University of Southern California Dissertations and Theses

Wastewater reclamation and potable reuse with novel processes: membrane performance and system integration

(USC Thesis Other)

Wastewater reclamation and potable reuse with novel processes: membrane performance and system integration

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Wastewater reclamation and potable reuse
with novel processes:
Membrane performance and system integration

A dissertation submitted in partial satisfaction of the requirements
for the degree of Doctor of Philosophy in
Environmental Engineering

Submitted by

Christopher P. Morrow

August, 2018

Degree approved by the faculty of the
University of Southern California Graduate School

a
Amy E. Childress, Ph.D. (Department chair; Ph.D. advisor)
a
Kelly T. Sanders, Ph.D.
a
Adam L. Smith, Ph.D.
a
Daniel McCurry, Ph.D.
b
Travis Longcore, Ph.D.

a
Viterbi School of Engineering; Sonny Astani Department of Civil and Environmental Engineering

b
School of Architecture and Dornsife Spatial Sciences Institute

i

Table of Contents

List of Tables .............................................................................................................................. iv
List of Figures ............................................................................................................................. v
Acknowledgements .................................................................................................................. vii
Support ...................................................................................................................................... viii
Chapter 1. Introduction ................................................................................................................. 1
1.1. Problem statement and significance ................................................................................... 1
1.2. Objectives and scope of work ............................................................................................. 4
1.3. Dissertation organization .................................................................................................... 4
Chapter 2. Integrating an osmotic membrane bioreactor system with membrane distillation for
potable reuse ................................................................................................................................ 6
2.0. Abstract .............................................................................................................................. 6
2.1. Introduction ......................................................................................................................... 6
2.2. Materials and methods ..................................................................................................... 10
2.2.1. Bench-scale FO system; CTA and TFC membrane testing ....................................... 10
2.2.2. OMBR subsystem; longer-term testing ...................................................................... 11
2.2.2.1. Bioreactor ............................................................................................................ 11
2.2.2.2. Forward osmosis subsystem ............................................................................... 12
2.2.3. Integration of OMBR and MD subsystems ................................................................. 13
2.2.3.1. Membrane distillation subsystem ........................................................................ 13
2.2.3.2. Integrated OMBR-MD system ............................................................................. 14
2.2.4. Analytical methods ..................................................................................................... 16
2.3. Results and discussion ..................................................................................................... 17
2.3.1. Membrane selection; CTA and TFC comparison ....................................................... 17
2.3.2. Long-term operation of the OMBR subsystem ........................................................... 19
2.3.2.1. Biological treatment with aerobic/anoxic cycling bioreactor ................................ 19
2.3.2.2. Forward osmosis; Water flux behavior ................................................................ 21
2.3.3. Integrated OMBR-MD system .................................................................................... 23
2.3.3.1. Productivity of the integrated system .................................................................. 23
2.3.3.2. Removal by the integrated system ...................................................................... 24
2.4. Conclusions ...................................................................................................................... 25

ii

Chapter 3. Submerged or sidestream? The influence of module configuration on fouling and
salinity in osmotic membrane bioreactors ................................................................................... 27
3.0. Abstract ............................................................................................................................ 27
3.1. Introduction ....................................................................................................................... 28
3.2. Materials and methods ..................................................................................................... 31
3.2.1. Forward osmosis sub-systems .................................................................................. 31
3.2.2. Forward osmosis biofouling experimental protocol .................................................... 34
3.2.3. Activated sludge characterization .............................................................................. 36
3.2.4. Fouling layer characterization; scanning electron microscopy and energy dispersive
X-ray spectroscopy .............................................................................................................. 36
3.2.5. Modeling the steady-state salt concentration of a continuous OMBR ....................... 37
3.3. Results and discussion ..................................................................................................... 38
3.3.1. Water flux ................................................................................................................... 38
3.3.2. Membrane surface characterization ........................................................................... 41
3.3.2.1. Cake layer thickness and morphology ................................................................ 42
3.3.2.2. Elemental composition of cake layers ................................................................. 45
3.3.3. Implications on reverse salt flux and OMBR salinity .................................................. 46
3.3.3.1. Specific reverse salt flux ...................................................................................... 46
3.3.3.2. Steady-state OMBR salinity ................................................................................ 48
3.3.4. Sidestream FO module with the pilot-scale OMBR-MD system ................................ 49
3.3.4.1. Sidestream OMBR-MD water flux and water production ..................................... 51
3.3.4.2. Sidestream OMBR water quality at pilot-scale .................................................... 54
3.3.4.3. Spatial distribution of MD membrane fouling ....................................................... 56
3.3.5. Specific energy of submerged and sidestream OMBR systems ................................ 59
3.4. Conclusions ...................................................................................................................... 60
Chapter 4. Membrane compaction and effects on membrane properties in FO ........................ 62
4.0. Abstract ............................................................................................................................ 62
4.1. Introduction ....................................................................................................................... 62
4.2. Materials and Methods ..................................................................................................... 64
4.2.1. Membranes and bench-scale FO system .................................................................. 64
4.2.2. Membrane characterization ....................................................................................... 65
4.2.2.1. RO-FO characterization method ......................................................................... 65
4.2.2.2. Stepwise FO characterization method ................................................................ 66
4.2.2.3. RO-FO and stepwise FO method comparison .................................................... 67
4.3. Result and discussion ....................................................................................................... 68

iii

4.3.1. Transport and structural parameters .......................................................................... 68
4.3.2. Accuracy of characterization methods ....................................................................... 69
4.3.3. FO membrane compaction under hydraulic pressure ................................................ 70
4.4. Conclusions ...................................................................................................................... 73
Chapter 5. Relating forward osmosis membrane properties to performance under fouling
conditions: Implications of the membrane-independent limiting flux ........................................... 74
5.0. Abstract ............................................................................................................................ 74
5.1. Introduction ....................................................................................................................... 75
5.2. Materials and methods ..................................................................................................... 80
5.2.1. Membranes and bench-scale FO system .................................................................. 80
5.2.2. Membrane characterization ....................................................................................... 81
5.2.3. Synthetic activated sludge feed solution .................................................................... 83
5.2.4. Experimental procedures ........................................................................................... 84
5.2.4.1. Stepwise draw solution osmotic pressure experiments ....................................... 84
5.2.4.2. Stepwise limiting flux and limiting osmotic pressure ........................................... 85
5.2.4.3. Validation of stepwise limiting flux; constant draw solution osmotic pressure ..... 86
5.2.4.4. Limiting flux with constant draw solution osmotic pressure ................................. 87
5.2.5. Membrane and foulant cake layer resistances .......................................................... 87
5.3. Result and discussion ....................................................................................................... 88
5.3.1. Transport and structural parameters .......................................................................... 88
5.3.2. Stepwise experiments; limiting water flux and limiting osmotic pressure ................... 89
5.3.3. Effect of osmotic pressure increases on RSF ............................................................ 92
5.3.4. Validation of stepwise limiting flux ............................................................................. 94
5.3.5. Membrane and cake layer resistances ...................................................................... 96
5.4. Conclusions ...................................................................................................................... 98
Chapter 6. Conclusions ............................................................................................................ 100
6.1. Research Synopsis ......................................................................................................... 100
6.1.1. Summary of OMBR-MD system integration ............................................................. 100
6.1.2. Summary of effects of forward osmosis membrane module configuration .............. 101
6.1.3. Summary of the effects of hydraulic pressure on forward osmosis membranes ..... 102
6.1.4. Summary of the limiting flux phenomenon in forward osmosis ................................ 102
References ............................................................................................................................... 104

iv

List of Tables

Table 3.1. Cake layer thickness and sodium content of cake layer surfaces measured by
SEM/EDS. Cake layer thickness error represents the standard deviation of ten measurements
from the cross-section image of one membrane and Na
+
error represents the standard deviation
of four EDS scans of the cake layer surface from one membrane. ............................................. 43
Table 3.2. Experimental and optimized energy consumption by submerged and sidestream
OMBRs. ....................................................................................................................................... 59
Table 4.1. Osmotic pressures (MPa) and corresponding NaCl draw solution concentrations
(g/L). ............................................................................................................................................ 66
Table 4.2. Water permeability (A), solute permeability (B), and structural parameters (S)
determined by RO-FO and FO methods. .................................................................................... 68
Table 5.1. Osmotic pressures (MPa) and corresponding NaCl draw solution concentrations
(g/L). ............................................................................................................................................ 82
Table 5.2. Concentrations of chemical components in synthetic activated sludge feed solution 83
Table 5.3. Water permeability (A), solute permeability (B), and structural parameter (S)
determined using the FO characterization method. .................................................................... 89
Table 5.4. Limiting flux values (J
w,h
) calculated from stepwise and constant draw solution
osmotic pressure (π
D
) experiments. ............................................................................................ 96

v

List of Figures

2.1. Graphical representation of an integrated OMBR-MD system in direct configuration. ........... 9
2.2. The integrated OMBR-MD system with the OMBR subsystem, MD subsystem, and transfer
pumps. ........................................................................................................................................ 15
2.3. Water flux for CTA and TFC membranes treating activated sludge from the pilot-scale
bioreactor. ................................................................................................................................... 17
2.4. A representative 4-hour period of bioreactor operation showing the effect of 30-minute
aerobic/anoxic cycling ................................................................................................................. 20
2.5. Long-term trends in NH
4
+
-N, COD, NO
3
-
-N, and NO
2
-
-N concentrations using a single
bioreactor with alternating aerobic/anoxic environments ............................................................ 21
2.6. The pilot-scale OMBR was seeded with activated sludge on Day 0 and operated
continuously for 200 days. .......................................................................................................... 22
2.7. Productivity and temperatures of the integrated OMBR-MD system during continuous
integrated operation. ................................................................................................................... 24
3.1. Submerged and sidestream FO testing configurations. ....................................................... 32
3.2. Experimental water flux for all configurations (control, submerged, sidestream) for 35 and
100 g/L NaCl draw solution concentrations. ................................................................................ 39
3.3. SEM images of selective layer surface of a virgin CTA membrane and membranes fouled in
control, submerged, and sidestream configurations with 35 g/L NaCl draw solution. ................. 42
3.4. SEM images of selective layer surface of a virgin CTA membrane and membranes fouled in
control, submerged, and sidestream configurations with 100 g/L NaCl draw solution. ............... 44
3.5. EDS spectra of the cake layers formed in the control, submerged, and sidestream
configurations. ............................................................................................................................. 46
3.6. Specific reverse salt flux for all membrane conditions in the control, submerged, and
sidestream configurations with 35 g/L and 100 g/L NaCl draw solution concentrations ............. 47
3.7. Modeled OMBR salinity as a function of time ...................................................................... 49
3.8. OMBR subsystem with mesh pre-filter and 4040 spiral-wound FO membrane module ...... 50
3.9. FO water flux, MD water flux, and bioreactor conductivity over 30 days of operation ......... 52
3.10. Daily FO and MD water production .................................................................................... 52
3.11. MD water flux and distillate conductivity during integrated OMBR-MD operation .............. 53
3.12. Feed and distillate inlet and outlet pressures during integrated OMBR-MD operation ...... 54
3.13. Soluble COD concentrations for the sidestream OMBR-MD system ................................. 55
3.14. Ammonium-nitrogen concentrations for the sidestream OMBR-MD system ...................... 55
3.15. Nitrate-nitrogen concentrations for the sidestream OMBR-MD system ............................. 56
3.16. SEM images and EDS spectra of used MD membranes; feed side ................................... 57

vi

3.17. SEM images and EDS spectra of used MD membranes; distillate side ............................. 58
3.18. Feed channel of the sidestream 4040 FO module with integral spacer ............................. 60
4.1. Comparison of membrane parameters obtained with the RO-FO and FO characterization
methods ...................................................................................................................................... 70
4.2. SEM images of FTS CTA membranes before and after RO testing .................................... 71
4.3. SEM images of HTI TFC membranes before and after RO testing ...................................... 72
4.4. SEM images of Toray TFC membranes before and after RO testing .................................. 72
5.1. Concentration polarization using limiting draw solution osmotic pressure (π
L
) without
membrane fouling and with membrane fouling ........................................................................... 78
5.2. Graphical representation of the bench-scale FO testing system ......................................... 81
5.3. Method for determining limiting osmotic pressure ................................................................ 86
5.4. Stepwise limiting flux and limiting osmotic pressure experiments ........................................ 90
5.5. Stepwise experiments and reverse salt flux ......................................................................... 93
5.6. Continuous osmotic pressure and limiting flux ..................................................................... 95
5.7. Resistances with stepwise osmotic pressure experiments .................................................. 97

vii

Acknowledgements

First and foremost, I would like to thank Jenna Cavelle for her tireless love and support over
the past decade. She has witnessed the day-to-day challenges and successes of the research
process and she has continuously revitalized my inspiration to continue. Without her, my
education would surely not have been possible and I am grateful that we can celebrate this
milestone together.
I would like to convey my most sincere appreciation to my advisor, Dr. Amy Childress. Her
unyielding drive to achieve perfection has been a source of inspiration throughout my research.
I take great pride in being among the first members of her group at USC and I have immensely
enjoyed watching it grow and flourish.
I also want to thank the other committee members, Prof. Kelly T. Sanders, Prof. Adam L.
Smith, Prof. Daniel McCurry, and Prof. Travis Longcore for their feedback and enthusiasm and
for bringing focus to the real-world significance of this dissertation. In addition, I would like to
acknowledge Prof. Sage Hiibel and Prof. Eric Marchand at the University of Nevada, Reno, and
Prof. Andrea Achilli at the University of Arizona for their mentorship and guidance throughout
the SERDP project.
I would also like to thank the graduate students in the Childress lab, Allyson McGaughey,
Ryan Gustafson, and Lauren Crawford at the University of Southern California, and Nicole
Furtaw and Derrick Satterfield at the University of Nevada, Reno, for their countless hours of
assistance both in the lab and in the office, and most importantly for their friendship throughout
this process.

viii

Support
This work was supported by the Strategic Environmental Research and Development
Program (SERDP ER-2237), the National Sciences Foundation Graduate Research Fellowship
Program (nsf13584), the University of Southern California’s Viterbi School of Engineering
Fellowship, and the 2016 American Membrane Technology Association/US Bureau of
Reclamation Fellowship Program. The authors would like to thank Porifera Inc., Oasys Water
Inc., and Toray Industries Inc. for supplying TFC membranes, Hydration Technologies Inc. for
supplying CTA and TFC membranes and the pilot-scale membrane module, and the Center for
Electron Microscopy and Microanalysis (CEMMA) at the University of Southern California for
their assistance.

1

Chapter 1. Introduction
1.1. Problem statement and significance
Urban reliance on imported water and increasing water supply variability due to climate
change have intensified efforts to increase the resiliency of water supplies on a global scale [1].
For reuse applications, membrane bioreactors (MBRs) with microfiltration (MF) or ultrafiltration
(UF) membranes submerged in a biological reactor have emerged as an efficient wastewater
treatment process to provide high-quality filtrate to a subsequent reverse osmosis (RO) process
followed by advanced oxidation [2-4]. More recently, osmotic membrane bioreactors (OMBRs)
with forward osmosis (FO) membranes submerged in the bioreactor are being investigated as
low-fouling alternatives to conventional MBRs [5-7]. If the opportunity exists to reconcentrate the
FO draw solution with waste-heat driven membrane distillation (MD), then the OMBR-MD
system can provide high quality product water with a low electrical energy requirement.
In an aerobic OMBR-MD system, wastewater is fed into a bioreactor that is aerated to
supply oxygen to the biomass and scour the membrane. Through osmosis, water diffuses from
the bioreactor, across a semi-permeable FO membrane, into the draw solution [8]. The FO
membrane acts as a barrier to solute transport and provides high rejection of contaminants in
the wastewater stream [9-11]. The diluted draw solution is sent to MD for reconcentration and
generation of product water [7, 12, 13].
Compared to the MF or UF process in a conventional MBR, the FO process in the OMBR
offers the advantage of much higher rejection (semi-permeable membrane versus microporous
membrane) at lower hydraulic pressure [14-16]. The FO membrane inside the bioreactor also
has much less fouling propensity than MF/UF membranes, and thus, requires less scouring and
much less frequent backwashing [17]. When comparing an integrated OMBR-MD system with a
conventional MBR-RO system, the higher rejection of the FO membranes also results in lower
fouling propensity for the downstream process (RO or MD) [18]. When comparing FO draw

2

solution recovery using RO or MD, both processes can utilize waste heat to reduce the required
energy input; however, MD utilizes waste heat directly with a heat exchanger, whereas RO
requires additional equipment (e.g., a pump, boiler, and turbine) to convert waste heat to
electrical energy [19], resulting in a larger footprint. Furthermore, the electrical energy
requirement for RO would increase as feed solution salinity increases whereas MD is only
minimally affected by feed solution salinity [20].
MF and UF membranes in MBRs pass dissolved solids while FO membranes in OMBRs
generally do not. However, a key concern with OMBRs is elevated bioreactor salinity caused by
retention of dissolved solutes from the influent wastewater and diffusion of draw solutes from
the FO draw solution into the bioreactor by a phenomenon known as reverse salt flux (RSF) [21-
23]. RSF is a critical concern of osmotic processes because it diminishes the salinity gradient,
thus reducing the osmotic driving force for water flux; in the case of OMBRs, RSF also
increases bioreactor salinity [24]. High bioreactor salinity can affect microbial processes, for
example decreasing the activity of nitrogen-removing bacteria [25, 26]. Ammonia-oxidizing
bacteria, which are responsible for ammonia removal via nitrification, are particularly sensitive to
salinity and sharp decreases in nitrification efficiency have been reported for bioreactors with
salinities ranging from 2-15 g/L NaCl [27-30].
The hydrodynamics and mixing that influence RSF and membrane fouling in OMBRs is
highly influenced by FO module configuration [7, 31]. In the submerged configuration, the draw
solution is circulated through the interior of the membrane module while the feed solution has no
direct crossflow [7]. Air bubbles can be used as a fouling control strategy with air scour of the
membrane surface reducing membrane fouling [23, 32]. In the sidestream configuration, the
membrane module is external to the bioreactor and the feed and draw solutions flow tangentially
across the membrane surfaces. High crossflow velocities can be used as a fouling control
strategy with hydraulic scour of the membrane surface reducing membrane fouling [33]. Large

3

ranges of steady-state salinities have been reported for both submerged (4.1-12.6 g/L NaCl)
and sidestream (4.0-33.5 g/L NaCl) OMBR configurations [5, 23, 25, 26, 34-40]. Despite the
different salinities reported for each configuration, the role of membrane module configuration
on RSF and bioreactor salinity is not well characterized.
In addition to system-dependent parameters such as membrane module configuration,
membrane-dependent parameters (e.g., water permeability (A), solute permeability (B), and the
structural parameter (S)) are used as inputs to predict membrane performance. Several recent
pilot-scale FO studies have reported a hydraulic pressure drop that occurs across larger
membrane modules that is not observed with smaller, bench-scale testing systems. Thus, when
FO processes are scaled-up, FO membrane performance will depend on a combination of
osmotic pressure and hydraulic pressure. While osmotic pressure is unlikely to affect FO
membrane structure and transport properties, hydraulic pressure may cause FO membrane
compaction, changing membrane performance. Therefore, understanding compaction with FO
membranes and how this affects membrane performance are also key aspects for designing
novel FO membrane materials.
Because FO is an osmosis-driven process, FO membrane fouling is less severe and more
reversible than fouling in pressure-driven membrane processes (i.e., nanofiltration (NF) and RO)
[41-43]. However, like pressure-driven membrane processes, FO membrane fouling can
significantly hinder water flux. For RO and NF membranes, only initial foulant deposition is
dependent on membrane properties and water flux is eventually restricted to the limiting flux,
which has been shown to be membrane-independent [44, 45]. The limiting flux concept may
also be applicable in FO, however, due to phenomenon unique to FO (e.g., internal
concentration polarization and RSF), FO membrane fouling is more complex than membrane
fouling in RO and NF. Thus, identification of membrane properties that influence performance

4

under fouling conditions is key for improving performance in wastewater reuse systems where
membrane fouling is expected to occur.
1.2. Objectives and scope of work
The overall goals of this work were to develop a wastewater reuse system with a small
footprint and low electrical energy consumption, and to use results of bench-scale studies to
refine the system and determine appropriate configurations and operating conditions. As part of
this research, the role of FO membrane properties on performance under fouling conditions was
investigated. First, a fully integrated pilot-scale OMBR-MD system with submerged FO
membranes and a unique dosing-and-transfer system was developed for potable reuse. Next,
submerged and sidestream FO membrane modules were tested at bench-scale to determine
the implications in OMBR applications. The scalability of these results was tested with an
additional investigation using a sidestream FO membrane module with the pilot-scale system. A
follow-up study was subsequently conducted to investigate the impact of hydraulic pressure on
FO membrane materials and transport properties. Finally, the importance of FO membrane
properties during membrane fouling was investigated. A limiting flux theory in FO was
developed to describe how hydraulic resistance due to membrane fouling is counterbalanced by
a changing magnitude of internal concentration polarization and a method was developed for
determination of the limiting flux in FO. This dissertation will cover these critical aspects of FO
membrane fouling and integration of FO membrane systems for wastewater reclamation and
potable reuse.
1.3. Dissertation organization
This dissertation is a compilation of papers written over the course of the dissertation
research. Chapter 2 is an entire paper that was published in Desalination. Chapter 3 consists of
an entire paper published in the Journal of Membrane Science as well as additional sections of

5

a manuscript that is in preparation. Chapter 4 represents a draft of a manuscript that is in the
final stages of preparation and will be submitted for publication. Chapter 5 represents a
completed manuscript of a paper that will be submitted for publication.

6

Chapter 2. Integrating an osmotic membrane bioreactor system with membrane
distillation for potable reuse

This chapter was published in Desalination [46].
2.0. Abstract
A novel OMBR-MD system was designed and fabricated to treat wastewater for potable
reuse. Two FO membranes were evaluated at bench-scale before proceeding with longer-term
operation to evaluate water flux and biological treatment of the pilot-scale OMBR subsystem.
There was no statistical difference between cellulose triacetate and thin-film composite
membrane performance for activated sludge feed solution. Also, FO water flux during long-term
operation was the same for 20 and 35 g/L NaCl draw solutions; however, the 35 g/L NaCl draw
solution resulted in greater reverse salt flux and higher conductivity in the bioreactor. The
OMBR subsystem was integrated with an MD subsystem to reconcentrate the draw solution and
produce high quality product water. Results from long-term testing using a high-strength
wastewater showed 98.4% COD removal and 90.2% NH
4
+
-N could be achieved in a single
bioreactor by alternating aeration on/off cycles to control the redox environment. An automated
dosing and transfer system was developed to maintain constant FO draw solution concentration
and prevent heat from being transferred to the bioreactor, which is critical for maintaining
biological nitrogen removal.
2.1. Introduction
Persistent drivers of water stress, such as climate change and urban population growth, are
motivating interest in potable reuse systems [1]. For extreme environments (e.g., in space) or in
locations with extenuating circumstances (e.g., military forward operating bases), water
treatment systems that are compact and easily deployable are required. Furthermore, at forward
operating bases, significant risk is associated with convoys supplying potable water and fuel to
the base as well as convoys taking wastewater away; thus, a system that can reclaim

7

wastewater for potable reuse with a low electrical energy requirement is of critical importance.
The development of potable reuse systems tailored to meet these unique challenges has direct
implications for development of municipal potable reuse systems.
For reuse applications, membrane bioreactors (MBRs) with microfiltration (MF) or
ultrafiltration (UF) membranes submerged in a biological reactor have emerged as an efficient
wastewater treatment process to provide high-quality filtrate to a subsequent reverse osmosis
(RO) process followed by advanced oxidation [2-4]. More recently, osmotic membrane
bioreactors (OMBRs) with forward osmosis (FO) membranes submerged in the bioreactor are
being investigated as low-fouling alternatives to conventional MBRs [5-7]. If the opportunity
exists to reconcentrate the FO draw solution with waste-heat driven membrane distillation (MD),
then the OMBR-MD system can provide high quality product water with a low electrical energy
requirement.
In an aerobic OMBR-MD system, wastewater is fed into a bioreactor that is aerated to supply
oxygen to the biomass and scour the membrane. Through osmosis, water diffuses from the
bioreactor, across a semi-permeable FO membrane, into the draw solution [8]. The FO
membrane acts as a barrier to solute transport and provides high rejection of contaminants in
the wastewater stream [9-11]. The diluted draw solution is sent to MD for reconcentration and
generation of product water [7, 12, 13].
Compared to the MF or UF process in a conventional MBR, the FO process in the OMBR
offers the advantage of much higher rejection (semi-permeable membrane versus microporous
membrane) at lower hydraulic pressure [14-16]. The FO membrane inside the bioreactor also
has much less fouling propensity than MF/UF membranes, and thus, requires less scouring and
much less frequent backwashing [17]. When comparing an integrated OMBR-MD system with a
conventional MBR-RO system, the higher rejection of the FO membranes also results in lower
fouling propensity for the downstream process (RO or MD) [18]. When comparing FO draw

8

solution recovery using RO or MD, both processes can utilize waste heat to reduce the required
energy input; however, MD utilizes waste heat directly with a heat exchanger, whereas RO
requires additional equipment (e.g., a pump, boiler, and turbine) to convert waste heat to
electrical energy [19]. The additional equipment would increase system transport and footprint
requirements. In the context of a military forward operating base, minimization of equipment
weight, shipping volume, and operational footprint are prioritized. Furthermore, the electrical
energy requirement for RO would increase as feed solution salinity increases whereas MD is
only minimally affected by feed solution salinity [20].
One unique challenge for OMBRs is a phenomenon known as reverse salt flux (RSF) where
a small amount of FO draw solute diffuses from the draw solution into the feed solution [21-23].
Due to RSF and high rejection of dissolved solutes by the FO membrane, bioreactor salinity
increases over time; increased bioreactor salinity may result in decreased water flux, greater
membrane fouling, and sub-optimal biological carbon and nitrogen removal [23-25]. To
accomplish removal of carbon and nitrogen species, conventional activated sludge systems
typically rely on physical separation of different zones (e.g., aerobic, anoxic, and anaerobic
zones) within a single reactor or within multiple reactors. If footprint is critical, single-reactor
systems with alternating redox environments may be used. In these single-sludge bioreactors,
aeration cycling is used to control the redox environment and achieve carbon and nitrogen
removal [47, 48].
OMBR and MD processes have been integrated in a direct configuration (Fig. 2.1) where
one tank is used to contain the solution that is simultaneously the FO draw solution and the MD
feed solution [49-51]. A key drawback of the direct configuration is transfer of heat from the MD
feed solution, across the FO membrane, and into the bioreactor. In general, higher bioreactor
temperature results in higher FO water flux due to decreasing water viscosity [52, 53]; however,
higher initial water fluxes can result in more fouling and scaling, and hence, severe flux decline

9

[54, 55]. Furthermore, as the growth rate of ammonia-oxidizing bacteria is temperature-sensitive
[56, 57], increased temperature in the bioreactor may inhibit biological nitrogen removal. For an
aerated, fixed-bed bioreactor treating saline wastewater, nitrification by ammonia-oxidizing
bacteria was very low at 6 °C, stable at temperatures between 12.5 and 40 °C, and nonexistent
at 50 °C [56]. For a sequencing aerobic/anoxic system designed for nitrogen and COD removal
in a single reactor, biological removal efficiency declined sharply at temperatures exceeding 37
°C [57]. For this reason, preventing heat transfer from the MD process to the bioreactor is
critical for integrated OMBR-MD systems designed for potable reuse where nitrogen removal is
essential.

