m3/year (Birch et al., 2023).

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/).
The chapter is from the book Experimental Methods for Membrane Applications in Desalination and
Water Treatment, Sergio G. Salinas-Rodriguez, Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Seawater is the main water source for desalination globally, with the exception of North
America, where the majority of applications is based on brackish water desalination. Japan,
South Korea, Taiwan, and China desalinate seawater, brackish water, and wastewater
effluent at relatively similar rates.

120
Total
Membrame based (RO) 100
Thermal (MSF + MED)
80

60

40

20

Capacity (million m3/d) 0

1970 1980 1980 2000 2010 2020
year

Figure 1a The growth of global desalination application in terms of online production capacity since
1970 (top) and current online seawater desalination by technology, region and end-user.
Produced with information from (DesalData, 2023). MSF = Multi-stage flash distillation,
MED = Multi-effect distillation, RO = Reverse osmosis

In terms of technology, membrane-based desalination using reverse osmosis (RO) dominates

the application (~74% of global capacity). This is mainly driven by the significantly lower
investment cost and energy requirements today are lower than thermal processes (e.g., MSF,
MED). A large majority (>75%) of the desalinated water are used for supplying drinking
water supply while about 20% are used in industries. Most of the Middle East countries rely
on desalination for municipal use, while countries such as China, India, South Korea, Brazil,
Taiwan, Chile, Indonesia use desalination to satisfy industrial demand.

2
 Chapter 1

Global online desalination 2023 (~115 Mm3/d)

Wastewater 9% Brackish water 19%

Seawater 60% Pure water 4%

Fresh water 8%

North America 9% Southern Asia 4%

Sub-Saharan Africa 2%
Western Europe 7%

Middle East / East Asia / Pacific 18%

North Africa 52%
Eastern Europe / Central Asia 42%

Latin America / Caribbean 6%

Irrigation 4% Drinking water 46%

Industry 50%

Seawater desalination 2023 (~70 Mm3/d)

MED 9% MSF 21%

RO 70%

North America 1% Southern Asia 3%

Sub-Saharan Africa 1%
Western Europe 7%
East Asia / Pacific 13%

Middle East / Eastern Europe / Central Asia 2%

North Africa 52% Latin America / Caribbean 4%

Industry (<10ppm) 21%

Irrigation (<1000ppm) 1%
Drinking water 78%
(10 ppm - 1000 ppm)

Figure 1b Current online global desalination (top 3) and online seawater desalination (bottom 3)
by technology, region and end-user. Produced with information from (DesalData, 2023)

3
Experimental Methods for Membrane Applications

Western Europe
Eastern Europe
7.7 Mm3/d
North America 2.3 Mm3/d
9.5 Mm3/d China
10.3 Mm3/d Japan, Korea,
Taiwan 2.0 Mm3/d

Caribbean Rest East Asia / Pacific

1.4 Mm3/d Middle East 2.0 Mm3/d
North Africa 46.7 Mm3/d
India
7.2 Mm3/d
3.4 Mm3/d
Singapore
2.1 Mm3/d
Latin America Sub-Saharan Africa
Seawater 4.5 Mm3/d 1.8 Mm3/d
Brackish water Australia
2.9 Mm3/d
Wastewater

Figure 2 Feedwater sources and (online, in construction) production capacities of desalination

plants in different geographic locations in 2023 (updated from Salinas Rodriguez and
Schippers (2021) with information from DesalData (2023)).

The price per cubic metre of desalinated water has reduced significantly over the years due
to more efficient membrane production, implementation of energy recovery devices, cost
of engineering, etc, and a more competitive market. The specific energy consumption has
already been reduced by at least 50% over the last 20 years and the overall carbon footprint
of desalination could be reduced down further by switching to renewable energy sources
(Birch et al., 2023). On the downside, membrane fouling and scaling are the main ‘Achilles
heel’ for the sustainable application of RO (Voutchkov, 2010, Salinas Rodriguez, 2021).

Fouling and scaling in membranes can lead to a variety of problems, such as the need for
(frequent) chemical cleaning, reduction of production capacity, higher energy consumption,
decrease in produced water quality, that it makes RO production facilities less reliable, and
require more frequent membrane replacement (Dhakal et al., 2020, Salinas Rodriguez et
al., 2021b). Fouling and scaling are broadly categorized into i) particulate/colloidal fouling
due to suspended and colloidal matter, ii) inorganic fouling due to iron and manganese, iii)
organic fouling due to organic compounds e.g., polymers, iv) biofouling due to growth of
bacteria, and v) scaling due to deposition of sparingly soluble compounds.

During RO operation, membrane fouling and scaling may manifest in three ways, namely:
i) increasing the differential pressure across the spacer in spiral wound elements due to
‘clogging’, resulting in potential membrane damage (such as telescoping, channelling, or
squeezing); ii) increasing membrane resistance (or decreasing the normalized permeability)
due to deposition and/or adsorption of materials on the membrane surface, resulting in
higher required feed pressure to maintain capacity; and iii) increasing in normalized salt
passage due to concentration polarization in the fouling layer, resulting in higher salinity in
the product water.

4
 Chapter 1

Particulate and colloidal fouling are mostly well controlled by the pre-treatment systems
(mostly media filtration or membrane filtration), but the occurrence of organic fouling and
biofouling is still a major issue in RO membranes, and is the main reason for the need for
frequent cleaning of the reverse osmosis membranes (Peña et al., 2022).

To minimize the occurrence of membrane fouling/scaling in RO, pre-treatment of the

feedwater is essential. Additionally, methods and tools can help significantly, by monitoring
the performance of the pre-treatment with regards to fouling/scaling control and process
optimization. Pre-treatment can take place in the form of media filters with or without
coagulation, membrane filtration with or without inline coagulation (e.g., ultrafiltration),
and dissolved air floatation in combination with the previous mentioned two options.

Along with the increase in the number of desalination plants (>22,800 plants in 2023),
the capacity of newly installed plants has also increased significantly over time. A growing
preference for extra-large (XL) plants (capacity >50,000 m3/d) has been reported in recent
years (Birch et al. (2023); Kurihara and Ito (2020). More XL seawater RO (SWRO) plants are
expected in the future. This means reliable pre-treatment systems and monitoring tools will
be essential for these XL plants, as Cleaning-in-Place (CIP) of membrane modules more than
once per year is rather challenging. The design and operational settings of such pre-treatment
systems will depend on the water quality and their temporal variations of the water source
alongside the feedwater quality guidelines provided by the membrane supplier.

For a long time, the silt density index (SDI) has served as a sum ‘king/ultimate’ parameter
for assessing RO feed water. DuPont (2020) for the first time introduced the MFI-0.45
in their RO feedwater guidelines. This is a major step forward due to the limitations of
the SDI in assessing fouling in RO (Schippers et al., 2014). In addition, the inclusion of
parameters like AOC and BFR bring relevance for the monitoring of the biofouling potential
of RO feedwater as several types of fouling take place simultaneously. Table 1 presents the
recommended guideline values for RO feedwater by RO manufacturers and literature. The
majority of RO membrane manufacturers are in agreement with the recommendations by
DuPont although they main guideline is the SDI value less than 4-5 and preferable less than
3 for RO feedwater.

5
Experimental Methods for Membrane Applications

Table 1 Parameters and recommended guideline values for RO feed water

Parameter Unit DuPont (2023) Other sources Standard Methods
(a) Particulate fouling indicators
SDI15 %/min <5 (target <3) <5 (target <3) (Wilf and Klinko, (ASTM D4189 - 07)
2016)
< 3 (Badruzzaman et al., 2019)
< 4 (Voutchkov, 2010)
MFI-0.45 s/L2 4 (target <1) (ASTM D8002 - 15)
MFI-UF s/L2 - <490 at 15 lmh (safe MFI*)
(Salinas Rodríguez, 2011)
Turbidity NTU <1 < 0.5 (Badruzzaman et al., 2019) (ASTM D1889-00)
< 0.1 (Voutchkov, 2010)
(b) Organic fouling indicators
Oil and grease mg/L 0.1 < 0.1 (Badruzzaman et al., 2019) (ASTM D7575-11)
< 0.02 (Voutchkov, 2010)
TOC mg-C/L 3 < 2 (Badruzzaman et al., 2019) (ASTM D2579-93e1)
< 2 (target <0.5)
(Voutchkov, 2010)
SUVA L/mg-m < 4 (USEPA, 2005)
COD mg/L 10 (ASTM D1252-
06(2020))
(c) Biological fouling indicators
AOC µg/L Ac-C 10 (target <5) <10 µg-C acetate/L (threshold for (NEN 6271:1995 nl)
biofouling in freshwater) (van der
Kooij et al., 1982)
BGP µg-C/L - - <70 (Abushaban, 2019)
BFR pg-ATP/ 5 (target <1) < 1 (Vrouwenvelder and van der
cm2 Kooij, 2001)
PO4-P µg/L 0.3 µg P/L (Vrouwenvelder et al.,
2010)
(d) Inorganic fouling and scaling indicators
Ferrous iron mg/L 4 < 2 (Badruzzaman et al., 2019) (ASTM D1068-15)
< 2 (Voutchkov, 2010)
Ferric iron mg/L 0.05 < 0.1 (Badruzzaman et al., 2019) (ASTM D1068-15)
<0.05 (Voutchkov, 2010)
Manganese mg/L 0.05 0.05 (Badruzzaman et al., 2019) (ASTM D858-17)
0.02 (Voutchkov, 2010)
Aluminium mg/L 0.05 (ASTM D857-17)
Silica mg/L 20 (Badruzzaman et al., 2019) (ASTM D859-
16(2021)e1)
pH - 4-11 (Voutchkov, 2010) (ASTM D1293-12)
LSI - Concentrate (ASTM D3739-19)
(freshwater) LSI < 0 (if no
antiscalant is
added)

6
 Chapter 1

Parameter Unit DuPont (2023) Other sources Standard Methods

S&DSI - Concentrate (ASTM D4582-
(seawater) S&DSI < 0 (if 91(2001))
no antiscalant is
added)
(e) Membrane material limits

Temperature °C < 35 (Voutchkov, 2010)

Free chlorine mg/L <0.1 < 0.1 (Badruzzaman et al., 2019) (ASTM D1253-
< 0.01 (Voutchkov, 2010) 14(2021)e1)
ORP mV <175-200 (ASTM D1498-
14(2022)e11)
Parameter Unit DuPont (2023) Other sources Standard Methods
* Safe MFI is a value for RO feedwater that will yield a 1 bar pressure increase in a 6 months period.

In addition to the established feedwater quality parameters in Table 1, tens of thousands of

scientific articles and patents were published over the past 30 years describing or applying
new assessment tools/indices for evaluating the fouling/scaling potential of RO feedwater
as well as to characterize the impact of specific feedwater components to RO operation
(Figure 3). Some of these tools were also applied to optimize the design and operation of
RO pre-treatment system, including MF/UF processes (see Chapter 2). The succeeding
sections review the advantages as well as the challenges of applying these assessment tools
in membrane-based desalination systems.

10 100

Cumulative publications (x1,000)

Yearly publications
9 90
Publications (year x1,000)

8 Cummulative publications 80
7 70
6 60
5 50
4 40
3 30
2 20
1 10
0 0
19 0
19 1
19 2
19 3
19 4
19 5
19 6
19 7
19 8
20 9
20 0
20 1
20 2
20 3
20 4
20 5
20 6
20 7
20 8
20 9
20 0
20 1
20 2
20 3
20 4
20 5
20 6
20 7
20 8
20 9
20 0
20 1
20 2
23
9
9
9
9
9
9
9
9
9
9
0
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
2
2
2
19

Figure 3 Number of scientific and patent publications related to fouling and scaling assessment
in reverse osmosis process from 1990 and 2023 (November). Data generated through
Google Scholar using the search string: “(fouling OR scaling) AND (characterization OR
assessment OR potential OR indicator OR index) AND (reverse osmosis)”.

1.2 PARTICULATE FOULING POTENTIAL

Fouling indices to measure the particulate fouling potential of RO feedwater have been
in development since the 1960’s (Figure 4). The oldest and most widely used index, the
silt density index (SDI) has been standardised by ASTM D4189 - 14 (2014), is applied
worldwide as it is simple to perform and with low-cost consumables (see Chapter 6).

7
Experimental Methods for Membrane Applications

However, increasingly the value of this test to predict the rate of fouling in RO systems due to
particle deposition is being questioned. The limitations of the SDI test are well documented
(Schippers and Verdouw, 1980, Nahrstedt and Camargo Schmale, 2008, Alhadidi et al.,
2011a, Alhadidi et al., 2011b, Rachman et al., 2013, Salinas Rodriguez et al., 2019) and
include no correction for feedwater temperature variation (higher SDI values at higher
temperatures); the result is heavily dependent on the permeability of the test membrane
filter; not applicable for testing high fouling feed water e.g., raw water – ASTM recommends
that turbidity should be < 1 NTU; not applicable for testing UF system permeate, which
is increasingly being used in desalination pre-treatment; no linear relation with colloidal/
suspended matter; fouling potential of particles smaller than 0.45 μm are not taken into
account; it is not based on any filtration mechanism.

MFI-UF CF
with flat MFI-0.45
MFI-UF membranes ASTM
constant pressure (10, 30, 50, 100) D8002-15

MFI-UF Flux effect, Improved fouling

constant Flux deposition prediction
with HF factor, (5 kDa membrane,
MFI-0.45 MFI-0.05 membrane safe MFI porosity effect)

1970 1980 1990 2000 2010 2020

FI SDI ASTM SDI ASTM SDI ASTM SDI ASTM

(SDI) D4189-95 D4189-02 D4189-07 D4189-14

Figure 4 Historical development of fouling indices for particulate fouling assessment (adapted from
Salinas Rodriguez et al. (2021a)). Legend: FI = fouling index, SDI = silt density index, MFI
= modified fouling index, MFI-UF = modified fouling index ultrafiltration, CF = constant
flux, ASTM = American society for testing and materials.

The modified fouling index (MFI-0.45) test, also standardized by ASTM D8002 - 15 (2015),
uses the same materials and equipment as the SDI test. It is based in the cake filtration
fouling mechanism. The obtained MFI value is corrected for temperature and pressure and
shows a linear relation with colloidal/suspended matter concentration (see Chapter 7). The
predicted rate of fouling turns out to be very low at a level of MFI = 1 s/L2, which is in the
range of SDI 1 to 3.

Based on the low sensitivity of MFI-0.45, the MFI-UF test with ultra-filtration membranes
was developed to capture these smaller particles (see Chapter 8). The strong dependence of
MFI on flux, means that to be able to predict accurately the potential of particulate fouling
in RO systems, the MFI should preferably be measured at a flux similar to a RO system
(15- 20 L/m2/h) or extrapolated from higher fluxes. A theoretical ‘safe MFI’ was proposed
assuming e.g., an allowable increase in NDP of 1 bar in 6 months (Salinas Rodríguez, 2011,
Salinas Rodriguez et al., 2019). The safe MFI calculated for a deposition factor Ω = 1 (worst
case) at a flux of 15 L/m2/h (average flux in RO), has been reported about 490 s/L2. And
could be used as a threshold value for assessing RO feed water quality. Good correlation with
RO membrane fouling development was observed when applying the MFI-UF prediction
model (Abunada et al., 2023).

8
 Chapter 1

1.3 INORGANIC FOULING AND SCALING POTENTIAL

Inorganic fouling occurs in RO processes due to the deposition of insoluble or sparingly

soluble inorganic compounds from the feedwater to the RO membrane. In many
literatures, inorganic fouling also includes scaling, however, deposition of metal ions, such
as iron, manganese and aluminium is also called as inorganic fouling. Iron, manganese and
aluminium ions are present naturally in surface water and groundwater. Fouling due to these
ions can be reduced by pre-treatment, such as aeration, oxidation, and filtration. Membrane
manufacturers provide guidelines for feedwater quality compatible with the membrane
which typically includes limits on iron, manganese, and aluminium concentrations
(Table 1). Analytical methods to quantify critical inorganic components in the feedwater
or to identify accumulation on the membrane surface are well established, which usually
includes ICP and SEM-EDX. There is no single analytical method than can completely
identify all inorganic foulants so it is important to get familiarized with these methods (see
full list of tools in Chapter 9) when studying inorganic fouling.

Scaling on RO membranes is specifically caused by precipitation and accumulation of

sparingly soluble salts usually as a consequence of up-concentrating these salts in the RO
process, eventually exceeding their solubility limit. The type of scaling depends on the ion
composition, temperature and pH of the feedwater. Scaling issues typically observed in RO
processes are calcium carbonate, calcium sulphate, calcium phosphate, barium sulphate,
and silica/metal silicates. Scaling can be avoided by limiting the RO recovery, acid dosing,
and antiscalant dosing. The latter is the most preferable approach because it does not
compromise the RO production and it is effective for different types of scaling with only
low dose requirement. Pretreatment of feedwater by ion exchange or lime softening can also
be applied to control scaling.

The scaling potential of feedwater in the RO system can be assessed based on scaling indices.
The commonly used indices for RO application are saturation index (SI) and supersaturation
ratio (Sr), Langelier saturation index (LSI), and Stiff-Davis stability index (S&DSI). SI and Sr
are applicable for all types of scaling species while LSI and S&DSI are specifically used for
calcium carbonate scaling. A positive value of the indices generally indicates supersaturation
condition where precipitation can occur. Calculation of saturation indices are available in
standardized ASTM methods (ASTM D4692-01, D3739-19, D4582-23) and in literatures
(e.g., Mangal et al. (2021)). In practice, the scaling potential of RO feedwater are usually
determined using the design software developed by antiscalant chemical suppliers and
membrane manufacturers, such as WAVE (DuPont), IMSDesign (Hydranautics) and
MembraneMaster MM5 (Genesys). A geochemical modelling program, PHREEQC, can also
be used to calculate SI.

An emerging parameter known as induction time has been increasingly used to predict or
study scaling in RO. Scaling in RO occurs when the lattice ions in supersaturated solution
start to agglomerate and form nuclei or clusters. If the size of the cluster is below the critical
size, then the formed crystal will re-dissolve into the solution and if the cluster size is above
the critical size, then the crystal become stable and grow bigger. Induction time is defined
as the time required for the supersaturated solution to form nuclei in the critical cluster size
dimensions just before it becomes stable and starts to accelerate growth (He et al., 1995,

9
Experimental Methods for Membrane Applications

Boerlage et al., 2000, Boerlage, 2001). Induction time can be determined by measuring the
change in pH, turbidity, conductivity, and specific ion concentration in the water over a
period in a controlled lab environment (Waly, 2011, Mangal, 2023). Measuring induction
time is potentially useful tool in developing new design, pretreatment and operational
strategies to control scaling in RO but better understanding is needed to measure impact
of other inorganics as well as organic components in the water on the induction time of
specific salts.

1.4 ORGANIC FOULING POTENTIAL

The accumulation of organic matter in membrane systems can directly cause substantial
decline in operational performance (e.g., permeability). Organic foulants are typically
abundant in surface waters but are also present in ground water sources. They can originate
from natural sources from human activities (e.g., sewage discharge) or from chemicals
used in the pre-treatment processes. Identifying organic foulants and understanding their
characteristics provides valuable insights on how to prevent organic fouling, by choosing
the appropriate pre-treatment and operational and cleaning strategies. Currently available
assessments tools for organic fouling can range from simple spectrophotometric methods
to more advanced chromatographic techniques (see Figure 5).

Total Organic Carbon (TOC)

Dissolved Organic Particulate Organic
Carbon (<0.45µm) Carbon, POC (>0.45µm)

Low molecular weight (LMW) organics Biopolymers (>> 1 kDa)

Building blocks Humic substances

(0.3 - 0.5 kDa) (0.5 - 1.2 kDa) Proteins (PT) Polysaccharides (PS)

LMW acids LMW neutrals Other Glyco- Acidic Other

(<0.35 kDa) (<0.35 kDa) PT proteins PS PS

TOC/POC UV254 TEP + precursors

FEEM TEP0.4µm/TEP10kDa

LC-OCD-UVD-OND
Analytical methods

Figure 5 Overview of the different fractions of organic matter in the water and the applicable
analytical methods to identify or quantify them (Villacorte et al., 2021). LC-OCD-UVD-
OND is liquid chromatography (LC) with inline detectors for organic carbon (OCD),
UV absorbance at 254 nm (UVD) and organic nitrogen (OND); FEEM is fluorescence
excitation-emission matrices; TEP refers to transparent exopolymer particles measured
with a 0.4 μm or 10 kDa membrane.

10
 Chapter 1

1.4.1 Organic carbon

Traditionally, the presence of organic matter in RO feedwater is assessed by measuring
the total organic carbon (TOC) or dissolved organic carbon (DOC). Routine TOC/
DOC measurements have been used for monitoring the bulk organic fouling potential
of the feedwater. However, not all organic carbon in the water can be directly associated
with organic fouling. So, in many cases, TOC/ DOC as such is not sufficient to assess the
variations in organic fouling potential of the feedwater. According to Voutchkov (2010),
if TOC concentration is below 0.5 mg/L then biofouling is unlikely; and above 2 mg/L,
biofouling is very likely. Nevertheless, as the TOC is a bulk concentration value, it is very
important to identify the fraction of the TOC responsible for bacterial growth. Cationic
organic polymers have also been reported with negative fouling effects on RO membranes,
as they may coprecipitate with negatively charged antiscalants and foul the membrane
irreversibly (Ekowati et al., 2014, Peña et al., 2015, DuPont, 2023).

1.4.2 UV absorbance and fluorescence

Hydrophobic and aromatic organic compounds such as humic substances can be
abundant in surface water sources. UV absorbance at 254 nm (UVA254) is typically used
as an indicator of their relative abundance. Specific UVA254 (SUVA), defined as the ratio
between UVA254 and DOC, is a parameter shown to correlate with the aromatic contents
and hydrophobicity of organic matter (Baghoth, 2012). Such measurement is simple,
fast and can be measured routinely. UV absorbing aromatic and hydrophobic organics are
typically removed by coagulation pretreatment more efficiently than the non-UV absorbing
hydrophilic components (Matilainen et al., 2011). Hence, a low SUVA after pretreatment
does not necessarily mean low organic fouling potential because hydrophilic compounds
may still remain in the RO feedwater. Moreover, humic substances were reported to be less
problematic foulant than hydrophilic organic substances (Amy et al., 2011). Therefore, it
is recommended that UVA254 or SUVA is supplemented with other measurements when
assessing organic fouling potential of RO feedwater.

An emerging technique to characterize organic foulants is by measuring differences

in fluorescence spectra associated with specific organic compounds in a sample using
spectrofluorometer (see Chapter 12). The technique can be applied on both liquid (e.g.,
feedwater) and solid (e.g., membrane surface) samples, and generates a three-dimensional
excitation-emission matrix (EEM). The location of EEM peaks provides a qualitative
indication of types of organic molecules present in water samples (Westerhoff et al., 2001).
For example, humic-like fluorescence peak could be clearly discriminated from protein-like
peak in the EEM spectra. In general, fluorescence spectroscopy can be used as a rapid and
sensitive method to characterize dissolved organic matter, but analysis is limited to organic
components in the water that contains fluorophores. For instance, other organic matter
components such as polysaccharides do not fluoresce and could not be analyzed using the
EEM spectra. Therefore, FEEM is typically not a standalone method for organic fouling
assessment.

1.4.3 LC-OCD
Liquid chromatography-organic carbon detection (LC-OCD) is an advanced method for
fractionation and measurement of organic carbon by size-exclusion chromatography

11
Experimental Methods for Membrane Applications

followed by inline analyses through multiple detectors (i.e., organic carbon, UV254 and
organic nitrogen). While DOC measures organic carbon in bulk, LC-OCD fractionates DOC
based on their molecular weight and hydrophobicity (see Chapter 11). LC-OCD measures
the concentration of organic carbon fractions as biopolymers, humic substances, building
blocks, low molecular weight acids, and low molecular weight neutrals (Huber et al., 2011).
The limit of detection of the method for each fraction is at ppb level and the method has
been adapted for high salinity water (Amy et al., 2011). LC-OCD method has been applied
in many studies to characterize organic matter in surface water for assessment of fouling
potential of the different fractions and their removal through the water treatment processes
(Lozier et al., 2009, Salinas Rodriguez et al., 2009, Villacorte et al., 2009, Simon et al., 2013,
Ho et al., 2015, Shanmuganathan et al., 2015, Jeong et al., 2016, Yin et al., 2019, Altmann
et al., 2023).

1.4.4 TEP
Transparent exopolymer particles (TEP) and their precursors can have a major role in organic
and biological fouling in membrane filtration processes. Berman and Holenberg (2005) first
proposed the potential role of TEP as a major initiator of biofilm leading to biofouling in
reverse osmosis (RO) membranes. Consequently, various experimental studies investigated
the role of TEP, on biofouling (Bar-Zeev et al., 2012) and organic fouling (Villacorte et al.,
2021) in membrane systems. Several methods have been developed, adopted, modified, and
demonstrated to quantify TEP and elucidate their role to membrane fouling (see Chapter 13).
The TEP0.4μm and TEP10kDa methods has been successfully used to semi-quantitatively
demonstrate the role of TEP on the operational performance of membrane processes
(Villacorte et al., 2015a). They have been applied as an indicator of fouling potential of RO
feedwater and showed significant correlation with MFI-UF (Villacorte et al., 2015b). So far,
no study has successfully determined the threshold level of TEP in the feedwater at which
membrane fouling will likely not occur. TEP methods still have some inherent limitations
(Discart et al., 2015, Bittar et al., 2018, Li et al., 2018), so it should be implemented with
proper attention to the protocol used and by someone who is experienced with laboratory
analytical techniques.

The quantification of algae in the RO feedwater source can act as an indication of the
occurrence of algal blooms, which can generate organic foulants like TEP. Algae can be
quantified directly through microscopic counting as cell density or indirectly through
chlorophyll-a measurement. Standard chlorophyll-a methods are widely available (Arar and
Collins, 1997; ASTM D3731-20, 2020; Lipps et al., 2023c). A spike in algae concentrations
can coincide with an increase of organic fouling mainly due to extracellular substances
released by algae. However, a spike in algae density or chlorophyll-a concentration in the
water does not necessarily result in high organic fouling because bloom-forming algal
species can vary in shapes/sizes, specific chlorophyll-a concentration, TEP/EPS production
and their characteristics that affects their removal in the pre-treatment process and their
organic fouling potential to RO.

1.4.5 Oil and grease

One of the most detrimental types of organic foulants are oily compounds which can impact
both the operation and integrity of the membrane units. Ideally, oil and grease should not
exceed 0.1 mg/L in the feed water of a RO system because it can attach and accumulate

12
 Chapter 1

on the membrane surface which may lead to irreversible organic fouling. Pre-treatment is
necessary in the case of treating produced water from oil and gas extraction and industrial
wastewater. The method to determine oil and grease in water consists of extraction by
liquid/liquid extraction, solid phase extraction, or microwave extraction and measurement
by gravimetric and infrared analysis. The standard method for determination of oil and
grease in water through these various extraction methods and analyses can be found in Lipps
et al. (2023b), USEPA (2010), ASTM D7066-04 (2010) and ASTM D7575-11 (2017).
Generally, the infrared methods are more sensitive compared to gravimetric methods
with detection limit of approximately 1 mg/L and even down to 0.1 mg/L when using
tetrachloroethylene as the extraction solvent (Farmaki et al., 2007).

1.5 BIOFOULING POTENTIAL

Biofouling occurs due to the growth of microorganisms on the membrane and feed spacer
of the RO system. Biofouling is a common issue in most RO desalination plants (Peña et
al., 2022) and is often inevitable even when bacteria/microorganisms are completely
removed through the pre-treatment system (i.e., using microfiltration or ultrafiltration
system). If a single bacteria/ microorganism finds their way to reach the RO system, it can
rapidly grow and form a biofilm layer on the membrane and/or feed spacer of the RO when
nutrient concentrations in the feedwater are limited. Biofouling occurs often in plants with
open water sources (e.g., sea, river, lake) as they typically contain higher concentrations of
organics and other nutrients. Thus, biofouling of brackish water RO is less frequent than
that of seawater RO system.

1.5.1 Bacterial growth potential

Given that RO biofouling is mainly due to bacterial growth on the membrane surface and
feed spacer, a method to measure the bacterial growth potential (BGP) in the RO feed water
and through the pre-treatment process was developed by Abushaban (2019). The full
description of the protocol and optimization of each step are discussed in Chapter 15. The
basic concept of the method is to measure the growth of constant number of indigenous
bacteria due to the presence of any nutrients (C: N: P) in a seawater sample. The limit of
detection (LoD) of the method is 10 μg-C (as glucose)/L which is low enough to measure
the BGP in the BWRO feedwater. However, this LoD might not be low enough in some
BWRO system where BGP can be much lower, especially after the pre-treatment process.
The correlation between BGP in the SWRO feed water and biofouling in selected SWRO
membrane systems in Australia, Europe, and the Middle East was investigated (Abushaban
et al., 2019b, Abushaban et al., 2020, Abushaban et al., 2021). Results show that a higher
BGP in the SWRO feedwater (100 - 950 μg-C/L) corresponds to a higher normalised
pressure drop or higher CIP cleaning frequency in the RO system, a demonstration of
the applicability of BGP as a biofouling indicator in RO systems. It was estimated that
a BGP value of 70 μg-C/L in the SWRO feedwater requires once per year CIP frequency
(Abushaban, 2019, Abushaban et al., 2019a, Dhakal et al., 2020, Salinas Rodriguez et al.,
2021b). Consequently, a safe level of BGP (below 70 μg-C/L as glucose equivalent) was
preliminarily proposed to control biofouling in SWRO desalination plants.

13
Experimental Methods for Membrane Applications

The advantage of the BGP method over other methods is that it measures the growth of
indigenous bacteria until when the biodegradable nutrients present in seawater sample is
depleted. Moreover, the duration of the test is around 4-5 days which is relatively short
compared to conventional assimilable organic carbon (12-14 days) and biomass production
potential (15-28 days) methods. A test duration of that takes days can still be a practical
limitation of the method, particularly, when the concentration of BGP varies significantly
(hourly or daily) in the water source. However, it should be noted that biofilm formation
usually takes couple of weeks to be grow on the membrane system. Thus, getting the results
within a few days can still be considered as an early detection of biofouling and a corrective
action can still be made to remove or control growth of microorganisms.

There are other limitations in the application of the BGP method. Firstly, the protocol itself
is quite complex. The complexity is due to the requirement of carbon-free glassware of each
step. Any introduction of carbon or organic matter during handling and measuring BGP will
negatively affect the results (Abushaban, 2019). Secondly, the cost of measuring BGP is high
as it needs a qualified and skilled technician to measure the sample for around two weeks
(including preparation and measurements) and the reagents used to measure microbial
adenosine triphosphate (ATP) are also expensive. The frequency of measuring the BGP in
SWRO feed water and/or along the pre-treatment processes, depends on the water source
quality and the expected variation in the quality from season/month/week/day to another.
Finally, another limitation of BGP method is that the results are influenced by the salinity
of the water which is normally not the case in many desalination plants where the salinity
of the water source is somehow constant. If the salinity of the water source is changed, it
means new calibration curves for both ATP and BGP should be established. The higher the
salinity the lower ATP signal is expected and thus lower slope of BGP.

In general, BGP method is a promising assessment method to control biofouling in SWRO

system. However, the method is rather complex and thus it is currently applied mainly for
research studies and not yet on a routine basis.

1.5.2 Assimilable organic carbon

Assimilable organic carbon (AOC) method has been developed 4 decades ago, mainly to
monitor growth potential in drinking water distribution systems. In the recent years, the
method has been adapted for seawater application. The difference between BGP and AOC
methods is that AOC uses a single strain of bacteria while BGP uses a mix of indigenous
bacteria (Abushaban et al., 2022). The use of single strain of bacteria enables standardization
the method (use one conversion value for samples from different location). Further
optimization of the inoculum used in the test led to substantial reduction the duration of
the test to 1-2 days.

The AOC method is that the method is simpler to implement than the BGP method.
However, it is considered less accurate than BGP as it represents bacterial growth of a single
strain of bacteria. It was also reported that the difference in bacterial growth of a single strain
of bacteria is typically at least 20% lower than the growth of indigenous bacteria in fresh
water (Ross et al., 2013).

14
 Chapter 1

Various methods of AOC have been developed over the years using different bacterial strains
such as Vibrio fischeri and Vibrio harveyi (Weinrich et al., 2011, Jeong et al., 2013). The LoD
of these two methods are 10 and 0.1 μg-C/L, respectively. However, the extremely low LoD
(0.1 μg-C/L) reported is questionable as it was calculated after subtracting the AOC of the
blank, which was >50 μg-C/L.

Weinrich et al. (2015) reported a good correlation between AOC and differential pressure
increase and specific flux decline at the Tampa Bay pilot seawater desalination plant
where the feedwater AOC concentrations measured were between 22 and 161 μg-C/L.
Consequently, a preliminary threshold concentration of AOC (50 μg-C/L) was proposed
using Vibrio harveyi bacteria in seawater (Weinrich et al., 2015). So far, the reported AOC
concentrations in SWRO feed water varies between 10 and 220 μg-C/L.

Overall, the AOC method is applicable to monitor biofouling potential along the SWRO
pre-treatment process and in the SWRO feed water. However, the accuracy of the reported
bacterial growth may not represent actual conditions in the RO system as only one single
strain is used to mimic the bacterial growth.

1.5.3 Biodegradable dissolved organic carbon

Not all DOC is bioavailable or can be directly utilized by microorganisms. Biodegradable
dissolved organic carbon (BDOC) represents the fraction of DOC that can be utilized by
microorganisms. It is calculated by subtracting the initial DOC of the water sample from the
final DOC at the end of incubation period. The incubation period in BDOC measurement
can be varied, depending on the time required to reach a stable DOC. Servais et al. (1987)
suggested 4 weeks of incubation period while shorter incubation periods were introduced
in various application of BDOC method (Joret et al., 1989, Frías et al., 1992, Kadjeski et al.,
2020). BDOC includes a larger fraction of total organic carbon (10-30%) with LoD of 0.1-
0.2 mg/L which is about 10 folds higher than LoD of BGP and AOC. In general, the BDOC
method is time consuming for routine monitoring and less sensitive compared to AOC and
BGP methods. Additionally, in membrane filtration, it is likely that a large portion of BDOC
is retained on the membrane while still allowing the majority of AOC to pass through
(Escobar et al., 2000, Escobar and Randall, 2001).

1.5.4 Phosphate
Biofilm formation in membranes can be largely influenced by the C:N:P nutrient mass ratio
in the water which is ideally 100:23:4.3. Based on this ratio, the requirement for phosphorus
(P) is lower than other substrates (carbon and nitrogen), so a small change in P can lead to
a significant change in microbial growth. Limiting P down to a low level can disrupt the
nutrient balance, restricting bacterial growth in the water and reducing biofouling (van
der Kooij et al., 2007, Galjaard et al., 2008, Jacobson et al., 2009, Vrouwenvelder et al.,
2010, Kim et al., 2014). Reliable analytical methods to measure phosphate down to sub-
ppb level is critical to this strategy. Standard phosphate analytical methods are available
such as the widely used phosphomolybdenum blue method (Lipps et al., 2023a). The
method has been applied and adapted in biofouling studies of water containing very low
phosphate concentrations (Abushaban et al., 2020, Javier et al., 2020). Ultra-low phosphate

15
Experimental Methods for Membrane Applications

concentration down to below 0.3 μg PO4-P/L was reported to limit biofouling even in water
with high concentration of organic carbon (Vrouwenvelder et al., 2010). Some antiscalants
(e.g., phosphonates) which are added to the feedwater to prevent scaling in RO, contain
phosphate which has been found to cause biofouling (Vrouwenvelder et al., 2000, Sweity
et al., 2013, Sweity et al., 2015, Hasanin et al., 2023). It is therefore recommended that the
dosage and type of antiscalants be taken into consideration when applied as pretreatment
for RO.

1.6 OUTLOOK AND OPPORTUNITIES

RO desalination of brackish and saline water sources is increasingly applied globally to

solve water scarcity challenges in the utility and industry sector. Assessing the fouling and
scaling potential of the feedwater source is highly critical when designing and operating the
RO desalination plant including its pre-treatment units. For many years, RO membrane
suppliers have provided specific guidelines of the ideal RO feedwater quality to minimize
possible particulate, inorganic, organic and biological fouling, and scaling issues in the
plant. However, some of these standard water quality parameters and indices have some
limitations so alternative/supplemental assessment and characterization tools has been
developed over the years for feedwater monitoring and experimental investigations.

Measuring the individual concentration of all potential foulants present in the RO feed
water is sort of a mission impossible due to costs, duration and specialized laboratory
facilities needed. Particulate fouling indices like the SDI or MFI-0.45 can be used in a daily
basis for monitoring of water quality with onsite measurements. Other parameters like the
MFI-UF constant flux are promising due to its sensitivity and ability to predict accurately the
rate of RO membrane fouling.

Standardized inorganic fouling and scaling assessment methods are widely available for lab-
based analyses. Online monitoring is currently a challenge and would be an important area
for further development. The induction time concept for predicting when scaling can occur
is a promising tool to developing RO plant design and control strategies to minimize scaling
issues (e.g., Mangal et al. (2022)).

The presence of organic foulants can be routinely measured with offline/online TOC
measurements. Routine chlorophyll-a measurement can be beneficial for feedwater sources
that are prone to seasonal algal blooms. For in-depth investigations of organic fouling,
more advanced or complex assessment methods (e.g., LC-OCD, FEEM, TEP) can be further
considered.

Biological fouling is currently the leading cause of operational challenges in RO applications.

Genomic tools (see Chapter 14) have been applied to identify bacterial communities often
associated with biofilm development in RO. Promising methods for assessing biofouling
potential of RO feedwater are either based on the bacterial growth capacity (see Chapters
15 and 16) or based on the concentration of specific limiting nutrient (phosphate) in the
feedwater. However, these methods either requires days to weeks of incubation to generate
results or have high sensitivity to contamination. Future developments should focus on
overcoming either one of these limitations.

16
 Chapter 1

Ideally, assessment methods and indices should be made practical for onsite monitoring
and low cost for consumables, enabling it for wide use. Testing conditions should be
standardized/calibrated and should specify the method LOD and their applicability in
fresh and saline water matrices. The need and opportunities for real time online monitoring
should be explored further. In some cases, a few analyses per day is sufficient and in other
cases once a day or once a week are considered sufficient.

Chemicals used in the treatment process may contribute to the membrane fouling
development due to introduction of nutrients or organic foulants. Hence, it is recommended
that operators also assess the fouling potential of such chemicals before applying it to the
RO system.

Current RO feedwater quality guideline values recommended by membrane manufacturers

may need to be verified under local conditions and modified/adjusted accordingly based
on latest developments. On the other hand, feedwater guidelines for emerging desalination
technologies such as forward osmosis (see Chapter 4) and membrane distillation (see
Chapter 5) are currently non-existent. Future developments of these new technologies
should also include understanding their propensity to fouling or scaling by applying current
or new feedwater quality assessment methods.

1.7 ABBREVIATIONS AND SYMBOLS

AOC Assimilable organic carbon

ATP Adenosine tri-phosphate
BGP Bacterial growth potential
BFR Biofilm formation rate
BWRO Brackish water reverse osmosis
CIP Cleaning in place
COD Chemical oxygen demand
DOC Dissolved organic carbon
EEM Excitation-emission matrix
LC-OCD Liquid chromatography-organic carbon detection
LoD Limit of detection
MED Multi-effect distillation
MFI-0.45 Modified fouling index, constant pressure, 0.45 μm filter
MFI-UF Modified fouling index, constant flux, 10 kDa or 100 kDa membrane filter
MSF Multi stage flash distillation
RO Reverse osmosis
SDI Silt density index, %/min
S&DSI Stiff and Davis saturation index
SWRO Seawater reverse osmosis
TEP Transparent exopolymer particles
TOC Total organic carbon

NB. This chapter is currently being prepared for submission as a scientific article in a journal.

17
Experimental Methods for Membrane Applications

1.8 REFERENCES

Abunada M (2023). Prediction of Particulate Fouling in Reverse Osmosis Systems: MFI-UF Method
Development and Application. PhD Dissertation, IHE Delft
Abunada M, Dhakal N, Gulrez R, Ajok P, Li Y, Abushaban A, Smit H, Moed D, Ghaffour N, Schippers
JC, Kennedy MD (2023) Prediction of particulate fouling in full-scale reverse osmosis plants
using the modified fouling index – ultrafiltration (MFI-UF) method. Desalination 553: 116478
DOI https://doi.org/10.1016/j.desal.2023.116478
Abushaban A (2019) Assessing Bacterial Growth Potential in Seawater Reverse Osmosis Pretreatment:
Method Development and Applications CRC Press9781000034707.
Abushaban A, Salinas-Rodriguez SG, Dhakal N, Schippers JC, Kennedy MD (2019a) Assessing
pretreatment and seawater reverse osmosis performance using an ATP-based bacterial
growth potential method. Desalination 467: 210-218 DOI https://doi.org/10.1016/j.
desal.2019.06.001
Abushaban A, Salinas-Rodriguez SG, Kapala M, Pastorelli D, Schippers JC, Mondal S, Goueli S, Kennedy
MD (2020) Monitoring Biofouling Potential Using ATP-Based Bacterial Growth Potential in
SWRO Pre-Treatment of a Full-Scale Plant. Membranes 10: 360
Abushaban A, Salinas-Rodriguez SG, Pastorelli D, Schippers JC, Mondal S, Goueli S, Kennedy MD
(2021) Assessing Pretreatment Effectiveness for Particulate, Organic and Biological Fouling in a
Full-Scale SWRO Desalination Plant. Membranes 11: 167
Abushaban A, Salinas-Rodriguez SG, Philibert M, Le Bouille L, Necibi MC, Chehbouni A (2022) Biofouling
potential indicators to assess pretreatment and mitigate biofouling in SWRO membranes: A short
review. Desalination 527: 115543 DOI https://doi.org/10.1016/j.desal.2021.115543
Abushaban A, Salinas-Rodriguez SG, Schippers JC, Kennedy MD (2019b) Assessing pretreatment and
seawater reverse osmosis performance using an ATP-based bacterial growth potential method.
Desalination 467: 210-218 DOI https://doi.org/10.1016/j.desal.2019.06.001
Alhadidi A, Blankert B, Kemperman AJB, Schippers JC, Wessling M, van der Meer WGJ (2011a) Effect of
testing conditions and filtration mechanisms on SDI. Journal of Membrane Science 381: 142-151
Alhadidi A, Kemperman AJB, Schippers JC, Wessling M, van der Meer WGJ (2011b) The influence of
membrane properties on the Silt Density Index. Journal of Membrane Science 384: 205-218
Altmann T, Rousseva A, Vrouwenvelder J, Shaw M, Das R (2023) Effectiveness of ceramic ultrafiltration
as pretreatment for seawater reverse osmosis. Desalination 564: 116781 DOI https://doi.
org/10.1016/j.desal.2023.116781
Amy GL, Salinas Rodríguez SG, Kennedy MD, Schippers JC, Rapenne S, Remize P-J, Barbe C, Manes
C-LdO, West NJ, Lebaron P, Kooij Dvd, Veenendaal H, Schaule G, Petrowski K, Huber S, Sim
LN, Ye Y, Chen V, Fane AG (2011) Water quality assessment tools. In: Drioli E, Criscuoli A,
Macedonio F (eds) Membrane-Based Desalination - An Integrated Approach (MEDINA): 192
10.2166/9781780400914
Arar EJ, Collins GB (1997) Method 445.0: In vitro determination of chlorophyll a and pheophytin a in
marine and freshwater algae by fluorescence United States Environmental Protection Agency, Office
of Research and Development, National Exposure Research Laboratory, Cincinnati, pp. 1-22
ASTM D857-17 (2017) Standard Test Method for Aluminum in Water ASTM International, West
Conshohocken, PA, pp. 5 https://www.astm.org/d0857-17.html.
ASTM D858-17 (2017) Standard Test Methods for Manganese in Water ASTM International, West
Conshohocken, PA, pp. 9 https://www.astm.org/d0858-17.html.
ASTM D859-16(2021)e1 (2021) Standard Test Method for Silica in Water ASTM International, West
Conshohocken, PA, pp. 5 https://www.astm.org/d0859-16r21e01.html.

18
 Chapter 1

ASTM D1068-15 (2016) Standard Test Methods for Iron in Water ASTM International, West
Conshohocken, PA, pp. 13 https://www.astm.org/d1068-15.html.
ASTM D1252-06(2020) (2020) Standard Test Methods for Chemical Oxygen Demand (Dichromate
Oxygen Demand) of Water ASTM International, West Conshohocken, PA https://www.astm.
org/d1252-06r20.html.
ASTM D1253-14(2021)e1 (2021) Standard Test Method for Residual Chlorine in Water ASTM
International, West Conshohocken, PA, pp. 6 https://www.astm.org/d1253-14r21e01.html.
ASTM D1293-12 (2018) Standard Test Methods for pH of Water ASTM International, West
Conshohocken, PA, pp. 10 https://www.astm.org/d1293-12.html.
ASTM D1889-00 (2017) Standard Test Method for Turbidity of Water (Withdrawn 2007) ASTM
International, West Conshohocken, PA https://www.astm.org/d1889-00.html.
ASTM D2579-93e1 (2021) Standard Test Method for Total Organic Carbon in Water (Withdrawn 2002)
ASTM International, West Conshohocken, PA https://www.astm.org/d2579-93e01.html.
ASTM D3731-20 (2020) Standard Practices for Measurement of Chlorophyll Content of Algae in
Surface Waters ASTM International, West Conshohocken, PA, pp. 5 www.astm.org.
ASTM D3739-19 (2019) Standard Practice for Calculation and Adjustment of the Langelier Saturation
Index for Reverse Osmosis ASTM International, West Conshohocken, PA, pp. 4 https://www.
astm.org/d3739-19.html.
ASTM D4189 - 07 (2007) Standard Test Method for Silt Density Index (SDI) of Water ASTM
International, West Conshohocken, PA
ASTM D4189 - 14 (2014) Standard Test Method for Silt Density Index (SDI) of Water ASTM
International, West Conshohocken, PA
ASTM D4582-91(2001) (2017) Standard Practice for Calculation and Adjustment of the Stiff and
Davis Stability Index for Reverse Osmosis ASTM International, West Conshohocken, PA, pp. 4
https://www.astm.org/d4582-91r01.html.
ASTM D7066-04 (2010) Standard Test Method for dimer/trimer of chlorotrifluoroethylene (S-
316) Recoverable Oil and Grease and Nonpolar Material by Infrared Determination ASTM
International, West Conshohocken, PA, pp. 9 www.astm.org.
ASTM D7575-11 (2017) Standard Test Method for Solvent-Free Membrane Recoverable Oil and
Grease by Infrared Determination ASTM International, West Conshohocken, PA, pp. 12 www.
astm.org.
ASTM D8002 - 15 (2015) Standard Test Method for Modified Fouling Index (MFI-0.45) of Water
ASTM International, West Conshohocken, PA www.astm.org.
Badruzzaman M, Voutchkov N, Weinrich L, Jacangelo JG (2019) Selection of pretreatment technologies
for seawater reverse osmosis plants: A review. Desalination 449: 78-91 DOI https://doi.
org/10.1016/j.desal.2018.10.006
Baghoth SA (2012) Characterizing natural organic matter in drinking water treatment processes and
trains CRC Press/Balkema, Delft. 9781138000261.
Bar-Zeev E, Berman-Frank I, Girshevitz O, Berman T (2012) Revised paradigm of aquatic biofilm
formation facilitated by microgel transparent exopolymer particles. Proceedings of the National
Academy of Sciences 109: 9119-9124 DOI doi:10.1073/pnas.1203708109
Bassa FGM (2021). Evaluating surface water pre-treatment with the MFI-UF for assessing fouling in
membrane systems. MSc, IHE Delft UWS-WSE/21.25.
Belila A, El-Chakhtoura J, Otaibi N, Muyzer G, Gonzalez-Gil G, Saikaly PE, van Loosdrecht MCM,
Vrouwenvelder JS (2016) Bacterial community structure and variation in a full-scale seawater
desalination plant for drinking water production. Water Research 94: 62-72 DOI https://doi.
org/10.1016/j.watres.2016.02.039

19
Experimental Methods for Membrane Applications

Berman T, Holenberg M (2005) Don’t fall foul of biofilm through high TEP levels. Filtration & Separation
42: 30-32 DOI https://doi.org/10.1016/S0015-1882(05)70517-6
Birch H, Pasture LdL, Gall M, Gasson C, Pankratz T, Quaresma M, Qureshi Z, Walker C (2023) Market focus
deck Desalination and reuse October 2023 Media Analytics Ltd., Oxford, UK. 978-1-907467-67-7.
Bittar TB, Passow U, Hamaraty L, Bidle KD, Harvey EL (2018) An updated method for the calibration of
transparent exopolymer particle measurements. Limnology and Oceanography: Methods 16: 621-
628 DOI https://doi.org/10.1002/lom3.10268
Boerlage SFE (2001) Scaling and Particulate Fouling in Membrane Filtration Systems Swets&Zeitlinger
Publishers, Lisse. 90 5809 242 9.
Boerlage SFE, Kennedy MD, Aniye MP, Abogrean EM, El-Hodali DEY, Tarawneh ZS, Schippers JC
(2000a) Modified Fouling Index ultrafiltration to compare pretreatment processes of reverse
osmosis feedwater. Desalination 131: 201-214
Boerlage SFE, Kennedy MD, Bremere I, Witkamp GJ, van der Hoek JP, Schippers JC (2000b) Stable
barium sulphate supersaturation in reverse osmosis. Journal of Membrane Science 179: 53-68
DOI 10.1016/s0376-7388(00)00504-4
DesalData (2023) 36th Desalination Plants Inventory In: Global Water Intelligence (ed)
Dhakal N, Abushaban A, Mangal MN, Abunada M, Schippers JC, Kennedy M (2020) Membrane Fouling
and Scaling in Reverse Osmosis. In: Sapalidis A (ed) Membrane Desalination: 325-344
Dhakal N, Salinas Rodriguez SG, Schippers JC, Kennedy MD (2015) Induction time measurements in
two brackish water reverse osmosis plants for calcium carbonate precipitation. Desalination and
Water Treatment 53: 285-293 DOI 10.1080/19443994.2014.903870
Discart V, Bilad MR, Vankelecom IFJ (2015) Critical Evaluation of the Determination Methods
for Transparent Exopolymer Particles, Agents of Membrane Fouling. Critical Reviews in
Environmental Science and Technology 45: 167-192 DOI 10.1080/10643389.2013.829982
DuPont (2020) FILMTEC™ Reverse Osmosis Membranes Technical Manual, Form No. 45-D01504-en,
Rev. 3Water solutions, pp. 207
DuPont (2023) FILMTEC™ Reverse Osmosis Membranes Technical Manual, Form No. 45-D01504-en,
Rev. 16Water solutions, pp. 188 https://www.dupont.com/content/dam/dupont/amer/us/
en/water-solutions/public/documents/en/RO-NF-FilmTec-Manual-45-D01504-en.pdf.
Ekowati Y, Msuya M, Salinas Rodriguez SG, Veenendaal G, Schippers JC, Kennedy MD (2014) Synthetic
organic polymer fouling in municipal wastewater reuse reverse osmosis. Journal of Water Reuse
and Desalination 4: 125-136 DOI doi:10.2166/wrd.2014.046
Escobar IC, Hong S, Randall AA (2000) Removal of assimilable organic carbon and biodegradable
dissolved organic carbon by reverse osmosis and nanofiltration membranes. Journal of Membrane
Science 175: 1-17 DOI https://doi.org/10.1016/S0376-7388(00)00398-7
Escobar IC, Randall AA (2001) Assimilable organic carbon (AOC) and biodegradable dissolved organic
carbon (BDOC):: complementary measurements. Water Research 35: 4444-4454 DOI https://
doi.org/10.1016/S0043-1354(01)00173-7
Farmaki E, Kaloudis T, Dimitrou K, Thanasoulias N, Kousouris L, Tzoumerkas F (2007) Validation of a
FT-IR method for the determination of oils and grease in water using tetrachloroethylene as the
extraction solvent. Desalination 210: 52-60 DOI https://doi.org/10.1016/j.desal.2006.05.032
Frías J, Ribas F, Lucena F (1992) A method for the measurement of biodegradable organic carbon in waters.
Water Research 26: 255-258 DOI https://doi.org/10.1016/0043-1354(92)90226-T
Galjaard G, Lampe M, Kamp PC (2008) (8) years of RO-experience at WTP Heemskerk biofouling aspects.
Paper presented at the Membrane Technology Conference, Naples, Florida2008

20
 Chapter 1

Hasanin G, Mosquera AM, Emwas A-H, Altmann T, Das R, Buijs PJ, Vrouwenvelder JS, Gonzalez-Gil
G (2023) The microbial growth potential of antiscalants used in seawater desalination. Water
Research 233: 119802 DOI https://doi.org/10.1016/j.watres.2023.119802
He S, Oddo JE, Tomson MB (1995) The Nucleation Kinetics of Barium Sulfate in NaCl Solutions up to 6 m
and 90°C. Journal of Colloid and Interface Science 174: 319-326 DOI https://doi.org/10.1006/
jcis.1995.1397
Ho JS, Ma Z, Qin J, Sim SH, Toh C-S (2015) Inline coagulation–ultrafiltration as the pretreatment for
reverse osmosis brine treatment and recovery. Desalination 365: 242-249 DOI https://doi.
org/10.1016/j.desal.2015.03.018
Huber SA, Balz A, Abert M, Pronk W (2011) Characterisation of aquatic humic and non-humic matter with
size-exclusion chromatography – organic carbon detection – organic nitrogen detection (LC-OCD-
OND). Water Research 45: 879-885 DOI http://dx.doi.org/10.1016/j.watres.2010.09.023
Jacobson JD, Kennedy MD, Amy G, Schippers JC (2009) Phosphate limitation in reverse osmosis: An
option to control biofouling? Desalination and Water Treatment 5: 198-206 DOI 10.5004/
dwt.2009.578
Javier L, Farhat NM, Desmond P, Linares RV, Bucs S, Kruithof JC, Vrouwenvelder JS (2020) Biofouling
control by phosphorus limitation strongly depends on the assimilable organic carbon concentration.
Water Research 183: 116051 DOI https://doi.org/10.1016/j.watres.2020.116051
Jeong S, Naidu G, Vigneswaran S, Ma CH, Rice SA (2013) A rapid bioluminescence-based test of
assimilable organic carbon for seawater. Desalination 317: 160-165 DOI http://dx.doi.
org/10.1016/j.desal.2013.03.005
Jeong S, Naidu G, Vollprecht R, Leiknes T, Vigneswaran S (2016) In-depth analyses of organic matters in a
full-scale seawater desalination plant and an autopsy of reverse osmosis membrane. Separation and
Purification Technology 162: 171-179 DOI https://doi.org/10.1016/j.seppur.2016.02.029
Joret C, Levi Y, Gilbert M (1989) The measurement of biodegradable organic carbon (BDOC): a tool in
water treatment. Water Supply 7: 41-45
Kadjeski M, Fasching C, Xenopoulos MA (2020) Synchronous Biodegradability and Production of
Dissolved Organic Matter in Two Streams of Varying Land Use. Frontiers in Microbiology 11 DOI
10.3389/fmicb.2020.568629
Khan MT, Busch M, Molina VG, Emwas A-H, Aubry C, Croue J-P (2014) How different is the composition
of the fouling layer of wastewater reuse and seawater desalination RO membranes? Water Research
59: 271-282 DOI https://doi.org/10.1016/j.watres.2014.04.020
Kim C-M, Kim S-J, Kim LH, Shin MS, Yu H-W, Kim IS (2014) Effects of phosphate limitation in feed
water on biofouling in forward osmosis (FO) process. Desalination 349: 51-59 DOI https://doi.
org/10.1016/j.desal.2014.06.013
Kurihara M, Ito Y (2020) Sustainable Seawater Reverse Osmosis Desalination as Green Desalination
in the 21st Century. Journal of Membrane Science and Research 6: 20-29 DOI 10.22079/
jmsr.2019.109807.1272
Li X, Skillman L, Li D, Ela WP (2018) Comparison of Alcian blue and total carbohydrate assays for
quantitation of transparent exopolymer particles (TEP) in biofouling studies. Water Research 133:
60-68 DOI https://doi.org/10.1016/j.watres.2017.12.021
Lipps WC, Braun-Howland EB, Baxter TE (2023a) 4500-P Phosphorus. In: American Public Health
Association, American Water Works Association, Water Environment Federation, (ed) Standard
Methods for the Examination of Water and Wastewater 24th Edition, 24th edn:
Lipps WC, Braun-Howland EB, Baxter TE (2023b) 5520 Oil and grease. In: American Public Health
Association, American Water Works Association, Water Environment Federation, (ed) Standard
Methods for the Examination of Water and Wastewater 24th Edition, 24th edn:

21
Experimental Methods for Membrane Applications

Lipps WC, Braun-Howland EB, Baxter TE (2023c) 10150 Chloropyll a. In: American Public Health
Association, American Water Works Association, Water Environment Federation, (ed) Standard
Methods for the Examination of Water and Wastewater 24th Edition, 24th edn:
Lozier JC, Bankston A, Beaty J, Garcia-Aleman J, Scharf R, Amy G, Salinas Rodríguez SG (2009) Use
of advanced NOM characterization methods to trace the fate of organic contaminants from a
membrane backwash recycle scheme. Paper presented at the Membrane Technology Conference,
Memphis, Tennessee United States2009
Mangal MN (2023) Controlling scaling in groundwater reverse osmosis: minimizing antiscalant
consumption Veenman, The Netherlands. 978-90-365-5588-3. 10.3990/1.9789036555890
Mangal N, Salinas Rodriguez SG, Yangali Quintanilla VA, Schippers JC, Kennedy MD (2021) Ch 08 -
Scaling. In: Salinas Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (eds) Seawater
Reverse Osmosis Desalination: Assessment and Pre-treatment of Fouling and Scaling, 1st edn:
207-242 10.2166/9781780409863_0207
Matilainen A, Gjessing ET, Lahtinen T, Hed L, Bhatnagar A, Sillanpää M (2011) An overview of the methods
used in the characterisation of natural organic matter (NOM) in relation to drinking water treatment.
Chemosphere 83: 1431-1442 DOI https://doi.org/10.1016/j.chemosphere.2011.01.018
Nahrstedt A, Camargo Schmale J (2008) New insights into SDI and MFI measurements. Water Science
and Technology: Water Supply 8: 401-412 DOI https://doi.org/10.2166/ws.2008.087
NEN 6271:1995 nl (1995) Bacteriological examination of water - Determination of the easily
assimilable organic carbon (AOC) content NEN, Netherlands https://www.nen.nl/en/nen-
6271-1995-nl-16389.
Peña N, del Vigo F, Salmerón O, Rodriguez J, Borrel A, Chesters SP (2015) Pre-coating of outside-
inside capillary UF membranes with iron hydroxide particles to limit non-backwashable
fouling during seawater algal blooms. Desalination and Water Treatment 55: 2973–2987 DOI
10.1080/19443994.2014.957950
Peña N, Rodriguez J, del Vigo F, Chesters SP (2022) 20 years of data from 500 seawater membrane
autopsies. Paper presented at the IDA World Desalination Congress, Sydney, Australia, 9-13
October 2022
Rachman RM, Ghaffour N, Waly F, Amy GL (2013) Assessment of Silt Density Index (SDI) as fouling
propensity parameter in Reverse Osmosis (RO) desalination systems. Desalination and Water
Treatment 51: 1091-1103 DOI doi:10.1080/19443994.2012.699448
Ross PS, Hammes F, Dignum M, Magic-Knezev A, Hambsch B, Rietveld LC (2013) A comparative
study of three different assimilable organic carbon (AOC) methods: results of a round-robin test.
Water Supply 13: 1024-1033 DOI 10.2166/ws.2013.079
Salinas Rodriguez SG (2021). Fouling and scaling in seawater reverse osmosis desalination. The Source:
4https://www.thesourcemagazine.org/fouling-and-scaling-in-seawater-reverse-osmosis-
desalination/
Salinas Rodríguez SG (2011) Particulate and organic matter fouling of SWRO systems: Characterization,
modelling and applications CRC Press/Balkema, Delft. http://dx.doi.org/10.1201/b11609
Salinas Rodriguez SG, Boerlage SFE, Kenedy MD, Schippers JC (2021a) Ch 04 - Particulate fouling.
In: Salinas Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (eds) Seawater Reverse
Osmosis Desalination: Assessment and Pre-treatment of Fouling and Scaling, 1st edn: 85-124
10.2166/9781780409863_0085
Salinas Rodriguez SG, Kennedy MD, Schippers JC, Amy GL (2009) Organic foulants in estuarine and bay
sources for seawater reverse osmosis – Comparing pre-treatment processes with respect to foulant
reductions. Desalination and Water Treatment 9: 155-164 DOI 10.5004/dwt.2009.766

22
 Chapter 1

Salinas Rodriguez SG, Schippers JC (2021) Ch 01 - Introduction to desalination. In: Salinas Rodriguez
SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (eds) Seawater Reverse Osmosis Desalination:
Assessment and Pre-treatment of Fouling and Scaling, 1st edn: 26 10.2166/9781780409863_0001
Salinas Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (2021b) Seawater Reverse Osmosis
Desalination: Assessment and Pre-treatment of Fouling and Scaling, 1st edn IWA Publishing,
London. 9781780409856. 10.2166/9781780409863
Salinas Rodriguez SG, Sithole N, Dhakal N, Olive M, Schippers JC, Kennedy MD (2019) Monitoring
particulate fouling of North Sea water with SDI and new ASTM MFI0.45 test. Desalination 454:
10-19 DOI https://doi.org/10.1016/j.desal.2018.12.006
Schippers JC, Salinas Rodriguez SG, Boerlage SFE, Kennedy MD (2014). Why MFI is edging SDI as a
fouling index. Desalination & Water Reuse: 28-32
Schippers JC, Verdouw J (1980) The modified fouling index, a method of determining the fouling
characteristics of water. Desalination 32: 137-148
Servais P, Billen G, Hascoët M-C (1987) Determination of the biodegradable fraction of dissolved
organic matter in waters. Water Research 21: 445-450 DOI https://doi.org/10.1016/0043-
1354(87)90192-8
Shanmuganathan S, Nguyen TV, Jeong S, Kandasamy J, Vigneswaran S (2015) Submerged membrane
– (GAC) adsorption hybrid system in reverse osmosis concentrate treatment. Separation and
Purification Technology 146: 8-14 DOI https://doi.org/10.1016/j.seppur.2015.03.017
Simon FX, Penru Y, Guastalli AR, Esplugas S, Llorens J, Baig S (2013) NOM characterization by LC-
OCD in a SWRO desalination line. Desalination and Water Treatment 51: 1776-1780 DOI
10.1080/19443994.2012.704693
Sweity A, Oren Y, Ronen Z, Herzberg M (2013) The influence of antiscalants on biofouling of RO
membranes in seawater desalination. Water Research 47: 3389-3398 DOI https://doi.
org/10.1016/j.watres.2013.03.042
Sweity A, Zere TR, David I, Bason S, Oren Y, Ronen Z, Herzberg M (2015) Side effects of antiscalants
on biofouling of reverse osmosis membranes in brackish water desalination. Journal of Membrane
Science 481: 172-187 DOI https://doi.org/10.1016/j.memsci.2015.02.003
USEPA (2005) Membrane Filtration Guidance Manual. EPA 815-R-06-009 USEPA, Online, pp. 35
https://nepis.epa.gov/Exe/ZyPDF.cgi/901V0500.PDF?Dockey=901V0500.PDF.
USEPA (2010) Method 1664 Revision B: n-hexane extractable material (HEM; oil and grease) and
silica gel treated n-hexane extractable material (SGT-HEM; non-polar material) by extraction
and gravimetry USEPA, Online, pp. 35 https://www.epa.gov/sites/default/files/2015-08/
documents/method_1664b_2010.pdf.
van der Kooij D (1992) Assimilable organic carbon as an indicator of bacterial regrowth. J American
water works association 84: Van der Kooij, D. 1992. Assimilable organic carbon as an indicator of
bacterial regrowth. J. AWWA 1984(1992):1957-1965.
van der Kooij D, Hijnen W, Cornelissen E, Van Agtmaal S, Baas K, Galjaard G (2007) Elucidation of
membrane biofouling processes using bioassays for assessing the microbial growth potential of
feed water. Paper presented at the Membrane Technology Conference, Tampa Bay, Florida2007
van der Kooij D, Visser A, Hijnen WAM (1982) Determining the concentration of easily
assimilable organic carbon in drinking water. Journal AWWA 74: 540-545 DOI https://doi.
org/10.1002/j.1551-8833.1982.tb05000.x
Villacorte LO, Boerlage SFE, Dixon MB (2021) Ch 06 - Algal Blooms and RO desalination. In: Salinas
Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (eds) Seawater Reverse Osmosis
Desalination: Assessment and Pre-treatment of Fouling and Scaling:

23
Experimental Methods for Membrane Applications

Villacorte LO, Ekowati Y, Calix-Ponce HN, Schippers JC, Amy GL, Kennedy MD (2015a) Improved
method for measuring transparent exopolymer particles (TEP) and their precursors in fresh and
saline water. Water Research 70: 300-312 DOI http://dx.doi.org/10.1016/j.watres.2014.12.012
Villacorte LO, Ekowati Y, Winters H, Amy G, Schippers JC, Kennedy MD (2015b) MF/UF rejection
and fouling potential of algal organic matter from bloom-forming marine and freshwater algae.
Desalination 367: 1-10 DOI https://doi.org/10.1016/j.desal.2015.03.027
Villacorte LO, Kennedy MD, Amy GL, Schippers JC (2009) The fate of Transparent Exopolymer
Particles (TEP) in integrated membrane systems: Removal through pre-treatment processes and
deposition on reverse osmosis membranes. Water Research 43: 5039-5052 DOI 10.1016/j.
watres.2009.08.030
Voutchkov N (2010) Considerations for selection of seawater filtration pretreatment system. Desalination
261: 354-364 DOI https://doi.org/10.1016/j.desal.2010.07.002
Vrouwenvelder JS, Beyer F, Dahmani K, Hasan N, Galjaard G, Kruithof JC, Van Loosdrecht MCM (2010)
Phosphate limitation to control biofouling. Water Research 44: 3454-3466 DOI https://doi.
org/10.1016/j.watres.2010.03.026
Vrouwenvelder JS, Manolarakis SA, Veenendaal HR, van der Kooij D (2000) Biofouling potential of
chemicals used for scale control in RO and NF membranes. Desalination 132: 1-10 DOI https://
doi.org/10.1016/S0011-9164(00)00129-6
Vrouwenvelder JS, van der Kooij D (2001) Diagnosis, prediction and prevention of biofouling of
NF and RO membranes. Desalination 139: 65-71 DOI https://doi.org/10.1016/S0011-
9164(01)00295-8
Waly TKA (2011) Minimizing the use of chemicals to control scaling in SWRO - Improved prediction
of the scaling potential of calcium carbonate CRC Press / Balkema, Leiden. 978-0-415-61578-5.
Waly TKA, Kennedy MD, Witkamp G-J, Amy G, Schippers JC (2012) The role of inorganic ions in the
calcium carbonate scaling of seawater reverse osmosis systems. Desalination 284: 279-287 DOI
http://dx.doi.org/10.1016/j.desal.2011.09.012
Weinrich L, LeChevallier M, Haas C (2015) Application of the bioluminescent saltwater assimilable
organic carbon test as a tool for identifying and reducing reverse osmosis membrane fouling
in desalination Water Reuse Research Foundation, Virginia, USA https://www.waterrf.org/
research/projects/application-bioluminescent-saltwater-assimilable-organic-carbon-test-tool
Weinrich LA, Schneider OD, LeChevallier MW (2011) Bioluminescence-based method for measuring
assimilable organic carbon in pretreatment water for reverse osmosis membrane desalination. Appl
Environ Microbiol 77: 1148-1150 DOI 10.1128/aem.01829-10
Westerhoff P, Chen W, Esparza M (2001) Fluorescence Analysis of a Standard Fulvic Acid and Tertiary
Treated Wastewater. Journal of Environmental Quality 30: 2037-2046 DOI https://doi.
org/10.2134/jeq2001.2037
Wilf M, Klinko K (2016) Effect of new pretreatment methods and improved membrane performance
on design of RO seawater systems Hydranautics, Online, pp. 1-17 https://membranes.com/
wp-content/uploads/Documents/Technical-Papers/Application/SWRO/Effect-Of-New-
Pretreatment-Methods-And-Improved-Membrane-Performance-On-Design-Of-RO-Seawater-
Systems.pdf.
Yin W, Li X, Suwarno SR, Cornelissen ER, Chong TH (2019) Fouling behavior of isolated dissolved
organic fractions from seawater in reverse osmosis (RO) desalination process. Water Research 159:
385-396 DOI https://doi.org/10.1016/j.watres.2019.05.038
Zhang M, Jiang S, Tanuwidjaja D, Voutchkov N, Hoek EMV, Cai B (2011) Composition and Variability of
Biofouling Organisms in Seawater Reverse Osmosis Desalination Plants. Appl Environ Microbiol
77: 4390-4398 DOI doi:10.1128/AEM.00122-11

24
 Part 1
Membrane processes
 doi: 10.2166/9781789062977_0027

Chapter 2

Microfiltration and
ultrafiltration
Morten Lykkegaard Christensen, Aalborg University, Denmark

Guillem Gilabert-Oriol, DuPont Water Solutions, Spain

The learning objectives of this chapter are the following:

• give an understanding of how microfiltration and ultrafiltration is used for water

and wastewater treatment

• give knowledge of how membranes system is best operated to reduce membrane

fouling and ensure high flux

• give an introduction to different cleaning options and how hydraulic cleaning

methods are optimized

• give knowledge of how microfiltration and ultrafiltration system are designed and
proper membrane is selected.

2.1 INTRODUCTION

Microfiltration and ultrafiltration are pressure-driven membrane filtration processes, where

a transmembrane pressure (TMP) is used to press water through the membrane. They are
operated at relatively low TMP compared with the other pressure-driven membrane
processes i.e., nanofiltration and reverse osmosis (Figure 1). Micro- and ultrafiltration
membranes are porous membranes and the separation based on sieving effects (size
exclusion). This means that large particles or macromolecules are rejected by the membrane,
whereas small molecules pass the membrane.

Microfiltration membrane typically have pores larger than 0.1 μm whereby bacteria and
particles can be removed by the membrane (Figure 1). Often a nominal pore size is given
from the manufacturer, but the performance of the membrane will also depend on pore size

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

distribution and membrane materials. Microfiltration is often used as pretreatment prior to

other membrane filtration processes such as RO, electrodialysis, and membrane distillation.
Ultrafiltration membranes have pores sizes below 0.1 μm and retains macromolecules
(Figure 1). Ultrafiltration membranes are used for removal of colloids, proteins, virus.
Molecular weight cut-off (MWCO) is used to describe the membrane, instead of pore size.
MWCO is the lowest molecular weight where more than 90% of the macromolecules are
rejected by the membrane.

Size Molecular weight Examples Process Pressure

100 µm Pollen

Starch
10 µm
Microfiltration 0.2 – 2 bar
Blood cells

Bacteria
1 µm

Latex emulsions

1000 Å

100,000 Albumin
100 Å
Ultrafiltration 1 – 10 bar
10,000 Pepsin

1,000 Vitamin B-12

10 Å

Glucose Nanofiltration 5 – 20 bar

Water
1Å RO 10 – 150 bar
Na+ Cl-

Figure 1 Pressure driven filtration processes

2.1.1 Advantages of ultrafiltration compared to conventional treatment

Micro- and ultrafiltration offers several advantages compared with conventional filtration
methods. As an example, ultrafiltration used as a pretreatment step to treat water has
experienced an impressive increase as a result of the continuous search for cost-effective
technologies which enable a sustainable production of water (Chu et al., 2009). Key benefits
associated to the ultrafiltration technology versus conventional pretreatment are a low
footprint, the ability to remove virus and bacteria and to significantly reduce colloids,
suspended particles, turbidity and some total organic carbon. Even more importantly, the
ability to reliably provide good quality filtrate water to the downstream reverse osmosis are
the most remarkable benefits associated with this technology (Mourato et al., 2003).

28
 Chapter 2

As another example Milwaukee, on the United States of America suffered back in 1993 on
of the worst modern water-born epidemies due to Cryptosporidium microscopic parasite
causing diarrhea. Cryptosporidium is known for being highly resistant to chlorination.
This can pose a risk when conventional water treatment schemes based on sand filter and
chlorination are used (Morris et al., 2005).

Since then, ultrafiltration has emerged as a preferred technology to treat drinking water, thus
replacing old water treatment schemes based on sand filters and chlorination. Ultrafiltration,
as it is based on an absolute pore-size filtration, can eliminate Cryptosporidium or other
chlorine-resistant organisms, as they cannot pass through the membrane (Pressdee, 2005).
Additional benefits of ultrafiltration versus sand filters are summarized in the table 1.

Table 1 Summary of the benefits of ultrafiltration over sand filters (DuPont, 2022)
Sand Filtration Ultrafiltration
Pathogenic bacteria removal (coliforms) ≤2 log ≥ 4 log
Pathogenic virus removal (enteric) ≤ 3 log ≥ 4 log
Water effluent quality (Turbidity) 0.1 NTU 0.01 NTU
Organics removal (TOC) 5% 34%
Lower Silt Density Index 3.2 2.6
Footprint reduction No 99%
Improve plant availably (reduce downtime) 95% > 98%
Operate reverse osmosis at higher flux < 14 L/m²h > 14 L/m²h

Most of the advantages of ultrafiltration rely on the small pores for water transport, typically
in the range of 20-30 nm. This allows the system to be significantly more compact, reducing
the required footprint by up to 99% compared to a conventional sand filter. Additionally,
thanks to this smaller pore size, it is very challenging for bacteria, parasites, protozoa and
even viruses to pass through the membrane. This greatly improves the water effluent
quality produced. When ultrafiltration is used as a pretreatment to the reverse osmosis, this
means that the reverse osmosis system can operate at a higher flux, as particulate fouling risk
is eliminated.

2.2 DESIGN AND OPTIMIZE MEMBRANE PROCESSES

Before setting up a micro- and ultrafiltration process, the following five points must be
considered (Raghunath et al., 2012)
1) Objective of the process and the success criteria
2) Pre-treatment of feed
3) Membrane selection
4) Module selection
5) Operation parameter optimization

29
Experimental Methods for Membrane Applications

It is usually not possible to follow the points chronologically, as e.g., a proper pre-treatment
depends on choice of membrane, membrane module and operational parameters. In the text
the objectives of the membrane will be discussed first, then membrane selection, operation,
membrane module and finally pre-treatment.

2.3 OBJECTIVE OF THE FILTRATION PROCESS

Ultra- and microfiltration are used in many industries, at water facilities and wastewater
plants. The objective of the process varies, and the relevant stream is not the same for all
filtration processes but can be
1. Concentrate e.g., for recovery of macromolecules,
2. Filtrate e.g., for removal of pathogenic microorganism from drinking water
3. Both concentrate and filtrate e.g., for harvesting valuable product from waste stream by
fractionation of macromolecules.

Feed Concentrate
Qf ,Cf Qr ,Cr

Filtrate Qp ,Cp

Figure 2 Membrane filtration process and mass balance.

This is important because it affect the optimal choice of membrane, modules, and operation
parameters. In order to define the success criteria different key equations and parameters
will be defined.

The optimal design of the membrane process depends on the flow and concentrations
of the inlet, the wanted flow and concentration of the outlets, and the operation time of
the membrane system (Figure 2). Q is the volumetric flow and C is the concentration of
the compounds of interest. A key factor for membrane operations, is most particles or
macromolecules (rejected materials) are concentrated in the concentrate stream, which can
be found by defining a concentration factor.

Qf Eq. 1
CF =
Qr

The concentration factor should be high and concentration factors up to 10 is typically set as
the goal. The concentration factor gives a maximum value for how much the product can be
concentrated and if the particles are fully rejected by the membrane, Cr = CF×Cf.
Another key parameter is the yield (Y)

Qr Cr Eq. 2
Y=
Qf C f

30
 Chapter 2

i.e., fraction of the wanted product that are left in the concentrate. The yield should be high
and ideally 1.

Filtrate may be the main product stream i.e., if the membrane is used for water treatment. CF
must then be high as well because as much water as possible have to be recovered, but often
the recovery is calculated instead, as it represents the water that is being produced compared
to the water that is being fed into the ultrafiltration membrane

Qp 1 Eq. 3
%Recovery = =1
Qf CF

Another important parameter is availability. Availability represents the amount of time, in

percentage, that the ultrafiltration is operating. Availability is defined as

toperation Eq. 4
% Availability =
toperation + tstopped

This is an important factor, since as an example, if the membrane needs a lot of time to be
properly cleaned, the overall ultrafiltration efficiency will be low.

%Efficiency = % Recovery % Availability Eq. 5

Filtrate flux (Jw) across a membrane can be described as the filtrate flow (Qp) normalized by
the membrane’s active area (A). Its units are typically expressed as liters/(m2·h) or LMH, or
as US gallons/(ft2·day) or GFD.

Op Eq. 6
Jw =
A

One useful parameter to calculate is the net flux (Jnet). This represents the net flux a micro-
and ultrafiltration installation is producing in a whole year. This takes into consideration
the time installation is operating after discounting the time the plant is stopped, and the net
water it produced after discounting the water that was produced but later consumed during,
for example, cleanings. It can be calculated by multiplying the design flux by the efficiency

Jnet = J w %Efficiency Eq. 7

This parameter is important since sometimes, in order to optimize the plant throughput
production, it might be, for example, more beneficial to increase the design flux while
decreasing a bit the recovery and availability, if this can help getting an overall higher net
flux. The net flux can be used to calculate the required membrane area

31
Experimental Methods for Membrane Applications

2.4 MEMBRANE TYPES

The selection of the membrane is important for the filtration process, and different thing
must be addressed. It is important to have a high flux through the membrane and a high
rejection of particles and macromolecules. However, high rejection usually results in
lower flux or higher energy demand, so often these two key parameters must be balanced
depending on the propose of the filtration process. It is important to have a membrane with
minimal risk of fouling or clogging, and a membrane with high lifetime, which depends on
the composition of the membrane, surface properties and pore size.

Different types of membrane modules exist, such as flat sheet membranes (e.g., 1×1 m and
200 μm thick), spiral wound membranes, hollow fiber membranes and tubular membranes.
Module length is typically around 1 m and the membrane area for one module can for large
installation be up to 40-80 m2.

Ultrafiltration and microfiltration can be operated either outside-in, or inside-out,

depending on the type of features that is mainly desired. Another classification criteria
depends on whether all the water that goes into the membrane is being treated, this is
called dead-end filtration; or if a portion of the water is not being treated, it is called cross-
flow filtration. Cross-flow filtration has the advantage of better managing fouling and
lower energy consumption, but at the expense of a reduction on water recovery. Dead-end
filtration has the advantage of having a higher water recovery and a smaller footprint, but
fouling is more difficult to control.

Outside-in membranes are hollow fibers characterized by having the water to be treated
outside the fiber, and the filtrated water collected inside the hollow fiber. These are typically
made of Polyvinylidene fluoride (PVDF). They are usually operated in dead-end mode. Key
features of this technology involve a higher resistant to chemicals, and the ability to perform
air scouring during cleanings.

Inside-out membranes are hollow fibers characterized by having the water to be treated
outside the fiber, and the filtrated water collected inside the hollow fiber. These are typically
made of Poly(ether-Sulfone) (PES). They are usually operated in dead-end mode. Key
feature of this technology is a higher membrane permeability.

Additionally, ceramic membranes are also used in the industry. Ceramic membranes are
usually made of Aluminum, Silica, Titanium, and Zirconium oxides. They are typically
operated in cross-flow filtration (Gruskevica and Mezule, 2021). One of the key challenges
of ceramic membranes are its cost, and managing fouling properly. Key features of this
technology are its thermal and chemical stability, as well as being able to operate at higher
fluxes (Sondhi et al., 2003).

Membranes system can also be classified as submerged and pressurized system. Submerged
membranes have the advantage that they can be visually observed, and they usually operate
at lower energy. Pressurized membranes have the advantage of operating at higher fluxes, a
lower footprint (Nick, 2019) , but might suffer from higher fouling rates, as a possible result
of higher compaction of the fouling layer. (Kim et al., 2015)

32
 Chapter 2

The use of coagulation in microfiltration and ultrafiltration can be used as a way to reduce
fouling rates and increase overall flux (Konieczny et al., 2009). Coagulants are typically iron
or aluminum based.

2.5 BASIC EQUATIONS

The filtrate flux (Jp) must be high and is a function of permeability and transmembrane
pressure (TMP)

TMP
J p = K TMP = Eq. 8
Rm

where K is the permeability (or more correctly the permeance) in m/(s·Pa), Rm is the
hydraulic resistance of the membrane in m-1, and η is the viscosity in Pa·s.

For filtration of pure water, the permeability is only dependent on membrane properties
and can be calculated from membrane thickness, pore numbers and size.

i
(ni di4 ) Eq. 9
K=
128 ητl

Where n is number of pores per square meter, di is pore diameter, or hydraulic diameter for
non-cylindrical pores (m), τ is tortuosity (-) and l is membrane thickness (m).

The ideal membrane is thin, to ensure high permeability, but thick enough to withstand
the transmembrane pressure. To improve permeability but still ensure the mechanical
properties asymmetric membrane or thin-film composite membrane are usually used
(Figure 3). Composite membrane consists of a thin active layer, where the separation
happens, and a support layer with low hydraulic resistance.

Asymmetric membranes Thin film composite

Figure 3 Anisotropic membranes

Notice that the number of pores per square meter is another important parameter and
implicit given from the porosity of the membrane.
Besides flux, membrane rejection (R) is important and is defined as

Cp
R =1 Eq. 10
Cr

where R is a number between 0 and 1, where all particles are rejected by the membrane if
R = 1 whereas the particles can freely pass the membrane if R = 0

33
Experimental Methods for Membrane Applications

For some applications , it is important that some molecules pass the membrane and other are
retained by the membrane. The selectivity (α) can then be calculated and used to compare
different type of membranes

1 Ri Eq. 11
i/ j
= and i/ j
1
1 Rj

Where i is the particle that should pass the membrane, and j is the particles that should be
rejected by the membrane. Several particles or macromolecules are partly rejected by the
membrane. The reason for this is that pores are not unisized, but a large pore size distribution
are often observed (Figure 4).

0.25

0.20

0.15

0.10

0.05

Distribution frequency 0
0.01 0.1 1.0 10 100
Pore size (µm)

Figure 4 Pore size distribution microfiltration membranes (Data obtained from Liu et al., 2018)

For water treatment and sterilization, it is important to ensure that bacteria cannot pass the
membrane. Here the size of the largest pores is critical. A method to determine the size of
the largest pores or ensure that all pores are below a given size is the bubble point method
(breakthrough pressure). An alternative method is the water intrusion procedure, which is
described elsewhere (see ‘water intrusion procedure’).

For wetted membranes, the air pressure must overcome the capillary pressure of the pores,
before liquid can pass the pore (Figure 5).

Experimental setup Individual pore

Water
Wetted membrane

N2

Figure 5 Bubble point method

34
 Chapter 2

The pressure required to overcome the capillary pressure is called breakthrough pressure
(P*)

4γ Eq. 12
P* =
d

where γ is the surface tension of the liquid. The surface tension for water is 0.072 N/m.
For ultrafiltration membrane it is more complicated to measure pores size, instead cut-off
values are often determined instead. MWCO can be determined by filtering mixtures of
macromolecules and measuring rejection of the molecules (Figure 6). It is thereby possible
to determine the size of the molecules where more than 90% of the molecules are retained
by the membrane.

100

80

60

40

20
cut-off
Rejection (%) 0
0 50 100 150 200 250 300 350 400 450
Molecular weight (kDa)

Figure 6 Molecular weight cut-off of PES/PSFNA membrane measured by using a mixture of PEG
molecules (Data obtained from Wu et al., 2018)

The selection of the membrane depends on feed composition and required quality of
permeate or molecules that must be concentrated. The following parameters for the
membrane must be considered

1. Selectivity (pore size and cut-off) so the membrane reject the particle that should be
concentrated or removed
2. Permeability (high flux through the membrane at low pressure)
3. Thermal stability
4. Mechanical properties
5. Chemical stability (resistance for cleaning process)
6. Membrane composition and surface properties (to avoid adsorption of foulant)

2.6 NORMALIZATION

Filtrate flow permeating through a micro- and ultrafiltration membrane is heavily influenced
by the water temperature. This happens as the process of water moving through convection
across the membrane pores is facilitated by a higher temperature, as water then becomes less

35
Experimental Methods for Membrane Applications

viscous, and less energy (TMP) is required for water to pass through the pores. On the other
side, a lower water temperature increases the water viscosity and therefore more energy
(TMP) is required for water to pass through the membrane pores.

Therefore, data normalization is of utmost importance in microfiltration and ultrafiltration

membranes to properly assess the performance of a membrane-based installation. As an
example, if data would not be normalized, one could observe an increase over time of raw
TMP. This could induce the plant engineer to think that there might be irreversible fouling
developing over time on the membrane system. If then this data would be operated, one
could see that this TMP increase over time is due to a decrease in temperature over time
as a result of moving to winter season. After normalizing this data considering water
temperature, it could be observed that TMP evolution over time, thus dismissing any
fouling issue.

The normalized TMP (TMP*) is determined by multiplying the raw TMP by the temperature
correction factor (TCF)

TMP* = TMP TCF Eq. 13

The purpose of the TCF is to take into consideration the effect of the Temperature (T) in
Celsius degrees and its influence on the viscosity of water, and normalizing its value from
the temperature the water has, to an ideal temperature of 25 ºC (Daucik and Dooley, 2008)
247.8
25 + 273.16 140 Eq. 14
TCF = 10 247.8
T + 273.16 140
10
It should be noticed that filtrate flux can also be normalized. This can be especially important
if a system is operated at constant TMP, as then Flux will change over time. It might also be
relevant in order to assess the stress that the ultrafiltration is operated, as the higher the flux,
the more stressed the system will be, and the more difficult it would be to be able to operate
the system under a sustainable way

TMP* = TMP TCF Eq. 15

Special attention should be put in order to not normalize the filtrate flux twice, as normalize
filtrate flux needs to be calculated using a non-normalized TMP, and then applying the TCF
in the equation described above.

2.7 MEMBRANE FOULING

Membrane fouling is an inevitable phenomenon in membrane filtration. Membrane

filtration occurs when particles are deposited on a membrane surface or in membrane pores
in a process. Fouling reduces the permeability of the membrane and thereby increase the
required pressure to keep the flux constant. Further, fouling may change the selectivity

36
 Chapter 2

of the membrane. This may in some cases be beneficial i.e., it may result in better water
treatment. However most often all type of fouling must be minimized, and different
methods exists to reduce fouling build up or remove already formed fouling. Microfiltration
is typically operated at low pressure (up to some bars) or flux. Too high flux increases the
risk of accumulation of materials at the membrane surface. Thus, pressure must be kept
low to avoid too high flux and thereby reduce fouling risk. Ultrafiltration membranes is
operated at higher pressure to obtain same flux through membrane due to the smaller pores
and higher hydraulic resistance. Micro- and ultrafiltration membrane are typically cleaned
hydraulically or chemically. Cleaning reduces the operation time of the membrane whereby
the availability of the membrane is reduced cf., Eq. (4).

Membrane fouling can be due to internal fouling such as adsorption of materials in the pores
or pore blocking, and it can be external such as gel formation, precipitation of salts (scaling)
and biofilm growth. The type of fouling is important for the optimal operational parameter,
cleaning strategy and different method exists to analyses fouling (later chapters). Besides
direct analysis of the fouling materials and the membrane, flux-pressure data can give some
hint of the type of fouling and is useful to optimize operational parameters such as flux,
pressure and cross-flow.

If flux-pressure data is used to quantify fouling, the resistance-in-series model is usually

used to determine individual contributors to resistance (Géasan-Guiziou et al., 1999) . One
method is to separate fouling into reversible fouling (Rrev) or irreversible fouling (Rirrev).
Reversible fouling can be removed by hydraulically cleaning whereas irreversible fouling
cannot.

For fouled membrane, the water flux through the membrane can be described as
P
Jp = Eq. 16
η i
R

Where the resistance is the sum of individual contributors to the hydraulic resistance

∑ R= R
i m
+ Rirrev + Rrev Eq. 17

Membrane resistance can be determined for a pristine membrane by filtering clean water
(i.e., RO treated water)
P
Rm = Eq. 18
ηJ w,0

Determination of membrane resistance is critical to determine the other resistances

Irreversible fouling can be determined by filtering the suspension, clean the membrane and
measure the water flux after cleaning.

37
Experimental Methods for Membrane Applications

P Eq. 19
Rirrev = Rm
ηJ w,cleaned

Irreversible fouling is often ascribed to internal fouling i.e., pore blocking, and adsorption,
and is problematic as it results in a permanent reduction of the membrane permeability.
Further, it is difficult to avoid by changing operation condition of the filtration process. It
depend on the type of membrane chosen i.e., pore size distribution and composition. Thus,
if the irreversible resistance is too high, another membrane may be chosen, or feed must be
pretreated prior to the filtration.

Reversible fouling usually increases continuously during filtration and can be calculated as
P
Rrev = Rm Rir Eq. 20
ηJ w

The reversible fouling strongly depends on the operational conditions and will be discussed
more deeply in next section.

2.8 SUSTAINABLE FLUX

Most filtration processes are operated at constant flux to have a stable output, i.e., a constant
cross-flow (CF). The flux must be set so the energy cost is low, and the required membrane
are is low. Often the system is operated so build-up of reversible fouling is low. Reversible
fouling is a result of external fouling and strongly dependent on the transport of particles
and molecules to the surface of the membrane. High flux increases the transport of material
to the surface and thereby increase fouling build-up. At low pressures, material transport to
the surface is low, and flux increase almost linearly with pressure also called the pressure-
controlled region (Figure 7). At high pressure, the effect of increasing the pressure becomes
less important for the flux. If pressure is increased, flux increase as well but decline to a lower
steady state flux. Thus, performance of the membrane process can no longer be improved
by increasing the pressure (pressure-independent region). Turbulence and high shear at the
membrane surface can remove part of the external fouling layer and thereby reduce fouling
build-up. A common method is to apply a high crossflow (up to 2-3 m/s) to reduce fouling
built up and increase flux. Besides high crossflow shear can be induced by using turbulence
promoters, rotating, or vibrating membrane or air scouring. If the concentration in the feed
is low, dead-end, or semi-dead-end setup can be used to lower the energy consumption.
At higher concentration feeds crossflow are required and for high viscous feed, rotating
membrane are may be the most energy efficient solution.

A critical flux has been defined as the transition from the pressure-controlled region to the
pressure independent region (Cleck et al., 2003, Bacchin et al., 2009). In the ideal case, the
filtrate flux follows that of clean water until the critical flux, but often the filtrate flux is
lower than the pure water flux due to irreversible fouling. The optimal flux is typical lower
than the critical flux; a number around 75% of the critical flux can be used (Raghunath et al
2012).

38
 Chapter 2

Water flux

Increased crossflow
Critical flux

Pressure-controlled region Pressure-independent region

Permeate flux (LMH)
Transmembrane pressure

Figure 7 Flux-pressure curve and definition of critical flux

For long operation, membrane performance will usually decline also if the flux is lower than
the critical flux. An alternative to the critical flux, is therefore the concept of sustainable flux.
The sustainable flux is the maximum flux at which the fouling rate is acceptable and can be
handled by hydraulic and chemical cleaning. At constant flux the fouling rate (FR) can be
defined as

ΔTMP Eq. 21
FR =
Δt

By assuming that the transmembrane pressure increases linearly with time. Notice at high
flux, a TMP jump may be observed where the fouling rate increases with time. At lower flux,
it is usually reasonable to assume a linear relationship between TMP and time at constant
flux. The sustainable flux can be determine using the flux step method, where the flux is
gradually increased, and FR measured. The sustainable flux is then defined as the point
where FR is exceeded a given threshold value for the fouling rate.

Threshold
fouling rate

Flux Fouling rate

Time Flux

Figure 8 Step-flux test for determination of sustainable flux (Modified from Wang et al., 2014)

39
Experimental Methods for Membrane Applications

Critical and sustainable flux is higher increased at higher cross-flow i.e., for microfiltration of
lactalbumin suspension, the critical flux increase a factor of three, when the cross-flow was
increased from 0.5 to 4-5 m/s (Vyas et al., 2002). However, a high crossflow is costly; thus
there is a balance with high flux vs. high cross-flow. Secondary effluent from wastewater
treatment have been treated by membrane filtration, and the membrane have been tested
at different flux and cross-flow. The optimal crossflow has been determined to 1.5 m/s for
polymeric membranes and 4.5 m/s for ceramic membrane as the ceramic membranes was
more expensive than polymeric one (Owen et al., 1994).

2.9 MEMBRANE DESIGN AND MODULE

A typical membrane setup is the feed-and-bleed process (Figure 9). The system consists
of a feed pump that ensure the transmembrane pressure and a constant flow of feed to
the system. The recirculation pump is added to ensure a high crossflow at the membrane
surface. The crossflow can be up to 2-3 m/s. At higher crossflow, the energy consumption
increases rapidly. Due to the high crossflow, the pressure drop along the membrane module
may decline significantly and in worst case, the transmembrane pressure becomes negative
at the end of the module. The permeate flux is usually kept constant and the transmembrane
pressure regulated to ensure a constant flux.

Retentate Valve 1

Qr ,Cr

Qp ,Cp
Membrane

Valve 2 Permeate

Valve 3

Qf ,Cf
Chemicals Backwash tank
Feed
Feed Prefilter Recirculation
pump pump

Figure 9 Feed-and-bleed filtration set-up.

For a feed-and-bleed system, the concentration of particles or macromolecules in the

retentate can then be calculated as

CF Eq. 22
Cr = C f
1− (1− R)(CF −1)

Where Cf and Cr is the concentration of particles in the feed and retentate, respectively. CF is
the concentration factor and given as the ratio between the feed flow (Qf) and the retentate
flow (Qr). If the particles are fully retained by the membrane, then R = 1 and Cr = Cf × CF .

40
 Chapter 2

The yield can be calculated as

1
Y= Eq. 23
1− (1− R)(CF −1)

2.10 PRETREATMENT

Pre-treatments are often required to prevent fouling, and is a necessary step for most
membrane filtration process. Often bigger solids must be removed. In this case,
macrofiltration is used. Its aim is to reduce big solids such as branches and leaves, that
could otherwise potentially clog the ultrafiltration membranes and reduce its effectivity.
Different methods exist, being the most typical screening, the use of hydro-cyclones,
prefiltration using cartridge filters or multimedia filters. Suspended particles and colloidal
particles can be problematic in filtration as such particles are difficult to remove from the
membrane surface. Thus, it may be necessary to flocculate particles before the pre-filtration.
Flocculation is usually done by adding salts (ferric, aluminum or poly-aluminum salts),
polymers or combinations, whereby particles aggregates and can be removed by filters.
As an example, coagulation and prefiltration have been used for treatment of raw water: a
coagulation using iron coagulant (FeCl3) with anionic polyelectrolyte in the first step and
aluminum coagulant in the second one was used before raw water enters the pre-filtration
(Sakola and Konierczny, 2004). The pre-treatment reduces the negative effect of membrane
fouling, but also improved the quality of the treated water (Sakola and Konierczny, 2004)

2.11 CLEANINGS

Microfiltration and ultrafiltration is a typically a semi-batch process, specially for dead-end

filtraiton applications. This is because it mainly handles particulate fouling, it gets fouled
quickly. This means that the system needs to operate for 20 to 90 min depending on the
water type and undergo a cleaning to restore its permeability. Optimizing cleanings remains
thus of outmost importance to maximize the efficiency of the filtration system (Gilabert-
Oriol, 2021). The main type of fouling that micro- and ultrafiltration membrane experience
is detailed below. The most frequent one is particulate fouling. This is the most common
and occurrent one. As the main task of an ultrafiltration membrane is to remove particles
represented by total suspended solids (TSS) that give turbidity to the water. As water is
filtrated through the membrane, it gets clogged by particles. Therefore, is it of utmost
importance to perform a backwash to the membrane, so these particles can get detached
from the membranes and its permeability can be restored.

A backwash (BW) typically consists of multiple steps. These steps can consist of an air scour,
which blows air across the membranes, and shakes them to create abrasion and detach foulants
accumulated on the top of the fibers. Later, a draining can be done to empty the module and
remove the detached foulants. Then, a backwash top can be performed, together with an
Air Scour. The main purpose of the backwash is to removed particles that are blocking the
pores. Water with foulants is removed through the concentrate side of the module located
at the top part of it. A backwash bottom follows the same approach, but water with foulants
is evacuated through the bottom part of the module. It is not practical to perform an air scour

41
Experimental Methods for Membrane Applications

in this step, as air is blown from the bottom of the element, and this would interfere with
the aeration which is also blown from the bottom of the module. Finally, a forward flush
is performed in order to provide a shear force and remove any remaining foulant from the
top of the membrane. Key steps in this backwash process are performing a backwash top
with and air scour, followed by a forward flush. By optimizing the duration of each one of
these sub steps, as well as the frequency of backwashes, the overall efficiency of the whole
ultrafiltration treatment can be greatly optimized (Gilabert-Oriol et al., 2021).

A chemical enhanced backwash (CEB) consists of a typical Backwash sequence, where

sodium hypochlorite (NaOCl) is dosed. As filtration cycles are done, bacteria start to grow
and develop a biofilm. This biological fouling, or biofilm, is noted typically after 1 to 2 days
of operation. It can be assessed as after certain number of backwashes, the TMP cannot be
recovered to its initial values.

A cleaning-in-place (CIP) consists of a tailor-made chemical cleaning. It is typically

performed every 3 to 4 months. As the membranes get cleaned on a regular basis, very
specific foulant specific to the water type being treated starts to slowly build up over time.
Typical CIPs performed are a caustic CIP, used to remove organics being accumulate on the
membrane, where NaOH is used; and acid CIPs, used to remove scaling or metallic-based
foulants, where HCl or oxalic acid is used.

2.11.1 Optimization of hydraulic cleaning

In order to remove fouling, hydraulic cleaning strategies can be used such as backwash,
backflush or relaxation. Relaxation is done by closing the valve 1 and maybe valve 2 at the
permeate side in Figure 9, whereby the transmembrane pressure drops to zero. Backwash
or backflush are done by opening valve 3 and pump permeate through the membrane in
Figure 9. Backwash and relaxation are typically done at regular intervals after 30-90 min
and the duration is typical 1-3 min. Backflush are shorter cycles e.g., 30-60 s flush every 15
min. Regular hydraulic cleaning removes the reversible fouling. The efficiency of backwash,
backflush or relaxation can be calculated as
TMPf TMP0
h= Eq. 24
TMPf TMPi

Where TMPi is the pressure at the beginning of the filtration cycle, TMPf at the end of the
filtration cycle and TMP0 after hydraulic cleaning.

Another method to find the best strategy for hydraulic cleaning is to calculate the net flux
(Jnet) defined in Eq. (9). An example of an optimization of relaxation time for a membrane
bioreactor shows that the optimal relaxation time is 3 min (Figure 10). Relaxation have been
used in a membrane bioreactor setup used to treat municipal wastewater. For example for
MBR systems with flat-sheet membrane, because backwash is not possible for these types
of membranes as it will destroy the membranes. Thus, relaxation have been used as a gentle
cleaning method to keep the flow high.

42
 Chapter 2

300 70
Optimum average flux
250 60

200
50
150

Net flux (LMH)

40
100
Flux (LMH)

30
50

0 20
0 1 2 3 4 5 0 5 10 15 20 25 30
Time (h) Relaxation time (min)

Figure 10 (a) Relaxation experiment at constant pressure during operation. (b) Net flux was
calculated including both operation time and relaxation time in calculation. Optimum flux
was obtained after 3 min. (Christensen et al., 2016) The process was operated at constant
transmembrane pressure.

At long term operations, the performance of the membrane will usually decline, and
hydraulic cleaning is not sufficient. In these cases, it is necessary to use chemical to clean
the membrane, one method is to add chemical to the backwash water known as chemically
enhanced backwash (Figure 9), but the membrane can also be chemical cleaned from the
feed side.

2.12 MEMBRANE CASCADES

Membrane system are designed to minimize the required membrane area. Membrane
fouling are more pronounced at higher dry matter concentration of the feed. Further, in feed
and-bleed system the concentration of feed near the membrane surface is almost the same
as in the retentate. Thus, it will sometimes beneficial to operate more membranes in series.
Modules in series reduce membrane area demand and lower energy consumption.

Stage 1 Stage 2 Stage 3

Figure 11 Feed-and-bleed in series to reduce required membrane area

An example is given for microfiltration of milk assuming the filtrate flux is given as

L L
J p = A+ B ln(CF ) = 40 2
−14 2 ln(CF )
mh mh

43
Experimental Methods for Membrane Applications

Where the flux decreases with increasing dry matter content in the feed and therefore with
the concentration factor (CF).

Table 2 Three stage feed-and-bleed system to treat 5000 L/h feed and concentrate it with a factor
of 5
Stage CF Qp (L/h) Jw (L/(m2h)) Area (m2)
1 1.5 1,667 34.3 48.0
2 2.5 1,333 27.2 49.1
3 5.0 1,000 17.5 57.2
Sum 5.0 4,000 154.9

Assuming a feed flow of 5000 L/h, rejection of SS is 1, and a required CF of 5, a single-stage

feed-and-bleed system will result in a permeate flow of 4000 L/h and filtrate flux of 17.5 L/
(m2h). Hence the required membrane area is 229.0 m2. This membrane area can be reduced
to 154.9 m2 for at three-stage feed-and bleed system (Table 2). This reduces the required
membrane area with more than 30%.

2.13 SUMMARY

Microfiltration and ultrafiltration are used to treat water, as well as for concentration of dry
matter or macromolecules. Membranes are selected based on their rejection, but it is also
important to consider the risk for membrane fouling, and the chemical and mechanical
stability of the membrane. Adsorption of molecules or pore blocking can sometimes be
avoided by selecting an alternative membrane.

Most operation is done at constant operating flux, since most plants are designed to produce
a certain amount of treated water per day. When designing the installation, it is important
to set up the operate at a flux below the sustainable flux. This helps to avoid a too steep
increase on trans-membrane pressure, and helps overall operating the microfiltration and
ultrafiltration plants in a sustainable way. Higher permeate flux can be obtained with a
high crossflow at the membrane surface, typical up to 2-3 m/s, but it also increases energy
consumption and increase cost. Membrane processes can be done as multi-stage operation
to reduce the required membrane area.

44
 Chapter 2

2.14 REFERENCES

Bacchin P. Aimar P., Field R.W. (2006) Critical and sustainable fluxes: Theory, experiments and
applications. Journal of membrane science 281, 42-69.
Christensen ML., Bugge TV. Hede BH, Nierychlo M. Larsen P. Jørgensen M. K. (2016) Effects of
relaxation time on fouling propensity in membrane bioreactors. Journal of Membrane Science.
504, 176-184
Chu R., J. Wei, M. Busch, Economic evaluation of UF+SWRO in seawater desalination. Chinese
Desalination Association Conference (2009), Qing Dao, China.
Daucik K., R.B. Dooley, Revised Supplementary Release on Properties of Liquid Water at 0.1 MPa, The
International Association for the Properties of Water and Steam, September 2008.
DuPont WaterApp, Pretreatment Advisor, Dupont (2022-09-09).
Gésan-Guiziou G., Boyaval E., Baufin G. (1999) Critical stability conditions in crossflow microfiltration
of skimmed milk: transition to irreversible deposition. Journal of membrane science 158
211-222.
Gilabert-Oriol, G. (2021). Ultrafiltration Membrane Cleaning Processes. In Ultrafiltration Membrane
Cleaning Processes. De Gruyter.
Gilabert Oriol, G., Hassan, M., Dewisme, J., Busch, M., & Garcia-Molina, V. (2013). High efficiency
operation of pressurized ultrafiltration for seawater desalination based on advanced cleaning
research. Industrial & Engineering Chemistry Research, 52(45), 15939-15945.
Gruskevica, K., & Mezule, L. (2021). Cleaning methods for ceramic ultrafiltration membranes affected
by organic fouling. Membranes, 11(2), 131.
Kim, Y. J., Jung, J. W., & Lee, S. (2015). Comparison of fouling rates for pressurized and submerged
ultrafiltration membranes. Desalination and Water Treatment, 54(13), 3610-3615.
Konieczny, K., Sakol, D., Płonka, J., Rajca, M., & Bodzek, M. (2009). Coagulation—ultrafiltration system
for river water treatment. Desalination, 240(1-3), 151-159.
Le Cleck P., Jefferson B., Chang I.S. Judd S.J. (2003) Critical flux determination by the flux-step method
in a submerged membrane bioreactor. Journal of membrane science 227, 81-93.
Liu, J., Huo, W., Zhang, X., Ren, B., Li, Y., Zhang, Z., Yang, J., (2018) Optimal design on the high-
temperature mechanical properties of porous alumina ceramics based on fractal dimension
analysis. Journal of Advanced Ceramic, 7, (2), 89-98.
Morris, Robert D., Elena N. Naumova, and Jeffrey K. Griffiths. ‘Did Milwaukee experience waterborne
cryptosporidiosis before the large documented outbreak in 1993?.’ Epidemiology (1998): 264-
270.
Mourato D., M. Singh, C. Painchaud, R. Arviv, Immersed membranes for desalination pre-treatment,
International Desalination Association (IDA) World Congress, Bahamas, Paradise Island,
Bahamas, 2003.
Nick, N., (2019) Pros And Cons Of Different Types Of Ultrafiltration Technology Configurations,
Weter Online.
Owen G., Bandi M. Howell J.A. Churchouse S.J. (1994) Economic assessment of membrane processes
for water and wastewater treatment. Journal of membrane science 102, 77-91.
Pressdee, J., Rezania, S., & Hill, C. (2005). Minneapolis Water Works’ ultrafiltration plant gets off to a
big start. Journal-American Water Works Association, 97(12), 56-63.

45
Experimental Methods for Membrane Applications

Raghunath B., Bin W,m Pattnaik P, Janssens J. (2012) Best practice for optimization and scale-up of
microfiltration TFF processes. Bioprocessing Journal. 11 (1) 30-39.
Sakola C, Konieczny K (2004) Application of coagulation and conventional filtration in raw water
pretreatment before microfiltration membranes. Desalination 162 61-67.
Sondhi, R., Bhave, R., & Jung, G. (2003). Applications and benefits of ceramic membranes. Membrane
Technology, 2003(11), 5-8.
Vyas H.K. Bennett R.J. Marshall A.D. (2002) Performance of crossflow microfiltration during constant
transmembrane pressure and constant flux operations. International dairy journal, 12, 473-479
Wang Z Field RW., Qu F., Han Y., Liang H Lia G., (2014) Use of threshold flux concept to aid selection
of sustainable operating flux: A multi-scale study from laboratory to full scale. Separation and
Purification Technology 123 (2014) 69–78.
Wu X., Xie Z., Wang H., Zhao C., Ng D., Zhang K (2018). Improved filtration performance and
antifouling properties of polyethersulfone ultrafiltration membranes by blending with
carboxylic acid functionalized polysulfone RSC ADV 2018, 8, 7774-7784.

46
 doi: 10.2166/9781789062977_0047

Chapter 3

Reverse Osmosis and

Nanofiltration
Guillem Gilabert-Oriol, DuPont Water Solutions, Spain

The learning objectives of this chapter are the following:

• give an understanding of the importance of reverse osmosis and nanofiltration

membranes and how they are used for water treatment

• give an introduction to the different equations that govern reverse osmosis

fundamentals and how to control, analyze and normalize a membrane installation

• give knowledge of how reverse osmosis and nanofiltration membranes are designed

• give knowledge on how membranes system is best operated to reduce membrane

fouling and ensure system availability.

3.1 THE RISE OF REVERSE OSMOSIS

Water scarcity is being recognized as one of the main threats that mankind is facing globally
(Fritzmann, et al., 2007). Reverse Osmosis (RO) membrane technology has developed as
a promising technology to address this problem, holding roughly 44% market share and
growing among all the desalinating technologies (Valavala, et al., 2011). This increased
market adoption has been driven as materials have been improved and costs have dropped
(Greenlee, et al., 2009. This is especially relevant in arid regions such in the Middle East
countries (ME), where population is located in arid and semi-arid regions, with a very limited
rainfall, and where due to high ambient temperatures, evaporation contributes to a higher
stress degree to the naturally available water sources. Moreover, water scarcity is aggravated
by the population increase this region is exposed, as well as the economic development
(Guo, et al., 2000). All these factors, together with the favorable energy to product quality
ratio that seawater reverse osmosis (SWRO) offers, has situated this technology as one key
driver to sustain population living standards in ME countries (Carroll, et al., 2000).

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

The first technologies used to desalinate seawater were using thermal processes where
seawater is evaporated, and then the steam, which is free of salts, is recondensed to obtain
fresh water. These thermal driven technologies that rely on distillation are multistage-flash
distillation (MSF), multiple-effect distillation (MED), and vapor compression desalination
processes (VCD). The main drawback of these methods is the significant amount of energy
per cubic meter of water produced, compared to modern reverse osmosis based desalination.
As Figure 1 shows, reverse osmosis (RO) desalination, specially when coupled with energy
recovery devices (ERD) is 10 times more energy efficient than multistage flash desalination,
and 4 times more efficient than vapor compression distillation (Kumar et al., 2017, Kim
et al., 2019).

30

25

20

15

10

5
3
Energy consumption (kWh/m ) 0
MSF MED VCD RO RO+ERD

Figure 1 Energy consumption of different desalination technologies

One of the key aspects that has allowed reverse osmosis desalination to be so energy efficient
is the introduction of energy recovery devices (Kadaj and Bosleman, 2018). These systems
are like heat exchangers operation units, but instead of exchanging thermal energy, then they
exchange pressure. This allows to recovery almost all the energy that was previously lost in
the concentrate water stream of a seawater reverse osmosis system, and use it to pressurize
the same volumetric flow in the feed of the reverse osmosis system. If we take seawater
desalination as an example, this can reduce the energy expense in a seawater desalination
plant by 55%. Also, a smaller high pressure pump is needed to pressurize the feed of a revere
osmosis system, as 55% of the flow is already pressurized coming from the energy recovery
device. It is worth mentioning that previously, the concentrate stream of a reverse osmosis
system that could come pressurized up to 80 bar was discharge into the open atmosphere
and all this energy was lost.

3.2 SUSTAINAIBLITY OF REVERSE OSMOSIS

Reverse osmosis membranes membrane technology offers a solution to achieve the

sustainability development goals that the United Nations has set up for 2030. This has
been stated and recognized by the United Nations (SDG, 2023). Thanks to the desalination
technology, it is possible to fight against water scarcity, and obtain water of high quality.

48
 Chapter 3

If the energy is powered through renewable energies, such as through solar power or
photovoltaics, it is possible to achieve drinking water of high quality. This allows to fight
against climate change, as well as to provide unlimited amount of drinking water for the
population, that it is not linked to whether it rains or note in the nature.

It is also important to make sure that desalination concentrate discharge is done properly,
and that tis brine is properly managed through diffusors or mixing it with seawater, so its
discharge does not affect marine species or the environment (Fernández-Torquemada et al.,
2019).

The use of brine can also be used as a resource, to get value out of the brine, and being able
to extract valuable minerals such as sodium chloride, magnesium compounds, bromide,
and rubidium, among others. This is important, as this allows reducing the cost of water of
the desalination technologies, as well as preserving natural resources such as landscape and
mountains from invasive mining extraction operations (Casas et al., 2014).

Finally, some endeavors such as the Water Positive initiative, aims to take the sustainability
impact of desalination and water reuse one step further. This is inspired by the carbon credits
system, but for water credits. It aims to help those companies that aim to become water
neutral in terms of its water footprint, so that they can compensate their water negative use,
with those companies that are net producers of water, such as the desalination and water
reuse installations. This can help driving awareness of the importance to reduce the water
footprint, and helping preserve this valuable resource, as well as to help making sustainable
water treatment processes more affordable.

3.3 UNDERSTANDING THE OSMOSIS PROCESS

In order to understand why reverse osmosis is called with this name, it is important to first
understand what osmosis means. Osmosis is a natural process that only takes places when
there is a semi- permeable membrane.

3.3.1 Semi-permeable membranes

A semi-permeable membranes is defined by letting a solvent like water pass through it, but
not letting a solute like salt pass through the membrane. In order to better understand this
process, it can be useful to imagine a simplified scenario, where only water is considered
for a solvent, and only sodium chloride (NaCl) is considered for a solvent. When sodium
chloride is dissolved in water, both sodium and chloride are separated in terms of Na+ ions
and Cl- ions, following the equation detailed in Equation 1.

NaCl → Na+ + Cl − Eq. 1

Although both atomic radius of water molecule (H2O) and Na+ ions and Cl- ions are similar,
and therefore it could be expected that both water and sodium and chloride ions can pass
through the membrane in a similar rate, this is not the case. The reason for this phenomenon,
is the solvation phenomena. In order to maintain electrical neutrality, when ions are
dissolved in water, they become surrounded by water molecules. Since water molecules are

49
Experimental Methods for Membrane Applications

polar, they tend to orient their mostly negative charge, with the sodium positively charged
ions. The same happens for chloride negatively charged ions, as they get surrounded by
the negatively charged side of the water molecules. The ultimate consequence for both
chloride and sodium ions is that their effective sizes drastically increase, as a result of being
surrounded by water molecules. This is the main reason why small molecular weight species
that are charged are much better rejected from a solute like water. Therefore, the smaller
a species is, and the more charged it is, the better it will be rejected by a reverse osmosis
membrane.

The solvation effect is represented in Figure 2, where it can be seen the solvation effect of
water molecules, represented by red (oxygen) and grey molecules (hydrogen) to sodium
ions (blue) and chloride ions (green) as they pass through a space in the reverse osmosis
membrane.

– – – – – –
–
–
– – – –
–
– –
– +
– – – –
– –

Figure 2 Solvation effect of water (red and grey molecules) to sodium ions (blue) and chloride ions
(green) as they pass through a space in the reverse osmosis membrane

It should be noted that this “pore” drawn in this diagram, represent a space that is created
in the polyamide chain. As polymers rotate as a result of being above 0º Kelvin temperature,
small “pores” like this drawn in the diagram are created, where water can pass through it
freely through a pressure driven convective flow (Wang et al., 2023).

3.3.2 The reverse osmosis process

Once the concept of a semi-permeable membrane is defined, and it remains clear that it
lets a solvent like water to pass, while a solute like sodium chloride cannot pass, the natural
process of osmosis can be defined. As mentioned, the osmosis process only makes sense
when a semi-permeable membrane is present. Osmosis is defined as a natural process when
a specie that is at a higher concentration goes to dilute the other side of the membrane that
has lower concentration of this species. This is typically the case of solutes like water. Water
can move through a semi-permeable membrane, while salt cannot. Therefore, water with
lower salt concentration like fresh water will always go to dilute the water with higher
salt concentration like seawater, as this will have a lower water concentration. Since this
natural phenomenon is not the goal of producing fresh drinking water, as it is not desired
to get fresh water consumed. In order to achieve the opposite result, and be able to generate
fresh drinking water from seawater, the reverse osmosis process was invented. This process
consist into applying a pressure on the seawater that is enough in order to reverse this
process, and being able to generate fresh water instead of consuming it.

50
 Chapter 3

Osmosis process happens in nature. One example are cherries. When it rains, rain droplets
cover the cherry skin. The cherry skin acts like a semi permeable membrane, it lets the water
to pass through, but not salts or sugars to escape. Therefore, after it rains, water travels from
the outside of the skin, where the water concentration is higher, towards the inside of the
cherry skin, where the water concentration is lower as a result of the fiber and fructose that
cherries contain. The ultimate result is that as water from rain enters the inside of the cherry,
it can eventually crack the cherry skin as its volume increases, and the cherry skin might
not be able to expand properly before cracking to account for the increase in the cherries’
volume.

In order to visualize this, two different solutions with the same volume each one. These
solutions can be put in contact through a semi permeable membrane. The first solution is
seawater, which it is assumed to have 30 g salt with 970 g of water. This represents a 3.1%
salt concentration. The second solution is fresh water, which it is assumed to have 1 g of salt
with 999 g of water. This represents a 0.1% salt concentration This set up can be observed
in Figure 3. In this case, two solutions are separated. Fresh water will go to dilute seawater,
while salt will not be able to pass.
Seawater Fresh water

30 g Salt 1 g Salt
970 g Water 999 g Water

Time = 0
3.1% salt 0.1% salt

Figure 3 Initial experimental set up

As salt cannot pass, water from the fresh water side will go to dilute the seawater. Water will
keep passing until both sides salt concentration are the same. This will happen after 935 g
of water have passed from the fresh water side to the see water side. At this equilibrium
point, concentrations in both sides will be the same at 1.6%. It is important to notice that
salt mass in both sides is kept the same, but water mass has changed. This change in water
volume, which increases in the seawater side but decreases in the fresh water side, leads to
a difference in water height between the seawater and fresh water side. This difference in
height is what is called osmotic pressure. This osmosis process be seen in Figure 4.

Seawater Fresh water

30 g Salt 10 g Salt
1,905 g Water 64 g Water

935 g Water
Time = [0,∞]
1.6% salt 1.6% salt

Figure 4 Osmosis process

51
Experimental Methods for Membrane Applications

If one aims to reverse this naturally occurring osmosis process, one needs to apply a pressure
that at least is the same as this osmotic pressure. Once this pressure is applied, the water
flow is reverses. If exactly the same pressure as the osmositc pressure is applied, it will be
possible to reverse the 935 g of water flow from the seawater side to the fresh water side.
This process can be observed in Figure 5.

Pressure

Seawater Fresh water

30 g Salt 1 g Salt
1,905 g Water 64 g Water

935 g Water
Time = [0,∞]
1.6% salt 1.6% salt

Figure 5 Reverse osmosis process

After applying the same pressure as the osmotic pressure, the equilibrium will again be
reached, and the initial state will be created. Since typically the goal is to produce fresh water
from seawater, and not to prevent osmosis to happen by reaching an equilibrium as the
one depicted in Figure 6, a pressure greater than the osmotic pressure will be needed to be
applied to produce more fresh water by consuming seawater.

Pressure

Seawater Fresh water

30 g Salt 1 g Salt
970 g Water 999 g Water

Time = 0
3.1% salt 0.1% salt

Figure 6 Equilibrium step

3.4 EQUATIONS

Reverse osmosis membranes are defined by a simple set of equations. These equations are
used to control the flow of a solvent like water through the membrane, as well as the flow
of solutes like salt through a membrane, and also finally to calculate the osmotic pressure
needed to start producing fresh water from a higher concentration water. Also, the equations
that are used to characterize a reverse osmosis system are presented.

52
 Chapter 3

3.4.1 Fundamental equations

Osmotic pressure is defined by the Greek letter pi (Π). Osmotic pressure can be calculating
by multiplying the temperature (T) in Kelvin degrees, with the ideal gas constant (R) and the
solute concentration (C) and the osmotic pressure coefficient (Φ). This formula is shown in
Equation 2.

3.4.1.1 Osmotic pressure

The osmotic pressure coefficient represents how well a solute dissociates in water. For
NaCl, which fully dissociates in sodium and chloride ions following Equation 1 it will be
equal to 1. For other species that do not dissociate at all in water, its value will be equal
to 0. Concentration is typically expressed in molar mas (mol/L). Ideal gas law is typically
expressed as 0.08314 L·bar/K·mol. Temperature is expressed in kelvin degrees (K).

π = ϕ CRT Eq. 2

One easy way to remember the osmotic pressure equations is thinking how for diluted
solutions, the osmotic pressure resembles the Van’t Hoff equation for ideal gas laws. Van’t
Hoff equation is shown in Equation 3.

PV = nRT Eq. 3

Rearranging terms of Van’t Hoff Equation 3 the same equation than osmotic pressure
Equation 2 can be obtained, as shown in Equation 4. It is important however to highlight
that the osmotic pressure dissociation coefficient needs to be factor in this equation.
n
P= RT = CRT Eq. 4
V

The net driving pressure (NDP) represents how much of an energy driving for it exists across
the membrane. This is obtained by discounting the osmotic pressure (Π) to the pressure that
is applied to make a membrane permeate water (P). Pressure units are typically expressed in
bar or psi. This formula is shown in Equation 5.

NDP = P − π Eq. 5

To better understand how to calculate the osmotic pressure, the following example can
be studied. To calculate the osmotic pressure of a solution that has 2 g/L sodium chloride
dissolved in water, Equation 2 can be used.

The first step is to transform the mass concentration to molar concentration. This is achieved
in Equation 6.

2 g NaCl 1 mol NaCl 0.0342 mol NaCl

= Eq. 6
L 58.44 g NaCl L

53
Experimental Methods for Membrane Applications

Since sodium chloride fully dissociates in water following Equation 1, osmotic pressure
dissociation coefficient can be assumed 1. Therefore it is possible to calculate the
concentration of each individual ion dissolved in water as described in Equation 7.

0.0342 mol NaCl → 0.0342 mol Na+ + 0.0342 mol Cl − Eq. 7

Finally, each contribution of the osmotic pressure needs to be calculated for each individual
ion. This is achieved by using Equation 2 in each individual sodium and chloride ion. This
can be seen in both Equation 8 and Equation 9 respectively, where it can be seen that both
sodium and chloride ions have the same osmotic pressure individual contribution of 0.85
bar each.

0.0342 mol Na+ 0.08314 L bar

PNa+ = 298K = 0.85 bar Eq. 8
L K mol

0.0342 mol Cl − 0.08314 L bar

PCl− = 298K = 0.85 bar Eq. 9
L K mol

Finally, tot total osmotic pressure of sodium chloride can be calculated adding each
individual ion osmotic pressures. This is shown in Equation 10, where it can be seen that the
total osmotic pressure of a 2 g/L sodium chloride solution equals 1.7 bar. A rule of thumb
to quickly estimate the osmotic pressure is to divide the total dissolved solids by 100. This
gives an approximation of the osmotic pressure in psi. To have it in bar, this resulting value
needs to be multiplied by 14.5.

+
PT = PNa + PCl = 1.7 bar Eq. 10

3.4.1.2 Water flux

Water flux across a membrane (Fw) is defined as the multiplication of the water permeability
value, called A-value (A) with the net driving pressure, which is obtaining by subtracting
the osmotic pressure gradient (Π) to the pressure gradient applied to the membrane (P).
Flux of water is typically expressed in US gallons divided per square feet per day (gfd) or
in liters divided per hour per square meter per day (LMH). A-value expresses membrane
water permeability, and it is typically expressed in US gallons divided per square feet per
day per psi (gfd/psi) or in liters divided per hour per square meter per bar (L/m2·h·bar or
simply LMH/bar). Pressure is typically expressed in psi or bar. This formula is shown in
Equation 11.

Fw = A(P − π ) Eq. 11

54
 Chapter 3

It is important to notice how this equation resembles the Darcy’s law equation, which states
that a flow across porous membrane is proportional to the pressure that is applied. Darcy’s
Law can be seen in Equation 12, where k represents the permeability coefficient, μ the
dynamic viscosity, and L the membrane thickness. All these parameters can be incorporated
into the A-value membrane permeability coefficient.

k
Fw = P Eq. 12
µL

This is the same equation that governs the transport ort of water across an ultrafiltration
membrane. Therefore, it can be concluded that for a solvent like water, when it faces a
semi-permeable membrane where it can pass freely through it, it acts as a pressure-driven
convective flow filtration.

3.4.1.3 Salt transport

The flux of salt across a membrane (Fs) is described as the multiplication of the salt
coefficient value, also referred as B-value, with the concentration gradient of solutes across
the membrane. The flux of salt across a membrane is typically expressed in pounds divided
per square feet per day (lbfd) or in grams divided per hour per square meter per day (GMH).
B-value, or salt diffusion coefficient is typically expressed in US gallons divided per square
feet per day (gfd) or in liters divided per hour per square meter (L/m2·h or simply LMH).
Concentration is usually expressed in pounds divided by US gallon (lb/gal) or grams divided
per liter (g/L). This formula is described in Equation 13.

Fs = BC Eq. 13

It is important to notice how this equation resembles the Fick’s law equation. Fick’s law
describes de diffusion transport of mass across a membrane. As salt cannot pass through the
cavities that are created in a membrane, it needs to pass through diffusion. It can be observed
how diffusion is noted as with a D, and this corresponds to the B-value diffusion coefficient.
Concentration gradient stays the same. This formula is shown in Equation 12.

Fs = DC Eq. 14

3.4.1.4 The difference between convective and concentration driven flows

Typically, a pressure driven convective flow is several orders of magnitude higher than the
mass transfer coefficient that can be achieved through diffusion. This is why reverse osmosis
membranes are able to separate so well a solute from a solvent. This mainly happens since
for a semi-permeable membrane, a solvent like water can travel across the membranes just
following a pressure gradient. This means that water sees ‘pores’ across the membrane.
However, salt cannot pass through these ‘pores’, as because of solvation, dissociated species
in water are too big to pass through these ‘pores’, and the only way they have to pass through
a membranes is through diffusion.

55
Experimental Methods for Membrane Applications

The following examples illustrates the different order of magnitude difference between a
convective flow like water, and a diffusion flow like sodium chloride across the membrane.

To calculate the flux of water across a membrane, it can be assumed a water permeability
A-value of 4 LMH/bar, a 15 bar feed pressure, a 1 bar osmotic pressure.

Fw = A(P ) = 4(15 1) = 56 L / m2 h = 56,000 gm2 h Eq. 15

To calculate the flux of salt across this same membrane, a salt diffusion coefficient of 0.2
LMH and a concentration of salts of 2 g/L can be assumed.

Fs = BC = 0.2 2 = 0.4g / m2 h Eq. 16

As it can be observed from this example, reverse osmosis membranes are really selective
to water mass transport across the membrane when compared to salt mass transport.
These serval order of magnitude different in mass transport clearly illustrate the difference
between convective and diffusive flow.

3.4.2 System equations

A reverse osmosis system is mainly composed by a feed flow (Qf), and then this feed flow
gets divided between the filtrated flow that is treated with the membrane active layer, which
is called the permeate flow (Qp), and the concentrate flow (Qc), which has all the rejected
salts or spices that could not permeate the membrane. Concentrate flow is sometimes also
referred as brine or retentate. Flows are usually specified with in cubic meters per hour or day
(m3/h or m3/d), or in US gallons per day (gfd). Plant capacity represents the permeate flow a
desalination plant can produces, and it is usually expressed in millions liters per day (MLD).
1,000 m3/d equals 1 MLD. A simple reverse osmosis diagram can be found in Figure 7.

Qp
Qf
Qc

Figure 7 Reverse osmosis diagram

A reverse osmosis system is characterized by having close water mass balance. This means
that all the water that is entering the reverse osmosis membrane (Qf) needs to exit the
reverse osmosis system through either the permeate (Qf) or through the concentrate flow
(Qc). This formula is depicted in Equation 16.

Q f = Q p + Qc Eq. 17

56
 Chapter 3

A reverse osmosis system is also characterizing by having a neutral salts mass balance. This
means that all salts that are entering the system will be also exiting the reverse osmosis
membrane by either the permeate of the concentrate. Individual salts concentration for the
feed (Cf), permeate (Cp) and concentrate (Cc) are typically represented in g/L or in mg/L
(ppm). This is represented in Equation 18.

Q f C f = Q pC p + QcCc Eq. 18

Flux (F) represents a flow relative (Q) to the membrane active area (A) it is permeating. Flux
is typically measured in cubic meters per hour or day per square meter (m3/h or m3/d)
or in US gallons per day per square feet (US gall/(d·ft2) or gfd). This formula is shown in
Equation 19.

Q
F= Eq. 19
A

Reverse osmosis system recovery (R) represents the process water yield, and is calculated
dividing the permeate flow (Qp) by the feed flow (Qf). It is expressed as a percentage. This
formula is shown in Equation 20.

