Major Organic Foulants in Ultrafiltration of

Treated Domestic Wastewater and their Removal

by Bio-filtration as Pre-treatment

vorgelegt von
Master of Science
Xing Zheng

an der Fakultät III – Prozesswissenschaften

der Technischen Universität Berlin
zur Erlangung des akademischen Grades

Doktor der Ingenieurwissenschaften

– Dr. Ing. –
genehmigte Dissertation

Promotionsausschuss
Vorsitzender: Prof. Dr. Ing. Sven-Uwe Geißen (Technische Universität Berlin)
Gutachter: Prof. Dr. Ing. Martin Jekel (Technische Universität Berlin)
Gutachter: Prof. Dr. Peter M. Huck (University of Waterloo/Ontario, Kanada)
Gutachter: Dr. Ing. Mathias Ernst (Technische Universität Berlin)

Tag der wissenschaftlichen Aussprache: 09.04.2010

Berlin 2010
D 83
Acknowledgements I

Acknowledgements
I would like to thank my two advisers Prof. M. Jekel and Dr. M. Ernst for providing me the
opportunity to conduct my research under their guidance in different projects. Several years
ago as I began to work on the Technical University of Berlin, the dissertation was so remote
to me that I doubted if I could finish it. Thanks to their suggestive advices, to their patient
corrections of my papers and presentations, and also to their inspirations, I could make some
progress in my work and enjoy the research as well. Thank you so much, Herr Jekel and
Mathias. Simultaneously, the finance agencies of the projects which are the Federal Ministry
for Education (BMBF, Germany) and the European Commission under the 6th Framework
Programme are gratefully acknowledged for the financial supports.

I would also give my thanks to Prof. Peter M. Huck for his suggestion to the writing of my
dissertation and the attendance of my examination. Prof. S. Geissen is acknowledged for be-
ing the chairman of my examination committee. My acknowlegement is also given to Prof.
Gary Amy for his suggestive advices each time when we met together and the coordination of
fluorescence measurement.

And of course I would like to thank all of my colleges at the Department of Water Quality
Control. This is a warm and creative research group. Either at the beginning of my work, or in
the difficult times during the research, I have got always kind helps from them. Especially, I
would acknowledge Alexander Sperlich for his kind help, and also the membrane group –
Jens Haberkamp and Renata Mehrez – for the discussions and introductions of the measuring
methods. Thanks also to Carsten Bahr, Steffen Grünheid, Uwe Hübner, Hendrik Paar, Ulf
Miehe, Andreas Albers, Michael Stieber, Arne Hein, Aki Sabastian Ruhl and Christian Peters
for their help.

The research work is based on experiments. Without the supports from lab assistances and the
factory of the institute would it be impossible to be finished. I would like to acknowledge Dr.
Anke Putschew for her always presence when the LC-OCD was defect, Angelika Kersten and
Katrin Noack for their maintenance of the instrument, Ulrike Förster for the DOC and protein
measurement, Jutta Jakobs, Gisela Sosna, Elke Profft, Hella Schmeisser for their patient helps
in the laboratory. All of my sand filter systems were set up by Thomas Thele and Hans Riet-
dorf. It was my luck to get their help. Finally, thanks also to Karin von Nordheim and Han-
nelore Meingast our secretary for all the help in practical things.

Besides the lab works, the pilot tests were carried out in different wastewater treatment plants.
The cooperation and help from Mr. S. Foth, Ms A. Krug and Mr. M. Kempf from the WWTP
Ruhleben in Berlin, Ms Yiping Gan, Mr. Jun Hu, Dr. Jun Zhou, Mr. Jinhua Gao from the Bei-
jing Drainage Group (BDG) and Dr. Xuzhou Cheng, Dr. Futai Chen from the Tsinghua Uni-
versity in Beijing, Mr. Jianhong Chen and Mr. Renyang Hong from the Industrial Technology
Research Institute (ITRI) in Taipei are sincerely acknowledged.

In different projects, the involved students attributed a lot to water quality measurement, lab
experiments, operation and maintenance of the pilot plants. Their efforts released me to a
large extent, either in Beijing or in Berlin. I would like to thank Jiahui Kang from the Univer-
II Acknowledgements

sity of Geology China, Tanja Ratuzny, Benito Martin, Nadin Drescher, Blanca Hernández
García, Zhichao Liang, Adriana González Tomaskovich, Florian Härtl, Stephan Plume and
Christopher Gabler from the Technical University of Berlin for their efforts.

I would thank Mr. Sun Kyu Maeng from TU Delft, Dr. S. Huber, Dr. M. Abert and Mr. A.
Balz from DOC Labor (DOC Labor Dr Huber, Germany) for their kind assistance in fluores-
cence and LC-OCD measurement. Special thanks to Martin Meingast from the ABU Consult
Berlin GmbH for commenting on the English grammar of this dissertation.

And finally thanks to my friends and family. The encouragement and help of my friends, es-
pecially from Dr. Weihua Peng in the United Water, Dr. Thomas Ludwig and Yongjun Zhang
from the Technical University of Berlin are sincerely acknowledged. In the past several years
the warm telephone communications with my parents in China inspired me to work forward.
Last but not least, I would like to express my love and gratefulness to my wife Bingxin for her
love, patience and support. It is her smile and insistence helping me to overcome the difficul-
ties and also enjoy the life. The past several years were also tough for you, Bingxin. We have
experienced a lot, sometimes maybe too much. But your patience and optimism gave me
courage to continue my research processes.
Abstract III

Abstract
Ultrafiltration (UF) membranes can be used after conventional wastewater treatment to pro-
duce particle free and hygienically safe water for reuse. However, membrane fouling affects
UF performance to a large extent. Without suitable pre-treatment and optimized operational
conditions, the membrane can loose its permeability quickly. Althogh hydraulic backwashing
is periodically used to remove the foulants, the accumulation of hydraulically irremovable
fouling reduces the efficiency of the membrane permanently. Thus, qualitative and quantita-
tive identification of the major foulants is necessary to understand and control fouling.

This research investigated the variation of fouling potential of domestic wastewater applying
bio-filtration as a pre-treatment process. The aim is to identify and quantify the major foulants
and understand fouling control effects using bio-filtration. Pilot-scale bio-filtration processes
were used to treat secondary effluent prior to UF. The fouling potential of various water sam-
ples was assessed by lab-scale stirred cell experiments. An UF pilot plant was used to verify
the correlation between the feedwater characteristics and membrane fouling.

The lab-scale stirred cell experiments demonstrated that the dissolved substances larger than
or comparable to the UF pore size are major foulants, in either secondary effluent or in bio-
filtrate. These substances were detected as biopolymers using liquid chromatography with
online organic carbon detection, UV absorption and organic nitrogen detection (LC-OCD-
UV-OND) which were characterized as macromolecular protein-like and polysaccharide-like
substances. The content of biopolymers influenced the filterability of water samples in UF
proportionally. It affected the reversibility of formed fouling crucially. The correlation be-
tween the biopolymer content in feedwater and the performance of UF was verified to be re-
producible by pilot-scale UF experiments.

Bio-filtration was demonstrated to be able to improve the performance of the downstream

connected UF due to the removal of particles, colloids and especially biopolymers from the
secondary effluent. The elimination of biopolymers was observed mainly within the upper
active filter layer and was revealed to be relevant to biological processes. The operational
conditions of bio-filtration, e.g. temperature, biopolymer concentration in feedwater and fil-
tration rate, influenced biopolymer removal efficiency.

After bio-filtraton, although the fouling potential of secondary effluent could be significantly
reduced, operational conditions of UF plant affected fouling development to a large extent.
Backwash interval and filtration flux were identified as the most important operational pa-
rameters affecting the severity of fouling. Severe fouling can appear due to the extension of
filtration time and/or the increase of filtration flux. To control fouling development under
these critical conditions, different chemicals were used to perform chemically enhanced
backwash. The results showed that NaClO and H2O2 were effective reagents in controlling
irreversible fouling. Chemical cleaning was carried out to restore the permeability of a fouled
UF membrane. In the present study, the combination of NaOH and NaClO as cleaning reagent
was identified as an effective choice.
IV Zusammenfassung

Zusammenfassung
In der weitergehenden Aufbereitung von Kläranlagenablauf bietet der Einsatz von Ultrafiltra-
tionsmembranen eine Energie schonende Aufbereitungsmöglichkeit bezüglich pathogener
Keime und Partikeln. Bei der Membranfiltration führen jedoch Ablagerungen auf der Memb-
ran sowie in den Membranporen (Fouling) sukzessive zu einer Reduzierung der Filtrations-
leistung. Für einen stabilen Betrieb der Ultrafiltration sind daher eine geeignete Vorbehand-
lung des Rohwassers und die optimale Betriebsführung der Filtration erforderlich, um den
Permeabilitätsverlust zu reduzieren. Eine qualitative und quanitative Analyse der für das
Fouling primär verantwortlichen Substanzen ist notwendig, um die dominierenden Fouling-
mechanismen zu verstehen und das Fouling kontrollieren zu können.

Im Rahmen dieser Arbeit wurde die Biofiltration zur Vorbehandlung von Kläranlagenablauf
und deren Auswirkung auf das Ultrafiltrationsverhalten untersucht. Dafür wurde eine Pilotan-
lage betrieben, die Biofiltration mit anschließender Ultrafiltration für die Aufbereitung von
Kläranlagenablauf kombinierte. Mit Hilfe von Laborfiltrationsversuchen wurde das Fouling-
potential des biologisch behandelten Abwassers vor sowie nach Durchlaufen der Biofiltration
betrachtet. Parallel konnte über das Filtrationsverhalten der UF-Pilotanlage der Zusammen-
hang zwischen den Rohwassereigenschaften und dem Membranfouling verifiziert werden.

Die Ergebnisse der Laborfiltrationsversuche zeigten, dass sowohl im Kläranlagenablauf als

auch im Biofiltrat gelöste Substanzen größer oder im Größenbereich einer Membranpore das
Fouling hauptsächlich beeinflussen. Bei der Analyse mittels Größenausschlusschroma-
tographie und anschließender Detektion des organischen Kohlenstoffs, der UV-Absorption
sowie des organischen Stickstoffs (LC-OCD-UV-OND) eluierten diese Stoffe als Biopolyme-
re. Steigende Biopolymerkonzentrationen im Rohwasser führten zu einer proportional sinken-
den Filtrationsleitung und wirkten sich ebenfalls auf die Reversibilität der Foulingschicht aus.

Die Biofiltration zeigte sich aufgrund der Entfernung von Partikeln, Kolloiden und insbeson-
dere Biopolymeren aus dem Ablauf der Kläranlage geeignet für die Verbesserung der Leis-
tungsfähigkeit der nachgeschalteten UF. Die Reduzierung von Biopolymeren fand hierbei
hauptsächlich in der oberen aktiven Filterschicht statt und ist damit auf biologische Prozesse
zurückzuführen. Betriebsparameter wie beispielsweise die Temperatur, die Biopolymerkon-
zentration im Rohwasser sowie die Filtrationsgeschwindigkeit beeinflussten die Effizienz der
Biopolymerentfernung.

Die Biofiltration reduzierte das Foulingpotential des Kläranlagenablaufs signifikant. Weiter

wirkten sich optimale Betriebsbedingungen der UF positiv auf ihre Leistung aus. Der Filtrati-
onsflux sowie das Rückspülintervall erwiesen sich als wichtigste Betriebsparameter für eine
Reduzierung des Foulings. Um einen stabilen Betrieb auch unter diesen kritischen Bedingun-
gen zu erreichen, wurden verschiedene Reinigungschemikalien für die Realisierung einer
chemisch unterstützten Rückspülung untersucht. NaOCl und H2O2 erwiesen sich als am effek-
tivsten, um dem irreversiblen Fouling entgegenzuwirken. Sobald die Filtrationsleistung der
UF auf ein Minimum sank, wurden die Membranmodule chemische gereinigt, um die Aus-
gangspermeabilität wiederherzustellen. Die effektivste Reinigungsleistung erreichte die Kom-
bination der Chemikalien NaOH und NaOCl.
Abbreviations V

Abbreviations
AFM atom force microscopy
BAC biological activated carbon adsorption
BC biopolymer concentration, mg C/L
BOD biological oxygen demand
BSA bovine serum albumin
BSF biological rapid sand filtration
BW backwash
BWF backwash flux, L/(m·h), LMH
BWI backwash interval, min
BWT backwash time, sec
CEBW chemically enhanced backwash
CIP cleaning in place
COD COD chemical oxygen demand
COP cleaning out place
DB delivered biopolymer load , mg C/m2
DBPs disinfect by-products
DCB divalent cation bridging
DIN Deutsches Institut für Normung (German Institute for Standardization)
DO dissolved oxygen, mg/L
DOC dissolved organic carbon, mg/L
DOM dissolved organic matter
EBCT empty bed contact time, h
EDTA polyamino carboxylic acid
EfOM effluent organic matter
EPS extracellular polymeric substances
FF forward flush
GAC granular activated carbon adsorption
GFH granular ferric hydroxide
HMW high molecular weight
HS humic substances
LBF lake bank filtratoin
LMH L/(m2·h)
LC-OCD-UVD-OND liquid chromatogram with online organic carbon detector, UV detector
VI Abbreviations

and organic nitrogen detector

LMW low molecular weight

LOQ limit of quantification
MBR membrane bioreactor
MF Microfiltraton
MW molecular weight, Da
MWCO molecular weight cut-off, Da
NF Nanofiltration
NOM natural organic matter
PACl poly aluminium chlorine
PE Polyethylene
PES Polyethylsulphone
POM particular organic matter
PP Polypropylene
PVDF polyvinylidene difluoride
RBF river bank filtration
RNA ribonucleic acid
RO reverse osmosis
SAT soil aquifer treatment
SE secondary effluent
SMP soluble microbial products
SOC synthetic organic compounds
SSF slow sand filtrate
SSFr slow sand filter
SSFn slow sand filtration
SUVA specific UV absorbance, L/(m·mg)
SWITCH EU Project: Sustainable Urban Water Management Improves Tomor-
row’s City’s Health
TDS total dissolved solids, mg/L
TMP trans-membrane pressure, mbar, bar
TMPend TMP at the end of each experiment, mbar
TMPstart TMP at the beginning of each experiment, mbar
TN total nitrogen, mg/L
TP total phosphorus, mg/L
Abbreviations VII

UF ultrafiltration
UV ultraviolet
WWTP wastewater treatment plant

Latin Symbols
A membrane filtration area m2
Cb biopolymer concentration mg C/L
Cf concentration of the subjected substance in the feed water mg/L
Cp concentration of the subjected substance in the permeate mg/L
Cs foulant concentration kg/m3
Da atomic mass unit
dh hydrodynamic diameter of organic molecule µm
H height m
J linear fluid velocity m3/(m2·s), L/(m2·h)
J20 flux normalized to 20 °C m3/(m2·s), L/(m2·h)
k rate constant depending on n
Kw permeability of a membrane L/(m2·h·bar)
Kw0 permeability of a membrane at the beginning of filtration L/(m2·h·bar)
Kw-c permeability after chemical cleaning L/(m2·h·bar)
Kw-end permeability after five days filtration L/(m2·h·bar)
Kw-start permeability before filtration L/(m2·h·bar)
L length m
Mb biopolymer mass loading per unit membrane area mg/m2
mc dry weight of foulants per unit membrane area kg/m2
n flow behaviour index characterizing fouling mechanisms
Pr permeate recovery %
R resistivity m ·cm
Rc hydraulic resistance of cake/gel layer 1/m
Rf total hydraulic resistance 1/m
Rfi hydraulic fouling resistance due to fouling component i 1/m
Rm hydraulic resistance of the membrane 1/m
Rp restoration of permeability %
Rr retention rate %
RTMP TMP increase per unit permeate mbar/m3
VIII Abbreviations

s compressibility coefficient
t time sec
T temperature ºC
V product volume of permeate m3
Vbw permeate volume used for backwash m3
W width m

Greek symbols
α specific cake/gel resistance m/kg
α0 specific cake/gel resistance at reference pressure m/kg
P TMP, the pressure difference between feed water
and permeate in dead-end filtration N/m2,bar or mbar
η viscosity of the fluid N·s/m2
Σ summation
Table of Contents IX

Table of Contents
Acknowledgements .....................................................................................................................I
Abstract .................................................................................................................................... III
Zusammenfassung.................................................................................................................... IV
Abbreviations ............................................................................................................................ V
Table of Contents ..................................................................................................................... IX
1 Low-pressure Membrane Filtration in Wastewater Reclamation ...................................... 1
1.1 Application of Low-pressure Membrane in Wastewater Reclamation ...................... 1
1.2 Research Objectives ................................................................................................... 3
1.3 Structure of the Dissertation....................................................................................... 4
2 Ultrafiltration Process and Fouling Phenomenon .............................................................. 5
2.1 Classification of Membrane Processes in Water Treatment ...................................... 5
2.2 Ultrafiltration in Water Reclamation.......................................................................... 5
2.3 Fouling Phenomenon and Fouling Mechanisms in UF Process................................. 9
2.4 Classification of Foulants......................................................................................... 13
2.4.1 Organic Fouling – Effluent Organic Matters (EfOM) ..................................... 14
2.4.2 Biofouling......................................................................................................... 16
2.4.3 Inorganic Fouling ............................................................................................. 16
2.4.4 Particulate and Colloidal Fouling..................................................................... 17
2.5 Complexity of Fouling Phenomena.......................................................................... 17
2.5.1 Influence of Membrane Property ..................................................................... 17
2.5.2 Solution Chemistry........................................................................................... 19
2.5.3 Co-effect of different Organic Foulants ........................................................... 20
2.6 Quantification of Fouling Potential.......................................................................... 20
2.7 Fouling Control ........................................................................................................ 22
2.7.1 Fouling Control by Pre-treatment Processes .................................................... 22
2.7.2 Influence of Operational Conditions on Fouling.............................................. 24
3 Material and Analysis....................................................................................................... 30
3.1 Lab and Pilot-scale UF Membrane Filtration Test................................................... 30
3.2 Pilot-scale Bio-filtration Systems............................................................................. 33
3.2.1 Short-term Bank-filtration in WWTP Beixiaohe-Beijing ................................ 33
3.2.2 Slow Sand Filtration in WWTP Ruhleben-Berlin............................................ 34
3.3 Analysis .................................................................................................................... 35
X Table of Contents

3.3.1 LC-OCD-UVD-OND ....................................................................................... 35

3.3.2 Fluorescence Exitation Emission Matrix ......................................................... 36
3.3.3 Photometric Protein and Polysaccharide Analysis........................................... 36
3.3.4 DOC Measurement........................................................................................... 37
3.3.5 On-line Measurement of Oxygen ..................................................................... 38
3.3.6 Analysis of Bulk Water Quality Parameters .................................................... 38
4 Identification and Quantification of Major Organic Foulants.......................................... 40
4.1 Identication of Major Foulants................................................................................. 40
4.1.1 Impact of the Size of Foulants on Filterability................................................. 40
4.1.2 Influence of Foulant Content on Water Filterability ........................................ 42
4.1.3 Effect of Foulant Content on the Reversibility of Fouling............................... 46
4.2 Fouling Mechanisms in UF ...................................................................................... 49
4.2.1 Fouling Mechanisms in Dead-end Filtration.................................................... 49
4.2.2 Fouling Mechanisms in Multi Cycle Filtration ................................................ 51
4.3 Comparison of Fouling Potential Measurements ..................................................... 52
4.3.1 Photometric Measurement of Polysaccharide and Protein............................... 52
4.3.2 Fouling Index and Hydraulic Fouling Resistance ............................................ 54
4.4 Discussion and Summary ......................................................................................... 56
5 Removal of Organic Foulants in Bio-filtration Processes................................................ 59
5.1 Fouling Control by Short-term Bank-filtration ........................................................ 59
5.2 Fouling Control by Slow Sand Filtration ................................................................. 61
5.3 Organic Foulants Removal in Slow Sand Filtration................................................. 62
5.3.1 Operation of Pilot-scale Slow Sand Filters ...................................................... 62
5.3.2 General Water Quality of Secondary Effluent and Slow Sand Filtrates .......... 62
5.3.3 Removal of Organic Compounds in Slow Sand Filtration............................... 63
5.4 Factors Affecting the Performance of Slow Sand Filtration .................................... 67
5.4.1 Filter Ripening.................................................................................................. 67
5.4.2 Temperature and Biopolymer Concentration in Secondary Effluent............... 67
5.4.3 Protective Layer ............................................................................................... 73
5.5 Discussion and Summary ......................................................................................... 73
5.5.1 Biopolymer Removal in Slow Sand Filtration ................................................. 73
5.5.2 Influencing Factors on the Performance of Slow Sand Filtration .................... 74
6 Pilot-scale Ultrafiltration of Bio-filtrate........................................................................... 76
6.1 Overview of the Experiments................................................................................... 76
Table of Contents XI

6.2 Influence of Flux on Fouling.................................................................................... 78

6.2.1 Pilot-scale Investigation ................................................................................... 78
6.2.2 Effect of Filtration Flux on Biopolymer Fouling ............................................. 79
6.3 Influence of Biopolymer Load Mass on Fouling Development............................... 82
6.3.1 Impact of Backwash Interval............................................................................ 82
6.3.2 Effect of Foulant Concentration in Feedwater ................................................. 82
6.3.3 Fouling Development Processes ...................................................................... 84
6.4 Influence of Backwash Time on Fouling Control .................................................... 87
6.5 Chemically Enhanced Backwash ............................................................................. 88
6.6 Membrane Cleaning ................................................................................................. 90
6.7 Discussion and Summary ......................................................................................... 94
7 General Conclusions ........................................................................................................ 99
7.1 Major Organic Foulants in UF and related Fouling Mechanisms ............................ 99
7.2 Removal of Organic Foulants in Bio-filtration Processes...................................... 100
7.3 Optimization of Operational Conditions ................................................................ 101
7.4 Research Recommendations .................................................................................. 101
8 Appendix ........................................................................................................................ 103
8.1 LC-OCD-UVD-OND System and LC-OCD Chromatogram ................................ 103
8.2 Amicon Cell Results............................................................................................... 107
8.3 Investigation of Slow Sand Filtration on Biopolymer Removal ............................ 111
8.4 Figure contents ....................................................................................................... 115
8.5 Table contents ........................................................................................................ 118
9 Literatures....................................................................................................................... 120
XII Table of Contents
Low-pressure Membrane Filtration in Wastewater Reclamation 1

1 Low-pressure Membrane Filtration in Wastewater Reclamation

1.1 Application of Low-pressure Membrane in Wastewater Reclamation

Wastewater Reuse and Low-pressure Membrane Filtration
Over the past decades there has been an increasing interest for wastewater reclamation and
reuse in the world, particularly in arid and semi-arid regions, to meet growing demand on
available water supplies. Driving forces of this trend arise from and are being strengthened
through water shortage caused by climate change or pollution, through increasing demand of
fresh water for urbanization, industrialization and increasing population (Jimenez and Asano,
2008). Different from long distance water transfer or exploitation of limited natural water
resources, large numbers of wastewater treatment plants (WWTP) can be utilized as reliable
and economical alternative water sources (Chu et al., 2004). After suitable treatment process,
treated wastewater can be used either for indirect potable purposes (Ernst et al., 2007), or
even for direct potable reuse (Wintgens et al., 2005).

In wastewater reclamation, suitable treatment processes are chosen fundamentally based on

water quality requirement, economical issues and reliability of the technique. As an innova-
tive technique, low-pressure membranes like microfiltration (MF) and ultrafiltration (UF)
have got more interest in this area due to the high water quality effluent, reliability in opera-
tion and small footprint (Wintgens et al., 2005). With the decrease of membrane prices, more
membrane equipments have been and will be set up for water reclamation (Sullivan, 2007).
Since the removal of contaminants through low-pressure membranes is predominantly due to
size exclusion of membrane pores in the lower µm range (Crittenden et al., 2006), MF and UF
have the ability to separate suspended solids, large dissolved organic compounds and patho-
gens like protozoan cysts, bacteria and/or viruses from reclaimed water for indirect water re-
use, such as urban purpose reuse (e.g., toilet flushing, auto washing and green irrigation, de
Koning et al., 2008). They can also be used to reduce fouling on the downstream high-
pressure membranes to produce high quality reuse water (U.S.EPA, 2004).

Based on a report about global perspective of low-pressure membranes (Furukawa, 2008), in

December 2006 the installed treatment capacity using MF and UF reached 13·106 m3/d, 25%
of which has been used in wastewater treatment and pre-treatment prior to reverse osmosis
membranes (RO).
Tab. 1-1 shows some current projects with large-scale low-pressure membrane facilities for
water reuse. More projects or studies on advanced treatment and reuse using low-pressure
membranes combined with other treatment processes e.g. flocculation, high-pressure mem-
brane filtration (NF and RO), soil aquifer treatment (SAT) can be found all over the world
(Fan et al., 2008; Kim et al., 2008; Jarusutthirak et al., 2002). With the increasing demand for
wastewater reclamation and more stringent water quality regulatory standards, the low-
pressure membrane technology exhibits as a reliable post-treatment choice after the
conventional treatment approaches (Furukawa, 2008).
2 Low-pressure Membrane Filtration in Wastewater Reclamation

Tab. 1-1 Case projects of wastewater reclamation applying low-pressure membrane filtration

Reuse purpose Location Capacity Treatment process Literature

(m3/d)
Direct potable Windhoek 21 000 Pre-O3 + coagulation + dual Wintgens et
reuse (Namibia) media filtration + O3 + BAC al., 2005
+ GAC + UF + chlorine
Indirect potable Newater Project 91 000 MF + RO + UV disinfection Qin et al.,
reuse (Singapore) 2006
Groundwater Factory 21 265 000 MF + RO + UV/ H202 Water21, 2008
recharge California (U.S.A.)
Irrigation, wa- Olympic park 2000 7 500 cross-flow MF de Koning et
ter fountains, Sydney al., 2008
domestic uses (Australia)
Domestic uses Olympic Park 2008 80 000 PACl flocculation + MF + Ernst et al.,
Beijing (China) GAC 2007
BAC: biological activated carbon adsorption, GAC: granular activated carbon adsorption, PACl: poly
aluminium chlorine

Fouling Problem in Low-pressure Membrane Filtration

Along with the wide spreading utilization of low-pressure membranes in wastewater reclama-
tion, the fouling problem has always been considered as one of the major limitations to in-
crease the acceptance of the process (Haberkamp, 2008). Fouling is ‘the process resulting in
loss of performance of a membrane due to the deposition of suspended or dissolved sub-
stances on its external surfaces, at its pore openings, or within its pores’ (Crittenden et al.,
2006). In fouling phenomena, all of the constituents in feed water that can be retained on the
membrane surface or in the membrane pores are called foulants (fouling materials). Due to
fouling, the productivity of membrane processes is reduced, membrane lifespan is shortened
and operation costs are increased. Therefore, the scientific understanding and control of
membrane fouling is of particular concern.

600

membrane fouling

membrane fouling in filtering secondary effluent

Number of publications

Source: Science Direct

400

200

0
2000 2002 2004 2006 2008 2010
Year

Fig. 1-1 Annual publications on membrane fouling (source: Science Direct)

Low-pressure Membrane Filtration in Wastewater Reclamation 3

During the last several years, with the expotential growth of installed MF and UF treatment
capacity (Furukawa, 2008), a considerable number of papers have been published about low-
pressure membrane fouling. According to Science Direct, the annual publication increased
from 200 by 2004 to over 600 by 2009. The number of studies investigating fouling in low-
pressure membrane processes filtering secondary effluent (SE) increases with a similar trend
(Fig. 1-1). The significant increase of publications indicates the increasing importance of un-
derstanding fouling phenomenon in wastewater reclamation using low-pressure membranes,
and moreover, the requirement of correspondingly suitable fouling control strageties.

In practice, biologically treated domestic wastewater always shows high fouling potentials in
membrane filtration. Direct filtration is identified as an unreasonable operational strategy as
the low-pressure membranes can be fouled seriously and loose its permeability quickly (Xie
et al., 2006; Fan et al., 2008). During long-term operation, hydraulically irremovable interac-
tions between foulants and membrane result in a continuous reduction of the membrane per-
meabilitiy although different fouling control methods, e.g. hydraulic flushing, are performed
periodically. Numerous studies have shown that effluent organic matters (EfOM) are the ma-
jor foulants in membrane filtration of SE (e.g. Jarusutthirak and Amy, 2006). In several previ-
ous studies, it is evident that a specific fraction of organic compounds in EfOM are principle
foulants in UF of treated domestic wastewater (Laabs, 2004; Haberkamp, 2008).
Nevertheless, EfOM represents a combination of various components. In terms of chemical
nature they could be proteins, polysaccharides, humic substances, lipids, nucleic acids, etc.
During membrane filtration these substances interact with the membrane in different manners.
Due to the complexity of fouling phenomenon and also conflicting results on the nature of the
foulants (Li et al., 2008), further studies on EfOM fouling in low-pressure membrane process
are necessary. To control EfOM fouling, different pre-treatment processes prior to membrane
filtration are commonly used, e.g., coagulation, adsorption, preoxidation and biofiltration
(Huang et al., 2009; Lehman and Liu, 2009; Sperlich et al., 2008). Althoug the foulants rem-
oval effect can be evaluated by comparing the performance of a downstream membrane unit,
a fundmental understanding of the character of major foulants and a quantitative assessment
of their content is necessarily required to assess the fouling control effect of different strate-
gies, and further more, to understand their removal mechanisms.

1.2 Research Objectives

The primary objectives of the present study are to identifiy the major foulants in UF of treated
wastewater and understand their removal in bio-filtration processes. Specially, the main ob-
jectives were:

1) Identify major foulants based on the quantitative correlation between the content of
foulants in feedwater and their fouling effect.
2) Compare the suitability of different foulant quantifying methods.
3) Understand fouling mechanisms during UF.
4) Demonstrate the effect of bio-filtration on the performance of UF.
5) Study foulant removal processes within a bio-filter.
6) Investigate operational factors of bio-filtration on its performance in foulant removal.
To support the major objectives in understanding fouling phenomena, additional objec-
tives were as follows:
4 Low-pressure Membrane Filtration in Wastewater Reclamation

7) Investigate the influence of operational conditions on fouling development in UF and

suggest sustainable operating strategy.
8) Understand fouling development in long-term membrane operation based on pilot-
scale experiments.
9) Develop membrane cleaning methods.

The study has been carried out within the scope of the German – Chinese cooperation project
entitled ‘Sustainable water management for Olympia Park 2008’ (Contract no.: 02WA0524,
02WA1016) subproject 4 ‘Wastewater treatment and reuse’ and within the European research
project ‘Sustainable Urban Water Management Improves Tomorrow’s City’s Health
(SWITCH)’. SWITCH is supported by the European Commission under the 6th Framework
Programme and contributes to the thematic priority area of “Global Change and Ecosystems”
[1.1.6.3] Contract n° 018530-2.

1.3 Structure of the Dissertation

This dissertation continues in chapter 2 with an introduction to UF membranes and a descrip-
tion of UF in water treatment. In relevance to membrane fouling, a review about the charac-
teristics of different foulants and their removal are presented. Chapter 3 introduces the mate-
rial and analysis applied in the study. In the present work pilot-scale bio-filtration (short-term
bank filtration) prior to UF was carried out in WWTP Beixiaohe (Beijing, China) in 2005.
From 2006 to 2008 fouling control effects of pilot-scale bio-filtration (slow sand filtration)
were investigated in WWTP Ruhleben in Berlin (Germany). For the pilot-scale experiments
the same UF pilot plant and the same type of membrane module was used. The water quality
analysis includes bulk water quality parameters and common organic foulant parameters, e.g.
proteins, polysaccharides and biopolymers.

Chapter 4 identifies firstly the attribution of size fractionated foulants on fouling in UF filter-
ing treated domestic wastewater. Subsequently, a quantitative correlation between specific
fractional substances and fouling is presented based on lab-scale experimental results. The
correlation is further validated based on pilot-scale membrane filtration results. Additionally,
the relationship of different methods determining fouling potential is quantitatively compared.
In Chapter 5, the fouling control effect by bio-filtration prior to UF is demonstrated. The
removal of different foulants within slow sand filter and factors affecting the removal effect
has been investigated. Although foulants are significantly removed by slow sand filtration,
pilot-scale UF results show that operational conditions play also an important role in fouling
development. In Chapter 6 several operational variables, e.g. filtration flux, back wash inter-
val (BWI) and back wash time (BWT) on fouling development in the UF pilot plant are inves-
tigated. In particular, the effect of filtration flux on fouling is quantitatively interpreted using
lab-scale results.

Finally, the main findings of the current study are discussed and presented in Chapter 7. In
this chapter some suggestions are also made for further studies.
Ultrafiltration Process and Fouling Phenomenon 5

2 Ultrafiltration Process and Fouling Phenomenon

2.1 Classification of Membrane Processes in Water Treatment

Membrane filtration is a filtration process in which membrane material acts as a selective bar-
rier between two phases (Mulder, 2000). As a result of the driving force and due to the selec-
tivity of the membrane, components transported to the membrane can pass through (in which
case they become the permeate stream) or be rejected by it (in which case they become the
retentate or concentrate stream, Fig. 2-1).

Retentate
Feed Permeate

Membrane

Fig. 2-1 Schematic illustration of a membrane separation process

For water and wastewater treatment, pressure is generally used as the driving force to push
water through the membrane; the degree of the selectivity of the applied membrane depends
on the membrane pore size. Based on the pore size range, the pressure driven membrane can
be subdivided into microfiltration (MF), ultrafiltration (UF), nanofiltration (NF) and reverse
osmosis (RO). In respect of driving pressure range, MF and UF are classified as low-pressure
membrane processes and high-pressure membranes include NF and RO membrane. Tab. 2-1
shows an overview of the classification of membrane processes available in water treatment.

Tab. 2-1 Membranes applied for water treatment (Mulder, 2000; Judd, 2006)
Membrane Pressure Pores Removable components Major separation
process (bar) (nm) in water treatment mechanism
Microfiltration 0.1 - 2 100 - 1000 Suspended solids, bacteria Sieving
Ultrafiltration 0.1 - 2 10 - 100 Macromolecules, viruses, Sieving
Nanofiltration 4 - 20 1- 10 Micropollutants, Charge rejection,
bivalent ions solution-diffusion and
sieving
Reverse Osmosis 10 - 30 <1 Monovalent ions Solution-diffusion

2.2 Ultrafiltration in Water Reclamation

Based on water quality requirements and energy consumption specifications, the use of low-
pressure membranes (MF/UF) in water and wastewater treatment has exploded since mid-
1990s. In which, UF is considered to be able to produce hygienically safe permeate compared
to MF due to its smaller pore size (Mulder, 2000). In the present dissertation UF process is
investigated in relation to wastewater reclamation.
6 Ultrafiltration Process and Fouling Phenomenon

Ultrafiltration Membranes
UF membranes typically consist of a thin surface layer (pore size: 0.01 to 0.1 µm, Mulder,
2000) which provides the required selectivity on the top of a more porous support layer (pore
size to 100 µm), and which provides mechanical stability. Different polymeric (organic) and
ceramic (inorganic) materials are used to form membranes to get the desired physical proper-
ties and reasonable thermal and chemical resistance. The most suitable and commonly used
polymers are polyvinylidene difluoride (PVDF), polyethylsulphone (PES), polyethylene (PE)
and polypropylene (PP, Judd, 2006). For inorganic membranes, alumina (α-/γ-Al2O3), zirco-
nia (ZrO2), titania (TiO2) and even metal like stainless steel are suitable membrane materials
(Li et al., 2008).

Membrane Properties
Pore size and pore size distribution
For low-pressure membranes, the separation character can be given as a relevant pore size,
normally in µm for MF, or as molecular weight cut-off (MWCO) in Daltons (Da, one Da
represents one g/mol) for UF. The nominal MWCO of a UF membrane is defined as the mo-
lecular weight of a certain solute which is 90% rejected by the membrane (Mulder, 2000).
The retention rate R can be calculated according to Eq. 2-1.

C f − Cp
Rr = × 100% Eq.2-1
Cf

Where
Rr = retention rate, %
Cp = concentration of the subjected substance in the permeate, mg/L
Cf = concentration of the subjected substance in the feed water, mg/L

R ranges from 0 to 100% to express no rejection (0%) to complete rejection (100%). Based on
the rejection curve using model solutes (frequently dextrans, Wickramasinghe et al., 2009),
the membrane pore size and its distribution can be obtained. For specific substances, the re-
jection rate by a membrane depends on the molecular size. However, the shape of the mole-
cules, their electrical charge and the pH of the feed solution play also important roles (Baker,
2004).

Pure water flux

Another important property of a membrane is its flow rate/flux operated at corresponding
pressures (Eq.2-2).

J
Kw = Eq.2-2
∆P
where
Kw = permeability of a membrane, m3/(m2·s·Pa) or in practice L/(m2·h·bar)
J = the linear fluid velocity, m3/(m2·s) or in practice L/(m2·h)
Ultrafiltration Process and Fouling Phenomenon 7

P = TMP = the pressure difference between feed water and permeate, N/m2, Pa; in practice
bar or mbar is commonly used)

The flow rate (J) of liquid passing through a membrane can be calculated according to
Darcy’s law (Eq.2-3).

dV ∆P
=J= Eq.2-3
Adt ηRm
where
V = the product volume of permeate, m3
t = the filtration time, s
A = the membrane filtration area, m2
η = viscosity of the fluid, N·s/m2
Rm = hydraulic resistance of the membrane, 1/m

The dynamic viscosity η is related to feed water temperature T (ºC) which can be determined
using the following equation (Roorda, 2004).

497 ⋅ 10 −3
η= Eq.2-4
(T + 42.5)1.5
where
T = temperature of the feed water, ºC

Membrane Configuration and Operation

Membrane module configuration
For practical operation, membranes are encapsulated in membrane modules. The term ‘mem-
brane module’ is used to describe a complete unit comprised of membranes, a pressure sup-
port structure, a feed inlet, a concentrate outlet and an overall support structure (Roorda,
2004). The principal types of membrane modules applied in water treatment at an industrial
scale are as following (Li et al., 2008):

Hollow fibre or capillary membranes: the most common configuration of MF and UF for wa-
ter treatment. The hollow fibres are 0.5 – 1.5 mm in diameter, several hundreds to thousands
of fibres are packed in a membrane module.
Tubular membranes are an extension of the hollow fibre configuration with a larger diameter
(up to 25 mm). This kind of module is suitable to treat feed water with a high level of sus-
pended solids.
Spiral wound membranes: two flat sheets of membrane are separated with a permeate collec-
tor channel material to form a leaf. The assembly is sealed on three sides and the forth side is
open to exit permeate.
Plate and frame membranes: are comprised of a series of flat membrane sheets and support
layers. Membranes sandwich support plates which are arranged in stacks. Permeate is col-
lected from each support plate.
8 Ultrafiltration Process and Fouling Phenomenon

The flow direction in tubular and hollow fibre membranes can be inside-out or outside-in. In
inside-out mode the feedwater flows from the inside of the membrane tube to the outside of
the tube. The outside-in configuration shows the opposite flow direction. Generally, the out-
side-in scheme of hollow fibre membranes offers a larger filtration area compared to a similar
model operated in inside-out mode (Crittenden et al., 2006).

Contained (pressurized) and submerged (suction) modules

In water and wastewater treatment processes, membrane modules can be classified into con-
tained pressure vessels or submerged membranes. Contained pressure vessel is the traditional
approach filtering water through low-pressure membranes. Permeate can be obtained due to
significant positive pressure from the feedwater side to the membrane. Trans-membrane pres-
sure (TMP) is developed by a feed pump that increases the feedwater pressure, while the per-
meate stays at a near-atmospheric pressure. Submerged membranes are an assembly of mem-
branes positioned in a feedwater tank typically open at the top. The liquid to be filtered is fed
to the tank and permeate is removed under suction, either by a pump or by a gravity effect.
While the contained pressure vessels can be operated in inside-out and/or outside-in mode,
filtering water using submerged membranes is always in outside-in mode (Li et al., 2008).

Operation of UF
Membrane systems can be operated at either constant flux or constant pressure. Operated at
constant flux mode TMP is increased to maintain the constant permeate flux as the membrane
fouls. During constant pressure filtration, the TMP is maintained constant but the productivity
decreases continuously as a result of fouling. In practical operation, membrane systems are
commonly operated at constant flux.

Considering feed water flux direction, UF and MF systems can be operated in cross-flow or
dead-end filtration. In cross-flow configuration, the feed water is supplied in parallel with the
membrane surface. A portion of the feed water passes through the membrane and forms per-
meate, the rest forms concentrate and is recirculated with additional feed water for filtration.
To control the concentration of the solids in the recirculation loop, the concentrate is dis-
charged at a specific rate. In dead-end filtration, all the feed water is forced pass through the
membrane. In practical operation, a fraction of permeate is used for periodical backwashing to
maintain the permeability of the membrane. As the energy consumption in dead-end mode
with backwashing is lower than that in cross-flow mode (Li et al., 2008), most of the low-
pressure membrane systems are operated in dead-end.

Application of Low-pressure Membranes in Wastewater Relamation

For wastewater reclamation, low-pressure membranes are being used either as a polishing
step after secondary/tertiary treatment (Fig. 2-2 a) or as a part unit of membrane bio-reactors
(MBR, Fig. 2-2 b). As a post-treatment process, MF/UF is considered primarily as a water
quality improvement unit, e.g. as a physical disinfection process removing bacteria and parti-
cles. In MBR systems, besides effluent quality control, low-pressure membranes are used to
concentrate up the biomass within the biological wastewater treatment unit. Subsequently, it
reduces the necessary tank size and increases the effciency of the biotreatment process. MBRs
thus tend to generate treated waters of higher purity with respect to dissolved constitutents.
Ultrafiltration Process and Fouling Phenomenon 9

Moreover, by removing the requirement for biomass sedimentation, the flow rate through a
MBR cannot affect product water quality through impeding solids settling, as is the case for
an activated sludge process (Judd, 2006). Currently, MBR has been actively employed for
wastewater treatment and reclamation and efforts have been mainly addressed on the im-
provement of its performance while decreasing treatment costs (e.g. Meng et al., 2009).

screened primary biological secondary membrane

influent sedimentaiton treatment sedimentation filtration

(a)

returned sludge

screened primary membrane bio-reactor

influent sedimentaiton (MBR)

(b)

Fig. 2-2 Application of low-pressure membranes (a) as a post-treatment unit after conventional bio-
logical wastewater treatment processes for water reclamation, (b) in a MBR system to produce reuse
water

2.3 Fouling Phenomenon and Fouling Mechanisms in UF Process

During membrane filtration the flux declines over time although all operating parameters are
kept constant. This phenomenon is caused by fouling. Fouling refers to the blockage of mem-
brane pores during filtration due to the combination of sieving and adsorption of particulates
and dissolved compounds. The accumulation of foulants can appear either on the membrane
surface which forms a cake/gel layer or within the membrane pores (Li et al., 2008). As foul-
ing decreases membrane performance and shortens membrane life, it is one of the most im-
portant issues hindering the widespread application of membrane filtration.

Classification of Fouling
Fouling can be broadly classified into reversible and irreversible fouling (Fig. 2-3). Reversible
fouling in this dissertation refers to the fouling that can be removed by backwashing or flush-
ing after each filtration cycle. Irreversible fouling is the fouling that cannot be removed by
hydraulic processes. This item can be further classified into chemically reversible and irre-
versible fouling, in which the chemically irreversible fouling is defined as the fouling that
cannot be removed through chemically enhanced backwash (CEBW) or chemical cleaning.
10 Ultrafiltration Process and Fouling Phenomenon

constant flux

hydraulic reversible fouling chemical cleaning

TMP

hydraulic irreversible fouling chemically irreversible fouling

Permeate volume / operating time

Fig. 2-3 Illustration of reversible and irreversible fouling in UF operated at constant flux

Based on the way the foulants accumulate, different fouling mechanisms can appear individu-
ally or simultaneously (Roorda, 2004; Bowen and Jenner, 1995; Böddeker, 2007; Foley,
2006). The following descriptions are the major fouling mechanisms in UF (Fig. 2-4):

Back diffusion
boundary layer
Concentration

Adsorption

Permeate

Feed
Pore blocking

Porous
membrane
Concentration Cake/gel layer
polarizition and compression

Fig. 2-4 Schematic diagram of a fouled membrane by various fouling mechanisms (adapted from
Roorda, 2004, revised)

Concentration polarisation due to high concentration of foulants (macromolecules) near to the

membrane surface (reversible) and/or subsequent solidification due to gel-polarization (less
reversible);
Adsorption inside the membrane pores (less reversible);
Ultrafiltration Process and Fouling Phenomenon 11

Blocking of the membrane pores due to sieving effect and followed by cake/gel layer forma-
tion (more reversible) and compression.