Figure 2.1. Graphical representation of an integrated OMBR-MD system in direct configuration.
FO and MD processes work simultaneously from one tank; the FO draw solution tank is also the
MD feed solution tank.

In the recent studies where bench-scale OMBR-MD systems were integrated in the direct
configuration [49-51], continuous aeration was used for nitrification, but no denitrification was
performed. To the author’s knowledge, the current study is the first study to integrate OMBR
and MD into a pilot-scale system that achieves both carbon and nitrogen removal. The novel

10

OMBR-MD system was designed and fabricated for testing under conditions typical of military
forward operating bases. The objective of the research was to address challenges associated
with integration and synergistic operation of larger scale OMBR and MD systems. Prior to
laboratory testing of the OMBR-MD system, two FO membranes were evaluated at bench-scale.
Subsequently, water flux and biological treatment of the OMBR subsystem was measured
during longer-term operation. Then, the OMBR subsystem was integrated with an MD
subsystem to reconcentrate the draw solution and produce high quality product water.
Experimental results for operation of the integrated OMBR-MD system required development of
unique system controls and automated processes.
2.2. Materials and methods
2.2.1. Bench-scale FO system; CTA and TFC membrane testing
CTA and TFC membranes (Hydration Technology Innovations, LLC, Albany, OR) were
tested using a custom-made crossflow module that holds three identically sized membrane
coupons with three independent draw solution loops. Each membrane coupon had an active
area of 42 cm
2
. All experiments were conducted with the selective layer facing the feed solution.
Additional details regarding this testing can be found in [58]. Briefly, the crossflow velocity was
20 cm/s, no spacers were used on the feed side, 31 mil (0.79 mm) spacers (Sterlitech
Corporation, Kent, WA) were used on the draw side, feed and draw solution temperatures were
24 ± 1 °C, and the draw solution was prepared from analytical grade NaCl (VWR International,
Radnor, Pennsylvania). The bench-scale testing system was equipped with an automatic dosing
system to maintain a constant draw solution concentration; water flux for each coupon was
calculated from the mass of draw solution that overflowed into a collection vessel on an
analytical balance (PA3102, OHAUS Corporation, Parsippany, NJ) with time. The standard
deviation between the coupons represents the error; overlapping standard deviation indicated a

11

statistically significant similarity and non-overlapping standard deviation indicated a statistically
significant difference.
The CTA and TFC membranes were initially tested with a NaCl feed solution that was
prepared to the same conductivity (10 mS/cm) as the activated sludge that would be tested.
Prior to testing, the membranes were equilibrated for one hour. The same initial flux (17 ± 1 L m
-
2
h
-1
) was achieved for both membranes by adjusting the individual draw solution concentration;
100 g/L NaCl was used for the CTA membrane and 64 g/L NaCl was used for the TFC
membrane. To compare flux decline from fouling, the feed solution was switched to activated
sludge that circulated from the pilot-scale bioreactor to the bench-scale module for 22 hours.
After that, the membranes were removed from the module and physically cleaned by rinsing the
visible fouling layer off with deionized water. In the final step, water flux recovery from physical
cleaning was tested using the NaCl feed solution from the initial step.
2.2.2. OMBR subsystem; longer-term testing
2.2.2.1. Bioreactor
A 250-L bioreactor was inoculated with a combination of return activated sludge and
biomass from a nitrifying trickling filter (Truckee Meadows Wastewater Reclamation Facility,
Reno, NV). Suspended solids in the bioreactor were continuously mixed with a submersible
pump and a solids retention time (SRT) of 30 days was maintained by wasting 8.4 L/d of the
mixed liquor through an automated solenoid valve (Automation Direct, Atlanta, GA). The
bioreactor was configured with an on-line recording system (IQ Sensor Net MIQ/TC 2020 XT)
connected to temperature, pH, dissolved oxygen (DO), chemical oxygen demand (COD), NH
4
+
-
N, NO
3
-
-N, and NO
2
-
-N probes (YSI, Yellow Springs, OH). The bioreactor was fed a synthetic
wastewater feed solution designed to simulate high-strength domestic wastewater; it was
prepared from sucrose and ammonium bicarbonate to achieve a strength of 1,350 mg/L COD

12

and 160 mg/L NH
4
+
-N. A float valve was used to feed the synthetic wastewater to the bioreactor
and maintain constant bioreactor volume. The OMBR was aerated by a diffuser positioned
beneath the submerged FO module. To achieve alternating aerobic/anoxic redox environments,
aeration was programmed for 30-minute cycles; the bioreactor was aerated for 30 minutes and
then not aerated for the next 30 minutes. Automated wasting, aeration cycling, and data logging
were executed in LabVIEW (National Instruments, Austin, TX).
2.2.2.2. Forward osmosis subsystem
CTA membranes were mounted in a plate-and-frame module that was submerged in the
bioreactor. Five double-sided membrane cassettes, each with 0.24 m
2
membrane area,
provided a combined membrane area of 1.2 m
2
. The membranes were mounted with the
selective layer facing the activated sludge feed solution. The draw solution was 20 g/L NaCl for
the first 140 days and the concentration was increased to 35 g/L NaCl thereafter. The draw
solution was circulated through each cassette at a flow rate of 100 mL/min using a peristaltic
pump (Cole-Parmer, Vernon Hills, IL). Pressure and conductivity sensors were installed on the
FO draw solution tank and the draw solution was continuously mixed with a magnetic stir bar
and stir plate (Cole-Parmer, Vernon Hills, IL). A brine tank containing concentrated NaCl
solution was connected to the FO draw solution tank and a solenoid valve (WIC Valve, San
Jose, CA) was installed between the tanks. The volume of each tank was calculated using
measured pressure values and standard curves with volume as a function of head pressure in
each tank. Similarly, the NaCl concentration in the draw solution tank was calculated using
measured conductivity values and a standard curve with NaCl concentration as a function of
conductivity. When conductivity fell below a set point, a dosing system was activated in
LabVIEW; LabVIEW triggered opening of the solenoid valve for 0.5 seconds to transfer
concentrated NaCl solution from the concentrate tank to the draw solution tank. After the draw
solution was mixed for 30 seconds, conductivity was checked again to determine if additional

13

dosing was required to reach the conductivity set point. FO water flux (L m
-2
h
-1
) and productivity
(L h
-1
) were calculated every 60 minutes by subtracting the change in concentrate volume from
the change in FO draw tank volume over time.
2.2.3. Integration of OMBR and MD subsystems
2.2.3.1. Membrane distillation subsystem
A custom-made crossflow DCMD module with stackable membrane plates containing six
membrane sheets (53.5 cm long × 23 cm wide), each with an area of 0.123 m
2
, for a total MD
membrane area of 0.74 m
2
was designed and constructed using results from a model
developed previously [59]. Six commercially available flat-sheet polytetrafluoroethylene (PTFE)
microfiltration membranes (Parker Performance Materials, Lee’s Summit, MO) were used. Tricot
mesh spacers (Hornwood Inc., Lilesville, NC) were installed on both sides of the membrane to
promote turbulence and mass transfer. Feed and distillate solutions were circulated in counter-
current flow using gear pumps (Standex International Corporation, Salem, NH) controlled by
variable frequency drives (Automation Direct, Atlanta, GA). In-line heat exchangers (Alfa,
Tampa, FL) were installed on the feed and distillate lines on the inlet side of the DCMD module.
On the feed side, electrical heaters were used to heat an ethylene glycol solution that was
pumped through the tube side of the heat exchanger. On the distillate side, a chiller (Cole-
Parmer, Vernon Hills, IL) was used to cool an ethylene glycol solution that was pumped through
the tube side of the heat exchanger. Temperature probes were installed at the feed and distillate
inlet and outlet ports on the MD module. The electrical heaters and temperature probe on the
feed side inlet were connected to a temperature controller (SOLO, Automation Direct, Atlanta,
GA) that was used to control the feed inlet temperature.
As a consequence of counter-current flow, heat exchange across the MD membrane causes
the outlet of the distillate solution to be warmer than the outlet of the feed solution. The higher

14

temperature of the distillate solution exiting the module provides opportunity for heat recovery;
the feed solution entering the module can be preheated with heat from the outlet of the distillate
stream. To accomplish this, a heat exchanger was installed between the two streams.
The feed and distillate solution tanks were equipped with pressure and electrical
conductivity sensors (Cole-Parmer, Vernon Hills, IL). Standard curves with volume as a function
of head pressure were created for each tank and the volume of each tank was calculated using
measured pressure values. MD water flux was calculated from the change in distillate tank
volume over time. A LabVIEW program was used to adjust pump speed, record temperature
and pressure at feed and distillate inlet and outlet ports, and record volumes and NaCl
concentrations of feed and distillate tanks. MD productivity (L h
-1
) was calculated every 60
minutes from the change in distillate volume over time.
2.2.3.2. Integrated OMBR-MD system
Coupling of the OMBR and MD subsystems was carried out to meet two main operating
objectives; 1) maintain continuous operation of FO and MD, and 2) ensure heat from the MD
process does not affect the temperature of the bioreactor. These objectives were addressed by
installing a secondary reservoir for storage of the concentrated MD feed solution (brine),
implementing a dosing system for the FO draw solution, and creating a transfer system between
the FO draw, MD feed, and brine tanks (Fig. 2.2). Utilizing the same dosing controls outlined in
Section 2.2.2, the FO draw solution was kept at a constant concentration by automatic dosing
with brine solution based on real-time conductivity measurements; this maintained continuous
FO operation. The brine solution was allowed to reach ambient temperature over time; this
prevented heat exchange between the MD process and the bioreactor. Periodically, a transfer
sequence was initiated that provided fresh feed solution to the MD feed tank; this maintained
continuous MD operation. When the transfer sequence began, concentrated MD feed solution
was transferred from the MD feed tank to the brine tank until a low volume set point in the MD

15

feed tank was reached. Thereafter, the accumulated volume of FO draw solution was
transferred to the MD feed tank until a low volume set point in the FO draw tank was reached.

Figure 2.2. The integrated OMBR-MD system with the (left) OMBR subsystem, (right) MD
subsystem, and (center) transfer pumps. The FO draw pump (P1) circulates draw solution
through the submerged FO module and the MD feed (P5) and distillate (P6) pumps circulate MD
feed and distillate solution in the MD module. A dual-channel pump (P4) facilitates heat transfer
through the main heat exchangers (HX) in the MD system. A third HX was used for heat
recovery. The transfer pumps (P2 and P3) are activated during the transfer sequence. For
simplicity, the MD module shown here contains only two membranes, however, six membranes
were used during the experiments.
Operation of the integrated system was fully automated using LabVIEW. Each tank (FO
draw, MD feed, and brine) had a 15-L capacity. A transfer sequence between the tanks was
initiated by the LabVIEW program when one of several user-defined set points was reached; the
transfer sequence was initiated by a high concentration set point (300 g/L NaCl) in the MD feed
tank, a low volume set point in the MD feed tank (4 L), or a high volume set point in the FO draw
tank (12 L). The high concentration set point prevented scaling in the MD module, the high

16

volume set point in the FO draw tank prevented overflow, and the low volume set point in the
MD feed prevented air entrapment and over heating. In this manner, FO driving force (salinity
gradient) and MD driving force (vapor pressure gradient) were maintained independently and
heat exchange with the bioreactor was prevented. When the integrated OMBR-MD system was
operated, the draw solution set point was 35 g/L NaCl, the MD feed inlet temperature was set to
80 °C, and the distillate inlet temperature was set to 10 °C.
2.2.4. Analytical methods
Samples (50 mL) were collected manually from the FO draw solution tank and MD distillate
tank every 12 hours during integrated OMBR-MD operation. NH
4
+
-N concentrations were
measured using Hach Salicylate method 10031 and COD was measured using Hach TNT 821
method with low-range vials (Hach Company, Loveland, CO). All samples were measured in
duplicate and the draw solution samples were diluted 20:1 to be within the NH
4
+
-N testing range
and to prevent chloride interference in the COD test (Hach Method 8000). NH
4
+
-N and COD in
the bioreactor were measured continuously with online sensors. NH
4
+
-N and COD rejection
(R
OMBR-MD
) for the integrated membrane system were calculated using:
(1)
where C
p
is the constituent concentration in the MD permeate and C
bio
is the concentration
within the bioreactor. As the MD permeate was collected in the distillate tank, C
p
was calculated
from the change in the constituent mass of over time using:
(2)
where C
dist,
is the distillate tank concentration, V
dist
is the distillate tank volume, and the
subscripts 1 and 2 correspond to time points 12 hours apart.
R
OMBR−MD
= 1−
C
p
C
bio
⎛
⎝
⎜
⎞
⎠
⎟
∗100%
C
p
=
C
dist,2
V
dist,2
−C
dist,1
V
dist,1
V
2
−V
1

17

2.3. Results and discussion
2.3.1. Membrane selection; CTA and TFC comparison
The CTA and TFC membranes were tested at bench-scale with an initial flux of 17 ± 1 L m
-2

h
-1
. This high initial flux was chosen because greater flux decline from membrane fouling [60]
ensures a measureable flux decline could be observed for the CTA and TFC membranes in a
relatively short time period. As can be seen in Fig. 2.3, flux decline from fouling was similar over
time for both membranes after exposure to activated sludge. Water flux after 22 hours was 3.3 ±
0.2 and 3.2 ± 0.2 L m
-2
h
-1
for the CTA and TFC membranes. This result indicates there was no
statistically significant difference in water flux between the two membranes after 22 hours of
exposure to activated sludge. Average flux after physical cleaning was 15.1 ± 0.4 and 15.1 ± 0.2
L m
-2
h
-1
for the CTA and TFC membranes, and flux recovery (100%*J
w,final
/J
w,initial
) after physical
cleaning was 91 and 89% for CTA and TFC membranes. From our results, there was no
significant difference in fouling or flux recovery of the CTA and TFC membranes for the
activated sludge feed solution.

Figure 2.3. Water flux for CTA and TFC membranes treating activated sludge from the pilot-
scale bioreactor. The bench-scale crossflow module was operated with a crossflow velocity of
20 cm/s, no spacers were used on the feed side, and diamond-style mesh spacers were used
on the draw side. Initial water flux with NaCl feed solution was set to 17 ± 1 L m
-2
h
-1
for both
membranes by adjusting draw solution concentrations; the draw solution was 100 and 64 g/L
NaCl for the CTA and TFC membranes, respectively. Error bars represent standard deviation of
triplicate membrane coupons in each experiment.

18

New-generation TFC membranes offer superior selectivity compared to CTA membranes,
however, a very high water permeability (A) that results in greater water flux also results in
greater internal concentration polarization, thus greater water flux through increased A becomes
self-limiting for the FO process [52, 61]. A recent study that stresses the importance of improved
solute permeability (B) over further improvements in A concludes that little benefit can be gained
by developing membranes with A values greater than 2 to 4 L m
-2
h
-1
bar
-1
, which are typical
values for high-performance TFC membranes [61]. Furthermore, when an FO membrane is
exposed to a fouling feed solution (e.g., activated sludge or wastewater) a foulant cake layer
likely forms on the membrane surface and A becomes less important because the additional
hydraulic resistance of the cake layer causes significant water flux decline for both TFC and
CTA membranes alike [37, 62, 63]. Higher water flux may also be achieved by optimizing
membrane structural parameter (S), which is an indicator of propensity for internal concentration
polarization [64, 65]; however, the coupled effects of internal concentration polarization,
membrane fouling, and RSF require further investigation [31, 58].
For long-term OMBR operation, low RSF is critical for maintaining FO driving force as well
as biological nitrogen removal as nitrifying bacteria have been shown to lose efficacy at NaCl
concentrations above 2% [25, 26, 30]. In a long-term OMBR study, CTA membranes exhibited
lower RSF and higher steady-state water flux compared to TFC membranes [66]. Another long-
term study showed that TFC membranes were more prone to biodegradation when exposed to
activated sludge [67]. Others have reported that fouling of TFC membranes is more severe and
less reversible than that of CTA membranes [33, 62]. These studies, and our experimental
results, underscore the importance of evaluating membrane performance in the presence of
feed solutions with high fouling potential and the relative unimportance of high A values for
OMBR applications.

19

2.3.2. Long-term operation of the OMBR subsystem
2.3.2.1. Biological treatment with aerobic/anoxic cycling bioreactor
COD and nitrogen species concentrations during a representative four-hour period of
aerobic/anoxic cycling are shown in Fig. 2.4. Aeration was on during the first 30 minutes of each
hour and off for the remaining 30 minutes. After aeration was turned on, an immediate decrease
in COD concentration was observed, indicating aerobic biological degradation of the carbon
components of wastewater. The COD decrease was lagged by an increase in dissolved oxygen
concentration as the oxygen demand of the bioreactor was met, and a subsequent decrease in
NH
4
+
-N concentration, indicating that biological nitrification was occurring. When aeration was
switched off, the environment quickly became anoxic as the remaining oxygen was consumed
by the aerobic species, with a corresponding increase in COD concentration followed by a
lagged increased in NH
4
+
-N concentration (Fig. 2.4a). Low concentrations of NO
3
-
-N (< 10 mg/L)
and lagged decreases of NO
3
-
-N concentrations during anoxic periods indicate that
denitrification was also occurring in the bioreactor (Fig. 2.4b). The trends in dissolved oxygen,
COD, and nitrogen species concentrations repeated every hour.

20

Figure 2.4. Representative 4-hour period of bioreactor operation showing effect of 30-minute
aerobic/anoxic cycling on (a) NH
4
+
-N and COD and (b) NO
3
-
-N and NO
2
-
-N concentrations.
Aeration was on during the first 30 minutes of each hour and off for the remaining 30 minutes.
Influent wastewater COD and NH
4
+
-N concentrations were 1,350 and 160 mg/L.

Long-term trends resulting from the aeration cycles are presented in Fig. 2.5. COD was
significantly reduced from the feed concentration of 1,350 to ~200 mg/L (Fig. 2.5a),
corresponding to ~85% biological removal. Initially, the NH
4
+
-N concentration in the bioreactor
(~250 mg/L) was greater than that in the influent wastewater (160 mg/L), indicating that
nitrification was not occurring. It was hypothesized that elevated NH
4
+
-N levels were due to a
low population of nitrifying bacteria in the initial inoculum, which was collected from an aeration
basin, and also due to biomass decay reactions as the microbial community adjusted to the
cycling aerobic/anoxic environment. In an effort to stimulate nitrification, the bioreactor was
inoculated with a microbial seed of nitrifying bacteria grown in a trickling nitrification filter on day
20. After that, the NH
4
+
-N concentration began to decrease and the final concentration after 53
days was 70 mg/L (Fig. 2.5a). Despite the high NH
4
+
-N concentrations during the first 20 days of
operation, NO
3
-
-N and NO
2
-
-N concentrations remained below 12 mg/L (Fig. 2.5b), indicating
that denitrification was occurring during this period.

21

Figure 2.5. Long-term trends in (a) NH
4
+
-N and COD and (b) NO
3
-
-N and NO
2
-
-N concentrations
for bioreactor with alternating aerobic/anoxic environments. Influent wastewater COD and NH
4
+
-
N concentrations were 1,350 and 160 mg/L, respectively. Biological COD and NH
4
+
-N removals
were 85 and 56% removal, respectively.

2.3.2.2. Forward osmosis; Water flux behavior
For the pilot-scale OMBR, the CTA membranes with 20 g/L NaCl draw solution had an initial
flux of 4.3 L m
-2
h
-1
(Fig. 2.6) that decreased to 1.6 L m
-2
h
-1
over the first 30 days. During that
time, bioreactor conductivity increased from 5.9 to 10.6 mS/cm. Flux decline was likely due to
membrane fouling and in part, increased bioreactor salinity from RSF that decreased the
osmotic pressure driving force. Between days 30 and 140, the average water flux and bioreactor
conductivity values were 1.8 ± 0.3 L m
-2
h
-1
and 8.9 ± 0.8 mS/cm. On day 140, in an effort to
increase steady-state water flux, the membranes were physically cleaned by rinsing with water
and the draw solution concentration was increased to 35 g/L NaCl. Water flux increased to 4.5 L
m
-2
h
-1
, but then decreased to 1.6 L m
-2
h
-1
by day 182. Between days 140 and 182, bioreactor
conductivity increased from 7.6 to 15.5 mS/cm; again, flux decline was likely due to membrane
fouling and in part, increased bioreactor salinity from RSF. The average water flux and
bioreactor conductivity between days 182 and 200 were 1.5 ± 0.1 L m
-2
h
-1
and 13.9 ± 0.9

22

mS/cm. While both draw solution concentrations resulted in a similar steady-state water flux, the
35 g/L draw solution increased average bioreactor conductivity by 58% (Fig. 2.6). Thus, the
higher draw solution concentration resulted in increased RSF without a corresponding increase
in steady-state water flux.

Figure 2.6. The pilot-scale OMBR was seeded with activated sludge on day 0 and operated
continuously for 200 days. The draw solution was 20 g/L NaCl between days 0 and 140 and 35
g/L NaCl after day 140.

Similarly, in our previous bench-scale work, two draw solution concentrations (35 and 100
g/L NaCl) resulted in the same steady-state water flux after membrane fouling [58]. The greater
flux decline for the higher draw solution concentration was attributed to greater foulant cake
layer resistance and greater cake-enhanced osmotic pressure due to RSF [42, 68]. Those
bench-scale results, as well as the pilot-scale OMBR results presented here, suggest the
existence of a “limiting flux” for fouled FO membranes, or a flux that cannot be overcome by
increasing osmotic pressure driving force because greater draw solution concentration also
results in greater RSF. These observations are in agreement with other studies that have

23

reported greater RSF with fouled membranes compared to clean membranes [38, 66]. Zou et al.
[69] observed “critical flux” behavior where increases in the FO draw solution concentration
beyond a certain value did not result in stable water flux and Luo et al. [38] showed that
increasing draw solution concentrations (from 1 to 2 M NaCl) resulted in higher RSF for fouled
membranes while water flux tended toward a singular value.
2.3.3. Integrated OMBR-MD system
2.3.3.1. Productivity of the integrated system
Prior to beginning the integrated experiment, FO membrane area was reduced from 1.2 to
0.24 m
2
so as to match OMBR water productivity (L h
-1
) with MD water productivity. Although
initial FO and MD productivities were similar, MD productivity had an early decrease from 0.97
to 0.49 L h
-1
(Fig. 2.7a), possibly due to MD membrane fouling. The average, steady-state MD
water flux between hours 15 and 48 of 0.8 L m
-2
h
-1
(calculated from productivity data in Fig.
2.7a) was lower than expected. It has been reported that larger scale MD systems tend to have
lower fluxes than smaller scale systems [70, 71]; this could be due in part to more rapid heat
utilization leading to reduced localized driving forces. FO productivity decreased slightly over 48
hours and varied between 1.05 and 1.37 L h
-1
(Fig. 2.7a). The slight declining trend in FO
productivity is the result of membrane fouling and to some extent, reduction in osmotic pressure
driving force due to an increase in bioreactor salinity (from 3.50 to 3.76 g/L NaCl) from RSF.
From the productivity data in Fig. 2.7a, an average FO water flux of 5.1 L m
-2
h
-1
was calculated
for the integrated operation. This average FO water flux value agrees with values from the
literature (ranging from 0.5 to 6.0 L m
-2
h
-1
) for OMBRs treating high fouling feed waters such as

24

activated sludge [25, 26, 35, 37, 39, 40].

Figure 2.7. (a) Productivity and (b) temperatures of the integrated OMBR-MD system during
continuous operation. The FO draw solution was 35 g/L NaCl and the MD feed and distillate
inlet temperatures were set to 80 and 10 °C.