Qp
R= Eq. 20
Qf

Salt passage (SP) represented the percentage concentration of salt that is passing through
the membrane compared to the initial salt concentration being treated. It is calculated by
dividing the concentration of salt (Cp) in the permeate by the concentration of salt in the
feed (Cf). This parameter is useful from a physics point of view as it lets directly comparing
two different membrane performances. This formula is shown in Equation 21.

Cp
SP = Eq. 21
Cf

Salt rejection (SR) represents the percentage on how much salt a membrane is rejecting. It is
calculated by subtracting 1 minus salt passage (SP). This parameter is useful as it enables to
quickly realize how much solute or salts are being rejected by a membrane system. However,
in order to perform comparative evaluation of two different membranes performance, it is
usually necessary to do any comparative evaluation using the salt passage parameter. This
formula is shown in Equation 22.

SR = 1− SP Eq. 22

57
Experimental Methods for Membrane Applications

Another factor that is calculated is the plant availability (Av). This represents the amount
of time in percentage that the plant is in operation producing water (top) and therefore
not stopped versus the total time of the time period being considered (tT). This formula is
shown in Equation 20.

top
Av = Eq. 23
tT

3.4.3 Factors affecting membrane performance

Several factors can affect membrane performance. The three factors that are usually most
relevant are the effect that feed pressure increase, feed concentration increase, and feed
temperature increase have on membrane performance. In order to understand how these
factors changes, only three equations are required. These are the osmotic pressure equation,
shown in Equation 5, the water transport equation, shown in Equation 11 and the salt
transport equation, shown in Equation 13.

A summary table highlighting these interactions can be found in Table 1.

Table 1 Summary table of the effect of feed pressure, concentration and temperature on the salt
rejection and flux of a reverse osmosis membrane
Flux Salt Rejection

Pressure Fw = A · (P -π)
Fs = B · C

Concentration Fw = A · (P-π )

Fs = B · C

Temperature Fw = A · (P-π )

Fs = B ·C

3.4.3.1 Feed pressure

When feed pressure increases, water flux also increases as a result of an increase in the
pressure. Salt passage across the membrane stays constant, but since more water is passing
across the membrane, whet the water flow is divided by the same amount of salt, the final
salt concentration in the permeate decreases. Therefore, salt rejection increases.

3.4.3.2 Feed concentration

As feed concentration increases, osmotic pressure also increases. This leads to a direct
reduction in the water flux across the membrane, as there is less net driving pressure
available for the membrane to permeate. Additionally, as concentration increases, the flux
of salt directly increases. This leads to a decrease in water passing through the membrane,
which is divided by a higher salt passing through the membrane, thus increasing the salt
passage and decreasing the salt rejection.

58
 Chapter 3

3.4.3.3 Feed temperature

As feed water temperature increases, the water becomes less viscous. This means that
with the same amount of energy, more water can permeate through the membrane. This
eventually leads to improving the water permeability value (A-value) and therefore the
permeate water flux. With regards to the salt rejection, an increase in temperatures improves
much more the salt diffusion factor (B-value). Therefore, there is a higher increase in salt is
passing through the membrane than water passing through the membrane, and as a result,
salt rejection decreases.

3.4.3.4 Concentration polarization

Concentration polarization is the phenomenon in which as membrane removes water from
the feed solution, solutes are pushed towards the boundary layer of the membrane. This
leads to a decrease in performance in the reverse osmosis membrane system as, since the
membrane is filtration, the membrane sees the concentration in the boundary layer and
not in the feed solution. As this salt concentration is higher, this means that the membrane
experiences a decline in flux and salt rejection due to this phenomenon (Sablani et al., 2001).
Concentration polarization can be minimizing by controlling the membrane recovery rate,
as well as through membrane element design elements such as a feed spacer that is able
to provide a proper mixing and therefore minimizes the accumulation of solutes in the
membrane boundary layer.

3.5 REVERSE OSMOSIS MEMBRANES

Reverse osmosis and nanofiltration membranes are pressure-driven membrane filtration

processes, where feed pressure bigger than the osmotic pressure is used to filtrate water
through the membrane.

Commercial elements used in large industrial installations are standardized. They are usually
referred as spiral-wound polyamide based membrane configured in a cylindrical shape with
a typical diameter is 8-in (20 cm), with a typical length of 40-in (1 m). For smaller industrial
application where the water capacity required is lower, elements with 4-in diameter and
40-in elements are also used. Finally, elements used in home drinking applications are less
standardized, and their size in terms of diameter and length can vary depending on each
manufacturer. Examples of these elements are 1.8-in and 2.5-in diameter elements with
12-in or 14-in length. An example of a DuPont FilmTec™ BW30 PRO-400 membrane
can be found in Figure 8. In this example, feed flow will enter the membrane through the
anti-telescoping device on the left. The permeate flow will be collected in each membrane
leave, and finally collected through the inner permeate water tube. The permeate flow will
be exiting the membrane through the permeate water channel located in the center of the
membrane, leaving through the right of the membrane. The concentrate flow will also be
exiting the membrane through the right part through the anti-telescoping device on the
right part. It is useful to realize that the remaining feed water that exits the membrane is
what it is called concentrate.

59
Experimental Methods for Membrane Applications

Figure 8 DuPont FilmTec™ BW30 PRO-400 membrane

A reverse osmosis membrane is typically composed by an active polyamide base layer that is
around 0.2 μm thick. This polyamide membrane is also referred as the active layer, as is the
one responsible from separating the salt solutes from the water solvent. As this membrane
is so thin, in order to be able to precipitate it during the phase inversion process, a support
layer typically consisting of polysulphone is used. This allows the proper precipitation of
polyamide on the polysulphone. The polysulphone layer has a thickness of around 40 μm
thick. As still this thickness is rather seen, in order to enhance its mechanical structure, this
layer is put in a polyesther reinforcing layer, typically consisting of 120 μm thick layer. This
three multi-layer structure is usually referred as thin-film (TFC) composite layer (DuPont,
2023). A schematic of this arrangement can be found in Figure 9.

Polyamide 0.2 µm
Polysulfone 40 µm

Polyester 120 µm

Figure 9 Thin-film composite reverse osmosis membrane multi-layer composition

A scanning electron microscopy (SEM) image courtesy of DuPont FilmTec™ membranes

depicting the main three layers in a reverse osmosis membrane can be found in Figure 10.

60
 Chapter 3

Polyamide 0.2 µm

Polysulfone 40 µm

Polyester 120 µm

Figure 10 SEM image of a reverse osmosis membrane

Nanofiltration membranes are very similar to reverse osmosis membranes. Their main
difference is that the active layer usually consists of a polypiperazine polymer. Typically,
nanofiltration membranes are used when only certain solutes are needed to be separated,
but not all of them. This allows to significant energy savings. Examples of their use are
sulphate removing nanofiltration membranes. These membranes can let sodium chloride
pass through their active layer, but they remove sulphates and other divalent ions. This is
especially useful for oilfield applications, where seawater is used for injecting it into the oil
wells. In this application, no sodium chloride needs to be removed. However, to prevent
multiple problems, sulphate needs to be removed. By using nanofiltration membranes, the
operating pressure can be reduced from 70 bar to 15 bar, therefore saving a lot of energy.

Typically a revers osmosis membrane spiral-wound element consists of multiple polyamide

sheets that are rolled together. Each membrane sheet is separated from the one on its top
by a feed spacer. Inside each membrane there is a permeate spacer. The role of the spacers
is to provide mechanical integrity into the reverse osmosis element so that it can be
properly folded. The feed spacer also plays a crucial role into minimizing the concentration
polarization effect, as well as controlling biofouling and saving energy. Minimizing the dead
spaces inside a membrane is important to prevent biological growth inside of a membrane.
This schematic shown in Figure 11 shows the main parts of a spiral wound reverse osmosis
element.

Permeate spacer Qp
Membrane
Feed spacer
Qf Qc

Qp

Qf Qc

Qp

Figure 11 A spiral wound reverse osmosis elements with its parts

61
Experimental Methods for Membrane Applications

3.5.1 The significance of desalination

As of 2023, there are more than 21,000 desalination plants in the world. All these plants
provide a total install capacity of newly created fresh water equivalent to 100,000,000 m3/d
(Climate ADAPT, 2023).

3.6 PERFORMANCE MONITORING

Typically, systems are designed to provide a constant yearly water production capacity.
Therefore systems are designed at a constant flux rate. So, when systems suffer from fouling
or changes in operating condictiones such as temperature decrease or salinity increase,
typically their flux rate would decrease. To prevent this, more energy is used, so that the
membrane system can compensate for the decrease in membrane permeability and be able
to provide the same operating flux.

Therefore, it is of utmost importance to periodically monitor the performance of a reverse

osmosis systems. There are three key parameters that need to be monitored in a reverse
osmosis.

The first one is the energy consumption, monitored through the high pressure pump
operating pressure. This is an important parameter because the energy consumption directly
impacts the operating expenses (OPEX) of a reverse osmosis pump. Additionally, the high-
pressure pump needs to have enough capacity to increase the pressure it is delivering to
the membrane system. If the pump cannot deliver enough pressure, the whole desalination
installation can start suffering from a decrease in the water it is delivering. This can lead
to water shortages and even being outside the offset contract. This is why typically plants
are designed in a conservative way. To properly size the high pressure pump, typically the
lowest yearly temperature is used to size the pump, as the lowest water temperature will
provide the highest pressure needed to sustain the targeted design installation permeate
flow capacity.

The second parameter that is key to monitor a membrane system is the permeate water
quality. Typically, the water conductivity is monitored. Conductivity is typically expressed
in μS/cm. Conductivity is used to estimate the total dissolved solids (TDS) salinity of water.
A good rule of thumb for low salinities water is that 2 μS/cm are equal to 1 mg/L (ppm)
salinity. For seawater types, a good rule of thumb is that 1.4 μS/cm are equal to 1 mg/L
(ppm). Sometimes, beside general salinity measured by conductivity, specific spices that are
important are also specified. This can involve measuring specific targets such as alkalinity,
boron, and pH, among others. Water quality is of utmost importance, because ultimately,
water treatment plants are usually designed to provide a warrantied water quality that is
typically limited to be below a certain limit. Therefore, water quality typically acts as
the independent variable when designing a membrane system. Typically, a membrane
installation is designed taking into account the highest water temperature thought the yearly

62
 Chapter 3

temperature cycle. This is because at the highest temperature is when the water quality will
be the worst, and therefore show the highest permeate water total suspended solids value.
The third important parameter to monitor is the membrane pressure drop (dP). Pressure
drop is a factor that is important to measure as it directly affects the energy consumption.
However, this parameter is key as when pressure drop increases, this means that the
membrane feed-concentrate channel is getting blocked. This can happen if the membrane is
experiencing fouling, as when the membrane gets fouled, this means it is getting obstructed,
and therefore it is more difficult for water to travel across the membrane. They key problem
this issue presents, is as if the membrane gets too much blocked, the membrane can start
to get mechanically damaged, and it can eventually lead to its irreversible damage and the
membrane can stop working as intended. Therefore, when pressure drop starts to increase
to higher values, it is a good habit to perform a cleaning-in-place (CIP), in order to try to
recover the membrane performance to the initial pressure drop values. It should be noted
that the more a membrane has higher pressure drop, the more difficult it becomes to clean
the membrane and restore its performance to its initial values.

This is why it is so important to clean the membranes early, so that it is easier to recover
the membrane performance. Because membranes systems typically suffer from fouling,
especially in the areas of the planet where temperature is usually higher and there is
therefore a higher water demand due to water scarcity issues, typically plants are designed
with redundant trains. A train is a collection of pressure vessels. Each pressure vessel usually
contains up to 7 membranes in series, and a train usually has dozens of pressure vessels. This
means that for example, in a plant that has 11 reverse osmosis trains, 1 of this trains can be
used to be put in operation when one train needs to be going through a chemical cleaning.
This designs are called in the industry N-1 designs.

3.7 NORMALIZATION

Monitoring feed pressure, permeate conductivity and pressure drop is of utmost importance
in order to be able to anticipate to possible problems that might arise from changes in
operating conditions and in fouling. As it was explained previously, as a result of water
temperature or water salinity changes, the energy consumption and permeate water quality
can change. Therefore, it is of utmost importance to understand if these changes in water
quality or energy consumptions are due to unexpected problems such as fouling, or they are
normal and expected as a result of the physical principles that were previously explained.
This is when membrane normalization comes into approach.

The key normalized parameters that are needed to be analyzed in a reverse osmosis
membrane system are the normalized permeate flow, the normalized salt rejection, and the
normalized pressure drop.

Normalization can be done using the equations listed below, or using computer assisted
programs such as the FT-Norm PRO that DuPont offers.

63
Experimental Methods for Membrane Applications

3.7.1 Why normalization matters

The example detailed in Table 2 is a good example of why normalization matters. This
is an example of a reverse osmosis plant where after 5 days of operation, the feed water
temperature decreases from 25 ºC to 21 ºC. As it can be seen in the monitoring parameters,
the plant is delivering a constant permeate production of 50 m³/h with a constant feed
flow of 100 m³/h. This represents a 50% recovery. Feed salinity is also constant at 1,000
mg/L. However, it can be observed how feed pressure increases from 8 bar to 8.81 bar,
and permeate quality decreases from 30 mg/L to 27 mg/L. Without normalizing the data,
it would be challenging to understand if this increase in feed pressure and improvement
in water quality is a result of expected thermodynamics, or it is a result of the membrane
getting for example fouled. Normalization is the tool that allows to properly perform this
assessment.

In this particular example, it can be seen that normalized permeate flow stays constant,
while normalized salt rejection also stays constant. This means that the increase in feed
pressure and improvement in water quality is not related to fouling, but do to they normal
behavior of the membrane system. This happens as seen previously, when the temperature
decreases, water quality improves as less salt passes through the membrane. Additionally,
more energy is needed to pump the water through the membrane as its viscosity increases.
If the normalized permeate flow would show, for example, a decrease over time, it could be
suspected that fouling is responsible for this loss in membrane permeability. If normalized
salt rejection would be, for example, decreasing, this could also indicate the likelihood of
issues in the membrane system.

Table 2 Normalization example

Normalized
Permeate Feed Feed TDS Permeate permeate Normalized
Feed flow flow pressure (mg/L) TDS Temperature flow salt
Days (m³/h) (m³/h) (bar) (mg/L) (ºC) (m³/h) rejection
0 100 50 8.00 1,000 30 25 50 97.0%
1 100 50 8.00 1,000 30 25 50 97.0%
2 100 50 8.19 1,000 29 24 50 97.0%
3 100 50 8.39 1,000 28 23 50 97.0%
4 100 50 8.59 1,000 27 22 50 97.0%
5 100 50 8.81 1,000 27 21 50 97.0%

3.7.2 Equations
Normalization can be done with operating software’s such as DuPont’s FT-Norm PRO
software, or using the equations found in (DuPont, 2023).

3.7.2.1 Normalized permeate flow

To normalize the permeate flow (Qpn) the following formula found in Equation 24 can be
used. It must be noted that the subscript 0 references to the conditions being normalized to.
This can typically be the first data when the system is stabilized, or even an ideal projection

64
 Chapter 3

that the system is designed to operate at normal or ideal conditions. The other data without
subscript refers to the data at the time instance being normalized to. Units are in bar for the
pressure drop, and in m³/h or m³/day for the flows.

NDP0 TCF0
Q pn = Q p Eq. 24
NDP TCF

Net driving pressure (NDP) is calculate taking the feed pressure, and subtracting half the
pressure drop or differential pressure (dP) to the permeate pressure (Pp) and the average
feed concentrate osmotic pressure (Πfc). All pressures are measured in bar. This is shown in
Equation 25. It should be noted that the osmotic pressure in the permeate has been omitted
for simplification.

dP
NDP = Pf − − Pp − π fc Eq. 25
2

Pressure drop (dP) is calculated using Equation 26. To calculate it, the concentrate pressure
(Pc) needs to be subtracted to the feed pressure (Pf).

dP = Pf − Pc Eq. 26

The average feed-concentrate osmotic pressure (Πfc) can be calculated using Equation 27
(DuPont, 2021). It should be noted that Temperature (T) is in kelvin.

fc
= 0.00265 ×C f c ×T Eq. 27

The average feed-concentrate concentration (Cfc) can be calculated using Equation 28. It uses
the feed concentration (Cf) and the membrane system recovery (R). Its units are in mg/L.

1
ln
Cf c = C f 1 R Eq. 28
R

To calculate the temperature correction factor, Equation 29 can be used. It should be noted
that also Temperature (T) needs to be in kelvin. Also, the k parameter depends on water
temperature. If temperature ≥ 25 ºC, k = 2640. If temperature ≤ 25 ºC k = 3020.

1 1
k 298 T
TCF = e Eq. 29

65
Experimental Methods for Membrane Applications

3.7.2.2 Normalized salt rejection

In order to normalize the salt rejection, the following formula is used. This is displayed in
Equation 30 (DuPont, 2021).

Q p ×TCF0 × C fc0 ×C f
SPn = SP Eq. 30
Q p0 ×TCF × C fc × C f0

3.7.2.3 Normalized pressure drop

Normalized pressure drop (dPn) can be calculated using the formula displayed in Equation
31. To do this calculation, the standard pressure drop (dP) is used, together with the division
of the baseline average feed-concentrate flow (Qfc0) and the average feed-concentrate flow
(Qfc). Units are in bar for the pressure drop, and in m³/h or m³/day for the flows.

C fc0
dPn = dP Eq. 31
C fc

3.8 FOULING

Fouling is the phenomenon in which membranes get dirty. Fouling is one of the major
challenges that is nowadays affecting the reverse osmosis industry (Kucera, 2011). The key
problem that fouling offers is that it is difficult to foresee, and also it is difficult to deal with
it once fouling starts to appear. It is sometimes also difficult to identify the type of fouling
present.

Fouling is typically characterized because of the accumulation of unwanted spices that

get deposited inside the membrane, leading to a decrease in its performance. Typically
fouling leads to higher operating costs in membrane systems, as it can lead to higher
energy consumption, lower water quality, lower water production, higher chemical costs
because of the cleanings, and even an irreversible damage to the membrane. Fouling can
be characterized into four major types, plus a fifth one that relates to membrane integrity
failure.

3.8.1 Biofouling
The main type of fouling is biological fouling or simply biofouling. Biofouling happens
as certain bacteria, that have evolved to form a biofilm when they found a solid surface,
meet the membrane and the feed spacer. When this happens, these bacteria attach on the
membrane, and they start to form a biofilm.

Biofouling is characterized by an exponential increase in the feed-concentrate pressure drop

of the reverse osmosis membrane. It can lead to an increase in the energy consumption and
if not deal properly, it can mechanically damage the reverse osmosis membrane. Biofouling
is typically present in the first elements of a pressure vessel.

66
 Chapter 3

Once a biofilm is established on the membrane, it is difficult to remove. Useful strategies

involve cleaning the membranes with caustic cleanings (CIP). These cleanings are most
effective the higher the temperature and pH. If possible by the membrane characteristics,
effective cleanings will involve pH around 13 at a temperature around 35 ºC using NaOH.
It is important to have enough time for the chemicals to get soaked in the membranes,
and then flush the soaked solution effectively. This combination of steps can be repeated
multiple times until the soaking solution appears clean.

Other more preventive approaches involve starving the bacteria from growing. This can
be achieved using pre-treatment technologies like the DuPont B-Free™ technology, that
is able to eliminate nutrients before they reach the reverse osmosis system (Kucera, 2011).
This system has been proved extremely effective in preventing biofouling development
into reverse osmosis systems.

Other concepts that are practiced involve using non-oxidizing biocides, such as DBNPA or
2,2-dibromo-3-nitrilopropionamide (DBNPA). This biocide is described as non-oxidizing
the polyamide active layer of the reverse osmosis membrane. Some considerations that
need to be taken into account with this solution is its limited compatibility with drinking
water application, and also the fact that bacteria can get used to the biocide, and at certain
point, it can stop being effective and biofouling can develop again.

3.8.2 Organic fouling

The second type of fouling is the organic fouling. Organic fouling happens when organics
molecules get accumulated on the membrane surface. It typically leads to decline in
normalized permeate flow.

The main way to deal with organic fouling is through performing caustic based chemical
cleanings (CIP). These cleanings are most effective the higher the temperature and pH. The
same cleaning procedure as with biofouling can be followed.

3.8.3 Particulate fouling

Particulate fouling, or colloidal fouling, refers to the accumulation of particles in the
membrane. This can be due to an inadequate pretreatment. This type of fouling leads to a
rapid increase in the feed-concentrate pressure drop, as the feed spacer channel fills quickly
with particles.

In order to clean particular fouling effectively, it might be necessary to perform a caustic based
chemical cleanings (CIP). These cleanings are most effective the higher the temperature and
the pH is. If possible by the membrane characteristics, effective cleanings will involve pH
around 13 at a temperature around 35 ºC.

A good solution to deal with this type of fouling can involve upgrading the pretreatment
from a conventional one to a membrane based one, like ultrafiltration. Another solution
could be making sure there are no fiber breakages in the ultrafiltration part, and if there are,
repairing them with glue and pins, or replacing the damage modules with new ones.

67
Experimental Methods for Membrane Applications

3.8.4 Scaling
Scaling is a type of fouling that typically occurs when non too soluble salts starts to
precipitate on the membrane module. This can happen if water recovery is too high, or
if temperature or water composition changes. A recommended way to prevent scaling is
consulting a specialized anitscalant company. They have powerful software that simulate
the operating conditions at the targets recoveries and temperatures, and recommend the
best antiscalants and their concentration to used, based on the spices that have higher risk to
precipitate as they have the risk to surpass its solubility limit. Scaling typically happens in the
last elements of a pressure vessel, as there is where there is less water, and the water is more
concentrated. So dissolved spices have a higher risk to precipitate. Scaling typically leads to
an increase in pressure drop, as well as to a decrease in water quality or salt rejection. As salts
start to precipitate on the membrane surface, this affects the concentration polarization,
and increases the effective concentration of salts on the boundary layer. Scaling can also
therefore lead to a decrease in normalized permeate flow, since the osmotic pressure is
greatly increase, thus reducing the net driving pressure.

In order to clean a scaled membrane, it is important, if possible, to autopsy a membrane, to

understand the type of scaling present. There are multiple companies that are specialized in
offering this service. To clean the membrane, it is recommended to perform an acid chemical
cleaning (CIP). If the membrane characteristics allows it, a pH of 1 at room temperature
is effective. It is recommended to use HCl to prevent any further scaling cause by other
species such as sulfuric acid. Sometimes, membrane can have some organic or biofouling
too. Therefore, it is also recommended to always start with a caustic cleaning as previously
described, and then follow the acid cleaning step.

3.8.5 Integrity failure

Integrity failure occurs when the membrane is suffering from a non-fouling related damage.

This can involve chemical oxidation of the membrane. This can happen if for example,
sodium hypochlorite (NaOCl) from the ultrafiltration pre-treatment manages to reach the
reverse osmosis membrane, as the polyamide can get damage and eventually eliminated.
Symptoms involve an increase in the normalized salt passage, as since the membrane does
not have an active layer, it stops separating salt from water. Once oxidation is detected, it is
important to eliminate the source that is causing the chemical that is leading to oxidation to
leach. A strategy to properly address the leaching of NaOCl in ultrafiltration membranes is
described here (Gilabert-Oriol, 2021).

Other types of mechanical failures might involve o-ring failure. This can happen when the
o-ring that is used to separate a membrane gets pinched, there is a by-pass of water from
the feed side to the permeate side. This leads to an increase in the normalized salt passage.
In order to fix this, a probing test needs to be done in each membrane connection inside a
pressure vessel. A methodology to perform this test is described elsewhere (DuPont, 2021).

68
 Chapter 3

Compaction can happen when a membrane is operating at a too high pressure and
temperature. Compaction can be reversible or irreversible, or a combination of both. Typicall
effects of compaction involve a high increase in energy consumption, observed by a decline
in normalized permeate flow. Compaction can lead to an improvement in water quality, as
the membrane becomes more dense, it is more difficult for the salt to pass through it. When
compaction is identified, it is important to either use compaction resistant reverse osmosis
membranes, or to decrease, if possible, the target permeate flux, specially in periods of high
temperature, with the aim to reduce the operating pressure.

69
Experimental Methods for Membrane Applications

3.9 REFERENCES

Carroll T., S. King, S. R. Gray, B. A. Bolto, N. A. Booker, The fouling of microfiltration membranes by
NOM after coagulation treatment. Water Research 34.11 (2000) 2861-2868
Casas, S., Aladjem, C., Larrotcha, E., Gibert, O., Valderrama, C., & Cortina, J. L. (2014). Valorisation
of Ca and Mg by products from mining and seawater desalination brines for water treatment
applications. Journal of Chemical Technology & Biotechnology, 89(6), 872-883.
Climate ADAPT, consulted on August 19, 2023, https://climate-adapt.eea.europa.eu/en/metadata/
adaptation-options/desalinisation
DuPont, FilmTec™ FT-Norm software, 2021
DuPont, FilmTec™ Reverse Osmosis Membranes Technical Manual, 2023.
Fernández-Torquemada Y., A. Carratalá, J. L. Sánchez Lizaso. Impact of brine on the marine environment
and how it can be reduced. Desalination and Water Treatment 167, 27–37. (2019).
Fritzmann C., J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse osmosis desalination,
Desalination 216.1 (2007) 1-76
Gilabert-Oriol G. Ultrafiltration Membrane Cleaning Processes: Optimization in Seawater Desalination
Plants. Walter de Gruyter GmbH & Co KG; 2021 Jun 8.
Greenlee L.F., D.F. Lawler, B.D. Freeman, B. Marrot, B., P. Moulin, P, Reverse osmosis desalination:
water sources, technology, and today’s challenges, Water research 43.9 (2009) 2317–2348
Guo W., H. H. Ngo,J. Li, A mini-review on membrane fouling, Bioresource technology 122 (2012),
27-34
Kadaj E., R. Bosleman, Energy recovery devices in membrane desalination processes. In Renewable
energy powered desalination handbook (pp. 415-444). Butterworth-Heinemann. (2018)
Kim J., K. Park, D. R. Yang, S. Hong, A comprehensive review of energy consumption of seawater
reverse osmosis desalination plants. Applied Energy, 254, (2019) 113652
Kucera J. Reverse Osmosis: Industrial applications and processes. Wiley; 2011.
Kumar M., T. Culp, Y. Shen, Water desalination: History, advances, and challenges. In Proc., Frontiers of
Engineering: Reports on Leading-Edge Engineering from the 2016 Symp,55-132 (2017)
Massons G., G. Gilabert-Oriol, S. Arenas-Urrera, J. Pordomingo, J.C. González-Bauzá, E. Gasia, M. Slagt.
Industrial scale pilot at Maspalomas I desalination plant demonstrates the efficiency of DuPont™
B-Free™ pretreatment–a new breakthrough solution against biofouling, Desalination and Water
Treatment, 1-5, 2022
Sablani S.S., M.F.A. Goosen, R. Al-Belushi, M. Wilf, (2001). Concentration polarization in ultrafiltration
and reverse osmosis: a critical review. Desalination, 141(3), pp.269-289.
Sustainable clean water through solar-powered desalination for water-scarce islands and coastal regions
(SDG: 2, 3, 6, 8, 11, 12, 14), Section #SDGAction42477, https://sdgs.un.org/partnerships/
sustainable-clean-water-through-solar-powered-desalination-water-scarce-islands-and,
consulted on August 19, 2023
Valavala R., J. Sohn, J. Han, N. Her, Y. Yoon, Pretreatment in Reverse Osmosis Seawater Desalination: A
Short Review, Environmental Engineering Research 16.4 (2011) 205-212
Wang L., J. He, M. Heiranian, , H. Fan,, L. Song, , Y. Li, M. Elimelech. (2023). Water transport in reverse
osmosis membranes is governed by pore flow, not a solution-diffusion mechanism. Science
Advances, 9(15), eadf8488.

70
 doi: 10.2166/9781789062977_0071

Chapter 4

Forward Osmosis
Maria Salud Camilleri-Rumbau, Aquaporin, Denmark/Eurecat, Spain
Xuan Tung Nguyen, Aquaporin, Singapore
Victoria Sanahuja-Embuena, Aquaporin, Denmark
Jan Frauholz, Aquaporin, Denmark/RWTH Aachen, Germany
Victor Augusto Yangali Quintanilla, Grundfos, Denmark
Alberto Tiraferri, Politecnico di Torino, Italy
Irena Petrinic, University of Maribor, Slovenia
Claus Hélix-Nielsen, Technical University of Denmark, Denmark

The learning objectives of this chapter are the following:

• To define principles of forward osmosis

• To define and apply forward osmosis parameters for assessing performance

• To present and discuss the basic equations governing forward osmosis performance
using typical experimental modes

• To understand the theoretical background of forward osmosis performance and

performance prediction using modeling tools.

4.1 INTRODUCTION: PRINCIPLES OF FORWARD OSMOSIS

Forward osmosis (FO) is an osmotically driven membrane technique which allows the
separation of water from a feed solution through a semi-permeable membrane using
osmotic pressure gradient as a driving force. Although during the last decade there have

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

been important advances in FO in terms of material development and processes, the few
commercially available products and best practices for effluent processing makes the
standardization of the FO applicability challenging.

In terms of materials used for fabrication of FO membranes, cellulose acetate (CTA)

membranes and thin film composite (TFC) polyamide (PA) membranes are the most widely
known, both being commercially available (Xiao et al., 2017). New approaches using new
materials for FO purposes have also been developed. To name a few, these are double-
skinned membranes, membranes obtained by layer-by-layer techniques, mixed organic-
inorganic membranes and aquaporin-based membranes, which consist of TFC PA with
embedded aquaporins (Suwaileh et al., 2020).

Despite FO being a developing technology, in the recent years, there has been an increasing
interest for its use in industry. This opens countless opportunities for further developing
membrane configurations that can be used in an industrial setting. In terms of applications,
FO is a versatile membrane technique with a broad applicability spectrum within the water
treatment sector (desalination, municipal wastewater, industrial wastewater, potable
and non-potable water reuse, etc.), the management of process water (biorefineries,
pharmaceutical processes, etc.) and food and beverage processes (concentration of flavours
and aromas, juices, etc.).

Regarding membrane performance, FO relies on the osmotic pressure of the two solutions
separated by the semi-permeable membrane. The pass of water is allowed due to osmotic
gradient resulting in a concentrating process for the lower osmotic pressure solution
(known as feed) and a dilution process for the higher osmotic pressure solution (known
as draw solution). As for any other membrane-based process, water flux (Jw) and forward
solute rejection (R) can be obtained from experimental data (see section 4.3.1); while a
parameter known as reverse solute flux (Js), unique for FO, allows to track the loss of draw
solute into the feed solution.

During FO processing, concentration polarization can severely affect membrane performance

in both feed and draw sides of the membrane. Both external concentration polarization
(ECP) at the feed side of the membrane as well as internal concentration polarization (ICP)
at the draw side of the membrane, play a detrimental role that cannot be overlooked when
interpreting the FO process performance and when evaluating its applicability (see section
4.3.2. for more details). Concentration polarization can act in combination with fouling,
scaling and/or a combination of both fouling mechanisms, to decrease the net driving force
available for mass transport.

With this Chapter, the authors have put together the most relevant experimental practices
in the FO field in order to guide researchers and engineers towards getting hands-on
experiences in FO.

72
 Chapter 4

4.2 MATERIALS AND EXPERIMENTAL SET-UP

4.2.1 Membrane configurations

Membranes for FO processes and generally for liquid-liquid separations, can be found
commercially either as flat-sheets or hollow tubes. The modules of these can be classified,
similarly as done for other membrane processes, in four module types: plate-and-
frame, spiral-wound, hollow fiber and tubular. The first two types are made of flat-sheet
membranes, while the last two types are composed of hollow tube membranes.

Plate-and-frame modules are the simplest module configuration where membrane sheets
are mounted in frames closely together. The feed solution to be treated passes alongside the
sheets surface and it gets collected at the end as a concentrated retentate, while the permeate
is collected in its own channel. Spiral-wound are membrane sheets rolled in alternating
order together with turbulence promoting plastic grids, called spacers. The feed solution
is introduced at one end of the module and flows axially on the active layer and feed spacer
side of the membrane. The permeate is collected in the envelope and led to the permeate
channel which is centrally located. Hollow fiber modules consist oppositely of tubes, packed
closely together and placed inside a vessel. Here, the feed solution passes through the lumen
of the hollow fiber, permeates through the membrane towards the shell side and exits the
module. When their inner diameter rises to 5 mm or above, and the module packing density
significantly decreases, the module is known as a tubular module.

There are intrinsic advantages and disadvantages of using each module configuration and
their usage would depend on the intended process application. For example, plate-and-
frame and tubular configurations are usually used with extremely high-particulate streams
and/or high-viscosity solutions. The simplicity of the plate-and-frame configuration
allows for high cross-flow velocities, reducing fouling by increasing longitudinal shear
stress. Additionally, these membranes can be easily cleaned which increases the lifetime of
the membrane. However, this module type is expensive to manufacture, and the packing
density is low, which increases the membrane installation footprint considerably. Similarly,
tubular modules can be more tolerant to fouling and clogging due to the large inner diameter
of their hollow tube and the possibility of operating at high cross-flow velocities. However,
it suffers from the same disadvantages as its counterpart, the plate-and-frame module due
to a low packing density. Spiral-wound and hollow fiber configurations are thus the most
commonly used module configurations for membrane-based liquid-liquid separation and
FO processes, as they can cover a broad range of applications by balancing effectiveness and
price.

4.2.2 Experimental modes

In general, there are three main experimental modes, regardless of process configuration
being single-stage or multi-stage, that can be defined when working with FO membrane
processes:
- Single pass mode
- Batch-batch mode
- Semi-batch mode

73
Experimental Methods for Membrane Applications

The main differences between these operational modes are related to how the feed and
draw solutions are being processed within the membrane module. For example, in single
pass mode, the feed and draw solutions enter the module and have a unidirectional contact
area across the membrane, where the solutions are not recirculated. The feed recovery
or concentration is achieved in one pass. Oppositely, in batch mode, both feed and draw
solutions are recirculated to their respective holding tanks. This means that the feed
solution gets increasingly concentrated in the feed tank, while the draw solution gets
diluted with time. During semi-batch mode operations, one of the two solutions, either
feed or draw solutions, are run in a batch mode while the other solution runs in single pass
mode. Typically, the feed will run in a batch mode, allowing for continuous concentration
during processing, while the draw solution will run in single pass to minimize loss in FO
performance due to dilution of the draw solution.

The advantages of the operational modes above depend on the type of feed, on the process
that needs to be undertaken, and on the objective of the FO system. For instance, during
food valuable concentration processes, a semi-batch mode will be preferred, while a batch-
batch mode might be preferred during concentration of secondary effluent by using sea
water brine as a draw solution.

Test setup – Semi-batch mode (feed in batch mode vs. draw in single pass mode)
Figure 1 shows the schematic outline of a semi-batch FO setup. The draw solution will
become diluted during the feed concentration process. The draw outlet can be discarded to
the drain. The feed solution will be continuously recirculated to the FO membrane module
and thus concentrated.

T T
Draw inlet
Sample C C tank D1
point Sample
F point
F
P Transmembrane
Draw outlet Pressure can
pressure: P be adjusted by
tank D2 minimum 0.2 bar
opening/closing Clean water
T C F P P
needle valve on tank C
retentate line
FO module

Sample
retentate line
needle valve on
opening/closing C F
be adjusted by
Pressure can
For auto
point logging
T
Sample
Feed point
Feed inlet
Inlet
tank
TankF1F For recirculation
For auto
retentate line
needle valve on
opening/closing

Weighing logging
be adjusted by

Clean water
Pressure can

tank C balance

Figure 1 Setup for FO semi-batch operation

74
 Chapter 4

Recommendations for running an FO application test:

1. To select the type and strength of the draw solution, refer to section 4.2.3.
2. Start the feed pump to fill in the system. Afterwards, start the draw pump and adjust to
the operating conditions as indicated by the FO membrane/FO system manufacturers,
always ensuring that the system operating conditions are in agreement with the
membrane manufacturer operating limits. Ensure removal of air in the system. The
operating conditions can be modified during the application test. For example, the feed
inflow can be increased to enhance shear on the membrane surface and delay fouling,
while the draw inflow can be increased if flux (Jw) falls below 1 L/m2h.
3. TMP must always be kept positive where possible. For this, it is crucial to monitor that
the feed inlet and outlet pressure readings do not overcome the maximum TMP and
inlet pressure specified by the membrane module manufacturer. See more details on the
effects of negative TMP in section 4.3.3.
4. To be able to calculate compound mass balances, weight of samples taken from the feed
inlet, feed outlet and draw outlet should be considered throughout the concentration
process. This will allow monitoring of Jw and water recovery values accurately during the
FO process.
For further recommendations on process parameters and constraints, refer to Table 1.

Table 1 Process parameters to be considered during FO application test

Process parameter Process values - considerations
Recommended (and maximum) application flow Refer to FO membrane/system manufacturer’s
rates, L/h recommendations.
Minimum feed outlet flow, L/h Refer to FO membrane/system manufacturer’s
recommendations. The user needs to make sure that the
feed outlet has a minimum flow to avoid module damage.
TMP, bar Refer to FO membrane/system manufacturer’s
recommendations. Generally, it should be around or just
above 0 bar.
Feed and draw inlet pressure, bar Refer to manufacturer’s recommendations on pressure
tolerance for the specific FO product.
Recommended (and maximum) temperature, ˚C Refer to manufacturer’s recommendations. Generally, the
FO module should be able to run feed and draw solutions
at room temperature. Be aware that increases in operating
temperature could affect FO performance, e.g., increased
Jw and Js. The module temperature tolerance should not be
surpassed. Refer also to manufacturer’s recommendations
for CIP procedures.

Data collection during the application test

The data to be collected during the test is shown in Table 2. For detailed calculations, refer to
equations on ‘Typical parameters and phenomena’ in section 4.3.1.

75
Experimental Methods for Membrane Applications

Table 2 Process parameters to be measured during the semi-batch application test and where to
make the related readings
Purpose/calculated values Process parameters Where to measure
Water flux (Jw) Flow/Weight Feed bulk
TMP Pressure Feed inlet/outlet, draw inlet/
outlet
Feed inlet pressure Pressure Feed inlet
Maintain stable temperature (T) Temperature Feed outlet (if measuring
conductivity at feed outlet)
Osmotic pressure (indirect Conductivity Feed outlet
measurement)
Ensure minimum feed outlet flow Flow Feed outlet

4.2.3. Draw solutions: properties, regeneration, types and selection criteria

The performance of FO applications depends on the draw solution, which provides the
driving force for water permeation. An adequate choice of draw solute agent can maximize
the water flux (Jw) and the water recovery of the system. In addition, the reverse solute flux
(Js) can be reduced and the regeneration costs can be lowered, which usually represents
the largest operational costs in FO applications. Therefore, this subsection will give a
short summary of the most important properties of draw agents, their influence on the
membrane performance, available draw regeneration methods, and discuss advantages and
disadvantages of different classes/types of draw agents. Eventually, guidelines regarding the
selection of suitable draw solutes are given.

Main properties of a draw solution

Osmotic pressure
The driving force in forward osmosis processes is provided by the difference in osmotic
pressure across the active layer of the membrane, as defined as follows:

RT RT
Π=− ⋅ ln(aw ) = − ⋅ ln(γ w χ w ) Eq. 1
Vm Vm

Where R is the gas constant, T the temperature, Vm the partial molar volume of water, aw the
water activity, γw the activity coefficient, and cw the mole fraction.

In practice, assumptions are made to simplify osmotic pressures estimation, often also due
to unknown activity coefficients. In very dilute solutions the solvent activity coefficient can
be assumed to be close to 1, resulting in the validity of the van’t Hoff equation, as follows:

= cm RT Eq. 2

where cm is the osmotic concentration (osmolarity). The osmolarity is the molar

concentration of osmotic active solutes. For a salt solution, such as NaCl which dissociates
into two ions, the osmolarity equals twice the molarity assuming complete dissociation.

76
 Chapter 4

Even though the van’t Hoff’s equation is formally only valid for very dilute solutions,
it often provides acceptable accuracy for salt-based draw solutions, especially if direct
measurements of the osmolarity (e.g., freezing-point or vapor pressure osmometry) are
available. In comparison, prediction of osmotic pressures of organics, polymers, or other
draw solutes can lead to significant deviations.

Diffusivity and viscosity

In most technical applications, the draw solution is applied on the support layer side of
the membrane leading inevitably to driving force losses due to concentration polarization
effects (see section 4.3.2). Besides the properties of the membrane (structural parameter),
the extent of ICP is mainly related to the draw solution concentration and diffusivity. The
use of a higher draw concentration, hence nominal driving force, does not correlate with a
linear increase in the water flux. A higher draw concentration leads to larger relative driving
force losses due to stronger ICP related to the convective transport of draw solutes away
from the active layer. Since the transport of draw solutes towards the active layer is diffusive
only, the diffusivity of the draw agent significantly influences the extent of polarization
effects. Diffusivity depends mainly on the solution’s viscosity and the diffusion coefficient.
Overall, draw solutions of low viscosity and high diffusion coefficient are the best choice
when considering FO system productivity.

Regeneration of draw solutions

Continuous FO processes require a reconcentration of the diluted draw solution. Up to
now, the regeneration process remains the bottleneck in the draw solution selection.
Among the most studied draw regeneration methods are membrane-based processes,
such as reverse osmosis (RO), nanofiltration (NF), and membrane distillation, as well as
evaporative technologies and electrodialysis. While pressure-driven membrane processes
are limited in achievable osmotic pressures due to allowable pressure, the evaporation of
water is highly energy intensive. This has led to ideas for the implementation of FO within
applications that do not require a draw regeneration. Important to mention are (so-called
direct) FO processes using seawater or a concentrated fertilizer as draw solution, which may
be discarded or beneficially used once diluted, to avoid costly regeneration. Furthermore,
novel types of draw solutions have been designed to overcome existing challenges in draw
regeneration. These so-called responsive draw solutions exploit drastic changes in physical
and chemical properties of the draw agent provided by external stimuli, such as heat or pH,
enabling a practical and efficient draw regeneration. Please note that the energy required
for draw agent regeneration is always somewhat related to its target osmotic pressure
due to thermodynamics considerations, thus simplicity of regeneration should not be
confused with cost of regeneration and the two issues should be considered separately and
simultaneously to design a feasible and effective FO system.

Types of draw solutes

In theory, any water-soluble component exhibiting an osmotic pressure can be used as a
draw solute. Considering the above-described influence of draw properties on the process
performance, small solutes of high osmotic pressures (high solubilities) are preferred. A
variety of different draw solutes including salts, small organic molecules (e.g., sugars),
volatile organic compounds, nanoparticles, polymers, or hydrogels have been investigated
up to now.

77
Experimental Methods for Membrane Applications

Among the most studied and applied draw solutes are inorganic and organic salts offering
the advantages of high osmotic pressures, low viscosities, high diffusivity, electrical charge,
and low toxicity. As a result, salt-based draw solution can reach a high water flux, exhibit
comparable low driving force losses, enable a hazard-free operation and simple regeneration
by pressure-driven membrane processes such as reverse osmosis. Most salts are inexpensive,
available as food grade quality, and their replenishment costs are low.

Due to the larger variety of salts, salt-based draw solutions can be selected with regard to the
specific process requirements. For example, multivalent ions of higher molecular weights
can be selected for food and beverage application to minimize the reverse solute flux into
the product stream. Furthermore, studies indicate that the rejection of feed compounds can
be enhanced by selecting an appropriate draw solute.

Worth mentioning is sodium chloride (NaCl), one of the most studied draw solutes. It is
often used as a benchmark draw agent to evaluate membrane and process performance.
Many membrane manufacturers and published articles use the specific reverse solute flux
(Js/Jw) of NaCl as key characteristic to account for the membrane’s selectivity for water
transport.

Besides the comparable high reverse solute flux of salts, their regeneration remains
the bottleneck of salt-based draw solutions due to osmotic pressure limitation. Novel
approaches such as osmotically-assisted reverse osmosis (OARO) may enable higher draw
concentrations in the future (Peters and Hankins, 2020).

To exceed the osmotic pressure limitation of conventional draw solutes, a variety of

responsive draw solutes which can switch solubility properties by external stimuli were
studied. Among the most studied responsive draw solutions are thermo- and CO2-
responsive draw agents. Thermo-responsive draw solutions exploit temperature-dependent
miscibility gaps between water and polymers or ionic liquids. By exceeding the lower
critical solution temperature, the diluted draw solution separates into two phases whereof
one phase is rich in draw agent and the other is water-rich. CO2-responsive draw solutions
undergo acid base reaction and are often amine-based. Amines can react reversibly with
CO2 and form bicarbonate salts which can be used as draw agents. Upon heating or purging
with inert gas, the diluted amine bicarbonate draw solute decomposes into either a water-
insoluble liquid or gaseous amines such as trimethylamine or ammonia. Disadvantages
of responsive draw solutions are related to the costs, toxicity (amines), or low membrane
performance due to severe ICP (polymers).

Selection of draw solutes

The right choice of draw agent depends on the specific application and feed stream, the
target recovery, the process configuration, availability and costs of different energy forms
(electrical, waste heat, ...), space requirements, commercial assessment, as well as further
considerations. The following guidance can assist in selecting an adequate draw agent:

78
 Chapter 5

1. Osmotic pressure of the draw solution

- Too low osmotic pressure of the draw solution induces low membrane performance,
prolonging the process time or requiring larger membrane areas. Both may contribute
to target compound losses from the feed solution as well as draw solute contamination
of the feed solution.
- Too high osmotic pressures can accelerate membrane fouling and scaling. Regeneration
costs can increase due to dissipation of osmotic potential.
- Rule of thumb: ratio in osmotic pressure between the draw solution and concentrate
should not be below 1.1 - 1.2
2. Draw regeneration method:
- The draw regeneration process is often limiting the choice of draw type and strength
- Reverse osmosis:
• RO is the most energy efficient process to regenerate draw solutions
• Limited by hydraulic pressure (65 bar, high-pressure RO: 120 bar)
• Osmotically-assisted reverse osmosis is not established but can overcome osmotic
pressure limitations
- Evaporators
• Energy intense regeneration with no limitations regarding osmotic pressures
• Corrosive draw solutes (e.g., chlorides) can drastically increase the CAPEX due to
material requirements and should be avoided
- Electrodialysis
• High CAPEX
• Limited to ionic draw solutes
• Energy efficient at low osmotic pressure range
- Membrane distillation
• Energy intensive regeneration, but often low grade heat sources can be used
• Not-yet-established technology due to currently low performance and specific
current limitations related to module configurations, fouling/scaling, and long-
term stability
- Responsive draw recovery
• Still under development, offering the potential to concentrate draw solution to
high osmotic pressures
• Energy intensive, but often enabling the utilization of low-grade heat sources
3. Applications:
- Food and beverage
• Only food-approved draw solutes are applicable (sugars, salts)
• Multivalent ions and agent of higher molecular weight can reduce unwanted
reverse solute flux
• Target compound rejection can be increased by selecting draw solutes which are
already present in the feed stream
- Wastewater concentration
• Lower concentration of draw agents may be beneficial to reduce fouling propensity
• Target compound rejection can be increased by selecting draw solutes which are
already present in the feed stream

79
Experimental Methods for Membrane Applications

4.3 EXPERIMENTAL METHODS

4.3.1. Typical parameters and phenomena

The most important process parameters are the water flux Jw, the reverse solute flux Js, the
specific reverse solute flux Js/Jw, the recovery, and the rejection (both forward and reverse).

The water flux Jw is defined as the areal permeation rate of water as follows:
Q permeate Q feed Qconcentrate Qdraw out Qdraw in
Jw = = = Eq. 3
Amembrane Amembrane Amembrane

Where Q is the flow rate and the active membrane area (A) is in the denominator. As seen
above, Jw can be calculated based on the difference in feed in- and outlet flow rates as well as

based on the difference on the draw side. Its unit is L/m2h. In batch operation, Jw can be
determined by measuring the change in feed or draw weight under the assumption that only
water permeates the membrane.

The reverse solute flux Js is defined as follows:

msolute
Js = Eq. 4
Amembrane

Where the mass flux of draw solute is in the numerator and the active membrane area
(A) is in the denominator. The reverse solute flux is determined by measuring the draw
solute concentration in the concentrate stream. Depending on the feed composition, an
appropriate measurement of draw solute concentration, such as conductivity, ICP-OES, or
HPLC, can be used. The reverse solute flux is usually given in g/m2h.

The specific reverse solute flux Js/Jw is defined as the ratio between Js and Jw. It is a measure
of the selectivity for water permeation over draw solute transport given in g/L.

The transmembrane pressure (TMP) is defined as the average hydraulic pressure between
the feed side and the the draw side of the membrane, given as follows:
( p feed + p feed out ) ( pdraw in + pdraw out )
TMP = Eq. 5
2

The recovery (Rec) defines the ratio of the volume of recovered water to the volume of feed
solution. In single-pass operation the membrane recovery is defined by using the permeate
and feed flow rates as follows:
Q permeate Qconcentrate Q Qdraw in
Recmembrane = ×100% = 1 100% = draw out 100% Eq. 6
Q feed Q feed Q feed

80
 Chapter 4

In batch processes where the feed solution is constantly concentrated, the recovery is
defined as follows:

Vfeed (t)
Rec(t) = 1 ×100% Eq. 7
Vfeed (t0 )

Assuming only water to permeate the membrane and a constant density of the feed solution
rfeed(t) allows calculating the recovery by weights instead of volumes.

In food and beverage processes, concentration factors (CF) are often used instead of recovery,
where CF is:

1
CF = Eq. 8
R
1−
100%
The average membrane forward rejection R of a compound i (moving forward from the feed
to the draw side) is commonly defined using the concentration ratio between permeate and
feed. To take the concentration difference between incoming feed and outgoing concentrate
stream into consideration, the average concentration on the feed side of the membrane is
often used:

ci, permeate
Ri = 1 ×100%
Eq. 9
ci,concentrate + ci, feed
2

In contrast to most other membrane applications, the permeate concentration cannot be

directly measured due to its dilution by the draw. Therefore, its average must be calculated
based on a mass balance of component i on the draw side.

Taking the draw flow rate into consideration leads to:

ci, permeate × Q permeate = ci,draw out × Qdraw out ci,draw in × Qdraw in

Inserting equation 1 (Jw) and rearranging leads to:

ci,draw out × Qdraw out ci,draw in × Qdraw in
ci, permeate = Eq. 10
J w × Amembrane

While in (draw) batch operation the ingoing target solute concentration in the draw solution
needs to be considered, in single-pass operation Ci,draw in is in most cases negligible.

The achieved membrane forward rejection depends on the membrane type, operation
conditions (e.g., flow rates of draw and feed solutions), the water flux, as well as the recovery,
and it is different for different compounds.

81
Experimental Methods for Membrane Applications

It is important to note that the above membrane rejection calculations consider the observed
rejection and not the real compound rejection, as it is calculated considering:
1) The total mass that has passed through the membrane during the entire pass and the
average Jw (or entire time in batch mode) and not the mass that is passing across the
membrane in each location along the module (or at any given time in batch mode);
2) The feed or draw bulk concentration and therefore not considering the higher compound
concentrations reached at the active layer interface due to the polarization phenomena.
Larger molecules are often better rejected than smaller ones. In addition, uncharged
organic molecules show lower rejection than charged molecules due to missing
electrostatic repulsion. Even during batch concentration processes the rejection may
change significantly as shown in the case of urea (Figure 2).

Feed: Urea 200mgL-1

100

80
HF-O

60

HF-C
40

20

Rejection (% ) 0
0 20 40 60 80 100
Recovery (%)

Figure 2 Urea rejection variation with recovery rate for HF–C (chlorinated membranes) and HF–O
(non-modified membranes). Adapted from Sanahuja-Embuena et al. (2019).

4.3.2 FO process design constraints and considerations

To design a specific FO process and experimental setup, users are strongly advised to refer
to the manufacturer’s FO module datasheet to understand the operating limits of the given
modules. It is also recommended that the user reads any other documentation provided by
the FO manufacturer.

Concentration polarization (ECP/ICP)

As seen in Figure 3, concentration polarization occurs on both sides of the membrane due to
the permeation of water concentrating the feed solution while diluting the draw solution.
A distinction is made between external concentration polarization (ECP) on the active layer
side of the membrane and polarization effects in the membrane’s support layer referred to
as internal concentration polarization (ICP). Depending on the membrane orientation, i.e.
FO mode (where the feed solution is in contact with the active layer) or PRO mode (where
feed solution is in contact with the membrane support layer), these polarization effects can
either be dilutive or concentrative. In most applications, the draw solution is applied on the
support layer side leading to dilutive ICP and concentrative ECP.

82
 Chapter 4

Concentration polarization reduces the difference in osmotic pressure across the active
layer and leads inevitably to driving force losses for water permeation. Besides driving force
losses, concentrative ECP increases the risk of membrane fouling and scaling. Lower water
fluxes as well as turbulent flow conditions can contribute to reducing these risks.

The intensity of dilutive ICP depends on the porosity, tortuosity, and thickness of the
support layer (see structural parameter in section 4.4.1) as well as on the diffusivity of the
draw solutes and the water flux. Since draw solutes diffuse against the convective water flux,
draw solutions of low viscosity and high diffusion coefficients can mitigate dilutive ICP (see
draw solution in section 4.2.3). Additionally, highly porous, and thin support layers can
lower the extent of driving force losses.

AL-DS membrane orientation can significantly decrease dilutive CP of the draw solution.
Since concentrative CP of feed solutes in the support layer is increased, this membrane
configuration might only be beneficial in specific applications, where the feed presents low
fouling potential).

a) Active layer b) Support layer

Support layer Active layer

Feed solution C5 Feed solution C4

C4 C3

Jw Jw
∆πm ∆πb ∆πm ∆πe ∆πb
C3
∆πc C2
Draw solution C1
C2 Draw solution
Js
C1
Js

Figure 3 ECP and ICP at a) AL-FS mode, and b) AL-DS mode. Adapted from Wang and Liu (2021).

Pressure limit
Pressure limit is one of many important factors to consider as it affects the choice of flow
rate and cross flow velocity sent into each element. This typically already translates into the
recommended flow rate range on both feed and draw side. In addition, how much water is
transported into the draw solution side is primarily a function of draw solution flow rate
and concentration. Even at low draw solution inlet flow, high osmotic pressure difference
may result in a large water permeation rate and hence a higher flow rate on the draw side.

System projections are therefore useful to predict the behavior of pressure drop on both
feed and draw lines. However, calculating pressure drops in a system can be complicated as
this will depend on several factors such as module geometry, array configuration and liquid
properties among other factors. Few considerations need thus to be taken when projecting:
1) permissible pressures given by membrane manufacturer should not be exceeded, 2) TMP

83
Experimental Methods for Membrane Applications

usually increases during batch concentration (e.g., viscosity increase of feed, feed outflow
rate increases due to a lower Jw), or in continuous mode due to fouling, 3) system arrays
require special considerations such as accounting for local changes in pressure drops,
pressure build-up when more-than-one modules are connected in series, draw solution fed
in the system in series or in parallel, etc.

Flow rate limit

Flow rate limit, by extension, is determined by the maximum pressure limit of the module.
Flow rate should be selected within manufacturer’s recommendation in order not to exceed
pressure limit. In addition, users are advised to check for any minimum feed reject flow
requirement by manufacturers. In a batch process or semi-batch, feed and bleed process,
recovery of feed is time-dependent and not flow-dependent. It is therefore possible to
maintain as high cross-flow velocity as possible, while staying within pressure limit,
to minimize risks of fouling and scaling. This is especially so when feed streams contain
medium to high degree of foulants. In a single-pass continuous process, recovery of the
feed is flow-dependent. Designers of the FO process should determine, through projection,
whether concentrate flow rate at the last in-series element is below recommendation.

Moreover, draw flow rate in operation should be carefully selected and monitored because
it influences transmembrane pressure, permeation flux across FO membrane, and the
concentration of draw agent and of possible compounds permeated from the feed side.
Usually, for polyamide-based FO membrane, manufacturers may recommend a safe limit of
negative TMP, beyond which there poses a risk of delamination of polyamide active layer.
Having a high draw flow rate increases the overall permeation. However, an excessively high
draw flow rate might raise pressure on the draw solution side and result in a high chance that
the negative TMP limit is exceeded.

Flow direction
Flow direction, whether counter-current or co-current, is also a tool available for FO process
designers. In co-current operation, feed and draw solutions enter the module through the
same end of the module, leading to a constantly reducing driving force along the module
length. Counter-current operation enables to maintain a more constant osmotic pressure
difference along the lengths of the module (see Figure 4). Additionally, counter-current
operation maximizes the average water flux across the FO module or system and the
permeate recovery, while minimizing local differences in water flux. This means that the
difference between water flux across FO membrane across inlet and outlet of FO system is
less for counter-current, as compared to co-current flow direction.

It should however be noted that the selection of the flow mode (i.e., co-current or counter.
current) depends on module type. For spiral wound or some plate-and-frame module type,
flow path is designed to be in cross flow, where feed and draw solutions are perpendicular
to each other. For hollow fiber and tubular membrane type, flow path can be selected to be
counter-current or co-current.

Conventionally, a counter-current flow path is the is the preferred option in most

applications and experimental setups as it allows maximization of the driving force.

84
 Chapter 4

In practice, FO process designers should pay attention to ease of filling up the shell side
chamber in counter current mode, assuming that the module is mounted vertically (i.e., feed
side flow is upwards and draw side flow is downwards). Modules of larger size which are
mounted vertically may however require ingoing streams to enter on the bottom side of
the module to remove any trapped air from the module. In such a case, if the draw inlet
flow rate is too small, partial filling of shell chamber may occur resulting in underutilized
membrane area. In this specific case, operating in co-current operation may be advantageous
even though process performance is reduced.

a)

Drawin Drawout Drawout Drawin

Feed Concentrate Feed Concentrate

Co-current Counter-current

Draw Draw
∆π ∆π
Feed
Feed

π π
Module length Module length

b) Feed: DI H2O
40 1.5

1.2
30
Jw-HF-C Jw-HF-C
0.9
JwHF-O JwHF-O
20
0.6
Js/Jw HF-C
10
0.3
Js/Jw HF-O

Jw (Lm-2h-1) 0 0.0 Js/Jw (gL-1)

Counter-current mode Co-current mode

Figure 4 a) Driving force for counter-current and co-current flow direction; b) Jw and Js/Jw for
HF–C (chlorinated membranes) and HF–O (non-modified membranes) in co-current and
counter-current when DI water was used as FS. Operating conditions were: Feed flow
rate was 100 L.h-1, draw flow rate was 25 L.h-1, draw concentration was 1 M NaCl and
TMP was 0.2 bar. (n = 2). Adapted from: Sanahuja-Embuena et al. (2019).

Limiting flux
Lastly, water permeation limit or design flux limit is a major factor affecting the FO design.
On the one hand, this is related to feasible feed inlet flow rate for FO module and system. A
lower FO feed inlet flow rate limit by manufacturers’ recommendation or by system design

85
Experimental Methods for Membrane Applications

indicates lower maximum design flux limit. However, this is in fact generally related to
fouling potential and reversibility of fouling.

While there is no consensus on what design flux limit for FO membrane should be, there
are research reports indicating limiting flux to be between 10-20 LMH and designed flux
for reversible fouling to be 5-15 LMH. Here, limiting flux is defined to be the starting flux
value at which there is decline of flux over time at constant osmotic driving force difference.
Design flux is defined to be the starting flux at which there is minimal flux decline and there
is flux restoration upon cleaning if there is any flux decline over time. These flux values vary
depending on membrane type and feed quality or foulants present in feed.

Water permeation limits may be controlled by draw inlet flow rate and draw inlet
concentration. As mentioned above, a higher draw inlet flow rate means less dilution effect
on the draw side, allowing osmotic driving force to be sustained from inlet to outlet of
module. This comes with the drawback of having a higher pressure drop on the draw side.

A higher draw inlet concentration means higher osmotic driving force across entire FO
modules or system, at the same draw inlet flow rate. The disadvantage is the high likelihood
of exceeding design or limit FO flux at certain sections of FO membrane within a module
or system. This may lead to sustained high ECP in those regions, increased likelihood of
fouling and scaling and premature module failure.

4.3.3 Best practices

Transmembrane Pressure (TMP)

Most forward osmosis membrane suppliers recommend running FO processes under low
positive transmembrane pressure. The positive TMP can hinder the transport of draw
solutes towards the feed solution due to the pressure gradient, which helps in preventing
the immediate contamination of feed solution by draw solutes in the event of membrane
breakage or defects on the selective layer. In the case of small defects on the polyamide layer,
a positive TMP will also be beneficial. However, a positive TMP may also aid the transport
of feed solutes into the draw solution and thus, the quality of the selective layer would need
to be checked. A tight and highly cross-linked polyamide layer should not be significantly
or drastically affected by slight positive and negative TMPs, and if this happens, it may be a
sign of membrane deterioration.

Nevertheless, a negative TMP should be strictly avoided, even for brief periods of time, due
to the polyamide layer configuration (where the layer is on the lumen side of the membrane).
When a negative TMP is applied, the pressure gradient direction can cause the delamination
of the polyamide layer, and consequently, the breakage of the membrane.

During the FO module operation, pressure losses from inlet to outlet for both feed and draw
side are expected, regardless of the flow mode selected (i.e., counter-current or co-current),
which could provoke negative TMP at the feed outlet or draw inlet locations. It is therefore
of paramount importance to maintain a positive TMP at the feed outlet ensuring following
the manufacturing guidelines on pressure limits.

86
 Chapter 4

In summary, since FO is a virtually pressure-less membrane process, membranes are not

designed for high hydraulic pressures on either side of the membrane. Therefore, commonly
recommended TMPs are around 0.2 bar. Allowable pressures given by the membrane
manufacturers should not be exceeded to ensure safe operation. Here, pressure relief valves
in the experimental setup can protect the membrane from maloperation.

Avoiding ‘over-recovery’
High recoveries of feed solution can lead to the precipitation and deposition of feed particles
on the membrane (fouling and scaling). While membrane fouling is characterized by the
deposition of (mainly organic) suspended solids, scaling refers to the precipitation and
crystallization due to exceeding salt solubilities. In process configuration consisting of serial
connected FO modules, fouling will occur in the first stages while scaling usually occurs in
the consequent stages.

Although FO is generally considered a low fouling propensity membrane technology which

can handle more difficult-to-treat feed stream, fouling and scaling will ultimately reduce
the membrane performance. Indications are a reduced water flux, increased pressure drops
on the feed side of the membrane, as well as reduced rejections. Besides an appropriate pre-
treatment of the feed solution to remove suspended solids, frequent cleaning-in-place (CIP)
can mitigate the deposition of solids on the membrane surface and performance detriment.
Scaling should be prevented by estimating the scaling risk of a certain feed composition by
using the Scaling Index and avoiding working at water recoveries that could provoke severe
compound precipitation.

4.4 DATA ANALYSIS: BASIC FO PROCESS DESIGN

4.4.1 FO Fundamental Equations

In a typical FO process, the equation for water flux flowing from feed side to draw side is
given by:

J w = A⎡⎣( PF − PD ) = π Dm − π Fm ⎤⎦
( ) Eq. 11

Where Jw is water permeation flux, A is the water permeability, PF and PD is hydraulic

pressure of feed side and draw side respectively, and π is osmotic pressure of draw side and
feed side at membrane surface, respectively. In FO operation, hydraulic pressure difference
tends to be zero or close to zero.
The salt flux equation is given by:

( )
JS = B C Dm C Fm = B Cm Eq. 12

where Js is sat flux from draw to feed, B is the salt permeability, and C is solute concentration
in draw and feed solution at membrane surface, respectively.

87
Experimental Methods for Membrane Applications

The salt transport across the FO membrane is also described by the convection-diffusion
model with a diffusive term proportionally related to solute concentration gradient and
a convective term related to water permeate flux across the membrane in the opposite
direction.
dC(x)
Js = D J wC(x) Eq. 13
dx
Where D is the solute diffusion coefficient. The solution of the transport equations above
differ depending on the orientation of the membrane.

In active layer facing feed side (AL-FS) mode or FO mode, water permeates from feed side into
the support layer on the draw side, leading to dilutive internal and external concentration
polarization (i.e., ICP and ECP, respectively). On the feed side, the convective water flux
carries solutes from bulk feed solution to membrane surface, at which they are rejected and
accumulate, causing concentrative ECP. The solution of the convective-diffusive equation
above, for AL-FS mode, become:

Jw 1 S
C Db exp C Fb exp J w +
kD k F DF Eq. 14
Cm,ALDS =
B 1 S Jw
1+ exp J w + exp
Jw k F DF kD
For active layer facing draw side (AL-DS) mode or PRO mode, water permeates from
feed side with solutes that are rejected and accumulate across the support layer, resulting
in concentrative ICP and ECP on the feed side. On the draw side, there is dilutive ECP as
pure water permeates into the draw side. The solution of the convective-diffusive equation
above, for AL-FS mode, becomes:

Jw 1 S
C Db exp C Fb exp J w +
kD k F DF
Cm,ALDS = Eq. 15
B 1 S Jw
1+ exp J w + exp
Jw k F DF kD

Where k is mass transfer coefficient, and the term ‘exp(-Jw/kD)’ indicates external
concentration polarization in general whereas the term exp[Jw(1/kF + S/Df)] denotes internal
concentration polarization with S being structural parameter of membrane, consisting of
porosity and tortuosity term used in modifying solute diffusion coefficient from the bulk
solution to the inside support layer.

The mass transfer coefficient k value is dependent on the type of membrane form factor and
module. In general, mass transfer coefficient is:
Sh × D
k= Eq. 16
dh

88
 Chapter 4

where Sh is Sherwood number and dh is hydraulic diameter, both being geometry-

dependent.

Combining above stated equations, one is able to calculate the expected water permeation,
Jw, and reverse solute flux, Js, of a FO membrane, given its bulk feed and draw solution
characteristics and some basic hydrodynamic information to obtain mass transfer
coefficients.

4.4.2 FO Module Mass Balance

To simulate transport inside a membrane module, mass balance equations should be
considered. In addition, the effect of volume change due to dissolved solute should also be
taken into account. This means the differential term of density and concentration of solute
cannot be neglected.

Typically, mass balance equations for pressure, velocity and concentration along module
length can be established. For instance, the velocity and concentration differential equation
on the feed side can be seen below for a rectangular flat plate channel type.

w d F× 1
Jw× + Js J ×
dv F
dc F s H Eq. 17
= F
dx F ,bulk d
×cF
F
dc

dv F 1
F cF × + ×J
dc dx H s Eq. 18
=
dx vF

Where rW is density of pure water, vF is the differential term to account for volume change
with solute concentration and H is the height of flat plate flow channel. In other geometries,
such as for hollow fiber or tubular types, these terms are referred to inner diameter of hollow
fiber (this assumes an inside out FO module with active layer being on the lumen side).

Similarly, the velocity and concentration differential equation on the draw side for a
rectangular flat plate channel type can be seen below.

w d D× 1
Jw × Js + J × Eq. 19
dv D
dc D s H
=
dx D d D D
×c
dc D

dv D 1
D cD× + ×J
dc dx H s Eq. 20
=
dx vD

89
Experimental Methods for Membrane Applications

For a hollow fiber bundle, the H hydraulic radius term becomes the following for lumen and
shell respectively:
Hlumen = 4/di Eq. 21

Hshell = (4 × n × di)/(n × do2 - Di2) Eq. 22

where di is fiber inner diameter, do is fiber outer diameter, Di is shell housing inner diameter
and n is the total number of fibers.

Similarly, the velocity and concentration differential equation on the draw side for a
rectangular flat plate channel type can be seen below.

w d D 1
Jw × Js + ×J ×
Eq. 23
dv D
dc D s H
=
dx D d D D
×c
dc D

dv D 1
D c D× + ×J
dc dx H s
= Eq. 24
dx vD
It should be noted that the sign of the velocity differential equation is reversed in the event
of counter current flow.

The pressure drop equation across the module strongly depends on the type of module
used. As an example, for the hollow fiber form factor, the analogy of flow through a packed
bed with the Ergun equation could be used to model pressure drop across the tube bundle
on the shell side.

150 (1 ) ×μ
2 D D 2
dP D ×v 1.75× (1 )× D
×vD Eq. 25
= × ×
3 2
+ 2
dx ×d o
× do

where θ is empirical pressure drop correction factor and α is flow direction (1 for counter
current, and -1 for co-current). ε is packing density of hollow fiber bundle, μ is fluid dynamic
viscosity, r is fluid density, and u is fluid velocity. D denotes draw solution side, which
typically flows on the shell side of a hollow fiber module.

Meanwhile, the Hagen-Poiseuille model for pressure drop across cylindrical tube is used for
the lumen side pressure drop:

dP F 32 ×μ F × v F Eq. 26
=
dx di2

where P is pressure, F denotes the feed solution side, which typically flows on the lumen
side of a hollow fiber bundle, μ is fluid dynamic viscosity, u is fluid velocity and di is the
fiber inner diameter.

90
 Chapter 4

4.4.3 FO Design Considerations

For the batch or feed-and-bleed type of system, feed solution is re-circulated and water
extraction happens over time. Sensors may be installed to automate the feed-and-bleed or
cycle shutdown operation. In such a system, because the feed side is being concentrated,
water flux will start high and decrease over the course of a cycle, assuming that draw inlet
flow rate and concentration are constant. Care should be taken to ensure that initial flux is
below design flux limit for the given process, and final flux is non-zero so that cycle time is
still productive and reverse salt flux into feed batch is minimized.

For a single pass process where the FO feed outlet is expected to reach a desired concentration
factor, the number of modules and their array should be designed to achieve recovery
outcome, while balancing all design constraints above.

For instance, FO modules may be arranged in parallel to sub-divide flow to be within

recommended flow rate. FO modules, and hence membrane area, may be added in series
to achieve recovery in single pass, while design flux limit is obeyed. For the same flow rate
extraction requirement, added area means lower operating flux, ensuring that it is within
design limit. The maximum number of modules in series is dictated by pressure drop across
the system, while the maximum number of lines of modules in parallel is dictated by the
minimum FO outlet flow rate for each line.

One way to circumvent minimum FO outlet flow rate being below limit is to implement
multi-stage design. That means, the flow rate of multiple lines of FO modules of the so-
called first stage are combined and redistributed over a smaller number of lines in the second
stage. This allows more flow per module when recovery is at the highest point and by design,
above module limit by manufacturers’ recommendation.

Lastly, it should be noted that process limits should be considered for both flushing or
cleaning process as well as FO process. In the former, cross flow velocity on the feed side is
highest, and on the later cross flow velocity on the draw side is highest. It should be ensured
that design considerations are met for both operation types for successful commissioning
of a FO system.

Other design considerations

Beside technical considerations, there are other parameters that FO process designers should
pay attention to as it influences the operating cost of such a system. Most directly, increasing
the number of FO modules used will increase the cost of membrane replacement and initial
capital investment on the system. This will also increase the hold-up volume and volume of
flushing water or chemicals required for the cleaning process, even though this tends to take
a small fraction of overall operating cost.

If operating flux is still within design flux limit, increasing draw solute concentration results
in less membrane required and reduces membrane initial investment. However, this would
result in increased reverse salt flux from draw to feed side, increased salt passage into draw
regeneration permeate stream and increased energy cost of draw regeneration step.

91
Experimental Methods for Membrane Applications

4.5 APPLICATION EXAMPLES

Textile industry application of FO for lowering water footprint

Textile production is estimated to be responsible for about 20% of global clean water
pollution from dyeing and finishing products (Morlet et al., 2017 ). Given the increasing
need for the textile industry to lower the environmental impact it is necessary not only
to design appropriate wastewater treatment technologies but also to enable reuse and
recycling of water. Here FO can be used for water reclamation using concentrated dyeing
salt solutions as draw solutions where the diluted dyeing salt solutions can be used in the
dyeing baths directly (see Figure 5).

Following the study of Sheldon et al., (2019) it was evaluated the potential of dye solutions
as a novel draw solution by screening, assessing and identifying suitable reactive dyes,
e.g., Reactive Black 5 and Basic Blue 41 GRL dyeing solutions were investigated as draw
solutions in FO with a dye-to-salt 1:10 mass ratio, see Figure 1. Synthetic seawater (SSW)
and two types of textile wastewater (TWW1 and TWW2) were evaluated as feed solutions
for water reclamation. Reactive Black 5 and Basic Blue 41 GRL were diluted 10 and 5 times
respectively.

With Reactive Black 5 as draw solution and SSW as feed solution a water recovery of 75%
was achieved. Using TWW1 and TWW2 as draw solutions, water recovery was around
30%. Using Basic Blue 41 GRL with SSW, TWW1, and TWW2 as feed solutions, water
recoveries of 50%, 20% and 20%, respectively, were achieved. The average reverse solute
fluxes were between 0.06 and 0.34 g/m2h. Results indicated the potential of FO in the
textile industry leading to substantial water savings.

Concentrated
salt solution
Concentrated batching
dyeing rolls
wastewater fabric

FO unit RO unit
Textile
dyeing
Water process

Diluted Dyeing
salt solution wastewater

Figure 5 Implementing forward osmosis (FO) into the textile wastewater treatment process can
provide high value to an industry segment which is a large consumer of fresh water and
one of the biggest polluters. The scheme shows the FO process integrated in a textile
wastewater treatment plant using inorganic salt as a draw solution. For using the salt
solution as a draw solution there is an integrated reverse osmosis unit for the reuse of the
diluted salt.

92
 Chapter 4

Concentrating distillery wastewater for subsequent antioxidant retrieval

Alcohol distillation from sugarcane molasses constitutes an important industry in several
countries. Molasses-based distillation is a water intensive method with a freshwater
consumption in the range of 9-21 L per alcohol and concomitant wastewater production
of 7-15 L per L alcohol (Gol, 2014). The resulting wastewater has a high organic load, low
pH, and high total dissolved solids. About 2% (w/v) of the wastewater is melanoidins,
a product of Maillard reaction obtained from reducing sugars and amino acids during
distillation. From a classical wastewater treatment point of view this makes this particular
stream problematic as melanoids are not readily biodegradable. However, melaniodins have
antioxidant properties which could be a valuable sub-product. The high organic load and
the high total dissolved solids makes separation based on classical filtration challenging but
due to the inherently low fouling potential FO has attracted attention as a method for up-
concentration of this potential antioxidant source.

Singh et al. (2018) studied the concentration of distillery wastewater by FO with magnesium
chloride hexahydrate (MgCl2.6H2O) as draw solution. They used a 10% v/v melanoidins
model feed solution to optimize the operational parameters. Subsequently they achieved
85-90% melanoidins rejections with as-received distillery wastewater and 3M MgCl2.6H2O
as draw solution. The water flux was 2.8 L m-2h-1 with water recovery over 24 h was around
70% which is significantly higher than reported for RO (35-45%). However, further
investigations on membrane fouling and draw solution recovery are required to establish
the superiority of FO over RO for the concentration of this type of wastewater.

Concentrating electroplating wastewater

Chromium plating and chromate processes are widespread technologies for electroplating
of pristine or nickel-coated plastics as chromium and chromate endow surfaces with special
properties such as hardness and corrosion resistance (Korzenowski et al, 2018; Sorme et
al., 2002). In this process, large quantities of wastewaters, residues, and sludge is generated
which can be categorized as problematic waste requiring extensive waste treatment (Sorme
et al. 2002).

In the study of Bratovcic et al. (2022) FO was investigated for concentration of hexavalent
chromium (Cr(VI)) in electroplating wastewater from processing plastics to enable the
reuse of recovered Cr(VI) in the plating baths, see Figure 6. The feed solution was chromium
galvanic wastewater, while the draw solution was an underground brine (close to the factory
location) with osmotic pressures of 28 and 226.8 bar, respectively.

Baseline and FO filtrations were performed using Aquaporin Inside(R) membrane hollow
fibre FO (AIM™ HFFO) modules with a sequence of baseline, filtration (1.5h) and cleaning
(30 min with DI water) steps. During the initial filtration (F1), the water flux decreased on
average from an initial value of 28.7 LMH at 46.7 % water recovery to 18.5 LMH. For the
second filtration (F2) the water flux decreased from 20.1 LMH at 28.4 % water recovery to
16.8 LMH. The corresponding feed solution (wastewater) volume reduction factors were
1.9 and 1.4 with a concomitant Cr(VI) concentration factor of 1.6 and 1.3 for F1 and F2,
respectively. After 1.5 h of filtration, the Cr(VI) rejection was 99.7 % and 95.8 % for F1 and
F2, respectively. As the AIM™ HFFO membrane is negatively charged electrostatic repulsion

93
Experimental Methods for Membrane Applications

between the membrane surface and the negative ions (HCrO4− and Cr2O72-) will contribute
to the rejection of Cr(VI). The appearance of Cr(VI) in the draw solution indicated a loss of
membrane integrity which was ascribed to chemical degradation of the membrane due to
oxidation from Cr(VI). Local guidelines for standard chromium discharge from industrial
wastewater into the environment is 0.5–1 mg L-1. Since the diluted brine draw solution
contained 0.07 gL-1 and 0.65 gL-1 of Cr(VI) for F1 and F2 respectively, it cannot be directly
discharged into the salt groundwater resource.

In conclusion, brine-driven FO could concentrate chromium galvanic wastewater taking

advantage of the high chemical potential gradient provided by the high salinity brine, but
the membrane material must be adapted to withstand harsh environments.

102

Cr(VI) Plating Factory 100

Cr(VI)
Cr(VI) 98
Cr(VI) 96
Cr(VI)
Plating bath 94

92
1h 1.5h 1h 1.5h
90
24 F01 F02
Rejection of Cr(VI)
Cr
Chromium
51.996 30 2.2
25 2.0
Iw (LMH)

20 1.8

CF
15 1.6
10 1.4
5 1.2

Chrome Underground 0 1
wastewater brine 0 10 20 30 40 50
R (%)
Hollow fibre FO
Water flux and concentration
membrane module
versus recovery

Water passage through

the membrane

Figure 6 FO tests using Aquaporin Inside membrane hollow fibre FO (AIM™ HFFO) modules
for concentration of hexavalent chromium (Cr(VI)) in electroplating wastewater from
processing plastics to enable the reuse of recovered Cr(VI) in the plating baths. Chromium
galvanic wastewater was used as feed solution while the draw solution was underground
brine close to the factory location. The results show that FO can be used in this type of
application, but the membrane material must be adapted to withstand harsh environments
(Bratovcic et al, 2022)

94
 Chapter 4

4.6 OUTLOOK

FO is a relatively new technology which presents numerous advantages, especially when

a direct FO system can be implemented (i.e., draw solution is available and regeneration is
not needed) or when the resulting feed concentrate can bring an added value to the final
product. However, FO presents the drawbacks of a developing technology. These are
mainly the scarce availability of FO membrane manufacturers, the development of materials
which ensure a high water flux, high compound rejection, withstand harsh environments,
high selectivity to water and a reduced concentration polarization. The unique system
design characteristics required by the FO technology (i.e., draw solution regeneration
and membrane configurations) also involve an additional level of system complexity.
The availability of non-expensive draw solutions with the desired characteristics and the
suitability of these in those applications that require high safety levels, such as in food and
pharma industries, are also challenging. However, overall, FO technology can still bring
unique advantages in niche applications, although more research in membrane materials
and processing are needed to fully understand its capabilities in industry.

95
Experimental Methods for Membrane Applications

4.7 REFERENCES

Baker, R. W. (2004). Membrane Technology and Applications. John Wiley & Sons, Ltd.
DOI:10.1002/0470020393
Bratovcic, A., Buksek, H., Helix-Nielsen, C., Petrinic, I., Concentrating hexavalent chromium
electroplating wastewater for recovery and reuse by forward osmosis using underground brine
as draw solution Chemical Engineering Journal, 431, #133918, 2022.
Gol, 2014. Report by Principal Scientific Advisor to Government of India. Electronic source. http://
psa.gov.in/publications-reports/opportunities-green-chemistryinitiatives-molasses-based-
distilleries-2014
Im Sung-Ju, Lee H., Jang A. (2021). Effects of co-existence of organic matter and microplastics on the
rejection of PFCs by forward osmosis membrane. Environmental Research, 194, 110597
Jørgensen, M. K., Keiding K., Christensen, M.L. (2014). On the reversibility of cake buildup and
compression in a membrane bioreactor. Journal of Membrane Science, 455, 152-161
Korzenowski C., Rodrigues M.A.S., Bresciani L., Bernardes A.M., Ferreira J.Z., Purification of spent
chromium bath by membrane electrolysis, J. Hazard. Mater. 152 (3) (2008) 960–967.
Morlet, A., Opsomer, R., Herrmann, S., Balmond, L., Gillet, C., and Fuchs, L. (2017). A new textiles
economy: Redesigning fashion’s future. Ellen MacArthur Foundation.
Mulder, M. (1996). Basic Principles of Membrane Technology. Springer. DOI: 10.1007/978-94-009-
1766-8
Peters, Christian D. and Hankins, Nicholas P. (2020). The synergy between osmotically assisted reverse
osmosis (OARO) and the use of thermo-responsive draw solutions for energy efficient, zero-
liquid discharge desalination, Desalination, Volume 493, 114630
Sanahuja-Embuena, V.; Khensir, G.; Yusuf, M.; Andersen, M.F.; Nguyen, X.T.; Trzaskus, K.; Pinelo, M.;
Helix-Nielsen, C. Role of Operating Conditions in a Pilot Scale Investigation of Hollow Fiber
Forward Osmosis Membrane Modules. Membranes 2019, 9, 66. https://doi.org/10.3390/
membranes9060066
Sheldon et al. Water Sci Technol (2019) 80 (6): 1053–1062
Singh et al. Water Research 130 (2018) 271-280
Sorme L. and Lagerkvist R., Sources of heavy metals in urban wastewater in Stockholm, Sci. Total
Environ. 298 (1-3) (2002) 131–145.
Suwaileh W., Pathak, N., Shon H., Hilal N., Forward osmosis membranes and processes: A
comprehensive review of research trends and future outlook, Desalination, Volume 485, 2020,
114455, ISSN 0011-9164, https://doi.org/10.1016/j.desal.2020.114455.
Wang J, Liu X (2021) Forward osmosis technology for water treatment: Recent advances and future
perspectives. Journal of Cleaner Production 280: 124354 DOI https://doi.org/10.1016/j.
jclepro.2020.124354
Xiao, T., Nghiem, L. D., Song, J., Bao, R., Li, X. & He, T. (2017). Phenol rejection by cellulose triacetate
and thin film composite forward osmosis membranes. Separation and Purification Technology,
186 45-54.

96
 doi: 10.2166/9781789062977_0097

Chapter 5

Membrane Distillation
Mohammad Mahdi A. Shirazi, Aalborg University, Denmark

Cejna Anna Quist-Jensen, Aalborg University, Denmark

Aamer Ali, Aalborg University, Denmark

The learning objectives of this chapter are the following:

• Introduction to membrane distillation

• An overview of the membrane materials and setup used in MD

• Main techniques to characterize the membranes for MD

• Present and discuss the main applications of MD

• Provide an overview of the outlook of the process.

5.1 INTRODUCTION

Membrane distillation (MD) is a non-isothermal membrane process (Lawson and Lloyd,

1997a). In MD, a hydrophobic membrane with porous structure separates the feed and
permeate channels. The feed channel contains a heated solution with an elevated temperature,
while the permeate channel contains a cooling solution with lower temperature (Curcio and
Drioli, 2005). This temperature difference can provide a vapor pressure difference between
the feed and permeate channels. As the membrane is hydrophobic, the liquid feed does not
penetrate the pores (El-Bourawi et al., 2006). However, due to the vapor pressure difference
between the hot side and the cold side (i.e., the driving force), the vapor molecules can
transfer across the membrane pores (Wang & Chung, 2015). Figure 1 illustrates a general
scheme of the MD process.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Tf

Tfm

Tpm

Tp

Water vapors
Ions
Other solids

Feed solution

Figure 1 A general scheme of the MD process (Tf: bulk temperature in the feed channel, Tfm:
temperature on the membrane surface in the feed channel, TPm: temperature on the
membrane surface in the permeate channel, and Tp: bulk temperature in the permeate
channel).

MD has numerous attractive advantages for water treatment applications. As only the vapor
molecules can pass through the membrane pores a complete solute rejection (i.e., ~100%,
theoretically) can be achieved by the MD process. MD uses membranes with pores in the
range of 0.1-0.5 μm, which is much larger than the pore size range in the pressure-driven
membranes, such as reverse osmosis (RO) and nanofiltration (NF) membranes. Moreover,
the operating pressure is much lower in MD compared with RO/NF. These can make
MD more cost-effective, less demanding on the membrane properties, and less sensitive
to fouling/scaling, as well (Shirazi et al., 2016; Tibi et al., 2020). Furthermore, the low
operating pressure allows MD to utilize less expensive materials with better anti-corrosive
properties, such as plastics, for module design and fabrication (Hussain et al., 2022). In MD,
the process liquid in the feed stream is not necessarily heated up to reach the boiling point
and the operating temperature ranges between 40-80 oC. This can reduce the required
energy for MD in comparison with the conventional distillation processes. Moreover,
working with low temperatures enables MD to utilize low-grade and renewable energy
sources, such as solar energy (Ahmed et al., 2020; Drioli et al., 2015b). All these can make
MD an attractive alternative for the desalination of seawater and brackish water, however,
in lower production rates in comparison with RO. Thus, MD may not still be competitive
with RO for freshwater production via seawater desalination for large scales but perform
effectively for small-scale desalination. Moreover, MD could find a way as an efficient
technology for treating challenging wastewater streams, such as RO brine, textile dyeing
wastewater, radioactive contaminated wastewater, etc., where other separation processes
cannot perform efficiently (Shirazi & Dumée, 2022). Table 1 highlights the advantages and
challenges associated with MD for various applications.

98
 Chapter 5

Table 1 Advantages and challenges associated with MD

Advantages Challenges
Low operating temperature (40-80 ˚C) Lack of specific membranes
High non-volatile solute rejection (~100%) High risk of pore wetting
Very low operating pressure (<0.2 bar) Heat loss
Less sensitivity to the feed concentration Temperature polarization
Modular and scalable Complexity in terms of process optimization
Integrate-able with other separation techniques Lack of experience in large scale

Competitive production cost with other processes

Possible to utilize renewable and low-grade energy
source
Less carbon footprint
Able to treat challenging wastewater streams

The very first MD experiments were conducted using commercially available microfiltration
(MF) membranes, which were made of hydrophobic polymers, such as polytetrafluorethylene
(PTFE), polypropylene (PP), and polyvinyl difluoride (PVDF) (Curcio & Drioli, 2005;
Shirazi et al., 2013a; Shirazi et al., 2012). However, they suffer from some drawbacks, such
as pore wetting, delamination of the active layer, etc., as they were not specifically fabricated
for MD (Wang & Chung, 2015). An extensive review of commercial membranes for MD
can be found in the literature (Khayet, 2011). The next generation of MD membranes
have been fabricated using commercial and synthesized polymers, such as PVDF. The MD
membranes can be fabricated using various techniques. The most investigated one is the
phase inversion technique (Eykens et al., 2017). However, the membranes by this technique
may suffer from some drawbacks, such as limited porosity, and dead-ended pores with
twisted structure (i.e., high tortuosity) (Qasim et al., 2021b). Thus, new research directions
have been focused on new fabrication techniques, such as electrospinning and 3D printing
(Tijing et al., 2014; Tijing et al., 2020). Moreover, extensive research was also carried out on
MD modelling and optimization (Ali et al., 2015a; Ali et al., 2016a; Ali et al., 2018; Hitsov
et al., 2015; Jantaporn et al., 2017; Olatunji & Camacho, 2018).

The new generation of membrane materials and fabrication techniques, and process
development and optimization have enabled MD to extend their applications in new
directions for more efficient and greener desalination towards zero-liquid discharge, treating
challenging wastewater streams (e.g., wastewater from biological and pharmaceutical
processes, metal finishing, and electronic industry), removal of specific gas streams (e.g.,
H2S) from process water, concentrating fruit juices in the food industry; and recovery of
minerals and value-added chemicals from wastewater streams (Tibi et al., 2020; Hussain et
al., 2022; Sanaeepur et al., 2022; Julian et al., 2022; Gontarek-Castro et al., 2022; Afsari et
al., 2021). Despite of all developments in this separation technique, MD still needs further
development for specific membranes with durable properties, high pore wetting resistance,
and long-term performance, new module design for higher energy efficiency and lower
polarization effect as well as experiencing new and challenging feed samples for treatment.

99
Experimental Methods for Membrane Applications

5.2 MATERIALS, EXPERIMENTAL SET-UP

5.2.1 MD membranes

5.2.1.1 Membrane properties

As MD is a non-isothermal membrane separation and only the water vapor molecules must
pass through the pores, the applied membrane should possess some essential requirements
(Alklaibi and Lior, 2005).

The most important characteristic of an MD membrane is surface hydrophobicity. The

membrane structure can possess a single layer or even multi-layers. However, the layer
which is in direct contact with the feed stream must be hydrophobic to repel the liquid and
only pass the vapor molecules (Shirazi et al., 2014).

To ensure a sustainable performance without pore wetting, the MD membrane should have
high liquid entry pressure (LEP). LEP defines as the minimum required pressure that allows
the feed liquid to enter the pores (He et al., 2011). To ensure a proper LEP value, further to
high surface hydrophobicity, the membrane pore size should be as small as possible. The
typical pore size for MD membranes has been reported in the range of 0.1-0.5 μm (Khayet,
2011). Moreover, the pore size distribution and the maximum pore size should be as narrow
as possible and as small as possible, respectively, to provide a high LEP value (McGaughey
et al., 2020a).

Porosity is another important parameter for MD membranes. It defines as the void volume
fraction which is openly available for transferring the vapor molecules (Gryta, 2007).
The membrane porosity is proportional to its permeability. Thus, the more porous the
membrane, the higher the permeate flux, regardless of the MD configuration. It is worth
noting that the fabrication technique can considerably affect the membrane porosity
(Ravi et al., 2020). For example, nanofiber membranes possess higher porosity (≥85%) in
comparison with phase-inverted membranes (40-80%) (Tijing et al., 2014).

The pore structure of MD membrane is often assumed to be a straight cylinder for modelling
purposes, while it is not true in real life. The deviation of the pore structure from the standard
cylindrical shape is defined by the tortuosity factor. Unlike porosity, tortuosity is inversely
proportional to the permeability of the MD membrane. Therefore, the lower the tortuosity
factor, the higher the permeate flux (Tai et al., 2019; Kim, 2021).

In MD, both mass and heat transfer through the membrane happen simultaneously
(Qtaishat et al., 2008). Although a high mass transfer (higher permeate flux) is favorable
for MD, high heat transfer through the membrane is considered as heat loss (Phattaranawik
et al., 2003a). With reference to the membrane thickness, a thin membrane can have
lower mass transfer resistance (higher permeate flux) and higher heat loss, while a thicker
membrane can have lower heat loss and higher mass transfer resistance (lower permeate
flux). Thus, the membrane thickness should possess an optimum value to compromise the
heat and mass transfer in MD. To achieve this, membranes with multi-layer structures have
been introduced for MD applications, where a thin hydrophobic membrane provides lower

100
 Chapter 5

mass transfer resistance and higher permeate flux, while the thicker and less hydrophobic
or hydrophilic support layer can reduce the heat loss (Bonyadi & Chung, 2007; Cheng et al.,
2018; Shirazi et al., 2020a; Zuo et al., 2017).

Table 2 Guideline for desired properties of MD membranes

Property Description Recommendation
Hydrophobicity The higher the hydrophobicity (i.e., the lower the As high as possible
surface energy), the higher the liquid retention.
Liquid entry pressure (LEP) The higher the LEP, the more pore wetting resistance. >250 kPa
Pore size Larger pore size can provide higher permeate flux; 0.1 – 0.45 µm
however, it reduces the LEP.
Porosity The higher the porosity; the higher the permeate flux, 60-80%
the lower the heat loss, and the lower the mechanical
strength.
Thickness The lower the thickness; the higher the permeate flux 30-450 µm
and heat loss.
Thermal conductivity The lower the thermal conductivity, the higher the As low as possible
permeate flux and energy efficiency.
Tortuosity The higher the tortuosity, the lower the permeate flux. As low as possible
Chemical resistance The higher the chemical resistance, the better the As high as possible
membrane integrity.
Mechanical strength The higher the mechanical strength, the better the As high as possible
membrane integrity.

As mentioned earlier, heat loss through the thermal conduction of the membrane is an
important challenge for MD (Phattaranawik et al., 2003b; Susanto, 2011a). Therefore, the
thermal conductivity of the membrane material should be as low as possible. Moreover,
the higher the porosity, the lower the heat loss through the membrane, as the heat transfer
coefficient of the trapped air in the membrane pores is much lower than that of the polymer
thermal conductivity (Eykens et al., 2016). Therefore, low thermally conductive polymers
can be used for fabricating a highly porous membrane to enhance the MD performance
(Shirazi et al., 2020a).

Further to all the above-mentioned characteristics, MD membranes should also possess anti-
fouling properties and should be chemically and thermally durable for stable performance
in the long-term operation (Chen et al., 2020). Table 2 provides a guideline for the desired
properties of MD membranes.

5.2.1.2 Membrane materials

The very first MD experiments were mostly focused on the process (e.g., proving the
concept, optimization of operating parameters, enhancing energy efficiency, etc.) (Lawson
and Lloyd, 1997b). These experiments were mostly carried out using commercial MF
membranes which were made of hydrophobic polymers (Franken et al., 1987a; Fane et al.,
1987; Schofield et al., 1987; Kimura et al., 1987; Schofield et al., 1990). However, these

101
Experimental Methods for Membrane Applications

membranes have not been specifically designed and fabricated for MD, which is a non-
isothermal separation based on the vapor-liquid interface equilibrium. This was found
as a serious challenge for optimizing and developing the application of MD. Therefore,
a considerable part of research in the next step has been focused on developing novel
membranes with enhanced performance in terms of anti-wetting properties, permeate flux,
solute rejection, and energy efficiency (Wang & Chung, 2015).

Table 3 Commercial polymers for the fabrication of MD membranes (Wang and Chung, 2015;
Qasim et al., 2021a)
Thermal
Chemical Surface energy conductivity Thermal Chemical
Polymer structure (×10-3 Nm-1) (Wm-1K-1) stability stability
PTFE F F 9-20 ~0.26 Very good Very good

C C

F F n

PVDF 30.3 ~0.18 Moderate Good

H F

C C

H F
n

PP 30 ~0.14 Moderate Good

CH3

PE H H 28-33 ~0.4 Poor Good

C C

H H n

MD membranes can be fabricated from both polymeric and inorganic materials. However,
the considered material must be hydrophobic (i.e., with low surface energy) intrinsically,
or by proper modification (Tibi et al., 2020). The common commercial polymers for the
fabrication of MD membranes include PTFE, PP, PVDF, and polyethylene (PE) (Qasim et
al., 2021b; Wang & Chung, 2015). Table 3 presents the characteristics of some common
hydrophobic polymers. As could be observed, PTFE has the lowest surface energy (9-
20*10-3 N.m-1), which can provide a relatively high surface hydrophobicity. Moreover, it is
chemically and mechanically stable. However, due to the insolubility of PTFE polymer in the
majority of chemical solvents, PTFE membranes should be fabricated using complicated and
expensive techniques, such as melt-extrusion (Feng et al., 2018). PVDF has higher surface
energy than that of PTFE, which means it is less hydrophobic. However, it has a lower
thermal conductivity in comparison with PTFE (Table 3). It is worth noting that PVDF and
their derivatives, such as poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP),

102
 Chapter 5

are the most investigated polymers for developing MD membranes (Qasim et al., 2021a).
This is due to the numerous advantages of these polymers, such as proper solubility in a
wide range of solvents and good processability, for membrane fabrication (Kang and Cao,
2014; Ji et al., 2015; Zou and Lee, 2022).

Recently, with developing new techniques for membrane fabrication, such as electrospinning
(which can provide nanofibers), new commercial polymers have also been investigated for
fabricating the MD membranes, such as polystyrene (PS), high-impact polystyrene (HIPS),
styrene-acrylonitrile (SAN), poly (methyl methacrylate) (PMM), acrylonitrile-butadiene-
styrene (ABS), styrene-butadiene-styrene (SBS), and polydimethylsiloxane (PDMS) (An et
al., 2017; Duong et al., 2018; Niknejad et al., 2021; Sadeghzadeh et al., 2020a; Shirazi et al.,
2020b).

Table 4 Comparison of polymeric and inorganic membranes for MD

Item Polymer membrane Inorganic membrane
Membrane
Material - Widely available - Available
Fabrication - Well-developed techniques - Developed techniques
- Inexpensive - Expensive
Mechanical strength - Flexible - Brittle
- Less durable in long term - Durable in long term
Fouling and cleaning - Sensible to fouling - Sensible to fouling
- Cleaning is challenging - Flexible in cleaning
Corrosion resistance - High - Moderate
Performance
perating temperature* - Low to moderate - Low to high
Permeate flux - Low (AGMD) to moderate (VMD) - Moderate (AGMD) to high (VMD)
Solute rejection - High - High

* Temperature range: Low (30˚C) to High (90˚C)

Further to polymers, inorganic materials can also be used for the fabrication of MD
membranes (Ramlow et al., 2019). The inorganic membranes can be made of a wide range of
materials, such as alumina, zirconia, titania, silicon nitride, and metal oxides of iron (Camacho
et al., 2013). However, these materials are hydrophilic in nature (e.g., due to the presence
of the hydroxyl group). Thus, inorganic membranes should be modified properly for their
surface to impart the required hydrophobicity for MD applications (Ferreira et al., 2021).
Inorganic membranes, in particular ceramic membranes, are more chemically, mechanically,
and thermally stable in comparison with polymeric membranes. Moreover, they can be
cleaned several times, even using extreme cleaning techniques, such as chemicals, steam, and
high-pressure backwash. However, inorganic membranes are expensive, and brittle, and show
higher fouling and scaling tendency in comparison with polymeric membranes (Omar et al.,
2022). Table 4 compares the polymeric and inorganic membranes for MD.

103
Experimental Methods for Membrane Applications

5.2.2 Experimental set-up

5.2.2.1 MD configurations
MD has four main configurations, as shown in Figure 2. All these configurations are identical
for the feed channel, where the hot solution stream is in direct contact with the hydrophobic
surface of the membrane (Adewole et al., 2022).

Condenser
Feed in Permeate out Feed in Sweeping
gas out

Permeate
Permeate

Permeate
Feed

Feed
(a) (b)

Feed out Permeate in Feed out Sweeping gas in

Feed in Feed in Cooling out

Permeate

Air gap
Feed

Feed
(c) (d)

Vacuum
pomp

Feed out Permeate Feed out Cooling in

Permeate

Figure 2 A general scheme of the main MD configurations: (a) direct contact membrane distillation
(DCMD), (b) sweeping gas membrane distillation (SGMD), (c) vacuum membrane
distillation (VMD), and (d) air-gap membrane distillation (AGMD).

When further to the feed side, the applied membrane is in direct contact with the liquid in
the permeate side, which is pure and cold, it is known as the direct contact MD (DCMD)
(Figure 2-a). This is the most investigated and simplest MD configuration (Khayet,
2011). DCMD has been used for different applications, such as seawater desalination and
wastewater treatment. However, the most important drawback of DCMD is the high heat
loss through the membrane thermal conduction (Ashoor et al., 2016; Niknejad et al., 2021).

104
 Chapter 5

Table 5 A comparative overview of MD configurations

Configuration Description Highlights
DCMD • Both feed and permeate channels • Simple modules design.
contains liquid streams which • Internal condensation.
are in direct contact with the • Possibility for internal heat recovery.
membrane surfaces. • Pure/fresh water is required for cooling stream.
• The most investigated MD configuration.
• Wide range of applications.
• High heat loss and low thermal efficiency.

SGMD • An inert gas stream with low • Lower permeate flux compared to VMD.
temperature is used for sweeping • External condenser is required.
the vapor molecules in the • The least investigated MD configurations.
permeate channel. • Promising for concentration application, where the
permeate can be vented.
• The operating condition in the permeate. channel is
more effective on flux compared to DCMD.
• Good when feed stream contains volatile
compounds.
• Low heat loss.
• Challenging in terms of heat recovery.

VMD • Vacuum is used to enhance the • High permeate flux.

permeation of vapor molecules in • Higher risk of pore wetting.
the permeate channel. • Low heat loss.
• Challenging in terms of heat recovery.
• External condenser is required.
• Good for processing solutions with low. vapor
pressure at a desire temperature.
• Feasible for large scale.

AGMD • An air gap is maintained between • High energy efficiency.

the membrane and a condensing • Internal condensation.
surface in the permeate channel. • Possibility for internal heat recovery.
• Seawater and non-hazardous (pre-treated)
wastewater can be used as cooling stream.
• Lower permeate flux compared to other
configurations.
• Wide range of applications.
• Feasible for large scale.

The water vapor molecules in the permeate channel can be collected either by imposing
a vacuum pressure or by imposing a sweeping gas flow. Under these circumstances, the
configurations are known as the vacuum MD (VMD) and the sweeping gas MD (SGMD),
respectively (Figures 2-b and 2-c, respectively) (Abu-Zeid et al., 2015; Said et al., 2020).
Although SGMD and VMD can provide relatively higher permeate flux in comparison with
DCMD along with moderately better energy efficiency, an external condenser is required
for both configurations, which can increase the operation cost and energy consumption
(Khayet et al., 2003; Huayan et al., 2011). To overcome this challenge, an air gap can be
imposed between the membrane and a condensing surface in the permeate channel (Figure
2-d). This MD configuration is known as the air gap MD (AGMD) (Khayet and Cojocaru,

105
Experimental Methods for Membrane Applications

2012). This module design for MD can considerably enhance energy efficiency without
needing an external condenser (Kalla et al., 2019). However, the stagnant air gap can impose
extra mass transfer resistance for transferring the water vapor molecules from the interface
of the membrane support in the permeate channel towards the condensing surface (Shahu
and Thombre, 2019a). Table 5 presents the comparison among the main configurations of
the MD process.

5.2.3 Process

5.2.3.1 MD system
Figure 3 illustrates a general scheme of the MD system. The system consists of various parts
which are introduced briefly.

Thermocouple
T T

Cooler

pH T Heater
Peristaltic meter EC meter
pump pH
meter EC meter

Permeate Peristaltic
pump

Balance
Computer Computer
Feed tank

Figure 3 A general scheme of the MD experimental system.

The MD system consists of at least two containers, one to store the feed solution (feed tank)
and another one to collect the product (permeate tank). The permeate tank should be placed
on a balance to record the variation in the permeate mass and calculate the permeate flux.

The mass transfer and permeate production take place inside the membrane module, which
can have different configurations, including plate and frame, tubular, hollow fiber, and spiral
wound (Francis et al., 2022). A pump is required to recirculate the feed stream in the close
loop of the feed tank-module-feed tank. Depending on the MD configuration, the cooling
flow in the permeate channel can be provided by another pump in DCMD and AGMD or
by a compressor or blower in SGMD. In the case of VMD, the permeate channel is under
vacuum pressure using a vacuum pump.

A heating system is required for heating up the feed stream to a desired temperature.
Likewise, a cooling system is required to keep the temperature of the cooling stream low
and constant. In the case of the SGMD and VMD configurations, an external condenser is

106
 Chapter 5

also required to condense and collect the permeate. To monitor and control temperature, at
least four thermal sensors should be placed as close as possible to the inlet and outlet points
of the MD module. Thermal sensors are shown by ‘T’ in Figure 3.

The quality of the product in the permeate tank can be monitored using an EC meter
and a pH meter. Likewise, the variation of the feed quality can be monitored in the feed
tank. Depending on the requirements of the research, other equipment, such as an in-line
microscope or an injection, can also be considered.

5.2.3.2 Operating parameters

MD is a non-isothermal separation process, and the feed temperature is a dominant operating
parameter in all MD configurations (Curcio and Drioli, 2005). The operating temperature
in the feed channel can vary from 40 to 80 ˚C, depending on the available energy source,
the thermal resistance of the membrane, and the system design (Ahmed et al., 2020).
Moreover, the temperature of the cooling stream in the permeate channel is also important
for DCMD, SGMD, and AGMD configurations. For VMD, however, the vacuum pressure
in the permeate channel is an important operating parameter further to feed temperature
(Mohammadi et al., 2015; Peydayesh et al., 2015).

Fluid flow rate (i.e., the recirculation rate) in both the feed and permeate channels is another
operating parameter. Different values were reported in the literature for the flow rate,
depending on the MD module and capacity of the system. One should be considered the
inlet pressure at high flow rate. When the fluid flow is provided by a peristaltic pump, the
pressure of the fluid flow in the feed and permeate channels is quite low, i.e., very close to the
atmospheric pressure. However, if other types of pumps, such as centrifugal or diaphragm
pumps, are used, a proper pressure reducer device/tool (e.g., the pressure regulator) should
be used before the MD module.

The feed solution in MD can contain various compounds, such as chemicals, alcohols in
dilute concentration, different salts (e.g., NaCl, MgCl, CaCO3, Na2CO3, Na2SO4, etc.),
sugars (lactose, sucrose, glucose, etc.), etc. (Shirazi & Kargari, 2019; Ali et al., 2021; Ali et
al., 2015, 2018; Park et al., 2020; Peydayesh et al., 2015; Quist-Jensen et al., 2016a; Quist-
Jensen et al., 2017; Quist-Jensen et al., 2016; Shirazi et al., 2014). Thus, the feed type and its
concentration are other operating parameters, which can affect the MD performance, then
they can also be considered.

Depending on the MD configuration, the vacuum pressure, sweeping gas flowrate, and the
distance of the air gap in VMD, SGMD, and AGMD configurations, respectively, can also
be investigated as other operating parameters. Moreover, membrane properties (i.e., type,
material, pore size, etc.) have also been investigated in various research (Abu-Zeid et al.,
2015; Chamani et al., 2021; Curcio & Drioli, 2005; Eykens et al., 2017; Hitsov et al., 2015;
Shahu & Thombre, 2019a; Shalaby et al., 2022a, 2022b).

5.2.4 MD modules
A typical membrane module is a unit consisting of a membrane mounted in a housing and
containing feed inlet, retentate outlet and permeate outlet channels (Yang et al., 2013). As
new membrane applications emerge and new module designs are developed, the definition

107
Experimental Methods for Membrane Applications

of modules evolves as well. For example, in submerged membrane modules, as the name
suggests, the module is directly immersed in the feed solution without an outer casing and
has only ports for permeate removal. Other examples are air-gap and permeate-gap MD
modules (AGMD and PGMD, respectively), which require additional channels at the inlet
and outlet of the cooling flow.

The main purpose of this module is to properly secure the membrane so that it can be used
in its designated application. However, a well-designed module must also meet several
other requirements. A proper module design should ensure a high packing density of the
membrane. Packing density is taken as the size of the surface area of a functional membrane
in a given volume. Generally, high packing density is desirable to avoid inefficient use of
module housing. However, it should be noted that for hollow fiber membranes, increasing
the packing density beyond a critical value can result in stagnant or ‘dead’ regions of poor heat
and mass transfer within the module. important. For plate and frame flat sheet membrane
modules, typical packing densities range from 100 to 400 m2/m3, whereas hollow fiber
membrane modules can have higher packing densities of up to 3000 m2/m3 (Peng et al.,
2012).

MD systems involve mass transport steps through the feed, membrane, and permeate, with
each region having a specific transfer coefficient. To mitigate mass transfer resistance at the
boundary layer, appropriate module designs should demonstrate good hydrodynamics,
minimize temperature and concentration polarizations, and minimize energy consumption.
Modules should maximize heat recovery, be easy and economical to fabricate, and minimize
leakage issues. They should also facilitate scale-up and integration into existing processes.
The module’s performance should be predictable using mathematical models under various
operating conditions and feed characteristics. The module must maintain integrity during
long-term operation, minimize foulant deposition, and be resistant to heat and chemical
degradation.

New variants of flat sheet and hollow fiber membrane modules have been introduced for
MD applications. Flat sheet membranes are typically assembled in plate and frame or spiral
wound configurations, while hollow fiber modules can be classified into shell and fiber/tube
and submerged configurations. These new variants aim to improve process performance by
improving heat and mass transport, heat recovery, and membrane area.

The simplest module design for MD experiments is the plate and frame module. In this
design, the membrane is placed between two frames and plates. The membranes for this
type of module are in flat sheet form. This module can possess different sizes and is useful
for lab-scale experiments. However, the membrane area is small, and this module does not
have that much chance for industrial applications. Various flow arrangements can also be
considered for this module, including the co-current, counter-current, and crossflow. The
efficiency of these flow arrangements in terms of the permeate flux in the plate and frame
modules is in the order of: crossflow>counter-current>co-current (Shirazi et al., 2014).

108
 Chapter 5

(A)
Cold stream

Hot stream

(B)
Cold stream

Hot stream

(C)

Hot stream
Cold stream

Figure 4 The flow arrangement of hot and cold streams in the plate and frame module; (a) co-current,
(b) counter-current, and (c) cross-current flows (Shirazi et al., 2014).

Vacuum multi-effect MD (VMEMD) systems operate under reduced pressure and can
achieve higher water recovery rates compared to AGMD. These systems use multiple stages
of MD in series, working at lower operating temperatures and pressures. They are compact,
high-efficiency systems with solar thermal collectors and solar-photovoltaic sources as heat
sources. VMEMD designs enable internal heating and condensation, saving heat energy
(Zhao et al., 2013).

The air gap width is a crucial parameter in AGMD module design, determining distillate
production rate. It aims to prevent condensing media from contacting the membrane surface,
reducing heat loss, and increasing vapor transport distance. The lower limit is determined
by thermal efficiency and water bridging. Various membrane configurations, including flat
sheet, tubular, hollow fiber, and spiral wound, have been applied in AGMD studies.

The fundamental module design of the AGMD has undergone numerous modifications,
including the introduction of spacers in the feed channel and the use of cooling plates on the
coolant channel. These modifications have improved the efficiency of heat removal from
the coolant and increased system flux, with the flat and channelled plates being effective
in increasing system flux. Multi-effect AGMD (ME-AGMD) is another novel approach to
improving module performance and industrialization of MD. Pangarkar and Deshmukh
developed a new ME-AGMD module for water treatment applications, which performed
better in terms of permeate flow and energy utilization (Pangarkar and Deshmukh, 2015).
The parallel stage MS-AGMD system generated 2.6 and 3 times more permeate volume
than a single-stage system, but its precise energy usage was only 1.5 times that of the
single-stage system. Operational modifications of the conventional AGMD module have
reduced pressure inside the gap below atmospheric pressure, increasing distillate flux up to
3 times. The generation of vacuum in the gap requires an additional vacuum pump and extra

109
Experimental Methods for Membrane Applications

electric energy input. The traditional AGMD module has been modified to an air-cooled
AGMD, which has the potential to significantly reduce energy consumption and costs for
desalination by minimizing or eliminating plant components associated with coolant flow
systems.

The spiral-wound module is a variation of the plate-and-frame configuration, where

a membrane envelope with a spacer is wound around a porous tube. It was first used
in pressure-driven processes like RO and UF and has since been used in research. The
Fraunhofer Institute for Solar Energy Systems has produced single and multi-channel spiral
wound membrane modules for MD. Fabrication involves rolling a membrane, condenser
foil, and spacer materials around a motorized main spindle, and sealing the channels with
resin. Multichannel spiral wound modules are built using multiple pairs of evaporator/
condenser flow channels. The main advantage of spiral wound modules is better counter-
current circulation of hot and cold streams, allowing for more efficient heat recovery
(Winter et al., 2012; Winter et al., 2011).

Traditional flat sheet membranes have been used for AGMD modules, but hollow fiber
has gained interest. Two strategies have been proposed for designing hollow fiber AGMD
modules: using the inside surface of the module to condense vapor passing through the
membrane or condensing the condensing surface inside the shell (Abu-Zeid & ElMasry,
2020; Alpatova et al., 2019; Shahu & Thombre, 2019b). Different variants of the second
approach have been proposed, including porous and dense hollow fibers packed inside
a module, water gap MD (WGMD), and multiple copper tubes enclosed in the shell of a
hollow fiber AGMD module. These approaches have shown promising results in improving
heat transfer efficiency and reducing cooling channel replacements.

The second conventional module design for MD is the hollow fiber module, which can have
hundreds of hollow fibers in a shell tube. Thus, the hollow fiber module can provide a much
larger surface area for MD experiments in comparison with the plate and frame modules. The
feed stream and the permeate stream can flow through either the fibers or the shell. In terms
of the high salinity brine, for example, it would be better to introduce the feed stream to
the shell instead of the fibers, as the membrane cleaning would be easier and more efficient
(Quist-Jensen et al., 2017). Same as the plate and frame modules, all MD configurations
can be performed using the hollow fiber modules. However, in the case of AGMD, special
design considerations may be necessary.

A hollow fiber module is a system consisting of a hollow fiber membrane bundle, cartridge,
tube sheets, and side caps. The bundle consists of hollow fiber membranes packed together,
with a liquid potting substance forming the tube sheet. The tube sheet acts as a fluid-tight
barrier, separating streams flowing through the lumen and shell sides of the module. The
packing density of the hollow fiber module is crucial for its productivity, as it directly affects
the module’s productivity(Mat et al., 2014). The packing density can be arranged in various
configurations, such as parallel, crisscross, or other precise geometric arrangements.

In DCMD, hollow fiber modules are typically in shell and tube heat exchanger configuration,
with feed flow on one side and permeate on the other. The feed compartment is based on
the feed solution properties. Axial flow can be divided into co-current and counter-current

110
 Chapter 5

flows, with the counter-current flow arrangement being the most widely used configuration
for MD applications. Cross flow is often used in axial flow designs to reduce stagnant regions
and concentration polarization effects.

Traditional parallel hollow fiber modules are susceptible to high temperature and
concentration polarizations, especially at low fluid flow rates. The non-uniformity of fiber
packing is a challenge due to the production of parallel fiber bundles. This results in sluggish
or dead zones, reduced separation efficiency, and channelling or bypassing in poorly packed
zones. To improve uniformity, fibers can be weaved into different structural geometries,
such as helical, wavy, or twisted shapes (Ali et al., 2015b; Shahu & Thombre, 2021).
This results in more uniform shell flow and less concentration polarization due to fluid
mixing. Studies have shown that using these geometries can increase flux enhancement
in membrane applications. Yang et al., (2012) compared five types of hollow fiber module
designs, revealing that the space-knitted fiber design showed the best performance, with
over 90% increase in permeate flux. This configuration improved fluid dynamics and even
flow distribution, increased vapor permeability, and reduced thermal polarization with
lower energy loss.

Submerged hollow fiber MD modules are increasingly popular in membrane bioreactors

(MBRs) due to their simplicity and ease of fabrication (Francis et al., 2015; Meng et al.,
2015; Gryta, 2020). These modules eliminate the need for feed stream circulation, reducing
electric energy consumption. They have been proposed for VMD and DCMD configurations,
allowing for recirculation of either feed or permeate stream while submerged in the other
stream. The MDBR system, combining MD with a thermophilic MBR, produces high-
quality permeate with a flux two orders higher than competitors. Submerged MD hollow
fiber modules have also been proposed for desalination of Red Sea water, using PTFE-based
hollow fibers immersed in clean water and hot feed introduced inside.

However, some inherent issues have been identified, such as high temperature and
concentration polarizations, fouling, and scaling at the membrane surface. Strategies such
as mixing feed solution with a magnetic stirrer, transmembrane vibrations, and low-power
ultrasound have been proposed to improve the efficiency of these modules.

Module design in membrane-based membranes (MD) has been influenced by overall

contact length, which refers to the membrane length over which hot feed streams stay in
contact without intermittent heating. High contact lengths can be achieved by increasing
membrane length, connecting multiple modules in series, or increasing membrane length
in each envelope of flat sheet membrane. Recovery of latent heat of condensation from
permeate can reduce the thermal energy consumption of MD, but in DCMD, a sufficient
length is required to keep the feed and permeate in contact. Recent studies have shown
that increasing flow channel length can reduce gain-to-output ratios, specific thermal
energy consumption, and channel depth (Abu-Zeid et al., 2021; Ali et al., 2016b). (Tsai
et al., 2023) proposed multipass hollow fiber membrane modules, which studied the
effect of operational modes, number of passes, length, and operating temperature on the
performance of the multipass modules. The results may open new windows for future MD
modules for better and more energy-efficient performance, particularly for desalination and
brine management purposes.

111
Experimental Methods for Membrane Applications

5.3 METHODS

5.3.1 Process measurements and calculations

5.3.1.1 Permeate flux

The permeate flux in MD can be measured as follow:

m
J= Eq. 1
A× t
where Δm, A, and t are the collected permeate mass (kg), membrane surface (m2), and the
interval time (sec), respectively (Drioli et al., 2013).

5.3.1.2 Solute rejection

The solute rejection in the permeate stream can be measured as follow:
Cp
Rejection(%) = (1 ) ×100 Eq. 2
Cf

where Cp and Cf are the solute concentrations in the permeate and feed sides (mg/L),
respectively. Cp can be calculated based on the following equation with reference to the
dilution effect:

C1m1 C0 m0
Cp = Eq. 3
m1 m0

where m0 and m1 are the initial and final masses of the cold stream. C0 and C1 are also the
initial and final salt concentrations of the cold stream, respectively (Lu et al., 2016).

5.3.1.3 Logarithmic temperature difference

The logarithmic temperature difference along the MD module is defined as follow (Quist-
Jensen et al., 2018):

(Tf i Tp0 ) (Tf 0 Tpi )

dTln = Eq. 4
(Tf i Tp0 )
ln
(Tf 0 Tpi )

where Tfi and Tfo are the inlet and outlet temperatures of the feed stream, and Tpi and Tpo
are the inlet and outlet temperatures of the permeate stream, in the membrane module,
respectively.

5.3.2 Membrane characterization

5.3.2.1 Physical and morphology properties

(a) Scanning electron microscopy
Membrane surface can affect the MD performance in terms of flux, fouling, etc. Thus, surface
morphology should be investigated, specifically for fabricated membranes (Talukder et al.,
2022; Wang et al., 2023). Scanning electron microscopy (SEM) is the most investigated

112
 Chapter 5

method for morphology observation of various materials, including membranes (Khulbe

and Matsuura, 2017). SEM has numerous advantages for determining the membrane
morphology, such as simplicity and ease of operation. Moreover, as SEM is a non-destructive
technique, the samples are not damaged during the analysis and can then be used many
times for further imaging (Naresh-Kumar et al., 2012). Figure 5 shows the example of SEM
images of a commercial PTFE and a fabricated membrane with nanofiber structure for MD.

Nanofiber membrane PTFE membrane

Figure 5 SEM images of a fabricated membrane with nanofiber structure (left) and a commercial
PTFE membrane (right) for MD (Shirazi et al., 2020a).

Before conducting an SEM test, sample preparation is required. To prepare the sample, a
small piece of membrane is cut, and placed on a stub. As the sample is in small size, tweezers
are usually used along with double-sided glue tape for fixing the sample on top of the
holder. To prevent charge accumulation, the sample should be sputter coated with a thin
layer of a highly conductive metal, such as gold or platinum (Sharma & Bhardwaj, 2019;
Vladár & Hodoroaba, 2020). Moreover, if cross-sectional images are required, the sample
should carefully be freeze-dried in liquid nitrogen, and then should properly be cut using a
razor blade (Zhu et al., 2012; Conners and Banerjee, 2020; Vladár and Hodoroaba, 2020). It
is worth noting that the sample should not be touched to avoid adding contamination and
footprint.

If more informative images are required, for example for morphology observation and
detection of nanoparticles on the membrane surface, field emission SEM (FESEM) with the
platinum coating for the sample are recommended, as it can provide images with higher
resolution (Lewczuk and Szyry ska, 2021; Kirk, 2017). The SEM or FESEM images can
be used for morphological observation, determination of pore size and its distribution,
thickness measurement (from the cross-sectional images), investigating the homogeneity,
and presence of particles or fouling layer on the membrane surface.

Also, SEM utilizes imaging software to measure the dimensions of, e.g., the size of particles,
on the surface at various magnification ranges. Moreover, various external software, such as
ImageJ which is an open-source software for image processing, can be used for measuring
the pore size, pore size distribution, and porosity (Guillen et al., 2010; Shirazi et al., 2013;
Ziel et al., 2008).

(b) Transmission electron microscopy

If clear view of the internal structure of the membrane sample is required, the transmission
electron microscopy (TEM) should be applied for imaging the membrane samples. TEM is
an alternative to SEM for investigating the structural morphology and crystalline (Sharma

113
Experimental Methods for Membrane Applications

et al., 2018; Shyam Kumar et al., 2017). For example, it can be used when nanoparticles are
incorporated into the membrane structure. Thus, the internal morphology and distribution
of nanoparticles can be investigated using TEM (Qin et al., 2015; Dadari et al., 2022; He et
al., 2023). Despite of SEM, which is more practical for surface observation, TEM can provide
accurate information about the structure and the body of the membrane sample (Mousa et
al., 2022; Talukder et al., 2022; Wang et al., 2023; Wiktor et al., 2017).

(c) Pore size and its distribution

Pore size is an important parameter for MD membranes, as only vapor molecules should
pass through the pores (Eykens et al., 2016). For example, although membranes with large
pore size are considered to provide higher permeate flux, they also suffer from high pore
wetting risk (McGaughey et al., 2020b). Membrane pore size can be measured using various
techniques (Nakao, 1994; Tanis-Kanbur et al., 2021; Tung et al., 2014).

As mentioned earlier, the pore size and its distribution can be measured using SEM images of
the membrane surface (Ahmed et al., 2015; Sadeghzadeh et al., 2020b; Shirazi et al., 2013a).
However, it should be noted that the obtained results are based on surface observation.
The membrane pore size can also be determined using the filtration test. To do this, fine
particles with a known particle size distribution can be filtered using the membrane sample.
The permeate sample should be tested for the particle content and their sizes. By comparing
particle size in the permeate with the original value of the particle size in the feed sample,
the pore size range can then be determined. The results of the particle filtration test can then
be compared with the obtained results based on the image processing of SEM images for
pore size measurement (Gopal et al., 2006; Sadeghzadeh et al., 2020).

More accurate data for pore size and pore size distribution of MD membranes can be provided
using the capillary flow porometry technique (Jena and Gupta, 2005a). In this technique, a
small piece of membrane sample should be placed in a holder and get wet using a proper
wetting liquid of known surface tension, such as Topor or Galwick (Jena and Gupta, 2010;
Kolb et al., 2018). Afterward, the different flows of inert gas should be used to displace the
liquid inside the pores on the membrane structure. Using this technique, pore size and the
pore size distribution can be obtained (AlMarzooqi et al., 2016; Jena and Gupta, 2005b; Li
et al., 2006).

(d) Wetting properties

(i) Water contact angle
In MD, the membrane pores must not get wet with the feed solution, and only vapor
molecules should be passed through the pores. Therefore, the wetting property of the
membrane surface is crucial for MD applications (Chamani et al., 2021). This can be
determined using the surface contact angle. According to the standard definition, the
membrane surface is hydrophobic if the water contact angle on the sample surface is greater
than 90o, while the contact angle lower than this represents the hydrophilicity of the
membrane surface (Ismail et al., 2022; Rezaei et al., 2018). Figure 6 illustrates this concept.
Figure 7 also shows a typical system for contact angle measurements.

114
 Chapter 5

θ > 90˚

θ < 90˚

Hydrophile Hydrophobe

Figure 6 A general scheme of the surface hydrophobicity.

Figure 7 A typical contact angle measurement system (www.kruss-scientific.com).

To measure the contact angle, a 5-μL liquid droplet (usually deionized water) is placed on
the membrane surface and a high-resolution speed camera takes the image of the shape of
the droplet. The sessile drop technique can then be used for calculating the water contact
angle and surface energy of the membrane sample (Franken et al., 1987b; Lu et al., 2019). To
have high accuracy in the reported results, it is recommended to conduct the contact angle
test at least for five different points on the membrane surface, and then report the average
value.

115
Experimental Methods for Membrane Applications

(ii) Liquid entry pressure

Liquid entry pressure (LEP) determines the minimum required pressure to penetrate the
feed solution to the pores. When the pressure of the feed stream is greater than LEP, the
liquid enters the pores and pore wetting happens (Eykens et al., 2016). Thus, the membranes
for MD should possess as high LEP value as possible (Sadeghzadeh et al., 2020b). LEP is
proportional to some parameters, including the surface hydrophobicity, surface tension of
the feed solution, pore structure, and the pore size. LEP can therefore be calculated as follow:
−2Bγ l cosθ
LEP = Eq. 5
rmax

where θ is the surface contact angle (can be measured using the water contact angle test), ϒl
is the surface tension of the feed solution, B stands for the pore geometry factor, and rrmax is
the maximum pore size (Silva et al., 2021; Rácz et al., 2014).

0.4 bar
6 5

3 4

1
Membrane
2

Figure 8 A general scheme of the typical setup for measuring the LEP value of MD membranes. The
system consists of (1) an inert gas cylinder, (2) a pressurized container, (3) a membrane
cell filled with water, (4) a flowmeter, (5) a digital manometer, and (6) a pressure
regulator (Essalhi and Khayet, 2013).

LEP can also be measured experimentally for the membrane samples. Figure 8 illustrates the
general scheme of an experimental LEP measurement set-up. To do this, a dry membrane
sample should be placed in a plate and frame module (for flat sheet membrane samples),
and distilled water should be exposed on the hydrophobic surface of the membrane. The
pressure of the module should then be increased stepwise (10 kPa per minute would be
recommended followed by a few seconds time laps) using an inert gas (e.g., nitrogen). As
soon as the first water droplet is observed on the other side of the membrane sample, the
corresponding pressure represents the LEP value (Khayet and Matsuura, 2001; Rácz et al.,
2014). It is also recommended to evaluate the LEP using the feed solution further to the
distilled water, as the membrane is in direct contact with the feed solution rather than the
distilled water in MD experiments (Silva et al., 2021).

116
 Chapter 5

(e) Porosity
The membrane porosity can be measured using the gravimetric method. In this technique,
a small piece of membrane should be cut and then the dry weight should be recorded. The
sample should then be immersed in a proper wetting liquid (e.g., isopropanol alcohol) to get
completely wet and re-weighted again (Khayet and Matsuura, 2001). The porosity (ε) of the
membrane sample can then be calculated as follow:

w1 w2
( )
= 1 m Eq. 6
w1 w2 w2
( )+( )
m l

where w1 and w2 are the weights of the dry and wet samples, respectively. Moreover, rm and
rl are the density of the membrane sample and the density of the wetting liquid, respectively
(Alkhudhiri et al., 2012). It is worth noting that the weight measurement of the wetted
membrane sample should be carried out carefully.

As it was mentioned earlier, the membrane porosity can also be measured using the image
processing of SEM images. However, it should be noted that this will be the surface porosity
(Sadeghzadeh et al., 2020).

(f) Thickness
Membrane thickness can directly be measured using a precise micro calliper. It is
recommended to measure the thickness at least for 10 points and then report the average
value, to be sure to minimize the compression effect (Zhang et al., 2017).

Further to this, the membrane thickness can also be measured using the optical microscope
along with a proper scale bar (Vicente et al., 2013). More accurate thickness data, however,
can be provided by SEM through the cross-sectional imaging (Attia et al., 2018a; Attia et al.,
2018b).

(g) Surface roughness and topography

Both the surface roughness and topography can affect the performance of MD membranes.
These parameters can be determined using the atomic force microscopy (Khayet et al.,
2004). To evaluate the membrane surface using atomic force microscopy (AFM), a sample
with defined dimensions should be cut, placed, and then fixed on the top of the holder using
the double-sided glue tape (Wyart et al., 2008).