Quantification of the hydraulic resistance caused by these fouling types can be obtained based
on resistance-in-series model. The model applies a resistance value to each component of
membrane fouling with the assumption that each contributes to hydraulic resistance and they
act independently from one another (Crittenden et al., 2006). Eq. 2-5 presents the resistance-
in-series relationship to describe the relationship of flux and the resistance of a fouled mem-
brane:

∆P
J= Eq.2-5
η ( Rm + ΣR fi )

where
Rm = the hydraulic resistance caused by membrane, 1/m
Rfi = hydraulic fouling resistance of fouling component i, e.g., reversible and irreversilbe
fouling, or due to adsorption, cake layer and gel layer, 1/m

Analysis of Fouling Mechanisms

Fouling caused by concentration polarization
During membrane filtration, the separation of solute and solvent takes place at the membrane
surface where the solvent passes through the membrane and the retained solute causes the
local concentration to increase, this effect is defined as concentration polarization (Bowen and
Jenner, 1995). The localized accumulation on the membrane could be either particle polariza-
tion or concentration polarization (applies to dissolved species) which leads to an osmotic
counter-pressure against the driving force. The basic equation describing the relationship be-
tween flux and driving force is as following:

∆P − ∆Π
J= Eq. 2-6
η ( R m + Rd )
where
Rd = hydraulic resistance due to reversible or irreversible deposition of solute onto the mem
brane surface, 1/m
П = П(cm) – П(cp), the osmotic pressure difference across the membrane, mbar
П(cm) = the osmotic pressure at the membrane surface, mbar
П(cp) = the osmotic pressure in the permeate, mbar

In low-pressure membrane processes for water and wastewater treatment, the osmotic pres-
sure is in general negligible. Therefore, osmotic effects are frequently ignored by calculation
(Bowen and Jenner, 1995). However, the concentration polarization effect is important in re-
spect to the formation of membrane fouling. During filtration, the solutes accumulated at the
membrane surface have a tendency to diffuse back into bulk solution according to Fick’s law
of diffusion. At a steady state condition the solute mass balance above the membrane surface
can be achieved which gives the rate of convective transport of solute towards the membrane
surface equal to the rate of solute leakage through the membrane plus the rate due to back-
12 Ultrafiltration Process and Fouling Phenomenon

diffusion. Withdrawing feed force leads to back-diffusion of foulants into the solution. When
the hydraulic resistance disappears after feed force withdrawing, the fouling is considered
reversible. However, under certain conditions concentration polarization can lead to an irre-
versible buildup. For instance, when the concentration of macrosolutes at the membrane sur-
face reaches its solubility limit and precipitates to form a solid or thioxotropic gel layer, the
consolidated structure is slow in diffusing into water and leads to a flux reduction, a fraction
of which could be irreversible. Current analysis (Wang and Waite, 2009) has revealed that the
formed gel layer in filtering wastewater effluent is a cross-linked three-dimensional network
consisting of macromolecular compounds. The interactions among these organics with mem-
brane are assisted or strengthened by bridging via multivalent metals and/or specific mole-
cules.

Fouling caused due to pore blocking

Based on the pore sealing effect a theoretical description of the fouling behaviour in dead-end
membrane filtration was introduced by Hermia (1982). In the physical model the pressure
applied is set constant. The membrane pores are assumed as straight cylindrical with identical
pore diameter and length. The flow through the pores is assumed as laminar. Based on the
modes in which the foulants accumulate (Fig. 2-5), the following four basic filtration laws can
be used to describe the fouling phenomena:

(a) (b)

(c) (d)

Fig. 2-5 Fouling mechanisms related with blocking laws (a) complete blocking, (b) intermediate
blocking, (c) standard blocking, (d) cake filtration

Complete pore blocking law: For complete blocking it is assumed that each particle reaching
the membrane participates in the blocking phenomenon by pore sealing (Fig. 2-5 a, i.e. no
flow at all through the pore), the particles are not superimposed upon another (i.e. the parti-
cles form a single particle layer).

Intermediate blocking law: Based on the complete blocking law and considering the probabil-
ity for a particle to block a pore, this law describes the filtration behaviour that the following
particles (the second particle layer) have an equal probability to settle on the first layer as on
the membrane surface which was left free (Fig. 2-5 b).
Ultrafiltration Process and Fouling Phenomenon 13

Standard blocking filtration law: It is assumed that pore volume decreases proportionally to
the filtrate volume by particles depositing/adsorption on the pore walls (Fig. 2-5 c). This im-
plicates a reduction of the pore diameter.

Cake filtration law: Assuming resistance in series and a constant superimposition, the resis-
tance of the fouling layer increases proportionally to the accumulated particle mass (Fig. 2-5
d).

In relevance to these possible physical phenomena, a unified equation describing the blocking
filtration laws was derived to represent the filtration characteristics (Eq.2-7, Hermia, 1982):

n
d 2t  dt 
2
= k  Eq.2-7
dV  dV 
where
t = filtration time
V = the total permeate volume
k = rate constant depending on n
n = flow behaviour index characterizing fouling mechanisms

Depending on the exponent n, the dominant fouling mechanism is:

n =2 complete pore blocking
n = 1.5 standard blocking
n =1 intermediate blocking
n =0 cake filtration

Similar derivations for constant flux filtration ( Huang et al., 2008a; Bolton et al., 2006) or for
varying flux and TMP filtration (Augstin Suarez and Veza, 2000) have also been presented.
The model has been successfully used to explain some fouling phenomena in dead-end mem-
brane filtration processes (Yuan et al., 2002; Costa et al., 2006; Roorda, 2004).

However, what the model describes is the behaviour of rigid, spherical and neutrally charged
particles in dilute solutions during filtration through straight cylindrical pores. The effects of
foulant-foulant and foulant-membrane interactions and fouling layer (cake/gel layer) com-
pression, which are significantly influenced by the physical and chemical characteristics of
foulants, are generally ignored. Nevertheless, these factors are considered to be important to
fouling phenomena in practical operation (Zhang et al., 2008; Roorda, 2004; Huisman et al.,
2000). To get a better understanding of fouling phenomena, it is necessary to characterize the
foulants and understand their interactions in detail.

2.4 Classification of Foulants

According to the type of foulants, the corresponding fouling in wastewater treatment and
reclamation can be generally classified into organic, inorganic, biological and particu-
lar/colloidal fouling (Amy, 2008; Li et al, 2008).
14 Ultrafiltration Process and Fouling Phenomenon

2.4.1 Organic Fouling – Effluent Organic Matters (EfOM)

In wastewater reclamation, organic fouling is considered to be a major constraint for the ap-
plication of UF. The EfOM which are considered as major organic foulants have been inves-
tigated intensively (Jarusutthirak et al., 2002 ; Laabs et al., 2006; Shon et al., 2006; Amy,
2008). Generally, EfOM in wastewater effluents originate from three different sources: 1)
natural organic matters (NOM) present in the drinking water, 2) synthetic organic compounds
(SOC) discharged by consumers and disinfection by-products (DBPs) generated during disin-
fection processes, and 3) soluble microbial products (SMP) generated during biological
wastewater treatment (Amy, 2008; Barker and Stuckey, 1999; Drewes and Fox, 1999). Typi-
cal organic constituents in treated domestic wastewater and their size ranges are shown in Fig.
2-6.

Dissolved organic matter (DOM) Particular organic matter (POM)

Recalcitrant
matter Humic acid, fulvic acid Bacteria

Nutrient RNA Cell fragment

Chlorophyl Polysaccharide Organic debris

Carbohydrate Protein

Amino acid Vitamin Virus Algae and protozoa

Fatty acid Extracellular enzyme DNA

Separation range of UF

10 102 103 104 105 106 107 108 109 Molecular weight (dalton)
102 (µm)

10-4 10-3 10-2 10-1 100 101 Nominal size (µm)

Fig. 2-6 Typical organic constituents in biological treated domestic wastewater and their size ranges
(Adapted from Levine et al., 1985)

Although NOM in drinking water (surface water as a source) attributes to fouling in low-
pressure membrane filtration (Hallé et al., 2009; Kennedy et al., 2008; Huang et al., 2007),
recent evidences have shown that the EfOM foulants in membrane filtration of treated domes-
tic wastewater are different to NOM foulants in drinking water. Huang (2007) compared the
fouling effect of UF treating surface water and biologically treated wastewater. It is evident
that the foulants from SE reduce the permeability of UF more significantly than that from
surface water but the formed fouling shows a higher reversibility. Further more, the high mo-
lecular organic compounds demonstrate much more severe fouling effect than other organic
fractions. This result coincides with the result reported by previous studies (Jarusutthirak,
2002; Laabs et al., 2006). These substances, based on the study made by Jarusutthirak and
Amy (2006), are mostly SMP and/or extracellular polymeric substances (EPS) which are
Ultrafiltration Process and Fouling Phenomenon 15

mainly formed during biological wastewater treatment processes (Jarusutthirak and Amy,
2007).

SMP is defined as ‘the pool of organic compounds that result from substrate metabolism
(usually with biomass growth) and biomass decay during the complete mineralization of sim-
ple substrates’ (Barker and Stuckey, 1999). This definition concerns soluble cellular compo-
nents in relation to different activities of micro-organisms e.g. excreted by micro-organisms
due to their interaction with the environment, produced as a result of substrate metabolism
and bacteria growth or released during the lysis and degradation of micro-organisms
(Chudoba, 1985). Some of SMP have been identified as humic and fulvic acid, polysaccha-
rides, proteins, nucleic acids, amino acids, structural components of cells and products of en-
ergy metabolism (Barker and Stuckey, 1999; Jarusutthirak and Amy, 2006).

EPS is used as a general term which encompasses all classes of autochthonous macromole-
cules such as carbohydrates, proteins, nucleic acids, phosphorous lipids and other polymeric
compounds found at or outside the cell surface and in the intercellular space of microbial ag-
gregates (Flemming and Wingender, 2001). Typically, the EPS solution is characterised ac-
cording to its relative content of proteins and polysaccharides. Although there is some differ-
ence between the definitions of SMP and EPS, it is now widely accepted that the concepts of
soluble EPS and SMP in treated domestic wastewater are identical (Judd, 2006).

The fouling effect of SMP/EPS in UF of EfOM-containing wastewater has been intensively

investigated (Haberkamp, 2008; Laabs et al., 2006). Laabs et al. (2006) compared the fouling
potential of treated wastewater after different pre-treatment processes prior to UF. Using liq-
uid chromatography with online detectors (LC-OCD) they found that the macromolecules
play a more crucial role in fouling than other dissolved fractions smaller than the membrane
pore size. Such substances block membrane pores, form cake/gel layer on membrane surface
and reduce membrane permeability seriously (Laabs and Amy, 2006). In terms of chemical
nature they can be characterized as polysaccharide-like and protein-like substances (Amy,
2008; Her et al., 2003; Huber, 1998). Studies investigating characteristics of foulants in both
conventional biological treatment processes and MBR systems support this conclusion
(Rosenberger et al., 2006; Jarusutthirak and Amy, 2006; Liang et al., 2007; Geng and Hall,
2007). Using inline attenuated total reflection (ATR) – fourier transform infrared (FTIR)
spectrometry, proteins and polysaccharides have been identified as the major organic foulants
in irreversible fouling (Shon et al., 2004). The important role of protein-like and polysaccha-
ride-like substances can be further confirmed using other analysis methods, e.g., fluorescence
excitation-emmison-matrix (EEM) for proteins (Her et al., 2003). Using these advanced
methods, Haberkamp (2008) investigated fouling phenomenon in a PVDF UF membrane fil-
tering treated domestic wastewater. Besides the identification of macromolecular protein-like
and polysaccharide-like compounds as primary foulants, protein-like substances have been
shown to play a more crucial role in irreversible fouling than other compounds (Haberkamp,
2008).

To understand fouling mechamisms, the fouling effect of model substances which have simi-
lar character like SMPs has also been investigated (Wang and Waite, 2008a; Torrestiana-
Sanchez et al., 2007). Frequently, dextran and alginate are used to represent polysaccharides
16 Ultrafiltration Process and Fouling Phenomenon

while bovine serum albumin (BSA), pepsin and myoglobin and are used to represent proteins.
It is evident that during membrane filtration, these organic compounds can accumulate and/or
be adsorbed on the membrane surface and/or in membrane pores. The interaction between
these organics and membranes leads to irreversible fouling (Susanto et al., 2007; Susanto and
Ulbricht, 2005). In long-term operation, model substances such as alginate can form irreversi-
ble fouling which results in continuous increase of hydraulic fouling resistance (Ye and Chen,
2005).

2.4.2 Biofouling
Biofouling is defined as the undesired development of microbial layers (biofilms) on mem-
brane surfaces (Li et al., 2008). Biofilm is an expression which applies to microbial life in
aggregates combined through EPS which contains mainly polysaccharides and proteins
(Flemming, 2002). Generally, biofouling mechanisms in membrane processes include the
adsorption of soluble and suspended EPS on membrane surfaces and in membrane pores, the
clogging of membrane pore structure by fine colloidal particles and cell debris, and the adhe-
sion and deposition of sludge cake on membrane surfaces (Liao et al., 2004). As bacteria
alone cannot foul the membrane seriously, the far more important foulants are the EPS se-
creted from them (Xu and Chellam, 2005). Due to the accumulation of EPS and their reaction
with solute ions, mass transfer coefficient in the water phase can be reduced and this contrib-
utes significantly to permeate flux decline in membrane filtration (Kim et al., 2006). More-
over, the secreted EPS decrease the effectiveness of backwashing and lead to irreversible
fouling (Xu and Chellam, 2005). The EPS gel structure protects also bacterial cells from hy-
draulic shearing and from chemical attack of biocides such as chlorine (Li et al., 2008). Based
on the characteristics of the active substances in biofouling, the classification of biofouling is
overlapping within organic fouling. Biofouling can then be considered as a biotic form of or-
ganic fouling (Amy, 2008). The severity of biofouling in wastewater reclamation is greatly
related to the characteristics of the feedwater, such as nutrient availability, turbulence, tem-
perature, particles, etc. The structure of microbial community and the surface condition of the
membrane surface influence the process also to some degree (Melo and Bott, 1997; Ahmed et
al., 2007; Huang et al., 2008b).

2.4.3 Inorganic Fouling

Inorganic fouling is caused by the accumulation of inorganic precipitates (such as metal hy-
droxides and carbonates) and scales on membrane surfaces or within the pore structure (Li et
al., 2008). Precipitates are formed when the concentration of these chemical species exceed
their saturation concentrations, which is a major concern for RO and NF as they reject most of
the solved inorganic species. For UF and MF, inorganic fouling due to concentration polariza-
tion is much less profound, but can exist most likely due to interactions between ions and
other fouling materials, e.g. organic foulants via chemical bonding (Liang et al., 2008; Costa
et al., 2006). If pre-treatment processes for membrane filtration such as coagulation or oxida-
tion are not designed or operated properly, it may introduce metal hydroxides into fouling
matrix (Zhao et al., 2005). Inorganic fouling/scaling can be a significant problem for make-up
water of caustic solutions prepared for chemically enhanced backwash or chemical cleaning
(Liu et al., 2006).
Ultrafiltration Process and Fouling Phenomenon 17

2.4.4 Particulate and Colloidal Fouling

In UF of water or wastewater, dissolved and colloidal organic matters lead to the most severe
fouling (Howe et al., 2006; Jarusutthirak and Amy, 2006). Fan (2008) showed that the colloi-
dal substances in the SE are major foulants to UF as the fouling potential of the water could
vary only a little after a pre-filter with a pore size of 1.5 µm prior to UF. Howe (2006) dem-
onstrated that the size of major foulants to UF is within the range of MW of 100 k Da to 1 µm
in diameter. In a pilot-scale study, Poele (2006) compared the fouling control effect using a
curved sieve with a mesh size of 0.45 µm and a MF with a nominal size of 0.1 µm prior to
UF. The results indicated that the major foulants in the SE to UF are within the size range of
0.1 to 0.45 µm. Particles are considered to have little influence on the fouling phenomenon in
UF as flux decline due to their accumulation is largely reversible by hydraulic cleaning proc-
esses (Liu et al., 2006). In conjunction to this statement, flocculation/coagulation processes
can be used to improve the filterability of water sample by enlarging the size of the foulants.
Several studies showed that after flocculation the water samples can be easier filtered, the
formed fouling layer can be more readily backwashed (Fan et al., 2008; Cho et al., 2006).
This leads to a better performance of the UF (Huang et al., 2009).

Although it is indicated that inorganic particles and colloids do not really foul the membrane
because of the better reversibility, under certain operational conditions the inorganic colloids
can also lead to irreversible fouling due to deposits and fouling layer compression (Bessiere et
al., 2005; Li et al., 2008). In drinking water treatment or wastewater reclamation, the colloidal
inorganic foulants could be irreversible adhered on the membrane surface through organic
foulants and lead to a complex irreversible fouling layer (Jermann et al., 2008; Zhang et al.,
2003). Zhang (2003) showed the effect of colloidal and particulate SiO2 (d = 0.5 to 10 µm)
and powdered active carbon (PAC, d = 1 to 80 µm) on the TMP development of UF filtering
NOM-containing water. It was illustrated that the dosage of both kinds of fine particles led to
more severe fouling. The combined particle-NOM layer at the membrane surface could not be
effectively back washed. Simultaneously, it was observed that although PAC is able to adsorb
much more NOM than SiO2 does in the feedwater, it reduces the reversibility of the fouling
more severely than the latter one at a similar dosage. After systematic investigation on the co-
effect of organic and inorganic colloidal foulants in dead-end UF, Jermann (2008) stated that
the combined fouling effect is determined by the characteristics of NOM and its interaction
with the particles in the feed water.

2.5 Complexity of Fouling Phenomena

2.5.1 Influence of Membrane Property

The previous studies indicate that in fouling phenomena organic foulants play a crucial role in
membrane filtration. In reality, although the physical and chemical interactions between or-
ganic foulants and membrane and among various types of foulants add the complexity of foul-
ing, the severity of fouling is primarily determined by the interaction between foulants and the
membrane (Yamamura et al., 2007a). Therefore, besides the characterization of foulants, the
18 Ultrafiltration Process and Fouling Phenomenon

property of membranes needs also to be identified (Huang et al., 2009). Based on the review
made by Wyart et al. (2008), the property of membranes can be described by two classes of
parameters which are membrane structure parameters (roughness, porosity, pore size, pore
shape, pore size distribution) and membrane/effluent coupling parameters (membrane mate-
rial, surface charge, hydrophobicity, etc.).

Influence of Membrane Structure

In respect to membrane structure, one important parameter relevant to fouling is the rough-
ness which is used to describe the surface topology of a membrane. Generally, rough surfaces
can be easily fouled because roughness increases surface area and affects macroscopic proper-
ties like zetal potential that can also influence fouling rates (Wong et al., 2009). This correla-
tion is always reproduciable in either MF/UF processes (Zhang et al., 2008; Evans et al.,
2008) or in RO and NF membrane filtration (Li et al., 2007; Peng et al., 2004), during memb-
rane filtration of either wastewater effluent, surface water or model substances (Lee et al.,
2005; Vrijenhoek et al., 2000).

For membranes made of identical material, pore size and its distribution are important pa-
rameters affecting membrane performance. It is suggested that a suitable pore size and a nar-
row pore size distribution are preferred to control membrane fouling (Kimura et al., 2006;
Meng et al., 2009). For instance, in filtering SE, Huang (2007) observed that the membrane
with a larger pore size (0.1 µm) is more prone to irreversible fouling than that with a smaller
pore size (0.02 µm) although they are both made of PVDF. This phenomenon could be attrib-
uted to the severe adsorption and pore blocking in the membrane with a larger pore size as
colloids penetrate more readily into the pores and interact with the membrane matrix (Costa et
al., 2006). Similar conclusion has been made by Fu et al. (2008). In their study it was simulta-
neously revealed that the membrane with a lower porosity, which indicates the ratio between
pore volume and that of the material, demonstrates much severe fouling. This phenomenon is
attributed to the fact that at similar delivered foulant load, the membrane with a higher poros-
ity can offer more spare void for the solutes.

Influence of Hydrophobicity and Charge

In relevance to the chemical characteristics of the membrane material, membrane surfaces
show hydrophilic or hydrophobic properties which reflect the interfacial tension between wa-
ter and the membrane material. To estimate the hydrophobicity of a membrane, contact angle
can be measured using the sessile drop method (Li et al., 2008). A membrane with a higher
contact angle indicates that it is of higher hydrophobicity. Hydrophobic membranes are re-
garded to be more prone to fouling than hydrophilic ones. That is because the constituents in
water that are sparingly soluble, nonpolar, or hydrophobic will accumulate readily at the hy-
drophobic surface to minimize the interfacial tension between the water and membrane (Li et
al., 2008). At this point, numerous studies have highlighted the correlation, either by filtering
surface water and wastewater effluent (e.g., Huang et al., 2007; Kim and Dempsey, 2008) or
using model substances (e.g., Susanto et al., 2007). The hydrophobicity of components in
water on fouling has also been intensively investigated (Xiao et al., 2009; Zularisam et al.,
2007a; Zularisam et al., 2007b). XAD-8 and XAD-4 resins have been used to separate and
isolate the hydrophobic and hydrophilic portions (Kim and Dempsey, 2008). Kim and
Dempsey (2008) claimed that hydrophobic/hydrophilic acid in EfOM attributes mostly to the
Ultrafiltration Process and Fouling Phenomenon 19

fouling of a PES UF membrane (more hydrophobic). Jarusutthirak et al. (2002) observed a

severe gel layer fouling caused by hydrophobic and transphilic components in EfOM by UF
using a polyamide (PA) membrane (more hydrophilic). But in a later study (Jarusutthirak and
Amy, 2006), major foulants in EfOM to a PA UF membrane were considered to be hydro-
philic macromolecules. In MF (hydrophobic PVDF) filtering wastewater effluent, Liang and
Song (2007) reported that the most important foulants in SMP are hydrophobic humic sub-
stances and hydrophilic neutrals.

Contrasting experimental results indicate that the interactions between membrane and foulants
are far more complicated than that can be interpreted only based on hydrophobic interacions.
For instance, the fouling phenomenon can also be impacted by electrostatic conditions of
NOM/EfOM foulants and membranes (Liang et al., 2008). Similar to hydrophobicity, electro-
static charge is also a membrane/effluent coupling parameter impacting fouling. As
NOM/EfOM in water is commonly negatively charged at neutral pH (Li et al., 2008; Schrader
et al., 2005), membranes are negatively charged or modified to be negatively charged to gen-
erate repulsive forces against organic fouling. Knowledge of the electrokinetic properties of
membranes can be obtained by zeta potential (ζ) measurement (Li et al., 2008). Numerous
studies have investigated the effect of electrostatic charge on fouling by changing the pH,
ionic strength of the solutions (e.g., Bouzid et al., 2008). Generally, it is believed that solu-
tions with a high ionic strength or in a certain range of pH reduces the electrostatic double
layer thickness of both membrane surface and foulants and leads to more severe fouling (Ahn
et al., 2008; Narong and James, 2006).

2.5.2 Solution Chemistry

The interacton between foulants and membrane is also influenced by solution chemistry, e.g.,
solution pH, ionic strength as reviewed in the previous section. In addition to the general ionic
strength effect, the divalent cations, especially calcium, have also been considered to be able
to enhance NOM/EfOM fouling due to bridging effects (Herzberg et al., 2009; Yamamura et
al., 2007b). For calcium ions, they alone play little effect in fouling during low-pressure
membrane filtration, but are shown to act as bridging material between carboxylic active
groups present in many NOM/EfOM and the negatively charged functional groups present at
the membrane surface. Additionally, they can promote the aggregation of organic compounds
in solution and lead to more severe pore blocking (Ahn et al., 2008). Using AFM the adhesion
force of model organic compounds, e.g., proteins (BSA, Ang and Elimelech, 2007), polysac-
charides (Alginate) and humic substances (Lee and Elimelech, 2006) on RO membranes at
different feed solution chemistry were compared. The results showed that Ca2+ concentration
influence the interaction between foulants and the membrane proportionally. The correlation
has been confirmed in RO filtering EPS-rich wastewater effluent (Herzberg et al., 2009).
Nevertheless, in different low-pressure membrane filtration processes calcium ions impact
fouling development in different manners. In contrast to a significant correlation between cal-
cium concentration and fouling development in a MBR system (Liang et al., 2008), no obvi-
ous effect can be observed in MF/UF filtering surface water (Yamamura et al., 2007b). In UF
of wastewater effluent, Hamberkamp (2008) observed a slight reduction of irreversible foul-
ing when the Ca2+ concentration increased from 2 to over 80 mg/L. It seems that in low-
pressure membrane filtration, within its typical concentration range, calcium does not to be an
20 Ultrafiltration Process and Fouling Phenomenon

influential factor in NOM/EfOM fouling as it performs in high-pressure membrane filtration

(Amy, 2008).

2.5.3 Co-effect of different Organic Foulants

When various foulant types are present in wastewater effluents, the possibility of one foulant
type binding to another will change the physical (size and molecular weight) and chemical
(charge and hydrophobicity) characteristics of the combined foulant type (Ang and Elimelech,
2008). Jermann et al. (2007) investigated the interaction effect between humic acid and poly-
saccharide (alginate) in dead-end UF. The results showed that the addition of humic acid into
polysaccharide solution led to more severe irreversible fouling due to the adsorption of humic
acid onto the membrane. Based on the finding that alginate (polysaccharide) can cause more
severe fouling compared to other substances in MF (Negaresh et al., 2006), Nagaresh (2007)
investigated the co-effect of polysaccharides (alginate) and proteins (bovine serum albumin –
BSA) and showed that the addition of protein results in different fouling behavior as filtering
polysaccharide samples alone.

This short review indicates that the fouling phenomena in UF membrane filtration for waste-
water reclamation are highly complicated. The interaction between foulants and membrane
depends on the characteristics of the foulants, membrane and solution chemistry. This sug-
gests that for the investigation of fouling phenomena, the experimental condition (e.g. mem-
brane material, source water) in either lab-scale or pilot-scale tests should be comparable. On
the other side, it has also been illustrated that organic foulants, in EfOM-containing water
referred to SMPs, which are mainly characterized as polysaccharides and proteins, play a cen-
tral role in fouling phenomenon.

2.6 Quantification of Fouling Potential

The content of proteins and polysaccharides can be quantified using different photometric
methods (Raunkjaer et al., 1994). Some methods have been widely used in MBR systems to
indicate the fouling potential of treated wastewater (e.g. Rosenberger et al., 2005). A good
correlation between the content of polysaccharides and/or proteins and fouling development
has also been reported by several studies (Al-Halbouni et al., 2008; Rosenberger et al., 2006).
Based on Lowry method (Lowry et al., 1951) and Dubois method (Dubois et al., 1956), an
automatic online measurement system has also been developed (Mehrez et al., 2007).

In contrast to that in a MBR system, little relationship between the content of the photometri-
cally determined SMP foulants and fouling has been found in membrane filtration of SE (e.g.
Poele et al., 2004). Poele et al. (2004) compared the filterability of pre-filtered SE and the
content of proteins and polysaccharides. It is evident that the improvement of filterability after
size fraction is mainly attributed to the the removal of macromolecules within the size range
of 0.1 to 0.45 µm. However, the photometrically quantified SMP cannot reflect a proportional
variation of the content of these organic compounds after size fraction. To test the usability of
photometrically quantified proteins and polysaccharides in presenting the fouling potential of
wastewater effluent, further investigations need to be performed.
Ultrafiltration Process and Fouling Phenomenon 21

The content of colloidal foulants can be reflected by the modified fouling index (MFI) which
is based on the filtration curve filtering colloid-containing water using a membrane with a
pore size of 0.45 µm under constant pressure (Schippers and Verdouw, 1980, Eq. 2-8). The
MFI is the slope of the linear regression between t/V plotted against V (t: filtration time, V:
permeate volume) during cake/gel filtration. Originally, MFI is used to assess the colloidal
fouling in RO systems. A similar method was used to quantify the content of high molecular
polymers in surface water lower than the µg/L range (Jekel and Boeckle, 1982). Compared to
the traditional silt density index (SDI, also known as fouling index, Dupont de Nemours and
Co. 1977) which can be poorly interpreted theoretically, this method is based on well defined
cake/gel filtration theory (Schippers and Verdouw, 1980).

t
d( ) η ⋅ Rf
MFI = V = Eq. 2-8
d (V ) 2 ⋅ ∆P ⋅ A 2

Where
MFI = membrane fouling index
t = filtration time, s
V = cumulative permeate volume, m3
Rf = resistance of the fouling, 1/m
A = area of the membrane, m2
P = trans-membrane pressure, N/m2

Adapted to UF membrane filtration, MFI is altered as MFI-UF and specific ultrafiltration re-
sistance (SUR) to fit the pore size of the correspondingly used UF membranes (Boerlage et
al., 2003; Roorda, 2004). Actually, these indicators present the hydraulic fouling resistance
plotted vs. permeate volume under different operational conditions (Roorda, 2004). Using
SUR as a parameter, it was identified that the major foulants in wastewater effluents are
within the size range of 0.1 to 0.2 µm. The water with a higher SUR value, which indicates a
lower filterability, is related to a more severe irreversible fouling problem (Roorda and van
der Graaf, 2005). Recently, a unified membrane fouling index (UMFI) has been developed to
assess the fouling potential of NOM-containing water in MF and UF processes operated at
constant flux (Huang et al., 2008a). Using this indictor, Amy (2008) showed a significant
positive correlation between the UMFI value and irreversible fouling. These results indicate
that under certain conditions the increase of major foulant content can be reflected by fouling
indicators.

Simultaneously, it should be noted that the nature of foulants play also an important role in
determining the interaction between foulants and the membrane (Amy, 2008; Haberkamp,
2008). When organic matter is the primary cause of membrane fouling, the indicators are
lacking in information about their properties. An understanding of the characteristics of major
organic foulants and their quantity in feedwater is otherwise required for fouling control and
selection of cleaning strategies. Therefore, a simultaneous characterization of these colloidal
organic foulants is necessarily required. This calls for a comprehensive parameter reflecting
size distribution, characteristics and quantity of organic foulants simultaneously.
22 Ultrafiltration Process and Fouling Phenomenon

Liquid chromatography with high sensitive and online multi-detectors, e.g. online organic
carbon detection, UV254 absorption detection and organic nitrogen detection (LC-OCD-UVD-
OND), can be applied to identify major organic foulants in treated domestic wastewater con-
sidering molecular size, aromatic characteristics and magnitude of organic carbon and nitro-
gen simultaneously (Huber, 1998). It has been reported that the fraction of large dissolved
organic molecules defined as biopolymers contribute more significantly to permeability re-
duction of a UF membrane than other DOC fractions in filtering wastewater effluent (Laabs et
al., 2006). They may be organic colloids, e.g. polysaccharides, proteins (Laabs, 2004; Her et
al., 2003). A significant correlationship between biopolymer content and hydraulic fouling
resistance in UF of treated wastewater was presented by Haberkamp (2008). In respect to ir-
reversible fouling, it is stated that the content of protein-like substances may play a more im-
portant role than other compounds. As the fouling potential of organic foulants is not univer-
sal but membrane-specific (Huang et al., 2007), further investigations are needed to get more
insight into fouling phenomena in UF of wastewater effluents.

2.7 Fouling Control

2.7.1 Fouling Control by Pre-treatment Processes

Along with the fouling problem in membrane filtration for wastewater reclamation, different
pre-treatment processes have been investigated to remove foulants and improve the perform-
ance of the downstream membrane processes (Li et al., 2008). The pre-treatment processes
can briefly be classified into coagulation / flocculation (pre-coagulation / sedimentation, in-
line coagulation, pre-flocculation), adsorption, oxidation, ion exchange, pre-filtration (rapid
sand filtration, sieve filtration), bio-filtration and their combinations. An overview of these
pre-treatment processes in respect to foulants removal prior to UF is shown in Tab. 2-2.

As simple and cost effective measure, bio-filtration demonstrates the potential to remove fou-
lants prior to membrane filtration (Huck and Sozanski, 2008; Hallé et al., 2009). Natural bio-
filtration processes such as river bank filtration (RBF) and soil aquifer treatment (SAT) have
been verified as effective processes reducing foulant content in SEs and improving the per-
formance of membrane filtration (Sperlich et al., 2008; Jarusutthirak, 2002). To reduce area
requirements bio-filtration can be operated at higher filtration rates. However, although bio-
logical (rapid) sand filtration (BSF, filtration rate > 5 m/h) can improve the performance of
MF in wastewater reclamation (Xie et al., 2008), it showed little effect on the improvement of
the performance of UF (Roorda, 2004; Poele, 2006). The foulant removal effect has been de-
termined to be influenced by the filtration rate. A suitable filtration rate range needs to be
identified to operate bio-filtration as fouling control process prior to UF.
Ultrafiltration Process and Fouling Phenomenon 23

Tab. 2-2 Overview of fouling control pre-treatment processes prior to membrane filtration for wastewater reclamation

Technical process Process description Possible fouling control mechanisms Shortages of the process in practice references
Haberkamp et
Coagulation Addition of chemicals Increase size of particles and colloids, Coagulate residual in membrane matrix al., 2007
/flocculation (coagulants) into feed promote cake layer formation and reduce Increase of solid loading to membrane Bagga et al.,
water / sedimentation irreversible fouling Requirement of sludge disposal 2008

Wang et al.,
Advanced oxidation Dosing oxidants into feed Break up large organic molecules into Requirement of anti-oxidation membrane 2007
water smaller ones Possible generation of disinfection by- Huang et al.,
products 2008c
Haberkamp et
Adsorption Feed water flow through Adsorb organics onto adsorbents Reduction of foulants removal efficiency al., 2007
adsorbent filter or dosing due to competitive adsorption Fan et al., 2008
adsorbents into feed wa- Disposal or regeneration of adsorbents
ter
Fan et al., 2008
Ion exchange Feed water flow through Remove low and medium MW fraction Low removal rate of large molecules Zhang et al.,
ion exchange bed molecules Sensitive to water quality 2006b
Regeneration and disposal of resins
Roorda, 2004
Pre-filtration Filtering feed water using Remove suspended and large colloids, Little effect on irreversible fouling control Poele, 2006
sand filter or sieve-filter protect membrane module from clogging Fan et al., 2008
Huck and So-
Bio-filtration Filtering feed water using Remove particles, colloids and dissolved Requirement of filtration area zanski, 2008
bio-filter substances through straining, adsorption van der Hoek et
and biological degradation al., 2000
Jarusutthirak,
2002
24 Ultrafiltration Process and Fouling Phenomenon

Compared to bank-filtration processes, slow sand filtration is an engineered bio-filtration unit

with higher filtration rate and shorter residence time (Tab. 2-3). In advanced wastewater
treatment it can be used as a tertiary filtration section for removing suspended solids or tur-
bidity, total microorganisms, biological oxygen demand (BOD), chemical oxygen demand
(COD) and ammonia nitrogen (Ellis, 1987). Although slow sand filtration decreases the col-
loidal fouling potential in membrane filtration (Downing et al., 2002; van der Hoek et al.,
2000), little is known about its effect on organic foulants removal and subsequent influence
on the performance of UF. To eliminate this knowledge gap is one of the major objectives of
the present research.

Tab. 2-3 Design/Operational Parameters/Conditions for biological (rapid) sand filtration (BSFn), slow
sand filtration (SSFn), river bank filtration (RBFn) and SAT (Adapted from Amy et al., 2006)

Condition/Process BSFn SSFn RBFn SAT

Travel Distance ≤2 m ≈1 m ≥10 m ≥30 m
Travel Time ≤10 min ≥2 hours ≥1 day ≥ 2 weeks
Filtration Rate ≥ 5 m/h ≤ 0.5m/h ≤ 1m/day ≤ 1m/day
Schmutzdecke* no yes yes yes
Fouling Control poor good very good excellent
Effect to UF (Roorda, 2004 (Zheng et al., (Sperlich et al., (Jarusutthirak,
Poele, 2006) 2009b) 2008) 2002)
Flow Regime saturated saturated un/saturated un/saturated
Redox oxic (aerobic) oxic (aerobic) an/oxic (aerobic) an/oxic (aerobic)
Schmutzdecke*: the so-called filter skin (dirt layer in German) on the sand bed which consists of
sediment and adjusted organic and inorganic particles, a biofilm made of different microorganisms,
protozoa and mesozoa (Ellis, 1985)

As a bio-filtration process, the performance of slow sand filtration is affected by operational

conditions. A study in advanced wastewater treatment using slow sand filtration (Ellis, 1987)
showed that filtration rate influences contaminant removal and the length of operational run
time for sand scraping. Investigations on slow sand filtration in drinking water treatment show
that temperature, source water characteristics, filtration rate and the extent of surface ripening
have the most significant effect on the performance in terms of water quality control (Amy et
al., 2006). So far, the influence of operational conditions on the removal of organic foulants
by slow sand filtration has little been investigated.

2.7.2 Influence of Operational Conditions on Fouling

Impact of Operational Conditions on Fouling Development
Besides the quality and quantity of foulants, properties of membranes and chemical circum-
stances, fouling extent in membrane filtration is also influenced by operational conditions
(Huang et al., 2007; Wang et al., 2008; Amy and Cho, 1999). The interactions between fou-
lants and membrane and among foulants in the fouling layer can be significantly changed in
terms of resistance and reversibility under different operating conditions (Le-Clech et al.,
2006). For instance, flux affects the development of fouling: ‘there exists a flux below which
a decline of flux with time does not occur; above this flux, fouling is observed. This flux is
Ultrafiltration Process and Fouling Phenomenon 25

termed as the critical flux’ (Field et al., 1995). Although in long-term operation fouling is
inevitable such as by filtering treated wastewater (Pollice et al., 2005), fouling development
can be mitigated by means of operating membrane filtration at a sustainable flux which is
defined as ‘the flux for which the TMP increases gradually at an acceptable rate, such that
chemical cleaning is not necessary’ (Ng et al., 2005). Choosing suitable operational strategies
is helpful to control the development of fouling and maintain membrane processes operated in
an economically acceptable level.

In dead-end UF membranes operated at constant flux, operational conditions such as flux and
duration of filtration (backwash interval: BWI) and relaxation time are the main parameters
investigated. Decarolis (2001) showed that BWI in dead-end UF of tertiary effluent could be
set between 15 to 30 minutes. The extension of BWI led to higher foulant load on the mem-
brane and more severe fouling. Based on the water quality of SE, the filtration flux was sug-
gested to be maintained at 50 to 60 LMH (Roorda, 2004; Poele, 2006). Higher flux results in a
higher foulant load on the membrane and leads to a more compact fouling layer (due to higher
pressure applied) which is more difficult to be removed (Kim and DiGiano, 2006; Decarolis
et al., 2001).

The removal of reversible fouling can be improved by membrane relaxation which can be
achieved by withdrawing pressure during filtration. Membrane relaxation is also called as
non-continuous operation of membrane (Le-Clech et al., 2006). Under relaxation, back trans-
port of foulants is naturally enhanced as non-irreversibly attached foulants can diffuse away
from the membrane surface through the concentration gradient. During UF of treated waste-
water the effect of relaxation was affected by the magnitude of solids and solvable fractions
accumulated at the membrane surface and the applied pressure during filtration (Smith et al.,
2006). Although combined with cross-flow and/or backwashing relaxation can mitigate foul-
ing phenomena, this type of operation is considered as economically unfeasible under certain
conditions (Le-Clech et al., 2006).

Characterization of Fouling Layer

The structure of a fouling layer is considered to play a significant role in the stability of the
fouling (Wang and Waite, 2008b). To characterize the fouling layers under pressure, the spe-
cific fouling resistance and the compressibility of the layer can be investigated based on cake
filtration theory. It is assumed that the macrosolutes rejected by the membrane form a solid
cake/gel layer, which causes fouling resistance to increases in proportion to the magnitude of
the delivered foulants to the membrane surface. According to this definition fouling layer re-
sistance can be calculated using Eq.2-9:

Rc = α ⋅ mc = α ⋅ cs ⋅ V Eq.2-9

where
Rc = resistance of cake/gel layer, 1/m
α = specific cake/gel resistance (fouling resistance per unit dry weight of foulants), m/kg
mc = dry weight of foulants per unit membrane area, kg/m2
cs = foulant concentration, kg/m3
V = permeate volume, m3
26 Ultrafiltration Process and Fouling Phenomenon

Tab. 2-4 Specific fouling resistance and compressibility of fouling layer in low-pressure membrane filtering EPS, activated sludge or NOM content waters

reference Membrane type Feed water and operating conditions Specific fouling resistance Compressibility
α (m/kg) s
15
Torrestiana- Polysulphone UF Xanthan solution 11-63 x 10 -
Sanchez et al., 2007 MWCO 500 k Da Constant TMP 3.45 bar (Variation due to concentration
of KCl and isopropanol)
Ye et al., 2005 Hydrophilic PVDF MF Sodium alginate solution
Pore size 0.22 µm Constant TMP 0.8 bar 5.6-8.9 x 1015 0.28
Constant flux 50 LMH 5.9-16 x 1015 0.41

Ho and Zydney, Four kinds of MF membranes Bovine serum albumin (BSA) solution
2001 Pore size 0.16-0.2 µm Constant pressure 0.02 - 0.7 bar 0.2-190 x 1015 0.97
Furukawa et al., Two types of Polysulphone UF MBR treating soy sauce lees 1.53x 1013 – 1.77x 1014
2008 MWCO 50 and 750 k Da Constant TMP 0.5 – 2 bar (Variation due to pressure and 1.81
membrane type)
Lodge et al., 2004 Hydrophilic PES UF Biologically treated wastewater 3.95-9.3 x 1015 0.68
MWCO 150-200 k Da TMP 0.125-1 bar (Depends on TMP)
Chellam et al., 1998 Hydrophobic polypropylene MF MBR treating domestic wastewater 0.46-8.22x 1014 0.65
Pore size 0.2 µm
Mousa, 2007 Six types of UF Humic acid solution -
14
MWCO from 10 to 150 k Da Constant TMP 0.35 bar 1.3-4.7 x 10
Costa et al., 2006 Cellulose acetate UF Humic acid solution -
15
MWCO 10 k Da Constant 1.4 bar 3.7 x 10
Ultrafiltration Process and Fouling Phenomenon 27

Many foulants are highly compressible, higher pressure therefore results in an increasing den-
sity of the cake/gel layer and thereby increases α. The effect of compressibility can be esti-
mated assuming specific fouling resistance to increase as a power law function with pressure
(Eq.2-10).

α = α 0 ⋅ ∆P s Eq.2-10
where
α0 = specific cake/gel resistance at reference pressure, m/kg
P = TMP, N/m2
s = compressibility coefficient (s = 0: no compression, increasing value of s represents in
creasing compressibility)

The compressibility coefficient can then be calculated as following:

d (log α )
s= Eq.2-11
d (log ∆P )
The specific cake resistance and its compressibility depend on the character of the foulants.
Foulants in SE have a biological background (EPS or SMP) and may behave similar to the
fractions of microorganims (Roorda, 2004). Tab. 2-4 presents a short summarisation of calcu-
lated specific fouling resistance and compressibilities of different organic foulants in low-
pressure membrane filtration.

Membrane Cleaning
Membrane cleaning is an integral part of the operation of membrane filtration and has signifi-
cant impact on the performance of membrane systems. Based on the usage of chemical clean-
ing reagents, membrane cleaning can be classified into hydraulic cleaning (permeate back-
washing) and chemical cleaning.

Backwashing involves reversing the flow through the membrane using the product water
(permeate) to remove the foulants accumulated on the membrane surface and clogging the
membrane pores. Under certain conditions, air can also be injected in the backwashing per-
meate in MF to enhance backwashing effect (Li et al., 2008). Normally, backwashing is con-
sidered to be effective in removing particulate and colloidal foulants (Liu et al., 2006). The
effect of backwashing is significantly correlated with the amount of solids and solvable frac-
tions deposited on the membrane surface (Wu et al., 2008).