The MD feed and distillate outlet temperatures were 17.3 ± 0.8 and 60.0 ± 4.1 °C (Fig. 2.7b),
indicating that a considerable amount of heat was recovered from the distillate solution by HX3.
Moreover, heat from the MD process did not affect the bioreactor temperature. A constant
bioreactor temperature of 30.8 ± 0.4 °C was maintained during the experiment (Fig. 2.7b),
indicating that the automated dosing and transfer system was effective in preventing transfer of
heat to the bioreactor.
2.3.3.2. Removal by the integrated system
Average NH
4
+
-N and COD concentrations in the bioreactor and MD distillate solution were
calculated from measurements taken every 12 hours over a 48-hr period of integrated OMBR-
MD operation and system removal was calculated using (1). NH
4
+
-N concentration in the
distillate solution was 24.6 ± 6.2 mg/L, corresponding to 84.6% removal from the synthetic
wastewater feed solution (160 mg/L NH
4
+
-N). It should be noted that the NH
4
+
-N concentrations
in the bioreactor (229 to 275 mg/L) were higher than in the feed solution during this 48-hr

25

period, which is indicative of slower acclimation of the ammonia-oxidizing bacteria than the
other bacteria in the freshly seeded inoculum, and that a lower steady-state NH
4
+
-N
concentration in the bioreactor is expected over time (see trend in Fig. 2.5a). Moreover, system
removal is expected to improve as bioreactor concentration decreases. If the elevated
bioreactor NH
4
+
-N concentration (average of 252 ± 20 mg/L) is used in place of the feed solution
concentration (C
f
) in (1), system removal becomes 90.2%. Previous FO-MD studies have
reported 99% and greater NH
4
+
-N removal [50, 72, 73]; in the current study, the NH
4
+
-N
concentrations in the distillate are slightly higher than would be expected. This is likely due to
higher NH
4
+
-N concentrations in the high-strength synthetic wastewater feed solution and
biological treatment with aerobic/anoxic cycling, where low dissolved oxygen concentrations can
lead to higher NH
4
+
-N concentrations compared to continuously aerobic bioreactors that do not
incorporate a denitrification process [74]. Furthermore, due to lower rejection of volatile
contaminants by MD membranes [75], the high MD feed temperature (80 °C) used in the current
study may have caused a disproportionate amount of volatile NH
3
to selectively pass through
the MD membranes [76, 77]. The average COD concentration in the distillate solution was 21.6
± 2.8 mg/L, corresponding to 98.4% removal from the OMBR feed solution (1,350 mg/L COD).
High carbon removal for FO-followed-by-MD systems has also been demonstrated elsewhere;
previous studies have shown 98.5% COD removal [50] and greater than 99% TOC removal [51,
73].
2.4. Conclusions
An integrated OMBR-MD pilot-scale system was tested for wastewater treatment and
production of reuse water. During membrane selection, no significant difference in water flux
between CTA and TFC membranes was observed for the activated sludge feed solution. After
long-term operation, FO water flux with CTA membranes was the same for 20 g/L and 35 g/L
NaCl draw solution, although the 35 g/L draw solution increased bioreactor conductivity due to

26

greater RSF. During long-term OMBR operation, carbon and nitrogen removal was achieved in
a single reactor by alternating between aerobic and anoxic bioreactor conditions. The OMBR
was coupled to DCMD with an automated dosing and transfer system. This allowed for
continuous water production by FO and MD without transfer of heat to the bioreactor, which is
critical for biological nitrogen removal. Results from the current study also highlight an added
benefit of the automated dosing and transfer system compared to directly-coupled FO-MD
configurations in the literature; in the event of an FO membrane failure, physical separation of
the FO draw solution and MD feed solution allows an operator or system response to prevent
MD membrane fouling and maintain high-quality product water. When treating a high-strength
wastewater feed solution, the integrated OMBR-MD system achieved 90.2% NH
4
+
-N removal
and 98.4% COD removal. Future efforts should focus on optimization of overall system
performance, specifically with regards to increasing FO and MD water fluxes while achieving
high water quality.

27

Chapter 3. Submerged or sidestream? The influence of module configuration on
fouling and salinity in osmotic membrane bioreactors

This chapter presents published work from the Journal of Membrane Science [58] as well as
unpublished work in Sections 3.3.4 to 3.3.5.
3.0. Abstract
The role of submerged and sidestream forward osmosis (FO) membrane module
configuration in osmotic membrane bioreactors (OMBRs) was investigated. Experiments were
performed under identical (solids retention time, bioreactor volume, feed solution, draw solute,
and draw solution concentration) conditions to isolate the effect of FO module configuration and
associated hydrodynamics on water flux, reverse salt flux, and membrane fouling. Steady-state
water flux of fouled membranes was the same for submerged and sidestream configurations
and two draw solution concentrations, leading to the concept of a limiting flux in OMBRs similar
to the critical flux in conventional membrane bioreactors. Despite a significant increase in driving
force, fouled membranes did not have higher steady-state water flux; instead, the higher draw
solution concentration resulted in higher specific reverse salt flux (SRSF) and increased fouling.
For the 35 g/L NaCl draw solution, SRSF was 1.61 ± 0.01 and 0.59 ± 0.07 g L
-1
for submerged
and sidestream configurations, respectively and for the 100 g/L NaCl draw solution, SRSF was
2.22 ± 0.25 and 1.05 ± 0.35 g L
-1
for submerged and sidestream configurations, respectively.
With 100 g/L draw solution, foulant cake layers were 2-4 times thicker, likely due to higher initial
water flux that resulted in more foulants being transported to the membrane surface.
Experimental results were used as model inputs to predict results for a larger scale system.
Model results predicted lower steady-state bioreactor salinities in the sidestream configuration,
particularly when longer solids retention times were used.

28

Graphical abstract
3.1. Introduction
Membrane bioreactor (MBR) systems are frequently being considered as upgrade and/or
retrofit options to conventional wastewater treatment systems [3]. Compared to conventional
activated sludge treatment, MBRs with submerged microfiltration (MF) or ultrafiltration (UF)
membranes offer a smaller footprint and higher quality product water for non-potable reuse
applications [3, 78]. Further treatment of MBR effluent (e.g., by reverse osmosis (RO) and
advanced oxidation) is required for potable reuse applications [4, 79]. However, membrane
fouling of both the MF/UF and RO membranes is a key concern for MBR processes [4, 80, 81].
Osmotic membrane bioreactor (OMBR) systems provide a low-fouling alternative to
conventional MBR systems for potable reuse applications [5-7]. In the OMBR process, a forward
osmosis (FO) membrane is submerged in, or placed sidestream to, the bioreactor. Water is
extracted from the activated sludge into a concentrated draw solution by an osmotic pressure
driving force [8]. Consequently, OMBRs require a secondary desalination step to reconcentrate
the draw solution and generate product water. Compared to conventional MBRs, OMBRs offer
higher rejection of low molecular weight organics, ions, and pharmaceuticals, as well as organic

29

nutrients [14-16]. Moreover, MF and UF membranes in MBRs pass dissolved solids while FO
membranes in OMBRs generally do not. However, a key concern with OMBRs is elevated
bioreactor salinity caused by retention of dissolved solutes from the influent wastewater and
diffusion of draw solutes from the FO draw solution into the bioreactor by a phenomenon known
as reverse salt flux (RSF) [21-23]. RSF is a critical concern of osmotic processes because it
diminishes the salinity gradient, thus reducing the osmotic driving force for water flux; in the
case of OMBRs RSF also increases bioreactor salinity [24].
High bioreactor salinity can affect microbial processes, for example decreasing the activity
of nitrogen-removing bacteria [25, 26]. Ammonia-oxidizing bacteria, which are responsible for
ammonia removal via nitrification, are particularly sensitive to salinity and sharp decreases in
nitrification efficiency have been reported for bioreactors with salinities ranging from 2-15 g/L
NaCl [27-30]. Notably, Holloway et al. [40] observed a period of nitrification inhibition followed by
recovery of nitrification that was attributed to growth of halotolerant microorganisms.
Nevertheless, maintaining low salinity is desirable because no shift in the microbial community
towards halotolerant microorganisms is required and higher bioreactor salinity may also lead to
scaling and increased membrane fouling.
Dissolved solutes within the foulant cake layer on the membrane surface have also been
shown to decrease water flux. This was first recognized in RO and nanofiltration (NF)
processes, where the applied hydraulic pressure must be greater than the osmotic pressure of
the feed solution for water to pass through the membrane. For fouled membranes, the cake
layer hinders back diffusion of dissolved solutes, causing the osmotic pressure to increase near
the membrane surface, resulting in decreased water flux by a phenomenon termed “cake
enhanced concentration polarization” (CECP) [82, 83]. For RO and NF, the osmotic pressure
due to CECP was systematically shown to be a greater hindrance to water flux than the
hydraulic resistance of the foulant cake layer itself [83]. Similarly, for FO, flux decline for fouled

30

membranes results not only from hydraulic resistance of the cake layer but also from
accumulation of draw solutes transported by RSF into the cake layer; this reduces the osmotic
pressure driving force by a phenomenon known as “cake enhanced osmotic pressure” (CEOP)
[42, 68]. Although fouling is less severe and more reversible in the FO process than in the RO
process [41-43], the accumulation of draw solutes in the cake layer from RSF in FO membrane
fouling results in a greater loss of driving force in FO than CECP does in RO [17, 84].
The hydrodynamics and mixing that influence RSF and membrane fouling in OMBRs is
highly influenced by FO module configuration [7, 31]. In the submerged configuration, the draw
solution is circulated through the interior of the membrane module while the feed solution has no
direct crossflow [7]. Air bubbles can be used as a fouling control strategy with air scour of the
membrane surface reducing membrane fouling [23, 32]. In the sidestream configuration, the
membrane module is external to the bioreactor and the feed and draw solutions flow tangentially
across the membrane surfaces. High crossflow velocities can be used as a fouling control
strategy with hydraulic scour of the membrane surface reducing membrane fouling [33]. Large
ranges of steady-state salinities have been reported for both submerged (4.1-12.6 g/L NaCl)
and sidestream (4.0-33.5 g/L NaCl) OMBR configurations [5, 23, 25, 26, 34-40]. In these long-
term studies, bioreactor salinity is not solely a function of the FO module configuration but is
also affected by the solids retention time (SRT), bioreactor volume, fouling behavior of the feed
solution, type of draw solute, and draw solution concentration.
The objective of the current work is to experimentally evaluate the effect of FO module
configuration on water flux, membrane fouling, RSF, and bioreactor salinity in OMBR systems.
The submerged and sidestream experiments were performed under identical (SRT, bioreactor
volume, feed solution, draw solute, and draw solution concentration) conditions to isolate the
effect of FO module configuration and associated hydrodynamics on RSF, bioreactor salinity,
and membrane fouling. A unique testing protocol was developed to evaluate water flux and RSF

31

through a fouled membrane in a continuous system; the physical and chemical properties of the
foulant cake layers were subsequently examined. All experiments were conducted using two
NaCl draw solution concentrations. Additionally, an iterative mass balance approach was
applied using experimentally derived water flux and RSF values to predict the effect of
configuration on scaled-up bioreactors operating with short and long SRTs. This study is the first
to provide a direct comparison of submerged and sidestream configurations and to
systematically evaluate short-term, steady-state water flux, RSF, and cake layer characteristics,
as these parameters will influence long-term bioreactor salinity and membrane fouling.
3.2. Materials and methods
3.2.1. Forward osmosis sub-systems
Novel submerged and sidestream membrane modules were designed to hold three
identically sized membrane coupons each. Each membrane coupon had an active membrane
area of 42 cm
2
and a dedicated draw solution dosing system. The draw solution was 35 or 100
g/L analytical-grade NaCl (VWR International, Radnor, Pennsylvania). The draw solution
channels for both submerged and sidestream modules were 11.2 cm long × 3.8 cm wide × 0.1
cm deep. All experiments were conducted with flat-sheet cellulose triacetate (CTA) membrane
coupons (Hydration Technology Innovations, LLC, Albany, OR) with the active layer facing the
feed solution. All three membrane coupons were exposed to the same biological environment,
which enabled replication typically not possible in MBR studies using real wastewaters with
temporal variations. For the sidestream module, the feed solution channel was identical to the
draw solution channel; for the submerged module, the feed side was exposed to the bioreactor.
Peristaltic pumps (Cole-Parmer, Vernon Hills, IL) were used to recirculate the draw solution
(for all configurations) and feed solution (for sidestream configuration) at a crossflow velocity of
20 cm/s. Spacers (0.787 mm) (Sterlitech Corporation, Kent, WA) were used in the draw solution

32

channels to provide structural support and enhance draw solution mixing; no spacers were used
in the feed solution channels of the sidestream module. The draw solution pumps were
positioned downstream of the membrane modules to provide negative pressure in the draw
solution channels. For the sidestream configuration, the feed solution pump was positioned
upstream of the module to provide positive pressure in the feed solution channel and to prevent
membrane deformation. The temperature of the feed and draw solutions was held at 24 ± 1 °C
by conducting experiments in a temperature controlled environment.

Figure 3.1. (a) Submerged and (b) sidestream FO testing configurations. Draw solution was
kept at constant concentration by dosing with 292 g/L (5M) NaCl solution. Temperature was 24
± 1 °C. Each module contained three membrane coupons, each with a dedicated draw solution
dosing system; only one dosing system is shown here for simplicity. The control configuration
was the same as the submerged configuration, however, no aeration (air scour) was used for
fouling control.

33

A 50-L aerobic bioreactor was filled with activated sludge and a wastewater feed tank was
filled with primary clarifier effluent collected from Hyperion Wastewater Treatment Facility (Los
Angeles, CA). A float valve was use to feed primary effluent to the bioreactor at the same rate
that water was extracted from the bioreactor by the FO process. The bioreactor was continuously
mixed with a stir bar and aerated with a diffuser.
The draw solution was recirculated from the draw solution reservoir (a 1-L sidearm flask) to
the membrane module and electrical conductivity and temperature were monitored in the draw
solution reservoir using a conductivity probe and transmitter (Cole-Parmer, Vernon Hills, IL).
Over time, FO water flux caused the draw solution reservoir to overflow; the mass of overflow
was monitored with an electronic balance (PA3102, OHAUS Corporation, Parsippany, NJ) and
used to calculate FO water flux. Conductivity was checked every 56 minutes. If conductivity
dropped below a set value, the dosing system was activated and a dosing pump transferred 292
g/L (5 M) NaCl to the draw solution reservoir until conductivity returned to the set point. The
conductivity set point was always reached within four minutes and draw solution conductivity
varied by less than ± 2% over the course of the experiments. The dosing period was excluded
from water flux calculations. Mass and conductivity data were recorded using data acquisition
devices (USB-6009, NI 9208) connected to a LabVIEW program (National Instruments, Austin
TX).
FO fouling can be reduced by air scour that creates shear stress at the membrane surface.
For flat-sheet membranes in the submerged configuration, this can be achieved using a baffle
and a bubble diameter equal to the spacing between the baffle and membrane surface [7, 85].
In the current study, the baffle was placed over the submerged module with a channel spacing
of 4.75 mm and the aeration rate through an air diffuser was adjusted to generate an average
bubble diameter approximately equal to the channel spacing. For the sidestream configuration,
fouling can be reduced by high crossflow velocity (from 17-30 cm/s) that creates shear stress

34

along the membrane surface [33, 42, 43, 66]. In the current study, a crossflow velocity of 20
cm/s was used. As a control, membrane coupons were tested without hydraulic crossflow or air
scour; the control membranes were mounted in the submerged module without the baffle.
3.2.2. Forward osmosis biofouling experimental protocol
Biofouling experiments were carried out in four stages. Stage one quantified water flux and
RSF of virgin membranes in the absence of fouling. Stage two quantified steady-state water flux
under fouling conditions. This required operation of the bioreactor in a continuous mode with
sufficient volume (50 L) such that RSF would not change bioreactor salinity significantly over the
course of the 20-hr fouling experiment. Stage three assessed steady-state RSF through fouled
membranes. This required operation of the bioreactor in batch mode with a smaller volume so
that changes in electrical conductivity could be quantified. After stage three, the membrane
coupons were removed from the module and the foulant cake was removed from the membrane
coupons by rinsing with deionized water. During stage four, the rinsed membranes were re-
tested to evaluate recovery of initial water flux and to determine if there was a change in water
flux and RSF after fouling and rinsing. Stages one through four were performed using both 35
and 100 g/L NaCl draw solutions. Operating details for all configurations in each stage are
described below.
During stage one, NaCl feed solution was prepared to the same conductivity as the
activated sludge solution (1.7 mS/cm) that would be used in stage two. This ensured that flux
decline in stage two could be attributed to fouling and not to a change in feed solution osmotic
pressure. In the submerged configuration, the membrane module was submerged in a 7-L
reactor. In the sidestream configuration, the three membrane coupons were tested with
individual 1.5-L feed solution volumes. Water flux and RSF were measured for one hour. Water
flux was calculated from the mass of draw solution overflow per membrane area per time; RSF
was calculated from [21, 86]:

35

RSF=
C
F
V
F
−C
i
V
i
A
M
Δt
(1)
where V
i
and V
F
are initial and final feed volumes; C
i
and C
F
are initial and final NaCl feed
concentrations; A
M
is membrane area; and Δt is elapsed time.
During stage two, the feed solution was switched to activated sludge and the membranes
were fouled for 20 hrs. During this stage, water flux reached steady state in all experiments.
Stage two was operated in continuous mode by circulating the activated sludge from the 50-L
bioreactor to the FO feed tank (Fig 1). Salinity accumulation in the bioreactor may change
fouling behavior and in general, higher salinity leads to greater fouling [87]. Steps were taken
during the fouling experiments to avoid salinity accumulation. The large bioreactor volume in
continuous mode enabled water flux measurements and resulted in insignificant conductivity
changes; this ensured the membranes were exposed to essentially the same salinity and hence,
same biofouling potential, in each experiment. Differences in water flux between each
configuration (control, submerged, sidestream) were directly attributable to hydrodynamics and
not to differences in fouling behavior from changes in salinity.
In stage three, the bioreactor circulation pumps (Fig. 3.1, dark gray pumps) were turned off
so that there were smaller feed volumes and quantifiable changes in conductivity. The feed
solution volume for each configuration was the same as in stage one. After stage three, the
membrane coupons were removed from the modules and rinsed with deionized water until all
visible membrane fouling was completely removed. After rinsing, the membranes were stored in
deionized water for 30 minutes. In stage four, the rinsed membranes were retested using the
same NaCl feed solution and conditions from stage one. Water flux and RSF were recorded for
one hour and the specific reverse salt flux (SRSF) was calculated for stages one, three, and
four using [21]:

36

SRSF=
J
S
J
W
(2)
where J
S
is reverse salt flux and J
W
is water flux.
3.2.3. Activated sludge characterization
Conductivity of the activated sludge was monitored using a bench-top meter (EC215, Hanna
Instruments, Woonsocket, RI) and dissolved oxygen, temperature, and pH were measured with
a handheld probe (SevenGo Duo Pro, Mettler Toledo Inc., Hightstown, NJ). Mixed liquor
suspended solids (MLSS) and mixed liquor volatile suspended solids (MLVSS) were analyzed
using Standard Method 2540D and 2540E, respectively [88]. The MLSS was 7.32 ± 0.08 g/L
and the MLVSS was 3.35 ± 0.04 g/L, with a MLVSS/MLSS ratio of 0.46. The bioreactor was
continuously aerated to maintain a dissolved oxygen concentration greater than 3.0 mg/L and
the pH was 7.4 ± 0.2 over the course of the experiments. These values and an MLVSS/MLSS
ratio between 0.4 and 0.6 are characteristic of OMBR systems with long SRTs for treatment of
domestic wastewater [5, 25, 26, 34, 35, 38, 40].
3.2.4. Fouling layer characterization; scanning electron microscopy and energy
dispersive X-ray spectroscopy
Fouled membrane coupons were characterized with scanning electron microscopy (SEM)
and energy dispersive X-ray spectroscopy (EDS). After stage three, one fouled membrane
coupon from each experiment was removed and stored at -20 °C until analysis. Prior to
analysis, the coupons were fixed with glutaraldehyde at 4 °C overnight, dried with a graded
ethanol series, and stored in a desiccator [89]. Cross-section images were obtained by freeze
fracturing the samples under liquid nitrogen. All samples were sputter-coated with a
palladium/gold mixture before SEM/EDS analysis (JEOL JSM-7001, JEOL Ltd., Tokyo, Japan).

37

Cake layer thickness was measured on each cross-section image using ImageJ software
(version 1.49, National Institutes of Health, Bethesda, MD).
3.2.5. Modeling the steady-state salt concentration of a continuous OMBR
Bioreactor salinity (C
R
) for a pilot-scale OMBR was modeled using the following equation
adapted from [40]:

C
R
= Q
in
C
in
+Q
S
−Q
W
C
R
i−1 ( )
⎡
⎣
⎤
⎦
t− t
i−1 ( )
V
R
+C
R
i−1 ( )
(3)
where Q
W
is wasting rate in L/d; Q
in
is FO productivity rate (Q
FO
) plus Q
W
in L/d; C
in
is influent
concentration in g/L; V
R
is reactor volume in L, Q
S
is reverse salt flux rate in g/d, and t is time
step (1 day). The measured bioreactor salinity at time 0 (0.9 g/L NaCl) was set as the initial
condition. For the next time step, bioreactor salinity (C
R
) at time t was calculated using the
experimental water flux and RSF at steady state with fouled membranes and the bioreactor
concentration from the previous time step (C
R(i-1)
). The concentration of dissolved solids in the
influent wastewater (C
in
) was set to 1,000 mg/L and discrete 24-hr time steps were used. The
iteration was repeated for a total duration of 120 days.
Due to the high rejection of FO membranes, the SRT of OMBRs is decoupled from the
hydraulic retention time (HRT) [5, 90]. For COD removal and nitrification, a 10-day SRT is
recommended and for reuse applications where nutrient removal requires a denitrification step,
longer SRTs are required [7]. Two SRTs (10 and 40 days) were chosen to represent different
reuse scenarios. The 10-day SRT represents COD removal with nitrification and the 40-day
SRT represents COD and nitrogen removal via nitrification and denitrification. Lay et al. [90]
suggest HRTs less than one day to minimize capital cost. The HRT was held at less than one
day by modeling a pilot-scale reactor with a 1,000-L volume (V
R
) and adjusting the membrane
area to achieve an FO productivity rate (Q
FO
) of 100 L/d.

38

3.3. Results and discussion
3.3.1. Water flux
Water flux results are shown for both draw solution concentrations in Fig. 3.2. Data are
shown for submerged and sidestream configurations as well as the control configuration
represented by a submerged membrane without fouling mitigation (no air scour or crossflow).
Water flux in all three configurations (control, submerged, sidestream) was measured using
triplicate membrane coupons. The standard deviation between the membrane coupons is
shown by the error bars in Fig. 3.2; overlapping standard deviation indicates statistically similar
water flux values; non-overlapping standard deviation indicates statistically different water flux
values. The first data points in both figures are from stage one and represent initial water flux of
the virgin membrane obtained using an NaCl feed solution prepared to the same conductivity as
the activated sludge (1.7 mS/cm). For the 35 g/L NaCl draw solution (Fig. 3.2a), all
configurations had the same initial water flux of approximately 9 LMH. As expected, the 100 g/L
NaCl draw solution resulted in higher initial water flux (~19 LMH for all configurations; Fig. 3.2b)
compared to the 35 g/L NaCl draw solution. Having the same initial water flux ensured that the
rate that foulants were brought to the membrane surface was the same for all configurations.

39

Figure 3.2. Experimental water flux for all configurations (control, submerged, sidestream) for
(a) 35 and (b) 100 g/L NaCl draw solution concentrations. Error bars represent standard
deviation for three replicate membranes in stages 1-3 and duplicate membranes in stage 4. The
boxed regions indicate the data points used to calculate average steady-state water flux for
fouled membranes.

In stage two, the membranes were fouled with an activated sludge feed solution. All
configurations (control, submerged, sidestream) reached steady-state with both draw solution
concentrations. Steady-state was defined as occurring when the water flux of the last four data
points of stage two (a total duration of 3.8 hours) varied less than 2% between each data point
(with 56-minute intervals) and less than 2% between the first and last data point in that period.
After steady-state was reached, an average steady-state water flux was calculated (Fig. 3.2,
boxed regions). For the 35 g/L NaCl draw solution, flux decline for the control configuration was
81%. Over the same period, flux decline was only 21% for the submerged configuration and
11% for the sidestream configuration. Interestingly, water flux with time for the submerged and
sidestream configurations was statistically similar. This was not expected because experimental

40

conditions were not adjusted to achieve similar results. When the draw solution was 100 g/L
NaCl, significant flux decline occurred for all configurations (Fig. 3.2b). This was likely due to the
higher initial water flux that resulted in more foulants being transported to the membrane
surface. Flux declined 51% for submerged and sidestream configurations and 91% for the
control configuration. Again, the statistically similar water flux with time in the submerged and
sidestream configurations was not expected. Fortuitously, the statistically similar water flux for
these configurations enabled direct comparison of cake layers and RSF.
The steady-state water flux values for the submerged and sidestream configurations were
also statistically similar for both draw solution concentrations (i.e., results for 35 and 100 g/L
NaCl draw solution were within standard deviation); despite the higher initial water flux with the
100 g/L NaCl draw solution, when the membranes were fouled, they had the same steady-state
water flux as that of the fouled membranes from the 35 g/L NaCl experiments. Similar results
were observed for the control configuration; the steady-state water flux results for both draw
solution concentrations were statistically similar. In other words, a three-fold increase in draw
solution concentration resulted in no increase in steady-state water flux under fouling conditions.
As has been seen in other FO fouling studies [38, 62, 91], this indicates that membrane fouling
may play a much more significant role in steady-state water flux than draw solution osmotic
pressure for OMBRs.
Luo et al. [38] reported no increase in water flux for an OMBR with CTA membranes when
the NaCl draw solution was increased above 1 M (58 g/L NaCl). Furthermore, Blandin et al. [62]
observed that water flux of fouled membranes converged to a common value for CTA and three
different TFC membranes, despite their very different initial fluxes. These results suggest that
FO membranes in OMBRs may have a critical flux analogous to the critical flux of MF and UF
membranes in MBRs. For MBRs, the critical flux is the flux at which increasing the
transmembrane pressure (TMP) driving force beyond a certain value does not result in stable

41

operation; it only results in greater membrane fouling [92]. The critical flux in MBRs is strongly
governed by the fouling potential of the feed solution and the reactor hydrodynamics [93, 94].
The same governing parameters would apply to OMBRs, however, the driving force in FO is
osmotic pressure and FO membranes reject not only particulate matter but also dissolved
solutes; thus, the critical flux concept becomes more complex. Given these key differences, it is
proposed that a “limiting” flux exists for OMBRs in which a state of dynamic constancy exists
despite changes in the driving force (e.g., higher draw solution concentration) and module
configuration (submerged and sidestream). Limiting flux is the same as critical flux in that
increased driving force does not result in higher steady-state water flux with fouled membranes.
However, limiting flux is different from critical flux in that both membrane fouling and CEOP are
exacerbated above the limiting flux due to RSF and concentration polarization. Flux decline from
FO membrane fouling reduces internal concentration polarization [17]; this increases the driving
force. However, due to RSF and CEOP, the cake layer resistance may still exceed the
increased driving force. Therefore, exceeding the osmotic pressure driving force at which the
limiting flux is achieved may increase membrane fouling without increasing water flux. The data
presented here support the limiting flux concept, however, a systematic investigation into this
phenomenon is required before definitive conclusions can be drawn.
In stage three (Figs. 3.2a and b, between 21 and 22 hrs), the system was operated in batch
mode and data from this stage were used to calculate SRSF through the fouled membranes
(Section 3.3.3.1). The final data point in each graph represents stage four water flux after the
membranes were rinsed and re-tested using the initial NaCl feed solution. After rinsing, greater
than 90% of the initial water flux was recovered for submerged and sidestream configurations.