AFM uses a nanometric prob to move along the membrane surface and collect the
topographical data using a laser diode and a detector. Imaging can be performed via three
different modes, i.e., contact mode, semi-contact mode, and non-contact mode. The
generated data should be analysed using a collector system, and then topography images
(with Angstrom resolution) can be made. It is worth quoting that the non-contact mode
can provide 3D topographic images with higher resolution (Johnson and Hilal, 2015;
Hilal et al., 2004). When performing the AFM analysis for a membrane sample, the type
of probe (e.g., silicon nitride), scanning environment (e.g., in air at ambient conditions),

117
Experimental Methods for Membrane Applications

the specifications of the cantilever and its tip (i.e., length, width, resonance frequency,
slope, etc.), the scanning speed (e.g., 5 μm/s at 1 Hz), the applied force (e.g., 0.15 nN), the
scan size, and the resolution (250 points per line) are important parameters (Shirazi et al.,
2013a). Figure 9 presents the 3D AFM images of three commercial MF membranes for MD
applications based on the non-contact mode imaging.

PP

Figure 9 3D AFM image of commercial MF membrane for MD applications (Shirazi et al., 2013b).

Table 6 Typical roughness parameters for evaluating the membrane surface topography (Shirazi
et al., 2013a)
Roughness parameter Expression
Average roughness (Ra)
1 n
Ra = ∑Z
n i−1 i

Root-mean-square roughness (Rq) n

1
Rq = Z i2
n i 1

Peak-to-valley height (Rz) Rz = Z max − Z min

Surface skewness (Rsk) n

1
Rsk = Z i3
nRq3 i 1

Surface kurtosis (Rku) n

1
Rku = Z i4
nRq4 i 1

Zi The height at point i

n Number of points in the image
Zmax and Zmin The highest and the lowest Z values

118
 Chapter 5

Table 6 introduces some AFM parameters which are useful for evaluating the membrane
topography. For example, the average roughness (Ra), which provides an overall view of the
surface roughness, is the most reported topography parameter for MD membranes. In other
words, the higher the Ra value, the rougher the membrane surface (Johnson & Hilal, 2015;
Shirazi et al., 2013b). The skewness factor (Rsk) represents the height distribution symmetry.
While the positive Rsk parameter shows the domination of peaks on the surface, the negative
Rsk values are associated with a porous surface. The kurtosis factor (Rku) corresponds to the
sharpness of the height distributions (Johnson and Hilal, 2015). Rku values greater than 3
represent a surface with sharper height distribution, while values lower than 3 indicate a flat
surface. Further detailed descriptions and applications of these parameters for characterizing
MD membranes can be found in the literature (Johnson & Hilal, 2015; Shirazi et al., 2013b;
Shirazi et al., 2013a, 2013b).

It is worth noting that the AFM parameters and their results are scale and mode dependent.
Therefore, AFM images that have been provided with the same scale and the same mode can
be compared together. Moreover, compared to SEM analysis, AFM is a non-vacuum analysis
technique, and the membrane sample is not coated. Therefore, the AFM results can be closer
to the real features of the membrane in real life (Shirazi et al., 2013b).

Mechanical properties
Although MD membranes do not require very strict mechanical properties compared to
membranes in pressure-driven processes, such as RO, a minimum mechanical strength is
still required for handling and modulation of MD membranes (Essalhi and Khayet, 2014;
Essalhi and Khayet, 2013). Tensile test can be employed to evaluate the mechanical strength
of MD membranes. Figure 10 illustrates the general scheme of a typical tensile measurement
system.

Force
measurement
Grips for Fixed head
holding
specimen
firmly

Test specimen

Fixed head Thickness 1/8”

Constant rate
of motion

Figure 10 A general scheme of the tensile test (Ismail et al., 2019).

119
Experimental Methods for Membrane Applications

To perform the tensile test, a piece of membrane sample should be cut according to ASTM
D882-10 and taped for both ends to secure the grip. The membrane sample should then
be placed and fixed in a grip between two jaws. To run the tensile test, some operating
parameters are important, such as the load cell (e.g., 10 N), and the cross-head speed of the
load cell (e.g., 5 or 10 mm/min). Based on the data recorded in the tensile machine, the stress-
strain curves can then be illustrated and investigated (Tijing et al., 2013; Wang et al., 2017).

5.3.2.2 Chemical properties

(a) Elemental composition

Energy-dispersive X-ray spectroscopy (EDS) is a well-known, analytical technique for
investigating the elemental composition of MD membranes. It usually performs along with
the SEM test. EDS can provide the elemental composition of the membrane surface after
modification (e.g., successful adsorption of nanomaterials on the membrane surface) or the
composition of the fouling layer on the membrane surface (Kamaz et al., 2019; Shirazi et
al., 2020a). It can also be used for investigating the uniform dispersion of nanoparticles on
surface or in the membrane matrix. To perform the EDS, when the surface of the membrane
sample is bombarded by SEM electron beams, the EDS detector can then detect the emitted
X-ray and analyse the elemental composition of the membrane surface (Yang et al., 2018).
X-ray photoelectron spectroscopy (XPS) is another practical technique for elemental
analysis of MD membranes (Khayet and Matsuura, 2003). XPS can provide useful and
detailed information such as the chemical formula and chemical composition of the
membrane surface. Thus, XPS can practically be employed to evaluate the efficiency of a
proposed surface modification technique (Suk et al., 2006; Wei et al., 2012; Zhao et al.,
2017).

(b) Functional groups

Fourier transform infrared (FTIR) spectroscopy is a practical, analytical technique to
investigate the functional groups on the membrane surface, either after surface modification
or deposition of foulants during the MD test (Korolkov et al., 2018). FTIR can detect
different bonding types as well as organic and inorganic functional groups in molecular level.
FTIR works based on the absorbance of the wavelength of the IR light of various functional
groups in a wide range (400 to 4000 cm-1). Thus, the detector can analyse the absorbed
wavelength and identify the target functional group, and its density, as well. When the FTIR
graph is plotted, the sharper peaks represent the higher amount of the specific functional
group. Likewise, the small peaks represent trace amount of the that group (Belfer et al.,
2000; Jhaveri & Murthy, 2016; Rahman et al., 2018). Moreover, same as SEM, FTIR is a
non-destructive technique, and thus samples can be used for further measurements.

(c ) Chemical structure
When a new membrane is fabricated or modified for investigating the MD performance, it
should also be evaluated for the chemical structure. This can be performed using the Raman
spectroscopy technique (Intrchom et al. 2018; Pouya et al., 2021). Raman technique can
also provide informative curves for determining molecular bonds, crystallinity, and the
orientation of polymeric chains. Moreover, Raman is a non-destructive analysis technique
(Bhadra et al., 2016; Dumée et al., 2011; Huang et al., 2020).

120
 Chapter 5

(d) Crystalline structure

Each material can be specified using special X-ray diffraction. This can then be used for
investigating the size of atoms, the crystal size, the length of chemical bonds, and the
layer spacing (Bunaciu et al., 2015). These have made X-ray diffraction (XRD) a powerful
method to determine the structural crystallinity of materials, such as inorganic substances
(e.g., metals, ceramics, etc.) (Kahle et al., 2002; Norby, 2006). Thus, this analytical technique
is practical for the characterization of inorganic and mixed matrix MD membranes, as
polymers usually have a low crystalline structure. Moreover, XRD can be applied for
investigating the inorganic scaling on the membrane surface and the produced crystals in
the MD crystallization process (Garofalo et al., 2016; Zuo and Chung, 2016; Gryta, 2011;
Mokhtar et al., 2014). Using XRD, the change in the chemical bonds and the crystallinity
of materials can be detected. When XRD data is plotted, the narrow and sharp peaks
indicate ordered material with large particle size, and vice a versa (Petkov, 2008; Flack and
Bernardinelli, 2008).

5.3.2.3 Thermal properties

(a) Thermal conductivity

Thermal properties of MD membranes are important for long term performance. This can
be highlighted in terms of heat loss through the membrane thermal conduction (Eykens et
al., 2016). The thermal conductivity of MD membranes can be calculated as follow:

km = (1− ε )ks + ε k g Eq. 7

where ε is the membrane porosity, and ks and kg are the thermal conductivity of polymer
and the trapped gas in the pores (e.g., air), respectively. As could be observed, the thermal
conduction is proportional to the membrane porosity. Thus, the higher the porosity, the
lower the membrane’s thermal conductivity. Therefore, membranes with higher porosity,
such nanofiber membranes, can perform better in MD experiments due to lower heat
loss (Alkhudhiri et al., 2012; Shirazi et al., 2020a). Moreover, the precise examination
of the membrane porosity along with accurate conductivity data for the used polymer
in membrane fabrication can provide better results for the thermal conduction of MD
membranes. However, in terms of mixed matrix MD membranes, other correlations should
be investigated (Eykens et al., 2016).

(b) Thermal stability

As the MD membrane works under a range of feed temperatures (40-80 ˚C), the thermal
stability matters in long-term performance (Saffarini et al., 2013; Susanto, 2011b; Gryta,
2005). The thermal stability of MD membranes can be investigated using the differential
scanning calorimetry (DSC) method (Ding et al., 2021; Prince et al., 2012). DSC provides
very useful structural information of the membrane sample at various temperatures (Khayet
et al., 2018; Essalhi et al., 2021; García-Fernández et al., 2015).

121
Experimental Methods for Membrane Applications

5.4 APPLICATIONS AND EXAMPLES

MD is traditionally used for desalination purposes as an alternative to RO or to overcome

the limited recoveries of RO and other thermal desalination techniques. Though the most
investigated application of MD has been desalination, the low fouling propensity of the
process, the ability to handle complex feed solutions, and the fact that separation occurs
through temperature-induced phase equilibria at specific locations have led to many other
interesting applications of MD being explored (Drioli et al., 2015a). The temperature
gradient-based nature of MD also opens new possibilities for use in vapor/gas separation
applications where the equilibrium composition contains more volatile components at
each temperature. As a result, the scope of the process has expanded beyond traditional
desalting applications. MD could also be operated in an osmotic configuration where the
mass transport through the membrane pores is driven by the osmotic pressure difference
across the membrane. This operational mode is interesting for temperature-sensitive
products such as pharmaceutical compounds, juices, dairy products, natural aromatic
compounds, and various chemical solutions. MD has also demonstrated the potential to
treat the solutions where the final effluent quality must meet strict criteria such as nuclear
waste, radioactive water, or water treatment for the semiconductor industry.

In the oil and gas sector, shale gas has been recognized as a game changer due to its abundant
availability in different parts of the world. However, the negative environmental impact of
shale gas exploration remains a major obstacle to large-scale adaptation. Water produced
along with the oil - so-called produced water - is a major contributor to the dangerous
environmental impacts of shale gas exploration. The produced water contains a very high
proportion of salts, various hydrocarbons, and production chemicals. Handling such
complex liquids using state-of-the-art processes is a real challenge. Additionally, the high
pressure and temperature of the water generated during the manufacturing process further
complicate the immediate handling. MD has been shown to be a potential candidate for
treating this water after certain physical processes that remove hydrocarbons from the
stream.

Traditional MBR also has the biggest pollution problem. MD as a standalone process or
integrated with other processes (such as FO) yielded very interesting results. Similarly,
the removal of heavy metals that act as trace contaminants is a challenge for existing RO
plants. For example, since boron exists as boric acid, under normal pH conditions it can
diffuse through RO membranes, so conventional RO fails to meet the required removal
criteria (0.5-2.5 ppm). Current alternative technologies are either expensive or not robust
to changing operating conditions. The application of MD successfully removed boron far
below the specified limit.

Another area of potential interest for MD is the recovery/removal of phosphorus from

agricultural, domestic, and industrial wastewater. The presence of phosphorus in soil
is essential for crop growth. Phosphorus influx, on the other hand, causes a condition
called eutrophication. Eutrophication is characterized by the excessive growth of algae
in the water, reducing oxygen levels and adversely affecting marine life. As phosphorus
reserves are limited, MD can be used as a stand-alone process or in combination with other

122
 Chapter 5

membrane-based processes to not only control the phosphorus content in wastewater,

but also to increase phosphorus enrichment. Phosphorus crystals can also be recovered
from large streams (Quist-Jensen et al., 2018b; Simoni et al., 2021). The same is true for
protein crystallization and crystallization of pharmaceutical compounds by membrane
crystallization.

5.5 OUTLOOK

MD has made great progress during the last two decades or so. It is expected that the process
will attract further research and commercial interest for sustainable desalination as well
as resource recovery from different liquid streams such as desalination and geothermal
brines and wastewaters. However, it should be highlighted that further road to progression
should not consider MD as the replacement of large-scale reverse osmosis units but rather
a complementary process to augment the deficiencies of RO. For instance, MD could be
used to concentrate the retentate of RO brine with the ambition to approach zero liquid
discharge in seawater and brackish water desalination. Due to its capability to run with
solar energy, MD can be a suitable candidate as a standalone desalination process in off-
grid water-scarce regions. Due to its capability to concentrate, the process could also be a
valuable tool to recover valuable dissolved components from different liquid streams. This
is evident from the current interest in the process of the concentration of lithium brines.
MD has also demonstrated the potential to produce electricity when operated in pressure
retarded mode. This could turn the process into the simultaneous producer of freshwater,
electricity, and raw materials (Rahman et al., 2023).

Realization of the full potential of the process, however, is associated with overcoming
some key challenges. On the membrane front, the development of membranes with long-
lasting hydrophobic and anti-scaling/fouling characteristics is desired. Further research and
development in material development and synthesis routes are needed to achieve this goal.
As MD is more feasible for the treatment of high-concentrated solutions, the membrane
scaling issues are expected to be more significant in MD than the conventional pressure-
driven membrane processes such as NF and RO. Therefore, the development of improved
techniques to overcome scaling issues is of paramount importance. In this context, the
development of anti-scaling membranes as well as appropriate pre-treatment strategies is
expected to offer an important contribution. MD is also gaining traction in food processing,
and treatment of oily water and organic-rich wastewater where the scaling issues will be
significantly higher than the conventionally investigated desalination applications. The
membrane and process development, therefore, should also consider the appropriate
strategies to tackle fouling issues when tackling such complex feed solutions. The presence
of natural organic matter like humic acid, carbohydrates, proteins, lipids, and other low
molecular weight species causes organic fouling in MD. Organic matter can adhere to the
membrane surface through hydrophobic interactions, chemical affinity, and electrostatic
forces. This adherence can cause reduced vapor permeability and can interfere with the
hydrophobic character of the membrane. However, currently, very little attention has
been devoted to developing strategies to tackle this type of fouling. Due to the low flux
and different separation mechanism than the conventional pressure-driven membrane
processes, particulate fouling has not received significant attention. However, the solid

123
Experimental Methods for Membrane Applications

particles inherently present in the feed or crystals precipitating from the feed can create
particulate fouling in MD as reported in the literature (Chimanlal et al., 2022). Therefore,
it becomes relevant to develop mitigation strategies for such fouling in MD. The possible
strategies include appropriate pre-treatments (e.g., filtration, chemical precipitation),
module designs, fouling-resistant membrane configurations and materials and optimized
process conditions. Biofouling, or biological fouling, is caused by the accumulation of
bacteria and living microorganisms on the surface of a membrane. It leads to the formation
of a biofilm, which can significantly reduce membrane performance and productivity.
Biofouling is less common in MD compared to other membrane processes, but it still
occurs, especially in MD bioreactors. Biofouling in MD inhibits the process through pore
wetting and pore blockage mechanisms, allowing particles to penetrate the permeate side
and causing distillate contamination. Factors such as feed flow rate, membrane properties,
microorganism properties, pH, and feed water source influence the attachment and growth
of microorganisms on the membrane surface. To control biofouling in MD, techniques
(appropriate pre-treatments, quorum quenching, membrane, and process design) developed
for other membrane processes could become of interest.

MD is also becoming increasingly relevant for the treatment of acidic wastewater (e.g., the
one originating from the battery recycling process). This will require the development of
membranes that are tolerable to exposure to the low-pH solutions.

Future efforts on the process front should focus on minimizing the electric as well as thermal
energy consumption of the process. This can be achieved by developing more energy-
efficient membranes, process configurations, and module designs. More rational integration
of different energy sources (solar, geothermal, and industrial) will also provide an important
contribution. In particular, the studies should focus more on the integration of the process
with geothermal heat, which is a more stable and broadly available source of energy.

124
 Chapter 5

5.6 REFERENCES

Abu-Zeid, M. A.E.R., X. Lu, and S. Zhang. 2021. Influence of Module Length on Water Desalination
Using Air Gap Membrane Distillation Process: An Experimental Comparative Study. Arabian
Journal for Science and Engineering 46, no. 8 (August 1): 7989–8008.
Abu-Zeid, M.A.E.-R., Y. Zhang, H. Dong, L. Zhang, H.-L. Chen, and L. Hou. 2015. A Comprehensive
Review of Vacuum Membrane Distillation Technique. Desalination 356 (January): 1–14.
Abu-Zeid, Mostafa Abd El Rady, and G. ElMasry. 2020. Experimental Evaluation of Two Consecutive
Air-Gap Membrane Distillation Modules with Heat Recovery. Water Science and Technology:
Water Supply 20, no. 5 (August 1): 1678–1691.
Adewole, J.K., H.M. al Maawali, T. Jafary, A. Firouzi, and H. Oladipo. 2022. A Review of Seawater
Desalination with Membrane Distillation: Material Development and Energy Requirements.
Water Supply 22, no. 12 (December 1): 8500–8526.
Afsari, M., H.K. Shon, and L.D. Tijing. 2021. Janus Membranes for Membrane Distillation: Recent
Advances and Challenges. Advances in Colloid and Interface Science 289 (March): 102362.
Ahmed, F.E., B.S. Lalia, and R. Hashaikeh. 2015. A Review on Electrospinning for Membrane
Fabrication: Challenges and Applications. Desalination 356 (January): 15–30.
Ahmed, F.E., B.S. Lalia, R. Hashaikeh, and N. Hilal. 2020. Alternative Heating Techniques in Membrane
Distillation: A Review. Desalination 496 (December 15): 114713.
Al-hadidi, A., Kemperman, A., Wessling, M. & van der Meer, W. (2008). The influence of membrane
properties on the silt density index. In: EDS (ed.) Membranes in drinking and industrial water.
Toulouse: EDS-INSA.
Ali, A., P. Aimar, and E. Drioli. 2015a. Effect of Module Design and Flow Patterns on Performance of
Membrane Distillation Process. CHEMICAL ENGINEERING JOURNAL 277: 368–377. http://
dx.doi.org/10.1016/j.cej.2015.04.108.
Ali, A., C.A. Quist-Jensen, F. Macedonio, and E. Drioli. 2016. Optimization of Module Length
for Continuous Direct Contact Membrane Distillation Process. Chemical Engineering and
Processing: Process Intensification 110 (December): 188–200.
Ali, Aamer, C. Quist-Jensen, F. Macedonio, and E. Drioli. 2015. Application of Membrane Crystallization
for Minerals’ Recovery from Produced Water. Membranes 5, no. 4 (November 24): 772–792.
Ali, Aamer, C.A. Quist-Jensen, E. Drioli, and F. Macedonio. 2018. Evaluation of Integrated
Microfiltration and Membrane Distillation/Crystallization Processes for Produced Water
Treatment. Desalination 434 (May): 161–168.
Ali, Aamer, C.A. Quist-Jensen, M.K. Jørgensen, A. Siekierka, M.L. Christensen, M. Bryjak, C. Hélix-
Nielsen, and E. Drioli. 2021. A Review of Membrane Crystallization, Forward Osmosis and
Membrane Capacitive Deionization for Liquid Mining. Resources, Conservation and Recycling
168 (May 1): 105273.
Ali, Aamer, C.A. Quist-Jensen, F. Macedonio, and E. Drioli. 2016. Optimization of Module Length and
Membrane Thickness for Membrane Distillation. In 2nd International Workshop on Membrane
Distillation and Innovating Membrane Operations in Desalination and Water Reuse, Ravello,
Italy. Ravello.
Ali, Aamer, J.-H. Tsai, K.-L. Tung, E. Drioli, and F. Macedonio. 2018. Designing and Optimization of
Continuous Direct Contact Membrane Distillation Process. Desalination 426 (January): 97–107.
Alkhudhiri, A., N. Darwish, and N. Hilal. 2012. Membrane Distillation: A Comprehensive Review.
Desalination 287 (February): 2–18.
Alklaibi, A.M., and N. Lior. 2005. Membrane-Distillation Desalination: Status and Potential.
Desalination 171, no. 2 (January 10): 111–131.

125
Experimental Methods for Membrane Applications

AlMarzooqi, F.A., M.R. Bilad, B. Mansoor, and H.A. Arafat. 2016. A Comparative Study of Image
Analysis and Porometry Techniques for Characterization of Porous Membranes. Journal of
Materials Science 51, no. 4 (February 27): 2017–2032.
Alpatova, A., A.S. Alsaadi, M. Alharthi, J.G. Lee, and N. Ghaffour. 2019. Co-Axial Hollow Fiber Module
for Air Gap Membrane Distillation. Journal of Membrane Science 578 (May 15): 172–182.
An, A.K., J. Guo, E.-J. Lee, S. Jeong, Y. Zhao, Z. Wang, and T. Leiknes. 2017. PDMS/PVDF Hybrid
Electrospun Membrane with Superhydrophobic Property and Drop Impact Dynamics for
Dyeing Wastewater Treatment Using Membrane Distillation. Journal of Membrane Science 525
(March): 57–67.
Anderson, D. M., Boerlage, S. F. E. & Dixon, M. B. (2017). Harmful Algal Blooms (HABs) and
Desalination: A Guide to Impacts, Monitoring and Management. In: UNESCO, I. O. C. o. (ed.)
IOC Manuals and Guides No. 78. Paris: UNESCO.
Ashoor, B.B., S. Mansour, A. Giwa, V. Dufour, and S.W. Hasan. 2016. Principles and Applications of
Direct Contact Membrane Distillation (DCMD): A Comprehensive Review. Desalination 398
(November): 222–246.
ASTM D4189 - 14 (2014). Standard Test Method for Silt Density Index (SDI) of Water. West
Conshohocken, PA: ASTM International.
Attia, H., D.J. Johnson, C.J. Wright, and N. Hilal. 2018a. Comparison between Dual-Layer
(Superhydrophobic–Hydrophobic) and Single Superhydrophobic Layer Electrospun Membranes
for Heavy Metal Recovery by Air-Gap Membrane Distillation. Desalination 439 (August): 31–45.
———. 2018b. Robust Superhydrophobic Electrospun Membrane Fabricated by Combination
of Electrospinning and Electrospraying Techniques for Air Gap Membrane Distillation.
Desalination 446 (November): 70–82.
Belfer, S., R. Fainchtain, Y. Purinson, and O. Kedem. 2000. Surface Characterization by FTIR-ATR
Spectroscopy of Polyethersulfone Membranes-Unmodified, Modified and Protein Fouled.
Journal of Membrane Science 172, no. 1–2 (July): 113–124.
Belfort, G., Davis, R. H. & Zydney, A. L. (1994). The behavior of suspensions and macromolecular
solutions in crossflow microfiltration. Journal of Membrane Science, 96, 1-58
Bhadra, M., S. Roy, and S. Mitra. 2016. Desalination across a Graphene Oxide Membrane via Direct
Contact Membrane Distillation. Desalination 378 (January): 37–43.
Bonyadi, S., and T.S. Chung. 2007. Flux Enhancement in Membrane Distillation by Fabrication of Dual
Layer Hydrophilic–Hydrophobic Hollow Fiber Membranes. Journal of Membrane Science 306,
no. 1–2 (December): 134–146.
Bunaciu, A.A., E. gabriela Udri tioiu, and H.Y. Aboul-Enein. 2015. X-Ray Diffraction: Instrumentation
and Applications. Critical Reviews in Analytical Chemistry 45, no. 4 (October 2): 289–299.
Camacho, L., L. Dumée, J. Zhang, J. Li, M. Duke, J. Gomez, and S. Gray. 2013. Advances in Membrane
Distillation for Water Desalination and Purification Applications. Water 5, no. 1 (January 25):
94–196.
Chamani, H., J. Woloszyn, T. Matsuura, D. Rana, and C.Q. Lan. 2021. Pore Wetting in Membrane
Distillation: A Comprehensive Review. Progress in Materials Science 122 (October 1): 100843.
Chen, L., P. Xu, and H. Wang. 2020. Interplay of the Factors Affecting Water Flux and Salt Rejection
in Membrane Distillation: A State-of-the-Art Critical Review. Water 12, no. 10 (October 13):
2841.
Cheng, D., J. Zhang, N. Li, D. Ng, S.R. Gray, and Z. Xie. 2018. Antiwettability and Performance Stability
of a Composite Hydrophobic/Hydrophilic Dual-Layer Membrane in Wastewater Treatment
by Membrane Distillation. Industrial & Engineering Chemistry Research 57, no. 28 (July 18):
9313–9322.

126
 Chapter 5

Chimanlal, I., L.N. Nthunya, C. Quist-Jensen, and H. Richards. 2022. Membrane Distillation
Crystallization for Water and Mineral Recovery: The Occurrence of Fouling and Its Control
during Wastewater Treatment. Frontiers in Chemical Engineering 4 (November 29).
Conners, T.E., and S. Banerjee. 2020. Surface Analysis of Paper. CRC Press.
Curcio, E., and E. Drioli. 2005. Membrane Distillation and Related Operations—A Review. Separation
& Purification Reviews 34, no. 1 (January): 35–86.
Dadari, S., M. Rahimi, and S. Zinadini. 2022. Novel Antibacterial and Antifouling PES Nanofiltration
Membrane Incorporated with Green Synthesized Nickel-Bentonite Nanoparticles for Heavy
Metal Ions Removal. Chemical Engineering Journal 431 (March): 134116.
Ding, Z., Z. Liu, and C. Xiao. 2021. Excellent Performance of Novel Superhydrophobic Composite
Hollow Membrane in the Vacuum Membrane Distillation. Separation and Purification
Technology 268 (August): 118603.
Drioli, E., A. Ali, S. Simone, F. Macedonio, S.A. AL-Jlil, F.S. al Shabonah, H.S. Al-Romaih, O. Al-Harbi,
A. Figoli, and A. Criscuoli. 2013. Novel PVDF Hollow Fiber Membranes for Vacuum and Direct
Contact Membrane Distillation Applications. Separation and Purification Technology 115
(August): 27–38.
Drioli, Enrico, A. Ali, and F. Macedonio. 2015. Membrane Distillation : Recent Developments and
Perspectives. Desalination 356: 56–84. http://dx.doi.org/10.1016/j.desal.2014.10.028.
Dumée, L., V. Germain, K. Sears, J. Schütz, N. Finn, M. Duke, S. Cerneaux, D. Cornu, and S. Gray. 2011.
Enhanced Durability and Hydrophobicity of Carbon Nanotube Bucky Paper Membranes in
Membrane Distillation. Journal of Membrane Science 376, no. 1–2 (July): 241–246.
Duong, H.C., D. Chuai, Y.C. Woo, H.K. Shon, L.D. Nghiem, and V. Sencadas. 2018. A Novel Electrospun,
Hydrophobic, and Elastomeric Styrene-Butadiene-Styrene Membrane for Membrane Distillation
Applications. Journal of Membrane Science 549 (March): 420–427.
El-Bourawi, M.S., Z. Ding, R. Ma, and M. Khayet. 2006. A Framework for Better Understanding
Membrane Distillation Separation Process. Journal of Membrane Science 285, no. 1–2
(November): 4–29.
Essalhi, M., and M. Khayet. 2013. Self-Sustained Webs of Polyvinylidene Fluoride Electrospun
Nanofibers at Different Electrospinning Times: 1. Desalination by Direct Contact Membrane
Distillation. Journal of Membrane Science 433 (April): 167–179.
———. 2014. Self-Sustained Webs of Polyvinylidene Fluoride Electrospun Nano-Fibers: Effects of
Polymer Concentration and Desalination by Direct Contact Membrane Distillation. Journal of
Membrane Science 454 (March): 133–143.
Essalhi, Mohamed, M. Khayet, N. Ismail, O. Sundman, and N. Tavajohi. 2021. Improvement of
Nanostructured Electrospun Membranes for Desalination by Membrane Distillation Technology.
Desalination 510 (August): 115086.
Eykens, L., K. de Sitter, C. Dotremont, L. Pinoy, and B. van der Bruggen. 2017. Membrane Synthesis for
Membrane Distillation: A Review. Separation and Purification Technology 182 (July): 36–51.
Eykens, Lies, K. de Sitter, C. Dotremont, L. Pinoy, and B. van der Bruggen. 2016. How To Optimize the
Membrane Properties for Membrane Distillation: A Review. Industrial & Engineering Chemistry
Research 55, no. 35 (September 7): 9333–9343.
Fane, A.G., R.W. Schofield, and C.J.D. Fell. 1987. The Efficient Use of Energy in Membrane Distillation.
Desalination 64 (January): 231–243.
Feng, S., Z. Zhong, Y. Wang, W. Xing, and E. Drioli. 2018. Progress and Perspectives in PTFE Membrane:
Preparation, Modification, and Applications. Journal of Membrane Science 549 (March): 332–
349.

127
Experimental Methods for Membrane Applications

Ferreira, R.K.M., H. Ramlow, C. Marangoni, and R.A.F. Machado. 2021. A Review on the Manufacturing
Techniques of Porous Hydrophobic Ceramic Membranes Applied to Direct Contact Membrane
Distillation. Advances in Applied Ceramics 120, no. 5–8 (November 17): 336–357.
Flack, H.D., and G. Bernardinelli. 2008. The Use of X-Ray Crystallography to Determine Absolute
Configuration. Chirality 20, no. 5 (May 15): 681–690.
Francis, L., F.E. Ahmed, and N. Hilal. 2022. Advances in Membrane Distillation Module Configurations.
Membranes 12, no. 1 (January 12): 81.
Francis, L., N. Ghaffour, A.S. Al-Saadi, and G.L. Amy. 2015. Submerged Membrane Distillation for
Seawater Desalination. Desalination and Water Treatment 55, no. 10 (September 4): 2741–
2746.
Franken, A.C.M., J.A.M. Nolten, M.H.V. Mulder, D. Bargeman, and C.A. Smolders. 1987. Wetting
Criteria for the Applicability of Membrane Distillation. Journal of Membrane Science 33, no. 3
(October): 315–328.
García-Fernández, L., M. Khayet, and M.C. García-Payo. 2015. Membranes Used in Membrane
Distillation: Preparation and Characterization. In Pervaporation, Vapour Permeation and
Membrane Distillation, 317–359. Elsevier.
Garofalo, A., M.C. Carnevale, L. Donato, E. Drioli, O. Alharbi, S.A. Aljlil, A. Criscuoli, and C. Algieri.
2016. Scale-up of MFI Zeolite Membranes for Desalination by Vacuum Membrane Distillation.
Desalination 397 (November): 205–212.
Gontarek-Castro, E., R. Castro-Muñoz, and M. Lieder. 2022. New Insights of Nanomaterials Usage
toward Superhydrophobic Membranes for Water Desalination via Membrane Distillation: A
Review. Critical Reviews in Environmental Science and Technology 52, no. 12 (June 18): 2104–
2149.
Gopal, R., S. Kaur, Z. Ma, C. Chan, S. Ramakrishna, and T. Matsuura. 2006. Electrospun Nanofibrous
Filtration Membrane. Journal of Membrane Science 281, no. 1–2 (September 15): 581–586.
Gryta, M. 2005. Long-Term Performance of Membrane Distillation Process. Journal of Membrane
Science 265, no. 1–2 (November 15): 153–159.
———. 2007. Influence of Polypropylene Membrane Surface Porosity on the Performance of Membrane
Distillation Process. Journal of Membrane Science 287, no. 1 (January 5): 67–78.
Gryta, M. 2011. The Influence of Magnetic Water Treatment on CaCO3 Scale Formation in Membrane
Distillation Process. Separation and Purification Technology 80, no. 2 (July): 293–299.
———. 2020. The Application of Submerged Modules for Membrane Distillation. Membranes 10, no.
2 (February 1).
Guillen, G.R., T.P. Farrell, R.B. Kaner, and E.M. v. Hoek. 2010. Pore-Structure, Hydrophilicity, and
Particle Filtration Characteristics of Polyaniline–Polysulfone Ultrafiltration Membranes. Journal
of Materials Chemistry 20, no. 22: 4621.
He, K., H.J. Hwang, M.W. Woo, and I.S. Moon. 2011. Production of Drinking Water from Saline
Water by Direct Contact Membrane Distillation (DCMD). Journal of Industrial and Engineering
Chemistry 17, no. 1 (January): 41–48.
He, Y., J. Wang, X. Fu, H. Lin, W. Zhang, X. Wang, and F. Liu. 2023. Electrosprayed Thin Film
Nanocomposite Polyamide Nanofiltration with Homogeneous Distribution of Nanoparticles
for Enhanced Separation Performance. Desalination 546 (January): 116206.
Hilal, N., H. Al-Zoubi, N.A. Darwish, A.W. Mohamma, and M. Abu Arabi. 2004. A Comprehensive
Review of Nanofiltration Membranes:Treatment, Pretreatment, Modelling, and Atomic Force
Microscopy. Desalination 170, no. 3 (November): 281–308.

128
 Chapter 5

Hitsov, I., T. Maere, K. de Sitter, C. Dotremont, and I. Nopens. 2015. Modelling Approaches in
Membrane Distillation: A Critical Review. Separation and Purification Technology 142 (March
4): 48–64.
Huang, J., Y. Hu, Y. Bai, Y. He, and J. Zhu. 2020. Novel Solar Membrane Distillation Enabled by a
PDMS/CNT/PVDF Membrane with Localized Heating. Desalination 489 (September): 114529.
Huayan, C., W. Chunrui, J. Yue, W. Xuan, and L. Xiaolong. 2011. Comparison of Three Membrane
Distillation Confi Gurations and Seawater Desalination by Vacuum Membrane Distillation.
Desalination and Water Treatment 28, no. 1–3 (April 3): 321–327.
Hussain, A., A. Janson, J.M. Matar, and S. Adham. 2022. Membrane Distillation: Recent Technological
Developments and Advancements in Membrane Materials. Emergent Materials 5, no. 2 (April
5): 347–367.
Intrchom, W., S. Roy, M. Humoud, and S. Mitra. 2018. Immobilization of Graphene Oxide on the
Permeate Side of a Membrane Distillation Membrane to Enhance Flux. Membranes 8, no. 3
(August 15): 63.
Ismail, A.F., K.C. Khulbe, and T. Matsuura. 2019. RO Membrane Characterization. In Reverse Osmosis,
57–90. Elsevier.
Ismail, M.F., M.A. Islam, B. Khorshidi, A. Tehrani-Bagha, and M. Sadrzadeh. 2022. Surface
Characterization of Thin-Film Composite Membranes Using Contact Angle Technique: Review
of Quantification Strategies and Applications. Advances in Colloid and Interface Science 299
(January): 102524.
Jantaporn, W., A. Ali, and P. Aimar. 2017. Specific Energy Requirement of Direct Contact Membrane
Distillation. Chemical Engineering Research and Design 128 (December): 15–26.
Jena, A., and K. Gupta. 2010. Advances in Pore Structure Evaluation by Porometry. Chemical
Engineering & Technology 33, no. 8 (June 25): 1241–1250.
Jena, Akshaya, and K. Gupta. 2005. Pore Volume of Nanofiber Nonwovens. International Nonwovens.
Journal os-14, no. 2 (June 10): 1558925005os–14.
Jhaveri, J.H., and Z.V.P. Murthy. 2016. A Comprehensive Review on Anti-Fouling Nanocomposite
Membranes for Pressure Driven Membrane Separation Processes. Desalination 379 (February):
137–154.
Ji, J., F. Liu, N.A. Hashim, M.R.M. Abed, and K. Li. 2015. Poly(Vinylidene Fluoride) (PVDF) Membranes
for Fluid Separation. Reactive and Functional Polymers 86 (January): 134–153.
Johnson, D., and N. Hilal. 2015. Characterisation and Quantification of Membrane Surface Properties
Using Atomic Force Microscopy: A Comprehensive Review. Desalination 356 (January): 149–
164.
Julian, H., N. Nurgirisia, G. Qiu, Y.-P. Ting, and I.G. Wenten. 2022. Membrane Distillation for
Wastewater Treatment: Current Trends, Challenges and Prospects of Dense Membrane
Distillation. Journal of Water Process Engineering 46 (April): 102615.
Kahle, M., M. Kleber, and R. Jahn. 2002. Review of XRD-Based Quantitative Analyses of Clay Minerals
in Soils: The Suitability of Mineral Intensity Factors. Geoderma 109, no. 3–4 (October): 191–
205.
Kalla, S., S. Upadhyaya, and K. Singh. 2019. Principles and Advancements of Air Gap Membrane
Distillation. Reviews in Chemical Engineering 35, no. 7 (October 25): 817–859.
Kamaz, M., A. Sengupta, A. Gutierrez, Y.-H. Chiao, and R. Wickramasinghe. 2019. Surface Modification
of PVDF Membranes for Treating Produced Waters by Direct Contact Membrane Distillation.
International Journal of Environmental Research and Public Health 16, no. 5 (February 26): 685.
Kang, G., and Y. Cao. 2014. Application and Modification of Poly(Vinylidene Fluoride) (PVDF)
Membranes – A Review. Journal of Membrane Science 463 (August): 145–165.

129
Experimental Methods for Membrane Applications

Khayet, M., and C. Cojocaru. 2012. Air Gap Membrane Distillation: Desalination, Modeling and
Optimization. Desalination 287 (February): 138–145.
Khayet, M., M.C. García-Payo, L. García-Fernández, and J. Contreras-Martínez. 2018. Dual-Layered
Electrospun Nanofibrous Membranes for Membrane Distillation. Desalination 426 (January):
174–184.
Khayet, M., M.P. Godino, and J.I. Mengual. 2003. Possibility of Nuclear Desalination through Various
Membrane Distillation Configurations: A Comparative Study. International Journal of Nuclear
Desalination 1, no. 1: 30.
Khayet, M., K. Khulbe, and T. Matsuura. 2004. Characterization of Membranes for Membrane
Distillation by Atomic Force Microscopy and Estimation of Their Water Vapor Transfer
Coefficients in Vacuum Membrane Distillation Process. Journal of Membrane Science 238, no.
1–2 (July 15): 199–211.
Khayet, M., and T. Matsuura. 2003. Application of Surface Modifying Macromolecules for the Preparation
of Membranes for Membrane Distillation. Desalination 158, no. 1–3 (August): 51–56.
Khayet, Mohamed. 2011. Membranes and Theoretical Modeling of Membrane Distillation: A Review.
Advances in Colloid and Interface Science 164, no. 1–2 (May): 56–88.
Khayet, Mohamed, and T. Matsuura. 2001. Preparation and Characterization of Polyvinylidene Fluoride
Membranes for Membrane Distillation. Industrial & Engineering Chemistry Research 40, no. 24
(November 1): 5710–5718.
Khulbe, K.C., and T. Matsuura. 2017. Synthetic Membrane Characterisation – a Review: Part I.
Membrane Technology 2017, no. 7 (July): 7–12.
Kim, W.-J., O. Campanella, and D.R. Heldman. 2021. Predicting the Performance of Direct Contact
Membrane Distillation (DCMD): Mathematical Determination of Appropriate Tortuosity Based
on Porosity. Journal of Food Engineering 294 (April): 110400.
Kimura, S., S.-I. Nakao, and S.-I. Shimatani. 1987. Transport Phenomena in Membrane Distillation.
Journal of Membrane Science 33, no. 3 (October): 285–298.
Kirk, T.L. 2017. A Review of Scanning Electron Microscopy in Near Field Emission Mode. In , 39–109.
Kolb, H.E., R. Schmitt, A. Dittler, and G. Kasper. 2018. On the Accuracy of Capillary Flow Porometry
for Fibrous Filter Media. Separation and Purification Technology 199 (June): 198–205.
Korolkov, I. v., Y.G. Gorin, A.B. Yeszhanov, A.L. Kozlovskiy, and M. v. Zdorovets. 2018. Preparation
of PET Track-Etched Membranes for Membrane Distillation by Photo-Induced Graft
Polymerization. Materials Chemistry and Physics 205 (February): 55–63.
Lawson, K.W., and D.R. Lloyd. 1997. Membrane Distillation. Journal of Membrane Science 124, no. 1
(February): 1–25.
Lewczuk, B., and N. Szyrynska. 2021. Field-Emission Scanning Electron Microscope as a Tool for Large-
Area and Large-Volume Ultrastructural Studies. Animals 11, no. 12 (November 27): 3390.
Li, D., M.W. Frey, and Y.L. Joo. 2006. Characterization of Nanofibrous Membranes with Capillary Flow
Porometry. Journal of Membrane Science 286, no. 1–2 (December): 104–114.
Lu, K.J., Y. Chen, and T.-S. Chung. 2019. Design of Omniphobic Interfaces for Membrane Distillation –
A Review. Water Research 162 (October): 64–77.
Lu, K.-J., J. Zuo, and T.-S. Chung. 2016. Tri-Bore PVDF Hollow Fibers with a Super-Hydrophobic
Coating for Membrane Distillation. Journal of Membrane Science 514 (September): 165–175.
Mat, N.C., Y. Lou, and G.G. Lipscomb. 2014. Hollow Fiber Membrane Modules. Current Opinion in
Chemical Engineering 4: 18–24.
McGaughey, A.L., P. Karandikar, M. Gupta, and A.E. Childress. 2020. Hydrophobicity versus Pore Size:
Polymer Coatings to Improve Membrane Wetting Resistance for Membrane Distillation. ACS
Applied Polymer Materials 2, no. 3 (March 13): 1256–1267.

130
 Chapter 5

Meng, S., Y.C. Hsu, Y. Ye, and V. Chen. 2015. Submerged Membrane Distillation for Inland Desalination
Applications. Desalination 361 (April 1): 72–80.
Mohammadi, T., P. Kazemi, and M. Peydayesh. 2015. Optimization of Vacuum Membrane Distillation
Parameters for Water Desalination Using Box–Behnken Design. Desalination and Water
Treatment 56, no. 9 (November 27): 2306–2315.
Mokhtar, N.M., W.J. Lau, A.F. Ismail, and B.C. Ng. 2014. Physicochemical Study of Polyvinylidene
Fluoride–Cloisite15A® Composite Membranes for Membrane Distillation Application. RSC
Adv. 4, no. 108: 63367–63379.
Mousa, H.M., M. Hamdy, M.A. Yassin, M.M. El-Sayed Seleman, and G.T. Abdel-Jaber. 2022.
Characterization of Nanofiber Composite Membrane for High Water Flux and Antibacterial
Properties. Colloids and Surfaces A: Physicochemical and Engineering Aspects 651 (October):
129655.
Nakao, S. 1994. Determination of Pore Size and Pore Size Distribution. Journal of Membrane Science
96, no. 1–2 (November): 131–165.
Naresh-Kumar, G., B. Hourahine, P.R. Edwards, A.P. Day, A. Winkelmann, A.J. Wilkinson, P.J.
Parbrook, G. England, and C. Trager-Cowan. 2012. Rapid Nondestructive Analysis of Threading
Dislocations in Wurtzite Materials Using the Scanning Electron Microscope. Physical Review
Letters 108, no. 13 (March 30): 135503.
Niknejad, A. S., S. Bazgir, M. Ardjmand, and M.M.A. Shirazi. 2021. Spent Caustic Wastewater Treatment
Using Direct Contact Membrane Distillation with Electroblown Styrene-Acrylonitrile
Membrane. International Journal of Environmental Science and Technology 18, no. 8 (August
14): 2283–2294.
Niknejad, Ali Sallakh, S. Bazgir, and A. Kargari. 2021. Desalination by Direct Contact Membrane
Distillation Using a Superhydrophobic Nanofibrous Poly (Methyl Methacrylate) Membrane.
Desalination 511 (September): 115108.
Niknejad, Ali Sallakh, S. Bazgir, A. Sadeghzadeh, and M.M.A. Shirazi. 2021. Evaluation of a Novel and
Highly Hydrophobic Acrylonitrile-Butadiene-Styrene Membrane for Direct Contact Membrane
Distillation: Electroblowing/Air-Assisted Electrospraying Techniques. Desalination 500
(March): 114893.
Norby, P. 2006. In-Situ XRD as a Tool to Understanding Zeolite Crystallization. Current Opinion in
Colloid & Interface Science 11, no. 2–3 (June): 118–125.
Olatunji, S.O., and L.M. Camacho. 2018. Heat and Mass Transport in Modeling Membrane Distillation
Configurations: A Review. Frontiers in Energy Research 6 (December 4).
Omar, N.M.A., M.H.D. Othman, Z.S. Tai, T.A. Kurniawan, T. El-badawy, P.S. Goh, N.H. Othman, M.A.
Rahman, J. Jaafar, and A.F. Ismail. 2022. Bottlenecks and Recent Improvement Strategies of
Ceramic Membranes in Membrane Distillation Applications: A Review. Journal of the European
Ceramic Society 42, no. 13 (October): 5179–5194.
Pangarkar, B.L., and S.K. Deshmukh. 2015. Theoretical and Experimental Analysis of Multi-Effect
Air Gap Membrane Distillation Process (ME-AGMD). Journal of Environmental Chemical
Engineering 3, no. 3 (September 29): 2127–2135.
Park, S.H., J.H. Kim, S.J. Moon, J.T. Jung, H.H. Wang, A. Ali, C.A. Quist-Jensen, F. Macedonio, E. Drioli,
and Y.M. Lee. 2020. Lithium Recovery from Artificial Brine Using Energy-Efficient Membrane
Distillation and Nanofiltration. Journal of Membrane Science 598 (March): 117683.
Peng, N., N. Widjojo, P. Sukitpaneenit, M.M. Teoh, G.G. Lipscomb, T.-S. Chung, and J.-Y. Lai. 2012.
Evolution of Polymeric Hollow Fibers as Sustainable Technologies: Past, Present, and Future.
Progress in Polymer Science 37, no. 10 (October): 1401–1424.

131
Experimental Methods for Membrane Applications

Petkov, V. 2008. Nanostructure by High-Energy X-Ray Diffraction. Materials Today 11, no. 11
(November): 28–38.
Peydayesh, M., P. Kazemi, A. Bandegi, T. Mohammadi, and O. Bakhtiari. 2015. Treatment of Bentazon
Herbicide Solutions by Vacuum Membrane Distillation. Journal of Water Process Engineering 8
(December): e17–e22.
Phattaranawik, J., R. Jiraratananon, and A.G. Fane. 2003. Heat Transport and Membrane Distillation
Coefficients in Direct Contact Membrane Distillation. Journal of Membrane Science 212, no. 1–2
(February): 177–193.
Pouya, Z.A., M.A. Tofighy, and T. Mohammadi. 2021. Synthesis and Characterization of
Polytetrafluoroethylene/Oleic Acid-Functionalized Carbon Nanotubes Composite Membrane
for Desalination by Vacuum Membrane Distillation. Desalination 503 (May): 114931.
Prince, J.A., G. Singh, D. Rana, T. Matsuura, V. Anbharasi, and T.S. Shanmugasundaram. 2012.
Preparation and Characterization of Highly Hydrophobic Poly(Vinylidene Fluoride) – Clay
Nanocomposite Nanofiber Membranes (PVDF–Clay NNMs) for Desalination Using Direct
Contact Membrane Distillation. Journal of Membrane Science 397–398 (April): 80–86.
Qasim, M., I.U. Samad, N.A. Darwish, and N. Hilal. 2021. Comprehensive Review of Membrane Design
and Synthesis for Membrane Distillation. Desalination 518 (December): 115168.
Qin, L., I.A. Mergos, and H. Verweij. 2015. Obtaining Accurate Cross-Section Images of Supported
Polymeric and Inorganic Membrane Structures. Journal of Membrane Science 476 (February):
194–199.
Qtaishat, M., T. Matsuura, B. Kruczek, and M. Khayet. 2008. Heat and Mass Transfer Analysis in Direct
Contact Membrane Distillation. Desalination 219, no. 1–3 (January): 272–292.
Quist-Jensen, C. A., J.M. Sorensen, A. Svenstrup, L. Scarpa, T.S. Carlsen, H.C. Jensen, L. Wybrandt, and
M.L. Christensen. 2018a. Membrane Crystallization for Phosphorus Recovery and Ammonia
Stripping from Reject Water from Sludge Dewatering Process. Desalination 440 (August 15):
156–160.
Quist-Jensen, C.A., F. Macedonio, C. Conidi, A. Cassano, S. Aljlil, O.A. Alharbi, and E. Drioli. 2016.
Direct Contact Membrane Distillation for the Concentration of Clarified Orange Juice. Journal of
Food Engineering 187 (October): 37–43.
Quist-Jensen, Cejna Anna, A. Ali, S. Mondal, F. Macedonio, and E. Drioli. 2016a. A Study of Membrane
Distillation and Crystallization for Lithium Recovery from High-Concentrated Aqueous
Solutions. Journal of Membrane Science 505 (May): 167–173.
Quist-Jensen, Cejna Anna, F. Macedonio, D. Horbez, and E. Drioli. 2017. Reclamation of Sodium Sulfate
from Industrial Wastewater by Using Membrane Distillation and Membrane Crystallization.
Desalination 401 (January): 112–119.
Rácz, G., S. Kerker, Z. Kovács, G. Vatai, M. Ebrahimi, and P. Czermak. 2014. Theoretical and
Experimental Approaches of Liquid Entry Pressure Determination in Membrane Distillation
Processes. Periodica Polytechnica Chemical Engineering 58, no. 2: 81–91.
Rahman, M.M., S. Al-Sulaimi, and A.M. Farooque. 2018. Characterization of New and Fouled SWRO
Membranes by ATR/FTIR Spectroscopy. Applied Water Science 8, no. 6 (October 26): 183.
Rahman, S.N., H. Saleem, and S.J. Zaidi. 2023. Progress in Membranes for Pressure Retarded Osmosis
Application. Desalination 549 (March): 116347.
Ramlow, H., R.K.M. Ferreira, C. Marangoni, and R.A.F. Machado. 2019. Ceramic Membranes Applied
to Membrane Distillation: A Comprehensive Review. International Journal of Applied Ceramic
Technology 16, no. 6 (November 13): 2161–2172.
Ravi, J., M.H.D. Othman, T. Matsuura, M. Ro’il Bilad, T.H. El-badawy, F. Aziz, A.F. Ismail, M.A. Rahman,
and J. Jaafar. 2020. Polymeric Membranes for Desalination Using Membrane Distillation: A
Review. Desalination 490 (September): 114530.
132
 Chapter 5

Rezaei, M., D.M. Warsinger, J.H. Lienhard V, M.C. Duke, T. Matsuura, and W.M. Samhaber. 2018.
Wetting Phenomena in Membrane Distillation: Mechanisms, Reversal, and Prevention. Water
Research 139 (August): 329–352.
Sadeghi Ghari, H., Z. Shakouri, and M.M.A. Shirazi. 2014. Evaluation of Microstructure of Natural
Rubber/Nano-Calcium Carbonate Nanocomposites by Solvent Transport Properties. Plastics,
Rubber and Composites 43, no. 6.
Sadeghzadeh, A., S. Bazgir, and M.M.A. Shirazi. 2020. Fabrication and Characterization of a Novel
Hydrophobic Polystyrene Membrane Using Electroblowing Technique for Desalination by
Direct Contact Membrane Distillation. Separation and Purification Technology 239.
Sadeghzadeh, Amirhossein, S. Bazgir, and M.M.A. Shirazi. 2020. Fabrication and Characterization of a
Novel Hydrophobic Polystyrene Membrane Using Electroblowing Technique for Desalination
by Direct Contact Membrane Distillation. Separation and Purification Technology 239 (May):
116498.
Saffarini, R.B., B. Mansoor, R. Thomas, and H.A. Arafat. 2013. Effect of Temperature-Dependent
Microstructure Evolution on Pore Wetting in PTFE Membranes under Membrane Distillation
Conditions. Journal of Membrane Science 429 (February): 282–294.
Said, I.A., T. Chomiak, J. Floyd, and Q. Li. 2020. Sweeping Gas Membrane Distillation (SGMD) for
Wastewater Treatment, Concentration, and Desalination: A Comprehensive Review. Chemical
Engineering and Processing - Process Intensification 153 (July): 107960.
Salinas Rodríguez, S. G., Amy, G. L., Schippers, J. C. & Kennedy, M. D. (2015). The Modified Fouling
Index Ultrafiltration Constant Flux for assessing particulate/colloidal fouling of RO systems.
Desalination, 365, 79-91.
Sanaeepur, H., A. Ebadi Amooghin, M.M.A. Shirazi, M. Pishnamazi, and S. Shirazian. 2022. Water
Desalination and Ion Removal Using Mixed Matrix Electrospun Nanofibrous Membranes: A
Critical Review. Desalination 521 (January): 115350.
Schofield, R.W., A.G. Fane, and C.J.D. Fell. 1987. Heat and Mass Transfer in Membrane Distillation.
Journal of Membrane Science 33, no. 3 (October): 299–313.
Schofield, R.W., A.G. Fane, C.J.D. Fell, and R. Macoun. 1990. Factors Affecting Flux in Membrane
Distillation. Desalination 77 (March): 279–294.
Shahu, V.T., and S.B. Thombre. 2019a. Air Gap Membrane Distillation: A Review. Journal of Renewable
and Sustainable Energy 11, no. 4 (July): 045901.
———. 2021. Experimental Analysis of a Novel Helical Air Gap Membrane Distillation System. Water
Science and Technology: Water Supply 21, no. 4 (June 1): 1450–1463.
Shalaby, S.M., A.E. Kabeel, H.F. Abosheiasha, M.K. Elfakharany, E. El-Bialy, A. Shama, and R.D. Vidic.
2022. Membrane Distillation Driven by Solar Energy: A Review. Journal of Cleaner Production
366 (September 15): 132949.
Sharma, S., C.N. Shyam Kumar, J.G. Korvink, and C. Kübel. 2018. Evolution of Glassy Carbon
Microstructure: In Situ Transmission Electron Microscopy of the Pyrolysis Process. Scientific
Reports 8, no. 1 (November 2): 16282.
Sharma, V., and A. Bhardwaj. 2019. Scanning Electron Microscopy (SEM) in Food Quality Evaluation.
In Evaluation Technologies for Food Quality, 743–761. Elsevier.
Shirazi, M. M.A., A. Kargari, S. Bazgir, M. Tabatabaei, M.J.A. Shirazi, M.S. Abdullah, T. Matsuura, and
A.F. Ismail. 2013. Characterization of Electrospun Polystyrene Membrane for Treatment
of Biodiesel’s Water-Washing Effluent Using Atomic Force Microscopy. Desalination 329
(November 15): 1–8.
Shirazi, A., M., and A. Kargari. 2019. Concentrating of Sugar Syrup in Bioethanol Production Using
Sweeping Gas Membrane Distillation. Membranes 9, no. 5 (May 1): 59. Shirazi, M.J.A., S.

133
Experimental Methods for Membrane Applications

Bazgir, M.M.A. Shirazi, and S. Ramakrishna. 2013. Coalescing Filtration of Oily Wastewaters:
Characterization and Application of Thermal Treated, Electrospun Polystyrene Filters.
Desalination and Water Treatment 51, no. 31–33 (September): 5974–5986.
Shirazi, M.M.A., A. Kargari, S. Bazgir, M. Tabatabaei, M.J.A. Shirazi, M.S. Abdullah, T. Matsuura, and
A.F. Ismail. 2013. Characterization of Electrospun Polystyrene Membrane for Treatment
of Biodiesel’s Water-Washing Effluent Using Atomic Force Microscopy. Desalination 329
(November): 1–8.
Shirazi, M.M.A., A. Kargari, and M.J.A. Shirazi. 2012. Direct Contact Membrane Distillation for Seawater
Desalination. Desalination and Water Treatment 49, no. 1–3.
Shirazi, M.M.A., A. Kargari, and M. Tabatabaei. 2014. Evaluation of Commercial PTFE Membranes in
Desalination by Direct Contact Membrane Distillation. Chemical Engineering and Processing:
Process Intensification 76.
Shirazi, Mohammad Mahdi A., D. Bastani, A. Kargari, and M. Tabatabaei. 2013. Characterization of
Polymeric Membranes for Membrane Distillation Using Atomic Force Microscopy. Desalination
and Water Treatment 51, no. 31–33 (September): 6003–6008.
Shirazi, Mohammad Mahdi A., S. Bazgir, and F. Meshkani. 2020a. A Novel Dual-Layer, Gas-Assisted
Electrospun, Nanofibrous SAN4-HIPS Membrane for Industrial Textile Wastewater Treatment
by Direct Contact Membrane Distillation (DCMD). Journal of Water Process Engineering 36
(August): 101315.
———. 2020b. A Dual-Layer, Nanofibrous Styrene-Acrylonitrile Membrane with Hydrophobic/
Hydrophilic Composite Structure for Treating the Hot Dyeing Effluent by Direct Contact
Membrane Distillation. Chemical Engineering Research and Design 164 (December): 125–146.
Shirazi, Mohammad Mahdi A., and L.F. Dumée. 2022. Membrane Distillation for Sustainable
Wastewater Treatment. Journal of Water Process Engineering 47 (June 1): 102670.
Shirazi, Mohammad Mahdi A., A. Kargari, A.F. Ismail, and T. Matsuura. 2016. Computational Fluid
Dynamic (CFD) Opportunities Applied to the Membrane Distillation Process: State-of-the-Art
and Perspectives. Desalination 377 (January 1): 73–90.
Shirazi, Mohammad Mahdi A., A. Kargari, M. Tabatabaei, A.F. Ismail, and T. Matsuura. 2014.
Concentration of Glycerol from Dilute Glycerol Wastewater Using Sweeping Gas Membrane
Distillation. Chemical Engineering and Processing: Process Intensification 78 (April): 58–66.
Shyam Kumar, C.N., V.S.K. Chakravadhanula, A. Riaz, S. Dehm, D. Wang, X. Mu, B. Flavel, R.
Krupke, and C. Kübel. 2017. Understanding the Graphitization and Growth of Free-Standing
Nanocrystalline Graphene Using in Situ Transmission Electron Microscopy. Nanoscale 9, no. 35:
12835–12842.
Silva, R. de S., H. Ramlow, B. de C. Santos, H.B. Madalosso, R.A.F. Machado, and C. Marangoni. 2021.
Membrane Distillation: Experimental Evaluation of Liquid Entry Pressure in Commercial
Membranes with Textile Dye Solutions. Journal of Water Process Engineering 44 (December):
102339.
Simoni, G., B.S. Kirkebæk, C.A. Quist-Jensen, M.L. Christensen, and A. Ali. 2021. A Comparison of
Vacuum and Direct Contact Membrane Distillation for Phosphorus and Ammonia Recovery
from Wastewater. Journal of Water Process Engineering 44 (December 1): 102350.
Suk, D., T. Matsuura, H. Park, and Y. Lee. 2006. Synthesis of a New Type of Surface Modifying
Macromolecules (NSMM) and Characterization and Testing of NSMM Blended Membranes for
Membrane Distillation. Journal of Membrane Science 277, no. 1–2 (June 1): 177–185.
Susanto, H. 2011a. Towards Practical Implementations of Membrane Distillation. Chemical
Engineering and Processing: Process Intensification 50, no. 2 (February): 139–150.

134
 Chapter 5

Tai, Z.S., M.H.A. Aziz, M.H.D. Othman, A.F. Ismail, M.A. Rahman, and J. Jaafar. 2019. An Overview of
Membrane Distillation. In Membrane Separation Principles and Applications, 251–281. Elsevier.
Talukder, M.E., Md.N. Pervez, W. Jianming, G.K. Stylios, M.M. Hassan, H. Song, V. Naddeo, and A.
Figoli. 2022. Ag Nanoparticles Immobilized Sulfonated Polyethersulfone/Polyethersulfone
Electrospun Nanofiber Membrane for the Removal of Heavy Metals. Scientific Reports 12, no.
1 (April 6): 5814.
Tanis-Kanbur, M.B., R.I. Peinador, J.I. Calvo, A. Hernández, and J.W. Chew. 2021. Porosimetric
Membrane Characterization Techniques: A Review. Journal of Membrane Science 619
(February): 118750.
Tibi, F., A. Charfi, J. Cho, and J. Kim. 2020. Fabrication of Polymeric Membranes for Membrane
Distillation Process and Application for Wastewater Treatment: Critical Review. Process Safety
and Environmental Protection 141 (September): 190–201.
Tijing, L.D., J.-S. Choi, S. Lee, S.-H. Kim, and H.K. Shon. 2014. Recent Progress of Membrane Distillation
Using Electrospun Nanofibrous Membrane. Journal of Membrane Science 453 (March): 435–
462.
Tijing, L.D., W. Choi, Z. Jiang, A. Amarjargal, C.-H. Park, H.R. Pant, I.-T. Im, and C.S. Kim. 2013. Two-
Nozzle Electrospinning of (MWNT/PU)/PU Nanofibrous Composite Mat with Improved
Mechanical and Thermal Properties. Current Applied Physics 13, no. 7 (September): 1247–1255.
Tijing, L.D., J.R.C. Dizon, I. Ibrahim, A.R.N. Nisay, H.K. Shon, and R.C. Advincula. 2020. 3D Printing
for Membrane Separation, Desalination and Water Treatment. Applied Materials Today 18
(March 1): 100486.
Tsai, J.H., C. Quist-Jensen, and A. Ali. 2023. Multipass Hollow Fiber Membrane Modules for Membrane
Distillation. Desalination 548 (February 15).
Tung, K.-L., K.-S. Chang, T.-T. Wu, N.-J. Lin, K.-R. Lee, and J.-Y. Lai. 2014. Recent Advances in the
Characterization of Membrane Morphology. Current Opinion in Chemical Engineering 4 (May):
121–127.
Vicente, J., Y. Wyart, and P. Moulin. 2013. Characterization (two-dimensional - three-dimensional)
of ceramic microfiltration membrane by synchrotron radiation: new and abraded membranes.
Journal of Porous Media 16, no. 6: 537–545.
Vladár, A.E., and V.-D. Hodoroaba. 2020. Characterization of Nanoparticles by Scanning Electron
Microscopy. In Characterization of Nanoparticles, 7–27. Elsevier.
Wang, K., A.A. Abdalla, M.A. Khaleel, N. Hilal, and M.K. Khraisheh. 2017. Mechanical Properties of
Water Desalination and Wastewater Treatment Membranes. Desalination 401 (January): 190–
205.
Wang, P., and T.-S. Chung. 2015. Recent Advances in Membrane Distillation Processes: Membrane
Development, Configuration Design and Application Exploring. Journal of Membrane Science
474 (January): 39–56.
Wang, X., Y. Liu, K. Fan, P. Cheng, and S. Xia. 2023. Nano-Striped Polyamide Membranes Enabled by
Vacuum-Assisted Incorporation of Hierarchical Flower-like MoS2 for Enhanced Nanofiltration
Performance. Journal of Membrane Science 668 (February): 121250.
Wei, X., B. Zhao, X.-M. Li, Z. Wang, B.-Q. He, T. He, and B. Jiang. 2012. CF4 Plasma Surface Modification
of Asymmetric Hydrophilic Polyethersulfone Membranes for Direct Contact Membrane
Distillation. Journal of Membrane Science 407–408 (July): 164–175.
Wiktor, C., M. Meledina, S. Turner, O.I. Lebedev, and R.A. Fischer. 2017. Transmission Electron
Microscopy on Metal–Organic Frameworks – a Review. Journal of Materials Chemistry A 5, no.
29: 14969–14989.

135
Experimental Methods for Membrane Applications

Winter, D., J. Koschikowski, and S. Ripperger. 2012. Desalination Using Membrane Distillation: Flux
Enhancement by Feed Water Deaeration on Spiral-Wound Modules. Journal of Membrane
Science 423–424 (December): 215–224. http://linkinghub.elsevier.com/retrieve/pii/
S0376738812006175.
Winter, D., J. Koschikowski, and M. Wieghaus. 2011. Desalination Using Membrane Distillation:
Experimental Studies on Full Scale Spiral Wound Modules. Journal of Membrane Science 375,
no. 1–2: 104–112. http://dx.doi.org/10.1016/j.memsci.2011.03.030.
Wyart, Y., G. Georges, C. Deumié, C. Amra, and P. Moulin. 2008. Membrane Characterization by Microscopic
Methods: Multiscale Structure. Journal of Membrane Science 315, no. 1–2 (May): 82–92.
Xing Yang, Hui Yu, Rong Wang, A.G.F., X. Yang, H. Yu, R. Wang, and A.G. Fane. 2012. Optimization
of Microstructured Hollow Fiber Design for Membrane Distillation Applications Using CFD
Modeling. Journal of Membrane Science 421–422 (December): 258–270. http://dx.doi.
org/10.1016/j.memsci.2012.07.022.
Yang, F., J.E. Efome, D. Rana, T. Matsuura, and C. Lan. 2018. Metal–Organic Frameworks Supported on
Nanofiber for Desalination by Direct Contact Membrane Distillation. ACS Applied Materials &
Interfaces 10, no. 13 (April 4): 11251–11260.
Yang, X., R. Wang, A.G. Fane, C.Y. Tang, and I.G. Wenten. 2013. Membrane Module Design and
Dynamic Shear-Induced Techniques to Enhance Liquid Separation by Hollow Fiber Modules: A
Review. Desalination and Water Treatment 51, no. 16–18: 3604–3627. https://doi.org/10.10
80/19443994.2012.751146.
Zhang, W., Y. Li, J. Liu, B. Li, and S. Wang. 2017. Fabrication of Hierarchical Poly (Vinylidene Fluoride)
Micro/Nano-Composite Membrane with Anti-Fouling Property for Membrane Distillation.
Journal of Membrane Science 535 (August): 258–267.
Zhao, D., J. Zuo, K.-J. Lu, and T.-S. Chung. 2017. Fluorographite Modified PVDF Membranes for
Seawater Desalination via Direct Contact Membrane Distillation. Desalination 413 (July): 119–
126.
Zhao, K., W. Heinzl, M. Wenzel, S. Büttner, F. Bollen, G. Lange, S. Heinzl, and N. Sarda. 2013.
Experimental Study of the Memsys Vacuum-Multi-Effect-Membrane-Distillation (V-MEMD)
Module. Desalination 323 (August): 150–160. http://linkinghub.elsevier.com/retrieve/pii/
S0011916412006509.
Zhu, Y., H. Inada, A. Hartschuh, L. Shi, A. Della Pia, G. Costantini, A.L. Vázquez de Parga, et al. 2012.
Scanning Electron Microscopy. In Encyclopedia of Nanotechnology, 2273–2280. Dordrecht:
Springer Netherlands.
Ziel, R., A. Haus, and A. Tulke. 2008. Quantification of the Pore Size Distribution (Porosity Profiles) in
Microfiltration Membranes by SEM, TEM and Computer Image Analysis. Journal of Membrane
Science 323, no. 2 (October): 241–246.
Zou, D., and Y.M. Lee. 2022. Design Strategy of Poly(Vinylidene Fluoride) Membranes for Water
Treatment. Progress in Polymer Science 128 (May): 101535.
Zuo, J., and T.-S. Chung. 2016. Metal–Organic Framework-Functionalized Alumina Membranes for
Vacuum Membrane Distillation. Water 8, no. 12 (December 8): 586.
Zuo, J., T.-S. Chung, G.S. O’Brien, and W. Kosar. 2017. Hydrophobic/Hydrophilic PVDF/Ultem®
Dual-Layer Hollow Fiber Membranes with Enhanced Mechanical Properties for Vacuum
Membrane Distillation. Journal of Membrane Science 523 (February): 103–110.

136
 Part 2
Particulate fouling
 doi: 10.2166/9781789062977_0139

Chapter 6

Silt Density Index

Steven J. Duranceau, University of Central Florida, USA

The learning objectives of this chapter are the following:

• Define the silt density index (SDI) and explain its significance

• Present a method that can be used to characterize the particulate fouling potential of
reverse osmosis feedwater

• Understand the theory behind the SDI and discuss the basic equations that are used
to calculate fouling indices

• To learn how to perform a SDI test using the appropriate equipment and procedures

• Identify and explain the limitations and deficiencies of the SDI as a measure of
particulate fouling in synthetic membrane processes.

6.0 ABSTRACT

The most widely accepted and historically used predictor of the fouling potential of reverse
osmosis feedwater is the plugging factor (PF), now commonly known as the Silt Density
Index (SDI). The SDI procedure was standardized by ASTM International and is intended
to be used as a measure of the fouling capacity of feedwater to reverse osmosis systems. The
SDI is an index calculated from a test that measures the rate at which a 0.45-micrometer
(μm) filter is plugged when subjected to a constant water pressure of 206.8 kPa (30 psi).
The SDI gives the percent drop per minute in the flow rate of the water through the filter,
averaged over a specified time-period, typically 15 minutes. Because the SDI is a relatively
simple procedure and inexpensive to implement, it has been universally applied since the
1960s to assess the particulate fouling tendency of a feedwater intended for treatment by
reverse osmosis (RO) membrane processes. Many facilities in the United States rely on
automated SDI systems that are microprocessor controlled and fully automatic to allow
operators to regularly monitor the feedwater. Care must be taken when employing the
SDI with regards to accuracy and reproducibility, as the index is not based on any filtration

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

mechanism and is not proportional with colloidal or particle concentration, in addition to

the fact that there is no correlation factor for temperature nor for variations in membrane
resistance. Even though there are legitimate concerns regarding the predictive value of the
SDI, the index continues to be successfully used in the planning of RO facilities to guide
the selection of pretreatment processes, and is often the basis of membrane guarantees and
other plant performance contracts.

6.1 DEVELOPMENT OF THE FOULING INDEX

Fouling is a major obstacle to the widespread use of membrane technology. Membrane

fouling has a direct impact on process performance because it decreases productivity
(permeate flux) over time, increases the amount of energy required to meet water demand,
and accelerates the need for membrane cleaning and replacement. Fouling is simply defined
as the accumulation of undesired materials, via deposition or adsorption of soluble and
particulate matter, onto the surface of the membrane. Membrane fouling can consist of
either non-living (inorganic and organic matter) or living organisms (bacteria) and is the
primary cause of permeate decline and loss of overall productivity in reverse osmosis and
nanofiltration processes.

Early in the development of semi-permeable membranes for water treatment, the need to
estimate membrane fouling potential of the raw water was found to be essential to identify
pretreatment requirement to prepare the feedwater prior to processing. This is important
because effective pretreatment can lower the number of required cleaning events and
extend the life of membrane elements. Attempts to correlate fouling propensity of water
with turbidity was only slightly successful.

To help solve these issues, the Du Pont de Nemours & Company (Du Pont) introduced
its first reverse osmosis permeators for water desalination in 1969 under the trade name
‘Permasep’ an outcome of the company’s research in polymer chemistry and synthetic fibers
(Hagley Library, 2022). By 1997, Du Pont had sold over 1.5 billion gallons of desalting
capacity, dominating the seawater desalination market for many decades. Since the first
hollow-fine fiber membranes were sold by DuPont it was initially believe that performance
was hampered by suspended and colloidal matter in the feedwater. Consequently, DuPont
developed the Fouling Index, which was later denoted as the Silt Density Index (Schippers
et al. 2014). Despite Permasep’s success, DuPont decided to discontinue the production
of hollow-fiber permeators in 1997, primarily attributed to the rise of the spiral-wound
membrane configuration’s success in the global desalination market (Hagley Library, 2022).

6.2 SILT AS A COMPONENT OF MEMBRANE FOULING

Assalay and colleages (1998) described silt as a solid granular material that is comprised of
suspended rock and mineral particles with a size between sand or clay that can accumulate
on the membrane surface. Although the SDI is termed as a silt index, this does not mean that
the measurement is for silt considerations alone. The SDI is a parameter used to determine
the fouling propensity of a source water intended to be processed using reverse osmosis
membranes. Sources of membrane fouling can be divided into four principal categories:

140
 Chapter 6

• Particulate (silt, inorganic colloids, oxidized iron and manganese, algae, aluminosilicates)
• Microbiological (bacteria)
• Organic fouling (natural or synthetic compounds, oils, grease)
• Scale (limiting salt chemistries)

Since reverse osmosis synthetic membrane processes were first introduced for the
treatment of water supplies, it was found that in most instances plugging of the elements
was due to blocking filtration by suspended particulate matter (Comstock, 1982). Fouling
by particulates (silt) generally impacts the lead membrane elements of any pressure vessel
process configuration unlike scale that concentrates in the flow channels of the tail-end
or last membrane elements located in a pressure vessel. Scaling is of greater concern with
more concentrated feed solutions, therefore the last modules in the process pressure vessel
configurations are most affected because they are exposed to the most concentrated feed
water. Microbiological and organic-type fouling can occur anywhere within the membrane
configuration depending on feedwater quality, pretreatment methods and process
operating conditions. Consequently, the SDI is a measurement that can determine the
fouling potential of a feedwater for particulate fouling, and may not be as predictive for
microbiological, organic or scale type conditions.

6.3 STANDARDIZATON OF THE SILT DENSITY INDEX

In 1982, ASTM International (West Conshohocken, PA), formerly known as the American
Society for Testing and Materials, developed the ‘Standard Test Method for Silt Density
Index (SDI) of Water’ designated as D4189-14 by ASTM International (2014) which was
first revised on January 30, 1987. According to ASTM International (2014), the SDI test
method can be used to ‘indicate the quantity of particulate matter in water and is applicable
to relatively low (<1.0 NTU) turbidity waters such as well water, filtered water, or clarified
effluent samples.’ The test is not applicable to RO, NF or ultrafiltration (UF) permeate. The
test essentially consists of filtering water through a 47-mm diameter cellulose-based filter
that possesses 0.45-μm at a constant pressure of 30 psi (210 kPa). The standard ASTM SDI
test does not contain any correction for testing parameters such as membrane resistance,
membrane area, feed temperature and applied pressure. The SDI increases with increasing
temperature since the water viscosity is affected; additionally, an increase in pressure and
decrease in membrane resistance will increase the measurements result. The SDI test is not
an absolute measurement of the quantity of particulate matter.

The method has essentially remained the same procedure since that time and has been
proven to be useful from an operating perspective for membrane plant operators (Ruiz-
Garcia et al. 2015). According to Harn R/O Systems, Inc. (2022), the SDI test gives a
calculated number in the range 0 – 6, where 0 is excellent and 6 denotes a very high fouling
potential. Most membrane manufacturers require the feed SDI to be below 3.0 to indicate
control of colloidal and particulate fouling. SDI values above 3.0 typically indicate periodic
cleaning will be required and values above 5.0 could indicate rapid fouling can indicate
the need to provide additional pretreatment to protect the membranes during process
operation. Although in practice a high SDI typically indicates that fouling may occur, a low
value does not guarantee that fouling will not occur.

141
Experimental Methods for Membrane Applications

6.4 METHODS AND PROCEDURES

Manual SDI testing typically requires the following components and items, portions of
which are illustrated in Figure 1 (Hydranautics, 2022) and shown in Figure 2:
• SDI test assembly made of high-quality stainless steel or plastic.
• Filter holder to withstand 50 psi (350 kPa) pressure and hold.
• 47 mm nominal, plain filter papers115 to 180 micrometer thickness rated to 0.45 μm,
typically white hydrophilic cellulose triacetate or mixed cellulose nitrate type materials.
• Pressure regulator and gauge able to measure 30 psi.
• Feedwater ball valve, plastic.
• Graduated cylinder, 500 mL capacity
• Stopwatch, graduated in hundredths of a minute.
• Thermometer, liquid-in-glass, suitable for measuring the temperature of the water
sample; capable of being read to within ±1°C.
• Dull tweezers.

Feed

Ball valve or 1st stage regulator

Pressure regulator 30 psi

Pressure gauge

Bleed
O-ring
0.45 micron filter

Figure 1 Typical assembly of the apparatus used for SDI measurements

Source: https://membranes.com/wp-content/uploads/Documents/TSB/TSB113.pdf

Several private corporations and original equipment manufacturer’s provide in-house

procedures, based on ASTM D4189-07, such as the information provided by AquaPhoenix
Scientific (2022), Hydranautics (2022) and Harn R/O Systems, Inc. (2022). Note that a
booster pump may be required for use in a manual SDI kit if insufficient pressure exists for
testing to proceed. To perform an SDI test, the following general procedure can be followed,
and is provided in more detail in ASTM International SDI testing method (2014). On-
line instruments are also available that automatically and consistently monitor the SDI of
RO feedwater that are typically controlled by a microprocessor and rely on a typical 4-20
mA interface. These devices contain side-stream or automatic flush sequences, and often
extrapolate the SDI to 75% of the filter plugging condition.

142
 Chapter 6

Figure 2 Photo depicting the components of the SDI apparatus: storage box for equipment,
500-mL graduated cylinder, Teflon tape, dull tweezers, SDI filter pads, stopwatch, cell
assembly and flexible tubing, pressure regulator and gauge, booster pump.
Source: Courtesy of Harn R/O Systems, Inc., a Division of Komline-Sanderson Corp.

Typically, most reverse osmosis and nanofiltration membrane water treatment facilities
make available a well flush valve on the raw water line upstream of the process building. The
flush valve allows the raw water to be flushed to waste at process start-up for an operational
pre-determined time period to reduce the SDI reading to below 3.0 prior to allowing water
to be transferred to the pretreatment system ahead of the membranes. An SDI sample point
can easily be installed on the raw water line upstream of the plant inlet valve so that SDI
tests can be performed during well flush events. Taking these SDI tests during plant start-up
allows operators to determine the length of time needed for individual and collective (or
intake) flushing cycles as each facility may have differing source water supply transmission
line configurations. It is recommended that an SDI test should be performed at least once a
day on the raw water when the process is in operation and the results should be recorded in
the operator’s daily log.

Step 1: Measure the time required to filter a fixed volume of water through a standard space:
0.45 μm pore size microfiltration membrane at a constant pressure of 30 psi (2.07 bar) per
the following procedure. Record this as Ti, or initial T.
a. Connect the test kit less filter paper for pretest flush.
b. Flush the test kit and supply line for 3 to 5 minutes to remove any possible contaminants.
c. Measure the temperature of the water and record the reading.
d. Make sure the O-ring on the filter is in good condition and properly placed. Set the
pressure regulator to 30 psig (210 kPa). The set screw on the regulator should be
adjusted while there is a small flow. Supply pressure to the regulator should be greater
than 40 psig (276 kPa).
e. Open the 47 mm in diameter filter holder and carefully place a 0.45 μm membrane
filter into the filter holder using the dull tweezers to avoid damage and touching the
filter paper. Screw loosely together.

143
Experimental Methods for Membrane Applications

f. Open the feed valve slightly and adjust the filter housing to overflow, displacing any
trapped air. Residual air trapped in the housing can be flushed by opening the small
‘bleed’ screw; care should be taken else this part can come loose and be easily lost.
Tighten the filter housing, open the feed valve completely and make final adjustments
to the pressure regulator as required; close the valve and tighten the filter holder the
remainder of the way without overtightening.
g. Prepare to take measurements. Open the ball valve and simultaneously, using the
stopwatch, begin measuring the time required to fill the 500 mL measuring cylinder.
Record the time (ti). Leave the valve open for continued flow; do not stop the watch or
the water flow.

Step 2: Take additional time measurements, normally after 5, 10 and 15 minutes (after silt
build up). Measure and record the times to collect additional 500 mL volumes of sample,
starting the collection at 5, 10, 15 minutes of total elapsed flow time. This value is recorded
as (tf) with f being the time used. Measure the water temperature and check the pressure
as each sample is collected. The pressure must remain constant at 30 psig (± 1 psig) and
the temperature must remain constant to 1˚C. After completion of the test, the membrane
filter may be retained for future reference or additional chemical evaluations of the filtered
deposit matter. It is recommended that the date, time, sample location, operator name, SDI
value and notes or comments be collected along with the filter pad.

Step 3: Calculate the Plugging Factor (PF) after 5, 10 and 15 minutes as determined as
shown in Equations 1, 2 and 3, respectively:

Ti
PF5 min = (1 ) ×100 Eq. 1
T5

Ti
PF10 = (1 ) ×100 Eq. 2
min
T10

Ti Eq. 3
PF15 min = (1 ) ×100
T15

Step 4: The SDI value is then determined at each interval as SDI = PF/T. Calculate the Silt
Density Index (SDI) as follows using Equation (4):

Ti
1 ×100
%P30 Tf
SDIT = = Eq. 4
T T
where SDIT is the Silt Density Index (%/min) at time T, ti is the initial time required to
collect 500 mL of sample, tf is the elapsed filtration time (min) required to collect 500 mL of
sample after a test time (typically 15 minutes after the initial measurement), and %P is the
percent at 30 psi feed pressure.

144
 Chapter 6

The ASTM method recommends that if the %P30 exceeds 75% after 5 min then other test
methods should be used to analyze for particulate matter. Considering that the %P30 is
essentially the percentage of plugging factor, Equation 4 can be rewritten as Equation 5:

T1
1 ×100
%PF T2 Eq. 5
SDIT = =
T T

SDI measures the percentage of the filtrate flow rate decline per minute and is expressed as
a percentage per minute but typically is reported without units. As an example, a SDI of 2.5
would indicate that the SDI filtrate flow was reduced by 2.5 percent per minute during the
test. This concept is illustrated graphically in Figure 3 where the filtration flow is presented
as a function of time, and V1 and V2 are the volumes of the first and second sample:

“Percentage flux decline per minute”

V1

V2=V1

Flow
t1 t2 Time (minutes)
T

Figure 3 Representation of the filtration flow as a function of time per the SDI test method.
(Adapted from Alhadidi and colleagues 2011)

Calculation Examples
1. Calculate the PF and SDI for a test where the time measurements indicated a Ti of one
minute and T15 of 4.0 minutes.
Solution: The plugging factor is calculated as PF = 1-(1/4)×100 = 75. On the other hand,
the SDI is then calculated as SDI = 75/15 = 5 as a percentage of flux decline per minute.
These results indicated that flow had decayed by a factor of four times, indicating that
75% of the 0.45-micron filter has been plugged. As the SDI test is a dead-end filtering
scenario, unlike the RO processes that have one input stream and two output streams,
the flow across the membrane in the feed-concentrate channel results in concentration
polarization that is not evaluated in the SDI; hence the SDI15 result should not be used to
predict an actual fouling rate for an operating RO process. The SDI value only represents
an indication that fouling may occur. For this example, an SDI result of 5 would indicate

145
Experimental Methods for Membrane Applications

that additional pretreatment is most likely required for this source water using RO
membrane processes.
2. A municipal utility is considering using a membrane process as an alternative treatment
method for its newly developed wellfield. The following information is collected using a
SDI testing apparatus on an unfiltered raw ground water sample. Determine the 15-min
SDI value.

Table 1 SDI Test Results for Water Utility

Test Time Water Volume Collected Time to Collect 500 mL
(min) (mL) (sec)
0.0 500 9.0
5.0 500 11
10 500 12
15 500 14

Solution: The SDI for the 15-min test is calculated as SDI = {100[1-(ti/tf)]/t} =
100[1-(9sec/14sec)]/15-min} = 2.4 units as a percentage flux decline per minute. Since
the SDI is less than 3.0 units, the specific water supply would be considered acceptable as
a feedwater for a reverse osmosis treatment process. However, the SDI result would not
predict fouling of an RO membrane due to such mechanisms as sparingly salt scaling,
the impact of dissolved iron, for example. Hence the use of the SDI results should be
supplemented with other predictive fouling factors, which include limiting salt chemistry,
contribution of metal oxidation, particle formation within the feed-concentrate channel,
and organic deposition due to natural or synthetic organic matter.

6.5 LIMITATIONS OF THE SDI

The procedure outlined in the ASTM International (2014) method should be followed
as closely as possible to collect data that has meaning and is reproducible. Test variability
(50 - 100%) has been a recognized problem with the SDI method and personnel training in
procedural details is a critical factor in obtaining precise and accurate test results. Although
the SDI is a popular test that has been empirically correlated with the fouling tendency of
desalination processes, its interpretation calls for some expertise on the part of the person
carrying out the test. The ASTM method procedure requires several actions to obtain an
accurate SDI that include:
• Requirement to flush the equipment prior to use,
• Need to wet the test pad filters prior to use,
• Purging air to avoid air entrapment within the test cell impacting the test pad surface,
• Efforts to avoid touching the membrane filters with hands, and
• Collecting the water temperature before and after each test.

Problems related to the SDI test have been documented by many (Rachman et al., 2013;
Alhadidi et al., 2011; Alhadidi et al., 2012; Boerlage 2008, Boerlage 2007; Yiantsios et al. ,

146
 Chapter 6

2005; Schippers and Verdouw, 1980). Issues are many and range from the understanding
that the SDI is not based on any one filtration mechanism to the fact that there is no
direct correlation between turbidity of a water and it’s measured SDI. Also, the SDI has
no linear relation with particulate matter and as is not corrected for temperature, pressure
and membrane resistance; consequently, values obtained from the method may not be
comparable and certainly variable.

Filtration, Flux and Fouling

Although the ASTM International method has proven valuable for determining the fouling
potential of seawater supplies for particulate fouling, the SDI may not be represented of
conditions that high-pressure RO membranes experience where concentration polarization
exists (350 psi to 1,000 psi). It is known that the SDI is not based on any single fouling
mechanism hence is not used to predict the rate of fouling in RO systems. This has been
illustrated in higher salinity feedwaters where SDI cake filtration is considered the
mechanism for particulate fouling as reported by Boerlage (2007).

Because the SDI method operates at 30 psi (210 kPa) in a ‘dead-end’ mode, cake compression
will influence the test results and may not be representative of actual conditions that exist at
the active layer of the RO membrane’s surface. Furthermore, the SDI makes use of 0.45 μm
filters, which are not represented of the pore size in RO and NF spiral-wound membranes
that approximate 0.001 μm (Fang, 2013 and Duranceau, 2013; Duranceau, 2021). Another
concern in comparing the SDI to actual RO plant operation, is that the flux rate of the test (>
1,600 L/m2/hr at the start) is far outside the typical values experienced in practice (20 to 25
L/m2/hr) as reported by Schippers and colleagues (1981).

As an example of one of the issues encountered with the testing, Alhadidi and colleagues
(2012) found that seawater sources with SDI values less than 3 may still foul the membranes
in practice; on the other hand, SDI values greater than 3 have been documented when
testing permeate of membrane filtration processing used as seawater pretreatment system
feeding a seawater RO process.

Membrane Filters and Holder

There are a variety of chemistries available for use as membrane filters used in collecting SDI
measurements, and can include mixed cellulose esters (MCE), mixed cellulose acetate and
cellulose nitrates), polyvinylidene fluoride (PVDF), Nylon, polytetrafluoroethylene (PTFE),
polyamide (PA) and polyethersulfone (PES). Prior work by Ando and colleagues (2003)
compared SDI measurements collected using hydrophilic (MCE, PVDF, PA) that yielded
greater values than the counterpart hydrophilic (PTFE) filters.

While there is some debate about which filter pad chemistry is appropriate for various
applications, often the membrane element warranty language does not stipulate the material
of the SDI filter pad required to maintain compliance with the warranty. Field experience at
a variety of municipal facilities has demonstrated that very different SDI results are observed
for different filter pad chemistries. Additionally, very different results can be obtained when
using the ‘same’ filter pad chemistries from different manufacturers. Also, SDI results can

147
Experimental Methods for Membrane Applications

produce different measurements when using ‘sterile’ versus ‘non-sterile’ pads. Furthermore,
pH has been shown to impact the SDI measurement, thought to be due to differences in pad
chemistry (Mosset et al., 2008).

Nahrstedt and Camargo (2008) studied the effect of filter support on SDI and MFI
measurements and found that the filter holder influenced test results, up to 100 percent
across measurements. A similar finding was reported by Escobar et al., (2009) as well as
Salinas-Rodriguez et al., (2019).

Turbidity, Suspended Solids and the SDI

Water turbidity is an optical characteristic that serves as a measure of the relative clarity
of a liquid and is commonly used as an indicator for the general condition of the drinking
water. Turbidity is regulated under the U.S. Environmental Protection Agency’s secondary
drinking water standard for aesthetic reasons, and it is used as an operational control measure
as it is an easy field water-quality parameter to measure. Turbidity in water is caused by
suspended matter such as clay, silt, and organic matter and by microscopic organisms that
interfere with the passage of light through the water (American Public Health Association,
2017). Turbidimeters are calibrated with a formazine standard solution, which was used
in experimentation using SDI measurements. Schippers and Verdouw (1980) determined
that the absence of a temperature correction for measurement method resulted in higher
SDI values at higher temperatures. As has earlier been noted, there exists a non-linear
relationship between colloidal particle concentrations and measured SDI values, such that
the water being tested appears to be less fouling as the test filter becomes progressively
plugged. This concept has been illustrated by Schippers and Verdouw (1980) and is shown
in Figure 4.

15

10

SDI 0
0 1 2 3 4 5 6
Formazin (mg/L)

Figure 4 SDI as a function of formazin concentration by test duration. (Schipper and Verdouw
1980)

148
 Chapter 6

Surface water often contains high levels of fine suspended solids and therefore can often
exhibit higher turbidity values as well as high SDI that can vary depending on rainfall and
runoff impacts and may require more extensive pretreatment to achieve acceptable values
of SDI and turbidity in feed water from brackish rivers, lakes, bays as well as the ocean.
Brackish and briny well water supplies may contain particulates and would be susceptible to
metal oxidation and colloid formation if anaerobic groundwater was exposed to an aerobic
condition.

Oftentimes water with very low turbidity, less than 0.5 Nephelometric turbidity unit
(NTU), can still cause unacceptable membrane fouling, due to the presence of many
particles having very small diameter (typically under 3 microns in diameter) which do not
efficiently reflect light to show up as higher turbidity. Nonetheless, these small particles can
quickly foul spiral wound RO membranes. Sources of silt can include organic colloids, iron
corrosion products, precipitated iron hydroxide, algae, and fine particulate matter.

6.6 ALTERNATIVES TO THE SDI

Fouling indices can be categorized in two groups: those operating at constant pressure (the
SDI method) and those operating at constant flux. Because of many long-term concerns
about the predictive value of the SDI, many have attempted to find either alternatives or find
methods that can overcome variations attributed to its use. However, the few recommended
alternatives to SDI have proven equally problematic in application as the SDI.

One of the more successful modifications to the SDI approach is the modified fouling
index (MFI) proposed by Schippers and co-workers (Schippers and Verdouw 1980). The
premise behind this alternative approach is to carry out dead-end filtration tests of relatively
short duration, using a permeable membrane rated at 0.45 μm as is the case with the SDI,
to establish conditions of cake filtration and determine a quantity indicative of fouling.
The MFI-0.45 test uses the same equipment as the SDI test and takes into account that as
flow commences through the test pad, initially pore blocking occurs, followed by cake or
gel filtration and finally, cake or gel blocking and compression. The modified fouling index
(MFI) was suggested to measure the rate of cake formation on the membrane surface that is
believed to provide a better prediction of RO fouling phenomena. The development of the
MFI in many cases has demonstrated to be a more effective alternative to the SDI because of
three primary reasons:
1. A linear relationship with particle concentration exists;
2. The index is corrected for temperature;
3. It is based on the cake filtration mechanism.

However, the MFI poses its own limitations as like the SDI, the alternative test method
also operates at constant pressure producing high initial flux values. To overcome this
issue, Boerlage et al. (2004) developed the concept of a constant flux MFI instead of
constant pressure filtration, which was further developed by Salinas-Rodriguez (2011)
using polyethersulfone UF membranes. The value of the MFI was confirmed when ASTM
International published in 2015 the ‘Standard Test Method for Modified Fouling Index
(MFI-0.45) of Water’ (ASTM International 2015).

149
Experimental Methods for Membrane Applications

A group of researchers developed the concept of a mathematical model that could quantify
the influence of pressure, temperature and membrane resistance of the test and normalize
SDI measurements (Alhadidi et al. 2013). An alternative filtration index was proposed and
referred to as the volume-based SDI measurement (SDI-v) that compared the test’s initial
flow rate to the flow rate after filtering a standard volume of feed water using a 0.45-μm
microfiltration membrane. It was found that the SDI-v was independent of the membrane
resistance, which could overcome some of the variation typically present in the existing
method.

As confirmed in the research by Sioutopoulos and Karabelas (2012), SDI indices operating
at constant pressure are affected by the high flux rates at the start of the filtration test and
by the compressibility of the formed cake resulting from the very high initial fluxes. The
authors point out that there is no method available to predict the rate of fouling in RO/
NF plant operation, which is characterized by axial variability of key parameters (local TMP,
cross-flow velocity and flux). It was determined that under constant flux filtration over a
broad range of fluxes, thin fouling layers, of practical interest in relation to RO membrane
operations, exhibited a linear increase with time of the pressure drop across the cake, as
demonstrated in Figure 5.

1016

UF-constant flux
RO-constant flux
UF-constant pressure
RO-constant pressure
1015

Specific cake resistance, α (m/kg)

0 100 1,000
∆Pc (kPa)

Figure 5 Specific cake resistance versus pressure drop across the cakeunder constant flux and
constant pressure conditions. Results shown are for an experimental test water containing
2,000 mg/L total dissolved solids and a mixture of 10 mg/L of 75% humic acid (HA) and
25% sodium alginate (SA). Source: Sioutopoulos and Karabelas (2012).

150
 Chapter 6

6.7 SUMMARY

For more than half a decade the SDI has been accepted as a valuable test parameter and applied
in the application of RO processes worldwide. The test was developed as an empirical means
to measure the fouling propensity of the feed water to RO and NF membrane processes.
The SDI represented the fouling potential of synthetic RO and NF membranes due to finely
suspended particles present in the feed water.

Despite the success of the SDI, there are disadvantages to the test that limit the accuracy
and reproducibility of the method. Limitations included the fact that no correction factor
for temperature exists for the method, the test is not based on any single filtration model,
there are impacts caused by variations in membrane resistance, and there remains no linear
correlation between the measurement and the colloidal or suspended particulate content of
the water. Efforts continue to develop enhancements to the SDI (Khirani et al., 2006; Jin et
al., 2015).

Regardless of the many identified weaknesses of the SDI, the test method remains a simple
procedure to perform at minimal cost that does not need highly skilled professionals to
conduct. For this reason, the test will continue to be used (or misused) in the future during
the planning and operation of RO and NF facilities worldwide.

151
Experimental Methods for Membrane Applications

6.8 REFERENCES

Alhadidi, A., A.J. B Kemperman, B. Blanket, J.C. Schipper, M. Wessling and W.G.J. van der Meer (2011).
Silt density index and modified fouling index relation and effect of pressure, temperature and
membrane resistance. Desalination. 273: 48-56.
Alhadidi, A., A.J.B. Kemperman, J.C. Schippers, M. Wessling, and W.G.J. van der Meer (2012). SDI: is it
a reliable fouling index? Desalination and Water Treatment. 42: 43-48.
Alhadidi, A., B. Blankert, A.J.B. Kemperman, R. Schurer, J.C. Schippers, M. Wessling, and W.G.J. van der
Meer (2013). Limitations, improvements and alternatives of the silt density index. Desalination
and Water Treatment. 51 (4-6): 1104-1113.
American Public Health Association (2017). Standard Methods for the Examination of Water and
Wastewater, 23rd Edition. R.B. Baird, A.D. Eaton, E.W. Rice (editors). Water Environment
Federation, American Water Works Association, American Public Health Association, January
1, 2017, 1,796 pages.
Ando, M, S. Ishiara, H. Iwahori and N. Tada (2003). Peculiar or unexpected behaviour of silt density
index of pretreated water for desalination. Proceedings of the International Desalting
Association (IDA), World Congress on Desalination and Water Reuse, along with the 12th
Annual Conference of the Caribbean Water and Wastewater Association from 28 September to
3 October. Paper BAH 03-071, Paradise Islands, Bahamas.
AquaPhoenix (2022). AquaPhoenix Scientific, 860 Gitts Run Road Hanover, PA 17331; collected
from Web [Source: https://www.aquaphoenixsci.com/wp-content/uploads/procedures/
processed/Silt%20Density%20Index%20(SDI)%20Test%20Procedure_TK8550-Z.pdf].
Assallay, A.; Rogers, C.D.F.; Smalley, I.J.; Jefferson, I.F. (1998). ‘Silt: 2–62 μm, 9–4 ’. Earth-Science
Reviews. 45 (1–2): 61–88.
ASTM International (2014). D4189-07(2014) Standard Test Method for Silt Density Index (SDI)
of Water (Revised), ASTM International, 100 Barr Harbor Drive, PO Box C700, West
Conshohocken, PA 19428-2959.
ASTM International (2015). D8002-15 (2015) Standard Test Method for Modified Fouling Index
(MFI-0.45) of Water, ASTM International, 100 Barr Harbor Drive, PO Box C700, West
Conshohocken, PA 19428-2959.
Boerlage, S.F.E., M.D. Kennedy, M.R. Dickson, D.E.Y. El-Hodali, J.C. Schippers, The modified fouling
index using ultrafiltration membranes (MFI-UF): characterization, filtration mechanisms and
proposed reference membrane, J. Membrane Science. 197 (2002) 1–21.
Boerlage, S. F. E., M. Kennedy, Z. Tarawneh, R.D. Faber, and J.C. Schippers (2004), Development of the
MFI-UF in constant flux filtration, Desalination, 161, 103-113.
Boerlage, S. F. E. (2007), ‘Understanding the SDI and Modified Fouling Indices (MFI0.45 and MFI-UF)’,
IDA World Congress-Maspalomas, Gran Canaria, Spain, October 21-26, 2007, IDAWC/MP07-
143.
Boerlage, S. F. E. (2008) Understanding the Silt Density Index and Modified Fouling Indices (MFI0.45
and MFI-UF). Desalination and Water Reuse Quarterly, May to June, 12-21.
Comstock, D. (1982). Testing the membrane plugging factor in reverse osmosis. Journal of the
American Water Works Association. 74(9): 486-490.
Duranceau, S.J. (2021). Membrane Fouling Indices: Silt Density Index, Modified Fouling Index and the
Mini Plugging Factor Index. Proceedings of the Southeast Desalting Association 2021 Spring
Symposium: Together Again for Membranes, The Westin Cape Coral Resort at Marina Village,
Cape Coral FL (June 6-9, 2021).

152
 Chapter 6

Escobar L, W. Sellerberg, D. Sanchez, F. Pastrana, and A. Wachinski (2009). Detailed analysis of the
silt density index (SDI) results on desalination and wastewater reuse applications for reverse
osmosis technology evaluation. Proceedings of the international forum on marine science and
technology and economic development. Asia-Pacific Desalination conference, Qingdao, China,
8-10 July 2009.
Fang, Y., and S.J. Duranceau (2022). Comparison of Non-Homogeneous and Homogeneous Mass
Transfer in Reverse Osmosis Membrane Processes. Desalination and Water Treatment. 51(34-
36): 6444-6458 (2013).
Hagley Library (2022). DuPont Permasep Products Records 1969-2003, Hagley Library, 298 Buck
Road, Wilmington, DE 19807.
Harn R/O Systems, Inc. (2022). Harn R/O Systems, Inc., a Division of Komline-Sanderson Corporation,
310 Center Ct., Venice, FL, 34285. Collected from the Web [https://blog.harnrosystems.com/
how-to-complete-a-silt-density-index-sdi-test-and-why-its-important].
Hydranautics (2022). Hydranautics, A Nitto Group Company, 401 Jones Road, Oceanside, CA 92058.
Collected from the Web [https://membranes.com/wp-content/uploads/Documents/TSB/
TSB113.pdf].
Jin, Y., J. Younggil, L. Hyunkyung, S. Hong (2015). Fouling potential evaluation by cake fouling index:
Theoretical development, measurements, and its implications for fouling mechanisms. Journal
of Membrane Science. 490: 57-64.
Khirani, S., R. Ben Aim, M.-H Manero (2006). Improving the measurement of the Modified Fouling
Index using nanofiltration membranes (NF–MFI). Desalination 191(2006)1–7.
Mosset, A., V. Bonnelye, M. Petry, and M.A. Sanz (2008). The sensitivity of SDI analysis: from RO feed
water to raw water. Desalination. 222: 17-23.
Nahrstedt, A. and J. Camargo-Schmale (2008). New insights into silt density index and modified
fouling index measurements. Water Science and Technology: Water Supply. 8(4): 401–411.
Rachman R.M., N. Ghaffour, F. Waly, G.L Amy (2013) Assessment of Silt Density Index (SDI) as
fouling propensity parameter in Reverse Osmosis (RO) desalination systems. Desalination and
Water Treatment 51: 1091-1103.
Ruiz-Garcia, A., E. Ruiz-Saavedra, S.O. Perez-Baez, J.E. Gonzalez-Gonzalez (2015). Evaluation of the
first nine years of operating data of a RO brackish water desalination plant in Las Palmas, Canary
Islands, Spain, Desalination and Water Treatment. 55: 2555-2561.
Salinas-Rodríguez, S.G. (2011). Particulate and organic matter fouling of SWRO systems:
Characterization, modelling and applications. CRC Press: Balkema, Delft.
Schippers, J. C., J.H. Hanemaayer, C.A. Smolders, and A/ Kostense (1981). Predicting flux decline or
reverse osmosis membranes. Desalination, 38, 339-348.
Schippers J.C. and J. Verdouw (1980). The modified fouling index, a method of determining the fouling
characteristics of water. Desalination 32: 137-148.
Schippers, J.C., S.G. Salinas-Rodriguez, M.D. Kennedy and S. Boerlage (2014). Why MFI is edging SDI
as a fouling index, Desalination and Water Reuse. May-June 2014: 28-32.
Sioutopoulos, D.C. and A.J. Darabelas (2012). Correlation of organic fouling resistances in RO and UF
membrane filtration under constant flux and pressure. Journal of Membrane Science. 407 (408):
34-46.
Yiantsios, S.G., D. Sioutopoulos, A.J. Karabelas (2005). Colloidal fouling of RO membranes: an
overview of key issues and efforts to develop improved prediction techniques. Desalination 183:
257-272.

153
 doi: 10.2166/9781789062977_0155

Chapter 7

Modified Fouling Index

(MFI-0.45)

Sergio G. Salinas-Rodriguez, IHE Delft, The Netherlands

Vanida A. Salgado-Ismodes, IHE Delft, The Netherlands

Almotasembellah Abushaban, UM6P, Morocco

The learning objectives of this chapter are the following:

• Define the theoretical principles of MFI-0.45 constant pressure and its prediction
model

• Describe the MFI-0.45’s testing set-up, testing protocol and calculation procedure

• Illustrate the application of the MFI-0.45 method

7.1 INTRODUCTION

The modified fouling index (MFI-0.45) was developed to overcome the limitations of the silt
density index (SDI). The MFI-0.45 has been standardized by ASTM in its method ‘Standard
Test Method for Modified Fouling Index (MFI-0.45) of Water’ (ASTM D8002 - 15, 2015)
and is based on filtration of feed water through a 0.45 μm microfiltration membrane filter
in dead-end mode at constant pressure (207 kPa), and importantly, it is based on the cake
filtration mechanism (Schippers and Verdouw, 1979, Schippers and Verdouw, 1980).

This MFI-0.45 test can be used to assess the fouling potential of reverse osmosis (RO) /
nanofiltration (NF) feed water due particulate matter and is applicable to low and high
turbidity waters. Similar to the SDI, the ASTM stipulates that this test is not suitable for
assessing ultra-pure water or effluents from most RO and ultra-filtration (UF) systems.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Some of the applications of the MFI-0.45 method are the following:

• It can serve as an indication of the quantity of particulate matter.
• It can be used to determine effectiveness of various treatment processes used to remove
particulate matter.
• It can be used to assess the clogging potential of water before infiltration in wells.

7.2 THEORY PARTICULATE FOULING

The flow through a porous medium like a membrane, based on Darcy’s law, can be described
by:
dV
Qw = = P ×K w × A Eq. 1
dt
where:
Qw = permeate flow (m3/hr)
V = total filtered volume water (permeate) (L or m3)
t = time (hour, minute, second)
∆P = differential pressure (pressure feed - pressure permeate)
Kw = permeability constant of porous media for water (m3/m2-s-bar)
A = surface area of the membrane(s) (m2)
In membrane technology, flux is defined as the ratio of the permeate flow and surface area of
the membrane. It is expressed as:
Qw 1 dV
J= = × = P× Kw Eq. 2
A A dt
Frequently, the concept of resistance (R) is used, instead of permeability:
1
Kw = Eq. 3
×R
T

Where: η is the viscosity of the water and RT is the total resistance [sum of membrane
resistance (Rm), pore blocking (Rp) and cake formation (Rc)].

RT = Rm + Rb + Rc Eq. 4

Replacing Eq. 3 and Eq. 4 in Eq. 2:

1 dV 1 P
J= × = × Eq. 5
A dt Rm + Rb + Rc

Permeability of the clean filter media (Rm) is a function of filter properties such as filter
thickness (Δx), surface porosity (ε), pore radius (rp), and tortuosity (τ) and can be defined
using Poiseuille’s Law:
× x×
Rm = 2
Eq. 6
× rp

156
 Chapter 7

Pore blocking resistance of the membrane (Rb) is defined as the restriction of flow during
filtration due to the particles deposited inside pores or blocking the pores entry. The cake
resistance (Rc) component in (membrane) filtration can be defined following the Ruth
equation (Ruth, et al., 1933), using the concept of ‘specific cake resistance’ per unit weight
(α) (Equation 10). Ruth showed that the resistance of the cake formed during constant
pressure filtration is proportional to the amount of cake deposited at the filter medium
provided the retention of particles and α are constant. Cake resistance is defined as:

V
Rc = I × Eq. 7
A
and the fouling index (I) is:

I= ×C
b Eq. 8

Where: I is a measure of the fouling characteristics of the water (m-2). The value of I is a
function of the nature of the particles and is proportional to their concentration. Cb is the
concentration of particles per unit volume of filtrate (e.g., mg/L) and α is the specific cake
resistance per mg cake per m2 membrane (m3/mg/m2).

The specific cake resistance is constant for incompressible cakes under constant pressure
filtration. Carman (Carman, 1937, Carman, 1938) derived Equation 9 for the specific
resistance of a cake composed of spherical particles of diameter dp from the Kozeny equation
including a factor for tortuosity of the voids within the cake. According to this relationship
a reduction in the porosity of the cake (ε) or a decrease in particle diameter size (dp) increases
the specific resistance of the deposited cake.

180 × (1 )
c
= 2 3
Eq. 9
p
×dp ×

Combining Eq. 7 and Eq. 5 and integrating at constant ΔP from t = 0 to t = t, assuming time
independent permeability and uniform porosity characteristics throughout the depth of the
cake (i.e., no compression of the cake), results in the well-known filtration equation:
t ×R ×I
= m
+ ×V Eq. 10
V P × A 2 × P × A2
Where Rm is considered constant throughout the filtration period. Equation 10 gives a
straight line when t/V vs. V is plotted which is used to test the formation of cake filtration.
Carmen defined the gradient of the line as:
dt / dV ×I
slope = tan = = Eq. 11
dV 2 × P × A2
The fouling index (I) can then be determined from the slope of the linear region in the plot
of time/volume vs. volume, which corresponds to cake filtration as illustrated in Figure 1.
The MFI is calculated considering the minimum I value. From this slope, the I value of the
water can be calculated from the actual testing conditions (η, A, ΔP).

157
Experimental Methods for Membrane Applications

Pore Cake
blocking Cake filtration compression

slope = tan α

t/V (s/L)
V (L)

Figure 1 Filtration curve t/V versus V. Adapted from Schippers (1989)

The modified fouling index MFI-0.45 can be obtained after normalizing to reference
conditions: pressure (∆Po) = 2 bar, membrane area (Ao) = 13.8×10-4 m2, water temperature
through water viscosity η at 20 °C.
×I
MFI = 0
Eq. 12
2 × P0 × A02

Reference conditions for normalization of MFI values

ΔPo = 2 bar = 200 kPa
Ao = 13.8×10-4 m2 (42 mm effective diameter of a 47 mm diameter filter)
At temperature 20 ˚C, the viscosity (ηo) is = 0.001 Ns/m2.
MFI is expressed in units of s/L2. By doing this the results will be in the same range order
of magnitude of SDI in the range 2 to 3 (Schippers, et al., 2014).
Replacing the reference values in the MFI formula helps us to find the conversion factor
of MFI into I, as follows:
MFI = (0.001 Ns/m2) × I / [2 × 200,000 N/m2 × (13.8×10-4 m2)2]
MFI (s/m6) = 13×10-4 × I (m-2)
MFI (s/L2) = 13×10-8 × I (m-2)
or
I (m-2) = 7.68×108 × MFI (s/L2)

An alternative method for calculating MFI is based on the equation:

dt × Rm ×I
= + ×V Eq. 13
dV P P ×A2
In this case, the calculated slope is two times higher than in Eq. 11. Thus, this factor needs
to be considered in the MFI calculation. This alternative approach has the advantage that
possible errors in time and volume at the start of the test will not influence the calculated
slope during the course of the test. Nevertheless, very accurate pressure regulator and
volume (or flow) measurement devices are needed to obtain accurate MFI values.

158
 Chapter 7

7.3 MEASURING MFI-0.45

7.3.1 Filtration set-up and materials

A concept schematic of the filtration apparatus is presented in Figure 2 as it was initially
proposed by Schippers and Verdouw (1979). Figure 3 shows the schematic of the filtration
set-up as available at the laboratory of IHE Delft.

All parts of the filtration set-up that are in contact with water should be made of high-
quality stainless or plastic to prevent contamination by corrosion.

Valve PI Valve

PC
PI

PI = pressure indicator
PC = pressure controller Air relief
Filter holder valve

Figure 2 Schematic of the filtration apparatus for measuring MFI-0.45 as initially proposed by
Schippers and Verdouw (1979)

The key components of the experimental set-up are: air compressor, pressure vessel,
pressure regulator, membrane filter holder, pressure transducer, thermometer, electronic
balance, computer and a three-way valve. A three-way valve is used to connect the feed tube
with the membrane holder and at the same time with the pressure transducer. The filtration
set-up needs to be verified for correct pressure / flow / weight readings, stable constant
pressure filtration, no leakages, and avoiding air/bubbles trapped in the system (tubing,
feed side of filter holder).

Pressure
transducer
Pressure
regulator 3-way
and gauge valve
Filter
holder
Compressed
air

Feed water Electronic Computer

reservoir mass balance

Figure 3 Example of laboratory apparatus for measuring MFI-0.45 at constant pressure with a
pressure vessel.

159
Experimental Methods for Membrane Applications

7.3.1.1 Membrane filters

The recommended membrane filters for the MFI-0.45 test are the same as the ones
recommended for the SDI test.

The membrane filters need to be white of colour, hydrophilic, with a mean pore size of 0.45
μm, mixed cellulose nitrate and mixed cellulose ester are allowed. The diameter of the filter
depends on the size of the membrane filter holder (25 mm or 47 mm).

Table 1 ASTM recommended membrane filter properties for MFI-0.45 and SDI tests
MFI-0.45 SDI
Property ASTM D8002 - 15 (2015) ASTM D4189 - 14 (2014)
Hydrophobic/ hydrophilic Hydrophilic
Material Mixed cellulose nitrate (50–75 Mixed cellulose nitrate (50–75
%), mixed cellulose ester* (MCE) %), cellulose acetate
Mean pore size 0.45 µm
Thickness 115-180 µm
Bubble point 179-248 kPa
Pressure Not mentioned 91.4 – 94.7 kPa
Pure water flow 25-50 s / 500 mL (10-20 mL/s)
Constant pressure 200 kPa (difference across filter) 207 kPa (gage pressure)

* The standard method mentions acetate but the acronym corresponds to mixed cellulose ester.

Considering that for a range of pressures (91.4-94.7 kPa) the water flow should be around
25-50 seconds per 500 mL, for a nominal filter diameter of 47 mm, the recommended
permeability of the filters at 20 °C ranges between 21,911 to 45,405 L/m2/h/bar; or in
terms of membrane resistance (Rm) varies between 7.93×109 and 1.64×1010 m-1.

7.3.1.2 Filter holder

Filter holders designed to withstand pressures of 3.5 bar are recommended for the test.
The filter holder should be equipped with a device for releasing air. The ideal filter support
surface (placed at the bottom of the filter holder) is a highly porous one so that it does not
influence the effective membrane area during the filtration procedure. Cytiva’s Whatman
(WHA-10461000 FP 025/1 Polyethersulfone) filter holder for 25 mm filter can be used,
although it does not have a fully porous support plate.

It is recommended to measure the head loss across the filter holder and correct for it in the
feed side to make sure that the feed pressure of 2 bar is effective in the test.

The filter holder should fit perfectly the size of the membrane filter. Filters are normally
available as 25 mm or 47 mm diameter. A smaller size will allow less volume of water
needed for the test, easing the sampling, transport and storage of the samples.

160
 Chapter 7

7.3.1.3 Feedwater reservoir

The water to be tested needs to be transferred to the feed reservoir which is a pressure
vessel like in Figure 4. Commercially there are various volume capacities. The minimum
recommended volume for the test is 3.8 L but a larger capacity one is preferred, especially
when clean water is tested. The required pressure is achieved by applying compressed air.
Nitrogen gas can also be used in case compressed air is not available.

Figure 4 Feedwater reservoir or pressure vessel for laboratory measurement (Sterlitech, 2023)

7.3.1.4 Electronic mass balance

The feed water that passes through the membrane filter needs to be collected in a beaker
set on an electronic balance, for instance a Sartorius model Entris with 0.1 g accuracy. The
capacity of the scale should be at least the same volume as the feedwater reservoir. The
scale needs to connect with a computer via a USB-C or RS232 port to acquire the permeate
weight from the balance. Considering the density of water, 1 gram equal to 1 millilitre.

Figure 5 Entris® II Advanced Line Precision Balance 12,200 g|100 mg (Sartorius, 2023)

7.3.1.5 Software for data acquisition

Data sets of filtrate weight collected over time need to be recorded and imported into an
MS Excel spread sheet by data acquisition software such as: WinWedge Standard (Taltech,
2023). The sampling frequency can be adjusted according to requirements for calculation
(e.g., between 2 and 10 seconds).

The MS Excel spreadsheet for data acquisition can be adapted to include a graph of MFI
versus time in order to set the criteria of filtration time; for instance, if the change of slope of
the MFI in 5 minutes is less than 5% per minute then the test should be finished.

161
Experimental Methods for Membrane Applications

7.3.1.6 Pressure regulator and gauge

The pressure in the feed water reservoir and just in front of the membrane filter holder
needs to be kept at a constant level of 2 bar. This can be set by means of a manually adjustable
pressure sustaining valve (Figure 6) which includes a pressure gauge (Figure 6). This valve is
manually adjustable by means of a turning wheel (0.02 bar step). The accuracy of adjustment
is depending on readings from the gauge.

Figure 6 Pressure regulator (0.02 bar step) (Festo, 2023a) and pressure gauge in the range 0-2.5
bar with 0.05 bar accuracy (Festo, 2023b)

7.3.1.7 Pressure transducer

The pressure transduce is commercially available (PXM409-001BGUSBH, Omega, USA),
made of 316L steel typically considered the minimum grade for use in marine environment.
The operational pressure range is 0-10 bar with a maximum deviation of 0.08 %.

The pressure transducer has the function to monitor that the pressure remains constant
while the test takes place. The location of this device is just in front of the filter holder.

Figure 7 Photo of the Omega pressure transducer with USB output (Omega, 2023)

The signal of the transducer can be read by the computer with help of software such as
‘Digital transducer application’ provided by the manufacturer of the pressure transducer
(Omega, 2023). The pressure transducer is connected to a computer via a USB port.

7.3.1.8 Non-plugging water

For each test a new membrane filter needs to be used. Although the ASTM does not mention
cleaning the filters before its use, it is recommended to measure the membrane resistance
of the filters and clean them by passing 0.5-1 L non-plugging water (e.g., lab water after
0.2 μm filtration) or preferably ultra-pure water (UPW) through them. UPW is water free
of colloidal particles, ions and organic matter.

162
 Chapter 7

7.3.2 MFI-0.45 testing procedure

The following procedure is recommended in the standard method (ASTM D8002 - 15, 2015):
1. Set the pressure regulator at 2 bar. The pressure must remain constant during the test (±
1%).
2. Measure temperature of the water.
3. Flush the water to be tested through the apparatus to remove contaminants.
4. Open the membrane filter holder and place a 0.45 μm membrane filter (25 mm or 47
mm in diameter) on the support plate of the holder. It is recommended to handle the
membrane filters only with tweezers and avoid touching them with fingers. Use only
filters that are packed in the same orientation.
5. Make sure the membrane holder’s O-ring is in good condition and properly placed.
6. Close the filter holder.
7. Release air by opening the pressure relief valve and open the small air relief valve on top
of the filter holder (in case the filter holder includes an air relief valve or make sure that
air is not trapped in the system).
8. Close the relief valve and start recording flow (and preferably pressure as well).
Recommended time interval for data acquisition is every 5 seconds.
9. Run the test for 30 minutes to 60 depending on the rate of flow decline. This depends
on the volume of the feed water container.
10. After completing the test, the membrane filter may be retained for future reference or
analysis.

At the start of the test, the initial flow needs to be controlled to quickly identify if the filter
might have been cracked or misplaced in the preparation. The initial flow (or permeability)
should be within 10 % of the flow recorded with ultra-pure water (UPW).

It is recommended to report together with the MFI value, the water temperature, the
membrane filter material and manufacturer.

7.3.3 MFI-0.45 calculation procedure

The ASTM recommends the following procedure for calculating the MFI-0.45. The volume
and time dataset should be plotted as t/V versus V to identify the linear relation between
resistance and the cumulative filtered water volume (see Figure 1, Eq 14), for which the
slope (b) describes the fouling tendency of a given water (Equation 15).

t 1 × Rm ×I
= = + ×V Eq. 14
V Qavg P A 2 P × A2
× ×

and

×I dt / dV
b= = Eq. 15
2× P × A2 dV

where: t is the filtrations time (s), V is the cumulated permeate volume (L), Qavg is the
average flow rate, η is the water viscosity (Ns/m²), I is the fouling index (m-2), Rm is the
membrane resistance (m-1), ΔP is the applied transmembrane pressure (bar or N/m²), and
A is the membrane surface area (m²).

163
Experimental Methods for Membrane Applications

The slope of the line (b) has been defined as the MFI. This value needs to be normalized to
reference conditions of ΔP0 (200 kPa, 2 bar), η (η20°C), and A0 (13.8×10-4 m2 equivalent
to 47 mm diameter membrane filter). The term I represents the fouling index for the
propensity of particles in water to form a layer with hydraulic resistance.
×I
MFI = 0
Eq. 16
2× P0 × A02

In conducting the MFI test, the MFI can be determined from the gradient (tan α, b, (dt/dV)/
dV) of the linear region of minimum slope determined in (a plot of) t/V versus V. Then
normalizing this slope to standard conditions of temperature (Tcorr), pressure (Pcorr) and
membrane area (Acorr) yields MFI as shown in Equation 17.
t t
d d
P
MFI = 20˚C
× × V = (Tcorr )× ( Acorr )× V Eq. 17
T
P0 dV dV

Example – Calculation of MFI-0.45 value

An MFI test was performed on raw seawater at 2 bar, at 20 ˚C ( = 0.001 Ns/m2) with a
filter of 25 mm nominal diameter (21 mm effective diameter). The results of the MFI
test are presented in the following figure. Calculate the MFI value.

1,200

1,000

800

600

400

200

t/V (s/L) 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
V (L)

The slope of the linear region (marked by the two red lines) can be calculated.
Slope = 140 s/L2.
The effective membrane area = π × (21/2/1000)2 = 0.0003464 m2.
Now we can calculate the MFI-0.45 value.
2
P A
MFI = 20˚C × × tan
T
P0 A0

By replacing the reference values and the ones used in the test, we have:
MFI = (0.001 Ns/m2 / 0.001 Ns/m2) × (2 bar / 2 bar) × (0.0003464 m2 / 0.00138 m2)2
× 140 s/L2
MFI = ~11 s/L2

164
 Chapter 7

7.4 MEMBRANE PROPERTIES OF COMMERCIAL MEMBRANES

Salinas Rodriguez et al., (2019) studied three 0.45 μm membrane filters as reported in
Table 2. The mixed cellulose nitrate (NC) filter had the highest permeability in comparison
to cellulose acetate (CA) and nylon (NN). The measured permeabilities suggest that all the
tested filters were according to the recommendations of the ASTM.

Table 2 Filter properties

Thickness1, Bubble point1, Permeability2,
Code Filter material µm bar L/m2/h/bar
CA Cellulose acetate3 106 >2.4 31,020 ± 926 (3%)
NN Nylon 6.6 144-170 2.2-2.5 22,764 ± 579 (3%)
NC Mixed cellulose nitrate3 150 >2.1 44,438 ± 1,259 (3%)

1 Information from manufacturer, 2 Measured using filter holder FH4 n=10,

3 ASTM recommended material.

Top view images of the membrane filters were obtained using a scanning electron microscope
(Jeol, JSM-6010 LA). Figure 8 shows similarities in the pore morphology among the three
filter materials. NN shows a smoother surface than the other NC and CA. The shape of the
pores is not well defined and they are not homogenously distributed over the surface.

Figure 8 SEM top view images of the three tested filters at different magnifications (1,000x for left
images, 10,000x for top right images, and 6,000x for bottom right image). Adopted from
Salinas Rodriguez, et al. (2019)

165
Experimental Methods for Membrane Applications

7.5 EFFECT OF FILTER MATERIAL ON MFI-0.45

Figure 9 presents the MFI-0.45 values of Delft canal water (DCW) and Formazin solution
(NTU= 15) measured with three different membrane materials. Filter properties influencing
the measurements are the mean pore size, pore size distribution, surface porosity, thickness,
tortuosity, surface charge.

252 300
Formazin
DCW
250

100

134
150
97
77 72 74 50

50

MFI-0.45 (s/L2) 0
CA NN NC

Figure 9 MFI-0.45 values measured with various filter materials (n=10) for Delft canal Water
(DCW) and Formazin NTU = 15.

For DCW, MFI0.45 value is 252 s/L2 for the CA, 134 s/L2 for NN, and 74 s/L2 for NC. The
relative error for CA, NN, and NC is 11 %, 8 %, and 6 %, respectively. NC filters show the
lowest variation for MFI-0.45. While for Formazin, the measured MFI-0.45 values are
similar for NN and NC (77 s/L2 for NN and 72 s/L2 for NC, with low relative errors 0.7 %
and 0.8 %, respectively). However, a larger MFI-0.45 value is obtained with the CA filter (97
s/L2 and 1.3 % relative error). The thinner thickness and clean water permeability of the CA
cannot directly explain the higher MFI0.45 value compared to the NC and NN filters when
testing Delft canal water. It is possible that the surface charge of the filter and the interaction
with the particulates and organic matter present in DCW influence the measured fouling
potential.

Membrane resistance (Rm) can be used as a general indicator of the membrane properties
(e.g., membrane porosity, pore size, and thickness). In the same mentioned study, the NC
membranes have the lowest Rm values (9.35×109 m-1 ± 5.4 %) followed by CA membranes
(1.39×1010 m-1 ± 6.6 %) and NN membranes (1.86×1010 m-1 ± 4.5 %) with the highest Rm
values. No correlation between MFI-0.45 and Rm was observed (Figure 10).

166
 Chapter 7

300
CA
NN
250
NC

100

150

100

50

MFI-0.45 (s/L2) 0
0.0E+00 5.0E+09 1.0E+10 1.5E+10 2.0E+10 2.5E+10
Rm (m-1)

Figure 10 MFI-0.45 as a function on Rm (n=10). Delft canal water.

7.5.1 Effect of membrane support holder

Nahrstedt and Camargo (2008) studied the effect of filter support on SDI and MFI values.
They reported that the filter holder had a strong influence on the obtained SDI values. The
type of filter holder will determine the effective membrane area during filtration. A difference
of more than 100 % was found for the same feedwater depending on the membrane holder
used. A similar conclusion was drawn by Escobar et al., (2009) when testing a Millipore
holder and a Pall membrane holder.

Salinas et al., (2019) also studied the effect of the filter holder, filter material in SDI and MFI
tests for seawater and fresh water samples. In this study they proposed a correction for the
effective filtration area by quantifying the actual effective membrane area after filtering a
solution of powder activated carbon. This is illustrated in Figure 11 where MFI-0.45 values
were measured for Delft Canal Water making use of 4 different filter support holders. Filter
holder (FH) 7 is a porous support type, FH 4 (Schleicher & Schuell) and FH2 (Sartorius) and
FH1 (Whatman) are concentrically channelled.

600
Without area correction
492
With area correction 468 500
393
400

281
97 97 243 300
216
200

100
2
MFI-0.45 (s/L ) 0
FH7 FH4 FH2 FH1

Figure 11 MFI-0.45 values measured with various filter holders (n=10) for Delft canal water with a
cellulose acetate filter. FH = filter holder.

The correction factor is the ratio of the actual effective membrane area over the initially
available membrane area. This is why a porous support for the membrane filter is
recommended for measuring the MFI-0.45 of water.

167
Experimental Methods for Membrane Applications

By correcting for the effective filter area, the MFI-0.45 results obtained with the different
filter holders (Figure 11) are closer to each other (247 s/L2 ± 10.8 %) in comparison with the
average without considering the area effect (400 s/L2 ± 27.6 %). In the MFI formula, the area
plays a significant role, hence the large variation in MFI values for filter holders without area
correction. Additionally, in the MFI the flow rate influences greatly the fouling potential of
a water sample, so any effect that increases the flow rate through the membrane (like the
channels in the filter support plates that reduce the effective filter area) will increase the
fouling load of the membrane and consequently the measured MFI-0.45 will be higher.

7.6 APPLICATION: WATER QUALITY MONITORING OF NORTH SEA WATER

North Sea water before and after filtration was monitored over almost one year. A summary
of the properties of the water is presented in Table 3. The pH and electrical conductivity
values remained fairly stable during the period, pH around 8 and EC values of 48-49 mS/
cm for the samples. Turbidity measurements ranged between 0.1-1.0 NTU for filtered
North Sea water and between 0.9-45 NTU for raw North Sea water. The elevated turbidity
values in the raw water can be attributed to high ocean tides and storms in the area.

Table 3 Summary of seawater quality properties in the study period

Filtered
Raw (glass media, EBCT 30 min)
Parameter North Sea water North Sea water
pH 8.0 ± 0.3
Turbidity, NTU Min = 0.9; Max = 45 Min = 0.1; Max = 1.0
Avg = 10.5 ± 11.0 Avg = 0.4 ± 0.2
Elec. conductivity, mS/cm 48.1 ± 1.6
DOC, mg/L 2.1 ± 0.5
SUVA, L/mg/m 1.8 ± 0.6
Total algal count, cell/mL Min = 11, Max = 1,039 -
Chlorophyll-a, µg/L < 5 (bdl) till end of March.
In May ~7.5 µg/L

bdl = below detection limit

Algal cell counts and chlorophyll-a are primarily used to indicate algal bloom occurrence.
The chlorophyll-a concentrations measured in the period were below the detection limit of
the method (LOD < 5 μg/L). According to the results presented in Figure 12, the minimum
and maximum algal count concentrations measured for raw North Sea water are 11 cell/
mL and 1,039 cell/mL, respectively. These values are considered normal due to the low
temperature during the period until the end of April when the values rapidly increased, most
likely due to a mild algal bloom. An algal cell counts lower than 1,000 cell/mL indicates that
there is no algal bloom. Based on the high algal cell numbers at the end of April until mid-
May, it can be concluded that a mild algal bloom event took place at the testing location.

168
 Chapter 7

From October 2016 to July 2017, the MFI-0.45 values were measured for both raw and
filtered North Sea water (Figure 12). SDI15, SDI10, and SDI5 could not be measured due to
clogging of the filter, but the SDI3 values were reported ranging between 6-26 %/min for
filtered and 9-28 %/min for raw North Sea water.

300 1,200
MFI-0.45 Raw NSW
MFI-0.45 Filtered NSW
250 1,000
Algal cells

Algal cell counts (Cells/mL)

200 800
MFI-0.45 (s/L2)

150 600

100 400

50 200

0 0
17

17
7

7
16

16

01

01

01
20

20
20

20

-2

-2

-2
6-

3-
7-

6-

15

25

14
3-

8-
-0

-2

1-

4-

6-
10

11

Figure 12 MFI-0.45 values of raw NSW and filtered NSW versus algal cell counts (n=3). CA filter.