Investigations on the effect of backwash strength and duration on the restoration of membrane
flux were conducted by several studies (Huang et al., 2007; Kim and DiGiano, 2006;
Kennedy et al., 1998). It is stated that either the extension of backwashing time or the en-
hancement of backwashing flux can improve fouling control effect. The efficiency of back-
washing is shown to be more dependent on duration than flux. However, due to the complex-
ity of fouling and in situ operational conditions, determination of backwashing conditions
needs considering fouling control effect and permeate recovery rate (Pr, Eq. 2-12, used for
dead-end filtration processes). They are generally obtained by trial and error procedure based
on suggestions of the membrane manufactures.
28 Ultrafiltration Process and Fouling Phenomenon

V − Vbw
Pr = × 100% Eq.2-12
V
Pr = permeate recovery, %
V = permeate volume, m3
Vbw = permeate volume used for backwash, m3

As irreversible fouling is inevitable, after some operating time the membrane must be chemi-
cally cleaned to restore its permeability. In practice, chemical cleaning can be conducted ei-
ther in place/in-line (CIP: the membrane module is cleaned without removing it from the in-
stallation) or out of place/off-line (COP: the membrane module is cleaned from the system
and soaked in chemicals). A generally used CIP measure is chemically enhanced backwash
(CEBW). Chemicals can be added into backwash water (with soaking 3 – 5 minutes,
(Teodosiu et al., 1999)/ or recirculation for 10 – 15 minutes, Li et al., 2008) to im-
prove/enhance foulant removal effect through backwashing. In practice, different chemicals
can be used as backwashing reagents based on different foulant removal mechanisms (Tab.
2-5). Considering the components in feed water, chlorine (Decarolis et al., 2001; Roorda,
2004), sodium hydroxide or critic acid (Teodosiu et al., 1999; Panglisch et al., 2007) can be
chosen. The major removal mechanisms are listed in Tab. 2-5. Recently, demineralised water
has also been tested as backwash water in UF filtering NOM-containing water (Li et al.,
2009). TMP development can be effectively controlled using demineralised water for back-
washing compared to using permeate. This is attributed to the reduction of charge-screening
effect around the negatively charged membrane and NOM due to the absence of cations. The
calcium-bridging effect between membrane and NOM can also be reduced, simultaneously.

Tab. 2-5 Major categories of membrane cleaning chemicals (Liu et al., 2006; Poele, 2006; Tian et al.,
2009)

Category Major functions Typical chemicals Removed foulants

Caustic Hydrolysis, solubilization, NaOH Polysaccharide, protein,
solutions Increase negative charges of fats, oils, humic sub-
functional groups of membrane stances
Oxidants Oxidation, disinfection NaClO, H2O2, Humic substances
Generate more oxygen contain- peroxyacetic acid and other organic foulants
ing function groups to improve
hydrophilicity
Acids, chelat- Solubilization, chelation Citric acid, nitric Scales, metal oxides and
ing agents acid and EDTA also organic foulants
Surfactants Emulsifying, dispersion, sur- Surfactants and Fat, oil, protein and
face conditioning detergents biofilms

Enzymes Hydrolyse specific functional Protease or Organic foulants (protein,

groups, cut long chains to sam- carbohydrases polysacchride, etc.)
ller parts

In selecting chemical reagents used for membrane cleaning, different reagents have been in-
vestigated. Poele (2006) tested the cleaning effect of alkaline, acid, enzyme protoase and
amylase on a fouled UF membrane after filtering SE. It is reported that using the combination
of alkaline and enzyme protease the permeability of a fouled membrane can be largely re-
stored. In contrast, the use of enzyme amylase after alkaline cleaning results in little further
 29

improvement. The results suggested that a combination of alkaline and suitable enzyme is an
effective method to remove the foulants. A consecutive use of NaOH (1%) and ethanol has
been developed by Tian et al. (2009) in removing NOM foulants in UF of surface water. For a
fouled membrane, the use of ethanol after alkaline cleaning can improve the cleaning effect
significantly. Additionally, ethanol shows positive effect in reducing the hydrophobicity of
the fouled polyvinyl chloride (PVC) membrane. Although the cleaning strategies of fouled
low-pressure membranes have been also investigated in other several studies (Yamamura et
al., 2007a; Kimura et al., 2004), the utilization of cleaning agents are mostly based on the rec-
ommendations from membrane manufactures due to the limited understanding of fouling
mechanisms. Liu et al. (2006) made an attempt to explain the interactions between foulants
and the membrane, among different foulants and between foulants and cleaning chemicals.
Their interpretation is based on the electrostatic and hydrophobic/hydrophilic interactions
theory. A summarization of the effects of different cleaning reagents and the removed fou-
lants is shown in detail in Tab. 2-5. These advances may help membrane operators to find out
suitable cleaning reagents.
30 Material and Analysis

3 Material and Analysis

3.1 Lab and Pilot-scale UF Membrane Filtration Test

Membranes used in Lab and Pilot-scale UF Experiments
In the present work, lab and pilot-scale membrane filtration tests have been carried out to
study the fouling potential of different wastewater effluents. To make the results more compa-
rable, UF membranes with similar properties in lab and pilot-scale experiments have been
applied. Important characteristics of the membranes are shown in Tab. 3-1.

Tab. 3-1 Properties of applied UF membranes in lab and pilot-scale membrane filtration tests

UF membrane property Used in lab-scale test Used in pilot-scale plant

Manufacturer Microdyn-Nadir GmbH Inge AG

Material Hydrohilized PES Hydrophilized PES
MWCO (Da) / pore size (µm) 150 k Daa / 0.026 µmb 100 k Daa / 0.02 µmb
Average permeability 650 530
2
(L/(m ·h·bar)) @20ºC
Zeta Potential (mV) -42c -20a
Contact angle 53 ± 2.5c n.m. (total wetting)a

Membrane configuration flat membrane (UP150) multi-bore hollow fibre membrane

(for module dizzer® 450)
Operational conditions Constant pressure Constant flux
Dead-end Dead-end
a: offered by the membrane manufacturers
b: calculated based on the relationship between the hydrodynamic diameter of a molecule (dextran)
and its MWCO (Crittenden et al., 2006)
c: measured by Haberkamp (2008)
n.m.: not measured

Lab-scale UF Tests
A filtration cell with a volume of 200 mL (Amicon 8200, Millipore, U.S.A) and UP150 mem-
brane (Microdyn-Nadir GmbH, Germany) with an effective filtration area of 2.87·10-3 m2
have been used to determine the fouling potential of different waters. As shown in Fig. 3-1,
due to an attached feed reservoir, up to 4 litres of water could be filtered in each experiment in
dead-end mode under a certain pressure offered by a nitrogen cylinder. The flux was meas-
ured using a balance (Sartorius-basic) connected to a computer. The weight of the collected
permeate is automatically stored by data loggers in a minute frequency.
Material and Analysis 31

Amicon cell 200 mL

V4
V1
magnetic membrane N2
stirrer
V2
V3

feed reservoir
4L

electronic scale

Fig. 3-1 Schematic illustration of the Amicon system (V1: nitrogen gas valve, V2: pressure reducer
valve, V3: relief valve of the feed reservoir, V4: relief valve of Amicon cell)

The fouling potential test includes the determination of the filterability of a water sample and
the reversibility of the correspondingly formed fouling. The filterability is determined by one
cycle filtration test which is standardized in the present work as the normalized flux decline
over 500 mL cumulative permeate volume using a clean membrane. The operation is carried
out under a trans-membrane pressure of 1 bar and at a stirring speed of 120 rpm. All experi-
ments are conducted at room temperature around 22°C. For each filtration test a new UP150
membrane is used.

Before commencing determination of the filterability, each membrane is soaked in pure water
(R >17.4 m ·cm, Elga) at least 24 hours and then filtrated with pure water (1m3 water / m2
membrane area) under 1 bar (water filled in the water reservoir is pressurized using nitrogen
gas, the pressure is measured using a pressure gauge and the variation of the pressure is con-
trolled within 5% using a valve) to remove wetting agents and production residues. Immedi-
ately before the stirred cell test the pure water flux of the membrane is determined by filtering
500 mL pure water. Only membranes within 20% pure water flux variation are chosen for the
experiment.

The reversibility of formed fouling in UF using Amicon cell system is defined as normalized
flux after several filtration cycles which has already been used in other studies (Jermann et al.,
2007; Jermann et al., 2008). The membrane is prepared in the same way as described for the
filterability test. Then the water sample is filtered in several filtration cycles. Each filtration
cycle included: filtration of 500 mL water sample using the prepared membrane, backwash
with 50 mL salt solution at one bar (CaCl2 2.8 m mol/L, NaCl 5.4 m mol/L in pure water,
conductivity around 1200 µS/cm), then filtration of 200 mL pure water at one bar to deter-
mine pure water flux. The ratio between the pure water flux at the last cycle and that of the
clean membrane is defined as flux recovery. Higher flux recovery indicates that the fouling
layer could be more easily removed by backwashing and represents higher fouling reversibil-
ity.
32 Material and Analysis

Pilot-scale UF Plant

(a)

(b) backwash water membrane module permeate

raw water tank

Coagulant
pre-filter pump for
200 µm P removal permeate
tank

raw water pump backwash pump

Fig. 3-2 (a) The photo of the W.E.T UF pilot plant system equipped with Inge Dizzer® 450 module
and (b) the schematic diagram of the UF system

An UF pilot plant (W.E.T, Germany) with a hydrophilized PES membrane (Dizzer® 450, Inge
AG, Germany) is used to filter different effluents (Fig. 3-2). The membrane module has a
filtration area of 4.5m2 and the MWCO of the membrane is around 100 kDa. The specific
permeability of the membrane at 20°C is about 530 L/(m2·h·bar) (tested with tap water). It is
operated in dead-end filtration mode. Based on the water quality of the feed water, the pilot
plant is suggested to be operated at 50 - 140 LMH with a filtration time from 10 – 60 minutes
and a backwash time of 15 – 60 seconds. The membrane is able to withstand pH range 1-13
and has a high tolerance to free chlorine (200 ppm) and hydrogen peroxide (500 ppm). The
arrangement of seven capillaries in one fiber results in high mechanical strength against fiber
breaking during intensive hydraulic backwash at a flux as high as 260 LMH (per Inge AG).
Material and Analysis 33

3.2 Pilot-scale Bio-filtration Systems

3.2.1 Short-term Bank-filtration in WWTP Beixiaohe-Beijing

Fig. 3-3 Illustration of short-term bank filtration in the artificial lake in WWTP Beixiaohe

The WWTP Beixiaohe is the first domestic wastewater treatment plant in Beijing, China.
With a treatment capacity of 70 000 m3/d it serves for more than 0.4 millions inhabitants. In
2008, it was decided to integrate a membrane bioreactor (MBR) system in the extension of the
plant. Based on the plans for the Olympic Green (Olympic Game 2008, Beijing) and the ex-
tension plan for the WWTP Beixiaohe, a wastewater reuse concept has been developed (Ernst
et al., 2007). Effluent of a MBR with in-situ P-precipitation would be treated to two different
qualities, depending on the reuse purpose. In the first treatment step, phosphate as an essential
nutrient is removed by adsorption onto granular ferric hydroxide (GFH) and the water is
pumped into an artificial lake for scenic impoundment. The lake water is then filtered using
lake-bank filtration for further removal of organic compounds as a fouling control step for a
downstream operated UF.

Tab. 3-2 Average water quality of the lake water in WWTP Beixiaohe in 2005

Parameter COD Turbidity TN NH4+-N TP pH

mg/L NTU mg/L mg/L mg/L
Average 11 0.6 10 0.8 < 0.05 7.9

The concept was realized in 2005. More information about the whole treatment process was
introduced in a previous study (Sperlich et al., 2008). For bank filtration, an artificial lake
(L×W×H: 5.5m×5.5m×2.5m) was constructed and was put into operation in 2005 (Fig. 3-3).
The applied filtration materials are sand (grain size 2 – 4 mm) and limestone (grain size 3 –
6 mm) which were set up in two separated containers to form embankments. At start-up, the
lake was filled with 95 % drinking water, supplemented by 5% treated waste water. In order
to facilitate the settlement of planktonic organisms, plankton samples were taken from four
34 Material and Analysis

lakes in Beijing and the lake was inoculated. On the surface of each bank, a layer of textile
was situated to ensure stability. The bank filtration was operated with a filtration passage in a
depth of 1.7 m, a filtration velocity of 1.55 m/d and a travelling time of about 1.1 to 1.5 days.
The lake water was the effluent from a MBR pilot plant (Toray, Japan) after flocculation and
GFH adsorption. The lake water quality during the experiment period is summarized in Tab.
3-2.

3.2.2 Slow Sand Filtration in WWTP Ruhleben-Berlin

WWTP Ruhleben
With a treatment capacity of 240 000 m3/d the WWTP Ruhleben is the largest sewage treat-
ment plant in Germany. Raw water, including 20 - 30% industrial wastewater is treated using
a conventional activated sludge treatment process with biological/chemical phosphorous re-
moval and biologic nitrification/denitrification (Laabs, 2004). Its annual average effluent
quality is summarized in Tab. 3-3.

Tab. 3-3 Wastewater effluent quality of the WWTP Ruhleben (Bahr et al., 2007)

Parameter COD BOD DOC TN NH4+-N NO-3 TP pH

mg/L mg/L mg/L mg/L mg/L mg/L mg/L
Average 40±7 3.8±0.5 11.4±0.9 10.3±2.1 0.9±0.6 6.9±1.7 0.28±0.1 7.4±0.2

Slow Sand Filtration

(b)
60 cm

Sand
d :1-2mm
72 cm

Gravel
layer

Fig. 3-4 (a) The photo of the slow sand filtration system in WWTP Ruhleben, (b) the schematic dia-
gram of a slow sand filter

Two pilot scale slow sand filters (SSFrs, Fig. 3-4) were operated from September 2006 to
December 2007 to treat SE after sieve filtration (pore size 100 m) from the WWTP Ruhle-
ben Berlin. The filters are operated in down flow mode with a filtration area of 2 m2 (SSFr Nr.
1) and 1 m2 (SSFr Nr. 2) respectively. Each filter is filled with silica sand (grain size 1 – 2
mm) for a layer with sickness of 0.7 m. Two gravel layers are placed below it as supporting
layer, each with a height of 0.1 m, consisting of gravel with grain size 4-8 mm and 8-12 mm,
respectively. The height of the supernatant is 0.6 m. On the outside wall of each filter there
are 6 probe sites vertically distributed, the distance of each probe site to the sand surface is 0,
0.1, 0.2, 0.4, 0.6 and 0.8 cm, respectively. In addition to the pilot scale filters, two lab-scale
Material and Analysis 35

filters (SSFr Nr. 3 and 4, each with a filtration area of 0.008 m2 at a filtration rate of 0.05 m/h
and 0.1 m/h, respectively) were run simultaneously from September 2006 to July 2007 to in-
vestigate the influence of the filtration rate on the performance of slow sand filtration. From
August 2007 till to March 2008 SSFr. Nr. 3 and 4 were operated with a protection layer con-
sisting of BioNet (Fig. 3-5 b) to improve the performance of slow sand filtration. The BioNet
carriers are porous compressible synthetic materials used in bioreactors. They offer large sur-
face area for adherence and growth of microorganisms and large void volumes for the reten-
tion of colloids and particles.

inlet (a)
(b)

54 cm

Bio Net
22 cm
(aeration)

sand
54 cm

gravel
19 cm
outlet

Fig. 3-5 (a) Diagram of a slow sand filter using BioNet as a protect layer (b) foto of BioNet carriers

3.3 Analysis

3.3.1 LC-OCD-UVD-OND
Biopolymer concentration in water samples is analyzed using liquid chromatography (LC,
manufacturer DOC-LABOR Dr. Huber, Karlsruhe, Germany) with a dissolved organic carbon
detector (OCD) using nondispersive infrared absorption (Ultramat 6, Siemens, Germany),
with a dissolved organic nitrogen detector (OND) using UV220 absorption (Well Chrom fixed
wavelength detector K2001- UV220 Knauer, Germany), with a UV254 detector (UVD, K-200-
UV254) and a size exclusion chromatography HW55S (GROM Analytik + HPLC GmbH,
Germany). The LC unit separates organic compounds according to their molecular size
through the size exclusion column. The underlying principle is the diffusion of molecules into
the resin pores. The larger molecules elute first as they cannot penetrate into the pores very
deeply, while the smaller molecules take more time to diffuse into the pores and out again.
The separated compounds are then detected by online detectors and corresponding content
36 Material and Analysis

could be converted into concentrations in mg C/L, mg N/L and SUV 1/m respectively. The
unit of the concentration in the present thesis is mg C/L (more detail information in appendix
8.1).

3.3.2 Fluorescence Exitation Emission Matrix

Additional identification of the presence of protein-like substances in biopolymer is con-
ducted by fluorescence measurement. Fluorescence of organic matter occurs when a loosely
held electron in an atom or a molecule is excited to a higher energy level by the absorption of
energy, and fluorescence occurs when energy is lost as light when the electron returns to its
original energy level (ground state). The wavelength at which absorption (excitation) and
emission occur is specific to the molecules. For protein-like substances, three fluorescent
amino acids (tryptrophan, tyrosine and phenylalanine) which contain aromatic ring structure
are specific indicators (Hudson et al., 2007).

The excitation emission matrix (EEM) spectroscopy is the state-of-the-art technique in fluo-
rescence measurement. The principal of EEM is that excitation, emission and fluorescence
intensity can be scanned over a range of wavelengths synchronously and plotted on a single
chart, developing a ‘map’ of optical space, an EEM. Recent advances in fluorescence spectro-
photometry permit the collection of fluorescence data from waters at high optical resolution
and the generation of excitation and emission data in the form of excitation-emission wave-
lengths.

To minimize the influence of the free amino acids and matrix effect in wastewater effluent
(Hudson et al., 2007), biopolymers used for the measurement of fluorescence EEM are frac-
tioned using SEC HW55S as following: a 2 mL water sample is injected into the size exclu-
sion column H55 and transported by eluent (consistent see A-Tab. 1). From elution time 20
minutes to the significant increase of UV254 signal (around elution time 50 to 55 minutes) the
fractioned water sample is collected. The sample is then analyzed using a JYHoriba/Spex
Fluoromax-2 fluorometer with a xenonlamp as the excitation source. EEM spectra are a col-
lection of a series of emission spectra over a range of excitation wavelengths. In this experi-
ment, EEM spectra are collected with subsequent scanning emission spectra from 240 to 500
nm at 2 nm increments by varying the excitation wavelength from 300 to 500 nm at 10 nm
increments. The presence of lower energy of the emitted photon than that of the excitation
energy is due to energy loss from the excited electron by collision, non-radiative decay and
other processes prior to emission. Spectral subtraction is performed to remove blank spectra
mainly caused by Raman scattering. The data is then treated by a program developed at the
Institute for Water Education (UNESCO-IHE, the Netherlands) based on MathLab®.

3.3.3 Photometric Protein and Polysaccharide Analysis

Measurement of Protein Concentration
Proteins are analysed according to the method developed by Rosenberger (2003) based on the
Lowry method (Lowry et al., 1951). Protein concentration is expressed in mg/L using bovine
serum albumin (BSA) as a standard. According to the calibration result (within the range of 0
to 30 mg/L) and DIN 32645, the limit of quantification (LOQ) of this method is 1.7 mg/L.
Material and Analysis 37

Reagents:
Reagent A: 143 mmol/L NaOH and 270 mM Na2CO3 in distilled water
Reagent B: 57 mmol/L CuSO4 in distilled water
Reagent C: 124 mmol/L Na2-Tartat in distilled water
Reagent D: mixture of reagent A, B and C in a ratio of 100:1:1 (shortly before the measure-
ment)
Reagent E: Folin-Cioucalteau-reagent. 1:2 diluted with dist. water
Reagent A, B, C and E are stable for around 30 d at 4 ºC. Reagent D was prepared shortly
before the measurement. The reagents and water samples were slowly warmed to room tem-
perature when needed.

Execution (duplicate determination)

1. 2.5mL water sample and 3.5 mL reagent D are placed into a test tube and quickly mixed for
5 seconds. The mixture stays for 10 minutes at room temperature.
2. 0.5 mL Reagent E is then added quickly into the tube and stirred strongly for 5 seconds
3. The mixture is in still at room temperature for 45 minutes.
4. The wave absorption of each sample is measured in a photometer at the wave length of
750nm (Perkin Elmer UV/VIS Spectrometer Lambda 12, Perkin-Elmer, U.S.A.).
5. The protein concentration is calculated based on the calibration.

Measurement of Polysaccharide Concentration

Polysaccharide were determined using the method developed by Rosenberger (2003) based on
the Dubois method (Dubois et al., 1956). The method is based on colour labelling, the absorp-
tion of the formed colour is measured at 490 nm in a 4 cm quartz cuvet by a photo spectrome-
ter. The concentration of polysaccharide is expressed in mg/L using D-glucose as a standard.
Chemical reagents are 5% phenol solution (in demineralized water) and 95% -97% sulphuric
acid (H2SO4). According to the calibration result (within the range of 0 to 10 mg/L) and DIN
32645, the LOQ of this method is 0.7 mg/L.

Execution (duplicate determination)

1. 2 mL water sample is added into a test cube and mixed with 2 mL phenol solution by shak
ing for 10 seconds. The mixture is then in still for 10 minutes at room temperature.
2. 10 mL sulphuric acid is added into the mixture. After in still for 5.5 minutes at room tem
perature the mixture is shaken for 10 seconds and then in still for 25 minutes at room tem
perature.
3. The wave absorption of the sample is measured in a photometer at the wave length of 490
nm (Perkin Elmer UV/VIS Spectrometer Lambda 12, Perkin-Elmer, U.S.A.).
4. The polysaccharide concentration is calculated based on the calibration result.

3.3.4 DOC Measurement

A High TOC II (Elementar Analysensysteme GmbH, Germany) is used to measure the DOC
concentration of treated wastewater. The DOC content is the difference of carbon content
between total carbon (TC) and total inorganic carbon (TIC) measured based on the thermal
catalytic method. For TC measurement, the water sample (0.45 µm filtered) is injected into a
38 Material and Analysis

cartridge which is heated from 700 º – 750 ºC to 850 º – 920 ºC. The combustion products are
then transported by CO2-free pressure air to the catalyst (CuO at 950 ºC) and oxidized. After
the removal of water vapour and halogens, the CO2 content is quantified by a nondispersive
infrared (IR) detector. To measure TIC, the water sample is acidified to lower than pH 2 when
it is injected into a sparger. After the removal of water vapour and halogens, the driven out
CO2 is transported by CO2-free pressure air to the IR detector. The difference of the two car-
bon contents is the DOC of a water sample. For the standardized measurement of TOC, the
detection limit is lower than 0.5 mg/L.

The DOC concentration can also be quantified using LC-OCD using by-pass operting mode
based on photochemical oxidation. During measurement, the injected water sample (after 0.45
µm filter) is firstly acided by ortho phosphoric acide to remove inorganic carbons. Subse-
quently, the water sample is transported by the eluent flow to a rotating Gräntzel thin-film
reactor. While being transported along the surface of the reactor, organic contents in water
sample are oxidized by UV radiation at a wave length of 185 nm (generated by a low-pressure
mercury-vapor lamp). The formed CO2 is blowed by pure nitrogen gas to the IR detector.
Corresponding carbon content can be calculated with the sofeware FIFIKUS® (DOC Labor Dr
Huber, Karlsruhe) according to the signal intensity. For DOC measurement, the detection
limit is lower than 0.5 mg/L. Compared to the thermal catalytic method which is assumped to
have almost 100% oxidation efficiency, the photochemical method demonstrates a DOC re-
covery over 90% with respect to secondary effluent. This suggests a requirement for improv-
ing the oxidation efficiency of the photochemical oxidation process.

3.3.5 On-line Measurement of Oxygen

The oxygen concentration in SE and slow sand filtrate (SSF) is measured using the oxygen
meter OXY-4 (Precision Sensing GmbH, Germany). The major components of the instrument
are a light source, a photodetector and optical fibers each connected with an oxygen-sensitive
minisensor. The principle of the measurement is based on the effect of dynamic luminescence
quenching by molecular oxygen. The collision between the luminophoren (on minisensors) in
its excited state (caused by light source) and the quencher (oxygen) results in radiationless
deactivation. After collision, energy transfer takes place from the excited indicator molecule
to oxygen which is consequently transferred from its ground (triplet state) to its excited
singlet state. As a result, the indicator molecule does not emit luminescence and the measur-
able luminescence signal descreases. Based on a proportional correlation between the lumi-
nescence intensity and oxygen concentration, the detected luminescence intensity by the
photodetector indicates the concentration of oxygen. Calibration of oxygen minisensors is
performed using a two-point calibration in oxygen free water (1 g sodium sulfite in 100 mL
water) and air-saturated water (20 min air blowing and 10 min stirring in 100 mL water).

3.3.6 Analysis of Bulk Water Quality Parameters

During the experiment, some other bulk water quality parameters were also tested. Tab. 3-4
exhibits the parameters and correspondingly used measurement equipments.
Material and Analysis 39

Tab. 3-4 Bulk water quality parameters and corresponding applied measurement equipments

Parameter Equipment Producer

UV254 (1/m) Perkin Elmer UV/VIS Spectrometer Perkin-Elmer GmbH
Lambda 12 U.S.A.
Turbidity (NTU) 2100 NIS Turbidimeter HACH
pH WTW pH 537 WTW, Germany
Temperature (ºC) and WTW LF340 WTW, Germany
conductivity (µS/cm)
40 Identification and Quantification of Major Organic Foulants

4 Identification and Quantification of Major Organic Foulants

In UF of treated domestic wastewater SMP/EPS have been identified as major organic fou-
lants (Jarusutthirak and Amy, 2006; Laabs et al., 2004; Amy, 2008). Different analysis meth-
ods have been applied to quantify the content of these substances and to correlate the results
with fouling development (Rosenberger et al., 2005; Laabs, 2004; Roorda, 2004). Neverthe-
less, due to the high complexity of the fouling phenomenon in UF there has no universal pa-
rameter been found reflecting the fouling potential. To detect foulants, the analysis method
should be capable of presenting their size, basic chemical characteristics and also quantity,
simultaneously.

In this section the influence of size fractioned substances on the filterability of water sample
and the reversibility of correspondingly formed fouling layers are quantitatively identified in
lab-scale experiments. Based on these results, the major organic foulants have been identifed.
The conclusion is verified to be reproducible according to pilot-scale UF results. Fouling
mechanisms due to organic foulants in UF of treated domestic wastewater are interpreted us-
ing the pore-blocking model. Based on quantitative results, different methods detecting the
fouling potential of treated wastewater are compared.

4.1 Identication of Major Foulants

4.1.1 Impact of the Size of Foulants on Filterability

It is considered that in low-pressure membrane filtration colloids and dissolved organic sub-
stances in treated wastewater are major foulants. In the current work the substances with a
nominal size larger than 1 µm are classified as particles, those with a diameter less than 1 µm
are classified as colloids and the materials able to pass through a filter with pore size of 0.45
µm are defined as dissolved substances (Haberkamp, 2008). In this section, the impact on
filterability of size fractioned substances is firstly tested (size-fraction using cellulose nitrate
filter with nominal pore size of 1.2 µm, 0.45 µm and PES UF membrane (UP150, nominal
pore size 0.026 µm)) and the results are displayed in Fig. 4-1. Fig. 4-1(a) shows that direct
filtration of SE causes a steep flux decline which indicates the low filterability of the water
sample. After the removal of suspend particles (due to pre-filtration with 1.2 µm filter), the
filterability is slightly improved. A further removal of the large colloids (due to pre-filtration
with 0.45 µm filter) results in an increasing improvement of the filterability. When the UF
permeate is filtrated using a new UF again, the normalized flux is significantly increased. The
results imply that compared to particles, large colloids and substances smaller than UF pore
size, the substances within the size of 0.45 – 0.026 m have the most significant influence on
the filterability of SE. Similar result can be obtained by filtering slow sand filtrate (SSF).
Identification and Quantification of Major Organic Foulants 41

(a) (b)
1.2 1.2

1 1

0.8 0.8

Noramlized flux
Noramlized flux

0.6 0.6

0.4 0.4

pure water SE-UF pure water SSF-UF

0.2 0.2
SE-0.45µm SE-1.2µm SSF-0.45µm SSF-1.2µm
secondary effluent slow sand filtrate
0 0
0 100 200 300 400 500 0 100 200 300 400 500
Permeate volume [ml] Permeate volume [ml]

Fig. 4-1 Influence of size fraction in SE and its SSF on filterability. (a) SE, SE after 1.2 µm filter,
after 0.45 µm filter and after UP150; (b) SSF at filtration rate 0.25m/h, SSF after 1.2 µm filter, after
0.45 µm filter and UP150

Fouling resistance (calculated based on Eq.2-3) contributions of different fractioned sub-

stances are shown in Fig. 4-2. As displayed, in either SE or SSF all fractions (particles, larger
colloids and organic residuals) lead to fouling resistance. But, the most pronounced fouling
was caused by dissolved substances within the fraction of 0.45 – 0.026 m. This fraction con-
tributes to more than 50% of the total fouling resistance in both water samples. Although the
fractionation neglects the coupling or synergistic effect which may influence the interaction
among different compounds, it reflects the importance of dissolved substances in respect to
fouling resistance during membrane filtration.

2.0E+12
dissolved substances ( <0.026 µm )

dissolved substances ( 0.026 - 0.45 µm )

Fouling resistance [1/m]

1.5E+12

large colloids ( 0.45 - 1.2 µm )

particles ( >1.2 µm )
1.0E+12

5.0E+11

0.0E+00
Secondary effluent SSF 0.25m/h

Fig. 4-2 Fouling resistance attributed from different fractioned substances in SE and its SSF

The filterability results show that in treated domestic wastewater the dissolved substances
larger than or comparable to UF membrane pore size (0.45 – 0.026 m) have a predominant
influence on the water filterability. These substances play an important role in the fouling
phenomenon and need to be identified. In subsequent sections, fouling tests and analytical
investigations were focused on these dissolved substances (water sample after 0.45 m filtra-
tion).
42 Identification and Quantification of Major Organic Foulants

4.1.2 Influence of Foulant Content on Water Filterability

Detection of Major Dissolved Organic Foulants Using LC-OCD-UVD-OND
As the DOC concentration in treated domestic wastewater was reduced by around 8% after
UF in a Amicon cell (Tab. 4-2), the removed DOC played a major role in the fouling phe-
nomenon and this part of organic matters must be characterized in more details. LC-OCD-
UVD-OND was used to determine the removed dissolved substances through UF.

Tab. 4-1 DOC and biopolymer content of SE, permeate of SE and SSFs

Biopolymer peak of LC-OCD-UVD

Water sample DOC DOC UV 254 SUVA

mg/L mg C/L 1/m L/(m·mg)

SE 12.2 0.48 0.31 0.65
Permeate of SE 11.2 < 0.05 n.d. -
SSF 0.25 m/h 10.0 0.23 0.14 0.61
SSF 0.5 m/h 10.5 0.26 0.16 0.62

SUVA: specific UV absorbance

secondary effluent
6
OC-Signal

UF permeate of
secondary effluent

0
30 60 90 120
Elution time [min]
Fig. 4-3 Liquid chromatography (LC-OCD) of SE and its UF permeate

A comparison of the chromatograms of SE and its UF permeate show that the most significant
difference appeared at the first peak after an elution time of about 40 minutes (Fig. 4-3).
While the SE exhibited a clearly detectable peak, in the UF permeate the peak was mostly
removed. As the difference after a retention time of 55 minutes was much smaller than that in
this peak, the substances detected in the first peak, which is called biopolymer peak, should
Identification and Quantification of Major Organic Foulants 43

present much higher responsibility for the fouling phenomenon than other organic fractions.
Similar results can be obtained by comparing the chromatograms of SSF and its correspond-
ing UF permeate. Based on the calibration results, most detectable substances in the biopoly-
mer peak are considered as large organic molecules. They show comparatively high organic
carbon content and low UV absorption (low aromaticity, Tab. 4-1). Organic nitrogen was also
detected (data not quantified). They are normally regarded as polysaccharide-like and protein-
like substances (Huber, 1998; Her et al., 2003; Amy, 2008).

Correlation between Biopolymer Content and Filterability

Filterability of SE, SSF (at filtration rate of 0.25 and 0.5 m/h) and UF permeate of SE was
compared (Fig. 4-4 (a)). It is shown that filtering SE (after 0.45 m) led to a 87% flux decline
for a cumulative permeate volume of 500 mL, while filtering SSFs (after 0.45 m) at filtration
the rate of 0.25 m/h and 0. 5 m/h showed 75% and 78% flux declines, respectively. After UF
when the permeate of the SE was filtrated again, it showed a strikingly high permeability
compared to SE and SSFs as the normalized flux showed only a 16% decline.

(a) 1 (b) 1
SE
0.8 0.8
UF permeate of SE SSF 0.5 m/h
Normalized flux [J/J0]

0.6 SSF 0.25 m/h 0.6 SSF 0.25 m/h

OC signal

SSF 0.5 m/h

0.4 UF permeate of SE
SE 0.4

0.2 0.2

0
0
0 100 200 300 400 500
35 40 45 50 55 60
Permeate volume [mL] Elution time [min]

Fig. 4-4 Correlation of filterability and biopolymer peak of different water samples (a) Normalized
stirred-cell UF flux decline curves using SE, SSFs and UF permeate over cumulative permeate volume
of 500 mL. (b) Biopolymer peaks of SE, UF permeate of SE and SSFs

The biopolymer peaks of corresponding water samples are shown in Fig. 4-4 (b). It is illus-
trated that SE was of the lowest filterability and it showed the largest biopolymer peak. After
slow sand filtration at a filtration rate of 0.5 m/h the filterability was to some extent improved
and the peak was reduced to some degree; the SSF at a filtration rate of 0.25 m/h showed a
higher filterability and it exhibited a further reduced biopolymer peak; the UF permeate of SE
exhibited a much higher filterability and there was much fewer biopolymer detected in it. The
comparison suggests that the content of biopolymer in dissolved fractions is negatively corre-
lated with the filterability of corresponding water samples using a UP150 membrane.
44 Identification and Quantification of Major Organic Foulants

Quantitative Relationship between Biopolymer Concentration and Filterability

Since the concentration of biopolymers influences the filterability, water samples with a
broader range of biopolymer concentration are tested to quantify the relationship. Fig. 4-5 (a)
shows the biopolymer peaks of different water samples. Larger biopolymer peak areas repre-
sent higher biopolymer concentrations; Fig. 4-5 (b) displays the filterability of corresponding
water samples. Comparison of the results demonstrates that the larger the area of the bio-
polymer peak is, i.e. the higher the biopolymer concentration in the water sample, the lower
the filterability. The fouling resistance results in Fig. 4-5 (c) is in agreement with the filter-
ability performance, water samples with higher biopolymer concentration lead to higher hy-
draulic resistance and show lower filterability.

(a) 1.0
Mar.06.2007 SE

Nov.16.2006 SE
0.8
Dec.16.2006 SSF

Apr.04.2007 SE

0.6 Mar.06.2007 SSF

OC-Signal

Jul.12.2007 SSF

0.4

0.2

0.0
35 40 45 50 55 60
Elution time [min]

(b) 1 (c) 6.E+12

0.8
Fouling resistance [1/m]
Noramlized flux

4.E+12
0.6

0.4
2.E+12

0.2
Mar.06.07. SE Nov.16.06. SE
Dec.16.06. SSF Apr.04.07. SE
Mar.06.07. SSF Jul.12.07. SSF
0 0.E+00
0 100 200 300 400 500 Mar.06.07. Nov.16.06. Dec.16.06. Apr.04.07. Mar.06.07. Jul.12.07.
Permeate volume [ml] SE SE SSF SE SSF SSF

Fig. 4-5 (a) Biopolymer concentration of different water samples, (b) normalized flux of correspond
ing water samples, UP150, TMP 1 bar, (C) fouling resistance of corresponding water samples (flux
data in detail see section 8.3, A-Tab.2)

To derive the relationship between biopolymer concentration and filterability or fouling resis-
tance, results of more water samples are plotted under permeate volume 150 ml, 250 ml and
500 ml and are displayed in Fig. 4-6. Fig. 4-6 (a) and (b) show that for each permeate vol-
ume, the relationship shows the same trend: in the range of biopolymer concentrations be-
tween 0.05 to 0.58 mg C/L in the present work, water samples with higher biopolymer con-
centrations show a lower filterability and lead to higher fouling resistance in UF. The rela-
 Identification and Quantification of Major Organic Foulants 45

tionship between biopolymer concentration and fouling resistance during the whole filtration
process is very striking as the R2 is higher than 0.95.

(a) 1 (b) 8.E+12

permeate 150 ml permeate 500 ml
0.8 y = 8E+12x - 4E+11
permeate 250 ml

Fouling resistance [1/m]

permeate 250 ml 2
R = 0.95
permeate 150 ml
Normalized flux

0.6 permeate 500 ml

4.E+12
2
0.4 R = 0.88
R² = 0.96

2
0.2 y = 0.0663x-0.8874 R = 0.91
2
R = 0.91 R² = 0.96
0 0.E+00
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Biopolymer concentration [mgC/l] Biopolymer concentration [mgC/l]

Fig. 4-6 Relationship between biopolymer concentration and (a) filterability (b) fouling resistance
under permeate volume 150 mL, 250 mL and 500 mL. (n=26, water samples were taken randomly
from Nov.2006 to July 2007 in WWTP Ruhleben, Berlin, Germany)

Verification of the Proportional Correlation

As biopolymers have been demonstrated as major foulants, it is reasonably to deduce that
under a similar delivered biopolymer load the normalized flux decline during UF of different
water samples should be comparable. To verify this hypothesis, different water samples were
filtrated with the same biopolymer load calculated based on Eq.4-1.

M b = Cb × V Eq. 4-1

where
Cb = biopolymer concentration, mg C/L
V = permeate volume, L

(a) (b)
1 1

0.8 0.8
Nomalized flux
Nomalized flux

0.6 0.6

0.4 0.4

0.2 0.2
slow sand filtrate slow sand filtrate
secondary effluent secondary effluent
0 0
0 1500 3000 4500 0 0.15 0.3 0.45
Permeate volume [ml] Delivered biopolymer [mgC]
Fig. 4-7 Influence of delivered biopolymer amount on the normalized flux decline of SE and SSF. (a)
Normalized flux decline under the same permeate volume; (b) Normalized flux decline under the same
delivered biopolymer amount

Relying on Eq.4-1 filtrating 500 mL SE (biopolymer 0.29 mg C/L) or 1350 mL SSF (0.11 mg
C/L) result in similar biopolymer loads to the membrane surface. The corresponding results
are shown in Fig. 4-7. Under the same permeate volume the normalized flux of SE is much
lower than that of SSF since under these conditions filtrating SE leads to more biopolymer
46 Identification and Quantification of Major Organic Foulants

load to the membrane surface (Fig. 4-7 (a)). But for the same load, the normalized flux de-
cline of the two different water samples is almost identical (Fig. 4-7 (b)), after approximately
0.145 mg C (50.5 mg C/m2 membrane area) delivered biopolymer to the UF membrane sur-
face in each filtration cycle, the normalized fluxes are always similar. The comparison verifies
that similar biopolymer load delivered to the membrane surface causes similar fouling effects.

4.1.3 Effect of Foulant Content on the Reversibility of Fouling

Although the hydraulic fouling resistance impacts the efficiency of membrane filtration, for a
long-term operation, the reversibility of the formed fouling layer plays a more crucial role on
the performance. To investigate the influence of biopolymer concentration on the reversibility
of the formed fouling layer in UF membranes, water samples with different biopolymer con-
centrations were filtered using Amicon cells with five filtration cycles. The results (Fig. 4-8
(a)) show that all of the treated wastewater lead to a flux decline – a permeability reduction
during the filtration process. As expected, in each cycle the water samples with higher bio-
polymer concentrations lead to a more severe flux decline. It was further observed that after
each backwash, the higher the biopolymer concentration in the feed water, the lower the re-
versibility the corresponding fouling was (Fig. 4-8 (b)). Although most of the biopolymers are
reversible as most of them are present in the backwash water under the present experimental
condition (Fig. 4-9), the results indicate that the more biopolymers are delivered to the mem-
brane surface, the more irreversible the fouling is.

(a)
UF permeate BC < 0.05 mgC/L
SSF BC 0.08 mgC/L
SE BC 0.42 mgC/L
SE BC 0.49 mgC/L
Normalized flux [J/J0]

1.0

0.5

0.0
0 500 1000 1500 2000 2500 3000
Permeate volume [mL]
Identification and Quantification of Major Organic Foulants 47

(b)
UF permeate BC < 0.05 mgC/L
BC 0.08 mgC/L
BC 0.42 mgC/L
BC 0.49 mgC/L
100
Reversibility [%]

50

0
cycle 1 cycle 2 cycle 3 cycle 4 cycle 5

Fig. 4-8 (a) Influence of biopolymer concentration on fouling effect, (b) flux recovery at different
biopolymer concentrations (BC: biopolymer concentration, Feed water is SE, SSF and SE UF perme-
ate, operating pressure 1 bar, permeate volume 500 mL; back wash using salt solution 50 mL at 1 bar)

From October 8th 2007 to May 8th 2009 ten water samples (SE, SSF and also UF permeate)
were randomly sampled and tested using Amicon cells with five filtration cycles. The correla-
tion between biopolymer concentration and the reversibility of correspondingly formed foul-
ing is displayed in Fig. 4-10 (a). As shown, with an increasing biopolymer concentration in
feed water the reversibility of the fouling decreases. This phenomenon is confirmed by Fig.
4-10 (b) which shows that the increase of irreversible fouling resistance is related to the rise
of biopolymer concentration. This phenomenon indicates that the more biopolymers delivered
to the membrane, the more difficult they can be removed by backwashing. Biopolymers play
thus a crucial role in the formation of irreversible fouling.

12

secondary effluent: BC 0.42 mg C/L

UF permeate: BC 0.08 mg C/L

8
concentrate: BC 0.55 mg C/L
OC-Signal

backwash water: BC 1.89 mg C/L

0
30 60 90
Elution time [min]

Fig. 4-9 Comparison of the LC-OCD chromatograms of SE, permeate after UP150, supernatant and
back wash water (Permeate volume 500 mL, concentrate 200 mL, backwash water volume 50 mL,
Dilution effect of different water samples was regulated)
48 Identification and Quantification of Major Organic Foulants

(a) 90 (b) 3.E+11

cycle 5

Irreversible fouling resistance [1/m]

cycle 3 R2 = 0.82
Reversibility [%]

80 2.E+11 cycle 1

R2 = 0.85 R2 = 0.79
70 1.E+11

R2 = 0.6
60 0.E+00
0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Biopolymer concentration [mg C/L] Biopolymer concentration [mg C/L]

Fig. 4-10 (a) linear correlation between biopolymer concentration and reversibility after filtration cy-
cle 5, (b) potential correlation between biopolymer concentration and irreversible fouling resistance
after different filtration cycles (from October 8th 2007 to May 8th 2009, n=10, biopolymer concen-
tration from 0.05 to 0.56 mg C/L).

To verify the influence of biopolymers on the performance of UF observed in the Amicon

experiment, TMP development of the UF pilot plant operated at identical operational condi-
tions (flux 50 LMH, BWI 10 minutes, BWT 10 seconds and BWF 260 LMH) was tested fil-
tering SE or SSF at different biopolymer concentrations. Results taken from filtering SE and
SSF were compared separately to minimize the influence of colloids and particles. Fig. 4-11
(a) shows the TMP development filtering SE at different biopolymer concentrations. It was
exhibited that filtering SE with a biopolymer concentration of 0.79 mg C/L resulted in a TMP
increase from 220 mbar to 700 mbar within 12 hours. The operating time with a biopolymer
concentration of 0.56 mg C/L was extended to 14 hours. When the biopolymer concentration
in the feed water was around 0.17 mg C/L, within the similar TMP increase scope, the operat-
ing time was extended to around 8 days. The comparison showed that in SE the biopolymer
concentration affects the performance of UF processes decisively, lower biopolymer concen-
trations lead to a slower TMP development.

(a) 800 4.0

SE BC 0.79 mg C/L, Dec.30 2007

Biopolymer concentration [mg C/L]

600 SE BC 0.56 mg C/L, May 29 2008 3.0

SE BC 0.17 mg C/L, Aug.12 - 20 2008

TMP [mbar]

400 2.0

200 1.0

0 0.0
0 1 2 3 4 5 6 7 8
Operating time [d]
Identification and Quantification of Major Organic Foulants 49

(b) 1000 2.5

SSF BC 0.49 mg C/L, Dec.13 - 29 2007

SSF BC 0.32 mg C/L, Nov.1 - Dec.10 2007

800 2.0

Biopolymer concentration [mg C/L]

SSF BC 0.09 mgC/L, Aug.23 - Oct.8 2008

600 1.5
TMP [mbar]

400 1.0

200 0.5

0 0.0
0 5 10 15 20 25 30 35 40 45 50
Operating time [d]

Fig. 4-11 TMP increase of UF pilot plant filtrating (a) SE and (b) SSFs at different biopolymer con-
centrations (BC: biopolymer concentration. Operational condition of UF: flux 50 LMH, BWI 10 min-
utes, BWT 10 seconds, BWF 260 LMH).