3.3.2. Membrane surface characterization

42

3.3.2.1. Cake layer thickness and morphology
SEM images of virgin and fouled membrane coupons for the 35 g/L NaCl draw solution
experiments are shown in Fig. 3.3. The embedded polyester mesh used for support in CTA
membranes is visible in surface images of the selective layers (Figs. 3.3a-d). The mesh is most
clear in Fig. 3.3a that shows the virgin membrane, and is obscured in Figs. 3.3b-d that show the
fouled membranes. The mesh support was more obscured for the control configuration (Fig.
3.3b) because no fouling mitigation strategy was used. In Fig. 3.3c-d, shear forces (air scour
and hydraulic crossflow) reduced cake layer thickness. This is further confirmed by SEM cross-
section images (Fig. 3.3f-h, Table 3.1) that show decreasing cake layer thickness in the order of
control (15.6 ± 3.2 µm; Fig. 3.3f), submerged (6.3 ± 1.3 µm; Fig 3.3g), and sidestream (2.7 ± 0.7
µm; Fig 3.3h) configurations.

Figure 3.3. SEM images of selective layer surface of (a) a virgin CTA membrane and
membranes fouled in (b) control, (c) submerged, and (d) sidestream configurations. Also, SEM
cross-section images of (e) a virgin CTA membrane and fouled membranes in (f) control, (g)
submerged, and (h) sidestream configurations for 35 g/L NaCl draw solution. Yellow bars on
cake layers of cross-section images indicate cake layer thickness.

43

Table 3.1. Cake layer thickness and sodium content of cake layer surfaces measured by SEM/EDS.
Cake layer thickness error represents the standard deviation of ten measurements from the cross-section
image of one membrane and Na
+
error represents the standard deviation of four EDS scans of the cake
layer surface from one membrane.

35 g/L NaCl Draw Solution

100 g/L NaCl Draw Solution
Configuration

Cake thickness
(µm)

Na
+
Content
(wt. %)

Cake thickness
(µm)

Na
+
Content
(wt. %)
Control

15.6 ± 3.2

2.2 ± 0.2

13.8 ± 2.3

1.3 ± 0.2
Submerged

6.3 ± 1.3

3.8 ± 0.3

18.3 ± 3.1

6.3 ± 0.6
Sidestream

2.7 ± 0.7

2.5 ± 0.1

12.5 ± 2.0

5.8 ± 0.4

For both draw solution concentrations, the sidestream configuration resulted in a thinner
cake layer than the submerged configuration. This may be the result of greater hydraulic scour
and/or cake layer compaction due to hydraulic pressure (3.5 psi) in the sidestream configuration
compared to air scour and hydrostatic pressure (0.2 psi) in the submerged configuration. Similar
cake layer thicknesses were measured for the control configuration for both draw solution
concentrations (Table 3.1, Fig. 3.4). In the absence of fouling mitigation by air or hydraulic
scour, foulant-foulant interactions and decreased foulant movement toward the membrane
surface resulted in the same cake layer thickness regardless of draw solution concentration.
This is supported by flux decline results that show a steady-state flux of 1.7 LMH for both draw
solution concentrations (Fig. 3.2). Conversely, in the submerged and sidestream configurations,
draw solution concentration had a substantial effect on membrane fouling and the 100 g/L draw
solution experiments resulted in thicker foulant cake layers (Table 3.1, Fig. 3.4) than the 35 g/L
draw solution experiments. For the 100 g/L NaCl draw solution experiments, cake layer
thickness was almost three times greater for the submerged configuration and almost four times
greater for the sidestream configuration compared to the 35 g/L NaCl experiments (Table 3.1).
Furthermore, higher RSF from the higher draw solution concentration is likely to influence cake
layer morphology and composition.

44

Figure 3.4. SEM images of selective layer surface of (a) a virgin CTA membrane and
membranes fouled in (b) control, (c) submerged, and (d) sidestream configurations. Also, SEM
cross-section images of (e) a virgin CTA membrane and fouled membranes in (f) control, (g)
submerged, and (h) sidestream configurations for 100 g/L NaCl draw solution. Yellow bars on
cake layers of cross-section images indicate cake layer thickness. Insets on (c) and (d) depict
scaling on cake layer surfaces.

Closer inspection (×17,000) of the cake layers revealed precipitation on the cake surfaces
(Fig. 3.4c and d insets). Amorphous solids formed on the cake surface in the submerged
configuration and solids with a symmetric lattice structure formed on the cake surface in the
sidestream configuration. No solids were observed on the cake surface for the control
configuration or on the cake layers formed in the 35 g/L NaCl draw solution experiments (Fig.
3.3). Inorganic crystallization (scaling) on the cake layer results from concentration of feed
solutes from forward water flux at the membrane surface and reverse diffusion of draw solutes
into the cake layer [42, 68, 95, 96]. Scaling on foulant cake layers in OMBRs has been reported
previously [39, 95, 97]. In the current study, scaling in the submerged and sidestream
configurations is likely the result of increased RSF with the 100 g/L NaCl draw solution. The
different module configurations (submerged and sidestream) and subsequent fouling mitigation
strategies (air scour and crossflow) likely give rise to different crystal morphologies.

45

3.3.2.2. Elemental composition of cake layers
Elemental compositions of the cake layers were analyzed using EDS (Fig. 3.5). The major
constituents for all cake layers were carbon and oxygen; nitrogen, sodium, magnesium,
phosphorous, chloride, and calcium, which are typical chemical constituents of mixed liquor [36,
66, 98], were also found. They were detected in similar amounts on each sample. Palladium
and gold peaks were generated from the SEM preparatory coating procedure. For the 100 g/L
NaCl draw solution experiments, when precipitation occurred on the cake layer surfaces, EDS
spectra of the precipitates (Fig. 3.5e and f: insets) confirmed they were primarily composed of
sodium and chloride. This means the NaCl concentration approached or exceeded the
saturation concentration at the cake layer surface due to concentration polarization and RSF.
Despite the formation of thicker fouling layers with higher NaCl content for the 100 g/L NaCl
draw solution experiments, water flux was the same as the 35 g/L NaCl draw solution
experiments. The EDS results, in conjunction with cake layer thickness and water flux results,
suggest that exceeding the osmotic pressure driving force beyond the point at which a limiting
flux occurs results in increased membrane fouling without increased water flux.

46

Figure 3.5. EDS spectra of the cake layers formed using 35 g/L NaCl draw solution in the (a)
control, (b) submerged, and (c) sidestream configurations and 100 g/L NaCl draw solution in the
(d) control, (e) submerged, and (f) sidestream configurations. Insets in (e) and (f) are EDS
spectra and SEM images of the scale that formed on the cake layer surfaces.

3.3.3. Implications on reverse salt flux and OMBR salinity
3.3.3.1. Specific reverse salt flux
SRSF values were calculated for virgin (stage one), fouled (stage three), and rinsed (stage
four) membranes with both draw solution concentrations (Fig. 3.6). For the 35 g/L NaCl
experiments (Fig. 3.6a), higher initial SRSF for the control and submerged configurations may
be the result of lower pressure compared to the sidestream configuration where some pressure
was generated by the recirculation pump. Indeed, lower RSF with increasing pressure on the
feed side has been reported in pressure-assisted osmosis studies [99-101]. In the current study,
for each configuration (control, submerged, sidestream), initial SRSF values with NaCl feed
solution were statistically similar for both draw solution concentrations (Figs. 3.6a-b). These
kV:9.0 Tilt:0.00 Tkoff:37.63 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
FS : 16026 LSec : 90.9 Prst:100C 9-Dec-2016 13:45:05
N
O
Na
Mg
P
Cl Pd
4.80 5.40 keV
kV:9.0 Tilt:0.00 Tkoff:36.92 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
FS : 12598 LSec : 91.7 Prst:100C 9-Dec-2016 13:21:27
C
O
Na
Mg
P
Au
Cl Pd
Ca
0.60 1.20 1.80 2.40 3.00 3.60
4.80 5.40 keV
FS : 3276 LSec : 95.1 Prst:100C 6-Mar-2017 16:18:59
N
O
Na
Mg
P
Au Cl
Pd
4.20
SE1 50µm 200x
EDS Quantitative Results
Element Wt% At%
CK 39.43 58.42
NK 5.32 6.76
OK 18.81 20.92
NaK 5.43 4.21
MgK 0.89 0.65
PK 5.53 3.18
AuM 10.33 0.93
ClK 6.62 3.32
PdL 6.45 1.08
CaK 1.17 0.52
C:\EDAX32\GENESIS\GENMAPS.SPC
kV:9.0 Tilt:0.00 Tkoff:42.23 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
7
C
N
O
Na
Mg
P
Au
Cl
Pd
Ca
4.20 4.80 5.40 keV
SE1 50µm 200x
EDS Quantitative Results
Element Wt% At%
CK 45.20 62.09
NK 6.97 8.21
OK 22.27 22.96
NaK 1.57 1.13
MgK 0.93 0.63
PK 3.94 2.10
AuM 10.63 0.89
ClK 1.85 0.86
PdL 6.25 0.97
CaK 0.39 0.16
C:\EDAX32\GENESIS\GENMAPS.SPC
kV:9.0 Tilt:0.00 Tkoff:43.15 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
C
N
O
Na
Mg
P
Au
Cl
Pd
4.20 4.80 5.40 keV
SE1 50µm 200x
EDS Quantitative Results
Element Wt% At%
CK 36.36 57.13
NK 4.47 6.03
OK 18.97 22.38
NaK 4.83 3.96
MgK 1.16 0.90
PK 5.07 3.09
AuM 12.98 1.24
ClK 6.08 3.24
PdL 9.27 1.64
CaK 0.81 0.38
C:\EDAX32\GENESIS\GENMAPS.SPC
kV:9.0 Tilt:0.00 Tkoff:42.86 Det: SDD Apollo 40 Reso:127.4
FS : 3693 LSec : 95.0 Prst:100C 6-Mar-2017 16:43:16
C
N
O
Na
Mg
P
Au
Cl
Pd
4.20
keV 0.60 1.20 1.80 2.40 3.00 3.60 keV 0.60 1.20 1.80 2.40 3.00 3.60 keV
0.60 1.20 1.80 2.40 3.00 3.60 keV 0.60 1.20 1.80 2.40 3.00 3.60 keV 0.60 1.20 1.80 2.40 3.00 3.60 keV
N
C
C
Au
SE1 1µm 14000x
EDS Quantitative Results
Element Wt% At%
CK 19.44 44.07
NK 1.52 2.95
OK 4.27 7.26
NaK 12.39 14.67
MgK 0.70 0.79
PK 3.66 3.21
AuM 18.89 2.61
ClK 27.30 20.96
PdL 10.75 2.75
CaK 1.09 0.74
C:\EDAX32\GENESIS\GENMAPS.SPC
kV:9.0 Tilt:0.00 Tkoff:42.74 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
FS : 1402 LSec : 95.8 Prst:100C 6-Mar-2017 17:03:02
C
N
O
Na
Mg
P
Au
Cl
Pd
Ca
0.60 1.20 1.80 2.40 3.00 3.60
4.20 4.80 5.40 keV
SE1 50.00nm 190000x
EDS Quantitative Results
Element Wt% At%
CK 21.70 40.73
NK 1.81 2.91
OK 6.67 9.40
NaK 21.05 20.65
MgK 0.61 0.57
PK 3.70 2.69
AuM 6.57 0.75
ClK 32.24 20.51
PdL 3.93 0.83
CaK 1.71 0.96
C:\EDAX32\GENESIS\GENMAPS.SPC
kV:9.0 Tilt:0.00 Tkoff:42.34 Det: SDD Apollo 40 Reso:127.4 Amp.T:12.80
FS : 4767 LSec : 94.7 Prst:100C 10-Jan-2017 12:56:00
C
N
O
Na
Mg
P
Au
Cl
Pd
Ca
0.60 1.20 1.80 2.40 3.00 3.60
4.80 5.40 keV
(a) (b) (c)
(d) (e) (f)
1µm
1µm

47

initial SRSF values, ranging from 0.26-1.02 g/L, are comparable to SRSF values for studies
using non-fouling feed solutions and/or membranes that were not fouled [12, 24].

Figure 3.6. Specific reverse salt flux for all membrane conditions in the control, submerged, and
sidestream configurations with (a) 35 g/L and (b) 100 g/L NaCl draw solution concentrations.
Error bars represent standard deviation for three replicate virgin and fouled membranes and
duplicate measurements for rinsed membranes.

In Fig. 3.6, it can be seen that SRSF was higher for fouled membranes (stage 3) than virgin
membranes (stage 1). This is in agreement with previous reports that have shown lower water
flux and higher RSF for fouled membranes in submerged [38] and sidestream [66]
configurations. Higher SRSF with fouled membranes is likely due to simultaneous water flux
decline and decreased internal concentration polarization that lead to higher RSF [17, 38]. For
both draw solution concentrations, SRSF for fouled membranes was highest for the control
configuration (Fig. 3.6), which was primarily due to much lower water flux (Fig. 3.2), and lowest
for the sidestream configuration, which may have been the result of slightly higher pressure
generated by the recirculation pump on the feed side. The SRSF of the rinsed membranes

48

approached the SRSF of the virgin membranes, which indicates that membrane fouling was for
the large part, reversible.
Interestingly, the SRSF of the fouled membranes with 100 g/L NaCl draw solution was lower
for submerged and sidestream configurations compared to the control configuration (Fig. 3.6b),
yet SEM and EDS data showed the cake layers in the submerged and sidestream
configurations had NaCl scaling whereas the control configuration did not. This result suggests
that the cake layers in the submerged and sidestream configurations sequestered salt. For the
submerged configuration, if in situ membrane cleaning (i.e., osmotic backwashing or physical
cleaning) were used, it could release the cake layer and the sequestered salt into the bulk
solution, increasing bioreactor salinity. An advantage of the sidestream configuration is that
membrane cleaning via osmotic backwashing or hydraulic scour with high crossflow velocity is
possible without releasing the cake layer and sequestered salt into the bioreactor so long as the
system does not return the wash solutions to the bioreactor.
3.3.3.2. Steady-state OMBR salinity
OMBR salinity was modeled using steady-state SRSF values for fouled membranes from
the 35 g/L NaCl draw solution experiment (Fig. 3.7). Membrane areas required to meet 100 L/d
FO productivity rate (Q
FO
) were 2.39, 0.56, and 0.52 m
2
for the control, submerged, and
sidestream configurations, respectively. Salinity reached a steady-state concentration when the
incoming salinity (from influent wastewater and RSF) was balanced with the sludge wasting rate
(determined by the SRT). For both SRTs, the steady-state salinity was lowest to highest in the
order of sidestream, submerged, and control configurations. Predicted OMBR salinities with a
10-day SRT after 120 days of operation were 5.9, 3.6, and 2.6 g/L NaCl for the control,
submerged, and sidestream configurations, respectively. When the SRT was 40 days, there
was more accumulation of solutes and bioreactor salinities were 19.8, 11.0, and 7.1 g/L NaCl for
the control, submerged, and sidestream configurations, respectively. Due to lower SRSF, which

49

may be attributed to higher pressure on the feed side, the sidestream configuration modeled in
this study offers the advantage of lower bioreactor salinity that may result in more efficient
carbon and nitrogen removal compared to the submerged configuration.

Figure 3.7. Modeled OMBR salinity as a function of time for control, submerged, and
sidestream configurations using 35 g/L NaCl draw solution with (a) 10-day and (b) 40-day solids
retention times.

3.3.4. Sidestream FO module with the pilot-scale OMBR-MD system
To validate the observations made at the bench-scale, a sidestream FO membrane module
that contained a 4040 spiral-wound CTA membrane element was installed into the integrated
OMBR-MD system described in Chapter 2. Previously, the OMBR was operated with a plate-
and-frame FO membrane module that was submerged in the bioreactor. The spiral-wound
configuration required one additional pump to circulate the FO feed solution and a 700 µm mesh
pre-filter was used to screen particulate matter from entering the module (Fig. 3.8). The
membrane was evaluated and compared to the submerged membrane module for potential
benefits in terms of water flux, reverse salt flux, and contaminant rejection and specific energy

50

consumption. In effort to improve MD membrane performance, the tricot mesh spacers used
previously were replaced with diamond-style spacers, the symmetric polytetrafluoroethylene
(PTFE) MD membranes were replaced with asymmetric PTFE membranes with a polypropylene
backing for support, and larger heat exchangers were installed to improve heat recovery and
energy efficiency. These changes were expected to decrease the pressure drop across the MD
membrane module and increase heat transfer into the feed solution, both of which lead to higher
MD water flux.

Figure 3.8. OMBR subsystem with mesh pre-filter and 4040 spiral-wound FO membrane
module.

The bioreactor was seeded with activated sludge collected from Cold Springs Water
Reclamation Facility (Washoe County, NV) and the reactor was fed a high-strength synthetic
4040 FO
membrane
module
FO feed
solution
pump
Draw
solution
Concentrated
NaCl
Bioreactor
Mesh
pre-filter

51

wastewater solution comprised of 1,350 mg/L COD and 160 mg/L NH
4
+
-N. To achieve
nitrification and denitrification in a single reactor, a solids retention time of 30 days was set by
wasting 9.2 L/d and the redox environment in the reactor was controlled by alternating aeration
and non-aeration cycles. The reactor was allowed to operate under anoxic conditions for 30
minutes and then it was aerated until the dissolved oxygen concentration reached 2 mg/L. The
microbial community was acclimated to these conditions for six weeks prior to installing the FO
membrane module.
3.3.4.1. Sidestream OMBR-MD water flux and water production
Water flux with the sidestream FO membrane module for 30 days of continuous operation is
shown in Fig. 3.9. For the first nine days, the draw solution was 20 g/L NaCl and the membrane
was osmotically backwashed with tap water for one hour every 24 hours. Flux declined due to
membrane fouling and flux recovery was observed after each backwashing cycle. FO water
production decreased from 38.5 to 4.0 L/d over this period (2.10). On day 10, the draw solution
concentration was increased to 35 g/L NaCl and FO water production increased to between
29.0 and 48.7 L/d between days 10 and 15. In effort to increase FO water flux and water
production, backwashing cycles were increased to 30 minutes every 12 hours on day 16. The
twice-daily backwash cycles did result in two periods of flux recovery per day; however, net
water production for the system did not increase due to flux decline from fouling and also due to
the additional downtime required for the second backwash cycle. Bioreactor conductivity
increased from 0.4 to 3.9 mS/cm over 30 days of OMBR operation (Fig. 3.9), corresponding to
an RSF less than 0.23 g m
-2
h
-1
. RSF with the spiral-wound membrane was substantially lower
than it was with the plate-and-frame module (~ 4 g m
-2
h
-1
).

52

Figure 3.9. FO water flux, MD water flux, and bioreactor conductivity over 30 days of operation.

Figure 3.10. FO and MD water production for the sidestream OMBR-MD system.

53

The MD subsystem was integrated with the OMBR subsystem on day 18. MD was operated
with feed inlet and distillate inlet temperatures of 70 and 30 °C, respectively, and feed and
distillate solutions were circulated at a flow rate of 2 L/min. Initial MD water flux (day 18) was ~
3.5 L m
-2
h
-1
(Fig. 3.11) and flux declined over time to ~ 2 L m
-2
h
-1
on day 20. Over the same
period, feed solution inlet pressure increased (Fig. 3.12), and between days 19 and 21, feed
solution inlet pressure increased from 10.4 to 16.8 psi, indicating MD membrane fouling was
occurring. On day 21, the MD subsystem was taken offline and a fouling mitigation strategy was
developed to mitigate over-pressurization in the MD membrane module. Beginning on day 22, a
fouling mitigation strategy that consisted of a 30-minute flush with 0.1 M NaOH followed by a
30-minute rinse with tap water was implemented (Fig. 3.11, black arrows). NaOH flushing on
day 22 resulted in 85% recovery of initial flux and similar pressure to the initial pressure on day
18. However, after subsequent NaOH flushes on days 23 to 26, the percent of flux recovery
decreased and inlet pressures continued to increase. In addition to membrane fouling,
decreasing MD water flux over time may also be attributable to pore wetting, and increasing
distillate concentration from 0.07 to 0.21 g/L as NaCl over time indicates some pore wetting may
have occurred (3.11).

Figure 3.11. MD water flux and distillate conductivity during integrated OMBR-MD operation.
0
0.2
0.4
0.6
0.8
1
0
1
2
3
4
17 18 19 20 21 22 23 24 25 26 27
Concentration (g/L NaCl)
Water flux (LMH)
Elapsed Time (d)
MD water flux Distillate concetration
0.1 M NaOH flush points

54

Figure 3.12. Feed and distillate inlet and outlet pressures during integrated OMBR-MD
operation.

3.3.4.2. Sidestream OMBR water quality at pilot-scale
COD removal and COD concentrations in the wastewater, bioreactor, FO draw solution tank,
and MD distillate tank are shown in Figure 3.13. Overall, the integrated system achieved
between 98.7 and 100% COD removal. COD concentrations in the bioreactor were lower than
those in the FO draw solution. COD accumulation in the FO draw solution is attributable to high
rejection by MD membranes. On day 18, when the MD subsystem was brought online,
ammonium removal was 96.2%. Due to the closed-loop between the FO draw solution and MD
feed solution, ammonium tended to accumulate in the FO draw tank and MD feed tank, leading
to a decreasing trend in ammonium removal over time (Fig. 3.14). However, a significant

55

amount of ammonium removal was observed in the bioreactor and the integrated system
achieved between 85.8 and 96.2% ammonium removal. Moreover, nitrification in the reactor,
and high nitrate rejection by FO and MD membranes, resulted in distillate solution with less than
0.05 mg/L nitrate-nitrogen (Fig. 3.15).

Figure 3.13. Soluble COD concentrations in the wastewater, bioreactor, FO draw solution tank,
and MD distillate tank during integrated operation. Blue circles represent system removal.

Figure 3.14. Ammonium-nitrogen concentrations in the wastewater, bioreactor, FO draw
solution tank, and MD distillate tank during integrated operation. Blue circles represent system
removal.

56

Figure 3.15. Nitrate-nitrogen concentrations in the bioreactor, FO draw solution tank, and MD
distillate tank during integrated operation.

3.3.4.3. Spatial distribution of MD membrane fouling
The MD membranes were removed on day 26 and samples were taken from the inlet,
center, and outlet sections for SEM and EDS analysis. Major constituents on the feed side inlet,
which are typically associated with biological fouling, include carbon, oxygen, sodium,
magnesium, chloride, and calcium (Fig. 3.16). The presence of fluorine is due to the PTFE
membrane material and the presence of aluminum and silicon may be attributable to
aluminosilicate, an anti-caking agent in the sucrose used to prepare the synthetic wastewater.
Platinum and palladium peaks are due to the membrane coating process required for SEM
analysis. The SEM image and EDS spectra at the feed side outlet are most representative of a
virgin PTFE membrane. Foulant concentrations were highest at the feed side inlet, lower at the
middle, and lowest at the outlet, indicating that foulant deposition along a lengthwise gradient
across the membrane.

57

Figure 3.16. SEM images and EDS spectra of used MD membranes taken from the feed side
(a,b) inlet, (c,d) middle, and (e,f) outlet sections of the membrane module.

Major constituents indicative of organic membrane fouling on the distillate side of the
membrane at the inlet include, carbon, oxygen, sodium, magnesium, and chloride (Fig. 3.17).
Similar to the feed side, SEM images and EDS spectra confirm foulant concentrations were
highest at the distillate side inlet, lower at the middle, and lowest at the outlet. These results

58

suggest that increasing feed and distillate inlet pressure over time was due to foulant
accumulation at the inlets.

Figure 3.17. SEM images and EDS spectra of used MD membranes taken from the distillate
side (a,b) inlet, (c,d) middle, and (e,f) outlet sections of the membrane module.