MFI-0.45 measured for filtered North Sea water is in the range of 12-170 s/L2 and for the
raw seawater sample between 20-310 s/L2, Figure 12. Consistently higher MFI-0.45 values
were obtained for the raw NSW (up to 8x higher) that has a higher particles content as
compared to the filtered NSW.

7.7 MONITORING OF MFI-0.45 IN A FULL-SCALE DESALINATION PLANT

A full-scale seawater desalination plant was monitored with MFI-0.45 over the period of 1
week with daily measurements. The treatment process consists of the following treatment
steps (Figure 13): Screens (3 mm) > 2 mg/L FeCl3 inline coagulation > 1st dual media
filtration (DMF1, 11-14 m/h) > 2nd dual media filtration (DMF2, 17-19 m/h) > 5 μm
cartridge filter (CF) > RO membrane (R = 43%) > Remineralization.

Shock FeCl3 Dual media filter Dual media filter RO membrane

chlorination (DMF) - 1st (DMF) - 2nd (1st pass)
Cartridge
filter (CF)

Open intake

Figure 13 The treatment processes of the 120,000 m3/d desalination plant.

169
Experimental Methods for Membrane Applications

MFI-0.45 of the influent seawater ranged between 13 and 26.5 s/L2 with an average of 19.8
s/L2. About 82% reduction in MFI was noticed through DMF1 where MFI declined from
19.8 s/L2 to 3.8 s/L2. A further reduction in MFI was recorded through DMF2 where the
reduction percentage increased to about 91%. MFI values after DMF2 ranged from 1.3 to 2.5
s/L2 while its value after CF ranged from 0.8 to 2.5 s/L2.

30
Influent
25
DMF1
DMF2
20
CF
15

10

MFI-0.45 (s/L2) 0

Figure 14 MFI-0.45 through the pre-treatment. Adapted from Abushaban, et al. (2021).

MFI after CF (2.1 s/L2) was slightly higher than the MFI value after DMF2 (1.8 s/L2). The
possible reason for the higher MFI after CF might be the detachment of particles from the
old CF.

The particles, larger than 1 μm, through the pre-treatment processes were counted using
particles counter device available at the lab of the plant. Particles counter device reports the
fractionation of particle sizes between 1 and 20 μm.

30,000
Influent
25,000
DMF1
DMF2
20,000
CF
15,000

10,000

5,000

>1 µm particle (count/mL) 0

Figure 15 Particle larger than 1 μm counts per mL along the treatment process. Adapted from
Abushaban (2019).

The total particles in the influent ranged from 18,777 to 26,561 CNT/mL with an average of
22,700 CNT/mL. The total counts decreased 95.2% through the DMF1 where the particles
count declined from 22,700 to 1,080 CNT/mL. This shows that DMF1 has a significant role
in the removal of particles which has the highest removal percentage. DMF2 reduced the
particles concentration from 1,080 to 470 CNT/mL.

170
 Chapter 7

7.8 REFERENCES

Abushaban A (2019) Assessing Bacterial Growth Potential in Seawater Reverse Osmosis Pretreatment:
Method Development and Applications CRC Press
Abushaban A, Salinas-Rodriguez SG, Pastorelli D, Schippers JC, Mondal S, Goueli S, Kennedy MD
(2021) Assessing Pretreatment Effectiveness for Particulate, Organic and Biological Fouling in a
Full-Scale SWRO Desalination Plant. Membranes 11: 167
ASTM D4189 - 14 (2014) Standard Test Method for Silt Density Index (SDI) of Water ASTM
International, West Conshohocken, PA.
ASTM D8002 - 15 (2015) Standard Test Method for Modified Fouling Index (MFI-0.45) of Water
ASTM International, West Conshohocken, PA.
Carman PC (1937) Fluid flow through granular beds Trans Instn Chem Engrs 15: 32-48
Carman PC (1938) Fundamental principles of industrial filtration (A critical review of present
knowledge). Trans Instn Chem Engrs 16: 168-188
Escobar L, Sellerberg W, Sanchez D, Pastrana F, Wachinski A (2009) Detailed analysis of the silt density
index (SDI) Results on desalination and wastewater reuse applications for reverse osmosis
technology evaluation. Paper presented at the International forum on marine science and
technology and economic development - Asia-Pacific Desalination conference, Qingdao, China,
8-10 July 2009 2009
Festo (2023). Precission pressure regulator https://www.festo.com/gb/en/a/162834/. Cited 02
November 2023 2023
Festo (2023). Pressure gauge https://www.festo.com/gb/en/a/162834/. Cited 02 November 2023
2023
Nahrstedt A, Camargo Schmale J (2008) New insights into SDI and MFI measurements. Water Science
and Technology: Water Supply 8: 401-412 DOI https://doi.org/10.2166/ws.2008.087
Omega (2023). USB Output Pressure Transducer https://www.omega.nl/pptst/PXM409-USBH.
html. Cited 27 October 2023
Ruth BF, Montillon GH, Montonna RE (1933) Studies in Filtration I. Critical Analysis of Filtration
Theory. Industrial & Engineering Chemistry 25: 76-82 DOI 10.1021/ie50277a018
Salinas Rodriguez SG, Sithole N, Dhakal N, Olive M, Schippers JC, Kennedy MD (2019) Monitoring
particulate fouling of North Sea water with SDI and new ASTM MFI0.45 test. Desalination 454:
10-19 DOI https://doi.org/10.1016/j.desal.2018.12.006
Sartorius (2023). Entris II Precision Balance https://shop.sartorius.com/nl/p/entris-ii-advanced-
line-precision-balance-12200-g100-mg-external-adjustment/BCA12201-1S. Cited 02
November 2023 2023
Schippers JC (1989) Vervuiling van hyperfiltratiemembranen en verstopping van infiltratieputten
Keuringinstituut voor waterleidingartikelen KIWA N.V., Rijswijk
Schippers JC, Salinas Rodriguez SG, Boerlage SFE, Kennedy MD (2014). Why MFI is edging SDI as a
fouling index. Desalination & Water Reuse: 28-32
Schippers JC, Verdouw J (1979). De membraanfiltratie-index als kenmerk voor de filtreerbaarheid van
water. 12: 104-109https://edepot.wur.nl/398518
Schippers JC, Verdouw J (1980) The modified fouling index, a method of determining the fouling
characteristics of water. Desalination 32: 137-148
Sterlitech (2023). Pressure filtration vessels https://www.sterlitech.com/pressure-filtration-
vessel-741060.html. Cited 02 November 2023
Taltech (2023). WinWedge Standard https://www.taltech.com/winwedge/. Cited 02 November
2023 2023

171
 doi: 10.2166/9781789062977_0173

Chapter 8

Modified Fouling Index

Ultrafiltration (MFI-UF)
Constant Flux
Sergio G. Salinas-Rodriguez, IHE Delft, The Netherlands

The learning objectives of this chapter are the following:

• Define the theoretical principles of MFI-UF constant flux and its prediction model

• Describe the MFI-UF’s testing set-up, testing protocol and calculation procedure

• Illustrate the application of the MFI-UF method

• Compare the MFI-UF with other fouling indices

8.1 INTRODUCTION

Membrane fouling due to particulate matter has plagued reverse osmosis (RO) systems
since their first use in desalination and remains a persistent issue today. Contrary to the
silt density index or modified fouling index (MFI-0.45), the MFI-UF is not, at this date,
an ASTM method. Nevertheless, its application has grown in the last 15 years in assessing
particulate fouling in full scale and pilot installations such as in seawater and brackish water
desalination plants, surface water treatment facilities, and wastewater reuse plants. The
application of the MFI-UF is expected to keep growing as a monitoring tool in membrane
treatment systems.

Boerlage (2001) proposed the constant flux MFI with a 13 kDa PAN hollow fibre UF
membranes and applied it to monitor particulate fouling removal in fresh surface water
treatment in a plant of Amsterdam Water Supply and of PWN. Constant flux mode was
obtained by manually adjusting the pressure reducing valve of the MFI-UF (constant
pressure) equipment to maintain the required applied flux at a constant value.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Since then, the testing set-up was further developed to perform constant flux dead-end
filtration, accurate at low flux rates, making use of flat UF membrane filters of various pore
sizes. The MFI-UF at constant flux has been applied for many years in its current form as
developed by Salinas Rodríguez (2011) and recently further developed by Abunada (2023).
The MFI-UF test, as developed by Salinas Rodríguez (2011), has been applied in several
research studies assessing particulate fouling in pilot and full scale treatment plants such
as presented in Table 1. Tabatabai (2014) applied the MFI-UF to measure algal released
organic matter (AOM) removal rates from RO feed water with conventional coagulation
(coagulation / flocculation / sedimentation / filtration).

Villacorte (2014) studied the fate of transparent exo-polymer particles (TEP) through the
treatment processes of 4 RO plants, where he observed significant correlations between TEP
concentration and MFI-UF, suggesting that TEPs likely have a major role in the fouling of UF
pre-treatment systems and possibly in seawater RO systems, if not effectively removed by
the pre-treatment.

Dhakal (2017) investigated the fouling potential and fouling behaviour of algae and
AOM in ultrafiltration membranes. MFI-UF was linearly related to algal cell density and
chlorophyll-a concentration, biopolymer concentration, TEP, during the growth phase of
the algal species. Abushaban (2019) also applied the MFI-UF to monitor particulate fouling
removal along the pre-treatment of full scale seawater RO plants. Zhan et al. (2020) applied
the MFI-UF as a diagnostic tool of fouling potential of a RO system in ultra-pure water
(UPW) production. Abunada (2023) further developed the MFI-UF test by using a 5 kDa
membrane and improved the prediction of particulate fouling rate in RO systems.

Some potential applications of the MFI-UF at constant flux are:

• Assessing RO and NF feed water quality with respect to particulate fouling potential.
• Predicting rate of fouling RO and NF membranes due to particles.
• Assessing performance of pre-treatment systems in terms of particulate fouling removal.
• Characterizing MF/UF feed water in predicting development pressure increase during a
filtration cycle
• Verifying membrane integrity of MF/UF/RO/NF membrane systems.

Legend Table 1
P = Pilot; FS = Full Scale; UF = ultrafiltration, RO = reverse osmosis; GBF = glass beads filter; FSW = fresh surface
water; SW = seawater; TDWW = treated domestic wastewater; On = Onsite; Off = offsite; DMF = dual media
filtration; BW = beach well; DAF = dissolved air flotation; Flo = Flotation; SWRO = seawater RO; BWRO =
brackish water RO; NF = nanofiltration; MSS = Marine science station; IEX = ion exchange.

174
 Chapter 8

Table 1 Examples of pilot and full-scale treatment plants where MFI-UF constant flux was applied
On-, off-
Location Water Utility Type site Treatment Year
Toulon, SW Veolia P On UF-RO, (Salinas Rodriguez and
France DMF-RO Kennedy, 2009)
Tarragona, SW Dow P On UF-RO (Salinas Rodriguez and
Spain Kennedy, 2009, Salinas
Rodriguez, et al., 2010,
Salinas Rodriguez, et al.,
2010)
Kamperland, SW Evides P On MS – UF (Althuluth, 2009, Salinas
Netherlands – RO Rodríguez, et al., 2009,
Salinas Rodríguez, et al.,
2010, Wahyudi, 2010, Salinas
Rodríguez, 2011, AlShuaili,
2012)
Emmen, TDWW Nieuwater FS Off UF-BACF- (Ekowati, 2011, Ekowati,
Netherlands RO-EDI et al., 2014, Gulrez, 2021)
Baanhoek, FSW Evides FS Off Coag, UF, (AlShuaili, 2012, Rathnayake,
Netherlands Industry RO 2013)
Ras Al SW Fewa DAF- (Dhakal, et al., 2016)
Khaimah, UF-RO,
UAE DMF-RO
Ras Abu SW Kahramaa FS On DAF-DF- (Dhakal, et al., 2018)
Fontas, Qatar UF-SWRO-
BWRO
Aqaba, Jordan SW MSS Jordan - On Raw water 2018 - 2022
Botlek, FSW Evides FS Off DAF-IEX- (Dhakal, et al., 2020)
Netherlands Industry RO
Heemskerk, FSW PWN FS Off UF-RO (Bassa, 2021, Abunada,
Netherlands et al., 2023)
Andijk, FSW PWN FS Off, Coag + Sed (Bassa, 2021, Lindsay, 2022)
Netherlands + RSF +
GAC
Andijk, FSW PWNT P Off, On UF-RO (Bassa, 2021, Lindsay, 2022)
Netherlands
Bahia de SW Suez On BW-DMF- (Le Bouille, et al., 2021)
Palma, Spain CF-RO
Perth , SW Veolia FS On Coag- (Le Bouille, et al., 2023)
Australia DMF-RO
Katwijk, FSW Dunea P On MS, DMF, (Malaichami, 2023, Tahtouh,
Netherlands UF, RO 2023)
Kamperland, SW Zeeschelp Off GBF, UF (Mirzaei, 2023)
Netherlands
Baanhoek, FSW Evides P Off, On Coag, Flo, (Ofori, 2023, Qatae, 2023,
Netherlands DMF, GAC, Yameen, 2023)
RO and NF

175
Experimental Methods for Membrane Applications

8.2 THEORY PARTICULATE FOULING

The flow through a RO membrane can be described by:

dV
Qw = = ( P − π )× Kw × A Eq. 1
dt
where:
Qw = permeate flow (m3/hr)
V = total filtered volume water (permeate) (L or m3)
t = time (hour, minute, second)
ΔP = differential pressure (pressure feed - pressure permeate)
Δπ = difference osmotic pressure
(osmotic pressure feed – osmotic pressure permeate)
Kw = permeability constant for water (m3/m2-s-bar)
A = surface area of the membrane(s) (m2)
Qw/A = permeate flow through membrane surface area (m3/m2/h)
= filtration rate (m3/m2/h), used in rapid sand filtration
= flux (L/m2/h) used in membrane filtration
(ΔP - Δπ) = net driving pressure (NDP)

In membrane technology, flux is defined as the ratio of the permeate flow and surface area of
the membrane. It is expressed as:
Qw 1 dV
J= = × Eq. 2
A A dt

To simplify the equations, we assume that Δπ is negligible. This assumption is valid for low
salinity water only. Then,
1 × dV
J= = P × Kw Eq. 3
A dt
Frequently the concept of resistance (R) is used, instead of permeability:
1
Kw = Eq. 4
η × RT

Where: η is the viscosity of the water and RT is the total resistance [sum of membrane
resistance (Rm), pore blocking (Rp) and cake formation (Rc)].

RT = Rm + Rb + Rc Eq. 5

Replacing Eq. 4 and Eq. 5 in Eq. 3:

1× P
J= Eq. 6
η Rm + R + Rc

176
 Chapter 8

When we assume that pore blocking does not play a dominant role in RO, then fouling is
mainly due to cake formation. As a consequence:
1× P
J= Eq. 7
η Rm + Rc

Permeability of the clean filter media (Rm) is a function of filter properties such as filter
thickness (Δx), surface porosity (ε), pore radius (rp), and tortuosity (τ) and can be defined
using Poiseuille’s Law:
θ × x ×τ
R
= Eq. 8
ε × rp2

The cake resistance (Rc) component in (membrane) filtration can be defined following
the Ruth equation (Ruth, et al., 1933), using the concept of ‘specific cake resistance’ per
unit weight (α) (Equation 9). Ruth showed that the resistance of the cake formed during
constant pressure filtration is proportional to the amount of cake deposited at the filter
medium provided the retention of particles and α are constant. Cake resistance is defined as:
V
Rc = I × Eq. 9
A
and the fouling index (I) is:

I = α ×Cb Eq. 10

Where: I is a measure of the fouling characteristics of the water (m-2). The value of I is a
function of the nature of the particles and is proportional to their concentration. Cb is the
concentration of particles per unit volume of filtrate (e.g., mg/L) and α is the specific cake
resistance per mg cake per m2 membrane (m3/mg/m2).

The specific cake resistance is constant for incompressible cakes under constant pressure
filtration and can be calculated according to the Carman-Kozeny relationship (Equation 11)
(Carman, 1937, Carman, 1938). Carman (Carman, 1937, Carman, 1938) derived Equation
11 for the specific resistance of a cake composed of spherical particles of diameter dp from
the Kozeny equation including a factor for tortuosity of the voids within the cake. According
to the Carmen relationship a reduction in the porosity of the cake (ε) or a decrease in particle
diameter size (dp) increases the specific resistance of the deposited cake.

180 × (1− ε )
α = Eq. 11
ρ p × dp2 × ε 3

As porosity is to the power three, it plays a dominant role. The more compact a cake, the
higher the specific cake resistance, and therefore the higher the cake resistance and pressure
required to overcome this resistance.

177
Experimental Methods for Membrane Applications

RO plants typically operate at constant capacity and recovery. So, the flux is constant. When
membranes foul, the pressure needs to be increased, in order to maintain a constant capacity
(and flux) in the system. Rewriting Eq. 7:
1× Pt
J= Eq. 12
η Rm + Rc

Where: ΔPt is the pressure at time ‘t’ (which will increase). Rearranging Eq. 2 because flux
is constant:
V
= J× t Eq. 13
A

and substituting Eq. 13 in Eq. 9:

V
Rc = I × = I × J × t Eq. 14
A
This results in:
1× Pt
J= Eq. 15
η Rm + I × J × t

Rearranging the previous equation, we obtain:

Pt = η × Rm × J + η × I × J 2 × t Eq. 16

Thus, the increase of pressure ΔPt across the membrane is linearly proportional with time,
with the fouling index (I) and with the flux to the power two (J2). As a consequence, flux has
a very dominant effect on the development of ΔPt.

The modified fouling index with ultrafiltration membranes can be calculated from Eq. 17,
where the following reference conditions are considered: pressure (ΔPo) = 2 bar, membrane
area (Ao) = 13.8×10-4 m2, water temperature through water viscosity η at 20°C. Therefore,
conversion of MFI into I results in I (m-2) = 7.6×108 × MFI (s/L2).

ηo × I
MFI = Eq. 17
2 × Po × Ao2

The fouling index (I) can then be determined from the slope of the linear region in a plot of
pressure vs. time, which corresponds to cake filtration as illustrated in Figure 1. The MFI is
calculated considering the minimum I value.

178
 Chapter 8

Pore Cake compression

blocking Cake filtration depth filtration

P
I

minimum I value
P, I (Slope = tan α = ΔP/Δt)
Time

Figure 1 Schematic of the constant flux MFI test illustrating pressure development over time and
the identification of the minimum I value for MFI calculation. Adapted from (Salinas
Rodríguez, 2011)

Substituting the Carman-Kozeny Eq. 11 in Eq. 16 gives Eq. 18. This equation shows that MFI
is a function of the dimension and nature of the particles forming a cake on the membrane,
and directly dependent on particle concentration in water, as illustrated in Figure 2.

ηo × 90 × (1− ε ) ×Cb
MFI = Eq. 18
ρ p × d p2 × ε 3 × Po × Ao2

1,000,000
1 nm
100,00
3 nm
10,000
10 nm
1,000
30 nm
100
100 nm
500 nm 10

MFI (s/L2) 0
0,3 0.4 0.5 0.6 0.7 0.8 0.9
Cake porosity (ε)

Figure 2 MFI as a function of particle size and cake porosity. Adopted from (Salinas Rodriguez,
et al., 2021)

179
Experimental Methods for Membrane Applications

Equation 16 is valid for ‘dead-end’ filtration. In ‘cross-flow’ filtration only a part of the
particles will deposit on the membrane surface due to the shear-force of the cross-flowing
water. Therefore, ‘I’ has to be corrected with a deposition factor ‘Ω’. This factor is the fraction
of particles which actually deposit on the membrane surface (Ω ≤ 1). Then, Eq. 16 becomes:

Pt = η × Rm × J + η × Ω × I ×J 2 × t Eq. 19

The phenomenon, that the increase of pressure is proportional to (flux)2 explains, why
manufactures of spiral wound elements recommend lower design fluxes with feedwater
having a higher fouling potential.

8.2.1 Deposition factor

Only a fraction of the RO feedwater is forced to pass through the membrane in cross flow
filtration. This fraction of water depends on the recovery at which the RO unit operates.
In dead-end filtration all the particles bigger than the membrane’s pores will be retained
while in the case of cross-flow, only the fraction of water passing through the membrane
is affected and the associated fraction of particles may or may accumulate on the membrane
surface.

The deposition factor was first proposed by Schippers et al. (1980, 1981) in a model to
predict flux decline in reverse osmosis systems. It was defined as the fraction of particles
deposited, which are present in the water that passes through the RO membrane.
Then, we can obtain the deposition factor equation as function of recovery,
1 MFI conc 1
= + ×(1− )
ΩR MFI feed R Eq. 20

Or as function of concentration factor (CF),

1 MFI conc
× CF −
Ω Eq. 21
(CF − 1) MFI feed

Where the concentration factor is:

1
CF = Eq. 22
1− R
The formula above assumes that the particle rejection by RO membranes is 100 %.
There are possible scenarios from previous equations:
• Ω = 0 means MFIconc = MFIfeed ×CF No particles deposit
• Ω = 1 means MFIconc = MFIfeed All particles deposit
• Ω > 1 means MFIconc < MFIfeed All particles deposited + retention inside
pressure vessel (e.g., spacer)
• Ω < 0 means MFIconc > MFIfeed ×CF Particles might be removed/sheared off
inside the pressure vessel; earlier deposited
particles released; particles formed by bacteria;
particle size distribution influence results.

180
 Chapter 8

A positive deposition factor indicates particles are being accumulated on the membrane
surface as they pass through the system while a negative factor indicates the number of
particles in the concentrate exceeds the concentration in the incoming feedwater (Schippers,
1989, Boerlage, 2001).

8.2.2 The particulate fouling prediction model

The particulate fouling models to predict fouling developed by Schippers are based on the
assumption that particulate fouling on the surface of reverse osmosis (or nanofiltration)
membranes can be described by the cake filtration mechanism (Belfort and Marx, 1979,
Schippers, et al., 1981). The relationship between the MFI measured for a feedwater and the
fouling rate predicted for a RO system are presented below. The relationship assumes that
scaling, adsorptive blocking and biofouling do not contribute to the fouling observed on the
RO membrane.

Generally, RO desalination plants operate at constant flux to meet production requirements.

Changes in feedwater temperature are compensated for by adjusting feed pressure. Similarly,
fouling resulting in an increase in membrane resistance is compensated for by increasing
the feed pressure and hence net driving pressure (NDP). In this case, increase in the NDP
can be predicted through equation 19 which already includes the deposition factor Ω. By
rearranging this equation, we have:

( ND Pr − NDP0 r )
= J 2 ×η × Ω × I r Eq. 23
tr

And replacing I, we have

∆ND Pt 2 × MFI × ∆P0 × A02 ×Ω × J r2 ×ηr Eq. 24

=
tr η20˚C

Where the subscript ‘r’ refers to RO system conditions. Based on this equation, a theoretical
‘safe MFI’ or threshold MFI value for RO feed water can be calculated assuming e.g., an
allowable increase in NDP of 15%. To accurately predict the rate of fouling of RO and NF
membrane systems, the MFI-UF value should be measured at the same flux rate as the one
the membrane system is operating (Salinas Rodríguez, 2011, Salinas Rodríguez et al., 2015,
Abunada et al., 2022).

8.3 MEASURING MFI-UF CONSTANT FLUX

8.3.1 Filtration set-up and materials

The MFI-UF constant flux test was proposed initially using a 13 kDa hollow fibre PAN
membrane (Boerlage, 2004) and further developed to its current configuration (Salinas
Rodríguez, 2011, Salinas Rodríguez et al., 2015, Abunada, 2023) using flat 25 mm diameter
PES UF membranes in dead-end filtration (see Figure 3).

181
Experimental Methods for Membrane Applications

Pressure
Thermometer transducer

Syringe pump
Permeate

3-way Membrane
valve holder
Computer

Figure 3 Schematic of the MFI-UF constant flux filtration set-up

The key components of the set-up are: syringe (infusion) pump, membrane holder,
pressure transducer, thermometer, computer and a three-way valve. A three-way valve is
used to connect the syringe (water sample) with the membrane holder and at the same time
with the pressure transducer. The filtration set-up was verified for correct pressure readings,
stable constant flux filtration, no leakages, and avoiding air/bubbles trapped in the system
(tubing, feed side of filter holder).

The readings from the pressure sensor were verified with a second manometer (pressure
gauge) by filtering ultra-pure water (UPW) at various flux rates. A maximum 3 % difference
was observed, which is considered acceptable.

The flux rate was verified by monitoring the permeate weight over time with help of an
electronic balance. Several flux rates (10 - 400 L/m2/h) were tested and a maximum
difference of 2.8 % was observed between the expected and measured flux with the lower
flux rate.

Verification of leakages in the set-up was performed by pressurizing the system (up to 4 bar)
without allowing filtration and monitoring the pressure change over time. No leaks were
observed at pressures less than 4 bars over time.

The presence of air in the system is not desirable. To verify the effect of air trapped in
the system, air was intentionally introduced and filtration was allowed. Erroneous high-
pressure values were observed by the effect of air; this could be related to the bubble point
of the membranes or related to the compression of air that will produce erratic pressure
development.

The parts of the MFI-UF filtration set-up are discussed in the following sections.

8.3.1.1 Membrane filters

Flat circular (25 mm diameter) UF membrane filters are used in the test with molecular
weight cut-off (MWCO) in the range 5 – 100 kDa. Two membrane materials are available
from Millipore, namely: poly ether sulfone (PES) and regenerated cellulose (RC);
nevertheless, the PES membranes have lower membrane resistance than the RC ones and
thus are preferred for the test. The specifications of the membrane filters are summarized
in Table 2.

182
 Chapter 8

Figure 4 Package of Biomax 10 kDa, 25 mm, PES membrane filters from Millipore. Each package
contains 10 membrane filters. Source: Bassa, (2021).

Table 2 Specifications of the UF PES membranes

MWCO, Mean pore Surface
kDa Code size1, nm porosity2, % Rm, m-1
5 PBCC02510 8.0 0.6 6.57×1012 ± 1.0×1012 (15%)3
2.05×1012 ± 1.36×1011 (7%) 6
10 PBGC02510 9.2 2.9 9.65×1011 ± 1.58×1011 (16%) 3
9.44×1011 ± 1.08×1011 (11%)4
1.12×1012 ± 1.1×1011 (10%)5
1.02×1012 ± 4.94×1010 (5%)6
100 PBHK02510 10.6 10.5 3.24×1011 ± 4.17×1010 (12.9%)7

1 Abunada, et al. (2022) ; 2 (Andyar, 2021, Abunada, et al., 2022); 3 (Bassa, 2021); 4 (Lindsay, 2022);
5 (Qatae, 2023); 6 (Sithole, 2017); 7 (Salinas Rodríguez, et al., 2015)

Salinas Rodríguez et al., (2015) reported that the PES membranes are hydrophilic, have a
contact angle of 64° ± 4.2°, and are negatively charged above a pH value of 4.5.

Membrane filters are available in packages containing 10 specimens. Each package

is numbered with a batch code and production date. For preservation purposes, the
manufacturer coats the membranes with glycerine and sodium azide; this coating needs to
be removed before the filtration test is performed. A new membrane needs to be used per
test. In 2023, the cost per package of membranes is 208 euros (Merck, 2023).

Abunada et al., (2022) using suspensions of polystyrene particles (75 nm) studied the effect
of membrane surface porosity on MFI-UF and proposed a preliminary correction factor of
0.5 for the 10 kDa PES UF membrane filters. Nevertheless, properties of real waters may
yield different effects than the ones found for inorganic particles.

8.3.1.2 Constant flow pump

An infusion syringe pump was found to be very reliable and accurate for delivering constant
flow in the filtration process. The PHD Ultra pump from Harvard Apparatus (see Figure 5)
can deliver up to 6 bar of pressure and work in the flow range suitable for the test.

183
Experimental Methods for Membrane Applications

Figure 5 Syringe pump model PHD ULTRA used in the MFI-UF test (Harvard Apparatus, 2023)

The characteristics of the pump are:

• Flow rate range: 1.5 pL/min - 216 mL/min
• Flow rate accuracy: ± 0.25 %
• Force: 75 lb
• Dimensions (H/W/D/) / Weight: 18.42 × 30.48 × 21.59 cm / 4.5 kg
• Battery: None
• Brand: Harvard Apparatus, infusion only PHD Ultra syringe pump
• In 2022 the cost of the pump was about 3,800 euros.

8.3.1.3 Pressure transducer

The chosen pressure transduce is commercially available (PXM409-001BGUSBH, Omega,
USA), made of 316L steel typically considered the minimum grade for use in marine
environment. The operational pressure range is 0-10 bar with a maximum deviation of 0.08 %.

The pressure transducer has the function to measure and transmit pressure values (up to
1000 readings/second) over time while filtration occurs.

Figure 6 Photo of the Omega pressure transducer with USB output (Omega, 2023)

Some of the properties of this pressure transmitter are:

• Accuracy: 0.08% BSL (linearity, hysteresis and repeatability combined)
• Resolution: Up to 5.5 significant figures
• Power Consumption: 0.35 W typical. USB powered from laptop.
• Wetted Parts: 316L stainless steel
• Weight: 200 g
• In 2023 the cost of the pressure transducer was about 800 euros (Omega, 2023).
The pressure transmitter was verified periodically with a manometer by pushing compressed
air.

184
 Chapter 8

8.3.1.4 Membrane filter holder

The membrane holder where the membrane filter is placed for the filtration test. It should
avoid leakages, and not damage the membrane at all. In this research a holder for 25 mm
diameter membranes was used. Several types were tested (Sterlitech - Stainless steel,
Whatman GE - Poly propylene, Schleicher & Schuell - Poly propelene) with the Cytiva’s
Whatman (WHA-10461000 FP 025/1 Polyethersulfone) syringe filter holder for 25 mm
filter selected.

Figure 7 Photo of the membrane filter holder. Right is the feed side. O-ring is placed on top of the
feed side of the membrane.

This membrane holder was slightly modified by removing the upper inner wall of the
membrane holder, so in this was the flow distribution towards the membrane is uniform
and only the membrane captures all the particles in the sample water. In the set-up, the filter
holder is connected with a three-way valve that connects with the syringe (sample water)
and with the pressure transducer.

The membrane holder and syringe should be rinsed with UPW before use. After a lot of use,
due to wear and tear, the filter holders need to be replaced as they may give inconsistent
results.

8.3.1.5 Syringe
The water sample is placed in a disposable syringe that is attached to the piston pump. The
used syringe is a BD Plastipak™ 60 mL. Syringes are cleaned by soaking with lab water
before testing. A new syringe can be used every 4 tests and it should be rinsed with UPW
before and after its use. The brand and volume of the syringe should be introduced in the
settings of the syringe pump.

8.3.1.6 Ultra-pure water

Ultra-pure water (UPW) or lab water was produced from Delft’s tap water by a MilliQ
instrument (IX 7005) with the following treatment steps: activated carbon filtration,
reverse osmosis, electro de-ionization, UV disinfection and 0.22 μm filtration.

UPW or lab water is water with organic matter content less than 5 μg/L and conductivity
0.055 μS/cm (or resistivity, 18.2 MΩ-cm). Besides the UPW being free of natural organic
matter and free of ions, it must be free of particles.

In case particles are present in the UPW water, they will be observed by a slight pressure
increase during the test to measure the clean water resistance of the membrane filters. If

185
Experimental Methods for Membrane Applications

particles are present in the UPW, then pores of the filter might be partly clogged and thus
influence the MFI-UF results by decreasing the effective membrane area.

8.3.1.7 Tubing
Tubing should be pressure resistant. The tested operational pressure values were up to 5
bars. Tubing is used to connect the three-way valve with the pressure transducer. Also,
it should be resistant to chemicals, aging and abrasion. The tube length is ~50 cm with a
diameter of 6 mm. Festo compressed air translucent tubing made of polyurethane or similar
product works well. The tubing needs to be filled with water and avoid the presence of air
inside of it.

8.3.1.8 Software
The measured signal of the pressure transducer needs to be recorded for further processing.
This is done via the software ‘Digital transducer application’ provided by the manufacturer
of the pressure transducer (Omega, 2023). The pressure transducer is connected to a
computer via a USB port. Data acquisition is set to every 10 seconds in the MFI-UF test.

8.3.2 Membrane cleaning and conditioning

Membrane filters must be cleaned before performing the MFI-UF test. A surface that is not
clean may affect the way that the fouling cake is formed on the membrane. According to the
operating instructions provided by the membrane manufacturer, the membranes (PES and
RC) are coated with glycerine to prevent the membrane drying out and also sodium azide
(NaN3) to preserve the membrane.

The suggested cleaning consists of flushing the membrane filter with ultra-pure water
(UPW) for at least 30 minutes until the pressure stabilizes over time at a high flux rate
such as 300 L/m2/h. This recommendation by Li (2019) was proposed after monitoring
the concentration of total organic carbon (TOC) in the permeate water after passing the
membrane filters. In Li’s study, the higher the flux rate the more TOC was released by the
membrane filter; for instance, for a 10 kDa PES membrane, at 300 L/m2/h about 4 mg-
C/L was measured after 30 minutes filtration while at 200 L/m2/h about 2.5 mg-C/L
was measured. Salinas Rodríguez, et al. (2015) cleaned the membranes by soaking them
24 h in UPW and flushing 200 mL of UPW which was effective for reducing the DOC to
concentrations less than 0.1 mg/L.

The data from the membrane cleaning (pressure and time) should be saved to calculate the
membrane resistance (Rm) and MFI-UF of the UPW water. Rm is measured with UPW at any
flux rate.

Membrane resistance can be calculated from the following equation:

P
Rm = Eq. 25
η⋅J

Where: J is the flux rate (m3/m2/s), P is the stable pressure (N/m2) measured while filtering
UPW, and η is the viscosity (Ns/m2) of the UPW, and Rm is the membrane resistance (m-1).
Membrane resistance should be normalized at 20 °C.

186
 Chapter 8

The calculated Rm should be within an accepted deviation. Filters with Rm values out of the
accepted deviation (e.g., 25 %) should not be used in the test. MFI-UF values more than 150
s/L2 suggest that the UPW contains particles that might block the pores of the membrane
filters.

8.3.3 MFI-UF testing procedure

After the cleaning of the membrane filters and determination of the membrane resistance,
the following steps, independently of the membrane MWCO or the flux rate for the test,
should be followed:
1. Membrane filter is placed into the membrane holder. The active layer of the membrane is
placed facing towards the feed side.
2. Measure the temperature of the water.
3. Place the water sample in the syringe.
4. Connect the syringe to the three-way valve.
5. When connecting the filter holder to the 3-way valve make sure that there are not
bubbles trapped inside the feed side of the filter holder. Bubbles can decrease the effective
filtration area when in contact with the membrane surface.
6. Filtration flux rate is controlled manually in the pump by defining the flow rate in mL/h.
The effective membrane area must be considered when calculating the flux rate.
7. The software for recording the pressure and time values should be started. Both, pump
and data logging, must start simultaneously.
8. Criteria to stop the test (either of the following):
a) When cake filtration is reached (linear trend between pressure and time, or the slope
of fouling index and time shows no change in time),
b) When a minimum fouling index (I) value is observed;
c) Change in MFI value in last 5 min filtration is less than 5 % per minute.

8.3.3.1 Selection of filtration flux rate

The MFI-UF test can be performed at any flux rate in the range 10-300 L/m2/h. A flux rate
of 100 L/m2/h was proposed as a default value for performing the test. Table 3 presents
recommended flux rates for the MFI-UF test depending on the assessment purpose.

The selection of the flux in the MFI test should reflect the treatment unit under consideration.
For instance, for profiling a treatment train, a flux rate of 100 L/m2/h is normally used
from source to product water. For purposes of modelling the fouling rate of a membrane
system, an MFI-UF flux rate similar to the one of the membrane unit being assessed should
be applied or correct for it afterwards. In average, UF systems operate in the range 60-110
L/m2/h, while RO systems operate in the range 15-25 L/m2/h. The measured MFI-UF
values depend on the applied flux rate during testing. The higher the flux rate, the higher
the measured MFI-UF value. This relationship is linear (Salinas Rodríguez, 2011, Salinas
Rodríguez, et al., 2012, Salinas Rodríguez, et al., 2015, Abunada, 2023) .

187
Experimental Methods for Membrane Applications

Table 3 Recommended flux rates for MFI-UF depending on purpose

Type of sample Flux of MFI-UF test, L/m2/h
Prediction of fouling rate of seawater RO Average of RO system (10-18)
Prediction of fouling rate of brackish water RO Average of RO system (18-25)
Prediction of fouling rate of RO systems 50, 100, 150, 200 and extrapolate to RO average
flux (e.g., 20 lmh).
Prediction of fouling rate of front RO element 25-30
Deposition factor (RO feed and RO concentrate) Average of RO system or extrapolate from MFI vs.
flux relations.
Modelling of UF system, first cycle after backwash Same flux as UF system
Profiling treatment train 100
Permeate of polymeric and ceramic MF & UF membranes 100
Other samples, e.g., TEP measurement 60 (see Chapter 13)

For the pump, we need to calculate the required flow by considering the effective membrane
area (Aeff) of the filter, as follows: Q = J × Aeff. Considering an effective diameter of 21 mm for
the membrane filter, the following flow rates (Table 4) should be used in the test.

Table 4 Fluxes rates, the required flow rates and estimated duration of the MFI-UF test
Flux rate, L/m2/h Corresponding flow rate, mL/h Estimated duration of the test, h
200 69.3 0.5
150 52.0 0.75
100 34.6 1
50 17.3 2-3
20 6.9 4-6
10 3.46 6-8

NB. 300 lmh is recommended for the cleaning of the UF membrane filters, while 100 lmh is set for measuring the
MFI-UF value.

8.3.4 Calculation procedure

The calculation of MFI-UF is done with the assistance of Microsoft Excel. The data (pressure
and time) collected by the software is exported to Excel.

From Eq. 16, the fouling index (I) is calculated by dividing the slope of the pressure vs. time
line (linear region) over square flux and water viscosity of the tested sample. As follows:
Slope
I= Eq. 26
J 2×η

Where I is the fouling index (m-2), the slope is the minimum slope value from the pressure
vs. time (bar/min) corresponding to cake filtration, J is the flux (L/m2/h), η is the dynamic
viscosity of the water being tested (Ns/m2).

188
 Chapter 8

To keep MFI-UF values comparable with MFI-0.45, the MFI-UF values are standardized to
reference conditions namely: viscosity at temperature of 20 °C (ηo), pressure of 2 bar (ΔPo)
and surface of area of a MFI-0.45 μm micro filter (Ao) as shown in Eq. 27.
η0 × I
MFI = Eq. 27
2 × P0 × A02

Where MFI-UF is the modified fouling index-ultrafiltration (s/L2), ηo is the dynamic

viscosity of water at 20 °C (Ns/m2), I is the fouling index (m-2), ΔPo is the reference pressure
(bar) = 207 kPa = 2 bar (reference condition from the SDI/MFI-0.45), Ao is the reference
membrane area (m2) = 13.8×10-4 m2.

8.3.4.1 Example of membrane resistance calculation of UPW

In the figure below the pressure values versus time are plotted after converting units from
milli bar to Pascal in the vertical axis and from minutes to seconds in the horizontal axis.
The pressure profile is horizontal suggesting that the UPW was free of particles. The average
pressure can be calculated from the time range between 840 and 2,400 seconds.

50,000

40,000

30,000

20,000

10,000

Pressure (Pa) 0
0 600 1,200 1,800 2,400
Time (s)

Figure 8 Filtration of UPW through a 10 kDa PES filter at 100 L/m2/h. Temperature 24 °C.

The average pressure is 24,995 N/m2. The flux (J) is 100 L/m2/h = 2.78×10-5 m3/m2/s.
The viscosity of the water at 24 °C is 0.000911 Ns/m2.

Replacing the values in the formula, we have:

Rm = (24,975 N/m2) / [(0.000911 Ns/m2) × (2.78×10-5 m3/m2/s)]

Then,
Rm = 9.87×1011 (m-1)

The membrane resistance at 20 °C.

The average Rm for the 10 kDa PES filters was reported as 9.65×1011 m-1 ± 1.58×1011; thus,
the calculated Rm falls within the average conditions.

189
Experimental Methods for Membrane Applications

8.3.4.2 Example of MFI-UF calculation

The data (pressure versus time) from the filtration test should be plotted as illustrated in
Figure 9.

200,000 0.8
0.7
160,000
0.6
0.5
120,000
0.4
80,000 0.3

0.2
40,000
0.1

Pressure (Pa) 0 0 I ×1012 (m2)

0 1,000 2,000 3,000 4,000 5,000 6,000 7,000
Time (s)

Figure 9 Filtration of Lake water through a 10 kDa PES filter at 100 L/m2/h and corresponding
fouling index (I ) value versus time.

The slope of the linear region in the range 1,000-2,000 seconds is 13.58 Pa/s. Considering
the testing conditions for flux (J) 100 L/m2/h = 2.78×10-5 m3/m2/s and viscosity of the
water at 24 °C is 0.000911 Ns/m2, the fouling index can be calculated:

I = (13.58 Pa/s) / [(0.000911 Ns/m2) × (2.78×10-5 m3/m2/s)2] = 19.3 ×1012 m-2.

The MFI (s/L2) value can now be calculated considering the reference conditions: η20°C =
dynamic viscosity of water at 20 °C = 0.001 (Ns/m2), ΔPo = reference pressure = 2 bar and
reference membrane filter area, Ao = 13.8 ×10-4 m2.

η0 × I
MFI =
2 × P0 × A02

Ns
0.0001 × (19.3×1012 m−2 )
m 2 1
MFI = ×
N L L
2 × 200,000 2 × (13.8 ×10−4 m2 )2 1,000 3 ×1,000 3
m m m

MFI = 25,370 s/ L2

Reporting decimals is not required due to the accuracy of the test.

190
 Chapter 8

The fouling index I over time as plotted in Figure 9 can be obtained by using the moving
average concept, taking for instance ranges of 3 min for calculating the slope (I) values.
Another possibility for calculating the I values is to use a polynomial fitting equation to fit
the pressure versus time development. By definition, the minimum (I) is the first derivative
of this equation.

8.3.5 Reproducibility
Replicate measurements of MFI-UF have been reported for various types of water samples.
The coefficient of variation ranged from 7% to 12 % as presented in Table 5. From these
measurements we can consider a variation of 10 % in MFI-UF values. This was obtained
from measurements in raw fresh water samples, samples after treatment and for RO feed
water.

Table 5 Average value, standard deviation and coefficient of variation for replicates of MFI-UF of
various samples measured with PES membrane filters
Flux, MWCO, MFI-UF,
Sample L/m2/h kDa n s/L2 Source
A 100 10 5 22,000 ± 2,200 (10 %) (Bassa, 2021)
B 100 10 6 21,800 ± 2,100 (10 %) (Bassa, 2021)
C 100 10 5 3,500 ± 400 (11 %) (Bassa, 2021)
D 100 10 5 2,100 ± 150 (7 %) (Bassa, 2021)
E 100 10 5 1,410 ± 140 (10%) (Bassa, 2021)
F 100 10 5 3,536 ± 430 (12%) (Qatae, 2023)
G 100 10 5 3,900 ± 418 (11%) (Qatae, 2023)
H 100 10 5 2,481 ± 232 (9%) (Yameen, 2023)
I 100 10 10 1,655 ± 236 (14%) (Sithole, 2017
J 50 5 10 2,911 ± 149 (5%) (Sithole, 2017)
K 100 100 10 3,880 ± 395 (10.3 %) (Salinas Rodríguez, 2011)
L CP* 13* 6 2,970 ± 180 (6.1%) (Boerlage, 2001)

* CP = Constant pressure test at 2 bar with a 13 kDa PAN hollow fibre capillary membrane

8.3.6 Blank and limit of detection

The blank level was measured by pushing lab water (type 2) through the membrane filters.
LOD is the concentration or amount corresponding to a measurement level (response,
signal) three standard deviation units above the zero analyte (Taverniers, 2004). To
measure the LOD the procedure is measuring a minimum of either 10 independent sample
blanks (LOD = Average + 3 × StdDev) or 10 independent samples blanks fortified at lowest
acceptable concentration (LOD = 3 × StdDev).

Detection/results below this LOD is possible, but has a higher level of uncertainty. By using
3 times (k) the standard deviation and a sample size (n) of at least 10, there is only a 1 %
chance that a blank sample will have a higher signal than the LOD. As both k and n decrease,
the probability that a blank sample has a higher signal that the LOD increases.

191
Experimental Methods for Membrane Applications

At least ten (10) blanks (lab water) were measured with various membrane MWCOs (e.g.,
10, 30, 50 and 100 kDa). The mean of the values of those blanks and the standard deviation
was calculated. Results are presented in Table 6.

Table 6 Reported blank and limit of detection values of MFI-UF constant flux
MWCO, Flux, Average LOD,
kDa L/m2/h MFI, s/L2 Std. Dev. n s/L2 Source
100 250 13.9 6.8 10 34.3 (Salinas Rodríguez,
2011, Salinas
100 250 22.5 6.5 27 42
Rodríguez, et al.,
50 250 25 12.5 10 62.5 2015)
30 250 35.2 10.5 10 66.7
10 250 15.3 8.7 30 41.4
10 100 0.6 1.9 10 6 (Lindsay, 2022)
10 100 108 46 100 246 (Qatae, 2023)
10 100 104.4 11.5 5 138.8 (Yameen, 2023)
100 50 25 34 5 127 (Li, 2019)
100 100 40 15 30 86
100 200 35 17 10 86
10 50 70 8 2 94
10 100 60 12 8 96

The LOD values ranged between 6 and 246 s/L2 for the 10 kDa membranes with an average
of 105.6 s/L2. A potential reason for the difference is due to the variation in water quality
of the lab water produced over time. Therefore, the LOD was set at 150 s/L2. Any MFI-
UF measurement with value lower than 150 s/L2 should be noted as below detection limit
(bdl).

8.3.7 Sample storage

If possible, it is recommended to measure the MFI-UF values on-site immediately after
taking the sample, thus avoiding having to transport and store the samples over time. If that
is unavoidable, then glass amber bottles should be applied for sampling and stored at 4 °C
for transportation and storage. Measurements should take placed within 24 hours.

Samples should be taken out of the fridge to bring the temperature of the water samples back
to the room temperature before MFI-UF testing; nevertheless, this is not strictly necessary
as the MFI-UF corrects for water temperature in the calculation process.

Storage effect was measured with water samples from two different locations in the
Netherlands (Dordrecht and Andijk) and results are reported in Table 7.

192
 Chapter 8

Table 7 MFI-UF values measured at 100 L/m2/h with 10 kDa PES membranes over time
Avg, Std. CV,
Sample 0 d4 1d 2d 3d 4d 7d s/L2 Dev. %
1A 18,220 27,500 18,300 25,760 22,445 4,884 22
1B 16,400 16,860 20,300 20,800 18,590 2,280 12
1C 5,470 4,320 3,970 4,470 4,558 643 14
1D 3,700 4,110 2,880 3,790 3,620 524 14
2 Reservoir 17,400 17,850 19,810 18,353 1,281 7
15,500 11,200 13,700 13,467 2,159 16
13,100 13,200 13,600 11,500 12,850 926 7
2 UF feed 4,200 3,200 3,490 3,630 514 14
4,250 4,300 3,000 3,850 737 19
2 UF perm 2,150 2,610 1,750 2,170 430 20
2,295 1,010 1,525 1,610 647 40
3 Reservoir 21,000 20,000 19,000 20,000 1,000 5
3 UF perm 1,800 1,900 1,600 1,767 153 9
100 lmh
3 UF perm 3,000 2,800 2,900 2,900 100 3
200 lmh

1 (Qatae, 2023); 2 (Lindsay, 2022); 3 (Bassa, 2021); 4 0 d = same day measurement

The coefficient of variation of the reported results for the 1-7 days of storage ranged between
3 to 40 %, thus, it is recommended to perform the MFI-UF test immediately after taking the
sample or within 24 hours.

8.3.8 Concentration of particles

Further evidence that cake filtration occurs during the MFI-UF test can be observed in the
results of the MFI-UF as a function of particles concentration in the feedwater. This premise
is based on the fouling index, I, being directly related to the concentration of particles Cb
(Eq. 10). Thus, I will decrease directly in proportion to an increase in the dilution factor of
Cb while the specific cake resistance component (α), characteristic of a feedwater type and
independent of concentration, remains constant.

In Figure 10, the results of the MFI-UF with dilutions of Delft canal water at an applied
flux of 100 L/m2/h are shown. Linearity was found for the feedwater, with the regression
coefficient calculated as 0.989.

193
Experimental Methods for Membrane Applications

14,000

12,000

10,000

8,000
y = 120.05x 6,000
R2 = 0.9891
4,000

2,000

MFI-UF (s/L2) 0
0 25 50 75 100
Concentration (%)

Figure 10. MFI values for dilutions of Delft canal water measured with 100 kDa RC membrane at 100
L/m2/h. Adopted from (Salinas Rodríguez, 2011)

Abunada, et al. (2023) also reported a direct linear correlation between MFI-UF values and
the concentration of 150 kDa Dextran, 75 nm polystyrene particles, and Delft canal water.

8.3.9 Membrane material

There are several materials used in ultrafiltration such as: PES, RC, PAN, and PVDF. PVDF is
produced mainly for tight MF membranes. PAN and PES are more available for hollow fibre
membranes and in various pore sizes. For this research, PES and RC membranes were tested
as the range of pore size available was wider.

5,000

4,000

3,000

1,000

1,000
PES
RC
MFI-UF (s/L2) 0
3.00E+11 4.00E+11 5.00E+11 6.00E+11 7.00E+11
Rm (1/m)

Figure 11 MFI values for Delft canal water measured with 100 kDa PES and 100 kDa RC at
100 L/m2/h. Adapted from Salinas Rodríguez (2011)

Figure 11 shows the measured MFI values for Delft canal water. For the PES membranes the
average was 3,880 s/L2 ± 395 (10.3 %), and for the RC membranes the average was 3,800
s/L2 ± 235 (6.3 %). Both membrane materials have an average value close to each other. RC
membranes are slightly more uniform than the PES membranes when measuring MFI-UF.

194
 Chapter 8

Bassa (2021) when studying MFI-UF values of Ijssel lake water and after ceramic membrane
filtration also did not observe a correlation between the 10 kDa PES membrane resistance
and MFI-UF. Qatae (2023) confirmed this observation when studying MFI-UF values after
dual media filtration.

8.4 VARIABLES AND APPLICATIONS OF THE MFI-UF

8.4.1 Plant profiling and water quality monitoring

The application of various UF membrane filters in a pilot scale desalination plant
(Jacobahaven, Netherlands) is illustrated in Figure 12 where 100 kDa, 50 kDa and 10 kDa
PES filters were used. The Amiad strainer showed only a small reduction in MFI-UF as
expected with a relatively large aperture size of 50 μm. Whereas, the reduction in MFI-UF
(and fouling) observed following UF (nominal MWCO of 150 kDa) was much larger i.e., of
94 %, 93 %, and 88 % reduction for 100 kDa, 50 kDa, and 10 kDa MFI-UF test membranes,
respectively (Salinas Rodríguez, et al., 2015).

70,000
100 kDa
60,000
50 kDa
10 kDa 50,000

40,000

30,000

20,000

10,000

MFI-UF (s/L2) 0
Raw strainer UF feed UF perm/ RO conc
water out a/Coag RO feed

Figure 12 Effect of pre-treatment on MFI-UF at the Jacobahaven seawater pilot plant using PES test
membranes of 100, 50 and 10 kDa size (Salinas Rodríguez, et al., 2015)

The percentages in reduction of MFI values after water passing through the ultrafiltration
units were 94.3 %, 93.4 % and 87.6 % for 100, 50 and 10 kDa, respectively.

From October 2016 to July 2017, the MFI-UF values were measured for both raw and 2
μm glass-media filtered North Sea water (Figure 13). The measured values with MFI-UF
were 75-28 times higher than the ones measured with the MFI-0.45. This illustrates the
high fouling potential of particles smaller than 0.45 μm. The high peak at the end of April
coincided for the MFI-UF measurements with the high algal cell numbers in the period.

195
Experimental Methods for Membrane Applications

1,200 20,000
Algal cells
18,000
1,000 MFI-UF Raw NSW

Algal cell counts (cells/mL)

16,000
MFI-UF Filtered NSW
800 14,000

MFI-UF (s/L2)
12,000
600 10,000
8,000
400
6,000
4,000
200
2,000
0 0
16

16

17

17

17

17

17
0

0
/2

/2

/2

/2

/2

/2

/2
0

8
/1

/1

/0

/0

/0

/0

/0
07

26

15

06

25

14

03
Figure 13 MFI-UF 10 kDa PES measured at 100 lmh for raw NSW and filtered NSW (Sithole, 2017)

8.4.2 Flux rate

The fouling potential of a water depends on the filtration flux during testing. To illustrate
this effect in practice, North Sea water was tested with 5 kDa, 10 kDa, 30 kDa and 100 kDa
PES membranes and the results showed higher MFI-UF values for the lower MWCO filters,
and a remarkably strong dependency on flux (see Figure 14).

12,000

5 kDa 10,000
10 kDa
8,000

30 kDa
6,000

4,000

2,000
100 kDa
2
MFI-UF (s/L ) 0
0 50 100 150 200 250
Flux (L/m2/h)

Figure 14 MFI-UF (PES) of North Sea water measured at various flux rates (Salinas Rodríguez,
et al., 2019)

When assessing the fouling potential of RO feedwater, it is important to measure the MFI-
UF value at the same flux at which the RO is operating, either the average flux for the whole
pressure vessel or the flux for the front element. An alternative approach is to find the MFI-
UF value at low flux rate by extrapolating from the MFI vs. flux relationship.

196
 Chapter 8

15,000
UF permeate DMF effluent
12,000
y = 32.884x - 160.1
R2 = 0.92
9,000

6,000

y = 18.375x - 219.39 3,000

R2 = 0.96
MFI-UF (s/L2) 0
0 100 200 300 400
Flux (L/m2/h)

Figure 15 MFI-UF 10 kDa PES values as function of filtration flux of Mediterranean seawater after
parallel treatment by dual media filtration and ultrafiltration (Salinas Rodríguez, et al.,
2015)

Figure 15 shows the MFI-UF values for Mediterranean seawater (left) and for UF permeate
(0.02 μm pore size) and dual media filtration effluent (right) measured at various flux
rates from ~50 L/m2/h up to 350 L/m2/h. Noticeably, the MFI-UF values for both pre-
treatments get closer to each other when measured at low flux rates than when measured at
high flux rates.

8.4.3 Predicting rate of fouling of seawater RO systems

Figure 16 illustrates the theoretical fouling rate, based on Equation 24, as function of MFI
values for various average RO system flux rates for a deposition factor equal to 1 (all particles
rejected by the RO membranes will remain on it forming a cake layer).
1.0
25 lmh 20 lmh
Seawater RO
Ω=1
0.8
T = 20˚C
15 lmh

0.6

0.4
10 lmh

Safe MFI 0.2

ΔNDP/tr (bar/month) 0
0 500 1,000 1,500 2,000
LOD
MFI-UF (s/L2)

Figure 16 Prediction of fouling rate (bar/month) as function of MFI values for a seawater RO system
operating at various flux rates considering Ω = 1 and temperature = 20 °C. Safe MFI = 1
bar NDP increase in 6 months. LOD = 150 s/L2. NB. MFI values should be measured at 10,
15, 20, or 25 lmh, or can be projected from the flux vs. MFI relationship when measuring
at higher flux rates.

197
Experimental Methods for Membrane Applications

A safe MFI value for RO feedwater was proposed considering an increase in NDP of 1 bar
in six months (Salinas Rodríguez, 2011, Salinas Rodríguez, et al., 2015, Salinas Rodriguez,
et al., 2019, Salinas Rodriguez, et al., 2021). Under this fouling rate (increase of 0.17 bar/
month), the safe MFI values for RO feedwater are 171 s/L2, 268 s/L2, 476 s/L2, and 1,071
s/L2 for RO operation at 10, 15, 20, and 25 L/m2/h, respectively.

The predicted fouling rate values, are significantly influenced by the particles deposition
factor, which need to be determined onsite. Figure 17 illustrates the effect of the deposition
factor on the predicted fouling rate for a RO system operating at 15 L/m2/h. The safe MFI
values for RO feedwater range from 476 s/L2 for particle deposition factor Ω = 1 to 1,904
s/L2 for Ω = 0.25.

1.0
Seawater RO
J = 15 lmh
0.8
T = 20˚C
Ω 1.0
0.6
Ω 0.75

Ω 0.50 0.4

Safe MFI Ω 0.25 0.2

ΔNDP/tr (bar/month) 0
0 500 1,000 1,500 2,000
LOD
MFI-UF (s/L2)

Figure 17 Prediction of fouling rate (bar/month) as function of MFI values for a seawater RO
system as function of the particle deposition factor (Ω), considering J = 15 L/m2/h and
temperature = 20 °C. Safe MFI = 1 bar NDP increase in 6 months. LOD = 150 s/L2.
NB. MFI values should be measured at 15 lmh or can be projected from the flux vs. MFI
relationship when measuring at higher flux rates.

8.4.4 Comparing fouling indices

Table 8 presents a comparison of the particulate fouling indices that are applied in practice.
Both SDI and MFI-0.45 operate at constant pressure, are accepted by the ASTM, but are not
sensitive enough to predict rate of fouling of RO systems. MFI-UF constant flux has the
potential of becoming a tool for monitoring fouling potential in membrane systems due to
its robustness, sensitivity, and sound testing procedure and calculation method.

198
 Chapter 8

Table 8 Comparison of particulate fouling indices: SDI, MFI-0.45, MFI-UF

Category SDI MFI-0.45 MFI-UF constant flux
Filtration mode Dead-end Dead-end Dead-end
Operation Constant pressure Constant pressure (2.00 kPa) Constant flux (100 L/
(2.07 kPa) m2/h)
Volume sample >3.8 L >3.8 L 60 mL
Pore size 0.45 µm 0.45 µm 10 kDa (others are
possible: 5, 50, 100 kDa)
ASTM D4189-07 (ASTM D8002-15 (ASTM D8002 - 15, -
D4189 - 14, 2014) 2015)
Filter Flat, 25 or 47 mm Flat, 24 or 27 mm Flat, 25 mm
Fouling None Pore blocking, cake filtration Cake filtration
mechanism
Flux rate >>> 1,000 L/m2/h >>> 1,000 L/m2/h 10 - 350 L/m2/h
Membrane Hydrophilic, Mixed Cellulose acetate, cellulose nitrate, Polyethersulfone (PES),
properties cellulose nitrate, mixed mixed cellulose ester regenerated cellulose (RC)
cellulose acetate, mixed
cellulose ester
Test recordings time t1, time t2 for a t/V vs. V (e.g., every 5 s) P vs. t (l vs. t). (e.g., every
given volume (V). 10 s).
Correction/ None Temperature, pressure, (effective) Temperature, pressure,
Normalization membrane area membrane area,
membrane surface porosity
t1
Formula 1− 1−ω 2
t2 η20 P A d(t / V ) η20 × I
SDIT = ×100 MFI = × MFI =
T η P0
×
A0
×
dV 2 × P0 × A02

Cake Not considered ω (compressibility coefficient Controlled by flux rate in

compression included in formula. It requires the test. Cake filtration
effect additional testing) identified in calculation.
Prediction model Theoretically impossible NDPr J (2 − J ) ( ND Pr − NDPor)
tr = × tr =
ηr × J 02 × Ω × I r 2 (1 − J )2 J 2 ×η × Ω × I r

RO feedwater <5 (preferably 3) < 4 (preferably 1) (DuPont, 2023) 476 s/L2 for SWRO at 15
guideline value (DuPont, 2023) lmh and Ω =1.

Others SDI cannot be used to MFI-0.45 is not sensitive enough Safe MFI = 1 bar increase
test UF permeate water for predicting actual RO fouling. in NDP in 6 months
Potential use for predicting clogging
of RO feed spacer. It can also be
used to predict rate of clogging in
infiltration wells.
Duration ~30 min ~30 min ~1 hour (excluding
membrane cleaning)

199
Experimental Methods for Membrane Applications

8.5 REFERENCES

Abunada M (2023). Prediction of Particulate Fouling in Reverse Osmosis Systems: MFI-UF Method
Development and Application. PhD Dissertation, IHE Delft
Abunada M, Dhakal N, Andyar WZ, Ajok P, Smit H, Ghaffour N, Schippers JC, Kennedy MD (2022)
Improving MFI-UF constant flux to more accurately predict particulate fouling in RO systems:
Quantifying the effect of membrane surface porosity. Journal of Membrane Science 660: 120854
DOI https://doi.org/10.1016/j.memsci.2022.120854
Abunada M, Dhakal N, Andyar WZ, Li Y, Ajok P, Ghaffour N, Schippers JC, Kennedy MD (2023)
Calibrating and Validating the MFI-UF Method to Measure Particulate Fouling in Reverse
Osmosis. Membranes 13 DOI 10.3390/membranes13050535
Abunada M, Dhakal N, Gulrez R, Ajok P, Li Y, Abushaban A, Smit H, Moed D, Ghaffour N, Schippers
JC, Kennedy MD (2023) Prediction of particulate fouling in full-scale reverse osmosis plants
using the modified fouling index – ultrafiltration (MFI-UF) method. Desalination 553: 116478
DOI https://doi.org/10.1016/j.desal.2023.116478
Abushaban A (2019) Assessing Bacterial Growth Potential in Seawater Reverse Osmosis Pretreatment:
Method Development and Applications CRC Press
AlShuaili HY (2012). Predicting fouling in seawater reverse osmosis system. MSc, IHE Delft
MWI/12.15.
Althuluth M (2009). Further Development of the Modified Fouling Index (MFI-UF) at Constant Flux
for SWRO Applications. MSc, UNESCO-IHE MWI 2009-032.
Andyar WZ (2021). Further development of the Modified Fouling Index-Ultrafiltration (MFI-UF)
method. MSc, IHIE Delft UWS-WSE/21.21.
ASTM D4189 - 14 (2014) Standard Test Method for Silt Density Index (SDI) of Water ASTM
International, West Conshohocken, PA.
ASTM D8002 - 15 (2015) Standard Test Method for Modified Fouling Index (MFI-0.45) of Water
ASTM International, West Conshohocken, PA.
Bassa FGM (2021). Evaluating surface water pre-treatment with the MFI-UF for assessing fouling in
membrane systems. MSc, IHE Delft UWS-WSE/21.25.
Belfort G, Marx B (1979) Artificial particulate fouling of hyperfiltration membranes II. Analyses and
protection from fouling. Desalination 28: 13-30
Boerlage SFE (2001) Scaling and Particulate Fouling in Membrane Filtration Systems Swets&Zeitlinger
Publishers, Lisse
Carman PC (1937) Fluid flow through granular beds Trans Instn Chem Engrs 15: 32-48
Carman PC (1938) Fundamental principles of industrial filtration (A critical review of present
knowledge). Trans Instn Chem Engrs 16: 168-188
Cytiva-Whatman (2023). Polyethersulfone syringe filter holder for 2.5 cm filter. https://www.
sigmaaldrich.com/NL/en/product/aldrich/wha10461000. Accessed 10 November 2023.
Dhakal N (2017) Controlling biofouling in seawater reverse osmosis membrane systems Delft
university of technology, CRC Press / Balkema - Taylor & Francis Group, the Netherlands
Dhakal N, Salinas Rodriguez SG, Schippers JC, Kennedy MD (2016) Investigation into Fouling of the
Ultrafiltration and Reverse Osmosis Systems: FEWA Desalination Plant, Ras Al Khaimah, UAE.
Prepared for Pentair X-Flow. IHE Delft, Delft, pp. 44.
Dhakal N, Schippers JC, Kennedy MD (2018) Investigation into fouling of the Ultrafiltration & reverse
osmosis systems: RAF desalination plant, QATAR IHE Delft, Delft, pp. 58.

200
 Chapter 8

Dhakal N, Abunada M, Schippers JC, Kennedy MD (2020) Water quality monitoring of a demineralized
water plant at Evides. IHE Delft, Netherlands.
DuPont (2023) FILMTEC™ Reverse Osmosis Membranes Technical Manual, Form No. 45-D01504-
en, Rev. 16Water solutions, pp. 211.
Ekowati Y (2011). Fouling Prediction in Reverse Osmosis Systems using The Modified Fouling Index
- Ultrafiltration (MFI-UF). MSc, IHE Delft MWI 2011/07.
Ekowati Y, Msuya M, Salinas Rodriguez SG, Veenendaal G, Schippers JC, Kennedy MD (2014) Synthetic
organic polymer fouling in municipal wastewater reuse reverse osmosis. Journal of Water Reuse
and Desalination 4: 125-136 DOI doi:10.2166/wrd.2014.046
Gulrez R (2021). Application of MFI-UF at constant flux in a full-scale RO plant. MSc, IHE Delft UWS-
WSE/21.33.
Harvard Apparatus (2023). Standard Infusion Only PHD ULTRA™ Syringe Pumps https://www.
harvardapparatus.com/standard-infusion-only-phd-ultra-syringe-pumps.html. Cited 26
October 2023
Le Bouille L, Bassa F, Salinas Rodriguez SG, Philibert M, Kennedy MD (2021) Water quality monitoring
at Bahia La Palma Desalination plant IHE Delft, Delft, pp. 20.
Le Bouille L, Salinas Rodriguez SG, Ekowati Y, Philibert M, Kennedy MD (2023) Water quality and
fouling monitoring at Perth II Desalination plant IHE Delft, Delft, pp. 22.
Li Y (2019). Further Development of the Modified Fouling Index – Ultrafiltration (MFI-UF) Method:
Calibration and Application in Full-Scale UF-RO. MSc, IHE Delft UWS-WSE.19-31.
Lindsay AL (2022). Assessing particulate fouling potential at PWNT Andijk pilot plant via on-site MFI-
UF testing. MSc MSc, IHE Delft UWS-WSE.22-17.
Malaichami V (2023). Operational optimization of the UF in terms of hydraulic performance and
permeate water quality - A case study for Lake Valkenburg pilot. MSc, IHE Delft UWS-
WSE.23-21.
Merck (2023). Ultrafiltration Discs, 10 kDa NMW PBGC02510 https://www.sigmaaldrich.com/NL/
en/product/mm/pbgc02510. Cited 27 October 2023
Mirzaei P (2023). Further development of the Seawater Bacterial Growth Potential method MSc, IHE
Delft UWS-WSE.23-23.
Ofori M (2023). Predicting Particulate Fouling in Reverse Osmosis systems. MSc, IHE Delft WSD.23-
1058145.
Omega (2023). USB Output Pressure Transducer https://www.omega.nl/pptst/PXM409-USBH.
html. Cited 27 October 2023
Qatae M (2023). Monitoring particulate fouling potential and organic matter characterization along the
pre-treatment line of an NF/RO surface water pilot plant. MSc, IHE Delft UWS/WSE.23.11.
Rathnayake SC (2013). Particulate fouling in reverse osmosis systems : Measuring the deposition factor
to predict rate of fouling. MSc, IHE Delft MWI 2013-31.
Ruth BF, Montillon GH, Montonna RE (1933) Studies in Filtration I. Critical Analysis of Filtration
Theory. Industrial & Engineering Chemistry 25: 76-82 DOI 10.1021/ie50277a018
Salinas Rodríguez SG (2011) Particulate and organic matter fouling of SWRO systems: Characterization,
modelling and applications CRC Press/Balkema, Delft
Salinas Rodríguez SG, Althuluth M, Schurer R, Kennedy MD, Amy GL, Schippers JC (2009) Modified
fouling index (MFI-UF) at constant flux for seawater RO applications. Paper presented at the
Desalination for the Environment: Clean water and Energy, Baden-Baden, Germany, 17-20 May
2009

201
Experimental Methods for Membrane Applications

Salinas Rodríguez SG, Amy GL, Schippers JC, Kennedy MD (2015) The Modified Fouling Index
Ultrafiltration Constant Flux for assessing particulate/colloidal fouling of RO systems.
Desalination 365: 79-91 DOI 10.1016/j.desal.2015.02.018
Salinas Rodriguez SG, Kennedy MD (2009) MFI-UF results for DOW Barcelona pilot plant UNESCO-
IHE, Delft, pp. 12.
Salinas Rodriguez SG, Kennedy MD (2009) Monitoring of particulate fouling after DMF and UF pre-
treatment at Veolia’s Toulon pilot plant UNESCO-IHE, Delft.
Salinas Rodríguez SG, Kennedy MD, Amy G, Schippers JC (2010) Flux dependency of particulate
fouling by MFI-UF measurements in seawater reverse osmosis systems. Paper presented at the
EuroMed 2010: Desalination for Clean Water and Energy, Tel-Aviv, Israel, 3-7 October 2010
Salinas Rodríguez SG, Kennedy MD, Amy GL, Schippers JC (2012) Flux dependency of particulate/
colloidal fouling in seawater reverse osmosis systems. Desalination and Water Treatment 42:
155-162 DOI DOI:10.1080/19443994.2012.683104
Salinas Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (2021) Seawater Reverse Osmosis
Desalination: Assessment and Pre-treatment of Fouling and Scaling, 1st edn IWA Publishing,
London
Salinas Rodriguez SG, Sithole N, Dhakal N, Olive M, Schippers JC, Kennedy MD (2019) Monitoring
particulate fouling of North Sea water with SDI and new ASTM MFI0.45 test. Desalination 454:
10-19 DOI https://doi.org/10.1016/j.desal.2018.12.006
Salinas Rodríguez SG, Sithole N, Dhakal N, Olive M, Schippers JC, Kennedy MD (2019) Monitoring
particulate fouling of north sea water with SDI, MFI0.45, and MFI-UF. Paper presented at the
IDA World Desalination Congress, Dubai, UAE, 20-24 October 2019
Salinas Rodriguez SG, Villacorte LO, Kennedy MD (2010) Water analysis results for Barcelona seawater
UF-RO plant. UNESCO-IHE, Delft, pp. 9.
Salinas Rodriguez SG, Villacorte LO, Wahyudi D, Kennedy MD (2010) Water analysis results for
DOW Barcelona seawater samples. UNESCO-IHE, Delft, pp. 6.
Schippers JC (1989) Vervuiling van hyperfiltratiemembranen en verstopping van infiltratieputten
Keuringinstituut voor waterleidingartikelen KIWA N.V., Rijswijk
Schippers JC, Folmer HC, Kostense A (1980) The effect of pre-treatment of river rhine water on fouling
of spiral wound reverse osmosis membranes7th International Symposium on Fresh water from
the Sea, pp. 297-306.
Schippers JC, Hanemaayer JH, Smolders CA, Kostense A (1981) Predicting flux decline or reverse
osmosis membranes. Desalination 38: 339-348
Sigma-Aldrich (2023). Whatman® FP 025/1 filter holder. https://www.sigmaaldrich.com/NL/en/
product/aldrich/wha10461000. Accessed 10 November 2023.
Sithole N (2017). Particulate fouling assessment of North Sea water. MSc, IHE Delft UWS-WSE 2017-
33.
Tabatabai SAA (2014) Coagulation and ultrafiltration in seawater reverse osmosis pretreatment CRC
Press/Balkema, Delft
Tahtouh L (2023). Monitoring the performance of in-line coagulation with rapid sand filters as a pre-
treatment unit for reverse osmosis and for dune’s infiltration A case study for lake Valkenburg
(Dunea). MSc, IHE Delft UWS-WSE.1085080.
Taverniers I, et al. (2004) Trends in quality in the analytical laboratory. II. Analytical method validation
and quality assurance. Trends in Analytical Chemistry 23: 535-552
Villacorte LO (2014) Algal Blooms and Membrane Based Desalination Technology CRC Press/
Balkema, Leiden

202
 Chapter 8

Wahyudi D (2010). Fouling Prediction with the Modified Fouling Index – Ultrafiltration (MFI-UF) at
Constant Flux for Seawater Reverse Osmosis (SWRO) Systems. MSc, IHE Delft MWI 2010 –
031.
Yameen A (2023). Prediction of particulate fouling in nanofiltration and reverse osmosis feed by pre-
treated surface water using modified fouling index – ultrafiltration (MFI-UF) method. MSc, IHE
Delft UWS-WSE.23-18.
Zhan M, Lee H, Jin Y, Hong S (2020) Application of MFI-UF on an ultrapure water production system
to monitor the stable performance of RO process. Desalination 491: 114565 DOI https://doi.
org/10.1016/j.desal.2020.114565

203
 Part 3
Inorganic fouling and scaling
 doi: 10.2166/9781789062977_0207

Chapter 9

Inorganic Fouling:
Characterization Tools
and Mitigation
Nuria Peña García, Genesys-PWT, Spain

The learning objectives of this chapter are the following:

• Define inorganic fouling in membrane systems

• Description of inorganic fouling components

• Presence and impact of inorganic fouling on membranes

• Tools to avoid inorganic foulant

• Inorganic fouling removal

9.1 INTRODUCTION

There is no doubt about the effect of fouling on the performance of the different membrane
processes that can be used in water treatment.

Depending on the type of membrane system used, the meaning of fouling can be different.
Microfiltration and ultrafiltration membranes work as barriers to retain suspended matter
and that matter should be removed during backwash and cleaning protocols established to
remove it. So, in those cases, fouling would correspond to the retained matter which is not
properly removed during routine/standard cleaning procedures/cycles.

The rest of water treatment membranes systems were mainly developed to remove salts
from water, so any suspended matter reaching their surface will produce fouling. On the
other hand, every feed water has a scaling potential which may scale membrane surface if it
is not protected by chemical treatment.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Main types of membranes fouling can be classified as shown in Table 1.

Table 1 Membrane fouling types and sub-categories

Organic fouling Inorganic fouling
Biofilm, organic matter Colloidal - particulate
Other organics (oil-greases, chemicals, etc.) Scaling

The difference between organic and inorganic compounds is mainly referred to the presence
of carbon which is as C-H in organics (with some exceptions). Inorganic substances or
compounds would include pure elements, salts, many acids and bases, metals and alloys and
minerals. There are also a few inorganic compounds containing carbon as oxides, carbides,
some carbonates and some cyanides.

Inorganic fouling could be defined as the deposition of inorganic compounds on the

membrane surface or inside the membrane pores. Thus, the deposits could be either
inorganic compounds with low solubility in water or solutes present in large amounts in
water.

When they form supersaturated solutions, eventually precipitate out of the solution and
onto the surface of the membrane. Scale formation on membrane surfaces would occur
through crystallization/precipitation (Figure 1b) and deposits would be from particulate
fouling (Figure 1a). The former mechanism involves the precipitation of ions and their
subsequent deposition on the membrane, whereas the latter involves the convective
transport of particulates from the bulk of the solution to the membrane surface. In rivers,
groundwater, seawater, and municipal wastewater, the main inorganic compounds that
contribute to scaling and inorganic fouling are hydroxides, sulfates, carbonates, calcium,
magnesium, iron, ortho-phosphates, silicic acids, and silica (Jiang et.al., 2017, Guo W. et al.,
2012).

1a Deposit 1b Scaling

Figure 1 Characteristic aspect of a deposit (1a) and scaling (1b) found on membranes surface
during autopsies (Images credit by Genesys Membrane Products, S.L.)

208
 Chapter 9

Experimental data obtained from the study of reverse osmosis (RO) and ultrafiltration (UF)
membranes surface establish that main types of foulant are: biofilm, organics, colloidal
matter/particulate, metals, scale.

Figure 2 includes data from autopsies, showing the percentage of membranes showing
these foulants as main component and as main reason of membrane failure.

As this figure indicates, approximately 60% of RO membranes show a major presence of

inorganic foulant while for UF this percentage is only around 30%. Thus, considering these
data, inorganic fouling is a bigger problem on desalination membrane systems than on other
membrane filtration process.

(a) RO membranes (b) UF membranes

Biofilm/ 31.3 Biofilm 45.7

organic matter

Collodial matter/ 29.0 Organic 21.4

particulate

Scale 22.1 Collodial matter/ 17.1

aluminosilicates

Metals 10.0 Metals 11.4

Other organic 7.7 Calcium carbonate 4.3

Figure 2 Percentage of membranes showing main presence of different types of foulant on RO

membranes (a-left) and UF membranes (b-right). Experimental data obtained from
membrane autopsies (1000 data from RO and 100 data from UF).

But even when fouling can be mainly organic on some systems, it is almost impossible to
find a pure foulant and they will mostly include an inorganic component (Peña et al., 2013).
Composite nature of foulants is the reason that makes them different from one system to
other and produces that the impact on membranes and their removal is also different than
expected in many cases. At this point, the identification of all the fouling components is
essential.

As Figure 1 shows, most common inorganic foulants on membranes are composed of

colloidal matter, scaling salts and metals. Same compounds are commonly found as
secondary components of biofilm and other organic foulants since they commonly trap
these particles coming with the water while they grow or settle on membranes surface.

In this chapter, main characteristics of inorganic foulant components will be covered and a
review of the different ways to determine them and how to remove them is included.

209
Experimental Methods for Membrane Applications

9.2 MAIN COMPONENTS OF INORGANIC FOULING

9.2.1 Colloidal matter/particulate

Solids are present in water in three main forms: suspended particles, colloids and dissolved
molecules. Figure 3 shows the size range of particles of concern in water treatment
(Koohestanian et al., 2008):

Typical diameter in metres

1E-10 1E-08 1E-06 1E-04 1E-02

1 nm 1 µm 1 mm
Molecules
Colloids

Suspended particles

Pathogenic:
Viruses protozoa

Bacteria

Algae

Figure 3 Size ranges of particles of concern in water treatment (Koohestanian, et.al. 2008)

Even when suspended particles may reach membranes surface depending on the
effectiveness of the pre-treatment, main components identified on membranes fouling are
hydrophobic colloids as clay and non-hydrated metal oxides.

Colloids have an assigned size range of 0,001 to 1 μm and constitute a significant component
of the particulate matter in natural waters (Koohestanian et al., 2008). In water treatment,
colloidal matter is commonly related to higher sizes (<2 μm).

The characteristics of particulate and colloidal material are largely dependent upon its
source and thus the unique environmental system under consideration (Miroslaw, 2007).
In open ocean systems the majority of particulate matter is biotic in origin and is generated
in the upper surface layer of the ocean. This is in contrast to riverine systems that are heavily
influenced by erosion of material in the watershed. Large lakes may be qualified as a mix of
these two extremes.

Most suspended solids smaller than 0.1 mm found in waters carry negative electrostatic
charges. Since the particles have similar negative electrical charges and electrical forces to
keep the individual particles separate, the colloids stay in suspension and small particles. The
magnitude of the zeta potential (Zp) is usually used to indicate colloidal particle stability.
The higher the zeta potential, the greater the repulsion forces between the colloidal particles

210
 Chapter 9

are and, therefore, the more stable is the colloidal suspension. Coagulation is commonly
used to neutralise the negative charges of the suspended solids by using a positively charged
element (coagulants based on Fe3+ and Al3+ for example) and form a gelatinous mass to trap
(or bridge) particles thus forming a mass large enough to settle or be trapped in the filter
(Kevin et al., Coagulation-Flocculation factsheet).

Pretreatment processes such us coagulation, flocculation and low pressure membrane

filtration (microfiltration and ultrafiltration) have been used in front of other membrane
processes to remove particles and large colloids, but fouling by small colloidal matter
(<2 μm) and fine suspended particles is a main problem (Armstrong et al., 2009).

There are different tools to quantify and determine the amount of matter coming with
the water which are very helpful: turbidity, suspended solids, SDI, MFI. SDI and MFI are
described in chapters 6 and 7-8, respectively.

Besides, it can be very helpful also to determine the size of the particles composing the
suspended matter to determine the best pre-treatment: particle counting.

The principal consequence of membrane fouling by colloidal matter is an increase in

hydraulic resistance resulting in a greater energy requirement to operate the process. The
formation of highly impermeable deposits on the membrane surface will result in significant
problems in maintaining permeate flux with frequent cleaning eventually being required
to maintain system operation. The primary effects of fouling by colloidal particles in a
membrane system will be seen mainly in the elements in the first positions. However, if this
problem remains untreated fouling will gradually affect all membrane elements. The effects
will include a reduction in membrane flux (reduction in product flow rate), an increase in
salt passage as well as an increase in feed channel pressure drop (ΔP).

Membrane damage through abrasion processes have also been identified during membrane
autopsies performed on systems fouled with clay mineral deposits due to the compression
of the crystalline structure against the membrane surface by increased operating pressures.
In other cases, the increase in ΔP when membrane operates in presence of foulant will
produce damage on the areas of contact between the spacer material and membrane surface.

From the different components of colloidal matter/particulate, clays/aluminosilicates,

silica (sand particles or colloidal silica) and microorganisms skeletons/inorganic structures
are mainly identified in membranes fouling.

Clays
Clay minerals are the most important component of the soil, they are usually ultra-fined
particles having less than 2 μm sized particles. Most of the clay minerals are known as
hydrous aluminosilicate or hydrous aluminum phyllosilicate (Nascimento et al., 2021).
Mineralogically, clays are divided into 3 principal groups, with more than 30 different
minerals within these groups (Armstrong et al., 2009).

211
Experimental Methods for Membrane Applications

Table 2 Main 3 types of clay minerals and their properties (Armstrong, et al. 2009)
Clay type Structure and property
Kaolinite The most common clay composed of silicate sheets
bonded to aluminium oxide/ hydroxide layers. Main composition:
aluminosilicates
Illite Structure contains a wide range of cations including
Al, Mg, Fe and potassium.
Montmorillonite/ smectite Structure includes Ca, Na, Al, Mg and is notable for
its ability to take up and lose water.

As already explained, one of the characteristics of clays as membrane foulant is that they
appear mixed with the rest of foulant components, mainly with the organic component.
During the operation of the membrane composite foulant compact and age in such a way that
it is very difficult to remove from membranes surface and the final impact is an irreversible
damage and the consequent lack of retention (Peña et al., 2013). Thus, considering that the
small size of the clays/aluminosilicates makes very difficult to achieve a complete removal
from water bulk, it is essential to establish preventive cleaning of the systems to avoid them.

a) Mixture of diatoms b) Characteristic mixture c) Detail of clay structure

and clays of clays and particles
Figure 4 Examples of foulant with presence of clays
(Images credit: Genesys Membrane Products S.L.)

Microorganisms with inorganic structures

There are some microorganisms that show a characteristic inorganic skeleton/structure and
are commonly found on membranes fouling. Even when the source could be considered
organic, they are included in this inorganic classification because what it is commonly
identified and detected during the study of membranes is their inorganic structure. Most
common microstructures are silica structures from diatoms and calcium carbonate structures
from marine algae or marine coralline microstructures.

These microstructures commonly appear embedded in the rest of the fouling as a minor
component and they could be removed mixed with the rest of components if the appropriate
cleaning procedure is applied. Main risk with these microstructures is that they may produce
a severe abrasion on membranes surface and the consequent irreversible damage. Since
these structures come in the water bulk, they are mostly found on first position membranes.

It is important to identify the presence of these microstructures during the study of

membranes to understand the source of silica and calcium carbonate and to determine if

212
 Chapter 9

these compounds are as scaling species or as deposits to determine the best way to avoid
them. For the removal of these structures it is very important to determine their size and to
adapt the pre-treatment to it if they appear in a significant amount.

• Silica structures-Diatoms (Bacillariophyta) are single cell, siliceous cell wall algae and
are the principal component from the phytoplankton division (Smol, et al., 2010). They
can exist in all aquatic ecosystems (marine, brackish, fresh waters) including in some moist
terrestrial ecosystems (Smol et al., 2010, Wynne et al., 1985). The photographs in Figure
5 show characteristic structures of this kind of algae structures found on membranes. It is
common to find not only these full structures, but fragments from them.

a) Diatom structure b) Massive mixture of diatoms

c) Massive presence of diatoms d) Diatom on organic foulant

mixed with organic foulant

e) Mixture of diatoms with f) Mixture of diatoms with

colloidal matter / clays organic foulants and
colloidal matter / clays

Figure 5 Examples of different diatom structures found in foulant.

(Images credit: Genesys Membrane Products S.L.)

213
Experimental Methods for Membrane Applications

• Calcareous structures. Calcium carbonate minerals are the building blocks for the
skeletons and shells of many marine organisms. It is very common also to find
coccolithophores, which are generally regarded as calcareous scale-bearing marine algae
(Figure 6, 2.0–75.0 μm in cell diameter Jordan et al., 2009). In many cases it is easier to
distinguish these structures on microfilters and SDI pads.

a) Calcareous structures retained b) Calcareous structures mixed with foulant

on a cartridge filter

c) Calcareous structures with foulant d) Calcareous structures mixed with foulant

on cartridge filter

e) Calcareous structures
retained on a fiber surface
of a cartridge filter

Figure 6 Examples of calcareous structures (Images credit: Genesys Membrane Products S.L.)

9.2.2 Metals
Presence of metals on membranes foulant may have many different sources:
• Metals can be dissolved or as a suspension in the originating body of sea water, well water
or surface water.
• Use of coagulants (aluminium and iron salts) during pretreatment
• Corrosion drags/products from pipe materials and equipment
• Complexes with natural organic matter (Schippers, 2021)
• Impurities from chemicals

214
 Chapter 9

Effect of metals on membranes is very similar to the rest of inorganic foulant components
related to suspended matter: main presence on first position membranes and the effect on
membrane performance will mainly depend on the rest of foulant components. On reverse
osmosis systems, it is important also to consider the catalytic characteristics of metals and
the effect that they many have in oxidation-reduction processes improving the damage of
the polyamide.

Considering bibliography and data obtained from autopsies (Peña et al., 2017), main types of
metals detected on membranes surface are: iron, manganese and aluminium. Figure 7 shows
the percentage of membranes which showed these and other metals as main component of
fouling on RO membranes.

Iron 55% Mixture of metals 16%

Chromium 1%

Aluminium 14%

Manganese 14%

Figure 7 Main metals detected during autopsies (Peña et al., 2017)

However as already explained, besides the main components of foulant, it is essential to

consider secondary or minor components. It is actually almost impossible to find a fouling
only composed of metals.

On the other hand, when studying different components of fouling on membranes it is very
common to detect small presence of metals, especially iron (both as iron oxide particles or as
part of aluminosilicates/clays) and also particles from corrosion drags (Fe-Cr-Ni).

Main effect of metallic particles from corrosion will be abrasion of membranes surface
(Figure 8).

215
Experimental Methods for Membrane Applications

a) Massive presence of corrosion drags b) Presence of corrosion drags on

at membrane feed end membrane surface feed side

c) Abrasion produced by metallic particles

Figure 8 Examples of membranes with presence of metallic particles from corrosion

(Images credit: Genesys Membrane Products S.L.)

A characteristic of main metals found on membranes is that they achieve characteristic color
to foulant. The photographs in Figure 9 show some examples and some microphotographs
obtained during the study of membranes with significant presence of iron, manganese and
aluminium.

a) RO membrane surface with main b) Fibers from a UF module with

presence of iron (as iron oxide). main presence of iron (as iron oxide).
It commonly shows a dark orange color. It commonly shows a dark orange color.

c) Detail of fouling with iron, which d) Detail of iron particles - Granular shape.
corresponds to the brighter component.
Iron commonly appears at SEM as very bright
and tiny particles mixed with the rest of components.

Figure 9 Membranes and foulant with main presence of iron

(Images credit: Genesys Membrane Products S.L.)
216
 Chapter 9

Presence of metals in foulant is directly related to a lack of or insufficient pre-treatment. For

example, when iron and aluminium are related to the dosing of coagulants, it is essential to
optimize dosing and to use the different tools available to control them when water shows
a very high variability and makes difficult to get the right concentration.

In the case of iron and manganese, there can be other different sources (Peña et al., 2017) and
in those cases it is necessary to use the different technologies available in water treatment for
metals removal.

a) Detail of membrane with manganese foulant: b) Detail of manganese fouling by SEM

it shows a characteristic dark brown color (MnO2) Manganese commonly shows very
characteristic spherical structures mixed
with the rest of fouling components.

c) Detail of fouling with main presence of d) Detail of fouling with aluminium by SEM
aluminium. It commonly appears mixed with It shows a very unspecific shape since it
organic matter and shows the color of appears mixed with organic matter.
the mixture due to the different components. It doesn’t show bright intensity as other metals.

Figure 10 Membranes and foulant with main presence of manganese (pictures a and b) and
aluminium (pictures c and d) (Images credit: Genesys Membrane Products S.L.)

To get a good control of metals it is necessary to identify the source of the metals and to have
data from water analyses from different points of the process treatment.

Following photographs in Figure 11 correspond to SDI pads from different points of pre-
treatment, in which a decrease on the initial orange color from iron is distinguished. As it can
be observed, the significant presence of iron on the raw water disc is significantly removed
after sand filtration but there is not much change after microfiltration.

217
Experimental Methods for Membrane Applications

a) Massive presence b) Presence of corrosion c) Abrasion produced

of corrosion drags at drags on membrane by metallic particles
membrane feed end surface - feed side

Figure 11 SDI pads obtained from different points of a pre-treatment process.

Besides, it is essential to distinguish dissolved metals from oxidized or suspended solids

and to consider the specifications of membrane manufacturers about the concentration of
metals that can reach the membrane.

It is very important also to consider concentration of metals in water since it is very

important to consider them for antiscalant projections. If a lower concentration is
considered for antiscalant projection, calculated dosing could be wrong, and it would have
scaling consequences.

9.2.3 Scaling
Scaling potential of water is one of the main issues to be considered in water treatment and
inevitably needs to be chemically controlled. When there is a failure on this scaling control,
the consequence is the precipitation of salts. On reverse osmosis membranes, this scaling
is mainly produced on last position membranes where concentrations of sparingly soluble
salts are the highest. Besides the issues on membrane performance as drop on production
and salt rejection, main risk with scaling is the irreversible damage that may produce by
abrasion on membrane surface.

The microphotographs in Figure 12 correspond to main scaling species identified on

RO systems and the percentage of membranes in which they were identified as the main
component in a study of 500 RO membranes (Peña et al., 2013).

218
 Chapter 9

a) Calcium carbonate – 31% b) Silica – 32%

c) Calcium phosphate – 14%

d) Calcium sulphate – 9% e) Barium sulphate – 4%

Figure 12 Main scaling species identified on RO systems.

(Images credit: Genesys Membrane Products S.L.)

Scaling species are not only identified when water saturation is exceeded. In many cases,
precipitation issues are produced due to increases in pH or the use of water with poor quality
in hardness during cleaning processes.

The photographs in Figure 13 correspond to UF membranes with calcium carbonate scaling.

a) Presence of calcium carbonate on b) Massive presence of calcium carbonate

an UF element fibers surface on an UF fiber surface

Figure 13 UF membranes with calcium carbonate scaling.

(Images credit: Genesys Membrane Products S.L.)
219
Experimental Methods for Membrane Applications

As a compliment, microphotographs in Figure 14 correspond to a UF hollow fiber

membrane showing scaling on the external surface due to some cleaning issues (blue color),
when membrane fouling is composed of aluminosilicates (yellow color from silicon) (Peña
et al., 2015):

Internal
surface

Support layer

b) Distribution of silica on
UF hollow fiber cross section
External
surface

a) Detail of UF hollow fiber cross section

c) Distribution of calcium on
UF hollow fiber cross section

Figure 14 UF hollow fiber membrane showing scaling on the external surface.

(Images credit: Genesys Membrane Products S.L.)

It is very important to determine the source of the scaling when it is not due to antiscalant
failure since it would be the only way to avoid it. Some tools that may help to identify scaling
source are the following:
• On UF systems it is common to find them on the side of the membrane which is directly
in contact with the cleaning solution during backwash: out surface for in-out filtration
mode and inner surface for out-in fibers. In these cases, scaling would be related to failed
cleaning procedures.
• On RO membranes, a scaling due to already precipitated structures would initially
be detected on lead positions since it would come from a deposition and not from a
precipitation due to an oversaturated solution.
• As previously explained, some species may come from microstructures as diatoms or sea
water calcareous structures (see Section 9.2.1).
• To identify the source of the scaling species when they are already formed, the analysis of
different components of the pre-treatment such as cartridge filters and SDI pads as well
as water samples will certainly help.

220
 Chapter 9

9.2.4 OTHER COMPONENTS

Besides the main inorganic elements of foulant already described, and depending on the
source of water and membrane applications, other elements could be detected. From them,
the following components are very common.

Filtration materials
In many cases, membranes show presence of big particles/grains from filtration materials:
silica grains, anthracite, etc.. They concentrate on feed end of first position elements (Figure
15). Unless their presence is massive or they even reach other membranes position, these
components are not an issue for membranes performance, but they commonly produce
damage by abrasion. Thus, it is very important to check filtration systems and to assure that
they do not reach membranes systems.

a) Massive presence of anthracite b) Massive presence of anthracite

on membrane feed on membrane feed

c) Massive presence of filtration d) Massive presence of anthracite

material on membrane feed on membrane feed creates channeling

Figure 15 Detail of different material leakage reaching membranes.

(Images credit: Genesys Membrane Products S.L.)

Sea water membranes-Sodium Chloride

Even when it is not an issue as foulant, sea water membranes commonly show sodium
chloride as the main inorganic component of foulant and it interferes on the quantification
of the organic/inorganic components of foulant (see Chapter 18 – membrane autopsy).
Mainly in biofilm foulant, sodium chloride precipitates during the drying process necessary
to carry out some analyses and it is very easily distinguished since it massively appears as
dendritic structures when membranes and foulants are studied by SEM-EDX for example.
Besides sodium chloride, common inorganic components of foulant on sea water
membranes are magnesium, calcium, phosphorus and sulphur in small percentages.

221
Experimental Methods for Membrane Applications

The microphotographs in Figure 16 show characteristic dendritic structures of sodium

chloride.

a) Characteristic dendritic structures of sodium chloride on sea water membranes foulant

b) Characteristic dendritic structures of sodium chloride with presence of other inorganic

components of foulant as magnesium and calcium

Figure 16 Characteristic dendritic structures from sodium chloride sea water on membranes foulant
(Images credit: Genesys Membrane Products S.L.)

9.3 METHODS FOR INORGANIC FOULING IDENTIFICATION

It is very important to remember that there is no single analytical tool which can give the
complete identification of foulant, mainly due to its composite nature. Then, it is necessary
to obtain information from different techniques to get the global information about foulant
components.

Tables 4 and 5 include the most common analytical methods used for the identification of
inorganic fouling.

To quantify the inorganic content of a foulant, main analytical methods are based on the
lower degradation temperature of the organic matter against the inorganic component.
Thus, loss of ignition and thermogravimetric analysis (TGA) are the most common
methods to quantify the amount or percentage of the organic/inorganic component. These
procedures are interesting also as complement for inorganic components identification,
since the analysis of the inorganic residue obtained after is a way to concentrate inorganic
components and, in some cases, it helps to determine if a specific element is related to the
organic or the inorganic component.

222
 Chapter 9

Besides, along this chapter it has been mentioned many times how important it is to
determine the source of some inorganic components (metals, scaling species, etc). For source
identification, it would be necessary to consider not only the analysis of the foulant, but also
the analysis of water and other process/pre-treatment components. Table 3 includes some
of the samples from water treatment for inorganic water components identification.

Table 3 Analysis of interest from water treatment components for inorganic components
identification
Inorganic elements/
Sample Objective Analytical technique components detected
Water Dissolved To determine Ionic chromatography Anions and cations
components scaling potential and Spectrophotometry Metals, silica, phosphates,
presence of metals. ICP sulphates, silica
Titration Metals, elements
Potentiometry (ISE) Calcium, magnesium,
chlorides, TA/TAC
Fluorides, others.
Suspended Identification of SEM-EDX By filtering water sample
matter suspended matter. ATR-FTIR by 0.45 or 0.22 µm,
Particle counting suspended solids can be
identified.
Determination of particles
size.
Filters Cartridge Identification of SEM-EDX Identification of inorganic
filter suspended matter. ATR-FTIR components.
SDI pads from raw water,
treated and microfiltered
water allows to determine
SDI pads components removal
and to get a complete
characterization of water.

223
Experimental Methods for Membrane Applications

Table 4 Analytical techniques for the identification of inorganic components of fouling – Elemental
identification
Analytical technique Observations
Spectrophotometry - Destructive technique.
- To be applied on foulant sample.
- It needs digestion of the fouling sample and individual analysis
of each component by different methods.
- It is necessary to determine in advance which elements are
necessary to analyse.
ICP – MS - Destructive technique.
(Inductively coupled plasma Mass - To be applied on foulant sample.
Spectrometry) - It needs digestion of the fouling sample.
- Even when this technique allows to make a scan of different
ICP – OES
elements, in some cases it is necessary to know in advance
(Inductively coupled plasma
which are the main since the system must be calibrated with the
Optical Emission
elements of interest.
Spectroscopy)
XRF - Non-destructive technique.
(X-ray fluorescence) - It can be applied both on the fouling and membrane surface.
- It can be applied on the raw but dried sample and to make a
scan which gives information about the elements from certain
concentration.
SEM-EDX - Non-destructive technique.
(Scanning Electronic Microscopy – Energy - It can be applied both on the fouling and membrane surface.
Dispersive X-ray spectroscopy) - It can be applied on the raw but dried sample and to make a
scan which gives information about the elements from certain
concentration.
For some scaling species, considering elements stoichiometric
ratio, the % atomic allows stablishing relation between some
elements to determine compounds.
For example, Ca/S in ratio 1/1 would be as CaSO4
- It allows also studying the distribution of the fouling on the
membrane surface and to distinguish different components on
the fouling or on different areas.
- Abrasion marks can be observed also by this technique.
XPS-ESCA - Non-destructive technique.
(X-ray Photoelectron Spectroscopy - - It can be applied both on the fouling and membrane surface.
Electron Spectroscopy for Chemical - It can be applied on the raw but dried sample and to make a
Analysis) scan which gives information about the elements from certain
concentration.
- Energy binding gives information about the oxidation state of
the detected elements.

224
 Chapter 9

Table 5 Analytical techniques for the identification of inorganic components of fouling –

Compounds identification
Analytical technique Observations
ATR-FTIR Non-destructive technique.
(Attenuated Total Reflection Fourier It can be applied both on the fouling and on membrane surface.
Transform Infrared spectroscopy) It can be applied on the raw sample and to make a scan which gives
information about functional groups.
There are available IR database which allows compounds
identification. These databases are commonly provided with IR
equipments, although there is a broad offer available on-line,
Identification is complicated when there is a mixture of different
components.
XRD Non-destructive technique.
(X-ray diffraction) It can be applied both on the fouling and membrane surface.
Since the identification is based on the diffraction pattern, it
cannot achieve accurate identification if the sample doesn’t show
crystalline structure. It hardly identifies silica on membranes
surface, for example.
Difficult identification for mixtures.

9.4 METHODS FOR INORGANIC FOULING REMOVAL

In many systems it is necessary to assume that due to inadequate chemical and physical
pre-treatment, some foulant components will continue entering the membrane system and
that it is necessary to optimize and to reduce the frequency of required cleaning. To reduce
cleaning frequency and minimise membrane damage the operator can achieve optimum
deposit removal by cleaning with a technically correct product (Peña et al., 2013).

Table 6 include some basics that need to be considered for inorganic foulant components
removal.

For specific cleaning recommendations, Tables 7 to 9 include specifications that can be

found at the technical manuals available from some of the main RO and UF membrane
manufacturers.

Besides commodity cleaners included in these tables, there are also many different
formulated multifunctional cleaners available in the market that can certainly be effective
against main inorganic foulants.

225
Experimental Methods for Membrane Applications

Table 6 Basics on cleaning procedures for inorganic foulants removal

Main foulant General cleaning recommendations
Colloidal matter Alkaline cleaner applied with temperature
Longer contact times will help for the removal
Use of optimum flow rates will help.
Metals Mild acid cleaner applied at room temperature.
For aluminium, alkaline cleaner could work also.
Calcium carbonate/ Strong acid cleaner at room temperature.
phosphate:
Sulphates: Specific alkaline with surfactant.
Temperature
Inorganic scaling
Depending on the amount of scaling higher contact time may
be needed.
Silica: Specific alkaline cleaner with high temperature and long
contact time.

Table 7 Cleaning recommendations of some of the leading membrane manufacturers for inorganic
colloids
Membrane RO membranes UF membranes
manufacturer * Recommended cleaner Recommended cleaner (CIP)
DUPONT- 0.1% (W) NaOH and 0.025% (W) 0.2% HCl, 2% citric acid /oxalic acid
DOW FILMTEC Na-DSS, pH 12, 35°C max. 0.1% NaOH + 0.2% NaOCl
HYDRANAUTICS 2.0% (w) STPP (sodium NaOH, HCl, H2SO4, or citric acid
tripoliphosphate) (Na5P3O10) and
0.8% (w) Na-EDTA, pH of 10.0
NaOH, HCl, H2SO4, or citric acid
0,1 % NaOH + 0,03% SDS,
pH 11,5
TORAY 1 – 2 % citric acid adjusted with
ammonia (NH3)
LG NANO H2O NaOH, EDTA / permeate RO
NaOH: until 0.1% weight
EDTA: until 1.0% weight
Citric acid 2.0% weight
INGE NaClO H2O2
NaOH / acid pH
PENTAIR NaOCl (active chlorine) 500 ppm max.
H2O2 1000 ppm max.
NaOH pH ≤ 11
Nitric acid pH ≥ 1
Phosphoric acid pH ≥ 1
EDTA pH ≤ 11
Citric acid
Enzyme compounds

226
 Chapter 9

Table 8 Cleaning recommendations of some of the leading membrane manufacturers for metals
Membrane RO membranes UF membranes (CIP)
manufacturer * Recommended cleaner Recommended cleaner
DUPONT- 1.0% Na2S2O4 (pH:5, 30º C) Citric acid, HCl,
DOW FILMTEC 2.0% Citric acid oxalic acid, sulphuric acid, pH= 2
0,5% H3PO4
1.0% sulfamic acid
NITTO-HYDRANAUTICS 2.0% Citric acid Citric acid or HCl
1.0% Na2S2O4 (pH: 4-6)
TORAY Citric acid 1 – 2 %, adjust with
ammonia (NH3), pH: 2-4
LG NANO H2O 2.0% Citric acid, pH: 2-4
INGE-DUPONT HCl, H2SO4, pH: 1 - 2,5
Citric acid pH: 4
PENTAIR 1% citric acid, 1% oxalic acid,
0,25% ascorbic acid

*Manufacturers technical manuals

Table 9 Cleaning recommendations of some of the leading membrane manufacturers for inorganic
salts
RO membranes
Salt species Membrane manufacturer * Recommended /preferred cleaner
Calcium carbonate / DUPONT DOW-FILMTEC 0.2 wt% HCl (pH 1 – 2, 35°C)
inorganic salts 2.0 wt% citric acid
(phosphate, etc) 0.5% H3PO4
1.0% Na2S2O4
NITTO-HYDRANAUTICS 2 v/v % citric acid
0,5 v/v % HCl, pH 2,5
TORAY Citric acid 1–2 wt%, adjust with
ammonia, pH: 2-4
Sulphate scales DUPONT DOW-FILMTEC 0.1 wt% NaOH 1.0 wt% and
Na4EDTA pH 12, 30°C maximum
NITTO-HYDRANAUTICS 2 % STPP + 0,8 % Na-DDBS, pH 10
0,5 v/v % HCl, pH 2,5
TORAY 1 wt% Sodium hexametaphosphate
(SHMP), pH 2
Silica DUPONT DOW-FILMTEC 0.1% (W) NaOH and 0.025%
(W) Na-DSS, pH 12, 35°C max.
0.1% (W) NaOH and 1.0% (W)
Na4EDTA, pH 12, 35°C max.
NITTO-HYDRANAUTICS 0,1 (w/v) % NaOH , pH 11,5

*Manufacturers technical manuals

227
Experimental Methods for Membrane Applications

9.5 REFERENCES

Armstrong M.W., S. Gallego, S.P. Chesters. Cleaning clay from fouled membranes. Desalination and
Water Treatment, 10 (2009) 108-114.
Gardner K. H. and Defne S. Apul, Influence of colloids and sediments on water quality, Environmental
and Ecological Chemistry – Vol. II – Influence of Colloids and Sediments on Water Quality –
Encyclopedia of Life Support Systems (EOLSS).
Guo, W.; Ngo, H.H.; Li, J. A mini-review on membrane fouling. Bioresour. Technol. 2012, 122, 27–34
Jiang, S.; Li, Y.; Ladewig, B.P. (2017). A review of reverse osmosis membrane fouling and control
strategies. Sci. Total Environ. 2017, 595, 567–583.
Jonasz M., Georges R. Fournier, The particles size distribution in Light Scattering by Particles in Water,
2007
Jordan R.W. Encyclopedia of Microbiology (Third Edition), 2009, Pages 593-605. https://doi.
org/10.1016/B978-012373944-5.00249-2
Koohestanian, M. Hosseini and Z. Abbasian, The Separation Method for Removing of Colloidal Particles
from Raw water. American-Eurasian J. Agric. & Environ. Sci., 4 (2): 266-273, 2008.
Kumari N. and Chandra Mohan, Basics of Clay Minerals and their characteristic properties. DOI:
10.5772/intechopen.97672
Nascimento G. M. D. Clay and Clay Minerals. London: IntechOpen; 2021 222 p. Available from:
https://www.intechopen.com/books/10949 doi: 10.5772/intechopen.95640.
Peña García N., Javier Rodriguez, Fernando del Vigo, Matt Armstrong, Max Fazel, Stephen Chesters,
(2017), Results of a neutral pH cleaner that removes complex fouling and metals from
membranes. IDA WWC-Sao Paulo Brazil
Peña García N., Javier Rodriguez, Fernando del Vigo, Stephen Chesters, (2015), UF membranes
autopsies: An approach to hollow fiber membranes surface. IDA WWC-San Diego Ca, USA
Peña N., S. Gallego, F. del Vigo & S.P. Chesters. Evaluating impact of fouling on reverse osmosis
membranes performance, Desalination and Water Treatment, 51 (2013) 958-961.
Schippers J.C. (2021), Inorganic fouling, Seawater Reverse Osmosis Desalination: Assessment
and Pre-treatment of fouling and Scaling (pp.187-206). IWA Publishing. doi:
10.2166/9781780409863_0187
Smol JP, Stoermer EF. The diatoms: applications for the environmental and earth sciences: Cambridge
University Press; 2010.
SSWM, Coagulation-Flocculation factsheet, University Course.
Wynne MJ, Bold H. Introduction to the Algae: Structure and Reproduction: Prentice-Hall, Incorporated;
1985.

228
 doi: 10.2166/9781789062977_0229

Chapter 10

Assessing Scaling Potential

with Induction Time and a
Once-through Laboratory Scale
RO System

M. Nasir Mangal, IHE Delft & Berghof Membranes, The Netherlands

Sergio G. Salinas-Rodriguez, IHE Delft, The Netherlands

The learning objectives of this chapter are the following:

• Describe experimental methods for assessing scaling potential

• Present the induction time protocol

• Propose the once-through RO as a system for assessing scaling potential of RO feed

water

10.1 INTRODUCTION

Membrane scaling is when one or more sparingly soluble salts (e.g., calcium carbonate,
calcium sulphate, silica/metal silicates, barium sulphate, calcium phosphate, etc.) precipitate
and form a dense layer on the membrane surface in reverse osmosis (RO) applications. In
Figure 1, the scanning electron microscopy (SEM) images of the RO membrane surface
without and with scaling are illustrated. Figure 1b is from the RO membrane where calcium
carbonate scaling occurred and Figure 1c is the membrane surface scaled with calcium
phosphate.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Figure 1 SEM images (500x magnification) of the (a) clean (virgin) RO membrane, (b) RO
membrane scaled with calcium carbonate, and (c) RO membrane scaled with calcium
phosphate. Adopted from (Mangal, 2023).

Scaling, like other types of membrane fouling, reduces permeate production (due to
decreased membrane permeability), raises operational costs (due to higher operating
pressure, cleaning costs, etc.), degrades permeate water quality (due to increasing salt
passage), and shortens the lifetime of membranes due to frequent membrane cleanings
(Kucera, 2010, Mangal, et al., 2021).

Membrane scaling can occur when sparingly soluble salts in RO concentrate become
supersaturated, meaning their concentrations exceed their equilibrium (solubility) levels.
In RO processes, the increased concentration of sparingly soluble salts in the concentrate
is primarily caused by the withdrawal of permeate water from the feedwater. The ratio of
permeate water to feedwater is known as recovery which is directly related to membrane
scaling. Recovery needs to be as high as possible in RO installations to minimize specific
energy consumption. However, at high recovery rates, the concentration of sparingly
soluble salts in the concentrate can increase dramatically. For example, for 80% and 90%
recovery, the concentration of salts in the concentrate can reach 5 and 10 times their
concentration in the feedwater, respectively. If the calcium and phosphate concentrations
in the RO feedwater are 200 mg/L and 5 mg/L, respectively, the concentrations in the
RO concentrate will be 1000 mg/L and 50 mg/L at 90% recovery, exceeding the calcium
phosphate solubility limit and resulting in calcium phosphate scaling. Therefore, in brackish
water reverse osmosis (BWRO) processes, scaling is typically the main barrier to operating
RO installations at high recoveries.

There are several methods for preventing scaling in RO applications, including acidification
of RO feed, lowering RO system recovery, and antiscalant addition. Acidification of RO
feedwater was one of the first methods for tackling calcium carbonate scaling in RO processes.
However, due to the risks associated with the use of acid, this method is becoming less
common. Furthermore, acidification may not be effective for all types of scales; for example,
it is very effective in preventing calcium carbonate scaling but not calcium sulphate scaling.

Another method of preventing scaling is to operate RO at low recovery (ratio of permeate

water to the feedwater). The recovery of the RO application is reduced in this approach to
reduce the supersaturation level of the concentrate water to undersaturated conditions. Low
recovery reduces the adverse effect of concentration polarization because there is less solute

230
 Chapter 10

concentration on the membrane surface, reducing the potential for scale formation. This
approach, however, is not very appealing or economical because it results in high specific
energy consumption. Furthermore, the large amount of concentrate disposal is a problem.
Antiscalant dosing in feedwater is one of the most extensively applied and effective scaling
prevention strategies in RO applications (Greenlee, et al., 2010, Kucera, 2010, Antony, et
al., 2011, van Engelen and Nolles, 2013, Yu, et al., 2020). Antiscalants are primarily organic
compounds containing sulphonate, phosphonate, or carboxylic acid functional groups that
hinder the crystallization process, i.e., nucleation and/or growth phase of scaling compounds
(Antony, et al., 2011, Boels and Witkamp, 2011, van Engelen and Nolles, 2013). In general,
antiscalant prevent scale formation by three mechanisms, namely threshold inhibition,
crystal modification and dispersion. Threshold inhibition is when antiscalant molecules
adsorb on crystal nuclei and halt their nucleation process, whereas crystal modification and
dispersion are the ability of antiscalants to stop the growth and/or agglomeration of crystals
and particles. The selection of antiscalants in RO applications depends on the feed water
composition as well as other factors such as recovery and discharge regulations.

With the use of antiscalants, the main question which arises is: How to determine the lowest
(optimum) dose of antiscalants to prevent scaling in RO applications? Operating the RO
with the lowest antiscalant dose at which scaling does not occur is highly desirable, since
high doses of antiscalant result in additional costs and pose environmental concerns (Boels
and Witkamp, 2011). In practice, the antiscalant dose for a given water composition is
generally determined using the antiscalant manufacturer’s proprietary programs. However,
the method used by the manufacturers to calculate the antiscalant dose is unknown and
therefore the end-users cannot verify their recommended doses. In general, the suppliers’
recommended antiscalant doses are in the range of 2–10 mg/L to prevent scaling in RO
processes (Singh, 2005).

Mangal (2023) studied the scaling potential of calcium carbonate and calcium phosphate in
brackish water reverse osmosis by assessing the theoretical scaling potential with manual
and computer programmes, by measuring the induction time and by developing a new
once-through RO system.

The chapter on membrane scaling by Mangal, et al. (2021) describes thoroughly the
principles of scaling, influencing factors, types of scaling, prediction calculations, scaling
indices, use of commercial software, monitoring tools in RO systems, scaling control, and
scaling in seawater applications. In this chapter we will focus on the description of the
induction time protocol as well as of the once-through lab-scale RO system for scaling
studies in RO applications.

10.2 INDUCTION TIME MEASUREMENTS

Induction time is defined as the time elapsed between the emergence of supersaturated
conditions and the detection of crystallization. It is composed of three time periods such as
relaxation time (tr), nucleation time (tn), and growth time (tg) (Kashchiev, 2000). The time
needed to initiate nucleation from time zero to steady state condition is called the relaxation
time(Guan, 2009). Nucleation time is defined as the time needed to form a stable nucleus

231
Experimental Methods for Membrane Applications

and the period in which detectable crystal are formed from the stable nucleus is known as
growth time (Kashchiev, 2000). Induction time depends on the supersaturation level of a
water solution, but it is mainly dependent on the precipitation kinetics.

To measure the induction time several methods have been developed, including but not
limited to the pH method (Waly, 2011), the conductivity method (Söhnel and Mullin,
1978), turbidity or scattered light method (Prisciandaro, et al., 2001, Abdel-Aal, et al.,
2004, Shih, et al., 2006), and the concentration of calcium (Verdoes, et al., 1992). Among
these methods, the pH method was reported to be the most accurate one for the induction
time measurement of CaCO3 (Waly, 2011). However, for measuring the induction time of
(amorphous) calcium phosphate, pH measurements are not that useful especially when high
bicarbonate concentrations are present. For calcium phosphate, turbidity measurements
could be used, but in the presence of some antiscalants, the formed calcium phosphate
particles may not be detected in turbidity measurements (due to their small size) (Mangal,
2023).

10.2.1 Experimental setup

Figure 2 depicts a schematic diagram of the experimental setup which was used by Mangal
(2023) to measure the induction time of calcium carbonate. The setup is composed of the
following components:

10.2.1.1 Glass reactor

Double wall glass reactor (Applikon, Netherlands) of 3.1 L volume in cylindrical shape (24
cm height and 12 cm internal diameter). It should be air tight with stainless steel lid and a
thin rubber.

It has a mixing shaft with two pedals fixed in the stainless-steel lid. The lid plate has 4 circular
holes with 13 mm diameter used for the pH probe or other instruments (e.g., turbidity,
dissolved oxygen, etc.) when required. Three inlet stainless steel pipes of 0.5 cm diameter.

Air/overflow: There is an air/overflow valve installed at the top of stainless-steel lid.

The main function of this valve is for ventilating the system during water circulation and
to check that the reactor is totally filled with the calcium chloride and sodium bicarbonate
solution by noticing the solution flooding from the overflow pipe.

NaOH addition: This element consists of one vertical pipe with a valve and a rubber on the
top of the reactor. The main function of NaOH is to adjust the pH of the targeted solution.
Feed line of saline water/demineralised water/acid: It is a 0.5 cm straight pipe inside the
reactor to feed the NaHCO3 solution or other saline solution. It connects the reactor with
the saline water flask and the demineralised water or acid when cleaning of the reactor is
required.

Feed line of CaCl2 /drainage line: This connection (L shaped pipe) allows feeding the
reactor with CaCl2.2H2O solution from the bottom in order to have homogeneous solution
by mixing the solution during feeding. This line also can be used for emptying the reactor
from the solution inside after each experiment.

232
 Chapter 10

CaCl2.2H2O NaHCO3
solution solution

1.5 L 1.5 L

Peristaltic
pump

pH & temperature
measurement pH 1)
probe
Overflow/
drain
HCl solution for
2) cleaning reactor
1) Stirrer controller
3)
2) Stainless steel straight tube
Thermostat Applikon glass reactor 3) Stainless steel L-shaped perforated tube

Figure 2 Experimental setup for induction time tests. Adopted from (Mangal, 2023).

10.2.1.2 Stirrer device

It includes Applikon controller and mixing shaft in the reactor. Applikon mixing controller
(Applikon type ADI 1032 motor controller) is mainly used for the adjustment of the stirring
rate which ranges between 0 and 1250 rpm.

For the induction time experiments the stirring rate is 150 rpm. For cleaning the reactor, the
stirring speed is increased up to 1200 rpm.

10.2.1.3 pH meter
The pH meter model is Endress and Hauser Memosens system (0.01 relative accuracy
for pH). The pH value and temperature sensor are in one probe. The interval time of pH
measurement can be set as low as 1 second.

10.2.1.4 Peristaltic pump

A Master flex peristaltic pumps model 77200 -50 is used to pump the HCl solution (or
nitric acid solution) to the reactor for the cleaning purposes.

10.2.1.5 Thermostat
A thermostat is used to keep the temperature at a desired value and constant during an
entire experiment. As the Applikon rector is a doubled wall, the thermostat water keeps
recirculating between the reactor and the thermostat to maintain a uniform temperature of
the solution in the reactor.

10.2.2 Experimental procedure

10.2.2.1 Preparation of artificial brackish water
For the CaCO3 induction time measurements, CaCl2.2H2O and NaHCO3 analytical grade
salts are mainly used. In addition, NaCl salt is used to achieve a certain TDS level of the
water and also to balance the cations concentrations with anions. The following procedure
is followed to prepare artificial brackish water:

233
Experimental Methods for Membrane Applications

1. The CaCl2.2H2O solution is prepared by dissolving the required amount of the salt for
10 L in 5 L of Milli-Q (2 times concentrated).
2. NaHCO3 and NaCl solution is also prepared by dissolving the required amount of the
salts for 10 L in 5 L of Milli-Q (2 times concentrated).
3. To ensure complete dissolving of the salts, the solution is mixed for at least 2 h using a
magnetic stirrer at an average speed of 300 rpm and at a room temperature of 20 °C.
4. The two solutions were filtered separately through 0.45 μm Millipore filter using
vacuum filtration. This step can be eliminated when Milli-Q water and high-grade
chemicals are used for the preparation of CaCl2.2H2O and NaHCO3 solutions.

10.2.2.2 Induction time measurement

The following steps are followed to measure induction time of artificial brackish water.
1. Prepare the stock solutions of CaCl2.2H2O and NaHCO3 as described in section 10.2.2.1.
2. Calibrate the pH electrodes using 2 standard buffer solutions. The calibration curve
can be checked by its slope which has to be between 55 and 59 (Endress Hauser
requirement).
3. Install the calibrated pH probe inside the reactor.
4. Check that the reactor is cleaned by filling the reactor with demineralised water
and measure its pH value inside the reactor which should be similar to the pH of
demineralised water.
5. Calculate the required volume of NaOH or HCL that needs to be added to the solution
in order to adjust the initial pH.
6. Switch on the stirring controller and set the stirring speed at 150 rpm.
7. Move 1.55 L of NaHCO3 and NaCl solution to the elevated flask using graduated
cylinder and let the solution feeding the reactor by gravity.
8. Switch on the thermostat and adjust the temperature. Keep the solution 10 minutes to
stabilize with the new temperature.
9. Add the required volume of NaOH or HCL by using pipette.
10. Close all opening in the reactor except the ventilation valve and calcium chloride
feeding line.
11. Switch on the computer and check that it is connected properly with the pH meter.
12. Switch on the pH meter software and set time interval and other settings.
13. Start measuring pH of the solution inside the reactor.
14. Move 1.55 L of CaCl2.2H2O solution to the elevated flask using graduated cylinder and
let the solution feeding the reactor by gravity.
NB: In case of testing antiscalant (AS), AS dose needs to be mixed with the calcium chloride
solution before moving the solution to the reactor.
15. Observe the ventilation valve. The reactor should be filled fully until water reaches to
the ventilation valve. Once it reaches the ventilation valve, close the ventilation valve,
and close the calcium chloride feeding line.

10.2.3 Calculation of induction time

Figure 3 illustrates an approximate method to determine induction time. As shown, the
induction time is the intersection point of the horizontal line of the curve (constant pH)
with the slope of the curve when pH starts declining.

234
 Chapter 10

Nucleation Growth Stability

Induction time
pH
Time

Figure 3 Determination of the induction time.

10.2.4 Cleaning of the reactor

After each experiment, the water inside the reactor has to be drained out from the reactor’s
stainless-steel pipe (L shaped) using peristaltic pump. Cleaning of the reactor includes all
components connected in the reactor such as; pH probe, water connection and fittings, etc.
Following is the procedure to clean the reactor:
1. Open the air release/ overflow valve.
2. Connect the drainage pipe with the drainage nozzle and open the valve of the drainage
pipe.
3. Once the all water in the reactor is drained, connect the acid line with the acid/
demineralised water feed line and switch on the pump.
4. Fill the reactor with 0.1 M HCl to dissolve any crystals formed during the experiment and
increase the stirring speed to 1,200 rpm.
5. Close the acid valve and switch off the pump. Allow acid cleaning for 30 minutes and
then empty reactor as mentioned in steps 1-2.
6. Connect the demineralised water line with the acid/demineralised water feed line and
open the demineralised water valve in order to flush the reactor.
7. Monitor pH of the demineralised water in the reactor.
8. Drain the water out and repeat step 6 and 7 until the pH of the solution inside the reactor
becomes same as the pH of the demineralized water.

10.2.5 Example of application of induction time

This section presents an application of the induction time in investigating the scaling
potential of calcium carbonate with and without antiscalant for an RO plant (treating
anaerobic groundwater for drinking water production) in the Netherlands and in exploring
if (and how much) antiscalant was needed for the plant at a certain recovery.

The detailed information about the RO installation and the composition of the anaerobic
groundwater is given elsewhere (Mangal, 2023). The RO plant was operating at 80 %
recovery. At this recovery, calcium carbonate (with Langelier saturation index (LSI) of 1.7)
was the primary compound which would cause scaling in the RO unit in the absence of
antiscalant according to the projection programs of antiscalant suppliers. Therefore, to
prevent calcium carbonate scaling in the RO, the supplier’s recommended dose of 2.0 mg/L

235
Experimental Methods for Membrane Applications

of a phosphonate antiscalant was added to the feedwater. Due to the stringent concentrate
discharge regulations, a phosphonate antiscalant was not preferred. It was desirable to lower
the antiscalant dose as much as possible as the supplier’s recommended dose was considered
to be greater than the optimum dose.

To determine the lowest possible antiscalant dose, the RO unit was operated at 80% recovery
and varying antiscalant dose. The starting antiscalant dose was 2.0 mg/L (equivalent to the
supplier’s recommended antiscalant dose) for the first day of the RO operation at 80 %
recovery which was afterwards lowered by 0.2 mg/L after every 12 h of operation to a final
dose of 0.2 mg/L. The normalized permeability (Kw) for the last element of the last stage
was recorded separately. The result is illustrated in Figure 4a. As can be seen, the normalized
Kw remained constant when the RO unit was operated for 12 days with an antiscalant dose
as low as 0.2 mg/L. Afterwards, in another test, the RO unit was operated at 80 % recovery
without antiscalant addition (Figure 4b). As illustrated, the normalized Kw of the last
element remained constant for an experimental period of 32 days at 80 % recovery which
indicated that there was no need to add antiscalant even when the LSI of the RO concentrate
was as high as 1.7. This suggested that calcium carbonate scaling might have been inhibited
by some constituents (possibly phosphate and humic substances) present in the feedwater
(anaerobic groundwater) that might have functioned as natural antiscalant.

6.0 2.4
a)
Normalized Kw (L[m2hbar]-1)

( )

Antiscalant dose (mg/L)

5.0 2.0

4.0 1.6

3.0 1.2

2.0 0.8

1.0 ( ) 0.4

0 0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Operational period (days)
6.0 2.4
Normalized Kw (L[m2hbar]-1)

b)
( )
Antiscalant dose (mg/L)

5.0 2.0

4.0 1.6

3.0 1.2

2.0 0.8

1.0 0.4
( )
0 0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
Operational period (days)

Figure 4 (a) RO operation at 80 % recovery with a phosphonate antiscalant, and (b) RO operation
at 80 % recovery without antiscalant addition: ( ) Normalized Kw of the last element of
the last stage, ( ) antiscalant dose. Adopted from (Mangal, 2023)

236
 Chapter 10

To understand why the RO unit did not scale at high supersaturation levels, induction time
measurements were performed with the anaerobic real RO concentrate at 80 % recovery in
the absence of antiscalant (Figure 5a). In parallel, induction time measurements were also
executed using artificial concentrate solutions. The artificial concentrate solutions were
prepared such that the Ca2+ and HCO3− concentrations were equivalent to the concentrate
concentration of real ground water at 80% recovery. In Figure 5a, we show that the measured
induction time of the real RO concentrate at 80 % recovery was longer than 168 h (7 days),
while for the artificial concentrates corresponding to 80 % recovery, the measured induction
time was approximately 1 h. Thus, at the same supersaturation level, the induction time
of the real RO concentrate at 80 % recovery was at least 168 times longer than that of the
artificial concentrate. When 10 mg/L of antiscalant concentration was added to the artificial
RO concentrate, the induction time became longer than 168 h (Figure 5b). This result
clearly showed that the formation of calcium carbonate was suppressed in the anaerobic real
RO concentrate by some constituents present in the RO feedwater (anaerobic groundwater)
which function as natural antiscalants.

7.8 7.8
a) b)
7.6 7.6
IT > 168 h ( ) ( )
7.4 7.4
IT > 168 h
7.6 7.2 7.2
IT ≈ 1 h ( )
7.4 7.0 7.0
pH 7.2
( )
( ) 7.0 6.8 6.8
( )
6.8 6.6 IT ≈ 1 h 6.6
0 1 2
Time (h)
6.4 IT ≈ 2 h ( ) 6.4
pH 6.2 pH 6.2
0 24 48 72 96 120 144 168 0 24 48 72 96 120 144 168

7.8 7.8
c) d)
( ) IT > 168 h 7.6 7.6
( ) IT > 168 h
7.4 7.4
( ) IT > 168 h ( ) IT > 168 h
7.2 7.2
7.0 7.0
6.8 6.8
6.6 6.6
6.4 6.4
pH 6.2 pH 6.2
0 24 48 72 96 120 144 168 0 24 48 72 96 120 144 168
Time (h) Time (h)

Figure 5 (a) Induction time of the: ( ) real RO concentrate without antiscalant, ( ) Artificial RO
concentrate without antiscalant, (b) Induction time of the artificial RO concentrate: ( )
with 10 mg/L phosphonate antiscalant, ( ) with 87 mg/L of magnesium ions, ( ) with
217 mg/L of sulphate ions, (c) Induction time of the artificial RO concentrate without
antiscalant: ( ) with 10 mg/L phosphate ions, ( ) with 5 mg/L phosphate ions, and (d)
Induction time of the artificial RO concentrate without antiscalant: ( ) with 10 mg/L
humic acid, ( ) with 10 mg/L fulvic acid. Adopted from (Mangal, 2023)

237
Experimental Methods for Membrane Applications

We consider that the difference between the induction times of the anaerobic RO
concentrate and the artificial concentrate is caused by phosphate (ca. 2.0 mg/L) and/or
humic substances (ca. 5.0 mg/L) present in the groundwater, as well as by magnesium and
sulphate, which are reported in literature to have a positive effect on the suppression of
calcium carbonate (Berner, 1975, Bischoff, 1968, Chen et al., 2005). We investigated this
hypothesis by varying the composition of the artificial concentrate, evaluating the effect of
magnesium and sulphate (Figure 5b), phosphate (Figure 5c), humic substances (Figure 5d).

It was found that magnesium and sulphate did not have a considerable effect on delaying
calcium carbonate formation (Figure 5b). The induction time of the artificial RO concentrate
at 80% recovery increased to 2 h and 1.2 h with 87 mg/L of magnesium and with 217
mg/L of sulphate, respectively. This showed that neither magnesium nor sulphate was
accountable for the long induction time (>168 h) of the real RO concentrate at 80 % recovery.
On the other hand, as can be seen from Figure 5c, in the presence of 10 mg/L of phosphate
(which was equal to the concentration of phosphate in the real RO concentrate of 80 %
recovery), induction time of the artificial concentrate increased from 1 h to a period longer
than 168 h which indicated that phosphate was one of the constituents of the feedwater
which was responsible for the long IT (>168 h) of the real RO concentrate. When 5 mg/L
of phosphate was added to the artificial concentrate solution, the measured induction time
was also longer than 168 h, which suggested that if the groundwater contained 1 mg/L of
phosphate, it would still reduce the need of antiscalants to control calcium carbonate scaling
at 80 % recovery.