The results filtering SSFs are shown in Fig. 4-11 (b). It can be seen that filtering SSF with a
biopolymer concentration of 0.49 mg C/L (on average) can be operated without chemical
cleaning for about 15 days. With a biopolymer concentration of 0.09 mg C/L in SSF the oper-
ating time was 47 days with a TMP increase only from 280 to 490 mbar. The results indicate
that in either SE or SSF, biopolymer content in the feed water influences the performance of
UF process crucially. Higher biopolymer loads on the membrane surface lead to more irre-
versible fouling.

4.2 Fouling Mechanisms in UF

4.2.1 Fouling Mechanisms in Dead-end Filtration

Relying on the results acquired by determining filterability and biopolymer concentration,
attempts were made to analyze the fouling mechanisms of biopolymers during dead-end UF
process. Mechanisms within filtration of water containing colloidal matter through a low
pressure membrane are well studied (Schippers and Verdouw, 1980; Roorda, 2004; Costa et
al., 2006; Katsoufidou et al., 2007). It is generally regarded that in dead-end ultrafiltration the
predominant filtration mechanisms are pore blocking and cake/gel filtration.

Fouling mechanisms relevant to constant pressure dead-end filtration experiments can be ana-
lyzed based on the approach summarized by Hermia (1982). Four major fouling mechanisms
are described as complete blocking, standard blocking, intermediate blocking and cake filtra-
tion. Different mechanisms can be present simultaneously and the flow behaviour index n
(section 2.3, Eq.2-7) may present a combination of different mechanisms by ranging in be-
tween given values (Crittenden et al., 2006).
50 Identification and Quantification of Major Organic Foulants

The experimental data of permeate volume versus time in Fig. 4-5 (b) and Fig. 4-7 (a) is ana-
lyzed to generate plots of log(d²t/dV²) versus log(dt/dV), the formed curve is regarded as a
characteristic curve reflecting fouling mechanisms as its slope is the flow behaviour index n.
Since the derivations depends strongly on the smoothness of t-V curves, fitting polynomials
of degree of six (Bowen et al., 1995) was conducted in each filtration process.

Fig. 4-12 presents characteristic curves reflecting fouling mechanisms under different bio-
polymer concentration ranges by filtrating 500 mL (standard condition in the present study) or
1350 mL water samples. Under standard condition, at high biopolymer concentrations (> 0.28
mg C/L) n starts within the range of 0.2 - 1 to 0 along the filtration process. Based on block-
ing laws, the fouling process can be explained as following: during filtration processes, bio-
polymers are firstly delivered to the membrane open pores. As the biopolymer concentration
is comparatively high, all of the pores can be blocked in a short time, successive delivered
biopolymers can only arrive on the biopolymers already present; this is the intermediate
blocking process. With more biopolymers accumulating on the membrane surface, a gel layer
then forms and cake/gel filtration dominates the following filtration process. As displayed in
Fig. 4-12, the characteristic curve transforms from intermediate blocking (or combination of
intermediate blocking) to cake filtration under the given operational condition.

5
Biopolymer: 0.59 mg C/l
n=0.4→0 (V=500 ml)
Biopolymer: 0.48 mg C/l
n=0.8→0 (V=500 ml)
4.5
Biopolymer: 0.38 mg C/l
n=1→0 (V=500 ml)
log(d²t/dV²)

Biopolymer: 0.28 mg C/l

n=0.2→0 (V=500 ml)
4
Biopolymer: 0.25 mg C/l
n=0.3→0.6 (V=500 ml)
Biopolymer: 0.17 mg C/l
n=0.9→0.1 (V=500 ml)
3.5
Biopolymer: 0.11 mg C/l
n=2→0 (V=1350 ml)
Biopolymer: 0.07 mg C/l
n=2.2→1.9 (V=500 ml)
3
3 3.5 4 4.5 5
log(dt/dV)

Fig. 4-12 Characteristic curves reflecting fouling mechanism for biopolymer fouling in UF at different
biopolymer concentrations (Data are taken from Fig. 4-5 (b) and Fig. 4-7 (a), treated with polynomials
of degree of six and analyzed with Eq.2-7. V: filtration volume. An example determining fouling
mechanisms is shown in detail by section 8.3 A-Tab.3).

When biopolymer concentration in influents is in the medium range (0.17 – 0.25 mg C/L), n
exhibits a value in the scope of 0.1 to 0.9 during filtration of 500 mL water samples. Fouling
mechanism can be assumed as a combination of intermediate blocking and cake filtration. In
filtrating waters containing a low concentration of biopolymers (< 0.11 mg C/L), n is always
around 2 at the beginning. This complies well with the behaviour predicated by the complete
Identification and Quantification of Major Organic Foulants 51

blocking model. The fouling process can be described as follows: Each delivered biopolymer
molecule arriving to the membrane participates firstly in blocking an open membrane pore.
As biopolymer concentration in the influent is comparatively low, after filtrating 500 mL wa-
ter sample the biopolymers are not sufficient to cover all of the membrane pores/area. How-
ever, an increase of biopolymer load results in an alteration of fouling mechanism. As shown
by filtering a 1350 mL water sample with a biopolymer concentration of 0.11 mg C/L, n
changes from 2 to 0. This demonstrates that at low biopolymer concentrations sufficient bio-
polymers delivered to membranes surface results in a transition of fouling mechanisms from
pore blocking to cake/gel filtration.

4.2.2 Fouling Mechanisms in Multi Cycle Filtration

In multifiltration cycles (section 4.1.3) it is shown that after the filtration process the formed
biopolymer cake/gel layer can be readily backwashed. Nevertheless, irreversible fouling de-
velops continuously along with the increase of filtration cycles. The influence of irreversible
fouling on fouling mechanism is further investigated based on the blocking laws.

(a) (b) 4.6

1 bar, BC 0.25 mg C/L 1 bar, BC 0.42 mg C/L
1 bar, BC 0.42 mg C/L
Normalized flux [J/J0]

1 0.5 bar, BC 0.42 mg C/L

log(d²t/dV²)

4.2

0.5
cycle 1, n=0.1→0 cycle 2, n=0.2→0
cycle 3, n≈0 cycle 4, n≈0
cycle 5, n≈0
3.8
0
0 500 1000 1500 2000 2500 3000 3.3 3.6 3.9 4.2
log(dt/dV)

(c) 4.6 (d) 4.6

0.5 bar, BC 0.42 mg C/L 1 bar, BC 0.25 mg C/L
cycle 1, n=1.0→0 cycle 2, n=0.9→0
cycle 3, n≈0 cycle 4, n=0.4→0
log(d²t/dV²)

log(d²t/dV²)

cycle 5, n≈0
4.2 4.2

cycle 1, n=1.2→0 cycle 2, n=0.3→0

cycle 3, n=0.1→0 cycle 4, n=0.4→0
cycle 5, n≈0

3.8 3.8
3.3 3.6 3.9 4.2 3.3 3.6 3.9 4.2
log(dt/dV) log(dt/dV)

Fig. 4-13 (a) Multifiltration of water under different conditions, characteristic curves in UF of water
(b) with a biopolymer concentration of 0.42 mg C/L under 1 bar, (c) with a biopolymer concentration
of 0.42 mg C/L under 0.5 bar, (d) with a biopolymer concentration of 0.25 mg C/L under 1 bar

Fig. 4-13 (b) presents the characteristic curves filtering a water sample with biopolymer con-
centration of 0.42 mg C/L in five filtration cycles. It is shown that n changes from 0.1 to 0
during the first cycle. The variation of n indicates the shift of fouling mechanisms from pore
blocking to cake/gel formation. After backwashing, in the second filtration cycle the alteration
of fouling mechanism is similar to that in the previous. Nevertheless, from filtration cycle
52 Identification and Quantification of Major Organic Foulants

three n presents around zero at the beginning of the filtration. This indicates that at the very
beginning of the filtration there is no open pore for biopolymers to block, they can only cu-
mulate at the surface of a fouling layer. In subsequent filtration cycles, the major fouling
mechanism is identified as cake/gel filtration. This phenomenon indicates that although the
formed cake/gel layer is readily backwashed, the irreversible attachment among biopolymers
and membrane is further promoted due to the interactions among them (Zhang et al., 2006a).
This leads to a further cumulation of biopolymers and subsequently the increase of irreversi-
ble fouling resistance. Similar trend can be observed filtering water at a lower operating pres-
sure (Fig. 4-13 b) or with a lower biopolymer concentration (Fig. 4-13 c).

4.3 Comparison of Fouling Potential Measurements

As biopolymers quantified using LC-OCD-UVD-OND are identified as major foulants, fur-
ther investigations were undertaken to find out a possible correlation between biopolymer
content and fouling potential parameters quantified by other measurement methods, e.g., pro-
tein and/or polysaccharide concentration analysed using photometric methods and fouling
index measured based on filtration curves.

4.3.1 Photometric Measurement of Polysaccharide and Protein

Several water samples analysed using LC-OCD-UVD-OND were also tested using the
photometric methods to determine the relevant concentration of polysaccharides and proteins
(Tab. 4-2).

Tab. 4-2 DOC, protein, polysaccharide concentration and biopolymer content of SE, permeate of SE
and SSFs

DOC Protein Polysac- Biopolymer Filterability Fouling

charide resistance
Water sample
(mg/L) (mg/L) (mg/L) (mg C/L) (1011 1/m)
SE 12.2 10.9 4.6 0.48 0.12 3.41
Permeate of SE 11.2 10.4 4.3 < 0.05 0.82 0.1
SSF 0.25 m/h 10.0 9.4 3.2 0.23 0.22 1.63
SSF 0.5 m/h 10.5 9.6 3.7 0.26 0.20 1.92
Identification and Quantification of Major Organic Foulants 53

(a) 12 0.6
R2 = 0.03

Biopolymer concentration [mg C/L]

protein
Protein polysaccharide

8 0.4
concentration [mg/L]

polysaccharide
R2 = 0.99 biopolymer

4 0.2
R2 = 0.06

0 0
0 0.2 0.4 0.6 0.8 1
Filterability

(b) 12 4

R2 = 0.99

Fouling resistance [10 1/m]

R2 = 0.13
Protein polysaccharide

11
8
concentration [mg/L]

2
R2 = 0.08

4
protein
polysaccharide
fouling resistance
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6
Biopolymer concentration [mg C/L]

Fig. 4-14 (a) Correlation between filterability and content of foulants (potential relationship with bio-
polymers), (b) correlation between fouling resistance and content of foulants (PR: protein, PS: poly-
saccharide, BC: biopolymer concentration)

It is shown that after UF the content of polysaccharides and proteins in the SE was reduced
from 4.6 mg/l and 10.9 mg/l to 4.3 mg/l and 10.4 mg/l, respectively. These substances were
removed by UF and seemed to be potential foulants. However, the concentration of polysac-
charides and proteins in the UF permeate of SE was higher than that in SSFs, although the UF
permeate showed a much higher filterability (Fig. 4-14 a). No quantitative correlation be-
tween biopolymer and protein and polysaccharides can be observed (Fig. 4-14 b). The com-
parison indicated that the photometrically detectable concentrations of polysaccharides and
proteins are not in accordance with the foulant content detected with LC-OCD-UVD-OND.
They show little information about the filterability of the water. In low-pressure membrane
filtration processes the rejection of foulants is primarily due to size exclusion (Amy and Cho,
1999), the substances which are larger than membrane pore size contribute predominantly to
the flux decline. The analysis using LC-OCD-UVD-OND can reflect the size distribution of
organics and the amount of detectable foulants which show a correlation with filterability of
water sample. Photometric methods reveal no information about molecular size distribution
54 Identification and Quantification of Major Organic Foulants

and the results show no correlation with filterability. This result is consistent with the obser-
vations in several other studies (Poele et al., 2004; Kimura et al., 2009). When the content of
biopolymers and photometricly quantified polysaccharides and proteins are plotted together
with more measurements (Fig. 4-15), as expected, little correlation can be found between
these two types of analysis methods.

(a) 0.8 (b) 0.8

2
R = 0.24
0.6 0.6
BC [mg C/L]

BC [mg C/L]
2
R = 0.03
0.4 0.4

0.2 0.2

0.0 0.0
2.0 3.0 4.0 5.0 5.0 10.0 15.0 20.0
Polysaccharide concentration [mg/L] Protein concentration [mg/L]

(c) 0.8

0.6 R2 = 0.07
BC [mg C/L]

0.4

0.2

0.0
10.0 12.0 14.0 16.0 18.0 20.0 22.0
Protein+polysaccharide [mg/L]

Fig. 4-15 Correlation between biopolymers and photometric measured polysaccharide (Dubois
method) and protein (Lowry method) (a) polysaccharide and organic carbon in biopolymers, (b) pro-
tein and organic carbon in biopolymers, (c) polysaccharide+protein and carbon in biopolymers (n =
82, BC: biopolymer concentration)

4.3.2 Fouling Index and Hydraulic Fouling Resistance

It is demonstrated that there is a proportional correlation between biopolymer concentration
and fouling resistance (Zheng et al., 2009a). This relationship suggests a possible connection
between fouling resistance measured in a filtration cycle and the fouling potential of the cor-
responding water sample. To validate this deduction, the filtration resistance of the ten water
sample taken from October 8th 2007 to May 8th 2009 (used in section 4.1.3) are plotted with
irreversible fouling resistance. As expected, Fig. 4-16 (a) shows a significant correlation be-
tween them. The result indicates that the higher the fouling resistance in a single filtration
cycle, the higher the irreversible fouling resistance is. Due to the presence of biopolymers, the
fouling resistance measured in a filtration cycle can reflect the fouling potential of the feed
water. Additionally, the correlation between biopolymer concentration and fouling resistance
shown in Fig. 4-16 (b) is similar to that investigated in a previous study between November
2006 and July 2007 (Fig. 4-5, Zheng et al., 2009a). The reproducible relationship indicates
that this correlation is constant throughout the whole experimental period (2006 – 2009).
Identification and Quantification of Major Organic Foulants 55

(a) 3.E+11 (b) 4E+12

cycle 5
Irreversible fouling resistance [1/m]

Fouling resistance after the

cycle 3 R = 0.84

first filtration cycle [1/m]

y = 6E+12x - 8E+10
2.E+11 cycle 1
R2 = 0.86
R2 = 0.91 2E+12

1.E+11

R2 = 0.68

0.E+00 0
0.E+00 1.E+12 2.E+12 3.E+12 4.E+12 0 0.2 0.4 0.6
Fouling resistance after one cycle [1/m] Biopolymer concentration [mgC/L]

Fig. 4-16 (a) Correlation between fouling resistance and development of irreversible fouling, (b) corre-
lation between biopolymer concentration and fouling resistance (from November 5. 2007 to May 08.
2009, n=10, flux and hydraulic resistance data in detail see section 8.3, A-Tab.4)

Tab. 4-3 Comparison of the operational conditions measuring SUR and fouling resistance in the pre-
sent work

SUR (Roorda, 2004) Fouling resistance

Feedwater SE pre-filtered SE (0.45µm)
Temperature effect nomalized to at 20 ºC nomalized to at 22 ºC
Operating pressure 0.5 bar 1 bar
Membrane type hollow fibre flat sheet membrane
Filtration dead-end dead-end
configuration
Membrane material PES/PVP PES
MWCO 70 – 150 kDa 150 kDa
Treatment of the NaOCl soaking and pure water flushing, flushing with pure water, varia-
membrane permeability variation within 10% tion of permeability within 10%
Specific permeate based on the applied membrane area and 174 L/m2
volume permeate volume after 30 min filtration
Filtration time 30 min varies from 30 to 90 min, de-
pends on biopolymer content
Fouling mechanism cake filtration cake/gel filtration

The measurement process of hydraulic resistance caused by biopolymer is compared with the
procedure determining the SUR of SE (Tab. 4-3). It is evident that the measurement of the
both parameters is based on cake/gel filtration in dead-end filtration mode. Besides minor
differences in operation, the operational conditions are also comparable. Similarity of deter-
mining biopolymer fouling resistance and SUR suggests a possible correlation between the
two parameters. In the Amicon UF cell filtration of biopolymer-containing water under con-
stant pressure at 1 bar, the predominant filtration mechanism is identified also as cake/gel
filtration, the duration of the filtration is generally longer than 30 min which fulfils the re-
quirement of the measurement of SUR value (Roorda, 2004). Based on the definition of SUR,
the specific fouling resistance per unit feedwater per m2 membrane area caused by biopoly-
mers in Amicon UF can be calculated according to Eq. 4-2. A significant relationship between
SUR and biopolymer concentration can be observed (Fig. 4-17). The result indicates that the
concentration of biopolymers can be proportionally reflected by SUR. Further more, com-
56 Identification and Quantification of Major Organic Foulants

pared to the slope of the linear relationship between fouling resistance and biopolymer con-
centration presented on Fig. 4-16 b, that presented on Fig. 4-17 is five times higher, which is
consistent with the theoretical relationship between these two slopes (Eq. 4-2).

A 28.7 × 10−4
SUR = Rcake = Rf × = R f × 5.74 Eq. 4-2
V 0.5 × 10 − 3
where
Rcake = Rf = hydraulic resistance of the biopolymer fouling, 1/m
V = the product volume of permeate, m3
A = the membrane filtration area, m2

2.0E+13

y = 3E+13x + 2E+11
1.5E+13
R2 = 0.96
SUR [m-2]

1.0E+13

5.0E+12

0.0E+00
0 0.1 0.2 0.3 0.4 0.5 0.6
Biopolymerkonzentration [mg C/L]

Fig. 4-17 Correlation between SUR and biopolymer concentration

Originally, SUR presents the total fouling resistance caused by particles, colloids and dis-
solved substances which can be retained by correspondingly used UF membranes (Roorda,
2004). The parameter has already been used in several studies and it shows a good correlation
with the fouling phenomena in UF of treated domestic wastewater (Roorda and van der Graaf,
2005). However, it is stated that no other water quality parameter is found to have a quantita-
tive correlation with it (Janssen and Van der Graaf, 2008). Considering that water quality
generally reflects the content of specific substances within a certain size range, the measure-
ment of SUR is however, conducted without any pre-treatment of the water sample. This may
be one of the reasons interpreting the absence of a correlation to other parameters. As dis-
solved substances (< 0.45 µm) are identified as major foulants (Zheng et al., 2009a; Poele,
2006), a pre-filtration of water samples with 0.45 µm filter prior to the determination of SUR
is therefore recommended. The SUR can be thereby defined as SUR0.45, indicating the meas-
urement of fouling potential of dissolved substances.

4.4 Discussion and Summary

Major organic foulants
In the present section, biopolymers are quantitatively identified as major organic foulants in
UF of treated domestic wastewater. Their mass load leads to a proportional increase of hy-
Identification and Quantification of Major Organic Foulants 57

draulic fouling resistance. Although most of them can be removed by backwashing, they im-
pact the development of irreversible fouling in the applied PES membrane crucially.

The crucial fouling effect of biopolymers is based on their size and chemical characteristics.
As shown in this section, biopolymers are with a nominal size larger than or comparable to
the UF pore size. They attribute to a much higher fouling resistance than that caused by part-
culate and large colloidal substances. In respect to fouling effects of different size fractioned
substances, Poele (2006) showed that using a 0.45 µm sieve prior to a pilot-scale UF couldn’t
improve its performance in filtering SE. By contrast, the application of a MF with a nominal
size of 0.1 µm led to a significant improvement in fouling control. It was concluded that the
major foulants in SE to UF are within the size range of 0.1 to 0.45 µm. The importance of
dissolved substances in UF membrane fouling is supported by other studies (Fan et al., 2008;
Laabs et al., 2006).

Despite the crucial role of dissolved biopolymers, the role of particulate and large colloidal
substances in fouling phenomena should not be ignored. For instance, it was observed that
direct UF of SE resulted in a much faster TMP development compared to treating SSF with a
similar biopolymer concentration (Fig. 4-11). It seems that particles and large colloids also
play an important role in fouling phenomenon. Jermann et al. (2008) investigated the influ-
ence of particles on the irreversibility of fouling by NOM. The results showed that the dosage
of inorganic particles (kaolinite) into polysaccharide (alginate) or other NOM (humic acid)
solutions could accelerate the development rate of irreversible fouling. A similar effect was
observed by the addition of PAC or SiO2 particles into NOM-containing water for UF (Zhang
et al., 2003). It is interpreted that the strong connection between particles (depends on their
nature and amount, Lohwacharin et al., 2009) and membrane through NOM binding results in
a continuous cake layer, which reduces the effectiveness of backwashing. Otherwise, a cake
layer formed by particles alone could be effectively backwashed and led to little irreversible
fouling (Zhang et al., 2003). The settlement of particular foulants on the membrane surface
was considered to reduce the available filtration area for dissolved foulants (Zhang et al.,
2003). The subsequent increase of local flux resulted in a more compacted fouling layer
which presented a lower reversiblility (Kim and DiGiano, 2006).

Based on the understanding of the role of particles, colloids and biopolymers in fouling phe-
nomenon, it is evident that biopolymers are the crucial foulants. They can adhere to mem-
banes strongly and bind other foulants with the membrane. Particles and large colloids are
entrapped in the sticky biopolymer fouling layer. The synergy effect of these foulants results
in a cake/gel layer which is stable against hydraulic backwashing or flushing.

Characterization of these dissolved large molecular foulants shows that biopolymers present a
low SUVA value and contain organic nitrogen, particularly identified as polysaccharides
(Huber, 1998) and proteins (Her et al., 2003). These organics have been stated as typical
components in irreversible fouling in low-pressure membrane filtering waters containing ef-
fluent organic matters (EfOM) (Jarusutthirak and Amy, 2006; Kimura et al., 2005). More re-
cently, Haberkamp (Haberkamp et al., 2008) compared the performance of a PVDF UF filter-
ing SE and its UF permeate dosed with a kind of extracted bacterial EPS. The more severe
fouling caused by SE than the prepared water was attributed to the higher content of protein-
58 Identification and Quantification of Major Organic Foulants

like substances in effluent biopolymers. Based on the investigation on the spatial components
of fouling on a PVDF membrane surface, Zhang (2009) stated that protein-like substances
attributed to more fouling than polysaccharide-like foulants in the fouling layer adjacent to
the membrane surface. Their adhesion increases the stickness and roughness of the membrane
surface and encourages the attachment of other foulants. At the outer fouling layer polysac-
charides were otherwise identified as the most important foulants. In contrast, Chen et al.
(2006a) investigated the foulant distribution on a cellulose ester membrane surface. Alpha
polysaccharide was revealed as the most concentrated EPS close to the membrane surface
which attributed to a high hydraulich resistance, although proteins and other polysaccharides
were also present in the feedwater. Filtering EPS-rich water with a hydrophilized PVDF MF
membrane, Shen et al. (2010) found that compared to other organic fractions in water, the
hydrophilic polysaccharides led to the steepest flux decline. The correspondingly formed foul-
ing showed also the lowest reversibility. It seems that in relevance to membrane properties,
different foulants play a different role in the formation of a fouling layer (Huang et al., 2007).

In general, biopolymers are considered to interact with the membrane and with each other
through hydrogen bonding (for polysaccharides) or dipole interactions (for proteins), and also
due to the contribution to the formation of a cake/gel layer (Amy, 2008). Variation of their
mass cumulated on the membrane surface can change the number of binding sites available
(Flemming and Wingender, 2001), and therefore, the total binding strength among foulants
and membrane.

Evaluation of Fouling Potential

Although fouling potential of treated wastewater can be reflected by the concentration of bio-
polymers, their carbon content alone may not be a general fouling indicator because the com-
ponents in terms of polysaccharides and proteins in different effluents can vary (Haberkamp,
2008). But, in regard to the effluent from a single wastewater treatment plant, as the case in
the present study, biopolymer content can represent its fouling potential. This result implies
that the variation of the components in biopolymers is within a certain ratio. Thus, it does not
impact the fouling phenomenon significantly.

Under the same assumption, the proportional correlation between hydraulic fouling resistance
(and also SUR0.45) and biopolymer concentration may be used to present the fouling potential
roughly. Nevertheless, as the indictor generally represents the filtration resistance caused by
dissolved substances, it is lacking in information about the nature and components of foulants.
The content of proteins and polysaccharides quantified by photometrical methods cannot re-
flect the fouling potential of treated domestic wastewater in UF, although typical function
groups of irreversible foulants have been always related to protein-like and polysaccharide-
like substances (Jarusutthirak and Amy, 2006; Shon et al., 2004). This is attributable to the
intrinsic property of the method which has no consideration of the size distribution of the fou-
lants despite the fact that these organic compounds have various molecular weight (size) dis-
tributions in different waters (Her et al., 2003; Garnier et al., 2005).
Removal of Organic Foulants in Bio-filtration Processes 59

5 Removal of Organic Foulants in Bio-filtration Processes

Bio-filtration is considered as a promising pre-treatment option to control membrane fouling
(Huck and Sozanski, 2008). In the present section the fouling control effects of short-term
bank filtration and slow sand filtration prior to UF are tested. Under varied operational condi-
tions the removal of biopolymers by slow sand filtration has been intensively investigated.

5.1 Fouling Control by Short-term Bank-filtration

To demonstrate fouling control effect using short-term bank-filtration, the UF pilot plant
equipped with the Inge module (Dizzer® 450) was set up behind an artificial lake to filter bio-
filtrate from July to November in 2005. Within this period the UF was operated at a flux of 89
LMH. Backwash interval (BWI) and backwash time (BWT) were 45 minutes and 30 seconds,
respectively. The backwash flux (BWF) was set at 260 LMH according to the instruction of
the membrane manufacturer. During each backwash, NaClO solution was dosed into the
backwash permeate to keep the concentration of active chlorine at around 10 mg/L.

1800
Trans-membrane pressure [mbar]

1200 chemical cleaning

600

phase1 phase2 phase3

0
7.

8.

8.

9.

9.

0.

0.

0.

1.
.0

.0

.0

.0

.0

.1

.1

.1

.1
27

15

28

12

24

07

18

29

09

Fig. 5-1 Trans-membrane pressure development during the experimental period in WWTP Beixiaohe
(July 27th to November 09th in 2005)

In the experimental period, TMP of the UF varied in different patterns depending on the char-
acteristics of influents through different filtration media (Fig. 5-1). In the first phase (July 27th
to August 15th) of the experiment, the influent to UF was a 1:1 mixture of sand and limestone
filtered water. Because of the impurity of limestone, effluent pH after limestone filtration was
always higher than ten and serious scaling was observed on the pipe wall to UF. Simultane-
ously, limestone filtration can only partly decrease humic substances (HS, retention time from
55 to 65 minutes, Fig. 5-2). The biopolymers (retention time from 35 to 55 minutes) could be
60 Removal of Organic Foulants in Bio-filtration Processes

hardly removed. In this phase scaling and organic fouling caused a steep TMP increase. As
shown in Fig. 5-1, TMP increased from 300 mbar to over 800 mbar within three days.

In phase 2 (August 16th to September 12th, 2005), the influent to UF was only sand filtrate.
Since the lake water contained high concentration of biopolymers (Fig. 5-2) and the biological
process in the sand seemed not totally adapt to removing them, TMP also increased quickly.
TMP increased much slower in phase 3 (September 12th to November 7th, 2005) compared to
phase 1 and 2. During this period, influent to UF was the effluent from sand filtration only. A
comparison of LC-OCD chromatogram of different water samples confirmed that in this
phase the biopolymer peak in the lake water was greatly reduced through sand bank filtration
prior to UF. With a low biopolymer concentration in the influent, UF could run more sustain-
ably under identical operational conditions. In this phase TMP increased around 400 mbar
within 55 days.

.
6

lake water

limestone filtrate

4
sand filtrate
OC-Signal

UF permeate

0
30 60 90
Elution time [min]

Fig. 5-2 LC-OCD diagram of lake water, effluent after limestone and sand filtration and UF permeate
(dilution 1:5)

These results demonstrate that the short-term bank filtration with silicate material performs
better than that with limestone material and can serve as an effective pre-treatment in fouling
control for UF. As it is operated at a comparatively low filtration rate (< 0.06 m/h), bio-
filtration with higher filtration rates (till to 0.5 m/h) was tested to identify the effect of filtra-
tion rate on fouling control. The experiments were conducted in WWTP Ruhleben (Berlin,
Germany).
Removal of Organic Foulants in Bio-filtration Processes 61

5.2 Fouling Control by Slow Sand Filtration

In WWTP Ruhleben the feedwater to the UF pilot plant was the filtrate from the slow sand
filter (SSFr) operated at a filtration rate of 0.25 m/h. After installation in September 2006, the
filter was continuously operated for about three months filtering SE to stabilize its bacterial
community and ensure constant removal of organic compounds. Since November 2006, the
treatment process showed a constant DOC removal.

1000

Secondary effluent
800

Slow sand filtrate

TMP [mbar]

600

400

200
0:00 3:00 6:00 9:00 12:00 15:00 18:00
Operating time [h]

Fig. 5-3 TMP development filtrating SE and SSF at a filtration rate of 0.25 m/h (Operational condi-
tion: flux 89 LMH, BWI 40 minutes, BWT 50 seconds, BWF 260 LMH)

12

secondary effluent in Ruhleben

8 SSF 0.25 m/h

OC-Signal

UF permeate of SSF 0.25 m/h

0
30 40 50 60 70 80 90 100
Elution time [min]

Fig. 5-4 LC-OCD diagram of SE, SSF (filtration rate at 0.25 m/h) and UF permeate of SSF (dilution
1:4)
62 Removal of Organic Foulants in Bio-filtration Processes

The effect of slow sand filtration on the performance of UF was investigated (Fig. 5-3). The
result exhibits that operated at identical operational conditions filtering SSF led to a slower
TMP improvement compared to treating SE. Due to the removal of particular, colloidal and
dissolved foulants (Fig. 4-1) - particularly the removal of biopolymers (Fig. 5-4), slow sand
filtration can improve the performance of UF significantly.

5.3 Organic Foulants Removal in Slow Sand Filtration

5.3.1 Operation of Pilot-scale Slow Sand Filters

The investigation of foulant removal during slow sand filtration was carried out from Sep-
tember 2006 to December 2007. In this period, SSFr Nr. 1 was continuously operated at a
filtration rate of 0.25 m/h; SSFr Nr. 2 was operated at a filtration rate of 0.5 m/h except the
period from February to July 2007, during which the filtration rate was 0.1 m/h. To recover
the permeability of slow sand filters, the top sand layer was removed by 3 to 5 cm and filled
with new sand after the water level in the supernatant was higher than 0.55 m. The time range
between sand change was defined as cycle time (days). To extend the cycle time of slow sand
filtration, a textile layer was put on to the surface of the filters and the influent was aerated
with a ratio between air and water at 1.4 : 1 Nm3/m3.

5.3.2 General Water Quality of Secondary Effluent and Slow Sand Filtrates
Tab. 5-1 Water quality of SE and SSFs on average (December 2006 to December 2007, SSF at 0.5
m/h: except February to July 2007)

Biopolymer DOC UV254 SUVA DO pH Turbidity

mg C/L mg/) 1/m L/(m·mg) mg/L NTU
SE 0.40±0.17 11.7±1.4 27.7±3 2.36±0.17 7.1±1 7.2±0.3 1.5±1.3
(n=54) (n=40) (n=45) (n=36) (n=30) (n=35) (n=44)

SSFr Nr.1 0.21±0.13 10.5±1.4 26.2±3 2.48±0.19 0.4±0.3 6.9±0.3 0.3±0.1

0.25 m/h (n=47) (n=40) (n=45) (n=34) (n=30) (n=35) (n=44)

SSFr Nr.2 0.22±0.14 10.6±1.3 26.4±3 2.51±0.18 1.5±1.2 6.9±0.3 0.3±0.1

0.5 m/h (n=25) (n=19) (n=25) (n=15) (n=14) (n=17) (n=23)

DO: dissolved oxygen

The two SSFrs began to operate in September 2006. After three months adaptation time both
showed constant DOC removal. The average water quality of SE and slow sand effluents
from December 2006 to December 2007 is shown in Tab. 5-1. The results show that during
the operation period slow sand filters removed biopolymers and DOC around 47 % and 10%
at both filtration rates, respectively. Humic substances were also removed as the UV254 ab-
sorption decreased about 5 - 6%. In contrast, specific UV absorption (SUVA) increased from
2.4 to around 2.5 L/(m·mg). The slight rise of SUVA indicates that after slow sand filtration
the proportion of humic substances in DOC increased. A significant decrease of dissolved
oxygen (DO) concentration (from 7.1 mg/L to 0.4 mg/L and 1.5 mg/L at filtration rates of
0.25 m/h and 0.5 m/h, respectively) after SSFn was observed. During the whole period SSFrs
Removal of Organic Foulants in Bio-filtration Processes 63

demonstrated effective removal of turbidity, the turbidity in SSF was always around 0.3 NTU
despite the large variations in influent.

5.3.3 Removal of Organic Compounds in Slow Sand Filtration

Removal Profile of Organics in Slow Sand Filter

To study the removal process of foulants in SSFrs different parameters next to biopolymer
concentration were measured simultaneously. The results investigated from March to No-
vember 2007 by SSFr. Nr.1 are shown in Fig. 5-5. The profile of biopolymer concentration
(Fig. 5-5 a, mean value (diamond) and standard deviation (horizontal line)) shows that in
SSFr the biopolymers were removed significantly in the upper 20 cm sand layer. In the filter
zone below this region they were also removed but to a much lower degree. The significant
reduction of DOC and humic substances (measured as UV254) occured also in the upper layer
(Fig. 5-5 b and d). In contrast, in this region SUVA increased from 2.33 to 2.42 L/(m·mg)
(Fig. 5-5 f) which indicates that the relative content of humic substances increases. The reduc-
tion of biopolymer concentration and the improvement of SUVA imply that in all of the DOC
fractions macromolecular polysaccharides and proteins are more preferentially removed than
aromatic and double-bond substances. Investigation on other parameters showed that in the
upper zone dissolved oxygen concentration decreased from around 7.5 mg/L to less than 0.5
mg/L (Fig. 5-5 b) and pH dropped also to some degree (Fig. 5-5 c). As conventional depth
filtration cannot remove dissolved substances effectively (Metcalf & Eddy, 2003), the correla-
tion of these phenomena implies that the removal of dissolved organic matters is related to
biological processes. The readily degradable organic substances are biologically oxidized
with the consumption of oxygen and lead to the formation of carbon dioxide and thus, the
reduction of pH. The humic substances which are normally considered as non-readily biode-
gradable organics are comparably persistent and this leads to only a slight increase of SUVA.

The LC-OCD diagram confirms the removal of different DOC fractions. As shown in Fig.
5-6, the comparison between the chromatograms of the SE and its SSF indicates that the most
fundamental change during slow sand filtration takes place within the fraction of biopolymers
(elution time at 35 – 55 min). They are significantly removed through slow sand filtration.
Also, a part of humic substances (second peak) could be removed which is consistent with the
results measured using UV254 (Fig. 5-5 d). Moreover, the chromatograms of the sand filtrate
and the water sample taken from 20 cm below the filter surface are almost identical, which
indicates that the removal of organic compounds takes place mostly at the upper sand zone.
This is consistent with the DOC measurement (Fig. 5-5 b).
64 Removal of Organic Foulants in Bio-filtration Processes

Biopolymer concentration [mg C/L] O2 and DOC concentration [mg/L]

0.0 0.2 0.4 0.6 0 4 8 12 16
Distance to surface [cm]

0 0

O2
30 30 DOC

60 60
a) b)

90 90

pH UV254 [1/m]
6.5 7.5 8.5 10 20 30 40
0 0
Distance to surface [cm]

30 30

60 c) 60 d)

90 90

Turbidity [NTU] SUVA [L/(m mg)]

0.0 1.0 2.0 2.0 2.2 2.4 2.6 2.8 3.0
0 0
Distance to surface [cm]

30 30

60 e) 60 f)

90 90

Fig. 5-5 Removal profile of a) biopolymer (n=13), b) dissolved oxygen and DOC (n=10), c) pH
(n=10), d) UV254 (n=12) and e) turbidity (n=12) f) SUVA (n=10) in slow sand filter operated at filtra-
tion of 0.25 m/h, period: March 2007 to November 2007, removal profile at filtration rate of 0.5 m/h is
shown in section 8.3, A-Fig. 5).

The significant removal of organics and turbidity (Fig. 5-5 e) in the upper sand layer is simi-
lar to the phenomenon in slow sand filtration for drinking water treatment (Ojha and Graham,
1996). In slow sand filtration it is normally considered that the schmutzdecke (thickness <
25mm) in association with the upper active layer plays an important role in removing con-
taminants (Amy et al., 2006). Head loss, maximum bio-film growth and fine particles removal
occur normally in this zone. Compared to slow sand filters used for drinking water treatment,
the removal of contaminants in treating SE also mostly occurred in the upper active layer
despite the high organic load and utilization of coarser sand (sand for drinking water
treatment ususally with grain size < 0.4 mm) (Rachwal et al., 1996). In this zone the
macromolecular proteins and polysaccharides can be effectively removed. The phenomenon is
similar to that taking place in water reclamation using bank-filtration or SAT system
(Gruenheid et al., 2005; Amy and Drewes, 2007). In those studies, the sustainable removal of
macro organic molecules was attributed to biodegradation.
Removal of Organic Foulants in Bio-filtration Processes 65

12
Humic substances

secondary effluent

9
slow sand filtrate
20 cm sample
OC-Signal

slow sand filtrate

80 cm sample
6

Low molecular weight

3
neutrals

Biopolymers

0
35 50 65 80 95
Elution time [min]
Fig. 5-6 LC-OCD chromatograms in different SSFr probe sites (SSFr 20 cm: the probe site 20 cm
below the sand filter surface)

Identification of Biological Degradation of Biopolymers

The biodegradability of biopolymers and humic substances was tested using the method de-
veloped by Servais et al. (1987) and Block et al. (1992) (detail information see section 8.4).
Fig. 5-7 shows the result of the biodegradability test. As displayed, after 30 days of inocula-
tion, around 45% of biopolymers (initial biopolymer concentration 0.4 mg C/L) can be bio-
logically degraded; simultaneously, humic substances were also degraded as the absorption
signal of UV254 (initial absorption 26.5 1/m) decreased around 8%. The reduction of these
organics is in good agreement with the result shown in (Tab. 5-1).

During the test, the degradation of biopolymers can be subdivided into three stages (Fig. 5-7).
In the initial stage, the biopolymer concentration decreased sharply - around 25% of the initial
value within one day of inoculation. Thereafter, the concentration increased from 0.3 to 0.37
mg C/L. Subsequently, it was decreasing continuously until the end of the test. The phenome-
non is similar to that in a biodegradability test of biofilm extracellular polymeric substances
(Zhang and Bishop, 2003). The initial decrease of the concentration of biopolymers can be
contributed to a fast degradation of easily degradable substances or biosorption (Esparza-Soto
and Westerhoff, 2003). Due to the production of extracellular biopolymers of inoculants (mi-
croorganisms) or desorption of absorbed substances, their concentration could rise to some
degree. Finally, biopolymers are consumed as substrates for microorganisms as they are under
a starvation situation (Zhang and Bishop, 2003; Flemming and Wingender, 2001) which leads
to the continuous decrease of biopolymer concentration.
66 Removal of Organic Foulants in Bio-filtration Processes

0.5 40

Biopolymer - blind
0.4

UV254 [1/m]
0.3 Biopolymer - inoculated
BC [mg C/L]

30

0.2 UV254 - blind

0.1 UV254 - inoculated

0 20
0 10 20 30
Sampling time [d]

Fig. 5-7 Biological degradation test of biopolymers and UV254-absorbing substances

Removal of the Protein-like Biopolymers

Next to the quantitative reduction of biopolymers, the impact of slow sand filtration on the
qualitative change of biopolymers, in terms of nitrogen to carbon ratio (N/C), is also investi-
gated. The N/C ratio of SE and SSF is within the range of 0.16 to 0.25 (n=10) and is altered
little after slow sand filtration. This implies that the nitrogen and carbon-containing com-
pounds are in general evenly removed by slow sand filtration. Suppose that the organic nitro-
gen is mostly from protein-like substances, the reduction of nitrogen content indicates their
removal. The EEM measurement of SE and corresponding SSF confirms the deduction. As
shown in Fig. 5-8, in both waters an obvious peak at 285-350 nm (Ex-Em) can be observed
which suggests the presence of protein-like substances (Henderson et al., 2009; Her et al.,
2003). Compared to that in the SE, the intensity of the peak in the SSF is significantly re-
duced. The decrease of the intensity reflects the removal of protein-like substances which is in
accordance with the reduction of organic nitrogen content in biopolymers detected using LC-
OCD-OND.

Fig. 5-8 Fluorescence EEM of fractioned biopolymer in (a) SE (b) corresponding SSF at a filtration
rate of 0.25 m/h (Em: Emission, Ex: Excitation)
Removal of Organic Foulants in Bio-filtration Processes 67

5.4 Factors Affecting the Performance of Slow Sand Filtration

5.4.1 Filter Ripening

As a biological and physical process, the ripening extent of slow sand filtration impacts its
performance (Amy et al., 2006). In terms of duration, filter ripening may take weeks to
months based on the influent water quality and the desired effluent quality (Weber-Shirk and
Chan, 2007). In the present work the filter ripening stage (adaption time) was three months
(September – November 2006). Within this period, slow sand filters achieved a comparatively
constant DOC removal.

Filter ripening can also be evaluated in regard to certain water quality parameters. With re-
spect to biopolymer removal, after three months adaption time SSFr.1 (0.25 m/h) achieved a
removal ratio around 30% from December 2006 to Feberary 2007. After a significant fluctua-
tion appeared from March to July 2007, the removal was constant at around 70% in warm
seasons. An extreme temperature drop appeared at the end of 2007 (Fig. 5-10), during which
the ratio fell to about 30%. It was indicated that under similar temperature conditions, the ra-
tio achieved after three months ripening time was comparable to that after one year of opera-
tion. This result demonstrated that three months filter ripening time were sufficient for the
microorganisms to adapt to biopolymers. Concerning turbidity control, SSFr.1 presented a
constant removal effect. The turbidity in the effluent was never higher than 0.6 NTU which
was independent of that in the influent (Fig. 5-9 b) throughout the experimental period.

(a) 2.7 (b) 9

turbidity in SE

Turbidity reduction by SSF [%]

BC in SE
biopolymer removal by SSF [%]

biopolymer removal by SSF turbidity in slow sand filtrate

90
turbidity reduction by SSF
BC in SE [mg C/L]

Turbidity [NTU]

1.8 6 90

45
0.9 3 45

0.0 0 0 0
29.11.06 09.03.07 17.06.07 25.09.07 03.01.08 29.11.06 09.03.07 17.06.07 25.09.07 03.01.08

Fig. 5-9 Removal effect by SSFr.1 in terms of (a) biopolymer, (b) turbidity due to slow sand filtration
(0.25 m/h)

5.4.2 Temperature and Biopolymer Concentration in Secondary Effluent

Variation of biopolymer concentrations in the SE was investigated within fifteen months (Oc-
tober 2006 to December 2007, Fig. 5-10). It is shown that the biopolymer concentration is
generally higher in winter time than in warmer seasons. In winter (November 2006 – Febru-
ary 2007, November 2007 – December 2007) the concentration was 0.55 mg C/L on average,
in warmer seasons (March – October 2007) it was around 0.31 mg C/L. In contrast to the
variation of biopolymer concentration, the DOC was comparatively constant during this pe-
riod. Throughout the whole sample period the variation of biopolymer concentration appears
to be associated with temperature change. As shown in Fig. 5-11, by trend biopolymer con-
centration decreased generally as the temperature increased.
68 Removal of Organic Foulants in Bio-filtration Processes

30 2.0
Temperature BC in SE
Biopolymer mass reduction Biopolymer percentage reduction

winter warm seasons winter

Biopolymer removal degree [ 0 - 1 ]

Biopolymer concentration [mg C/L]
1.5

20
Temperature [oC]

1.0

10

0.5

0 0.0
26.10.06 26.12.06 26.02.07 26.04.07 26.06.07 26.08.07 26.10.07 26.12.07

Fig. 5-10 Variation of temperature, biopolymer concentration in SE (October 2006 – December 2007)
and removal effect by SSFr.1 (filtration rate 0.25 m/h, December 2006 – December 2007, removal
degree: 0 - 1 represents 0 - 100%)

0.9

R2 = 0.55
BC in SE [mg/L]

0.6

0.3

0.0
10 15 20 25 30
Temperature [oC]

Fig. 5-11 Trend of biopolymer concentration in association with temperature in SE (October 2006 –
December 2007, n=63)

Similar effects on the variation of biopolymer concentration by temperature was observed by

Laabs (2004) operating two MBR systems treating domestic wastewater. Al-Halbouni et al.
(2008) noted seasonal variations of bound EPS content in a MBR treating municipal wastewa-
ter. Since either proteins, polysaccharides or biopolymers are mostly SMP (Jarusutthirak and
Amy, 2006) or synthetic/waste organic substances from wastewater (Barker and Stuckey,
1999; Al-Halbouni et al., 2009; section 8.3, A-Fig. 6) variation of temperature may put stress
on bacteria to produce more EPS/SMP and release them into the water phase or influence the
activity of microorganisms to degrade them.
Removal of Organic Foulants in Bio-filtration Processes 69

The influence of temperature and biopolymer concentration on biopolymer removal by SSFr

Nr. 1 is depicted in Fig. 5-10. The results show that from February to June 2007 the removal
of biopolymers was increased from around 34% to 65% while the temperature increased from
16 °C to 22 °C and the average biopolymer concentration decreased from 0.5 mg C/L to
around 0.4 mg C/L. During the period from July to November 2007, the biopolymer concen-
tration in the SE was around 0.3 mg C/L – even though the temperature decreased from 21 °C
to 15 °C – and the removal was rather constant at about 75 %. From the end of November to
the end of December 2007, as the temperature dropped to lower than 13 °C and the biopoly-
mer concentration jumped to over 0.6 mg C/L, the removal fell from 74% to below 35%. Si-
multaneously, the cycle time of the SSFr. 1 was dramatically reduced from around 80 days to
shorter than 18 days (Tab.5-2).