59

3.3.5. Specific energy of submerged and sidestream OMBR systems
Measured water flux and electrical energy of the circulation pumps used in the submerged
and sidestream OMBRs at the pilot-scale were used to calculate the experimental specific
energy (SE) (Table 3.2). SE was calculated using
SE=
P
Q
P
(4)
where P is the electrical power consumption of the pumps and Q
P
is the permeate flow rate. The
pilot-scale systems were not operated with pumps designed for high efficiency at the operational
flow rates, which resulted in high experimental SE values. Energy consumption is expected to
be lower for a system operated with pumps designed to operate at much higher efficiencies for
the desired flow rates. Experimentally measured water flux and flow rates were used to model
an optimized system operating with 80% pumping efficiency, which is typical of full-scale
membrane systems [102]. The required pumping power was calculated using
P=
QΔP
( )
F
+ QΔP
( )
D
η
(5)
where Q is the flow rate; ΔP is the pressure drop across the membrane module, η is the pump
efficiency, and the subscripts F and D correspond to feed and draw solutions. The optimized SE
was 7.3 and 33.7 kWh/m
3
for submerged and sidestream OMBRs. SE of the MD subsystem
would be the same for both system, thus it was not included in the comparison of SE between
the configurations.
Table 3.2. Experimental and optimized energy consumption by submerged and sidestream OMBRs.
Experimental OMBR Optimized OMBR
Submerged Sidestream Submerged Sidestream
Feed pump energy (kW) - 1.60E-01

- 2.42E-05
Draw pump energy (kW) 2.16E-02 2.16E-02

4.83E-06 2.42E-06
Total pumping energy (kW) 2.16E-02 1.82E-01 4.83E-06 2.66E-05
Specific Energy (kWh/m
3
) 17.4 112.6 4.9 20.6

60

SE of the sidestream OMBR was higher than the submerged OMBR, which is likely due to
the additional pump required to circulate the feed solution and to low water flux through with the
spiral-wound membrane module. The spiral-wound module is designed with an integral spacer
(Fig. 3.18) that may cause accumulation of foulants at the membrane surface. This
accumulation would increase the pressure drop across the module, causing an increase in
energy requirement. Indeed, the feed solution pump in the sidestream configuration accounts
for 88% and 91% of the experimental and optimized SE requirement for the sidestream
configuration. However, an improved sidestream module design with an open channel that
results in a lower pressure drops across may significantly reduce the SE with the sidestream
configuration.

Figure 3.18. Feed channel of the sidestream 4040 FO module with integral spacer.
3.4. Conclusions
Overall, the initial water flux and fouled water flux with time was the same for submerged
and sidestream configurations. Furthermore, the steady-state water flux of the fouled

61

membranes was the same for both (submerged and sidestream) configurations and both draw
solution concentrations, leading to the concept of a limiting flux in OMBRs similar to the critical
flux in conventional MBRs. However, for fouled membranes, the sidestream configuration
resulted in lower SRSF at steady-state with both draw solution concentrations. The 100 g/L
NaCl draw solution experiments did not result in higher steady-state water flux for the fouled
membranes despite the significant increase in driving force; instead, the higher draw solution
concentration did result in the formation of thicker fouling layers with NaCl scaling. Thinner cake
layers were formed in the sidestream configuration where fouling is mitigated with hydraulic
crossflow compared to the submerged configuration where fouling is mitigated via air scour.
Hydraulic pressure from recirculation pumping on the feed side of the sidestream configuration
may have also resulted in a more compact cake layer over time. For the experimental conditions
tested in this study, model results predicted that the sidestream configuration would result in
lower steady-state OMBR salinity, particularly when longer SRTs required for nitrogen removal
are used.
Lower SRSF in the sidestream configuration is advantageous, although it comes at the cost
of additional recirculation pumping, which can add significantly to specific energy consumption
(kWh m
-3
) [103]. On the other hand, in the submerged configuration, the aeration required to
achieve air scour exceeds that required to supply the biological community with oxygen [38,
104]. How much additional aeration is required depends on membrane length, bubble size, and
width of the bubble channel [7]. In the end, additional energy requirements for operation (e.g.,
recirculation pumping) as well as for fouling mitigation (e.g., increased crossflow velocity,
increased aeration rate, chemical cleaning, and osmotic backwashing) must be considered to
comprehensively assess the tradeoffs between FO module configurations.

62

Chapter 4. Membrane compaction and effects on membrane properties in FO

This chapter is currently in preparation for submission to a peer-reviewed journal. Final results
are presented in this manuscript.

4.0. Abstract
Dense membranes can be used to separate water from dissolved solutes with a hydraulic
pressure gradient, an osmotic pressure gradient, or a combination of the two. However, recent
publications have emphasized that fundamental transport mechanisms are different and the
testing environment may alter the water permeability, solute permeability, and structural
parameters of forward osmosis (FO) membranes. Three commercially available FO
membranes, one CTA and two TFC membranes, were characterized with the conventional
reverse osmosis-forward osmosis (RO-FO) method and a recently developed method that uses
FO testing only. Deviations in FO membrane transport and structural parameters occurred
between methods. Overall, there was better agreement between experimental results and
modeled FO fluxes using parameters obtained with the FO method, particularly for RSF. SEM
images revealed that compaction under hydraulic pressure led to decrease membrane
thickness. CTA membrane was resistant to membrane compaction and TFC membranes
compacted significantly (36%). For TFC membranes, compaction of macro-void spaces resulted
in thinner, more permeable (higher A and B values) membrane and compaction of finger-like
porous structures resulted in increased selectivity (similar A and lower B values).
4.1. Introduction
Forward osmosis (FO) membranes consist of a thin, dense selective layer with a thicker,
porous structural layer for support. The selective layer is typically comprised of either cellulose
triacetate (CTA) or polyamide. Prior to their use in FO, these dense materials were used as the
selective layers for reverse osmosis (RO). Although the selective layers are the same, the

63

driving force with RO is a hydraulic pressure gradient whereas the driving force with FO is an
osmotic pressure gradient. Regardless of the different driving forces, the solution-diffusion
model is commonly applied to predict water flux for both processes. In addition to system-
dependent parameters such as mass transfer coefficients, temperature, and diffusivity,
membrane-dependent parameters (e.g., water permeability (A), solute permeability (B), and the
structural parameter (S)) are used as inputs to the solution-diffusion model.
The standard methodology for determination of intrinsic FO membrane parameters is done
with RO experiments for transport parameters [105]; however, due to the different driving forces
for separation between RO (hydraulic pressure) and FO (osmotic pressure) processes,
determination of A, B, and S solely under FO testing conditions may be more relevant for
predicting FO performance. Recently, an alternative method utilizing FO experiments with
stepwise increases in draw solution concentration for determination of A, B, and S was
developed by Tiraferri et al. [106]. It is well known that for clean membranes with non-fouling
feed solutions, increasing the draw solution concentration results in greater water flux, greater
RSF, and a consistent specific reverse salt flux (SRSF; g/L); however, we have found this not to
be true for fouled membranes. We assert that the limiting flux concept applies to fouled
membranes.
The objectives were to determine if membrane transport and structural parameters were
dependent on the different driving forces with RO (hydraulic) and FO (osmotic) testing
environments. We also sought to the effect of hydraulic pressure FO membrane properties (i.e.,
thickness and porosity). One CTA membrane and two TFC membranes were tested. First,
membrane parameters A, B, and S were obtained using both the conventional RO-FO method
[105] and the stepwise FO method [106]. SEM image analysis was used to quantify changes in
membrane thickness and changes in support layer porosity due to compaction under hydraulic
pressure. Next, parameters obtained by both methods were used to model FO water flux and

64

RSF and model results were compared to experimental results for goodness of fit. Finally, the
difference between transport mechanisms between hydraulic and osmotic pressure gradients
was tested by performing FO experiments with membranes previously compacted by RO
testing.
4.2. Materials and Methods
4.2.1. Membranes and bench-scale FO system
Three commercially available FO membranes were tested; a CTA membrane from Fluid
Technology Solutions, Inc. (Albany, OR), a TFC membrane from Hydration Technology
Innovations, LLC (Albany, OR), and a TFC membrane from Toray Industries, Inc. (Tokyo,
Japan). Membranes were tested in the FO orientation with the selective layer facing the feed
solution. The bench-scale FO system has been described previously [58]. Briefly, the
membrane module houses three identically sized (42 cm
2
) coupons and each coupon has a
dedicated 2-L feed solution reservoir and an independent 1-L draw solution reservoir equipped
with an automated dosing system to maintain a constant osmotic pressure. The draw solution
and dosing solution were prepared from analytical grade NaCl (VWR International, Radnor,
Pennsylvania) and the membranes were tested with a crossflow velocity of 10 cm/s at 24 ± 1
°C. No feed spacers were used and 31-mil (0.79 mm) diamond-style spacers (Sterlitech
Corporation, Kent, WA) were used on the draw solution side. The draw solution was kept at
constant osmotic pressure based on real-time measurements from electrical conductivity probes
(Cole-Parmer, Vernon Hills, IL) submerged in the draw solution reservoirs. When the
conductivity fell below the set point, the dosing system was activated and 5-M NaCl was
transferred to the reservoir until the set point was reached. Conductivity measurements from a
probe in the feed solution (EC215, Hanna Instruments, Woonsocket, RI) were used to calculate
RSF with

65

(1)
where C
i
and C
F
are initial and final NaCl feed concentrations (calculated from conductivity
measurements); V
i
and V
F
are initial and final feed volumes; A
M
is membrane area; and Δt is
elapsed time. Error in water flux and RSF values were calculated from the standard deviation
between triplicate membrane coupons.
4.2.2. Membrane characterization
4.2.2.1. RO-FO characterization method
The RO test cell had dimensions of 15 cm long × 10 cm wide × 0.2 cm deep. A 31 mil (0.79
mm) diamond-style spacer was used on the feed side and a 17 mil (0.43 mm) permeate carrier
spacer was used on the permeate side (Sterlitech Corporation). Membranes were tested under
0.5, 1.0, 2.0, and 3.0 MPa (72.5, 145, 290, and 435 psi) transmembrane pressures (TMP) with a
crossflow velocity of 20 cm/s at 24 ± 1 °C. A and B were obtained using the same membrane
coupon; deionized water was the feed solution for determination of A and a 1 g/L NaCl solution
was the feed solution for determination of B. A and B were calculated from at least three
replicate experiments at each pressure using Eqn. (2) and (3) [105].
(2)
(3)
where J
w
RO
is RO water flux; ΔP is transmembrane pressure; R is rejection (calculated using
NaCl concentrations of the feed and permeate solutions); and k is the mass transfer coefficient
of the RO test cell. The structural parameter (S) was determined from 30-min FO experiments
using the FO membrane module and testing conditions described in 4.2.1. The feed solution
RSF=
C
F
V
F
−C
i
V
i
A
M
Δt
A=
J
RO
W
ΔP
B=J
w
RO
1−R
R
⎛
⎝
⎜
⎞
⎠
⎟
exp −
J
w
RO
k
⎛
⎝
⎜
⎞
⎠
⎟

66

was 1 g/L NaCl (0.1 MPa osmotic pressure) and water flux was recorded with four draw solution
concentrations corresponding to 0.5, 1, 2, and 3 MPa osmotic pressure. The average S value
was calculated from the four experiments using [105, 107, 108]
(4)
where D is draw solute diffusivity from [109]; J
w
is FO water flux; π
D
is osmotic pressure of the
draw solution; and π
F,m
is osmotic pressure at the membrane surface on the feed solution side
which was calculated using the concentrative external concentration polarization modulus [52].
Osmotic pressures of NaCl draw solutions (Table 4.1) were calculated using OLI Stream
Analyzer (OLI Systems, Inc., Morris Plains, NJ).

Table 4.1. Osmotic pressures (MPa) and corresponding NaCl
draw solution concentrations (g/L).
Draw Solution Osmotic
Pressure (π D, MPa)
g/L NaCl
0.5 6.4
1 12.9
2 25.5
3 37.7

4.2.2.2. Stepwise FO characterization method
Results from the four FO experiments measured with the RO-FO method (4.2.1.) were also
used for determination of A, B, and S with the stepwise FO method. Water flux and conductivity
values were used with the governing solution-diffusion transport equations for water flux (5) and
reverse salt flux (6) [106, 110]
(5)
S=
D
J
w
ln
B+Aπ
D
B+J
w
+Aπ
F,m
⎛
⎝
⎜
⎜
⎞
⎠
⎟
⎟
J
w
=A
π
D
exp
−J
w
S
D
⎛
⎝
⎜
⎞
⎠
⎟
−π
F
J
w
k
⎛
⎝
⎜
⎞
⎠
⎟
1+
B
J
w
exp
J
w
k
⎛
⎝
⎜
⎞
⎠
⎟
−exp −
J
w
S
D
⎛
⎝
⎜
⎞
⎠
⎟
⎡
⎣
⎢
⎤
⎦
⎥
⎧
⎨
⎪
⎪
⎩
⎪
⎪
⎫
⎬
⎪
⎪
⎭
⎪
⎪

67

(6)
where C
D
and C
F
are the NaCl concentrations of the bulk draw solution and bulk feed solution.
The system of eight transport equations was fit to experimental data using a least squares non-
linear regression with A, B, and S as regression parameters. A, B, and S were determined by
minimizing the global error calculated from the sum of the offsets from non-linear equations
[106].
4.2.2.3. RO-FO and stepwise FO method comparison
The A, B, and S values determined by RO-FO and stepwise FO methods were used in Eqn.
(5) and (6) to predict water flux and RSF values for four draw solution osmotic pressures (0.5, 1,
2, and 3 MPa). Coefficient of determination (R
2
) values were used to determine goodness of fit
for model predictions utilizing parameters obtained by both methods. For example, coefficient of
determination values for water flux using FO parameters (R
2
w,FO
) were calculated using [106,
111]
(7)
where SS
res
is the residual sum of squares; SS
tot
is the total sum of squares; n equals the
number of data points (n = 4); and superscripts Exp and Model indicate experimental values
and model predictions. Coefficients of determination for RSF (R
s
2
) were calculated similarly. An
R
2
value of 1 indicates perfect model fit and values less than 1 indicate variance in model
predictions.
J
S
=B
c
D
exp
−J
w
S
D
⎛
⎝
⎜
⎞
⎠
⎟
−c
F
J
w
k
⎛
⎝
⎜
⎞
⎠
⎟
1+
B
J
w
exp
J
w
k
⎛
⎝
⎜
⎞
⎠
⎟
−exp −
J
w
S
D
⎛
⎝
⎜
⎞
⎠
⎟
⎡
⎣
⎢
⎤
⎦
⎥
⎧
⎨
⎪
⎪
⎩
⎪
⎪
⎫
⎬
⎪
⎪
⎭
⎪
⎪
R
w,FO
2
≡1−
SS
res
SS
tot
=1−
J
w,i
Exp
−J
w,i
FO−Model
( )
2
i
n
∑
J
w,i
Exp
−J
i→n
Exp
( )
2
i
n
∑

68

4.3. Result and discussion
4.3.1. Transport and structural parameters
A, B, and S values obtained using the RO-FO method and the FO method are given in
Table 3. Dissimilarities in the values are attributable to the effects of different driving forces in
RO (hydraulic) and FO (osmotic) on the membrane [106, 112]. Despite the differences in
magnitude, similar trends in A, B, and S values were expected between the two methods for
each membrane; however, no consistent trends were observed. The A/B ratio is an indicator of
selectivity; higher A/B indicates greater selectivity and lower A/B indicates lower selectivity [22,
113]. The inconsistencies in A and B values inevitably led to differences in S values.
Discrepancies in individual A, B, and S values between the methods brings into question the
assumption that membrane parameters are conservative properties independent of hydraulic
(RO) and osmotic (FO) pressure environments.
Table 4.2. Water permeability (A), solute permeability (B), and structural parameters
(S) determined by RO-FO and FO methods.

RO-FO Method
Membrane A (L m
-2
h
-1
bar
-1
) B (L m
-2
h
-1
) S (µm) A/B (bar
-1
)
FTS CTA 0.69 ± 0.06 0.38 ± 0.14 297 ± 11 1.82
HTI TFC 1.12 ± 0.42 0.15 ± 0.05 146 ± 10 7.34
Toray 3.87 ± 0.36 0.46 ± 0.14 254 ± 36 8.47

FO Method
Membrane A (L m
-2
h
-1
bar
-1
) B (L m
-2
h
-1
) S (µm) A/B (bar
-1
)
FTS CTA 0.77 ± 0.07 0.30 ± 0.03 403 ± 33 2.72
HTI TFC 1.36 ± 0.10 1.21 ± 0.14 222 ± 39 1.12
Toray 2.29 ± 0.34 0.30 ± 0.08 198 ± 44 7.63

Membrane compaction under hydraulic pressure may decrease membrane thickness,
particularly for porous support structures, which can result in additional mass transfer resistance

69

[112]. However, reported effects of compaction on transport parameters have been inconsistent,
especially with regard to solute permeability, B. In some instances, compaction resulted in lower
A and lower B [114]. Conversely, lower A and higher B have also been attributed to support
layer compaction under hydraulic pressure [112]. No change in A accompanied with lower B has
also been reported [100]. Differing effects of compaction on A and B values are likely related to
the propensity of a given membrane material to undergo structural changes that alter water and
solute permeability, and/or to the compacted morphology of the structural layer (e.g., more
dense versus more porous). This assertion, which is supported by the discrepancies between
the RO-FO and FO methods found in the current study, suggests a strong dependence between
membrane parameters and the driving force under which they are obtained.
4.3.2. Accuracy of characterization methods
Water flux and RSF were modeled with Eqns. (5) and (6) using parameters obtained by both
RO-FO and FO methods and compared to experimental values with 0.5, 1, 2, and 3 MPa draw
solution osmotic pressure (Fig. 4.1). Errors in model water flux using RO-FO parameters (R
2
w,RO-
FO
) were between 0.86 and 0.99 and errors using FO parameters (R
2
w,FO
) were also between
0.86 and 0.99, demonstrating that both methods resulted in similar accuracy. However, errors in
model RSF using RO-FO parameters (R
2
s,RO-FO
) were between -2.4 and 0.81 while errors using
FO parameters (R
2
w,FO
) were between 0.91 and 0.98; thus, the FO method predicted RSF with
greater accuracy, particularly for HTI TFC membranes. A previous study comparing RO-FO and
FO characterization methods found that for CTA membranes, A/B values calculated using RO
parameters were lower [106]. In the current study, A/B values calculated using RO-FO
parameters were also lower for FTS CTA. This suggests that hydraulic pressure in RO testing
may cause lower rejection due to defects in the selective layer that either formed during testing,
or previously existed and became more problematic. Conversely, A/B values were higher for
HTI TFC and Toray TFC membranes. Clearly, the testing conditions under which transport and

70

structural parameters are obtained influences model predictions. This brings into question the
underlying assumption that mass transport with non-porous membranes under hydraulic
pressure is governed purely by solution-diffusion [112].
Figure 4.1. Comparison of membrane parameters obtained with the RO-FO and FO
characterization methods. Model predictions and experimental water flux (top) and RSF
(bottom) values for (a,d) FTS CTA, (b,e) HTI TFC, and (c,f) Toray TFC membranes. R
2
values
represent the errors between model predictions and experimental results.

4.3.3. FO membrane compaction under hydraulic pressure
Figs. 4.2a-f depict the selective layer surfaces, structural layer surfaces, and cross-sections
of virgin membranes and membrane after RO characterization. For all membranes, there was
no visible change in the characteristics of the membrane surfaces after RO characterization
(i.e., no visible defects manifested into larger pores as a result of hydraulic pressure). However,

71

some differences were observed in membrane thickness as determined by the difference in the
cross-section images. CTA membrane had the least amount of compaction (4%) and TFC
membranes had a higher degree of compaction (36%). For the HTI TFC membrane, compaction
of both finger-like and macro-void spaces was observed (Fig. 4.4c, f). For the Toray TFC
membrane, only compaction of the macro-void spaces was observed while the finger-like pores
remained intact (Fig. 4.4c, f).
For the CTA membrane, similar A and B values with both characterization methods is
attributed the structural resistance of the support layer to resist compaction during RO testing.
For the HTI TFC membrane, higher B with the RO method arises from uniform compaction of
the finger-like and macro-void pore structures, resulting in a denser material that effectively
behaves as a selective layer. For the Toray TFC membrane, increased A and B with the RO
characterization method indicate that decreased thickness due to compaction of macro-void
spaces without compaction of finger-like structure causes increasing A and B.

Figure 4.2. SEM images of FTS CTA membrane (a,d) selective layer surface, (b,e,) support
layer surface, and (c,e) cross-sections before and after RO characterization.

72

Figure 4.3. SEM images of HTI TFC membrane (a,d) selective layer surface, (b,e,) support
layer surface, and (c,e) cross-sections before and after RO characterization.

Figure 4.4. SEM images of Toray TFC membrane (a,d) selective layer surface, (b,e,) support
layer surface, and (c,e) cross-sections before and after RO characterization.

73

4.4. Conclusions
SEM analysis revealed that CTA membranes resist compaction under hydraulic pressure
(up to 435 psi), leading to similar A and B values with both methods. Conversely, TFC
membranes underwent severe compaction during RO characterization. Collapsed macro-void
and finger-like pore structures lead to decreased solute permeability and increased selectivity
with the HTI TFC membrane during RO characterization. For the Toray TFC membrane,
collapsed macro-void spaces with intact finger-like structures increased permeability during RO
characterization. Model FO water flux was accurate using parameters obtained with either
characterization method. However, model RSF was more accurate using parameters obtained
with the FO characterization method than the RO-FO method. These results suggest that FO
membranes which are expected to be subjected to a combination of hydraulic and osmotic
pressures during scaled-up applications may undergo compaction and longer-term studies are
required.

74

Chapter 5. Relating forward osmosis membrane properties to performance
under fouling conditions: Implications of the membrane-independent limiting flux

This chapter is currently in preparation for submission to Environmental Science and
Technology.
5.0. Abstract
Recent observations with fouled FO membranes suggests that a similar phenomenal to the
critical flux with pressure driven membrane processes exists for osmotically driven membrane
processes. However, the critical flux phenomenon in FO, termed the limiting flux, is more
complex due to reverse salt flux, internal concentration polarization, and cake enhanced
osmotic pressure. While much attention has been given to attaining high FO water flux by
improving membrane transport and structural parameters (e.g., higher A, and lower B and S),
the role of membrane parameters on critical flux behavior with FO is not well understood. Thus,
a systematic investigation into the limiting flux phenomenon was conducted with one CTA and
four TFC membranes. Membranes were characterized and tested using stepwise increases in
draw solution osmotic pressure, then experiments were repeated under fouling conditions. CTA
membranes had the lowest selectivity (A/B = 0.58) and TFC membranes had higher A/B (2.04
to 7.63). Baseline water flux increased according to increasing A/B and baseline RSF
decreased according to increasing A/B. However, during fouling experiments, an upper limit to
water flux (the limiting flux) was observed for all membranes despite increasing draw solution
osmotic pressure. Conversely, there was not an upper limit to RSF. Limiting water flux values
were validated with a second set of fouling experiments conducted with constant, higher draw
solution osmotic pressures. Membranes with higher selectivity values had the highest initial flux,
although, over time, membrane fouling limited flux to a singular value for all membranes. Using
a resistance-in-series model, hydraulic resistance of the cake layer was higher for membranes
with higher selectivity, indicating that greater selectivity was completely offset by membrane
fouling.

75

5.1. Introduction
Osmotically driven membrane processes such as forward osmosis (FO) and pressure-
retarded osmosis are being investigated for their potential to reduce energy consumption and
cost of water reuse [1, 8, 115, 116]. Osmotic membrane bioreactors (OMBRs), which combine
FO with biological wastewater treatment, represent a low-fouling alternative to conventional
membrane bioreactors (MBRs) that use microfiltration (MF) or ultrafiltration (UF) membranes [5-
7, 53, 117, 118]. In FO, the driving force for water flux is the osmotic pressure difference
between a high salinity draw solution and a low salinity feed solution across the selective layer
of a semi-permeable membrane [8]. Because FO is an osmosis-driven process, FO membrane
fouling is less severe and more reversible than fouling in pressure-driven membrane processes
(i.e., nanofiltration (NF) and reverse osmosis (RO)) [41-43]. However, like pressure-driven
membrane processes, FO membrane fouling can significantly hinder water flux. Additionally,
dilution of the draw solution within the porous support structure (dilutive internal concentration
polarization (ICP)) and diffusion of draw solutes into the feed solution (reverse salt flux (RSF))
represent unique phenomena with FO. Thus, FO membrane fouling becomes more complex
than fouling with pressure-driven membrane processes.
For pressure-driven membranes with high salt rejection (i.e., NF and RO), solutes in the
feed solution are rejected at the membrane surface and subsequently back diffuse into the bulk
feed solution. The presence of a foulant cake layer hinders this back diffusion and increases the
osmotic pressure at the membrane surface on the feed side by “cake enhanced concentration
polarization” (CECP), resulting in significant flux decline [82, 83]. This also occurs with FO,
although FO processes are typically operated at lower water fluxes where CECP is less
significant. However, draw solutes also diffuse into the cake layer due to RSF. This reduces the
osmotic pressure driving force across the selective layer by “cake enhanced osmotic pressure”
(CEOP), causing water flux to decline [42, 68]. However, as water flux declines, draw solute

76

concentration in the support layer increases, causing the driving force across the selective layer
to increase by the “ICP self-compensation effect” [17, 31, 119]. Tang et al. [119] conducted
humic acid fouling experiments to investigate ICP self-compensation; higher draw solution
concentrations (0.5, 1, 2, and 4 M NaCl) caused a greater amount of foulant deposition and
greater flux decline. These observations suggest that the “critical flux” concept with MF/UF
membranes, which asserts that a critical water flux exists below which flux decline from fouling
does not occur with pressure-driven membranes [93, 94], might also be applicable with FO
processes.
Fouling studies with FO [62, 69, 91, 119-122], PRO [123], and OMBR [38, 58] applications
have also observed critical flux behavior with FO. In an FO fouling study with sodium alginate
[122], flux decline did not occur unless initial flux was above a certain “critical” value. The critical
flux value depended on the choice of draw solute and magnitude of the critical flux was highest
in the order of NaCl > MgCl
2
> CaCl
2
> Ca(NO
3
)
2
, suggesting dependencies on RSF, CEOP,
and foulant/draw solute interactions (i.e., alginate bridging with divalent ions). This dependency
was supported in a systematic investigation into FO critical flux behavior [69] utilizing stepwise
increases in NaCl and MgCl
2
draw solution concentrations; critical flux values were lower and
fouling was less reversible with MgCl
2
draw solution. Similarly, no flux decline was observed
with an OMBR when 0.5 M NaCl draw solution was used, and flux declined toward a singular
value with 1.0, 1.5, and 2-M NaCl draw solution concentrations [38]. In an OMBR study, steady-
state water flux with fouled membranes was the same with two different draw solution
concentrations (35 and 100 g/L NaCl) [58]. However, a thicker foulant cake layer formed on the
membrane and higher RSF was observed when 100 g/L NaCl draw solution was used. The
observations in this study led to the concept of a limiting flux for FO based on the critical flux
concept for pressure-driven membranes and extended to incorporate RSF and ICP
compensation.