Furthermore, as illustrated in Figure 5d, humic substances also have a substantial role
in hindering the formation of calcium carbonate. The effect of humic substances was
investigated by comparing the induction time of the artificial concentrate of 80% in
the absence and in the presence of 10 mg/L humic acid and fulvic acid (obtained from
International Humic Substances Society (IHSS)). With 10 mg/L humic acid and fulvic
acid, induction times of the artificial concentrate of 80 % recovery increased from 1 h to a
period longer than 168 h. This result showed that the presence of humic substances in the
anaerobic groundwater could also be one of the reasons for the long induction time of the
real RO concentrate at 80 % recovery.

In summary, the induction time measurements revealed that both phosphate and humic
substances considerably hinder the formation of calcium carbonate and therefore can
prevent calcium carbonate scaling in RO applications. When they are present in the RO feed,
the required dose of antiscalants for calcium carbonate scaling can be lowered substantially
or can be completely eliminated as in the case of the RO installation in the Netherlands.

10.3 ONCE THROUGH LAB-SCALE RO SYSTEM

10.3.1 Experimental set-up

To evaluate the performance of antiscalants in preventing calcium phosphate scaling in RO
systems, a lab-scale RO setup (Figure 6) was used.

238
 Chapter 10

a)
Demi-water 37 L/h 45 L/h 45 L/h 43 L/h Demi-water

2 L/h 2 L/h 2 L/h 2 L/h pH probe 2 L/h

Residence
time
< 1 min
HCO3- PO43- NaOH
Antiscalant solution solution solution Ca2+ solution
(0.01-0.02 M)
90 L/h
Permeate
Feed
Concentrate
OSMO Unit
Feed

Permeate/concentrate
to waste
TW30-1812 RO element

b) QC = Quick connection
T = Temperature P-2
P = Pressure
CTRL = Control valve
P.R. = Pressure relief valve CTRL-1 QC-5
QC-3

T-1 P-1 P-3 T-2

0-30 L/h

QC-1 QC-2 Membrane QC-4 Flow meter QC-6

Pump-1 element
P.R.-1 P.R.-2

QC-7

Figure 6 (a) Once-through lab-scale RO setup for calcium phosphate scale inhibition studies,
and (b) Piping and instrumentation diagram (P&ID) of the OSMO unit. Adopted from
(Mangal, 2023)

The artificial solution was fed at a rate of 90 L/h to a TW30-1812-50 RO element

(OsmoPure Water Systems) with the use of an OMSO Inspector unit (Convergence
Industry B.V., Netherlands).

The OSMO unit was equipped with a very sensitive flow meter (with high accuracy) which
could measure the permeate flow rate of 2 mL/min (0.12 L/h) to 500 mL/min (30 L/h).
For each experiment, a new RO element was used.

In all experiments, the initial recovery of the membrane element was in the range of
5–6 % and the permeate flux was between 13–15 L/m2/h. According to the membrane
manufacturer, the minimum ratio of the concentrate flow to the permeate flow should not
drop below 5.

In these conditions, the concentrate flow is about 18 times greater than the permeate flow.
The cross-flow velocity was in the 10–12 cm/s range. Both permeate and concentrate were
directed to the drain.

239
Experimental Methods for Membrane Applications

All experiments should be performed at temperature-controlled conditions of stable

conditions like room temperature (20–23 °C).

10.3.2 Experimental protocol

In this setup, as illustrated, antiscalant, HCO3−, PO43− and NaOH were dosed from stock
solutions each at 2 L/h to a stream of demineralized-water (demi-water) with a flow rate of
37 L/h, resulting in a final flow rate of 45 L/h.

The dosage of NaOH was executed from a 0.01–0.02 M stock solution to adjust the pH of
the final solution to 7.6. To another stream of demi-water (with a flow rate of 43 L/h), Ca2+
was dosed from the stock solution at 2 L/h, also resulting in a final flow rate of 45 L/h.
Both streams were then connected to a single pipe resulting in the final flow rate of 90 L/h
which had nearly the same composition of the artificial concentrate of 85 % recovery. The
final solution (artificial concentrate solution) was introduced to a 4 L reactor in which the
artificial concentrate solution was stirred at a rate of 200 rpm with a residence time shorter
than 1 min.

The residence time of less than 1 min was achieved by maintaining equal flow rates (90
L/h) of the artificial concentrate solution entering and leaving the reactor and by keeping
the volume of the artificial concentrate solution in the reactor to approximately 1.5 L.

10.3.3 Example of application

This section presents an application of the once-through lab-scale RO system to investigate
the effectiveness of available calcium phosphate antiscalants against calcium phosphate
scaling in the RO installation in the Netherlands. As discussed in the previous section
(section 10.2.5), the RO installation could operate at 80% recovery even without antiscalant
as calcium carbonate scaling was inhibited by phosphate and humic substances. The
drinking water company wanted to increase the RO recovery to 85% recovery (or even
higher). However, the permeability of the last stage of the RO unit decreased due to calcium
phosphate scaling which was identified in membrane autopsy (Mangal, 2023). Several
antiscalants (which were effective against calcium phosphate scaling according to the
antiscalant suppliers) were tested. However, none of the antiscalants could prevent calcium
phosphate scaling in the RO unit at 85% recovery as the membrane permeability of the last
element of the last stage decreased with all antiscalants. As the anaerobic RO concentrate
at 85% recovery contained high concentrations of ferrous ion (Fe2+ = 55 mg/L), some
antiscalant suppliers claimed that the effectiveness of their antiscalants is reduced when
iron (II) is present in the RO feed. Therefore, to understand if the available antiscalants
can prevent calcium phosphate scaling in the absence of iron (II), once-through lab-scale
RO tests (as well as induction time tests) were performed with the artificial concentrate
solutions having the same calcium and orthophosphate concentrations that were present
in real RO concentrate at 85 % recovery. More precisely, the artificial concentrate of 85 %
recovery contained 767 mg/L of Ca2+, 14 mg/L of PO43−, and had a pH of 7.6.

240
 Chapter 10

Table 1 Induction time measurements (of calcium phosphate) using the artificial concentrate of
85 % recovery without antiscalant addition and with 5.0 mg/L of various antiscalants.
Adopted from (Mangal, 2023)
Time Turbidity (NTU)
(min) No AS AS–A AS–B AS–C AS–D AS–E AS–F AS–G AS–H
0 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08
1 2.29 1.13 1.01 0.69 0.08 0.08 0.08 0.08 0.08
15 2.94 1.97 2.57 1.62 0.08 0.08 0.12 0.08 0.08
30 3.16 2.36 3.23 2.26 0.08 0.08 0.18 0.08 0.08
45 3.30 2.83 3.82 2.91 0.08 0.08 0.26 0.08 0.08
60 3.38 3.25 4.19 3.60 0.08 0.08 0.39 0.08 0.08

In Table 1, the results of the induction time tests (of calcium phosphate) with and without
antiscalant are presented. The actual names of the eight tested antiscalants (dispersant) are
replaced with arbitrary names in Table 1. It is worth mentioning that for the induction time of
calcium phosphate, turbidity of the artificial solution was measured to detect the formation
of calcium phosphate as their formation could not be detected with the pH measurement.
As can be seen from Table 1, the turbidity of the artificial RO concentrate of 85% increased
without antiscalant as well as with AS–A, AS–B, AS–C, and AS–F which indicated that these
antiscalants were not able to prevent the formation of calcium phosphate. Nevertheless,
the turbidity of the artificial RO concentrate remained constant with AS–D, AS–E, AS–G,
and AS–H which initially gave the impression that the formation of calcium phosphate
was inhibited in the presence of those antiscalant. This, however, was not the case. The
artificial concentrate solutions with no increase in turbidity were filtered through 100 kDa
PES filters. The filters were then examined with SEM (Figure 7). As can be seen, the filter
surface was covered with calcium phosphate particles. This revealed that the formation of
calcium phosphate in the artificial concentrate of 85 % recovery was not inhibited by AS–D,
AS–E, AS–G and AS–H even when no increase in turbidity was observed. The reason for
not observing an increase in the turbidity of the artificial concentrate of 85 % recovery could
be due to the formation of fewer calcium phosphate particles with very small size by the
aforementioned antiscalants.

Figure 7 SEM images (500x magnification) of the 100 kDa filter after filtering the artificial
concentrate of 85 % recovery in the presence of 33 mg/L of a) AS–D, b) AS–E, c) AS–G,
and d) AS–H. Adopted from (Mangal, 2023)

241
Experimental Methods for Membrane Applications

According to some antiscalant suppliers, the induction time results of Table 1 and Figure
7 were not conclusive enough to conclude if those antiscalants are not able to prevent
calcium phosphate scaling in RO where the filtration mode is not dead-end but cross-flow.
It could be that the adsorbed antiscalants (dispersants) on the formed calcium phosphate
particles may diminish their tendency to deposit on the membrane surface in a cross-flow
operation. Based on this, it was necessary to test the performance of these antiscalants in a
once-through RO system.

1.2 1.2
(9)
1.0 1.0
(6)
(8)
Normalized flux (J/J0)

Normalized flux (J/J0)

0.8 (7) 0.8
(5)
(3)
(2) 0.6 (4) 0.6

0.4 0.4
(1)
0.2 0.2

0 0
0 0.5 1.0 1.5 2.0 2.5 3.0 0 0.5 1.0 1.5 2.0 2.5 3.0
Time (h) Time (h)

Figure 8 Normalized flux of the membrane element (of the once-through lab-scale RO setup) when
fed with the artificial concentrate of 85 % recovery (1) without antiscalant addition, and
with 5 mg/L of (2) AS–A, (3) AS–B, (4) AS–C, (5) AS–D, (6) AS–E, (7) AS–F, (8) AS–G,
and (9) AS–H. Adopted from (Mangal, 2023)

In Figure 8, the normalized flux of the RO element when fed with artificial concentrate
of 85 % recovery without antiscalant addition and with 5 mg/L of various antiscalants is
shown. As can be seen, none of the antiscalants could completely prevent the deposition of
calcium phosphate particles on the membrane surface since the permeate flux decreased in
the presence of each antiscalant. However, it is evident that the rate of flux-decline decreased
in the presence of antiscalants. Additionally, one can see that some antiscalants had better
performance than others in slowing down the flux decline. For instance, the permeate flux
decreased by 25 % with AS-A, approximately 17 % with each AS–B, AS–C and AS–F and
about 7 % with each AS–D, AS–G and AS–H in 1.5 h, while no decrease was observed with
AS–E in the same duration. However, after 3 h of operation, the permeate flux with AS–E
decreased by approximately 15 %. The possible reasons for some antiscalants performing
better than others might be due to (i) the formation of fewer particles (which needed longer
time to foul the membrane) in the presence of such antiscalants, and/or (ii) less deposition
of the formed particles due to the reduction in the deposition propensity of the particles by
the antiscalants.

In brief, the results obtained with the once-through lab-scale RO revealed that the available
calcium phosphate antiscalants are not effective against calcium phosphate scaling.
Furthermore, the once-scale RO system is a useful tool for scaling studies and to identify the

242
 Chapter 10

performance of antiscalants in preventing a given scaling compound. The once-through lab-

scale RO system cannot only be limited to scaling studies, but it can also be implemented
to investigate other types of membrane fouling such as organic fouling, biofouling and
particulate fouling.

10.4 OUTLOOK AND FINAL COMMENTS

In this chapter, we presented the induction time protocol and the once-through lab-scale
RO system for scaling-related research in full-scale RO installations. While there are benefits
to both systems, there are limitations as well. The induction time measurements are really
useful in investigating if the formation of a scaling specie (e.g., calcium carbonate, calcium
phosphate, etc.) can be hindered in the presence of an antiscalant or any other substance
(e.g., humic substances, etc.). However, as the induction time measurements are performed
in a glass reactor and it doesn’t have a membrane or filtration step, it is difficult to correlate
its results with the scaling in RO. More precisely, if induction time for a scaling compound
is found to be 1 h, it doesn’t mean that the induction time (or the onset of crystallization) of
that compound in RO installation would be 1 h as well. It could be that scaling in RO occurs
much faster than in the glass reactor due to different material, hydrodynamics, etc.

On the other hand, the once-through lab scale RO system is equipped with an RO element
and it has similar conditions/properties to an RO element in the full-scale RO system.
Therefore, the scaling findings obtained with the once-through lab-scale RO system could
be well representative to a full-scale RO plant. Furthermore, as the lab-scale RO system is a
once-through system which means that the concentrate and permeate are not recirculated
back to the RO feed. This makes the setup more attractive for research, as studies conducted
with concentrate/permeate recirculation are often a matter of debate among researchers
for not being representative to the conditions of full scale RO systems because of: (i) the
residence time in a recycled system is much longer (in the range of hours) than the residence
time (< 1 min) of the concentrate in the last stage of full scale RO plants, and (ii) recycling
of concentrate back to the feed tank may accelerate the process of scaling as micro(crystals)
may be formed. Furthermore, antiscalant manufacturers emphasize that antiscalants may
not be as effective in recycled systems as they should be in once-through flow systems (like
RO systems) and therefore the performance of antiscalants assessed in recycled systems
may not be representative. The drawback of the once-through approach is that it requires
an extensive amount of chemicals for longer tests. Nonetheless, by running experiments at
greater saturation levels, which means rapid scaling, this problem can be addressed.

It is worth mentioning that the once-through lab-scale RO setup is not limited to laboratory-
scale or pilot-scale experiments, as it has the potential for application in full-scale RO systems
(as a scale monitor) in identifying the plant’s maximum recovery as illustrated in Figure 9.
A scale monitor is an additional RO element that is fed with the concentrate of the last stage
of a full-scale RO installation, and because the scale monitor provides additional recovery,
scaling occurs in the scale-guard prior to the final stage of the full-scale RO installation. The
concept of the scale monitor (or scale-guard) is described in more detail elsewhere (van de
Lisdonk et al., 2000). Recently, the once-through lab-scale RO unit has been applied in the

243
Experimental Methods for Membrane Applications

RO installation in the Netherlands. The preliminary results (not shared here) are promising.
However, the unit needs to be further tested and validated in full-scale RO plants to verify
its application as a scale monitor.

This chapter is based on previously published data by Nasir Mangal and co-authors in his
Ph.D. dissertation and also from work at IHE Delft. The findings of his dissertation have
also been published in peer-reviewed journals (Mangal et al., 2021a; Mangal et al., 2021b;
Mangal et al., 2022a; Mangal et al., 2022b).

RO unit RO permeate
rd
2nd stage 3 stage
1st stage

RO feed

RO concentrate
to waste

Scaling monitor

Permeate
Feed
Concentrate
OSMO Unit
Feed

Permeate/concentrate
TW30-1812 RO element to waste

Figure 9 Application of the once-through lab-scale RO setup as a scaling monitor for the
identification of maximum recoveries in full-scale RO installations

244
 Chapter 10

10.5 REFERENCES

Abdel-Aal EA, Rashad MM, El-Shall H (2004) Crystallization of calcium sulfate dihydrate at different
supersaturation ratios and different free sulfate concentrations. Crystal Research and Technology
39: 313-321 DOI 10.1002/crat.200310188
Antony A, Low JH, Gray S, Childress AE, Le-Clech P, Leslie G (2011) Scale formation and control in
high pressure membrane water treatment systems: A review. Journal of Membrane Science 383:
1-16 DOI http://dx.doi.org/10.1016/j.memsci.2011.08.054
Boels L, Witkamp G-J (2011) Carboxymethyl Inulin Biopolymers: A Green Alternative for
Phosphonate Calcium Carbonate Growth Inhibitors. Crystal Growth & Design 11: 4155-4165
DOI 10.1021/cg2007183
Greenlee LF, Testa F, Lawler DF, Freeman BD, Moulin P (2010) The effect of antiscalant addition
on calcium carbonate precipitation for a simplified synthetic brackish water reverse
osmosis concentrate. Water research 44: 2957-2969 DOI http://dx.doi.org/10.1016/j.
watres.2010.02.024
Guan X (2009) Kinetics Studies of Reactions at Solid-Liquid Interface: Simulation of Calcification
VDM Verlag
Kashchiev D (2000) Nucleation Elsevier Science
Kucera J (2010) Reverse Osmosis Membrane Fouling ControlThe Science and Technology of Industrial
Water Treatment:247-270.
Mangal MN (2023) Controlling scaling in groundwater reverse osmosis: minimizing antiscalant
consumption Veenman, The Netherlands
Mangal MN, Salinas-Rodriguez SG, Blankert B, Yangali-Quintanilla VA, Schippers JC, van der Meer
WGJ, Kennedy MD (2021) Role of phosphate and humic substances in controlling calcium
carbonate scaling in a groundwater reverse osmosis system. Journal of Environmental Chemical
Engineering 9: 105651 DOI https://doi.org/10.1016/j.jece.2021.105651
Mangal MN, Salinas-Rodriguez SG, Dusseldorp J, Blankert B, Yangali-Quintanilla VA, Kemperman AJB,
Schippers JC, van der Meer WGJ, Kennedy MD (2022) Foulant Identification and Performance
Evaluation of Antiscalants in Increasing the Recovery of a Reverse Osmosis System Treating
Anaerobic Groundwater. Membranes 12: 290
Mangal MN, Salinas-Rodriguez SG, Dusseldorp J, Kemperman AJB, Schippers JC, Kennedy MD, van
der Meer WGJ (2021) Effectiveness of antiscalants in preventing calcium phosphate scaling
in reverse osmosis applications. Journal of Membrane Science 623: 119090 DOI https://doi.
org/10.1016/j.memsci.2021.119090
Mangal MN, Yangali-Quintanilla VA, Salinas-Rodriguez SG, Dusseldorp J, Blankert B, Kemperman
AJB, Schippers JC, Kennedy MD, van der Meer WGJ (2022) Application of a smart dosing
pump algorithm in identifying real-time optimum dose of antiscalant in reverse osmosis
systems. Journal of Membrane Science 658: 120717 DOI https://doi.org/10.1016/j.
memsci.2022.120717
Mangal N, Salinas Rodriguez SG, Yangali Quintanilla VA, Schippers JC, Kennedy MD (2021) Ch 08 -
Scaling. In: Salinas Rodriguez SG, Schippers JC, Amy GL, Kim IS, Kennedy MD (eds) Seawater
Reverse Osmosis Desalination: Assessment and Pre-treatment of Fouling and Scaling, 1st
edn:207-242.
Prisciandaro M, Lancia A, Musmarra D (2001) Calcium Sulfate Dihydrate Nucleation in the Presence
of Calcium and Sodium Chloride Salts. Industrial & Engineering Chemistry Research 40: 2335-
2339 DOI 10.1021/ie000391q

245
Experimental Methods for Membrane Applications

Shih W-Y, Gao J, Rahardianto A, Glater J, Cohen Y, Gabelich CJ (2006) Ranking of antiscalant
performance for gypsum scale suppression in the presence of residual aluminum. Desalination
196: 280-292 DOI http://dx.doi.org/10.1016/j.desal.2006.04.001
Singh R (2005) Chapter 2 - Water and membrane treatment. In: Singh R (ed) Hybrid Membrane
Systems for Water Purification:57-130.
Söhnel O, Mullin JW (1978) A method for the determination of precipitation induction periods. Journal
of Crystal Growth 44: 377-382 DOI http://dx.doi.org/10.1016/0022-0248(78)90002-7
van Engelen G, Nolles R (2013) A sustainable antiscalant for RO processes. Desalination and water
treatment 51: 921-923 DOI 10.1080/19443994.2012.700787
Verdoes D, Kashchiev D, van Rosmalen GM (1992) Determination of nucleation and growth rates from
induction times in seeded and unseeded precipitation of calcium carbonate. Journal of Crystal
Growth 118: 401-413 DOI http://dx.doi.org/10.1016/0022-0248(92)90089-2
Waly T (2011) Minimizing the Use of Chemicals to Control Scaling in Sea Water Reverse Osmosis:
Improved Prediction of the Scaling Potential of Calcium Carbonate CRC Press/Balkema. Leiden.
Netherlands
Waly T, Kennedy MD, Witkamp G-J, Amy G, Schippers JC (2012) The role of inorganic ions in the
calcium carbonate scaling of seawater reverse osmosis systems. Desalination 284: 279-287 DOI
http://dx.doi.org/10.1016/j.desal.2011.09.012
Waly T, Munoz R, Kennedy MD, Witkamp GJ, Amy G, Schippers JC (2010) Effect of particles on the
induction time of calcium carbonate in synthetic SWRO concentrate. Desalination and water
treatment 18: 103-111 DOI 10.5004/dwt.2010.1351
Yu W, Song D, Chen W (2020) Antiscalants in RO membrane scaling control. Water research 183:
115985 DOI 10.1016/j.watres.2020.115985

246
 Part 4
Organic fouling
 doi: 10.2166/9781789062977_0249

Chapter 11

Practical Considerations of
Using LC-OCD for Organic
Matter Analysis in Seawater
Barun Lal Karna, Helen Rutlidge, Rita Kay Henderson, Pierre Le-Clech,

University of New South Wales, Australia

The learning objectives of this chapter are the following:

• Understand basic principles of LC-OCD measurement

• Assess the impact of salinity on LC-OCD analysis

• Appreciate the benefits and limitations of LC-OCD measurement for marine samples

11.1 INTRODUCTION

Reverse osmosis (RO) is a well-established treatment process used for the production
of drinking water from brackish water and seawater. However, RO systems suffer from
fouling of the membrane by dissolved solids, micro-organisms, and dissolved organic
matter (Matin et al., 2011; Prihasto et al., 2009). While an integrated membrane system
approach involving pre-treatment options has been partially successful in controlling most
particulate fouling problems, the more persistent problems of organic and biological fouling
persists, influencing the operation and maintenance costs (Prihasto et al., 2009).

Many studies have investigated the characterisation of organic matter (OM) in surface water
(Fan et al., 2001; Kennedy et al., 2005; Matilainen et al., 2011; Swietlik et al., 2004); however,
the lower concentration of organic matter in seawater initially presented some analytical
challenges in performing OM characterisation in seawater (Spyres et al., 2000). In seawater,

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

organics are mainly produced by micro-organisms rather than discharge from industrial
effluents, municipal wastewater, runoff rainwater and naturally decaying vegetation (Azam
et al., 1983; Cai et al., 2012; Parsons and Strickland, 1961). Both bacteria and algae release
neutral and charged polysaccharides, proteins and other organic compounds including
lipids, nucleic acids into the seawater environment (Voutchkov, 2008). As OM comprises
a heterogeneous mix of molecules at very low concentrations in seawater, high-resolution
techniques are necessary to qualify and quantify the marine organic fractions. Furthermore,
the high concentrations of Na+ and Cl- ions (together ~85%), along with other species of ions
(Ca2+, Mg2+, SO42-, K+, HCO3-, CO32-, Br- etc.) at low concentration (altogether ~15%) in the
seawater matrix adds an additional challenge to their analytical characterisation (Mopper et
al., 2007).

Many techniques, such as fluorescence spectroscopy, UV-vis spectroscopy, infrared

spectroscopy, nuclear magnetic resonance (NMR), size exclusion chromatography (SEC)
have been used to analyse the composition of OM of seawater. The main advantages and
disadvantages associated with these techniques are summarised in Table 1. The minimal
sample preparation, and range of quantitative composition information obtained from SEC
makes it an ideal technique for OM analysis in seawater.

SEC is a widespread method that allows the separation of macromolecular complexes,

such as proteins and carbohydrate polymers OM, by exploiting the size and chemical
functionality (Mecozzi et al., 2001). Liquid chromatography-organic carbon detection
(LC-OCD) is an advanced technique that uses SEC principles and is well suited to measure
organic compounds present in a range of aquatic water samples (Huber et al., 2011). This
established technique provides wide-ranging information regarding the nature of the organic
matter [e.g., size (molecular weight distribution), structure (aromatic or aliphatic), and
functionality (charge density i.e., humic and fulvic acids)] with minimal sample preparation
(Filloux et al., 2012; Henderson et al., 2011). LC-OCD is a proprietary technique and there
is only one supplier of the instrumentation and the associated integration software (DOC-
Labor GmbH, 2023).

250
 Chapter 11

Table 1 Advantages and disadvantages of analytical techniques of organic carbon for the analysis
of marine samples.
Analytical
techniques Advantages Disadvantages References
Infrared - comprehensive structural - requires the isolation of OM (Chon et al.
spectroscopy information. from abundant salt from 2020; Lee
- quick analysis. seawater. et al. 2020;
- accurate wave number. - the influence of the refractive Al-Juboori and
index of the surrounding Yusaf, 2012)
environment on the analyses.
- only qualitative determination
of lipids and hydrocarbons for
algae.
Nuclear magnetic - simple and fast method for - requires the isolation of OM (Al-Juboori and
resonance (NMR) samples with sufficient from abundant salt from Yusaf 2012;
concentration. seawater. Benner et al.
- comprehensive structural - low sensitivity with nominal 1992; Mopper
information. detection limits in the millimolar et al. 2007;
- a non-destructive and range. Jeong et al.
non-invasive mean to obtain 2013)
information regarding the
metabolic pathways.
UV-vis - quick test, - detects aromatic compounds (Chon et al.
spectroscopy - requires minimal sample only. 2020; Lee et al.
preparation - qualitative composition of humic 2020; Murphy
substance. et al. 2008; Liu
et al. 2022
Fluorescence - high level of sensitivity, - does not detect aliphatic (Lee et al.
spectroscopy - quick test. compounds. 2020; Murphy
- requires minimal sample - cannot differentiate between et al. 2008; Liu
preparation. overlapping fluorescing et al. 2022; Liu
- characterises marine and components. et al. 2023)
terrestrial DOM.
- can study the aggregation
mechanism of OM, and OM
dynamics with the use of flow
through cells.
Size exclusion - requires minimal sample - a relative molecular weight (Lee et al. 2020;
chromatography preparation. technique. Valladares
(SEC) - qualitative and quantitative - column must be calibrated Linares et al.
determination. with polymer standards of 2012; Jeong
- separation of organic fractions known molecular weight, data et al. 2022;
based on molecular weights. acquisition and processing is Tansakul et al.,
- determination of fatty acid, critical. 2011)
analysis of carbohydrates and
amino acids.
- lower temperature during
analysis.

251
Experimental Methods for Membrane Applications

The principles behind organic fractionation by LC-OCD are based on three separation
processes namely: size exclusion, ion interaction and hydrophobic interaction. Since OM
is very heterogeneous, size exclusion is the most dominant mechanism of separation.
Chromatography by size exclusion uses the difference in speed of diffusion for smaller and
larger molecules. The stationary phase is a packing of porous beads which allows smaller
molecules to diffuse into the bead interior. Consequently, large molecules travel faster than
small molecules. According to molecular size/weight and polarity, the column separates
OM into five fractions: (a) biopolymers (BP), comprising polysaccharides and proteins,
(b) humic substances (HS) such as humic and fulvic acids, (c) building blocks (BB) such as
hydrolysates of humics, (d) low molecular weight acids (LMWA) such as fraction for all
aliphatic low-molar-weight organic acids and (e) low molecular weight neutrals (LMWN)
such as alcohols, aldehydes, ketones and amino acids (Huber et al., 2011). A sixth fraction,
hydrophobic organic carbon (HOC) is determined as the difference between the total
DOC (that passes through a bypass column) and the total amount that elutes from the
chromatographic column. These fractions were originally identified and named based on
surface water analysis by Huber et al. ( 2011) and have become the conventional labels
for the fractions which will be followed in this work, however it is acknowledged that
these labels may not be the most appropriate for different water types, including seawater
samples. Due to the low concentration of organics in seawater, and the saline matrix that
is challenging for traditional single column LC-OCD, the use of dual column LC-OCD has
become more common for seawater samples, which is explored further in Section 11.2.2.

In this chapter, the impact of salinity on LC-OCD is explored and some analytical
considerations for the use of LC-OCD for saline samples is discussed, including the
reproducibility, reliability and sensitivity of LC-OCD in saline environments. Previous
applications are also explored to show the potential of LC-OCD for organic characterisation
in RO applications.

11.2 LC-OCD ANALYSIS

11.2.1 Instrumentation and chromatogram integration

A LC-OCD instrument uses a phosphate buffer solution that is irradiated in a UV batch
reactor to eliminate organic impurities prior to sample addition. A mobile phase (buffer
solution at pH of 6.85 requiring 2.5 g KH2PO4 + 1.5 g Na2HPO4 • 2H2O to 1 L) is delivered
at a flow rate of 1.1 mL/min to an autosampler with 1 mL injection volume for samples with
a concentration higher than 1 mgC/L (Huber et al., 2011); otherwise, an injection volume
up to a maximum of 4 mL can be used. Original systems used a TSK HW-50S (Toso, Japan)
chromatographic column for separation, however due to a change in manufacturing this
column is no longer suitable for LC-OCD systems. The first detector after chromatographic
separation is nondestructive, fixed wavelength UV-detection (UVD 254 nm), after which
the sample passes through an organic carbon detector (OCD) and a dissolved organic
nitrogen (DON) detector (Huber and Frimmel, 1991). The OCD uses a Grantzel thin-
film reactor in which organics are oxidized to CO2 by UV before measurement by infrared
detection.

252
 Chapter 11

Quantification of the fractions is conducted by area integration of chromatograms with

ChromCALC software (DOC-LABOR, Germany) (Huber et al., 2011). This software
has limitation that raw data integration is done either assuming sample contains humic
substances or without humic substances (biopolymers). Evaluation of chromatograms
containing HS should be assessed by checking that there is a corresponding humic peak
in the UVD chromatogram. The area under the curves is calculated as area units (AU) and
then is converted into corresponding concentration in μg/L (ppb) using calibration curves
where the calibration of OCD, UVD and OND are undertaken with a mixture of standards
comprising potassium hydrogen phthalate and potassium nitrate. More details of the data
integration process can be obtained from the literature (Huber et al., 2011).

In addition to the quantitative assessment of concentration by integration, the aromaticity

and nominal molecular weight of the HS fraction is calculated, and these values can be plotted
in what is known as a HS-diagram (Huber and Frimmel, 1991). This HS-diagram includes
a range of water samples, including HS standards of the International Humics Substances
Society (IHSS) and natural surface waters, and this is known as the humification pathway,
with transitions from aquagenic fulvic acids in the lower left of the diagram to pedogenic
fulvic acids to pedogenic humic acids in the upper right of the diagram (Huber et al., 2011).
Seawater samples tend to have values in the ranges 0-1.2 L/(mg·m) for aromaticity and
430-610 g/mol nominal molecular weight of the HS fraction.

As an example of the ChromCALC software, the chromatograms of OCD, UVD and OND
from a mixture of model compounds are shown in Figure 1 as black, blue and green lines
respectively. The model compounds were selected to simulate OM found in seawater and
represent the different LC-OCD fractions as per Table 2.

Table 2 Composition of DOM model mixture

Model Compound LC-OCD Fraction Concentration (mgC/L)
Bovine serum albumin (BSA) BP (protein) 0.76
Alginate BP (polysaccharide) 0.11
Xanthan BP (polysaccharide) 0.05
Humic acid HS 0.48
Tryptophan LMWN 7x10-5
Oxalic acid LMWA 0.06

253
Experimental Methods for Membrane Applications

5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0

Building blocks
1.5
A
Biopolymers B 1.0
LMW-Acids
0.5
C D
1.0
Humics

E 0.5
OCD
(UVD@254 nm, OCD)

B LMW-Neutrals 0.0
A C
D E UVD -0.5
Signal response

A
C -1.0
B
-1.5
OND -2.0
-2.5
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 ¹35 140 145 150
Time (min)

Figure 1 Chromatograms of OCD, UVD and OND for mixture of six model compounds with
Milli-Q®, using single column LC-OCD, with each fraction identified. The letters (A-E)
represent the various points that are set by the operator as part of the integration, where
A is humics peak max., B is humics left slope, C is humics right slope (division between
humics and building blocks), D is division between building blocks and acids, and E is
division between LWN-acids and LMW-neutrals.

11.2.2 Effect of salinity on organic characterization and calibration

There are known interferences from salts that can impact the LC-OCD calibration and OM
characterization in saline matrices (Baghoth et al., 2008; Huber and Frimmel, 1994). The
salt content in seawater samples has been shown to reduce the resolution of the individual
fractions (Baghoth et al., 2008). To overcome this issue, the use of two chromatographic
columns in series (dual LC-OCD) connected in series has been suggested. To better illustrate
the impact of the duplication of columns, a comparison between single and dual LC-OCDs
for a sample containing model compounds in a NaCl (32 g/L) matrix is shown in Figure
2 (Karna, 2014). The use of dual column leads to longer retention times, and this allows
the resolution to be improved with the HS, BB and LMWA peaks clearly separated. As
seawater samples often contain low levels of organics, it is therefore recommended to use a
dual column LC-OCD to obtain better qualitative and quantitative analysis (Dulaquais et al.
2018; Jeong et al., 2016; Karna, 2014).

The LMWN region of a LC-OCD chromatogram can be impacted due to the presence of salts
in a sample. It is well understood that chloride causes an interference in nitrate quantification
using UV detection (Sah, 1994). This prevents the UVD chromatogram being used to
determine nitrate or total nitrogen in salty water. However, there is no salt interference in
the BP or HS regions of the OND chromatogram, and the organic nitrogen content can still
be accurately determined by LC-OCD (Dulaquais et al., 2018). The LMWN region can also

254
 Chapter 11

be influenced by salt-induced ‘column bleeding’, where the salts allow the partial elution of
the accumulated hydrophobic material (from previous samples) from the column (Huber
and Frimmel, 1994). This can lead to the measured negative concentrations of the HOC
fraction that have previously been observed (Karna, 2014).

7
a)

LMW-Acids and HS
5
Humics

Biopolymers 4
DOC (by-pass)
3

2
LMW-Neutrals
1

Relative signal response 0

0 20 40 60 80 100 120

3
b)
LMW-Acids and HS
Building blocks
Humics
Biopolymers
DOC (by-pass)
2

LMW-Neutrals

Relative signal response 0

0 20 40 60 80 100 120

Figure 2 Typical spectrum OCD (blue line) and OND (dotted line) from LC-OCD of model mixture
with NaCl (32 g/L) a) in single column and b) in dual column. Figure from (Karna, 2014).

The effect of salinity on the calibration of the individual fractions was investigated using
the same model compounds from Section 11.2.1 (Karna, 2014). Mixtures of the model
compounds at different concentrations were prepared in Milli-Q® water and NaCl and the
calibration in the different matrices was compared, as shown in Figure 3 and Table 3. There
were no significant differences observed between the slope of the calibration lines for all
model compounds in Milli-Q® water, except for BSA, which was attributed to the poorer
oxidation efficiency of the Granztel reactor for the higher molecular weight compounds
(Lankes et al., 2009; Li et al., 2019). When the model compounds were mixed in saline

255
Experimental Methods for Membrane Applications

solution, there was significant decrease of the gradient observed for humic acid, in addition
to BSA. It was suggested that this decrease in the humic acids in the presence of NaCl was
due to microflocculation/precipitation of humic substances and subsequent capture on
the in-line filter before passing through the LC-OCD column. Dulaquais et al. (2018), also
found calibrations differences due to salinity and recommended calibrating the LC-OCD
system with the same salinity as the samples (if the samples have similar salinity levels) or
determining the salinity dependence of the detector to estimate the adjusted calibrations.

1.5

– 1.0

0.5

Measured DOC (mgC/L)

0
0 0.5 1.0 1.5

Dosed concentration (mgC/L)

1.5

– 1.0

Oxalic acid
Xanthan
Humic acid
Tryptophan 0.5
BSA
Alginate
– PHP
Measured DOC (mgC/L)
0
0 0.5 1.0 1.5
Dosed concentration (mgC/L)

Figure 3 Calibration of individual model compounds measured from chromatographic peak,

compared with PHP standard (dotted line in green) in (a) Milli-Q® water (single) and (b)
NaCl (32g/L) solution (dual column).

256
 Chapter 11

Table 3 Theoretical and measured carbon content of model compounds assessed in freshwater
(Milli-Q®) and saline (NaCl) matrix
Average carbon content measured
Theoretical from LC-OCD (mg/L)
Carbon content
in Milli-Q® in NaCl (dual)
Model compounds Chemical formulae mg/L
(single)
BSA (BP) n/a n/a 0.23 0.18
Alginate (BP) (NaC6H7O6)n* 0.73 0.67 0.61
Xanthan (BP) (C35H49O29)n * 0.90 0.63 0.56
Humic acid (HS) n/a n/a 0.91 0.55
Tryptophan (LMWN) H12C11O2N2 1.29 1.19 1.22
Oxalic acid (LMWA) H2C2O4 0.53 0.63 0.46

*assuming n=1

11.2.3 LEVEL OF DETECTION

The instrument detection limit (IDL) and lower level of detection (LLD) of organic
compounds have been assessed in two studies one using model compounds (Karna, 2014)
and different types of blank samples (Dulaquais et al., 2018), shown in Table 4. In each study,
to determine the IDL, ten samples of Milli-Q® water as blank were assessed through LC-
OCD and the IDL was calculated as three times the standard deviation of the blank sample. In
addition, Dulaquais et al. (2018) also assessed the IDL for a seawater blank, through using an
UV-irradiated seawater sample. Both studies had a similar IDL for total DOC with Milli-Q®
water of 20 and 30 μg/L, and the LLD were similar for most samples, except for alginate and
tryptophan, this could be due to lower oxidation efficiencies for higher molecular weight
compounds and N-heterocyclic compounds (Li et al., 2019). For the Milli-Q® samples the
IDL and LLD for the BP, HS, BB and LMWA were all below 10 μg/L, which indicates that the
quantification of these fractions is accurate, even at low concentrations. The higher IDL and
LLD for the saline matrix was due to salt interference, which is discussed in Section 11.2.5.

11.2.4 REPRODUCIBILITY OF LC-OCD

The reproducibility of LC-OCD for the low concentration of organics present in seawater
has been assessed in a couple of studies, and the results are given in Table 3 (Dulaquais
et al., 2018; Karna, 2014). For the Dulaquais et al., (2018) study, the reproducibility was
assessed using 10 non-consecutive analyses of a coastal seawater sample; while in the Karna
(2014) study, triplicate analysis of model compounds were used. The reproducibility for
all fractions was lower than 10%, indicating that LC-OCD can provide reasonably accurate
quantification OM. The %RSD values were higher for the model compounds when NaCl
was added, indicating the impact of a saline matrix.

257
Experimental Methods for Membrane Applications

Table 4 Instrument detection limit (IDL) and lower level of detection (LLD) for LC-OCD, using
Milli-Q® water, and UV-irradiated seawater (Bay of Brest, salinity = 35 g/L). Data taken
from (Dulaquais et al., 2018; Karna, 2014). N.B. for the Karna study (Karna, 2014) the
LLD was only reported for total DOC instead of the relevant fraction.
Sample DOC BP HS BB LMWN LMWA Reference
Milli-Q® water 20 2 2 1 10 2 [1]
Irradiated seawater 104 1 1 0 96 6 [1]
Milli-Q® water 30 N/A N/A N/A N/A N/A [2]
LLD (µg/L)
Milli-Q® water 30 5 5 4 19 5 [1]
Irradiated seawater 131 5 3 1 131 18 [1]
BSA (BP) 40 N/A N/A N/A N/A N/A [2]
Alginate (BP) 90 N/A N/A N/A N/A N/A [2]
Xanthan (BP) 30 N/A N/A N/A N/A N/A [2]
Humic acid (HS) 30 N/A N/A N/A N/A N/A [2]
Tryptophan 80 N/A N/A N/A N/A N/A [2]
(LMWN)
Oxalic acid (LMWA) 40 N/A N/A N/A N/A N/A [2]

[1] Dulaquais et al., 2018, [2] Karna, 2014

Table 5 Reproducibility for LC-OCD, reported as % relative standard deviation (RSD), using
coastal seawater (Bay of Brest, salinity = 35 g/L) and model compounds. BDL indicates
that signal was below the detection limit. Data taken from (Dulaquais et al., 2018; Karna,
2014). N.B. for the Karna study (Karna, 2014) the reproducibility was only reported for
total DOC instead of relevant fraction.
Reproducibility (% RSD)
Sample DOC BP HS BB LMWN LMWA Study
Sample DOC BP HS BB LMWN LMWA Study
Coastal seawater 2.9 8.3 2.5 7.5 5.2 BDL [1]
Alginate (BP) 1.6 0.7 BDL BDL BDL BDL [2]
Humic acid (HS) 0.9 BDL 0.5 BDL BDL BDL [2]
Alginate (BP) + NaCl 3.3 N/A N/A N/A N/A N/A [2]
Humic acid (HS) + NaCl 3.5 N/A N/A N/A N/A N/A [2]
Tryptophan (LMWN) + NaCl 12.7 N/A N/A N/A N/A N/A [2]
Oxalic acid (LMWA) + NaCl 1.4 N/A N/A N/A N/A N/A [2]

[1] Dulaquais et al., 2018, [2] Karna, 2014

258
 Chapter 11

11.2.5 CHARACTERISATION OF ORGANIC MIXTURES

The impact of the salt matrices on the organic characterization by LC-OCD has been
previously studied, using a mixture of 3 mg/L of six model compounds (as listed in Table 4,
and prepared in equal proportion, i.e., 0.5 mg/L by weight) in Milli-Q® and saline matrixes
(NaCl, NaCl with added Ca2+ and Mg2+, and Red Sea salt) (Karna, 2014). The concentration
of NaCl and red sea salt was 32 g/L. There was no significant difference observed between
the fraction concentrations prepared in Milli-Q® and NaCl. However, a significant
decrease of BP, HS and BB were observed for the red sea salt and NaCl with the added ions
in comparison with the NaCl solution alone, while the LMWN and LMWA concentration
increased. These differences were attributed to the presence of divalent ions in red sea
salt solution, such as Ca2+, which promoted the formation of organic (especially alginate)
complexes (Kye et al., 2021) that were potentially retained on the membrane surface (0.45
μm filter) before entering the LC-OCD chromatographic columns. In addition, divalent
cations interact specifically with humic carboxyl functional groups and, thus, substantially
reducing the humic charge and the electrostatic repulsion between humic macromolecules.
Reduced organic interchain repulsion can result in increased organics deposition on the
membrane surface (0.45 μm filter) and formation of a densely packed fouling layer before
entering the LC-OCD chromatographic columns (Hong and Elimelech, 1997). Therefore, it
is important to note that potentially some natural seawater organics will not be measured by
LC-OCD as they would be removed prior to entering the LC-OCD column due to fouling of
the filter and/or formation of complexes.

0.7

0.6

0.5

0.4
LMW-N&A
0.3
BB
HS 0.2

BP 0.1
Corrected concentration 0
(mg/L) MQ NaCl NaCl with Ca2+ Red sea salt
and Mg2+

Figure 4 Effect of background matrix on quantification of organic fractions for model mixture in
Milli-Q® (single column), in 32 g/L NaCl (dual column), NaCl with added divalent ions
(dual column) and red sea salt solution (dual column).

11.2.6 Applications
LC-OCD has been applied to brackish and seawater samples, often with other characterization
techniques, with the aim to improve the effectiveness of treatment processes. This has
included investigating the fouling behaviour of organic matter in RO and other membrane

259
Experimental Methods for Membrane Applications

processes (Fortunato et al., 2020; Jeong et al., 2016; Park et al., 2019; Yin et al., 2019), and
the effectiveness of pretreatment techniques in the removal of organic matter (Fortunato et
al., 2020; Jeong et al., 2016; Kye et al., 2021; Simon, Rudé et al., 2013; Simon, Penru et al.,
2013).

11.2.6.1 OM composition in seawater

In seawater samples, HS and LMWN tend to be the most abundant fractions, with BP
and LMWA present in the lowest concentrations (Table 5). Jeong et al., (2016) reported
differences in the DOM concentration due to an increase in LMWN in the warm season
compared to the cool season. This increase was attributed to increased algal/microbial
action in the surface seawater due to higher temperatures which is then photodegraded to
LMWN (Jeong et al., 2016).

Due to differing aromaticity and nominal molecular weight values there was grouping of
samples based different sea areas at the Korean Peninsula, enabling them to be differentiated
on the HS-diagram (Kye et al., 2021).

Table 6 DOM composition from LCOCD for different seawater samples. Low Molecular Weight
(LMW) represents the combined LMWN and LMWA fractions.
DOC BP HS BB LMWN LMWA LMW
Sample (mgC/L) (%) (%) (%) (%) (%) (%) Ref
Rottnest Island, 0.98 12.2 52.0 8.2 26.5 1.0 27.5 Rutlidge et al.,
Western Australia 2021
Sydney Harbour 1.01 14.9 41.6 13.9 26.7 4.0 30.7 Karna, 2014
NW 1.14 5.6 33.2 17.6 38.4 5.2 43.6 Simon, Rudé
Mediterranean et al., 2013
Sea
NW 1.40 7.5 25.9 15.8 45.6 5.2 50.8 Simon, Penru
Mediterranean et al., 2013
Sea (Barcelona)
North Sea 1.35 4.7 38.2 16.0 33.9 7.2 41.1 Salinas, 2011

Singapore 1.18 6.7 42.1 15.5 NR NR 30.2 Yin et al.,

2019
Red Sea 1.33 6.8 10.5 1.5 81.2 NR Fortunato et
al., 2020
Kwinana, Perth, 1.50 5.9 34.0 NR 50.3 NR Jeong et al.,
Western Australia 2016
Korean Peninsula 2.02 NR 36.1 NR 32.5 NR Kye et al.,
2021

NR: not reported

260
 Chapter 11

11.2.6.2 Fouling behaviour of organic matter

Organic constituents of extracted foulants from RO autopsies have been investigated by
LC-OCD, with different fractions being identified. For a specific Australian study, the main
organic foulants on the cartridge filters were BP and HS, while the dominant fractions in the
incoming raw seawater, HS and LMWN, were the main organic foulants for RO (Jeong et
al., 2016). While another study from a desalination plant located on the Red Sea found that
LMWN was the main organic foulant for RO, followed by either HS or BP depending on the
position in the module (Fortunato et al., 2020). However, Yin et al. (2019) found that BP
followed by LMWN had a greater impact on RO membrane fouling.

11.2.6.3 Effectiveness of pretreatment methods

LC-OCD has also been used to assess the effectiveness of various pretreatment methods for
RO of seawater as per some examples below:
• Ozonation reduced the aromaticity and molecular weight of the OM in seawater samples,
with the aromaticity rapidly decreasing and then stabilising after 12 hr (Kye et al., 2021).
• The use of a biofilter in Mediterranean desalination pilot plant resulted in the reduction
of the LMWN fraction, followed by BB and BP fractions (Simon et al., 2013).
• The existing pretreatment process (i.e., coagulation, followed by dual media filtration and
cartridge filtration) at a Perth Seawater Desalination Plant presented only small removal
of OM, with approximately 16% DOM removal, which was predominately the larger BP
and HS, as expected (Jeong et al., 2016).

11.3 CONCLUSIONS

LC-OCD is a suitable technique for measuring the organic matter concentration and
composition in seawater samples with minimal sample preparation. LC-OCD can
reliably measure the low concentrations of organics generally present in seawater, with
the concentration of individual fractions being well above the LLD. However, due to the
fouling of the filter and the formation of complexes a small portion of natural seawater
OM may not be measured by LC-OCD. Due to improved resolution of the chromatogram,
it is recommended to use dual chromatographic columns than a single column for saline
samples.

261
Experimental Methods for Membrane Applications

11.4 REFERENCES

Al-Juboori, Raed A., and Talal Yusaf. 2012. ‘Biofouling in RO system: Mechanisms, monitoring and
controlling’, Desalination, 302: 1-23.
Azam, F, T Fenchel, JG Field, JS Gray, LA Meyer-Reil, and F Thingstad. 1983. ‘The Ecological Role of
Water-Column Microbes in the Sea’, Marine Ecology Progress Series, 10: 257-63.
Baghoth, S. A., S. K. Maeng, S. G. Salinas Rodríguez, M. Ronteltap, S. Sharma, M. Kennedy, and G.
L. Amy. 2008. ‘An urban water cycle perspective of natural organic matter (NOM): NOM in
drinking water, wastewater effluent, storm water, and seawater’, Water Science & Technology,
8: 701-07.
Benner, Ronald, J. Dean Pakulski, Matthew McCarthy, John I. Hedges, and Patrick G. Hatcher. 1992.
‘Bulk Chemical Characteristics of Dissolved Organic Matter in the Ocean’, Science, 255: 1561-
64.
Cai, Yihua, Laodong Guo, Xuri Wang, Allison K. Mojzis, and Donald G. Redalje. 2012. ‘The source and
distribution of dissolved and particulate organic matter in the Bay of St. Louis, northern Gulf of
Mexico’, Estuarine, Coastal and Shelf Science, 96: 96-104.
Chon, Kangmin, Namjo Jeong, Hojung Rho, Joo-Youn Nam, Eunjin Jwa, and Jaeweon Cho. 2020.
‘Fouling characteristics of dissolved organic matter in fresh water and seawater compartments of
reverse electrodialysis under natural water conditions’, Desalination, 496: 114478.
DOC-Labor GmbH. 2023. ‘Screening of Organics in Natural and Technical Waters’, Accessed
9/07/2023. http://doc-labor.de/.
Dulaquais, Gabriel, Johann Breitenstein, Matthieu Waeles, Rémi Marsac, and Ricardo Riso.
2018. ‘Measuring dissolved organic matter in estuarine and marine waters: size-exclusion
chromatography with various detection methods’, Environmental Chemistry, 15: 436-49.
Fan, Linhua, John L. Harris, Felicity A. Roddick, and Nic A. Booker. 2001. ‘Influence of the characteristics
of natural organic matter on the fouling of microfiltration membranes’, Water Research, 35:
4455-63.
Filloux, E., J. Labanowski, and J. P. Croue. 2012. ‘Understanding the fouling of UF/MF hollow fibres
of biologically treated wastewaters using advanced EfOM characterization and statistical tools’,
Bioresource Technology, 118: 460-68.
Fortunato, Luca, Abdullah H. Alshahri, Andreia S. F. Farinha, Islam Zakzouk, Sanghyun Jeong,
and TorOve Leiknes. 2020. ‘Fouling investigation of a full-scale seawater reverse osmosis
desalination (SWRO) plant on the Red Sea: Membrane autopsy and pretreatment efficiency’,
Desalination, 496: 114536.
Henderson, Rita K., Nashida Subhi, Alice Antony, Stuart J. Khan, Kathleen R. Murphy, Greg L. Leslie,
Vicki Chen, Richard M. Stuetz, and Pierre Le-Clech. 2011. ‘Evaluation of effluent organic matter
fouling in ultrafiltration treatment using advanced organic characterisation techniques’, Journal
of Membrane Science, 382: 50-59.
Hong, Seungkwan, and Menachem Elimelech. 1997. ‘Chemical and physical aspects of natural organic
matter (NOM) fouling of nanofiltration membranes’, Journal of Membrane Science, 132: 159-
81.
Huber, S. A., and F. H. Frimmel. 1994. ‘Direct Gel Chromatographic Characterization and Quantification
of Marine Dissolved Organic Carbon Using High-Sensitivity DOC Detection’, Environ Sci
Technol, 28: 1194-7.
Huber, Stefan A., Andreas Balz, Michael Abert, and Wouter Pronk. 2011. ‘Characterisation of aquatic
humic and non-humic matter with size-exclusion chromatography – organic carbon detection –
organic nitrogen detection (LC-OCD-OND)’, Water Research, 45: 879-85.

262
 Chapter 11

Huber, Stefan A., and Fritz H. Frimmel. 1991. ‘Flow injection analysis for organic and inorganic carbon
in the low-ppb range’, Analytical Chemistry, 63: 2122-30.
Jeong, Ganghyeon, Hyeonho Lee, Chang-Min Kim, and Am Jang. 2022. ‘Size-dependent transport and
fouling formation of organic matters in a pilot-scale PFFO–RO hybrid system for real wastewater
treatment’, Journal of Cleaner Production, 361: 132233.
Jeong, Sanghyun, Sung-Jo Kim, Chang Min Kim, Saravanamuthu Vigneswaran, Tien Vinh Nguyen, Ho-
Kyong Shon, Jaya Kandasamy, and In S. Kim. 2013. ‘A detailed organic matter characterization
of pretreated seawater using low pressure microfiltration hybrid systems’, Journal of Membrane
Science, 428: 290-300.
Jeong, Sanghyun, Gayathri Naidu, Robert Vollprecht, TorOve Leiknes, and Saravanamuthu
Vigneswaran. 2016. ‘In-depth analyses of organic matters in a full-scale seawater desalination
plant and an autopsy of reverse osmosis membrane’, Separation and Purification Technology,
162: 171-79.
Karna, Barun Lal. 2014. ‘Advanced characterisation techniques to assess seawater organic matter
removal by dissolved air flotation (DAF)’, UNSW Australia.
Kennedy, Maria D., Hyoung K. Chun, Victor A. Quintanilla Yangali, Bas G. J. Heijman, and Jan
C. Schippers. 2005. ‘Natural organic matter (NOM) fouling of ultrafiltration membranes:
fractionation of NOM in surface water and characterisation by LC-OCD’, Desalination, 178: 73-
83.
Kye, Homin, Kiho Kim, Youmi Jung, Yirga Weldu Abrha, Seong-Nam Nam, Il-hwan Choi, Joon-Wun
Kang, and Yeojoon Yoon. 2021. ‘Characterization of marine dissolved organic matter and its
effect on ozonation’, Chemosphere, 277: 130332.
Lankes, Ulrich, Margit B. Müller, Matthias Weber, and Fritz H. Frimmel. 2009. ‘Reconsidering the
quantitative analysis of organic carbon concentrations in size exclusion chromatography’, Water
Research, 43: 915-24.
Lee, Yong-Gu, Sangwon Kim, Jaegwan Shin, Hojung Rho, Younggeun Lee, Young Mi Kim, Yongeun
Park, Sang-Eun Oh, Jaeweon Cho, and Kangmin Chon. 2020. ‘Fouling behavior of marine
organic matter in reverse osmosis membranes of a real-scale seawater desalination plant in South
Korea’, Desalination, 485: 114305.
Li, X., N. R. H. Rao, K. L. Linge, C. A. Joll, S. Khan, and R. K. Henderson. 2019. ‘An evaluation of
measurement techniques for algal-derived organic nitrogen’, Water Research, 165: 114998.
Liu, Jiajian, Li Ling, Qing Hu, Chao Wang, and Chii Shang. 2022. ‘Effects of operating conditions on
disinfection by-product formation, calculated toxicity, and changes in organic matter structures
during seawater chlorination’, Water Research, 220: 118631.
Liu, Yang, Xiaofang Liu, Yujian Wen, and Jun Sun. 2023. ‘A snapshot on vertical variability of dissolved
organic matter in the epilagic zone of the eastern Indian Ocean’, Marine Pollution Bulletin, 192:
114985.
Matilainen, Anu, Egil T. Gjessing, Tanja Lahtinen, Leif Hed, Amit Bhatnagar, and Mika Sillanpää. 2011.
‘An overview of the methods used in the characterisation of natural organic matter (NOM) in
relation to drinking water treatment’, Chemosphere, 83: 1431-42.
Matin, Asif, Z. Khan, S. M. J. Zaidi, and M. C. Boyce. 2011. ‘Biofouling in reverse osmosis membranes
for seawater desalination: Phenomena and prevention’, Desalination, 281: 1-16.
Mecozzi, Mauro, Demetria Cardarilli, Eva Pietrantonio, and Marina Amici. 2001. ‘Estimation of
Similarity in the Qualitative Composition of Humic Substance in Marine Sediments By Means of
an Uv-Spectroscopic Library’, Chemistry and Ecology, 17: 239-54.

263
Experimental Methods for Membrane Applications

Mopper, Kenneth, Aron Stubbins, Jason D. Ritchie, Heidi M. Bialk, and Patrick G. Hatcher. 2007.
‘Advanced Instrumental Approaches for Characterization of Marine Dissolved Organic Matter:
Extraction Techniques, Mass Spectrometry, and Nuclear Magnetic Resonance Spectroscopy’,
Chemical Reviews, 107: 419-42.
Murphy, Kathleen R., Colin A. Stedmon, T. David Waite, and Gregory M. Ruiz. 2008. ‘Distinguishing
between terrestrial and autochthonous organic matter sources in marine environments using
fluorescence spectroscopy’, Marine Chemistry, 108: 40-58.
Park, Sanghun, Taewoo Nam, Jeongyeop You, Eun-Sik Kim, Ilhwan Choi, Jongkwan Park, and Kyung
Hwa Cho. 2019. ‘Evaluating membrane fouling potentials of dissolved organic matter in brackish
water’, Water Research, 149: 65-73.
Parsons, T. R., and J. D. H. Strickland. 1961. ‘On the production of particulate organic carbon by
heterotrophic processes in sea water’, Deep Sea Research (1953), 8: 211-22.
Prihasto, Noka, Qi-Feng Liu, and Seung-Hyun Kim. 2009. ‘Pre-treatment strategies for seawater
desalination by reverse osmosis system’, Desalination, 249: 308-16.
Rutlidge, Helen, Liza K. McDonough, Phetdala Oudone, Martin S. Andersen, Karina Meredith,
Khorshed Chinu, Mark Peterson, and Andy Baker. 2021. ‘Characterisation of groundwater
dissolved organic matter using LCOCD: Implications for water treatment’, Water Research, 188:
116422.
Sah, R. N. 1994. ‘Nitrate-nitrogen determination—a critical review’, Communications in Soil Science
and Plant Analysis, 25: 2841-69.
Salinas Rodríguez SG. 2011. Particulate and organic matter fouling of SWRO systems: Characterization,
modelling and applications CRC Press/Balkema, Delft. http://dx.doi.org/10.1201/b11609
Simon, F. Xavier, Ywann Penru, Andrea R. Guastalli, Santiago Esplugas, Joan Llorens, and Sylvie Baig.
2013. ‘NOM characterization by LC-OCD in a SWRO desalination line’, Desalination and Water
Treatment, 51: 1776-80.
Simon, F. Xavier, Elisabet Rudé, Joan Llorens, and Sylvie Baig. 2013. ‘Study on the removal of
biodegradable NOM from seawater using biofiltration’, Desalination, 316: 8-16.
Spyres, Georgina, Malcolm Nimmo, Paul J. Worsfold, Eric P. Achterberg, and Axel E. J. Miller. 2000.
‘Determination of dissolved organic carbon in seawater using high temperature catalytic
oxidation techniques’, TrAC Trends in Analytical Chemistry, 19: 498-506.
Swietlik, J., A. Dubrowska, U. Raczyk-Stanisławiak, and J. Nawrocki. 2004. ‘Reactivity of natural
organic matter fractions with chlorine dioxide and ozone’, Water Research, 38: 547-58.
Tansakul, Chatkaew, Stéphanie Laborie, and Corinne Cabassud. 2011. ‘Adsorption combined with
ultrafiltration to remove organic matter from seawater’, Water Research, 45: 6362-70.
Valladares Linares, Rodrigo, Victor Yangali-Quintanilla, Zhenyu Li, and Gary Amy. 2012. ‘NOM and
TEP fouling of a forward osmosis (FO) membrane: Foulant identification and cleaning’, Journal
of Membrane Science, 421-422: 217-24.
Voutchkov, Nikolay. 2008. Pre-treatment Technologies for Membrane Seawater Desalination
(Australian Water Association).
Yin, Wenqiang, Xin Li, Stanislaus Raditya Suwarno, Emile R. Cornelissen, and Tzyy Haur Chong.
2019. ‘Fouling behavior of isolated dissolved organic fractions from seawater in reverse osmosis
(RO) desalination process’, Water Research, 159: 385-96.

264
 doi: 10.2166/9781789062977_0265

Chapter 12

Fluorescence Excitation
Emission Matrix (EEM)
Spectroscopy
Adam C. Hambly & Urban J. Wünsch,

Technical University of Denmark, Denmark

The learning objectives of this chapter are the following:

• Understand the theoretical and historical background of fluorescence EEM

spectroscopy

• Outline EEM instrumentation and best practice for method development, with
consideration of potential shortcomings and interferences

• Present and discuss research literature for fluorescence EEM spectroscopy as applied
to the fouling of membrane-based water treatment systems

12.1 INTRODUCTION

Fluorescence is a form of luminescence, whereby light (energy) is absorbed by a substance at

a particular wavelength (excitation), and then emitted at a longer wavelength (lower energy,
emission). The difference between the absorption and fluorescence maxima is known as the
Stokes shift, and the entire process typically takes place over a very short timeframe. This is
one of the characteristics which separates it from phosphorescence, which typically takes
place over longer timeframes. These processes are best described in general by a Jablonski
diagram (Figure 1), named after the Polish physicist Aleksander Jabłonski (1933).

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

S2
Internal
conversion Intersystem
crossing
S1

T1
Absorption Fluoresence
hvA hvF hvP
hvA Phosphoresence
S0 21
0

2,000 DENS in 0.1 M

Phosphate buffer 1.0
pH 7.0 Et
N
1,500 Et

Fluorescence intensity
-
SO3

1,000
0.5
ε (m-1cm-1)

500

0 0
350 400 450 500 550 600
Wavelenght (nm)

Figure 1 Left Jabłonski diagram; and Right: An example of a compounds absorption (A) and
fluorescence emission (F) spectra, where the Stokes shift is the distance between the
peaks of the two spectra (Adapted from Lakowicz, 2006b).

Not all substances or compounds are capable of exhibiting fluorescence. However, since it
was first referred to in 1852 (Stokes, 1852), fluorescence-based methods have been used
to analytically detect and quantify specific compounds with high sensitivity. Fluorescence
spectroscopy is often described as being 10 to 1000 times more sensitive than absorbance
spectroscopy, due to the nature of the measurement having a ‘true’ zero. There are many
ways which fluorescence can and has been used analytically, from single point to simple
2D emission measurements at a single excitation wavelength, to synchronous fluorescence
and excitation-emission matrix (EEM) spectroscopy. Depending on the complexity of
the sample, spectra of mixtures such as organic matter, can often be deconstructed into
independently varying components using a variety of analytical tools (see section: 11.6
Data Processing). This chapter will focus on fluorescence EEM spectroscopy, and its practical
relevance in relation to membrane-based water treatment and fouling in general.

Broadly speaking, the analysis of FDOM (fluorescent dissolved organic matter) with EEM
spectroscopy found early use in the natural sciences, the bulk of which was first carried out
within oceanography. The methods have also been transferred to freshwater and estuarial
research applications, and subsequently have seen applications tied to water treatment
within a number of different fields. Fluorescence EEM spectroscopy has now been used to
detect and characterise organic content within the aquatic sciences, across a large variety of
natural and engineered systems. Oceans (Catalá et al., 2015; Stedmon and Nelson, 2015),

266
 Chapter 12

seas, lakes (Kothawala et al., 2014; Osburn et al., 2017), and rivers (Baker and Inverarity,
2004), as well as drinking water (Bridgeman et al., 2011), wastewater (Carstea et al., 2016),
water recycling (Hambly et al., 2010; Hambly et al., 2015; Henderson et al., 2009; Murphy
et al., 2011), aquaculture (Hambly et al., 2015; Spiliotopoulou et al., 2017; Yamin et al.,
2017), and desalination (Drozdova et al., 2017; Shutova et al., 2016) are some of the varied
aquatic settings in which fluorescence has been applied, from small scale research studies to
large scale industrial uses.

Within the context of this chapter, fluorescence EEM spectroscopy has shown particular
use in the analysis and understanding of membrane-based systems. As an optimised target
compound removal is paramount to the performance of any membrane system, any sort
of membrane fouling can therefore limit the systems performance. Feed waters often
contain a high level of organic compounds, and as such various forms of organic fouling of
the membranes can occur. The performance of a membrane system can be evaluated on the
FDOM analysis of different aspects of it, and depending on which aspect it is measuring,
the appropriate analytical method will require tweaking. Specific details of fluorescence
methods are thus contained in the following sections.

Numerous other aspects of the measurement also need to be addressed before fluorescence
data is ready for interpretation. While fluorescence measurements are somewhat simpler
when compared to e.g., liquid chromatography coupled to mass spectrometry, analysts must
consider many questions before measurements can take place. For example, how should
samples be taken and how long can they be stored? How exactly should fluorescence be
measured, and which analysis strategy is the best for a given study? The following sections
11.2 to 11.7 provide guidance for the practical aspects behind fluorescence measurements
within an aquatic context.

12.2 SAMPLING & STORAGE

Samples taken from different stages of membrane filtration consist of particulate and soluble
material in water. A fraction of this material is highly bioavailable to microorganisms, while
other fractions resist biodegradation for longer periods (Hu and Ren, 2019; Urgun-Demirtas
et al., 2008). When it comes to sampling and subsequent storage prior to measurements, the
more bioavailable material requires special attention, as microbes naturally target the most
available compounds first and can thus alter the sample quickly (Heinz and Zak, 2018).
In general, it is advisable to process samples and perform fluorescence measurements as
quickly as possible to avoid storage artefacts. However, the constraints of sampling do not
always allow for fast sample processing. In such cases, preserving the sample and knowing
about possible storage effects is important. Preservation strategies include filtration, storage
in cold and dark conditions, freezing, and chemical poisoning. In contrast, autoclaving
introduces changes in FDOM (Andersson et al., 2018).

Good practices for characterizing dissolved materials include 1) removing living organisms
as quickly as possible through filtration; 2) storing the sample at temperatures below
10 °C in the dark to minimize biological activity; 3) measuring as quickly as possible but
preferably at most within 5 days of sampling, and; 3) maintaining the same procedure for
sample processing within a study to keep potential biases constant.

267
Experimental Methods for Membrane Applications

Different filter materials and pore sizes are available for filtration. Amongst the available
materials, glass fiber filters are the safest option regarding contamination since they can
be ashed (> 4 h at > 450 °C, (Coble et al., 2014)). However, glass fibre filters generally
only exist with pore sizes of larger than 0.3 μm. The classic GF/F filter with a pore size of
0.7 μm was used to distinguish particulate from dissolved matter. However, at that pore
size, studies have reported bacterial passage rates between 10 and 25 % in marine samples
(Morán et al., 1999). It should however be noted that passage of microbes through filters
with all common diameters can be observed (Obayashi and Suzuki, 2019; Wang et al.,
2007, 2008). If initial cell counts are high and assimilable carbon is abundant, microbial
regrowth can quickly change sample character despite the usage of ‘sterile’ filters (< 0.2 μm).
These observations emphasize that storages times should be kept short and effects of
microbial passage will depend on the original sample. Lastly, the leaching of filter material
into the sample should be investigated for the specific batch of filters used in each study.
In the context of wastewater, such leaching is likely negligible but can affect primarily the
UVA fluorescence emission range due to the leaching of production-related impurities.
Filters should be rinsed with ultrapure water followed by sample prior to obtaining a filtrate
for analysis. Filtration should occur slowly to avoid the bursting of cells (Rosenstock and
Simon, 1993).

Freezing as a means to slow down biological processes can help to facilitate longer sample
storage. While some studies recommend freezing as suitable for samples with low carbon
concentrations, significant changes in optical indices, sample absorbance, and fluorescence
emission characteristics have been observed (Fellman et al., 2008; Heinz and Zak, 2018;
Spencer et al., 2007) concluded that the effects cannot be predicted from the composition of
the original sample. Thus, the effects cannot be corrected post-measurement.

12.3 BENCHTOP INSTRUMENTATION

Since fluorescence analyses are increasingly popular, users have a range of choices concerning
benchtop instrumentation. However, specifications regarding instrument and software
can differ significantly between instruments and affect the measurement experience and
resulting data quality.

Commercial instruments on the market today usually feature incandescent or pulsed Xenon
lamps. These lamps provide excitation light in the entire ultraviolet-visible spectrum
(approx. 220 – 800 nm) and have a relatively continuous emission spectrum. To understand
the instrument’s limitations, it is important to keep in mind the lamp’s output spectrum.
For example, incandescent Xe lamps provide little light at wavelengths shorter than
240 nm and resulting emission scans are generally noisier and can be difficult to interpret.
Xe flash lamps can provide more light in the ultraviolet range resulting in a wider usable
excitation range and a more uniform signal-to-noise relationship across EEMs (Lakowicz,
2006a). However, Xe flash lamps also contain more distinct emission bands, that need to
be addressed to obtain spectrally calibrated EEMs. Lastly, incandescent Xe lamps have a
lifetime of < 2000 h and thus require more maintenance compared to pulsed light sources.

268
 Chapter 12

The optical configuration of spectrofluorometers can differ significantly due to the

requirements dictated by the detection system. The classic photomultiplier tube (PMT) is a
sensitive photon detector that lacks the ability to distinguish light of different wavelengths
but offers superior sensitivity thanks to signal amplification and low noise levels.
Spectrofluorometers that utilize PMTs require two monochromators: The first selects a
narrow band of light for sample excitation, the second permits a narrow band of fluorescence
emission to pass through to the PMT. By scanning through all desired emission wavelengths
at all desired excitation wavelengths, an EEM is constructed with typical speeds of 500 nm/
min (total time typically between 20-40 min). On the other hand, charge-coupled device
detectors (CCDs) allow the simultaneous detection of the entire wavelength range of
interest and reduce measurement times considerably.

Due to the necessity to consider non-linearity in fluorescence observations due to the optical
density of a sample (discussed below), it is also important to consider the availability of
spectrophotometers during the measurement of fluorescence. If potential projects involve
field measurements, the use of instruments with a combined absorbance-fluorescence
detection system can be advantageous since all required measurements are made within one
unit (see Figure 2).

Figure 2 An example of a modern spectrofluorometer instrumentation setup in a research

laboratory – in this case an iteration of the Horiba Aqualog with a CCD detector.

As mentioned above, samples generally contain particulate and dissolved material. While
both fractions contain fluorescent moieties, different instrument configurations are required
to characterize the material. For example, dissolved fluorescent material is quantified after
filtration of a water sample through filters in the classic right-angle geometry with a quartz
cuvette (Figure 3, left). On the other hand, the measurement of particulate material occurs
either directly in the unfiltered, optically thick suspension or by exciting particles directly
on a surface. In both cases, the non-transparent nature of the particulate sample necessitates
a front-face illumination (Figure 3, right). Front-face measurements of thick suspensions
can also occur in cuvettes, but require adapters to either adjust the angle of the incident light
relative to the cuvette face or to capture and direct the light at a specific angle toward the
cuvette. The remainder of this section will discuss right-angle fluorescence of optically thin
solutions since this is by far the most common application of fluorescence.

269
Experimental Methods for Membrane Applications

Right-angle fluoresence Front-face fluoresence

– transparant solution – thick solution
– filtered samples – unfiltered samples & particles

Figure 3 Schematic view (top) of the geometric position of samples and cuvettes.

12.4 QUALITY ASSURANCE

Most of today’s spectrofluormeters are spectrally calibrated in full from the factory.
However, to generate comparable results, wavelength accuracy and spectral calibration are
especially important to monitor over time as the instrument ages. Wavelength accuracy
refers to the deviation between true and detected wavelength in nanometres; either the
excitation light intercepted by the cuvette or the fluorescence emission captured by the
detector. Manufacturers commonly list the specification in the instrument manual and
a precision of ± 1 nm is typical. Deviations are monitored by detecting the peak position
of Rayleigh scatter at e.g., 467 nm, while the accuracy of the emission detector can be
verified by determining the peak position of the 435.8 nm Hg band emitted by common
low-pressure mercury vapor lamps (Sansonetti et al., 1996). Monitoring changes in peak
positions is especially important after instrument transport.

Spectral calibration refers to the elimination of spectral biases that arise from a biased lamp
emission spectrum, and wavelength-dependent monochromator and detector biases. Some
of these biases are eliminated with the use of reference detectors, but the remainder of the
bias is removed with the use of excitation and emission correction factors that come pre-
installed from the factory. It can however be a good idea to verify their appropriateness
from time to time. A triangular cuvette of Rhodamine B produces a flat excitation spectrum
after the successful elimination of spectral biases (Kopf and Heinze, 1984). The emission
calibration is commonly verified with standards available from the National Institute
of Standards and Technology (NIST). Different standards cover the ultraviolet-visible
emission range and the recorded spectra should fall within the certified values at all times
(Gilmore, 2014).

To obtain valid results, the measured sample needs to meet certain criteria to ensure that
the instrument is capable of delivering the best results possible. For example, fluorescence
counts should not exceed the linear range of the detector. Fortunately, the instrument
control software usually warns users when the linearity threshold is exceeded. In such cases,
settings can be adjusted (e.g., integration time) or samples diluted.

In cases with high concentrations of chromophores, the sample transmission can be too low
to deliver quantitative fluorescence results. Kothawala et al., (2013) defined an absorbance
of 0.05 cm-1 (89 % transmittance) as the cutoff below which such effects can be safely

270
 Chapter 12

neglected (Kothawala et al., 2013). On the other hand, an absorbance of 1.5 cm-1 (3 %
transmittance) was found to be the upper limit after which no quantitative fluorescence
results can be obtained even if correction methods are applied (see below). It is thus
important that samples meet the second criterion during the fluorescence measurements, as
no post-measurements for linearity will be possible.

Prior to every study, choosing appropriate measurement settings is important to ensure

appropriate fluorescence counts in the relevant ranges of the EEM are accumulated. Regarding
range and resolution, the types of observed fluorophores and their properties will govern
which parts of the EEM should be captured. For example, if protein-like material, phenolic
compounds, and / or oils are present a high resolution in the excitation range below 300 nm
is especially important. To distinguish these highly similar fluorophores and quantify their
fluorescence, it can also be important to capture emission spectra with a resolution below
3 nm if bandwidth characteristics of monochromators permit this. For instruments with
incandescent Xe lamps, it generally makes little sense to capture emission below 240 nm
even if sample fluorophores exhibit strong absorbance bands since signal-to-noise ratios
deteriorate in the UV region. Capturing the emission up until 800 nm is necessary should
the sample contain algae or fluorescent pigments. Moreover, to enable the correction of
inner filter effects, it is important to measure the samples absorbance spectrum covering all
excitation and emission wavelengths. Otherwise, such corrections can become difficult to
implement.

12.5 INTERFERENCES

As an extrinsic property of fluorophores in solution, fluorescence fingerprints are

vulnerable to changes due to interferences. Such changes can impact fluorescence yields (per
mol of substance) and spectra and are caused by physicochemical properties of the sample.
When comparisons between samples are made, it is thus important to consider whether
physicochemical properties remain stable or are subject to change. In the following, we will
list some examples (not all) of the physicochemical parameters known to influence DOM
fluorescence.

The sample’s temperature can affect the fluorescence intensity observed for a given sample.
When solvent temperatures increase, observed fluorescence generally decreases. However,
there is no evidence to suggest that spectral characteristics change due to temperature
(McKay et al., 2018). This effect is of great importance for in situ measurements with
sensors since temperature can vary systematically over longer periods of time. However,
a compensation is trivial if the sample’s temperature is known (Watras et al., 2011). For
benchtop instruments, temperature effects are usually not an issue since jacketed cuvette
holders and climate-controlled laboratories eliminate the chance for systematic biases.

Changes in pH can lead to spectrally-dependent changes in a sample’s fluorescence.

Numerous studies have investigated the effect of pH on organic matter fluorescence, and
the different results reported in each study hints at sample-dependent, complex effects
(e.g. Esteves et al., 1999; Groeneveld et al., 2022; Mobed et al., 1996; Murphy et al., 2018;
Spencer et al., 2007). These complex changes make it effectively impossible to compensate

271
Experimental Methods for Membrane Applications

the pH-induced interference. It is thus best to avoid sample-to-sample differences in pH to

facilitate comparisons between samples.

Beyond temperature and pH, ionic strength, the presence of metal ions, and particle
attenuation have also been reported to affect fluorescence measurements. As with all
physicochemical properties, it is advisable to obtain reference measurements for samples
coming from the system that is subject of a study. This will help to ascertain if issues with
certain parameters are to be expected and if so, whether great variation (e.g., pH or ionic
strength) might introduce artefacts that complicate interpretation of the fluorescence
readings in a given dataset.

12.6 DATA PROCESSING

Fluorescence measurements require several steps of processing before further analyses can
take place (Figure 4). While some software offers comprehensive features that contain the
most critical steps, we believe it is most advisable to export data from proprietary formats
and subsequently use open software environments to process and analyze fluorescence
data. This gives the user more control over processing steps, freedom of choice regarding
analysis strategies, and maximizes the impact of the conducted research by extracting
as much information as possible. Amongst the most common languages for statistical
computing, Matlab and R have community-driven software packages (Matlab: drEEM,
EEMlab; R: eemR, StaRdom, albatross) that facilitate all steps in Figure 4 and enable a range
of multivariate analyses (Micó et al., 2019; Murphy et al., 2013; Pucher et al., 2019). While
theoretically possible, an analysis of EEMs in spreadsheet software is not recommended
since workflows are not easily automatable.

Data export Data import

Export files from software & convert into - Choose analysis platform
standardized, open delimited-seperated format - Assembly & import scans

Bias correction Inspect data set

- Blank subtraction - Remove excessive noise
- Correct inner-filter effect - Indentify & handle outliers
- Calibrate fl. data to R.U./QSU - Identify best anlysis strategy
- Remove scatter
- Consider other biases

Figure 4 Steps involved in the processing of fluorescence measurements.

After the successful import into the programming environment of choice, the next
processing steps concerns the correction or removal of different measurement biases. For
example, a blank should be subtracted from each sample fluorescence landscape to remove
the impact of background signals and reduce the abundance of scatter. Such blanks should
be measured daily and always be made from the sample solvent (e.g. water, buffered water,

272
 Chapter 12

organic solvent). Blanks can also function as a standard for the calibration of fluorescence
signals into Raman Units (Lawaetz and Stedmon, 2009). Next, inner-filter effects (IFEs) are
compensated since they disturb the linearity of observed fluorescence due to the partial or
complete absorbance of emitted fluorescence by chromophores (Parker and Rees, 1960).
Such effects are easily corrected by applying correction factors derived from the sample’s
absorbance scan – provided the maximum absorbance be below approx. 1.5 cm-1 (Kothawala
et al., 2013).

The removal of Rayleigh and Raman scatter can be an important step if the subsequent
analysis strategy (see below) includes the decomposition of fluorescence EEMs into
statistical components according to Beer Lambert’s law (Bahram et al., 2006). The open-
source software packages mentioned above include functions for scatter removal and thus
simplify this task considerably.

12.7 DATA ANALYSIS

Once data is measured and fully processed, the data analysis can occur. Analysis strategies
(overview in Table 1) can range from simply comparing fluorescence intensities (known
as ‘peak picking’) and fluorescence indices to multivariate analysis such as parallel factor
analysis (PARAFAC).

Table 1 Overview of most common strategies to analyse fluorescence EEMs.

Analysis strategy Description References
Peak picking Extraction of fluorescence intensities (Coble 2007)
from EEMs at defined wavelengths.
Fluorescence regional Integration of fluorescence in (Chen et al., 2003)
integration (FRI) wavelength regions with predefined
interpretation.
Fluorescence indices Qualitative descriptors of FDOM with (Huguet et al., 2009; Maie et al.,
insights into humification, aromaticity, 2006; Ohno 2002; Parlanti et al.,
freshness and microbial processing. 2000)
Parallel factor analysis Multivariate decomposition of EEMs into (Murphy et al., 2013)
(PARAFAC) components.

The comparison of fluorescence intensities usually occurs at predefined wavelengths that

typically have letters assigned to them (see Figure 5, table 2). For example, peak T, extracted
at excitation / emission 275 / 340 is typically ascribed to tryptophan- or protein-like
material. Peaks A and C on the other hand are often described as humic-like material. It
should be noted that the interpretation of fluorescence peaks should only be informed by
comparison with studies of the same sample material and take into account potential issues
(e.g. pharmaceuticals fluorescing like amino acids).

273
Experimental Methods for Membrane Applications

Table 2 Position of predefined peaks and fluorescence indices as listed or described in Coble
(2007), Maie et al. (2006), Huguet et al. (2009), Ohno (2002).
Tentative interpretation in
Peak / index λEx / λEm natural environments
A 260/400-460 Humic-like, terrestrial
B 275/305 Autochthonous
T 275/340 Autochthonous
M 290-310/370-410 Anthropogenic contaminants
C 320-360/420-460 Humic-like, terrestrial
D 390/509 Humic-like, ubiquitous
Fluorescence index (FI) 370/470 Distinguishes microbial and
370/520 terrestrial inputs
Biological index (BIX) 310/380 Contribution of biological
310/430 transformations
Humification index (HIX) 254/ ∫ 435– 480 Ratio between protein- and
254 / ∫ 300 − 345 + 254 / ∫ 435 − 480 humic-like fluorescence

Fluorescence regional integration (FRI) is a particularly popular technique in engineered

systems by which integrals of wavelength regions in the EEM (Figure 5, yellow lines) are
tracked across samples. The assignment of these regions is based on model compounds
and natural environmental samples (Chen et al., 2003). According to Chen et al., (2003),
the five regions as illustrated in Figure 5 are aromatic protein-like material (I + II), fulvic
acid-like compounds (III), microbial by-product-like fluorescence (IV), and humic-like
material (V). Subsequent to the publication of the FRI approach, multivariate modelling has
indicated that underlying fluorescence spectra in regions I, II and IV, as well as III and V
overlap spectrally and regional integrals are likely not as specific as the names above suggest
(Stedmon et al., 2003).
550

D
500
FI
HIX
450 V C
III A BIX
400 M
N
350 II IV
T
Emission (nm) I B
250 300 350 400
Excitation (nm)

Figure 5 Emission-Excitation Matrix of a sample from a boreal river (Öre river, northern Sweden).
Yellow lines and text refer to fluorescence regional integration areas (FRI). Black dots
represent fluorescence indices: FI: Fluorescence Index; BIX: Biological Index; HIX:
Humification Index). Blue lines and text refer to predefined peak locations.

274
 Chapter 12

Several fluorescence-based indices also find frequent application (Figure 5, thick black
lines, Table 2). For example, the humification index (HIX) informs about the ratio between
protein- and humic-like fluorescence and thus can help to understand qualitative shifts
between samples (Ohno, 2002). The biological index (BIX) is a ratio between ultraviolet and
visible fluorescence and can indicate the importance of recent biological transformations
of material (Huguet et al., 2009). Lastly, the fluorescence index (FI) distinguishes between
microbial and terrestrial inputs in aquatic environments (Maie et al., 2006; McKnight et
al., 2001). As with peak picking, one should be careful to extrapolate interpretations from
studies performed on different sample types to membrane samples. The above-mentioned
fluorescence indices were defined in studies of natural aquatic environments but can help to
identify qualitative differences between samples.

12.7.1 PARAFAC
Amongst the multivariate analysis techniques, PARAFAC is the most popular technique
in the analysis of DOM and this section will provide a short overview over this technique.
For tutorials on MCR and PCA, we refer the reader to Bro and Smilde (2014), and de Juan
et al., (2014). PARAFAC is a particularly popular model for the decomposition since it
naturally follows the analytical principals of fluorescence (Bro, 1997; Murphy et al., 2013;
Stedmon and Bro, 2008). Each analyte (termed ‘component’ in the model) is described as
a product of an excitation and emission spectrum multiplied by a concentration factor. The
entire EEM is the sum of the fluorescence arising from each of the components. PARAFAC
is particularly attractive since it can distinguish spectrally overlapping components and thus
allows insights into components that may not be distinguishable in the raw fluorescence
data. Also, PARAFAC can isolate systematic signals in noisy measurements and thus help
to improve the quality of the results. Moreover, the component spectra can be compared
between studies and help to inform the chemical interpretation of the signals.

As a multivariate modelling approach, PARAFAC analyses work best if a number of criteria

are met. For example, a minimum number of samples with meaningful compositional
variability helps to identify meaningful and robust models. If two or more peaks covary
perfectly, the approach may produce questionable models. Similarly, if a dataset consists
of too few samples, the algorithm can struggle to identify the underlying components.
As discussed above, the fluorescence occurring in each sample can be altered due to
interferences. In such cases, it is most likely more fruitful to rely on peak picking and/or the
interpretation of fluorescence indices.

12.8 APPLICATION IN MEMBRANE SYSTEMS

As outlined above, there are myriad ways in which fluorescence measurements can be
applied to aquatic systems that utilise some form of membrane. Organic matter is a common
source of membrane fouling, and it follows that fluorescence measurement of organic
matter has gained traction in the analysis and investigation of how organic fouling occurs
in these systems. Whether it is microfiltration, ultrafiltration, nanofiltration, or reverse and
forward osmosis, using flat sheet, hollow fiber, tubular, or spiral wound constructions, there
is nearly always a way in which fluorescence measurements can be, and has been applied.

275
Experimental Methods for Membrane Applications

These methods can assist in evaluating the membrane structure, the process performance in
general, and of course the level and character of fouling. However, the construction, matrix,
and analytical targets for each of these specific applications will determine how fluorescence
EEM spectroscopy can, and cannot, be used in each case.

Section 11.3 has highlighted two physical application differences: (1) front-face fluorescence
spectroscopy; and (2) right-angle fluorescence spectroscopy. Both methods can be used to
assess and/or predict organic fouling, though in different ways and typically for different
applications. Front-face fluorescence EEM spectroscopy currently finds its most common
(membrane-related) use for systems such as MBRs (membrane bioreactors), or direct
measurement from fouled membrane surfaces. Regardless of whether front-face or right-
angle fluorescence is used, the organic character of any fouling will be dependent on both
the feed, and the physical and/or chemical characteristics of the membrane in use.

The direct, in situ fluorescence analysis of membranes has to date been carried out in
different ways in order to understand the main organic components behind the organic
fouling of membrane surfaces. For example, Yamamura et al., (2019) used in situ front-face
EEM spectroscopy to investigate the organic fouling of PVDF membranes from secondary
treated wastewater, in a bench-scale study. They observed increasing intensities with time
over three main peaks, and by combining intermittent backwashing with EEM analysis,
the authors observed peaks which were predominantly associated with reversible and/
or irreversible fouling. Yu et al., (2019), also utilised front-face fluorescence to detect and
quantify model foulants on UF membranes, and concluded that it was a better method than
liquid right-angle fluorescence in this particular study due to lower standard deviations
observed between repeated measurements (see Figure 6). Pawlowski et al., (2016) used
front-face fluorescence to monitor fouling deposition on ion-exchange membranes. In this
study, the authors found it to be a useful tool in evaluating reverse electrodialysis processes,
and particularly for increasing membrane cleaning efficiencies.

Similar front-face techniques have also been used to investigate activated sludge systems
(Huaorng, 2022), to quantify biomass and bioactivity amongst other parameters. From
a practical spectroscopy perspective, these can be likened to the sludge and mixed liquor
components of MBRs. The fluorescence character of the sludge over time can be linked
to the evolution of fouling on the membranes, as the organic matter in particular EPS
(extracellular polymeric substances) has been found to be closely related to TMP (Chen et
al., 2018). Various iterations of front-face spectroscopy have been used for assessing both
the sludge and membrane components of MBR systems in situ (Galinha and Crespo, 2022).

276
 Chapter 12

a) 10 mg/L BSA

2,500 8x1011

2,000
6x1011
Foulant mass (mg/m2)
1,500
4x1011

Resistance (m-1)
1,000
2x1011
500 BSA (liquid EEM)
BSA (FFEEM)
0 0 Resistance
0 1.000 2.000 3.000 4.000 5.000
Time (s)

b) 10 mg/L HA
140 8x1011
120

100 6x1011
Foulant mass (mg/m2)

80
4x1011
Resistance (m-1)
60

40
2x1011
HA (liquid EEM)
20 HA (FFEEM)
0 0 Resistance
0 1.000 2.000 3.000 4.000
Time (s)

Figure 6 Comparison of fouling by 10 mg/L of (a) Bovine Serum Albumin and (b) Humic Acid
measured by liquid right-angle EEM and front-faced fluorescence EEM measurements
(adapted from Yu et al. (2019)).

Looking beyond front-face fluorescence EEM spectroscopy, the more common method
of analysing membrane performance has historically been by a direct measurement of
the different liquid streams with right-angle fluorescence. This is due to the higher signal
and higher sensitivity that is achieved by this method, though front-face fluorescence
spectroscopy has been described and utilised for high absorbance liquid samples for nearly
half a century (Eisinger and Flores, 1979). In the case of membrane systems, using the
right-angle fluorescence method typically means a comparison of one or more out of the
feed, permeate, and concentrate streams. Whether some, or all, of these streams can be
analysed with right-angle fluorescence will depend on whether the target matrix adheres
to the requirements set out in section 11.4 Quality Assurance. Primarily, the liquid matrix
must exhibit low absorbance values, which either are below the threshold where they can
be considered negligible, or within the mathematically correctable range. In the latter case,
an absorbance measurement must accompany the fluorescence measurement to guarantee
accurate values.

277
Experimental Methods for Membrane Applications

For example, Poojamnong (2020) used right-angle fluorescence to compare the feed
and permeate EEMs of an MBR treating pulp and paper wastewater. This study found
protein-like fluorescence to be the most reduced region from the feed to the permeate,
which correlated to the main component of the fouling formed on the UF membranes. A
comparison of feed and permeates from RO processes of water recycling plants have also
shown that right-angle fluorescence EEM could be used to monitor organics rejection and
membrane integrity (Pype et al., 2013; Singh et al., 2009; Singh et al., 2015). Bagastyo et
al., (2011) identified humic and fulvic-like organics, as well as soluble microbial products
as the main constituents of the concentrate stream within an RO system treating secondary
wastewater effluent.

A direct analysis of already fouled membranes can also be performed with right-angle
fluorescence EEM spectroscopy. Stripping off individual foulant layers by, for example,
backwashing, or acid/base washing, has been used to understand differences in the formation
of membrane foulant layers. In 2004, Kimura et al. (2004) showed that alkaline cleaning
removed primarily protein-like fluorescence from membranes fouled by surface water.
Henderson et al. (2011) also found that protein-like fluorescence was the predominant
fluorescence region that was removed through a three-step UF membrane cleaning
procedure. More specifically, in this case a low UV ‘tyrosine-like’ fluorescence component
found within 5 different sources of wastewater, was found to be highly correlated to
membrane fouling potential. In a study that investigated the role of DOM in the fouling
of membrane bioreactors (MBR) treating wastewater, Tang et al. (2010) also found that
two protein-like fluorescence components were most correlated to membrane fouling.
These and other similar studies highlight how EEM spectroscopy can be used effectively
to gain insight into how different foulant layers form, and therefore how to minimise their
formation.

Further to the direct observation of various fluorescent membrane fouling components,

studies have also investigated different pre-treatment steps to remove these components
and hence minimise organic fouling. For example, Wang et al. (2017) utilised liquid EEM
analysis to compare the performance and effect of 8 different types of powder activated
carbon (PAC), as a pre-treatment to UF membrane treatment. They showed that whilst
initial fouling was linked to the humic-like fluorescence region, ultimately the ability
of PAC to minimise irreversible fouling was linked to how well it absorbed protein-like
fluorescence. Aftab et al. (2020) applied combinations of different pre-treatment processes
to change the FDOM character of landfill leachate, in order to compare how the resulting
organics character ultimately affected NF fouling. This study concluded that both fulvic-like
and protein-like fluorescence was more linked to irreversible fouling, than was humic-like
fluorescence. In slight contrast to this, Xu et al. (2022) concluded that both proteins and
humics contributed to the initial pore blocking stage, though the study was conducted on
synthetic mixtures and model foulant compounds. Through a combination of size exclusion
chromatography and liquid fluorescence EEM analysis, Haberkamp et al. (2011) also showed
that protein-like fluorescence correlated with the extent of hydraulically irreversible fouling
of UF membranes by secondary effluents. In this case, the authors also showed that chemical
coagulation and biological sand filtration as pre-treatment were both able to significantly
reduce membrane fouling. Furthermore, in a study that investigated dissolved air flotation
as a potential pre-treatment for membrane desalination, Shutova et al. (2016) showed

278
 Chapter 12

that DAF treatment removed higher proportions of humic-like fluorescence than protein-
like fluorescence, yet it still was able to remove between 28% and 58% of protein like
fluorescence from real water samples.

Table 3 An overview of publications which have utilised fluorescence within membranerelated

treatment studies. N.B. MF (Microfiltration); UF (Ultrafiltration); RED (Reverse
electrodialysis); CE (cation exchange); AE (anion exchange); MBR (membrane
bioreactor); RO (reverse osmosis); AS (Activated sludge); FF (front-face), RA (right
angle), PP (peak picking); PARAFAC (Parallel factor analysis); FRI (fluorescence regional
integration); FI (fluorescence indices); PCA (principal components analysis); and PLS
(partial least squares)

Reference Application Method Highlights or main findings

Kimura et al. (2004) UF RA + PP Polysaccharide-like organic matter, Fe and Mn
in surface water responsible for irreversible
fouling.
Singh et al. (2009) RO RA + PP Humic-like fluorescence most appropriate for
distinguishing between stage 1 and stage 2 RO.
Tang et al. (2010) MBR RA + PP Protein-like fluorescence correlated positively
with membrane fouling.
Henderson et al. (2011) UF RA + PP/ Tyrosine-like fluorescence monitoring could be
PARAFAC used as an indicator of fouling potential from
domestic wastewater.
Bagastyo et al. (2011) RO RA + PP Advanced oxidation of RO concentrates more
efficient than coagulation & MIEX adsorption.
Haberkamp et al. (2011) UF RA + PP Removal of protein-like substances by sand
filtration or coagulation resulted in reduced
irreversible fouling.
Galinha et al. (2011) MBR RA/FF + 3 fluorescence components could be used to
PLS predict COD concentration in MBR permeate.
Pype et al. (2013) RO RA + FRI Fluorescence proposed as surrogate for
pathogen removal in RO systems.
Singh et al. (2015) RO RA + PP Peak C linked to TMP/fouling.
Shutova et al. (2016) RO RA + Humics concentration used to determine optimal
PARAFAC coagulant dose.
Pawlowski et al. (2016) RED - CE/AE FF / RA + Fluorescence able to monitor fouling
PCA development of ion-exchange membrane
surfaces.
Vera et al. (2017) UF/RO RA + Quantified OM removals through treatment
PARAFAC plant, monitoring FDOM composition can
optimise treatment conditions due to seasonal
variation.
Wang et al. (2017) UF RA + PP Humic-like substances contributed to initial
membrane fouling, protein-like correlated with
irreversible fouling.
Cai et al. (2017) MBR RA + Protein-like substances more readily
PARAFAC biodegradable than humic-like substances.

279
Experimental Methods for Membrane Applications

Reference Application Method Highlights or main findings

Jacquin et al. (2017) MBR RA + FRI Correlations established between LC-OCD-
OND and EEM data to quantify protein-like and
humic-like substances.
Xiao et al. (2018) MBR RA + FRI Identified correlations between characteristic
EEM wavelength regions and hydrophobic/
hydrophilic DOM components.
Yamamura et al. (2019) MF FF / RA Proteinaceous substances responsible for
+ PP reversible and irreversible fouling, gels mainly
contributed to irreversible fouling.
Yu et al. (2019) UF FF/RA + FF-EEM method superior to RA-EEM coupled
PARAFAC with mass balance for UF foulant determination.
Aftab et al. (2020) NF RA + Different pre-treatment led to different
PARAFAC quantities and qualities of membrane foulants .
Poojamnong et al. MBR RA + FRI Irreversible foulants mainly comprised of
(2020) protein-like substances.
Yu et al. (2021) MBR RA + PP/FI Combination of protein-like fluorescence and
UV280 used to predict fouling MBR potentials.
Xu et al. (2022) UF RA + PP Proteins and humics mainly participate in
initial pore blocking, polysaccharides mainly
participate in later gel/cake layer stage
Yu et al. (2022) AS FF + Protein-like substances, NADH, and humic-like
PARAFAC substances correlated with MLVSS, intracellular
NADH, and humic-like substances in SMP,
respectively.
Cifuentes-Cabezas et al. NF RA/FF + PP Fluorescence showed different fouling
(2023) development between different NF membrane
products.

The studies that have been mentioned in this section are but a small proportion of the
many research applications of fluorescence EEM spectroscopy within membrane treatment
systems to date. While they highlight the method as a clear and practical use for measuring
organic fouling on membranes, it is nevertheless imperative to keep in mind that only a
fraction of OM is fluorescent. Although the fluorescent fraction is typically considered to
be representative of OM as a whole, a better overall picture of OM and OM-based fouling
will nearly always be attained when it is applied in combination with other analytical tools.

Within the near future, further advances in optical technology and computer processing
power look to be the main impetus’ behind further development and wider application of
fluorescence sensors in membrane systems. As optical sensors become cheaper and more
sensitive, and light sources become more stable with higher output and tighter bandwidths
(particularly at lower wavelengths) fluorescence spectroscopy will become even more
practical and accessible for both research and industry applications.

280
 Chapter 12

12.9 REFERENCES

Aftab B, Cho J, Shin HS, and Hur J (2020) Using EEM-PARAFAC to Probe NF Membrane Fouling
Potential of Stabilized Landfill Leachate Pretreated by Various Options. Waste Management.
102: 260–269.
Andersson MGI, Catalán N, Rahman Z, Tranvik LJ, and Lindström ES. (2018) Effects of Sterilization
on Dissolved Organic Carbon (DOC) Composition and Bacterial Utilization of DOC from Lakes.
Aquatic Microbial Ecology. 82(2): 199–208.
Bagastyo AY, Keller J, Poussade Y, and Batstone DJ (2011) Characterisation and Removal of
Recalcitrants in Reverse Osmosis Concentrates from Water Reclamation Plants. Water Research
45(7): 2415–2427.
Bahram M, Bro R, Stedmon CA, and Afkhami A (2006) Handling of Rayleigh and Raman Scatter for
PARAFAC Modeling of Fluorescence Data Using Interpolation. Journal of Chemometrics. 20(3–
4): 99–105.
Baker A, and Inverarity R (2004) Protein-like Fluorescence Intensity as a Possible Tool for Determining
River Water Quality. Hydrological Processes. 18(15): 2927–45.
Bridgeman J, Bieroza M, and Baker A (2011) The Application of Fluorescence Spectroscopy to Organic
Matter Characterisation in Drinking Water Treatment. Reviews in Environmental Science and
Biotechnology. 10(3): 277–290.
Bro R (1997) PARAFAC. Tutorial and Applications. Chemometrics and Intelligent Laboratory Systems.
38(2): 149–171.
Bro R, and Smilde AK (2014). Principal Component Analysis. Analytical Methods. 6(9): 2812–2831.
Cai W, Liu J, Zhu X, and Liu Y (2017) Fate of Dissolved Organic Matter and Byproducts Generated from
On-Line Chemical Cleaning with Sodium Hypochlorite in MBR. Chemical Engineering Journal.
323: 233–242.
Carstea EM, Bridgeman J, Baker A, and Reynolds DM (2016) Fluorescence Spectroscopy for Wastewater
Monitoring: A Review. Water Research. 95: 205–219.
Catalá TS, Reche I, Fuentes-Lema A, Romera-Castillo C, Nieto-Cid M, Ortega-Retuerta E, Calvo
E, Álvarez M, Marrasé C, Stedmon CA, and Álvarez-Salgado XA (2015) Turnover Time of
Fluorescent Dissolved Organic Matter in the Dark Global Ocean. Nature Communications. 6(1):
5986.
Chen W, Qian C, Zhou K-G, and Yu H-Q. (2018) Molecular Spectroscopic Characterization of
Membrane Fouling: A Critical Review. Chem. 4(7): 1492–1509.
Chen W, Westerhoff P, Leenheer JA, and Booksh K (2003). Fluorescence Excitation− Emission Matrix
Regional Integration to Quantify Spectra for Dissolved Organic Matter. Environmental Science
& Technology. 37(24): 5701–5710.
Cifuentes-Cabezas M, Galinha CF, Crespo JG, Vincent-Vela MC, Mendoza-Roca JA, and Álvarez-
Blanco S (2023) Nanofiltration of Wastewaters from Olive Oil Production: Study of Operating
Conditions and Analysis of Fouling by 2D Fluorescence and FTIR Spectroscopy. Chemical
Engineering Journal. 454: 140025.
Coble PG (2007) Marine Optical Biogeochemistry: The Chemistry of Ocean Color. Chemical Reviews.
107(2): 402–418.
Coble P, Lead J, Baker A, Reynolds D, and Spencer R (2014). Aquatic Organic Matter Fluorescence. In P.
Coble, J. Lead, A. Baker, D. Reynolds, & R. Spencer (Eds.), Aquatic Organic Matter Fluorescence
(Cambridge Environmental Chemistry Series, p. V). Cambridge: Cambridge University Press.

281
Experimental Methods for Membrane Applications

Drozdova AN, Patsaeva SV, and Khundzhua DA (2017) Fluorescence of Dissolved Organic Matter
as a Marker for Distribution of Desalinated Waters in the Kara Sea and Bays of Novaya Zemlya
Archipelago. Oceanology. 57(1): 41–47.
Eisinger J, and Flores J (1979) Front-Face Fluorometry of Liquid Samples. Analytical Biochemistry.
94(1): 15–21.
Esteves VI, Santos EBH, and Duarte AC (1999) Study of the Effect of PH, Salinity and DOC on
Fluorescence of Synthetic Mixtures of Freshwater and Marine Salts. Journal of Environmental
Monitoring. 1(3): 251–254.
Fellman JB, D’Amore DV, and Hood E (2008) An Evaluation of Freezing as a Preservation Technique
for Analyzing Dissolved Organic C, N and P in Surface Water Samples. Science of The Total
Environment. 392(2): 305–312.
Galinha CF, Carvalho G, Portugal CAM, Guglielmi G, Oliveira R, Crespo JG, and Reis MAM (2011)
Real-Time Monitoring of Membrane Bioreactors with 2D-Fluorescence Data and Statistically
Based Models. Water Science and Technology. 63(7): 1381–1388.
Galinha CF, Carvalho G, Portugal CAM, Guglielmi G, Oliveira R, Reis MAM and Crespo JG (2011)
Two-Dimensional Fluorescence as a Fingerprinting Tool for Monitoring Wastewater Treatment
Systems. Journal of Chemical Technology & Biotechnology. 86(7): 985–992.
Galinha CF and Crespo JG (2022) Development and Implementation of MBR Monitoring: Use of 2D
Fluorescence Spectroscopy. Membranes. 12: 1218.
Gilmore AM (2014) How to Collect National Institute of Standards and Technology (NIST) Traceable
Fluorescence Excitation and Emission Spectra. In Fluorescence Spectroscopy and Microscopy:
Methods and Protocols, Methods in Molecular Biology, eds. Yves Engelborghs and Antonie
J.W.G. Visser. Totowa, NJ: Humana Press, 3–27.
Groeneveld M, Catalán N, Einarsdottir K, Bravo AG, and Kothawala DN (2022) The Influence of PH on
Dissolved Organic Matter Fluorescence in Inland Waters. Analytical Methods. 14(13): 1351–
1360.
Haberkamp J, Ernst M, Paar H, Palliscgeck D, Amy G, and Jekel M (2011) Impact of Organic Fractions
Identified by SEC and Fluorescence EEM on the Hydraulic Reversibility of Ultrafiltration
Membrane Fouling by Secondary Effluents. Desalination and Water Treatment. 29(1–3): 73–86.
Hambly A, Henderson RK, Baker A, Stuetz RM, and Khan SJ (2010) Probabilistic Analysis of
Fluorescence Signals for Monitoring Dual Reticulation Water Recycling Schemes. Water Science
and Technology. 62(9): 2059–2065.
Hambly AC, Arvin E, Pedersen L-F, Pedersen PB, Seredynska-Sobecka B, and Stedmon CA (2015)
Characterising Organic Matter in Recirculating Aquaculture Systems with Fluorescence EEM
Spectroscopy. Water Research. 83: 112–120.
Hambly AC, Henderson RK, Baker A, Stuetz RM, and Khan SJ (2015) Application of Portable
Fluorescence Spectrophotometry for Integrity Testing of Recycled Water Dual Distribution
Systems. Applied Spectroscopy. 69(1): 124–129.
Heinz M, and Zak D (2018) Storage Effects on Quantity and Composition of Dissolved Organic Carbon
and Nitrogen of Lake Water, Leaf Leachate and Peat Soil Water. Water Research. 130: 98–104.
Henderson RK, Baker A, Murphy KR, Hambly A Stuetz RM and Khan SJ (2009) Fluorescence as a
Potential Monitoring Tool for Recycled Water Systems: A Review. Water Research. 43(4): 863–
881.
Henderson RK, Subhi N, Antony A, Khan SJ, Murphy KR, Leslie GL, Chen V, Stuetz RM, and Le-Clech
P (2011) Evaluation of Effluent Organic Matter Fouling in Ultrafiltration Treatment Using
Advanced Organic Characterisation Techniques. Journal of Membrane Science. 382(1): 50–59.

282
 Chapter 12

Hu H, and Ren H (2019) Removal of Bioavailable Dissolved Organic Nitrogen in Wastewater by

Membrane Bioreactors as Posttreatment: Implications for Eutrophication Control. Bioresource
Technology. 271: 496–499.
Huguet A, Vacher L, Relaxans S, Saubusse S, Froidefond JM, and Parlanti E (2009) Properties of
Fluorescent Dissolved Organic Matter in the Gironde Estuary. Organic Geochemistry. 40(6):
706–719.
Jablonski A (1933) Efficiency of Anti-Stokes Fluorescence in Dyes. Nature. 131(3319): 839– 840.
Jacquin C, Lesage G, Traber J, Pronk W, and Heran M (2017) Three-Dimensional Excitation and
Emission Matrix Fluorescence (3DEEM) for Quick and Pseudo-Quantitative Determination of
Protein- and Humic-like Substances in Full-Scale Membrane Bioreactor (MBR). Water Research.
118: 82–92.
de Juan A, Jaumot J, and Tauler R (2014) Multivariate Curve Resolution (MCR). Solving the Mixture
Analysis Problem. Analytical Methods. 6(14): 4964–4976.
Kimura K, Hane Y, Watanabe Y, Amy G, and Ohkuma N (2004) Irreversible Membrane Fouling during
Ultrafiltration of Surface Water. Water Research. 38(14): 3431–3441.
Kopf U, and Heinze J (1984) 2,7-Bis(Diethylamino)Phenazoxonium Chloride as a Quantum Counter
for Emission Measurements between 240 and 700 Nm. Analytical Chemistry. 56(11): 1931–
1935.
Kothawala DN, Murphy KR, Stedmon CA, Weyhenmeyer GA, and Tranvik LJ (2013) Inner Filter
Correction of Dissolved Organic Matter Fluorescence. Limnology and Oceanography: Methods
11(12): 616–630.
Kothawala DN, Stedmon CA, Müller RA, Weyhenmeyer GA, Köhler SJ, and Tranvik LJ (2014) Controls
of Dissolved Organic Matter Quality: Evidence from a Large-Scale Boreal Lake Survey. Global
Change Biology. 20(4): 1101–1114.
Lakowicz JR, ed. (2006a) Instrumentation for Fluorescence Spectroscopy. In Principles of Fluorescence
Spectroscopy, Boston, MA: Springer US, 27–61.
Lakowicz JR, ed. (2006b) “Introduction to Fluorescence.” In Principles of Fluorescence Spectroscopy,
Boston, MA: Springer US, 1–26.
Lawaetz AJ, and Stedmon CA (2009) Fluorescence Intensity Calibration Using the Raman Scatter Peak
of Water. Applied Spectroscopy. 63(8): 936–940.
Maie N, Parish KJ, Watanabe A, Knicker H, Benner R, Abe T, Kaiser K, and Jaffé R (2006) Chemical
Characteristics of Dissolved Organic Nitrogen in an Oligotrophic Subtropical Coastal Ecosystem.
Geochimica et Cosmochimica Acta 70(17): 4491–4506.
Mangal MN, Salinas-Rodriguez SG, Dusseldorp J, Blankert B, Yangali-Quintanilla VA, Kemperman
AJB, Schippers JC, van der Meer WGJ, and Kennedy MD (2022) Foulant Identification and
Performance Evaluation of Antiscalants in Increasing the Recovery of a Reverse Osmosis System
Treating Anaerobic Groundwater. Membranes. 12(3): 290.
McKay G, Korak JA, and Rosario-Ortiz FL (2018) Temperature Dependence of Dissolved Organic
Matter Fluorescence. Environmental Science & Technology. 52(16): 9022–9032.
McKnight DM, Boyer EW, Westerhoff PK, Doran PT, Kulbe T, and Andersen DT (2001)
Spectrofluorometric Characterization of Dissolved Organic Matter for Indication of Precursor
Organic Material and Aromaticity. Limnology and Oceanography. 46(1): 38–48.
Micó Tormos P, Garcia-Ballesteros S, Mora Carbonell M, Vincente Candela R, Amat Payá AM, and
Arqués Sanz A (2019) EEMlab: A Graphical User-Friendly Interface for Fluorimetry Experiments
Based on the DrEEM Toolbox. Chemometrics and Intelligent Laboratory Systems. 188: 6–13.

283
Experimental Methods for Membrane Applications

Mobed JJ, Hemmingsen SL, Autry JL, and McGown LB (1996) Fluorescence Characterization of IHSS
Humic Substances: Total Luminescence Spectra with Absorbance Correction. Environmental
Science & Technology. 30(10): 3061–3065.
Morán XAG, Gasol JM, Arin L, and Estrada M (1999) A Comparison between Glass Fiber and Membrane
Filters for the Estimation of Phytoplankton POC and DOC Production. Marine Ecology Progress
Series. 187: 31–41.
Murphy KR, Timko SA, Gonsior M, Powers LC, Wünsch UJ, and Stedmon CA (2018) Photochemistry
Illuminates Ubiquitous Organic Matter Fluorescence Spectra. Environmental Science &
Technology. 52(19): 11243–11250.
Murphy KR, Hambly A, Singh S, Henderson RK, Baker A, Stuetz RM, and Khan SJ (2011) Organic
Matter Fluorescence in Municipal Water Recycling Schemes: Toward a Unified PARAFAC
Model. Environmental Science & Technology. 45(7): 2909–2916.
Murphy KR, Stedmon CA, Graeber D, and Bro R (2013) Fluorescence Spectroscopy and Multi-Way
Techniques. PARAFAC. Analytical Methods. 5(23): 6557– 6566.
Obayashi Y, and Suzuki S (2019) High Growth Potential of Transiently 0.2-Mm- Filterable Bacteria
with Extracellular Protease Activity in Coastal Seawater. Plankton and Benthos Research. 14(4):
276–286.
Ohno T (2002) Fluorescence Inner-Filtering Correction for Determining the Humification Index of
Dissolved Organic Matter. Environmental Science & Technology. 36(4): 742–746.
Osburn CL, Anderson NJ, Stedmon CA, Giles ME, Whiteford EJ, McGenity TJ, Dumbrell AJ, and
Underwood GJC (2017) Shifts in the Source and Composition of Dissolved Organic Matter in
Southwest Greenland Lakes Along a Regional Hydro-Climatic Gradient. Journal of Geophysical
Research: Biogeosciences. 122(12): 3431–3445.
Parker CA, and Rees WT (1960) Correction of Fluorescence Spectra and Measurement of Fluorescence
Quantum Efficiency. Analyst. 85(1013): 587–600.
Parlanti E, Wörz K, Geoffroy L, and Lamotte M (2000) Dissolved Organic Matter Fluorescence
Spectroscopy as a Tool to Estimate Biological Activity in a Coastal Zone Submitted to
Anthropogenic Inputs. Organic Geochemistry. 31(12): 1765–1781.
Pawlowski S, Galinha CF, Crespo JG, and Velizarov S (2016) 2D Fluorescence Spectroscopy for
Monitoring Ion-Exchange Membrane Based Technologies – Reverse Electrodialysis (RED).
Water Research. 88: 184–198.
Pucher M, Wünsch U, Weigelhofer G, Murphy KR, Hein T, and Graeber D (2019) StaRdom: Versatile
Software for Analyzing Spectroscopic Data of Dissolved Organic Matter in R. Water. 11(11):
2366.
Poojamnong K, Tungsudjawong K, Khongnakorn W, and Jutaporn P (2020) Characterization of
reversible and irreversible foulants in membrane bioreactor (MBR) for eucalyptus pulp and paper
mill wastewater treatment using fluorescence regional integration. Journal of Environmental
Chemical Engineering, 8(5): 104231
Pype M-L, Patureau D, Wery N, Poussade Y, and Gernjak W (2013) Monitoring Reverse Osmosis
Performance: Conductivity versus Fluorescence Excitation–Emission Matrix (EEM). Journal of
Membrane Science. 428: 205– 211.
Rosenstock B, and Simon M (1993) Use of Dissolved Combined and Free Amino Acids by Planktonic
Bacteria in Lake Constance. Limnology and Oceanography. 38(7): 1521–1531.
Sansonetti CJ, Salit ML, and Reader J (1996) Wavelengths of Spectral Lines in Mercury Pencil Lamps.
Applied Optics. 35(1): 74–77.

284
 Chapter 12

Shutova Y, Karna BL, Hambly AC, Lau B, Henderson RK, and Le-Clech P (2016) Enhancing Organic
Matter Removal in Desalination Pretreatment Systems by Application of Dissolved Air Flotation.
Desalination. 383: 12–21.
Singh S, Henderson RK, Baker A, Stuetz RM, and Khan SJ (2009) Distinguishing Stage 1 and 2 Reverse
Osmosis Permeates Using Fluorescence Spectroscopy. Water Science and Technology. 60(8):
2017–2023.
Singh S, Henderson RK, Baker A, Stuetz RM, and Khan SJ (2015) Online Fluorescence Monitoring of
RO Fouling and Integrity: Analysis of Two Contrasting Recycled Water Schemes. Environmental
Science: Water Research & Technology. 1(5): 689–698.
Spencer RGM, Bolton L, and Baker A (2007) Freeze/Thaw and PH Effects on Freshwater Dissolved
Organic Matter Fluorescence and Absorbance Properties from a Number of UK Locations. Water
Research. 41(13): 2941–2950.
Spiliotopoulou A, Martin R, Pedersen L-F, and Andersen HR (2017) Use of Fluorescence Spectroscopy
to Control Ozone Dosage in Recirculating Aquaculture Systems. Water Research. 111: 357–365.
Stedmon CA, and Bro R (2008) Characterizing Dissolved Organic Matter Fluorescence with Parallel
Factor Analysis: A Tutorial. Limnology and Oceanography: Methods. 6(11): 572–579.
Stedmon CA, Markager S, and Bro R (2003) Tracing Dissolved Organic Matter in Aquatic Environments
Using a New Approach to Fluorescence Spectroscopy. Marine Chemistry. 82(3): 239–254.
Stedmon CA, and Nelson NB (2015). Chapter 10 - The Optical Properties of DOM in the Ocean. In
Biogeochemistry of Marine Dissolved Organic Matter (Second Edition), eds. Dennis A. Hansell
and Craig A. Carlson. Boston: Academic Press, 481–508.
Stokes GG (1852) XXX. On the Change of Refrangibility of Light. Philosophical Transactions of the
Royal Society of London. 142: 463–562.
Tang S, Wang Z, Wu Z, and Zhou Q (2010) Role of Dissolved Organic Matters (DOM) in Membrane
Fouling of Membrane Bioreactors for Municipal Wastewater Treatment. Journal of Hazardous
Materials. 178(1): 377–384.
Urgun-Demirtas M, Sattayatewa C, and Pagilla KR (2008) Bioavailability of Dissolved Organic
Nitrogen in Treated Effluents. Water Environment Research: A Research Publication of the
Water Environment Federation. 80(5): 397–406.
Vera M, Cruz S, Boleda MR, Mesa J, Martí-Alonso J, Casas S, Gibert O, and Cortina JL (2017)
Fluorescence Spectroscopy and Parallel Factor Analysis as a Dissolved Organic Monitoring Tool
to Assess Treatment Performance in Drinking Water Trains. Science of The Total Environment.
584–585: 1212–1220.
Wang H, Ding A, Gan Z, Qu F, Cheng X, Bai L, Guo S, Li G, and Liang H (2017) Fluorescent Natural
Organic Matter Responsible for Ultrafiltration Membrane Fouling: Fate, Contributions and
Fouling Mechanisms. Chemosphere. 182: 183–193.
Wang Y, Hammes F, Boon N, and Egli T (2007) Quantification of the Filterability of Freshwater Bacteria
through 0.45, 0.22, and 0.1 Mm Pore Size Filters and Shape-Dependent Enrichment of Filterable
Bacterial Communities. Environmental Science & Technology. 41(20): 7080–7086.
Wang Y, Hammes F, Düggelin M, and Egli T (2008) Influence of Size, Shape, and Flexibility on Bacterial
Passage through Micropore Membrane Filters. Environmental Science & Technology. 42(17):
6749–6754.
Watras CJ, Hanson PC, Stacy TL, Morrison KM, Mather J, Hu Y-H, and Milewski P (2011) A
Temperature Compensation Method for CDOM Fluorescence Sensors in Freshwater. Limnology
and Oceanography: Methods. 9(7): 296–301.

285
Experimental Methods for Membrane Applications

Xiao K, Shen Y, Liang S, Tan J, Wang X, Liang P, and Liang X (2018) Characteristic Regions of the
Fluorescence Excitation–Emission Matrix (EEM) To Identify Hydrophobic/Hydrophilic
Contents of Organic Matter in Membrane Bioreactors. Environmental Science & Technology.
52(19): 11251–11258.
Xu H, Xu Y, Xiao K, Gao T, Liu Z, Xue W, Wei C, and Huang X (2022) Interplay of Organic Components
in Membrane Fouling Evolution: Statistical Evidence from Multiple Spectroscopic Analyses.
Journal of Membrane Science 661: 120913.
Yamamura H, Ding Q, and Watanabe Y (2019) Solid-Phase Fluorescence Excitation Emission Matrix
for in-Situ Monitoring of Membrane Fouling during Microfiltration Using a Polyvinylidene
Fluoride Hollow Fiber Membrane. Water Research. 164: 114928.
Yamin G, Borisover M, Cohen E, and van Rijn J (2017) Accumulation of Humic-like and Proteinaceous
Dissolved Organic Matter in Zero-Discharge Aquaculture Systems as Revealed by Fluorescence
EEM Spectroscopy. Water Research. 108: 412–421.
Yu H, Li Y, Yang H, Lv Y, Rong H and Qu F (2022) Characterization of Activated Sludge in Wastewater
Treatment Processes Using Front-Face Excitation–Emission Matrix (FF-EEM) Fluorescence
Spectroscopy. Environmental Science: Water Research & Technology. 8(10): 2265–2276.
Yu H, Wu Z, Zhang X, Qu F, Wang P, and Liang H (2019) Characterization of Fluorescence Foulants on
Ultrafiltration Membrane Using Front-Face Excitation-Emission Matrix (FF-EEM) Spectroscopy:
Fouling Evolution and Mechanism Analysis. Water Research. 148: 546–555.
Yu J, Xiao K, Qi T, Li Y, Tan J, Wen X, and Huang X (2021) Spectroscopic Sensing of Membrane Fouling
Potential in a Long-Term Running Anaerobic Membrane Bioreactor. Chemical Engineering
Journal. 426: 130799.

286
 doi: 10.2166/9781789062977_0287

Chapter 13

Transparent Exopolymer
Particles
Loreen O. Villacorte, Grundfos, Denmark

Yuli Ekowati, Grundfos, Denmark

Helga Calix Ponce, Denmark

The learning objectives of this chapter are the following:

• Understand the relevance of transparent exopolymer particles (TEP) to membrane

filtration processes

• Learn the different methods to measure TEP in fresh and saline water sources

• Describe experimental protocols to quantify TEPs and their precursors

• Discuss application of TEP methods for membrane filtration applications

13.1 INTRODUCTION

Transparent exopolymer particles (TEP) and their precursors has been reported to cause
organic or biological fouling in membrane filtration processes such as microfiltration (MF),
ultrafiltration (UF), nanofiltration (NF) and reverse osmosis (RO). As the name suggest,
TEPs are transparent organic substances, seasonally abundant in marine and fresh surface
water environments, particularly during algal blooms. TEPs has been known to exist in
marine and lake environments since the early 90’s but its link to membrane processes has
only been studied since the mid-2000’s. Since then, experimental methods have been
adopted, modified, developed, and demonstrated to quantify these substances and elucidate
their impact to membrane systems.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

TEPs largely originate from exudates or detritus of phytoplankton (micro-algae) and

bacterioplankton but they can also originate from macro-algae and some species of
oysters, mussels, scallops and sea snails. TEPs are generally sticky and highly hydrated gels,
comprising mainly hydrophilic, negatively charged, acidic polysaccharides (Mopper et al.,
1995). They tend to associate with or absorb proteins, lipids, trace elements and heavy
metals from the water (Passow, 2002). This makes them a good platform and hotspot for
bacterial growth and likely have an important role in the formation of aquatic biofilms
(Alldredge et al., 1993; Passow, 2002; Bar-Zeev et al., 2012a).

Berman and Holenberg (2005) initially proposed the potential role of TEP as a major initiator
of biofilm leading to biofouling in reverse osmosis (RO) membranes. Consequently, various
studies were conducted to investigate the link between TEP and biofouling in membranes
(Figure 1; Bar-Zeev et al., 2009; Villacorte et al., 2009a,b; Villacorte et al., 2017a). Further
studies have also demonstrated that TEPs can directly cause organic fouling in MF/UF
(Figure 1; Kennedy et al., 2009; Villacorte et al., 2010a,b; 2013; 2015a; Schurer et al., 2012,
2013) and forward osmosis membranes (Valladares Linares et al., 2012).

Filtration

TEP gels

Planktonic
bacteria
Polymers/ TEP algae cells
colloids
g MF/UF
Protobiofilm Membrane

a b c d e

h
Minutes Backwashing
NF/RO
Time f membrane

Minutes to hours i
Non-
backwashable
Feed pressure fouling
Time

Figure 1 Possible contribution of TEP on biological and organic fouling in membrane systems. In
RO/NF, some planktonic bacteria (first colonizers) can attach (d) reversibly on clean
surfaces or (e) irreversibly on TEP-conditioned surfaces. When nutrients are available in
the water, (f) contiguous coverage of mature biofilm can develop within a short period of
time (minutes to hours). In MF/UF, organic fouling can occur during (g) filtration of algal
bloom impacted water. TEPs strongly adhere to membrane pores and surfaces causing (h)
incomplete removal of cake layer during backwashing and leading to (i) gradual increase
in feed pressure or permeability over time. Figures adapted from Bar-Zeev et al. (2012a)
and Villacorte et al. (2021).

288
 Chapter 13

Early analytical methods operationally defined TEPs as particles larger than 0.4 μm
considering that they were first identified through retention on 0.4 μm pore size membrane
filters (Alldredge et al., 1993). They typically contain more than 99% water, which means
they can bulk-up to more than 100 times their dried volume (Azetsu-Scott and Passow,
2004; Verdugo et al., 2004). Large TEPs can be directly produced through sloughing of
algal cell coatings or through disintegration of large algal colonies (Figure 2). Other TEPs
are produced indirectly from colloidal polymers (1-10 kDa) released by phyto-/bacterio-
planktons which eventually grow into TEPs (>0.4 μm) through the subsequent process
of annealing, gelation, and aggregation (Verdugo et al., 2004; Chin et al., 1998). These
sub-micron components (<0.4 μm) which have similar chemical properties as TEPs are
collectively known as TEP precursors (Passow, 2000).
AB-XG complex

Alcian blue (AB)

H3C
H3C CH3
Cl- N+
H3C S
Xanthan gum (XG)

H 0H CH2OH
H H O
N N 0H H H O
N
CH3 CH3 H O H
Cu2+N
N S N S N CH2OH H OH
H3C CH3 n
N CH3COOCH2
N + N N N +
O O
H3C CH3 H 3C CH3 H H
Cl- Cl- H
OH
OH
-OOC H H
S CH3 O
O CH2 COO-
N+ Cl-
CH2C H O O
H3C N CH3 H
H H
CH3 0H O 0H H
O H H
H H H 0H

a b c

Figure 2 Top figures show the molecular structures of alcian blue (AB) and a standard acidic
polysaccharide xanthan gum (XG) and precipitates formed by the reaction of the two
compounds at pH 2.5 (adopted from Villacorte et al., 2015b). Bottom figures are optical
microscope images of Alcian Blue stained algal cells of (a) Alexandrium tamarense, (b)
Lepidodinium chlorophorum and (c) Chaetoceros affinis, where TEPs (stained blue) were
released through shedding of cell mucus (b and c) and membrane coatings (a and b).
Images a and c adopted from Villacorte et al. (2015c) and b from Claquin et al. (2008).

289
Experimental Methods for Membrane Applications

13.2 QUANTIFICATION METHODS

Since the discovery of TEPs, various quantification methods have been developed, all of
which involved staining with Alcian Blue (AB) dye). The dye is known to be highly selective
and forms insoluble complexes with target compounds that cannot be easily reversed by
subsequent treatments. AB is also widely available and has been routinely used in medical
and biological research. However, despite being one of the most widely used biological
stains, the mechanisms involved during reaction of the dye with a specific substrate is still
not well understood.

An AB molecule is a tetravalent cation with a copper atom at the center of its core (Figure 2). In
aqueous solutions without extra electrolytes, AB can bind with anionic carboxyl, phosphate
and half-ester sulphate groups of acidic polysaccharides, resulting in the formation of
neutral blue precipitates (Figure 2). It is also known to react with carbohydrate-conjugated
proteins such as acidic glycol-proteins and proteoglycans but does not stain nucleic acids
and neutral biopolymers.

AB staining can be largely impacted by the type and density of anionic functional groups
associated with the material in the sample. Selectivity depends on the pH and ionic strength
of the sample solution. In high ionic strength solutions, better interactions between anionic
polymers and cationic AB can be expected due to compression of the electrical double
layer surrounding the AB molecule. It can also spontaneously aggregate in saline solutions
forming AB precipitates not associated with TEP. This is a major drawback of the application
of AB for TEP measurements in seawater. To minimize this, AB staining solutions should be
pre-filtered and should not be directly applied to solutions with high salinity.

Table 1 shows an overview of the available TEP methods described in literature. The first
ever TEP method is a direct quantification through filter retention, AB staining and optical
microscopic enumeration (Alldredge et al., 1993). The method can provide information of
the size-frequency distribution of TEP in the water, but not feasible for quantifying TEPs <
2 μm and TEP precursors. Currently, the most widely used TEP method was developed by
Passow and Alldredge (1995), also referred to this work as TEP0.4μm. With additional sample
preparation steps (e.g., bubble adsorption, laminar shear), it may be possible to measure TEP
precursors using this method (Zhou et al., 1998). Villacorte et al. (2009a) proposed a slight
modification for TEP0.4μm by using a smaller pore size filter (0.1 μm) to capture some of the
TEP precursors (TEP0.1μm). To reduce the interference of salinity, a rinsing step was later
introduced to the TEP0.4μm filtration protocol to dilute or minimize salts on the filter before
AB staining (Villacorte et al., 2015b).

Arruda-Fatibello et al. (2004) introduced a different approach to measuring TEP by direct

staining on water samples. However, it is only applicable for freshwater samples because
salts interfere with AB staining. The method by Thornton et al. (2007) introduced a similar
approach but added a dialysis step for saline samples. Further modification of the method,
known as TEP10kDa, was later introduced to address a major practical limitation of the two
previous methods by introducing a sample concentration step through 10 kDa membrane
filtration (Villacorte et al., 2015b; 2017b). This method can substantially reduce the TEP
analysis time for brackish or saline water samples. It also enables the size fractionation

290
 Chapter 13

of TEPs in the water by filtering through series of membranes with decreasing pore sizes
during the concentration step.

More recent developments have shown that TEP can be measured online using an auto-
imaging technique (Thuy et al., 2017) or a crossflow filtration unit with integrated
spectrophotometer (Sim et al., 2019). Online measurement techniques would be the next
logical step towards routine TEP monitoring especially during algal blooms. However,
further studies are still needed to verify replicability and reliability of these advanced
techniques in the field, particularly regarding the impact of salinity during staining.

Table 1 Overview of quantification methods for TEP and their precursors.