Based on the results shown in Fig. 5-10, temperature or biopolymer concentration results are
separately plotted versus biopolymer removal (Fig. 5-12). The results show that either tem-
perature or biopolymer concentration in the SE impacts the biopolymer removal in slow sand
filtration. Higher temperatures (Fig. 5-12 a) or lower biopolymer concentrations (Fig. 5-12 b)
result in higher biopolymer removal. In general, SSFr Nr. 1 could control biopolymer effec-
tively, with a removal over 65% on average during warm seasons (Fig. 5-12 c, around 20 °C,
n=30) and comparatively low biopolymer concentrations (0.15 – 0.45 mg C/L). At lower tem-
peratures (Fig. 5-12 c, on average 15 °C, n=16) and higher biopolymer concentrations (> 0.4
mg C/L), the removal was lower than 40%.

(a) 90 (b) 100

Biopolymer removal [%]

Biopolymer removal [%]

60 R2 = 0.39

50

30
R2 = 0.67

0 0
10 15 20 25 0.0 0.3 0.6 0.9
Temperature [oC] BC in SE [mg C/L]

(c) 0.50
14 Mar. - 30 Okt. 2007, T 20.2 ºC

24 Nov. 2006 - 08 Mar. 2007

BC reduction by SSF [mg C/L]

+ 01 - 21 Nov.2007, T 15.2 ºC

0.25

0.00
0.0 0.2 0.4 0.6 0.8
BC in SE [mg C/L]

Fig. 5-12 Influence of a) temperature, b) biopolymer concentration and c) temperature and biopolymer
in the SE on biopolymer removal by SSFn (filtration rate 0.25 m/h, Dec. 2006 – Dec. 2007, n=46)
70 Removal of Organic Foulants in Bio-filtration Processes

As the percentage removal of biopolymers depends highly on the feedwater quality, a further
investigation on their mass reduction under different conditions was carried out (Fig. 5-10 and
Fig. 5-12 c). It is evident that with a similar feedwater quality higher temperature promotes
more biopolymers removal (Fig. 5-12 c). The results indicate that increasing temperature can
create a more suitable circumstance to the microorganism and therefore, they can remove
more biopolymers. Furthermore, it is observed that an increase in biopolymer content in the
feedwater within a certain scope results in an almost proportional rise in their mass removal
(Fig. 5-11 c). Nevertheless, the rise is much less than linear when the biopolymer concentra-
tion is beyond a certain value. For instance, at the end of 2006 and 2007 the variation of bio-
polymer content from 0.5 to 0.8 mg C/L resulted in a similar mass removal by 0.21 mg C/L,
which is also similar to the mass removal in April and July 2007 (Fig. 5-10). These phenom-
ena suggest that overloaded biopolymers cannot be effectively degraded, which is attributable
to a restricted degrading activity of microorganisms. At this point, environmental factors, e.g.,
available oxygen, surface area available for biofilms (Metcalf & Eddy, 2003; Huck and So-
zanski, 2008) may play an important role.

Tab. 5-2 Performance of bio-filtration operated under different temperature and biopolymer concen-
tration in SE

Dec.01.2006 Feb.10.2007 Jul.17.2007 Nov.20.2007

-Feb.07.2007 -Jul.10.2007 -Nov.13.2007 -Dec.28.2007
o
Average temperature ( C) 16.8 (n=11) 19.6 (n=25) 19.1 (n=19) 13.1 (n=9)
Average BC in SE (mg C/L) 0.46 (n=11) 0.33 (n=25) 0.3 (n=19) 0.68 (n=9)
Biopolymer SSFr Nr.1 (0.25 m/h) 37% (n=6) 58% (n=15) 73% (n=15) 32% (n=5)
removal SSFr Nr.2 (0.5 m/h) 24% (n=8) -- 67% (n=17) 28% (n=3)

BioN+SSFn (0.5 m/h) -- -- 45% (n=6)* 31% (n=5)**

Cycle time SSFr Nr.1 (0.25m/h) ~ 40 ~ 80 ~ 80 ~ 18
(days) SSFr Nr.2 (0.5 m/h) ~ 20 -- ~ 30 ~8
BioN+SSFn (0.5 m/h) -- -- > 120 (test till March 2008)
BC: biopolymer concentration
BioN: BioNet as filter material
Cycle time: the period between upper sand layer cleaning
*Water samples from September to November 2007
**Water samples from November 2007 to January 2008

Filtration Rate

Tab. 5-2 shows the biopolymer removal and cycle time of SSFr Nr. 1 (filtration rate 0.25 m/h)
and SSFr Nr. 2 (filtration rate 0.5 m/h) during the experimental period. The results indicate
that during winter time in 2006, biopolymer removal at a filtration rate of 0.25 m/h and 0.5
m/h was around 37% and 24%, in the warmer seasons in 2007 the removal was 73% and
67%, respectively. As lower filtration rates lead to longer empty bed contact times (EBCT),
higher organic removal effects can be achieved (Huck and Sozanski, 2008; Urfer et al., 1997).
The comparison of biopolymer removal profile within slow sand filters operated at different
filtration rates is shown in Fig. 5-13 (a) and (c). It is evident that at lower filtration rates the
Removal of Organic Foulants in Bio-filtration Processes 71

active layer in slow sand filter can remove more biopolymers due to the prolonging of EBCT.
At similar EBCT similar biopolymer removal efficiency can be achieved (Fig. 5-13 b and d).

(a) Normalized biopolymer concentraton [%] (b) Normalized biopolymer concentraton [%]
0 20 40 60 80 100 0 20 40 60 80 100
Distance to surface [cm]

0 0

EBCT [min]
50
30
SSF 0.5 m/h 100 SSF 0.5 m/h
SSF 0.25 m/h SSF 0.25 m/h
60
150

90 200

(c) Normalized biopolymer concentration [%] (d) Normalized biopolymer concentration [%]
0 20 40 60 80 100 0 20 40 60 80 100
0 0
Distance to surface [cm]

EBCT [min]

100
30
200
SSF 0.1 m/h SSF 0.1 m/h
SSF 0.25 m/h SSF 0.25 m/h
300
60
400

90 500

Fig. 5-13 Normalized biopolymer concentration profile within slow sand filters operated at filtration
rate of 0.5 and 0.25 m/h in terms of (a) depth and (b) contact time (test period: July to November
2007, biopolymer n=13 for SSF 0.5 m/h and n= 6 for SSF 0.25 m/h), operated at filtration of 0.25 and
0.1 m/h in terms of (c) depth and (d) contact time (test period: April to June 2007, n = 6 for both fil-
ters)

Although a higher biopolymer removal effect can be obtained by reducing the filtration rate, a
decrease of filtration speed cannot lead to a proportional increase of the removal efficiency.
The investigation carried out at filtration rates from 0.05 m/h to 0.5 m/h from November 2006
to July 2007 (Fig. 5-14) shows that biopolymer removal rates are improved from 32 % to
50 % when filtration rates are decreased from 0.5 m/h to 0.25 m/h. But if filtration rates are
reduced from 0.1 m/h to 0.05 m/h, corresponding biopolymer removal rate increases only
from 62 % to 65 %. It has been indicated (Zhang and Huck, 1996) that although the filtration
rate (based on EBCT) is a key parameter determining organics removal effect in bio-filtration
processes, other factors, e.g. reactor specific surface area, biomass density, substrate diffusiv-
ity and kinetic parameters, are also important. The variation of one parameter cannot result in
a proportional change on the performance of the bio-filtration. Therefore, an unified parame-
ter, e.g. dimensionless contact time, which contains other factors important for biofiltration
performance is more suitable to describe and optimize the bio-filtration process (Huck and
Sozanski, 2008).
72 Removal of Organic Foulants in Bio-filtration Processes

100

80

60

removal [%] 40

20

0
-1 -1 -1 -1
0.05m h 0.1m h 0.25m h 0.5m h
(n=10) (n=12) (n=12) (n=10)

Fig. 5-14 Biopolymer removal at different filtration rates (November 2006 to July 2007 in WWTP
Ruhleben, Median value (horizontal centre line), average value (the square in box), the 25%- and
75%-quartiles (box) and the minimum and maximum values (vertical line))

The influence of filtration rates on the cycle time was much more significant than on bio-
polymer removal. In the investigated period, the cycle time of SSFr Nr. 2 was always around
50% of that of SSFr Nr. 1 (Tab.5-2). During operation, the increase of hydraulic resistance is
observed mainly in the upper sand layer (Fig. 5-15) and is considered generally as a result of
the accumulation of particles and the growth of microorganisms (Leverenz et al., 2009;
Maelzer and Gimbel, 2006). Compared to the active upper sand layer, the layer deeper than
0.2 m presents little variation in terms of water headloss during the operational period (Fig.
5-15). It implies that due to the removal of particles and biopolymers in the upper layer, the
deeper layer works more like a support layer than as a foulants removal layer. The difference
of cycle time at different filtration rates indicates that higher filtration rates lead to higher sur-
face loading rate within the top zone. This results in subsequently more accumulation of sub-
stances (Leverenz et al., 2009) pre unit filtration area and leads to a faster increase in hydrau-
lic resistance.

(a) 0.6 (b) 0.6

Apr.18.2007 Sep.23 2007
Water head [m]

Water head [m]

Apr.12.2007
0.4 0.4 Sep.09 2007
Mar.21.2007
Aug.28 2007
Feb.28.2007
0.2 Jan.17.2007 0.2 Aug.07 2007

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Filter depth [m] Filter depth [m]

Fig. 5-15 Water head in (a) SSFr Nr. 1 operated at a filtration rate of 0.25 m/h during a cycle time
from Jan.17 to Apr.18 2007 (b) SSFr Nr. 2 operated at a filtration rate of 0.5 m/h during a cycle time
from Aug.07 to Sep.23 2007
Removal of Organic Foulants in Bio-filtration Processes 73

5.4.3 Protective Layer

To extend the cycle time of slow sand filtration at high filtration rates, e.g., 0.5 m/h, a protec-
tive layer consisting of BioNet was put on the sand layer surface (Maelzer and Gimbel, 2006)
in SSFr Nr. 3. A 0.2 m thick BioNet layer is constructed above a 0.5 m sand layer. The dual-
media filter was operated from July 2007 to March 2008. After two months adaption time, the
biopolymer removal effect was noted (Tab.5-2).

The operational results show an astonishing extension of the cycle time due to the protect
layer. During the experiment period (July 2007 to March 2008), the filter was only cleaned
once. The cycle time is longer than four months under the present operational condition.
Compared to a slow sand filter, the combined BioNet and sand filter can remove biopolymers
to an extent of about 45 % from September to November 2007. The removal rate is lower than
that achieved by the SSFr Nr. 2 at the same filtration rate. This might be due to the much
shorter operation time of the filter. After several months of operation, the combined BioNet
and sand filter achieved similar biopolymer removal effect as the SSFr Nr. 2 during the cold
season (November to December 2007). As BioNet carriers are made of porous compressible
material, they can offer large surface areas for adherence and growth of microorganisms and
also large void volumes for the retention of colloids and particles. With the removal of bio-
polymers, colloids and particles in the BioNet layer, the loads to the sand layer could be sig-
nificantly reduced. Therefore, the permeability of the sand layer can be maintained for a
longer time.

5.5 Discussion and Summary

5.5.1 Biopolymer Removal in Slow Sand Filtration

Slow sand filtration removed particles, colloids and biopolymers from SE and improved the
performance of UF to a large extent. Within the slow sand filter, the removal of biopolymers
was demonstrated to be related to the reduction of oxygen concentration and the decrease of
pH. As dissolved substances cannot be retained due to straining by the filter material, their
removal was attributable to biological processes (Morales et al., 2007). Further tests demon-
strated that biopolymers are biodegradable. It indicated that the removal of biopolymers
within slow sand filter depended on biodegradation. Nevertheless, it should be noted that in
biological processes biosorption of high molecular weight substances to live biomass is a pre-
liminary step previous to enzymatic hydrolysis breakdown and biological uptake (Esparza-
Soto and Westerhoff, 2003). Therefore, organic compounds are also removed from water
phase due to biosorption. But, due to the exhaustion of adsorption capacity no adsorption sys-
tem can offer a sustainable removal effect. In an investigation on macromolecular organics
removal through bioadsorbtion to live biomass (Esparza-Soto and Westerhoff, 2003), it was
observed that within several hours the bioadsorption reached a steady state and no more re-
moval effect could be achieved. By contrast, the biofilters could remove biopolymers con-
tinuously for hundred to thousand hours at a constant rate. This phenomenon indicated that
although biosorption is a necessary step transporting biopolymers from water phase to bio-
mass, the ultimate elimination of these substances was biodegradation.
74 Removal of Organic Foulants in Bio-filtration Processes

Due to biological removal, biopolymer content can be reduced in terms of organic carbon or
nitrogen. The N/C ratio of biopolymers showed little variation after slow sand filtration. It
seemed that the carbon and nitrogen-containing compounds were mostly evenly removed. The
removal of nitrogen compounds by slow sand filtration was further validated based on the
EEM spectra measurement of fractioned biopolymers in feedwater and effluent. Nevertheless,
several studies reported that the protein-like substances (generally organic nitrogen-
containing organics) are preferably removed by biofiltration than polysaccharides. For in-
stance, based on the comparison of EEM spectra and LC-OCD chromatograms of SE and
SSF, Haberkamp (2008) stated that protein-like substances were more significantly removed
by slow sand filtration than other biopolymer compounds. This phenomenon was attributed to
their higher biodegradability compared to polysaccharides (Flemming and Wingernder,
2002). According to this deduction, the N/C ratio of biopolymers after slow sand filtration
should decrease, which is inconsistent to the observation in the present work. The conflicting
results may lie on the difference in protein measurement scope. In the present study, the
measurement focused only on the proteins within biopolymers as water sample fractionation
using chromatography was conducted prior to LC-OCD-OND and fluorescence EEM. In the
study undertake by Haberkamp (2008), all the proteins removed by slow sand filtration were
detected by the fluorescence EEM measurement, which were considered as a component of
biopolymers. Therefore, a higher removal ratio of proteins can be deduced.

5.5.2 Influencing Factors on the Performance of Slow Sand Filtration

The removal of biopolymers as well as other DOC fractions took place mostly in the active
upper sand layer. In the deeper sand layer, they could be further reduced, but to a much
slower rate. This phenomenon is similar to that in several studies investigating DOC removal
in bank filtration and/or artificial recharge (Gruenheid et al., 2005; Brinkmann et al., 2004;
Hua et al., 2003). Having identified biopolymer removal mainly due to biological processes,
the factors such as temperature, biopolymer concentration in feedwater and filtration rate (de-
pends on EBCT) are considered to influence its performance.

As a primary environmental factor, temperature affects the microbial growth and activity in
slow sand filter. Generally, a higher temperature leads to a more suitable circumstance to the
microorganism and therefore, can result in a higher biopolymer removal. Additionally, it was
evident that temperature impacted the biopolymer concentration in SE, an increase of tem-
perature led to a decrease of biopolymer concentration. To understand the end effect of tem-
perature on biopolymer removal, a more direct indicator is the removed biopolymer mass. As
shown in Fig. 5-12 (c), it is evident that under higher temperature more biopolymer can be
removed when biopolymer concentration in SE is similar.

Besides biopolymer removal, temperature seemed to influence the cycle time of the filters.
Lower temperature led always to a shorter running time due to filter clogging at the upper
sand layer. A similar phenomenon was observed by Le Binhan and Lessard (2000) while they
studied biofilter clogging in treating snowmelt wastewater. The reduction of the operating
time was primarily attributed to a weakened hydrolytic activity of the bacteria due to low
temperature. This led to the abundance of EPS in biofilm which had a clogging effect in the
biofilter. In the present work, slow sand filter received more biopolymers (EPS) in cold time
Removal of Organic Foulants in Bio-filtration Processes 75

than in warm seasons. It has been shown that the accumulated biopolymer cannot be removed
by bacteria as quickly as under higher temperatures. Thus, filter clogging appears more fre-
quently in winter.

To control slow sand filter clogging at low temperatures, the application of a protective layer,
e.g., BioNet, was identified as an effective method. The use of BioNet could not improve the
removal effect of biopolymers, but offers large void space for the accumulation of bacteria as
well as biofilm and particles. Therefore, the biofilm inhabited not only in the sand layer, but
also in BioNet structure. This leads to a partial removal of biopolymers before they reach the
sand filter, and may reduce the accumulation of biofilm. The permeability of the upper sand
zone can thus be maintained for a much longer time. A further verfication of the effect of a
protective layer consisting of BioNet has been conducted in pilot-scale. Operated at a filtra-
tion rate of 0.25 m/h, the biofilter has worked from Febrary 2009 to February 2010 without
sand scraping.

Despite the fluctuation of biopolymer concentration throughout the whole experimental pe-
riod, a significant correlation between the concentration in the feedwater and the amount re-
moved by slow sand filtration (0.25 m/h, Fig. 5-12 b) could be observed when biopolymer
content is within a certain range, such as lower than 0.4 mg C/L observed in the present work.
This correlation is reproducible at different temperature scopes (Fig. 5-12 c). The proportional
relevance suggests that the removal of biopolymer may be a first order process (Rooklidge et
al., 2005; Huck and Anderson, 1992). Nevertheless, there was no unified apparent first order
constant found suitable to describe the biopolymer removal rate. The major reason for this
phenomenon is that biopolymers contain components with different properties, e.g., proteins,
polysaccharides, etc. During biodegradation they present different removal rates. A first order
constant describing the removal rate of a signal substance cannot be adapted to present the
rate of a mixture with different biodegradation characteristics (Gruenheid et al., 2008). An
overload of biopolymer mass to the filter changes the removal to a zero order process. As the
substrates (biopolymers) are always available, the almost exhaustion of oxygen after active
layer suggests that oxygen availability may restrict the biomass activity significantly.

The performance of slow sand filtration is also impacted by operating variables in which
EBCT is considered to be a key parameter controlling the removal effect (Huck and Sozanski,
2008; Urfer et al., 1997). The EBCT in the present work was changed by the variation of hy-
draulic loading (filtration rate) for the filters had a similar filter depth. For a given EBCT, as
shown in Fig. 5-13 (b and d), biopolymer removal was little dependent on hydraulic loading
in the range of applied filtration rates (0.1 to 0.5 m/h). This empirical finding suggests that
external mass transfer may play a minor role in biopolymer removal under conditions applied
in the present work because for a given EBCT the increase in flow velocity did not increase
the percentage of removal (Urfer et al., 1997). Moreover, the decrease of filtration rates from
0.5 to 0.05 m/h resulted in an improvement in biopolymer removal. This is mainly attributed
to the manipulation of the duration of interaction between biofilm and organic substance
(Simpson, 2008). Nevertheless, the correlation was less than proportional (Fig. 5-14). It seems
that over a certain level, further increase of EBCT offers only a very limited improvement of
biopolymer removal.
76 Pilot-scale Ultrafiltration of Bio-filtrate

6 Pilot-scale Ultrafiltration of Bio-filtrate

Although the application of biofiltration prior to UF improved the performance of membrane
to a large extent, it was observed that TMP development in UF was significantly influenced
by operational conditions. The effect of filtration force (flux), filtration duration (backwash
interval: BWI), backwash duration (backwash time: BWT) and chemically enhanced back-
wash (CEBW) were investigated using a pilot-scale UF plant. The impact of filtration flux on
fouling was further explored in lab-scale tests. Several cleaning reagents were tested to restore
the permeability of the fouled membrane.

6.1 Overview of the Experiments

In the previous chapter it is evident that due to the reduction of foulant content by biofiltration
the performance of UF can be improved significantly. Nevertheless, UF treating slow sand
effluent leads to a fast TMP increase when the plant was operated under high permeate flux
and long BWI, e.g., flux 89 LMH, BWI 40 minutes, the TMP increased from 300 to 700 mbar
within 16 hours (Fig. 5-3). Subsequent chemical cleaning had to be carried out to restore the
permeability of the membrane. To slow down TMP development, the effect of operational
conditions needs to be investigated.

The investigation was carried out from November 2006 to August 2007. An overview of the
tests is displayed in Tab. 6-1. During this period, the effect of flux, BWI, BWT and CEBW on
TMP development rate were tested. In each experiment, a test ended as soon as the TMP
reached a specific level (maximum 800 mbar per manufacturer’s guidelines). The time the
plant operated from the beginning until the TMP increased to the specific value is defined as
operating time. The corresponding TMP development rate is calculated in terms of per unit
permeate volume using Eq.6-1:

TMPend − TMPstart
RTMP = Eq.6-1
V
where
RTMP = TMP increase per m3 permeate, mbar/m3
V = permeate volume, m3
TMPend = TMP at the end of each experiment, mbar
TMPstart = TMP at the beginning of each experiment, mbar

The TMP data used in the present work was the TMP value (mbar) after 5 minutes filtration
in each filtration cycle at identical times on each operating day. After each experiment the
module was subsequently cleaned using NaClO (active chlorine 50 mg/L) and NaOH (pH 12)
/ H2SO4 (pH 2) solution until the permeability (tested with tap water) was recovered to over
90% of the initial value.
Pilot-scale Ultrafiltration of Bio-filtrate 77

Tab. 6-1 Influence of operational conditions on the performance of UF filtering SSF (BWF was always 260 LMH during the whole experiment)

Comprison Experiment date Flux TMP BWI BWT BWT arrangement Permeate Produc- RTMP Biopolymer Turbidity
3
increase volume tivity (mbar/m ) Concentration
3
(LMH) (mbar) (min.) (s) (FF/BW/FF/still - s) (m ) (%) (mg C/L) (NTU)

Flux 02 15 - 02 26 2007 50 228 - 280 10 20 10 / 20 / 10 / 0 51 83 1.2 0.19 0.22

07 29 - 08 04 2007 65 259 - 709 10 20 10 / 20 / 10 / 0 40 87 13.1 0.12 -
04 18 - 04 20 2007 80 284 - 708 10 20 10 / 20 / 10 / 0 12 89 39.1 0.17 0.2

BWI 02 15 - 02 26 2007 50 228 - 280 10 20 10 / 20 / 10 / 0 51 83 1.2 0.19 0.22

02 28 - 03 10 2007 50 216 - 676 20 20 10 / 20 / 10 / 0 58 92 8.7 0.32 0.38
04 03 - 04 05 2007 50 268 - 712 40 20 10 / 20 / 10 / 0 11 96 43.7 0.19 0.27

BWT 03 14 - 04 02 2007 50 250 - 570 20 40 10 / 40 / 10 / 0 57 83 6.8 0.18 0.29

02 28 - 03 10 2007 50 216 - 676 20 20 10 / 20 / 10 / 0 58 92 8.7 0.32 0.38
04 03 - 04 05 2007 50 268 - 712 40 20 10 / 20 / 10 / 0 11 96 43.7 0.19 0.27
04 12 - 04 14 2007 50 300 - 720 40 60 10 / 60 / 10 / 0 11 87 46.1 0.17 0.3
CEBW
Permeate 11 07 2006 89 308 - 660 40 50 10 / 50/ 10 / 300 6 94 78.3 0.2 -
NaOH 08 22 – 08 29 2007 89 300 - 658 40 50 10 / 50/ 10 / 300 61 94 7.5 0.1 0.28
H2O2 09 17 – 10 18 2007 89 195 - 405 40 50 10 / 50/ 10 / 300 270 94 1.7 0.11 0.2
NaClO 11 20 2006 – 01 23 89 210 - 390 40 50 10 / 50/ 10 / 300 561 94 0.8 0.29 0.31
2007
BWI: backwash interval
BWT: backwash time
CEBW: chemically enhanced backwash
FF: forward flush
BW: backwash
78 Pilot-scale Ultrafiltration of Bio-filtrate

6.2 Influence of Flux on Fouling

6.2.1 Pilot-scale Investigation

The effect of operating flux was investigated by filtering SSF at 50, 65 and 80 LMH. Each
experiment was performed at BWI 10 minutes and BWT 20 seconds, the permeability of the
membrane at the beginning of each experiment (Kw0) was around 260 L/(m2·h· ·bar) (tested
with tap water, normalized to at 20°C) which was lower than that of a new membrane as it
was chemically cleaned in previous tests. Fig. 6-1 (a) shows the result of TMP development
with respect to operating time at different fluxes. At 80 LMH TMP increased from 284 to 708
mbar within 40 hours. At 65 LMH the operating time was extended to longer than 120 hours.
When the flux was reduced to 50 LMH, TMP increased only 50 mbar within 240 hours. In
terms of permeability reduction, the declining slopes indicate clearly that Kw decreased with
time/permeate volume at each flux. The comparison of permeability decline rates at different
operational conditions illustrated that the permeability decreased steadily as operating flux
increased, either due to operating time (Fig. 6-1 b) or permeate volume (Fig. 6-1 c).

(a) 900 (b) 1.2

Flux 50 LMH Flux 50 LMH Flux 65 LMH Flux 80 LMH

Flux 65 LMH
600 0.8
TMP [mbar]

Kw / Kw0

Flux 80 LMH slope = -0.0011

300 0.4
slope = -0.0066

slope = -0.0225
0 0
0 60 120 180 240 0 60 120 180 240
Operating time [h] Operating time [h]

(c) 1.2 (d)

Flux 50 LMH Flux 65 LMH Flux 80 LMH Flux 50 LMH Flux 65 LMH Flux 80 LMH
1
Normalized permeability

0.8
Kw / Kw0

slope = -0.0063 slope = -0.36

0.5
0.4
slope = -0.0253 slope = -0.76
slope = -0.0747 slope = -1.98
0 0
0 10 20 30 40 50 0 0.5 1 1.5 2
Permeate volume [m3] Biopolymer load [g C/m2]

Fig. 6-1 (a) TMP development vs. operating time at different fluxes, (b) Kw/Kw0 vs. operating time,
(c) Kw/Kw0 vs. permeate volume, (d) Kw/Kw0 vs. biopolymer load mass

The more severe decrease of permeability at higher flux is primarily attributed to a higher
foulant mass load (biopolymers mass load) during a filtration cycle which can lead to more
irreversible fouling. According to the water quality data shown in Tab. 6-1, more biopolymer
was delivered to the membrane at a flux of 80 LMH than at 50 LMH within a filtration cycle.
The more severe fouling can be attributed to the biopolymer mass load.
Pilot-scale Ultrafiltration of Bio-filtrate 79

However, compared to operation at a flux of 50 LMH, filtering SSF at a flux of 65 LMH de-
livered less biopolymer to the membrane in each filtration cycle due to a lower biopolymer
concentration in the feed water (Tab. 6-1), the permeability of the membrane decreased
sharper, either versus operating volume or biopolymer load mass (Fig. 6-1 c and d). This phe-
nomenon cannot be interpreted according to biopolymer load mass. Therefore, lab-scale ex-
periments were carried out investigating flux on fouling development.

6.2.2 Effect of Filtration Flux on Biopolymer Fouling

Influence of TMP on Biopolymer Fouling Effect
(a)
at 1 bar at 0.5 bar
1.0
Normalized flux [J/J0]

0.5

0.0
0 500 1000 1500 2000 2500 3000
Permeate volume [mL]

(b)
at 1 bar at 0.5 bar

100
Reversibility [%]

50

0
cycle 1 cycle 2 cycle 3 cycle 4 cycle5

Fig. 6-2 (a) Influence of operating pressure on fouling effect, (b) flux recovery at different operating
pressures (feed water is SE after 0.45µm filter with a biopolymer concentration of 0.42 mg C/L, oper-
ating pressure at 1 bar and 0.5 bar, permeate volume 500 mL; back wash using salt solution 50 mL at 1
bar)

Lab-scale multi cycle filtration tests, each operated at constant pressure throughout the test,
were used to investigate the effect of different operating pressures. The operation was con-
ducted by filtering identical SE samples at 1 bar and 0.5 bar. The results in Fig. 6-2 (a) show
that each filtration cycle operated at 1 bar leads to a more serious flux decline than at 0.5 bar;
the correspondingly formed fouling is also more irreversible as the reversibility is always
lower than that formed at lower operating pressure (Fig. 6-2 b). As the feedwater is identical,
the difference in performance of the UF was interpreted by comparing the biopolymer mass in
 80 Pilot-scale Ultrafiltration of Bio-filtrate

the fouling layer under the two different pressures. Based on a mass balance, the biopolymer
mass in each fouling layer can be calculated (the dry weight of biopolymers is assumed to be
twice their carbon content as carbon accounts for around 50 % of the mass of the protein-like
and polysaccharide-like substances (Thullner et al., 2002)). The comparison (Tab. 6-2) shows
little difference in biopolymer mass at the two operating pressures. Further analysis was then
focused on the fouling mechanisms at different filtration pressures. The results shown in Fig.
4-13 (c) indicated that during each filtration at 0.5 bar the predominant fouling mechanism
was cake/gel filtration which was similar to that at a filtration pressure of 1 bar (Fig. 4-13 d).
The difference in the performance of the UF thus needs to be further investigated, taking into
consideration the structure of the fouling layer as influenced by the filtration pressure.

Tab. 6-2 Calculated biopolymer mass content in formed cake/gel layer and the corresponding specific
fouling resistance (biopolymer concentration in SE is 0.42 mg C/L, permeate volume 0.5 L, super-
natant 0.2 L).

cycle1 cycle2 cycle3 cycle4 cycle5 Average

Biopolymer in fouling layer at 1 bar(mg) 0.27 0.28 0.27 0.30 0.27 0.28
Biopolymer in fouling layer at 0.5bar (mg) 0.27 0.29 0.29 0.25 0.26 0.27

Specific resistance α1 at 1.0 bar (1016 m/kg) 1.97 1.93 2.0 1.83 2.13 1.98

Specific resistance α 0.5 at 0.5 bar (1016 m/kg) 1.40 1.26 1.25 1.58 1.51 1.40

Based on the experimental results displayed in Fig. 6-2 (a), the specific fouling resistance of
the biopolymer fouling layer was calculated according to Eq.2-5 and Eq.2-9. The calculated
results (Tab. 6-2) show that under the present experimental conditions the specific fouling
resistance of the biopolymer fouling layer is about 1.4 x 1016 and 2.0 x 1016 m/kg at operating
pressure 0.5 bar and 1 bar, respectively. The formed biopolymer fouling layer is moderately
compressible as the calculated compressibility was around 0.54 at corresponding pressure
range (calculation according to Eq.2-11 based on the results shown in Fig. 6-2 (a). The values
are similar to the results presented by Lodge et al. (2004). In their research a UF membrane
similar to the one used in the present study was operated to treat wastewater effluent.

The comparison of the reversibility of fouling formed at TMP 1 bar and 0.5 bar is shown in
Fig. 6-2 (b). As displayed, during the five cycles of the filtration process the declined mem-
brane permeability at 1 bar can always be restored to a lower extent than that under 0.5 bar.
The phenomenon indicates that the permeability reduction caused at a higher filtration pres-
sure is more difficult to be recovered. As higher operating pressure leads to a more condensed
fouling layer and results in a higher α, the stronger compressed biopolymer fouling layer leads
to a more serious normalized flux decline and a decrease of the reversibility.

Validation of the Effect of Compression on Biopolymer Fouling

Further tests filtering SSF with a biopolymer concentration of 0.51 mg C/L were conducted to
validate the reproducibility of the effect of operating pressure. The filtration was carried out at
1.0, 0.8, 0.6 and 0.4 bar, using five filtration cycles (Fig. 6-3).
Pilot-scale Ultrafiltration of Bio-filtrate 81

(a)
at 0.4 bar at 0.6 bar at 0.8 bar at 1.0 bar
1
Normalized flux [J/J0]

0.5

0
0 500 1000 1500 2000 2500 3000
Permeate volume [mL]

(b)
0.4 bar 0.6 bar 0.8 bar 1.0 bar

100
Reversibility [%]

50

0
cycle 1 cycle 2 cycle 3 cycle 4 cycle 5

(c) 16.4 (d) 84

0.8 bar - 1 bar
s = 0.72 at 0.4 bar
at 0.6 bar
Reversibility [%]

78
at 0.8 bar
log(a)

16.2 0.4 bar - 0.6 bar 0.6 bar - 0.8 bar

s = 0.61 s = 0.66 72
at 1 bar

66
16
1.1E+16 1.65E+16 2.2E+16
3.55 3.8 4.05
Specific fouling resistance [m/kg]
log ( P)

Fig. 6-3 (a) Influence of operating pressure on fouling effect, (b) flux recovery at different operating
pressures, (c) compressibility under different pressure ranges, (d) correlation between specific fouling
resistance and reversibility of corresponding fouling (feed water is slow sand effluent with biopolymer
concentration of 0.51 mg C/L, operating pressure at 1.4 bar, 1 bar and 0.5 bar, permeate volume 0.5 L;
back wash using salt solution 0.05 L at 1 bar)

As expected, the results show that the calculated α (Fig. 6-3 c) is comparable to the value
shown in the previous experiment (Tab. 6-2) at operating pressure from 0.5 bar to 1 bar. Si-
multaneously, α values also show a correlation with the reversibility (Fig. 6-3 d). This correla-
tion verifies the observation that more compact biopolymer fouling layers decrease the effi-
ciency of backwash and the layer is less reversible. More over, the compressibility calculated
at 0.4 to 0.6 bar, 0.6 to 0.8 bar and 0.8 to 1 bar increased with the increase of operating pres-
82 Pilot-scale Ultrafiltration of Bio-filtrate

sures (Fig. 6-3 c). This indicates that at higher pressures the biopolymer fouling layer can be
further compressed (Foley, 2006). This is also consistent with the increase of specific fouling
resistance results.

6.3 Influence of Biopolymer Load Mass on Fouling Development

6.3.1 Impact of Backwash Interval

The foulant load mass can be changed by backwash interval or due to the variation of foulant
concentration in feedwater. The influence of BWI on TMP development was tested at a flux
of 50 LMH and a backwashing time of 20 seconds. In each experiment, BWI was set at 10, 20
or 40 minutes. At the beginning of a test the specific permeability of the membrane was
around 260 L/(m2·h·bar) (tested with tap water, normalized to at 20°C). Operational condi-
tions, average biopolymer concentration and turbidity in feed water are shown in Tab. 6-1.

(a) 900 (b) 1.2

BWI 40 min BWI 10 min BWI 20 min BWI 40 min
BWI 20 min
600 0.8
TMP [mbar]

BWI 10 min
Kw / Kw0

slope = -0.0063

300 0.4

slope = -0.0191
slope = -0.08
0 0
0 60 120 180 240 0 15 30 45
Operating time [h] Permeate volume [m3]

Fig. 6-4 (a) TMP development vs. operating time at different BWI (b) Kw/Kw0 vs. permeate volume
at different BWI (Other operating parameters were identical: BWI 10 minutes, BWT 20 seconds, BWF
260 LMH)

The comparison of TMP increase with respect to operating time is displayed in Fig. 6-4 (a).
As shown, TMP increased more quickly when BWI was extended from 10 to 40 minutes. At
BWI 10 minutes TMP increased 50 mbar within 240 hours, the extention of BWI to 20 min-
utes led to a TMP increase of 600 mbar in 210 hours. When the BWI was 40 minutes, TMP
raised over 600 mbar after 40 hours operation. Fig. 6-4 (b) shows the slopes of permeability
reduction versus permeate volume under these conditions. It is indicated that at a longer BWI
the membrane suffers more severe fouling while treating similar water volume. As the feed-
water quality in each experiment was similar to each other (Tab. 6-1), the severity of fouling
was therefore mainly caused by the variation of BWI. The extention of BWI leads to more
foulants accumulating at the membrane surface. This decreases the efficiency of backwashing
and results in more severe fouling.

6.3.2 Effect of Foulant Concentration in Feedwater

Considering biopolymers as major organic foulants in UF of treated domestic wastewater, it is
expected to find a connection between the variation of biopolymer concentration in feed water
Pilot-scale Ultrafiltration of Bio-filtrate 83

and TMP development along membrane filtration. A logical correlation can offer useful in-
formation in fouling control.

Correlation between Biopolymer Concentration and TMP Development

The investigation on the relationship was conducted in three tests. In each test, the UF pilot
plant was used to filter slow sand effluent. During the first two succession tests (test 1 and 2),
the UF was operated at a flux of 50 LMH, BWI of 20 minutes (Fig. 6-5 a, b). In test 3, the
operational conditions were flux at 75 LMH and BWI in 14 minutes. Comparison of the TMP
development in these tests indicates that TMP proceeded in a similar pattern. After a rise at
the beginning of filtration, TMP increased slowly for several days. Thereafter, it increased
more quickly.

A significant correlation between biopolymer concentration and TMP variation was observed
in operating day 16 to 17 by test 1 (Fig. 6-5 a), during which TMP decreased sharply while
biopolymer concentration fell in the feedwater. During the end phase of test 3 (from operating
day 9 to 11, Fig. 6-5 c), a slower than expected TMP increase was related to a very low bio-
polymer concentration in the SSF. Thereafter, TMP rose significantly in parellel to an in-
crease of biopolymer content in feedwater.

In contrast, no significant relevance between these two parameters was found when TMP de-
velopment was in the slow increase phase. For instance, at day 7 by test 2 (Fig. 6-5 b) and day
6 by test 3 (Fig. 6-5 c) a significant variation of biopolymer concentration in the feedwater led
no immediate change in TMP variation.

(a) 800 0.8

Biopolymer concentration
TMP
Biopolymer concentration [mg C/L]

Operational condition:
600 0.6
flux 50 LMH, BWI 20 min
TMP [mbar]

400 0.4

200 0.2

0 0
0 5 10 15 20
Operating time [d]
84 Pilot-scale Ultrafiltration of Bio-filtrate

(b) 800 0.8

Biopolymer concentration
TMP

Biopolymer concentration [mg C/L]

600 Operational conditions: 0.6
flux 50 LMH, BWI 20 min
TMP [mbar]

400 0.4

200 0.2

0 0
0 4 8 12 16 20
Operating time [d]

(c) 800 0.8

Biopolymer concentration Operational condition:
TMP flux 74 LMH, BWI 14 min

Biopolymer concentration [mg C/L]

600 0.6
TMP [mbar]

400 0.4

200 0.2

0 0
0 3 6 9 12
Operating time [d]

Fig. 6-5 Correlation between biopolymer concentration and TMP development in (a) test 1 at flux 50
LMH, BWI 20 min, (b) test 2 at flux 50 LMH, BWI 20 min and (c) test 3 at flux 74 LMH, BWI 14
min

6.3.3 Fouling Development Processes

Multi-phase Fouling Development
It is evident that although an increase of biopolymer load mass in each filtration cycle leads to
generally more severe fouling (Fig. 4-11), a correlation between biopoylmer concentration in
feedwater and TMP variation cannot always be observed in daily operation. Particularly, no
relevance was found when the operation was in the TMP slow increase phase. It implies that
the interactions between foulants and membrane and between foulants vary over the whole
period of membrane operation. The effect of biopolymer load mass on fouling development
may be different at different TMP development phases.

Based on pilot-scale results, a similar TMP development pattern at different operational con-
ditions can be observed. It contains an initial rise of TMP, followed by a steady TMP in-
crease. In the end of a test there is a sudden TMP rise. A similar fouling development has
been observed by Ye et al. (2005) and Zhang et al.(2006a) while operating membrane filtra-
tion at sustainable operational conditions and in constant flux mode. A three-stage fouling
process in MBR was well summarized by Zhang et al. (2006a). Although the conditions are
Pilot-scale Ultrafiltration of Bio-filtrate 85

not identical to that in this work, the similar organic fouling behaviour in MBR and in low-
pressure membrane processes of treated wastewater (Amy, 2008) render their research helpful
to understand the fouling phenomenon in UF as a post treatment process.

Biological floc Feed

EPS SMP Colloids Particulates

Stage 1
Biomaterial residue Passive adsorption Initial pore blocking Conditioning
fouling

Cake formation Biofilm growth Further pore blocking

Stage 2
Steady fouling
Irregular fouling distribution

Redistribution of permeability
Stage 3
TMP jump
Local flux > Critical flux

Severe TMP increase

Fig. 6-6 Fouling mechanisms for MBR operation at constant flux (adapted from Zhang et al., 2006a)

Fig. 6-6 shows an overview about the TMP development stage and correspondingly related
fouling mechanisms. Based on their report the initial fouling stage is defined as conditioning
fouling which appears at the beginning of the filtration when strong interactions take place
between the membrane surface and the SMP/EPS. The fouling is mainly due to pore blocking
or adsorption (Jiang et al., 2005). This leads to an initial TMP increase. In the following
steady fouling stage the membrane surface is expected to be mostly covered by SMP/EPS,
promoting further attachment of particulates, colloids and dissolved substances. Although the
membrane filtration is operated at sustainable conditions, adsorption and/or deposition of or-
ganics may also occur during this stage. As these effects take place also across the whole
membrane surface besides directly on the membrane pores, the accumulation of foulants may
not affect TMP significantly. Therefore, TMP develops more slowly compared to the condi-
tioning fouling phase. Because of the uneven distribution of the delivered foulants along a
membrane fibre, the fouling layer is assumed to develop from location to location alterna-
tively (Wang and Waite, 2008b). At last, most of the fibres are covered with a fouling layer to
a critical extent. The critical fouling level would lead to a jump of TMP when the membrane
is operated at a constant flux mode. Judd (2006) summarized several mechanisms to explain
the TMP jump condition. Based on this summary, the jump would be a result of filtration area
loss (in a macroscopic view) or pore loss (in a microscopic view) and the structure modifica-
86 Pilot-scale Ultrafiltration of Bio-filtrate

tion of the fouling layer due to increasing driving force effects. Above the critical fouling
level, the permeability of the membrane is difficult to be restored by hydraulic cleaning.

Fouling Development in UF
Based on the proposed mechanisms of the fouling effect of biopolymers and the above men-
tioned review on fouling phenomena, a three-stage fouling process (Fig.6-8) is used to inter-
pret fouling development in UF of treated domestic wastewater. The experimental result of
UF of SSF displayed in Fig. 4-11 (b) (operated at 50 LMH, BWI 10 minutes, BWT 10 sec-
onds, biopolymer concentration is around 0.32 mg C/L on average) is recalled and analyzed
as an example (Fig. 6-7).

800 1

Stage 3

Stage 2
0.8
600
Stage 1

Normalized permeability
0.6

BC [mg C/L]
TMP [mbar]

400

0.4

200
TMP development
0.2
Permeability reduction

Biopolymer concentration

0 0
0 5 10 15 20 25 30 35
Operating time [d]

Fig. 6-7 Three-stage fouling development process during UF of SSF (0.5 m/h)

As shown, an increase of TMP (TMP increase rate 20 mbar/day) occurred within the initial
three days. Correspondingly, the permeability of the membrane was reduced by 24%. This
phase can be classified as conditioning fouling phase, during which rapid irreversible fouling
occurs due to pore blocking by biopolymers (Fig. 4-12 a) and/or passive adsorption
(Haberkamp, 2008; Susanto et al., 2007).