77

In the absence of membrane fouling (Fig. 5.1a), external concentration polarization (ECP)
occurs along the membrane surface/bulk solution boundary layers, ICP occurs within the porous
support layer, and mass transfer is quantified by the membrane parameters (e.g. water
permeability (A), solute permeability (B)) and structural parameter (S)). With a fouling feed
solution (Fig. 5.1b), convective water flux brings foulants into contact with the membrane
surface. Limiting flux is the maximum steady-state water flux that is obtainable in the presence
of foulants, which cannot be overcome by increasing the osmotic driving force. The limiting
osmotic pressure can be defined as the minimum osmotic pressure required to achieve limiting
flux. When the osmotic pressure is greater than the limiting osmotic pressure (Fig. 5.1c), water
flux is temporarily greater than limiting flux; this causes foulants to accumulate into a cake layer
on the membrane surface and water flux declines due to additional hydraulic resistance and
CEOP. As water flux declines, the driving force adjusts due to the ICP self-compensation effect,
and ultimately, the system achieves equilibrium at the limiting flux. Increasing the draw solution
osmotic pressure beyond the limiting osmotic pressure does not increase steady-state water
flux; however, the increased draw solution concentration at the selective/structural layer
interface (π D,i) results in greater RSF and CEOP. The dynamics between critical flux and limiting
flux are quite different, although the core principle remains the same; greater driving force does
not result in greater water flux.

78

Figure 5.1. Concentration polarization with limiting draw solution osmotic pressure difference
(Δπ
H
) for (a) non-fouling feed solution and (b) fouling feed solution and membrane fouling and
(c) with membrane fouling and osmotic pressure difference (Δπ) greater than Δπ
H
. Δπ
represents osmotic difference between bulk feed solution (π
F,bulk
) and bulk draw solution
(π
D,Bulk
), concentrative ECP is between π
F,Bulk
and the osmotic pressure at the selective layer
surface (π
F,Mem
), dilutive ECP is between π
D,Bulk
and the osmotic pressure at the structural layer
surface (π
D,Mem
), dilutive ICP is between π
D,Mem
and the selective/structural layer interface (π
D,i
).
The magnitude of CEOP is the osmotic pressure difference across the foulant cake layer. Water
flux is the same for all cases; in (a) there is no membrane fouling; in (b) the presence of foulants
does not cause flux decline; and in (c) the increased driving force causes membrane fouling,
increased CEOP, and ICP self-compensation resulting in higher π
D,I
and greater RSF.

Stepwise and long-term methods for determination of critical flux values were developed for
MBRs and the fouling potential of the feed solution was found to be a major determinant of the
critical flux value [92]. In addition to the fouling potential of the feed solution, membrane type
and selectivity (i.e., greater selectivity with thin film composite (TFC) membranes than cellulose
triacetate (CTA) membranes) may also influence limiting flux, limiting osmotic pressure, and
RSF with fouled membranes. The standard methodology for determination of intrinsic FO
membrane parameters is done with RO experiments for transport parameters and FO
experiments for the structural parameter (S) [105]; however, due to the different driving forces
for separation with RO (hydraulic pressure) and FO (osmotic pressure), determination of A, B,
π
D,bulk
∆π
Eff.
J
w
π
D,i
π
F,bulk

π
F,mem
J
S
Selective
Layer
Structural
Layer
Non-fouling feed
∆π =

∆π
L
Draw
Solution
Feed
Solution
π
D,Mem
Dilutive ECP
Dilutive ICP
Concentrative
ECP
Fouling feed
∆π =

∆π
L
Fouling feed
∆π >

∆π
L
π
D,bulk
∆π
Eff.
J
w
π
D,i
J
S
Selective
Layer
Structural
Layer
Draw
Solution
Feed
Solution
π
D,Mem
Dilutive ECP
Concentrative
ECP
Cake
Layer
Greater
RSF
Cake enhanced
osmotic pressure
a) b) c)
π
F,bulk

π
F,mem
π
D,bulk
∆π
Eff.
J
w
π
D,i
π
F,bulk

π
F,mem
J
S
Selective
Layer
Structural
Layer
Draw
Solution
Feed
Solution
π
D,Mem
Dilutive ECP
Dilutive ICP
Concentrative
ECP
Foulants
ICP
self-compensation Dilutive ICP

79

and S solely under FO testing conditions may be more appropriate for predicting FO
performance. Recently, an alternative method utilizing FO experiments with stepwise increases
in draw solution concentration for determination of A, B, and S was developed by Tiraferri et al.
[106]. It is well known that for clean membranes with non-fouling feed solutions, increasing the
draw solution concentration results in greater water flux, greater RSF, and a consistent specific
reverse salt flux (SRSF; g/L); however, we have found this not to be true for fouled membranes
[38, 58]. While high A/B and low S indicate high selectivity and low propensity for dilutive ICP
[65] with non-fouling feed solutions, it is unclear what impact the membrane parameters have
fouled performance.
The objectives of this research were to determine if a limiting flux value and a corresponding
limiting osmotic pressure could be identified for fouled membranes with very different transport
and structural parameters. We also sought to determine what effect exceeding the limiting
osmotic pressure would have on RSF and membrane fouling. One CTA membrane and four
TFC membranes were tested. First, membrane water permeability (A), solute permeability (B),
and structural parameters (S) were obtained using both the conventional RO-FO method [105]
and the stepwise FO method developed by Tiraferri, et al. [106]. Next, a stepwise method of
successively increased draw solution osmotic pressures was used to measure baseline water
flux and baseline RSF. Then, stepwise experiments were repeated with membrane fouling using
a synthetic activated sludge feed solution to measure fouled water flux and fouled RSF. Results
from fouling experiments were used to calculate limiting flux values and limiting osmotic
pressures for each membrane. A resistance in series model was used to compare the relative
contributions of membrane resistance and foulant cake layer resistance. Additionally, a longer-
term fouling experiment with a constant draw solution osmotic pressure was used to validate
limiting flux values from the stepwise experiments.

80

5.2. Materials and methods
5.2.1. Membranes and bench-scale FO system
Five commercially available FO membranes were tested; a CTA and a TFC membrane from
Hydration Technology Innovations, LLC (Albany, OR), a TFC membrane from Porifera Inc.
(Hayward, CA), a TFC membrane from Oasys Water Inc. (Cambridge, MA), and a TFC
membrane from Toray Industries Inc. (Tokyo, Japan). All membranes were tested with the
selective layer facing the feed solution. Details of the bench-scale FO system used in this study
have been described previously [58]. Briefly, the membrane module contained three membrane
coupons, each with an effective area of 42 cm
2
, and each membrane coupon had independent
draw solution (1-L sidearm flasks) and feed solution (2-L tanks) reservoirs. 31-mil (0.79 mm)
diamond-style spacers (Sterlitech Corporation, Kent, WA) were used on the draw solution side
and no spacers were used on the feed solution side. Membranes were tested at 24 ± 1 °C with
a crossflow velocity of 10 cm/s. Analytical grade NaCl (VWR International, Radnor,
Pennsylvania) was used to prepare the draw solution. Conductivity probes (Cole-Parmer,
Vernon Hills, IL) in the draw solution reservoirs were used to monitor draw solution conductivity.
An automatic dosing system (Fig. 5.2) kept the draw solution reservoirs at constant
concentration; when conductivity fell below the set point the dosing system was activated and
292 g/L NaCl (5 M) solution was transferred to the appropriate draw solution reservoir until the
set point was reached. The mass of overflow from each draw solution reservoir was measured
with an analytical balance (PA3102, OHAUS Corporation, Parsippany, NJ) and used to
calculate water flux in 5-min increments. Feed solution conductivities were measured with a
conductivity probe (EC215, Hanna Instruments, Woonsocket, RI) and used to calculate RSF
with
𝑅𝑆𝐹=
𝐶
!
𝑉
!
−𝐶
!
𝑉
!
𝐴
!
𝛥𝑡
(1)
where C
i
and C
F
are initial and final NaCl feed concentrations (calculated from conductivity

81

measurements); V
i
and V
F
are initial and final feed volumes; A
M
is membrane area; and Δt is
elapsed time. Data acquisition and control devices (USB-6009, NI 9208, NI ER-8) were
connected to LabVIEW (National Instruments, Austin, TX) and used to record data and trigger
the dosing system. Error in water flux and RSF values was calculated from the standard
deviation between triplicate membrane coupons.

Figure 5.2. Graphical representation of the bench-scale FO testing system. Feed and draw
solutions were circulated counter-currently, feed solution was periodically replenished with DI
water, and draw solution was kept at constant concentration by dosing with 292 g/L (5 M) NaCl
solution. The membrane module contained three membrane coupons, each with an
independent feed solution and a dedicated draw solution dosing system; for simplicity, only one
testing system is depicted.

5.2.2. Membrane characterization
Water flux, RSF, and NaCl concentrations from four FO experiments were used to calculate
A, B, and S values with the FO characterization method [106]. The feed solution was dilute NaCl
(1.0 g/L; 0.1 MPa) and experiments with four draw solution concentrations corresponding to 0.5,

82

1, 2, and 3 MPa osmotic pressure were conducted for 30 minutes each. Osmotic pressures and
NaCl concentrations (Table 5.1) were calculated using OLI Stream Analyzer (OLI Systems, Inc.,
Morris Plains, NJ). Average water flux, RSF, and NaCl concentrations from the four experiments
were substituted into the governing solution-diffusion transport equations for water flux (Jw) and
RSF (Js) [106, 110]
𝐽
!
=𝐴
𝜋
!
exp −
𝐽
!
𝑆
𝐷
−𝜋
!
𝐽
!
𝑘
1+
𝐵
𝐽
!
exp
𝐽
!
𝑘
−exp −
𝐽
!
𝑆
𝐷
(2)
𝐽
!
=𝐵
𝐶
!
exp −
𝐽
!
𝑆
𝐷
−𝐶
!
𝐽
!
𝑘
1+
𝐵
𝐽
!
exp
𝐽
!
𝑘
–exp −
𝐽
!
𝑆
𝐷
(3)
where π
D
and π
F
are osmotic pressures of the bulk draw and bulk feed solutions; k is the mass
transfer coefficient of the membrane module [124], D is draw solute diffusivity; and C
D
and C
F

are NaCl concentrations of the bulk draw and bulk feed solutions. The system of eight transport
equations was fit to experimental data using a least squares non-linear regression with A, B,
and S as regression parameters. A, B, and S were determined by minimizing the global error
calculated from the sum of the offsets of non-linear equations [106].
Table 5.1. Osmotic pressures (MPa) and corresponding NaCl
draw solution concentrations (g/L).
Draw Solution Osmotic
Pressure (MPa)
Draw Solution
Concentration (g/L NaCl)
0.5 6.4
1 12.9
2 25.5
3 37.7
5 60.9
10 112.1
15 153.2
20 184.3

83

5.2.3. Synthetic activated sludge feed solution
A synthetic activated sludge feed solution (Table 5.2) was designed to simulate the fouling
potential and chemical environment of activated sludge in an OMBR treating domestic
wastewater. Sodium alginate, bovine serum albumin, and humic acid (Sigma Aldrich, St. Louis,
MO) were used as model foulants representing polysaccharides, protein, and NOM, respectively
[125, 126]. CaCl
2
was used to form cross-linked alginate gel layers that simulate the structure of
biologically formed foulant cake layers [127, 128]. The chemical environment of an OMBR was
simulated using ACS reagent grade ionic salts (Sigma Aldrich, St. Louis, MO); NH
4
Cl,
Ca(NO
3
)
2
4H
2
O, KH
2
PO
4
, and MgSO
4
were sources of ammonium, nitrate, phosphate, and
sulfate; NaHCO
3
and NaCl were sources of hardness and salinity. The feed solution was
prepared to final NH
4
+
, NO
3
-
, PO
4
3-
, SO
4
2-
, and HCO
3
-
concentrations representative of OMBR
systems treating domestic wastewater [5, 34, 40, 46, 66]. Ionic concentrations of the
counterions, Ca
2+
, K
+
, Mg
2+
, Na
+
, and Cl
-
were also representative of commonly reported values
in OMBRs [23, 40, 66].
Table 5.2. Concentrations of chemical components in synthetic activated sludge feed solution
Activated sludge component Chemical component Concentration (mg/L)
Polysaccharides Sodium alginate 4,000
Protein Bovine serum albumin 200
Naturally occurring organic matter Humic acid

200
Influent ammonia (NH 4
+
) NH 4Cl 40
Nitrification product (NO 3
-
) Ca(NO 3) 24H 2O 40
Phosphate (PO 4
3-
) KH 2PO 4 20
Sulfate (SO 4
2-
) MgSO 4 20
Hardness NaHCO 3 125
Salinity NaCl 50
*Biofouling; alginate cross-linking CaCl 2 54
*CaCl 2 was added to bring the total Ca
2+
concentration (2 mM) to facilitate alginate cross-linking.

84

Stock (100-mM) solutions of NH
4
Cl, Ca(NO
3
)
2
4H
2
O, KH
2
PO
4
, MgSO
4
, NaHCO
3
, NaCl, and
CaCl
2
were prepared. Four liters of concentrated (8 g/L) sodium alginate solution was mixed
overnight. Then, sufficient volumes of stock solutions required for a final 8-L volume of synthetic
feed solution were added to the sodium alginate solution and mixed for one hour. Subsequently,
solid bovine serum albumin and humic acid (1.6 g each) were added to the solution, the total
volume was brought to 8 L with deionized water, and the mixture was stirred for two additional
hours until all constituents were completely solubilized. Total solids and volatile solids were
analyzed in duplicate using Standard Methods 2540B and 2540E [88]. Total solids
concentration was 5.15 ± 0.04 g/L and volatile solids concentration was 2.49 ± 0.07 g/L.
Conductivity and pH were measured using a handheld probe (SevenGo Duo Pro, Mettler Toledo
Inc., Hightstown, NJ); conductivity and pH of the synthetic foulant solution were 1.7 mS/cm and
7.2.
5.2.4. Experimental procedures
5.2.4.1. Stepwise draw solution osmotic pressure experiments
A stepwise-style experiment with roots in MBR critical flux determination [92, 129-131],
direct microscopic observation of FO membrane fouling [69], and FO transport and structural
parameter determination [106] was designed to identify limiting flux values for fouled FO
membranes. Baseline water flux and RSF experiments were conducted using an NaCl feed
solution prepared to the same conductivity (1.7 mS/cm) as the synthetic foulant solution. NaCl
draw solution osmotic pressure was increased in a stepwise manner corresponding to 0.5, 1, 2,
3, 5, and 10 MPa. Each step was carried out for two hours and water flux values were
determined as in 5.2.1. Initial and final feed solution conductivity and mass values were
recorded and used to calculate RSF during each step (Eqn. 1).

85

Experiments were repeated with synthetic activated sludge feed solution and results were
used to calculate water flux and RSF under fouling conditions. Feed solution volumes
decreased during each step. This would cause the foulant concentration of the feed solution to
increase over time; however, this was prevented by refilling the feed solution reservoirs to the
initial volume (2 L) with deionized water between each step. This ensured that each step began
with the same foulant concentration in the feed solution.
5.2.4.2. Stepwise limiting flux and limiting osmotic pressure
For stepwise experiments, membranes were considered fouled when water flux declined by
at least 10% during a given step. Limiting flux values were calculated from a minimum of three
steps with membrane fouling. It was necessary to use higher draw solution osmotic pressures
(15 and 20 MPa) to observe three steps with membrane fouling for CTA membranes. Final
water flux was calculated over the last 15 minutes of each step. Limiting flux was the average of
the final water flux values from steps where membrane fouling was observed. For each
membrane, final water flux values in each step were plotted as a function of draw solution
osmotic pressure and a fourth order quadratic equation with an R
2
value of 0.99 or greater was
fit to experimental data. Limiting flux values for each membrane were used in the corresponding
quadratic equations to solve for the minimum osmotic pressure (limiting osmotic pressure)
required to achieve homoeostatic flux (see example in Fig. 5.3).

86

Figure 5.3. Method for determining limiting osmotic pressure. Final water flux in each step was
plotted as a function of draw solution osmotic pressure and a fourth order quadratic equation
(trendline) was fit to water flux values (open circles). By using the limiting water flux as the y
value in the quadratic equation (not shown), the limiting osmotic pressure (diamond) was
determined by solving for x.

5.2.4.3. Validation of stepwise limiting flux; constant draw solution osmotic pressure
A second set of fouling experiments was conducted to validate limiting flux values obtained
from the stepwise experiments. The validation experiments were carried out for 12 hours with a
constant draw solution osmotic pressure (constant π
d
) that was approximately twice that of the
calculated limiting osmotic pressure from the stepwise experiments. Draw solution conductivity
was monitored continuously and, if necessary, the dosing system was activated every 60
minutes. Feed solution reservoirs were replenished with deionized water every four hours to
prevent the foulant concentration in the feed solution from increasing over the duration of the
experiments.

87

5.2.4.4. Limiting flux with constant draw solution osmotic pressure
During experiments with constant draw solution osmotic pressure, flux decline occurred due
to 1) additional mass transfer resistance from foulant deposition, and 2) decreased driving force
from RSF into the feed solution. Flux decline from foulant deposition occurred until the cake
layer was fully formed. However, flux decline from RSF would continue until the system reaches
equilibrium. At the true equilibrium state, water and solute diffusion would occur in both
directions at the same rate (water flux and RSF equal zero). For this reason, the system was not
operated until the true equilibrium state was reached. Instead, a t-test comparing the rate of flux
decline between two successive 2-hr periods was performed to determine the significance of
flux decline. Water flux for each of the three coupons was plotted as a function of time and
slopes from linear regressions were calculated over discrete two-hour periods. Membrane
fouling was considered fully developed when the slopes between two successive two-hour
periods were statistically similar, indicating no significant difference in flux decline occurred
thereafter. The slopes of successive periods were compared using a paired t-test (n = 3
(triplicate coupons), α = 0.05, µ = 0). A t value below the critical value (t* = 2.92) indicated a
statistical similarity between the slopes for two successive periods. Water flux at the beginning
of the first two-hour period was taken as the limiting flux.
5.2.5. Membrane and foulant cake layer resistances
The relative contributions of membrane resistance and cake layer resistance in the stepwise
fouling experiments was calculated using the resistance-in-series model [98, 132]
𝐽
!
=
𝛥𝜋
𝜇𝑅
!
=
𝛥𝜋
𝜇 𝑅
!
+𝑅
!
(4)
where Δπ is osmotic pressure difference between bulk feed and bulk draw solutions; µ is
dynamic viscosity of water; R
m
is membrane resistance; R
c
is cake layer resistance; and R
t
is
total resistance. The resistance-in-series model was expanded to account for RSF,

88

concentrative external concentration polarization (ECP), and dilutive ICP using [62]
𝑅
!
=
1
𝜇𝐽
!
𝜋
!
𝑒
!!
!
!
−𝜋
!
𝑒
!
!
!
!
1−
𝐵
𝐽
!
𝑒
!!
!
!
−𝑒
!
!
!
!
(5)
where K is solute resistivity of the membrane support structure and k
f
is the mass transfer
coefficient of the feed channel. R
m
was calculated using Eqn. (5) with results from baseline
experiments, R
t
was calculated using Eqn. (5) with results from fouling experiments, and R
f
was
the difference between R
t
and R
m
. K was calculated using [105, 107, 108]
𝐾=
𝑆
𝐷
=
1
𝐽
!
ln
𝐵+𝐴𝜋
!
𝐵+ 𝐽
!
+𝐴𝜋
!,!
(6)
where πF,m is osmotic pressure at the membrane surface on the feed solution side which was
calculated using the concentrative ECP modulus [52].
5.3. Result and discussion
5.3.1. Transport and structural parameters
A, B, and S values determined using the FO characterization method are given in Table 5.3
for the five membranes. The A/B ratio is also shown to indicate selectivity of the membranes,
with higher A/B indicating higher selectivity and lower A/B indicating lower selectivity [22, 113].
CTA membranes exhibited the lowest A/B, and A/B increased in the order of: HTI TFC, Oasys
TFC, Porifera TFC, and Toray TFC membranes. Although FO membranes are assumed to be
non-porous, it was previously suggested that defects in the selective layer result in a
combination of pore flow and diffusion when hydraulic pressure is applied (i.e., with the
conventional RO-FO membrane characterization method) [112]. Furthermore, compaction of the
porous support structure under hydraulic pressure may reduce permeability by increasing the
membrane density. Thus, discrepancies between A, B, and S values reported here and

89

literature values obtained with the conventional RO-FO method are likely due to the different
driving forces in RO (hydraulic pressure) and FO (osmotic pressure) [106, 112].
Table 5.3. Water permeability (A), solute permeability (B), and structural parameter (S)
determined using the FO characterization method.
Membrane A (L m
-2
h
-1
bar
-1
) B (L m
-2
h
-1
) S (µm) A/B (bar
-1
)
HTI CTA 0.58 ± 0.05 0.49 ± 0.06 385 ± 18 1.18
HTI TFC 2.78 ± 0.21 1.36 ± 0.14 513 ± 36 2.04
Oasys TFC 5.26 ± 0.88 2.44 ± 0.68 554 ± 50 2.16
Porifera TFC 3.35 ± 0.51 1.18 ± 0.18 398 ± 35 2.84
Toray TFC 2.29 ± 0.34 0.30 ± 0.08 198 ± 44 7.63

5.3.2. Stepwise experiments; limiting water flux and limiting osmotic pressure
Water flux as a function of time is shown in Figs. 5.4a-e for experiments with stepwise
increases in draw solution osmotic pressure (solid green line). Baseline water flux (dashed blue
line) is with sodium chloride feed solution and water flux from fouling experiments (open blue
circles) is with synthetic activated sludge feed solution. For each membrane, baseline water flux
increased with each osmotic pressure step. Baseline water flux was higher for membranes with
successively higher A/B values (Toray TFC > Porifera TFC > Oasys TFC > HTI TFC > HTI CTA)
and it was relatively constant during each step. Water flux with activated sludge was constant
during steps with low draw solution osmotic pressures (0.1 and 1.0 MPa), but decreased over
the duration of steps at higher draw solution osmotic pressures. This observation indicates that
membrane fouling only occurred when water flux exceeded a certain value. This value, termed
the limiting flux, is the maximum steady-state water flux that is obtainable in the presence of
foulants, which cannot be overcome by increasing the osmotic driving force. For all TFC
membranes (Figs. 5.4b-e), water flux with activated sludge feed solution declined during the last
four steps (2, 3, 5, and 10 MPa). Due to the comparatively lower water flux of the HTI CTA
membrane (Fig. 5.4a), water flux decline was not observed until higher draw solution osmotic

90

pressures (10, 15, and 20 MPa) were reached. Interestingly, for the CTA membrane, for steps
when flux decline from fouling occurred (between 10 and 20 MPa), final water flux (Fig. 5.4a,
dark circles) varied by less than 0.5 L m
-2
h
-1
. The same convergence was observed for all TFC
membranes (Figs. 5.4b-e, dark circles). In fact, the final flux values were within standard
deviation for all membranes except the Toray TFC (Fig 5.4f). These observations indicate that
the increases in draw solution osmotic pressure at each step were completely offset by flux
decline from membrane fouling; as a result, final water flux tended toward a singular value: the
limiting flux.

Figure 5.4. Baseline water flux, water flux with synthetic activated sludge feed solution, and
final water flux values for fouled membranes with stepwise increases in draw solution osmotic
pressure for (a) HTI CTA, (b) HTI TFC, (c) Oasys TFC, (d) Porifera TFC, and (e) Toray TFC
membranes. Also, (f) stepwise limiting flux and limiting osmotic pressure for each membrane.
Error bars represent the standard deviation between triplicate membrane coupons.