Experimental
methods Description of main steps TEP analysed Water type References
Offline measurement (grab samples)
Microscopic Filtration, AB staining, TEP>2µm fresh/saline Alldredge et al.
enumeration microscopic counting (1993)
TEP0.4µm Filtration (0.4µm), AB TEP>0.4µm fresh/saline Passow &
staining, acid extraction, Alldredge (1995);
absorbance measurement Villacorte et al.
(2015)
Rapid AB staining, centrifugation, TEP + precursors fresh Arruda-Fatibello
spectrophotometric absorbance measurement et al. (2004);
Acidic dialysis (1 kDa; if saline TEP + precursors fresh/saline Thornton et al.
polysaccharide water), AB staining, filtration (2007)
(APS) (0.1µm), absorbance
measurement
TEP0.05µm Filtration (0.4 and 0.05µm), TEP>0.05µm fresh/saline Villacorte et al..
AB staining, acid extraction, (2009a)
absorbance measurement
TEP10kDa Filtration (10kDa), TEP + precursors fresh/saline Villacorte et al.
ultrasonication, AB staining, (2015b)
acid extraction, absorbance
measurement
Inline/Online measurement
Flow-CAM imaging AB staining, FlowCAM TEP > 5µm fresh Thuy et al. (2017)
imaging, image processing
and counting
Crossflow TEP crossflow filtration, TEP + precursors fresh/saline Sim et al. (2018)
monitor AB staining, fiber optic
spectrophotometry

The succeeding sections include detailed descriptions of two relevant methods for measuring
TEPs in and fresh and saline water, namely TEP0.4μm and TEP10kDa. The TEP0.4μm method
measures TEPs retained by membrane filters having pores of 0.4 μm and conventionally
known as TEP (Passow and Alldredge, 1995). The TEP10kDa method covers transparent
exopolymer particles retained by membrane filters with molecular weight cut-off of 10
kDa. Consequently, this method covers both TEP and most (if not all) of their colloidal
precursors.

291
Experimental Methods for Membrane Applications

13.2.1 Alcian blue dye preparation

Spectrophotometric TEP methods involve absorbance analyses of AB in either acetic acid
solution (pH 2.5) or 80% sulfuric acid solution. Figure 3a shows the spectral scans of AB
dissolved in these two matrices. The maximum absorbance of AB in 80% sulphuric acid
solution is at 787 nm wavelength while the maximum absorbance of AB in acetic acid
solution (pH 2.5) is at 610 nm within the visible light spectrum. The typical absorbance
value of AB in sulphuric acid at 787 nm is around twice that of AB of similar concentration
dissolved in acetic acid solution at 610 nm (Figure 3b). Regardless of the method used,
preparation of AB solution is usually prepared by dissolving AB in acetic acid solution at pH
2.5. The succeeding sections describe the recommended procedure in preparing the AB dye
solutions for TEP measurements.

(a) (b)
1.4
4.0
787
1.2 3.5

1.0 3.0
2.5
Absorbance (abs/cm)

Absorbance (abs/cm)
0.8
y = 0.0657x 2.0
610 0.6 R2 = 0.9988
1.5
0.4
1.0
0.2 y = 0.0286x 0.5
R2 = 0.9985
0 0
0 300 400 500 600 700 800 900 0 10 20 30 40 50 60
Wavelenght (nm) Alcian blue concentration (mg AB/L)

AB in sulfuric acid AB in sulfuric acid @ 787 nm

AB in acetic acid buffer (pH 2.5) AB in acetic acid buffer @ 610 nm

Figure 3 (a) Absorption spectra of Alcian blue 8GX (16 mg AB/L) dissolved in sulphuric acid
and acetic acid solutions and (b) peak absorbance values of Alcian blue (AB) solutions at
various concentrations (adopted from Villacorte et al, 2015b).

Materials and analytical set-up

• Alcian blue 8GX, 0.05 g
• Ultrapure water (UPW), 200 mL
• Magnetic stirrer
• Acetic acid
• pH meter
• Glass beaker
• Filter 0.05 μm (polycarbonate track etched membrane filters 47 mm)
• Clean glass tube (for capturing the filtered solution)
• Vacuum filtration set-up for 47 mm filter
• Vacuum pump
• Demineralised water
• Glass container for the final AB solution

292
 Chapter 13

Preparation of AB stock solution (0.025%)

1. Add 200 mL of UPW in a clean glass beaker and stir using a magnetic stirrer at 500 rpm.
2. Add acetic acid to the UPW drop by drop until pH lowers to 2.5.
3. While stirring at 1000 rpm, add 50 mg of AB powder and continue stirring for 18 to
24 hrs at 30 °C. Make sure to cover the top of the beaker with parafilm during this period.
4. Store the stock solution at 4 °C in a closed container. In practice, the stock solution
should be used only within 4 weeks after it was prepared.
5. Calculation of the amount of alcian blue powder needed:

m(g)
% AB = ×100%
V (mL)
0.025% AB = x (g) / 200 mL UPW,
then x = 0.05 g AB is added per 200 mL of ultrapure water

Preparation of AB working solution

In each time performing TEP measurement, the amount of AB solution needed should be
filtered the same day of the measurement (around 1 mL per measurement plus 10 mL extra).
1. Clean the vacuum filtration set up by filtering 100 mL of UPW at pressure ≤ 0.2 bar.
Dispose the filtered water.
2. Place the clean glass tube inside the filter flask to capture the filtered AB stock solution.
3. Mount a 0.05 μm PC filter on the filter holder of the vacuum filtration set-up.
4. Filter the amount of AB stock solution needed at a vacuum pressure ≤ 0.2 bar. Throw
away the first few drops of filtered, light blue colored filtrate or until dark blue color
filtrate appears.
5. Filter the collected filtrate again through a clean 0.05 μm filter on same filtration set-up.
6. Transfer the twice filtered AB stock solution in a glass container and cover.

13.2.2 TEP0.4µm measurement

TEP0.4μm was originally developed in the mid-90s by Passow and Alldredge (1995) and later
improved by Villacorte et al. (2015b) as illustrated in Figure 5 and further described in the
succeeding sections.

Materials and analytical set-up

• Alcian blue working solution (see section 13.2.1)
• 47 mm polycarbonate track-etched membrane filters (0.4 μm pore size)
• Clean glass tube (for collecting filtered solution)
• Vacuum filtration set-up for 47 mm filter (see Figure 4)
• Vacuum pump
• Demineralised water (demiwater), for cleaning filtration set-up
• Ultrapure water (UPW), for rinsing membrane filters.
• 50 ml glass beaker
• 80% sulfuric acid solution
• Shaker

293
Experimental Methods for Membrane Applications

• Spectrophotometer
• 1 cm cuvette
• Pipette 5 mL and 1 mL
• Tweezer
• Measuring cylinder

Figure 4 Overview of analytical setup for TEP0.4μm measurement.

Sample filtration and staining

1. Collect 0.5-1 L of water samples to be analyzed. A volume (typically >20 mL) of the
sample is filtered through a 47 mm diameter PC filter (0.4 μm pore size) by applying a
vacuum of 0.2 bar. Note: To remove possible contaminants, rinse PC filters by flushing
>200 ml of UPW through it prior to sample filtration.
2. Filter (≤0.2 bar vacuum) 2 mL of UPW through the retained TEP to wash the remaining
sample moisture through the filter and replace it with very low salinity water.
3. Pipette 1 mL of the working AB dye solution, apply over the filter, allowed to react with
TEP for 10 seconds, and then flush the unreacted dye through by vacuum filtration (<0.2
bar). To remove the remaining unreacted dye, perform a rinsing step by filtering 2 mL of
UPW.
4. Using a tweezer, fold the rinsed filter two-fold (while on the filter holder) with the
stained TEP in the inner part, and transfer to a 50 mL glass beaker.
5. Add 6 mL of 80% sulfuric acid solution on top on the filter in the beaker, cover beaker
with parafilm and mix on an auto-shaker for 2 hours.

294
 Chapter 13

6. Transfer part of the acid solution to a 1-cm cuvette and measure the absorbance (At) at
787 nm wavelength.
7. Measure the filter blank (Af) in the same way as Steps 1-6 but filtering TEP-free blank
samples (e.g., synthetic water with similar ion concentration as the water sample) instead
of actual water samples.
8. Measure sample blank correction (As) by filtering water samples in the same way as Steps
1-6 but skipping the AB staining procedure (Step 3).
9. Repeat at least three times Steps 1-8 per sample, per batch of filters and per batch of
working AB solution.

To prevent cross-contamination of samples, it is recommended to clean the glass filter

holder with sulfuric acid and then with UPW in between tests until no more visible AB
remains on the holder. The filtered sample volume in Step 1 can be adjusted (increased
or reduced) depending on the initial absorbance results. To get more reliable absorbance
results, it is recommended to reduce sample volume if initial absorbance result is ≥0.8 cm-1
and increase sample volume if initial absorbance is very close to or lower than the blank
absorbance.

Concentration calculation without calibration

The concentration of TEP0.4μm in terms of abs/cm/L is calculated as follows:

At − Af − As
TEP0.4 µm = Eq. 1
Vf

where (At) is the total absorbance of the dye that reacted with TEP and the filter (abs/cm);
(Af) is the absorbance of the dye adsorbed to the filter (abs/cm); (As) is the absorbance of
unstained sample (abs/cm) and Vf is the volume of sample filtered (L).

Concentration calculation with calibration

TEP0.4μm can be further calibrated and expressed in terms of equivalent weight of standard
acid polysaccharide - Xanthan gum – as mg Xeq/L:

At Af As
TEP0.4 m
= Eq. 2
m787V f

where m787 is the slope of the calibration line [(abs/cm)/mg Xeq] which is determined by
calibrating the absorbance of AB corresponding to the mass of the Xanthan gum stained as
described in Sections 12.2.4.2 or 12.2.4.3.

295
Experimental Methods for Membrane Applications

water sample synthetic water

1 (similar salinity
with sample)
filtered
volume (vf)
0.4 µm PC 0.4 µm PC 0.4 µm PC

TEP0.4 µm
0.2 bar vacuum

2
2 mL ultra-pure water

At As Af
< 0.2 bar
vacuum

3 10 sec.
reaction time 1 mL 0.025% Alcian Blue, pH 2.5
0.05 µm pre-filtered
At As Af
< 0.2 bar
vacuum

4
2 mL ultra-pure water

At As Af
< 0.2 bar
vacuum

5
50 mL glass soak in 6 mL 80% H2SO4
beakers
At As Af

autoshaker
for 2 hours

6 787 nm
Measure absorbance
at 787 nm (At, Af and As )

7
Absorbance

Slope = m787
@787 nm
abs/cm

At-Af-As Calibration with

TEP0.4µm = Xanthan gum
m787Vf
Xanthan gum (µg)

• TEP0.4µm = TEP concentration (mg Xeq/L)

• At = total absorbance (abs/cm)
• Af = filter blank absorbance (abs/cm)
• As = sample blank absorbance (abs/cm)
• m787 = slope of the calibration line [(abs/cm)/mg Xeq]
• Vf = total volume of filtered sample (L)

Figure 5 Procedural diagram for measuring TEP0.4μm (adapted from Villacorte, 2015).

296
 Chapter 13

13.2.3 TEP10kDa measurement

The TEP10kDa method was developed to measure TEP and their precursors (down to 10
kDa). This method was described by Villacorte et al. (2015b) and partly based on the method
developed by Thornton et al. (2007). Figure 7 illustrates the protocol for measuring TEP10kDa
and further described in the following sections.

Materials and analytical set-up

• Alcian blue working solution (see Section 13.2.1)
• 10 kDa MWCO flat sheet membrane (25 mm regenerated cellulose)
• 60 mL plastic syringe
• Syringe pump (see Figure 6a)
• Filter holder 25 mm
• Ultrapure water (UPW), for rinsing membrane filters.
• Acetic acid
• Clean plastic cups (for collecting 10 kDa filtered samples)
• Vortex mixer
• Sonicator (see Figure 6b)
• Vacuum filtration set-up for 25 mm filter with stainless steel filter support (see Figure 6c)
• Vacuum pump
• Filter 0.1 μm polycarbonate membrane filters (47 mm diameter)
• clean plastic cups (for capturing 0.1 μm filtered stained sample)
• Demineralised water (demi-water), for cleaning filtration set-up
• Glass beaker
• Spectrophotometer
• Pipette 1 mL, 0.1 mL
• Tweezer
• Measuring cylinder

Figure 6 Overview of analytical setup for TEP10kDa measurement: (a) syringe filtration unit, (b)
ultrasonication unit and (c) vacuum filtration unit.

297
Experimental Methods for Membrane Applications

1 water sample 2 10 mL air 3 5 mL UPW 4 10 mL air

Constant flux 60 (L/m2)/h 60 (L/m2)/h 60 (L/m2)/h
[J = 60 (L/m2)/h]

10 kDa RC 10 kDa RC 10 kDa RC 10 kDa RC

Total filtered volume

Vf>10 mL)

5 Place membrane feed 6 Vortex for 10 sec + 7 4 mL 4 mL

side down and submerge sonicate for 60 min sample UPW
in UPW water

Disposable plastic container Sonicator bath sample blank

(40 ml capacity) with Disposable plastic container
10 ml (Vr) of UPW (20 ml) capacity

8 0.05 mL acetic acid 9 1 mL 0.025% AB, pH = 2.5 10 Filtration to remove

(0.05 µm pre-filtered) TEP-AB precipitates
4 mL sample 4 mL blank

0.1 µm PC 0.1 µm PC

sample blank sample blank

pH adjustment + mixing Alcian blue staining +

(pH = 2.5) mixing (10 min reaction time)
Ae Ab

Filtrate containing excess AB

• TEP10kDa = TEP concentration (mg Xeq/L)

• m610 = neg. slope of the calibration line [(abs/cm)/mg Xeq]
• Vr = total volume of re-suspended TEP solution in UPW (10 mL)
• Vf = total volume of filtered sample (L)
• Ae = absorbance of the excess or non-precipitated dye (abs/cm)
11 • Ab = absorbance of the blank (abs/cm)

610 nm Calibration with 12

Absorbance
@610 nm

Xanthan gum
abs/cm

Slope = m610

XG (mg/L)

1 Vr
13 TEP10 kDa= (Ae -Ab)
m610 Vf

Figure 7 Procedural diagram for measuring TEP10kDa (adapted from Villacorte, 2014.)

298
 Chapter 13

Sample filtration and TEP extraction

1. Place a clean 10 kDa MWCO regenerated cellulose (RC) membrane (25 mm diameter)
to the syringe filter holder and filter 10-100 mL water sample at constant flux (60 L/
m2/h) using a syringe pump while collecting the filtrate in a container. Note: To remove
possible contaminants, soak RC membranes for at least 24 hours in UPW and then
flushing 5-10 ml UPW prior to sample filtration.
2. After filtering sample, replace the syringe with a new clean syringe containing 5-10 mL
of air. Filter air through to the filter (60 L/m2/h) until all the remaining water in the feed
side of the membrane holder has passed through the membrane. Measure total volume of
sample in the filtrate container using a measuring cylinder.
3. To rinse out residual saline moisture, filter 5 mL of UPW through the filter holder at
60 L/m2/h. Filter air again as in step 2 until all the rinse water on the feed side of the
membrane holder has passed through the membrane.
4. Carefully remove the membrane from the filter holder and place feed side down, in a
clean disposable plastic container filled with 10 mL of UPW.
5. Cover the container, vortex for 10 s and sonicate for 60 min.
6. Transfer 4 mL of the re-suspended TEP solution to a clean 20 mL disposable plastic
container. Adjust the sample pH to 2.5 by adding 0.05 mL of acetic acid solution the
sample solution.
7. Add 1 mL of the working AB dye solution to the sample, mix vigorously and allowed to
react for 10 min.
8. Rinse PC filter by filtering 4 mL of UPW try to remove all the remaining UPW inside of
the funnel. Any extra UPW might dilute the TEP-AB filtrate and lower the absorbance
reading. Filter 4 mL sample of the TEP-AB solution through a 0.1 μm PC filter by vacuum
filtration (0.2 bar). Collect the filtrate in a plastic container (10 mL) as shown in Figure 8.
9. Transfer part of the filtrate to a 1-cm cuvette and measure absorbance (Ae) at 610 nm
wavelength using a spectrophotometer.
10. Measure the blank absorbance (Ab) to determine
amount of AB stain adsorbed on PC filter by following
steps 6-9 but replacing the sample with UPW.

To prevent cross-contamination of samples, it is

recommended to clean the syringe filter holder and
vaccum filter holder with UPW in between tests
until no more visible AB remains on the holders.
The filtered sample volume in Step 1 can be adjusted
(increased or reduced) depending on initial absorbance
results. To get more reliable absorbance results, it is
recommended to reduce sample volume if initial
absorbance result is ≤0.05 cm-1 and increase sample
volume if initial absorbance is very close to or higher
than the blank absorbance.

Figure 8 An example of how to assemble the setup to collect

residual AB solution directly into a plastic sample container
during vacuum filtration.

299
Experimental Methods for Membrane Applications

Concentration calculation (without calibration)

The TEP10kDa concentration in absorbance per cm per liter of filter water (abs/cm/L) is
calculated as follows:
Ab − Ae
TEP10kDa = Eq. 3
Vf

where Ab is the absorbance of filtered blank (abs/cm), Ae is the absorbance of the excess or
un-reacted dye (abs/cm) and Vf is the volume of filtered sample (mL).

Concentration calculation (with calibration)

Ideally, the TEP10kDa concentration can be calibrated and expressed in terms of mg Xanthan
equivalent per litre (mg Xeq/L):
1 Vr
TEP10kDa = ( A − Ab ) Eq. 4
m610 V f e

where Vr is the total volume of the re-suspended TEP sample solution (i.e., 10 mL) and m610
is the slope of the calibration curve [(abs/cm)(mg Xeq/L)] which is determined by calibrating
the absorbance of residual AB corresponding to the mass of the Xanthan gum stained on the
filter (see section 13.2.4.4).

13.2.4 Method calibration

calibration of absorbance results from TEP analyses (Li et al., 2018). XG is by far the most
widely accepted standard among surrogates. However, Thornton et al. (2007) argued that
alginic acid is a more suitable standard for TEP precursors. For consistency and comparability
of results, it is recommended to use XG as the standard for quantifying both TEPs and their
precursors.

The first TEP calibration protocol introduced by Passow and Alldredge (1995) involves
dry weight measurements to determine the mass of Xanthan gum retained on PC filters.
The procedure is tedious and prone to inaccuracies particularly during drying (dust
contamination) and weighing (electrostatic force interference) of very low quantities (5-50
μg) of XG. It can be also challenging to prepare a homogeneous and artifact-free solution
of XG for the calibration. Various TEP studies have skipped the calibration step entirely,
whereby concentrations of TEP are expressed in terms of abs/cm/L (see Section 13.2.3).
Without calibration, TEP results cannot be directly compared with results from analyses
using different batches of AB staining solution. To overcome the above challenges,
improved TEP calibration protocols have been introduced without the drying and weighing
steps. These new protocols are described in the succeeding sections.

13.2.4.1 Xanthan gum standard preparation

A standard XG solution is prepared by dissolving XG in UPW solution. For example, to
prepare 100 mg/L solution, 50 mg of XG is added to 500 mL of UPW while rapidly stirring
with a magnetic stirrer. Rapid stirring is maintained for at least 1 hour until no flocs are
visible. Typically, the solution is further homogenized with a tissue grinder. Each volume of
50-100 mL of Xanthan gum solution is homogenized 3 times by fully rotating the pestle 5

300
 Chapter 13

times for each batch. A recent study by Bittar et al. (2018) found that current commercially
available XG is easier to dissolve and forms negligible number of gel-like particles in solution
compared with the earlier versions of the XG powder used in the original calibration
method (Passow and Alldredge, 1995). Hence, the use of a tissue grinder may no longer
be necessary to homogenize the standard XG solution. The following sections describe the
different calibration protocols for TEP0.4μm and TEP10kDa.

13.2.4.2 TEP0.4µm calibration 1

Villacorte et al. (2015b) introduced a simpler and replicable standard calibration procedure
than the original method for TEP0.4μm. The calibration steps are as follows:
1. Prepare 4 ml standard solutions containing different concentrations of XG (0, 1, 2, 3, 4
and 5 mg/L) by diluting of XG stock solution (100 mg/L) with UPW.
2. Adjust pH of solutions to pH 2.5 by adding 0.05 mL acetic acid to each solution and then
briefly agitated.
3. Take one solution and add 1 mL of pre-filtered AB solution, mix for 10 s and incubate fo
10 min.
4. Filter 4 mL of the reacted solution through a 0.1 μm PC membrane by vacuum filtration
(0.2 bar).
5. While in the filter holder, fold the PC membrane with feed side in and carefully transfer
to a 50 mL beaker.
6. Add 6 mL of 80% sulfuric acid solution to the beaker, cover it with a parafilm and mix on
an auto-shaker for 2 hours.
7. Transfer part of the acid solution to a 1-cm cuvette and measure absorbance at 787 nm
using a spectrophotometer.
8. Repeat steps 3-7 for each of the remaining standard solutions.
9. To determine the calibration slope (m787), the mass of Xanthan gum retained on the PC
membrane is calculated by multiplying the volume filtered (4 mL) with the concentration
of Xanthan in the stained standard solution. The calculated mass is then plotted against
the corresponding AB absorbance measured at 787 nm wavelength, whereby the average
linear slope is the m787.

13.2.4.3 TEP0.4µm calibration 2

Bittar et al. (2018) introduced an alternative TEP0.4μm calibration procedure as follows:
1. Prepare 75 mg/L of XG stock solution in UPW. To make XG standard solution, pipette
0.125, 0.250, 0.500, 0.750, and 1 mL of XG stock solution, in triplicates, into clean
5-mL polypropylene tubes. Top-up with UPW to bring the final volume of the solutions
to 1 mL. The prepared XG standard solutions should contain 9.37, 18.75, 37.50, 56.25,
and 75 μg-XG
2. Stain procedural blanks (1 mL ultrapure water in triplicates) and standard solutions by
adding 0.5 mL of AB solution (400 mg/L) to the polypropylene tubes, for a final volume
of 1.5 mL, and mix by manually agitating the tubes for 1 min.
3. Pour the stained standard solutions directly to the vacuum filtration funnel and filter
through 0.22 μm or 0.45 μm polycarbonate filters (25 mm) at <175 mm Hg.
4. Transfer the filters to clean glass vials/beakers with caps/covers.
5. Add 6 mL of extraction solution (80% sulfuric acid) to the vials/beakers and cover
immediately. Soak filters in in the solution for 2–20 h while vials are agitated regularly.

301
Experimental Methods for Membrane Applications

6. Measure the absorbance of AB extracted from the filters at 787 nm in a spectrophotometer

using a 1-cm cuvette. During absorbance measurement, the spectrophotometer is
first blanked with ultrapure water and then the absorbance of acid extraction solution
is measured at 787 nm to check for consistency and potential contaminations of each
extraction solution batch (0.008–0.150 ± 0.004). The instrument is then further blanked
with extraction solution.
7. To determine the calibration slope (m787), the mass of XG retained on the PC membrane
is calculated by multiplying the XG stock solution concentration (75 mg/L) with the
volume of XG stock solution used for each point in the calibration (0.125, 0.250, 0.500,
0.750, and 1 mL). The calculated mass is then plotted against the corresponding AB
absorbance measured at 787 nm wavelength, whereby the average linear slope is the
m787.

13.2.4.4 TEP10kDa calibration

For TEP10kDa, Villacorte et al. (2015b) developed a new calibration protocol modified from
the protocol described by Thornton et al. (2007). This new calibration protocol can be
performed simultaneously with TEP0.4μm calibration as illustrated in Figure 9.
1. Prepare standard solutions (4 mL) containing different concentrations (0, 1, 2, 3, 4 and 5
mg/L) of XG from standard stock solution (Section 12.2.4.1).
2. Adjust sample pH to 2.5 by adding 0.05 mL acetic acid to each solution and then agitated
briefly. The solution is then stained by adding 1 mL of pre-filtered AB staining solution,
mixed for 10 seconds, and incubated for 10 min.
3. Filter 4 mL of the resulting solution through a 0.1 μm PC membrane by vacuum filtration
(0.2 bar). The filtrate is collected, transferred to 1-cm cuvette and absorbance is measured
at 610 nm.
4. The concentration of standard XG solution stained with AB is plotted against the
measured absorbance (excess dye absorbance) and the average linear line is the calibration
slope (m610). Since concentration is inversely proportional to the excess dye absorbance,
the calibration slope (m610) is a negative value.

13.2.5 Other considerations

13.2.5.1 Limit of detection

The lower limit of detection (LODmin) of the TEP methods depends on the variability of
the blank absorbance. Villacorte (2014) calculated LODmin as follows:

TEP0.4μm : LODmin = 3σb (1/m787) (1/Vf)

TEP10kDa : LODmin = 3σb (1/m610) (Vr /Vf)

where σb is the standard deviation of 10 independently measured blank absorbance (abs/

cm). The factor 3 corresponds to a significance level of 0.00135, which means that only
0.135% of blank measurements will statistically yield results that fall above the computed
detection limit (Harvey, 2000).

302
 Chapter 13

The upper limit of detection (LODmax) is the upper threshold of the absorbance which
can yield reliable concentration results. For TEP0.4μm, this limit is determined based on
the maximum absorbance at which a linear correlation between absorbance and filtered
volume can be observed. For TEP10kDa, this limit is the minimum absorbance at which the
excess stain absorbance and the standard concentration has a significant linear correlation.
In practice, the recommended absorbance thresholds are 0.8 abs/cm and 0.05 abs/cm for
TEP0.4μm and TEP10kDa, respectively (Villacorte, 2014)

0.05 mL 1 mL AB
4 mL samples acetic acid solution 10 sec mixing +
prefiltered 10 min incubation

Soak filter in
6 mL 80% H2SO4

Blank XG solutions
(1-5 mg/L)
Filter 4 mL of 0,1 µm PC 0,1 µm PC
stained samples
(0.2 bar vacuum)
autoshaker
for 2 hours

Collect filtrate solution

787 nm
Slope = m787
Absorbance
@787 nm
abs/cm

Measure absorbance at 787 nm Xanthan gum (µg)

Absorbance

610 nm
@610 nm
abs/cm

Slope = m610

Measure absorbance at 610 nm Xanthan gum (mg/L)

Figure 9 Overview of analytical steps for combined calibration of TEP0.4μm and TEP10kDa

Bittar et al. (2018) applied a different approached to calculate LOD for TEP0.4μm. The
LODmin, in μg-XG, for each individual curve was calculated according to Corley (2003),

Y1
TEP0.4 m
: LOD min = 3 •
m

where Y1 is the mean square error of all calibration points and m is the slope of the regression.

303
Experimental Methods for Membrane Applications

The LODmax is defined as the raw absorbance values not exceeding 1, given that the amount
of transmitted light detected by the spectrophotometer decreases exponentially (Beer–
Lambert Law), corresponding to 10 and 1% for absorbance values between 1 and 2. For
TEP0.4μm, absorbance is typically lower than 1 at ~ 75 μg of AB-stained XG.

13.2.5.2 Impact of storage on TEP concentration

Storing water samples for a prolong period before analysis can lead to significant disparity
between the measured and in-situ TEP concentration. TEP loss/gain during storage
may be attributed to coagulation, bacterial release, bacterial degradation, adsorption on
walls of sample bottles or a combination. Bottle tests conducted by Villacorte (2014)
revealed that TEP0.4μm concentration may increase over a long period in storage either
due to coagulation of TEP precursors or through bacterial TEP release. On the other hand,
TEP10kDa concentration can rapidly decrease by up to 45% within the first 3 days of storage
resulting from either bacterial degradation or adsorption to walls of sample bottles. Further
investigations are necessary to fully understand the mechanisms involved in the TEP loss or
increase as well as to develop reliable measures to preserve TEP samples (e.g., sample bottle,
freezing, preservative addition). It is therefore important that samples should be analyzed
immediately (within 24 hours) after sampling to obtain reliable TEP concentrations. For
TEP10kDa, it is recommended to filter and the sample immediately after sampling so the
membrane can be stored at 4˚C until further analysis.

13.2.6 Application and interpretation

Over the last 15 years, a significant number of experimental studies applied various TEP
methods to investigate the impact of TEPs and their precursors in membrane processes. A
non-exhaustive overview of these studies is shown in Table 2.

If implemented properly, TEP methods such as TEP0.4um and TEP10kDa can be effective tools
in quantifying the presence of TEP and their precursors in the s he pretreatment process or
in the feed/product/concentrate streams of a membrane system. TEP0.4μm measures mainly
TEP while TEP10kDa can measure both TEP and their colloidal precursors. TEP0.4μm is a
relatively more rapid and cheaper method than TEP10kDa which means it is ideal for routine
TEP monitoring in untreated water sources particularly during algal blooms. If the objective
is to assess the removal efficiencies of TEP and their precursors through the treatment
processes, TEP10kDa measurement is more appropriate method because it covers both TEP
and their colloidal precursors.

Villacorte (2014) measured TEP0.4μm regularly for 3 years in a seawater desalination plant
and illustrated that the occurrence of TEP generally coincide with the seasonal algal bloom
based on chlorophyll-a concentration (Figure 10). The study also demonstrated that
chlorophyll-a concentration is not a reliable indicator of the abundance of TEP, because
some bloom-forming algal species produce more TEP than others (Villacorte et al., 2015c).

304
 Chapter 13

Table 2 Overview of membrane related studies applying TEP quantification methods.

Membrane Scope/highlights of the
Reference TEP method processes Water type study
De la Torre et al. Rapid spectro- MF/MBR Wastewater, TEP as potential fouling
(2008) photometric activated sludge indicator for MBR systems
Kennedy et al. TEP0.4µm UF Surface water, TEP removal in UF
(2009) treated wastewater membrane with inline
coagulation
Villacorte et al. TEP0.4µm MF/UF, RO lake water, seawater TEP removal in 6 integrated
(2009b) TEP0.05µm membrane systems including
pretreatment
Berman et al. TEP0.4µm UF, RO Lake water Investigate role of TEP in
(2011) aquatic biofilm initiation and
membrane fouling
Bar-Zeev et al. TEP0.4µm RO Seawater Removal efficiency of
(2012b) pretreatment for RO
van Nevel et al. TEP0.4µm UF, RO Treated wastewater, Removal efficiencies of
(2012) TEP0.05µm surface and ground UF-RO system and other
water treatment methods
Valladares Linares Direct AB FO Treated wastewater TEP identified as major
et al. (2012) staining foulant in FO
Discart et al. TEP0.4µm MF Freshwater Role of TEP in membrane
(2014) TEP0.05µm fouling during algae broth
filtration
Villacorte et al. TEP0.4µm MF/UF, RO Fresh/saline Characterization of TEP
(2015c) TEP10kDa produced by bloom forming
algae, from a membrane
fouling perspective
Villacorte et al. TEP0.4µm MF/UF Fresh/saline MF/UF rejection and
(2015a) TEP10kDa fouling potential of algal
organic matter from bloom-
forming algae
Villacorte et al. TEP0.4µm RO Saline TEP quantification of biofilm
(2017a) extracted from biofouled
capillary spiral wound
membranes
Li et al. (2016) TEP0.4µm RO Saline TEP/TEP precursors
TEP0.1µm monitoring through
pretreatment and SWRO
processes
Meng and Liu TEP0.04µm UF Model fresh/saline TEP-associated UF fouling is
(2017) TEP0.05µm water more severe with freshwater
that with seawater
Zhang et al. TEP0.4µm UF Freshwater bacterial TEP-induced irreversible
(2023) culture fouling in UF can be reduced
by the MIEX pretreatment

305
Experimental Methods for Membrane Applications

0.8
40
60 chlorophyll-a 0.7
water temperature 35

watertemperature (˚C)
50 0.6
Chlorophyll-a (µg/L)
TEP 0.4 µm 30

TEP0.4 µm (mg Xeq/L)

40 0.5
25
0.4
30 20
0.3
15
20
0.2 10
10 0.1 5
0 0 0
Dec-08 May-09 Oct-09 Mar-10 Aug-10 Jan-11 Jun-11 Nov-11 Apr-12 Aug-12

Figure 10 Variations of TEP0.4μm, chlorophyll-a and water temperature at the intake water of
seawater RO plant. Adopted from Villacorte (2014).

Measuring both TEP0.4μm and TEP10kDa on the same water samples can offer a better
understanding of the TEPs produced by different species of algae in pure culture applications
(Villacorte et al., 2015c). As shown in Figure 11, The TEP production and concentrations
over time varies substantially for 3 species of bloom-forming fresh and saline water algae.

a) b)
30 80 100
Algal concentration (x103 cells/mL)

Algal concentration (x103 cells/mL)

60 cell concentration cell concentration
TEP + precursors (mg Xeq/L)

TEP + precursors (mg Xeq/L)

70
25 TEP + precursors TEP + precursors 80
50 TEP TEP 60
20 60
40 50
15 40 40
30
30
10 20 20
20
5 10 10 10

0 0 0 0
0 10 20 30 40 50 60 0 5 10 15 20 25 30

c) d)
7 160 75
Algal concentration (x103 cells/mL)

cell concentration Alexandrium tamarense

TEP + precursors (mg Xeq/L)

TEP + precursors (mg Xeq/L)

6 140
TEP + precursors Chaetoceros affinis
60
5 120 TEP Microcystis sp.
100 45
4
80
3 30
60
2 C. affinis: y = 4.7x+9.9; R2=0.91
40
A. tamarense: y=8.0x+2.2; R2=0.83 15
1 20 Microcystis sp.: y=2.0x+1.6; R2=0.84
cell concentration
0 0 0
0 4 8 12 16 20 24 28 32 0 2 4 6 8 10 12

Figure 11 TEP concentration profile over days of incubation for three species of bloom-forming
algae: (a) Alexandrium tamarense, (b) Chaetoceros affinis, (c) Microcystis sp.; and (d)
correlation between TEP0.4μm (TEP) and TEP10kDa (TEP + precursors) concentrations.
Adopted from Villacorte et al. (2015c).

306
 Chapter 13

Monitoring TEP0.4um and TEP10kDa through the pretreatment processes of membrane

plant can reveal TEP removal efficiencies of the treatment steps. As shown in Figure 12, TEP
can be fully removed by UF, but some TEP precursors can still pass through. A significant
correlation was also found for TEP and organic biopolymers measured using the LC-OCD
method (see chapter 11).

1.2
TEP + precursors (mg Xeq/L)

a)
TEP + precursors
Biopolymers (mg C/L)

TEP 1.0
Biopolymers
0.8
0.6
0.4
0.2

0
Raw water After After coag/ After UF RO
intake microstrainer sed/RSF GAC permeate permeate

b) c) d)

TEP + precursors (mg Xeq/L)

0.8 0.8 1.4
Biopolymers (mg C/L)

Biopolymers (mg C/L)

1.2
0.6 0.6
1.0
0.8
0.4 0.4
0.6
y = 12.1x+0.13 y = 22.68x+0.10 0.4
0.2 y = 0.54x+0.08 0.2
R2 = 0.89 R2 = 0.93
R2 = 0.97
0.2
0 0 0
0.0 0.02 0.04 0.06 0.0 0.4 0.8 1.2 0.0 0.02 0.04 0.06
TEP (mg Xeq/L) TEP + precursors (mg Xeq/L) TEP (mg Xeq/L)

Figure 12 Concentrations of biopolymer (measured by LC-OCD), TEP (TEP0.4μm) and TEP +

precursors (TEP10kDa) in (a) samples collected over the treatment processes of an RO
plant plant and (c-d) linear regressions between measured parameters. Note: coag =
coagulation + flocculation; sed = sedimentation; RSF = rapid sand filtration; GAC =
granular activated carbon; UF = ultrafiltration. Adopted from Villacorte et al. (2015b).

Quantifying TEP together with LC-OCD to measure biopolymers and MFI-UF (see chapter
8) to measure the bulk fouling potential, has also been implemented. As shown in Figure
13, a significant correlation was reported between TEP10kDa and MFI-UF, which has better
observed correlation than between biopolymers and MFI-UF (Villacorte, 2014). TEP0.4μm
showed lower correlation with MFI-UF likely because TEP precursors (< 0.4μm) were not
measured in the method.

307
Experimental Methods for Membrane Applications
a) b) c)
24 24 24
20 20 20
16 16 16
12 12 12

MF-UF (s /L2)

MF-UF (s /L2)
MF-UF (s /L2)

8 8 8
y = 104514x+3569.5 4 y = 12911x+1352.3 4 y = 19230x+1527.3 4
R2 = 0.371 R2 = 0.669 R2 = 0.449
0 0 0
0.0 0.02 0.04 0.06 0.06 0.0 0.4 0.8 1.2 1.6 0.0 0.2 0.4 0.6 0.8 1.0
TEP0.4µm (mg Xeq/L) TEP10kDa (mg Xeq/L) Biopolymers (mg C/L)

Figure 13 Correlation between membrane fouling potential as MFI-UF in relation to (a) TEP0.4μm,
(b) TEP10kDa and (c) biopolymers. Adopted from Villacorte (2014).

TEP accumulation on membranes can be quantified from foulant or biofilm extracted from
a fouled membrane (Villacorte et al., 2017a). Figure 14 shows TEP0.4μm concentrations on
membrane and spacers in a series of membrane fouling simulator (MFS) experiments. In
this application, TEP results can be directly correlated to the hydraulic performance of the
membrane (i.e., pressure drop) or bulk organic parameter (i.e., TOC).

MFS1 MFS2

MFS3 MFS4

0.8 90 350
pressure drop 80
0.7 TEP 300
Total organic carbon (µg/cm2)
TOC 70
0.6
250
Final pressure drop (bar)

60
0.5
TEP 0.4 µm (µg Xeq/cm2)

50 200
0.4
40 150
0.3 30
100
0.2 20
0.1 10 50

0 0 0
MFS 1 MFS 2 MFS 3 MFS 4
AOM pre-conditioned Reference/control

Figure 14 TEP0.4μm and TOC concentrations in biofilms on membranes and spacers from membrane
fouling simulator (MFS) experiments in relation to the measured feed channel pressure
drop. Top images are showing biofilm accumulation on initially clean RO membranes
(MFS 3+4) and on algal organic matter (AOM) pre-conditioned RO membrane at day
10 of the experiments. Biofilm samples extracted for TEP/TOC measurements were
taken on day 10 for MFS 1 and on day 20 for MFS 2, 3 and 4. Adopted from Villacorte
et al. (2017a).

308
 Chapter 13

13.3 SUMMARY AND OUTLOOK

Various TEP quantification methods were developed over the years, and some are widely
applied in membrane filtration studies. The experimental protocols to measure TEP0.4μm
and TEP10kDa are described in detail in this chapter as a promising tool to semi-quantitatively
or qualitatively investigate the impact of TEP to the operational performance of membrane
processes. Despite recent improvements, these TEP methods still have some limitations
so it should be implemented with proper attention to the protocol used and by someone
who is experienced with laboratory analytical techniques. Critical analyses of the results,
particularly when comparing results from different sets of samples, different TEP methods
and different studies, should include assessing what steps were taken to minimize the
impact of the following (Discart et al., 2014; Li et al. 2018; Meng et al., 2020):

• variability of staining capacity of AB solutions relative to age or chemical supplier,

• variability of filter characteristics used to retain TEP (e.g., pore size and distribution),
• variability of surrogates or standard solutions used to calibrate the method,
• variability of salinity in the water samples.

Further improvements of the TEP methods should therefore focus on minimizing the
impacts of above-mentioned variabilities to the results.

Another practical limitation of current TEP methods is that they are time consuming.
TEP0.4μm and TEP10kDa analyses can take up 2-4 hours and 3-4 hours per sample, respectively.
APS method has shorter analytical time than TEP10kDa for freshwater samples, but it can take
up to 24h for saline samples due to the additional dialyses step. If current TEP methods
are used as parameter to optimize operation of a membrane processes, the analytical time
delay in addition to transport time of samples to the laboratory may prevent the operator
to timely mitigate potential fouling issues. New online TEP measurement methods may cut
down analytical time to less than 1 hour, but further verifications are needed if such rapid
methods have similar/better accuracy than offline methods.

309
Experimental Methods for Membrane Applications

13.4 REFERENCES

Alldredge, A. L., Passow, U., & Logan, B. E. (1993). The abundance and significance of a class of large,
transparent organic particles in the ocean. Deep Sea Research Part I: Oceanographic Research
Papers, 40(6), 1131-1140.
Arruda-Fatibello, S. H. S., Henriques-Vieira, A. A., & Fatibello-Filho, O. (2004). A rapid spectrophotometric
method for the determination of transparent exopolymer particles (TEP) in freshwater. Talanta,
62 (1), 81-85.
Azetsu-Scott, K., & Passow, U. (2004). Ascending marine particles: Significance of transparent
exopolymer particles (TEP) in the upper ocean. Limnology and Oceanography, 49(3), 741-748.
Bar-Zeev, E., Belkin, N., Liberman, B., Berman, T., & Berman-Frank, I. (2012b). Rapid sand filtration
pretreatment for SWRO: Microbial maturation dynamics and filtration efficiency of organic
matter. Desalination, 286, 120-130.
Bar-Zeev, E., Berman-Frank, I., Girshevitz, O., & Berman, T. (2012a). Revised paradigm of aquatic
biofilm formation facilitated by microgel transparent exopolymer particles. PNAS, 109(23),
9119-24.
Bar-Zeev, E., Berman-Frank, I., Liberman, B., Rahav, E., Passow, U., & Berman, T. (2009). Transparent
exopolymer particles: Potential agents for organic fouling and biofilm formation in desalination
and water treatment plants. Desalination and Water Treatment, 3, 136–142.
Berman, T., & Holenberg, M. (2005). Don’t fall foul of biofilm through high TEP levels. Filtration and
Separation, 42(4), 30-32.
Berman, T., Mizrahi, R., & Dosoretz, C.G. (2011). Transparent exopolymer particles (TEP): A critical
factor in aquatic biofilm initiation and fouling on filtration membranes. Desalination, 276, 184-
190.
Bittar, T. B., Passow, U., Hamaraty, L., Bidle, K. D., & Harvey, E. L. (2018). An updated method for the
calibration of transparent exopolymer particle measurements. Limnology and Oceanography:
Methods, 16(10), 621-628.
Chin, W. C., Orellana M. V., & Verdugo P. (1998). Spontaneous assembly of marine dissolved organic
matter into polymer gels. Nature, 391, 568–572.
Claquin, P., Probert I., Lefebvre S., & Veron B. (2008). Effects of temperature on photosynthetic
parameters and TEP production in eight species of marine microalgae. Aquatic Microbial Ecololy,
51, 1-11.
Corley, J. (2003). Best practices in establishing detection and quantification limits for pesticide residues
in foods, p. 59–74. In P. W. Lee [ed.] Handbook of residue analytical methods for agrochemicals,
v. 1. Wiley.
de la Torre, T., Lesjean, B., Drews, A., & Kraume, M. (2008). Monitoring of transparent exopolymer
particles (TEP) in a membrane bioreactor (MBR) and correlation with other fouling indicators.
Water Science and Technology, 58 (10), 1903–1909.
Discart, V., Bilad, M. R., van Nevel, S., Boon, N., Cromphout, J., & Vankelecom, I. F. J. (2014). Role of
transparent exopolymer particles on membrane fouling in a full-scale ultrafiltration plant: Feed
parameter analysis and membrane autopsy. Bioresource Technology, 173, 67-74.
Harvey, D. (2000) Modern analytical chemistry. McGraw-Hill, USA.
Kennedy, M. D., Muñoz-Tobar, F. P., Amy, G. L., & Schippers, J. C. (2009). Transparent exopolymer
particle (TEP) fouling of ultrafiltration membrane systems. Desalination and Water Treatment,
6(1-3) 169-176.

310
 Chapter 13

Li, S., Sinha S., Leiknes T., Amy G. L., & Ghaffour N. (2016). Evaluation of potential particulate/
colloidal TEP foulants on a pilot scale SWRO desalination study. Desalination, 393, 127–134.
Li, X., Skillman, L., Li, D., & Ela, W. P. (2018). Comparison of Alcian blue and total carbohydrate
assays for quantitation of transparent exopolymer particles (TEP) in biofouling studies. Water
Research, 133, 60-68.
Meng, S., & Liu, Y. (2017). Transparent exopolymer particles (TEP)-associated membrane fouling at
different Na+ concentrations. Water Research, 111, 52-58.
Meng, S., Meng, X., Fan, W., Liang, D., Wang, L., Zhang, W., & Liu, Y. (2020). The role of transparent
exopolymer particles (TEP) in membrane fouling: A critical review. Water Research, 181,
115930.
Mopper, K., Zhou, J., Sri Ramana, K., Passow, U., Dam, H. G., & Drapeau, D. T. (1995). The role of
surface-active carbohydrates in the flocculation of a diatom bloom in a mesocosm. Deep-Sea
Research Part II, 42(1), 47-73.
Passow, U. (2000). Formation of transparent exopolymer particles (TEP) from dissolved precursor
material. Marine Ecology Progress Series, 192, 1-11.
Passow, U. (2002). Transparent exopolymer particles (TEP) in aquatic environments. Progress in
Oceanography, 55(3), 287-333
Passow, U., & Alldredge, A. L. (1995). A dye-binding assay for the spectrophotometric measurement
of transparent exopolymer particles (TEP). Limnology and Oceanography, 40(7), 1326-1335.
Schurer, R., Janssen, A., Villacorte, L. O., & Kennedy, M. D. (2012). Performance of ultrafiltration and
coagulation in an UF-RO seawater desalination demonstration plant. Desalination and Water
Treatment, 42(1-3), 57-64.
Schurer, R., Tabatabai, A., Villacorte, L. O, Schippers, J. C., & Kennedy, M. D. (2013). Three years
operational experience with ultrafiltration as SWRO pretreatment during algal bloom.
Desalination and Water Treatment, 51 (4-6), 1034-1042.
Sim, L. N., Suwarno, S. R., Lee, D. Y. S., Cornelissen, E. R., Fane, A. G., & Chong, T. H. (2019).
Online monitoring of transparent exopolymer particles (TEP) by a novel membrane-based
spectrophotometric method. Chemosphere, 220, 107-115.
Thornton, D. C., Fejes, E. M., DiMarco, S. F., & Clancy, K. M. (2007). Measurement of acid polysaccharides
in marine and freshwater samples using alcian blue. Limnology and Oceanography: Methods,
5(2), 73-87.
Thuy, N. T., Huang, C. P., & Lin, J. L. (2017). Visualization and quantification of transparent exopolymer
particles (TEP) in freshwater using an auto-imaging approach. Environmental Science and
Pollution Research, 24, 17358-17372.
Valladares-Linares, R., Yangali-Quintanilla, V., Li, Z., & Amy, G. L. (2012). NOM and TEP fouling of
a forward osmosis (FO) membrane: Foulant identification and cleaning. Journal of Membrane
Science, 421–422 (2012) 217–224.
van Nevel, S., Hennebel, T., De Beuf, K., Du Laing, G., Verstraete, W., & Boon, N. (2012). Transparent
exopolymer particle removal in different drinking water production centers. Water Research,
46(11), 3603-3611.
Verdugo, P., Alldredge, A. L., Azam, F., Kirchman, D. L., Passow, U., & Santschi, P. H. (2004). The
oceanic gel phase: a bridge in the DOM–POM continuum. Marine Chemistry, 92, 67-85.
Villacorte, L. O. (2014). Algal blooms and membrane-based desalination technology. ISBN 978-1-
138-02626-1, CRC Press/Balkema, Leiden.

311
Experimental Methods for Membrane Applications

Villacorte, L. O., Boerlage, S. F. E., & Dixon, M. B. (2021). Ch 6 – Algal blooms and RO desalination. In:
Salinas-Rodriguez, S. G., Schippers, J. C., Amy, G. L., Kim, I. S., & Kennedy, M. D. (eds.) Seawater
Reverse Osmosis Desalination: Assessment & Pre-Treatment of Fouling and Scaling (pp. 145–
176). IWA Publishing, London.
Villacorte, L. O., Ekowati, Y., Calix-Ponce, H. N., Kisielius, V., Kleijn, J. M., Vrouwenvelder, J. S.,
Schippers, J. C., & Kennedy, M. D. (2017a.). Biofouling in capillary and spiral wound membranes
facilitated by marine algal bloom. Desalination, 424, 74-84.
Villacorte, L. O., Schippers, J. C. & Kennedy, M. D. (2017b). Appendix 3. Methods for measuring
transparent exopolymer particles and their precursors in seawater. In: Anderson, D. M.,
Boerlage, S. F., & Dixon, M. B. (eds.) Harmful Algal Blooms (HABs) and Desalination: A Guide
to Impacts, Monitoring and Management. IOC Manuals and Guides No. 78 (pp. 501–507).
Intergovernmental Oceanographic Commission of UNESCO, Paris.
Villacorte, L. O., Ekowati, Y., Winters, H., Amy, G. L., Schippers, J. C., & Kennedy, M. D. (2015a).
MF/UF rejection and fouling potential of algal organic matter from bloom-forming marine and
freshwater algae. Desalination, 367, 1-10.
Villacorte, L. O., Ekowati, Y., Calix-Ponce, H. N., Schippers, J. C., Amy, G. L., & Kennedy, M. D. (2015b).
Improved method for measuring transparent exopolymer particles (TEP) and their precursors in
fresh and saline water. Water Research, 70, 300-312.
Villacorte, L. O., Ekowati, Y., Neu, T. R., Kleijn, J. M., Winters, H., Amy, G. L., Schippers, J. C., &
Kennedy, M. D. (2015c). Characterisation of algal organic matter produced by bloom-forming
marine and freshwater algae. Water Research, 73, 216-230.
Villacorte, L. O., Ekowati, Y., Winters, H., Amy, G. L., Schippers, J. C., & Kennedy, M. D. (2013).
Characterisation of transparent exopolymer particles (TEP) produced during algal bloom: a
membrane treatment perspective. Desalination and Water Treatment, 51 (4-6), 1021-1033.
Villacorte, L. O., Schurer, R., Kennedy, M. D., Amy, G. L., & Schippers, J. C. (2010a). Removal and
deposition of Transparent Exopolymer Particles (TEP) in seawater UF-RO system. IDA Journal,
2, 45-55.
Villacorte, L. O., Schurer, R., Kennedy, M. D., Amy, G. L., & Schippers, J. C. (2010b). The fate of
transparent exopolymer particles in integrated membrane systems: a pilot plant study in
Zeeland, the Netherlands. Desalination and Water Treatment, 13, 109-119.
Villacorte, L. O., Kennedy, M. D., Amy, G. L., & Schippers, J. C. (2009a). Measuring Transparent
Exopolymer Particles (TEP) as indicator of the (bio)fouling potential of RO feed water.
Desalination and Water Treatment, 5, 207-212.
Villacorte, L. O., Kennedy, M. D., Amy, G. L., & Schippers, J. C. (2009b). The fate of Transparent
Exopolymer Particles (TEP) in integrated membrane systems: Removal through pretreatment
processes and deposition on reverse osmosis membranes. Water Research, 43 (20), 5039-5052.
Zhang, Z., Zhang, T., Wang, L., Chen, M., Zhao, B., Li, J., Ma, C., Chu, X., & Zhang P. (2023). Irreversible
membrane fouling caused by free TEP: Mitigation performance and mechanism of the integrated
MIEX/UF process. Journal of Water Process Engineering, 54, 103919.
Zhou, J., Mopper, K., & Passow, U. (1998). The role of surface-active carbohydrates in the formation
of transparent exopolymer particles by bubble adsorption of seawater. Limnology and
Oceanography, 43(8), 1860-1871.

312
 Part 5
Biological fouling
 doi: 10.2166/9781789062977_0315

Chapter 14

Genomics Tools to Study

Membrane-Based Systems
Muhammad Ali, The University of Dublin, Ireland

Lucia Ruiz Haddad, KAUST, Saudi Arabia

Mohamed Fauzi Haroon, Moderna, USA

Pascal E. Saikaly, KAUST, Saudi Arabia

The learning objectives of this chapter are the following:

• Designing an experimental plan for omics-based study for membrane-based

systems

• Omics tools to investigate microbial communities of membrane-based systems

• Introduction to state-of-the-art bioinformatics tools and reference database

resources

• Application of genome-resolved metatranscriptomics approach to study the activity

of microorganisms

14.1 INTRODUCTION

The performance of membrane-based operations is significantly decreased by the

accumulation of microbes and chemical foulants on the membrane surface. Thus,
biofouling is the main operational challenge for membrane-based processes (Flemming
et al., 2011). Various molecular biological techniques are utilized to understand the
behaviour of microbes in membrane-based water treatment systems. These techniques
enable the analysis of microbes’ taxonomy, morphology, physiology, and ecology. The
molecular methods can be applied to study membrane-based systems for fast, reliable and
cheap identification of relevant microorganisms. The most common molecular methods

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

include polymerase chain reaction (PCR), quantitative polymerase chain reaction (qPCR),
fluorescence in situ hybridization (FISH), denaturing gradient gel electrophoresis (DGGE),
and terminal restriction fragment length polymorphism (T-RFLP) (Lovley, 2003; Nielsen
et al., 2009). However, these techniques fail to address the function of microorganisms and
the mechanisms underlying their interactions due to limited throughput as compared to
their high-throughput sequencing-based counterparts and cannot resolve abundance and
activity for functionally important microbes in low concentrations (Hugenholtz, 2002).
Now, the cost of culture-independent next-generation sequencing (NGS) technologies
rapidly decreases (Figure 1), significantly improving our understanding and functions in
various natural (Alqahtani et al., 2019; Bougouffa et al., 2013; Garcias-Bonet et al., 2018;
Speth et al., 2017; Wang et al., 2013) and engineered (Ali et al., 2020b; Ali et al., 2019;
Matar et al., 2021; Rehman et al., 2019) ecosystems.

Most environments harbour a diverse range of organisms (genomes). The metagenome is the
entire DNA content of an environment, so it includes the collection of individual genomes.
Metagenomes allow to 1) Obtain a gene catalogue of the environment; 2) Screen for genes
that confer novel functions (Ali et al., 2020a; Garcia Martin et al., 2006); 3) Discover and
characterize new bacterial candidate divisions, and 4) Reconstruct microbial genomes via
metagenome assembly and binning (Albertsen et al., 2013), offering novel insights into
the microbial functions and metabolic pathways involved in complex microbial systems.
On the other hand, metatranscriptomics is the entire RNA content of a given environment.
High-throughput sequencing can be applied for metagenomics or metatranscriptomics,
where all the DNA or expressed genes (mRNA) from a certain community are sequenced to
resolve abundance and activity for functionally important microbes in low concentrations.
Metatranscriptomics analysis presents unique challenges due to RNA’s short half-life
and the variability in sequencing coverage caused by the molecule’s secondary structure.
Unlike metagenomes, metatranscriptomes can shed light on the activity of environmental
populations. As a result, metagenomes and metatranscriptomes from the same samples can
be analyzed together to identify ‘active’ microorganisms and their gene expression patterns
caused by the metabolic activities of the microbial community (Ali et al., 2020b; Shaw et
al., 2020).

Metagenomics, metatranscriptomics, and next-generation sequencing (NGS) revolutionize

microbiology and ecology research. However, the most challenging aspect, particularly for
newcomers, is the bioinformatics analysis of the massive sequencing data (Desai et al., 2012).
Therefore, this chapter covers high-performance bioinformatics computing methods,
tools, and pipelines. Besides, the experimental design and sample preparations are critical
preliminary steps in determining the dependability, comparability, and (cost-)effectiveness
of the sequence data and analytical result (Desai et al., 2012; Goodrich et al., 2014; Knight
et al., 2012; Kunin et al., 2008; Thomas et al., 2012). For example, several factors can cause
bias in sequence data or render all experimental efforts less significant (Kunin et al., 2008;
Scholz et al., 2012). As a result, this chapter will provide a comprehensive technical guide
for using metagenomics in the study of microbiology and ecology of membrane-based
systems (Figure 2).

316
 Chapter 14

a)

b) 4,500

4,307
4,000

3,500

3,414
3,000

2,633
2,500

2,284
2,000

1,894
1,646
1,500

1,432
938
774

1,126
1,000
595
412
356
286

500
180
101
76
30
14
7
2

Publications 0
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
5,500
5,292

5,000

4,500

4,000
3,656

3,500

3,000

2,500
2,608

2,000

1,500
1,217

1,000
884

0.015
0.385
0.155
0.073
0.061
0.054

0.015

0.015
0.009
0.030

0.007
0.007

500
642

1.23
479
32

Average costs ($/Mb) 0

2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022

Year

Figure 1 A) Timeline for sequencing technology development. The data is adopted from this link
https://figshare.com/articles/dataset/developments_inNGS/100940?file=5470844.
B) Cost of per raw megabase of DNA Sequencing. https://www.genome.gov/about-
genomics/fact-sheets/DNA-Sequencing-Costs-Data and Number of publications using
metagenomics tools. Data obtained from PubMed: metagenom*

317
Experimental Methods for Membrane Applications

This chapter presents the relevant concepts and protocols required for the experimental
design, sample preparation, library construction, and NGS sequencing. These steps
are crucial for metagenomics analysis of microbial communities in membrane-based
bioengineered systems operated for research or industrial aims (e.g., bioenergy production
and bioremediation). Additionally, the methods for annotating metagenomic and
metatranscriptomics data are compared, and their conditions of applicability are discussed.
Likewise, the chapter provides an overview of the most cited, user-friendly, rigorous, and
complete bioinformatics analysis tools and reference database resources, along with some
analysis examples. These metagenomic data mining strategies and resources will make it
simpler for users to select the best resources to meet their needs. Finally, the significant
pitfalls and limitations of applying metagenomics are discussed.

1 1 1 1 1 1 1 1

Membrane Sampling DNA Sequencing Data Assembly Genome Genome

fouling isolation pretreatment binnig annotation

Figure 2 Step-by-step workflow for metagenomics experiment. ‘Created with BioRender.com’

14.2 EXPERIMENTAL DESIGN AND SAMPLE PREPARATION

14.2.1 Experimental Design in a Metagenomics

A rigorous design of an experimental plan is vital for acquiring data to answer the scientific
question of interest in membrane-based systems. The following key aspects should be
considered when designing omics experiments. First, an adequate number of technical
and biological replicates of experimental samples is necessary for statistically meaningful
analyses. Technical replicates are repeated measurements of the same sample, demonstrating
the protocol’s variability. Therefore, technical replicates measure the reproducibility of
the data generated using identical NGS protocols (e.g., library construction and sample
multiplexing) and molecular methods (e.g., DNA extraction and PCR amplification).
Besides, biological replicates are parallel measurements of biologically distinct samples (at
least three) within one experimental group (e.g., the same treatment or condition). This
captures the random biological variation within a group, which may be a subject of study
or a noise source. Understanding biological variation allows an accurate assessment of the
effectiveness and differences between different experimental groups. Besides replicates,
positive and negative controls may be required to ensure no contamination occurs among
samples. Many metagenomics-based studies failed to address this issue adequately. Now
that the sequencing costs have been significantly reduced, the economic constraints should
not be used to compromise experimental rigour.

318
 Chapter 14

Moreover, decisions must be made at different experimental stages, including selecting the
following: physiochemical parameters, biological parameters, sequencing platform, sample
preservation, quality control, bioinformatics tools, and reference databases. Physiochemical
parameters include temperature, pH, conductivity, alkalinity, turbidity, dissolved oxygen,
metals, nutrients, etc. Biological parameters include the sample size and the scales (temporal
or spatial). The sequencing platform will be determined by the desired depth, sequence
length, and error rate (Fig. 1A). While sampling preparation includes methods for fixing
or storing samples on-site (RNA latter or freezing). Finally, quality control allows for
increased amount and purity of DNA or RNA, especially when yield is low or microbial
diversity is high. This could be accomplished through amplification, assembly, and binning
steps. On the other hand, compiling metadata is an equally important step in the process.
Correlating reactor performance with the structure and functions of microbial communities
requires synchronous preservation of biological samples and regular monitoring of reactor
performance (removal rates, production rate, biodegradation, and accumulation) in different
operating conditions.

14.2.2 Sample Collection and Preservation

Sample collection and preservation are the first critical steps in an omics-based study. For
sampling from a membrane-based system, the following factors should be considered:
biomass type (e.g., an attached vs. suspended biomass), bioreactor configuration (e.g., plug
flow vs. mixed), modification (e.g., presence of carriers or baffles), mixing state and operating
time and condition. These factors affect biomass distribution and evenness, granule size,
substrate gradients, and mass transfer, leading to a heterogeneous microbial composition
within a membrane-based system. In certain membrane systems, when extracting DNA
material, it may be necessary to obtain a sample by cutting a piece of the membrane, typically
measuring at least 1 – 5 cm2. Multiple samples can be collected separately to reflect microbial
heterogeneity, or a composite sample can be taken to represent average microbial profiles.
DNA extraction of sludge/slurry samples is preserved at −80 °C (Mason et al., 2012) or fixed
in 50% ethanol before storage at −20 °C (Zhang et al., 2012). In contrast, centrifugation and
membrane filtration are viable methods for recovering microbial cells in samples containing
less biomass such as effluent samples (Thomas et al., 2012). For mRNA-based studies, two
preservation methods are frequently used: 1) Flash freezing with liquid nitrogen and 2)
Immersion in RNAlater, which is widely used to recover high-quality mRNA (Riesgo et al.,
2012).

14.2.3 DNA Extraction

An optimized protocol for DNA/RNA extraction is essential to any analysis of microbial
composition and function. The microbes differ enormously in their resistance to different
lysing methods (Thomas et al., 2012). Hence, microbes with cell walls that are difficult to
lyse will effectively seem less abundant if sub-optimal extraction protocols are used. Thus,
the method for DNA/RNA extraction needs to be robust to cope with the challenges.

The protocols and kits used for DNA/RNA extraction should be consistent throughout
a study to increase the reliability and comparability of data (Goodrich et al., 2014). These
kits should be used following the manufacturer’s instructions. Likewise, in 2012 Thomas
and co-workers established a guide for isolating high-quality nucleic acid for NGS library

319
Experimental Methods for Membrane Applications

preparation and sequencing. Following the extraction of DNA and RNA, their quality must
be determined using various quality indicators. Generally, the ratio of absorbance at 260 and
280 nm is used to assess the purity of DNA and RNA. A ratio of 1.8 is commonly accepted
as pure DNA, and a value of 2.0 is recognized as pure RNA. (Lucena-Aguilar et al., 2016).
At the same time, lower ratios indicate the presence of non-nucleic acid content, such as
protein, phenol, or other contaminants that absorb strongly at or near 280 nm. Besides, the
absorbance ratio of 260/230 can be used as a secondary measure of nucleic acid purity. The
typical values for nucleic acid samples are between 2.0-2.2 (Lucena-Aguilar et al., 2016). If
the ratio is lower, it may indicate the presence of contaminants that absorb at 230 nm. These
absorbance parameters should be checked with spectrophotometers or qubit.

14.2.4 Library Preparation

Library construction transforms the RNA/DNA into a format compatible with the
sequencing platforms. Various physical, enzymatic, and chemical methods can be used for
fragmentation and size selection required for library preparation on the different sequencing
platforms (see Fig. 1A for desired read length). The library preparation strategy, the sequence
length, and the sequencing platforms’ length and depth are all essential considerations for
sequencing. Moreover, technical details on sequencing library construction have been well
documented (Head et al., 2014). However, library preparation generally shares similar
principles and considerations regardless of the sequencing platform. For example, two
objectives are to maximize library complexity (e.g., a lower ratio of artificial duplicate reads
may indicate higher complexity) and to minimize PCR or other amplification-based biases
(e.g., less amplification and more sample RNA/DNA). Therefore, a paired-end (PE) library
is recommended over a single-end library. A PE improves the performance of metagenome
(e.g., scaffolding and chimera detection) (Peng et al., 2012), binning (e.g., tracking multiple-
copy genes) (Albertsen et al., 2013), and enables the use of computational tools designed to
consider PE relationships (Imelfort et al., 2014).

14.2.5 Sequencing platforms

The expected sequence depth is related to the biodiversity and complexity of microbial
samples. Generally, the soil (Roesch et al., 2007) and sediments (Roesch et al., 2007)
harbor more diverse microbial species than the bioengineered ecosystems (Dueholm et al.,
2022; Wang et al., 2012a; Wu et al., 2019). For biological wastewater treatment systems,
higher biodiversity is generally detected in 1) full-scale than lab-scale bioreactors (Matar
et al., 2021), 2) biofilm than suspended sludge and 3) activated sludge than anaerobic
sludge (Ali et al., 2019; Trego et al., 2020). Recent attempts to assemble large complex soil
metagenomes suggest that 80% of the sequencing data could not be assembled due to the
low coverage. For instance, even 300 Gb sequencing data are insufficient to deeply cover a
localized soil sample (Howe et al., 2014). In contrast, in an enriched microbial system, more
than 45% of the metagenomics reads could be effectively assembled (Albertsen et al., 2013).

The following equation can calculate the required sequencing depth for a given genome
coverage in metagenomic samples.

Sequencing Depth = (Genome Size • Coverage × 100)/(Relative Abundance)

320
 Chapter 14

Here,
Genome Size = Bacterial genome length (usually between 2.5- 5 Mbp)
Coverage = Required depth coverage (between 5 to 10×)
Relative Abundance = Expected relative abundance of targeted bacteria species. The
relative abundance value might be estimated from previous studies on similar ecosystems.

For example, a sample with a relative abundance of 0.5% for the target bacteria species and
a genome size of 3Mbp will require 6 Gb of sequencing depth to achieve 10× coverage.
Furthermore, a sequencing depth of 6Gb and a PE read length of 150bp will produce 300bp
(150bp × 2) per reading, 20 million of paired reads((6Gb)⁄(300bp)), and 40 million of
single reads((6Gb)⁄(150bp)). Illumina has recently been the most popular commercial
sequencing platform due to its high data throughput and low per-base cost (van Dijk et
al., 2014). Notably, the short but high-quality PE sequences (100−150 bps) generated by
Illumina’s platforms could be used to rectify (> 99.9% accuracy) (Koren et al., 2012) and
concatenate the low-quality but exceptionally long reads (e.g., 20 kb) from third-generation
sequencing platforms, (van Dijk et al., 2014) such as PacBio RS and nanopore technologies.

14.3 BIOINFORMATICS ANALYSIS

Bioinformatics studies biological data in conjunction with computer science and statistics.
This field is constantly evolving and developing new tools to make metagenomics/
metatranscriptomics analysis and genetic engineering simpler and more reliable. The main
steps required for bioinformatics analysis are listed below

14.3.1 Data Pre-treatment

Raw sequencing reads must be pre-treated to ensure that only high-quality sequences (clean
reads) are used for the downstream analysis. Before pre-treatment, FASTQC can be used
to check the data quality (base quality, GC content, ambiguous bases, length distribution,
sequence duplication levels, and adapter content). In general, data pre-treatment includes
the following steps: adaptors/linkers removal, demultiplexing (assign reads to samples
using index reads or barcodes), quality control, dereplication (identifying unique sequences
and abundances), and reads overlapping (for PE library sequencing) etc.

14.3.2 Amplicon-based approach

Amplicon-based sequencing is a useful tool for understanding microbial community
composition and diversity. It is a cost-effective and computationally simple method (Bodilis
et al., 2012). However, an amplicon-based analysis could not always classify microbes to
a lower taxonomic level, such as genus or species. Also, generally, it targets hypervariable
region(s) of the 16S rRNA gene, which does not provide the functional identity of the
microbe (Siegwald et al., 2017). Though, efforts are made to manually curate the existing
16S rRNA gene database (SILVA, (Pruesse et al., 2012)) to provide information about
the physiology (function) and ecology of the microorganisms present in bioengineered
ecosystems (Albertsen et al., 2015; Dueholm et al., 2022). For amplicon sequencing data
processing, there are numerous open-source packages available, such as QIIME (Caporaso et
al., 2010), USEARCH (Edgar, 2013), MOTHUR (Schloss et al., 2009), etc.

321
Experimental Methods for Membrane Applications

14.3.3 Metagenomics, read-based approach

Bioinformatics analyses of metagenomic data can be performed using short-reads, assembled
contigs (assembled overlapping reads), or reconstructed draft genomes through a self-
accelerating data mining circle. The first strategy for metagenomic analysis is to directly use
unassembled clean reads for the quantitative analysis of microbial community composition
and function. The assembly-free approach is advantageous when studying rare organisms
(with low sequencing depth or coverage) to avoid bias, such as their removal due to their
inability to be assembled. In an assembly-free (read-based) approach, reads are directly
compared against a reference database for taxonomic profiling and functional analyses.
Generally, unassembled short reads retain the original abundance information and enable
quantitative comparisons of microbial taxa, functional genes, and metabolic profiles (Yang
et al., 2013). However, they have a large amount of data and lack resolution for taxonomic
and functional annotations.

14.3.4 Metagenomics, assembly-based approach

The assembly is a computational process that connects the clean short-reads sequences to
form long contigs (>1,000 bp). This is especially relevant to 1) Recover genomic sequences
(via binning as discussed in sections 15.3.5 and 15.3.6; 2) Analyze full-length protein-
coding genes; 3) Identify strain-specific genomes (Langille et al., 2010); 4) Analyze the
genetic content (e.g., at the strain or species level), especially for uncultured microorganisms.
Moreover, a metagenomic assembly reduces the data size to be analyzed in the downstream
processing, though assembly requires substantial computational resources (Howe et al.,
2014). Two assembly strategies are used depending on whether a reference database is
used: reference-based and de novo. Reference-based assembly aligns short-read fragments
with reference genomes. However, this analysis cannot capture the differences between the
genomes of novel species, resulting in an underestimation of microbial diversity in an open
microbial system. (e.g., activated sludge, soil) (Howe et al., 2014). As a result, a de novo
strategy should be used when genetic novelty and diversity are high.

14.3.5 Metagenome-assembled Genome (MAG) Binning

The continuously decreasing sequencing cost has allowed researchers to access
environmental metagenomes at increasing sequencing depths (e.g., > 50 Gbp). As a result,
sufficient resolution can be obtained to retrieve partial or near-complete genomes of rare
(1%), novel, and/or uncultured microorganisms from complex communities (Albertsen
et al., 2013). Binning is the computational process of clustering the assembled contigs or
the short reads into groups representing an individual genome/taxon or genomes/taxa of
closely related microorganisms (Albertsen et al., 2013; Wrighton et al., 2012). Binning can
be based on genomic signatures such as 1) sequence composition, 2) homology, 3) coverage
(abundance), or 4) a combination of these. Composition-based binning algorithms typically
group DNA sequences based on their conserved nucleotide compositions, such as the
tetranucleotide frequencies and GC content (McHardy et al., 2007). These methods can be
improved by providing sample-specific training data sets (e.g., a long DNA fragment with
marker genes). The homology-based binning methods are based on a similarity search against
existing genomes. However, this method is limited by the quality and representativeness of
reference databases, the poor taxonomic resolution of short reads, and the accuracy and/or
sensitivity of alignment tools. As a result, they are unreliable for assigning short reads. They

322
 Chapter 14

often require longer assembled contigs (e.g., > 1 kb for the expert-trained PhyloPythiaS
package) and manual efforts to ensure high assignment accuracies. Homology-based tools
include MetaPhlAn2, MetaPhyler, and CARMA3. These are commonly applied for the
taxonomic classification of shotgun metagenomic reads based on similarity comparisons
with reference marker genes such as 16S rRNA, rpoB, or clade-specific markers. Besides,
some packages consider the composition and homology of sequences for taxonomic
classification and clustering, such as MEGAHIT (Li et al., 2016), Metabat (Kang et al., 2015)
and MetaCluster (Wang et al., 2012b).

14.3.6 Supervised and unsupervised binning

Two unsupervised approaches have been widely applied to reconstruct high-quality
genomes of uncultured organisms from metagenomes: Tetra-ESOM (Dick et al., 2009)
and ‘differential coverage binning’ (Albertsen et al., 2013). Tetra-ESOM analyzes the
composition of DNA signatures by clustering tetranucleotide frequencies using emergent
self-organizing maps (ESOM). While ‘differential coverage binning’ categorizes contigs
based on their differential coverage profiles across multiple related metagenomes with
the assumption that contigs from the same microorganisms will have similar abundance
(coverage) profiles in a single metagenome. These draft genomes can be refined further
using composition-based paired-end tracking, reassembly, and manual curation techniques.
Several automated pipelines for platforms like CONCOCT (Alneberg et al., 2014), MaxBin
(Wu et al., 2016), and Metabat (Kang et al., 2015) allow genome reconstruction based on the
coverage profiles and composition (tetranucleotide frequency patterns). The completeness
and contamination in reconstructed genomes have been estimated by the presence/absence
of marker genes, such as essential single copy marker genes conserved in 95% of bacteria
(Dupont et al., 2012), conserved phylogenetic marker genes (Wrighton et al., 2012), or
clusters of orthologous groups (COGs) (Raes et al., 2007). Currently, CheckM is the only
automated tool to assess the quality of a genome recovered from isolates, single cells, and
metagenomes based on conserved marker genes (Parks et al., 2015).

14.3.7 Functional annotation

Annotation is the process of analysing the structure and functions of assembled
metagenomic contigs. Compared to unassembled short reads, assembled contigs are longer
and more compact, allowing for a more robust and rapid analysis of specific species and their
functional genes. Functional annotation of assembled contigs allows the prediction of the
functional capacities of a microbial community. After genome assembly, binning, and gene
calling are finished, several tools enable functional annotations. Gene function is commonly
determined through similarity searches while using established tools like BLAST. However,
it is computationally expensive and time-consuming to conduct a similarity search using
BLASTX or PSI-BLAST, particularly for large query data sets and reference databases (such
NCBI’s RefSeq). Tools such as USEARCH (Edgar, 2010) and DIAMOND (Buchfink et al.,
2015) have been developed to overcome these computational challenges faster.

14.3.8 Genome-resolved Metatranscriptomics

Genome-resolved metatranscriptomics approach combines data from metagenomics and
metatranscriptomics from the same environment to characterize ‘active’ and ‘non-active’
microorganisms and to compare gene expression patterns under different conditions such
as treatment v/s control (Frias-Lopez et al., 2008; Haroon et al., 2013; Yu and Zhang,

323
Experimental Methods for Membrane Applications

2012). MAGs recovered from metagenomics are used to facilitate genome-resolved

metatranscriptomics analyses for profiling the gene expression across the recovered
population genomes (Ali et al., 2020b).

14.4 DATA SHARING AND STORAGE

Sharing sample metadata, sequence data, and computational results is a widespread and
effective method for exchanging knowledge. Data exchange enables comparative studies
and avoids the needless repetition of processing the same data sets or sequencing similar
microbial ecosystems. Several public access databases have been established to promote
sequencing data sharing and storage, such as the National Center for Biotechnology
Information (NCBI) and Genomes OnLine Database (GOLD).

14.5 BIOINFORMATICS ANALYSIS WORKFLOW EXAMPLES

The following sections describe the simple bioinformatics workflow examples for process
amplicon, metagenomics and metatranscriptomics datasets.

14.5.1 Amplicon Sequences Processing Workflow

The amplicon sequences analysis workflow is shown in Fig. 3A. The first step of the
process is to re-label the sequences in the raw FASTQ files with the corresponding sample
name/ID. This enables the identification of the sequences from a specific sample in the
later processing stages. The re-labelling can be performed using USEARCH (Edgar, 2013)
with the command fastx_relabel. Subsequently, re-labelled forward and reserve sequences
of different samples are concatenated separately into a single file. Then, the concatenated
forward and reserve sequences are quality-filtered using trimomatic (Bolger et al., 2014)
with the settings SLIDINGWINDOW:5:3 and MINLEN:275, these settings should be
adjusted based on the raw quality of the dataset determined through FASTQC package. The
quality filtered sequences are hits to the PhiX genome USEARCH command filter_phix.
PhiX genome-based library is an ideal sequencing control (typically with > 1% spike-in)
for run quality monitoring (cluster generation, sequencing, and alignment). Later, the
trimmed and filtered forward and reverse reads were merged using FLASH (Fast Length
Adjustment of Short Reads) (Magoc and Salzberg, 2011) with the settings –m 25 –M 200,
these parameters should be adjusted based on the input data quality. FLASH merges paired-
end reads to create consensus sequences. Following that, the merged reads are compared
letter by letter to the set of unique sequences, a process known as dereplication, and
carried out with the USEARCH command fastx uniques. Then, the cluster_otus command
performs 97% Operational Taxonomic Unit (OTU) clustering of unique sequences using
the UPARSE-OTU algorithm (Edgar, 2013). Taxonomy is assigned to OTUs using sintax
command available in USEARCH package and MiDAS database (Dueholm et al., 2022).
OTU abundances table is generated by mapping merged reads against the OTUs using otutab
command available in USEARCH package. The OTU count and corresponding taxonomy
table are imported into the Rstudio IDE environment using the ampvis2 package by the
amp_load command (Andersen et al., 2018). The ordination plot (Fig 3B) and heatmap (Fig.
3C) were generated using ampvis2 package. All the scripts, metadata and exercise files are
available at this link (https://drive.google.com/drive/folders/1swJIv9Z1pyD52Jo630d-
ivBTOhxt7dHu?usp=share_link).

324
 Chapter 14

14.5.2 Genome-resolved Metagenomics

In general, metagenomics is used to extract the genomes of a microbial population in the
sample. The recovered MAGs from metagenomics analysis facilitate genome-resolved
metatranscriptomics analyses and gene expression profiling across the recovered
population genomes. Genome-resolved metagenomics workflow is presented in Fig. 4. In
this exercise, two independent biomass samples for metagenomics were collected from the
MBR in a previous study (Ali et al., 2020b). Raw reads are processed for quality filtering
using Cutadapt (Martin, 2011). Trimmed forward and reserve sequences of different
samples are concatenated separately into a single file. Concatenated forward and reserve
sequences were assembled using MEGAHIT (Li et al., 2016). The assembled contigs were
reformatted to simplify deflines and to remove short contigs (<1000 bp) using anvi’o
command anvi-script-reformat-fasta (Eren et al., 2021). The filtered reads were mapped
back to the reformatted assembly file using Burrows-Wheeler Aligner (Li and Durbin,
2010) to generate coverage files for metagenomics binning. These files were converted to
the sequence alignment/map (SAM) format using samtools (Li and Durbin, 2009). The
SAM files are converted into a sorted and indexed BAM-file using samtools. Later, anvi’o
uses sorted and indexed BAM-file in anvi-profile command to create a single profile that
reports properties for each contig in a single sample based on mapping results. These
profiles are merged into one anvi’o profile using anvi’o command anvi-merge. MAGs were
obtained from assembled scaffolds by binning based on sequence composition, differential
coverage, and read-pair linkage using metabat2 program (Kang et al., 2019). Generated
MAGs collections are imported into the merged profile database using anvi’o command
anvi-import-collection. The command anvi-interactive provides an interactive interface
that allows to visualize the results of unsupervised binning, perform supervised binning,
or refine existing bins. The MAGs are manually refined if needed by the command anvi-
refine provided in anvi’o package. Once the binning collection is ready, the anvi-summarize
command provides a summary. The obtained summary, as shown in Table 1, includes
details about the MAG completion as well as statistics like mean coverage, variability, etc.
Subsequently, MAGs are functionally annotated using Prokka (Seemann, 2014). Finally,
taxonomic classifications are assigned to MAGs based on the Genome Database Taxonomy
(GTDB) using the GTDB-Tk program (Chaumeil et al., 2022). Relative abundance values of
metagenomics read that mapped to each MAG generated from ‘bins_percent_recruitment.
txt’. All the scripts and exercise files are available at this link (https://drive.google.com/
drive/folders/1DQVaqD2VpSx6YbZUjOMHI97XcNP4Go_I?usp=share_link).

Table 1 Files obtained from the anvi-summarize command

File Name Description
bins_summary.txt Basic statistics of the recovered MAGs.
bins_across_samples/bins_percent_recruitment.txt Coverage and detection of the MAGs.
bins_across_samples/mean_coverage.txt The average number of sequencing reads that map to
each MAG.
bin_by_bin A folder with all the MAGs extracted from the
metagenomics processing.

325
Experimental Methods for Membrane Applications

Figure 3 A) 16S rRNA gene Amplicon Sequencing Workflow.

B) Ordination plot of influent wastewater, flocs, small granules, and large granules based
on Principal Components Analysis (PCA). Hellinger transformation is used to perform
PCA as it will produce a more ecologically meaningful result.
C) Heatmap distribution of the dominant OTUs classified down to the genus level (f,
o, c, and p represent family, order, class, and phylum, respectively) for the influent
wastewater, flocs, small granules, and large granules. Figures in Panel B and C are
adapted with permission from (Ali et al. Environmental Science & Technology, 2019,
53, 8291−8301). Copyright (2019) American Chemical Society.

326
 Chapter 14

Figure 4 Genome-resolved Metagenomics Workflow

14.5.3 Genome-resolved metatranscriptomics

In this exercise, metatranscriptomics samples are collected from the same MBR
where metagenomics samples were collected (Ali et al., 2020b). Genome-resolved
metatranscriptomics workflow is presented in Fig. 5A. The reads are mapped to the
predicted protein-coding genes generated from Prokka for each MAG using USEARCH
command usearch_global. Reads with a sequence identity below 0.98 were discarded. The
count tables are used to generate gene expression of microorganisms represented by the
MAGs recovered from the MBR. The gene expression is based on transcription per million
(TPM) metatranscriptomics reads that are mapped to each MAG (Fig. 5B&C). TPM values
are calculated by dividing the read counts by the MAG size (kilobases), which gives reads
per kilobase (RPK) value. The RPK values in a sample are then added, and the resultant value
is divided by 1,000,000 to obtain a ‘per million’ scaling factor. Finally, the individual RPK
values were divided by the ‘per million’ scaling factor, yielding the TPM value. All the scripts
and exercise files are available at this link (https://drive.google.com/drive/folders/1I_
wbxtSBsYaL_OIy1wPlWqnkGIhin4Zg?usp=share_link). For further comparative analysis,
the count tables can be imported to RStudio, processed using the default DESeq2 workflow
(Love et al., 2014), and visualized using ggplot2. Principle component analysis (PCA) and
hierarchical clustering analysis (HCA) of overall sample similarity can be performed using
DESeq2 normalized count in the R platform.

327
Experimental Methods for Membrane Applications

a)

1,000,000
b)

750,000

500,000

250,000

Metagenomics
reads (million) 0
AMX
PNC
BCD1
CLD1
GFX1
CLD2
CLD3
BCD2
GFX2
PRO1
PRO2
CLD4
ATM
CLD5
PRO3
PRO4
BCD3
1,000,000
c)

750,000

500,000

Control
250,000
Treatment
Metagenomics
reads (million) 0
AMX
PNC
BCD1
CLD1
GFX1
CLD2
CLD3
BCD2
GFX2
PRO1
PRO2
CLD4
ATM
CLD5
PRO3
PRO4

Mag ID BCD3

Figure 5 A) Genome-resolved Metatranscriptomics Workflow.

B) Relative abundance and
C) gene expression based on transcription per million (TPM) values of metagenomics
and metatranscriptomics reads that mapped to each MAG recovered from the MBR.
Figures in Panel B and C are adapted with permission from (Ali et al. Water Research,
2020, 170, 115345). Copyright (2019) Elsevier Ltd.

328
 Chapter 14

14.6 APPLICATIONS OF GENOMICS IN MEMBRANE FILTRATION RESEARCH

The use of 16S rRNA gene amplicon sequencing has been valuable for understanding
the diversity and abundance of microorganisms in membrane-based water systems
(Figure 3C) (Vries, 2022). For instance, it was used to create a microbiological database
of bacteria detected in biofilms on rubber-coated valves in drinking water systems and
to identify the specific bacterial phylum present in contaminated reverse osmosis (RO)
membranes (Bereschenko, 2010; Beyer, 2019; Vries, 2022). Furthermore, by obtaining a
thorough understanding of the bacterial communities within membranes, scientists have
gained valuable insights regarding the types of microorganisms that are prone to causing
fouling. These insights have been instrumental in refining membrane-cleaning strategies,
specifically by targeting particular microorganisms or proactively preventing the formation
of bacterial biofilms (Møllebjerg et al., 2023). Additionally, 16S rRNA and other omics
data can be used for ordination analysis to reveal similarities and clustering patterns among
samples in membrane processes (Figure 3B). The findings from such studies can provide
valuable information about the microbial ecology within membranes, helping researchers
to gain insights into the microbial communities present in membranes and understand how
they relate to one another.

Genome-centric metagenomics processing provides a more comprehensive and refined

approach than 16S rRNA analysis. For example, MAGs enabled the identification of
microorganisms at the species level (Rehman et al., 2019). In Figure 6, two main aspects of
genome-centric metagenomics analysis are presented. Firstly, it enables the construction of
phylogenetic trees (Figure 6A). This allows us to understand the evolutionary relationships
between different organisms and their role in the ecosystem. Secondly, genome-centric
metagenomics processing allows the reconstruction of metabolic pathways using
metagenome-assembled genomes (MAGs) based on gene presence or absence (Figure 6B).
By analyzing the genetic content of MAGs, we can identify the presence or absence of specific
genes involved in metabolic pathways. This information helps us understand the functional
potential of microbial communities and their contributions to various biogeochemical
processes in membrane-based systems. Overall, genome-centric metagenomics processing
is a powerful tool that provides insights into the taxonomic and functional composition of
microbial communities. It helps us understand the complex interactions between different
organisms and their roles in ecosystem functioning. It also enables the discovery of genes
that are enriched in biofilm development as well as the investigation of specific genes
within microbial communities. These insights have played a crucial role in developing
effective strategies to prevent biofouling. For example, antifouling strategies for seawater
reverse osmosis (RO) membranes involve targeting the enriched Planctomycetes bacteria
or employing inhibitory compounds such as azide, chlorate, cyanide, and thiocyanide to
specifically target nitrate-reducing enzymes, which have been identified as enriched in the
biofilm metagenome (Rehman et al., 2019).

329
Experimental Methods for Membrane Applications

a) Phylogenetic tree b) Metabolic pathways modelling

Figure 6 Genome-centric metagenomics processing enables the construction of A) Phylogenetic

trees and B) Reconstruction of metabolic pathways using metagenome-assembled
genomes (MAGs) based on gene presence or absence. Figures in Panel A is adapted with
permission from (Ali et al. Water Research, 2020, 170, 115345). Copyright (2019)
Elsevier Ltd. Figures in Panel B is adapted with permission from (Ali et al. Frontiers in
Microbiology, 2020, 11, 1637.).

14.7 OUTLOOK

Recently, metagenomics has been successfully employed to discover novel microorganisms.

However, the following points should be considered in future omics-based studies,
including membrane-based filtration systems.

1. The omics-based study reports a relative abundance of microbial species as an actual

abundance in the system. However, the relative abundance derived from such analysis
should not be considered or interpreted as the system’s actual abundance.
2. Assembled genomes (MAGs) have clear advantages for further functional analyses.
However, obtaining correct assemblies is still challenging. The presence of genomic
repeats, short overlap lengths, and phylogenetically close organisms can result in false-
positive assembly outputs, such as joining non-overlapping fragments from different
parts of the genome or producing chimeric contigs from different organisms. Besides
the difficulty in assembling low-abundance species and closely related strains (micro
diversity), and the exclusion of significant amounts of unassembled data, an assembly-
based annotation strategy can introduce biases for quantitative analysis.
3. NGS-based sequencing is still expensive because the cost rises with sequencing depth.
Most omics-based studies use quantitative analysis due to the high cost of technical/
biological replicates. This makes it challenging to determine whether observed differences
(particularly small ones) within or between experimental groups are statistically
significant and robust (after considering replicate variance). Therefore, future research
should consider the reproducibility of omics data before delving into differentially
expressed features.

330
 Chapter 14

4. The completeness of database resources inevitably affects the taxonomic and functional
annotation of sequence data. However, the databases improve as researchers continue to
investigate microbial dark matter.

14.8 DATA AVAILABILITY

All raw sequencing data associated with amplicon sequencing analysis taken from the
previous study (Ali et al., 2019) are available at NCBI Sequence Read Archive under accession
number SRP115069. All sequencing data associated with genome-resolve metagenomics
analysis taken from a previous study (Ali et al., 2020b) is available at NCBI under BioProject
PRJNA482223.

331
Experimental Methods for Membrane Applications

14.9 REFERENCES

Albertsen, M., Hugenholtz, P., Skarshewski, A., Nielsen, K.L., Tyson, G.W. and Nielsen, P.H. 2013.
Genome sequences of rare, uncultured bacteria obtained by differential coverage binning of
multiple metagenomes. Nat Biotechnol 31(6), 533-538.
Albertsen, M., Karst, S.M., Ziegler, A.S., Kirkegaard, R.H. and Nielsen, P.H. 2015. Back to Basics--The
Influence of DNA Extraction and Primer Choice on Phylogenetic Analysis of Activated Sludge
Communities. PLoS One 10(7), e0132783.
Ali, M., Shaw, D.R. and Saikaly, P.E. 2020b. Application of an enrichment culture of the marine
anammox bacterium “Ca. Scalindua sp. AMX11” for nitrogen removal under moderate salinity
and in the presence of organic carbon. Water Res 170, 115345.
Ali, M., Shaw, D.R., Albertsen, M. and Saikaly, P.E. 2020a. Comparative Genome-Centric Analysis
of Freshwater and Marine ANAMMOX Cultures Suggests Functional Redundancy in Nitrogen
Removal Processes. Front Microbiol 11, 1637.
Ali, M., Wang, Z., Salam, K.W., Hari, A.R., Pronk, M., van Loosdrecht, M.C.M. and Saikaly, P.E. 2019.
Importance of Species Sorting and Immigration on the Bacterial Assembly of Different-Sized
Aggregates in a Full-Scale Aerobic Granular Sludge Plant. Environ Sci Technol 53(14), 8291-
8301.
Alneberg, J., Bjarnason, B.S., de Bruijn, I., Schirmer, M., Quick, J., Ijaz, U.Z., Lahti, L., Loman, N.J.,
Andersson, A.F. and Quince, C. 2014. Binning metagenomic contigs by coverage and
composition. Nat Methods 11(11), 1144-1146.
Alqahtani, M.F., Bajracharya, S., Katuri, K.P., Ali, M., Ragab, A., Michoud, G., Daffonchio, D. and Saikaly,
P.E. 2019. Enrichment of Marinobacter sp. and Halophilic Homoacetogens at the Biocathode
of Microbial Electrosynthesis System Inoculated With Red Sea Brine Pool. Front Microbiol 10,
2563.
Andersen, K.S., Kirkegaard, R.H., Karst, S.M. and Albertsen, M. 2018. ampvis2: an R package to analyse
and visualise 16S rRNA amplicon data. bioRxiv.
Bereschenko, L.A. (2010) Biofilm development on new and cleaned membrane surfaces, Wageningen
University.
Beyer, F. (2019) Microbiological and process technological aspects of nanofiltration and reverse osmosis
membrane biofouling, Wageningen University.
Bodilis, J., Nsigue-Meilo, S., Besaury, L. and Quillet, L. 2012. Variable copy number, intra-genomic
heterogeneities and lateral transfers of the 16S rRNA gene in Pseudomonas. PLoS One 7(4),
e35647.
Bolger, A.M., Lohse, M. and Usadel, B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence
data. Bioinformatics 30(15), 2114-2120.
Bougouffa, S., Yang, J.K., Lee, O.O., Wang, Y., Batang, Z., Al-Suwailem, A. and Qian, P.Y. 2013.
Distinctive microbial community structure in highly stratified deep-sea brine water columns.
Appl Environ Microbiol 79(11), 3425-3437.
Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Peña,
A.G., Goodrich, J.K., Gordon, J.I., Huttley, G.A., Kelley, S.T., Knights, D., Koenig, J.E., Ley, R.E.,
Lozupone, C.A., McDonald, D., Muegge, B.D., Pirrung, M., Reeder, J., Sevinsky, J.R., Turnbaugh,
P.J., Walters, W.A., Widmann, J., Yatsunenko, T., Zaneveld, J. and Knight, R. 2010. QIIME
allows analysis of high-throughput community sequencing data. Nature Methods 7, 335–336.
Chaumeil, P.A., Mussig, A.J., Hugenholtz, P. and Parks, D.H. 2022. GTDB-Tk v2: memory friendly
classification with the Genome Taxonomy Database. Bioinformatics.

332
 Chapter 14

Desai, N., Antonopoulos, D., Gilbert, J.A., Glass, E.M. and Meyer, F. 2012. From genomics to
metagenomics. Curr Opin Biotechnol 23(1), 72-76.
Dick, G.J., Andersson, A.F., Baker, B.J., Simmons, S.L., Thomas, B.C., Yelton, A.P. and Banfield, J.F.
2009. Community-wide analysis of microbial genome sequence signatures. Genome Biol 10(8),
R85.
Dueholm, M.K.D., Nierychlo, M., Andersen, K.S., Rudkjobing, V., Knutsson, S., Mi, D.A.S.G.C.,
Albertsen, M. and Nielsen, P.H. 2022. MiDAS 4: A global catalogue of full-length 16S rRNA
gene sequences and taxonomy for studies of bacterial communities in wastewater treatment
plants. Nat Commun 13(1), 1908.
Dupont, C.L., Rusch, D.B., Yooseph, S., Lombardo, M.J., Richter, R.A., Valas, R., Novotny, M., Yee-
Greenbaum, J., Selengut, J.D., Haft, D.H., Halpern, A.L., Lasken, R.S., Nealson, K., Friedman,
R. and Venter, J.C. 2012. Genomic insights to SAR86, an abundant and uncultivated marine
bacterial lineage. ISME J 6(6), 1186-1199.
Edgar, R.C. 2013. UPARSE: highly accurate OTU sequences from microbial amplicon reads. Nat
Methods 10(10), 996-998.
Eren, A.M., Kiefl, E., Shaiber, A., Veseli, I., Miller, S.E., Schechter, M.S., Fink, I., Pan, J.N., Yousef, M.,
Fogarty, E.C., Trigodet, F., Watson, A.R., Esen, O.C., Moore, R.M., Clayssen, Q., Lee, M.D.,
Kivenson, V., Graham, E.D., Merrill, B.D., Karkman, A., Blankenberg, D., Eppley, J.M., Sjodin,
A., Scott, J.J., Vazquez-Campos, X., McKay, L.J., McDaniel, E.A., Stevens, S.L.R., Anderson,
R.E., Fuessel, J., Fernandez-Guerra, A., Maignien, L., Delmont, T.O. and Willis, A.D. 2021.
Community-led, integrated, reproducible multi-omics with anvi’o. Nat Microbiol 6(1), 3-6.
Frias-Lopez, J., Shi, Y., Tyson, G.W., Coleman, M.L., Schuster, S.C., Chisholm, S.W. and Delong, E.F.
2008. Microbial community gene expression in ocean surface waters. Proc Natl Acad Sci U S A
105(10), 3805-3810.
Garcia Martin, H., Ivanova, N., Kunin, V., Warnecke, F., Barry, K.W., McHardy, A.C., Yeates, C., He,
S., Salamov, A.A., Szeto, E., Dalin, E., Putnam, N.H., Shapiro, H.J., Pangilinan, J.L., Rigoutsos, I.,
Kyrpides, N.C., Blackall, L.L., McMahon, K.D. and Hugenholtz, P. 2006. Metagenomic analysis
of two enhanced biological phosphorus removal (EBPR) sludge communities. Nat Biotechnol
24(10), 1263-1269.
Garcias-Bonet, N., Fusi, M., Ali, M., Shaw, D.R., Saikaly, P.E., Daffonchio, D. and Duarte, C.M. 2018.
High denitrification and anaerobic ammonium oxidation contributes to net nitrogen loss in a
seagrass ecosystem in the central Red Sea. Biogeosciences 15(23), 7333-7346.
Goodrich, J.K., Di Rienzi, S.C., Poole, A.C., Koren, O., Walters, W.A., Caporaso, J.G., Knight, R. and Ley,
R.E. 2014. Conducting a microbiome study. Cell 158(2), 250-262.
Haroon, M.F., Hu, S., Shi, Y., Imelfort, M., Keller, J., Hugenholtz, P., Yuan, Z. and Tyson, G.W. 2013.
Anaerobic oxidation of methane coupled to nitrate reduction in a novel archaeal lineage. Nature
500(7464), 567-570.
Head, S.R., Komori, H.K., LaMere, S.A., Whisenant, T., Van Nieuwerburgh, F., Salomon, D.R. and
Ordoukhanian, P. 2014. Library construction for next-generation sequencing: overviews and
challenges. Biotechniques 56(2), 61-64, 66, 68, passim.
Howe, A.C., Jansson, J.K., Malfatti, S.A., Tringe, S.G., Tiedje, J.M. and Brown, C.T. 2014. Tackling soil
diversity with the assembly of large, complex metagenomes. Proc Natl Acad Sci U S A 111(13),
4904-4909.
Hugenholtz, P. 2002. Exploring prokaryotic diversity in the genomic era. Genome Biology 3(2), 1-8.
Imelfort, M., Parks, D., Woodcroft, B.J., Dennis, P., Hugenholtz, P. and Tyson, G.W. 2014. GroopM:
an automated tool for the recovery of population genomes from related metagenomes. PeerJ 2,
e603.

333
Experimental Methods for Membrane Applications

Kang, D.D., Froula, J., Egan, R. and Wang, Z. 2015. MetaBAT, an efficient tool for accurately
reconstructing single genomes from complex microbial communities. PeerJ 3, e1165.
Kang, D.D., Li, F., Kirton, E., Thomas, A., Egan, R., An, H. and Wang, Z. 2019. MetaBAT 2: an adaptive
binning algorithm for robust and efficient genome reconstruction from metagenome assemblies.
PeerJ 7, e7359.
Knight, R., Jansson, J., Field, D., Fierer, N., Desai, N., Fuhrman, J.A., Hugenholtz, P., van der Lelie,
D., Meyer, F., Stevens, R., Bailey, M.J., Gordon, J.I., Kowalchuk, G.A. and Gilbert, J.A. 2012.
Unlocking the potential of metagenomics through replicated experimental design. Nat
Biotechnol 30(6), 513-520.
Koren, S., Schatz, M.C., Walenz, B.P., Martin, J., Howard, J.T., Ganapathy, G., Wang, Z., Rasko, D.A.,
McCombie, W.R., Jarvis, E.D. and Adam, M.P. 2012. Hybrid error correction and de novo
assembly of single-molecule sequencing reads. Nat Biotechnol 30(7), 693-700.
Kunin, V., Copeland, A., Lapidus, A., Mavromatis, K. and Hugenholtz, P. 2008. A bioinformatician’s
guide to metagenomics. Microbiol Mol Biol Rev 72(4), 557-578, Table of Contents.
Li, D., Luo, R., Liu, C.M., Leung, C.M., Ting, H.F., Sadakane, K., Yamashita, H. and Lam, T.W. 2016.
MEGAHIT v1.0: A fast and scalable metagenome assembler driven by advanced methodologies
and community practices. Methods 102, 3-11.
Li, H. and Durbin, R. 2009. Fast and accurate short read alignment with Burrows-Wheeler transform.
Bioinformatics 25(14), 1754-1760.
Li, H. and Durbin, R. 2010. Fast and accurate long-read alignment with Burrows-Wheeler transform.
Bioinformatics 26(5), 589-595.
Love, M.I., Huber, W. and Anders, S. 2014. Moderated estimation of fold change and dispersion for
RNA-seq data with DESeq2. Genome Biol 15(12), 550.
Lovley, D.R. 2003. Cleaning up with genomics: applying molecular biology to bioremediation. Nat
Rev Microbiol 1(1), 35-44.
Magoc, T. and Salzberg, S.L. 2011. FLASH: fast length adjustment of short reads to improve genome
assemblies. Bioinformatics 27(21), 2957-2963.
Martin, M. 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads.
EMBNet Journal, 10-12.
Mason, O.U., Hazen, T.C., Borglin, S., Chain, P.S., Dubinsky, E.A., Fortney, J.L., Han, J., Holman, H.Y.,
Hultman, J., Lamendella, R., Mackelprang, R., Malfatti, S., Tom, L.M., Tringe, S.G., Woyke, T.,
Zhou, J., Rubin, E.M. and Jansson, J.K. 2012. Metagenome, metatranscriptome and single-cell
sequencing reveal microbial response to Deepwater Horizon oil spill. ISME J 6(9), 1715-1727.
Matar, G.K., Ali, M., Bagchi, S., Nunes, S., Liu, W.T. and Saikaly, P.E. 2021. Relative Importance of
Stochastic Assembly Process of Membrane Biofilm Increased as Biofilm Aged. Front Microbiol
12, 708531.
McHardy, A.C., Martin, H.G., Tsirigos, A., Hugenholtz, P. and Rigoutsos, I. 2007. Accurate phylogenetic
classification of variable-length DNA fragments. Nat Methods 4(1), 63-72.
Møllebjerg, A., Zarebska, A., Nielsen, H.B., Hansen, L.B.S., Sørensen, S.R., Seredynska-Sobecka, B.,
Villacorte, L.O., Gori, K., Palmén, L.G. and Meyer, R.L. 2023. Novel high-throughput screening
platform identifies enzymes to tackle biofouling on reverse osmosis membranes. Desalination
554.
Nielsen, P.H., Daims, H., Lemmer, H., Arslan-Alaton, I. and Olmez-Hanci, T. (2009) FISH Handbook
for Biological Wastewater Treatment, IWA Publishing, London, UK.
Parks, D.H., Imelfort, M., Skennerton, C.T., Hugenholtz, P. and Tyson, G.W. 2015. CheckM: assessing
the quality of microbial genomes recovered from isolates, single cells, and metagenomes.
Genome Res 25(7), 1043-1055.

334
 Chapter 14

Peng, Y., Leung, H.C., Yiu, S.M. and Chin, F.Y. 2012. IDBA-UD: a de novo assembler for single-cell and
metagenomic sequencing data with highly uneven depth. Bioinformatics 28(11), 1420-1428.
Pruesse, E., Peplies, J. and Glockner, F.O. 2012. SINA: accurate high-throughput multiple sequence
alignment of ribosomal RNA genes. Bioinformatics 28(14), 1823-1829.
Raes, J., Korbel, J.O., Lercher, M.J., von Mering, C. and Bork, P. 2007. Prediction of effective genome
size in metagenomic samples. Genome Biol 8(1), R10.
Rehman, Z.U., Ali, M., Iftikhar, H. and Leiknes, T. 2019. Genome-resolved metagenomic analysis
reveals roles of microbial community members in full-scale seawater reverse osmosis plant.
Water Res 149, 263-271.
Riesgo, A., Perez-Porro, A.R., Carmona, S., Leys, S.P. and Giribet, G. 2012. Optimization of preservation
and storage time of sponge tissues to obtain quality mRNA for next-generation sequencing. Mol
Ecol Resour 12(2), 312-322.
Roesch, L.F., Fulthorpe, R.R., Riva, A., Casella, G., Hadwin, A.K., Kent, A.D., Daroub, S.H., Camargo,
F.A., Farmerie, W.G. and Triplett, E.W. 2007. Pyrosequencing enumerates and contrasts soil
microbial diversity. ISME J 1(4), 283-290.
Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hartmann, M., Hollister, E.B., Lesniewski, R.A.,
Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B., Thallinger, G.G., Van Horn, D.J.
and Weber, C.F. 2009. Introducing mothur: open-source, platform-independent, community-
supported software for describing and comparing microbial communities. Appl Environ
Microbiol 75(23), 7537-7541.
Scholz, M.B., Lo, C.C. and Chain, P.S. 2012. Next generation sequencing and bioinformatic bottlenecks:
the current state of metagenomic data analysis. Curr Opin Biotechnol 23(1), 9-15.
Seemann, T. 2014. Prokka: rapid prokaryotic genome annotation. Bioinformatics 30(14), 2068-2069.
Shaw, D.R., Ali, M., Katuri, K.P., Gralnick, J.A., Reimann, J., Mesman, R., van Niftrik, L., Jetten, M.S.M.
and Saikaly, P.E. 2020. Extracellular electron transfer-dependent anaerobic oxidation of
ammonium by anammox bacteria. Nat Commun 11(1), 2058.
Siegwald, L., Touzet, H., Lemoine, Y., Hot, D., Audebert, C. and Caboche, S. 2017. Assessment of
Common and Emerging Bioinformatics Pipelines for Targeted Metagenomics. PLoS One 12(1),
e0169563.
Speth, D.R., Lagkouvardos, I., Wang, Y., Qian, P.Y., Dutilh, B.E. and Jetten, M.S.M. 2017. Draft
Genome of Scalindua rubra, Obtained from the Interface Above the Discovery Deep Brine in the
Red Sea, Sheds Light on Potential Salt Adaptation Strategies in Anammox Bacteria. Microb Ecol
74(1), 1-5.
Thomas, T., Gilbert, J. and Meyer, F. 2012. Metagenomics - a guide from sampling to data analysis.
Microb Inform Exp 2(1), 3.
Trego, A.C., O’Sullivan, S., Quince, C., Mills, S., Ijaz, U.Z. and Collins, G. 2020. Size Shapes the Active
Microbiome of Methanogenic Granules, Corroborating a Biofilm Life Cycle. mSystems 5(5).
van Dijk, E.L., Auger, H., Jaszczyszyn, Y. and Thermes, C. 2014. Ten years of next-generation
sequencing technology. Trends Genet 30(9), 418-426.
Vries, H.J.d. (2022) Characterizing bacteria involved in fouling of spiral wound membranes,
Wageningen University.
Wang, X., Hu, M., Xia, Y., Wen, X. and Ding, K. 2012a. Pyrosequencing analysis of bacterial diversity
in 14 wastewater treatment systems in China. Appl Environ Microbiol 78(19), 7042-7047.
Wang, Y., Cao, H., Zhang, G., Bougouffa, S., Lee, O.O., Al-Suwailem, A. and Qian, P.Y. 2013.
Autotrophic microbe metagenomes and metabolic pathways differentiate adjacent Red Sea brine
pools. Sci Rep 3, 1748.

335
Experimental Methods for Membrane Applications

Wang, Y., Leung, H.C., Yiu, S.M. and Chin, F.Y. 2012b. MetaCluster 4.0: a novel binning algorithm for
NGS reads and huge number of species. J Comput Biol 19(2), 241-249.
Wrighton, K.C., THOMAS, B.C., SHARON, I., MILLER, C.S., CASTELLE, C.J., VERBERKMOES, N.C.,
WILKINS, M.J., HETTICH, R.L., LIPTON, M.S., WILLIAMS, K.H., LONG, P.E. and BANFIELD,
J.F. 2012. Fermentation, Hydrogen, and Sulfur Metabolism in Multiple Uncultivated Bacterial
Phyla. Science 337(6102), 1661-1665.
Wu, L., Ning, D., Zhang, B., Li, Y., Zhang, P., Shan, X., Zhang, Q., Brown, M.R., Li, Z., Van Nostrand,
J.D., Ling, F., Xiao, N., Zhang, Y., Vierheilig, J., Wells, G.F., Yang, Y., Deng, Y., Tu, Q., Wang, A.,
Global Water Microbiome, C., Zhang, T., He, Z., Keller, J., Nielsen, P.H., Alvarez, P.J.J., Criddle,
C.S., Wagner, M., Tiedje, J.M., He, Q., Curtis, T.P., Stahl, D.A., Alvarez-Cohen, L., Rittmann,
B.E., Wen, X. and Zhou, J. 2019. Global diversity and biogeography of bacterial communities in
wastewater treatment plants. Nat Microbiol 4(7), 1183-1195.
Wu, Y.-W., Tang, Y.-H., Tringe, S.G., Simmons, B.A. and Singer, S.W. 2016. MaxBin: an automated
binning method to recover individual genomes from metagenomes using an expectation-
maximization algorithm. Microbiome 2, 26.
Yang, Y., Li, B., Ju, F. and Zhang, T. 2013. Exploring variation of antibiotic resistance genes in activated
sludge over a four-year period through a metagenomic approach. Environ Sci Technol 47(18),
10197-10205.
Yu, K. and Zhang, T. 2012. Metagenomic and metatranscriptomic analysis of microbial community
structure and gene expression of activated sludge. PLoS One 7(5), e38183.
Zhang, T., Shao, M.F. and Ye, L. 2012. 454 pyrosequencing reveals bacterial diversity of activated
sludge from 14 sewage treatment plants. ISME J 6(6), 1137-1147.

336
 doi: 10.2166/9781789062977_0337

Chapter 15

Biofouling Potential
Measurement using Bacterial
Growth Potential (BGP)
Almotasembellah Abushaban, UM6P, Morocco
Karima Bakkali, UM6P, Morocco
Léonie Le Bouille, IHE Delft / CIRSEE-Suez, France
Mohamed ChakerNecibi, UM6P, Morocco
Sergio G. Salinas-Rodriguez, IHE Delft, The Netherlands

The learning objectives of this chapter are the following:

• Define biological fouling in membrane systems

• Apply bacterial growth potential measurements as a biofouling indicator

• Define and apply microbial adenosine triphosphate for assessing biomass activity
and bacterial growth

• Present and discuss the latest regulation to limit biofouling in SWRO systems

• Present several cases studies on assessing biofouling potential along SWRO pre-
treatment and feedwater.

15.1 INTRODUCTION

Numerous seawater reverse osmosis (SWRO) desalination plants still struggle to control
biological and organic fouling because there are no standard methods to monitor these
types of fouling in desalination plants. Biological fouling results from microbial growth
in membrane systems, which may lead to operational problems such as increased head

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

loss across feed spacers in spiral wound elements and decreased permeability of SWRO
membranes. Biofilm formation in SWRO is inevitable if the feed water supports significant
bacterial growth due to the presence of easily biodegradable (dissolved) nutrients.

While there is currently no standard test method for measuring the progression of biofilm
formation in the plant, it is still possible to identify the formation of biofilms by testing and
monitoring water quality, membrane fouling, and a variety of other factors.

Several approaches are being studied to monitor and control biofouling in SWRO. A good
approach for biofouling control is to lower biological fouling potential in SWRO feed water
through the pre-treatment. By pre-treating feed water, it can be reduced to a condition where
bacteria and other contaminants cannot grow as well or kill off the type of microorganism
that causes bio fouling. This approach is attractive because it can be used as an early warning
system allowing adjustment of the operational conditions of the pre-treatment processes to
meet the required quality in SWRO feed water and consequently better control of biofouling
in SWRO systems. For this purpose, few methods were developed to measure biological/
organic fouling potential such as assimilable organic carbon (AOC), biodegradable dissolved
organic carbon (BDOC) and bacterial growth potential (BGP). The use of AOC and BGP
methods has gained interest as high levels of AOC/BGP are directly linked to biofilm
formation and thus more severe biofouling is expected at higher AOC/BGP in the SWRO
feed water.

In general, to measure AOC or BGP in any water sample, four steps should be followed as
following (Abushaban et al., 2022; Salinas-Rodriguez et al., 2021):
- Bacterial inactivation by pasteurization, filtration or sterilization. This step allows
the standardization of the initial microbial population by adding a constant inoculum
concentration.
- Bacterial inoculum by adding constant concentration of specific bacteria or indigenous
bacteria.
- Incubation at a constant temperature.
- Bacterial enumeration: Different enumeration methods can be used depending on the
bacterial culture such as total direct cell count, bioluminescence, turbidity, microbial
electrolysis, flow cytometry (FCM) and ATP.

15.2 MATERIALS

15.2.1 Laboratory equipment

• Eppendorf single channel pipette (adjustable volume) 10-100 μL, 100-1000 μL, 0.5-5 mL
• Eppendorf Biopur® pipette tips (sterile, pyrogen-free, DNA-free and ATP-free) for 10-
100 μL, 100-1000 μL, 0.5-5 mL
• Eppendorf Biopur® 1.5 ml Safe-Lock micro test tubes (sterile, pyrogen-free, human-
DNA free, bacterial-DNA free, RNase-free, DNase-free, PCR-inhibitor free and ATP-
free), individually sealed
• 50 mL centrifuges tubes with screw caps, Roth (sterile)
• 0.1 μm PVDF membrane Filters, Millex-GP, Merck Millipore (sterile)
• Syringes 10 mL, 5 mL, BD (sterile)
• Tube holders

338
 Chapter 15

Remark:
All materials and consumables that is used for microbial ATP measurement should be sterile
including pipette, pipette tips, centrifuge tubes, Eppendorf tubes, etc. The consumable
should be used for one single use.

15.2.2 CHEMICALS

Chemicals for ATP measurement

• Bacteria lysis is used to extract ATP from the microbial cell by destroying the microbial
cell membrane. It can be stored at room temperature.
• ATP Water-Glo: is the light generating reagent containing luciferin and luciferase. ATP
Water-Glo is stored at room temperature and after reconstitution with buffer the reagent
is stored at 4 °C.
• ATP standard is a liquid stable of 100 nM (50,000 ng/L) (Biothema or Promega Corp.)
of ATP concentration.
Staining chemicals for flow cytometry:
• SYBR Green I (SG): It is used to measure the total cell counts (TCC) in a water sample. It
needs to be stored at 4 °C.
• Propidium iodide (PI): It is used to measure the intact cell counts in a water sample. It
needs to be stored at 4 °C.

15.2.3 Instrumental equipment

• AccuBlock™ Digital Dry Baths, Block
• Vortex mixer (Heidolph Reax 2000 shaker)
• Incubator (30 °C)
• Autoclave
• Muffle furnace (up to 550 °C)

Figure 1 The GloMax® 20/20 Single Tube Luminometer.

• Promega GloMax®-20/20 single tube Luminometer: Different companies have different

Luminometer. GloMax® 20/20 single tube Luminometer (Figure 1) developed by the
Promega Company measures the light emitted during Luciferin-Luciferase reaction. To
measure luminescence of a sample, the sample should be transferred into a 1.5-Micro

339
Experimental Methods for Membrane Applications

centrifuge tube before inserting it inside the Luminometer. The cover of the equipment
should be closed during the measurement to avoid any interference of any other external
light. The result of the measurements is displayed within 10 seconds on the computer
screen in relative light units (RLU).
• BD Accuri C6 or C6Plus Flow Cytometer system: Flow cytometer (FCM) is a device used
to measure particles or cells which are present in the fluid and pass the laser intercept
of the equipment and results in light forwarding and scattering. The flow cytometer
can also measure the size of the cell which can be determined by the amount of light
scattered by individual cell (Prest et al., 2016). Since the flow cytometer is based on light
forwarding and scattering, which can be due to cells or any other particle present in the
sample, a standard flow cytometric staining protocol was developed at Eawag (Gatza et
al., 2013) to distinguish between the bacterial cells and other particles present in sample.
The staining protocol is employed by BD AccuriTM C6 and C6 Plus (Figure 2) software
analysis template which can be used to measure not only the total cells present in water
but also the viable cells. The template consists a single, fixed gate that can determine
total bacterial cells when stained with SYBR® Green I, or only viable bacteria cells when
co-stained with SYBR® Green I and propidium iodide (Abushaban, 2020; Gatza et al.,
2013).

Figure 2 The BD Accuri C6 Flow Cytometer system.

15.3 METHODS AND EXPERIMENTAL PROCEDURE

15.3.1 Sample collection and storage

Seawater sample needs to be collected in sterile glass and stored in cooler box (4 °C)
during transportation. Amber color glass sampling bottles with tight sealing screw caps are
preferred. The volume of the collected sample should be at least 100 mL.

340
 Chapter 15

15.3.2 Cleaning glassware

Vials, bottles, caps and all glassware materials must be washed before any usage. Glassware
may contain some external sticky contaminants such as particles and bacteria from inside
and outside, which can lead to inaccurate results if not cleaned.
1. Wash all glassware and caps using lab detergent such as Alconox Ultrasonic Cleaner.
2. Rinsed all glassware with Milli-Q water (Milli-Q® water Optimized purification, 18.2
MΩ.cm at 25°C, EC < 10 μS/cm, TOC < 30 μg/L, Millipore, USA).
3. Submerge all glassware and caps overnight in 0.2 M HCl.
4. Once again, rinse all glassware and caps three times with Milli-Q and let them dry in the
air.
5. After draying, vials need to be covered with aluminum foil to avoid any potential organic
contaminants.
6. Heat vials and bottles in oven furnace at 550 °C for 6 hours (effective time to expose
glassware to heat).
7. On the other hand, vials caps are bathed in sodium persulfate solution (100 g/L ) for 1
hour at 60 °C. Then, rinse all caps with Milli-Q water and let them air dried. Powder free
gloves were used during the handling process.
8. Finally, take out the aluminum foil and close all vials and bottles with caps.

15.3.3 Preparation of artificial seawater

Artificial seawater (ASW) needs to be prepared with similar characteristic of the real
seawater in terms of ion compositions, electrical conductivity and pH. Table 1 shows the ion
composition of the average global seawater conditions (Villacorte, 2014). For preparation of
the ASW, all glassware needs to be cleaned before using them as mentioned in the previous
section. The salts used in the preparation of ASW may be contaminated. Therefore, the salts
with high enough melting points must be burned at 550 °C for 6 hours to make carbon
volatile. (Table 1). The amount of each salt is then dissolved in Milli-Q water (Milli-Q®
water Optimized purification, 18.2 MΩ.cm at 25 °C, EC < 10 μS/cm, TOC < 30 μg/L,
Millipore, USA) and stirred for at least 24 hours. Moreover, once the ASW is prepared, it
needs to be autoclaved again to ensure that the prepared ASW is bacteria free.

Table 1 Ion concentrations of artificial seawater to mimic the average global seawater (Abushaban,
2020)
Concentration Heating temperature
Component name Chemical formula (g/L) (°C)
Sodium carbonate Na2CO3 0.002 100 °C
Sodium hydrogen carbonate NaHCO3 0.213 NA
Potassium chloride KCl 0.739 550 °C
Calcium chloride di hydrate CaCl2.2H2O 1.540 100 °C
Sodium sulfate Na2SO4 3.993 550 °C
Magnesium chloride hexahydrate MgCl2.6H2O 10.873 100 °C
Sodium chloride NaCl 22.3 550 °C

341
Experimental Methods for Membrane Applications

The autoclaving temperature of ASW should be set to 100 °C for 15 minutes, as the melting
point of magnesium chloride hexahydrate (MgCl2.6H2O) is around 117 °C. It is worth
mentioning that Sodium hydrogen carbonate (NaHCO3) should not be heated because its
melting point is 50 °C and should therefore be added after the autoclaving step.

15.3.4 Intact cell count by flow cytometry

To measure the number of intact cells using FCM in a water sample, two staining solutions
can be used such as SYBR Green I (SG) and Propidium Iodide (PI). The former is used to
distinguish the total bacterial cells from debris and the latter is used to differentiate the intact
cells from damaged bacterial cells. The staining should be only done for samples containing
bacterial cells lower than 107 cells/mL. For samples containing bacterial cells more than
107 cells/mL, they should be diluted with filtered ASW before staining. The procedure to
measure intact cell concentration of a water sample is as follows:
• Preheat the samples before staining for 5 min at 35 °C.
• Add 5μL of the stock solution (SG/PI) in 500 μL sample.
• Place the stained samples in dark place and at 35 °C for exactly 10 minutes.
• From computer screen, choose 50 μL run limits, with medium speed and threshold 700
on F1 was selected.
• Insert the sample into the FCM device and start the measurement.

Remark:
The maximum number of cells should not exceed 2 × 107 cells/mL. In case the sample
contains higher number of cells, but lower than 107 cells/mL, the sample can be diluted
after staining, as mentioned earlier and measured again.

15.3.5 Measurement of Bacterial growth potential

To measure BGP of a seawater sample, four steps should be followed as including, bacterial
inactivation, bacterial inoculation, sample incubation, and bacterial growth monitoring/
bacterial enumeration.
I. Bacterial inactivation: to inactivate bacteria in the seawater sample, the collected sample
in glass bottle needs to be placed in an autoclave at a temperature between 70 and 100 °C
and for 15-20 min. The bottle should be closed properly with a screw cap. It is important
not to increase the temperature to more than 100 °C because higher temperature could
negatively affect the Free ATP concentration in the seawater sample. After autoclaving,
take the sample out of the autoclave and let it cool down (at room temperature).
II. Bacterial inoculation:
1. Transfer 20 mL of the autoclaved sample to a small cleaned vial (40 mL) in triplicate (3
vials). The vials should be cleaned following the pre-mentioned protocol for cleaning.
The vial should not be filled completely to avoid any contamination form the cap.
Sterilized syringe or pipette can be used to transfer the sample. It is recommended to
do all steps on a clean surface of closed room (to avoid contamination from air).
2. Close the vial immediately after sample transfer using the cleaned vial cap.
3. Count the intact cell concentration of the same non-autoclaved sample (inoculum
concentration) using flow cytometry (FCM).
4. Calculate the volume of inoculum to be added to the autoclaved sample (20 mL) using
the following formula (equation 1). The final concentration of the inoculum should
be 10,000 cells/mL.

342
 Chapter 15

C2 ⋅V2
V1 = Eq. 1
C1

Where,
C1 is the concentration of the sample measured by FCM minus the concentration of the
inoculum (10,000 cells/mL),
V1 is the volume of the inoculum to be added to the sample,
C2 is the final concentration of the inoculum (10,000 cells/mL), and
V2 is the final volume of the sample (20 mL).
Example: Assume the concentration of intact cells in the sample is 1.5 M cells/mL. The
volume of the inoculum is (10,000 cells/mL × 20 mL ÷ 1,490,000 cells/mL) = 134.2 μL.
5. Transfer the calculated inoculum volume to the vials which have 20 mL autoclaved
samples using a sterilized pipette tip.
Note: The sample should be vortexed before pipetting the inoculum. Moreover, open
the vial just before transferring the inoculum.
6. Once the inoculum is transferred in triplicate, the vials should be closed with a cleaned
cap and vortexed.
III. Sample incubation: Place the inoculated sample at the incubator at 30 ˚C. Make sure
that each vial is labelled including the date of sample collection and date of inoculation.
The number of replications of each sample should be mentioned on the label to avoid
mixing samples.
IV. Sample enumeration: The bacterial growth of inoculum needs to be monitored daily for
at least 5 to 7 days. Microbial ATP or intact cell concentration using FCM can be used to
monitor bacterial growth. The detailed protocol of these methods is presented in section
15.7. The microbial concentration on the day of inoculation should be measured as well
as a reference concentration (Day 0). The concentration of day 0 (day of inoculation)
should be around 10,000 cells/mL (intact cells measured by FCM) or less than 10 ng/L
(Microbial ATP concentration).

15.3.6 Bacterial yield and calibration line

Establishing a calibration line of BGP is necessary to convert microbial growth to carbon
concentration. Calibration curve is prepared by monitoring the bacterial growth of ASW of
different known glucose concentrations (0, 10, 25, 50, 75 and 100 μg-C/L). To establish
the curve, a correlation is made between added glucose concentration and its corresponding
maximum bacterial growth of each concentration (Figure 3). It is worth mentioning that
the bacterial growth may vary depending on the microorganism present in inoculum of
indigenous microbial consortium (Wang et al., 2014; Weinrich et al., 2010), therefore the
bacterial yield needs to be determined for each location.

The bacterial growth can be also inhibited due to absence of essential nutrients such as
nitrogen and phosphorus. Therefore, nitrogen and phosphorous concentrations are added
according to C: N: P proportional weight of 20:4:1 in artificial seawater.
The conversion of microbial growth to carbon concentrations is only possible if the bacterial
yield is known. The bacterial yield is determined by calculating the slope of the established
calibration curve which equals to the average bacterial growth that corresponds to 1 μg/L of
glucose.

343
Experimental Methods for Membrane Applications

To prepare BGP calibration curve, the following procedure should be followed:

1. Prepare 1 L of artificial seawater following the above-mentioned procedure.
2. Prepare 20 mL of different glucose concentrations (for example; 0, 10, 25, 50, 75 and
100 μg-C/L) in triplicate (3×6 concentrations) using the sterilized artificial seawater
(point 1). To prepare these concentrations, one stock solution of glucose with 2 mg/L
concentration can be firstly prepared (using ASW of point 1) before preparing the low
glucose concentration samples.
3. To allow bacterial growth, nitrogen and phosphorous concentrations should be added
similar to point 2 and according to C: N: P proportional weight of 20:4:1.
4. Vortex the prepared samples containing C, N, and P before adding the inoculum
concentration. To calculate the inoculum concentration, follow step II.4 under section
15.3.4.
5. Follow the same steps of BGP measurement to incubate and enumerate the samples.
6. After 7 days of bacterial growth monitoring, draw the relationship between the added
glucose concentration and the corresponded maximum bacterial growth. The relationship
should be a linear between them.
7. Bacterial yield can be calculated by identifying the slope of the correlation between
maximum growth and the added glucose concentration.
8. Use the linear correlation to convert the maximum bacterial growth in ng/L to carbon
concentration μg-C/L.
Note: A correlation needs to be established between bacterial growth and carbon
concentration for a specific location.
Remark: To make sure that the calibration line is valid for your specific seawater sample, a
correlation can be investigated using a real seawater (RSW). The slope using RSW should
be similar or close to the slope prepared using ASW.

240

210
y = 1.56x + 66.8
R2 = 0.98 180

150

120

y = 1.43x + 15.98 90
R2 = 0.99
Seawater 60
Artificial seawater
30

Maximum growth (mg-ATP/L) 0

0 20 40 60 80 100 120
Glucose concentration (µg/L)

Figure 3 The correlation between added glucose concentration and the BGP in seawater and
artificial seawater (Abushaban et al., 2019a).

344
 Chapter 15

15.4 APPLICATIONS

15.4.1 Example A: BGP monitoring of an SWRO pre-treatment

BGP concentration were measured in a full-scale membrane-based desalination plant
in Australia. The pre-treatment processes of SWRO system includes a drum screen,
coagulation and flocculation, dual media filter (DMF), and cartridge filter. Four samples were
collected in October 2016 through the pre-treatment as following; just before coagulation
(raw seawater), after coagulation and flocculation, after DMF, and after cartridge filter.

Results showed that bacteria started to grow immediately in seawater and reached a
maximum growth within 2 days (Figure 4a). Afterwards, microbial ATP gradually decreased
due to the partial decay of bacteria or because bacterial activity decreased due to the depletion
of nutrients. The maximum bacterial growth was observed (305 ng-ATP/L) in raw seawater
(Figure 4b), indicating the highest potential of bacterial growth. Slightly lower potential of
bacterial growth (262 ng-ATP/L) was observed after coagulation and flocculation, while a
significant reduction (> 55 %) of the bacterial growth potential was found after DMF. The
high reduction in BGP through DMF could indicate that the DMF is a biologically active
media filter. The maximum bacterial growth decreased modestly through the cartridge
filter to 86 ng-ATP/L. This result shows that the seawater after pre-treatment still supports
further bacterial growth when it is compared with the BGP of the blank.

a) b)
400 400
350 350
Microbial ATP (ng-ATP/L)

300 300
250 250
200 200
150 150
100 100
50 50
0 0
0 1 2 3 4 5 Blank Raw After After After
Time (day) (ASW) seawater flocculation filtration cartidge
(DMF) filter

After flocculation Reported in ng-ATP/L

Raw seawater Reported in µ-C/L
After cartidge filter
After filtration (DMF)
Blank (ASW)

Figure 4 (a) bacterial growth over time and (b) maximum growth of different seawater samples
collected through the pre-treatment processes of an SWRO desalination plant in Australia.
(Abushaban et al., 2021, 2018).

15.4.2 Example B: BGP in the intake and SWRO feed water

BGP was monitored at the intake and feedwater of an SWRO desalination plant in the
Middle East. High BGP variations were observed in the seawater intake, in which the average
monthly BGP ranged between 200 and 2,500 μg-C/L as glucose (Figure 5). Low BGPs were
measured in the seawater intake during the Autumn, whereas extremely high BGPs were

345
Experimental Methods for Membrane Applications

observed in September due to algal blooms which is widely reported in the Arabian Sea in
September and October. The high BGP during summer might be attributed to carbon release
from the algal cells present in seawater.

On average, the removal of BGP along the SWRO pre-treatment is 62 %, in which the
maximum BGP removal (85%) was observed in July. However, low BGP removal or even
negative removal were noted from October to December which could be attributed to
the addition of antiscalant or the make-up water used for diluting antiscalant. The higher
organic concentration after antiscalant addition has been observed in several SWRO and
RO plants(Abushaban et al., 2021; Vrouwenvelder et al., 2000). Although reasonable
concentration of organic and biological fouling potential was removed through the pre-
treatment, still considerable concentration remains in the SWRO feed water (> 100 μg-
C/L) which could lead to biofouling operational problem in the SWRO.