After the initial significant rise, TMP increased steadily at an average rate of 5.5 mbar per day
from day four to day 27. During this period, the permeability was reduced continuously from
76% to 52% of the initial value. It is observed that although from day 14 to day 27 the bio-
polymer concentration is much higher than before, the variation of TMP was not significant.
During this phase the TMP increased averagely 6 mbar per day. This phenomenon could be
Pilot-scale Ultrafiltration of Bio-filtrate 87

attributable to the sustainable operational conditions. As shown in section 6.2.1, operated at a

flux of 50 LMH, BWI of 10 minutes, TMP could be kept stable for a long period. Under these
conditions, it can be supposed that the biopolymer load mass is under a critical level, most of
the accumulated foulants can be effectively backwashed. Therefore, the increase of biopoly-
mer concentration may not affect TMP to a large extent. At this point, TMP development in
this phase does not seem to reflect the change of biopolymer concentration in the feedwater.

When the biopolymer concentration could be maintained at lower levels, e.g. lower than 0.1
mg C/L in SSF, the operating time could be extended significantly (Fig. 4-11 b). However, as
the concentration was above 0.5 mg C/L from day 27, the high biopolymer load to UF led to a
TMP jump. Within four days the TMP increased from 364 mbar to around 700 mbar (TMP
increase rate 82 mbar/d). The permeability decreased from 53% to 28% of the initial capacity.

Conditioning Steady fouling TMP jump

fouling stage stage stage

Permeability

Permeabiltiy
TMP

TMP

Permeate volume / operating time

Fig.6-8 Illustration of an ideal three-stage fouling process in UF of treated domestic wastewater

(adapted from Zhang et al., 2006a)

Based on the above case study and previous observation, a graphic illustration of the fouling
process in UF of treated domestic wastewater is suggested (Fig.6-8). In general, fouling proc-
esses include conditioning fouling, steady fouling and TMP jump phase when the UF is oper-
ated in constant flux mode and under sustainable operational conditions. In UF of SSF, due to
the removal of particles, colloids and biopolymers, the three-stage fouling process can be
readily identified. Nevertheless, due to the high fouling potential of SE, the three-stage proc-
ess can only be observed at very low biopolymer concentrations in feed water (Fig. 4-11 a,
biopolymer concentration in SE is 0.1 mg C/L). The reduction of the foulant load mass and/or
optimization of operational conditions can extend the duration of these phases significantly.
Otherwise, the distinct differences between the different fouling phases cannot be identified
(Fig. 6-1 a, Fig. 6-4 a, Fig. 4-11 a).

6.4 Influence of Backwash Time on Fouling Control

Since the backwash flux was set at 260 LMH which is significantly higher than the filtration
flux (from 50 to 90 LMH) in the present work, a shorter BWT is preferred to improve the
88 Pilot-scale Ultrafiltration of Bio-filtrate

permeate recovery (Pr, Eq.2-12). Therefore, it is necessary to investigate the influence of

BWT on fouling control effects.

Flux 50 LMH, BWI 20 min, BWT 20 s

Flux 50 LMH, BWI 20 min, BWT 40 s
1 Flux 50 LMH, BWI 40 min, BWT 20 s
Flux 50 LMH, BWI 40 min, BWT 60 s
Linear (Flux 50 LMH, BWI 40 min, BWT 20 s)
0.8
Linear (Flux 50 LMH, BWI 40 min, BWT 60 s)
Kw / Kw0

0.6
slope = -0.0169

0.4

slope = -0.077 slope = -0.0191

0.2
slope = -0.08

0
0 10 20 30 40 50
Permeat volume (m3)

Fig. 6-9 Comparison of the influence of BWT on the performance of UF

The comparison was conducted by filtering slow sand effluent at a flux of 50 LMH. The ef-
fect of BWT of 20 and 40 seconds at BWI of 20 minutes, effect of BWT 20 and 60 seconds at
BWI 40 minutes were investigated, respectively. The biopolymer concentration on average
throughout the experiments was similar (Tab. 6-1). Fig. 6-9 displays the permeability reduc-
tion at corresponding operational conditions. In the case of BWI at 20 min, the BWT at 20 s
and at 40 s demonstrates no significant influence on the performance of UF, although longer
BWT led to a slightly slower permeability reduction. The extension of BWT from 20 s to 40 s
resulted in, however, an approximate eight percent decrease of permeate recovery. When BWI
was set at 40 min, an extention of BWT from 20 s to 60 s led to no improvement in fouling
control but a significant reduction in respect of permeate productivity. The result implies that
most of the hydraulic reversible fouling can be removed at the beginning of a backwashing.
Based on the experimental results, the backwash time in filtering SSF in the present work was
set to around 20 seconds.

6.5 Chemically Enhanced Backwash

In practice, membrane filtration is preferred to be operated at high flux and long BWI to re-
duce the membrane area and improve the permeate productivity. Under these conditions, as
shown previously, fouling develops quickly. Therefore, chemically enhanced backwash is
necessary to stabilize the performance of membrane systems. Based on the chemical charac-
teristics of fouling and feed water quality different reagents can be used (Decarolis et al.,
2001; Teodosiu et al., 1999; Liu et al., 2006).
Pilot-scale Ultrafiltration of Bio-filtrate 89

(a) 800
Backwash with permeate
NaOH (pH 12)
600 H2O2 (20 mg/L)
NaClO (active Cl 10 mg/L)
TMP [mbar]

400

200

0
0 400 800 1200 1600
Operating time [h]

(b) 1.2
Backwash with permeate
NaOH (pH 12)
H2O2 (20 mg/L)
NaClO (active Cl 10 mg/L)
0.8
Kw / Kw0

0.4 slope = -0.0037 slope = -0.0004

slope = -0.0008
slope = -0.0379

0
0 400 800 1200 1600
Operating time [h]

Fig. 6-10 (a) TMP development vs. operating time using permeate backwash and enhanced backwash
reagents (b) Kw/Kw0 vs. operating time using permeate backwash and enhanced backwash reagents

In the present work sodium hydroxide, hydrogen peroxide and sodium hydrochloride were
tested individually as enhanced backwash reagents. The UF was operated at flux 89 LMH,
BWT 40 minutes. To improve backwash effect, CBWT was set at 50 seconds followed by a
soaking time of 5 minutes. During each chemically enhanced backwash, the dosage was con-
trolled by keeping the backwash permeate at pH of 12 or by holding the concentration of
H2O2 at 20 mg/L or active chlorine at 10 mg/L. Each experiment was conducted until the
TMP increased to 700 mbar. The operational conditions and water qualities are shown in Tab.
6-1.

TMP control results using different chemicals are shown in Fig. 6-10 (a) and Tab. 6-1. The
results show that dosing different backwash reagents extends the operating time of the pilot
90 Pilot-scale Ultrafiltration of Bio-filtrate

plant in different patterns. Within the same TMP increase scope, the dosing of NaOH into the
backwash permeate extended the operating time of the UF pilot plant from 16 hours to around
150 hours. It is indicated that the hydraulic irreversible fouling can be removed by bases.
However, after several days the pilot plant was out of operation due to serious scaling in the
membrane module and dosing system. The inorganic precipitates scaled the outside of the
membrane surface so seriously that the whole system was stopped due to high pressure losses.
Best TMP control effects were achieved by using NaClO (active Cl concentration 10 mg/L).
The TMP rose from 200 to 450 mbar after 1600 hours operation. In detail, Fig. 5-3 shows the
effect of NaClO. It can be seen that each time after enhanced backwash using NaClO, the
TMP was almost back to its original level. This result indicated that NaClO removed most of
the hydraulic irreversible fouling during backwashing. In contrast, backwash using permeate
resulted in continuous increase of TMP. Althoug it recovered the permeability to some extent,
it was unable to remove the hydraulic irreversible fouling.

Dosing H2O2 (concentration 20 mg/L) into backwash water was also identified as an effective
fouling control method. This led to a TMP increase from 200 to 500 mbar after about 800
hours operation. Normalized permeability reduction versus time (Fig. 6-10 b) shows that the
usage of NaClO leads to the flattest permeability reduction compared to other enhanced
backwash reagents. Although using H2O2 shows a sharper permeability decreases, it is pre-
sented as a reliable chlorine-free fouling control reagent compared to the usage of chlorine
which causes disinfection by-product problems in water treatment (Hua and Reckhow, 2007).

6.6 Membrane Cleaning

As an integrated part of membrane operation, chemical cleaning is important to recover the
permeability of a fouled membrane. Based on the characteristics of major foulants in mem-
brane filtration, different cleaning reagents can result in different cleaning effects. For UF of
SSF in the present work, cleaning effects of acid (H2SO4 pH 2), caustic (NaOH pH 12), oxi-
dants (NaClO active Cl 50 mg/L, H2O2 125 mg/L) and different waters (UF permeate, tap
water and demineralised pure water) on the permeability restoration were tested (Tab. 6-3).

Tab. 6-3 Cleaning effect of different chemical reagents on permeability restoration of UF of SSF

Experiment date Flux TMPstart TMPend Chemical TMP c Permeability

(2007) (LMH) (mbar) (mbar) reagents (mbar) Restoration
23.05.~27.05 60 220 680 H2SO4 pH 2 520 15%
01.06.~06.06 60 240 688 NaOH pH 12 312 65%
08.06.~13.06 60 270 468 NaClO active Cl 324 61%
-1
50 mg L
15.06.~20.06 60 224 400 Tap water 336 24%
22.06.~27.06 60 240 576 Permeate 460 18%
29.06.~04.07 60 260 708 H2O2 125 mg/L 500 24%
06.07.~11.07 60 280 600 Pure Water 396 45%
Pilot-scale Ultrafiltration of Bio-filtrate 91

To compare the cleaning effect of different reagents, membrane filtration of SSF (filtration
rate at 0.25 m/h) was operated at a flux of 60 LMH, BWI 20 minutes and BWT 20 seconds
for five days. After that, the membrane was cleaned using a single cleaning reagent. During
cleaning, 10 L cleaning reagent was firstly backwashed into the membrane module to remove
the cake layer on the membrane. The membrane was then soaked for 24 hours. Subsequently,
10 L cleaning reagent was backwashed again to remove irreversible foulants. Thereafter, tap
water was used to forward flush the membrane for two minutes. At the end, the TMP was
tested by filtering tap water at 60 LMH for five minutes. All the cleaning procedures were
performed at room temperature (around 20 ºC). After cleaning the corresponding permeability
restoration was calculated according to Eq.6-2. In the present study cleaning conditions would
not damage the membrane since the concentration studied was within the ranges recom-
mended by the membrane manufacturer.

K w−c − K w−end
Rp = × 100 o o Eq.6-2
K w− start − K w−end

where
Rp = restoration of permeability, %
Kw-start = permeability before filtration, L/(m²·h·bar)
Kw-end = permeability after five days operation, L/(m²·h·bar)
Kw-c = permeability after cleaning, L/(m²·h·bar)

The results are shown in Fig. 6-11. It can be seen that the highest permeability recovery was
obtained by using NaOH (pH 12) while acid (H2SO4, pH 2) exhibited the lowest cleaning ef-
fect. This result indicates that inorganic foulants (scales and metal dioxides) play a minor role
in hydraulic irreversible fouling during UF of treated wastewater. The fouling is mainly of
organic property. Caustic reagents can hydrolyze the organic binding more readily than acids
and solve them into the solution (Liu et al, 2006).

Although the concentration of NaClO (active Cl 50 mg/L, fourfold diluted as the recom-
mended concentration per the membrane manufacturer) used in the present work was much
lower than the recommend value, it presented also a capacity restoring the membrane perme-
ability. At a higher concentration NaClO is expected to perform better (Kimura et al., 2004;
Liu et al., 2006). H2O2 (125 mg/L, fourfold diluted as the recommended value) was also a
recommended cleaning reagent, but it showed a permeability recovery lower than 30% at the
present concentration. The result suggests that with respect to removing EfOM fouling, chlo-
rine performs better than hydrogen peroxide.

A significant permeability restoration was obtained by using demineralized pure water in

membrane cleaning. After 24 h soaking, the permeability of the fouled membrane was recov-
ered by 50%. In contrast, tap water and permeate showed a recovery by 20% (Fig. 6-11). The
DOC content in these waters before and after membrane was measured. It is evident that (Tab.
6-4) using pure water more organic substances can be removed from the fouled membrane.
Considering that calcium in backwash water influences the fouling control efficiency (Li et
al., 2009), calcium content in these waters before and after membrane cleaning was measured
(Atom Absorption Spectrometry, GBC Secientific Equipment, Australia). In the used pure
92 Pilot-scale Ultrafiltration of Bio-filtrate

water it reached 106 mg/L. In the tap water and permeate the concentration was little changed
after membrane cleaning.

As no single reagent can recover the permeability to over 90%, a mixture of NaOH, NaClO in
demineralised water (pH 12, active Cl 50 mg/L) was then tested to remove organic foulants.
The result (Fig. 6-12) showed that using this mixture reagent the permeability of a fouled
membrane can be restored to 90 – 95 %. It suggests that a cleaning strategy utilizing the co-
effect of different reagents can improve the cleaning effect to a large extent.

80
Permeability restoration [%]

60

40

20

0
NaOH NaClO Pure H2O2 Tap Permeate H2SO4
pH 12 active Cl water 125 mg/L water pH 2
50 mg/L

Fig. 6-11 Comparison of permeability restoration effect of UF using different cleaning reagents

Biopolymer content in feedwater (SSF) to UF and in cleaning water after membrane cleaning
was quantified (Tab. 6-4). The results show that in feedwater to UF biopolymers contributed
to around 1.1 - 2.4% carbon of the total DOC. In the dissolved fraction in cleaning water,
their contribution increased to 30 – 63%. The significant enrichment of biopolymers in clean-
ing water indicates that biopolymer fraction is preferred to be retained by UF and forms irre-
versible fouling.

The biopolymer concentration in different cleaning waters shows an apparent correlation with
the membrane permeability restoration (Fig. 6-12). Generally, the higher the biopolymer con-
centration in the cleaning water – the more biopolymers are removed and/or solubilized in the
solution, the higher the permeability of the membrane can be recovered. It suggests that the
understanding of the property and fouling mechanisms of biopolymers can improve the effec-
tiveness of membrane cleaning. Simultaneously, it is noted that other fractions with a size
smaller than biopolymers play also a role in irreversible fouling. Within the increased DOC
fraction, which represents the removed organics from the fouled membrane, these nonbio-
polymer organics attributed to over 37% of the DOC increase (except the case using permeate
as cleaning reagent, Tab. 6-4).
Pilot-scale Ultrafiltration of Bio-filtrate 93

Tab. 6-4 Comparison of biopolymer content in SSF to UF and that in cleaning water using different
cleaning reagents

Cleaning reagent BC in DOC of Ratio BC in DOC increase in Ratio

SSF SSF BC/DOC cleaning water cleaning water BC/DOC
increase
(mg C/L) (mg/L) (%) (mg C/L) (mg/L) (%)
H2SO4 pH 2 0.14 12.4 1.1 0.9 2.9 32
NaOH pH 12 0.1 8.5 1.1 11.3 26.7 42
NaClO active 0.1 8.6 1.2 1.4 4.0 34
Cl 50 mg/L
Tap water 0.08 7.1 1.1 0.7 2.2 30
Permeate 0.08 7.3 1.1 0.3 0.31 97
H2O2 125 mg/L 0.09 7.8 1.2 - - -
pure water 0.11 7.5 1.5 2.1 5.2 40
Standard cleaning 0.32 13.3 2.4 19.4 30.5 63
reagent*
Standard cleaning reagent*: NaOH pH 12 + NaClO active Cl 50 mg/L + demineralized water; BC:
biopolymer concentration, DOC increase: DOC increase in cleaning water after membrane cleaning

100
Demineralised pure water + NaOH
pH 12 + NaClO active Cl 50 mg/L

80
Permeability restoration [%]

NaClO active Cl 50 mg/L

60 R2 = 0.70 NaOH pH 12

Demineralised pure water

40

Tap water

20 H2SO4 pH 2

Permeate
0
0 5 10 15 20 25 30 35
Biopolymer concentration in cleaning water [mg C/L]

Fig. 6-12 Relation between permeability restoration using different cleaning reagents and biopolymer
concentration in corresponding cleaning waters (same volume of cleaning water was used in all the
experiments)

Optical impressions of several cleaning water samples are shown in Fig. 6-13 (a) and (b).
Generally, suspended tender fragments can be seen in cleaning waters. In NaClO solution
large numbers of fragments were observed (Fig. 6-13 b). The breadth of a compact fragment
on a slide glass (Fig. 6-13 c) was measured as 1.4 mm (Microscopy ALTRA 20, Zeiss, Ger-
94 Pilot-scale Ultrafiltration of Bio-filtrate

many). The diameter of the corresponding fragment was calculated as 0.9 mm. It was identi-
cal to the diameter of a single UF fibre (Fig. 6-13 d, Dizzer 450, Inge AG). As the result was
reproducible, it can be logically proposed that the tender fragments are mainly irreversible
cake/gel layers which are formed by aggregated or compacted foulants on the membrane sur-
face.

(a) (b)

(c) (d)
d=0.9mm

1.4 mm

Fig. 6-13 (a) Optical impression of demineralised water after membrane cleaning, (b) optical impres-
sion of NaClO solution (active Cl 50 mg/L) after membrane cleaning, (c) diameter measurement of the
suspended orbicular fragment in NaClO solution after membrane cleaning, (d) diameter of a mem-
brane fibre of Inge membrane (dizzer® 450, source: Inge AG)

6.7 Discussion and Summary

Influence of BWI and Flux
The present section shows that operational conditions, in particular BWI and flux, impact the
performance of UF to a large extent. Under similar feedwater quality, they affect the extent of
fouling by influencing foulant mass load and/or the structure of the fouling. In lab-scale tests
(section 4.1) it is evident that an increase of biopolymer mass load to a membrane leads to
more fouling. The increase of the interaction between foulants and membrane and between
foulants is primarily attributed to the multiplication of binding sites available of the macro-
molecules. Although the interactions are based on weak physico-chemical adhesion and cohe-
sion, the total binding energies of weak interactions between EPS molecules multiply with the
Pilot-scale Ultrafiltration of Bio-filtrate 95

increase of binding sites and add up to binding values exceeding those of strong covalent C-C
bonds (Flemming and Wingender, 2001). The strengthened interaction due to load mass in-
crease can be used to interpret the promotion effect on fouling caused by the extension of
BWI by the UF pilot plant.

The increase of filtration flux can also lead to more fouling. It is evident that an increase of
flux results in a more than proportional TMP development. The main reason has been identi-
fied as the alteration of the fouling layer structure. Under a higher filtering flux, which is
caused by a higher filtration pressure, the biopolymer layer is further compressed. The fouling
layer induces higher specific fouling resistance and lower reversibility. In membrane filtration
compressibility is presented as an intrinsic property of the deposited foulants (Singh and
Song, 2006). Deformable colloids (soft colloids) e.g. biopolymers are more compressible than
rigid ones. While for rigid colloids the compression is mainly due to closer packing and rear-
rangement, biological colloids can be significantly deformed under filtration pressures. The
deformation can increase the contact area among foulants, fluid and membrane (Foley, 2006).
Thus, both the available binding sites and the cross-linked network structure of the fouling
layer can be subsequently changed (Wang and Waite, 2008b). In particular, the foulants adja-
cent to the membrane are exposed to the highest pressure, present the highest specific fouling
resistance and induce the severest fouling effect (Tansel et al., 2006; Ye et al., 2005). For a
membrane filtration operated at constant flux mode, the pressure on newly delivered foulants
in each cycle is similar (based on flux applied). According to pressure gradients within the
fouling layer, the TMP measured for the membrane module is acting mainly on the existing
fouling, instead of the newly accumulated foulants (Chen et al., 2006b; Ye et al., 2005).

Main Fouling Mechanisms and Fouling Development

In this section, the fouling mechanisms discussed in section 4.2 are integrated to the fouling
development investigated by a UF pilot plant to get an overview of the fouling phenomenon.
In lab-scale experiments (section 4.2) it is evident that during the initial filtration phase the
major fouling mechanism during is pore blocking. This leads to a significant permeability
reduction. Simultaneously, adsorption is regarded to play also an important role. In an adsorp-
tion experiment, Haberkamp (2008) showed that adsorption of biopolymer onto a membrane
surface resulted in a permeability reduction by 5 %. Continuous delivering biopolymers to the
membrane surface alters the fouling mechanisms from pore blocking to the formation of a
cake/gel layer (section 4.2).

Lab-scale multi cycle filtration experiments (section 4.1) show that backwashing can recover
the permeability of a fouled membrane to a large extent, which indicates the high reversibility
of the fouling layer. Nevertheless, irreversible fouling increases continuously. As the major
fouling mechanism is identified as cake/gel filtration, this implies a constant growth of a
cake/gel layer along the operation. The development of the fouling layer is caused by the in-
crease of the biopolymer mass and also due to the change of the fouling structure under pres-
sures (Wang and Waite, 2008b; Ye et al., 2005). Considering that major foulants are protein-
like and polysaccharide-like substances, an increase of their accumulated mass promotes the
interaction among them and with the membrane significantly (Amy, 2008). As a result, the
development of the cake/gel layer impacts the effectiveness of hydraulic backwashing.
96 Pilot-scale Ultrafiltration of Bio-filtrate

In practical operation of a UF pilot plant, the variation of biopolymer concentration in feed-

water cannot always result in a consequent TMP alteration. This can be partly attributed to the
high complexity of fouling phenomenon in the reality. In addition, the influence of the fouling
development phase on the relationship should also be taken into consideration. Fouling
mechanism analysis in previous studies has been generally focused on the initial interactions
between foulants and membrane (Listiarini et al., 2009; Susanto et al., 2008), little has been
addressed on fouling development during a long-term UF operation. Under this condition, the
fouling mechanism is different to that in the initial phase. Ye et al. (2005) found that fouling
presented in a long-term membrane filtration, even under subcritical flux, showed a lower
reversibility compared to that presented in a short-term dead-end experiment.

According to the concept describing fouling development in MF of model EPS (Ye et al.,
2005) and in MBR systems (Zhang et al., 2006a), and based on the pilot-scale experimental
results in the present work, the fouling process in UF of treated domestic wastewater is sug-
gested to be separated into three stages. It is evident that in the first stage, which is called
conditioning fouling stage, the development of irreversible fouling is mainly caused by pore
blocking and adsoption due to a direct interaction between foulants and membrane. In gen-
eral, the fouling in this stage is formed within a short time. It presents a limited fouling resis-
tance (Haberkamp, 2008) but is of low reversibility. During the second stage, the deposition
of foulants may take place mostly on the fouling layer already present on the membrane sur-
face. Although small foulants can pass through this layer and be transported into the pores,
the adsorption capability of the pore wall could be exhausted due to saturation (Xiao et al.,
2009; Haberkamp, 2008; Susanto et al., 2007). Therefore, the main mechanism in this stage is
considered to be cake/gel filtration. The cake/gel layer shows a high reversibility. This has
been revealed by the lab-scale experiment (section 4.1). In pilot-scale plant the high reversi-
bility of the fouling can be detected by the slow increase of TMP. This phenomenon indicates
that most of the newly cumulated foulants are hydraulically removable. In a more detailed
investigation, Wessling (2001) visulized that in cake/gel filtration the density of the fouling
layer above the pore opening is lower than that above the flat adsorption surface. This kind of
fouling can be readily removed by backflushing when its density is lower than a critical level.
In this phase, a short term variation of foulant load may not influence TMP development sig-
nificantly. But, when biopolymer concentration in feedwater is kept high for a longer time,
the accumulation of irreversible fouling can accelerate the TMP increase rate significantly.
Otherwise, a long-term operation under sustainable conditions offers also time for the reloca-
tion of foulants and consolidation of the links between foulants and membrane (Chen et al.,
2006b; Ye et al., 2005). This leads to a chronic accumulation of irreversible fouling until a
sudden TMP increase, which indicates the appearance of the third stage. In this phase, the
accumulation of irreversible fouling is over a critical condition. The effectiveness of back-
washing is reduced to a large extent. Therefore, the variation of foulants content in feedwater
can be immediately reflected by TMP development.

The results based on pilot-scale tests suggest that an overview of fouling development phases
is helpful to understand the correlation between foulants in feedwater and TMP development.
At critical operational conditions, e.g. at high flux, the duration of the three-stage TMP devel-
opment can be largely influenced. The correlation between flux and the severity of fouling
indicates that the compression of a cake/gel fouling layer plays an important role in fouling
phenomenon.
Pilot-scale Ultrafiltration of Bio-filtrate 97

Membrane Cleaning
Backwashing was used as a physical cleaning process to maintain the permeability of the UF.
Generally, key parameters in the design of backwashing are frequency (presented by back-
wash interval), duration and intensity. Based on the previous discussion, it is evident that the
frequency impacts the severity of fouling mainly due to the variation of the accumulated fou-
lant mass. At a fixed backwash intensity (flux), section 6.4 demonstrated that an extension of
backwashing duration from 20 to 60 s led to only a very limited improvement of permeability
recovery. By contrast, it reduced the productivity of the membrane system significantly. In
addition, the result suggested that at a high backwashing flux, e.g. 260 LMH in the present
case, most of the removable foulants were backwashed within a short time. As backwashing is
mainly used to remove more reversible foulants, e.g. particles or pore blocking foulants, an
extension of its duration is not expected to improve irreversible fouling removal. Thus, an
optimization of backwashing is suggested to be undertaken with regard to reversible fouling
control and energy or permeate consumption.

The irreversible fouling can be effectively mitigated by chemically enhanced backwash dur-
ing operation, as shown in section 6.5, or by chemical cleaning which is investigated in sec-
tion 6.6. In coincidence to the results of other studies (Guo et al., 2009; Le-Clech et al., 2006;
Kimura et al., 2004), caustic reagents and oxidants are effective membrane cleaning chemi-
cals. The effective separation of fouling and membrane is considered to be due to the func-
tions of the agents when they interact with fouling meterials. According to the report pre-
sented by Liu et al. (Liu et al., 2006), the functions of caustics are hydrolysis and solubiliza-
tion. Organic materials including polysaccharides and proteins can be hydrolyzed by caustics.
For humic substances, caustic can improve their negative charge as well. Therefore, the bind-
ing between membrane and these organics can be weakened. This results in a more reversible
fouling.

In either CEBW or in membrane chemical cleaning experiments, oxidant performs effectively

in removing fouling. Besides the function of hydrolization (Le-Clech et al., 2006), another
major fouling removal mechnism of an oxidant is considered as generation of oxygen-
containing functional groups, e.g. carboxylic acids, during oxidation of organic polymers (Liu
et al., 2006). The existence of these functional groups generally increases hydrophilicity of
their parent compounds. Therefore, oxidation reduces the adhesion of fouling materials to
membranes. As a prevalent oxidant, NaClO appears in most of the cleaning recipes (Le-Clech
et al., 2006). However, the selection of a suitable reagent is not only based on its fouling re-
moval effect and economic items, its influence on the environment needs also to be taken into
consideration. Due to the disinfection by-products caused by chlorine, other chlorine-free
oxidants, e.g. H2O2 , are more recommended to be used.

Using demineralized water in membrane cleaning can recover the permeability of a fouled
membrane more significantly than using tap water or UF permeate. In the used pure water a
higher biopolymer content and a much more significant increase of calcium concentration
than that in tap water or UF permeate was observed (section 6.6). These phenomena indicated
that the removal of irreversible fouling was relevant to the release of cations into pure water.
Demineralized water has been used as backwash water in several studies to reduce fouling in
98 Pilot-scale Ultrafiltration of Bio-filtrate

UF membranes (Li et al., 2009; Abrahamse et al., 2008). It was found that due to the absence
of cations, either divalent or monovalent, the removal efficiency of irreversible fouling was
largely improved. These results suggested that the absence of cations leads to a weakened link
between fouling and membrane (Abrahamse et al., 2008). Therefore, more irreversible fouling
could be backwashed. The mechanisms are considered similar to that in membrane cleaning
using demineralised water. The function of demineralised water in fouling removal was
summarized by Li et al (2009) in detail. It is stated that during membrane filtration, due to the
screen effect caused by cations organic foulants can easily deposit on the membrane surface
and cause fouling. The absence of cations in cleaning water reduces the charge-screening ef-
fect. The repulsion force between the negatively charged membrane and organic compounds
can be increased. This subsequently results in a more removable fouling layer. In particular,
due to dissolution of calcium into pure water, the calcium-bridging strength between mem-
brane and organic foulants can also be partly reduced. The weakened cohesion of organic
compounds on the membrane improves the reversibility of the fouling layer.
General Conclusions 99

7 General Conclusions
This study aims at relating the content of principal organic foulants to the fouling phenomena
in ultrafiltration of treated domestic wastewater and investigating fouling control effects using
bio-filtration as a pre-treatment process. In lab and pilot-scale UF experiments, the fouling
potential of various water samples was assessed. A quantified correlation between the content
of biopolymers and fouling in UF has been set up and verified. Pilot-scale bio-filtration proc-
esses (bank-filtration, slow sand filtration) were operated to reduce the fouling potential of
wastewater effluent prior to UF. Foulants removal mechanisms and factors affecting the per-
formance of biofiltration process have been investigated.

7.1 Major Organic Foulants in UF and related Fouling Mechanisms

The identification of major foulants in UF of treated domestic wastewater is undertaken in lab
and pilot-scale experiments. In either secondary effluent or slow sand filtrate, the fraction of
dissolved organics larger than or comparable to the correspondingly used UF pores (hydrophi-
lized PES UF membrane, MWCO 100-150 k Da, nominal pore size 0.026 µm) have been
identified as major foulants affecting the filterability of the water sample and the reversibility
of the fouling in UF. These organics can be roughly characterized using LC-OCD-UVD-
OND. The measurements show that they are large organic molecules with low UV absorbance
and some organic nitrogen content, commonly regarded as macromolecular polysaccharide-
like and protein-like biopolymers.

Using the carbon content to represent the concentration of biopolymers, it is evident that the
biopolymer concentration affects the filterability of water proportionally. In lab-scale multi
cycle filtration experiments it is further shown that the higher the biopolymer load to the
membrane, the lower the restoration of the membrane permeability by hydraulic backwash-
ing, although most of the cumulated biopolymers can be backwashed. A significant negative
correlation between biopolymer concentration in feedwater and the reversibility of fouling
layer is identified. Based on the quantified correlation, it can be stated that the content of bio-
polymers in treated wastewater can reflect its fouling potential in UF. The statement is veri-
fied to be reproducible using pilot-scale UF filtering secondary effluent or slow sand filtrates.
Further tests show that the fouling potential of treated wastewater cannot be quantified using
photometric methods for proteins and polysaccharides. But, under certain conditions fouling
index parameters, e.g. SUR, show a reasonable correlation with it.

Besides the effect of biopolymer load mass on fouling development, the structure of the bio-
polymer fouling layer influences the extent of fouling as well. Pilot-scale UF experiments
show that an increase of flux leads to a more than proportional development of fouling. Lab-
scale experiment results demonstrate that enhanced flux (due to increasing TMP in constant
pressure tests) leads to a further compression of biopolymer fouling layer which presents
higher specific cake/gel layer resistance. The more compact fouling decreases the backwash-
ing efficiency.

Fouling mechanism analysis is carried out based on the filtration curves in lab-scale experi-
ments. The analysis indicates that according to the delivered biopolymer mass, fouling
100 General Conclusions

mechanisms vary from pore blocking to cake/gel filtration. Although the fouling layer is hy-
draulically reversible, cake/gel filtration is identified as the primary fouling mechanism in
multi cycle filtration test. Based on the observation of TMP development in a UF pilot-plant,
fouling processes can be devided into conditioning fouling, steady fouling and TMP jump
phase when the UF is operated at sustainable operational conditions. The three stages can be
significantly influenced by foulant load mass and/or filtration flux. High foulant load mass or
filtration force (caused by flux) increases the severity of fouling.

7.2 Removal of Organic Foulants in Bio-filtration Processes

In the present work bio-filtration (short-term bank filtration: filtration rate around 0.06 m/h,
retention time around 24 h; slow sand filtration: filtration rate from 0.1 to 0.5 m/h, retention
time from 2 to 8 h) of treated domestic wastewater can remove the particulate, colloidal and
soluble foulants effectively. The performance of the downstream UF can be improved to a
large extent.

The removal mechanism of biopolymers within slow sand filters is investigated in pilot-scale
tests. The removal is observed mainly taking place at the top layer of slow sand filters. In this
active zone, the decrease of biopolymer concentration is found to be relevant to the reduction
of oxygen concentration and pH. Biological processes appear to be important during the re-
moval. Tests on the bio-degradability of biopolymers show that these macromolecular poly-
saccharide-like and protein-like substances can be biologically degraded. Compared to humic
substances they are more readily removed by bio-filtration processes.

Several factors, e.g., filter ripening, filtration rate, temperature, biopolymer concentration in
feedwater and the application of a protective layer are investigated in terms of biopolymer
removal in slow sand filtration processes. It is evident that three months ripening time suffices
for a constant biopolymers removal effect. However, the removal efficiency is not stable un-
der temperature fluctuations. A one-year pilot-scale investigation demonstrated a positive
influence of temperature on the removal. As temperature impacts the biopolymer concentra-
tion in secondary effluent negatively, the winter season leads to a poor performance of slow
sand filter in terms of both biopolymer removal and cycle time. To improve the performance,
lowering filtration rate or constructing a protective layer on the filter surface can be applied.
A lower filtration rate extends the contact time between biopolymers and biofilms in sand
filter, and thus leads to a higher removal effect. Nevertheless, the filtration rate and the re-
moval efficiency are of a less than proportional relationship. To extend the cycle time of a
filter, the effect of BioNet as protective layer material is tested in the present work. Although
the protective layer cannot improve biopolymer removal efficiency, it offers large void space
for foulants and bacteria and mitigates the treatment load in the sand surface layer. Therefore,
the cycle time of a biofilter is extended significantly.

In general, low filtration rates, high temperatures, low biopolymer concentrations in the feed-
water and the application of suitable protective layers can improve the performance of slow
sand filtration.
General Conclusions 101

7.3 Optimization of Operational Conditions

Although the performance of UF is higher in filtering bio-filtrate compared to treating secon-
dary effluent, the operational conditions (filtration flux, filtration duration) influence fouling
development to a large extent. It has been demonstrated that unsuitable filtration durations or
high filtration fluxes lead to a fast TMP increase. The severe fouling is attributed to an in-
creased foulant mass load or more compressed fouling layers, respectively. Based on the pi-
lot-scale tests, a sustainable operation of UF treating slow sand filtrate is achieved using the
Inge Dizzer® 450 module at a filtration flux of 50 L/(m2·h), backwash interval of 10 minutes,
backwash flux of 260 L/(m2·h) and backwash time of 10 -15 seconds.

Different chemicals are used as enhanced backwash reagents in UF. Considering fouling con-
trol and also environmental protection, operational conditions of UF treating slow sand filtrate
can be suggested as following: filtration flux at 90 L/(m2·h), backwash interval of 40 minutes,
backwash flux at 260 L/(m2·h), backwash time around 30 seconds using H2O2 (active H2O2 20
mg/L) as an enhanced backwash chemical. After each backwash the disinfection time can be
set to around 5 minutes. Different cleaning reagents are also used to clean the fouled mem-
brane filtering treated domestic wastewater. Based on the effect on permeability restoration,
the most effective reagent for the removal of irreversible organic fouling is a combination of
base (NaOH, pH 12) and NaClO (active Cl 50 mg/L).

7.4 Research Recommendations

The present work shows a coincidence between the fouling potential tested in lab-scale multi
cycle filtration experiments and that obtained based on pilot-scale UF tests. It indicates that
the lab-scale test can reflect the fouling potential of different water samples in the reality, al-
though it presents mainly the fouling developed in the initial filtration phase. Compared to a
pilot-scale experient, this method is a simple, time saving and cost-effective approach deter-
mining fouling potential. As irreversible fouling depends on the qualitative composition of
biopolymers (Haberkamp, 2008), a multi cycle filtration test is suggested to be used to test the
fouling effect caused by different biopolymers, e.g., proteins, polysaccharides and biopoly-
mers from treated wastewater. Based on the results, a further understanding of fouling
mechanisms by biopolymers is expected to be achieved.

Although several studies investigated the fouling potential of model substances using a simi-
lar method (Jermann et al., 2008; Ye et al., 2005), the organic compounds used were always
bovine serum albumin for proteins, dextran or alginate for polysaccharides with an assump-
tion that these substances are representive in fouling phenomenon. This may lead to a negli-
gence of other important foulants. In addition, certain compounds may play a specific role in
fouling development. For instance, using a confocal laser scanning microscopy (CLSM)
Zhang et al. (2009) identified the important role of proteins adjacent to the membrane surface.
In another study, Chen et al. (2006a) showed that a certain type of polysaccharides played a
more crucial role in fouling than proteins. Thus, several kinds of proteins and polysaccharides
are suggested to be tested using multi cycle tests to evaluate their fouling potential and the
function in fouling development.
102 General Conclusions

The second prospect is to optimize the chemical cleaning procedure (choice of suitable re-
gents). In the present work, a combination of NaOH and NaClO soaking of membrane module
for 24 h is demonstrated as an effective method removing irreversible fouling. However, due
to the disinfection by-products using chlorine, the cleaning water can be a source of new con-
taminants. Based on the understanding of the contribution from biopolymers in irreversible
fouling, more environmental friendly and specialized reagents, e.g., enzymes (Poele, 2006)
and ethanol (Tian et al., 2009), are suggested to be tested in further membrane cleaning ex-
periments. To improve foulant removal efficiency, circulating cleaning solutions through the
system instead of soaking is also suggested for dynamic cleaning promotes more mass trans-
fer of chemicals to the fouling layer and of the reaction products back to the bulk liquid phase
than static soaking.
Appendix 103

8 Appendix

8.1 LC-OCD-UVD-OND System and LC-OCD Chromatogram

autosampler by pass

Gräntzel thin IR
film reactor detector
size
exclusion
column

wastewater
eluent
UV220
detector
nitrogen UV254
gas detector

additional
oxidation

acid

A-Fig. 1 Set-up of the LC-OCD-UVD-OND system

Measurement Process
During the chromatogram measurement one mL water sample is injected into the eluent flow
(A-Fig. 1) and is carried into the size exclusion column (Alltech-Grom GmbH, resin material:
Toyopearl® HW-55s which are synthesized by co-polymerization of ethylene glycol and
methacrylate-type polymers, Tosoh Bioscience, Tokio/Japan). The separated organic com-
pounds in the column are transported into the UV254 detector (Wellchrom K-200, Knauer
GmbH, Berlin) to measure the UV absorbance. Subsequently, the eluent flow and acid flow
enter the reactor at the upper end where they are thoroughly mixed by means of the Teflon
pins in the rotor. The upper third of the reactor is UV-shielded and the carbon dioxide from
inorganic source is stripped out upwards using the carrier gas (nitrogen). The residence time
of the liquid in this area is about 20 seconds. The liquid continues way down due to gravity.
The UV oxidation (λ = 185 nm) forms carbon dioxide from the dissolved organic compounds.
This carbon dioxide is carried downwards stripped through the nitrogen gas and is measured
in a no-dispersive infrared detector cuvette (SIEMENS ULTRAMAT 6E, Siemens AG, Mu-
nich). The residence time of the liquid in the reactor is about 60 seconds. During the oxidation
process, the thinness of the liquid film (10 – 20 µm) and the active stirring by the Teflon pins
(500 rpm) guarantees a rapid oxidation and stripping process. The wastes after the thin film
oxidation is collected in a small reservoir and then pumped to the additional oxidation process
(185 nm, Well chram K-2000). The oxidized N-content organic form nitrate (NO3-) and the
104 Appendix

corresponding nitrogen content can be quantified measuring the absorbance at λ of 220 nm

(Wellchrom K-2000, Knauer GmbH, Berlin). Besides the measurement of a chromatogram,
the total C, N contents and UV absorbance of a water sample can be determined by injecting
0.1 mL water sample into eluent flow through by pass (A-Fig. 1).

A-Tab. 1 Eluent and acid solution used in LC-OCD system

Eluent di-sodium hydrogen phosphate dehydrate (Na2 HPO4·2H2O) 8.4 mmol/L

Potassium di-hydrogen phosphate KH2PO4 18.4 mmol/L
pH 6.6
flux 1 mL/min
Acid Ortho-phosphoric acid 60 mmol/L
pH 1.8
flux 0.5 mL/min

Quantification of Biopolymer Content

Biopolymers generally consist of organics with high molecular weights (HMW), e.g. polysac-
charides, proteins and other organic colloids. The fractions after the biopolymer peak are hu-
mic substances, building blocks, acids and low molecular weight (LMW) humics and LMW
neutrals (A-Fig. 2). For LC-OCD analysis water samples are pre-filtered using cellulose ni-
trate (Sartorius Stedim Biotech GmbH) with a pore size of 0.45 µm. The DOC concentration
in the water sample is suggested to be diluted using pure water into the range of 2 to 5 mg/L
(per manufacturer).

12
secondary effluent - OCD humic
substances
secondary effluent - UVD
9
building
OCD, UVD Signal [AU]

secondary effluent - OND 100

blocks
OND Signal [AU]

6
Biopolymers

50
3 LMW acids
and neutrals

0 0
10 50 90
Elution time [min]

A-Fig. 2 LC-OCD-UVD-OND chromatogram of treated domestic wastewater

Appendix 105

Using the software FIFIKUS® (DOC Labor Dr Huber, Karlsruhe) the acquired data can be
integrated to quantify the content of biopolymers. In the present dissertation the biopolymer
peak begins at elution time of 20 minutes till to the time point where UV254 signal jumps (A-
Fig. 2). As in the biopolymer peak there is only a weak UV254 signal peak, the jump of the
absorbance strength indicates the presence of humic peak due to the high contents of UV254
active substances in the corresponding region. The carbon, nitrogen content and UV254 ab-
sorbance strength of biopolymers can then be integrated in the determined boundary.