91

This observation is consistent with our understanding of membrane fouling; after initial
foulant deposition (driven by foulant-membrane interactions), only foulant-foulant adhesion,
convective water flux toward the membrane surface (drag force), and operational parameters
(crossflow velocity and pressure) dictate foulant deposition and the corresponding hydraulic
resistance [45, 133]. Initially, membrane-foulant interactions influence foulant deposition [134],
thus, the onset of flux decline from fouling is dependent on membrane parameters and
morphology. However, similar to the “limiting flux” of RO and NF membranes [44, 45], FO water
flux is eventually independent of membrane properties and membrane fouling limits water flux to
the limiting flux. Thus, the limiting flux is a function of adhesion forces, drag forces, and
operational parameters. As can be seen by the data in Fig. 5.4f, limiting flux values were within
standard deviation for HTI CTA, HTI TFC, Oasys TFC and Porifera TFC membranes; the
average value for these four membranes was 13.90 ± 0.54 L m
-2
h
-1
. However, Toray TFC
membranes had a higher limiting flux value (17.7 ± 0.66 L m
-2
h
-1
). This is likely due to the Toray
TFC membranes having higher initial water flux during each step with activated sludge feed
solution compared to all other membranes. Higher initial water flux causes more foulant
deposition [60]. Thus, it is likely that foulant deposition, and the resulting flux decline, were not
complete when the experiment was terminated due to high initial flux with Toray TFC
membranes.
The symbols in Fig. 5.4f represent limiting osmotic pressures for the five membranes.
Limiting osmotic pressure was between 2.2 and 2.5 MPa for all TFC membranes and 8.6 MPa
for HTI CTA membranes. Thus, CTA and TFC membranes required significantly different draw
solution concentrations (30.3 ± 2.4 vs. 111 ± 0.1 g/L NaCl) to achieve limiting flux. This is not
unexpected because higher permeability TFC membranes require lower osmotic pressure to
achieve similar water fluxes as lower permeability CTA membranes [64, 135]. The lower limiting

92

osmotic pressure of the TFC membranes is advantageous due to reduced requirements for
draw solute replenishment, although the need to replenish the draw solute also depends on
losses due to RSF [24].
5.3.3. Effect of osmotic pressure increases on RSF
For the same membranes, water flux and RSF were evaluated as a function of increasing
draw solution osmotic pressure (Fig. 5.5). For non-fouling feed solution (dilute NaCl), baseline
water flux (solid blue lines) and baseline RSF (solid red lines) increase proportionally with draw
solution osmotic pressure [12, 24]. However, when tested with activated sludge feed solution,
water flux and RSF increase disproportionately with draw solution osmotic pressure [34, 58, 66];
water flux (dashed blue lines) increases prior to reaching a maximum value (the limiting flux)
and RSF (dashed red lines) increases continually with draw solution osmotic pressure. For the
HTI CTA and Toray TFC membranes, RSF with activated sludge feed solution was greater than
or equal to baseline RSF. This may have been caused by electrostatic attraction between the
negatively charged foulant layers and positively charged draw solution cations (sodium) [34, 58,
66]. Conversely, RSF with activated sludge feed solution was less than or equal to baseline
RSF at lower draw solution osmotic pressures (0.5 to 5 MPa) for HTI TFC, Oasys TFC, and
Porifera TFC membranes. Lower RSF with membrane fouling may be attributed to “fouling
reduced concentration polarization” [31, 136], a phenomenon where hydraulic resistance from
the foulant cake layer improves NaCl rejection and results in lower RSF. However, even for
these membranes, RSF with activated sludge feed solution increases continually with draw
solution osmotic pressure whereas water flux is limited to limiting flux.

93

Figure 5.5. Baseline water flux, baseline RSF, water flux with synthetic activated sludge feed
solution, and RSF with A.S. as a function of stepwise increases in draw solution osmotic
pressure for (a) HTI CTA, (b) HTI TFC, (c) Oasys TFC, (d) Porifera TFC, and (e) Toray TFC
membranes. Error bars represent standard deviations between triplicate membrane coupons.

The membranes with the lowest B values (HTI CTA and Toray TFC membranes) had
significantly higher RSF at higher osmotic pressures with activated sludge feed solution than
membranes with higher B values (HTI TFC, Oasys TFC and Porifera TFC membranes). For the
CTA membrane, higher RSF is attributable to low A/B and a higher limiting osmotic pressure to
achieve similar water flux as TFC membranes. However, the membrane with the highest A/B,
the Toray TFC membrane, also had significantly higher RSF at higher osmotic pressures ( > 5
MPa) with activated sludge feed solution compared to the other TFC membranes with lower
A/B. This disparity between B, A/B values, and RSF may be because the HTI CTA and Toray

94

TFC membranes also have the lowest S values (385 ± 18 and 198 ± 44 µm). Low S indicates a
shorter diffusive path length through the support structure and a low propensity for dilutive ICP
[65], which is calculated from the dilutive ICP modulus [52]
𝜋
!,!
=𝜋
!
exp −
!
!
!
!
(11)
where πD,i is the solute concentration at the selective layer/support layer interface. According do
Eqn. (11), for a fixed J
w
, πD,i increases exponentially as S decreases and πD,i increases linearly
as πD
increases [22]. At the limiting flux, increasing πD does not increase water flux (fixed Jw).
Thus, at or above the limiting osmotic pressure, πD,i is solely dependent on S. RSF increases
with πD,i because rejection depends on solute concentration at the selective layer surface.
Consequently, RSF increases more severely with πD for membranes with small S than for
membranes with large S. Higher S values result in lower RSF, however, dilutive ICP associated
with higher S is recognized as the major impediment to achieving high FO water flux [64, 124].
This suggests achieving low RSF requires not only low B and high A/B, but that moderate S
values (~250 to 500 µm) can also be beneficial for applications where membrane fouling occurs.
5.3.4. Validation of stepwise limiting flux
To validate the limiting flux values obtained from the stepwise experiments, a second set of
experiments was conducted. These experiments, also with activated sludge feed solution, were
run with constant draw solution osmotic pressures that were approximately twice that of the
limiting osmotic pressures calculated from stepwise experiments. 15 MPa was used with the
HTI CTA membrane and 5 MPa was used with the TFC membranes. This was done to ensure
that initial fluxes were significantly higher than the limiting fluxes obtained from the stepwise
experiments and to induce flux decline from membrane fouling. Water flux as a function of time
for all membranes is shown in Fig. 5.6. Over time, water flux declined and tended toward the
singular, limiting flux (13.80 ± 0.35 L m
-2
h
-1
). P values (Fig. 5.6 inset) represent the significance

95

of flux decline between each successive 2-hour period. P values less than 0.05 (red values)
indicate flux decline was significant and P values greater than 0.05 (black values) indicate flux
decline was not significant. It is only the periods with significant flux decline for each membrane
that are shown in Fig. 5.6 (e.g., 0 to 6 hours for CTA membrane and 0 to 8 hours for Toray TFC
membrane). The average of the final water flux values for all membranes (the limiting flux) was
13.80 ± 0.35 L m
-2
h
-1
. Remarkably, for each membrane, the percent difference between the
limiting flux value and the average value was less 3.0%.

Figure 5.6. Water flux for HTI CTA, HTI TFC, Oasys TFC, Porifera TFC, and Toray TFC
membranes with synthetic activated sludge feed solution. Continuous draw solution osmotic
pressures of 15 MPa for HTI CTA membrane and 5 MPa for all TFC membranes were used. P-
values represent significance of flux decline between successive 2-hr periods. P ≥ 0.05
indicates no further significant flux decline was observed.

As can be seen by the data in Table 5.4, comparison of the limiting flux values from the
constant osmotic pressure and stepwise osmotic pressure methods shows less than 5%
difference for the HTI CTA, HTI TFC, Oasys TFC, and Porifera TFC membranes. For the Toray

96

membrane, the limiting flux from the constant osmotic pressure method was 22.7% lower than
with the stepwise osmotic pressure method. Membrane fouling was not fully developed until
after 8 hrs of operation with the constant osmotic pressure method for Toray TFC membranes
whereas all other membranes required 6 hrs or less. These observations support our earlier
conclusion (5.3.2) that flux decline from fouling was not complete with the stepwise method for
the Toray TFC membrane. However, water flux tended toward a singular value for all
membranes with the constant osmotic pressure method.
Table 5.4. Limiting flux values (J w,h) calculated from stepwise and constant draw solution
osmotic pressure (π D) experiments.
Membrane
J w,h (L m
-2
h
-1
);
Stepwise π D
J w,h (L m
-2
h
-1
);
Constant π D
% Difference
between methods
HTI CTA 14.30 ± 0.77 14.22 ± 0.14 0.54
HTI TFC 14.40 ± 0.45 13.70 ± 0.37 4.99
Oasys 13.31 ± 0.80 13.57 ± 0.16 1.92
Porifera 13.59 ± 0.45 13.41 ± 0.27 1.31
Toray 17.73 ± 0.66 14.12 ± 0.27 22.68
Average 13.90 ± 0.54* 13.80 ± 0.35 0.68
*Does not include Toray TFC membrane from stepwise method.

5.3.5. Membrane and cake layer resistances
Results from stepwise experiments were used along with transport and structural
parameters to calculate resistances (Eqn. 9). The data in Figs. 5.7a-e show resistance as a
function of draw solution osmotic pressure. R
m
was lower for membranes with higher A values.
R
m
was highest ( > 6.7 * 10
14
m
-1
) for the HTI CTA membrane, lower ( < 1.8 * 10
14
m
-1
) for the
HTI TFC, Porifera TFC, and Toray TFC membranes, and lowest (< 0.8 * 10
14
m
-1
) for the Oasys
TFC membranes. For all membranes, R
f
was near zero with low draw solution osmotic
pressures. R
f
increased as membrane fouling occurred (10 MPa for HTI CTA membranes, 2

97

MPa for TFC membranes). Eventually, R
f
surpasses R
m
, indicating that the foulant layer was
exerting a greater resistance to water flux than the membrane itself.

Figure 5.7. Total resistance (R
t
), membrane resistance (R
m
), and foulant cake layer resistance
(R
f
) as a function of draw solution osmotic pressure for (a) HTI CTA, (b) HTI TFC, (c) Oasys
TFC, (d) Porifera TFC, and (e) Toray TFC membranes. (f) Foulant cake layer resistance as a
function of draw solution osmotic pressure for all membranes.

The data in Fig. 5.7f compares R
f
for all membranes. Two key observations can be made: 1)
R
f
continually increases with draw solution osmotic pressure, and 2) membranes with higher A/B
have higher R
f
. These observations support our hypothesis that, under fouling conditions, high
water flux afforded by high water permeability is offset by additional hydraulic resistance due to
membrane fouling. Thus, for osmotic processes where membrane fouling occurs, membrane

98

transport parameters A and B may not be accurate performance indicators and obtaining high
water flux by using high permeability, high selectivity membranes may not be effective. Instead,
optimizing operational parameters such as fouling mitigation strategies (e.g., air scour, hydraulic
scour, and osmotic backwashing), hydrodynamics at the membrane surface (e.g., spacer or
module design), and membrane module configuration (e.g., flat-sheet or tubular membranes)
may be more viable alternatives toward improved performance for fouling applications.
5.4. Conclusions
FO experiments with stepwise increases in draw solution osmotic pressure revealed
baseline water flux increased and baseline RSF decreased according to increasing A/B values;
membranes with higher A/B values exhibited higher baseline water flux and lower baseline RSF
compared to membranes with lower A/B values. However, when stepwise experiments were
repeated under fouling conditions, there was an upper limit to water flux, termed the limiting flux,
that could not be overcome by increasing the osmotic pressure beyond the corresponding
minimum osmotic pressure (limiting osmotic pressure). Instead, further increases in draw
solution osmotic pressure only resulted in higher RSF. Limiting flux values with the stepwise
method were validated using a constant draw solution osmotic pressure that was greater than
the limiting osmotic pressure. Limiting flux values from the constant osmotic pressure
experiments were similar for all membranes despite very different A/B values, indicating that A
and A/B are poor indicators of water flux once membrane fouling occurs. Interestingly, achieving
low RSF with membrane fouling was dependent on S. This dependency was due to higher
osmotic pressure at the selective layer/structural layer interface for very low S values compared
to higher S values.
A resistance in series model revealed that hydraulic resistance of the foulant cake layer
surpassed membrane resistance at high draw solution osmotic pressures. Additionally, cake
layer resistance was higher for membranes with higher A/B. These results confirm that a

99

dynamic constancy exists for osmotic processes; when membrane fouling occurs, hydraulic
resistance of the cake layer limits water flux to a singular value regardless of membrane
selectivity and structural parameters within the ranges tested here. This suggests other means
of increasing water flux such as high-efficiency membrane cleaning procedures and improved
membrane module configurations may be more promising avenues than improved membrane
parameters when membrane fouling is anticipated.

100

Chapter 6. Conclusions
6.1. Research Synopsis
This dissertation represents several investigations into membrane fouling with FO
membranes and the use of FO membranes for wastewater treatment and potable reuse. These
investigations include: (1) integration of OMBR and MD membrane systems, (2) the role of FO
membrane module configuration on water flux, RSF, and membrane fouling, (3) the effects of
FO membrane compaction on transport and structural parameters, and (4) the limiting flux
phenomenon extended to incorporate membrane fouling and concentration polarization with FO
membranes.
6.1.1. Summary of OMBR-MD system integration
An integrated OMBR-MD system was tested at the pilot-scale for wastewater treatment and
production of high-quality reuse water. During membrane selection, no significant difference in
water flux between CTA and TFC membrane was observed when activated sludge feed was the
feed solution. After long-term operation, FO water flux with CTA membrane was the same for 20
g/L and 35 g/L NaCl draw solution, although bioreactor conductivity increased when 35 g/L was
used due to greater RSF. During long-term OMBR operation, carbon and nitrogen removal was
achieved in a single-reactor by alternating between aerobic and anoxic bioreactor conditions.
The OMBR was coupled to DCMD with a three-tank configuration and automated controls for
transferring solutions between FO and MD processes. This allowed for continuous water
production by FO and MD without transferring heat to the bioreactor, which is critical for
biological nitrogen removal. When treating a high-strength wastewater feed solution, the
integrated OMBR-MD system achieved up to 91% rejection of both NH
4
+
-N and COD. Similar
productivity (1-1.4 L h
-1
) for both FO and MD processes was achieved by adjusting the FO
membrane area. While the current work has demonstrated a novel method for integration of
OMBR and MD systems for wastewater reuse at the pilot-scale, future efforts should focus on

101

optimization of the overall system performance, specifically with regards to the low FO and MD
water fluxes observed here.
6.1.2. Summary of effects of forward osmosis membrane module configuration
Overall, the initial water flux and fouled water flux with time was the same for submerged
and sidestream configurations. Furthermore, the steady-state water flux of the fouled
membranes was the same for both (submerged and sidestream) configurations and both draw
solution concentrations, leading to the concept of a limiting flux in OMBRs similar to the critical
flux in conventional MBRs. However, for fouled membranes, the sidestream configuration
resulted in lower SRSF at steady-state with both draw solution concentrations. The 100 g/L
NaCl draw solution experiments did not result in higher steady-state water flux for the fouled
membranes despite the significant increase in driving force; instead, the higher draw solution
concentration did result in the formation of thicker fouling layers with NaCl scaling. Thinner cake
layers were formed in the sidestream configuration where fouling is mitigated with hydraulic
crossflow compared to the submerged configuration where fouling is mitigated via air scour.
Hydraulic pressure from recirculation pumping on the feed side of the sidestream configuration
may have also resulted in a more compact cake layer over time. For the experimental conditions
tested in this study, model results predicted that the sidestream configuration would result in
lower steady-state OMBR salinity, particularly when longer SRTs required for nitrogen removal
are used.
Lower SRSF in the sidestream configuration is advantageous, although it comes at the cost
of additional recirculation pumping, which can add significantly to specific energy consumption
(kWh m
-3
) [103]. On the other hand, in the submerged configuration, the aeration required to
achieve air scour exceeds that required to supply the biological community with oxygen [38,
104]. How much additional aeration is required depends on membrane length, bubble size, and
width of the bubble channel [7]. In the end, additional energy requirements for operation (e.g.,

102

recirculation pumping) as well as for fouling mitigation (e.g., increased crossflow velocity,
increased aeration rate, chemical cleaning, and osmotic backwashing) must be considered to
comprehensively assess the tradeoffs between FO module configurations.
6.1.3. Summary of the effects of hydraulic pressure on forward osmosis membranes
SEM analysis revealed that CTA membranes resist compaction under hydraulic pressure
(up to 435 psi), leading to similar A and B values with both methods. Conversely, TFC
membranes underwent severe compaction during RO characterization. Collapsed macro-void
and finger-like pore structures lead to decreased solute permeability and increased selectivity
with the HTI TFC membrane during RO characterization. For the Toray TFC membrane,
collapsed macro-void spaces with intact finger-like structures increased permeability during RO
characterization. Model FO water flux was accurate using parameters obtained with either
characterization method. However, model RSF was more accurate using parameters obtained
with the FO characterization method than the RO-FO method. These results suggest that FO
membranes which are expected to be subjected to a combination of hydraulic and osmotic
pressures during scaled-up applications may undergo compaction and longer-term studies are
required.
6.1.4. Summary of the limiting flux phenomenon in forward osmosis
FO experiments with stepwise increases in draw solution osmotic pressure revealed
baseline water flux increased and baseline RSF decreased according to increasing A/B values;
membranes with higher A/B values exhibited higher baseline water flux and lower baseline RSF
compared to membranes with lower A/B values. However, when stepwise experiments were
repeated under fouling conditions, there was an upper limit to water flux, termed the limiting flux,
that could not be overcome by increasing the osmotic pressure beyond the corresponding
minimum osmotic pressure (limiting osmotic pressure). Instead, further increases in draw

103

solution osmotic pressure only resulted in higher RSF. Limiting flux values with the stepwise
method were validated using a constant draw solution osmotic pressure that was greater than
the limiting osmotic pressure. Limiting flux values from the constant osmotic pressure
experiments were similar for all membranes despite very different A/B values, indicating that A
and A/B are poor indicators of water flux once membrane fouling occurs. Interestingly, achieving
low RSF with membrane fouling was dependent on S. This dependency was due to higher
osmotic pressure at the selective layer/structural layer interface for very low S values compared
to higher S values.
A resistance in series model revealed that hydraulic resistance of the foulant cake layer
surpassed membrane resistance at high draw solution osmotic pressures. Additionally, cake
layer resistance was higher for membranes with higher A/B. These results confirm that a
dynamic constancy exists for osmotic processes; when membrane fouling occurs, hydraulic
resistance of the cake layer limits water flux to a singular value regardless of membrane
selectivity and structural parameters within the ranges tested here. This suggests other means
of increasing water flux such as high-efficiency membrane cleaning procedures and improved
membrane module configurations may be more promising avenues than improved membrane
parameters when membrane fouling is anticipated.

104

References

[1] R.L. McGinnis, M. Elimelech, Global challenges in energy and water supply: the promise of
engineered osmosis, in, ACS Publications, 2008.
[2] T. Stephenson, S. Judd, B. Jefferson, K. Brindle, Membrane bioreactors for wastewater treatment,
IWA Publishing, 2000.
[3] F.I. Hai, K. Yamamoto, C.-H. Lee, Membrane Biological Reactors, IWA Publishing, 2013.
[4] J. Ottoson, A. Hansen, B. Bjorlenius, H. Norder, T.A. Stenstrom, Removal of viruses, parasitic protozoa
and microbial indicators in conventional and membrane processes in a wastewater pilot plant, Water
Res, 40 (2006) 1449-1457.
[5] A. Achilli, T.Y. Cath, E.A. Marchand, A.E. Childress, The forward osmosis membrane bioreactor: A low
fouling alternative to MBR processes, Desalination, 239 (2009) 10-21.
[6] E. Cornelissen, D. Harmsen, E. Beerendonk, J. Qin, H. Oo, K. De Korte, J. Kappelhof, The innovative
osmotic membrane bioreactor (OMBR) for reuse of wastewater, Water Science and Technology, 63
(2011) 1557-1565.
[7] R.W. Holloway, A. Achilli, T.Y. Cath, The osmotic membrane bioreactor: a critical review, Environ. Sci.:
Water Res. Technol., 1 (2015) 581-605.
[8] T. Cath, A. Childress, M. Elimelech, Forward osmosis: Principles, applications, and recent
developments, Journal of Membrane Science, 281 (2006) 70-87.
[9] J.L. Cartinella, T.Y. Cath, M.T. Flynn, G.C. Miller, K.W. Hunter, A.E. Childress, Removal of Natural
Steroid Hormones from Wastewater Using Membrane Contactor Processes, Environmental Science and
Technology, 40 (2006) 7381-7386.
[10] T.Y. Cath, S. Gormly, E.G. Beaudry, M.T. Flynn, V.D. Adams, A.E. Childress, Membrane contactor
processes for wastewater reclamation in space. I. Direct osmotic concentration as pretreatment for
reverse osmosis, Journal of Membrane Science, 257 (2005) 85-98.
[11] R.J. York, R.S. Thiel, E.G. Beaudry, Full-scale experience of direct osmosis concentration applied to
leachate management, in: Sardinia ’99 Seventh International Waste Management and Landfill
Symposium, S. Margherita di Pula, Cagliari, Sardinia, Italy, 1999.
[12] A. Achilli, T.Y. Cath, A.E. Childress, Selection of inorganic-based draw solutions for forward osmosis
applications, Journal of Membrane Science, 364 (2010) 233-241.
[13] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, A forward osmosis-membrane distillation hybrid
process for direct sewer mining: system performance and limitations, Environ Sci Technol, 47 (2013)
13486-13493.
[14] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Comparison of the removal of hydrophobic trace
organic contaminants by forward osmosis and reverse osmosis, Water Res, 46 (2012) 2683-2692.
[15] A.A. Alturki, J.A. McDonald, S.J. Khan, W.E. Price, L.D. Nghiem, M. Elimelech, Removal of trace
organic contaminants by the forward osmosis process, Separation and Purification Technology, 103
(2013) 258-266.
[16] M. Xie, W.E. Price, L.D. Nghiem, Rejection of pharmaceutically active compounds by forward
osmosis: Role of solution pH and membrane orientation, Separation and Purification Technology, 93
(2012) 107-114.
[17] W.C. Lay, T.H. Chong, C.Y. Tang, A.G. Fane, J. Zhang, Y. Liu, Fouling propensity of forward osmosis:
investigation of the slower flux decline phenomenon, Water Sci Technol, 61 (2010) 927-936.
[18] W. Luo, H.V. Phan, M. Xie, F.I. Hai, W.E. Price, M. Elimelech, L.D. Nghiem, Osmotic versus
conventional membrane bioreactors integrated with reverse osmosis for water reuse: Biological
stability, membrane fouling, and contaminant removal, Water Res, 109 (2017) 122-134.

105

[19] A. Schuster, S. Karellas, E. Kakaras, H. Spliethoff, Energetic and economic investigation of Organic
Rankine Cycle applications, Applied thermal engineering, 29 (2009) 1809-1817.
[20] H. Luo, Q. Wang, T.C. Zhang, T. Tao, A. Zhou, L. Chen, X. Bie, A review on the recovery methods of
draw solutes in forward osmosis, Journal of Water Process Engineering, 4 (2014) 212-223.
[21] N.T. Hancock, T.Y. Cath, Solute coupled diffusion in osmotically driven membrane processes,
Environmental science & technology, 43 (2009) 6769-6775.
[22] W.A. Phillip, J.S. Yong, M. Elimelech, Reverse draw solute permeation in forward osmosis: modeling
and experiments, Environmental science & technology, 44 (2010) 5170-5176.
[23] W.C.L. Lay, Q. Zhang, J. Zhang, D. McDougald, C. Tang, R. Wang, Y. Liu, A.G. Fane, Study of
integration of forward osmosis and biological process: Membrane performance under elevated salt
environment, Desalination, 283 (2011) 123-130.
[24] K.S. Bowden, A. Achilli, A.E. Childress, Organic ionic salt draw solutions for osmotic membrane
bioreactors, Bioresour Technol, 122 (2012) 207-216.
[25] G. Qiu, Y.P. Ting, Osmotic membrane bioreactor for wastewater treatment and the effect of salt
accumulation on system performance and microbial community dynamics, Bioresour Technol, 150
(2013) 287-297.
[26] X. Wang, Y. Chen, B. Yuan, X. Li, Y. Ren, Impacts of sludge retention time on sludge characteristics
and membrane fouling in a submerged osmotic membrane bioreactor, Bioresour Technol, 161 (2014)
340-347.
[27] A. Uygur, Specific nutrient removal rates in saline wastewater treatment using sequencing batch
reactor, Process Biochemistry, 41 (2006) 61-66.
[28] L. Ye, C.-y. Peng, B. Tang, S.-y. Wang, K.-f. Zhao, Y.-z. Peng, Determination effect of influent salinity
and inhibition time on partial nitrification in a sequencing batch reactor treating saline sewage,
Desalination, 246 (2009) 556-566.
[29] S. Aslan, E. Simsek, Influence of salinity on partial nitrification in a submerged biofilter, Bioresource
technology, 118 (2012) 24-29.
[30] A. Dincer, F. Kargi, Salt inhibition of nitrification and denitrification in saline wastewater,
Environmental Technology, 20 (1999) 1147-1153.
[31] Q. She, R. Wang, A.G. Fane, C.Y. Tang, Membrane fouling in osmotically driven membrane
processes: A review, Journal of Membrane Science, 499 (2016) 201-233.
[32] J.J. Qin, K.A. Kekre, M.H. Oo, G. Tao, C.L. Lay, C.H. Lew, E.R. Cornelissen, C.J. Ruiken, Preliminary
study of osmotic membrane bioreactor: effects of draw solution on water flux and air scouring on
fouling, Water Sci Technol, 62 (2010) 1353-1360.
[33] B. Mi, M. Elimelech, Organic fouling of forward osmosis membranes: Fouling reversibility and
cleaning without chemical reagents, Journal of Membrane Science, 348 (2010) 337-345.
[34] E. Cornelissen, D. Harmsen, K. Dekorte, C. Ruiken, J. Qin, H. Oo, L. Wessels, Membrane fouling and
process performance of forward osmosis membranes on activated sludge, Journal of Membrane
Science, 319 (2008) 158-168.
[35] A. Alturki, J. McDonald, S.J. Khan, F.I. Hai, W.E. Price, L.D. Nghiem, Performance of a novel osmotic
membrane bioreactor (OMBR) system: flux stability and removal of trace organics, Bioresour Technol,
113 (2012) 201-206.
[36] B. Yuan, X. Wang, C. Tang, X. Li, G. Yu, In situ observation of the growth of biofouling layer in
osmotic membrane bioreactors by multiple fluorescence labeling and confocal laser scanning
microscopy, Water Res, 75 (2015) 188-200.
[37] X. Wang, Y. Zhao, B. Yuan, Z. Wang, X. Li, Y. Ren, Comparison of biofouling mechanisms between
cellulose triacetate (CTA) and thin-film composite (TFC) polyamide forward osmosis membranes in
osmotic membrane bioreactors, Bioresour Technol, 202 (2016) 50-58.