The monitored BGP based on microbial ATP illustrate its applicability to evaluate the pre-
treatment processes of desalination systems.

2,500

2,000

1,500

1,000

Seawater intake 500

SWRO feedwater
BGP concentration (µg-C/L) 0
July Aug Sep Oct Nov Dec
Time (month)

Figure 5 Monitored bacterial growth potential in the intake and SWRO feed water of an SWRO
desalination plant located in the Middle East (Abushaban et al., 2020).

15.5 DATA DISCUSSION AND INTERPRETATION

The correlation between BGP in the SWRO feed water and the normalized pressure drop/
permeability in the SWRO membrane system has been demonstrated (Abushaban, 2020).
It was reported that the higher BGP in SWRO feed water corresponds to higher normalized
pressure drop. In general, biofouling was observed in SWRO membrane systems when the
measured BGP in the SWRO feed water is more than 100 μg-C/L.

Abushaban et al. (2020) preliminary investigated the correlation between the measured
BGP in SWRO feed water and the RO membrane cleaning in place (CIP) frequency in the
four SWRO plants. The CIP frequency (CIPs per year) was used as a surrogate parameter
for biofouling. It is reported that a higher CIP frequency corresponded to a higher BGP of
SWRO feed water (Figure 6). Accordingly, a tentative threshold concentration of BGP (< 70
μg/L) was proposed for SWRO feed water in order to ensure a chemical cleaning frequency

346
 Chapter 15

of once/year or lower. However, to establish a real correlation, more data needs to be

collected and many more SWRO plants need to be monitored for longer periods of time
with different operating conditions.
6
Plant A
5

4
Plant D
3

2
Plant B
Plant C 1
CIP frequency (CIPs/year) 0
0 50 100 150 200 250
BGP in SWRO feed (µg-C/L as glucose)

Figure 6 The correlation between bacterial growth potential in the feed water and cleaning
frequency of SWRO membrane systems (Abushaban, 2019; Dhakal et al., 2020).

15.6 ATP MEASURMENT

15.6.1 Introduction
ATP is a substance present in all living cells (including bacteria) that provides energy for
many metabolic processes. In particular, it is used as a coenzyme in living cells and it is often
called the ‘molecular unit of currency’ of intracellular energy transfer (Knowles, 1980).
The structure of ATP has an ordered carbon compound as shown in Figure 7. ATP consists
of adenosine and three phosphate groups. Adenosine composes of an adenine ring and
a ribose sugar. The critical part of ATP is the phosphorous part - the triphosphate. Three
phosphorous groups are connected by oxygen, and also there are side oxygen connected
to the phosphorous atoms which leads to high repulsion between the negative charges of
oxygen in the normal condition. Therefore, ATP has a lot of potential energy. ATP converts
to adenosine diphosphate (ADP) if one of the three phosphates is broken down. This
conversion is an extremely crucial reaction due to the realization of the energy after the
reaction. The reaction and realized energy of one Mole ATP is shown in eq (2).

ATP + H 2O ADP + Pi G˚= 30.5kJ /mol ( 7.3kcal / mol) Eq. 2

Adenine
NH2
3 phosphate groups
N C C N
C H O- O- O-
H C
C O H 2C O P O P O P O
N
N O O O
H H
H H
OH OH

Ribose

Figure 7 Schematic diagram showing the structure of ATP.

347
Experimental Methods for Membrane Applications

Several methodologies are used for ATP determination, but so far, the most successful
technique is the bioluminescent method, because of its sensitivity and the wide range
concentration that can be measured (Van der Kooij et al., 2003). ATP bioluminescence has
been used for determining levels of ATP in many different cell types. The bioluminescence
method involving the Luciferase enzyme is a multistep process which mainly requires
Luciferin substrate, oxygen (O2), magnesium (Mg2+) and ATP. Enzyme luciferase uses as
catalyst of reaction which is extracted from firefly (Photinus pyralis). Luciferase converts in
presence of ATP and magnesium firefly D-luciferin into the corresponding enzyme-bound
luciferil adenylate which converts to oxyluciferin in the presence of oxygen. This process
occurs according to the following chemical equations:

D luceferin + luciferase + ATP Mg

luceferil adenylate complex + PPi Eq. 3
O2
Luceferin adenylate complex Oxyluceferin + AMP *+CO2 + light Eq. 4

*AMP = Adenosine monophosphate

Light is emitted from a rapid loss of energy of the oxyluciferine molecule. The light emission
is in the range between 500 to 700 nm wavelengths (Riemann, 1979). Under optimum
conditions, light intensity is linearly related to ATP concentration. Cellular ATP can be
measured by direct lysis of the cells with suitable detergent. In general, the determination
of ATP concentration includes the following procedures (Van der Kooij et al., 2003) which
also presented in Figure 8:
• Collection of a representative sample.
• Extraction of ATP from the microorganisms.
• Addition of reagents for luciferine – luciferase assay.
• Recording the light emitted.
• Calculation of the ATP concentration from calibrated data.

ATP
ATP ATP
Extraction of ATP ATP Luciferin/luciferase Light
ATP ATP ATP
ATP
ATP ATP
ATP

Figure 8 The simple protocol of bioluminescent method.

15.6.2 Material and methods

Material
The laboratory equipment, instrumental equipment and chemicals are listed in section 15.2
Note: All materials and consumables to be used for microbial ATP measurement should be
sterile including pipette, pipette tips, centrifuge tubes, Eppendorf tubes, etc. The consumable
should be used for one single use.

348
 Chapter 15

Methods
Preparation of Water-Glo detection reagent
• Clean the bench, pipette and the cover of the pipette tips with Ethanol (70 %),
• Carefully open the WaterGlo™ Substrate vial and do not touch the inside part of the
stopper,
• Transfer the contents of the WaterGlo™ Buffer vial to the WaterGlo ™ Substrate vial,
• Replace the stopper and slowly invert the vial several times to dissolve the contents, and
• Allow reconstituted detection reagent to stand at room temperature for 1 hour before
use.
Note: The prepared Water-Glo detection reagent should be Stored at the fridge at 5 °C and
it can be taken out of the fridge 15 min before the measurement.

Direct ATP method

a. Calibration curve
Since the concentration of ATP is unknown in the sample, a calibration curve needs to
be established by measuring the relative light unit (RLU) of different ATP concentration.
To prepare different ATP concentrations, ATP standard (100 nM, 50,000 ng-ATP/L,
Biothema) needs to be diluted in the water medium. To prepare the calibration curve, the
following procedure is followed:
1. Biothema ATP Standard (100 nM) is taken out of the fridge 5 min before using it.
2. Use sterilized artificial seawater as water medium to dilute standard ATP.
3. Use the table 2 below to prepare different ATP standard solutions.
4. Switch on the heating block and adjust the heating temperature at 38 ˚C.
5. Distribute the prepared concentration to 6 Eppendorf tubes (3 for total ATP and 3 for
free ATP) in which 100 μL is pipette into each tube.
6. Add 100 μL of Lysis reagents on each Eppendorf tubes prepared for total ATP
measurements.
7. Place your prepared samples into the heating block and start measuring the samples as
per described in the protocol below.

Table 2 The required volume of ATP standard and medium to prepare a calibration curve using
ATP direct method.
Final Concentrate Volume of the ATP Original concentrate Volume of medium
(ng ATP/L) standard (µL) (ng ATP/L) (µL)

5,000 60 50,000 540

500 100 5,000 900
250 45 5,000 855
100 18 5,000 882
50 100 500 900
25 18 500 882
5 100 50 900
0.5 100 5 900
0 0 0 1,000

349
Experimental Methods for Membrane Applications

b. Measurement protocol
The ATP measurement to be used is the method developed by Abushaban et al., (2018) for
seawater samples. The protocol of ATP measurement using the direct method is illustrated
in Figure 9.

a) Total ATP

ATPTotal

100 µL of Bacteria Lysis Heating at 38˚C + 200 µL of Water-Glo RLU

+ 100 µL of seawater for 4 minutes

b) Free ATP

ATPFree

100 µL of seawater Heating at 38˚C + 100 µL of Water-Glo RLU

for 4 minutes

Microbial ATP = Total ATP – Free ATP

Figure 9 Scheme of Direct Method Set up (Abushaban et al., 2018).

Total ATP
• Transfer 100μL of seawater sample into 1.5mL Eppendorf tubes in triplicates.
• Use pipette to transfer 100μL of Bacteria Lysis into the Eppendorf tubes containing the
sample and mix with the pipette.
• Heat the sample in a heating block for 4 minutes and at a temperature of 38 ˚C together
with the Water-Glo reagents in separate Eppendorf tubes.
• Add 200μL of heated Water-Glo reagent to the heated sample using a pipette.
• Mix the sample by using the sucking and release the mixture once and then immediately
place the tube into the GloMax Luminometer (measurement to be conducted within 20
second of contact time).
• The measured RLU of the sample will be recorded directly in the excel sheet connected
with the luminometer.

Free ATP Measurement:

• Transfer 100μL of seawater sample into 1.5mL Eppendorf tubes in triplicates.
• Heat the sample in a heating block for 4minutes at 38 ˚C together with the WaterGlo
reagents in separate Eppendorf tubes.
• Add 100μL of heated Water-Glo reagent to the heated sample using a pipette.
• Mix the sample by using the sucking and release the mixture once and then immediately
place the tube into the GloMax Luminometer (measurement to be conducted within 20
second of contact time).
• The measured RLU of the sample will be recorded directly in the excel sheet connected
with the luminometer

350
 Chapter 15

• Calculate the ATP concentration based on the calibration curve. Bacterial ATP = Total
ATP – Free ATP

Note: The lid cover of the luminometer should be immediately closed after inserting the
Eppendorf tube to avoid any external light influence.

ATP Filtration method

a. Preparation of the Calibration Curve:

ATP calibration line is needed to convert the measured relative light unit to a concentration
of ATP. For this purpose, different solutions of different ATP concentrations need to be
prepared and measured using Luminometer. For the ATP filtration method, ATP standard
of 2 nM (1,000 ng-ATP/L, Promega Corp.) is used. Moreover, bacterial lysis is used as a
medium solution to dilute ATP standard. To prepare the calibration curve of ATP filtration
method, the following procedure is followed:
1. Promega ATP Standard (2 nM) is taken out of the fridge 5 min before using it.
2. Use the sterile bacterial lysis reagent as water medium to dilute standard ATP.
3. Use the table 3 below to prepare different ATP standard solutions.
4. Switch on the heating block and adjust the heating temperature at 38 ˚C.
5. Distribute the prepared concentration to 3 Eppendorf tubes in which 100 μL is pipette
into each tube.
6. Place your prepared samples into the heating block and start measure the samples as per
described protocol below.

Table 3 Dilution of ATP standard in bacterial lysis for ATP filtration method calibration curve
(Abushaban, 2020)
Final Concentrate Volume of bacterial lysis ATP standard Volume of the ATP
(ng-ATP/L) reagent (µL) (ng ATP/L) standard (µL)

500 250 1,000 250

250 375 1,000 125
100 450 1,000 50
50 950 1,000 50
25 250 50 250
5 450 50 50
0 1,000 0 0

b. Protocol of the ATP filtration method

The followings are the steps to measure microbial ATP using ATP filtration method which
is also presented in Figure 10:
1. Filter 5 mL of seawater sample through 0.1 μm filter in order to accumulate bacterial cells
on the filter surface. The filtrate is discarded.
2. Filter 2 mL of rinsing solution (sterilized ASW) through the same filter to eliminate free
ATP which could be retained in the filter holder from the 1st step. The filtrate is discarded.

351
Experimental Methods for Membrane Applications

3. Filter 5 mL of bacteria lysis through the same filter to extract ATP from bacterial cells.
Additionally, inject air to the filter holder in order to push the remaining bacteria lysis
solution.
4. Transfer 100 μL of collected sample into 1.5 mL Eppendorf tubes (triplicate).
5. Heat all the samples and ATP Water-Glo detection reagent at 38 ˚C using a heating block.
6. Transfer 100 μL of the heated ATP Water-Glo detection reagent to the heated sample
and mix the two solutions (reagent and sample).
7. Immediately insert the mixed sample in the luminometer to measure RLU value of the
sample.
8. Calculate the ATP concentration from the calibration curve using the below formula:

1
a. ATP = ⋅ (RLU − b)
m
Where:
RLU: the relative light unit obtained from Luminometer
b: The intercept with y-axis of calibration curve.
m: The slope of calibration curve.
RLU: Relative light units value recorded by the luminometer
ATP: The calculated ATP concentration based on calibration curve

Heat the filtrate solution Mix Promega BacTiter Glo

and the lighting reagent reagent with the filtrate
(BacTiter Glo) at 38˚C for 4 min solution to measure RLU

Concentrate seawater Flush out the

bacteria on the free ATP from real
membrane surface seawater sample

Extract ATP from collected ATP = 1 x (RLU–b)

m
bacterial using bacterial lysis

Figure10 The protocol of microbial ATP measurement in seawater (Abushaban et al., 2019b).

352
 Chapter 15

15.7 REFERENCES

Abushaban, A., 2019. Assessing Bacterial Growth Potential in Seawater Reverse Osmosis Pretreatment:
Method Development and Applications. CRC Press.
Abushaban, A., Mangal, M.N., Salinas-Rodriguez, S.G., Nnebuo, C., Mondal, S., Goueli, S.A., Schippers,
J.C., Kennedy, M.D., 2018. Direct measurement of ATP in seawater and application of ATP to
monitor bacterial growth potential in SWRO pre-treatment systems. Desalination and Water
Treatment 99, 91–101.
Abushaban, A., Salinas-Rodriguez, S.G., Dhakal, N., Schippers, J.C., Kennedy, M.D., 2019a. Assessing
pretreatment and seawater reverse osmosis performance using an ATP-based bacterial growth
potential method. Desalination 467, 210–218. https://doi.org/10.1016/j.desal.2019.06.001
Abushaban, A., Salinas-Rodriguez, S.G., Kapala, M., Pastorelli, D., Schippers, J.C., Mondal, S.,
Goueli, S., Kennedy, M.D., 2020. Monitoring Biofouling Potential Using ATP-Based Bacterial
Growth Potential in SWRO Pre-Treatment of a Full-Scale Plant. Membranes 10. https://doi.
org/10.3390/membranes10110360
Abushaban, A., Salinas-Rodriguez, S.G., Mangal, M.N., Mondal, S., Goueli, S.A., Knezev, A.,
Vrouwenvelder, J.S., Schippers, J.C., Kennedy, M.D., 2019b. ATP measurement in seawater
reverse osmosis systems: Eliminating seawater matrix effects using a filtration-based method.
Desalination 453, 1–9.
Abushaban, A., Salinas-Rodriguez, S.G., Pastorelli, D., Schippers, J.C., Mondal, S., Goueli, S., Kennedy,
M.D., 2021. Assessing Pretreatment Effectiveness for Particulate, Organic and Biological Fouling
in a Full-Scale SWRO Desalination Plant. Membranes 11, 167–167.
Abushaban, A., Salinas-Rodriguez, S.G., Philibert, M., Le Bouille, L., Necibi, M.C., Chehbouni,
A., 2022. Biofouling potential indicators to assess pretreatment and mitigate biofouling in
SWRO membranes: A short review. Desalination 527, 115543. https://doi.org/10.1016/j.
desal.2021.115543
Dhakal, N., Abushaban, A., Mangal, N., Abunada, M., Schippers, J.C., Kennedy, M.D., 2020. Membrane
Fouling and Scaling in Reverse Osmosis, in: Membrane Desalination. CRC Press, pp. 325–344.
Gatza, E., Hammes, F., Prest, E., 2013. Rapid and accurate quantitation of bacteria in drinking water
is essential to monitor, control, and optimize water treatment processes, and to illuminate the
biology of low nutrient water systems. Historically, laboratories have relied on heterotrophic pl.
Knowles, J.R., 1980. Enzyme-catalyzed phosphoryl transfer reactions. Annual Review of Biochemistry
49, 877–919. https://doi.org/10.1146/annurev.bi.49.070180.004305
Prest, E.I., Hammes, F., van Loosdrecht, M.C.M., Vrouwenvelder, J.S., 2016. Biological stability of
drinking water: controlling factors, methods, and challenges. Frontiers in Microbiology 7, 45–
45. https://doi.org/10.3389/fmicb.2016.00045
Riemann, B.O., 1979. The occurrence and ecological importance of dissolved ATP in fresh water.
Freshwater Biology 9, 481–490. https://doi.org/10.1111/j.1365-2427.1979.tb01532.x
Salinas-Rodriguez, S.G., Mangal, M.N., Villacorte, L.O., Abushaban, A., 2021. Methods for Assessing
Fouling and Scaling of Saline Water in Membrane-Based Desalination, in: Removal of Pollutants
from Saline Water. CRC Press.
Van der Kooij, Albrechtsen H., D., Corfitzen, C., Ashworth, J., Parry, I., Enkiri, F., Hambsch, B.,
Hametner, C., Kloiber, R., H., Veenendaal., 2003. Assessment of the microbial growth support
potential of products in contact with drinking water. CPDW Project: European Commission
Joint Research Centre.

353
Experimental Methods for Membrane Applications

Villacorte, L.O., 2014. Algal blooms and membrane based desalination technology. Environmental
engineering and water technology department. Delft University of Technolog, CRC Press/
Balkema.
Vrouwenvelder, J.S., Manolarakis, S.A., Veenendaal, H.R., Van der Kooij, D., 2000. Biofouling potential
of chemicals used for scale control in RO and NF membranes. Desalination 132, 1–10.
Wang, Q., Tao, T., Xin, K., Li, S., Zhang, W., 2014. A review research of assimilable organic carbon
bioassay. Desalination and Water Treatment 52, 2734–2740. https://doi.org/10.1080/1944
3994.2013.830683
Weinrich, L.A., Jjemba, P.K., Giraldo, E., LeChevallier, M.W., 2010. Implications of organic carbon in
the deterioration of water quality in reclaimed water distribution systems. Water Research 44,
5367–5375. https://doi.org/10.1016/j.watres.2010.06.035

354
 doi: 10.2166/9781789062977_0355

Chapter 16

Assessing Biological Stability

of Ultra-low Nutrient Water by
Measuring Bacterial Growth
Potential
Mohaned Sousi, IHE Delft, The Netherlands

The learning objectives of this chapter are the following:

• Define biological stability of drinking water and the factors influencing it

• Present and discuss the method for measuring bacterial growth potential of ultra-
low nutrient drinking water

• Present a case study for applying the developed bacterial growth potential method.

16.1 INTRODUCTION

Water utilities do not only aim at producing high quality drinking water that is safe for
human consumption, but also maintaining this quality during distribution until water
reaches the consumer’s tap. Water distribution systems provide a complex environment
for bacterial growth either in the form of planktonic bacteria or biofilm (Liu et al., 2013),
where such systems usually contain pipes of a wide range of diameters and made of various
materials, in addition to the fluctuating water flow throughout the day. Several studies have
shown that bacterial growth could take place in water distribution systems whether residual
chlorine was added or not (Prest et al., 2016b).

The concept of biological stability of drinking water appeared by the end of the twentieth
century, where the issue of bacterial growth in water distribution networks has gained
increasing attention. Biologically stable drinking water does not promote excessive bacterial
growth in distribution systems and until water reaches the consumption point (van der
Kooij and Veenendaal, 2014; Liu et al., 2013; Huck, 1990). Several serious problems are

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

associated with biologically unstable water, including threats posed to human health due
to the growth of (opportunistic) pathogens, deterioration of the aesthetic aspects of water
(taste, odour, and colour), and operational problems related to bio-corrosion of pipes and
fittings (Volk and LeChevallier, 1999; Berry et al., 2006). Biologically stable drinking water
can be achieved by applying a multi-barrier treatment strategy to reduce the concentration
of nutrients that promote bacterial growth in water (Smeets et al., 2009; van der Kooij and
Veenendaal, 2014).

Biological stability of drinking water is traditionally assessed with the presumption that
a small fraction of organic carbon is promoting bacterial growth. Several laboratory-based
methods were developed to measure this fraction of organic carbon, namely assimilable
organic carbon (AOC) method (van der Kooij et al., 1982; van der Kooij and Hijnen, 1984)
and biodegradable dissolved organic carbon (BDOC) method (Servais et al., 1987). One of
the main disadvantages of these traditional methods is the use of pure bacterial strains that
might not completely consume the available organic carbon present in water, resulting in
underestimation of bacterial growth that could take place in distribution systems where
diverse bacterial strains are present.

Major developments in the field of microbiological methods have occurred in the past years,
allowing for rapid, less laborious, and more accurate measurements of bacteria in water
samples compared with the traditional HPC method (van Nevel et al., 2017). One of these
major developments is flow cytometry (FCM) which can be coupled with DNA staining to
enable complete enumeration of bacterial cells in a water sample (Prest et al., 2016a). FCM
has been applied for biological stability assessment to measure AOC using natural bacterial
consortium (Hammes and Egli, 2005), allowing for more accurate estimation of AOC.
Additionally, FCM has been used for direct measurement of bacterial growth potential (BGP)
of water expressed as cell count (cells/mL) (Sousi et al., 2020a; Nescerecka et al., 2018;
Farhat et al., 2018; Prest et al., 2016a), without converting the obtained growth into AOC.
BGP of water can also be measured with adenosine triphosphate (ATP) as an alternative
bacterial parameter (Abushaban et al., 2019; Vital et al., 2012; Farhat et al., 2018). In
addition, combining BGP measurements with 16S rRNA gene sequencing enables in-depth
understanding of bacterial growth characteristics of water during treatment and distribution
(Liu et al., 2020; Liu et al., 2018; Li et al., 2017).

Current methods to assess bacterial growth potential are suitable for drinking water
that is produced using conventional water treatment technologies such as coagulation,
flocculation, sedimentation, (rapid or slow) sand filtration and activated carbon filtration.
However, this chapter focuses on assessing bacterial growth potential of water with ultra-
low nutrient content produced by advan0ced water purification systems, such as reverse
osmosis (RO). Post-treatment is a key process to make RO permeate potable and suitable
for distribution. The process of re-adding the essential minerals to RO permeate is called
remineralisation, which can be conducted using several methods, including: blending RO
permeate with source water, direct dosing of chemicals, calcite contactors, and micronized
calcite dosing. Remineralisation by calcite contactors is widely applied in practice since it
is a simple and cost-efficient method, where RO permeate percolates through a calcite bed
to dissolve calcium carbonate into calcium (Ca2+) ions and hydrogen carbonate (HCO3-)
(Ruggieri et al., 2008; Hasson and Bendrihem, 2006).

356
 Chapter 16

Research has shown that bacterial growth-promoting nutrients could be considerably

removed by RO filtration, resulting in a very low level of bacterial growth in RO permeate
(Escobar et al., 2000; Thayanukul et al., 2013; Park and Hu, 2010; Dixon et al., 2012).
However, this may be influenced by post-treatment, more specifically by remineralisation
since it involves the addition of substances to RO permeate.

16.2 MATERIALS AND EXPERIMENTAL SET-UP

Bacterial regrowth potential (BGP) method is applied to assess biological stability in drinking
water. This method implies monitoring of total and/or intact bacterial counts in water
samples over time using flow cytometry concept (FCM). To perform BGP measurements,
the following materials and equipment are required:

16.2.1 Equipment

A. Flow cytometer (FCM)

Flow cytometry (FCM) coupled with fluorescent staining has emerged as a leading tool for
single-cell analysis in microbiology. It is used as a rapid and accurate enumeration tool for
total bacteria or specific bacterial groups in water samples. Figure 1 describes the concept of
flow cytometry.

Sample Bandpass filters

(taken op from tube or well)

Photomultiplier tubes

Forward scatter detector

Monochromatic Side scatter detector

light source
Dichroic mirrors

Figure 1 Flow cytometry principle (Safford and Bischel, 2019)

The labelled cells with a fluorochrome are transported through a flow cell where they are
subjected to a laser beam of a specific wavelength. The flow cell is designed to allow only one
fluorescently labelled cell to pass through the laser beam at a time where it gets excited and
emits light which is captured by the detectors. The data gathered can be analysed statistically
by flow cytometry software.

FCM measurements are performed using a BD Accuri C6® flow cytometer equipped with
a 50 mW laser emitting at a fixed wavelength of 488 nm. Green fluorescence intensity is
collected in FL1 channel (533 ± 30 nm) and red fluorescence is collected in FL3 channel
(> 670 nm), while sideward and forward scattered light intensities were collected as well.
All parameters are collected in logarithmic signals. The FCM is equipped with volumetric
counting hardware, calibrated to measure the number of particles in 50 μL of a 500 μL

357
Experimental Methods for Membrane Applications

sample. Measurements were performed at pre-set flow rate of 35 μL/min. The BD Accuri
CFlow® software is used to process all data. An electronic gate on green/red fluorescence
density plots was used to distinguish stained microbial cells from instrument noise or water
sample background.

The following chemicals are used for FCM calibration and cells staining.

1- Partec calibration beads, 3 µm (for daily calibration).

Calibration beads for laser position with blue 488 nm emission, which is used for daily
control before starting with the actual measurements to check the performance of the FCM
and to give an indication if the maintenance is needed. According to the product supplier,
5 times diluted Partec calibration bead should give a count of 55,000 Events/mL on a
specific provided gate with 10% standard deviation (the recommended range is between
50,000 - 60,000 Events/mL):
a) Shake Partec calibration beads bottle gently before pipetting 100 μL into 400 μL of
filtered Evian water in an eppendorf 1.5 mL plastic vial (500 μL total sample volume of
5 times diluted beads sample).
b) Vortex gently the vial to mix its content and then fix it under the flow cytometry SIP.
c) Run the 500 μL of the bead control solution on medium speed and threshold of 500 on
Fl1 using the gate template provided by the producer.

2- Spherotech 8-peak validation beads (for calibration after maintenance works):

Concentration: 1 × 107 particles/mL
Storage buffer: deionized water with 0.02% sodium Azide and 0.01% NP40
Storage temperature: 2-8 °C, expirers 1 year after opening
Use: routine validation of the flow cytometer performance.

Description: 8-peak validation beads product contains Rainbow particles (3.0-3.4 μm)
with 8 different fluorescent intensities. Each Rainbow particle contains a mixture of
fluorophores that enable their excitation with the blue laser (488 nm) to validate FL1, FL2
and FL3 channels of the flow cytometer.

Procedures: the beads sample should be prepared and measured as follows:

a) Prepare the beads solution by diluting 4 drops of 8-peaks bead in 1 mL of 0.22 μm filtered
Evian water and diluting 4 drops of 6-peaks/peak 1 beads and 4 drops of the 6-peaks/
peak 2-6 beads in 1 mL of 0.22 μm filtered Evian water.
b) Pipette 1 mL of the 8-peaks calibration bead solution in an eppendorf 1.5 mL plastic
vial and vortex it properly before fixing it under the flow cytometry SIP and starting the
measurement.
c) Run the beads solution with the settings of 50,000 Events run limit and slow speed.
d) Wipe the SIP and proceed the same way with the prepared 6-peaks beads solution.
e) Run 0. 22 μm filtered Evian water at the end of the procedure.
f) Check the results by counting the peaks numbers.

Recommended range: see Figure 2.

358
 Chapter 16

Plot 2 Plot 6 Plot 3

105.9
6,000 6,000

4,000 4,000

2,000 2,000

Count

Count
SSC-H

M1 M2
R1
12.3% 10.9%
84.5%
105.4 0 0
105.6 106 101 102 103 104 105 106 107.2 101 102 103 104 105 106 107.2
FSC-H Green-H Orange-H

Figure 2 recommended range for 8-peak validation beads (own data)

3- Spherotech 6-peak validation beads (for calibration after maintenance works):

Concentration: 5 × 106 particles/mL
Storage buffer: deionized water with 0.02% sodium Azide and 0.01% NP40
Storage temperature: 2-8 °C, expirers 1 year after opening
Use: routine validation of the flow cytometer performance.

Description: 6-peak validation beads product contains a mixed population of 3.2 μm

particles in six different fluorescent intensities. The particles can be excited at wavelengths
in the 600-650 nm range.

Procedures: see 8-peak validation beads.

Recommended range: see Figure 3.

Plot 1 Plot 2 Plot 3

Gate No Gating Gate (R1 in all) Gate (R1 in all)
106.3 2,000 2,000

1,500 1,500
Count

Count
SSC-H

1,000 1,000

500 500
M1 M2
0.0% 0.0%
105.2 0 0
10 5.4
10 6.0 6.2
10 101 102 103 104 105 106 107.2 101 102 103 104 105 106 107.2
FSC-H FL1-H FL2-H
Plot 4 Plot 5 Plot 6
Gate (R1 in all) Gate No Gating Gate (R2 in all)
2,000 106.2 2,000

1,500 1,500
Count

Count
SSC-H

1,000 1,000

500 500
M3 M4
0.0% 16%
0 105.4 0
1 2 3
10 10 10 10 10 10 10 4 5 6 7.2
10 5.5
106.0
10 6.2
101 102 103 104 105 106 107.2
FL3-H FSC-H FL4-H

Figure 3 recommended range for 6-peak validation beads (own data)

359
Experimental Methods for Membrane Applications

4- Spherotech AccuCount fluorescent particles, 7.0-7.9 µm (for weekly calibration).

Concentration: 102,000 beads/mL
Storage buffer: 0.016 M PBS, pH 7.4 with 0.02% Sodium Azide and 0.2% BSA
Storage temperature: 2-8 °C, expirers 3 years from date of manufacturing
Use: to check the counting accuracy of the flow cytometer on monthly basis (can be used
more frequently if needed).

Description: AccuCount validation beads product is designed to be used as stand-alone,

quality control reagent to validate accurate volume measurement by the flow cytometer
running with CFlow Plus software.

Procedures: the beads sample should be prepared and measured as follows:

a) Prepare 10 times diluted spherotech beads solution in an eppendorf 1.5 mL plastic vial
(500 μL total sample volume of 450 μL filtered Evan water and 50 μL spherotech beads
stock), the sample should be at room temperature.
b) Vortex the eppendorf plastic tube gently and then invert it a few times before putting it
under the machine SIP.
c) Run the analysis of the sample with the following settings: run limits = 50 μL on medium
speed.
d) Repeat steps b and c two more times with the same eppendorf plastic tube (the same
sample will be analyzed in triplicate).
e) Check the count on Q1-UR gate and perform the calculations.

Recommended range: the difference between the obtained average count (the beads to be
counted in triplicate every time) and the expected count of 102,000 events/mL should not
exceed 10 %.

5- SYBR Green I (SG) for total cells counts.

SYBR Green I Nucleic Acid Gel Stain, 10,000 times concentrated in DMSO Invitrogen
(Molecular Probes), was used.

A 1 mL working SG stock solution is prepared by diluting the Invitrogen 10,000 times

concentrate SG solution 100 times in 0.22 μm filtered DMSO (IC Millex – LG, 0.2 μm,
Millipore), where 10 μL of SG concentrate solution was added to 990 μL of filtered DMSO.
The prepared stock solutions can be kept at -20 °C for future measurements. The working
SG stock solution can be used for total cell count (TCC) using FCM.

6- SYBR Green I + Propidium iodide (SG+PI) for intact cells counts

A mix of SYBR Green I (SG) stain (described above) and Propidium iodide (PI) stain.

Propidium Iodide (PI) is a ready-to-use stock solution for the exclusion of nonviable cells
in flow cytometry analysis. PI binds to double stranded DNA by intercalating between base
pairs, but is excluded from cells with intact plasma membranes.

A working SG + PI stock solution is prepared by adding 10 μL of Invitrogen 10,000 times

concentrate SG solution and 10 μL of PI ready-to-use stock solution in 980 μL of 0.22
μm filtered DMSO (IC Millex – LG, 0.2 μm, Millipore), so that the working SG + PI stock

360
 Chapter 16

solution contains 1:100 diluted SG and 0.3 mM of PI. The prepared stock solutions can be
kept at -20 °C for future measurements. The working SG + PI stock solution can be used for
intact cell count (ICC) using FCM.

B. Autoclave
An autoclave is a pressure chamber which is used to sterilize water samples and supplies
by subjecting them to a high pressure saturated steam at 121 °C for 15 to 20 minutes
depending on the size of the load and the contents.

C. Water bath
Water bath is a system to control water temperature in where bottles containing water
samples are placed for wet-pasteurization.

16.2.2 Materials and Methods

A. Glassware (Figure 4):
• Duran® graduated clear glass bottles with screw solid plastic cap for sampling.
• Clear glass vials with screw silicone septum cap for incubating.
B. Eppendorf plastic tubes of 1.5 mL.
C. Plastic syringes.
D. Pipettes and plastic tips of 1 mL, 200 μL and 5 μL.
E. Stopwatch.

Figure 4 Glassware used for bacterial growth potential measurement (Own photos)

16.2.3 Method
1. Preparation of glassware: Glassware should be made AOC-free by applying the following
cleaning protocol in a clean laboratory environment to avoid air-borne contamination
(Hammes et al., 2006).

A. Rinse the glassware and the caps with a cleaning solution (Alconox, 10 g/L in demi
water) with brushing the internal side thoroughly.
B. Rinse it 3 times with Milli-Q water and leave it to air dry.
C. Put all the glassware (excluding the plastic caps) inside the muffle furnace. Glass bottles
and vials should be wrapped with aluminium foil to avoid contamination.
D. Set the muffle furnace at 550 °C, the glassware stays inside for 24 hours as follows:

361
Experimental Methods for Membrane Applications

• 1.5 hours for the furnace to warm up and to reach the targeted temperature.
• 6 hours of effective exposure to heat at the targeted temperature.
• The rest of the time is for cooling down the temperature.
E. Clean the plastic caps by soaking them in heated (60 °C) sodium persulfate (Na2S2O8,
100g/L) in water bath for 1 hour, thereafter rinse them 3 times with Milli-Q water and
leave them to air dry.
F. After taking the cleaned glassware out of the muffle furnace, close them with the cleaned
caps and keep them in a clean and closed place away from contamination.

2. Preparation of stock solutions: For blank preparation, four different inorganic stock
solutions are prepared in AOC-free bottles using ultrapure water (i.e., Milli-Q water)
with final concentrations of 67.2 g/L NaHCO3 (for pH adjustment and buffer addition),
combined 294 g/L CaCl2.2H2O and 67 g/L MgCl2·6H2O (for calcium and magnesium
addition), 0.219 g/L KH2PO4 (for phosphate addition), and 3.607 g/L KNO3 (for nitrogen
addition). Additionally, organic stock solutions of 1,000 ± 50 mg-C/L are prepared using
Milli-Q water in AOC-free bottles for researching bacterial inocula, as follows: sodium
acetate and glucose representing readily-available organic carbon, and laminarin (from
Laminaria digitata) and gelatin (type B, from bovine skin) representing complex organic
carbon. The prepared stock solutions were kept in the fridge at 4 °C and were used for
multiple experiments. Reagent grade chemicals (>99% purity) were used throughout this
study (J.T.Baker® Reagents Salts, ACS Grade, the USA).

3. Preparation of blanks: An ultra-pure blank was prepared by adjusting the pH and

mineral content of RO permeate at the laboratory to 122 mg/L HCO3− (final pH of
7.8 ± 0.2), 40 mg/L Ca , 4 mg/L Mg2+, 5 μg/L PO4-P, and 50 μg-N/L. These concentrations
were obtained by the addition of 2.5 μL/mL of NaHCO3, 0.5 μL/mL of CaCl2 and MgCl2
and 0.1 μL/mL of both KH2PO4 and KNO3 stock solutions. The BGP of ultra-pure blank
(laboratory-remineralised RO permeate) is in the range of 50 ± 20 × 103 ICC/mL (ICC:
intact cell counts as measured by FCM). For chemical addition, pipettes are used after rinsing
sterilised plastic tips 10 times with ultra-pure water to prevent AOC leaching into the water
samples.

Moreover, a broth of trace elements can be used for growth limitation experiments, where
two stock solutions should be prepared (pH ~7): stock solution A containing 5 mg/L
CoCl2.6H2O and 10 mg/L H3BO3; and stock solution B containing 500 mg/L MnSO4.7H2O,
10 mg/L ZnSO4.7H2O, and 300 mg/L FeSO4.7H2O. The stock solutions should be kept in
the dark at room temperature. Aliquots of 4 and 3.7 mL/L from stock solutions A and B,
respectively, were added in water samples, resulting in final concentrations of 5 μg/L Co,
6.5 μg/L B, 359 μg/L Mn, 8.5 μg/L Zn, 215 μg/L Fe, and 345 μg/L S. Moreover, adding
phosphate and nitrogen was accompanied with the addition of 29.2 μg/L K.

The ultra-pure blank is made carbon-limited to ensure detecting any potential carbon
contaminations during the handling of samples, which comes in different forms such as:
(i) carbon attached to the glassware and caps, (ii) volatile carbon present in the laboratory
environment, and (iii) carbon contamination present in reagent grade chemicals used in the
laboratory, in addition to the original carbon content of the blanks.

362
 Chapter 16

4. Choice of inoculum: the following steps are applied to test the suitability of a
certain bacterial inoculum for BGP measurement, where a natural bacterial inoculum is
recommended to ensure the consumption of available organic compounds to a large extent:
A. Prepare sample for BGP measurement as shown in part 6 of this section.
B. After inoculating with the target bacterial consortium, add the different organic stock
solutions mentioned in part 2 of this section in separate test bottles, considering the
following final concentrations per organic stock solution: 0, 10, 20, 50, 100, and 200
μg-C/L.
C. Perform the above-mentioned test procedure using different bacterial consortium, if
available.
D. Select the bacterial inoculum that yields the maximum BGP in all organic carbon type.

5. Sampling procedures: water samples are collected as follows:

E. Open sampling tap for at least 10 minutes to flush the piping before sampling.
F. Collect 200 mL of the water sample in an AOC-free Duran® graduated clear glass bottle.
G. Keep the sampling bottle in a cooling box away from potential contaminations (Note:
prevent any contact between water sample and cap by handling the bottle gently).
H. Samples are transported to the laboratory within 3 hours.

6. Bacterial growth potential (BGP) test procedure: water samples are pre-treated,
inoculated, and measured at the laboratory, as follows:
A. Pre-treat water samples by pasteurisation at 70 °C using a water bath to inactivate
indigenous bacteria. Water level inside sampling bottle and in water bath should be
comparable to ensure effective pasteurisation. Sampling bottle should be air-tight to avoid
contamination with water vapour. Effective pasteurization time is 30 min excluding the
time needed for water sample to reach the target temperature (the warming up time is
determined using a reference sample with similar volume to actual water sample).
B. Put them in an ice bath to cool down to room temperature.
C. Inoculated them with ~104 ICC/mL of a natural bacteria consortium originating from
water that contains diverse bacterial species, e.g., tap water. For inoculation, pipettes are
used after rinsing sterilised plastic tips (at 121 °C for 15 minutes) 10 times with ultra-
pure water to prevent AOC leaching into the water samples.
D. Distribute each water sample in three individual AOC-free vials (i.e., triplicate
measurements per sample) by direct pouring from sampling bottle into the vials. Use a
reference vial for volume measurement (20 ± 2 mL sample per vial).
E. Incubate the vials in the dark at 30 °C under static conditions during a test period of 20
days.
F. Aliquots of ~1 mL are directly poured from the incubated vials into 1.5 mL Eppendorf
tubes to perform FCM measurement, either intact cell count (ICC), total cell count (TCC),
or both. FCM measurements are performed daily in the first test week, and then every
two days. The FCM measurements are performed as follows:
• Pipette 500 μL of the sample in 1.5 mL eppendorf plastic tube. Samples have to be
diluted using 0.1 μm filtered bottle water (e.g., Evian) at >200,000 TCC or ICC/mL
(dilution rates are 5, 10 or 20). For instance, to achieve 5 times dilution, 400 μL of
filtered bottle water should be added to 100 μL of sample.
• Preheat the sample in dark to 35°C for 5 min.

363
Experimental Methods for Membrane Applications

• Stain the sample by adding 5 μL of SG (for TCC) or SG + PI (for ICC). Dilution ratio is
1:100.
• Incubate the stained sample in dark at 35°C for exactly 10 minutes.
• Run the FCM at the following settings: run limit: 50 μL (identical settings to
volumetric calibration of FCM); fluidics: medium speed; and threshold on FL1-H
channel of green fluorescence to 700.
• Data acquisition was performed using BD Accuri CFlow® software where a digital
gate was set on FL1/FL3 density plot to distinguish the stained bacterial cells from
inorganic particles and instrument noise. The FCM detection limit is 103 TCC or ICC/
mL. Figure 5 shows an example of data acquisition using FL1/FL3 density plot.
G. Measuring each incubating vial for TCC and/or ICC implies that triplicate measurements
per sample are performed. Statistical analysis based on Dixon’s Q-test is conducted to
define the outlier. Thereafter, average TCC and/or ICC is reported per measurement
day and standard deviation is presented as an error bar. Student’s t-test and one-way
analysis of variance (ANOVA) were applied to compare BGP of different sample types. A
confidence level of 95 % was considered (alpha of 0.05).
H. BGP results are expressed as the maximum TCC or ICC obtained during the whole test
period of 20 days.

Figure 6 represents the complete sample handling and BGP test procedure.

Background Gate
Typically inorganic particles The lower limit for bacteria is 2,000 on FL1.
The gate may be extended doagonally to
the right in case of large bright cells.

104.5

104

Background
Typically non-bacterial organic
particles and free DNA
103

P1
69.0%
102
101.7 FL3-A
2.8 4 5 5.6
10 10 10 10
FL1-A

LNA Bacteria HNA Bacteria

Bacteria with low nucleic acid content are Bacteria with high nucleic acid content are
small and stain weakly with SYBR®Green1 large and stain brightly with SYBR®Green1

Figure 5 An example of data acquisition using FCM BD Accuri CFlow® software (FL1/FL3
density plot) (own data)

16.3 EXAMPLES OF APPLICATION

This method can be applied in studies about water treatment plants that are based on
advanced technologies to produce ultra-low nutrient water, such as, membrane technology
(e.g., RO filtration) and distillation. RO filtration is a superior barrier for bacteria and

364
 Chapter 16

growth-promoting nutrients in water, as a result of which RO-treated has a high degree of

biological stability, which was the main application while developing the proposed method
in this chapter. The following paragraph shows an example of using this method in a case-
study in the Netherlands.

Sampling

Sample
pretreatment (70˚C, 30 minutes)
(pasteurisation)

Inoculating
with natural
bacterial

Dividing
sample in
triplicate

Incubation
at 30˚

maximum growth

Die-off
Exponential growth
Cell counts

Cell count using

Flow Cytometry
Lag phase

Time (days)

Figure 6 illustration of the steps to follow for measuring BGP of ultra-pure water

The Kamerik drinking water treatment plant (Oasen Drinking Water Company, Gouda,
Netherlands) currently produces 340 m3/h of drinking water from anaerobic groundwater
(AGW) by conventional water treatment processes, which are given in Figure 8A in the
following order: spray aeration on the surface of rapid sand filters (so-called dry sand
filtration, DSF), tower aeration, pellet softening (SOF), carry-over submerged rapid sand
filtration (RSF), granular activated carbon filtration (ACF), and UV disinfection (UVD)
before storing the conventionally treated water (CTW) in the clean water reservoir.

Installed in parallel for research purposes, a pilot-scale advanced treatment line with a
capacity of 7 m3/h treats the same source water with the following processes (Figure 8B):
anaerobic RO filtration (RO) with a total recovery of 75%, followed by post-treatment
comprising anaerobic ion exchange (IEX) to remove residual ammonium, remineralisation

365
Experimental Methods for Membrane Applications

using anaerobic calcite contactors (CC) to correct the calcium and bicarbonate concentrations
to the required level (40 mg/L Ca2+, 122 mg/L HCO3-), magnesium dosing (MgCl2, 4 mg/L
Mg2+), and tower aeration for the introduction of oxygen and the removal of methane and
excess carbon dioxide. The finished drinking water after RO filtration and all post-treatment
processes is denoted as site-Remin and has a final pH of 7.8 ± 0.2. Water samples were
collected after each treatment step in both the conventional and RO-based treatment lines.
To identify bacterial growth-limiting nutrients, BGP of water samples was measured with
the addition of different combinations of nutrients as previously described by Prest et al.
(2016a), and shown in Table 1. The used nutrient stocks were carbon (1.07 g/L C2H3NaO2),
phosphate (0.219 g/L KH2PO4), nitrogen (3.61 g/L KNO3), and a broth of trace elements (5
mg/L CoCl2.6H2O, 10 mg/L H3BO3, 10 mg/L CaSO4.5H2O, 500 mg/L MnSO4.7H2O, 10
mg/L ZnSO4.7H2O, 300 mg/L FeSO4.7H2O). Nutrients were added according to the ratio
of C:N:P = 100:10:1 (Hammes and Egli, 2005). The blank (lab-Remin) and samples of the
finished drinking water produced by the RO-based and conventional treatment lines (site-
Remin and CTW, respectively) were tested.

Table 1 BGP test matrix to identify the bacterial growth-limiting nutrient in water samples
C P N
Test # (C2H3NaO2) (KH2PO4) (KNO3) TE* Investigation
1 – – – – actual BGP
2 – + + + C-limited BGP
3 + – + + P-limited BGP
4 + + – + N-limited BGP
5 + + + – TE-limited BGP
6 + + + + positive control

The profiling of the two treatment lines showed considerably different degrees of BGP and
nutrient removal. The conventional treatment line reduced the BGP by ~60% (from 1,250
± 100 ×103 in AGW to the range of 450 ×103 – 550 ×103 ICC/mL across the different
treatment steps), where the BGP of conventionally treated water (CTW) was 515 ± 5 ×103
ICC/mL (Figure 8a). Meanwhile, DOC decreased from 7.2 mg/L in AGW to 6.0 mg/L
in CTW (Table 2). Notably, the humic substances, which accounted for >70% of DOC in
AGW, showed the highest removal in the conventional treatment line (from 5.2 mg/L to
4.3 mg/L).

366
 Chapter 16

Anaerobic Dry sand Tower Softening Carry-over Activated UV Reservoir of

groundwater filtration aceration rapid sand carbon desinfection conventionally
filtration filtration treated water
Acid
(a) dosing

(DSF) (RSF) (ACF) (UVD) (CTW)

RO Ion Calcite Magnesium Tower

filtration exchange contactors dosing aceration

(b)

(AGW) (RO, (IEX) (CC) (site-Remin)

lab-Remin)

Figure 7 Full-scale conventional (a) and pilot-scale RO-based (b) water treatment lines at the
drinking water treatment plant. Sampling locations (dashed arrows) and codes (between
brackets) are indicated.

Phosphate was also considerably reduced, mainly during DSF (>98%, from 553 μg/L PO4-P
in AGW to 7 μg/L PO4-P in DSF), reaching down to 1 μg/L PO4-P in CTW (Table 2).
Similarly, ammonium was also reduced below 0.02 mg/L NH4-N (limit of detection) by
the conventional treatment (Table 2). The results showed that nitrification was the main
mechanism for ammonium removal, where ammonium (NH4+) in AGW (2.90 ± 0.10
mg/L NH4-N) was completely converted into nitrate (NO3-) in CTW (2.77 ± 0.40 mg/L
NO3-N). Methane, which was present at 2,000–4,000 μg-CH4/L in AGW, was reduced to
10–20 μg-CH4/L in CTW.

a) b)
BGPmax (intact cells/mL)

BGPmax (intact cells/mL)

1.5E+06 1.5E+06
1.2E+05 1.2E+05

9.0E+05 9.0E+05
6.0E+05 6.0E+05
3.0E+05 3.0E+05

0.0E+00 0.0E+00
AGW DSF SOF RSF ACF UVD CTW AGW lab- IEX CC site-
Remin Remin

Figure 8 Bacterial growth potential (BGP) at each step of the conventional (A) and RO-based
(B) treatment lines. BGPs of RO permeate and ion exchange effluent were measured
after remineralisation at the laboratory (i.e., lab-Remin and IEX respectively). Error bars
represent the standard deviation of triplicate measurements.

367
 Source

368
Treatment water Conventional line RO-based line
Table 2

Sample type AGW DSF RSF ACF CTW RO IEX CC site-

Remin
Carbon DOC 7,242 7,237 6,636 6,105 5,987 36 32 20 27
(µg-C/L)a
Bio-polymers 3 16 8 6 10 4 0 3 13
treatment lines

Humic Substances 5,202 5,170 4,610 4,486 4,323 0 0 0 0

Building Blocks 1,110 1,095 1,151 1,027 994 2 1 5 6
Neutrals 869 809 801 652 636 12 66 7 22
Acids 0 0 0 0 0 3 2 2 3
Nitrogen Ammonium 2.90 ± 0.82 ± <0.02 n.a. <0.02 0.17 ± <0.02 <0.02 <0.02
(mg-N/L)b (NH4+) 0.10 0.15 0.02
Nitrite (NO2-) <0.003 0.022 ± <0.003 n.a. <0.003 n.a. n.a. n.a. <0.003
0.004
Experimental Methods for Membrane Applications

Nitrate (NO3-) <0.23 n.a. n.a. n.a. 2.77 ± n.a. n.a. n.a. 0.23 ±
0.40 0.05
Methane 2,000– n.a. n.a. n.a. 10–20 n.a. n.a. n.a. <5–14
(µg-CH4/L) c 4,000
Phosphate 553 6.8 ± 0.6 1.1 ± 0.7 ± 1.1 ± 0.1 0.9 ± 0.9 ± 7.0 ± 7.3 ±
(µg/L PO4-P)d ±17 0.1 0.1 0.1 0.1 0.5 0.1
a The reporting limit is 100 µg-C/L for biopolymers and 200 µg-C/L for the other LC-OCD fractions
b Limit of detection: 0.02 mg-N/L for ammonium, 0.003 mg-N/L for nitrite, and 0.23 mg-N/L for nitrate
c Limit of detection: 5 µg-CH /L
4
d Limit of detection is 0.3 µg/L PO -P. The concentrations after SOF and UVD are 3.8 ± 0.1 and 1.1 + 0.1, respectively. n.a., not measured
4
The concentration of carbon (LC-OCD fractions), phosphate, nitrogen (ammonium,
nitrite, and nitrate), and methane at each step of the conventional and RO-based
 Chapter 16

The RO-based treatment showed a substantial BGP reduction (>96%) from ~1,250 ± 100
×103 ICC/mL in AGW to ~50 ± 12 ×103 ICC/mL in lab-Remin (i.e., RO permeate after
remineralisation at the laboratory, Figure 8b). However, the BGP increased by 160% after
remineralisation using calcite contactors (CC) and tower aeration (site-Remin), reaching
130 ± 10 ×103 ICC/mL. The LC-OCD analysis revealed that all organic matter fractions
were considerably retained by RO filtration to levels below the reporting limit (Table 2).

Despite the increase in BGP after post-treatment, there was no detectable increase in any
DOC fraction by the LC-OCD. For phosphate, a sharp decrease from 553 to 1 μg/L PO4-P
was observed after RO filtration, followed by an increase across the post-treatment to 7
μg/L PO4-P (Table 2). In contrast to conventional treatment, nitrification was insignificant
within the RO-based treatment line, where ammonium in AGW was mostly retained by
the RO membrane (0.17 ± 0.02 mg/L NH4-N in RO permeate), and was further removed by
absorption in ion exchange resins (<0.02 mg/L NH4-N). This resulted in a low concentration
of nitrate in RO-treated water (0.23 ± 0.05 mg/L NH4-N) (Table 2). Methane in RO-treated
water was at similar concentrations as in CTW.

The investigation of the growth-limiting nutrient (Figure 9) revealed that the growth in the
examined water types was limited by organic carbon. For all samples, the difference between
the actual BGP (i.e., without nutrient addition to the sample) and the C-limited BGP (i.e.,
samples spiked with all nutrients except for carbon) was insignificant (p > 0.05). Contrarily,
the BGP of samples with limited phosphate, nitrogen, and trace elements (Fe, Mn, Zn, Co,
and B) was significantly (p < 0.05) higher than the actual BGP of the corresponding sample.
Interestingly, the P-limited BGP of site-Remin was 50% higher than that of CTW and lab-
Remin.

a) b) c)

Combined limitation

–TE

–N

–P

–C

Actual BGP
IE+4

IE+5

IE+6

IE+7

IE+4

IE+5

IE+6

IE+7

IE+4

IE+5

IE+6

IE+7

BGPmax (ICC/mL)

Figure 9 BGP of lab-Remin (the blank, A), site-Remin (B), and CTW (C) with the addition of
different nutrients as given in Table 1 (Lower nutrient concentrations were added in
the case of lab-Remin and site-Remin). Actual BGP: no nutrients added, –C: no carbon
added, –P: no phosphate added, –N: no nitrogen added, –TE: no trace elements added,
Combined limitation: all nutrients added. Error bars represent the standard deviation of
triplicate tests.

369
Experimental Methods for Membrane Applications

The initial intact cell count and BGP of lab-Remin, site-Remin, and CTW were monitored
for a period of 2 years (Figure 10). The results demonstrated superior performance of the
RO-based treatment line, where the initial cell count of lab-Remin (<103 ICC/mL) and site-
Remin (25 × 103 – 200 × 103 ICC/mL) were systematically lower than that of CTW (400
× 103 – 600 × 103 ICC/mL). Similarly, the BGP was subsequently reduced by >75% with
the RO-based treatment line compared with the conventional one, where no pronounced
seasonal variations were observed and the BGP was stable around 35 × 103 – 60 × 103, 90 ×
103 – 150 × 103, and 500 × 103 – 700 × 103 ICC/mL for lab-Remin, site-Remin, and CTW,
respectively.

800,000
a)
600,000

400,000

200,000

Initial cell count (ICC/ML) 0

800,000
b)

600,000

400,000

200,000

BGPmax (ICC/ML) 0
oct. 2016
nov.
dec.
Jan. 2017
Feb.
Mar.
April
May
June
July
Aug.
Sep.
Oct.
Nov.
dec.
Jan. 2017
Feb.
Mar.
April
May
June
July
Aug.
Sep.
Oct.
Nov.
dec.
CTW
site-Remin
lab-Remin

Figure 10 Initial intact cell count (A) and maximum bacterial growth potential
(BGPmax, B) of lab-remineralised RO permeate (lab-Remin, the blank), site-remineralised
RO permeate (site-Remin), and conventionally treated water (CTW). All samples were
pasteurised (70 °C for 30 min) and inoculated with CTW. Error bars represent the
variations of triplicate measurements.

16.4 ADDITIONAL CONSIDERATIONS

This chapter is based on previously published data by Mohaned Sousi and co-authors in
his Ph.D. dissertation. The findings of his dissertation have also been published in peer-
reviewed journals (Sousi et al., 2018; Sousi et al., 2020a; Sousi et al., 2020b; Sousi et al.,
2021).

The proposed method in this chapter can be further developed, e.g., the limit of detection
of the BGP method might be further lowered by testing different types of water as a blank
other than RO permeate, for instance, distilled water produced at the laboratory under

370
 Chapter 16

controlled conditions. Moreover, further investigation on the impact of various steps in the
procedure can be carried out, such as glassware materials and methods of collecting aliquots
from incubated glassware. Another factor that could be critical for other types of ultra-low
nutrient water is the contamination caused by inoculation, even though this effect was
negligible for the inoculum concentration used in this dissertation. Preparing AOC-free
inoculum could be, therefore, considered, where regular validation is required by testing
the ability of this inoculum to consume readily available as well as complex organic carbon
in addition to performing 16S rRNA gene sequencing analysis for diversity control.

371
Experimental Methods for Membrane Applications

16.5 REFERENCES

Abushaban, A., Salinas-Rodriguez, S.G., Mangal, M.N., Mondal, S., Goueli, S.A., Knezev, A., et al.
(2019). ATP measurement in seawater reverse osmosis systems: Eliminating seawater matrix
effects using a filtration-based method. Desalination 453. https://doi.org/10.1016/j.
desal.2018.11.020
Berry, D., Xi, C., Raskin, L. (2006). Microbial ecology of drinking water distribution systems. Curr.
Opin. Biotechnol. 17:3. https://doi.org/10.1016/j.copbio.2006.05.007
Dixon, M.B., Qiu, T., Blaikie, M., Pelekani, C. (2012). The application of the bacterial regrowth potential
method and fow cytometry for biofouling detection at the Penneshaw desalination plant in
south Australia. Desalination 284. https://doi.org/10.1016/j.desal.2011.09.006
Escobar, I.C., Hong, S., Randall, A.A. (2000). Removal of assimilable organic carbon and biodegradable
dissolved organic carbon by reverse osmosis and nanofiltration membranes. Journal of Membrane
Science 175:1. https://doi.org/10.1016/S0376-7388(00)00398-7
Farhat, N., Hammes, F., Prest, E., Vrouwenvelder, J. (2018). A uniform bacterial growth potential assay
for different water types. Water Res. 142. https://doi.org/10.1016/j.watres.2018.06.010
Hammes, F., Salhi, E., Köster, O., Kaiser, H.-P., Egli, T., von Gunten, U. (2006). Mechanistic and kinetic
evaluation of organic disinfection by-product and assimilable organic carbon (AOC) formation
during the ozonation of drinking water. Water Res. 40:12. http://dx.doi.org/10.1016/j.
watres.2006.04.029
Hammes, F.A., Egli, T. (2005). New method for assimilable organic carbon determination using flow-
cytometric enumeration and a natural microbial consortium as inoculum. Environ. Sci. Technol.
39:9. https://doi.org/10.1021/es048277c
Hasson, D., Bendrihem, O. (2006). Modeling remineralization of desalinated water by limestone
dissolution. Desalination 190:1–3. http://dx.doi.org/10.1016/j.desal.2005.09.003
Huck, P.M. (1990). Measurement of biodegradable organic matter and bacterial growth potential in
drinking water. J. Am. Water Works Assoc. 82:7. https://doi.org/10.1002/j.1551-8833.1990.
tb06995.x
Li, Q., Yu, S., Li, L., Liu, G., Gu, Z., Liu, M., et al. (2017). Microbial communities shaped by treatment
processes in a drinking water treatment plant and their contribution and threat to drinking water
safety. Front. Microbiol. 8:2465. https://doi.org/10.3389/fmicb.2017.02465
Liu, G., Verberk, J.Q.J.C., van Dijk, J.C. (2013). Bacteriology of drinking water distribution systems:
An integral and multidimensional review. Appl Microbiol Biotechnol 97:21. https://doi.
org/10.1007/s00253-013-5217-y
Liu, G., Zhang, Y., Liu, X., Hammes, F., Liu, W.-T., Medema, G., et al. (2020). 360-degree distribution
of biofilm quantity and community in an operational unchlorinated drinking water distribution
pipe. Environ. Sci. Technol. 54:9. https://doi.org/10.1021/acs.est.9b06603
Liu, G., Zhang, Y., van der Mark, E., Magic-Knezev, A., Pinto, A., van den Bogert, B., et al. (2018).
Assessing the origin of bacteria in tap water and distribution system in an unchlorinated drinking
water system by SourceTracker using microbial community fingerprints. Water Res. 138.
https://doi.org/10.1016/j.watres.2018.03.043
Nescerecka, A., Juhna, T., Hammes, F. (2018). Identifying the underlying causes of biological instability
in a full-scale drinking water supply system. Water Res. 135. https://doi.org/10.1016/j.
watres.2018.02.006
Park, S.K., Hu, J.Y. (2010). Assessment of the extent of bacterial growth in reverse osmosis system for
improving drinking water quality. J. Environ. Sci. Health, Part A: Tox. Hazard. Subst. Environ.
Eng. 45:8. https://doi.org/10.1080/10934521003772386

372
 Chapter 16

Prest, E.I., Hammes, F., Kötzsch, S., van Loosdrecht, M.C.M., Vrouwenvelder, J.S. (2016a). A systematic
approach for the assessment of bacterial growth-controlling factors linked to biological stability
of drinking water in distribution systems. Water Sci. Technol. Water Supply 16:4. https://doi.
org/10.2166/ws.2016.001
Prest, E.I., Hammes, F., van Loosdrecht, M.C.M., Vrouwenvelder, J.S. (2016b). Biological stability of
drinking water: Controlling factors, methods, and challenges. Front. Microbiol. 7:45. https://
doi.org/10.3389/fmicb.2016.00045
Ruggieri, F., Fernandez-Turiel, J.L., Gimeno, D., Valero, F., García, J.C., Medina, M.E. (2008). Limestone
selection criteria for EDR water remineralization. Desalination 227:1–3. http://dx.doi.
org/10.1016/j.desal.2007.07.020
Safford, H.R., Bischel, H.N. (2019). Flow cytometry applications in water treatment, distribution, and
reuse: A review. Water Res. 151. https://doi.org/10.1016/j.watres.2018.12.016
Servais, P., Billen, G., Hascoët, M.C. (1987). Determination of the biodegradable fraction of dissolved
organic matter in waters. Water Res. 21:4. https://doi.org/10.1016/0043-1354(87)90192-8
Smeets, P.W.M.H., Medema, G.J., van Dijk, J.C. (2009). The Dutch secret: How to provide safe
drinking water without chlorine in the Netherlands. Drink. Water Eng. Sci. 2:1. https://doi.
org/10.5194/dwes-2-1-2009
Sousi, M., Liu, G., Salinas-Rodriguez, S.G., Chen, L., Dusseldorp, J., Wessels, P., et al. (2020a). Multi-
parametric assessment of biological stability of drinking water produced from groundwater:
Reverse osmosis vs. conventional treatment. Water Res. 186:116317. https://doi.
org/10.1016/j.watres.2020.116317
Sousi, M., Liu, G., Salinas-Rodriguez, S.G., Knezev, A., Blankert, B., Schippers, J.C., et al. (2018). Further
developing the bacterial growth potential method for ultra-pure drinking water produced by
remineralization of reverse osmosis permeate. Water Res. 145. https://doi.org/10.1016/j.
watres.2018.09.002
Sousi, M., Salinas-Rodriguez, S.G., Liu, G., Dusseldorp, J., Kemperman, A.J.B., Schippers, J.C., et al.
(2021). Comparing the bacterial growth potential of ultra-low nutrient drinking water assessed
by growth tests based on flow cytometric intact cell count versus adenosine triphosphate. Water
Res. https://doi.org/10.1016/j.watres.2021.117506
Sousi, M., Salinas-Rodriguez, S.G., Liu, G., Schippers, J.C., Kennedy, M.D., van der Meer, W. (2020b).
Measuring bacterial growth potential of ultra-low nutrient drinking water produced by reverse
osmosis: Effect of sample pre-treatment and bacterial inoculum. Front. Microbiol. 11:791.
https://doi.org/10.3389/fmicb.2020.00791
Thayanukul, P., Kurisu, F., Kasuga, I., Furumai, H. (2013). Evaluation of microbial regrowth potential
by assimilable organic carbon in various reclaimed water and distribution systems. Water Res.
47:1. https://doi.org/10.1016/j.watres.2012.09.051
van der Kooij, D., Hijnen, W.A.M. (1984). Substrate utilization by an oxalate-consuming Spirillum
species in relation to its growth in ozonated water. Appl. Environ. Microbiol. 47:3.
van der Kooij, D., Veenendaal, H.R. (2014). “Regrowth problems and biostability assessment in the
Netherlands” in Microbial growth in drinking-water supplies: Problems, causes, control and
research needs, ed. D. van der Kooij, P.W.J.J. van der Wielen (London: IWA Publishing.), 291–
337.
van der Kooij, D., Visser, A., Hijnen, W.A.M. (1982). Determining the concentration of easily
assimilable organic carbon in drinking water. J. Am. Water Works Assoc. 74:10. https://doi.
org/10.1002/j.1551-8833.1982.tb05000.x

373
Experimental Methods for Membrane Applications

van Nevel, S., Koetzsch, S., Proctor, C.R., Besmer, M.D., Prest, E.I., Vrouwenvelder, J.S., et al. (2017).
Flow cytometric bacterial cell counts challenge conventional heterotrophic plate counts for
routine microbiological drinking water monitoring. Water Res. 113. https://doi.org/10.1016/j.
watres.2017.01.065
Vital, M., Dignum, M., Magic-Knezev, A., Ross, P., Rietveld, L., Hammes, F. (2012). Flow cytometry
and adenosine tri-phosphate analysis: Alternative possibilities to evaluate major bacteriological
changes in drinking water treatment and distribution systems. Water Res. 46:15. https://doi.
org/10.1016/j.watres.2012.06.010
Volk, C.J., LeChevallier, M.W. (1999). Impacts of the reduction of nutrient levels on bacterial water
quality in distribution systems. Appl. Environ. Microbiol. 65:11.

374
 doi: 10.2166/9781789062977_0375

Chapter 17

Optical Coherence Tomography

(OCT) as a Tool for
(Bio)-fouling Assessment
in Desalination Systems
Johannes S. Vrouwenvelder, KAUST, Saudi Arabia

Luca Fortunato, KAUST, Saudi Arabia

The learning objectives of this chapter are the following:

• Present and discuss the use of Optical Coherence Tomography in membrane-based

desalination systems

• Monitor biofilm formation non-invasively in-situ in spacer filled channels

• Define and apply fouling descriptors to quantify biofouling development by using

Optical Coherence Tomography

• Understand the impact of biofouling on performance decline in spacer filled

channels.

This chapter is based, with permission from copyright holder, on two papers previously
published in Journal of Membrane Science Volume 524, 15 February 2017, Pages 673 doi:
10.1016/j.memsci.2016.11.052 and in Bioresource Technology Volume 229, April 2017,
Pages 231-235 doi: 10.1016/j.biortech.2017.01.021

17.1 INTRODUCTION

In the last decades the use of membrane filtration to produce high quality drinking water
has increased. One of the major problems of membrane filtration systems is biofouling
(Ridgway and Flemming, 1996; Vrouwenvelder et al., 2008a). Biofilm formation is caused
by the accumulation of microorganisms, including extracellular polymeric substances

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment,
Sergio G. Salinas-Rodriguez, Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

(EPS) produced by microorganisms, on a surface due to either deposition and/or growth. A

biofilm causing an unacceptable decline in membrane performance is defined as biofouling.
Performance losses are caused by increase in feed channel pressure drop, permeate flux
reduction and/or salt passage (Matin et al., 2011).

The complex configuration of the spiral-wound membrane modules makes it difficult to

study biofouling in-situ. Lab-scale monitors have been developed to allow easier access and
better analyses of biofilm development in spiral wound membrane modules (Flemming,
2003; Hemming et al., 1998). Membrane fouling simulator (MFS) was proved to be a suitable
tool to study biofouling in spiral wound membrane systems (Vrouwenvelder et al., 2006).

A key aspect of biomass studies involves the analysis of biomass structure (Halan et al.,
2012), which can predict the biomass behavior, and thus, the impact on membrane filtration
performance. Several approaches are reported in literature to study biomass, most often
involving destructive methods (Flemming and Wingender, 2010; Herzberg and Elimelech,
2007). Microscopic techniques are considered an important tool for biomass structure
investigation. However, these techniques involve sample preparation, and are less suitable
to study the biomass development in-situ.

To better understand the biomass development in membrane systems, in-situ qualitative

and quantitative analyses of the biomass under operational conditions are needed). Several
techniques are currently available to study the biomass formation under membrane
operational conditions, such as nuclear magnetic resonance spectroscopy (NMR), planar
optodes and optical coherence tomography (OCT) (Valladares Linares et al., 2016).

Optical coherence tomography (OCT) can investigate biomass formation and 3D structure
in-situ, without any biomass staining procedures. The OCT has been used to study
biofouling in membrane filtration systems (Derlon et al., 2012; Wibisono et al., 2015). The
biofilm time-resolved deformation was calculated in real-time from OCT cross sectional
scans (Blauert et al., 2015). Fortunato et al. (2017b) monitored in real-time the fouling
layer evolution in a submerged membrane bioreactor. West et al (2015a) correlated the
biomass accumulation to the feed channel pressure drop increase in time using OCT. Yang
et al. (2000) demonstrated the importance of 3D structural analyses for biofilms grown on
a membrane surface. The 3D image analysis offers several advantages with respect to the
2D analysis, such as quantification of biomass growth defined by biovolume, porosity,
heterogeneity, thickness, and spatial distribution. In the last years, the use of OCT has been
extended to several membrane processes and configurations (Fortunato et al., 2019; Im
et al., 2021; Jang et al., 2022; Pathak et al., 2018; Ranieri et al., 2022; Ricceri et al., 2022;
Scarascia et al., 2021).

The objective of this study was to assess the biomass formation in a spacer filled flow
channel under representative conditions for spiral wound membrane filtration systems.
A novel approach is proposed to process 3D datasets acquired with OCT and to visualize
and quantify the biomass distribution over the feed spacer and membrane surfaces. The
proposed approach allows to evaluate the impact of accumulated biomass on membrane
filtration performance measured by feed channel pressure drop and permeate flux.

376
 Chapter 17

17.2 MATERIALS, EXPERIMENTAL SET-UP

For all the experiments the biomass was grown on sheet of membrane and spacer in
membrane fouling simulator (MFS) (Vrouwenvelder et al., 2007). To enable in-situ non-
destructive observation of the biomass formation by OCT, the MFS cover contained five
millimeter thick glass window. For each experiment a 20 cm × 10 cm ultrafiltration (PAN
UF, with a molecular cut-off of 150 kDa) membrane coupon and 31 mil (787 μm, Trisep,
USA) thick feed spacer was inserted into the MFS. The ultrafiltration (UF) membrane was
necessary to allow water permeation at one bar through the membrane due to the low
hydraulic pressure thereby mimicking the flux through the system and resulting hydraulic
resistance. Moreover, the use of this membrane enabled the investigation of the biofouling
in spacer filled channel without any influence of concentration polarization or other types
of fouling.

C
OCT
A B C E D

Inlet MFS

permeate
∆P
C

Nutrient D
dosage

Figure 1 Schematic representation of the experimental setup consisting of carbon (A) and cartridge
filters (B), a tank containing nutrient solution, pump (C), dosing pump, flow meter (D),
pressure reducing valve (E), differential pressure transmitter, membrane fouling simulator
(MFS), and optical coherence tomography (OCT) device.

The MFS was operated under constant hydraulic pressure of one bar at ambient temperature
(20 ˚C). The MFS was fed with tap water by a gear pump (Cole Palmer, USA) at a flow rate of
45.5 L·h-1, resulting in a 0.16 m·s-1 linear flow velocity at the inlet side of the flow channel,
representative for practice (Vrouwenvelder et al., 2009). The tap water was filtered through
carbon and cartridge filters (5 μm pore size) to remove residual chlorine and to avoid
larger particles entering the MFS (Figure 1). Water permeation though the UF membrane
was accomplished with one bar pressure. The hydraulic pressure was regulated by a back-
pressure valve (Hydra cell, Wanner Engineering Inc., USA) located on the outflow of the
MFS. During the five days experimental period the biomass development was monitored
by OCT imaging and its impact on performance was evaluated by the feed channel pressure
drop (Deltabar, Endress + Hauser PMD75, Germany) (Bucs et al., 2015), and permeate flux
(Sensirion, Switzerland) measurements.

17.3 METHODS

17.3.1 Imaging with Optical Coherence Tomography

An OCT (Thorlabs GANYMEDE GmbH, Dachau, Germany) with a central wavelength
of 930 nm equipped with a 5× telecentric scan lens (Thorlabs LSM 03BB) was used to
investigate the biomass growth in the MFS flow channel containing membrane and feed

377
Experimental Methods for Membrane Applications

spacer sheets. The MFS was mounted on a stage under the OCT probe in order to monitor the
biomass development over time in a fixed area (one feed spacer square element) positioned
at 5 cm from the feed inlet over time (Figure 2). The monitored area corresponds to 5.3 mm
× 5.3 mm with 2.7 μm axial resolution. The OCT lens depth of field was adjusted to 950
μm (slightly higher than the total flow channel height of 787 μm) to allow capturing a part
of the membrane and cover glass window. The resulting image stack resolution was (545 ×
545 × 482) pixels, with a lateral resolution of 11 μm.

Figure 2 Orthogonal view of OCT images of accumulated biomass (orange color) on the feed
spacer, membrane, and cover glass window (5.3 mm × 5.3 mm × 0.95 mm) in the MFS
after one day of operation. The yellow lines show the location of the orthoslices.

17.4 DATA ANALYSIS

17.4.1 Biovolume calculation

The OCT images were processed using ImageJ software (Version 1.48). A multi-step
processing sequence was applied, consisting of (1) subtraction the initial image t0 from
the image taken at any given time (tX), (2) adjustment of contrast and brightness of the
resulting image (3) application of a median filter and (4) binarization of the image with
Otsu algorithms (Otsu, 1979). This approach allows the elimination of the cover glass,
membrane, and feed spacer from the OCT image stack, and allowing the quantification of
the accumulated biomass (Figure 3).

The initial scan was subtracted from the successive scans (step 1) in order to eliminate the
over or under estimation of the accumulated biomass in the scanned area the feed spacer
geometry and other structures present in the flow channel need to be eliminated from the
scans. The resulting stack was then processed with a customized MATLAB code to obtain
the thickness map.

378
 Chapter 17

a) t0

b) tx

c) tx- t0

Figure 3 OCT scans at different times at the same position: (a) image before biomass formation
at t0, (b) image with accumulated biomass after certain time tX and spatially-resolved
biomass quantification (c) after subtracting the image at time 0 from the image taken after
a certain time period (tX – t0). The final image shows only the biomass (orange color)
without the background signals (glass, membrane, and feed spacer).

The binarized datasets were then further analyzed to assess the accumulated biomass
volume (VTot) using the ImageJ plug-in voxel counter. Two different biomass descriptors
were used to quantify the biomass development in the flow channel. The total biovolume
(mm3/cm2) for the scanned (monitored) area was calculated with the following equation:

VTot
VScanned = Eq. 1
AScanned

where VTot is the total biomass volume and AScanned is the scanned area (in this case 0.53 cm
× 0.53 cm). The specific biovolume (VSpecific) was calculated using the following equation:

VSpecific =
∑V i
biomass
=
VTot
Eq. 2
∑A i
∑ Ai
where Vbiomass is the biomass volume, Ai the covered area of the investigated element (i) of
the flow channel (membrane, feed spacer, cover glass). The total biomass VTot is the sum
of biomass accumulated on the membrane, spacer, and cover glass surface. The specific
biovolume (ViSpecific) for each element was calculated using the following equation:
i
Vbiomass
i
VSpecific = Eq. 3
Ai
where ViSpecific is the specific biovolume of each individual flow cell element (i.e., membrane,
feed spacer, cover glass). The developed approach allows to separately evaluate the
accumulated biomass on the membrane, feed spacer and cover glass surface respectively.

379
Experimental Methods for Membrane Applications

A B
C

Figure 4 2-D view of the spatial distribution of the biomass on the three elements (membrane, feed
spacer and cover glass). Three masked areas A, B and C (the boundaries of the masked
area are represented by dashed lines) are distinguished in correspondence of the three
elements. Biofilm is represented by brown color.

Three different masks (A, B, C) were created for the three elements one for the spacer (B)
and two for the glass (A) and membranes (C) (Figure 4). The size of masks was determined
according to the maximum thickness of the biomass observed on the surface of the elements.
For the cases where the biomass is attached simultaneously to two elements (Figure 5), the
biomass volume is calculated by equally distributing the biomass over the two elements.
First the voxels are counted in the areas where the masks belonging to two different elements
intersect (A∩B and B∩C) and the total number of voxels are divided by two and subtracted
from the total number of voxels counted in each mask (Eqs. 4 -11).

Figure 5 2-D representation of areas where the biomass is simultaneously attached to two elements
(membrane, feed spacer and cover glass). The hatched regions represents the areas where
the biomass is attached to two elements. The dashed lines represent the boundaries of the
three different masked areas and the brown color symbolizes the biomass.

Voxel counting in the three different masks: glass (A), spacer (B) and membrane (C). For
the cases where the biomass is attached simultaneously to two elements (Figure 5), the
biomass volume is calculated by equally distributing over the two elements. First the voxels
are counted in the areas where the masks belonging to two different elements intersects
(A∩B and B∩C) and the total number of voxel are divided by two and subtracted to the total
number of voxels counted in each mask (Eq. 4 – 11).

A = ∑Voxel A Eq. 4

B = ∑VoxelB Eq. 5

C = ∑VoxelC Eq. 6

A∩ B = ∑Voxel A∩B Eq. 7

380
 Chapter 17

B ∩ C = ∑VoxelB∩C Eq. 8

A∩ B B ∩ C
VoxelSPACER = B − − Eq. 9
2 2

A∩ B
VoxelGLASS = A − Eq. 10
2

B∩C
Voxel MEMBRANE = C − Eq. 11
2

17.4.2 Image Processing

The OCT was used to monitor the biomass formation at a fixed position in the spacer filled
channel two times per day throughout the five days experimental period. To quantify the
biomass development the accumulated biomass volume was calculated from the OCT
scans. The feed spacer was not transparent for the OCT. When the feed spacer was present a
shift of the location of the membrane and possible biomass below the feed spacer filaments
were observed (Figure 3a,b). The applied image processing method allows visualization of
the biomass only and thus excludes the membrane, feed spacer and cover glass structure
from the collected scans (Figure 6). The rendered volume development over time shown in
Figure 6 represents only the biomass.

Figure 6 Three-dimensional (3D) rendered OCT image with biomass (brown color), the spacer,
membrane and cover glass were eliminated by using the scan at time zero as baseline.

381
Experimental Methods for Membrane Applications

17.5 DATA DISCUSSION AND INTERPRETATION

17.5.1 Biomass Quantification

The OCT scans confirmed the presence of biomass after one day of operation with nutrient
dosage (Figure 8a). As reported in the material and methods, the biomass grown in a
specific area can be quantified with different descriptors as biomass volume (Vtot), scanned
biovolume (V) and specific biovolume (VSpecific). The scanned biovolume normalizes the
biomass volume for the scanned area, allowing comparison of data obtained with the same
feed spacer (and flow channel height). However, the specific biovolume is the only descriptor
that allows comparing the biomass volume with different feed spacers, normalizing the
biomass volume for the available surfaces (membrane, feed spacer and glass window) in the
flow cell. In Table 1 are reported the biomass values over the time according to different
descriptors.

The OCT scans taken periodically during the experimental period confirm the exponential
biomass growth (r2 = 0.97). In the first two days of nutrient dosage only a small amount of
biomass was detected. A specific biovolume of 0.22 mm3·cm-2 was detected in the position
monitored after one day, corresponding to 0.9 percent of the available volume. From the
third day a steep increase in biomass volume was observed (Figure 8a). Towards the end
of the study the rate of increase in biomass volume started to decrease. At the end of the
experimental period, the final biomass volume occupied 24.9% of the monitored area
reaching a specific biovolume of 6.29 mm3·cm-2.

Table 1 Biomass development in the flow cell in time with the four descriptors
Biomass Volume Scanned Biovolume Specific Biovolume Feed channel void
Time (hours) (mm3) (mm3·cm-2) (mm3·cm-2) volume %*

27 0.17 0.61 0.22 0.9

39 0.22 0.78 0.28 1.1
45 0.45 1.60 0.58 2.3
54 0.93 3.31 1.19 4.7
63 1.69 6.02 2.17 8.6
72 2.7 9.61 3.47 13.7
81 3.19 11.36 4.09 16.2
93 3.96 14.10 5.08 20.1
102 4.50 16.02 5.78 22.9
114 4.90 17.44 6.29 24.9

*Percentage of the occupied volume occupied by the biomass from the total available volume.
The fixed area = 5.3 mm x 5.3 mm; flow channel height = 0.787 µm; feed channel volume = 22.1 mm3;
feed spacer porosity = 0.89; available feed channel volume = 19.7 mm3.

382
 Chapter 17

Day 1 Day 2

Day 3 Day 4

Figure 7 Biomass development over time. Flow direction from bottom to top (arrow).

17.5.2 Membrane Performance

Pressure drop over the feed channel and permeate flux through the membrane were
monitored throughout the experimental period. Additionally, biomass volume was
calculated from the OCT scans.

Biomass accumulation was confirmed by the feed channel pressure drop. A rapid increase in
feed channel pressure drop was observed after two days of operation with nutrient dosage
(Figure 8b). By the end of the experimental period the normalized feed channel pressure
drop reached a value of 980 mbar/m due to biomass accumulation.

Feed channel pressure drop was measured over the whole flow channel length (20 cm)
while the OCT data is from a fixed position. In the present study, the OCT scans covered a
much smaller area than pressure drop measurements, 5.3 mm × 5.3 mm in our case with a
2.7 μm resolution positioned at 5 cm from the feed inlet. This gives the possibility to detect
biomass deposition and growth at an early stage with micrometers resolution.

Because of the use of a UF membrane, the initial permeate flux of the clean membrane was
105 L·m-2·h-1. With nutrient dosage a small flux decline was observed at the first day of the
experimental period, followed by a rapid decrease (1.27 L·m-2·h-1) on the second and third
days (Figure 8c). On the fourth day the rate of flux decline was slowed down (0.21 L·m-2·h-1)
and reached a final permeate flux of 30 L·m-2·h-1 at the end of the experimental period.

383
Experimental Methods for Membrane Applications

N. Feed channel pressure drop (mbar)

Specific biovolume (mm3/cm2) a) b) b)
7 250 120
6

Permeate flux (LMH)

200 100
5
80
4 150
60
3 100
2 40
50 20
1
0
0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5
Time (days) Time (days) Time (days)

Figure 8 Development of biomass and membrane performances over time. (a) Specific biovolume
calculated from the OCT scans. (b) Normalized pressure drop over the MFS feed channel
due to biomass development. (c) Permeate flux.

17.6 APPLICATIONS, EXAMPLES

17.6.1 Biomass Distribution

Pressure drop over the feed channel and permeate flux through the membrane was monitored
throughout the experimental period. Additionally, biomass volume was calculated from the
OCT scans.

Biomass accumulation was confirmed by the feed channel pressure drop. A rapid increase in
feed channel pressure drop was observed after two days of operation with nutrient dosage
(Figure 8b). By the end of the experimental period the normalized feed channel pressure
drop reached a value of 980 mbar/m due to biomass accumulation.

Feed channel pressure drop was measured over the whole flow channel length (20 cm)
while the OCT data is from a fixed position. In the present study, the OCT scans covered a
much smaller area than pressure drop measurements, 5.3 mm × 5.3 mm in our case with a
2.7 μm resolution positioned at 5 cm from the feed inlet. This gives the possibility to detect
biomass deposition and growth at an early stage with micrometers resolution.

Because of the use of a UF membrane, the initial permeate flux of the clean membrane was
105 L·m-2·h-1. With nutrient dosage a small flux decline was observed at the first day of the
experimental period, followed by a rapid decrease (1.27 L·m-2·h-1) on the second and third
days (Figure 8c). On the fourth day the rate of flux decline was slowed down (0.21 L·m-2·h-1)
and reached a final permeate flux of 30 L·m-2·h-1 at the end of the experimental period.

384
 Chapter 17

a) b)
1.8

Specific biovolume (mm3/cm2)

Glass 1.6 Glass 6
Spacer Spacer
1.4 5
Biomass volume (mm3)

Membrane Membrane
1.2
4
1.0
0.8 3
0.6 2
0.4
0.2 1

0 0
1 2 3 4 1 2 3 4
Time (day) Time (day)

Figure 9 Biomass volume (a) and specific biovolume (b) in time on the feed spacer, and membrane
surface in the MFS. Specific biovolume is the biomass volume over the available surface
area (area of both membrane and cover glass was each 28 mm2 and of feed spacer was
21.9 mm2).

17.6.2 Biomass and performance decline

Based on the OCT images, the accumulated biomass volume was calculated for each
measurement time thus enabling to quantify changes in biomass volume. As the biomass
volume increased the feed channel pressure drop increased (Figure 10a) and the permeate
flux decreased (Figure 10b). The two performance indicators feed channel pressure drop
and permeate flux, were seen to respond differently by the increasing biomass volume.
During the biomass accumulation in the flow cell two phases were observed in the rate
of permeate flux decline (a sharp decrease followed by less sharp decrease), while the feed
channel pressure drop increased with increasing biomass.

Increase in the channel pressure drop can be explained by the biomass distribution in the
flow channel (Figure 10a). Quantification of the accumulated biomass volume on the
membrane and feed spacer surfaces showed more biomass accumulation on the feed spacer
than on the membrane surface (Figure 9a).

The impact of the accumulated biomass on the different flow channel elements (membrane,
feed spacer and cover glass) on pressure drop increase is shown in Figure 11. The biomass
accumulated on the feed spacer and on the cover glass had a higher impact on pressure drop
increase than the biomass accumulated on the membrane surface.

385
Experimental Methods for Membrane Applications

Normalized permeate flux (LMH)

a) b)
250
1.0
Normalized feed channel 200
pressure drop (mbar) 0.8

150
0.6

100
0.4

50 0.2

0 0
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Specific biovolume (mm3/cm2) Specific biovolume (mm3/cm2)

Figure 10 Normalized feed channel pressure drop (a), and permeate flux (b) as function of the
accumulated biovolume during the 5 day experimental period

1.8
1.6
1.4
1.2
1.0
0.8
Glass 0.6
Spacer 0.4
Membrane 0.2
Biomass volume (mm3) 0
0 20 40 60 80 100 120
Pressure drop (mbar)

Figure 11 Accumulated biomass volume on the three different elements (membrane, feed spacer
and cover glass) in function of feed channel pressure drop increase.

17.6.3 Biomass Thickness Map

Biofilm thickness maps are presented in this study as a new tool to assess the biofilm spatial
distribution on a surface. It is a similar approach as the classical distribution map that depicts
the distribution of a phenomenon on a surface, where different colors are used to show and
evaluate the distribution in a physical map.

386
 Chapter 17

Figure 12 Biofilm thickness map generated from a 3D OCT dataset. The calibration bar allows
estimating the biofilm thickness deposited in the spacer filled channel.

In Figure 12 the biofilm thickness map generated from a 3D OCT dataset is shown for a
spacer filled channel representative of a spiral-wound membrane. Applying 3D OCT image
analysis to a specific area enables the analysis of information related to the spatial distribution
and to the homogeneity of the biofilm. In Figure 12, areas shown in red represent higher
amounts of biofilm, and are more easily and rapidly detected. In this study, the approach
was used to assess the biomass thickness and the spatial distribution over a single frame
(Figure 12), and to evaluate the biofouling development over the time (Figure 13). The same
approach can be extended to the evaluation of the biofilm distribution over any surfaces.

Figure 13 Biofilm development in spacer filled channel (20 hours, 30 hours, 40 hours). Biofilm
thickness map. Flow direction is from left to right.

17.7 ADDITIONAL CONSIDERATIONS

In this study a novel approach for 3D reconstruction, assessment and visualization of OCT
images was presented. The method presented enables monitoring and quantification of
biomass growth during operation. The approach was used to evaluate the effect of the (i)
biomass on membrane performance and evaluate the (ii) biomass spatial distribution in the
flow channel.

17.7.1 OCT Image Analysis

The novel image processing method presented in this study (i) eliminates the background
signal (feed spacer, membrane, and cover glass) from the images and (ii) enables reduction of
the noise of the OCT scans. By applying the scan at time zero as a baseline, all changes in the

387
Experimental Methods for Membrane Applications

subsequent images can be normalized to time zero. Besides subtracting the signals due to
the three elements (spacer, cover glass and membrane) it also removes the signal due to the
water present in the flow cell. Therefore, the proposed approach reduces the background
noise and facilitates the binarization.

West et al. (2015b) used an image masking process on OCT scans to avoid the structures
not corresponding to the biomass. The data presented in this study are similar with the
results shown by West et al. (2015), a different OCT scan processing method has been
applied. The method presented in this study enabled the detailed visualization of the
biomass deposition in the monitored area.

Other imaging techniques used to study biofilms such as confocal laser scanning microscopy
(CLSM) and scanning electron microscopy (SEM) generate images with a higher resolution,
however, OCT enables studying larger areas necessary to gain knowledge on biofouling
behavior and how it may influence the performance of membrane filtration systems. As
reported by Wagner et al. (2010), the structural information at micro-scale and nano-
scale level might be of minor relevance to characterize the behavior of macro-scale biofilm
processes as they occur in membrane filtration systems.

Meso-scale investigation of the biomass by OCT gives insight to the biofouling distribution
in a spiral wound membrane module. Biomass formation under different conditions, like
various spacer geometry, hydrodynamic conditions or cleaning strategies can be evaluated
at a meso-scale range, due to the repetitive geometry of the feed spacer (Bucs et al., 2015;
Radu et al., 2014). The possibility to evaluate biomass development under operational
conditions, in-situ, at a meso-scale range (mm3) is one of the advantages of OCT compared
with other imaging techniques.

Obtaining 3D biomass structures formed under representative conditions for spiral

wound membrane systems may be used as additional tool to better understand the impact
of different operational conditions on the biomass formation and to evaluate the effect of
control strategies on the biomass structure. In-situ real-time detailed image analysis on
the acquired biomass morphology could be used to evaluate how the biomass structure
responds to the operational conditions (i.e., feed pressure).

The proposed approach for analyzing OCT scans can be used to evaluate biomass
development: (i) under various operating conditions, (ii) on different membranes and
spacers (e.g., coatings/modifications) and (iii) in the presence of biocides.

17.7.2 Biomass accumulation and membrane performance

The delay in increasing feed channel pressure drop with respect to the biomass increase
as detected by OCT scans can be explained by the higher sensitivity of the OCT and the
position of the scanned area. When biofilm starts to form and grow in the feed channel
the pressure drop starts to increase. However, in an early stage of biomass accumulation
the biomass may not have an immediate impact on pressure drop. Bucs et al. (2014)
demonstrated that a 5 μm thin biofilm and small biofilm patches in the flow channel may
not be detected by pressure drop measurements. Conversely, OCT imaging allows to
capture and visualize these thin biofilms.

388
 Chapter 17

The higher impact on pressure drop increase of biomass accumulation on the feed spacer
has been observed in other studies as well (Bucs et al., 2014). As shown in Figure 9 the
biomass accumulated on the membrane has lower impact on the feed channel pressure drop
in respect to the biomass accumulated on the other elements.

The effect of biomass on permeate flux in spiral wound elements (reverse osmosis,
nanofiltration) depends on the membrane surface coverage, flow channeling, biofilm
hydraulic resistance, biofilm porosity and thickness (Dreszer et al., 2013a). At the initial
phase of biofilm formation, studies showed that a thin biofilm layer is deposited on the
surface (Flemming et al., 1997). At this phase the biofilm is a thin and porous structure
with low hydraulic resistance, meaning that the membrane surface coverage will be the
main factor which impacts water flux (West et al., 2015b). As the biofilm grows (i.e. more
biomass volume), thickness, porosity and hydraulic resistance change. Studies have shown
that young biofilms are less porous and tend to have a low hydraulic resistance compared to
a mature biofilm (Dreszer et al., 2013b; Martin et al., 2014). The rapid flux decline observed
in the early stage of biomass accumulation may be attributed to the pore blocking fouling
mechanism in UF membranes (Wang and Tarabara, 2008). Once the biomass layer is formed
on the membrane surface the flux depends mainly on its properties and the flux decline rate
decreases (Figure 8c, days 3 and 4). As shown in figure 9a, on the third and fourth day the
biomass volume only slightly increases on the membrane surface while sharply increases on
the other two elements (feed spacer and glass).

The biomass accumulation in the flow channel had different impact on the membrane
performance parameters. While the pressure drop increases as the biomass increases, the
permeate flux decrease is significantly affected in the initial phase of biomass accumulation.

17.7.3 Biomass location in the flow channel

The biomass accumulation occurred mainly on the feed spacer in the early stages may
be an indication of either a higher affinity of bacteria to attach to the feed spacer material
(polypropylene) or preferential deposition due to the hydrodynamics of the system.
Other studies have also reported that at initial stages of biomass formation, more biomass
accumulates on the feed spacer than on the membrane surface (Baker et al., 1995; Van
Paassen et al., 1998; Vrouwenvelder et al., 2008b). As reported in Vrouwenvelder et al.
(2008) feed spacers play an important role in biofouling development and in membrane
cleanability (Creber et al., 2010).

A lower biomass volume was measured on the membrane surface compared to the cover
glass surfaces (Figure 9). The difference in the biomass volume can be attributed to the water
flux through the membrane. For this study a UF membrane was used, resulting in a high-
water flux (105 LMH). It was shown previously that the biomass compacts, decreasing in
thickness and thus in biomass volume under high flux conditions (Dreszer et al., 2014;
Valladares Linares et al., 2016b). This may have affected the measured biomass volume
and underestimated the amount on the membrane surface. However, in a spiral wound
membrane system the flow channel is delimited by membranes on both sides therefore the
biomass accumulated on the membrane surface may have a lower impact on pressure drop.

389
Experimental Methods for Membrane Applications

17.7.4 Use of OCT in biofouling studies

The main advantage of OCT is that it allows observation and monitoring of biomass
development during MFS operation without sample preparation such as the use of stains or
contrast agents. The effect of various cleaning strategies (e.g., chemical cleaning, air flushing,
back washing etc.) on biomass developed can also be evaluated. Moreover, the reconstructed
3D biomass structures can be further imported into modeling software for mathematical
modeling to increase the understanding of biofouling processes. The 3D biomass analysis
and mapping presented in this study shows that OCT is a promising tool to study biofouling
in membrane systems.

17.7.5 Mapping The Biofouling

In biofouling monitoring, there is a need to quickly assess and evaluate the spatial
development of the biofilm on the membrane over time. Compared to the method used for
direct observation through the membrane (DOTM) (Chen et al., 2004), 3D visualization
and thickness maps allow a better understanding of the biofilm deposition, providing a
depth-resolved biofilm structure. As matter of fact, cross-sectional analysis is necessary to
enable the distinction of biofilm accumulation in different elements (i.e., membrane, feed
spacer and glass for the spacer filled channel) and for quantifying the thickness. With respect
to the 3D volume rendering image analysis (Fortunato et al., 2017a), thickness mapping
images require less imaging skills, less computational resources and automated data
handling is therefore more feasible. Furthermore, interpretation of fouled systems applying
3D rendered volume images is more difficult compared to the biomass thickness maps. In
the 3D rendering images, biofilm deposited on the glass may obscure visualization of lower
areas and thus impact the image results. Thickness maps can be obtained directly from raw
images without any correction (e.g., time zero as baseline) or data segmentation, where the
image data is not only due to the biofilm but also the other elements (e.g., feed spacer, glass,
and membrane).

Another advantage of mapping images is the calibration of the color scales, making it
possible to relate each color with a corresponding thickness value. In this way, quantifying
the amount of biofilm deposited over a specific area can be done while evaluating the spatial
distribution. An example of calibrated maps for the system studied in this paper is shown
in Figure 13.

The approach proposed can easily be applied to any membrane configuration. It is possible
to evaluate the biofilm thickness distribution on the membrane surface and distinguish
between different fouled areas. The colored scale allows identifying zones with lower
biofilm deposition, which can further be related to the water flow (i.e., lower hydraulic
resistance).

390
 Chapter 17

17.8 SUMMARY

The use of optical coherence tomography (OCT) to investigate biomass in membrane

systems has increased with time. OCT enables characterizing the (bio)-fouling in-situ and
non-destructively. In this study, a novel approach to process three-dimensional (3D) OCT
scans is proposed. The approach allows obtaining spatially-resolved detailed structural
biomass information. The 3D biomass reconstruction enables analysis of the biomass
only, obtained by subtracting the time zero scan to all images. A 3D time series analysis of
biomass development in a spacer filled channel under representative conditions (cross-flow
velocity) for a spiral wound membrane element was performed. The flow cell was operated
with monitoring of ultrafiltration membrane performance: feed channel pressure drop and
permeate flux. The biomass development in the flow cell was detected by OCT before a
performance decline was observed. Feed channel pressure drop continuously increased with
increasing biomass volume, while flux decline was mainly affected in the initial phase of
biomass accumulation. The novel OCT imaging approach enabled the assessment of spatial
biomass distribution in the flow cell, discriminating the total biomass volume between
the membrane, feed spacer and glass window. Biomass accumulation was stronger on the
feed spacer during the early stage of biofouling, impacting the feed channel pressure drop
stronger than permeate flux.

391
Experimental Methods for Membrane Applications

17.9 REFERENCES

Baker, J., Stephenson, T., Dard, S., Cote, P., 1995. Characterisation of fouling of nanofiltration
membranes used to treat surface waters. Environmental Technology 16, 977–985.
Blauert, F., Horn, H., Wagner, M., 2015. Time-resolved biofilm deformation measurements using
optical coherence tomography 112, 1893–1905. https://doi.org/10.1002/bit.25590
Bucs, Szilárd S., Farhat, N., Siddiqui, A., Valladares Linares, R., Radu, A., Kruithof, J.C., Vrouwenvelder,
J.S., 2015. Development of a setup to enable stable and accurate flow conditions for membrane
biofouling studies. Desalination and Water Treatment 57, 12893–12901. https://doi.org/10.1
080/19443994.2015.1057037
Bucs, Szilard S., Linares, R.V., Marston, J.O., Radu, A.I., Vrouwenvelder, J.S., Picioreanu, C., 2015.
Experimental and numerical characterization of the water flow in spacer-filled channels
of spiral-wound membranes. Water Research 87, 299–310. https://doi.org/10.1016/j.
watres.2015.09.036
Bucs, Sz.S., Radu, A.I., Lavric, V., Vrouwenvelder, J.S., Picioreanu, C., 2014. Effect of different
commercial feed spacers on biofouling of reverse osmosis membrane systems: A numerical
study. Desalination 343, 26–37. https://doi.org/10.1016/j.desal.2013.11.007
Chen, V., Li, H., Fane, a. G., 2004. Non-invasive observation of synthetic membrane processes - A
review of methods. Journal of Membrane Science 241, 23–44. https://doi.org/10.1016/j.
memsci.2004.04.029
Creber, S.A., Vrouwenvelder, J.S., van Loosdrecht, M.C.M., Johns, M.L., 2010. Chemical cleaning of
biofouling in reverse osmosis membranes evaluated using magnetic resonance imaging. Journal
of Membrane Science 362, 202–210. https://doi.org/10.1016/j.memsci.2010.06.052
Derlon, N., Peter-Varbanets, M., Scheidegger, A., Pronk, W., Morgenroth, E., 2012. Predation influences
the structure of biofilm developed on ultrafiltration membranes. Water Research 46, 3323–
3333. https://doi.org/http://dx.doi.org/10.1016/j.watres.2012.03.031
Dreszer, C., Vrouwenvelder, J.S., Paulitsch-Fuchs, A.H., Zwijnenburg, A., Kruithof, J.C., Flemming,
H.C., 2013a. Hydraulic resistance of biofilms. Journal of Membrane Science 429, 436–447.
https://doi.org/10.1016/j.memsci.2012.11.030
Dreszer, C., Vrouwenvelder, J.S., Paulitsch-Fuchs, A.H., Zwijnenburg, A., Kruithof, J.C., Flemming,
H.C., 2013b. Hydraulic resistance of biofilms. Journal of Membrane Science 429, 436–447.
https://doi.org/10.1016/j.memsci.2012.11.030
Dreszer, C., Wexler, A.D., Drusová, S., Overdijk, T., Zwijnenburg, A., Flemming, H.-C., Kruithof, J.C.,
Vrouwenvelder, J.S., 2014. In-situ biofilm characterization in membrane systems using Optical
Coherence Tomography: formation, structure, detachment and impact of flux change. Water Res
67, 243–54. https://doi.org/10.1016/j.watres.2014.09.006
Flemming, H.C., 2003. Role and levels of real-time monitoring for successful anti-fouling strategies -
An overview, in: Water Science and Technology. pp. 1–8.
Flemming, H.-C., Schaule, G., Griebe, T., Schmitt, J., Tamachkiarowa, A., 1997. Biofouling—the
Achilles heel of membrane processes. Desalination 113, 215–225. https://doi.org/10.1016/
S0011-9164(97)00132-X
Flemming, H.-C., Wingender, J., 2010. The biofilm matrix. Nat Rev Micro 8, 623–633.
Fortunato, L., Bucs, S., Linares, R.V., Cali, C., Vrouwenvelder, J.S., Leiknes, T., 2017a. Spatially-resolved
in-situ quantification of biofouling using optical coherence tomography (OCT) and 3D image
analysis in a spacer filled channel. Journal of Membrane Science 524, 673–681. https://doi.
org/10.1016/j.memsci.2016.11.052

392
 Chapter 17

Fortunato, L., Li, M., Cheng, T., Rehman, Z.U.Z.U., Heidrich, W., Leiknes, T., 2019. Cake layer
characterization in Activated Sludge Membrane Bioreactors: Real-time analysis. Journal of
Membrane Science 578, 163–171. https://doi.org/10.1016/J.MEMSCI.2019.02.026
Fortunato, L., Qamar, A., Wang, Y., Jeong, S., Leiknes, T., 2017b. In-situ assessment of biofilm formation
in submerged membrane system using optical coherence tomography and computational
fluid dynamics. Journal of Membrane Science 521, 84–94. https://doi.org/10.1016/j.
memsci.2016.09.004
Halan, B., Buehler, K., Schmid, A., 2012. Biofilms as living catalysts in continuous chemical syntheses.
Trends in Biotechnology 30, 453–465. https://doi.org/10.1016/j.tibtech.2012.05.003
Hemming, H., Tamachkiarowa, A., Klahre, J., Schmitt, J., 1998. Monitoring of fouling and biofouling
in technical systems. Water Science and Technology 38, 291–298. https://doi.org/10.1016/
S0273-1223(98)00704-5
Herzberg, M., Elimelech, M., 2007. Biofouling of reverse osmosis membranes: Role of biofilm-enhanced
osmotic pressure. Journal of Membrane Science 295, 11–20. https://doi.org/10.1016/j.
memsci.2007.02.024
Im, S.J., Fortunato, L., Jang, A., 2021. Real-time fouling monitoring and membrane autopsy analysis in
forward osmosis for wastewater reuse. Water Research 197, 117098. https://doi.org/https://
doi.org/10.1016/j.watres.2021.117098
Jang, Y., Lee, J.G., Fortunato, L., Lee, J., Lee, Y., An, A.K., Ghaffour, N., Lee, S., Jeong, S., 2022. Colloidal
silica fouling mechanism in direct-contact membrane distillation. Desalination 527. https://
doi.org/10.1016/j.desal.2022.115554
Martin, K.J., Bolster, D., Derlon, N., Morgenroth, E., Nerenberg, R., 2014. Effect of fouling layer spatial
distribution on permeate flux: A theoretical and experimental study. Journal of Membrane
Science 471, 130–137. https://doi.org/10.1016/j.memsci.2014.07.045
Matin, A., Khan, Z., Zaidi, S.M.J., Boyce, M.C., 2011. Biofouling in reverse osmosis membranes
for seawater desalination: Phenomena and prevention. https://doi.org/10.1016/j.
desal.2011.06.063
Otsu, N., 1979. Threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern.
Pathak, N., Fortunato, L., Li, S., Chekli, L., Phuntsho, S., Ghaffour, N., Leiknes, T.O., Shon, H.K.H.K.,
2018. Evaluating the effect of different draw solutes in a baffled osmotic membrane bioreactor-
microfiltration using optical coherence tomography with real wastewater. Bioresource
Technology 263, 306–316. https://doi.org/10.1016/j.biortech.2018.04.123
Radu, A.I., van Steen, M.S.H., Vrouwenvelder, J.S., van Loosdrecht, M.C.M., Picioreanu, C., 2014.
Spacer geometry and particle deposition in spiral wound membrane feed channels. Water Res
64, 160–76. https://doi.org/10.1016/j.watres.2014.06.040
Ranieri, L., Vrouwenvelder, J.S., Fortunato, L., 2022. Periodic fouling control strategies in gravity-
driven membrane bioreactors (GD-MBRs): Impact on treatment performance and membrane
fouling properties. Science of The Total Environment 838, 156340. https://doi.org/https://
doi.org/10.1016/j.scitotenv.2022.156340
Ricceri, F., Blankert, B., Ghaffour, N., Vrouwenvelder, J.S., Tiraferri, A., Fortunato, L., 2022. Unraveling
the role of feed temperature and cross-flow velocity on organic fouling in membrane distillation
using response surface methodology. Desalination 540, 115971. https://doi.org/https://doi.
org/10.1016/j.desal.2022.115971
Ridgway, H.F., Flemming, H.-C., 1996. Membrane biofouling in water treatment membrane processes.
Water Treatment Membrane Processes.
Scarascia, G., Fortunato, L., Myshkevych, Y., Cheng, H., Leiknes, T., Hong, P.-Y., 2021. UV
and bacteriophages as a chemical-free approach for cleaning membranes from anaerobic

393
Experimental Methods for Membrane Applications

bioreactors. Proceedings of the National Academy of Sciences 118, e2016529118. https://doi.

org/10.1073/PNAS.2016529118
Valladares Linares, R., Fortunato, L., Farhat, N.M., Bucs, S.S., Staal, M., Fridjonsson, E.O., Johns, M.L.,
Vrouwenvelder, J.S., Leiknes, T., 2016a. Mini-review: novel non-destructive in situ biofilm
characterization techniques in membrane systems. Desalination and Water Treatment 1–8.
https://doi.org/10.1080/19443994.2016.1180483
Valladares Linares, R., Wexler, A.D., Bucs, S.S., Dreszer, C., Zwijnenburg, A., Flemming, H.C., Kruithof,
J.C., Vrouwenvelder, J.S., 2016b. Compaction and relaxation of biofilms. Desalination and Water
Treatment 57, 12902–12914. https://doi.org/10.1080/19443994.2015.1057036
Van Paassen, J.A.M., Kruithof, J.C., Bakker, S.M., Kegel, F.S., 1998. Integrated multi-objective membrane
systems for surface water treatment: pre-treatment of nanofiltration by riverbank filtration and
conventional ground water treatment. Desalination 118, 239–248. https://doi.org/10.1016/
S0011-9164(98)00137-4
Vrouwenvelder, J., Van Paassen, J., Wessels, L., Van Dam, A., Bakker, S., 2006. The Membrane Fouling
Simulator: A practical tool for fouling prediction and control. Journal of Membrane Science 281,
316–324. https://doi.org/10.1016/j.memsci.2006.03.046
Vrouwenvelder, J.S., Bakker, S.M., Wessels, L.P., van Paassen, J.A.M., 2007. The Membrane Fouling
Simulator as a new tool for biofouling control of spiral-wound membranes. Desalination 204,
170–174. https://doi.org/10.1016/j.desal.2006.04.028
Vrouwenvelder, J.S., Hinrichs, C., Van der Meer, W.G.J., Van Loosdrecht, M.C.M., Kruithof, J.C., 2009.
Pressure drop increase by biofilm accumulation in spiral wound RO and NF membrane systems:
role of substrate concentration, flow velocity, substrate load and flow direction. Biofouling 25,
543–55. https://doi.org/10.1080/08927010902972225
Vrouwenvelder, J.S., Manolarakis, S.A., van der Hoek, J.P., van Paassen, J.A.M., van der Meer, W.G.J., van
Agtmaal, J.M.C., Prummel, H.D.M., Kruithof, J.C., van Loosdrecht, M.C.M., 2008a. Quantitative
biofouling diagnosis in full scale nanofiltration and reverse osmosis installations. Water Research
42, 4856–4868. https://doi.org/10.1016/j.watres.2008.09.002
Vrouwenvelder, J.S., Manolarakis, S.A., van der Hoek, J.P., van Paassen, J.A.M., van der Meer, W.G.J., van
Agtmaal, J.M.C., Prummel, H.D.M., Kruithof, J.C., van Loosdrecht, M.C.M., 2008b. Quantitative
biofouling diagnosis in full scale nanofiltration and reverse osmosis installations. Water Research
42, 4856–4868. https://doi.org/10.1016/j.watres.2008.09.002
Wagner, M., Taherzadeh, D., Haisch, C., Horn, H., 2010. Investigation of the mesoscale structure
and volumetric features of biofilms using optical coherence tomography. Biotechnology and
Bioengineering 107, 844–853. https://doi.org/10.1002/bit.22864
Wang, F., Tarabara, V. V, 2008. Pore blocking mechanisms during early stages of membrane fouling by
colloids. J Colloid Interface Sci 328, 464–9. https://doi.org/10.1016/j.jcis.2008.09.028
West, S., Wagner, M., Engelke, C., Horn, H., 2015a. Optical coherence tomography for the in situ three-
dimensional visualization and quantification of feed spacer channel fouling in reverse osmosis
membrane modules. Journal of Membrane Science 498, 345–352. https://doi.org/10.1016/j.
memsci.2015.09.047
West, S., Wagner, M., Engelke, C., Horn, H., 2015b. Optical coherence tomography for the in situ three-
dimensional visualization and quantification of feed spacer channel fouling in reverse osmosis
membrane modules. Journal of Membrane Science 498, 345–352. https://doi.org/10.1016/j.
memsci.2015.09.047

394
 Chapter 17

Wibisono, Y., El Obied, K.E., Cornelissen, E.R., Kemperman, a. J.B., Nijmeijer, K., 2015. Biofouling
removal in spiral-wound nanofiltration elements using two-phase flow cleaning. Journal of
Membrane Science 475, 131–146. https://doi.org/10.1016/j.memsci.2014.10.016
Yang, X., Beyenal, H., Harkin, G., Lewandowski, Z., 2000. Quantifying biofilm structure using image
analysis. Journal of Microbiological Methods 39, 109–119. https://doi.org/10.1016/S0167-
7012(99)00097-4

395
 Part 6
General applications
 doi: 10.2166/9781789062977_0399

Chapter 18

Membrane Autopsy
Javier Rodriguez Gómez, Genesys - PWT, Spain

Nuria Peña García, Genesys - PWT, Spain

The learning objectives of this chapter are the following:

• Describe the membrane autopsy procedure

• Present the experimental set-up for membrane autopsies

• Present the membrane autopsy protocol

• Mention methods for fouled membrane characterisation

• Describe tests for membrane damage characterisation

• Illustrate membrane performance characterisation

• Present tests for chemical cleaning efficiency

18.1 INTRODUCTION

Membrane systems suffer foulant in a higher or lower extent during their performance
life. This foulant is related to feed water quality, pre-treatment efficiency and other factors
related to design and plant operation. On the other hand, membranes may suffer irreversible
damage which is necessary to identify.

Membrane autopsy is the main tool for identifying fouling nature and membrane failures.
It is based on a complex group of analyses and tests which are used to find out the cause of
membrane failure or performance decline. In this process there are some analyses and tests
that are basic to understand the status of the membrane, but there are additional tests which
can be used depending on the case.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment,
Sergio G. Salinas-Rodriguez, Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

Table 1 Main causes of membrane failures and their impact

Damage
Foulant Physical / mechanical issues | Chemical issues

Reversible Irreversible
↓ Permeability ↑ Permeability
↓ Salt rejection ↓ Salt rejection
↑ Differential pressure (dp) in some cases

Considering the main issues that membranes may suffer, autopsies should involve those
analyses and tests which allow to determine which is the main cause of membrane failure,
how it affects membrane performance and if it is possible to recover it.

Membrane autopsies can provide critical information needed to help troubleshoot and/
or to optimize membrane systems performance, but considering that they include many
results, it is necessary that an expert is reading the data to get a more reliable and useful
diagnosis.

Identification of foulant can be carried out on any membrane type. But to check the impact
of that failure on membrane performance, it is necessary to reproduce its performance.
There are systems available to check membranes performance, but those mainly used are for
NF and RO spiral bound membranes and UF hollow fibers. In this chapter NF and RO spiral
wound membranes are mainly covered.

To give a good diagnosis about membrane failure, it is very important to have information
about the problem detected in plant and some details about the site. This will also help to
achieve the final diagnosis and to select the membranes to study.

Main details to consider would be:

- Water type
- Plant stages/pass
- Elements per pressure vessel
- Recirculation Yes/No
- Recovery
- Feed system (storage tanks) opened / covered
- Pretreatment details: Settings, flotation, UF, filtration,
- Microfilters: Type (wound/expanded/pleated), replacement, dp design/current,
micron, etc.
- Reagents dosing: Coagulant, flocculant, pH adjustment, Chorine, reducer (eg., SBS),
antiscalant
- SDI values
- CIP frequency and protocols

400
 Chapter 18

A complete study of the membrane would include the characterization of the full element
first and the study and characterization of the membrane once unrolled. Through this way,
it will be possible to distinguish if some issues are due to membrane configuration or other
components different than membrane. The analysis and characterization of membrane
samples will mainly give information about the membrane itself.

For a useful autopsy, membrane selection is essential. For example, for RO systems, it is
recommended to autopsy one membrane from lead position of first stage and one from tail
position from last stage to get a complete understanding of the plant. Lead element in the
system will contain highest concentration of suspended solids and organics. Biofouling is
also worst at the first stage where the bacteria attach to the first available hydrophobic surface
that they find. Last membrane in the system receives water with a higher concentration of
salts so it is the most likely to scale.

18.2 MATERIALS, EXPERIMENTAL SET-UP

Spiral wound membranes are built with an outer glass fiber covering which must be
removed to unroll the element. For this purpose, an autopsy table and cutting machine are
the basic tools (Figure 1).

Figure 1 Full membrane element ready for autopsy.

The best surface to perform an autopsy is a level table, accessible from all sides and with
a size slightly wider than the axial length of the element and at least the leaf/membrane
envelopes length (Franz Leitz, 1996).

Considering the cylindrical shape of the membranes, it is necessary to fix the membrane to
the table in such a way that it keeps blocked during the external housing cutting, unrolling
and membrane sampling. For it, different shape cradles or systems to fix membrane to the
table can be used.

Cutting of the external housing produces a significant amount of glass fiber dust, so it is very
important to carry out this process considering personnel safety. Then, it is mandatory to
follow some cautions:

401
Experimental Methods for Membrane Applications

Table 2 Safety during membrane autopsy

It is advisable to use an electric cast saw
Cutting machine better than grinder machines
Personnel protection - Wear ample eye and face protection.
- Wear gloves.
- Wear long-sleeved garments to avoid skin contamination
with fiberglass particles.

For sampling of the different components that can be analysed for the autopsy, it is necessary
to arrange some simple tools which allow taking membrane samples, foulant and particles.
Some of these tools are plastic bags, plastic bottles, scalpel, tweezers, spatula, scissors, etc.

18.3 MEMBRANE AUTOPSY PROTOCOL

The diagram in Figure 2 show main steps involved in a membrane autopsy:

Step 1
Step 2 Step 5
Data collection
External Step 4 Foulant
inspection identification
Visual
Sampling
inspection
Internal Integrety tests /
inspection damage identification

Step 3
Membrane
Cleaning tests
performance

Figure 2 Main steps for implementing a membrane autopsy.

Before the autopsy starts it is important to have the following information (data collection):
- Plant failure to know the objective of the study. That information is essential also to
determine the kind of tests and analyses to be carried out and the necessary samples.
- Membrane position. It will help to understand the results and to give more useful
recommendations. For this it is essential also that the autopsied membrane is the correct
for the purpose of the study. This point is essential for RO systems.
- Manufacturer, model and type of membrane. It is necessary to know the reference
values of membrane performance, pH limits for cleaning tests, etc.
- Membrane serial number. It will help in some cases to verify in plant membrane position
or to check performance parameter values with membrane manufacturer.

Since there are many details to consider during the autopsy, it is advisable to use a check
list to be sure that all the items are reviewed, especially during the visual inspection of the
membrane.

402
 Chapter 18

As summary, there are two main procedures during membrane autopsy:

- Steps 2 and 3: Visual inspection which reviews macroscopic issues.
- Step 5: Analyses and tests which give information about microscopic issues, foulant
composition, structural changes, recovery of membrane performance, etc.
All the analysis and tests carried out during the autopsy should be selected to get the most
accurate diagnosis.

18.4 METHODS

18.4.1 Visual inspection

External inspection
Before proceeding to open the membrane, some details of the membrane must be considered.
Table 3 includes the main membrane details that must be checked.

Table 3 Details to consider during external inspection

Weight By comparing to manufacturer reference, it gives an
idea about the amount of foulant. Element must be
drained to assure that retained water is removed.
It is mainly useful when scaling is suspected

External housing integrity. In some cases affected by overpressure issues.

Presence of particles/deposits on external housing. Failures on element seal (O-ring).
Antitelescoping devices (ATD) condition. In some cases damage is produced during transport.
Presence of particles/deposits on element ends. Related to pre-treatment issues.
Permeate tube condition. If there is a damage, there will be salt rejection issues.
Telescoping
Spacer protrusion

The best way to document these details is taking photographs. Some photographs are
included in Figure 3, showing examples of main membrane failures that can be observed
during visual inspection (Images credit by Genesys Membrane Products S.L.U).

Depending on the membrane manufacturer, it will be necessary or not to remove ATD to

make element ends inspection.

If deposits and particles are observed during external inspection, samples should be taken
just in case it is necessary to carry out additional analyses. These samples would provide
information about plant performance.

During external inspection, it is recommendable to carry out also an integrity test of the full
element. This integrity test can be carried out in two ways:
- Bubble test: During this test, a small pressure (3-5 psi) is put into the permeate tube
while the element is submerged vertically under water. If the element continually emits
bubbles and cannot hold the air pressure, then the element exhibits compromised
mechanical integrity (Technical Service Bulletin Nitto-Hydranautics, 2017).

403
Experimental Methods for Membrane Applications

- Vacuum decay test: In this test, after element drain to remove water, permeate tube of
the element is evacuated and isolated. A vacuum decay higher than 100 mbar/minute
indicates mechanical integrity or a leak on the membrane element (Dupont, 2020).

Membrane failures to consider during external inspection

(a) Telescoping (b) Spacer protrusion

(c) Damaged external housing (d) Damaged permeate tube

(e) Damaged antitelescoping device (f) Particle / grain at feed end producing
separation between envelopes

Figure 3 Examples of different issues that can be observed during external inspection of membranes
during autopsy. (Images credit: Genesys Membrane Products S.L.)

Internal inspection

To proceed with the internal inspection of the membrane, it is necessary to remove the
outer casing (Figure 4). Before unrolling the membrane, it is necessary to verify the feed side
of the element and it is recommendable to mark it. It is the best way to verify the feed side
once the membrane is unrolled.

404
 Chapter 18

For outer casing removal, at least four cuts are necessary to make easier the procedure:
two cross sections to remove ATDs (cuts 1 and 2 at the following photograph) and two
longitudinal sections (cuts 3 and 4). It is mandatory to follow the safety recommendations
described in Section 18.2.

Cut 3

Cut 4*

Cut 1 Cut 2

Figure 4 Recommendable cuts for external housing removal during autopsy.

*Note: cut 4 should be made approximately on the opposite side of cut 3.

With this procedure, two pieces of external housing with a quite similar size must be
removed before unrolling the membrane (Figure 5). Once the membrane is unrolled,
inspection can be carried out.

(a) Outer housing removal (b) Membrane unrolling

(c) Unrolled membrane ready

for internal inspection
Figure 5 Membrane unrolling during autopsy. (Images credit: Genesys Membrane Products S.L.)

As during external inspection, some details must be checked during this internal inspection.
Table 4 includes these details, which should be checked in the number of leaves/envelopes
that assure a representative inspection of the membrane element.

405
Experimental Methods for Membrane Applications

Table 4 Details to consider during internal inspection

Odour Odour description is important to distinguish some
foulant components.
Presence of foulant /particles During visual inspection it is possible to distinguish
deposits from scaling, organics from inorganics, and if
it will be necessary to make specific analyses to identify
them.
Distribution of foulant/particles It is very important to check if there is a higher
concentration of foulant on feed or reject side, if there
are different components and if there is presence of
particles.
Failures on membrane surface Creases, delamination, and other failures.
Spacer prints If spacer prints are visually detected a significant
damage from increases in dp can be expected.
Spacer material condition Presence of foulant, wrinkles or deformation are
important details to consider if detected.
Permeate carrier material condition If presence of foulant is observed on permeate carrier it
will be evidence of damage on membrane material.
Membrane colour It is important to know in advance the colour of a new
membrane to determine if there is a failure/problem
or not.
Membrane leaves insertion
Failures on these components can lead on salt
Glue lines condition rejection reduction and permeate flow increase.

Similar to external inspection, the best evidence of the details and failures observed during
membrane internal inspection is by taking photographs. The images in Figure 6 correspond
to some of the main membrane failures that should be considered during this inspection.

406
 Chapter 18

(a) Membrane with heterogeneous distribution of foulant

(b) Particles at feed end (c) Wrinkles / crease

(d) Delamination on glue lines and (e) Scaled spacer material

membrane active layer

Figure 6 Examples of different issues that can be observed during internal inspection of membranes
during autopsy. (Images credit: Genesys Membrane Products S.L.)

During internal inspection of the membrane element, the different samples needed for the
rest of analyses and tests included in the autopsy procedure should be taken. Each autopsy
may involve different samples depending on the objective of the study, but in general terms
the minimum samples to consider are listed in Table 5.

407
Experimental Methods for Membrane Applications

Table 5 Recommended sampling during autopsy

Sample Analysis / test

Particles / deposits from element ends They will allow distinguishing elements from
suspended matter reaching the membranes from
foulant components developed on membrane surface.
Foulant By analysing foulant removed from membrane surface,
foulant components will be concentrated and it will be
possible to distinguish some foulant components from
membrane components during analyses.
Membrane Raw sample Microbiology
Foulant identification
Performance parameters
Clean surface Cleaning tests
Presence of halogens/oxidant agents
Polyamide bands conditions

18.4.2 Analytical methods for foulant and damage identification

Some of the analytical techniques commonly used for foulant identification are described
in other chapters. The aim of this chapter is to guide during membrane autopsy proceeding
the analytical techniques that can be used for membrane failure diagnosis. Those considered
more simple, accessible, and more commonly used are described in Table 6.

Table 6 Analytical techniques for the identification of fouling components and membrane damage
Autopsy step Method Items to check

Foulant identification Analytical techniques LOI / TGA Quantification organic/inorganic based

in the loss of weight after thermal
degradation.
SEM-EDX Elementary identification of foulant
components.
It is possible to distinguish different foulant
components.
ATR-FTIR Identification of functional groups and
compounds from data base.
Microbiological It gives quantification of specific
counts microorganisms.
LC-OCD Separation and identification of foulant
components, mainly from NOM.

408
 Chapter 18

Autopsy step Method Items to check

Damage Analytical techniques SEM-EDX Abrasion marks from particles and spacer
material.
ATR-FTIR Polyamide bands changes indicate
structural changes/degradation.
XPS/ESCA Halogens detection Elements oxidation
state.
Tests Dye tes Damage, mainly physical.
Fujiwara test Contact with halogens if they are on high
concentration.

Table 6 includes some of the most common techniques used during autopsies but any
technique which could be applied on solids and surfaces to provide information about both
foulant identification and membrane conditions could be used.

Other techniques to be considered: optical microscopy, confocal microscopy, chemical

solubility of foulant, humic and acids testing, contact angle testing, Zeta potential, FRX,
DRX, ICP-MS, GC-MS, etc.

During autopsies, it is also common to calculate foulant density (mg of foulant/cm2).

Reference for this parameter may change depending on the characteristics of foulant, but
it gives quite useful information about the distribution of foulant on a system if different
membranes from same pressure vessel (PV) are analysed.

LOI and TGA

- Loss on ignition (LOI):
LOI is a common and widely used method to estimate the organic content of membrane
foulant. Organic matter is oxidised at 500–550 °C to carbon dioxide and ash. The weight
loss during this reaction is easily measured by weighing the dry sample (LOI 105 ˚C) before
and after heating and is closely correlated to the organic matter.
- Thermogravimetric analysis (TGA)
For TGA, the mass of a substance is monitored as a function of temperature or time as
the sample specimen is subjected to a controlled temperature program in a controlled
atmosphere (Perkin Elmer, 2010-2015).

Both techniques provide the content of organic/inorganic component in percentage. This

information is very valuable to determine which is the main cause in failure and also about
the performance of the plant comparing results from different membrane position.

SEM-EDX
Scanning Electronic Microscopy with X-ray dispersive energy analysis (SEM-EDX) is used
to study the membrane surface and to verify the elemental composition of its fouling.
Elemental determination with the SEM-EDX system is based on analysis of X-rays
produced via electron beam excitation of a sample area. This technique allows analysis of
a sample in selective areas. The limited depth of analysis (typically a few microns), and the

409
Experimental Methods for Membrane Applications

possibility to select the area of interest under the electron beam, allows for local analysis to
reveal differences in composition. The identification and measurement of individual peak
intensities in the X-ray spectrum is done with a computerized multichannel analyser.

SEM-EDX is one of the most powerful and basic tools used during membrane autopsies.
It is useful not only for foulant identification, but also for studying foulant distribution on
membrane surface and to distinguish different foulant components. Some damage can also
be identified, mainly abrasion marks and marks from spacer. Concerning chemical damage,
it is very difficult to detect them by this technique (Peña et al., 2013).

The microphotographs in Figures 7a and 7b show characteristic membrane structure studied

by SEM-EDX at high magnifications and different types of foulant and damage.

(a) Charateristic membrane structure (b) Membrane structure

with a thin organic covering

(c) Membrane with biofilm (d) Different inorganic components

(e) Composite foulant composed or an (f) Different structures of diatoms

organic component with both

Figure 7a Examples of different observations and details obtained during analysis of membranes by
SEM-EDX. (Images credit: Genesys Membrane Products S.L.)

410
 Chapter 18

(a) Example of a membrane with (b) Mapping allows to apply color to

different components on areas of different elements detected during
contact with spacer material and EDX analysis, wich makes much easier
the rest of the membrane surface to study element distribution on membrane surface

(c) Marks from scaling on a clean (d) Marks from spacer material on
membrane surface a clean membrane surface

Figure 7b Examples of different observations and details obtained during analysis of membranes by
SEM-EDX. (Images credit: Genesys Membrane Products S.L.)

ATR-FTIR
In the mid-infrared, absorption of radiation is related to fundamental vibrations of the
chemical bonds. Then, fourier transform infrared spectroscopy with attenuated total
reflectance (ATR-FTIR) Spectrometry can provide valuable information related to the
chemical structure of membrane or characterize the fouling layer that may be present on the
membrane surface. IR spectra can be studied both in absorbance or transmittance.

As examples, Figures 8 and 9 include IR spectra in absorbance showing the characteristic

spectrum of a clean polyamide-polysulphone membrane in good chemical conditions (blue
lines) compared to spectra of membranes with thick covering and thin covering (red line).
Membranes with thin covering will show most of the membrane bands and membranes
with thick covering will mainly show bands from foulant composition. Identification of
foulant will be obtained both by the identification of functional groups or by using IR data
base. It is important to remind that the composite nature of the foulants may be difficult in
some cases the identification and that commonly only the main component of foulant will
be identified.

411
Experimental Methods for Membrane Applications

Thick covering Membrane reference

0.30
0.25

1635
0.20

1238

1073
1103

556
1046
0.15

1585
1544

1014
1505
2923

1487
2959
0.10

1323

689
1411

832
715
634
0.05

2852

853
794
0.00

1488

1242

1151
-0.05

16091541

1106
-0.10

558
1169
-0.15

1449

834
1663
Absorbance

-0.20

873
1414

691
1014
3324

1081

716
1738
-0.20
3071

602
3157
2967

664
795
2873

917
-0.20
-0.25
4,000 3,000 2,000 1,500 1,000 500
Wavenumbers (cm-1)

Figure 8 Example of membrane with thick foulant (red line) compared to a reference membrane
(blue line): only some of the membrane bands are detected since there is a main presence
of foulant.

0.32

1241
1488

Thin covering 0.30

Membrane reference 0.28
1150 0.26
1105
0.24
0.22
0.20
1585

0.18
1169

557
1609

0.16
833
1324

0.14
1414

1294
1542

853
1660

691

0.12
1081
3320

1447

873

0.10
3039 3075
2969

735
1364

664

0.08
636

470
795
Absorbance

0.06
2868

919

0.04
0.02
0.00
4,000 3,000 2,000 1,500 1,000 500
Wavenumbers (cm-1)

Figure 9 Example of membrane with thin foulant (red line) compared to a reference membrane
(blue line): most of the membrane bands are detected.

Since IR spectrometry provides information related to the presence or absence of specific

functional groups, shifts in the frequency of absorption bands and changes in relative band
intensities indicate changes in the chemical structure or changes on the membrane surface.

412
 Chapter 18

By this technique it is possible to check the presence of polyamide layer bands in an RO

membrane and to determine if there are any structural changes on it (Figure 10).

Clean membrane surface Membrane blank

0.4

0.3

0.2

0.1

Intensity 0.0
4,000 3,000 2,000 1,500 1,000 500
Wavenumbers (cm-1)

Figure 10 Example of membrane with significant chemical damage: Comparison of used membrane
IR spectrum after a mechanical cleaning (red line) compared to membrane reference
(green line).

There is an absence of polyamide bands.

In some case in which the decrease is not significant and it is necessary to check membrane
conditions, quantification of the polyamide bands can be carried out to achieve additional
information (Sandin et al., 2013).

Fujiwara test
Fujiwara test (FJ) test detects significant levels of polyhalogen compounds so it will detect if
the membrane is oxidised. This is a colorimetric test in which a pink colour in the analytical
solution, indicates organically bound halogens. FJ test must be carried out on RO water
rinsed membrane samples and without deposit (physically removed).
It is a qualitative test that can be easily carried out by the following procedure:
1. Introduce a sample of membrane in a test tube.
2. Add pyridine, add approximately the same volume of sodium hydroxide. Shake to
mix the two layers.
3. Heat the mixture in a boiling water bath.
4. If a reddish-pink colour appears in the organic layer, it indicates that the membrane
has been in contact with a halogen (normally chlorine)*.

413
Experimental Methods for Membrane Applications

*It is also advisable to carry out a blank test (without membrane) to verify that the
mixture of pyridine and sodium hydroxide does not generate any colour without the
sample.

The photographs in Figure 11 show a positive and a negative Fujiwara test. This is a qualitative
test that only detects contact with halogens when they are present in a high extent. To
detect halogens in low concentration or to know the impact on membrane performance, it
is necessary to make additional tests and analysis (ATR-FTIR, XPS/ESCA, etc.).

(a) Negative Fujiwara test (b) Positive Fujiwara test

Figure 11 Examples of a negative (picture a) and positive results (picture b) obtained during Fujiwara
test. (Images credit: Genesys Membrane Products S.L.)

XPS/ESCA
X-ray photoelectron spectroscopy (XPS), also known as electron spectroscopy for chemical
analysis (ESCA) is a quantitative spectroscopic technique that measures the empirical
formula, chemical state and electronic state of the elements that exist within a material. XPS
spectra are obtained by irradiating a material with a beam of X-rays while simultaneously
measuring the kinetic energy and number of electrons that escape from the top 1 to 10
nm of the material being analysed. This technique is able to detect and identify different
halogens (chlorine, bromine, iodine) and it can distinguish the oxidation state of the
detected elements.

Then, for example if the oxidant agent is sodium hypochlorite, this technique allows
distinguishing between chloride (198,7 eV) and chlorine related to carbon by a quasi-
covalent bound (200 eV) (Beverly et al., 2000, Hiraoka et al., 2011).

For seawater membranes, even if the disinfection process is based in NaOCl, the oxidant
agent is bromine. Figure 12 includes a characteristic spectra of a clean membrane surface
with a significant presence of bromine.

Depending on the detected halogen, different impact on membrane performance can be

expected (Maugin, 2013).

414
 Chapter 18

80
Name Pos. FWHM Area At%

C 1s
C 1s 285.0 3.48 8726.68 68.68
N 1s 400.0 2.65 2172.54 9.50 70

O 1s
O 1s 532.0 3.24 5703.11 15.32
Na 1s 1072.0 2.71 255.64 0.24 60
Ca 2p 347.0 2.62 450.82 0.70
Br 3p 184.0 3.52 3560.94 5.57
50

N 1s
40

Ca 2p
Na 1s

30

Br 3p
20

10

CPS (103) 0
1,200 900 600 300 0
Binding energy (eV)

Figure 12 XPS spectrum (clean surface) of a membrane with presence of bromine.

Dye test
This test is based on the passage of a high molecular weight compound solution through the
membrane. A high molecular weight compound should be retained by the polyamide layer,
so when presence of this compound is detected on the permeate side of the membrane, it
means that the polyamide layer is damaged on that area. This test indicates damage but not
the source, although it is more sensitive to physical damage (Peña et al., 2013).Different
dyes can be used for this test, but one of the most common is Methylene blue. This test can
be done on the full element or on membrane samples.

For this test, it is necessary to circulate a solution of aprox. 0.01% dye, applying the minimum
pressure at which it permeates the membrane during 20-30 minutes.

If the surface of the membrane is physically damaged, the dye will pass through the
membrane, showing colour on the permeate side (Figure 13).

Feed side Feed side

Permeate side
Permeate side

(a) Negative methylene blue test (b) Massive positive methylene blue test

Figure 13 Examples of a negative (picture a) and positive results (picture b) obtained during
Methylene Blue test. (Images credit: Genesys Membrane Products S.L.)
415
Experimental Methods for Membrane Applications

18.4.3 Membrane performance

Membrane performance tests must be carried out considering standard conditions from
membrane manufacturer. These tests can be carried out on a full element before the autopsy
and on flat test rig during the autopsy.

Besides the flat test rig, total system design should include parts listed in Table 7 (weebly.
com):

Table 7 Main components to be included in a flat test rig system for membranes autopsy
Part of the rig design Details
Water tank Good chemical and temperature.
Thermostatic system Membrane characterization must be carried out at a temperature as constant and
close to 25˚C as possible.
Heating element To apply temperature during cleaning tests.
Pump Both pressure and flow rates must be variable.
Piping It must be pressure and corrosive resistant.
Pressure gauge 0 to 70-80 bar.
Collection tank A measuring cylinder is the most appropriate since it will provide volume reading.

There are various flat test rig systems available in the market. Main detail to consider for a
correct flat test rig selection is to choose the one that allows with brackish and/or sea water
samples depending on the needs.

The diagrams in Figure 14 obtained from references correspond to some examples of rig and
system configuration that can be used for this purpose:

Feed/outlet channels Permeate Concentrate

Pressure gauge

PTFE gasket Valve

Feed

Sintered disc Pump Pressure gauge 4” x 6” flat sheet

(membrane membrane
sits on top)

Permeate outlet

14a) 14b)

Figure 14 Examples of flat test rigs by (14a) Robinson et al, (2004) and (14b) Cartwright (2012).

416
 Chapter 18

Permeate
collection
vessel

Membrane cell

Feed tank
Brine Bypass valve Pressure
pressure relief valve
gauge Flow meter

Brine
valve/ Feed pressure gauge Hydro cell pump
pressure
regulator

Figure 15 Example of system for membrane characterization (Farhat et al., 2013).).

To characterize membrane coupons during autopsy, it is necessary to select representative

samples. It is recommended also to use spacer material and permeate carrier from same
membrane model.

Feed water must be prepared with sodium chloride (reverse osmosis membranes) or
magnesium sulphate (nanofiltration membranes) dissolved in deionized water. The
concentration of each salt will be specified by the manufacturer for the design values.

Salt solutions must be recirculated through membrane coupons by applying specified

pressure. Most common characterization conditions are shown in Table 8.

Table 8 Most common conditions for membrane characterization (this info is available from
membrane manufacturers data sheets for each specific type and membrane model)
NF membrane BW RO membrane SW RO membrane
Salt 2,000 mg/L MgSO4 1,500 mg/L NaCl 32,000 mg/L NaCl
concentration 2,000 mg/L NaCl
Pressure, psi 70 150 / 225 600 / 800

The characterization procedure is implemented as follows:

1. Once solution is circulating through the membrane by applying the specified pressure
and it is producing permeate, let it get stabilized during 20-30 minutes.
2. Measure permeate flow and register temperature to normalize the data to 25ºC. For this
normalization, membrane manufacturers commonly include tables or equations to make
needed calculations.
3. Take both feed water sample and permeate sample to measure conductivity or a selected
ion to check salt rejection.
4. Take at least three measurements to get a representative average.

417
Experimental Methods for Membrane Applications

Comparing the results obtained during these tests with reference values established by
membrane manufacturer. different conclusions will be obtained. The following figures
show some examples of characteristic results of fouled (Figure 16) and damaged membranes
(Figure 17).

a) Characteristic permeate flow values b) Characteristic salt rejection values

obtained from membranes with foulant obtained from membranes with foulant
50 Reference 100
Reference Minimum
40 98
Minimum
Flux permeate (L/m2h)

30 96

Salt rejection (%)

20 94

10 92

0 90
GA200608-Feed GA200607-Tail GA200608-Feed GA200607-Tail
Flux Salt rejection

Figure 16 Characteristic results of membranes with foulant with (a) permeate flow lower than
reference and (b) salt rejection lower but close to reference Both parameters can be
recovered during cleaning tests / foulant removal.

a) Characteristic permeate flow values b) Characteristic salt rejection values

obtained from damaged membranes obtained from damaged membranes
90 100
Reference
80 Minimum
70 98
Flux permeate (L/m2h)

6 0
96
50
Salt rejection (%)

Reference
40
94
Minimum 30
20 92
10
0 90
GA200715-First GA200716-Last GA200715-First GA200716-Last
Flux Salt rejection

Figure 17 Characteristic results of damaged membranes where (a) permeate flux higher than
reference and (b) salt rejection lower than reference. These parameters cannot be
recovered during cleaning tests / foulant removal.

18.4.4 Cleaning tests

In the event of a reduction in productivity, it may be interesting to carry out individual
cleaning tests in pilot plants/test rigs to find out the cleaning chemical products and the
ideal cleaning conditions.

418
 Chapter 18

The choice of the most suitable chemical product for cleaning depends on the type of
membrane fouling. For this reason, it is advisable to start the cleaning tests when foulant
composition is already identified.

To carry out cleaning tests, performance data obtained from the characterization of the
membrane are the baseline, but it is necessary to apply the different cleaners on different
samples to obtain information about flux and salt rejection before and after the application
of each product. Then, the procedure should be:
- Characterization of permeate flux and salt rejection as described in section 18.4.3. If
cleaning tests are carried out on a full element, dp should be considered also.
- Application of the cleaner considering cleaner and/or membrane manufacturer
recommendations.
- Rinsing with water until the complete removal of the cleaning solution. pH control can
be used to be sure about cleaning solution removal.
- Characterization of permeate flux and salt rejection as described in section 18.4.3. If
cleaning tests are carried out on a full element, dp measurement should also be considered.

By using this procedure with different cleaners it will be possible to select the most
suitable product, which will be the one achieving the best results in terms of parameters
improvement.

When working with samples at test rigs, if membrane initially gives a flux lower than
reference, the aim of the cleaning test will be to reach that value.

If membrane is damaged, it is probable that in some cases the flux is already equal or higher
than reference. Then, best cleaner will be the one achieving the highest flux increase in terms
of percentage. In this case, it won’t be possible to recover salt rejection and it may happen
that the value gets even worse after foulant removal.

Considering the composite nature of foulants, it is very common to need a multi-step

cleaning procedure based on alkaline-acid cleaners, alkaline-biocide-acid cleaners, etc. In
these cases, once the best cleaners are selected, additional tests must be carried out to check
their performance applied on a full cleaning protocol.

The application of each cleaner must take into account the recommendations of the
product manufacturer and the membrane manufacturer, to follow the guidelines about
pH, temperature, contact time, etc. In some case it will be necessary to improve contact
time, concentration, etc., but always working within conditions that preserve membrane
integrity.

419
Experimental Methods for Membrane Applications

18.5 REFERENCES

Beverly, Seal, Hong (2000), Identification of surface chemical functional groups correlated to failure of
reverse osmosis polymeric membranes, J. Vac. Sci. Technol. 18 1107–1113.
Cartwright (2012) Reverse Osmosis Technology-Not just for Water Purification. Water conditioning
& Purification International Magazine, 2012.
Farhat A., Ahmad F., Hilal N., Arafat H.A. (2013) Boron removal in new generation reverse osmosis
(RO) membranes using two-pass RO without adjustment. Desalination 310, 50-59.
FilmtecTM (2020) Reverse Osmosis Membranes. Technical Manual. Page 143. From No.609-00071-
0416
Leitz (1996) Membrane element autopsy manual. Water Treatment Technology Program Report No.
17. Water Treatment Engineering and Research Group, Environmental Resources Services. U.S.
Department of the Interior.
Hiraoka, Iijima, Sakai (2011) Surface analysis of polyvinylchloride bombarded by Ar+ and charged
water droplets, Surf. Interface Anal. 43, 236–240.
List of sourced parts for the final rig design. - Design of a Desalination by Reverse Osmosis Laboratory
Test Facility (weebly.com)
Maugin, (2013) Fate of Reverse Osmosis (RO) membranes during oxidation by disinfectants used
in water treatment: Impact on membrane structure and performances. Doctoral Thesis. King
Abdullah University of Science and Technology, Thuwal, Kingdom of Saudi Arabia.
Peña García, del Vigo, Chesters, Armstrong, Wilson, Fazel (2013), A study of the physical and
chemical damage on reverse osmosis membranes detected by autopsies. IDA World Congress on
Desalination and Water Reuse, Tianjing, China. IDAWC/TIAN13-184
Perkin Elmer 2010-2015, A Beginner’s Guide.
Robinson, J.P. et al., 2004. Evidence for swelling-induced pore structure in dense PDMS nanofiltration
membranes. Filtration, 4 (1), pp. 50-56.
Sandin et al., (2013) Reverse osmosis membranes oxidation by hypochlorite and chlorine dioxide:
spectroscopic techniques vs Fujiwara test. Desalination and Water Treatment. 51(1-3), p. 318-
327.
Technical Service Bulletin, Nitto-Hidranautics. February 2017. TSB 101.03

420
 doi: 10.2166/9781789062977_0421

Chapter 19

CFD as a Tool for Modelling

Membrane Systems
Gustavo A. Fimbres Weihs, The University of Sydney, Australia
Francisco Javier García Picazo, City of San Diego, USA
Yie Kai Chong, Wen Yew Lam, Jia Xin Tan, Kathleen Foo,
Weng Fung Twong, Yong Yeow Liang,
Universiti Malaysia Pahang Al-Sultan Abdullah, Malaysia

The learning objectives of this chapter are the following:

• Define basic principles of CFD

• Define and apply CFD for modelling membrane systems

• Present and discuss the equations involved and steps for building a CFD model of a
membrane system

• Understand the theoretical background of mass transport and permeate flux and
how they interplay to result in concentration polarisation and fouling.

19.1 INTRODUCTION

Computational fluid dynamics (CFD) is a computer-based numerical method used to analyse

systems that involve fluid flow and/or heat and mass transfer (Versteeg & Malalasekera,
2007). CFD bridges the two different approaches for solving engineering problems before
the computer era, theoretical and experimental; it relies on mathematical models while
being easy to adapt to almost any realistic condition (Anderson & Wendt, 1995). Another
feature of CFD is its versatility, as it allows the analysis of systems for a variety of applications
such as chemical reactions (Salehi et al., 2016), aerodynamics (Snel, 2003), dispersion of
pollutants (Chu et al., 2005), blood flows (Byun & Rhee, 2004), among many others.

© 2024 The Authors. This is an Open Access book chapter distributed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License (CC BY-NC-ND 4.0),
(https://creativecommons.org/licenses/by-nc-nd/4.0/). The chapter is from the book Experimental
Methods for Membrane Applications in Desalination and Water Treatment, Sergio G. Salinas-Rodriguez,
Loreen O. Villacorte (Eds).
Experimental Methods for Membrane Applications

The popularity of CFD has increased as the processing power of computers has become
more capable in the last decades. Some of the advantages of CFD over experiment-based
approaches for analysing and understanding systems are (Versteeg & Malalasekera, 2007):
• significantly higher resolution,
• excellent reproducibility and repeatability,
• ability to analyse complex systems,
• capacity to extract multiple characteristics of the fluid (velocity, concentration, vorticity,
etc.),
• non-intrusiveness, and
• set-up cost and time reduction.

Theoretical analysis of fluids is typically intractable for realistic conditions, so typically

several assumptions are considered in order to be able to solve these models (Wiley &
Fletcher, 2003). In addition, these simplified analytical models are very difficult to validate
in the laboratory due to the presence of external sources of noise (Liang et al., 2020b). In
those cases, CFD stands out a useful technique as it allows the study of complex models that
are difficult to analyse mathematically or to set up experimentally, albeit at the expense of
computational power.

There are multiple software options available to perform CFD analysis, including commercial
offerings such as ANSYS-Fluent®, ANSYS-CFX® and COMSOL-Multiphysics®, as well as
freely available open-source options such as OpenCFD-OpenFOAM® (Table 1). The most
general differences between them are the user interface, the flexibility to edit the simulation
parameters and the numerical approach used to obtain the results. In general, they all make
use of the computer CPU (ANSYS-CFX®) to perform calculations, while some may use the
GPU or both (ANSYS-Fluent®). The selection of software should be based on the type of
analysis being performed.

Table 1 Commercially available software for CFD analysis and their characteristics
Software Developer Type Method Reference Used by

Fluent® ANSYS Proprietary Cell-centred ANSYS Inc. Gavelli et al. (2008),

finite volume or (2020) Liu et al. (2013),
finite element Su et al. (2019)
CFX® ANSYS Proprietary Cell-vertex ANSYS Inc. Liang et al. (2018b),
finite volume (2012) Toh et al. (2020b)
Multiphysics® COMSOL Proprietary Finite element COMSOL AB Baghel et al. (2020),
(2008) Brunner et al. (2018)
OpenFOAM® OpenCFD Open source Cell-centred OpenCFD Haddadi et al. (2018),
finite volume Ltd. (2016) Liu and Hinrichsen
(2014), Kone et al.
(2018)

422
 Chapter 19

19.1.1 What is NOT modelled

As any other analysis method, CFD presents some limitations that are worthy of note when
considering it for the analysis of fluid flows. For instance, some simulations may require
very large computational time to perform even using specialised hardware. The use of
high-performance computing hardware can significantly reduce the computational time,
but this is often expensive (Jamshed, 2015). In addition, CFD is not suitable for all cases
or purposes; for example, simulation of the transport of individual molecules in a fluid
domain can be extremely difficult (or even impossible) to model. Determining the effect
of the physicochemical properties of a membrane on its transport performance is also not
within the applications of CFD, as the relevant phenomena do not occur in the fluid phase
but within the membrane matrix. Moreover, CFD is not useful to determine the effect of
the membrane roughness on the boundary layer, as roughness is in order of 10 nm (Boussu
et al., 2005), while the concentration boundary layer typically ranges from 1 to 100 μm in
thickness (Rodrigues et al., 2013). Given the nature of CFD, variables cannot be explicitly
computed at every possible point of the domain, but they are rather approximated. Major
aspects of the numerical methods that form the basis of CFD are discussed in section 18.1.2.

19.1.2 How modelling can assist membrane systems

Membrane separation processes (MSP) are operations where a fluid is forced through a thin
semipermeable barrier, called a ‘membrane’. Transmembrane pressure and electrochemical
potential are generally the driving forces used to operate membrane systems (Baker, 2012).
One of the key components of MSPs is the membrane itself, as its properties determine
the efficiency of separation of the components in the fluid. There are many applications
for MSPs such as gas purification, crystallisation and membrane reactors; nonetheless, this
chapter focuses on applications in the field of water treatment.

Desalination and tertiary wastewater treatment stand out as the most widely spread
applications of membrane systems. These processes are very effective to separate small
solutes from water mainly using pressure as the driving force. Nevertheless, they require
very high operating pressure (of the order of 10 to 70 bar), which comes with high operation
costs. Another issue faced by MSPs is concentration polarisation (CP) which results in a
significant decrease in efficiency. CP occurs when the solutes rejected by the membrane
accumulate in the vicinity of its surface, forming a region of high concentration, i.e., the
concentration boundary layer. Given the nature of MSPs, CP is impossible to prevent,
although it can be mitigated so its effect on efficiency is minimised, most commonly via
mass transfer enhancement strategies. Other than CP, particulate fouling is also one of the
challenging issues that persist in reverse osmosis systems since the technology of RO was
first introduced into the desalination process and other pressure-driven operations such as
microfiltration, ultrafiltration and nanofiltration.

Over the last decades, CFD has been used to study mass transfer in MSPs and to find
innovative approaches to enhance it. Modifying the geometry and location of spacers
(Fimbres-Weihs & Wiley, 2008), imposing an unsteady flow at the inlet (Liang et al., 2018b)
and vibration-assisted modules (Su et al., 2018), are among the approaches investigated
using CFD for mass transfer enhancement in MSPs. Moreover, CFD has also been employed

423
Experimental Methods for Membrane Applications

to study different types of fouling in membrane modules (Fimbres Weihs & Wiley, 2014;
Radu et al., 2010). The following sections take a closer look into some specific application
cases of CFD analysis of membrane systems.

19.2 METHODS

Using CFD for the analysis of MSPs involves multiple steps that need to be performed to
draw valuable conclusions on the model. A comprehensive CFD study comprises three
main stages: set-up, solution and conclusion, with an additional stage for experimental
validation often performed based on the scope of the study. Figure 1 summarises the typical
workflow for CFD modelling. The steps required for setting up the model are in yellow,
the ones involved in solving the model are in red and finally, in green, the postprocessing
steps. The set-up stage involves the problem definition, generation of the geometry of the
fluid domain, definition of the meshing strategy and the establishment of the boundary
conditions. Despite seeming trivial, defining the goal and subsequent approach to model a
membrane system is crucial, as it sets the foundation of the method.

The solution of the model comes in the next step. CFD uses a numerical approach to solve
the Navier-Stokes equations for continuity, momentum and mass transfer. These equations
are summarised as follows:

=0 Eq. 1

t
+ ( ) = 2
– p Eq. 2

w 2
+ (w ) = D w Eq. 3
t

where r, μ and D are the density (kg m−3), dynamic viscosity (kg m−1 s−1) and diffusivity (m2
s−1), respectively. These physicochemical properties of the fluid are typically considered to
remain constant, although, some variations may be caused by temperature and concentration
fluctuations. CFD methods solve (numerically) these equations for the velocity vector ( ),
the pressure field (p) and the solute mass fraction field (w).

It is worthy of note that there are currently no generally applicable fully-analytical solutions
for the general case of the Navier-Stokes equations, hence the importance of computational
methods. Given the numerical nature of CFD methods, they require boundary and initial
conditions as inputs to the solution algorithm. Determining a solution to the model is
probably the most time-consuming step of modelling, as it involves solving equations -
iteratively until the solution error is below a specified tolerance. It is at this point, when
the error is sufficiently small and the solution variables are not changing significantly after
further iterations, that the solution is said to have converged.

424
 Chapter 19

Once the model solution has converged, it should be compared against previous results for
the same or a similar model, in a step known as verification. Data processing and assessment
come after the results are verified. These two steps are key parts of modelling as they are used
to draw conclusion about the model and, ultimately, for decision making. The assessment
step is crucial to suggest improvements and correct deficiencies in a real-world system.

Validation is not necessarily part of modelling, although, it is useful to prove that the model
is valid for real-world applications. The set-up starts with the problem definition, this is,
conceptualising the model that is to be solved. The steps followed for CFD modelling are
detailed in the following section.

CFD modelling

Problem
Results Verification Validation
definition
Mesh independency study

Yes
Experimental
Generate the Error< No
geometry (1d, tolerance Next iteration Data processing
2D or 3D

Meshing Solver run Assessment Conclusions

Error tolerances
e.g. inlet velocity,
Differentation scheme
zero flux, non-slip, Boundary
symmetry,
Solver setup Additional variables
conditions Timestep scheme
interfaces
Initial conditions

Set-up Solution

Figure 1 Overall process for CFD modelling of membrane systems.

19.2.1 Geometry
One of the first steps for a CFD simulation of an MSP consists in defining the domain. The
geometry of the domain is important as it should represent the system to be modelled.
Spirally wound membrane (SWM) modules are the most industrially spread, therefore,
many of the CFD models are based in this type of modules. In SWM modules the membrane
layers are separated using a spacer mesh with a small width (typically ~1 mm) to create a path
for the flow (Scott, 1995). Thus, most CFD analysis of SWM modules consist of domain
modelling a spacer-filled narrow channel in order to approximate this geometry. The length
of commercial SWM modules is around 1 m which can be difficult to model due to large
computational requirements. Nevertheless, the geometry of a SWM module consists of
multiple repetitions of small units with the same geometry, called unitary cells. Thus, a
common approach in CFD studies is to reduce the length of the channel to be modelled by
considering a small number of unitary cells.

425
Experimental Methods for Membrane Applications

Membrane modules can be divided into four categories, namely flat sheet (FS) membrane,
spiral-wound membrane (SWM), hollow fibre (HF) and tubular membrane as shown in
Figure 2. Among these modules, HF (6000 to 8000 m2/m3) is the most compacted (highest
surface area per volume), followed by SWM (600 to 800 m2/m3), FS (50 to 100 m2/m3)
and tubular membrane (50 to 70 m2/m3) (Martín, 2016). Nevertheless, HF is susceptible to
fouling and clogging, thus it can only be used to treat low viscous water (Berk, 2009).

On the other hand, tubular membrane modules have better antifouling properties compared
to HF and SWM modules because of their relatively larger diameter (10 to 25 mm), so it can
maintain high tangential velocity in the feed stream (Berk, 2009). Thus, it is widely used to
treat wastewater with a high content of suspended solids, or viscous oil and water mixtures
(Xue et al., 2021). In terms of robustness, the SWM is stronger against the membrane
breakage compared to HF (Lu & Chung, 2019). Moreover, FS membranes offer simplicity in
design, but they cannot withstand higher pressures, thus they are only suitable for MF and
UF (Berk, 2009).

1) 2)
Membrane Permeate collection holes Anti-telescoping
device
Feed solution
Permeate Concentrate
Permeate out
Permeate Concentrate

Permeate
channel Feed flow across
feed channel spacer
Retentate Retentate
channel
Membrane
Permeate collection Covering Permeate flow
3) material (after passage
Inside feed Membrane through membrane
into permeate
Retentate Feed channel spacer collection material
Feed

4) Permeate Permeate
Permeate

Outside feed
Feed
Permeate Permeate

Concentrate Feed

Retentate

Figure 2 Schematic diagram of (1) flat sheet membrane module (Berk, 2009), (2) spiral-wound
membrane module (Sparks & Chase, 2016), (3) hollow fibre module (Balster, 2016) and
(4) tubular membrane module (Xue et al., 2021).

426
 Chapter 19

19.2.2 Flow types

19.2.2.1 1D, 2D and 3D

As fluid flow is a 3D phenomenon, the most representative way to model it is using a 3D
model. However, this may be computationally intensive. In some cases, it is useful to reduce
the dimension of the model sacrificing information in order to reduce the computational
requirements of a simulation. CFD allows to analyse a system as a 1D, 2D or 3D model and
the choice between them depends on the applications intended for a study. For example, a 1D
analysis may be useful to analyse a laminar flow between two steady plates, but it overlooks
the variations of the flow properties along the other two dimensions. Furthermore, 2D
models can be used for flow pattern identification disregarding other phenomena like
vortex stretching. In the early stages of CFD, 2D models were the most widely used since the
requirements to simulate in 3D were almost inaccessible. Nevertheless, as the capability of
computers has improved over the last decades, 3D studies are gaining ground against their
2D analogues (Fimbres-Weihs & Wiley, 2010).

19.2.2.2 Laminar, transient, turbulent

A fluid flow can be classified in different ways according to its characteristics, such as
laminar, transient or turbulent. In laminar flow, fluid particles move along a smooth path in
parallel trajectories or layers with very low energy losses. On the other hand, turbulent flow
is characterised by irregular paths for the flow particles and by large energy losses (Streeter
et al., 1985). A fluid flow must have two features in order to be considered turbulent,
randomness and auto-similarity (Landahl et al., 1989). Transient flow is an intermediate
between laminar and turbulent flow, therefore, it shares some characteristics of these
types of flow. Irregular flow trajectory and intermediate energy loss are characteristics of a
transient flow.

The hydraulic Reynolds number (Reh) is almost ubiquitous when characterising a fluid flow.
It is the ratio of inertial forces and viscous forces in a fluid flow. The Reh is described by:
ueff d h
Re h = Eq. 4
μ
where ueff is the effective velocity of the flow, dh is the hydraulic diameter, r is the density
of the fluid and μ is its dynamic viscosity. A flow with Reh > 2100 is typically considered
as turbulent under specific conditions (Rajaratnam, 1976). The hydraulic diameter of the
channel is described by:

4Vch
dh = Eq. 5
aws

The volume of the channel (Vch) and the area of the wetted surface (aws) depend on the
geometry of the channel. Typical values for the dh are in the order of 1×10−3 m.

427
Experimental Methods for Membrane Applications

Operating an MSP under turbulent flow conditions may enhance mass transfer, but implies
excessive pressure drop caused by energy dissipation (Burn & Gray, 2015). Conversely,
laminar flow may improve mass transfer at relatively low pressure drop. For this reason,
most SWM modules operate at a Reh between 300 and 400 (for reverse osmosis) (Liang et
al., 2020b).

19.2.2.3 Single phase

CFD has been used extensively to demonstrate the CP phenomena in a membrane channel
for single phase applications. Common membrane applications that involve only liquid
phase include desalination and oily-waste water treatment. Since the 2000s, CFD has been
used to simulate the hydrodynamic and concentration profiles in the feed channel of SWMs
(Fimbres Weihs & Wiley, 2007; Li et al., 2016; Liang et al., 2019) and HFs (Cancilla et al.,
2021; Junker et al., 2021; Kaya et al., 2014) for desalination processes. The concentration
profile can be visualised in the channel by fixing a solute concentration on an impermeable
wall or incorporating Darcy’s law on a permeable wall (Pak et al., 2008; Wardeh & Morvan,
2008). As the variations of density and viscosity are insignificant in a horizontal RO
membrane system, a Newtonian fluid with constant properties is normally assumed in a
SWM feed channel for the purposes of CFD simulation (Foo et al., 2021).

The simulation of oil-in-water MF through CFD approaches has also been presented in the
literature (Behroozi et al., 2019; Lotfiyan et al., 2014; Zare et al., 2013). The Eulerian model
(described in the following Section) is commonly used to solve the governing equations
for the oil and water portions separately. Zare et al., (2013) and Lotfiyan et al., (2014)
conducted 2D CFD studies of oil-in-water MF and the simulation could predict the CP
profile accurately. However, the permeate flux was underpredicted due to the simplification
of the model (e.g., neglecting pore blockage, size distribution of oil droplets, and interaction
between oil droplets and the membrane surface). Later, Behroozi et al., (2019) coupled the
pore blocking phenomena into a 2D CFD model. The permeate flux predicted by the CFD
simulation was comparable to the experiment with 4.62% error. In addition, it was found
that the pore-blocking model reduced the relative error by 15% compared to the non-pore
blocking model. Thus, considering the pore-blocking phenomena in the simulation of oil-
in-water MF can improve the precision of the result.

19.2.2.4 Multiphase
Air sparging in liquid solutions has been extensively used to induce shear stress and
mitigate fouling on the membrane surface (Martinelli et al., 2010). This two-phase flow
can be modelled using CFD with fluid models such as the Eulerian two-fluid model, or
the volume-of-fluid (VOF) model incorporating into the governing equations. In the case
of Eulerian two-fluid method, the gas and liquid phases are treated with their respective
velocity fields, but both share a common pressure field. Hence, the governing equations
are solved separately. VOF model on the other hand, is capable of tracking the gas/liquid
interface throughout the computational domain by assuming a no-slip condition between
the phases, and that all fluid properties can be calculated as weight-averaged volume
fractions (Ndinisa et al., 2005). The governing equations for Eulerian two-fluid (Asefi et al.,
2019) and VOF (Javid et al., 2017) methods have been reported elsewhere.

428
 Chapter 19

The two-phase flow regime can be either slug or bubbly (Figure 3), which is determined by
the superficial velocities of gas phase and liquid phase according to the channel geometry
(Golrokh Sani et al., 2021; Gupta et al., 2009). Since the 2000s, modelling two-phase flow
in membrane channels through CFD has been extensively studied to understand its impacts
on shear stress and flux enhancement (Gupta et al., 2009; Ndinisa et al., 2005; Taha & Cui,
2002, 2006a, 2006b). From the simulation, several approaches were found to be effective
for minimizing fouling while enhancing flux such as maintaining high gas flow rate and low
liquid rate (Ratkovich et al., 2009), maintaining bigger bubble size (Radaei et al., 2018), and
controlled pulse injection (Taha et al., 2006; Taha & Cui, 2002).

1) 2)

(a) (b) (c) (a) (b) (c)

Time: 0 sec. Time: 0 sec.

(a) (b) (c) (a) (b) (c)

Time: 0.2 sec. Time: 0.2 sec.

Figure 3 Surface profiles of (a) wall shear, (b) bubbles and (c) vorticity for (1) slug flow and (2)
bubbly flow reported by (Javid et al., 2017).

429
 Table 2 Recent 5-years CFD multiphase simulation in membrane process for water treatment

430
Two
phases
flow Software Modelling Membrane Membrane
patterns used Dimension approaches configuration process Application Main findings of spacer performances Ref

Bubbly In-house 2D Eulerian two FS NF MgSO4 The bubbly flow had marginal impact on Asefi et al.
and slug fluid solution the flux enhancement, but a change of the (2019)
flows flow regime to slug flow resulted in a flux
enhancement.
Bubbly OpenFOAM 3D VOF FS UF whey The velocity gradient of fluid flow near Javid et al.
and slug protein bubbles resulted in the formation of vorticity, (2017)
flows concentrate which increased shear stress in the membrane
solution system.
A significantly larger shear stress was
reported in slug flow when compared with the
bubbly flow.
Bubbly ANSYS 3D Eulerian two FS MF Surface Slug flow causes a larger shear stress than Du et al.
and slug FLUENT fluid and water the case in bubbly flow. However, the shear (2017)
Experimental Methods for Membrane Applications

flows VOF induced by bubbly flow was more effective

for reducing fouling in pre-treated feed water.
Bubbly ANSYS 3D VOF HF MF/MBR Water and The fouling rate was found to be inversely Radaei et al.
flow FLUENT Xanthan proportional to the calculated RMS shear (2018)
gum stress. Non-Newtonian fluid showed a
solution reduced tendency in shear fluctuations for
membrane cleaning than those observed in
water. A larger pulse bubble size yielded more
RMS shear stress, which is more effective in
fouling control.
 Chapter 19

In recent years, the effects of slug and bubbly flow in a flat sheet channel on shear stress
and flux enhancement have been compared through experimental and CFD studies. Air
sparging was found to be effectively in reducing the CP by 73% for cross flow NF, leading
to flux enhancement (Asefi et al., 2019). However, only slug flow (0.8 min-1 to 1 min-1) has
significant effects to permeate flux enhancement but not for the bubbly flow (0.2 min-1 to
0.6 min-1) (Asefi et al., 2019). Furthermore, an experimental study showed that the slug
flow and bubbly flow yield higher flux than single liquid phase flow by 78% and 30%
respectively in UF (Javid et al., 2017). This is because the slug flow and bubbly flow result
in a higher velocity, which leads to a higher wall shear stress as shown in Figure 3 (Javid et
al., 2017). Further, the shear stress caused by the slug flow is higher than those reported
in bubbly flow (Javid et al., 2017). In addition, the peak shear stress was found higher for
slug flow (15 Pa) than bubbly flow (1.4), although the shear stress is more even distributed
for bubbly flow than slug flow (Du et al., 2017). Table 2 summarises recent findings on
multiphase for membrane proceses.

19.2.3 Boundary conditions

A set of boundary conditions is required to solve a CFD model due to its numerical nature.
Boundary conditions are required at each interface of the model (i.e., inlet, outlet, walls,
membrane surface and interfaces). Figure 4 shows a schematic of a membrane channel with
the boundary conditions typically used for modelling of MSPs. Other boundary conditions
may apply when analysing chemical reaction or heat transfer.

The boundary condition for the incoming flow can be set as inlet or periodic. Setting a flow
profile is required for the inlet condition, while a periodic boundary wraps the outcoming
flow velocity profile as the inflow. A fully developed profile giving a laminar flow is
generally set as the boundary condition at the inlet. The solute concentration profile in the
incoming flow can be adjusted according to the application of the model (e.g., constant or
periodic). There are three options for boundary conditions at the exit. The outlet condition
is the simplest but does not account for backward flow. The opening condition accounts for
any recirculation re-entering the channel, but it requires value to be set for any transported
variables (e.g., mass fraction and temperature) at the exit. In some cases, the area-averaged
concentration at the exit can be calculated and assigned to the backward flow. Setting an
opening at the exit usually involves a zero-pressure condition (p = 0).

Walls and spacer surfaces are considered impermeable walls with zero-flux, zero-
concentration gradient and non-slip velocity conditions. Perhaps the most critical boundary
condition is located at the membrane surface, where different approaches may be considered
depending on the intended applications.

i. Impermeable wall with constant concentration. The concentration at the vicinity of

the membrane surface increases due to the CP phenomenon. Under fully developed
conditions for the concentration at the membrane surface is roughly twice the
concentration at the bulk flow; hence a common assumption includes a constant
concentration at the membrane surface. Other assumptions made include a zero flux
throughout the membrane (impermeable wall). This assumption is not realistic for MSPs,
but it has been shown that the flux across the membrane has very low impact on the

431
Experimental Methods for Membrane Applications

characteristics of the boundary layer (Schwinge et al., 2002b). In addition, the flux across
the membrane may be estimated using a correlation developed by Geraldes and Alfonso
(2006). These considerations are useful to analyse the mass transfer and hydrodynamics
inside the channel while reducing computational time.
ii. Semi-permeable wall. This is a more realistic approach to membrane systems modelling
as it accounts for the separation capacity of the membrane (Ahmad et al., 2005). In
this case, the flux throughout the membrane is computed using the physicochemical
parameters of the membrane such as permeance and intrinsic solute rejection. This
model is suitable for cases where the ratio between permeate flux and bulk flow velocity
is higher. Calculating the flux is especially important for techno-economic assessment
since it is used to determine the productivity of the system.

Symmetry or periodic

Inlet Outlet

ll
Wa
Membrane
ll

Opening
Wa

Periodic
Periodic
Symmetry or periodic

Figure 4 Schematic of a narrow spacer-filled membrane channel and possible boundary conditions
for CFD analysis.

19.2.3.1 Steady-state and transient-state

There are two types of CFD simulations that can be used to model membrane systems,
steady- and transient-state. The solution obtained in a steady-state simulation describes the
behaviour of the system in the long term, when there is no variation over time (dw/dt = 0). In
general, the variables of the system are recalculated over the domain until some convergence
criterion is reached. Steady-state simulations require relatively low computational time;
therefore, they are suitable for most of the engineering problems where the goal is to
determine the final state of the system.

Transient-state simulations are used to determine the evolution of a system over time. These
provide more information on the phenomena taking place in a membrane system and can
be used for fundamental analysis (e.g., flow pattern analysis). In transient-state simulations
a steady solution is computed at each timestep which involves large computational time.

19.2.4 Initial conditions

The finite volume method requires an initial condition for each element as any numerical
method. The selection of the initial condition is normally based upon previous information
on the model and using an arbitrary set of initial conditions may lead to excessive
computational time. Using a good approximation of the transient-state as an initial
condition can significantly speed up the simulations. This can be achieved by setting a
previous steady-state solution as the initial condition for the transient case.

432
 Chapter 19

19.2.5 Meshing and algorithms

CFD is essentially a numerical technique that commonly uses a finite volume scheme
to calculate the properties of the flow within the domain. The finite volume method
is based on discretising the fluid domain in a finite number of smaller volumes, which
collectively form the mesh. All assigned variables are calculated for these elements and
some interpolation scheme is used to estimate their values at mid-points. The mesh can
be structured or unstructured depending on requirements of the application. In a structure
mesh the elements share the same pattern of construction, whereas in an unstructured mesh
has multiple patterns. Unstructured meshes are typically suitable to model MSPs since mass
transfer occurs differently at different locations of the domain. For MSPs, more elements
are required at regions where the concentration gradient is higher, such as close to the
membrane surface. The structured mesh is useful for regions where the flow is tangential to
the boundary. Most CFD models combine both types of meshing to increase the resolution
of the method at a low computational time. Figure 5 shows an example of the mesh used for
CFD modelling including an unstructured pattern for the bulk flow region and a structured
pattern for the region corresponding to the boundary layer (zoomed-in). A coarse mesh
is shown as an example for better visualisation; however, it is recommended that meshes
where the largest elements are about 5% of the channel height are used for CFD studies of
membrane channels under typical operation conditions.

Figure 5 Typical meshing approach for mass transfer analysis on a narrow membrane channel using
CFD.

Another meshing approach is using an adaptive mesh, which changes with every iteration
on an established criterion. The solute concentration gradient may be used as a refinement
criterion as it is advisable to have more elements in the regions with higher mass transfer.
When using adaptive meshing the software will automatically refine those regions where
the criterion is met at a certain number of iterations. Using adaptive refinement increases the

433
Experimental Methods for Membrane Applications

computational time required to solve each iteration, however, it can reduce significantly
the number of iterations required by the model. Both steady-state and transient-state cases
can be solved using adaptive meshing on ANSYS FLUENT and ANSYS CFX.

19.2.6 Convergence
The results obtained via CFD simulation change with every iteration or timestep and a
criterion is required to determine when to stop the simulation. The convergence criterion
for steady-state simulations is the residual which measures the imbalance of a conserved
variable for each control volume (ANSYS Inc., 2021). Given the numerical nature of the
method, the residuals cannot be zero, but an acceptable tolerance is set arbitrarily depending
on the precision required for the analysis. The residual tolerance can be set as maximum or
root mean square (RMS) with the former being stricter. A solution is considered converged
when the residual is equal or less than the tolerance setpoint. The residuals of the velocity
and pressure fields are computed by default in most of the commercial software, however,
the user can add additional criteria for other variables. It is important to note that using
stringent criteria for the residual, can significantly increase the required computational
time.

The velocity and pressure fields are typically sufficient to determine convergence in
terms of hydrodynamics, but not for mass transfer. Additional variables such as solute
concentration or mass transfer coefficient are suitable as a convergence criterion when
analysing mass transfer phenomena.

Transient-state simulations require the solution to be converged in time, therefore, the

relative error is computed between two different time-dependant states of the system,
often referred as ‘snapshots’. The error between two snapshots depends on the timestep
size; for example, using a very small timesteps will misleadingly reduce the relative error
significantly. The ratio between the relative error and the timestep size, which is analogous
to the time derivative, may be used to account for the effect of the timestep size. Another
approach is to compute a time-averaged variable within a time window and use it as
convergence criterion. The later approach is useful for oscillatory states where the variables
are not converging to a single value over time, but rather they oscillate within a limit cycle.

19.3 DATA ANALYSIS

19.3.1 Verification
Another essential step of CFD modelling is the verification of the results. At this stage
we already have precise results and now the main goal is to determine whether they
are accurate or not. Verification is done by comparing the results obtained against some
previously reported for a same (or similar) model or method. In most of the cases there is no
true value to aim for, unless there are explicit analytical solutions available for the variables
to be compared. Instead, there is a range within the results are considered acceptable. The
––
results used for verification are commonly the area-averaged Sherwood number (Sh)
and the global friction factor (fglob), which are indicators of mass transfer and energy loss,
––
respectively. The Sh and the fglob are described by the following equations:

434
 Chapter 19

dh w
Sh = Eq. 6
1 y
ww – (w – w )
2 b,out b,in
w,AA

d h ΔP
f glob = Eq. 7
2 ρuave
2
L

In Equation 6, ww is the solute concentration at the membrane surface and, wb,out and wb,in
are the solute concentration at the bulk at the outlet and the inlet, respectively. The partial
derivative term is the area-average of the y-component of the solute concentration gradient
vector at the membrane surface (m−1). Moreover, in Equation 7, r is the density of the fluid
(kg m−3), uave is the average velocity of the bulk flow (m s−1), ΔP is the pressure drop across
the channel (kg m−1 s−2) and L is the channel length (m). Despite being good indicators of
––
the channel performance, using Sh and fglob for verification, overlooks the local aspects of
the flow.

Another type of verification that is typically carried out in CFD modelling is mesh
independency analysis. Mesh independency studies are used to quantify the error
introduced by the meshing strategy by computing the flow variables using different meshes.
The error for a variable is computed for meshes with different number of elements until
certain tolerance is reached. The variables selected should be representative for the whole
domain and for the phenomena that is being studied (e.g., hydrodynamics, mass transfer or
heat transfer). In addition, the effect of the number of elements should be taken into account
as, for example, the error between two meshes may be small because they have relatively the
same number of elements. First developed by Roache (1997), the grid convergence index
(GCI) is one of the first introduced criterion for mesh independency studies as it accounts
for the entire domain and for the number of elements of the mesh. The GCI for a fine mesh
(GCIfine) is described by:

3 | e | Rη Eq. 8
GCI fine =
Rη – 1

where η is the order of accuracy (or the dimension number), e is the relative error for an
integral function between two meshes and R is the growth factor for the number of
elements. Values for the GCIfine less than 5% are typically considered acceptable but, of
course, this depends on the application of the model. Although this criterion is very
suitable for structured meshes, it presents some issues when used for unstructured meshes,
as increasing the number of elements does not necessarily decrease the error. Refining in
regions where the gradients (either velocity or concentration) are very small may cause an
‘artificial’ decrease in the GCI, while increasing the number of elements in a relatively small
amount can lead to a higher GCI.

The GCI as a verification metric has more recently been replaced by alternative criteria for
mesh independency analysis, as it has been shown that unstructured meshes are more
suitable for modelling membrane systems. Recent CFD studies focus on the behaviour of

435
Experimental Methods for Membrane Applications
––
integral functions (e.g., Sh and fglob) in an effort to circumvent the effect of the number of
elements of the mesh on the GCI. A typical approach consists in computing these integral
functions for multiple discretisation meshes with an increasing number of elements, while
also computing the relative error between each other until an asymptotical behaviour of the
integral function is observed. An arbitrary threshold for the relative error is set depending on
the level of accuracy required for the study, with 5% being acceptable for most applications.
It is important to highlight that this verification approach requires using the same meshing
strategy for all the meshes compared.

19.3.2 Validation
Validation assessment determines if the computational results agree with physical
reality (NASA, 2022). The results obtained throughout a CFD simulation are compared
against experimental data in order to validate them. In another words, validation is used
to determine if the computational model is actually representative of the physical system
that was intended to analyse. Some physical systems may be very difficult to setup or to
monitor, this is why in some cases CFD analysis comes before and its validation can become
an issue. In these cases, the validation may be performed by comparing against experimental
results from a different model with similar characteristics.

Existing techniques for experimental analysis of flows include particle image velocimetry
(PIV), micro-particle image velocimetry (μPIV) and holographic particle image velocimetry
(HPIV). These techniques use optical devices such as lasers, cameras and lenses to capture
the flow pattern structures in the fluid. PIV is based on the light scattering capacity of small
particles that often need to be added to the fluid. A high-speed camera captures snapshots
of the tracer particles distribution to visualise the flow pattern (Lindken & Burgmann,
2012). μPIV refers to the PIV variation where the fluid motion can be determined by the
resolved length scales, ranging from 10−4 m to 10−7 m. A specialised high-resolution camera
is required to achieve microscopic length scales for μPIV. The application of μPIV include
visualisation of fluid patterns in microchannels, which makes this technique suitable for
flow analysis in membrane narrow channels. Figure 6a shows the typical components of the
experimental set-up for PIV and μPIV. A flow field captured using PIV is shown in Figure 6b.

Flow with
tracer particles

Lens Light-sheet optics

Laser
Camera beam

Double-pulse
laser
1) 2)

Figure 6 (1) Experimental set-up for PIV/μPIV (Lindken & Burgmann, 2012) and (2) jet flow
image captured using PIV (Yu et al., 2021).

436
 Chapter 19

HPIV stores the amplitude and phase of a light wave in a two-dimensional film, the
hologram. Based on the wave interference phenomenon, the interference pattern of this
scattered wave is compared against a second wave refer to as the reference wave (Hinsch,
2002). The interference pattern is then used to reconstruct the flow field by illuminating
the hologram with a replica of the reference wave. Figure 7a depicts a basic set-up for HPIV
with its components. In general, the laser beam is split into an object wave and a reference
wave using an arrange of optical devices. The object wave is directed towards the flow and
the reference wave to the holographic film (or device, for digital HPIV). Figure 7b shows an
example of a HPIV hologram, whereas the corresponding particle distribution is shown in
Figure 7c (Sun et al., 2020).

Pulsed laser
M1

M3
L3
Flow
2)

M2
BS L1 L2

Holographic plate

1) 3)

Figure 7 (1) Schematic of a HPIV set-up (Bryanston-Cross et al., 1992), (2) hologram of an
interference pattern and (3) corresponding particle distribution (Sun et al., 2020).

Despite the excellent performance of the aforementioned experimental methods for flow
characterisation, the applications for membrane systems are limited. For example, the small
solutes (salts or organic molecules) common in membrane systems are difficult to emulate
for PIV since, the latter requires particles to have certain characteristics to be able to scatter
light.

19.4 DATA DISCUSSION AND INTERPRETATION

19.4.1.1 Data processing and assessment

Simulations may yield large amounts of data and their interpretation is often the
cornerstone of CFD analysis. Numerical data has very little to no value if it is not processed
and interpreted correctly to draw information about a physical system, as that is the
purpose of CFD modelling. When presenting results, data is most often used to verify or
validate the model, to explain physical phenomena taking place in the system or to propose
improvements. CFD may also be coupled with different mathematic tools such as, calculus,
Fourier analysis, statistics and linear algebra to produce valuable data for membrane system
modelling. In addition, data science (Holemans et al., 2022) and machine learning (Vinuesa
& Brunton, 2022) are increasingly used to make CFD a more powerful tool for modelling.

437
Experimental Methods for Membrane Applications

19.5 APPLICATIONS, EXAMPLES

19.5.1 Flow stability

19.5.1.1 Laminar steady
The first uses of CFD for membrane modelling where focused on optimising the design
of mesh spacers (Da Costa et al., 1994). These studies used laminar flow in spacer-filled
membrane channels to gather information about the CP phenomena. Cao et al. (2001)
found that recirculation zones are formed before and after the spacers in a narrow membrane
for laminar flow. The presence of recirculation zones significantly increases the mass
transfer enhancement as they introduce low concentration flow to the boundary layer.
Nevertheless, the effect of recirculation zones on mass transfer is limited as they remain
attached to the boundary and eventually stabilise. Ahmad et al. (2005) went a step further
by studying the mechanism by which the recirculation zones increase mass transfer, finding
that increasing the shear rate at the membrane surface is critical to mitigate CP. As the role
of shear rate (velocity in the normal direction) on the boundary layer development became
more evident, the efforts were diverted towards finding ways to destabilise the boundary
layer by increasing shear rate.

19.5.1.2 Laminar unsteady – oscillating vs vortex shedding

The fluid flow pattern may change over time under specific conditions without being
considered turbulent. This flow regime is called laminar unsteady and it is characterised by
unsteady recirculation zones. Recirculation zones may contract and expand or move while
remaining attached to the wall affecting the boundary layer. Figure 8a shows a flow pattern
for a recirculation zone. Destabilising recirculation zones may cause them to detach and
get convected by the bulk flow in a phenomenon called vortex shedding. The efficiency of
vortex shedding to mitigate CP is higher than that of recirculation zones as these only affect a
small portion of the channel. Vortex shedding is characterised by two separate regions with
swirl motion with one of them detaching from the boundary and moving downstream. The
double swirl pattern for vortex shedding is shown in Figure 8b.

Recirculation zone

Vortex shedding

Figure 8 Fluid flow patterns for (top) recirculation zones and (bottom) vortex shedding in a
spacer-filled membrane channel (Fimbres-Weihs et al., 2006).

438
 Chapter 19

Vortex shedding occurs in flows at higher Reynolds number translating into more energy
requirements. Because of this, some researchers have focused on finding the optimal
Reynolds number that can cause vortex shedding, but at a lower energy requirement.
Alexiadis et al. (2007) found that the critical Reynolds number (Recr) for the transition from
recirculation zones to vortex shedding in spacer-filled membrane channels is within the
range of 526-841.

Fimbres-Weihs et al. (2006) studied mass transfer under unsteady-flow conditions at

hydraulic Reynolds numbers up to 1683, paying special attention to the causes of mass
transfer enhancement. Wall shear was identified as having a significant impact on mass
transfer enhancement; however, the inflow of lower concentration towards the membrane
surface was found to dominate mass transfer phenomena. Thus, recent CFD studies on mass
transfer enhancement are focused on maximising the inflow towards the membrane surface
(i.e., normal velocity).

19.5.1.3 Quasiperiodic flow

The study of fluid flow instabilities is crucial as the effect of disturbance on the hydrodynamic
stability and how the flow system responds to it remain unclear. Fluid periodicity is one of
the indicators to measure the flow stability. Fluid periodicity in flow refers to a flow system
that displays a recurring behaviour given at regular intervals. Quasiperiodic flow, on the
other hand, refers to a recurrence behaviour with a component of uncertainty, such that it
could be periodic on a small scale but unpredictable on a large scale, which would eventually
result in imprecise measurements.

From fluid mechanic point of view, the flow can be classified into three states namely the
laminar, transition and turbulent flow. The fluid flow states are usually determined by a
dimensionless number known as Reynolds number. Reynolds number (Re) is defined
as the ratio of inertial forces to viscous forces in a fluid flow. As Re increases, the inertial
forces become larger, leading to flow instability (Masuda & Tagawa, 2019). Transition
flow is a mixture of laminar and turbulent flow occurring simultaneously in a fluid channel
with a range of Re between 2,000−4,000. During the transition state, aperiodic oscillatory
flow or quasiperiodic flow usually occurs due to the unstable flow condition with mixed
characteristics of laminar and turbulent flow characteristics. Nowadays, Computational
Fluid Dynamics (CFD) has been widely used to study the hydrodynamics behaviour of
quasiperiodic flow. One of the advantages of flow simulation by CFD is that it can help to
generate different physical data such as vorticity or energy dissipation rate which cannot be
measured easily in practical experiments.

Schwinge et al., (2002a) investigated the effect of unsteady 2D flow in narrow spacer-filled
channels for spiral wound membrane modules. Their simulation results show that when the
membrane channel is filled with obstacles (spacer), an unsteady flow pattern is observed at
Re as low as 200, depending on the geometry of the obstacles. Figure 9 shows the transition
of flow from stable to unsteady conditions occurring at Re above 300 for a single filament
located at the centre of a narrow channel. At this Reynolds number, the recirculation begins
to form behind the filament and as Re increases, the extent of flow unsteadiness increases
until it reaches a turbulent condition. For a single filament located close to the bottom wall,
the flow is found to be stable for Re below 600 and it becomes unstable when Re increases

439
Experimental Methods for Membrane Applications

above 600 as the flow disturbance induced by the recirculation behind the filament spreads
the unsteadiness downstream the membrane channel (as shown in Figure 10). Nevertheless,
their transient results also found that the channel walls close to the cylindrical filament tend
to stabilise the fluid flow and slow down the transition of flow to the unsteady state.

0.10 Unsteadiness caused as

the effect of the recirculation
region is convected downstream
0.05
V-velocity (ms-1)

0.0

–0.05

–0.10
0.32 0.33 0.34 0.35 0.36
Time (s)

Figure 9 Unsteady flow caused by a cylinder located in the centre of the narrow channel for
Re = 500 (Schwinge et al., 2002a)

0.04 -5.00E-04
V-velocity (ms-1)

V-velocity (ms-1)

0.03

0.02 -1.00E-03

0.01

0.00 -1.50E-03
0.15 0.16 0.17 0.18 0.19 0.20 0.21 0.22 0.10 0.12 0.14 0.16 0.18 0.20 0.22 0.24
Time (s) Time (s)

Figure 10 Unsteady flow caused by the cylinder close to the bottom wall for Re = 1,000 and small
flow disturbances are convected downstream (Schwinge et al., 2002a)

440
 Chapter 19

19.5.1.4 Turbulent flow

Turbulent flow generally occurs at a Reynolds number greater than 2,000. As Reynolds
number increases, the fluid flow undergoes irregular fluctuations or mixing that eventually
leads to an aperiodic oscillation. Not only that, the direction and magnitude of fluid flow are
constantly changing at a turbulent condition. The characteristics of a turbulent flow include
higher velocities, low viscosity and higher characteristic linear dimension compared to a
laminar flow. Due to the random nature and irregularity of turbulent flow, the flow pattern
is extremely difficult to understand. Hence, the governing equations for turbulent flow
condition are not easy to develop due to the unsteady flow continuously changing with time,
which increases the difficulty level for studying flow turbulence. Although the analysis of
turbulent flow is very challenging, analysis of flow turbulence is important for industries as
currently there are more membrane filtration processes are using turbulent flow to enhance
mixing and reduce the extent of concentration polarisation in the membrane module.

In general, there are three main ways to simulate turbulent flow using CFD technique,
including the Direct Numerical Simulation (DNS), Large Eddy Simulation (LES), and
Reynolds-Averaged Navier-Stokes (RANS) equations. The computational cost associated
with solving the flow simulation increases ascendingly with the order of RANS, LES,
followed by the DNS technique in CFD modelling. For DNS and LES methods, CFD will
take into account the majority of fluid scales in simulation and provide a comprehensive
flow data. In contrast to DNS and LES, RANS do not require fine details of all the turbulent
eddies, hence lower computational cost. However, the accuracy of flow simulation can be
impacted for RANS method because the flow is only simply modelled and not fully (DNS)
or partially (LES) resolved.

Jafarkhani et al. (2012) developed a 3D model with semi-circular baffles incorporated into
the membrane tube to study the hydrodynamic behaviour of turbulent flow. Their results
found that the intense fluctuations induced by the baffles increase the local wall shear stress
and velocity in the membrane channel. The induced turbulence to the bulk flow at Re up to
7,500 results in a rapid change in the flow directions, which enhances the flow fluctuations
and reduces the formation of concentration boundary layer on the membrane surface,
consequently leading to a potential fouling reduction while improving the filtration flux
performance.

19.5.2 Mass transfer and vortex shedding

Vortex shedding occurs as an oscillating flow at specific velocities when a fluid flow passed
a bluff body depending on the size and shape of the body. The alternate formation and
shedding of vortices produce alternating forces, which happen more frequently as flow
velocity increases. This phenomenon plays a vital role in membrane technology due to the
significant impact of vortex shedding on membrane efficiency, as well as the mass and heat
transfer of a fluid flow passed a bluff body (Korinek et al., 2017). Research has shown that
vortex shedding is capable of enhancing membrane separation efficiency by increasing the
transmembrane flow, while disrupting the formation of thermal and solute boundary layer
on the membrane surfaces.

441
Experimental Methods for Membrane Applications

Su et al., (2018) investigated the performance of a vibration-enhanced reverse osmosis

membrane module on the membrane separation efficiency. Their CFD results showed that
the vibration force results in more vortices along time at the downstream and upstream faces
of the membrane, compared to that of non-vibration case (Figure 11). The vortex generation
induced by the vibration increases the boundary shear rate in the membrane channel, which
significantly reduces CP.

8
3
4 5 2

8
3
2

(c) t = 5 ms

Figure 11 Velocity vector profile of vibration case (top) and non-vibration case (bottom) (Su et al.,
2018).

19.5.3 Spacer design

19.5.3.1 Two-Dimensional Feed spacer

The accumulation of solute due to membrane rejection leads to concentration polarisation
(CP), which later causes the deterioration of membrane flux. Thus, intensive studies such
as optimisation of feed spacer geometry in the membrane channel have been performed
to mitigate CP by enhancing the mixing of fluid. The first two-dimensional (2D) models
of spacer-filled membrane channels appeared in the early 2000s, aiming to understand its
hydrodynamic (Chong et al., 2022a) and concentration profiles. Later, the configurations
of the spacer (e.g., cavity, submerged and zigzag) and other parameters (filament diameter,
mesh length, etc.) have been studied thoroughly to understand their impacts to the flux
enhancement and pressure loss.

Early 2D feed spacer study found that zig-zag spacer was the most efficient spacer for SWM
module (Schwinge et al., 2002b). Subsequent findings include 1) dependence of formation
of recirculation region on spacer geometry and flow condition (Schwinge et al., 2002a);
2) significance of vortex shedding for mass transfer enhancement. It was also concluded
that the mass transfer enhancement depends on two important mechanism: 1) flow of
low solute concentration to the membrane boundary layer and 2) increase of wall shear
(Fimbres-Weihs et al., 2006).

19.5.3.2 Three-Dimensional Feed Spacer

3D modelling of spacer-filled channel emerged rapidly around 2010 due to improvement
of computer processing and storage of capacity. During the first decade of 21st century,
the research direction regarding CFD analysis in spacer studies mainly focused on the
hydrodynamic performance caused by the spacer. Later, the effect of various geometries such

442
 Chapter 19

as the filament size, mesh length and flow attack angle on the mass transfer performance and
pressure loss reduction were studied and investigated (Gu et al., 2017). In the past decade,
the research direction concerning CFD analysis of spacer-filled channels has shifted and
focused more on the novel spacer geometries for further improvement on the mass transfer
performance and reduction in the pressure loss (Park et al., 2021).

Chong et al. (2022) studied the effects of submerged spacers with variations in the node
geometries and sizes (as shown in Figure 12a) on the hydrodynamics and mass transfer
performance through CFD. It is found that conventional spacers have similar or higher Sh
than the submerged type spacers at lower Reh (<100) because of the sideway flow. However,
due to a greater vortex mixing effect, submerged type spacers performed better than the
conventional spacers at higher Reh (>200). In addition, the mass transfer performance of the
spherical node spacer is inferior to the column node spacer as the flow can pass through the
gap between the spherical nodes and membrane, thereby resulting in a lower local velocity
at filament and mass transfer. In fact, as node size increases, pressure loss increases as well
because of the larger nodes significantly impeding the flow, creating more form drag and
skin friction in the membrane channel, thus increasing the pressure loss.

Recently, a honeycomb shape spacer (Figure 12b) was proposed in which it has the ability to
generate high-magnitude turbulent kinetic energy in the area of spacer filaments, resulting
in a smaller fouling deposit (Park et al., 2021). Furthermore, optical coherence tomography
(OCT) scans showed the fouling layer thickness could be decreased by 33% using the
honeycomb spacer.

a) b)
df/hch= 0.6 dn/hch=1.2
Spherical Column

ds

Lx

Ly

Lz

Figure 12 Schematic diagram of (a) spherical nodes submerged and column nodes submerged
spacers and (b) honeycomb spacer (Park et al., 2021)

19.5.4 Flow perturbation

One of the efforts to minimise concentration polarisation (and eventually fouling) in a
membrane system is by introducing a disturbance into the fluid flow through an external
flow perturbation technique. With the help of CFD, the technique is performed in an
attempt to induce flow unsteadiness for promoting fluid mixing and therefore, reducing
the fouling tendency in membrane modules (Fimbres-Weihs et al., 2006; Schwinge et al.,

443
Experimental Methods for Membrane Applications

2002a). The flow perturbation technique generally involves creating unsteadiness to the
bulk flow, which can be done by introducing an oscillating flow to the system; or causing a
disturbance to the boundary layer adjacent to the membrane surface by using vibration or
electro-osmosis approach. Further descriptions regarding the flow perturbation techniques
as well as their effectiveness in enhancing membrane performance are elaborated in the
following sections.

19.5.4.1 Electro-osmosis
Electro-osmosis is an electrokinetic phenomenon involving the movement of a thin fluid
layer adjacent to a charged surface in response to an external electric field (Asadi et al., 2013;
Hu & Li, 2007; Jagannadh & Muralidhara, 1996; Ouyang et al., 2013). The electrokinetic
phenomenon occurs due to the electrostatic interaction between the similarly charged and
oppositely charged ions in the vicinity of a solid/liquid surface, which then results in the
motion of a thin fluid layer near the surface, as illustrated in Figure 13. Electro-osmosis has
great potential to enhance mass transfer while minimising fouling tendency, particularly
for membrane separation processes, such as reverse osmosis and nanofiltration because a
disruption to the flow near the membrane surface tends to promote back-transport of solute
and reduce polarisation effect (Liang et al., 2014a). Further, the approach is suitable for water
treatment and desalination processes as there are salts and charged species involved in the
system that can respond to the applied electric field (Jagannadh & Muralidhara, 1996).

Direction of bulk flow

– – – – – – – – – – – – – – – – – – – –
+ + + + + + + + + + + + + + + + + + + +

u(y)

+ + + + + + + + + + + + + + + + + + + +
– – – – – – – – – – – – – – – – – – – –
y
Charged surface
x

Figure 13 Schematic of electro-osmosis technique with red arrows representing the electro-osmotic
flow (EOF) adjacent to a charged surface in an empty flow channel.

Several theories have been developed to describe electro-osmosis, including the Helmholtz-
Smoluchowski (HS) theory, the Spiegler friction model, the Schmid theory, and ion
hydration theory. Among those early theoretical approaches, an early study of Spiegler
and Macleish (1981) investigated the electro-osmosis technique on a reverse osmosis
desalination process by filling the feed stream with ferric hydroxide. Their results showed
that the electro-osmotic backwashing of the membrane managed to recover a range of
30–100% of flux loss in the system. Despite the great potential of electro-osmosis, the
technique is difficult to analyse experimentally in a membrane system due to various technical

444
 Chapter 19

constraints. Hence, there is significant value in utilising CFD as a tool for understanding the
mechanisms of electro-osmotic flow (EOF) in enhancing mass transfer and fluid flow in a
membrane process. The following sections further discuss the modelling of electro-osmosis
in membrane system studies and the many learnings obtained using CFD.

19.5.4.2 Modelling electro-osmosis in CFD

Electro-osmotic flow refers to the motion of a thin fluid layer that carries a net electric charge
acting upon a fluid/solid interface (i.e., membrane surface) in response to an electric field.
This layer of net charge is known as the electric double layer (EDL) and is often characterised
by the Debye length (λD) (Cummings et al., 2000), which is formed due to the charge
separation occurred near the fluid/solid interface. The movement of fluid layer (EOF) is
then driven by the excess charged ions in the double layer field via viscous effect under the
influence of external electric field (Hu & Li, 2007; Probstein, 1989; Russel et al., 1991).
For an incompressible fluid flow in a membrane system, the electro-osmotic effects can be
introduced into the Navier-Stokes momentum equation by means of an external force given
as follows (Hu & Li, 2007):
v
t
+ (v ) v = μ 2
v – p+ e
E+ g Eq. 9

where r, v, p, μ, re, E, and g are the density, velocity vector, pressure, fluid viscosity, electric
charge density, electric field, and gravitational acceleration, respectively. The electric field
(E) can be calculated as follows:

E=– ( + ) Eq. 10

where f and y are the potentials due to the double layer and external electric field,
respectively (Patankar & Hu, 1998; Rawool & Mitra, 2006). The charge density (re) can be
related to the electric potential by the Poisson equation (Hunter, 2013):

–
2
= e Eq. 11

where ε is the fluid permittivity. Excluding any disturbance to the double layer, the charge
density can be solved by the Boltzmann distribution or other similar correlations (Hunter,
2013; Probstein, 1989).

Nevertheless, the high computational requirement and cost is one of the challenges for
solving the Poisson and Navier-Stokes equations at the scale of Debye length ranging up
to several hundred nanometres in numerical simulation of electro-osmosis considering a
typical membrane channel height of ~10−3 m. For the sake of simplicity, the thickness of
double layer is generally assumed to be neglected in most EOF studies (Hu & Li, 2007).
In fact, the coupled relations between the Poisson and Navier-Stokes equations can be
simplified by dropping the EOF term reE from equation 9 and instead, replace the no-slip
boundary condition at channel surfaces by using the Helmholtz-Smoluchowski (HS) slip
boundary (Anderson, 1989; Hu & Li, 2007; Ren et al., 2003; Santiago, 2001; Zhang et al.,
2006).

445
Experimental Methods for Membrane Applications

Using the simplified HS approximation, the electro-osmotic flow can be incorporated

as an artificial forced slip velocity applied along the surface of the membrane channel in
a separation module. The HS slip velocity is applied outside the edge of the double layer
assuming the thickness of EDL much smaller compared to the channel height (Probstein,
1989). In addition, the slip velocity equation assumes a 1D charge distribution when the
fluid velocity is small and/or the inertial terms in the momentum equation do not dominate
(Patankar & Hu, 1998). Assuming time constant with negligible pressure gradient and
gravitational acceleration, the coupled system of Poisson and Navier-Stokes equation from
equation 9 can be simplified to the following HS equation, which corresponds linearly with
the magnitude of electric field, expressed as follows (Probstein, 1989):
Ex
us = Eq. 12
μ

where us is the forced slip velocity, ζ is the zeta potential, and Ex is the magnitude of electric
field in the x-direction. The permittivity (ε) is assumed to be uniform for the case of an RO
system, such that the permittivity value of the membrane can be regarded as similar to that
of water (Liang et al., 2014a). This is because of the structure of RO membrane which is
mostly comprised of the high porosity support layers (Antony et al., 2013), along with
an extremely fine selective membrane layer of 1×10−7 m or less (Baker, 2004), in order
to assume a uniform permittivity value and to safely neglect any effect related with the
nonuniformity in this case.

The HS approximation was compared against a more rigorous charge density distribution
(CD) solution under the influence of uniform and non-uniform electric fields (Liang et
al., 2014b). The case study with uniform electric field showed that the HS approximation
agrees well with the CD solution at increasing solute concentration in a 2D unobstructed
membrane channel. Their results found that the HS approximation is more accurate at higher
solute mass fractions of 0.001 or more, which are the typical salt profiles encountered in RO
desalination process. Hence, the HS forced-slip model is suitable for modelling the electro-
osmotic effect in an RO membrane system.

19.5.4.3 Significant learnings of EOF slip velocity in CFD studies

The HS slip velocity model has been extensively validated and is suitable to be applied
for the typical flow conditions encountered in real RO membrane modules (Liang et al.,
2014a; Liang et al., 2014b, 2016b). It is reported that the EOF perturbation induced near
the membrane surface results in the shedding of vortices in a 2D spacer-filled membrane
channel, which tends to enhance fluid mixing and reduce concentration polarisation (as
shown in Figure 14). In addition, the EOF induced vortex shedding also increases wall shear
along the membrane that could potentially reduce the fouling tendency in the membrane
system (Liang et al., 2016a; Liang et al., 2016b). The effectiveness of EOF slip velocity in
enhancing membrane performance and the significant findings in CFD membrane studies
are further elaborated in Table 3.

446
 Chapter 19

Table 3 Significant findings of application of electro-osmosis in membrane system

Membrane model Reference Main findings

2D unobstructed RO Chan et al. (2020b) A reduced-order model was developed for fast predictions of
rectangular channel concentration polarisation in an RO membrane system under
permeation condition.
2D unobstructed RO Chan et al. (2020a) A reduced-order model was proposed to study the effect
rectangular channel of permeation in an RO membrane system with EOF slip
velocity considered.
2D spacer-filled Liang et al. (2016b) First CFD study incorporating steady and unsteady EOF for
channel for RO SWM enhancing mass transfer in 2D spacer-filled channels, using
modules HS forced-slip approximation.
2D unobstructed RO Ratnayake et al. A spatio-temporal frequency response analysis was
membrane channel (2016) performed to investigate the effect of waves of different
frequencies for an EOF forced slip velocity and the impact of
changes on solute concentration gradients.
2D spacer-filled Liang et al. (2018a) CFD study of non-sinusoidal waveforms of EOF slip velocity
channel for RO SWM reported a similar membrane performance as those obtained
modules by sinusoidal slip velocity, in terms of both mass transfer and
wall shear.
2D spacer-filled RO Liang et al. (2020a) A comparison study between forced slip velocity and
membrane channel oscillating feed perturbation found that both approaches
predict similarly in terms of hydrodynamics and flux
performance.
2D spacer-filled RO Foo et al. (2020) CFD study of the effect of varying feed spacer geometries
membrane channel on membrane performance revealed that the resonant
slip frequency increases as spacer size is increased due to
stronger shear layer interactions. An increased in distance
between spacers leads to a greater flux due to forced-slip,
albeit the actual flux is smaller.
2D spacer-filled Foo et al. (2021) CFD study of varying spacer configurations found that the
channel for RO SWM submerged configuration results in the largest flux increase
modules under forced-slip effect at the expense of a relatively larger
pressure drop.

19.5.4.4 Oscillating flow

Inducing oscillations in the bulk flow is an approach to enhance mass transfer in membrane
systems by generating instabilities. Li et al. (1998) studied the use of oscillating inflow to
promote vortex shedding at lower Re in microfiltration systems. Their results suggested that
using a time-dependant inflow reduces the cake layer resistance and, therefore, increases
the flux. CFD modelling was used by Liang et al. (2018b) to show that an oscillating inflow
may promote vortex shedding increasing the magnitude of the flow velocity towards the
membrane surface which was later validated by Liang et al. (2020b).

The oscillatory inflow (OI) approach is implemented in CFD by changing the inlet velocity
from time-independent to time-dependant, giving place to a waveform. Sine and cosine
functions are typically used as waveforms for OI, though, others like square and sawtooth
waves are also possible to use. The sine waveform for OI is described by:

447
Experimental Methods for Membrane Applications

uosc = uave ⎡⎣1+ Asin(2π ft)⎤⎦ Eq. 13

where uosc is the oscillatory inlet velocity (m s−1), uave is the average inlet velocity, A is the
oscillation normalised amplitude, f is the oscillation frequency (s−1) and t is time (s). Figure
15 shows different waveforms previously used for the OI technique (plot as normalised
velocity).

No EOF slip
velocity

No vortex shedding High concentration

polarisation

With EOF
forced-slip

Vortex shedding Less concentration

polarisation

0 0.25 0.50 0.75 1.0 0.025 0.028 0.030 0.033 0.036

Velocity (ms-1) Solute mass fraction

Figure 14 Effect of EOF slip velocity on velocity and solute concentration profiles in a unit cell of
spacer-filled membrane channel (Foo et al., 2020).

-A
(a)
A

-A
(b)
A

-A
(c)
A
Figure 15 Waveforms for CFD analysis of OI technique:
0
(a) sine, (b) square, (c) triangular and (d)
-A sawtooth wave (Liang et al., 2018b).
(d)

448
 Chapter 19

The optimal frequency of the oscillation to maximise mass transfer enhancement lies
between 10 Hz and 1,000 Hz for typical spacer-filled membrane channels. A frequency
response analysis can be used to identify the optimal response frequency for a membrane
system. In a frequency response analysis, an input stimulus (commonly a short pulse) is
introduced in the system and the response in velocity or concentration is recorded to find
the maximum response. Recent approaches have undertaken the challenge of promoting
vortex shedding at the minimum energy expenses by optimising the oscillating inflow
amplitude and frequency.

19.5.4.5 Vibrations
Another approach to destabilise the boundary layer, instead of causing fluctuations in the
flow, is to constantly move the module. Vibration-assisted modules use a mechanical device
to induce oscillations on the channel enhancing mass transfer. Su et al. (2018) conducted a
CFD study on concentration polarisation and permeate flux in a vibration enhanced system
obtaining a reduction of up to 10% on the CP modulus when applying vibrations to the
module. In addition, both the Sherwood number and permeate flux increased in about
15% and 5%, respectively, compared to the case without vibrations. In the same work, the
results for permeate flux were validated experimentally. Studies on the effectiveness of
vibration-assisted modules to mitigate CP are relatively recent and they are mainly focused
on desalination processes (Li et al., 2017; Su et al., 2019). Thus, multiple aspects may be
analysed to improve this technique such as optimal frequency and amplitude, waveform,
etc.

19.5.5 Fouling modelling

Fouling phenomenon occurs in membranes due to solute deposited into the membrane
pores or onto the surface. The phenomenon usually begins with a condition known as
concentration polarisation (CP), which involves the solute accumulation near the membrane
surface due to a greater applied pressure, compared to the osmotic pressure difference in
the membrane process. There are two major types of fouling namely particulate fouling and
biofouling phenomena. Particulate fouling occurs when foreign particles such as proteins,
carbohydrates, oil, silt or clay are deposited into the membrane pores, leading to pore-
blocking. The biofouling phenomenon, on the other hand, occurs due to the precipitation
of microorganisms on the membrane surface, which leads to the formation of biofilm on the
solid material. It is known that a prolonged fouling phenomenon would reduce the overall
performance of a membrane module as permeate flux and membrane separation efficiency
decrease (Fritzmann et al., 2007).

19.5.5.1 Particulate fouling

Particulate fouling generally occurs when a solution containing particles with sizes ranging
from few nanometres to micrometres dispersed evenly in the solution, eventually clogging
the membrane pores. Particulate fouling condition can be reversed or becomes non-reversible
depending on the types of foulant binding to the membrane surface. The phenomenon is
said to be reversible if the particulate removal can be achieved via physical cleaning, and it is
irreversible when foulant requires a chemical cleaning (Leam et al., 2020). The particulates
can be categorized into organic and inorganic substances, such that the organic particles are
usually made up of proteins, carbohydrates, oils, etc., while the inorganic particles consist

449
Experimental Methods for Membrane Applications

of silt, clay, silica sediments, etc (Qasim et al., 2019). A continuous deposition of particles
near the membrane surface over time results in the formation of a cake layer adjacent to the
surface, which would impose a cake resistance onto the mass flow and eventually reduce
the mass transport across the membrane, leading to a decrease in permeate flux (Jiang et al.,
2017).

For the past few decades, CFD has been used to analyse the effects of particulate fouling on
membrane performance as the software is capable of providing analysis for numerous flow
and heat transfer mechanism, mass transport of soluble substrate, and the hydrodynamics
of fluid flowing through the membrane for optimisation of membrane design through
model simulations. A study on the particle deposition in a spiral wound membrane module
was conducted to identify the optimum spacer type for preventing particulate fouling (Li
et al., 2012). Flow that has lower velocity will lead to lower shear stress across membrane
surface that promotes particulate fouling. Their study addressed the effects of curvature on
the flow pattern for four different types of spacer configurations, namely zigzag, submerged,
i-cavity, and o-cavity, respectively by changing the dimensionless radius of curvature, η
given as follows:
Ro + Ri
η= Eq. 14
Rave

where Ro is the outer radius, Ri is the inner radius, and Rave is their arithmetic mean (Li &
Tung, 2008).

Based on the simulation results for empty channels and four spacer configurations of
spacer-filled channels, a higher shear stress has the potential for a lower tendency for
particle deposition as well as fouling. For submerged spacer-filled channels, it was found
that the deposition ratio is greater at the outer membrane compared with that at the surface
of the inner membrane. This can be explained by the lower shear stress observed at the
surface of the outer membrane compared with that at the inner membrane due to a lower
flow velocity near the outer membrane surface, which promotes the particulate fouling at
the outer membrane surface. An illustration that depicts the deposition ratio at different
positions of membrane in the submerged spacer-filled channels is shown in Figure 16.

In the case of asymmetric spacer-filled channels, it is reported that curve channels result in
a higher particulate deposition ratio at the outer membrane surface compared with that of
the flat channel because of a lower shear stress. This occurs because the curve channel tends
to direct the fluid flow into the inner membrane surface, which restrains the recirculation
size and decreases the velocity near the inner surface of the membrane. A summary on the
significant findings of particulate fouling in membrane studies by CFD is listed in Table 4.

450
 Chapter 19
0.0 0.540

0.473
Outer membrane (η = 0)
Deposition ratios at the re;ative position (%) Inner membrane (η = 0)
1.5 Outer membrane (η = 0.18) 0.405
Inner membrane (η = 0.18)

0.338

3.0 0.270

0.203

1.5 0.135

0.068

0.0 0.000
1C
1D
2A
2B
2C
2D
3A
3B
3C
3D
4A
4B
4C
4D
5A
5B
5C
5D
6A
6B
6C
6D
7A
7B
(m/s)
Position along spacer-filled channel

Figure 16 The deposition ratios on the inner and outer membranes in the flat and curved channel
filled with submerged spacers at various positions (Li et al., 2012).

Table 4 Significant findings of CFD studies on particulate fouling in membranes

Author Research types Main findings Observations
Lin et al. (2022a) CFD and Response Diagonal-flow feed channel results in The study is limited to
surface methodology a higher salt rejection and water flux feed channels without
(RSM) with an average crossflow velocity considering the effects of
in the channel increased by ~50%, feed spacer.
compared to that in the conventional
feed channel.
Rahimi et al. CFD and Fouling on membrane is not uniform The study is limited to
(2009) experimental and the possibilities of fouling to an incompressible flow
happen are higher in regions with system.
lower shear stress.
Li et al. (2012) CFD only Recirculation occurred adjacent This study is limited to a
to the membrane and behind the low volume fraction of
filament causing a higher shear discrete phase that ranges
stress, which potentially reduces less than 10-12%. It
particulate fouling in the membrane is also limited to a two-
system. phase flow.

451
Experimental Methods for Membrane Applications

19.5.5.2 Tracer test

Tracer test or commonly known as the Salt Tracer Response Technique (STRT), is
frequently used to evaluate the effects of cake-enhanced osmotic pressure (CEOP) on the
development of concentration polarisation in a membrane system. It is reported that a larger
extent of CEOP indicates an elevated solute concentration near the membrane surface due
to the formation of a cake-layer, which obstructs the back-diffusion of solute into the bulk
solution (Taheri et al., 2015). This eventually causes an increase in the transmembrane
pressure (TMP), in which more energy is required to push the fluid through the membrane.

CFD technique is used to assess and interpret the results from tracer test for estimating the
fouling resistance and concentration polarisation. A study conducted by Fimbres Weihs
& Wiley (2014) focuses on the effect of cake-enhanced osmotic pressure on particulate
fouling. The tracer test technique is used to monitor the TMP, permeate flux, and solute
concentration in the permeate by injecting sodium chloride as the tracer into the feed stream
of a membrane separation unit (Chong et al., 2007). A CP index is used to measure the extent
of CP, which can be defined in a one-dimensional mass balance differential equation, given
as follows:

ww – wp ⎛J ⎞
CP = = exp ⎜ v ⎟ Eq. 15
wb – wp ⎝ kmt ⎠

where CP is the local CP index based on local solute bulk concentration, kmt is the mass
transfer coefficient, and Jv is the volumetric permeate flux.

Results from the tracer test found that a step change in feed concentration does not affect the
degree of concentration polarisation across the membrane. The CP index was found to be
over-estimated, which can be explained by the entrance effects limited by spacers, such that
the presence of spacers separating the membrane leaves would eventually shift the velocity
profile of fluid towards the module inlet. The over-estimation in the fouled membrane is
greater and the overestimation of CP index decreases with higher tracer concentration but
increases with fouling layer mass, which reduces the membrane performance in terms of
reducing particulate fouling. However, it is important to make assumptions to maintain a
constant CPM index to identify precise error results on constant pressure tracer response
tests because changing salt concentration may alter the specific cake resistance that will lead
to deviation in over-estimation.

19.5.5.3 Biofouling
Biofouling phenomena occur due to the accumulation and adhesion of microorganisms on
the membrane surface, leading to the formation of a biofilm on the membrane. The biofilm
layer is commonly composed of an extracellular polymeric substance (EPS) matrix, which
is a polymer-like material enclosed with microorganisms on the surface (Unal, 2022).
The extent of biofouling can depend on several factors such as membrane characteristics,
influent composition and microorganism types. Most biofouling instances occur as a result

452
 Chapter 19

of a symbiotic interaction between bacteria, algae and fungus. Another fact to consider when
dealing with biofouling, is that a significant amount of the cells attached to the membrane are
dead cells, meaning that biofouling leads to organic fouling. Recent advances on membrane
science have focused on producing membranes with high surface hydrophobicity and low
surface roughness, as it has been found that membranes with these characteristics have a
lower tendency for biofouling, due to smaller exposed areas and less active sites for microbial
adhesion (Maddah & Chogle, 2016). Furthermore, the development of a biofilm depends,
in general, on the substrate availability and the type of microorganisms on the feed stream.

A recent study by Lin et al. (2022b) numerically and experimentally investigated the effects
of feed spacer geometries and channel porosity on the degree of biofouling in the membrane
system. Feed spacer geometries were varied in terms of the filament length, diameter, mesh
angle, as well as spacer thickness to obtain different channel porosities. The porosity of the
feed spacer membrane channel is calculated using the following equation:

D2 L
Vspacer 4 D2
channel
=1– = 1– = 1– Eq. 16
Vchannel HL2 sin 2HLsin

where D is filament diameter (m), L is filament length (m), H is spacer thickness (m), and α
is mesh angle. The channel porosity is an important factor in determining the feed channel
pressure (FCP) drop. A larger channel porosity will generally result in less FCP drop, as there
are less flow obstructions per unit of channel volume. Furthermore, larger porosity also
generally leads to less biomass accumulation, due to less stagnant flow regions and larger
shear.

Lin et al. (2022b) noted that several experimental studies report that higher shear stress
conditions tend to promote biofouling because of a higher nutrient load enhances biomass
accumulation, although the biofilm forming on high shear stress regions would be thinner
and more compacted. In addition, their analysis shows that the average cross-flow velocity
decreases with increasing filament length and spacer thickness due to reduced turbulence
caused by the number of spacer filament per unit channel length and the increment of space
between spacer filament and membrane surface, respectively. However, the average velocity
increases at a larger mesh angle or larger filament diameter. This is because a decrease in the
distance between the neighbouring spacer meshes and a narrow space between the spacer
filament and the membrane surface cause more turbulence in the fluid flow. It was also
found that an increase in the filament diameter and/or a decrease in the spacer thickness
promotes the accumulation of biomass, which favours the growth of bacteria on the
membrane surface as the interspace between the membrane and spacer filaments is reduced.
Therefore, a thicker feed spacer is recommended as higher feed channel porosity tends to
reduce biofouling in the membrane system (Bucs et al., 2014). A summary of significant
findings related to biofouling in membrane systems by CFD is presented in Table 5.

453
Experimental Methods for Membrane Applications

Table 5 Significant findings of CFD studies on biofouling in membrane system

Research
References types Main findings Observations

Vrouwenvelder CFD only The presence of spacer strongly affects This study is limited to one
et al. (2010) the level of pressure drop as an increment spacer geometry, specifically
of 10 times the pressure drop in the feed on the spacer thickness of
channel is reported, when compared with the 31 mil.
experimental case without spacer. Fouling
on the feed spacer is more important than
fouling on the membrane.
Gu et al. (2017) CFD only Fully woven spacers result in the highest The study is limited to
water flux and have lower average hydrodynamics and ignores
concentration polarisation moduli. mass transfer along the
channel.
Li et al. (2016) CFD only Regions with high concentration of biomass The study is limited to one
are isolated to zones near the spacer type of spacer geometry
filaments in which the flow is relatively static. without considering the
The fouling tendencies are higher in the effects of spacer geometry
stagnant regions because the particles can on CP.
settle down more easily.

Chen and Wu CFD only A larger average pore size of the membrane This study lacks focus on the
(2021) increases the nucleation frequency and effects of pressure drop in
growth rate of membrane biofouling. At determining the degree of
lower velocity, the flow rate decreases concentration polarisation.
more than 50%, partly contributed by the
decreased low permeate flux caused by
fouling.

An approach to model biofouling development and its effect on mass transfer in membrane
systems is to consider the biofilm as a different phase. The biofilm phase is considered to
have different mass transfer properties for the models. The solute concentration and the
biofilm are linked by introducing a mathematical model to describe the biomass growth as a
function of the substrate concentration. The Monod equation is possibly the simplest model
to describe the cell growth rate in terms of the substrate concentration. This differential
equation states that the cell growth rate is directly proportional to the number of cells, but
limited by the substrate availability. The Monod equation is described as:

dCX CS Eq. 17
= μ max C
dt K S + CS X

where CX is the biomass concentration (kg m−3), µmax is maximum growth rate (s−1),
CS (kgm−3) is the substrate concentration and KS is the half-velocity constant (kg m−3).

More sophisticated models may take into account the cell growth phase, the inhibitory
effect of the components of the media (e.g., sodium chloride, chlorine, reactive oxygen
species, etc.). Radu et al. (2010) conducted a numerical study on the biofilm formation in

454
 Chapter 19

a spacer-filled membrane channel coupling COMSOL and MATLAB. Figure 17 shows a

graphic overview of the process for biofouling modelling used by Radu et al. (2010). The
biomass growth rate is modelled using Monod equation with a random seeding on the
filament wall or membrane surface. A finite element mesh is used to discretise the fluid
and the biofilm sub-domains. The hydrodynamics, solute concentration and substrate
concentration are solved via CFD. The biofilm growth is modelled according to the substrate
distribution and then updated into the CFD model. Finally, an attachment/detachment step
is added based on the mechanical stress. This process is repeated for each timestep to follow
the time evolution of the biofilms.

(a) + (b)

Cx

(c)
C - Flux
1
u, v, p
- Salt passage
Biofilm development loop

- FCP
2

Cs

(d) 1+2 u, v, p, σ
Detachment loop

Cx

Cx

Cx

Figure 17 Adaptive algorithm for modelling of biofilm evolution using a hybrid CFD/numerical
approach (Radu et al., 2010).

19.6 ADDITIONAL CONSIDERATIONS

19.6.1 Multi-scale modelling

Despite of being a powerful tool for analysis, CFD studies typically focus only on a
small-scale model in order to facilitate the modelling of an MSP. Multi-scale modelling
refers to the analysis of fluids at different scales of space and/or time (Steinhauser, 2017).
While analysing the operational advantages of a certain technique in a small-scale model
may be useful to determine its effectivity, in some cases, it is worth to analyse the large-scale
effects as they are directly linked to the feasibility of the approach. For example, using an
instantaneous pulse as the inflow for a membrane system may enhance mass transfer over a
short period of time, but this effect is negligible compared to the large operation periods of
actual systems. Indicators such as pressure drop and permeate flux can be used to extrapolate

455
Experimental Methods for Membrane Applications

the efficiency of a full-length membrane system. Furthermore, optimising the separation

performance of the system can lead to energy savings during operation; although, these may
be negligible when compared to the energy requirements of the whole process.

19.6.1.1 Techno-economics
Techno-economic analyses are used to determine the cost-effectiveness of a process or a
modification made to it, by estimating the overall production cost of the product (Toh et al.,
2020a). A unitary cost per volume (e.g., dollars per m3) is used to reflect the processing cost
for producing treated water. A comprehensive techno-economic analysis would include
both capital and operational costs which are difficult to determine as they depend on
geographical and time factors (Toh et al., 2020a). The results obtained via CFD analysis are
useful to carry out a simplified techno-economic assessment considering only operational
costs related to the membrane operation such as pre-treatment, operating pressure, pressure
drop and permeate flux (Liang et al., 2019).

19.7 OUTLOOK

Instead of focusing on the spacer geometry, recent CFD studies on membrane systems are
focusing on new strategies to promote transient laminar flow and induce vortex shedding
such as unsteady inflow, vibration-assisted modules and electroosmosis. In addition, the
number of studies including experimental validation is increasing due to the necessity of
assessing the feasibility for real-life applications. At the time, there is a gap between CFD
studies and real-life applications which demands for more effective ways to scale-up the
models. Comprehensive techno-economic studies are increasingly being included to assess
the economic benefits of implementing a new technique.

Over the last decade, 2D CFD models are becoming sparse due to the increasing capability
of modern computers that significantly reduced the computational time required for 3D
simulations. The soaring development of new technology allows higher resolution with
less computational time and has introduced new approaches for data analysis. New studies
including advanced modelling techniques are frequently appearing in an effort to extract
the most information from CFD analysis. This modelling techniques include tools like
big data, machine learning and reduced-order modelling. The arrival of techniques such as
micro- and holographic-particle image velocimetry is allowing the validation of micro-scale
phenomena studied by CFD during the recent years.

456
 Chapter 19

19.8 REFERENCES

Ahmad, A. L., Lau, K. K., Bakar, M. A., & Shukor, S. A. (2005). Integrated CFD simulation of concentration
polarization in narrow membrane channel. Computers & Chemical Engineering, 29(10), 2087-
2095. doi:https://doi.org/10.1016/j.compchemeng.2005.06.001
Alexiadis, A., Wiley, D., Fletcher, D., & Bao, J. (2007). Laminar flow transitions in a 2D channel with
circular spacers. Industrial & Engineering Chemistry Research, 46(16), 5387-5396. doi:https://
doi.org/10.1021/ie0607797
Anderson, J. D., & Wendt, J. (1995). Computational fluid dynamics (Vol. 206): Springer.
Anderson, J. L. (1989). Colloid transport by interfacial forces. Annual review of fluid Mechanics, 21(1),
61-99.
ANSYS Inc. (2012). ANSYS CFX Solver theory guide. Canonsburg.
ANSYS Inc. (2020). ANSYS Fluent Users Guide. Canonsburg.
ANSYS Inc. (2021). ANSYS CFX Solver Manager User’s Guide. Canonsburg.
Antony, A., Chilcott, T., Coster, H., & Leslie, G. (2013). In situ structural and functional characterization
of reverse osmosis membranes using electrical impedance spectroscopy. Journal of Membrane
science, 425-426, 89-97. doi:https://doi.org/10.1016/j.memsci.2012.09.028
Asadi, A., Huat, B. B., Nahazanan, H., & Keykhah, H. A. (2013). Theory of electroosmosis in soil.
International Journal of Electrochemical Science, 8(1), 1016-1025.
Asefi, H., Alighardashi, A., Fazeli, M., & Fouladitajar, A. (2019). CFD modeling and simulation
of concentration polarization reduction by gas sparging cross-flow nanofiltration. Journal
of Environmental Chemical Engineering, 7(5), 103275. doi:https://doi.org/10.1016/j.
jece.20r19.103275
Baghel, R., Kalla, S., Upadhyaya, S., Chaurasia, S., & Singh, K. (2020). CFD modeling of vacuum
membrane distillation for removal of Naphthol blue black dye from aqueous solution using
COMSOL multiphysics. Chemical Engineering Research and Design, 158, 77-88. doi:https://
doi.org/10.1016/j.cherd.2020.03.016
Baker, R. W. (2004). Membrane Technology and Applications. England: John Wiley & Sons Ltd.
Baker, R. W. (2012). Membrane technology and applications: John Wiley & Sons.
Balster, J. (2016). Hollow Fiber Membrane Module. In E. Drioli & L. Giorno (Eds.), Encyclopedia of
Membranes (pp. 955-957). Berlin, Heidelberg: Springer Berlin Heidelberg.
Behroozi, A. H., Kasiri, N., & Mohammadi, T. (2019). Multi-phenomenal macroscopic investigation
of cross-flow membrane flux in microfiltration of oil-in-water emulsion, experimental
& computational. Journal of Water Process Engineering, 32, 100962. doi:https://doi.
org/10.1016/j.jwpe.2019.100962
Berk, Z. (2009). Chapter 10 - Membrane processes. In Z. Berk (Ed.), Food Process Engineering and
Technology (pp. 233-257). San Diego: Academic Press.
Boussu, K., Van der Bruggen, B., Volodin, A., Snauwaert, J., Van Haesendonck, C., & Vandecasteele,
C. (2005). Roughness and hydrophobicity studies of nanofiltration membranes using different
modes of AFM. Journal of colloid and interface science, 286(2), 632-638. doi:https://doi.
org/10.1016/j.jcis.2005.01.095
Brunner, D., Khawaja, H., Moatamedi, M., & Boiger, G. (2018). CFD modelling of pressure and shear
rate in torsionally vibrating structures using ANSYS CFX and COMSOL Multiphysics. The
International Journal of Multiphysics, 12(4), 349-358. doi:https://doi.org/10.21152/1750-
9548.12.4.349

457
Experimental Methods for Membrane Applications

Bucs, S. S., Valladares Linares, R., van Loosdrecht, M. C., Kruithof, J. C., & Vrouwenvelder, J. S. (2014).
Impact of organic nutrient load on biomass accumulation, feed channel pressure drop increase
and permeate flux decline in membrane systems. Water Res, 67, 227-242. doi:10.1016/j.
watres.2014.09.005
Burn, S., & Gray, S. (2015). Efficient Desalination by Reverse Osmosis: A guide to RO practice (Vol.
14). London, UK: IWA Publishing.
Byun, H. S., & Rhee, K. (2004). CFD modeling of blood flow following coil embolization of
aneurysms. Medical engineering & physics, 26(9), 755-761. doi:https://doi.org/10.1016/j.
medengphy.2004.06.008
Cancilla, N., Gurreri, L., Marotta, G., Ciofalo, M., Cipollina, A., Tamburini, A., & Micale, G. (2021).
CFD prediction of shell-side flow and mass transfer in regular fiber arrays. International
Journal of Heat and Mass Transfer, 168, 120855. doi:https://doi.org/10.1016/j.
ijheatmasstransfer.2020.120855
Cao, Z., Wiley, D., & Fane, A. (2001). CFD simulations of net-type turbulence promoters in a narrow
channel. Journal of Membrane science, 185(2), 157-176. doi:https://doi.org/10.1016/s0376-
7388(00)00643-8
Chan, F. S., Tan, C. K., Ratnayake, P., Junaidi, M. U. M., & Liang, Y. Y. (2020a). Reduced-order
modelling of concentration polarization with varying permeation: Analysis of electro-osmosis
in membranes. Desalination, 495, 114677. doi:https://doi.org/10.1016/j.desal.2020.114677
Chan, F. S., Tan, C. K., Ratnayake, P., & Liang, Y. Y. (2020b). Reduced-order modeling of flow and
concentration polarization in membrane systems with permeation. AIChE journal, 66(4),
e16851. doi:https://doi.org/10.1002/aic.16851
Chen, L., & Wu, B. (2021). Research Progress in Computational Fluid Dynamics Simulations
of Membrane Distillation Processes: A Review. Membranes (Basel), 11(7). doi:10.3390/
membranes11070513
Chong, T. H., Wong, F. S., & Fane, A. G. (2007). Enhanced concentration polarization by unstirred
fouling layers in reverse osmosis: Detection by sodium chloride tracer response technique.
Journal of Membrane science, 287(2), 198-210. doi:10.1016/j.memsci.2006.10.035
Chong, Y. K., Liang, Y. Y., Lau, W. J., & Fimbres Weihs, G. A. (2022). 3D CFD study of hydrodynamics
and mass transfer phenomena for spiral wound membrane submerged-type feed spacer with
different node geometries and sizes. International Journal of Heat and Mass Transfer, 191,
122819. doi:https://doi.org/10.1016/j.ijheatmasstransfer.2022.122819
Chu, A., Kwok, R. C.-W., & Yu, K. (2005). Study of pollution dispersion in urban areas using
Computational Fluid Dynamics (CFD) and Geographic Information System (GIS). Environmental
Modelling & Software, 20(3), 273-277. doi:https://doi.org/10.1016/j.envsoft.2004.05.007
COMSOL AB. (2008). COMSOL Multiphysics User’s Guide. Burlington.
Cummings, E. B., Griffiths, S. K., Nilson, R. H., & Paul, P. H. (2000). Conditions for Similitude between
the Fluid Velocity and Electric Field in Electroosmotic Flow. Analytical Chemistry, 72(11),
2526-2532. doi:10.1021/ac991165x
Da Costa, A., Fane, A., & Wiley, D. (1994). Spacer characterization and pressure drop modelling
in spacer-filled channels for ultrafiltration. Journal of Membrane science, 87(1-2), 79-98.
doi:https://doi.org/10.1016/0376-7388(93)e0076-p
Du, X., Liu, X., Wang, Y., Radaei, E., Lian, B., Leslie, G., Li, G., & Liang, H. (2017). Particle deposition
on flat sheet membranes under bubbly and slug flow aeration in coagulation-microfiltration
process: Effects of particle characteristic and shear stress. Journal of Membrane science, 541, 668-
676. doi:https://doi.org/10.1016/j.memsci.2017.07.023

458
 Chapter 19

Fimbres-Weihs, G., & Wiley, D. (2008). Numerical study of two-dimensional multi-layer spacer
designs for minimum drag and maximum mass transfer. Journal of Membrane science, 325(2),
809-822. doi:https://doi.org/10.1016/j.memsci.2008.09.005
Fimbres-Weihs, G., & Wiley, D. (2010). Review of 3D CFD modeling of flow and mass transfer in
narrow spacer-filled channels in membrane modules. Chemical Engineering and Processing:
Process Intensification, 49(7), 759-781. doi:https://doi.org/10.1016/j.cep.2010.01.007
Fimbres-Weihs, G. A., Wiley, D. E., & Fletcher, D. F. (2006). Unsteady Flows with Mass Transfer in
Narrow Zigzag Spacer-Filled Channels: A Numerical Study. Industrial & Engineering Chemistry
Research, 45(19), 6594-6603. doi:10.1021/ie060243l
Fimbres Weihs, G., & Wiley, D. (2007). Numerical study of mass transfer in three-dimensional
spacer-filled narrow channels with steady flow. Journal of Membrane science, 306, 228-243.
doi:10.1016/j.memsci.2007.08.043
Fimbres Weihs, G. A., & Wiley, D. E. (2014). CFD analysis of tracer response technique under
cake-enhanced osmotic pressure. Journal of Membrane science, 449, 38-49. doi:10.1016/j.
memsci.2013.08.015
Foo, K., Liang, Y. Y., & Fimbres Weihs, G. A. (2020). CFD study of the effect of SWM feed spacer
geometry on mass transfer enhancement driven by forced transient slip velocity. Journal of
Membrane science, 597. doi:10.1016/j.memsci.2019.117643
Foo, K., Liang, Y. Y., Tan, C. K., & Fimbres Weihs, G. A. (2021). Coupled effects of circular and
elliptical feed spacers under forced-slip on viscous dissipation and mass transfer enhancement
based on CFD. Journal of Membrane science, 637, 119599. doi:https://doi.org/10.1016/j.
memsci.2021.119599
Fritzmann, C., Löwenberg, J., Wintgens, T., & Melin, T. (2007). State-of-the-art of reverse osmosis
desalination. Desalination, 216(1-3), 1-76. doi:10.1016/j.desal.2006.12.009
Gavelli, F., Bullister, E., & Kytomaa, H. (2008). Application of CFD (Fluent) to LNG spills into
geometrically complex environments. Journal of hazardous materials, 159(1), 158-168.
doi:https://doi.org/10.1016/j.jhazmat.2008.02.037
Geraldes, V., & Afonso, M. D. (2006). Generalized mass transfer correction factor for nanofiltration and
reverse osmosis. AIChE journal, 52(10), 3353-3362. doi:https://doi.org/10.1002/aic.10968
Golrokh Sani, A., Najafi, H., & Azimi, S. S. (2021). CFD simulation of air-sparged slug flow in the flat-
sheet membrane: A concentration polarization study. Separation and Purification Technology,
270, 118816. doi:https://doi.org/10.1016/j.seppur.2021.118816
Gu, B., Adjiman, C. S., & Xu, X. Y. (2017). The effect of feed spacer geometry on membrane performance
and concentration polarisation based on 3D CFD simulations. Journal of Membrane science,
527, 78-91. doi:https://doi.org/10.1016/j.memsci.2016.12.058
Gupta, R., Fletcher, D. F., & Haynes, B. S. (2009). On the CFD modelling of Taylor flow in microchannels.
Chemical Engineering Science, 64(12), 2941-2950. doi:https://doi.org/10.1016/j.
ces.2009.03.018
Haddadi, B., Jordan, C., Miltner, M., & Harasek, M. (2018). Membrane modeling using CFD: Combined
evaluation of mass transfer and geometrical influences in 1D and 3D. Journal of Membrane
science, 563, 199-209. doi:https://doi.org/10.1016/j.memsci.2018.05.040
Hinsch, K. D. (2002). Holographic particle image velocimetry. Measurement Science and Technology,
13(7), R61-R72. doi:10.1088/0957-0233/13/7/201
Holemans, T., Yang, Z., & Vanierschot, M. (2022). Efficient Reduced Order Modeling of Large Data
Sets Obtained from CFD Simulations. Fluids, 7(3), 110. doi:https://doi.org/10.3390/
fluids7030110

459
Experimental Methods for Membrane Applications

Hu, G., & Li, D. (2007). Multiscale phenomena in microfluidics and nanofluidics. Chemical Engineering
Science, 62(13), 3443-3454. doi:https://doi.org/10.1016/j.ces.2006.11.058
Hunter, R. J. (2013). Zeta potential in colloid science: principles and applications (Vol. 2): Academic
press.
Jafarkhani, M., Moraveji, M. K., Davarnejad, R., Moztarzadeh, F., & Mozafari, M. (2012). Three-
dimensional simulation of turbulent flow in a membrane tube filled with semi-circular baffles.
Desalination, 294, 8-16. doi:https://doi.org/10.1016/j.desal.2012.02.031
Jagannadh, S. N., & Muralidhara, H. S. (1996). Electrokinetics Methods To Control Membrane Fouling.
Industrial & Engineering Chemistry Research, 35(4), 1133-1140. doi:10.1021/ie9503712
Jamshed, S. (2015). Using HPC for Computational Fluid Dynamics: A Guide to High Performance
Computing for CFD Engineers: Academic Press.
Javid, S. M., Passandideh-Fard, M., Faezian, A., & Goharimanesh, M. (2017). Slug and bubble flows in
a flat sheet ultrafiltration module: Experiments and numerical simulation. International Journal
of Multiphase Flow, 91, 39-50. doi:https://doi.org/10.1016/j.ijmultiphaseflow.2016.12.006
Jiang, S., Li, Y., & Ladewig, B. P. (2017). A review of reverse osmosis membrane fouling and control
strategies. Sci Total Environ, 595, 567-583. doi:10.1016/j.scitotenv.2017.03.235
Junker, M. A., de Vos, W. M., Lammertink, R. G. H., & de Grooth, J. (2021). Bridging the gap between
lab-scale and commercial dimensions of hollow fiber nanofiltration membranes. Journal of
Membrane science, 624, 119100. doi:https://doi.org/10.1016/j.memsci.2021.119100
Kaya, R., Deveci, G., Turken, T., Sengur, R., Guclu, S., Koseoglu-Imer, D. Y., & Koyuncu, I. (2014).
Analysis of wall shear stress on the outside-in type hollow fiber membrane modules by CFD
simulation. Desalination, 351, 109-119. doi:https://doi.org/10.1016/j.desal.2014.07.033
Kone, J.-P., Zhang, X., Yan, Y., Hu, G., & Ahmadi, G. (2018). CFD modeling and simulation of PEM
fuel cell using OpenFOAM. Energy Procedia, 145, 64-69. doi:https://doi.org/10.1016/j.
egypro.2018.04.011
Landahl, M. T., Mollo Christensen, E., & Korman, M. S. (1989). Turbulence and random processes in
fluid mechanics. In: Acoustical Society of America.
Leam, J. J., Bilad, M. R., Wibisono, Y., Hakim Wirzal, M. D., & Ahmed, I. (2020). Membrane Technology
for Microalgae Harvesting. 97-110. doi:10.1016/b978-0-12-817536-1.00007-2
Li, H. y., Bertram, C. D., & Wiley, D. E. (1998). Mechanisms by which pulsatile flow affects cross
flow microfiltration. AIChE journal, 44(9), 1950-1961. doi:https://doi.org/10.1002/
aic.690440903
Li, M., Bui, T., & Chao, S. (2016). Three-dimensional CFD analysis of hydrodynamics and concentration
polarization in an industrial RO feed channel. Desalination, 397, 194-204. doi:https://doi.
org/10.1016/j.desal.2016.07.005
Li, W., Su, X., Palazzolo, A., Ahmed, S., & Thomas, E. (2017). Reverse osmosis membrane, seawater
desalination with vibration assisted reduced inorganic fouling. Desalination, 417, 102-114. doi:.
https://doi.org/10.1016/j.desal.2017.05.016
Li, Y.-L., & Tung, K.-L. (2008). The effect of curvature of a spacer-filled channel on fluid flow in spiral-
wound membrane modules. Journal of Membrane science, 319(1-2), 286-297. doi:10.1016/j.
memsci.2008.03.069
Li, Y.-L., Tung, K.-L., Chen, Y.-S., & Hwang, K.-J. (2012). CFD analysis of the initial stages of particle
deposition in spiral-wound membrane modules. Desalination, 287, 200-208. doi:10.1016/j.
desal.2011.10.001

460
 Chapter 19

Liang, Y. Y., Chapman, M. B., Fimbres-Weihs, G. A., & Wiley, D. E. (2014a). CFD modelling of
electro-osmotic permeate flux enhancement on the feed side of a membrane module. Journal of
Membrane science, 470, 378-388. doi:https://doi.org/10.1016/j.memsci.2014.07.039
Liang, Y. Y., Fimbres-Weihs, G. A., & Fletcher, D. F. (2018a). CFD study of the effect of unsteady slip
velocity waveform on shear stress in membrane systems. Chemical Engineering Science, 192,
16-24. doi:https://doi.org/10.1016/j.ces.2018.07.009
Liang, Y. Y., Fimbres-Weihs, G. A., Setiawan, R., & Wiley, D. E. (2016a). CFD modelling of unsteady
electro-osmotic permeate flux enhancement in membrane systems. Chemical Engineering
Science, 146, 189-198. doi:https://doi.org/10.1016/j.ces.2016.02.028
Liang, Y. Y., Fimbres-Weihs, G. A., & Wiley, D. E. (2014b). Approximation for modelling electro-
osmotic mixing in the boundary layer of membrane systems. Journal of Membrane science, 450,
18-27. doi:https://doi.org/10.1016/j.memsci.2013.08.031
Liang, Y. Y., Fimbres-Weihs, G. A., & Wiley, D. E. (2016b). CFD modelling of electro-osmotic permeate
flux enhancement in spacer-filled membrane channels. Journal of Membrane science, 507, 107-
118. doi:https://doi.org/10.1016/j.memsci.2016.02.012
Liang, Y. Y., Fimbres-Weihs, G. A., & Wiley, D. E. (2020a). Comparison of oscillating flow and slip
velocity mass transfer enhancement in spacer-filled membrane channels: CFD analysis and
validation. Journal of Membrane science, 593, 117433. doi:https://doi.org/10.1016/j.
memsci.2019.117433
Liang, Y. Y., Fimbres Weihs, G. A., & Fletcher, D. F. (2018b). CFD study of the effect of unsteady slip
velocity waveform on shear stress in membrane systems. Chemical Engineering Science, 192,
16-24. doi:https://doi.org/10.1016/j.ces.2018.07.009
Liang, Y. Y., Fimbres Weihs, G. A., & Wiley, D. E. (2020b). Comparison of oscillating flow and slip
velocity mass transfer enhancement in spacer-filled membrane channels: CFD analysis and
validation. Journal of Membrane science, 593, 117433. doi:https://doi.org/10.1016/j.
memsci.2019.117433
Liang, Y. Y., Toh, K. Y., & Fimbres Weihs, G. A. (2019). 3D CFD study of the effect of multi-layer
spacers on membrane performance under steady flow. Journal of Membrane science, 580, 256-
267. doi:https://doi.org/10.1016/j.memsci.2019.02.015
Lin, W., Lei, J., Wang, Q., Wang, X.-m., & Huang, X. (2022a). Performance enhancement of spiral-
wound reverse osmosis membrane elements with novel diagonal-flow feed channels.
Desalination, 523, 115447.
Lin, W., Wang, Q., Sun, L., Wang, D., Cabrera, J., Li, D., Hu, L., Jiang, G., Wang, X.-m., & Huang, X.
(2022b). The critical role of feed spacer channel porosity in membrane biofouling: Insights and
implications. Journal of Membrane science, 649, 120395.
Lindken, R., & Burgmann, S. (2012). 14 - Laser-optical methods for transport studies in low temperature
fuel cells. In C. Hartnig & C. Roth (Eds.), Polymer Electrolyte Membrane and Direct Methanol
Fuel Cell Technology (Vol. 2, pp. 425-461): Woodhead Publishing.
Liu, D., Bu, C., & Chen, X. (2013). Development and test of CFD–DEM model for complex geometry:
A coupling algorithm for Fluent and DEM. Computers & Chemical Engineering, 58, 260-268.
doi:https://doi.org/10.1016/j.compchemeng.2013.07.006
Liu, Y., & Hinrichsen, O. (2014). CFD modeling of bubbling fluidized beds using OpenFOAM®: Model
validation and comparison of TVD differencing schemes. Computers & Chemical Engineering,
69, 75-88. doi:https://doi.org/10.1016/j.compchemeng.2014.07.002
Lotfiyan, H., Zokaee Ashtiani, F., Fouladitajar, A., & Armand, S. B. (2014). Computational fluid
dynamics modeling and experimental studies of oil-in-water emulsion microfiltration in a flat

461
Experimental Methods for Membrane Applications

sheet membrane using Eulerian approach. Journal of Membrane science, 472, 1-9. doi:https://
doi.org/10.1016/j.memsci.2014.08.036
Maddah, H., & Chogle, A. (2016). Biofouling in reverse osmosis: phenomena, monitoring, controlling
and remediation. Applied Water Science, 7(6), 2637-2651. doi:10.1007/s13201-016-0493-1
Martín, M. M. (2016). Chapter 4 - Water. In M. M. Martín (Ed.), Industrial Chemical Process Analysis
and Design (pp. 125-197). Boston: Elsevier.
Martinelli, L., Guigui, C., & Line, A. (2010). Characterisation of hydrodynamics induced by air injection
related to membrane fouling behaviour. Desalination, 250(2), 587-591. doi:https://doi.
org/10.1016/j.desal.2009.09.029
NASA. (2022). Overview of CFD Verification & Validation. Retrieved from https://www.grc.nasa.
gov/www/wind/valid/tutorial/overview.html
Ndinisa, N. V., Wiley, D. E., & Fletcher, D. F. (2005). Computational Fluid Dynamics Simulations
of Taylor Bubbles in Tubular Membranes: Model Validation and Application to Laminar
Flow Systems. Chemical Engineering Research and Design, 83(1), 40-49. doi:https://doi.
org/10.1205/cherd.03394
OpenCFD Ltd. (2016). OpenFOAM: User Guide. Retrieved from https://www.openfoam.com/
documentation/guides/latest/doc/guide-schemes.html
Ouyang, H., Bao, J., Fimbres-Weihs, G. A., & Wiley, D. E. (2013). Control study on mixing
enhancement in boundary layers of membrane systems. Journal of Process Control, 23(8), 1197-
1204. doi:https://doi.org/10.1016/j.jprocont.2013.07.005
Pak, A., Mohammadi, T., Hosseinalipour, S. M., & Allahdini, V. (2008). CFD modeling of porous
membranes. Desalination, 222(1), 482-488. doi:https://doi.org/10.1016/j.desal.2007.01.152
Park, S., Jeong, Y. D., Lee, J. H., Kim, J., Jeong, K., & Cho, K. H. (2021). 3D printed honeycomb-shaped
feed channel spacer for membrane fouling mitigation in nanofiltration. Journal of Membrane
science, 620, 118665. doi:https://doi.org/10.1016/j.memsci.2020.118665
Patankar, N. A., & Hu, H. H. (1998). Numerical Simulation of Electroosmotic Flow. Analytical
Chemistry, 70(9), 1870-1881. doi:10.1021/ac970846u
Probstein, R. F. (1989). Physiochemical Hydrodynamics: An Introduction. Hoboken, New Jersey: John
Wiley & Sons.
Qasim, M., Badrelzaman, M., Darwish, N. N., Darwish, N. A., & Hilal, N. (2019). Reverse osmosis
desalination: A state-of-the-art review. Desalination, 459, 59-104. doi:10.1016/j.
desal.2019.02.008
Radaei, E., Liu, X., Tng, K. H., Wang, Y., Trujillo, F. J., & Leslie, G. (2018). Insights on pulsed bubble
control of membrane fouling: Effect of bubble size and frequency. Journal of Membrane science,
554, 59-70. doi:https://doi.org/10.1016/j.memsci.2018.02.058
Radu, A. I., Vrouwenvelder, J. S., van Loosdrecht, M. C. M., & Picioreanu, C. (2010). Modeling the
effect of biofilm formation on reverse osmosis performance: Flux, feed channel pressure drop
and solute passage. Journal of Membrane science, 365(1), 1-15. doi:https://doi.org/10.1016/j.
memsci.2010.07.036
Rahimi, M., Madaeni, S. S., Abolhasani, M., & Alsairafi, A. A. (2009). CFD and experimental studies
of fouling of a microfiltration membrane. Chemical Engineering and Processing: Process
Intensification, 48(9), 1405-1413. doi:10.1016/j.cep.2009.07.008
Rajaratnam, N. (1976). Turbulent jets: Elsevier.
Ratkovich, N., Chan, C. C. V., Berube, P. R., & Nopens, I. (2009). Experimental study and CFD modelling
of a two-phase slug flow for an airlift tubular membrane. Chemical Engineering Science, 64(16),
3576-3584. doi:https://doi.org/10.1016/j.ces.2009.04.048

462
 Chapter 19

Ratnayake, P., Setiawan, R., Bao, J., Fimbres-Weihs, G. A., & Wiley, D. E. (2016). Spatio-temporal
frequency response analysis of forced slip velocity effect on solute concentration oscillations
in a reverse osmosis membrane channel. Computers & Chemical Engineering, 84, 151-161.
doi:https://doi.org/10.1016/j.compchemeng.2015.08.016
Rawool, A. S., & Mitra, S. K. (2006). Numerical simulation of electroosmotic effect in serpentine
channels. Microfluidics and Nanofluidics, 2(3), 261-269. doi:10.1007/s10404-005-0076-1
Ren, L., Sinton, D., & Li, D. (2003). Numerical simulation of microfluidic injection processes in
crossing microchannels. Journal of Micromechanics and Microengineering, 13(5), 739-747.
doi:10.1088/0960-1317/13/5/329
Roache, P. J. (1997). Quantification of uncertainty in computational fluid dynamics. Annual review of
fluid Mechanics, 29(1), 123-160. doi:https://doi.org/10.1146/annurev.fluid.29.1.123
Rodrigues, C., Garcia-Algado, P., Semião, V., de Pinho, M. N., & Geraldes, V. (2013). Concentration
boundary layer visualization in nanofiltration by holographic interferometry with light deflection
correction. Journal of Membrane science, 447, 306-314. doi:https://doi.org/10.1016/j.
memsci.2013.07.035
Russel, W. B., Russel, W., Saville, D. A., & Schowalter, W. R. (1991). Colloidal dispersions: Cambridge
university press.
Salehi, M.-S., Askarishahi, M., Godini, H. R., Görke, O., & Wozny, G. n. (2016). Sustainable process
design for oxidative coupling of methane (OCM): comprehensive reactor engineering via
computational fluid dynamics (CFD) analysis of OCM packed-bed membrane reactors. Industrial
& Engineering Chemistry Research, 55(12), 3287-3299. doi:https://doi.org/10.1021/acs.
iecr.5b03292
Santiago, J. G. (2001). Electroosmotic Flows in Microchannels with Finite Inertial and Pressure Forces.
Analytical Chemistry, 73(10), 2353-2365. doi:10.1021/ac0101398
Schwinge, J., Wiley, D. E., & Fletcher, D. F. (2002a). A CFD study of unsteady flow in narrow
spacer-filled channels for spiral-wound membrane modules. Desalination, 146(1), 195-201.
doi:https://doi.org/10.1016/S0011-9164(02)00470-8
Schwinge, J., Wiley, D. E., & Fletcher, D. F. (2002b). Simulation of the Flow around Spacer Filaments
between Channel Walls. 2. Mass-Transfer Enhancement. Industrial & Engineering Chemistry
Research, 41(19), 4879-4888. doi:10.1021/ie011015o
Scott, K. (1995). Handbook of industrial membranes: Elsevier.
Snel, H. (2003). Review of aerodynamics for wind turbines. Wind Energy: An International Journal for
Progress and Applications in Wind Power Conversion Technology, 6(3), 203-211. doi:https://
doi.org/10.1002/WE.97
Sparks, T., & Chase, G. (2016). Section 2 - Filter Media. In T. Sparks & G. Chase (Eds.), Filters and
Filtration Handbook (Sixth Edition) (pp. 55-115). Oxford: Butterworth-Heinemann.
Spiegler, K. S., & Macleish, J. H. (1981). Molecular (osmotic and electro-osmotic) backwash of cellulose
acetate hyperfiltration membranes. Journal of Membrane science, 8(2), 173-192. doi:https://
doi.org/10.1016/S0376-7388(00)82089-X
Steinhauser, M. O. (2017). Computational multiscale modeling of fluids and solids: Springer.
Streeter, V., Wylie, E., & Bedford, K. (1985). Fluid Mechanics (8th ed.): McGraw-Hill.
Su, X., Li, W., Palazzolo, A., & Ahmed, S. (2018). Concentration polarization and permeate flux
variation in a vibration enhanced reverse osmosis membrane module. Desalination, 433, 75-88.
doi:https://doi.org/10.1016/j.desal.2018.01.001
Su, X., Li, W., Palazzolo, A., & Ahmed, S. (2019). Permeate flux increase by colloidal fouling control in
a vibration enhanced reverse osmosis membrane desalination system. Desalination, 453, 22-36.
doi:https://doi.org/10.1016/j.desal.2018.12.003

463
Experimental Methods for Membrane Applications

Sun, B., Ahmed, A., Atkinson, C., & Soria, J. (2020). A novel 4D digital holographic PIV/PTV
(4D-DHPIV/PTV) methodology using iterative predictive inverse reconstruction. Measurement
Science and Technology, 31(10), 104002. doi:10.1088/1361-6501/ab8ee8
Taha, T., Cheong, W. L., Field, R. W., & Cui, Z. F. (2006). Gas-sparged ultrafiltration using horizontal
and inclined tubular membranes—A CFD study. Journal of Membrane science, 279(1), 487-494.
doi:https://doi.org/10.1016/j.memsci.2005.12.063
Taha, T., & Cui, Z. F. (2002). CFD modelling of gas-sparged ultrafiltration in tubular membranes. Journal
of Membrane science, 210(1), 13-27. doi:https://doi.org/10.1016/S0376-7388(02)00360-5
Taha, T., & Cui, Z. F. (2006a). CFD modelling of slug flow in vertical tubes. Chemical Engineering
Science, 61(2), 676-687. doi:https://doi.org/10.1016/j.ces.2005.07.022
Taha, T., & Cui, Z. F. (2006b). CFD modelling of slug flow inside square capillaries. Chemical
Engineering Science, 61(2), 665-675. doi:https://doi.org/10.1016/j.ces.2005.07.023
Taheri, A. H., Sim, L. N., Chong, T. H., Krantz, W. B., & Fane, A. G. (2015). Prediction of reverse osmosis
fouling using the feed fouling monitor and salt tracer response technique. Journal of Membrane
science, 475, 433-444. doi:10.1016/j.memsci.2014.10.043
Toh, K., Liang, Y., Lau, W., & Weihs, G. F. (2020a). The techno-economic case for coupling advanced
spacers to high-permeance RO membranes for desalination. Desalination, 491, 114534.
doi:https://doi.org/10.1016/j.desal.2020.114534
Toh, K. Y., Liang, Y. Y., Lau, W. J., & Fimbres Weihs, G. A. (2020b). A review of CFD modelling and
performance metrics for osmotic membrane processes. Membranes, 10(10), 285. doi:https://
doi.org/10.3390/membranes10100285
Unal, B. O. (2022). Membrane autopsy study to characterize fouling type of RO membrane used in an
industrial zone wastewater reuse plant. Desalination, 529, 115648.
Versteeg, H. K., & Malalasekera, W. (2007). An introduction to computational fluid dynamics: the
finite volume method: Pearson education.
Vinuesa, R., & Brunton, S. L. (2022). Enhancing computational fluid dynamics with machine learning.
Nature Computational Science, 2(6), 358-366. doi:https://doi.org/10.1038/s43588-022-
00264-7
Vrouwenvelder, J. S., Picioreanu, C., Kruithof, J. C., & van Loosdrecht, M. C. M. (2010). Biofouling in spiral
wound membrane systems: Three-dimensional CFD model based evaluation of experimental
data. Journal of Membrane science, 346(1), 71-85. doi:10.1016/j.memsci.2009.09.025
Wardeh, S., & Morvan, H. P. (2008). CFD simulations of flow and concentration polarization in spacer-
filled channels for application to water desalination. Chemical Engineering Research and Design,
86(10), 1107-1116. doi:https://doi.org/10.1016/j.cherd.2008.04.010
Wiley, D. E., & Fletcher, D. F. (2003). Techniques for computational fluid dynamics modelling of
flow in membrane channels. Journal of Membrane science, 211(1), 127-137. doi:https://doi.
org/10.1016/s0376-7388(02)00412-x
Xue, Y. L., Zhang, R., Cao, B., & Li, P. (2021). Chapter 20 - Tubular membranes and modules. In T.-S.
Chung & Y. Feng (Eds.), Hollow Fiber Membranes (pp. 431-448): Elsevier.
Zare, M., Zokaee Ashtiani, F., & Fouladitajar, A. (2013). CFD modeling and simulation of concentration
polarization in microfiltration of oil–water emulsions; Application of an Eulerian multiphase
model. Desalination, 324, 37-47. doi:https://doi.org/10.1016/j.desal.2013.05.022
Zhang, J.-b., He, G.-w., & Liu, F. (2006). Electro-osmotic flow and mixing in heterogeneous
microchannels. Physical Review E, 73(5), 056305. doi:10.1103/PhysRevE.73.056305

464
Experimental Methods for Membrane Applications
in Desalination and Water Treatment
Water quality is a critical issue for the water industry today, and membrane filtration a highly
effective treatment used to provide clean water for the global population. Experimental processes
and methods are being developed for assessing fouling, scaling, performance and modelling of
membrane systems, and research is needed to further improve the sustainability and feasibility of
these technologies, and to mitigate the challenges of water scarcity for billions of people, particularly
in developing countries.

This book aims to address this vital issue by bringing together experts in the field to share their
learning, including:
• Membrane processes: microfiltration (MF), ultrafiltration (UF), reverse osmosis (RO), forward
osmosis (FO) and membrane distillation (MD)
• Particulate fouling: silt density index (SDI), modified fouling index (MFI-0.45 and MFI-UF)
• Inorganic fouling and scaling: assessment, characterization tools and mitigation
• Organic fouling: size exclusion chromatography (LC-OCD), fluorescence spectroscopy (FEEM)
and transparent exopolymer particles (TEP)
• Biological fouling: genomic tools, bacterial growth potential (BGP) of seawater and low-
nutrient water and optical coherence tomography (OCT)
• General applications: membrane autopsy and computational fluid dynamics (CFD) modelling

This book will be an important resource for undergraduate and graduate

engineering students and researchers, academics, plant operators, consultants,
professionals and practitioners in the water sector.
Contributors:
Aamer Ali Karima Bakkali Pierre Le-Clech
Adam C. Hambly Kathleen Foo Poul Toft Frederiksen
Alberto Tiraferri Léonie Le Bouille Pouyan Mirzaei Vishkaei
Almotasembellah Abushaban Loreen O. Villacorte Rita Kay Henderson
Barun Lal Karna Luca Fortunato Sergio G. Salinas-Rodriguez
Cejna Anna Quist-Jensen Lucia Ruiz Haddad Steven J. Duranceau
Claus Hélix-Nielsen Maria Salud Camilleri-Rumbau Urban J. Wünsch
Francisco Javier García Picazo Mohamed Chaker Necibi Vanida A. Salgado-Ismodes
Guillem Gilabert-Oriol Mohamed Fauzi Haroon Victor A. Yangali Quintanilla
Gustavo A. Fimbres Weihs Mohammad Mahdi A. Shirazi Victoria Sanahuja-Embuena
Helen Rutlidge Mohaned Sousi Wen Yew Lam
Helga Calix Ponce Morten L. Christensen Weng Fung Twong
Irena Petrinic Muhammad Ali Xuan Tung Nguyen
Jan Frauholz Muhammad Nasir Mangal Yie Kai Chong
Javier Rodriguez Gómez Nuria Peña García Yuli Ekowati
Jia Xin Tan Pascal E. Saikaly Yong Yeow Liang
Johannes S. Vrouwenvelder

www.iwaponline.com
@IWAPublishing

ISBN 9781789062960 (Hardback)

ISBN 9781789062977 (eBook)
ISBN 9781789062984 (ePub)