Calibration of the Detectors and Size Exclusion Column

90

y = 0.0413x
2
R = 0.9996
60
Area unit

30

0
0 500 1000 1500 2000 2500
DOC [µg]

A-Fig. 3 Calibration of the carbon detector of the LC-OCD system (standard solution of potassium
hydrogen phthalate: 2.1254g/L = 1000mg DOC/L, injection volumes are 0.05, 0.1 and 0.2 mL within
concentration range of 0.005 to 10 mg/L)

The organic carbon, nitrogen and UV254 absorbance detectors can be calibrated using potas-
sium hydrogen phthalate (PHP, C8H5KO4), potassium nitrate (KNO3) and standard humic acid
(HA) and fulic acid (FA) from international humic substance society (IHSS). The calibration
can be carried out using by-pass with injecting specific volume of corresponding standard
solutions. In A-Fig. 3 is the DOC content and corresponding area unit of the OCD chroma-
togram.
106 Appendix

Molecular weight [Da] 45000

30000

15000

0
30 35 40 45 50 55 60 65 70
Elution time [min]

A-Fig. 4 Molecular weight distribution and corresponding elution time of polyethylenglycol (MW
range of 106 to 40 000 da) using size exclusion column HW 55s

The size exclusion effect the column HW 55s can be calibrated using polyethylenglycol
(PEG) with a molecular weight range from 106 to 40 000 da (g/mol). A-Fig. 4 demonstrates
that there is a negative correlation of molecular weight and elution time using the correspond-
ing PEG standard. Since in treated domestic wastewater the characteristics of molecules are
different to that of the PEG standards, in the present work the biopolymers are preferred to be
presented using elution time instead of molecular weight. Simultaneously, it should be stated
that within the elution time of biopolymer peak the detected organics are generally large
molecules.
Appendix 107

8.2 Amicon Cell Results

A-Tab. 2 Amicon cell filtration determining the filterability of water samples (corresponding to Fig. 5-
8)
Date Mar.06.2007 07.08.2007 Nov.16.2006 Dec.16.2006 Apr.08.2007 Apr.04.2007 Mar.06.2007 July 12 2007
BC [m g C/L] 0.59 0.11 0.48 0.38 0.28 0.25 0.17 0.07
water sam ple SE SSF SE SE SE SE SSF SSF
filtration tim e [s] V [m L] V [m L] V [m L] V [m L] V [m L] V [m L] V [m L] V [m L]
0 0 0 0 0 0 0 0 0
60 27.36 38.11 29.37 25.83 31.76 30.65 29.05 25.71
120 48.82 74.13 52.42 48.56 59.08 58.35 56.04 51.43
180 66.38 108.9 72.05 68.27 83.38 83.82 81.27 76.5
240 81.4 142.61 89.3 86.03 105.17 107.09 105.04 101.14
300 95.05 174.91 104.8 102.48 125.26 128.77 127.42 125.44
360 107.59 206.01 119.21 117.75 143.7 149.23 148.78 149.3
420 119.16 236.09 132.58 131.98 161.13 168.43 168.95 172.83
480 129.9 265.05 144.53 145.9 177.41 186.69 188.34 196.05
540 140.07 293.1 156.45 158.31 192.83 204.07 206.92 218.84
600 149.68 320.3 167.59 170.52 207.77 220.62 224.74 241.46
660 158.82 346.52 178.28 182.21 222.13 236.42 241.96 263.64
720 167.47 372.22 188.4 193.33 235.92 251.7 258.67 285.52
780 175.86 396.99 198.06 204.18 249.14 266.25 274.78 307.03
840 183.82 421.09 207.44 214.57 261.62 280.39 290.35 328.32
900 191.55 444.56 216.46 224.72 273.76 294.02 305.53 349.28
960 198.93 467.52 225.24 234.25 285.61 307.15 320.27 370.02
1020 206.12 489.75 233.69 243.9 297.11 319.98 334.56 390.52
1080 213.14 511.62 241.96 253.18 308.25 332.46 348.59 410.69
1140 219.93 532.92 249.98 262.16 319.16 344.52 362.25 430.72
1200 226.58 553.76 257.84 270.98 329.77 356.29 375.57 450.44
1260 233.05 574 265.58 279.49 340.13 367.76 388.59 469.85
1320 239.21 593.91 273.03 287.82 350.23 378.96 401.32 489.07
1380 245.35 613.41 280.39 296.06 360.13 389.94 413.86 508.11
1440 251.39 632.6 287.64 304.1 369.78 400.68 426.02
1500 257.25 651.3 294.66 311.81 379.24 411.12 438.11
1560 262.97 669.75 301.49 319.53 388.53 421.42 449.89
1620 268.64 687.78 308.32 327.14 397.63 431.4 461.38
1680 274.16 705.71 314.9 334.59 406.51 441.26 472.78
1740 279.62 723.36 321.43 341.96 415.32 450.98 484
1800 284.92 740.59 327.89 349.23 423.86 460.43 494.87
1860 290.19 757.52 334.14 356.18 432.41 469.79 505.81
1920 295.33 774.22 340.43 363.17 440.71 478.97
1980 300.42 790.69 346.43 370.12 448.86 488.02
2040 305.38 806.85 352.42 376.88 456.99 496.84
2100 310.36 822.88 358.46 383.58 464.87 505.61
2160 315.23 838.65 364.28 390.13 472.77
2220 320.05 854.14 369.97 396.65 480.44
2280 324.7 869.52 375.71 403 488.04
2340 329.37 884.55 381.25 409.32 495.54
2400 333.94 899.46 386.82 415.58 503
2460 338.51 914.15 392.32 421.74 510.3
2520 343.04 928.66 397.69 427.92
2580 347.46 943.1 403.03 433.93
2640 351.92 957.19 408.34 439.9
2700 356.19 971.2 413.51 445.69
2760 360.5 985.06 418.71 451.66
2820 364.7 998.77 423.88 457.4
2880 368.9 1012.35 428.89 463.04
2940 373.04 1025.76 434.01 468.77
3000 377.19 1039.01 438.98 474.49
3060 381.28 1052.09 443.88 480.17
3120 385.22 1065.24 448.81 485.71
3180 389.23 1077.95 453.72 491.21
3240 393.24 1090.67 458.59 496.67
3300 397.1 1103.32 463.4 502.11
3360 400.97 1115.86 468.14
3420 404.88 1128.12 472.89
3480 408.67 1140.29 477.54
3540 412.45 1152.45 482.2
3600 416.19 1164.46 486.84
3660 419.89 1176.46
3720 423.59 1188.31
3780 427.31 1199.93
3840 430.93 1211.59
3900 434.54 1223.18
3960 438.11 1234.55
4020 441.69 1245.98
4080 445.15 1257.23
4140 448.67 1268.42
4200 452.19 1279.44
4260 455.59 1290.44
4320 459.01 1301.38
4380 462.42 1312.16
4440 465.83 1323.01
4500 469.24 1333.79
4560 472.52 1344.33
4620 475.9
4680 479.17
4740 482.46
4800 485.74
4860 488.94
4920 492.22
4980 495.36
5040 498.58
108 Appendix

A-Tab. 3 Analysis of fouling mechanisms during a dead-end filtration (corresponding to Fig. 5-10)
Mechanism analysis based on pore blocking model
Experiment date: November 16.2006
Biopolymer concentration in secondary effluent: 0.48 mg C/L, filtration volume 500 mL

Filtration in the test calculated for simulation calculation

time: s volume: L t (s) v (L) dt/dv d2t/dv2 log(dt/dv) log(d2t/dv2) n

0 0 0.0E+00 0
60 0.03 1.9E+01 0.01 1889.71
120 0.05 4.0E+01 0.02 2066.32 17661.01 3.32 4.25
180 0.07 6.2E+01 0.03 2255.76 18943.73 3.35 4.28 0.80
240 0.09 8.7E+01 0.04 2456.10 20034.52 3.39 4.30 0.66
300 0.10 1.1E+02 0.05 2665.63 20952.46 3.43 4.32 0.55
360 0.12 1.4E+02 0.06 2882.78 21715.66 3.46 4.34 0.46 4000
420 0.13 1.7E+02 0.07 3106.20 22341.34 3.49 4.35 0.38
The relationship between filtration time
480 0.14 2.1E+02 0.08 3334.65 22845.77 3.52 4.36 0.31 and permeate volume used to analyse
540 0.16 2.4E+02 0.09 3567.10 23244.31 3.55 4.37 0.26 fouling mechanism
600 0.17 2.8E+02 0.1 3802.61 23551.38 3.58 4.37 0.21
660 0.18 3.2E+02 0.11 4040.42 23780.50 3.61 4.38 0.16 3000
y = -128434x6 + 178046x5 - 96934x4 +
720 0.19 3.6E+02 0.12 4279.86 23944.24 3.63 4.38 0.12
26761x3 + 8092.9x2 + 1806.2x

Filtration time [s]

780 0.20 4.1E+02 0.13 4520.40 24054.26 3.66 4.38 0.08 R2 = 1
840 0.21 4.6E+02 0.14 4761.61 24121.29 3.68 4.38 0.05
900 0.22 5.1E+02 0.15 5003.17 24155.14 3.70 4.38 0.03 2000
960 0.23 5.6E+02 0.16 5244.81 24164.68 3.72 4.38 0.01
1020 0.23 6.1E+02 0.17 5486.39 24157.88 3.74 4.38 -0.01
1080 0.24 6.7E+02 0.18 5727.81 24141.77 3.76 4.38 -0.02
1140 0.25 7.3E+02 0.19 5969.03 24122.46 3.78 4.38 -0.02
1000
1200 0.26 7.9E+02 0.2 6210.09 24105.12 3.79 4.38 -0.02
1260 0.27 8.6E+02 0.21 6451.03 24094.02 3.81 4.38 -0.01
1320 0.27 9.2E+02 0.22 6691.95 24092.48 3.83 4.38 0.00
1380 0.28 9.9E+02 0.23 6932.98 24102.93 3.84 4.38 0.01
1440 0.29 1.1E+03 0.24 7174.25 24126.84 3.86 4.38 0.03 0
1500 0.29 1.1E+03 0.25 7415.90 24164.77 3.87 4.38 0.05 0 0.1 0.2 0.3 0.4 0.5
1560 0.30 1.2E+03 0.26 7658.06 24216.36 3.88 4.38 0.07 Permeate volume [L]
1620 0.31 1.3E+03 0.27 7900.86 24280.31 3.90 4.39 0.08
1680 0.31 1.4E+03 0.28 8144.41 24354.41 3.91 4.39 0.10
1740 0.32 1.5E+03 0.29 8388.76 24435.53 3.92 4.39 0.11
1800 0.33 1.5E+03 0.3 8633.96 24519.59 3.94 4.39 0.12
1860 0.33 1.6E+03 0.31 8879.97 24601.61 3.95 4.39 0.12
1920 0.34 1.7E+03 0.32 9126.73 24675.67 3.96 4.39 0.11
1980 0.35 1.8E+03 0.33 9374.08 24734.93 3.97 4.39 0.09
2040 0.35 1.9E+03 0.34 9621.80 24771.64 3.98 4.39 0.06
2100 0.36 2.0E+03 0.35 9869.57 24777.11 3.99 4.39 0.01
2160 0.36 2.1E+03 0.36 10116.98 24741.71 4.01 4.39 -0.06
2220 0.37 2.2E+03 0.37 10363.53 24654.92 4.02 4.39 -0.15 1.E+05
2280 0.38 2.3E+03 0.38 10608.59 24505.27 4.03 4.39 -0.26
2340 0.38 2.4E+03 0.39 10851.39 24280.37 4.04 4.39 -0.41 flow behavior index n
2400 0.39 2.5E+03 0.4 11091.06 23966.92 4.04 4.38 -0.59
d2t/dv2

2460 0.39 2.7E+03 0.41 11326.57 23550.67 4.05 4.37 -0.83

2520 0.40 2.8E+03 0.42 11556.73 23016.47 4.06 4.36 -1.14
2580 0.40 2.9E+03 0.43 11780.21 22348.22 4.07 4.35 -1.54
2640 0.41 3.0E+03 0.44 11995.50 21528.93 4.08 4.33 -2.06
2700 0.41 3.1E+03 0.45 12200.91 20540.65 4.09 4.31 -2.77 1.E+04
2760 0.42 3.3E+03 0.46 12394.55 19364.52 4.09 4.29 -3.74 1.E+03 1.E+04 1.E+05
2820 0.42 3.4E+03 0.47 12574.36 17980.76 4.10 4.25 -5.15 dt/dv
2880 0.43 3.5E+03 0.48 12738.05 16368.66 4.11 4.21 -7.26
2940 0.43 3.6E+03 0.49 12883.11 14506.59 4.11 4.16 ####
3000 0.44
3060 0.44
3120 0.45
3180 0.45
3240 0.46
3300 0.46
3360 0.47
3420 0.47
3480 0.48
3540 0.48
3600 0.49
Appendix 109

A-Tab. 4 Amicon cell test with five filtration cycles for setting up the relationship between biopolymer concentration in the influent and irreversible fouling filter-
ing corresponding water sample (corresponding to Fig. 4-16)

Experiment date May 8 2009 April 29 2009 April 17 2009 Mar. 17 2009 Feb. 25 2009 Dec. 31 2008 Dec.05 2008 April 1 2008 Nov. 5 2007 Oct. 8 2007
Water sample SSF SE SSF SE SE SE SE SSF-UF permeate SSF SE
Biopolymer concentration (mg C/L) 0.19 0.56 0.5 0.43 0.23 0.25 0.49 0.05 0.08 0.42
filtration TMP (bar) 1 1 1 1 1 1 1 1 1 1

Flux filtering pure water (LMH) 698.3 725.9 666.8 625.9 630.4 583.3 696.1 765.9 695.8 706.9
Flux of pure water (m3/m2S) 1.94E-04 2.02E-04 1.85E-04 1.74E-04 1.75E-04 1.62E-04 1.93E-04 2.13E-04 1.93E-04 1.96E-04
Filtration temperature ((°C) 21 21 21 21 21 21 21 21 21 21
Water viscosity ( ŋ) 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04 9.82E-04
Filterability 0.38 0.15 0.16 0.14 0.35 0.30 0.17 0.83 0.50 0.21
Flux recovery (%) 80.3 66.1 69.7 69.0 82.1 82.0 76.0 87.0 83.6 76.8

R mem (1/m) 5.30E+10 5.10E+10 5.55E+10 5.91E+10 5.87E+10 6.35E+10 5.32E+10 4.83E+10 5.32E+10 5.24E+10
Flux after 500 mL (LMH) 268.0 109.3 108.9 89.9 218.9 176.0 118.5 634.5 347.0 150.7
R total (fouling after 500+ R mem) (1/m) 1.38E+11 3.39E+11 3.40E+11 4.12E+11 1.69E+11 2.10E+11 3.12E+11 5.83E+10 1.07E+11 2.46E+11
R fouling after 500 mL (1/m) 8.51E+10 2.88E+11 2.84E+11 3.53E+11 1.10E+11 1.47E+11 2.59E+11 1.00E+10 5.35E+10 1.93E+11
Pure water flux after backwashing (500mL)(LMH) 629.0 590.8 589.8 572.5 573.2 548.5 590.8 744.3 632.5 650.7
Irreverble fouling after 500 mL (1/m) 5.84E+09 1.17E+10 7.25E+09 5.51E+09 5.86E+09 4.02E+09 9.48E+09 1.41E+09 5.32E+09 4.53E+09

Flux after 1000 mL (LMH) 258.8 106.2 106.8 91.6 212.4 171.0 113.9 627.2 368.6 147.6
R total (fouling after 1000+ R mem) (1/m) 1.43E+11 3.49E+11 3.47E+11 5.77E+11 5.77E+11 5.77E+11 3.25E+11 5.90E+10 1.00E+11 2.51E+11
R fouling after 1000 mL (1/m) 9.00E+10 2.98E+11 2.91E+11 4.94E+11 4.94E+11 4.94E+11 2.72E+11 1.07E+10 4.72E+10 1.98E+11
Pure water flux after backwashing (1000mL) (LMH) 604.4 554.4 560.0 532.3 561.9 528.5 536.7 725.3 615.3 619.1
Irreverble fouling after 1000 mL (1/m) 8.23E+09 1.58E+10 1.06E+10 1.04E+10 7.16E+09 6.58E+09 1.58E+10 2.71E+09 6.96E+09 7.43E+09

Flux after 1500 mL (LMH) 250.2 103.3 108.5 89.9 203.2 171.4 113.7 616.5 355.2 150.5
R total (fouling after 1500+ R mem) (1/m) 1.48E+11 3.58E+11 3.41E+11 4.12E+11 1.82E+11 2.16E+11 3.26E+11 6.00E+10 1.04E+11 2.46E+11
R fouling after 1500 mL (1/m) 9.49E+10 3.07E+11 2.86E+11 3.53E+11 1.23E+11 1.52E+11 2.72E+11 1.17E+10 5.10E+10 1.94E+11
Pure water flux after backwashing (1500mL) (LMH) 581.7 523.0 528.4 479.7 534.6 507.0 521.6 722.0 596.9 588.7
Irreverble fouling after 1500 mL (1/m) 1.06E+10 1.98E+10 1.45E+10 1.80E+10 1.05E+10 9.54E+09 1.78E+10 2.94E+09 8.82E+09 1.05E+10

Flux after 2000 mL (LMH) 249.8 102.0 108.1 85.9 205.3 163.9 112.1 602.3 352.7 145.7
R total (fouling after 2000+ R mem) (1/m) 1.48E+11 3.63E+11 3.43E+11 4.31E+11 1.80E+11 2.26E+11 3.30E+11 6.15E+10 1.05E+11 2.54E+11
R fouling after 2000 mL (1/m) 9.52E+10 3.12E+11 2.87E+11 3.72E+11 1.22E+11 1.62E+11 2.77E+11 1.31E+10 5.18E+10 2.02E+11
Pure water flux after backwashing (2000mL) (LMH) 574.4 490.3 515.7 478.4 515.8 496.8 528.3 692.6 584.2 562.1
Irreverble fouling after 2000 mL (1/m) 1.14E+10 2.45E+10 1.63E+10 1.82E+10 1.30E+10 1.10E+10 1.69E+10 5.11E+09 1.02E+10 1.35E+10

Flux after 2500 mL (LMH) 243.1 101.2 104.1 84.9 200.5 156.6 111.6 571.4 367.5 144.5
R total (fouling after 2500+ R mem) (1/m) 1.52E+11 3.66E+11 3.56E+11 4.36E+11 1.85E+11 2.36E+11 3.32E+11 6.48E+10 1.01E+11 2.56E+11
R fouling after 2500 mL (1/m) 9.92E+10 3.15E+11 3.00E+11 3.77E+11 1.26E+11 1.73E+11 2.78E+11 1.65E+10 4.75E+10 2.04E+11
Pure water flux after backwashing (2500mL) (LMH) 560.6 479.8 464.6 431.5 517.4 478.4 528.9 663.2 581.7 543.2
Irreverble fouling after 2500 mL (1/m) 1.30E+10 2.62E+10 2.42E+10 2.67E+10 1.28E+10 1.39E+10 1.68E+10 7.49E+09 1.04E+10 1.58E+10
110 Appendix

A-Tab. 5 Calculation of biopolymer mass in fouling layer and corresponding specific fouling resistance during UF

Biopolymer concentration Biopolymer mass Fouling resistance Specific fouling resistance

Pressure Cycle (mg C/L) (mg) (1011 1/m) (1016 m/kg)
Feedwater Concentrate Permeate
(bar) (0.7 L) (0.2 L) (0.5 L)
0.5 1 0.42 0.54 0.1 0.27 1.35 1.43
2 0.42 0.52 0.09 0.29 1.30 1.28
3 0.42 0.50 0.1 0.29 1.29 1.27
4 0.42 0.52 0.1 0.28 1.37 1.41
5 0.42 0.54 0.11 0.26 1.42 1.54
1 0.42 0.44 0.14 0.27 1.91 2.01
1.0 2 0.42 0.52 0.1 0.28 1.92 1.97
3 0.42 0.55 0.1 0.27 1.94 2.07
4 0.42 0.44 0.11 0.30 1.96 1.87
5 0.42 0.50 0.12 0.27 2.03 2.18
Appendix 111

8.3 Investigation of Slow Sand Filtration on Biopolymer Removal

Biopolymer concentration [mg C/L] O2 and DOC concentration [mg/L]

0.0 0.2 0.4 0.6 0 5 10 15
Distance to surface [cm]

0 0

O2
30 30
DOC

60 a) 60 b)

90 90

pH UV254 [1/m]
6.5 7.5 8.5 10 20 30 40
Distance to surface [cm]

0 0

30 30

60 c) 60 d)

90 90
Turbidity [NTU] SUVA [L/m-mg]
0.0 1.0 2.0 2.0 2.2 2.4 2.6 2.8 3.0
Distance to surface [cm]

0 0

30 30

60 e) 60 f)

90 90

A-Fig. 5 Removal profile of a) biopolymer (n=14), b) dissolved oxygen and DOC (n=10), c) pH
(n=14), d) UV254 (n=14) and e) turbidity (n=14) f) SUVA (n=10) in slow sand filter operated at filtra-
tion of 0.5 m/h, period: July 10 2007 to November 20 2007)
112 Appendix

1.5
Biopolymer concentration [mg C/L]

0.5

0
influent anaerobic zone anoxic zone aerobic zone secondary effluent

A-Fig. 6 Biopolymer concentration variation in water phase along wastewater treatment procedure in
WWTP Ruhleben (Average value of three series of water samples, July 2007 and November 2007)

A-Fig. 6 presents the variations of biopolymer concentration along the wastewater treatment
process in WWTP Ruhleben. The biopolymer concentration is always higher in the influent
than in the effluent. The results imply that part of the biopolymers detected in the SE could be
biopolymers from untreated wastewater.

Measurement of Biological Degradability Biopolymers

Biological degradability of biopolymers and humic substances was tested using BDOC de-
termination method developed by Servais et al., 1987 and Block et al., 1992 with some revi-
sions. Suspended biomass inoculation (SBI) and blinds test were conducted simultaneously.
SE 900 mL is sterilized by filtration through 0.2 µm cellulose acetate membrane (Sartorius).
Inoculation was undertaken as following: 300 g sand was taken from sand filter at depth 10
cm and strongly shaked in 900 mL SE. After 1 hour sedimentation, 9 mL supernatant was
filtered through 1.2 µm cellulose nitrate membrane (Sartorius) and then added to the sterilized
water sample. Incubation and blind test occurred in darkness at 20±1 oC for a period of 30
days, during which biopolymer concentration and UV254 was measured daily at the first two
days and then weekly.
Appendix 113

A-Tab. 6 Biopolymer concentration in SE and SSFs (Oct.26 2006 to Dec. 31 2007)

Date Temperature BC in SE BC in SSF 0.1 m/h BC in SSF 0.2-0.25m/h BC in SSF 0.4-0.5m/h

ºC mg C/L mg C/L mg C/L mg C/L
10.26.06 15 0.52 0.24 0.21 0.25
11.01.06 16 0.43 0.33 0.29 0.294
11.13.06 15 0.40 0.23 0.12 0.14
11.18.06 17 0.48 0.23 0.22 0.26
11.24.06 16 0.43 0.31 0.25 0.32
12.01.06 18 0.68 0.56 0.43 0.42
12.06.06 19 0.40 0.35 0.33 0.32
12.14.06 18 0.43 0.38 0.29 0.32
12.20.06 19 0.44 0.25 0.24 0.32
01.05.07 16 0.62 0.41 0.46
01.10.07 17 0.54 0.36 0.30 0.49
01.17.07 15.3 0.34 0.32 0.24 0.3
01.24.07 18 0.30 0.21 0.21 0.21
01.31.07 12 0.42 0.24 0.27 0.31
02.07.07 16.1 0.41 0.24 0.29
02.14.07 17 0.25 0.13 0.11
02.21.07 17 0.37 0.13 0.27
03.01.07 15 0.58 0.18 0.37
03.08.07 18 0.59 0.22 0.41
03.14.07 19 0.41 0.19 0.21
03.21.07 18 0.31 0.12 0.18
03.25.07 17.5 0.30 0.12
04.01.07 18.2 0.23 0.10 0.14
04.04.07 18.3 0.41 0.14 0.19
04.10.07 18.8 0.41 0.17 0.14
04.12.07 19.2 0.48 0.17
04.24.07 20.9 0.33 0.13 0.15
05.01.07 18.3 0.36 0.12 0.12
05.13.07 19.8 0.49 0.19 0.18
05.28.07 22.6 0.33 0.09 0.12
05.30.07 20.9 0.17 0.08 0.09
06.07.07 24.2 0.18 0.11 0.10
06.12.07 25.6 0.19 0.08 0.10
06.25.07 22 0.13 0.09 0.08
06.29.07 20.7 0.15 0.07
07.03.07 21.5 0.30 0.10 0.09
07.05.07 19.9 0.26 0.10
07.10.07 19.8 0.38 0.12 0.25
07.19.07 23.1 0.32 0.07 0.11
07.29.07 21.4 0.34 0.10 0.16
07.31.07 19.4 0.27 0.05 0.1
08.04.07 21 0.26 0.09 0.09
08.14.07 22 0.23 0.08 0.07
08.21.07 21.6 0.36 0.11 0.14
08.28.07 21 0.23 0.07 0.05
09.04.07 20.0 0.20 0.04 0.05
09.17.07 19.6 0.35 0.08 0.08
09.18.07 19.6 0.32 0.08 0.09
10.01.07 17.9 0.23 0.07
10.08.07 18 0.32 0.09 0.12
10.23.07 15.7 0.33 0.09
10.30.07 15.5 0.46 0.17 0.09
11.05.07 15 0.45 0.08 0.17
11.13.07 12.5 0.44 0.07 0.15
11.20.07 13 0.79 0.38 0.31
11.27.07 14 0.59 0.36 0.43
12.07.07 12.5 0.70 0.48 0.62
12.12.07 12.1 0.69 0.46
12.19.07 12 0.74 0.50
12.28.07 13 0.75 0.56
12.31.07 13.9 0.79
114 Appendix

(a) 12 (b) 12
secondary effluent secondary effluent
9 Jan.31.2007 9 Mar.22.2009
OC-Signal

OC-Signal
biofiltrate 8 m/h biofiltrate 8 m/h
6 Jan.31.2007 6
Mar.22.2009

3 3

0 0
30 60 90 120 30 60 90 120
Elution time [min] Elution time [min]

(c) 12 (d) 12
secondary effluent secondary effluent
9 Mar.22.2009 9 Mar.22.2009

OC-Signal
OC-Signal

biofiltrate 0.25 m/h

biofiltrate 4 m/h Mar.22.2009
6 6
Mar.22.2009
biofiltrate 0.05 m/h
3 3 Mar.22.2009

0 0
30 60 90 120 30 60 90 120
Elution time [min] Elution time [min]

A-Fig. 7 Influence of filtration rate on biopolymer removal effect (a) LC-OCD chromatogram of SE
and bio-filtrate operated at filtration rate of 8 m/h (Jan. 31 2007, start up November 2007), (b) LC-
OCD chromatogram of SE and bio-filtrate operated at filtration rate of 8 m/h (March 22 2009, start up
November 2007), (c) LC-OCD chromatogram of SE and bio-filtrate operated at filtration rate of 4 m/h
(March 22 2009, start up April 2008), (d) LC-OCD chromatogram of SE and bio-filtrate operated at
filtration rate of 0.25 m/h (March 22 2009, start up March 1 2009) and 0.05 m/h (March 22 2009, start
up Jan. 1 2009)

A-Fig. 7 shows biopolymer removal effect using bio-filtration operated at filtration rates from
0.05 m/h to 8 m/h. The bio-filters operated at filtration rate of 8 m/h and 4 m/h were operated
within the project ‘Barrieren’ which was supported by Berliner Wasser Betriebe (BWB). The
bio-filter operated at 8 m/h is consisted of 1 m anthracite layer (size: 1.4 – 2.5 mm) and a 0.8
m sand layer (size: 0.7 – 1.1 mm). During operation the filter was periodically backwashed
using filtrate. The comparison of A-Fig. 7 (a) and (b) show that after longer than 16 months
operation, bio-filter operated at the corresponding conditions cannot remove biopolymers ef-
fectively although they show effective removal of turbidity (on average the turbidity in SE
was 2 – 4 NTU, in bio-filtrate it was lower than 0.6 NTU). The bio-filter operated at 4 m/h
was filled with filtralite (Maxit group, Norway) which is expended clay with a core size of 2 –
5 mm. The filter layer was around 0.9 m high. In the filter was operated in anoxic condition.
A-Fig. 7 (c) shows that after longer than half year operation, the filter exhibited little effect in
biopolymer removal. A-Fig. 7 (d) displays the biopolymer removal effect of slow sand filtra-
tion operated at filtration rate of 0.25 m/h (started from Feb. 01 2009, filter material BioNet
0.2 m in depth, sand 0.6 m in depth (core size: 2 – 4 mm)) and 0.05 m/h (started from Jan. 01
2009, sand layer 0.6 m with sand core size of 2 – 4 mm)).
Appendix 115

8.4 Figure contents

Fig. 1-1 Annual publications on membrane fouling (source: Science Direct).......................... 2
Fig. 2-1 Schematic illustration of a membrane separation process........................................... 5
Fig. 2-2 Application of low-pressure membranes (a) as a post-treatment unit after
conventional biological wastewater treatment processes for water reclamation, (b) in
a MBR system to produce reuse water ....................................................................... 9
Fig. 2-3 Illustration of reversible and irreversible fouling in UF operated at constant flux ... 10
Fig. 2-4 Schematic diagram of a fouled membrane by various fouling mechanisms (adapted
from Roorda, 2004, revised) ..................................................................................... 10
Fig. 2-5 Fouling mechanisms related with blocking laws (a) complete blocking, (b)
intermediate blocking, (c) standard blocking, (d) cake filtration.............................. 12
Fig. 2-6 Typical organic constituents in biological treated domestic wastewater and their size
ranges (Adapted from Levine et al., 1985) ............................................................... 14
Fig. 3-1 Schematic illustration of the Amicon system (V1: nitrogen gas valve, V2: pressure
reducer valve, V3: relief valve of the feed reservoir, V4: relief valve of Amicon cell)
................................................................................................................................... 31
Fig. 3-2 (a) The photo of the W.E.T UF pilot plant system equipped with Inge Dizzer® 450
module and (b) the schematic diagram of the UF system ......................................... 32
Fig. 3-3 Illustration of short-term bank filtration in the artificial lake in WWTP Beixiaohe . 33
Fig. 3-4 (a) The photo of the slow sand filtration system in WWTP Ruhleben, (b) the
schematic diagram of a slow sand filter .................................................................... 34
Fig. 3-5 (a) Diagram of a slow sand filter using BioNet as a protect layer (b) foto of BioNet
carriers....................................................................................................................... 35
Fig. 4-1 Influence of size fraction in SE and its SSF on filterability. (a) SE, SE after 1.2 µm
filter, after 0.45 µm filter and after UP150; (b) SSF at filtration rate 0.25m/h, SSF
after 1.2 µm filter, after 0.45 µm filter and UP150................................................... 41
Fig. 4-2 Fouling resistance attributed from different fractioned substances in SE and its SSF
................................................................................................................................... 41
Fig. 4-3 Liquid chromatography (LC-OCD) of SE and its UF permeate ............................... 42
Fig. 4-4 Correlation of filterability and biopolymer peak of different water samples (a)
Normalized stirred-cell UF flux decline curves using SE, SSFs and UF permeate
over cumulative permeate volume of 500 mL. (b) Biopolymer peaks of SE, UF
permeate of SE and SSFs .......................................................................................... 43
Fig. 4-5 (a) Biopolymer concentration of different water samples, (b) normalized flux of
correspond ing water samples, UP150, TMP 1 bar, (C) fouling resistance of
corresponding water samples (flux data in detail see section 8.3, A-Tab.2) ............ 44
Fig. 4-6 Relationship between biopolymer concentration and (a) filterability (b) fouling
resistance under permeate volume 150 mL, 250 mL and 500 mL. (n=26, water
samples were taken randomly from Nov.2006 to July 2007 in WWTP Ruhleben,
Berlin, Germany)....................................................................................................... 45
Fig. 4-7 Influence of delivered biopolymer amount on the normalized flux decline of SE and
SSF. (a) Normalized flux decline under the same permeate volume; (b) Normalized
flux decline under the same delivered biopolymer amount ...................................... 45
Fig. 4-8 (a) Influence of biopolymer concentration on fouling effect, (b) flux recovery at
different biopolymer concentrations (BC: biopolymer concentration, Feed water is
SE, SSF and SE UF permeate, operating pressure 1 bar, permeate volume 500 mL;
back wash using salt solution 50 mL at 1 bar) .......................................................... 47
Fig. 4-9 Comparison of the LC-OCD chromatograms of SE, permeate after UP150,
supernatant and back wash water (Permeate volume 500 mL, concentrate 200 mL,
116 Appendix

backwash water volume 50 mL, Dilution effect of different water samples was
regulated)................................................................................................................... 47
Fig. 4-10 (a) linear correlation between biopolymer concentration and reversibility after
filtration cycle 5, (b) potential correlation between biopolymer concentration and
irreversible fouling resistance after different filtration cycles (from October 8th
2007 to May 8th 2009, n=10, biopolymer concentration from 0.05 to 0.56 mg C/L).
................................................................................................................................... 48
Fig. 4-11 TMP increase of UF pilot plant filtrating (a) SE and (b) SSFs at different
biopolymer concentrations (BC: biopolymer concentration. Operational condition of
UF: flux 50 LMH, BWI 10 minutes, BWT 10 seconds, BWF 260 LMH). .............. 49
Fig. 4-12 Characteristic curves reflecting fouling mechanism for biopolymer fouling in UF at
different biopolymer concentrations (Data are taken from Fig. 4-5 (b) and Fig. 4-7
(a), treated with polynomials of degree of six and analyzed with Eq.2-7. V: filtration
volume. An example determining fouling mechanisms is shown in detail by section
8.3 A-Tab.3). ............................................................................................................. 50
Fig. 4-13 (a) Multifiltration of water under different conditions, characteristic curves in UF of
water (b) with a biopolymer concentration of 0.42 mg C/L under 1 bar, (c) with a
biopolymer concentration of 0.42 mg C/L under 0.5 bar, (d) with a biopolymer
concentration of 0.25 mg C/L under 1 bar ................................................................ 51
Fig. 4-14 (a) Correlation between filterability and content of foulants (potential relationship
with biopolymers), (b) correlation between fouling resistance and content of foulants
(PR: protein, PS: polysaccharide, BC: biopolymer concentration)........................... 53
Fig. 4-15 Correlation between biopolymers and photometric measured polysaccharide
(Dubois method) and protein (Lowry method) (a) polysaccharide and organic carbon
in biopolymers, (b) protein and organic carbon in biopolymers, (c)
polysaccharide+protein and carbon in biopolymers (n = 82, BC: biopolymer
concentration)............................................................................................................ 54
Fig. 4-16 (a) Correlation between fouling resistance and development of irreversible fouling,
(b) correlation between biopolymer concentration and fouling resistance (from
November 5. 2007 to May 08. 2009, n=10, flux and hydraulic resistance data in
detail see section 8.3, A-Tab.4)................................................................................. 55
Fig. 4-17 Correlation between SUR and biopolymer concentration ........................................ 56
Fig. 5-1 Trans-membrane pressure development during the experimental period in WWTP
Beixiaohe (July 27th to November 09th in 2005) ....................................................... 59
Fig. 5-2 LC-OCD diagram of lake water, effluent after limestone and sand filtration and UF
permeate (dilution 1:5).............................................................................................. 60
Fig. 5-3 TMP development filtrating SE and SSF at a filtration rate of 0.25 m/h (Operational
condition: flux 89 LMH, BWI 40 minutes, BWT 50 seconds, BWF 260 LMH) ..... 61
Fig. 5-4 LC-OCD diagram of SE, SSF (filtration rate at 0.25 m/h) and UF permeate of SSF
(dilution 1:4).............................................................................................................. 61
Fig. 5-5 Removal profile of a) biopolymer (n=13), b) dissolved oxygen and DOC (n=10), c)
pH (n=10), d) UV254 (n=12) and e) turbidity (n=12) f) SUVA (n=10) in slow sand
filter operated at filtration of 0.25 m/h, period: March 2007 to November 2007,
removal profile at filtration rate of 0.5 m/h is shown in section 8.3, A-Fig. 5). ....... 64
Fig. 5-6 LC-OCD chromatograms in different SSFr probe sites (SSFr 20 cm: the probe site
20 cm below the sand filter surface) ......................................................................... 65
Fig. 5-7 Biological degradation test of biopolymers and UV254-absorbing substances .......... 66
Fig. 5-8 Fluorescence EEM of fractioned biopolymer in (a) SE (b) corresponding SSF at a
filtration rate of 0.25 m/h (Em: Emission, Ex: Excitation) ....................................... 66
Fig. 5-9 Removal effect by SSFr.1 in terms of (a) biopolymer, (b) turbidity due to slow sand
filtration (0.25 m/h)................................................................................................... 67
Appendix 117

Fig. 5-10 Variation of temperature, biopolymer concentration in SE (October 2006 –

December 2007) and removal effect by SSFr.1 (filtration rate 0.25 m/h, December
2006 – December 2007, removal degree: 0 - 1 represents 0 - 100%) ....................... 68
Fig. 5-11 Trend of biopolymer concentration in association with temperature in SE (October
2006 – December 2007, n=63) .................................................................................. 68
Fig. 5-12 Influence of a) temperature, b) biopolymer concentration and c) temperature and
biopolymer in the SE on biopolymer removal by SSFn (filtration rate 0.25 m/h, Dec.
2006 – Dec. 2007, n=46)........................................................................................... 69
Fig. 5-13 Normalized biopolymer concentration profile within slow sand filters operated at
filtration rate of 0.5 and 0.25 m/h in terms of (a) depth and (b) contact time (test
period: July to November 2007, biopolymer n=13 for SSF 0.5 m/h and n= 6 for SSF
0.25 m/h), operated at filtration of 0.25 and 0.1 m/h in terms of (c) depth and (d)
contact time (test period: April to June 2007, n = 6 for both filters) ........................ 71
Fig. 5-14 Biopolymer removal at different filtration rates (November 2006 to July 2007 in
WWTP Ruhleben, Median value (horizontal centre line), average value (the square
in box), the 25%- and 75%-quartiles (box) and the minimum and maximum values
(vertical line))............................................................................................................ 72
Fig. 5-15 Water head in (a) SSFr Nr. 1 operated at a filtration rate of 0.25 m/h during a cycle
time from Jan.17 to Apr.18 2007 (b) SSFr Nr. 2 operated at a filtration rate of 0.5
m/h during a cycle time from Aug.07 to Sep.23 2007 .............................................. 72
Fig. 6-1 (a) TMP development vs. operating time at different fluxes, (b) Kw/Kw0 vs.
operating time, (c) Kw/Kw0 vs. permeate volume, (d) Kw/Kw0 vs. biopolymer load
mass........................................................................................................................... 78
Fig. 6-2 (a) Influence of operating pressure on fouling effect, (b) flux recovery at different
operating pressures (feed water is SE after 0.45µm filter with a biopolymer
concentration of 0.42 mg C/L, operating pressure at 1 bar and 0.5 bar, permeate
volume 500 mL; back wash using salt solution 50 mL at 1 bar) .............................. 79
Fig. 6-3 (a) Influence of operating pressure on fouling effect, (b) flux recovery at different
operating pressures, (c) compressibility under different pressure ranges, (d)
correlation between specific fouling resistance and reversibility of corresponding
fouling (feed water is slow sand effluent with biopolymer concentration of 0.51 mg
C/L, operating pressure at 1.4 bar, 1 bar and 0.5 bar, permeate volume 0.5 L; back
wash using salt solution 0.05 L at 1 bar)................................................................... 81
Fig. 6-4 (a) TMP development vs. operating time at different BWI (b) Kw/Kw0 vs. permeate
volume at different BWI (Other operating parameters were identical: BWI 10
minutes, BWT 20 seconds, BWF 260 LMH)............................................................ 82
Fig. 6-5 Correlation between biopolymer concentration and TMP development in (a) test 1 at
flux 50 LMH, BWI 20 min, (b) test 2 at flux 50 LMH, BWI 20 min and (c) test 3 at
flux 74 LMH, BWI 14 min ....................................................................................... 84
Fig. 6-6 Fouling mechanisms for MBR operation at constant flux (adapted from Zhang et al.,
2006a)........................................................................................................................ 85
Fig. 6-7 Three-stage fouling development process during UF of SSF (0.5 m/h).................... 86
Fig.6-8 Illustration of an ideal three-stage fouling process in UF of treated domestic
wastewater (adapted from Zhang et al., 2006a) ........................................................ 87
Fig. 6-9 Comparison of the influence of BWT on the performance of UF............................. 88
Fig. 6-10 (a) TMP development vs. operating time using permeate backwash and enhanced
backwash reagents (b) Kw/Kw0 vs. operating time using permeate backwash and
enhanced backwash reagents..................................................................................... 89
Fig. 6-11 Comparison of permeability restoration effect of UF using different cleaning
reagents ..................................................................................................................... 92
118 Appendix

Fig. 6-12 Relation between permeability restoration using different cleaning reagents and
biopolymer concentration in corresponding cleaning waters (same volume of
cleaning water was used in all the experiments)....................................................... 93
Fig. 6-13 (a) Optical impression of demineralised water after membrane cleaning, (b) optical
impression of NaClO solution (active Cl 50 mg/L) after membrane cleaning, (c)
diameter measurement of the suspended orbicular fragment in NaClO solution after
membrane cleaning, (d) diameter of a membrane fibre of Inge membrane (dizzer®
450, source: Inge AG) ............................................................................................... 94

A-Fig. 1 Set-up of the LC-OCD-UVD-OND system............................................................. 103

A-Fig. 2 LC-OCD-UVD-OND chromatogram of treated domestic wastewater ................... 104
A-Fig. 3 Calibration of the carbon detector of the LC-OCD system (standard solution of
potassium hydrogen phthalate: 2.1254g/L = 1000mg DOC/L, injection volumes are
0.05, 0.1 and 0.2 mL within concentration range of 0.005 to 10 mg/L) ................. 105
A-Fig. 4 Molecular weight distribution and corresponding elution time of polyethylenglycol
(MW range of 106 to 40 000 da) using size exclusion column HW 55s ................ 106
A-Fig. 5 Removal profile of a) biopolymer (n=14), b) dissolved oxygen and DOC (n=10), c)
pH (n=14), d) UV254 (n=14) and e) turbidity (n=14) f) SUVA (n=10) in slow sand
filter operated at filtration of 0.5 m/h, period: July 10 2007 to November 20 2007)
................................................................................................................................. 111
A-Fig. 6 Biopolymer concentration variation in water phase along wastewater treatment
procedure in WWTP Ruhleben (Average value of three series of water samples, July
2007 and November 2007)...................................................................................... 112
A-Fig. 7 Influence of filtration rate on biopolymer removal effect (a) LC-OCD chromatogram
of SE and bio-filtrate operated at filtration rate of 8 m/h (Jan. 31 2007, start up
November 2007), (b) LC-OCD chromatogram of SE and bio-filtrate operated at
filtration rate of 8 m/h (March 22 2009, start up November 2007), (c) LC-OCD
chromatogram of SE and bio-filtrate operated at filtration rate of 4 m/h (March 22
2009, start up April 2008), (d) LC-OCD chromatogram of SE and bio-filtrate
operated at filtration rate of 0.25 m/h (March 22 2009, start up March 1 2009) and
0.05 m/h (March 22 2009, start up Jan. 1 2009) ..................................................... 114

8.5 Table contents

Tab. 1-1 Case projects of wastewater reclamation applying low-pressure membrane filtration2
Tab. 2-1 Membranes applied for water treatment (Mulder, 2000; Judd, 2006) ......................... 5
Tab. 2-2 Overview of fouling control pre-treatment processes prior to membrane filtration for
wastewater reclamation ............................................................................................. 23
Tab. 2-3 Design/Operational Parameters/Conditions for biological (rapid) sand filtration
(BSFn), slow sand filtration (SSFn), river bank filtratin (RBFn) and SAT (Adapted
from Amy et al., 2006) .............................................................................................. 24
Tab. 2-4 Specific fouling resistance and compressibility of fouling layer in low-pressure
membrane filtering EPS, activated sludge or NOM content waters ......................... 26
Tab. 2-5 Major categories of membrane cleaning chemicals (Liu et al., 2006; Poele, 2006;
Tian et al., 2009) ....................................................................................................... 28
Tab. 3-1 Properties of applied UF membranes in lab and pilot-scale membrane filtration tests
................................................................................................................................... 30
Tab. 3-2 Average water quality of the lake water in WWTP Beixiaohe in 2005 .................... 33
Tab. 3-3 Wastewater effluent quality of the WWTP Ruhleben (Bahr et al., 2007) ................. 34
Appendix 119

Tab. 3-4 Bulk water quality parameters and corresponding applied measurement equipments
................................................................................................................................... 39
Tab. 4-1 DOC and biopolymer content of SE, permeate of SE and SSFs ............................... 42
Tab. 4-2 DOC, protein, polysaccharide concentration and biopolymer content of SE, permeate
of SE and SSFs.......................................................................................................... 52
Tab. 4-3 Comparison of the operational conditions measuring SUR and fouling resistance in
the present work ........................................................................................................ 55
Tab. 5-1 Water quality of SE and SSFs on average (December 2006 to December 2007, SSF
at 0.5 m/h: except February to July 2007)................................................................. 62
Tab. 5-2 Performance of bio-filtration operated under different temperature and biopolymer
concentration in SE ................................................................................................... 70
Tab. 6-1 Influence of operational conditions on the performance of UF filtering SSF (BWF
was always 260 LMH during the whole experiment) ............................................... 77
Tab. 6-2 Calculated biopolymer mass content in formed cake/gel layer and the corresponding
specific fouling resistance (biopolymer concentration in SE is 0.42 mg C/L,
permeate volume 0.5 L, supernatant 0.2 L). ............................................................. 80
Tab. 6-3 Cleaning effect of different chemical reagents on permeability restoration of UF of
SSF ............................................................................................................................ 90
Tab. 6-4 Comparison of biopolymer content in SSF to UF and that in cleaning water using
different cleaning reagents ........................................................................................ 93

A-Tab. 1 Eluent and acid solution used in LC-OCD system ................................................. 104
A-Tab. 2 Amicon cell filtration determining the filterability of water samples (corresponding
to Fig. 5-8)............................................................................................................... 107
A-Tab. 3 Analysis of fouling mechanisms during a dead-end filtration (corresponding to Fig.
5-10) ........................................................................................................................ 108
A-Tab. 4 Amicon cell test with five filtration cycles for setting up the relationship between
biopolymer concentration in the influent and irreversible fouling filtering
corresponding water sample (corresponding to Fig. 5-16) ..................................... 109
A-Tab. 5 Calculation of biopolymer mass in fouling layer and corresponding specific fouling
resistance during UF ............................................................................................... 110
A-Tab. 6 Biopolymer concentration in SE and SSFs (Oct.26 2006 to Dec. 31 2007)........... 113
120 Literatures

9 Literatures

Abrahamse, A.J., Lipreau, S., Li, S. and Heijman, S.G.J., 2008, Removal of divalent cations
reduces fouling of ultrafiltration membranes. J Membrane Sci 323, 153 - 158.

Ahmed, Z., Cho, J., Lim, B.R., Song, K.G. and Ahn, K.H., 2007, Effects of sludge retention
time on membrane fouling and microbial community structure in a membrane bioreactor. J
Membrane Sci 287(2), 211-218.

Ahn, W.Y., Kalinichev, A.G. and Clark, M.M., 2008, Effects of background cations on the
fouling of polyethersulfone membranes by natural organic matter: Experimental and molecu-
lar modeling study. J Membrane Sci 309(1-2), 128-140.

Al-Halbouni, D., Traber, J., Lyko, S., Wintgens, T., Melin, T., Tacke, D., Janot, A., Dott, W.
and Hollender, J., 2008, Correlation of EPS content in activated sludge at different sludge
retention times with membrane fouling phenomena. Water Res 42(6-7), 1475-1488.

Al-Halbouni, D., Dott, W. and Hollender, J., 2009, Occurence and composition of extracellu-
lar lipids and polysaccharides in a full-scale membrane bioreactor. Water Res 43(1), 97-106.

Amy, G. and Cho, J., 1999, Interactions between natural organic matter (NOM) and mem-
branes: Rejection and fouling. Water Science and Technology 40(9), 131-139.

Amy, G., Carlson, K., Collins, M.R., Drewes, J., Gruenheid, S. and Jekel, M.,2006, Integrated
comparison of biofiltration in engineered versus natural systems. In: Recent progress in slow
sand and alternative biofiltration processes, Gimbel, R., Graham, N.J.D. and Collins, M.R.
(eds), pp. 3-11, TJ International Ltd, Cornwall, UK.

Amy, G. and Drewes, J., 2007, Soil aquifer treatment (SAT) as a natural and sustainable
wastewater Reclamation/Reuse technology: Fate of wastewater effluent organic matter
(EfOM) and trace organic compounds. Environmental Monitoring and Assessment 129(1-3),
19-26.

Amy, G., 2008, Fundamental understanding of organic matter fouling of membranes. Desali-
nation 231, 44-51.

Ang, W.S. and Elimelech, M., 2007, Protein (BSA) fouling of reverse osmosis membranes:
Implications for wastewater reclamation. J Membrane Sci 296(1-2), 83-92.

Ang, W.S. and Elimelech, M., 2008, Fatty acid fouling of reverse osmosis membranes: Impli-
cations for wastewater reclamation. Water Res 42(16), 4393-4403.

Augstin Suarez, J. and Veza, J.M., 2000, Dead-end microfiltration as advanced treatment for
wastewater. Desalination 127(1), 47-58.

Bagga, A., Chellam, S. and Clifford, D.A., 2008, Evaluation of iron chemical coagulation and
electrocoagulation pretreatment for surface water microfiltration. J Membrane Sci 309(1-2),
82-93.

Bahr, C., Ernst, M. and Jekel, M. 2007 Pilotuntersuchungen zur kombinierten oxidativbiolo-
gischen Behandlung von Klärwerksabläufen für die Entfernung von organischen Spuren- und
Wirkstoffen und zur Desinfektion, Berlin.

Baker, R.W., 2004, Membrane technology and applications, Second Edition, John Wiley &
Sons Ltd., Chichester, England.
Literatures 121

Barker, D.J. and Stuckey, D.C., 1999, A review of soluble microbial products (SMP) in
wastewater treatment systems. Water Res 33(14), 3063-3082.

Bessiere, Y., Abidine, N. and Bacchin, P., 2005, Low fouling conditions in dead-end filtra-
tion: Evidence for a critical filtered volume and interpretation using critical osmotic pressure.
J Membrane Sci 264(1-2), 37-47.

Block, J.C., Mathieu, L., Servais, P., Fontvieille, D. and Werner, P., 1992, Indigenous Bacte-
rial Inocula for Measuring the Biodegradable Dissolved Organic-Carbon (BDOC) in Waters.
Water Res 26(4), 481-486.

Böddeker, K.W., 2007, Liquid seperations with membranes-An Introduction to Barrier Inter-
ference, Springer, Glinde, Germany.

Boerlage, S.F.E., Kennedy, M.D., Aniye, M.P., Abogrean, E., Tarawneh, Z.S. and Schippers,
J.C., 2003, The MFI-UF as a water quality test and monitor. J Membrane Sci 211(2), 271-289.

Bolton, G., LaCasse, D. and Kuriyel, R., 2006, Combined models of membrane fouling: De-
velopment and application to microfiltration and ultrafiltration of biological fluids. J Mem-
brane Sci 277(1-2), 75-84.

Bouzid, H., Rabiller-Baudry, M., Paugam, L., Rousseau, F., Derriche, Z. and Bettahar, N.E.,
2008, Impact of zeta potential and size of caseins as precursors of fouling deposit on limiting
and critical fluxes in spiral ultrafiltration of modified skim milks. J Membrane Sci 314(1-2),
67-75.

Bowen, W.R., Calvo, J.I. and Hernandez, A., 1995, Steps of Membrane Blocking in Flux De-
cline during Protein Microfiltration. J Membrane Sci 101(1-2), 153-165.

Bowen, W.R. and Jenner, F., 1995, Theoretical Descriptions of Membrane Filtration of Col-
loids and Fine Particles - an Assessment and Review. Advances in Colloid and Interface Sci-
ence 56, 141-200.

Brinkmann, T., Abbt-Braun, G., Karle, E., Hesse, S., Lungar, N. and Frimmel, F.H., 2004,
Transformation of wastewater-derived dissolved organic matter below leaky sewers - Fate of
amino acids and carbohydrates. Acta Hydroch Hydrob 32(4-5), 316-327.

Chellam, S., Jacangelo, J.G. and Bonacquisti, T.P., 1998, Modeling and experimental verifi-
cation of pilot-scale hollow fiber, direct flow microfiltration with periodic backwashing. En-
vironmental Science & Technology 32(1), 75-81.

Chen, M.Y., Lee, D.J. and Tay, J.H., 2006a, Extracellular polymeric substances in fouling
layer. Separ Sci Technol 41(7), 1467-1474.

Chen, M.Y., Lee, D.J., Yang, Z., Peng, X.F. and Lai, J.Y., 2006b, Fluorecent staining for
study of extracellular polymeric substances in membrane biofouling layers. Environmental
Science & Technology 40(21), 6642-6646.

Cho, M.H., Lee, C.H. and Lee, S., 2006, Effect of flocculation conditions on membrane per-
meability in coagulation-microfiltration. Desalination 191(1-3), 386-396.

Chu, J.Y., Chen, J.N., Wang, C. and Fu, P., 2004, Wastewater reuse potential analysis: impli-
cations for China's water resources management. Water Res 38(11), 2746-2756.

Chudoba, J., 1985, Quantitative Estimation in Cod Units of Refractory Organic-Compounds

Produced by Activated-Sludge Microorganisms. Water Res 19(1), 37-43.

Costa, A.R., de Pinho, M.N. and Elimelech, M., 2006, Mechanisms of colloidal natural or-
ganic matter fouling in ultrafiltration. J Membrane Sci 281(1-2), 716-725.
122 Literatures

Crittenden, J.C., Trussell, R.R., Hand, D.W., Howe, K.J. and Tchobanoglous, G. (eds),2006,
Water treatment principles and design, John Wiley and Sons, Inc., New Jersey, U.S.A.

de Koning, J., Bixio, D., Karabelas, A., Salgot, M. and Schafer, A., 2008, Characterisation
and assessment of water treatment technologies for reuse. Desalination 218(1-3), 92-104.

Decarolis, J., Hong, S.K. and Taylor, J., 2001, Fouling behavior of a pilot scale inside-out
hollow fiber UF membrane during dead-end filtration of tertiary wastewater. J Membrane Sci
191(1-2), 165-178.

Downing, J.B., Bracco, E., Green, F.B., Ku, A.Y., Lundquist, T.J., Zubieta, I.X. and Oswald,
W.J., 2002, Low cost reclamation using the Advanced Integrated Wastewater Pond Systems
(R) Technology and reverse osmosis. Water Science and Technology 45(1), 117-125.

Drewes, J.E. and Fox, P., 1999, Behavior and characterization of residual organic compounds
in wastewater used for indirect potable reuse. Water Science and Technology 40(4-5), 391-
398.

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A. and Smith, F., 1956, Colorimetric
Method for Determination of Sugars and Related Substances. Analytical Chemistry 28(3),
350-356.

Ellis, K.V., 1985, Slow sand filtration. CRC Critical Reviews in Environmental Control
15(4), 315-354.

Ellis, K.V., 1987, Slow Sand Filtration as a Technique for the Tertiary-Treatment of Munici-
pal Sewages. Water Res. 21(4), 403-410.

EPA, U.S. 2004 Guide lines for water reuse. Agency, U.S.E.P. (ed), Camp. Dresser &
McKee, Inc., Washington, DC.

Ernst, M., Sperlich, A., Zheng, X., Gan, Y., Hu, J., Zhao, X., Wang, J. and Jekel, M., 2007,
An integrated wastewater treatment and reuse concept for the Olympic Park 2008, Beijing.
Desalination 202(1-3), 293-301.

Esparza-Soto, M. and Westerhoff, P., 2003, Biosorption of humic and fulvic acids to live ac-
tivated sludge biomass. Water Res 37(10), 2301-2310.

Evans, P.J., Bird, M.R., Pihlajamaki, A. and Nystrom, M., 2008, The influence of hydropho-
bicity, roughness and charge upon ultrafiltration membranes for black tea liquor clarification.
J Membrane Sci 313(1-2), 250-262.

Fan, L.H., Nguyen, T., Roddick, F.A. and Harris, J.L., 2008, Low-pressure membrane filtra-
tion of secondary effluent in water reuse: Pre-treatment for fouling reduction. J. Membr. Sci.
320(1-2), 135-142.

Field, R.W., Wu, D., Howell, J.A. and Gupta, B.B., 1995, Critical Flux Concept for Microfil-
tration Fouling. J Membrane Sci 100(3), 259-272.

Flemming, H.-c. and Wingernder, J., 2002, Was Biofilme zusammenhält. chemie in unserer
Zeit.

Flemming, H.C. and Wingender, J., 2001, Relevance of microbial extracellular polymeric
substances (EPSs) - Part I: Structural and ecological aspects. Water Science and Technology
43(6), 1-8.

Flemming, H.C., 2002, Biofouling in water systems - cases, causes and countermeasures. Ap-
plied Microbiology and Biotechnology 59(6), 629-640.
Literatures 123

Foley, G., 2006, A review of factors affecting filter cake properties in dead-end microfiltra-
tion of microbial suspensions. J Membrane Sci 274(1-2), 38-46.

Fu, X.Y., Maruyama, T., Sotani, T. and Matsuyama, H., 2008, Effect of surface morphology
on membrane fouling by humic acid with the use of cellulose acetate butyrate hollow fiber
membranes. J Membrane Sci 320(1-2), 483-491.

Furukawa, D. 2008 A global perspective of low pressure membranes, p. 13, National Water
Research Institute, Fountain Valley, CA, USA.

Furukawa, T., Kokubo, K., Nakamura, K. and Matsumoto, K., 2008, Modeling of the perme-
ate flux decline during MF and UF cross-flow filtration of soy sauce lees. J Membrane Sci
322(2), 491-502.

Garnier, C., Gorner, T., Lartiges, B.S., Abdelouhab, S. and de Donato, P., 2005, Characteriza-
tion of activated sludge exopolymers from various origins: A combined size-exclusion chro-
matography and infrared microscopy study. Water Res 39(13), 3044-3054.

Geng, Z.H. and Hall, E.R., 2007, A comparative study of fouling-related properties of sludge
from conventional and membrane enhanced biological phosphorus removal processes. Water
Res 41(19), 4329-4338.

Gruenheid, S., Amy, G. and Jekel, M., 2005, Removal of bulk dissolved organic carbon
(DOC) and trace organic compounds by bank filtration and artificial recharge. Water Res.
39(14), 3219-3228.

Gruenheid, S., Huebner, U. and Jekel, M., 2008, Impact of temperature on biodegradation of
bulk and trace organics during soil passage in an indirect reuse system. Water Science and
Technology 57(7), 987-994.

Guo, X., Li, Q.L., Hu, W., Gao, W. and Liu, D., 2009, Ultrafiltration of dissolved organic
matter in surface water by a polyvinylchloride hollow fiber membrane. J Membrane Sci (In
press).

Haberkamp, J., Ruhl, A.S., Ernst, M. and Jekel, M., 2007, Impact of coagulation and adsorp-
tion on DOC fractions of secondary effluent and resulting fouling behaviour in ultrafiltration.
Water Res 41(17), 3794-3802.

Haberkamp, J., 2008, Organic membrane fouling in ultrafiltration of treated domestic waste-
water: causes, mechanisms and methodes to reduce (in German). Ph.D. Dissertation, Tech-
nische Univesität Berlin, Berlin.

Haberkamp, J., Ernst, M., Bockelmann, U., Szewzyk, U. and Jekel, M., 2008, Complexity of
ultrafiltration membrane fouling caused by macromolecular dissolved organic compounds in
secondary effluents. Water Res 42(12), 3153-3161.

Hallé, C., Huck, P.M., Peldszus, S., Haberkamp, J. and Jekel, M., 2009, Assessing the Per-
formance of Biological Filtration As Pretreatment to Low Pressure Membranes for Drinking
Water. Environ Sci Technol 43(10), 3878-3884.

Henderson, R.K., Baker, A., Murphy, K.R., Hamblya, A., Stuetz, R.M. and Khan, S.J., 2009,
Fluorescence as a potential monitoring tool for recycled water systems: A review. Water Res
43(4), 863-881.

Her, N., Amy, G., McKnight, D., Sohn, J. and Yoon, Y.M., 2003, Characterization of DOM
as a function of MW by fluorescence EEM and HPLC-SEC using UVA, DOC, and fluores-
cence detection. Water Res. 37(17), 4295-4303.
124 Literatures

Hermia, J., 1982, Constant Pressure Blocking Filtration Laws - Application to Power-Law
Non-Newtonian Fluids. Transactions of the Institution of Chemical Engineers 60(3), 183-187.

Herzberg, M., Kang, S. and Elimelech, M., 2009, Role of Extracellular Polymeric Substances
(EPS) in Biofouling of Reverse Osmosis Membranes. Environmental Science & Technology
43(12), 4393-4398.

Ho, C.C. and Zydney, A.L., 2001, Protein fouling of asymmetric and composite microfiltra-
tion membranes. Industrial & Engineering Chemistry Research 40(5), 1412-1421.

Howe, K.J., Marwah, A., Chiu, K.P. and Adham, S.S., 2006, Effect of coagulation on the size
of MF and UF membrane foulants. Environmental Science & Technology 40(24), 7908-7913.

Hua, G.H. and Reckhow, D.A., 2007, Comparison of disinfection byproduct formation from
chlorine and alternative disinfectants. Water Res 41(8), 1667-1678.

Hua, J.M., An, P.L., Winter, J. and Gallert, C., 2003, Elimination of COD, microorganisms
and pharmaceuticals from sewage by trickling through sandy soil below leaking sewers. Wa-
ter Res 37(18), 4395-4404.

Huang, H., Lee, N., Young, T., Gary, A., Lozier, J.C. and Jacangelo, J.G., 2007, Natural or-
ganic matter fouling of low-pressure, hollow-fiber membranes: Effects of NOM source and
hydrodynamic conditions. Water Res 41(17), 3823-3832.

Huang, H., Young, T.A. and Jacangelo, J.G., 2008a, Unified membrane fouling index for low
pressure membrane filtration of natural waters: Principles and methodology. Environmental
Science & Technology 42(3), 714-720.

Huang, H., Schwab, K. and Jacangelo, J.G., 2009, Pretreatment for Low Pressure Membranes
in Water Treatment: A Review. Environmental Science & Technology 43(9), 3011-3019.

Huang, L.N., Wever, H.D. and Diels, L., 2008b, Divers and distinct bacterial communities
induced biofilm fouling in membrane bioreactors operated under different conditions. Envi-
ron. Sci. Technol 42(22), 8360-8366.

Huang, X.H., Leal, M. and Li, Q.L., 2008c, Degradation of natural organic matter by TiO2
photocatalytic oxidation and its effect on fouling of low-pressure membranes. Water Res
42(4-5), 1142-1150.

Huber, S.A., 1998, Evidence for membrane fouling by specific TOC constituents. Desalina-
tion 119(1-3), 229-234.

Huck, P.M. and Anderson, W.B., 1992, Quantitative relationship between the removal of
NVOC, chlorine demand and AOX formation potential in biological drinking water treatment.
Vom Wasser 78, 281-303.

Huck, P.M. and Sozanski, M.M., 2008, Biological filtration for membrane pre-treatment and
other applications: towards the development of a practically-oriented performance parameter.
Journal of Water Supply Research and Technology - Aqua 57(4), 203-224.

Hudson, N., Baker, A. and Reynolds, D., 2007, Fluorescence Analysis of Dissoved Organic
Matter in Natural, Waste and Polluted Waters - a Review. River Research and Applications
23, 631 - 649.

Huisman, I.H., Pradanos, P. and Hernandez, A., 2000, The effect of protein-protein and pro-
tein-membrane interactions on membrane fouling in ultrafiltration. J Membrane Sci 179(1-2),
79-90.
Literatures 125

Janssen, A.N. and Van der Graaf, J., 2008, Dead-end ultrafiltration of wwtp effluent: pre-
treatment and performance. Acdamic sumer school particle separation in water and wastewa-
ter treatment 2008, Delft, the Netherlands.

Jarusutthirak, C., 2002, Fouling and flux decline of reverse osmosis (RO), nanofiltration (NF),
and ultrafiltration (UF) membranes associated with effluent organic matter (EfOM) during
wastewater reclamation/reuse. Ph.D. Thesis, University of Colorado, Colorado.

Jarusutthirak, C., Amy, G. and Croue, J.P., 2002, Fouling characteristics of wastewater efflu-
ent organic matter (EfOM) isolates on NF and UF membranes. Desalination 145(1-3), 247-
255.

Jarusutthirak, C. and Amy, G., 2006, Role of soluble microbial products (SMP) in membrane
fouling and flux decline. Environ Sci Technol 40(3), 969-974.

Jarusutthirak, C. and Amy, G., 2007, Understanding soluble microbial products (SMP) as a
component of effluent organic matter (EfOM). Water Res. 41(12), 2787-2793.

Jekel, M. and Boeckle, R., 1982, Investigation on the filtration of dilute suspensions in the
presence of high molecular weight polymers. Vom Wasser 59, 195-206.

Jermann, D., Pronk, W., Meylan, S. and Boller, M., 2007, interplay of different NOM fouling
mechanisums during UF for drinking water production. Water Res 41(8), 1713-1722.

Jermann, D., Pronk, W. and Boller, M., 2008, Mutual influences between natural orgnaic mat-
ter and inorganic particles and their combined effect ultrafiltration membrane fouling. Envi-
ronmental Science & Technology 42(24), 9129-9136.

Jiang, T., Kennedy, M.D., Guinzbourg, B.F., Vanrolleghem, P.A. and Schippers, J.C., 2005,
Optimising the operation of a MBR pilot plant by quantitative analysis of the membrane foul-
ing mechanism. Water Science and Technology 51(6-7), 19-25.

Jimenez, B. and Asano, T. (eds),2008, Water Reuse - An International Survey of current prac-
tice, issues and needs, IWA, London.

Judd, S., 2006, The MBR book: Principles and Applications of Membrane Bioreactors in Wa-
ter and Wastewater Treatment, Elsevier Ltd., Great Britain.

Katsoufidou, K., Yiantsios, S.G. and Karabelas, A.J., 2007, Experimental study of ultrafiltra-
tion membrane fouling by sodium alginate and flux recovery by backwashing. J Membrane
Sci 300(1-2), 137-146.

Kennedy, M., Kim, S.-M., Mutenyo, I., Broens, L. and Schippers, J., 1998, Intermittent cross-
flushing of hollow fiber ultrafiltration systems. Desalination 118(1-3), 175-187.

Kennedy, M.D., Kamanyi, J., Heijman, B.G.J. and Amy, G., 2008, Colloidal organic matter
fouling of UF membranes: role of NOM composition & size. Desalination 220(1-3), 200-213.

Kim, A.S., Chen, H.Q. and Yuan, R., 2006, EPS biofouling in membrane filtration: An ana-
lytic modeling study. J Colloid Interf Sci 303(1), 243-249.

Kim, H.C. and Dempsey, B.A., 2008, Effects of wastewater effluent organic materials on
fouling in ultrafiltration. Water Res 42(13), 3379-3384.

Kim, J. and DiGiano, F.A., 2006, A two-fiber, bench-scale test of ultrafiltration (UF) for in-
vestigation of fouling rate and characteristics. J Membrane Sci 271(1-2), 196-204.
126 Literatures

Kim, J., DiGiano, F.A. and Reardon, R.D., 2008, Autopsy of high-pressure membranes to
compare effectiveness of MF and UF pretreatment in water reclamation. Water Res 42(3),
697-706.

Kimura, K., Hane, Y., Watanabe, Y., Amy, G. and Ohkuma, N., 2004, Irreversible membrane
fouling during ultrafiltration of surface water. Water Res 38(14-15), 3431-3441.

Kimura, K., Yamato, N., Yamamura, H. and Watanabe, Y., 2005, Membrane fouling in pilot-
scale membrane Bioreactors (MBRs) treating municipal wastewater. Environmental Science
& Technology 39(16), 6293-6299.

Kimura, K., Yamamura, H. and Watanabe, Y., 2006, Irreversible fouling in MF/UF mem-
branes caused by natural organic matters (NOMs) isolated from different origins. Separ Sci
Technol 41(7), 1331-1344.

Kimura, K., Naruse, T. and Watanabe, Y., 2009, Changes in characteristics of soluble micro-
bial products in membrane biorectors associated with different solid retention times: relation
to membrane fouling. Water Res. 43(2), 1033-1039.

Laabs, C., 2004, Fouling of low-pressure membranes by municipal wastewater: identification

of principal foulants and underlying fouling mechanisms, Technische Universität Berlin, Ber-
lin.

Laabs, C., Amy, G. and Jekel, M., 2004, Organic colloids and their influence on low-pressure
membrane filtration. Water Science and Technology 50(12), 311-316.

Laabs, C. and Amy, G.,2006, Membrane Treatment for Drinking Water and Reuse Applica-
tions, edited by Kerry J. and Howe P.E. Kerry, J. and Howe, P.E. (eds), pp. 565-580, AWWA,
Denver, U.S.A.

Laabs, C.N., Amy, G.L. and Jekel, M., 2006, Understanding the size and character of fouling-
causing substances from effluent organic matter (EfOM) in low-pressure membrane filtration.
Environ Sci Technol 40(14), 4495-4499.

Le-Clech, P., Chen, V. and Fane, T.A.G., 2006, Fouling in membrane bioreactors used in
wastewater treatment. J Membrane Sci 284(1-2), 17-53.

Le Bihan, Y. and Lessard, P., 2000, Monitoring biofilter clogging: Biochemical characteris-
tics of the biomass. Water Res 34(17), 4284-4294.

Lee, N., Amy, G., Croue, J.P. and Buisson, H., 2005, Morphological analyses of natural or-
ganic matter (NOM) fouling of low-pressure membranes (MF/UF). J Membrane Sci 261(1-2),
7-16.

Lee, S. and Elimelech, M., 2006, Relating organic fouling of reverse osmosis membranes to
intermolecular adhesion forces. Environmental Science & Technology 40(3), 980-987.

Lehman, S.G. and Liu, L., 2009, Application of ceramic membranes with pre-ozonation for
treatment of secondary wastewater effluent. Water Res 43(7), 2020-2028.

Leverenz, H.L., Tchobanoglous, G. and Darby, J.L., 2009, Clogging in intermittently dosed
sand filters used for wastewater treatment. Water Res. 43(2), 695-705.

Levine, A.D., Tchobanoglous, G. and Asano, T., 1985, Characterization of the size distribu-
tion of contaminants in wastewater: Treatment and reuse implications. J. Water Pollut. Con-
trol Fed. 57(7), 805-816.

Li, N.N., Fane, A.G., Ho, W.S.W. and Matsuura, T., 2008, Advanced Membrane Technology
and Applications, John Wiley and Sons, Inc., New Jersey.
Literatures 127

Li, Q.L., Xu, Z.H. and Pinnau, I., 2007, Fouling of reverse osmosis membranes by biopoly-
mers in wastewater secondary effluent: Role of membrane surface properties and initial per-
meate flux. J Membrane Sci 290(1-2), 173-181.

Li, S., Heijman, S.G.J., Verberk, J.Q.J.C., Verliefde, A.R.D., Kemperman, A.J.B., van Dijk,
J.C. and Amy, G., 2009, Impact of backwash water composition on ultrafiltration fouling con-
trol. J Membrane Sci 344(1-2), 17-25.

Liang, S., Liu, C. and Song, L.F., 2007, Soluble microbial products in membrane bioreactor
operation: Behaviors, characteristics, and fouling potential. Water Res. 41(1), 95-101.

Liang, S. and Song, L.F., 2007, Characteristics and fouling behaviors of dissolved organic
matter in submerged bioreactor systems. Environ. Eng. Sci. 24, 652 - 662.

Liang, S., Zhao, Y., Liu, C. and Song, L., 2008, Effect of solution chemistry on the fouling
potential of dissolved organic matter in membrane bioreactor systems. J Membrane Sci
310(1-2), 503-511.

Liao, B.Q., Bagley, D.M., Kraemer, H.E., Leppard, G.G. and Liss, S.N., 2004, A review of
biofouling and its control in membrane separation bioreactors. Water Environment Research
76(5), 425-436.

Listiarini, K., Chun, W., Sun, D.D. and Leckie, J.O., 2009, Fouling mechanism and resistance
analyses of systems containing sodium alginate, calcium, alum and their combination in dead-
end fouling of nanofiltration membranes. J Membrane Sci 344(1-2), 244-251.

Liu, C., Caothien, S., Hayes, J., Caothuy, T., Otoyo, T. and Ogawa, T. 2006 Membrane
chemical cleaning: from art to science, Pall Corporation, Asahi Chemical Industry Co., Ltd.

Lodge, B., Judd, S.J. and Smith, A.J., 2004, Characterisation of dead-end ultrafiltration of
biotreated domestic wastewater. J Membrane Sci 231(1-2), 91-98.

Lohwacharin, J., Oguma, K. and Takizawa, S., 2009, Ultrafiltration of natural organic matter
and black carbon: Factors influencing aggregation and membrane fouling. Water Res 43(12),
3076-3085.

Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J., 1951, Protein Measurement with
the Folin Phenol Reagent. Journal of Biological Chemistry 193(1), 265-275.

Maelzer, H.-J. and Gimbel, R.,2006, Extension of Slow Sand Filter Running Times by Protec-
tion Layers. In: Recent progress in slow sand and alternative biofiltration processes, Gimbel,
R., Graham, N.J.D. and Collins, M.R. (eds), pp. 251-259, TJ International Ltd, Cornwall, UK.

Mehrez, R., Ernst, M. and Jekel, M., 2007, Development of a continuous protein and polysac-
charide measurement method by Sequential Injection Analysis for application in membrane
bioreactor systems. Water Science and Technology 56(6), 163-171.

Melo, L.F. and Bott, T.R., 1997, Biofouling in water systems. Experimental Thermal and
Fluid Science 14(4), 375-381.

Meng, F.G., Chae, S.R., Drews, A., Kraume, M., Shin, H.S. and Yang, F.L., 2009, Recent
advances in membrane bioreactors (MBRs): Membrane fouling and membrane material. Wa-
ter Res 43(6), 1489-1512.

Metcalf & Eddy, I., 2003, Wastewater engineering; treatment and reuse, McGraw-Hill, New
York.
128 Literatures

Morales, C.F.L., Strathmann, M. and Flemming, H.C., 2007, Influence of biofilms on the
movement of colloids in porous media. Implications for colloid facilitated transport in subsur-
face environments. Water Res 41(10), 2059-2068.

Mousa, H.A., 2007, Investigation of UF membranes fouling by humic acid. Desalination

217(1-3), 38-51.

Mulder, M., 2000, Basic principles of membrane technology [2. Ed.], Kluwer academic Pub-
lisher, Dordrecht.

Narong, P. and James, A.E., 2006, Effect of the zeta-potential on the micro/ultra-filtration of
yeast suspensions using ceramic membranes. Sep Purif Technol 49(2), 149-156.

Negaresh, E., Le-Clech, P. and Chen, V., 2006, Fouling mechanisms of model extracellular
polymeric substances in submerged membrane reactor. Desalination 200(1-3), 715-717.

Negaresh, E., Le-Clech, P. and Chen, V., 2007, Fouling mechanisms of model polymeric sub-
stances. Asia-Pacific Journal of Chemical Engineering 2(5), 394-399.

Ng, C.N., Sun, D., Zhang, J., Chua, H.C., Bing, W., Tay, S. and Fane, A.G., 2005, Strategies
to improve the sustainable operation of membrane bioreactors. Proccedings of the Interna-
tional Desalination Association Conference, Singapore.

Numours, D.d. and Co. 1977 Determination of silt density index, Technical Balletin Number
491.

Ojha, C.S.P. and Graham, N.J.D.,1996, Modelling the role of the schmutzdecke layer in slow
sand filtration. In: Advances in slow sand and alternative biological filtration, Graham, N. and
Collins, M.R. (eds), pp. 277-286, John Wiley & Sons Ltd, Chichester, England.

Panglisch, S., Loi-Brugger, A. and Gimbel, R., 2007, Particle removal with membranes in
water treatment in Germany - state of the art and further developments. J Water Supply Res T
56(6-7), 375-383.

Peng, W.H., Escobar, I.C. and White, D.B., 2004, Effects of water chemistries and properties
of membrane on the performance and fouling - a model development study. J Membrane Sci
238(1-2), 33-46.

Poele, S.T., Roorda, J.H. and van der Graaf, J.H.J.M., 2004, Influence of the size of mem-
brane foulants on the filterability of WWTP-effluent. Water Science and Technology 50(12),
111-118.

Poele, S.T., 2006, Foulants in ultrafiltration of WWTP effluent. Ph.D. Dissertation, Delft
University of Technology, Delft, the Netherlands.

Pollice, A., Brookes, A., Jefferson, B. and Judd, S., 2005, Sub-critical flux fouling in mem-
brane bioreactons - a review of recent literature. Desalination 174(3), 221-230.

Qin, H.J., Kekre, K.A., Tao, G.H., Oo, M.H., Wai, M.N., Lee, T.C., Viswanath, B. and Seah,
H., 2006, New option of MBR-RO process for production of NEWater from domestic sew-
age. J Membrane Sci 272(1-2), 70-77.

Rachwal, A.J., Bauer, M.J., Chipps, J.S., Colbourne, J.S. and Foster, D.M.,1996, Comparison
between slow sand and high rate biofiltration. In: Advances in slow sand and alternative bio-
logical filtration, Graham, N. and Collins, M.R. (eds), pp. 3-10, John Wiley & Sons Ltd,
Chichester England.

Raunkjaer, K., Hvitvedjacobsen, T. and Nielsen, P.H., 1994, Measurement of Pools of Pro-
tein, Carbohydrate and Lipid in Domestic Waste-Water. Water Res 28(2), 251-262.
Literatures 129

Rooklidge, S.J., Burns, E.R. and Bolte, J.P., 2005, Modeling antimicrobial contaminant re-
moval in slow sand filtration. Water Res 39(2-3), 331-339.

Roorda, J.H., 2004, Filtration characteristics in dead-end ultrafiltration of wwtp-effluent.

Ph.D. Dissertation, Delft University of Technology, Delft.

Roorda, J.H. and van der Graaf, J.H.M., 2005, SUR test used for optimisation of membrane
filtration plants treating wastewater effluents. Desalination 179(1-3), 131-150.

Rosenberger, S., 2003, Charakterisierung von belebtem Schlamm in Membranbelebungs-reaktoren zur Ab-
wasserreinigung (in German), Technische Universität Berlin, Berlin.

Rosenberger, S., Evenblij, H., Poele, S.T., Wintgens, T. and Laabs, C., 2005, The importance
of liquid phase analyses to understand fouling in membrane assisted activated sludge proc-
esses - six case studies of different European research groups. J Membrane Sci 263(1-2), 113-
126.

Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M. and Schrotter, J.C.,
2006, Impact of colloidal and soluble organic material on membrane performance in mem-
brane bioreactors for municipal wastewater treatment. Water Res 40(4), 710-720.

Schippers, J.C. and Verdouw, J., 1980, Modified Fouling Index, a Method of Determining the
Fouling Characteristics of Water. Desalination 32(1-3), 137-148.

Schrader, G.A., Zwijnenburg, A. and Wessling, M., 2005, The effect of WWTP effluent zeta-
potential on direct nanofiltration performance. J Membrane Sci 266(1-2), 80-93.

Servais, P., Billen, G. and Hascoet, M.C., 1987, Determination of the Biodegradable Fraction
of Dissolved Organic-Matter in Waters. Water Res. 21(4), 445-450.

Shen, Y.X., Zhao, W.T., Xiao, K. and Huang, X., 2010, A systematic insight into fouling pro-
pensity of soluable microbial products in membrane bioreactors based on hydrophobic inter-
action and size exclusion. J Membrane Sci 346, 387-393.

Shon, H.K., Vigneswaran, S., Kim, I.S., Cho, J. and Ngo, H.H., 2004, Effect of pretreatment
on the fouling of membranes: application in biologically treated sewage effluent. J Membrane
Sci 234(1-2), 111-120.

Shon, H.K., Vigneswaran, S., Kim, I.S., Cho, J. and Ngo, H.H., 2006, Fouling of ultrafiltra-
tion membrane by effluent organic matter: A detailed characterization using different organic
fractions in wastewater. J. Membr. Sci. 278(1-2), 232-238.

Simpson, D.R., 2008, Biofilm processes in biologically active carbon water purification. Wa-
ter Res 42(12), 2839-2848.

Singh, G. and Song, L.F., 2006, Cake compressibility of silica colloids in membrane filtration
processes. Industrial & Engineering Chemistry Research 45(22), 7633-7638.

Smith, P.J., Shon, H.K., Vigneswaran, S., Ngo, H.H. and Nguyen, H., 2006, Productivity en-
hancement in a cross-flow ultrafiltration membrane system through automated de-clogging
operations. J Membrane Sci 280(1-2), 82-88.

Sperlich, A., Zheng, X., Ernst, M. and Jekel, M., 2008, An integrated wastewater reuse con-
cept combining natural reclamation techniques, membrane filtration and metal oxide adsorp-
tion. Water Sci Technol 57(6), 909-914.

Sullivan, F., 2007, Membrane filtration technologies tackle water reuse and purification.
Membrane technology 2007(1), 9-11.
130 Literatures

Susanto, H. and Ulbricht, M., 2005, Influence of ultrafiltration membrane characteristics on

adsorptive fouling with dextrans. J Membrane Sci 266(1-2), 132-142.

Susanto, H., Franzka, S. and Ulbricht, M., 2007, Dextran fouling of polyethersulfone ultrafil-
tration membranes - Causes, extent and consequences. J Membrane Sci 296(1-2), 147-155.

Susanto, H., Arafat, H., Janssen, E.M.L. and Ulbricht, M., 2008, Ultrafiltration of polysaccha-
ride-protein mixtures: Elucidation of fouling mechanisms and fouling control by membrane
surface modification. Sep Purif Technol 63(3), 558-565.

Tansel, B., Sager, J., Garland, J., Xu, S.H., Levine, L. and Bisbee, P., 2006, Deposition of
extracellular polymeric substances (EPS) and microtopographical changes on membrane sur-
faces during intermittent filtration conditions. J Membrane Sci 285(1-2), 225-231.

Teodosiu, C.C., Kennedy, M.D., Van Straten, H.A. and Schippers, J.C., 1999, Evaluation of
secondary refinery effluent treatment using ultrafiltration membranes. Water Res 33(9), 2172-
2180.

Thullner, M., Mauclaire, L., Schroth, M.H., Kinzelbach, W. and Zeyer, J., 2002, Interaction
between water flow and spatial distribution of microbial growth in a two-dimensional flow
field in saturated porous media. Journal of Contaminant Hydrology 58(3-4), 169-189.

Tian, J.Y., Chen, Z.L., Yang, Y.L., Liang, H., Nan, J. and Li, G.B., 2009, Consecutive chemi-
cal cleaning of fouled PVC membrane using NaOH and ethanol during ultrafiltration of river
water. Water Res (in press).

Torrestiana-Sanchez, B., Balderas-Luna, L., Brito-De la Fuente, E. and Lencki, R.W., 2007,
The use of membrane-assisted precipitation for the concentration of xanthan gum. J Mem-
brane Sci 294(1-2), 84-92.

Urfer, D., Huck, P.M., Booth, S.D.J. and Coffey, B.M., 1997, Biological filtration for biode-
gradable organic matter and particle removal: a critical review. Journal American Water
Works Association 89(12), 83-98.

van der Hoek, J.P., Hofman, J.A.M.H., Bonne, P.A.C., Nederlof, M.M. and Vrouwenvelder,
H.S., 2000, RO treatment: selection of a pretreatment scheme based on fouling characteristics
and operating conditions based on environmental impact. Desalination 127(1), 89-101.

Vrijenhoek, E.M., Elimelech, M. and Hong, S.K., 2000, Influence of Membrane Properties,
Solution Chemistry and Hydraudynamics on Colloidal Fouling of RO and NF Membranes.
Membrane Separation Processes in Aquatic Systems, Washington DC.

Wang, L., Wang, X.D. and Fukushi, K.I., 2008, Effects of operational conditions on ultrafil-
tration membrane fouling. Desalination 229(1-3), 181-191.

Wang, X.D., Wang, L., Liu, Y. and Duan, W.S., 2007, Ozonation pretreatment for ultrafiltra-
tion of the secondary effluent. J Membrane Sci 287(2), 187-191.

Wang, X.M. and Waite, T.D., 2008a, Impact of gel layer formation on colloid retention in
membrane filtration processes. J Membrane Sci 325(1), 486-494.

Wang, X.M. and Waite, T.D., 2008b, Gel layer formation and hollow fiber membrane filter-
ability of polysaccharide dispersions. J Membrane Sci 322(1), 204-213.

Wang, X.M. and Waite, T.D., 2009, Role of gelling soluble and colloidal microbial products
in membrane fouling (in press). Environmental Science & Technology.
Literatures 131

Water21, 2008, Online in Orange County: Early results from the Groundwater Replenishment
system. WATER21 August 2008, 22.

Weber-Shirk, M.L. and Chan, K.L., 2007, The role of aluminum in slow sand filtration. Water
Res 41(6), 1350-1354.

Wessling, M., 2001, Two-dimensional stochastic modeling of membrane fouling. Sep Purif
Technol 24(3), 375-387.

Wickramasinghe, S.R., Bower, S.E., Chen, Z., Mukherjee, A. and Husson, S.M., 2009, Relat-
ing the pore size distribution of ultrafiltration membranes to dextran rejection. J Membrane
Sci 340(1-2), 1-8.

Wintgens, T., Melin, T., Schafer, A., Khan, S., Muston, M., Bixio, D. and Thoeye, C., 2005,
The role of membrane processes in municipal wastewater reclamation and reuse. Desalination
178(1-3), 1-11.

Wong, P.C.Y., Kwon, Y.N. and Criddle, C.S., 2009, Use of atomic force microscopy and
fractal geometry to characterize the roughness of nano-, micro-, and ultrafiltration mem-
branes. J Membrane Sci 340(1-2), 117-132.

Wu, J.L., Le-Clech, P., Stuetz, R.M., Fane, A.G. and Chen, V., 2008, Effects of relaxation
and backwashing conditions on fouling in membrane bioreactor. J Membrane Sci 324(1-2),
26-32.

Wyart, Y., Georges, G., Demie, C., Amra, C. and Moulin, P., 2008, Membrane characteriza-
tion by microscopic methods: Multiscale structure. J Membrane Sci 315(1-2), 82-92.

Xiao, K., Wang, X.M., Huang, X., Waite, T.D. and Wen, X.H., 2009, Analysis of polysaccha-
ride, protein and humic acid retention by microfiltration membranes using Thomas' dynamic
adsorption model. J Membrane Sci 342(1-2), 22-34.

Xie, R.J., Gomez, M.J. and Xing, Y.J., 2006, Permeate flux optimisation of a pilot microfiltra-
tion plant for cost-effectiveness of water reclamation for reuse. Journal of Environmental Sci-
ence and Health Part A -Toxic/Hazardous Substances & Environmental Engineering 41(7),
1171-1183.

Xie, R.J., Gomez, M.J. and Xing, Y.J., 2008, Understanding permeability decay of pilot-scale
microfiltration in secondary effluent reclamation. Desalination 219(1-3), 26-39.

Xu, W.D. and Chellam, S., 2005, Initial stages of bacterial fouling during dead-end microfil-
tration. Environmental Science & Technology 39(17), 6470-6476.

Yamamura, H., Kimura, K. and Watanabe, Y., 2007a, Mechanism Involved in the Evolution
of Physically Irreversible Fouling in Microfiltration and Ultrafiltration Membranes Used for
Drinking Water Treatment. Environ. Sci. Technol.,
41(19), 6789 -6794.

Yamamura, H., Okimoto, K., Kimura, K. and Watanabe, Y., 2007b, Influence of calcium on
the evolution of irreversible fouling in microfiltration/ultrafiltration membranes. J Water Sup-
ply Res T 56(6-7), 425-434.

Ye, Y. and Chen, V., 2005, Reversibility of heterogeneous deposits formed from yeast and
proteins during microfiltration. J Membrane Sci 265(1-2), 20-28.

Ye, Y., Le Clech, P., Chen, V. and Fane, A.G., 2005, Evolution of fouling during crossflow
filtration of model EPS solutions. J Membrane Sci 264(1-2), 190-199.
132 Literatures

Yuan, W., Kocic, A. and Zydney, A.L., 2002, Analysis of humic acid fouling during microfil-
tration using a pore blockage-cake filtration model. J Membrane Sci 198(1), 51-62.

Zhang, G.J., Ji, S.L., Gao, X. and Liu, Z.Z., 2008, Adsorptive fouling of extracellular poly-
meric substances with polymeric ultrafiltration membrances. J Membrane Sci 309(1-2), 28-
35.

Zhang, J., Chua, H.C., Zhou, J. and Fane, A.G., 2006a, Factors affecting the membrane per-
formance in submerged membrane bioreactors. J Membrane Sci 284(1-2), 54-66.

Zhang, M.M., Li, C., Benjamin, M.M. and Chang, Y.J., 2003, Fouling and natural organic
matter removal in adsorben/membrane systems for drinking water treatment. Environmental
Science & Technology 37(8), 1663-1669.

Zhang, R., Vigneswaran, S., Ngo, H.H. and Nguyen, H., 2006b, Magnetic ion exchange
(MIEX (R)) resin as a pre-treatment to a submerged membrane system in the treatment of
biologically treated wastewater. Desalination 192(1-3), 296-302.

Zhang, S.L. and Huck, P.M., 1996, Parameter estimation for biofilm processes in biological
water treatment. Water Res 30(2), 456-464.

Zhang, X., LaFleur, L. and Zhou, H., 2009, Mapping biofilm formation in large-scale sub-
merged membrane bioreactor to treat municipal wastewater. IWA Membrane Technology
Conference and Exhibition 2009, Beijing, China.

Zhang, X.Q. and Bishop, P.L., 2003, Biodegradability of biofilm extracellular polymeric sub-
stances. Chemosphere 50(1), 63-69.

Zhao, P., Takizawa, S., Katayama, H. and Ohgaki, S., 2005, Factors causing PAC cake foul-
ing in PAC-MF (powdered activated carbon-microfiltration) water treatment systems. Water
Science And Technology 51(6-7), 231-240.

Zheng, X., Ernst, M. and Jekel, M., 2009a, Identification and Quantification of Major Organic
Foulants in Treated Domestic Wastewater Affecting Filterability in Dead-end Ultrafiltration.
Water Res. 43(1), 238-244.

Zheng, X., Mehrez, R., Ernst, M. and Jekel, M., 2009b, Effect of slow sand filtration of
treated domestic wastewater as pre-treatment to UF. Desalination 249(2), 591-595.

Zularisam, A.W., Ismail, A.F., Salim, M.R., Sakinah, M. and Hiroaki, O., 2007a, Fabrication,
fouling and foulant analyses of asymmetric polysulfone (PSF) ultrafiltration membrane fouled
with natural organic matter (NOM) source waters. J Membrane Sci 299(1-2), 97-113.

Zularisam, A.W., Ismail, A.F., Salim, M.R., Sakinah, M. and Ozaki, H., 2007b, The effects of
natural organic matter (NOM) fractions on fouling characteristics and flux recovery of ul-
trafiltration membranes. Desalination 212(1-3), 191-208.