106

[38] W. Luo, F.I. Hai, W.E. Price, L.D. Nghiem, Water extraction from mixed liquor of an aerobic
bioreactor by forward osmosis: Membrane fouling and biomass characteristics assessment, Separation
and Purification Technology, 145 (2015) 56-62.
[39] J. Zhang, W.L.C. Loong, S. Chou, C. Tang, R. Wang, A.G. Fane, Membrane biofouling and scaling in
forward osmosis membrane bioreactor, Journal of Membrane Science, 403-404 (2012) 8-14.
[40] R.W. Holloway, A.S. Wait, A. Fernandes da Silva, J. Herron, M.D. Schutter, K. Lampi, T.Y. Cath, Long-
term pilot scale investigation of novel hybrid ultrafiltration-osmotic membrane bioreactors,
Desalination, 363 (2015) 64-74.
[41] Y. Kim, M. Elimelech, H.K. Shon, S. Hong, Combined organic and colloidal fouling in forward osmosis:
Fouling reversibility and the role of applied pressure, Journal of Membrane Science, 460 (2014) 206-212.
[42] S. Lee, C. Boo, M. Elimelech, S. Hong, Comparison of fouling behavior in forward osmosis (FO) and
reverse osmosis (RO), Journal of Membrane Science, 365 (2010) 34-39.
[43] M. Xie, J. Lee, L.D. Nghiem, M. Elimelech, Role of pressure in organic fouling in forward osmosis and
reverse osmosis, Journal of Membrane Science, 493 (2015) 748-754.
[44] C.Y. Tang, J.O. Leckie, Membrane independent limiting flux for RO and NF membranes fouled by
humic acid, Environmental science & technology, 41 (2007) 4767-4773.
[45] C.Y. Tang, Y.-N. Kwon, J.O. Leckie, The role of foulant–foulant electrostatic interaction on limiting
flux for RO and NF membranes during humic acid fouling—theoretical basis, experimental evidence, and
AFM interaction force measurement, Journal of Membrane Science, 326 (2009) 526-532.
[46] C.P. Morrow, N.M. Furtaw, J.R. Murphy, A. Achilli, E. Marchand, S.R. Hiibel, A.E. Childress,
Integrating an aerobic/anoxic osmotic membrane bioreactor with membrane distillation for potable
reuse, Desalination, 432 (2018) 46-54.
[47] R.J. Zeng, R. Lemaire, Z. Yuan, J. Keller, Simultaneous nitrification, denitrification, and phosphorus
removal in a lab‐scale sequencing batch reactor, Biotechnology and bioengineering, 84 (2003) 170-178.
[48] E.V. Münch, P. Lant, J. Keller, Simultaneous nitrification and denitrification in bench-scale
sequencing batch reactors, Water Research, 30 (1996) 277-284.
[49] M.A. Shahzad, S.J. Khan, M.S. Siddique, Draw solution recovery using direct contact membrane
distillation (DCMD) from osmotic membrane bioreactor (Os-MBR), Journal of Water Process
Engineering, (2017).
[50] M.S. Siddique, S.J. Khan, M.A. Shahzad, M.S. Nawaz, N.P. Hankins, Insight into the Effect of Organic
and Inorganic Draw Solutes on the Flux Stability and Sludge Characteristics in the Osmotic Membrane
Bioreactor, Bioresource Technology, (2017).
[51] W. Luo, H.V. Phan, G.X. Li, F.I. Hai, W.E. Price, M. Elimelech, L.D. Nghiem, An Osmotic Membrane
Bioreactor–Membrane Distillation System for Simultaneous Wastewater Reuse and Seawater
Desalination: Performance and Implications, Environmental science & technology, (2017).
[52] J.R. McCutcheon, M. Elimelech, Influence of concentrative and dilutive internal concentration
polarization on flux behavior in forward osmosis, Journal of Membrane Science, 284 (2006) 237-247.
[53] E.R. Cornelissen, D. Harmsen, E.F. Beerendonk, J.J. Qin, H. Oo, K.F. de Korte, J.W.M.N. Kappelhof,
The innovative Osmotic Membrane Bioreactor (OMBR) for reuse of wastewater, Water Science &
Technology, 63 (2011) 1557.
[54] S. Zhao, L. Zou, Effects of working temperature on separation performance, membrane scaling and
cleaning in forward osmosis desalination, Desalination, 278 (2011) 157-164.
[55] Y. Kim, S. Lee, H.K. Shon, S. Hong, Organic fouling mechanisms in forward osmosis membrane
process under elevated feed and draw solution temperatures, Desalination, 355 (2015) 169-177.
[56] U. Sudarno, J. Winter, C. Gallert, Effect of varying salinity, temperature, ammonia and nitrous acid
concentrations on nitrification of saline wastewater in fixed-bed reactors, Bioresource Technology, 102
(2011) 5665-5673.

107

[57] Q. Fontenot, C. Bonvillain, M. Kilgen, R. Boopathy, Effects of temperature, salinity, and carbon:
nitrogen ratio on sequencing batch reactor treating shrimp aquaculture wastewater, Bioresource
Technology, 98 (2007) 1700-1703.
[58] C.P. Morrow, A.L. McGaughey, S.R. Hiibel, A.E. Childress, Submerged or sidestream? Module
configuration influence on fouling and salinity in forward osmosis membrane bioreactors, Journal of
Membrane Science, 548 (2018) 583-592.
[59] R.D. Gustafson, J.R. Murphy, A. Achilli, A stepwise model of direct contact membrane distillation for
application to large-scale systems: Experimental results and model predictions, Desalination, 378 (2016)
14-27.
[60] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Impact of organic and colloidal fouling on trace
organic contaminant rejection by forward osmosis: Role of initial permeate flux, Desalination, 336
(2014) 146-152.
[61] J.R. Werber, A. Deshmukh, M. Elimelech, The Critical Need for Increased Selectivity, Not Increased
Water Permeability, for Desalination Membranes, Environmental Science & Technology Letters, 3 (2016)
112-120.
[62] G. Blandin, H. Vervoort, P. Le-Clech, A.R.D. Verliefde, Fouling and cleaning of high permeability
forward osmosis membranes, Journal of Water Process Engineering, 9 (2016) 161-169.
[63] N.M. Mazlan, P. Marchetti, H.A. Maples, B. Gu, S. Karan, A. Bismarck, A.G. Livingston, Organic
fouling behaviour of structurally and chemically different forward osmosis membranes – A study of
cellulose triacetate and thin film composite membranes, Journal of Membrane Science, 520 (2016) 247-
261.
[64] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, M. Elimelech, High performance thin-film
composite forward osmosis membrane, Environmental science & technology, 44 (2010) 3812-3818.
[65] S.S. Manickam, J.R. McCutcheon, Model thin film composite membranes for forward osmosis:
Demonstrating the inaccuracy of existing structural parameter models, Journal of Membrane Science,
483 (2015) 70-74.
[66] E.A. Bell, R.W. Holloway, T.Y. Cath, Evaluation of forward osmosis membrane performance and
fouling during long-term osmotic membrane bioreactor study, Journal of Membrane Science, 517 (2016)
1-13.
[67] W. Luo, M. Xie, F.I. Hai, W.E. Price, L.D. Nghiem, Biodegradation of cellulose triacetate and
polyamide forward osmosis membranes in an activated sludge bioreactor: Observations and
implications, Journal of Membrane Science, 510 (2016) 284-292.
[68] C. Boo, S. Lee, M. Elimelech, Z. Meng, S. Hong, Colloidal fouling in forward osmosis: Role of reverse
salt diffusion, Journal of Membrane Science, 390-391 (2012) 277-284.
[69] S. Zou, Y.-N. Wang, F. Wicaksana, T. Aung, P.C.Y. Wong, A.G. Fane, C.Y. Tang, Direct microscopic
observation of forward osmosis membrane fouling by microalgae: Critical flux and the role of
operational conditions, Journal of Membrane Science, 436 (2013) 174-185.
[70] N. Dow, S. Gray, J. Zhang, E. Ostarcevic, A. Liubinas, P. Atherton, G. Roeszler, A. Gibbs, M. Duke,
Pilot trial of membrane distillation driven by low grade waste heat: Membrane fouling and energy
assessment, Desalination, 391 (2016) 30-42.
[71] H.C. Duong, P. Cooper, B. Nelemans, T.Y. Cath, L.D. Nghiem, Evaluating energy consumption of air
gap membrane distillation for seawater desalination at pilot scale level, Separation and Purification
Technology, 166 (2016) 55-62.
[72] I.-S. Chang, P. Le Clech, B. Jefferson, S. Judd, Membrane fouling in membrane bioreactors for
wastewater treatment, Journal of environmental engineering, 128 (2002) 1018-1029.
[73] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Toward Resource Recovery from Wastewater:
Extraction of Phosphorus from Digested Sludge Using a Hybrid Forward Osmosis–Membrane Distillation
Process, Environmental Science & Technology Letters, 1 (2014) 191-195.

108

[74] S.M. Hocaoglu, G. Insel, E.U. Cokgor, D. Orhon, Effect of low dissolved oxygen on simultaneous
nitrification and denitrification in a membrane bioreactor treating black water, Bioresource technology,
102 (2011) 4333-4340.
[75] K.A. Salls, D. Won, E.P. Kolodziej, A.E. Childress, S.R. Hiibel, Evaluation of semi-volatile contaminant
transport in a novel, gas-tight direct contact membrane distillation system, Desalination, 427 (2018) 35-
41.
[76] D. Qu, D. Sun, H. Wang, Y. Yun, Experimental study of ammonia removal from water by modified
direct contact membrane distillation, Desalination, 326 (2013) 135-140.
[77] Z. Xie, T. Duong, M. Hoang, C. Nguyen, B. Bolto, Ammonia removal by sweep gas membrane
distillation, Water research, 43 (2009) 1693-1699.
[78] S. Judd, The MBR book: principles and applications of membrane bioreactors for water and
wastewater treatment, Elsevier, 2010.
[79] G.K. Pearce, UF/MF pre-treatment to RO in seawater and wastewater reuse applications: a
comparison of energy costs, Desalination, 222 (2008) 66-73.
[80] G. Pearce, UF/MF pre-treatment to RO in seawater and wastewater reuse applications: a
comparison of energy costs, Desalination, 222 (2008) 66-73.
[81] A. Achilli, E. Marchand, A. Childress, A performance evaluation of three membrane bioreactor
systems: aerobic, anaerobic, and attached-growth, Water Science and Technology, 63 (2011) 2999-3005.
[82] M. Herzberg, M. Elimelech, Biofouling of reverse osmosis membranes: Role of biofilm-enhanced
osmotic pressure, Journal of Membrane Science, 295 (2007) 11-20.
[83] E.M. Hoek, M. Elimelech, Cake-enhanced concentration polarization: a new fouling mechanism for
salt-rejecting membranes, Environmental science & technology, 37 (2003) 5581-5588.
[84] S.E. Kwan, E. Bar-Zeev, M. Elimelech, Biofouling in forward osmosis and reverse osmosis:
Measurements and mechanisms, Journal of Membrane Science, 493 (2015) 703-708.
[85] L. Bohm, A. Drews, H. Prieske, P.R. Berube, M. Kraume, The importance of fluid dynamics for MBR
fouling mitigation, Bioresour Technol, 122 (2012) 50-61.
[86] M. Xie, E. Bar-Zeev, S.M. Hashmi, L.D. Nghiem, M. Elimelech, Role of Reverse Divalent Cation
Diffusion in Forward Osmosis Biofouling, Environ Sci Technol, 49 (2015) 13222-13229.
[87] W. Luo, F.I. Hai, J. Kang, W.E. Price, W. Guo, H.H. Ngo, K. Yamamoto, L.D. Nghiem, Effects of salinity
build-up on biomass characteristics and trace organic chemical removal: Implications on the
development of high retention membrane bioreactors, Bioresource technology, 177 (2015) 274-281.
[88] E.W.B. Rice, R.B. Eaton, A.D. Clesceri, L.S. Bridgewater, Standard methodsfor the examination of
water and wastewater, in, 2012.
[89] J.H. Priester, A.M. Horst, L.C. Van De Werfhorst, J.L. Saleta, L.A. Mertes, P.A. Holden, Enhanced
visualization of microbial biofilms by staining and environmental scanning electron microscopy, Journal
of Microbiological Methods, 68 (2007) 577-587.
[90] W.C. Lay, Y. Liu, A.G. Fane, Impacts of salinity on the performance of high retention membrane
bioreactors for water reclamation: A review, Water Res, 44 (2010) 21-40.
[91] Y. Wang, F. Wicaksana, C.Y. Tang, A.G. Fane, Direct microscopic observation of forward osmosis
membrane fouling, Environmental science & technology, 44 (2010) 7102-7109.
[92] P. Le Clech, B. Jefferson, I.S. Chang, S.J. Judd, Critical flux determination by the flux-step method in a
submerged membrane bioreactor, Journal of membrane science, 227 (2003) 81-93.
[93] R. Field, D. Wu, J. Howell, B. Gupta, Critical flux concept for microfiltration fouling, Journal of
Membrane Science, 100 (1995) 259-272.
[94] J.A. Howell, Sub-critical flux operation of microfiltration, Journal of Membrane Science, 107 (1995)
165-171.

109

[95] S. Phuntsho, F. Lotfi, S. Hong, D.L. Shaffer, M. Elimelech, H.K. Shon, Membrane scaling and flux
decline during fertiliser-drawn forward osmosis desalination of brackish groundwater, Water Res, 57
(2014) 172-182.
[96] E.W. Tow, J.H. Lienhard V, Quantifying osmotic membrane fouling to enable comparisons across
diverse processes, Journal of Membrane Science, 511 (2016) 92-107.
[97] Y. Gu, L. Chen, J.-W. Ng, C. Lee, V.W.C. Chang, C.Y. Tang, Development of anaerobic osmotic
membrane bioreactor for low-strength wastewater treatment at mesophilic condition, Journal of
Membrane Science, 490 (2015) 197-208.
[98] Y. Sun, J. Tian, Z. Zhao, W. Shi, D. Liu, F. Cui, Membrane fouling of forward osmosis (FO) membrane
for municipal wastewater treatment: A comparison between direct FO and OMBR, Water Res, 104
(2016) 330-339.
[99] G. Blandin, A.R.D. Verliefde, C.Y. Tang, A.E. Childress, P. Le-Clech, Validation of assisted forward
osmosis (AFO) process: Impact of hydraulic pressure, Journal of Membrane Science, 447 (2013) 1-11.
[100] B.D. Coday, D.M. Heil, P. Xu, T.Y. Cath, Effects of transmembrane hydraulic pressure on
performance of forward osmosis membranes, Environ Sci Technol, 47 (2013) 2386-2393.
[101] Y. Oh, S. Lee, M. Elimelech, S. Lee, S. Hong, Effect of hydraulic pressure and membrane orientation
on water flux and reverse solute flux in pressure assisted osmosis, Journal of Membrane Science, 465
(2014) 159-166.
[102] G. O'Toole, L. Jones, C. Coutinho, C. Hayes, M. Napoles, A. Achilli, River-to-sea pressure retarded
osmosis: Resource utilization in a full-scale facility, Desalination, 389 (2016) 39-51.
[103] S. Zou, H. Yuan, A. Childress, Z. He, Energy consumption by recirculation: a missing parameter
when evaluating forward osmosis, in, ACS Publications, 2016.
[104] Q. Zhang, Y.W. Jie, W.L.C. Loong, J. Zhang, A.G. Fane, S. Kjelleberg, S.A. Rice, D. McDougald,
Characterization of biofouling in a lab-scale forward osmosis membrane bioreactor (FOMBR), Water
research, 58 (2014) 141-151.
[105] T.Y. Cath, M. Elimelech, J.R. McCutcheon, R.L. McGinnis, A. Achilli, D. Anastasio, A.R. Brady, A.E.
Childress, I.V. Farr, N.T. Hancock, J. Lampi, L.D. Nghiem, M. Xie, N.Y. Yip, Standard Methodology for
Evaluating Membrane Performance in Osmotically Driven Membrane Processes, Desalination, 312
(2013) 31-38.
[106] A. Tiraferri, N.Y. Yip, A.P. Straub, S. Romero-Vargas Castrillon, M. Elimelech, A method for the
simultaneous determination of transport and structural parameters of forward osmosis membranes,
Journal of Membrane Science, 444 (2013) 523-538.
[107] S. Loeb, L. Titelman, E. Korngold, J. Freiman, Effect of porous support fabric on osmosis through a
Loeb-Sourirajan type asymmetric membrane, Journal of Membrane Science, 129 (1997) 243-249.
[108] J. Wang, D.S. Dlamini, A.K. Mishra, M.T.M. Pendergast, M.C.Y. Wong, B.B. Mamba, V. Freger,
A.R.D. Verliefde, E.M.V. Hoek, A critical review of transport through osmotic membranes, Journal of
Membrane Science, 454 (2014) 516-537.
[109] V. Lobo, Mutual diffusion coefficients in aqueous electrolyte solutions (technical report), Pure and
applied chemistry, 65 (1993) 2613-2640.
[110] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, L.A. Hoover, Y.C. Kim, M. Elimelech, Thin-film
composite pressure retarded osmosis membranes for sustainable power generation from salinity
gradients, Environmental science & technology, 45 (2011) 4360-4369.
[111] A.C. Cameron, F.A. Windmeijer, An R-squared measure of goodness of fit for some common
nonlinear regression models, Journal of Econometrics, 77 (1997) 329-342.
[112] S. Kook, C. Swetha, J. Lee, C. Lee, T. Fane, I. Kim, Forward Osmosis Membranes under Null-
pressure Condition: Do Hydraulic and Osmotic Pressures Have Identical Nature?, Environmental science
& technology, (2018).

110

[113] X. Lu, L.H. Arias Chavez, S. Romero-Vargas Castrillon, J. Ma, M. Elimelech, Influence of active layer
and support layer surface structures on organic fouling propensity of thin-film composite forward
osmosis membranes, Environ Sci Technol, 49 (2015) 1436-1444.
[114] M.T.M. Pendergast, J.M. Nygaard, A.K. Ghosh, E.M. Hoek, Using nanocomposite materials
technology to understand and control reverse osmosis membrane compaction, Desalination, 261 (2010)
255-263.
[115] A. Achilli, T.Y. Cath, A.E. Childress, Power generation with pressure retarded osmosis: An
experimental and theoretical investigation, Journal of Membrane Science, 343 (2009) 42-52.
[116] S. Zhao, L. Zou, C.Y. Tang, D. Mulcahy, Recent developments in forward osmosis: Opportunities
and challenges, Journal of Membrane Science, 396 (2012) 1-21.
[117] N.C. Nguyen, S.-S. Chen, H.T. Nguyen, H.H. Ngo, W. Guo, C.W. Hao, P.-H. Lin, Applicability of a
novel osmotic membrane bioreactor using a specific draw solution in wastewater treatment, Science of
the Total Environment, 518 (2015) 586-594.
[118] W. Luo, F.I. Hai, W.E. Price, M. Elimelech, L.D. Nghiem, Evaluating ionic organic draw solutes in
osmotic membrane bioreactors for water reuse, Journal of Membrane Science, 514 (2016) 636-645.
[119] C.Y. Tang, Q. She, W.C.L. Lay, R. Wang, A.G. Fane, Coupled effects of internal concentration
polarization and fouling on flux behavior of forward osmosis membranes during humic acid filtration,
Journal of Membrane Science, 354 (2010) 123-133.
[120] S. Zhao, L. Zou, D. Mulcahy, Effects of membrane orientation on process performance in forward
osmosis applications, Journal of Membrane Science, 382 (2011) 308-315.
[121] S. Zou, Y. Gu, D. Xiao, C.Y. Tang, The role of physical and chemical parameters on forward osmosis
membrane fouling during algae separation, Journal of Membrane Science, 366 (2011) 356-362.
[122] Q. She, X. Jin, Q. Li, C.Y. Tang, Relating reverse and forward solute diffusion to membrane fouling
in osmotically driven membrane processes, Water Res, 46 (2012) 2478-2486.
[123] Q. She, Y.K.W. Wong, S. Zhao, C.Y. Tang, Organic fouling in pressure retarded osmosis:
Experiments, mechanisms and implications, Journal of Membrane Science, 428 (2013) 181-189.
[124] N.-N. Bui, J.T. Arena, J.R. McCutcheon, Proper accounting of mass transfer resistances in forward
osmosis: Improving the accuracy of model predictions of structural parameter, Journal of Membrane
Science, 492 (2015) 289-302.
[125] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of forward osmosis
membranes, Journal of Membrane Science, 320 (2008) 292-302.
[126] A. Tiraferri, Y. Kang, E.P. Giannelis, M. Elimelech, Superhydrophilic thin-film composite forward
osmosis membranes for organic fouling control: fouling behavior and antifouling mechanisms, Environ
Sci Technol, 46 (2012) 11135-11144.
[127] S. Lee, M. Elimelech, Relating organic fouling of reverse osmosis membranes to intermolecular
adhesion forces, Environmental science & technology, 40 (2006) 980-987.
[128] B. Mi, M. Elimelech, Chemical and physical aspects of organic fouling of forward osmosis
membranes, Journal of Membrane Science, 320 (2008) 292-302.
[129] V. Chen, A. Fane, S. Madaeni, I. Wenten, Particle deposition during membrane filtration of colloids:
transition between concentration polarization and cake formation, Journal of Membrane Science, 125
(1997) 109-122.
[130] L. Defrance, M. Jaffrin, Comparison between filtrations at fixed transmembrane pressure and fixed
permeate flux: application to a membrane bioreactor used for wastewater treatment, Journal of
Membrane Science, 152 (1999) 203-210.
[131] B. Cho, A. Fane, Fouling transients in nominally sub-critical flux operation of a membrane
bioreactor, Journal of membrane science, 209 (2002) 391-403.
[132] X. Zhang, Z. Ning, D.K. Wang, J.C.D. da Costa, Processing municipal wastewaters by forward
osmosis using CTA membrane, Journal of Membrane Science, 468 (2014) 269-275.

111

[133] J.A. Brant, A.E. Childress, Assessing short-range membrane–colloid interactions using surface
energetics, Journal of Membrane Science, 203 (2002) 257-273.
[134] Y. Liu, B. Mi, Combined fouling of forward osmosis membranes: Synergistic foulant interaction and
direct observation of fouling layer formation, Journal of membrane science, 407 (2012) 136-144.
[135] D.L. Shaffer, J.R. Werber, H. Jaramillo, S. Lin, M. Elimelech, Forward osmosis: Where are we now?,
Desalination, 356 (2015) 271-284.
[136] M. Xie, L.D. Nghiem, W.E. Price, M. Elimelech, Impact of humic acid fouling on membrane
performance and transport of pharmaceutically active compounds in forward osmosis, Water research,
47 (2013) 4567-4575.

Abstract (if available)

Abstract Urban reliance on imported water and increasing water supply variability due to climate change have intensified efforts to increase the resiliency of water supplies on a global scale. For reuse applications, membrane bioreactors (MBRs) with microfiltration (MF) or ultrafiltration (UF) membranes submerged in a biological reactor have emerged as an efficient wastewater treatment process to provide high-quality filtrate to a subsequent reverse osmosis (RO) process followed by advanced oxidation. More recently, osmotic membrane bioreactors (OMBRs) with forward osmosis (FO) membranes submerged in the bioreactor are being investigated as low-fouling alternatives to conventional MBRs. If the opportunity exists to reconcentrate the FO draw solution with waste-heat driven membrane distillation (MD), then the OMBR-MD system can provide high quality product water with a low electrical energy requirement. ❧ In an aerobic OMBR-MD system, wastewater is fed into a bioreactor that is aerated to supply oxygen to the biomass and scour the membrane. Through osmosis, water diffuses from the bioreactor, across a semi-permeable FO membrane, into the draw solution. The FO membrane acts as a barrier to solute transport and provides high rejection of contaminants in the wastewater stream. The diluted draw solution is sent to MD for reconcentration and generation of product water. ❧ Compared to the MF or UF process in a conventional MBR, the FO process in the OMBR offers the advantage of much higher rejection (semi-permeable membrane versus microporous membrane) at lower hydraulic pressure. The FO membrane inside the bioreactor also has much less fouling propensity than MF/UF membranes, and thus, requires less scouring and much less frequent backwashing. When comparing an integrated OMBR-MD system with a conventional MBR-RO system, the higher rejection of the FO membranes also results in lower fouling propensity for the downstream process (RO or MD). When comparing FO draw solution recovery using RO or MD, both processes can utilize waste heat to reduce the required energy input

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Optimizing biomembrane reactor systems for water reclamation and reuse applications

PDF

The roles of surface and pore properties in wetting resistance for membrane distillation membranes

PDF

Integrating systems of desalination and potable reuse: reduced energy consumption for increased water supply

PDF

Fate of antibiotic resistance in anaerobic membrane bioreactors

PDF

Advancing energy recovery from food waste using anaerobic biotechnologies: performance and microbial ecology

PDF

Thermally driven water treatment with membrane distillation: membrane performance, waste heat integration, and cooling analysis

PDF

The formation, fate and transformation of nitromethane in potable reuse processes

PDF

Water desalination: real-time membrane characterization for performance prediction and system analysis for energetic enhancement

PDF

Integrated technologies, blending schemes, and reuse practices to address contaminant and energy challenges in water reclamation

PDF

Developing frameworks to quantify the operational and environmental performance of energy systems within the context of climate change

PDF

Bioelectrochemical treatment of anaerobic process effluents: mitigation of dissolved methane and sulfide

Asset Metadata

Creator Morrow, Christopher Paul (author)

Core Title Wastewater reclamation and potable reuse with novel processes: membrane performance and system integration

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Environmental Engineering

Publication Date 02/02/2019

Defense Date 06/08/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag forward osmosis,membrane bioreactor,membrane distillation,membrane fouling,OAI-PMH Harvest,osmotic membrane bioreactor,potable reuse,wastewater reuse

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Chuildress, Amy Elizabeth (committee chair)

Creator Email christopher.morrow@usc.edu,cmorrow8@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-50541

Unique identifier UC11671949

Identifier etd-MorrowChri-6610.pdf (filename),usctheses-c89-50541 (legacy record id)

Legacy Identifier etd-MorrowChri-6610.pdf

Dmrecord 50541

Document Type Dissertation

Format application/pdf (imt)

Rights Morrow, Christopher Paul

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA