University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Water desalination: Real-time membrane characterization for
performance prediction and system analysis for energetic enhancement
By
Weijian Ding
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ENVIRONMENTAL ENGINEERING)
August 2024
Copyright [2024] Weijian Ding

ii
Acknowledgments
I would like to express my heartfelt gratitude to my advisor, Dr. Amy Childress. Since the
first day of my academic journey as a Ph.D. student, Dr. Childress has been recognized for being
not just a mentor but also a dear friend and a guiding light in my life/research during the highs
and lows. She played a pivotal role in my personal and academic growth, imparting invaluable
lessons on how to navigate the complexities of research and interpersonal relationships.
I am deeply grateful to my wife for her love, unwavering support, scarification, and
constant companionship throughout my Ph.D. Her presence has been a source of strength and
solace, especially during the trying times we faced, including illnesses, the challenges of
navigating the COVID-19 pandemic, the trials of being separated by distance, and more. I also
wish to extend my sincere appreciation to my parents, who, despite the limited time I spent
with them during my Ph.D., always stood by my side, offering unconditional trust and support,
and encouragement in every decision I made, shaping my life in more ways than I can express.
I wish to express my appreciation to my Ph.D. Defense Committee, comprised of Dr. Amy
Childress, Dr. Adam Smith, Dr. Daniel McCurry, and Dr. Andrea Hodge. Their dedication in
investing time and effort in my research and guiding me through the challenging process has
been invaluable. I am immensely grateful for their contributions to my academic journey.
Additionally, I want to extend gratitude to my research colleagues that I have worked
closely: Dr. Ryan Gustafson, Dr. Allyson McGaughey, Dr. Xin Wei, Dr. Sarah Philo, Shih-Hsun
Nien, Shounak Joshi, Martijn Bindels, Bana Dahdah, Connie Devenport, with special thanks to
Kexin Ma for her contributions.

iii
Support
This work was supported by the National Science Foundation under Grant No. 1820389 and
by the California Department of Water Resources under Water Desalination Grant Program
Proposition 1 Agreement No. 4600013148.
This work was also supported by a Cooperative Agreement (W9132T-23-2-0002) with the
U.S. Army Corps of Engineers, Engineer Research and Development Center, Construction
Engineering Research Laboratory (USACE ERDC-CERL).
The author acknowledges the University of Southern California Astani Department's Teh Fu
“Dave” Yen Fellowship in Environmental Engineering. And Sonny Astani Civil and Environmental
Engineering Department Das Family Travel Award.
The SEM images were acquired at the University of Southern Core Center for Excellence in
Nano Imaging.

iv
Table of Contents
Acknowledgments .........................................................................................................................ii
Support..........................................................................................................................................iii
List of Tables ............................................................................................................................... viii
List of Figures................................................................................................................................ix
Abstract....................................................................................................................................... xvi
Chapter 1 Introduction ..................................................................................................................1
1.1 Membrane Distillation Overview..........................................................................................1
1.1.1 Water treatment membranes .......................................................................................3
1.1.2 MD and waste heat integration.....................................................................................3
1.1.3 MD cooling technology and replenishment cooling......................................................6
1.2 Objectives and Scope of Work..............................................................................................7
1.3 Dissertation Organization.....................................................................................................9
Chapter 2 A Novel Method to Quantify Real-time Compaction of Water Treatment
Membranes..................................................................................................................................11
2.1 Abstract ..............................................................................................................................11
2.2 Introduction and Background.............................................................................................12
2.3 Materials and Methods ......................................................................................................15
2.3.1 Compaction experiments ............................................................................................15
2.3.1.1 Measurement system...........................................................................................15

v
2.3.1.2 Variable-condition short-term experiments.........................................................17
2.3.1.3 Time-dependent long-term experiments.............................................................17
2.3.2 Method validation .......................................................................................................18
2.4 Results and Discussion........................................................................................................20
2.4.1 Membrane compaction at serval temperatures and pressures..................................20
2.4.2 Membrane creep and recovery ...................................................................................23
2.4.3 Membrane hysteresis and in situ validation of method..............................................27
2.4.4 Ex situ validation of method using SEM images..........................................................28
2.5 Implications........................................................................................................................30
Chapter 3 Compaction of High-Pressure Water Treatment Membranes: Real-Time
Quantification and Analysis.........................................................................................................31
3.1 Abstract ..............................................................................................................................31
3.2 Introduction and Background.............................................................................................33
3.3 Materials and Methods ......................................................................................................37
3.3.1 NF and RO membranes................................................................................................37
3.3.2 Real-time compaction and creep measurement system.............................................37
3.3.3 Compressive experiments ...........................................................................................37
3.3.3.1 Short-term compressive experiments..................................................................38
3.3.3.2 Creep-and-recovery experiments.........................................................................38
3.3.3.3 Hysteresis experiments ........................................................................................40
3.3.4 SEM images .................................................................................................................40
3.3.5 Tensile experiments.....................................................................................................42

vi
3.4 Results and Discussion........................................................................................................43
3.4.1 Membrane compaction under compression ...............................................................43
3.4.1.1 NF membrane.......................................................................................................43
3.4.1.2 RO membrane.......................................................................................................47
3.4.2 Membrane creep-and-recovery behavior ...................................................................50
3.4.2.1 Creep-and-recovery experiments with constant pressure ...................................50
3.4.2.2 Creep-and-recovery experiments with incrementally increasing pressure..........54
3.4.3 Membrane hysteresis and method validation.............................................................58
3.4.3.1 Membrane hysteresis and in-situ validation ........................................................58
3.4.3.2 Ex-situ method validation.....................................................................................60
3.4.4 Difference between compressive and tensile tests for water desalination membranes
..............................................................................................................................................62
3.5 Implications........................................................................................................................66
Chapter 4 Analysis of Anthropogenic Waste Heat Emission from an Academic Data Center...68
4.1 Abstract ..............................................................................................................................68
4.2 Introduction........................................................................................................................69
4.3 Materials and Methods ......................................................................................................74
4.3.1 Server room selected for simulation ...........................................................................74
4.3.2 Air-cooling system in the server room.........................................................................75
4.3.3 Analysis framework and collection of primary data ....................................................78
4.3.4 Calculation of cooling and energy efficiency indices...................................................83
4.3.5 Characterization of operating trends and thermal conditions....................................84

vii
4.3.6 Development of CFD model.........................................................................................84
4.3.7 Estimation of server room waste heat quality and quantity .......................................86
4.4 Results and Discussion........................................................................................................88
4.4.1 Server room CSE and PUE............................................................................................88
4.4.2 Server room operating trends and temperature distribution .....................................90
4.4.3 Server room waste heat estimation and reuse analysis from CFD model results.......96
4.4.3.1 Current server room configuration with open aisles............................................96
4.4.3.2 Example of improved server room configuration with HAC system...................101
4.5 Conclusions and Implications ...........................................................................................105
Chapter 5 Conclusions ...............................................................................................................107
5.1 Research Synopsis ............................................................................................................107
5.2 Summary of Real-time Water Treatment Membrane Characterization...........................107
5.3 Summary of Waste Heat Characterization and Integration in MD Systems.....................108
References..................................................................................................................................110

viii
List of Tables
Table 3.1: Percentages of NF membrane compaction contribution from instantaneous
compaction and creep. .................................................................................................................56
Table 3.2: Percentages of RO membrane compaction contribution from instantaneous
compaction and creep. .................................................................................................................57
Table 4.1: Rack group and CRAC information. .............................................................................75
Table 4.2: Average temperatures and airflow rates at random locations in the hot aisles of
the server room. ...........................................................................................................................81
Table 4.3: Average temperatures and flowrates entering the hot vents and exiting the cold
vents. ............................................................................................................................................81
Table 4.4: Computational mesh with finer cells in the critical regions of the hot rack faces,
hot vents, and cold vents..............................................................................................................86
Table 4.5: Average temperature for hot and cold aisles in the server room with open-aisle
configuration ..............................................................................................................................100
Table 4.6: Average temperature for hot and cold aisles in the server room with HAC system ..103

ix
List of Figures
Figure 1.1: Schematic diagram of a bench-scale direct contact membrane distillation
(DCMD) system composed of a DCMD module, auxiliary heater and chiller, feed and
distillate pumps, feed and distillate tanks, and heat exchanger (HX). ...........................................2
Figure 1.2: Schematic diagram of the temperature gradient across the hydrophobic porous
membrane in DCMD process. .........................................................................................................5
Figure 2.1: Schematic of the novel measurement system used to measure membrane
compaction. Stable force and temperature control are provided by dynamic mechanical
analysis; real-time cross-membrane impedance is detected by electrochemical impedance
spectroscopy. The integrated system enables quantification of membrane compaction. ...........16
Figure 2.2: Impedance and compaction for pressures of 1, 5, and 10 psi at: (a to c)
temperature of 25 °C, (d to f) temperature of 45 °C, and (g to i) temperature of 65 °C. The
shading around each line represents the standard deviation of the data. Impedance
difference (and compaction) at higher pressure is larger.............................................................21
Figure 2.3: Impedance results during temperature sweeping from 25 to 80 °C with constant
5 psi pressure. Impedance (and compaction) decrease linearly with increased temperature
at a very slow rate. .......................................................................................................................22
Figure 2.4: Continuous creep testing with stepwise pressure increases from 1 to 12.5 psi
over a 3 h period. The pressure was then released (dropped to 0 psi). The table embedded
in the graph shows percentages of creep/total compaction, instant/total compaction, and
creep/instant compaction for each creep step. The creep/instant compaction ratio
increases from 2.4 to 49% from steps 1 to 6. The average irrevocable compaction is 16%. ........24

x
Figure 2.5: Impedance across three creep and recovery cycles under a constant pressure
of 5 psi. The gradual occurrence of creep results in an ultimate irrevocable compaction
of 12.5%........................................................................................................................................26
Figure 2.6: Hysteresis of the membrane is demonstrated through (a) strain hysteresis
detected by dynamic mechanical analysis and (b) impedance hysteresis detected by
electrochemical impedance spectroscopy. Comparison of the average irrevocable strain
(2.9%) with the average irrevocable impedance difference (2.7%) provides validation of
the novel method..........................................................................................................................27
Figure 2.7: SEM images for (a) the pristine membrane and (b) the compacted membrane
imaged in a fully recovered state after compaction under 4 psi pressure. The thickness
percent difference of the compacted membrane compared with the pristine membrane
is 2.6%...........................................................................................................................................29
Figure 3.1: (a) SEM image of PE backing layer with void spaces and solid material
indicated and (b) schematic of PE backing layer showing the void space and solid
material areas as well as their representative thicknesses. .........................................................41
Figure 3.2: SEM images of the cross-sectional morphology of the (a) pristine NF
membrane, (b) NF membrane tested under 80-psi pressure, and (c) NF membrane tested
under 300-psi pressure. Corresponding values of layer/component and total membrane
thickness for (d) the pristine NF membrane, (e) the NF membrane tested under 80-psi
pressure, and (f) the NF membrane tested under 300-psi pressure. A continuous air pocket is
observed clearly in the pristine membrane; it becomes disjointed and less visible during
compaction – especially at 300 psi. ..............................................................................................44

xi
Figure 3.3: SEM images of the cross-sectional morphology of the (a) pristine NF membrane,
(b) NF membrane tested under 80-psi pressure, and (c) NF membrane tested under 300-psi
pressure. Corresponding values of layer/component and total membrane thickness for
(d) the pristine NF membrane, (e) the NF membrane tested under 80-psi pressure, and
(f) the NF membrane tested under 300-psi pressure. A continuous air pocket is observed
clearly in the pristine membrane; it becomes disjointed and less visible during compaction –
especially at 300 psi......................................................................................................................46
Figure 3.4: SEM images of the cross-sectional morphology of the (a) pristine RO membrane,
(b) RO membrane tested under 80-psi pressure, and (c) RO membrane tested under 300-psi
pressure. Corresponding values of layer/component and total membrane thickness for
(d) the pristine RO membrane, (e) the RO membrane tested under 80-psi pressure, and
(f) the RO membrane tested under 300-psi pressure. Compared to the pristine NF
membrane, the pristine RO membrane appears denser with less void space and all layers
undergo more compaction at 21 bar (300 psi) than 80 psi. .........................................................48
Figure 3.5: (a) Relative thickness and (b) relative compaction of RO membrane
layers/components. The greatest compaction occurs in the PE solid material, which further
compacts at 300-psi to counterbalance the applied pressure......................................................49
Figure 3.6: Normalized impedance across seven creep and recovery cycles of (a) the NF
membrane under a constant pressure of 80 psi and (b) the RO membrane under a constant
pressure of 200 psi. The gradual occurrence of creep is an indication of the development of
membrane fatigue, which leads to a final irrevocable compaction of 5.1% for the NF
membrane and 5.0% for the RO membrane.................................................................................52

xii
Figure 3.7: Normalized impedance and corresponding compaction of pristine and aged RO
membrane samples under a constant pressure of 200 psi. Membrane samples were aged by
exposing them to UV radiation for 4, 6, 24, 48, and 72 h. RO membrane deformation
(compaction) increases with the duration for which the samples were exposed to UV
radiation likely due to the development of membrane fatigue....................................................53
Figure 3.8: Creep-and-recovery behavior with incrementally increasing pressure for
(a) the NF membrane with stepwise pressure increases from 10 to 80 psi. The role of creep
increases with each step, while the instantaneous compaction decreases with each and
(b) the RO membrane with stepwise pressure increases from 80 to 150 psi from steps 1-8;
two additional steps (steps 9 and 10) were applied at a higher-pressure (200 and 250 psi).
For both membranes, the role of instantaneous compaction decreases, and the role of
creep increases with each step.....................................................................................................56
Figure 3.9: Hysteresis of the NF membrane is demonstrated through (a) strain hysteresis
detected by dynamic mechanical analysis and (b) impedance hysteresis detected by a
potentiostat. Comparison of the average irrevocable strain (2.2%) with the average
irrevocable impedance difference (2.6%) provides in-situ method validation..............................59
Figure 3.10: Hysteresis of the NF membrane is demonstrated through (a) strain hysteresis
detected by dynamic mechanical analysis and (b) impedance hysteresis detected by a
potentiostat. Comparison of the average irrevocable strain (2.2%) with the average
irrevocable impedance difference (2.6%) provides in-situ method validation..............................60
Figure 3.11: SEM images for (a) the pristine NF membrane, (b) the compacted NF
membrane imaged in a fully recovered state after hysteresis experiments with 80 psi applied

xiii
pressure; the thickness percent difference of the compacted membrane compared with
the pristine membrane is 3.4%, (c) the pristine RO membrane, and (d) the compacted RO
membrane imaged in a fully recovered state after hysteresis experiments with 200 psi
applied pressure; the thickness percent difference of the compacted membrane compared
with the pristine membrane is 5.2%. ............................................................................................61
Figure 3.12: Tensile and compressive deformation for (a) low-pressure membranes,
(b) NF membranes, and (c) RO membranes. Average compaction measured by
compressive tests is different from the average strain measured by tensile tests for all
three types of membranes............................................................................................................65
Figure 4.1: (a) Section view of the server room. Air is circulated by computer room airconditioning (CRAC) units that draw hot exhaust generated by the servers into the
warm return air channel and blow chilled air back into the server room via ceiling and
ground vents. (b) Schematic diagram of server room indicating the temperature difference
between the warm returned air and chilled air in the CRAC units (∆T). .......................................77
Figure 4.2: Analysis framework to estimate energy efficiency and anthropogenic
waste heat generated by a representative academic server room. .............................................78
Figure 4.3: (a) Photograph of fabricated thermo-anemometer installed in a rack facing
a hot aisle; (b) infrared photograph of the rack taken by FLIR ONE Pro, which shows
the highest temperature as 41.5 ℃..............................................................................................80
Figure 4.4: Server room geometry with the hot rack faces and outlet vents colored red
and the cold vents colored blue....................................................................................................84

xiv
Figure 4.5: Computational mesh with finer cells in the critical regions of the hot rack faces,
hot vents, and cold vents..............................................................................................................85
Figure 4.6: (a) Cooling efficiency of the server room given as cooling system efficiency
(CSE). The server room has an average CSE of 1.28 kW/ton, which is higher (less efficient)
than the good practice benchmark of 0.8 kW/ton. (b) Energy efficiency of the server
room given as power usage effectiveness (PUE). The server room has an average PUE
value of 2.5, which is higher (less efficient) than the US average PUE of 1.6...............................89
Figure 4.7: Computing load data collected from the server room for three scenarios.
a) A daily scenario for mean CPU occupation during weekdays and weekends in mid- and
off-semester. b) A weekly scenario for average mean CPU occupation during mid- and
off-semester. The inset graphs show the variance in a typical daily scenario.
c) A typical monthly scenario for average mean CPU occupation during mid- and
off-semester. The data circled in red were collected during a maintenance period. ....................93
Figure 4.8: Row-wise temperature data and average row-wise temperature for
(a) rack group A, (b) rack group B, (c) rack group C, (d) rack group D, and (e) rack group E. ......95
Figure 4.9: Temperature contour plots showing the horizontal temperature distribution
for several rack heights: (a) Z = 0.5m,(b) Z = 1.0m, (c) Z = 1.5m, (d) Z = 2.0m, (e) Z = 2.5m,
(f) Z = 3.0m, (g) Z = 3.5m, and (h) configuration of server room rack groups and aisles;
hot spots were identified horizontally in the middle and at the ends of rack groups...................97
Figure 4.10: (a) Air flow distribution given by vectors. Chilled air can be seen flowing
from the ceiling cold vents directly to the hot vents due to bypassing and short-circuiting.

xv
(b) Vertical temperature distribution in the server room. Hot spots are shown in red
on the faces of the rack groups in the hot aisles. .........................................................................98
Figure 4.11: Vertical average temperature distribution for the server room with open-aisle
configuration (a) hot aisle and (b) cold aisle. ...............................................................................99
Figure 4.12: (a) Air flow distribution given by vectors. Chilled air can be seen flowing
from the ceiling and floor cold vents to the servers without bypassing occurring.
(b) Vertical temperature distribution in the server room. Temperature in the hot aisles
is higher than open-aisle configuration......................................................................................102
Figure 4.13: Vertical average temperature distribution for the server room with a basic
hot air containment system: (a) hot aisle and (b) cold aisle.......................................................104

xvi
Abstract
Membrane distillation (MD) is a thermally driven water desalination process gaining
interest for its ability to treat challenging saline waters, achieve high rejection of non-volatile
contaminants, and contribute to zero-liquid-discharge (ZLD) desalination. However, MD is often
criticized for its high thermal energy demand and inefficiency compared to pressure-driven
processes like reverse osmosis (RO). Additionally, MD performance can be significantly
degraded due to membrane compaction, scaling, and biofouling. This work aims to advance the
MD process by achieving deeper insights into membrane compaction and integrating waste
heat to improved energy efficiency. A novel method combining electrical impedance
spectroscopy (EIS) with dynamic mechanical analysis (DMA) is developed to characterize
membrane compaction in real-time under varying temperatures and pressures. Additionally, an
analysis framework is created to identify and utilize low-grade waste heat from an academic
data center. A CFD model simulates the quality and quantity of waste heat emissions, which can
be used for MD process. This study provides valuable insights into understanding real-time
membrane compaction and the development of membrane fatigue, as well as strategies to
enhance MD system efficiency through innovative heat reuse techniques.

1
Chapter 1 Introduction
1.1 Membrane Distillation Overview
Membrane distillation (MD), a thermally driven desalination process, is gaining interest due
to its capability of desalting challenging water with high saline (e.g., produce water), high
rejection of non-volatile contaminants, and the potential to contribute to achieving zero-liquiddischarge (ZLD) desalination [1, 2]. In MD, represented in Figure 1.1 as a direct contact
membrane distillation (DCMD) system, a warm feed solution circulates on one side of a
microporous, hydrophobic membrane, while a cool pure water solution circulates on the other
side. The temperature difference induces a vapor pressure difference across the membrane,
which causes water at the feed side to vaporize then pass through the membrane pores and
condense upon contact with the cool distillate solution. However, to maintain the obligatory
vapor pressure difference across the hydrophobic membrane during operation, MD, especially
DCMD without heat recovery is generally criticized for the intensive demand of thermal energy
and inherent inefficiency compared with conventional pressure-driven desalination process
(e.g., reverse osmosis (RO)), which is well-known for its high energy efficiency and is the
benchmark for other desalination processes [3, 4].

2
Figure 1.1: Schematic diagram of a bench-scale direct contact membrane distillation (DCMD) system
composed of a DCMD module, auxiliary heater and chiller, feed and distillate pumps, feed and distillate tanks,
and heat exchanger (HX).
MD systems can be categorized according to their configurations, which include DCMD, air
gap MD (AGMD), sweeping gas MD (SGMD), and vacuum MD (VMG) [1, 5, 6]. Among these MD
configurations, DCMD stands out as the simplest with minimal complexity, while AGMD is the
most commonly employed configuration [1, 7]. The major challenges that MD systems are
facing include membrane performance issues (i.e., fouling, scaling, and especially membrane
compaction) as well as system energy consumption and efficiency, particularly for heating and
cooling processes.

3
1.1.1 Water treatment membranes
Membranes are critical to water treatment applications (e.g., seawater desalination,
wastewater treatment, potable water reuse, and resource recovery) [8-12]. Water treatment
membranes are polymeric thin films that are characterized by one dimension (the thickness)
being significantly smaller than the other two dimensions [13]. For example, for thermally
driven water desalination applications (e.g., MD), hydrophobic polymeric membranes like
polytetrafluoroethylene (PTFE), polyvinylidene fluoride (PVDF), and polypropylene (PP) are used
[1, 7, 14]. For pressure-driven desalination applications (e.g., ultrafiltration (UF), nanofiltration
(NF), and RO), except for the active polyamide (PA) layer, one or more support layers usually
composed of cellulose acetate (CA), polyethylene (PE), polyacrylonitrile (PAN), and polysulfone
(PSf) [15-17]. Polymeric membranes’ performance can be easily impacted by membrane
compaction, scaling, and biofouling [1, 6, 7, 12]. Compaction of desalination and water
treatment membranes can significantly reduce membrane transport properties, damage the
membrane surface/support layer, and weaken membrane selectivity [6, 18, 19]. Especially for
thermally driven desalination applications (e.g., MD), membrane compaction results in
increased heat loss [5, 6, 18, 19] and reduced water permeability [5, 6, 18-23]. Therefore,
understanding water treatment membrane performance, especially real-time compaction, is
crucial to characterizing water treatment systems' performance.
1.1.2 MD and waste heat integration
One of MD’s most criticized shortcomings is its energy (especially thermal energy)
consumption and low energy efficiency [1, 5-7]. As the feed water flows through the membrane
in all MD configurations, it absorbs the latent heat of vaporization and transfers it to the

4
condensation side. Simultaneously, heat is conducted through the membrane from the feed
side to the condensation side, which induces heat loss [24]. Take DCMD as an example (shown
in Figure 1.2), three major heat transfer mechanisms occur – (1) convention in the bulk feed
and distillate solutions, (2) conduction through the hydrophobic membrane, and (3) transfer of
the latent heat of evaporation, which is carried by the water vapor passing through the
membrane pore [1, 25-27]. Under typical conditions, the temperature gradient within the bulk
feed and distillate solutions is negligible compared to MD (as illustrated in Fig. 1.2). Therefore,
the impacts of convective heat transfer within the bulk feed and distillate solutions can be
considered negligible compared to heat transfer through the membrane [27]. The dominant
heat transfer mechanisms in DCMD are conduction and latent heat loss through the membrane
[1, 25-29]. Conductive heat loss through the membrane can be reduced by increasing
membrane thickness; however, increasing membrane thickness negatively impacts water flux
and water production rates [30]. In contrast, latent heat loss through the membrane is
inevitable. Due to the inevitable heat loss, the MD system is considered a low-efficient system.

5
Figure 1.2: Schematic diagram of the temperature gradient across the hydrophobic porous membrane in
DCMD process.
To address the energy demand for feed heating of MD systems, two major approaches (i.e.,
utilization of alternative thermal energy and/or heat recovery techniques) are proposed [24,
31-38]. The utilization of alternative thermal energy (e.g., solar [31-35], waste heat from a
diesel engine exhaust [35, 36], and the waste heat from a power station [37]) as the heating
source can address the energy concern and minimize MD’s electrical energy consumption [35].
While the utilization of heat recovery techniques helped improve the MD configurations'

6
thermal efficiency and reduce the cooling needs from the alternative cooling system [38]. The
lost heat across the membrane can be reclaimed through either external or internal heat
recovery methods.
External heat recovery is applicable in DCMD, VMD, and SGMD processes. In DCMD, an
external heat exchanger is usually positioned between the incoming cold feed and the outgoing
warm distillate, enabling the heat lost to the distillate stream to preheat the feed before being
heated by the main heat source [2, 28, 39, 40]. And in VMD and SGMD, external heat recovery
is implemented by using the incoming cool feed as the cooling fluid in the condenser [41]. In
AGMD and PGMD, internal heat recovery is implemented by passing cool feed water into the
coolant channel, which absorbs heat through the condensation foil; the preheated feed is then
passed through the heat source to heat when entering the membrane module fully [31, 35, 36,
38, 42-44].
1.1.3 MD cooling technology and replenishment cooling
MD research on energy consumption has focused on system heating, while none explicitly
focuses on energy needs for cooling. However, in MD, as water vapor transferred from the feed
side to the distillate side and brings the latent heat through the hydrophobic membrane, just
like the heating in the feed side mentioned in Section 1.1, a continuous cooling source is
needed to maintain the obligatory temperature difference to provide sufficient vapor pressure
gradient across the membrane. For bench-scale MD systems, electrical chillers or heat pumps
are usually used as cooling method [31, 35, 36, 38, 42-44], while for pilot-scale MD systems,
cooling requirements are fulfilled by cooling approaches with auxiliaries (e.g., dry/wet cooling

7
with chiller or refrigerators) and/or system heat recovery through internal or external heat
exchanging process [35, 38, 41].
When heat recovery occurs during feed recirculation, the term “replenishment cooling”
can be proposed to describe the cooling effect of heat recovery in MD systems that use a cool
feed stream. It is also used to describe the continuous or periodic replenishment of the feed
stream with cool feed water, providing cooling in a system that utilizes heat recovery.
Replenishment cooling involves the transfer of heat from the higher-temperature distillate to
the lower-temperature feedwater. This method is considered promising for MD systems due to
its potential for lower energy consumption, a smaller footprint, and stand-alone operation.
Pilot-scale MD systems that exclusively use replenishment cooling with a single-pass feed
stream configuration have been reported by Banat et al. [31], Hagedorn et al. [38], and AndrésMañas et al. [45]. However, the replenishment cooling method used in existing MD systems is
rarely discussed, and a systematic assessment is still needed.
1.2 Objectives and Scope of Work
The first objective of this work is to achieve greater insight into membrane compaction and
its role in materials fatigue. A novel method is developed to monitor and characterize
membrane compaction by combining the ability to detect membrane deformation by electrical
impedance spectroscopy with dynamic mechanical analysis. Short-term mechanical tests across
a range of temperatures and pressures (up to 330 psi) to observe real-time compaction of the
water treatment membranes (i.e., MD, NF, and RO membranes). Creep, recovery, and
hysteresis properties exhibited during longer-term mechanical tests are investigated. The novel

8
method is validated in two ways: with in situ hysteresis results by comparing the irrevocable
strain with the irrevocable impedance difference and with ex situ thickness measurements by
correlating the compaction observed in SEM images with that detected by our novel method.
The development of membrane fatigue is characterized.
The second objective is to advancing MD systems’ performance based on – waste heat
integration. To investigate the potential novel waste heat source, a critical assessment of
potential low-grade waste heat sources has been conducted, which identified a novel waste
heat source from an academic data center. An analysis framework to estimate energy efficiency
and determine quality and quantity of waste heat from a typical academic data center is
developed. Thermal data are collected from the data center using two arrays of thermometers
and thermos-anemometers. A CFD model is then developed to simulate and estimate the
quality and quantity of anthropogenic waste heat emission from the data center.

9
1.3 Dissertation Organization
This dissertation presents four distinct chapters developed based on the investigation of
the water treatment membrane characterizations and thermally driven desalination process
(i.e., MD). It serves as a compilation of papers written on these projects and reports on the
ongoing experiments.
Chapter 2 comprises an entire paper that has been published in Environmental Science and
Technology Letters. In this chapter, a novel method is introduced to quantify real-time
compaction of water treatment membranes under low pressure (up to 12.5 psi). This method is
validated using in situ and ex situ measurements and shows promise for detecting and
predicting performance and fatigue of water treatment membranes in various applications.
Chapter 3 comprises an entire paper that has been submitted for publication in
Environmental Science and Technology. In this chapter, the novel method developed in Chapter
2 has been upgraded to quantify real-time compaction of multi-layer heterogenous
nanofiltration and reverse osmosis membranes under higher pressures, reaching up to 330 psi.
Additionally, this chapter demonstrates the contributions of the void space and solid material
components of the backing layer to entire membrane compaction and characterizes the
development of membrane fatigue.
Chapter 4 comprises an entire paper that has been published in Energies. In this chapter,
an investigation is conducted to identify and characterize a potential novel industrial waste
heat source from an academic data center. This investigation involves on-site data collection
and CFD model simulation. Furthermore, an improved server room configuration is proposed,

10
featuring a hot air containment (HAC) system to concentrate potential waste heat and enhance
server room cooling efficiency.

11
Chapter 2 A Novel Method to Quantify Real-time Compaction of Water
Treatment Membranes
2.1 Abstract
Membranes play critical roles in seawater desalination, wastewater treatment, potable
water reuse, and resource recovery. Their performance can be adversely affected by
compaction, which can limit their practical use and durability. While previous studies have
employed scanning electron microscopy to measure thickness before and after compaction,
real-time compaction measurement has not been possible. This study introduces a novel
method to quantify compaction of membranes under low pressure (up to 12.5 psi) by
combining electrical impedance spectroscopy with dynamic mechanical analysis. Short- and
long-term mechanical tests were conducted to investigate instant compaction, creep, recovery,
and hysteresis. The method was validated using in situ and ex situ measurements. Results
indicate that the initial instant compaction (98% of total compaction) contributes more to total
compaction than subsequent instant compaction (from 83 to 67% of total compaction). The
ratio of creep to instant compaction is introduced as a key indicator of material resilience; its
increase from 2.4 to 49% indicates that as pressure increases, creep contributes more to total
compaction. While limited to low pressures and a dry testing environment, this novel method
shows promise for detecting and predicting performance and fatigue of water treatment
membranes in various applications.

12
2.2 Introduction and Background
Seawater desalination, wastewater treatment, potable water reuse, and resource recovery
applications are critically dependent on membrane technology [8-11]. Water treatment
membranes are thin films that are characterized by one dimension (the thickness) being
significantly smaller than the other two dimensions [13]. When pressure is applied to thin-film
materials during use or during manufacturing, the thin-film material often undergoes
compaction [20, 46-49]. For water treatment membranes, a hydraulic pressure differential
results in membrane compaction that can decrease performance and limit practical use of the
membrane [48]. In pressure-driven applications (e.g., ultrafiltration, nanofiltration, and reverse
osmosis), compaction changes the membrane morphology and internal structure [49-51] and
can cause decreased water flux, higher energy consumption [47, 48, 52, 53], and lower salt
rejection [50, 51]. In thermally driven applications (e.g., membrane distillation), membrane
compaction results in increased heat loss [5, 6, 18, 19] and reduced water permeability [5, 6,
18-23].
Compaction can occur instantly or over longer time periods and, depending on the
recovery properties of the membrane, can cause irreversible, partially reversible, or reversible
changes in membrane morphology and mechanical properties [6, 18, 19]. The recovery
properties of a membrane are often characterized by creep-recovery measurements [6, 54, 55],
whereby the membrane is exposed to long-term compressive stress (causing time-dependent
deformation), and then is allowed to recover. Recovery can also be characterized by hysteresis
curves, whereby the membrane is exposed to repeated loading and unloading forces [56].

13
Measuring deformation due to creep separately from deformation due to instant compaction is
essential to distinguish the effects of sudden pressure changes versus the effects of long-term
compressive stress on membrane performance and resiliency in water treatment applications
[6]. For example, in nanofiltration and reverse osmosis processes, the membranes are usually
pre-compacted by instant force for a short time period to prevent early flux decline and provide
stable flux during normal operation [48, 57, 58]; however, over longer-term operation, the
performance of these membranes can still be impacted by creep.
In past studies, scanning electron microscopy (SEM) has been used to measure membrane
thickness before and after compaction [6, 22, 46, 48, 59-61]. Although SEM images are useful to
visualize and quantify changes in membrane morphology and thickness, SEM images represent
the membrane state after compaction and recovery and cannot distinguish the effects of
compaction due to creep from the effects of instant compaction. Studies of water treatment
membranes have also used mechanical testing to characterize compaction [6, 49, 62, 63]. Using
dynamic mechanical analysis (DMA), material elongation due to shear strain has been used to
represent compaction of homogeneous membranes [6]. However, if membranes are not
homogeneous, compaction cannot be represented by material elongation. Also, for water
treatment membranes with low stiffness, compaction cannot be measured by DMA because
the membrane cannot support the load of the clamp during measurements. Severe noise is
typically observed and prevents acquisition of meaningful data [64]. Thus, a method to measure
compaction of water treatment membranes is needed, and even better, a method to measure
compaction in real-time is needed. Reinsch et al. [65] and Stade et al. [53] used a real-time
ultrasonic time-domain reflectometry method to measure fouling and compaction of

14
ultrafiltration membranes. However, system complexity, insufficient penetration depth of the
ultrasonic waves, and heterogeneous surface roughness (especially for polymeric thin-film
materials) reduced the accuracy and limited the application of this method.
Electrochemical impedance spectroscopy (EIS) has commonly been used to detect and
quantify membrane fouling/scaling [66-69] and membrane wetting [68-73] by measuring
system and/or cross-material electrochemical impedance changes. The potential of using EIS to
detect deformation has been preliminarily evaluated in two studies (Antony et al. [68] and
Chilcott et al. [69]) that used a dielectric model along with material and system dielectric and
conductivity measurements to detect thickness changes in reverse osmosis membranes. Due to
the limitations and complexity of the dielectric model, membrane compaction could not be
quantified.
Membrane compaction changes the internal structure of a membrane by swelling,
twinning, and kinking the polymer chains [74, 75]. Compaction brings the polymer chains closer
together, which reduces voids and air pockets, increases density, and decreases thickness [71,
72, 76]. The decreased thickness enhances the formation of conducting pathways within the
membrane and facilitates the flow of electrical current [76]. By minimizing barriers to electron
movement, the compaction process increases the overall conductivity and decreases the
electrical resistance of the membrane. Membrane conductivity changes can be detected by EIS,
and the conductivity changes can be used to quantify membrane compaction.
In this work, we combine the ability to detect membrane deformation by EIS with the
ability to apply a precisely controlled force by DMA and propose a novel method to quantify the

15
compaction of water treatment membranes. Linking EIS with DMA, we perform short-term
mechanical tests across a range of temperatures and pressures to observe real-time
compaction of the membrane. We also investigate creep, recovery, and hysteresis properties
exhibited during longer-term mechanical tests. We introduce the creep/instant compaction
ratio as an indicator of the contribution of creep to membrane compaction. Finally, we validate
our novel method in two ways: with in situ hysteresis results by comparing the irrevocable
strain with the irrevocable impedance difference and with ex situ thickness measurements by
correlating the compaction observed in SEM images with that detected by our novel method.
2.3 Materials and Methods
2.3.1 Compaction experiments
2.3.1.1 Measurement system
The measurement system (Figure 2.1) is composed of a mechanical analyzer (DMA 850, TA
Instruments, New Castle, DE), a potentiostat (Model SP-150e, BioLogic Science, Knoxville, TN),
and a pair of electrodes fabricated in our laboratory. The mechanical analyzer was used to
apply a precisely controlled force to a membrane in a temperature-controlled environment and
the potentiostat was used to detect the real-time cross-material impedance at a frequency of
100 kHz. This frequency was selected based on a previous study (Chen et al. [73]) that found
that this frequency accurately describes the impedance change of a similar membrane. In our
study, a commercial hydrophobic expanded polytetrafluoroethylene (ePTFE) membrane that is
generally used for membrane distillation applications was tested.

16
For each experiment, a membrane coupon (15 by 15 mm) was placed between the
electrodes in the compaction clamps of the mechanical analyzer and a pre-load force was
applied while initial impedance (Ω! in Ohms) was measured. Pre-loading ensures that contact is
maintained between the electrodes and the membrane surface. Subsequently, constant or
sweeping temperatures and pressures were applied to the membrane, real-time temperatures
and pressures were recorded, and final impedance (Ω" in Ohms) was measured. Compaction
(%) was calculated from impedance percent difference (%) using:
compaction = impedance percent difference = #!$#"
#!
× 100% (2.1)
Figure 2.1: Schematic of the novel measurement system used to measure membrane compaction. Stable force
and temperature control are provided by dynamic mechanical analysis; real-time cross-membrane impedance
is detected by electrochemical impedance spectroscopy. The integrated system enables quantification of
membrane compaction.
Electrochemical
Impedance
Spectroscopy
Dynamic
Mechanical
Analysis
Applies
stable force
Detects
impedance
Impedance
Electrode
Pressure
Electrode
Membrane
coupon

17
2.3.1.2 Variable-condition short-term experiments
The effects of the temperature and pressure on membrane compaction were measured in
two sets of experiments; each set was repeated four times. In the first set, nine experiments
were conducted at three temperatures (25, 45, and 65 °C) and three pressures (1, 5, and 10
psi). For each experiment, the initial impedance was measured, the force was applied for 10
min, the final impedance was measured, and the compaction was calculated.
The second set of experiments was conducted under dynamic conditions. In constant
pressure experiments, the surface pressure was kept at 5 psi, while the temperature was
“swept” from 25 to 80 °C at a ramping rate of 5 °C min-1
. The duration of each experiment was
approximately 12 min. Impedance was measured, and compaction was calculated the same as
for the first set of experiments.
2.3.1.3 Time-dependent long-term experiments
Continuous creep and recovery experiments were carried out following ASTM E139 [77]
(with a room temperature of 25 °C); experiments were repeated four times. The surface
pressure was incrementally increased from 1 to 12.5 psi over a 3 h period. The first pressure
increase was from 1 to 2.5 psi, and subsequently, pressure increases were 2.5 psi. The pressure
was then released (dropped to 0 psi), and the membrane was allowed to relax and recover.
Instant compaction (θ in %) was differentiated from compaction due to creep (ε in %). The
creep/instant compaction ratio (φ) was then calculated to assess the contribution of creep to
membrane deformation:
φ = %
& (2.2)

18
Hysteresis experiments were conducted to further evaluate the membrane's ability to
recover. A pressure-sweeping approach was used with a final pressure of 4 psi (representing a
typical operating pressure for this membrane). In these experiments, the pressure was
gradually increased from 0 to 4 psi at a ramping rate of 0.5 psi min-1
. Immediately thereafter,
the pressure was decreased back to 0 psi at a buckling rate of 0.5 psi min-1
. This ramping and
buckling process was repeated ten times per experiment, and experiments were repeated four
times. Strain was measured as the pressure was increased to 4 psi and decreased back to 0 psi.
The final strain values (when the pressure was returned to 0 psi) were averaged to determine
the irrevocable strain. Similarly, the lowest (at maximum pressure) and final (when pressure
was returned to 0 psi) impedance values were measured and the difference was calculated; the
impedance difference values were averaged to determine the irrevocable impedance
difference.
2.3.2 Method validation
The method was validated using in situ hysteresis results and ex-situ thickness
measurements. First, the irrevocable strain measured by the DMA was compared with the
irrevocable impedance difference measured by EIS. The difference between the irrevocable
strain and the irrevocable impedance difference values indicates the method's accuracy.
Additionally, after the membrane coupon was removed from the compaction clamps, it
was prepared for SEM imaging using the cryo-snap freeze-fracture method of Jones et al. [78].
The fractured membrane was then sputter-coated with nanocarbon to render the surface
conductive. SEM images were obtained by using a scanning electron microscope (Nova

19
NanoSEM 450, FEI, Hillsboro, OR). ImageJ software (version 1.51j8, National Institutes of
Health, Bethesda, MD) was used to determine the membrane thickness in three locations on a
sample; the average thickness for each sample was determined from three measurements. The
thickness of the compacted membrane (d' in µm) was subtracted from the thickness of the
pristine membrane (d( in µm) to calculate thickness percent difference (%). Compaction (%)
was calculated from thickness percent difference (%) using
compaction = thickness percent difference = )#$)$
)#
× 100% (2.3)
This thickness percent difference was then compared with the impedance percent
difference calculated from the impedance measurements. The difference between the two
values provides on additional indicator of the method's accuracy.

20
2.4 Results and Discussion
2.4.1 Membrane compaction at serval temperatures and pressures
The data in Figure 2.2 show impedance and compaction at three temperatures and under
three pressures. At 25 °C (Figure 2.2a-c), compaction increases from 16 to 34% under pressures
of 1, 5, and 10 psi. A similar trend in the effect of higher pressures resulting in greater
compaction is observed at 45 °C (Figure 2.2d-f) and 65 °C (Figure 2.2g-i). Comparing Figure 2.2a,
d, and g with temperatures of 25, 45, and 65 °C at 1 psi, it is observed that there is no clear
trend in the effect of higher temperatures on compaction at this relatively low pressure. This is
also the case at 5 psi (Figure 2.2b, e, and h) and 10 psi (Figure 2.2c, f, and i).

21
Figure 2.2: Impedance and compaction for pressures of 1, 5, and 10 psi at: (a to c) temperature of 25 °C, (d to
f) temperature of 45 °C, and (g to i) temperature of 65 °C. The shading around each line represents the
standard deviation of the data. Impedance difference (and compaction) at higher pressure is larger.
To further consider the effect of temperature on compaction, impedance was measured
continuously as temperature was increased at a rate of 5 °C /min (Figure 2.3) under a pressure
of 5 psi. As temperature increases from 25 to 80 °C, impedance decreases, but only at a rate of
76 Ohm/min. This corresponds to a decrease in compaction of 0.2%/min. Thus, impedance and
compaction decrease linearly with increased temperature but at a very slow rate. It is likely that
this trend was not apparent when analyzing the data in Figure 2.2 due to the relatively large
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Dimensionless Time
a) 25 o
C, 1 psi
Compaction 16% Compaction 26% Compaction 34%
Compaction 26% Compaction 27% Compaction 36%
Compaction 11% Compaction 29% Compaction 32%
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
b) 25 o
C, 5 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
c) 25 o
C, 10 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Dimensionless Time
d) 45 o
C, 1 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
e) 45 o
C, 5 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
f) 45 o
C, 10 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Dimensionless Time
g) 65 o
C, 1 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
h) 65 o
C, 5 psi
0.0 0.2 0.4 0.6 0.8 1.0 20000
24000
28000
32000
36000
40000
44000
Dimensionless Time
i) 65 o
C, 10 psi

22
deviation of the data points compared to the very low rate of impedance difference. For
example, if three random points are extracted from a line with a steep slope, the trend of the
slope is likely to be visible, whereas if the slope is very low and three random points are
extracted, discerning the slope's trend may be difficult. Thus, although the temperature does
directly affect membrane impedance difference, the effect is minimal at the pressures tested in
this study.
Figure 2.3: Impedance results during temperature sweeping from 25 to 80 °C with constant 5 psi pressure.
Impedance (and compaction) decrease linearly with increased temperature at a very slow rate.
0.0 0.2 0.4 0.6 0.8 1.0
20
30
40
50
60
70
80
20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Temperature o
C
Dimensionless Time
Temperature
Impedance
As temperature increases
impedance decreases at 76 Ohm/min
and compaction decreases at ~0.2% /min)
Linear fitting curve for impedance change, R2
= 0.81

23
2.4.2 Membrane creep and recovery
Membrane compaction is observed during six stepwise pressure increases (Figure 2.4).
During each step, impedance decreases (and compaction increases). When pressure is exerted
on a polymeric material, the material undergoes deformation (i.e., compaction) to absorb the
impact energy and offer resistance. This deformation is likely due to polymer fiber and
nanocomposite deformation or destruction of the crystal structure [75, 79, 80]. The impedance
decrease between the first point of a new step and the last point of the previous step
represents instant compaction. As shown in Figure 2.4, the initial instant compaction is high
(98% of total compaction) and the subsequent instant compaction values have a decreasing
trend (from 83 to 67% of total compaction). This trend signifies that the membrane initially has
the capacity to deform significantly to offset the applied pressure but as pressure increases,
irreversible damage may cause irrevocable compaction of the membrane. After the sixth stepincrease in pressure, the membrane is given a chance to recover. The membrane only partially
recovers with a final irrevocable compaction of 16%.

24
Figure 2.4: Continuous creep testing with stepwise pressure increases from 1 to 12.5 psi over a 3 h period. The
pressure was then released (dropped to 0 psi). The table embedded in the graph shows percentages of
creep/total compaction, instant/total compaction, and creep/instant compaction for each creep step. The
creep/instant compaction ratio increases from 2.4 to 49% from steps 1 to 6. The average irrevocable
compaction is 16%.
After the membrane’s instant response to compaction, the membrane deforms more
gradually as pressure continues; this time-dependent deformation is referred to as creep. The
data in Figure 2.5 show membrane creep and recovery over three cycles of operation under 5
psi pressure. In the first cycle, the membrane compacts 21.9% during the creep process and
recovers by 10.2% (resulting in 11.7% irrevocable compaction). In the second and third cycles,
Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 (%)
Instant/total
compaction 98 83 82 71 69 67
Creep/total
compaction 2.3 17 18 29 31 33
Creep/instant
compaction 2.4 21 21 41 46 49
0.0 0.2 0.4 0.6 0.8 1.0
0
2
4
6
8
10
12
14
12000
16000
20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Pressure (psi)
Dimensionless Time
Pressure
Impedance
16%
Average
Average initial impedance
Average final
Creep
Instant compaction
impedance
irrevocable
compaction

25
slightly higher compaction is observed during creep along with slightly higher irrevocable
compaction. The gradual occurrence of creep results in an ultimate irrevocable compaction of
12.5%. Membrane creep (and recovery) is observed under constant pressure. Figure 2.3
demonstrates the gradual manifestation of creep on the membrane and showcases the
method's capability of detecting subtle deformation that results from creep. Returning to
Figure 2.2, the increasing contribution of creep to total compaction (from 2.3 to 33%) over the
six-step increases indicates that the role of creep on membrane compaction increases
significantly with pressure. The increasing contribution of creep combined with the decreasing
contribution of instant compaction results in the creep/instant compaction ratio increasing
from 2.4 to 49% over the six-step increases.

26
Figure 2.5: Impedance across three creep and recovery cycles under a constant pressure of 5 psi. The gradual
occurrence of creep results in an ultimate irrevocable compaction of 12.5%.
Creep Recovery
Cycle 1
Creep Recovery
Cycle 2
Creep Recovery
Cycle 3
3 hours 6 hours 3 hours 6 hours 3 hours 6 hours
0.0 0.2 0.4 0.6 0.8 1.0
0
1
2
3
4
5
20000
24000
28000
32000
36000
40000
44000
Impedance (Ohm)
Pressure (psi)
Dimensionless Time
Pressure
Impedance
Average
irrevocable
compaction
during recovery
during creep cycle 1
12.4% 12.5%
21.9% 22.2% 22.4%
11.7%
Average initial impedance
Average
compaction
during creep cycle 1

27
2.4.3 Membrane hysteresis and in situ validation of method
The strain hysteresis of the membrane detected by DMA is shown in Figure 2.6a. After
repeated loading and unloading of (4 psi) stress, the average maximum strain is 6.6%.
Additionally, the average irrevocable strain is 2.9%. For the same experiments, the impedance
hysteresis detected by EIS is shown in Figure 2.6b. By subtracting the average minimum
impedance from the average initial impedance, the average maximum impedance difference is
18%. Furthermore, when the average final impedance is subtracted from the average initial
impedance, the average irrevocable impedance is 2.7%.
Figure 2.6: Hysteresis of the membrane is demonstrated through (a) strain hysteresis detected by dynamic
mechanical analysis and (b) impedance hysteresis detected by electrochemical impedance spectroscopy.
Comparison of the average irrevocable strain (2.9%) with the average irrevocable impedance difference
(2.7%) provides validation of the novel method.
Comparison of the impedance hysteresis, shown for the first time in this paper, with the
strain hysteresis allows for in-situ method validation. The significant difference between the
average maximum impedance difference (18%) and the average maximum strain (6.6%) is not
0 1 2 3 4 5 6 7
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Stress (psi)
Strain (%)
2.9%
6.6%
Average irrevocable strain
a) Strain hysteresis behavior b) Impedance hysteresis behavior
Average maximum strain
Average
maximum
impedance
change
0.0 0.2 0.4 0.6 0.8 1.0
20000
22000
24000
26000
28000
30000
Impedance (Ohm)
Dimensionless Time
18.2%
Average
irrevocable
impedance change
Average initial impedance
Average final impedance
Average minimum impedance
2.7%

28
unexpected because of the inherent detection limitation of the mechanical analyzer. Precise
control of the clamp is challenging when compressing membranes with low stiffness and
thickness; this leads to a discrepancy in the maximum strain measured during thin-film
compaction. On the other hand, comparison of the average irrevocable impedance (2.7%) with
the average irrevocable strain (2.9%) results in a low percent difference (approximately 7%).
During the recovery phase, no external force is applied to the clamp, which means that the
strain measurements represent the membrane properties solely. Therefore, the matched
average irrevocable strain and average irrecoverable impedance data are expected since both
values describe the same membrane hysteresis. Thus, comparison of irrevocable impedance
from EIS with irrevocable strain from DMA provides in-situ validation of the novel detection
method.
2.4.4 Ex situ validation of method using SEM images
The average thickness of the pristine membrane used in this study is 54.1 ± 4 µm (Figure
2.7a), and the average thickness of the compacted membrane, imaged in a fully recovered
state, is 52.7 ± 3 µm (Figure 2.7b). The thickness percent difference between the compacted
and pristine membranes is 2.6% or, in other words, the membrane is irrevocably compacted by
2.6%. This percent difference is very similar to the irrevocable compaction values determined
by the hysteresis tests (i.e., an irrevocable strain of 2.9% with an approximately 12% difference
and an irrevocable impedance of 2.7% with an approximately 4% difference).

29
Figure 2.7: SEM images for (a) the pristine membrane and (b) the compacted membrane imaged in a fully
recovered state after compaction under 4 psi pressure. The thickness percent difference of the compacted
membrane compared with the pristine membrane is 2.6%.
Average thickness: 54.1 ± 4 µm
Average thickness: 52.7 ± 3 µm
Thickness decrease: 2.6%
a) Pristine membrane b) Compacted membrane

30
2.5 Implications
Our novel method provides the capability of detecting compaction, even subtle
deformation caused by creep. Stepwise pressure increases that induced membrane compaction
due to both instant compaction and creep, enabled differentiation of the contribution of creep
to total compaction. The initial instant compaction contributed more to total compaction than
the subsequent instant compaction.
Comparison of hysteresis using EIS with that using DMA provided an in situ validation of
the method. The average irrevocable strain and average irrevocable impedance were found to
be in close agreement, which further validated the proposed detection method. Ex situ
validation was provided by using thickness measurements from SEM images.
Results from real-time compaction testing are essential for detecting and predicting
membrane fatigue. Due to limitations of the system components, our novel method was
applied only up to 12.5 psi, which is appropriate for membrane distillation, but represents a
relatively low pressure for water treatment applications in general. Also, our method was
tested only in a dry environment. Further investigations should build on the current method to
further develop and validate it for a broader range of water treatment membranes and
operating conditions.

31
Chapter 3 Compaction of High-Pressure Water Treatment Membranes:
Real-Time Quantification and Analysis
3.1 Abstract
Water treatment membranes play crucial roles in applications such as seawater
desalination, wastewater treatment, potable water reuse, and resource recovery. Their
performance, including transmembrane water flux and salt rejection, can be adversely affected
by compaction. Distinguishing between membrane compaction and other factors (e.g.,
membrane scaling, fouling, or reduced surface tension) affecting performance degradation,
especially during long-term higher-pressure operations, is challenging. In a prior study, we
introduced a novel method, combining electrical impedance spectroscopy with dynamic
mechanical analysis, to quantify single-layer homogenous membrane compaction under low
pressure (up to 12.5 psi). In this study, we extend the method's capabilities to quantify realtime compaction of multi-layer heterogenous nanofiltration and reverse osmosis membranes
under higher pressures, reaching up to 330 psi. By measuring membrane creep-and-recovery
and hysteresis behaviors, we quantify the longer-term response of the membrane.
Unexpectedly, our findings reveal that air pockets between the membrane layers contribute
significantly to total compaction, contrary to previous studies that generally neglected the
existence of air pockets and their contribution to membrane compaction. Finally, we compare
the measurement results obtained from our method with the membrane tensile test, a
common representation of membrane compaction in homogenous membranes. Contrary to

32
prior findings, our research reveals that this test underestimates the compaction properties of
both homogenous (e.g., single-layer membrane distillation membranes) and heterogeneous
(e.g., multi-layer NF and RO) membranes.

33
3.2 Introduction and Background
Water treatment membranes play crucial roles in applications such as water desalination,
wastewater treatment, potable water reuse, and resource recovery [8-10, 12, 81]. During
normal operations, water treatment membranes experience a hydraulic pressure differential
from the feed solution to the permeate side [48]. This hydraulic pressure differential creates a
transmembrane pressure (TMP), which is particularly significant in pressure-driven water
treatment processes such as microfiltration (MF), ultrafiltration (UF), nanofiltration (NF), and
reverse osmosis (RO) [12, 48, 61]. In these processes, TMP is essential for driving water through
the membrane, overcoming resistance, and achieving effective separation and purification [48].
However, long-term exposure to TMP can lead to membrane deformation, particularly
compaction [6, 12, 48]. This deformation, along with other factors such as membrane scaling,
fouling, and reduced surface tension, contributes to performance degradation [12, 48]. While
membrane compaction can alter membrane morphology [12, 50, 51, 53], resulting in decreased
transmembrane water flux and salt rejection [12, 48, 50, 51, 53, 79], and increased energy
consumption [47, 53]; distinguishing between the effects of membrane compaction and other
factors on performance degradation, especially during long-term higher-pressure operations, is
challenging.
Under normal operating conditions, applying higher pressure to membranes (represented
by NF and RO membranes) can result in two types of deformation: instantaneous compaction
and creep [12]. Creep is a gradual time-dependent deformation that occurs under constant
pressure [6, 75, 82]. When pressure is released, the membranes can recover, and depending on

34
their recovery properties, the combination of instantaneous compaction and creep may lead to
irrevocable, partially revocable, or fully revocable changes in membrane morphology and
mechanical properties [6, 18, 75, 82, 83]. The recovery properties of a membrane are often
characterized by creep-recovery measurements [75, 82].
In contrast to gradual, time-dependent deformations (i.e., creep), material fatigue leads to
structural and mechanical failure under cyclic loading, even when the experienced stress range
remains significantly below the static material strength [82]. While widely mentioned in metals,
material fatigue also applies to polymeric materials like membranes and plays a crucial role in
degrading membrane mechanical properties [82]. It unfolds in three distinct stages: microscopic
damage accumulates within the material's internal structure over stress loading and unloading
cycles, gradually developing until a macroscopic crack forms. As the cyclic loading continues,
this macroscopic crack grows incrementally with each stress cycle. Eventually, it reaches a
critical length, leading to component failure and material breakdown [75, 82]. As the creep
represents the constant TMP suffered by the membrane during normal operating conditions,
the effects of TMP variations and the pressure loading and unloading processes (e.g.,
membrane backwashing) can be represented by membrane fatigue. Both phenomena
significantly influence structural integrity and are crucial to ensure the longevity and reliability
of water treatment membranes.
However, unlike metal, where fatigue cracks are more evident, detecting the development
of fatigue for polymeric materials (especially thin-film composite) is challenging [84, 85].
Furthermore, material failure occurs more gradually under compressive pressure compared

35
with tensile stress (often leads to rapid fatigue development) due to the slower progression of
damage [85]. One possible way to accelerate the polymeric fatigue process is by exposing the
polymer to ultraviolet (UV) radiation [86-89]. Under UV radiation, the molecular structure of
certain polymeric bonds (e.g., C-O bond) can undergo fracture (or breakage), leading to changes
in the material’s mechanical properties [87, 89]. Adjusting the UV radiation strength and
wavelength based on the material type [86, 88] allows for a significantly shorter fatigue process
compared to material degradation under operating pressure (e.g., TMP for water treatment
membranes). An alternative approach to characterizing membrane fatigue is to analyze
deformation trends in membrane compaction resulting from both creep and membrane
hysteresis tests (subjecting the membrane to cyclic loading and unloading stress cycles), which
are proposed and are the focus herein.
Membrane compaction can be characterized by using ex-situ methods, such as scanning
electron microscopy (SEM) to measure membrane thickness before and after compaction [6,
14, 21, 22, 48, 61] and focused iron-beam SEM (FIB-SEM) [61, 83] for materials’ 3-dimensional
structure analysis, which cannot capture real-time membrane performance. Or the in-situ
dynamic mechanical analysis (DMA) to measure the material’s deformation [5, 6, 62, 63]. Given
the limitations of DMA highlighted by Ding et. al. [12], the tensile test, which measures
membrane elongation, serves as a common representation of membrane compaction in
homogenous membranes (i.e., low-pressure MF or membrane distillation membranes without a
support layer) [5, 6, 62, 63]. It has also been shown in past studies [61, 63] that compaction of
the support/backing layer is the majority of the total membrane compaction; however,
whether the contribution of the support/backing layer comes from compaction of void space or

36
solid material has not been considered. In our previous study (i.e., [12]), we introduced a novel
method that combines electrical impedance spectroscopy (EIS) with dynamic mechanical
analysis (DMA) to quantify single-layer homogenous membrane compaction under low pressure
(up to 12.5 psi). This method provides real-time and in-situ detection of even subtle material
compaction.
In this work, we extend the capabilities of our novel method to quantify real-time
compaction and recovery of water treatment membranes under higher pressures. By reaching
330 psi, our method can now characterize NF and RO membranes used in applications such as
brackish water desalination and potable water reuse. We also extend validation of the novel
method to 330 psi by correlating the irrevocable impedance difference with the irrevocable
strain and by correlating real-time compaction with that observed in SEM images. Compressive
tests that measure creep-and-recovery and hysteresis are conducted to depict immediate and
longer-term responses of the membrane to pressure. The hypothesis that membrane
compaction is solely attributable to support/backing layer compaction is tested by calculating
the thickness change of each membrane layer and comparing the results with the thickness
change of the entire membrane. Also, the contributions of the void space and solid material
components of the backing layer are delineated. Compressive testing is performed on RO
membrane samples that were age-accelerated through exposure to UV radiation. The
compaction results from compressive testing are compared to elongation results from tensile
testing, which have often been used to estimate membrane compaction.

37
3.3 Materials and Methods
3.3.1 NF and RO membranes
Commercial NF and RO membranes were used in this study. The NF270 membrane
(Sterlitech Corporation, Auburn, Washington, USA) is a thin-film composite NF membrane
comprised of a polyamide (PA) selective layer above a microporous polysulfone (PSf) support
layer, which is reinforced with a polyester (PE) non-woven backing layer. The SP475 membrane
(SUEZ Group, Paris, France) is an RO membrane comprised of the same three (PA-PSf-PE) layers.
3.3.2 Real-time compaction and creep measurement system
The real-time compaction measurement system developed in our previous study (i.e., [12])
that combines a mechanical analyzer and a potentiostat was upgraded to measure compaction
and recovery of membranes at higher pressures. The upgraded system utilizes compaction
clamps (TA Instruments, New Castle, Delaware, USA) and 3-mm CNC lathe-carved electrodes
that enable measurement at pressures up to 330 psi. Similar to our previous study, a crossmaterial impedance scanning frequency of 100 kHz and a constant temperature of 25 °C were
used for all experiments.
3.3.3 Compressive experiments
For each experiment, a 6 by 10 mm sample of membrane was placed between the
electrodes in the compressive clamps of the mechanical analyzer. A pre-load pressure of 1 psi
was applied and initial impedance (Ω! in Ohms) was measured. Short-term, creep-andrecovery, and hysteresis experiments were performed as described below. Final impedance (Ω"

38
in Ohms) was used to calculate the impedance percent difference (%) and compaction (%)
according to:
compaction = impedance percent difference =
Ω0 - Ωf
Ω0
× 100% (3.1)
3.3.3.1 Short-term compressive experiments
In the short-term compressive experiments, impedance was measured with time while
pressures of 80 and 300 psi were applied to both the NF and RO membranes for 10 min each.
Each experiment was repeated four times, and the average of the final impedance values was
used in equation 3.1.
Short-term compressive experiments were also performed on age-accelerated RO
membrane samples. To accelerate aging, pristine RO membrane samples were placed inside a
self-fabricated UV chamber and directly exposed to a 4-W UV source (UVP UVGL-25, Analytik
Jena, Upland, California, USA) for 4, 6, 24, 48, and 72 hours. A wavelength of 365 nm was
selected for use based on ASTM G151-19 [88]. The age-accelerated membrane samples were
then placed in the compressive clamp and subject to short-term compressive measurements
with 300 psi applied pressure.
3.3.3.2 Creep-and-recovery experiments
Creep-and-recovery experiments were carried out with constant pressure following ASTM
E139 [77]. In these experiments, a pressure (80 psi) for the NF membrane and 200 psi for the
RO membrane) was applied for 8 hours. After that, the pressure was released (dropped to 1 psi)
for a 16-hour recovery period. The creep-and-recovery cycle was repeated seven times. The

39
average initial and recovered impedance values were used to calculate the average irrevocable
compaction for each cycle. Increasing irrevocable compaction indicates the development of
membrane fatigue.
Creep-and-recovery experiments were also carried out with incrementally increasing
pressure. In these experiments, pressure was incrementally increased from 1 to 80 psi at 10-psi
intervals (i.e., 10, 20, …, 80 psi) for the NF membrane. For the RO membrane, pressure was
increased from 80 to 150 psi at 10-psi intervals (i.e., 80, 90, …, 150 psi) and from 150 to 250 psi
at 50-psi intervals (i.e., 150, 200, and 250 psi). At each step, pressure was kept constant for 1
hour. After the last hour at 80 psi for the NF membrane and 250 psi for the RO membrane, the
pressure was released (dropped to 0 psi) for a one-hour recovery period. The average initial and
final impedance values were used to calculate the average irrevocable compaction.
Instantaneous compaction (θ in %) was calculated as the difference between the impedance at
the last point of the previous pressure step and the impedance at the first point of the current
pressure step; compaction due to creep (ε in %) was calculated as the change in impedance
from the beginning to the end of each pressure step. The creep/instantaneous compaction ratio
(φ) was calculated to assess the importance of creep compared to instantaneous compaction
[12]:
φ = ε
θ (3.2)
Observations of φ with time are used to provide an indication of membrane fatigue with
increasing φ indicating the development of membrane fatigue.

40
3.3.3.3 Hysteresis experiments
Hysteresis experiments were conducted using a pressure-sweeping approach with a final
pressure of 80 psi for the NF membrane and 200 psi for the RO membrane. For the NF
membrane, pressure was gradually increased from 0 to 80 psi at a ramping rate of 12 psi min-1
and immediately thereafter, the pressure was decreased back to 0 psi at a buckling rate of 12
psi min-1
. For the RO membrane, the pressure was gradually increased from 0 to 200 psi at a
ramping rate of 30 psi min-1 and immediately thereafter, the pressure was decreased back to 0
psi at a buckling rate of 30 psi min-1
. For both membranes, strain and impedance were
measured as the pressure was increased to the maximum and decreased back to 0 psi. The
ramping-and-buckling cycle was repeated 10 times. The average maximum impedance
difference was calculated by subtracting the average initial impedance from the average
minimum impedance values (at maximum pressure). The average of the final strain values was
calculated and taken as the average irrevocable strain of the membrane. Increasing average
irrevocable strain and average maximum strain indicate the development of membrane fatigue.
3.3.4 SEM images
After a membrane sample was removed from the compressive clamps, it was prepared for
SEM imaging using the cryo-snap freeze-fracture method of Ferlita et al. [78]. The fractured
samples were sputter-coated with Pt/Pd for 45 s and SEM images were obtained using a
scanning electron microscope (Nova NanoSEM 450, FEI, Hillsboro, Oregon, USA). ImageJ
software (version 1.51j8, National Institutes of Health, Bethesda, MD, USA) was used to
determine the thickness of the entire membrane and the thicknesses of the individual
membrane layers in six locations on each sample. The average thickness of the pristine and

41
compacted membrane or membrane layer (d' in µm) was calculated from the six values. Using
the average thickness of the pristine membrane or membrane layer (d( in µm), thickness
percent difference (%) and compaction (%) were calculated according to:
compaction = thickness percent difference =
dp- dc
dp
×100% (3.3)
Cross-sectional SEM images were also used to identify and quantify the area in the PE
backing layer that is occupied by void space and by solid material. Void space and solid material
areas could not be identified in the PSf support layer. Figure 1a shows void spaces in the PE
backing layer (labeled as “void space 1, void space 2… void space n”). By representing the void
spaces as polygons, their areas were calculated and summed for a total void space area.
Subtracting the total void space area from the PE backing layer area results in the total solid
material area. These areas are shown as rectangular areas in Figure 1b. By using the SEM image
length (equal to 345 µm in this study), a representative void space thickness and representative
solid material thickness are shown.
Figure 3.1: (a) SEM image of PE backing layer with void spaces and solid material indicated and (b) schematic
of PE backing layer showing the void space and solid material areas as well as their representative
thicknesses.
length
representative
solid material
thickness
total void space area
length
void space 1
void space 2
… void space n
total solid material area
representative
void space
thickness
solid material 1 solid material 2 … solid material n
(a) (b)

42
3.3.5 Tensile experiments
Tensile tests were performed using film clamps (TA Instruments) instead of compressive
clamps in the mechanical analyzer. Similar to compressive testing, a pre-load pressure of 1 psi
was applied and initial deformation (X! in %) was measured. Then tensile experiments were
performed whereby stress was measured with time while a pressure of 80 psi was applied to
the NF membrane and 300 psi to the RO membrane for 5 min each. Each experiment was
repeated four times and the average of the final stress values (X" in %) was used to calculate
elongation (%) according to:
elongation = (Xf - X0) ´ 100% (3.4)

43
3.4 Results and Discussion
3.4.1 Membrane compaction under compression
3.4.1.1 NF membrane
SEM images of the pristine and compacted NF membrane are shown in Figures 2a-c. After
being subject to pressures of approximately 80 psi (Figure 3.2b) and 300 psi (Figure 3.2c), total
membrane thickness decreases and void space in the PE backing layer decreases. The PE fibrils
that can be clearly distinguished in the pristine membrane (Figure 3.2a) are no longer visible in
the compacted membranes (Figures 3.2b and 3.2c). Based on the work of Gustafson et al. [6],
the PE fibrils have likely widened, flattened, and aggregated. Also, a continuous air pocket is
observed clearly in the pristine membrane (Figure 3.2a). The air pocket is likely due to
incomplete merging of the PSf support layer when it is coated over the previously cured PE
backing layer during membrane fabrication [83]. The air pocket was unexpected, and has not
been discussed in previous compaction studies, but becomes disjointed and less visible during
compaction – especially at 300 psi (Figure 3.2c).
Thickness values of the four main components (the PSf support layer, air pocket, PE void
space, and PE solid material) corresponding to the SEM images are shown in Figures 3.2d-f.
(The thickness of the PA selective layer is on the nm scale and is ignored in this study.) The PE
void space thickness was distinguished from the PE solid material thickness in the backing layer
using the method described in Section 3.3.3.1. The same method was used to quantify the
thickness of the air pocket. Comparing results at the two applied pressures (i.e., comparing
Figures 3.2e and f), the only membrane component that undergoes significantly more

44
compaction at the higher pressure is the PE void space. Total membrane thickness is also given;
it decreases by 22 µm (19%) and 29 µm (22%) at 80- and 300-psi applied pressures.
Figure 3.2: SEM images of the cross-sectional morphology of the (a) pristine NF membrane, (b) NF membrane
tested under 80-psi pressure, and (c) NF membrane tested under 300-psi pressure. Corresponding values of
layer/component and total membrane thickness for (d) the pristine NF membrane, (e) the NF membrane
tested under 80-psi pressure, and (f) the NF membrane tested under 300-psi pressure. A continuous air pocket
is observed clearly in the pristine membrane; it becomes disjointed and less visible during compaction –
especially at 300 psi.
The relative thickness of each membrane component is determined by dividing the
thickness of each component by total membrane thickness (Figure 3.3a). Thickness values for
the compacted membranes are normalized by dividing by total compacted membrane
thickness. As can be seen in the Figure, the relative contributions of the PSf layer and PE solid
material to total membrane thickness increase with compaction and the relative contributions
of the air pocket and PE void space generally decrease. An increase in relative contribution
indicates that the component thickness is increasing at a slower rate than total membrane
thickness, meaning that the thickness of these layers/components decreases relatively less than
PSf support layer
PE backing layer
continuous
air pocket
disjointed air pockets
PE fibril void space aggregated PE fibrils void space
becomes smaller
PE solid material
PE void space
air pocket
PSf support layer 46 μm
12 μm
20 μm
61 μm
41 μm
8 μm 9 μm
59 μm
40 μm
7 μm 6 μm
57 μm
Pristine 80 psi applied pressure 300 psi applied pressure
(a) (b) (c)
(d) (e) (f)
PA selective layer
138 ± 5 μm 116 ± 2 μm 109 ± 2 μm
PE backing layer

45
the entire membrane. A decrease in relative thickness contribution indicates the opposite,
meaning that the thickness of these layers/components decreases relatively more than the
entire membrane.
In Figure 3.3b, the relative compaction of each membrane component is determined by
dividing compaction values for each component by total membrane compaction. The relative
compaction of the layers/components are generally similar for 80 and 300 psi. It can be seen
that the greatest relative compaction (up to 45%) occurs in the PE void space, which only
accounts for 14% of the total pristine membrane thickness. The combined compaction of the PE
void space and the air pocket accounts for over 60% of total membrane compaction under 300
psi applied pressure. The fact that the majority of the compaction occurs in the void space and
air gap is not necessarily a surprise, it just has not been demonstrated in the past. And with 18%
relative compaction, the air pocket is significant and cannot be ignored.

46
Figure 3.3: SEM images of the cross-sectional morphology of the (a) pristine NF membrane, (b) NF membrane
tested under 80-psi pressure, and (c) NF membrane tested under 300-psi pressure. Corresponding values of
layer/component and total membrane thickness for (d) the pristine NF membrane, (e) the NF membrane
tested under 80-psi pressure, and (f) the NF membrane tested under 300-psi pressure. A continuous air pocket
is observed clearly in the pristine membrane; it becomes disjointed and less visible during compaction –
especially at 300 psi.

47
3.4.1.2 RO membrane
SEM images of the pristine and compacted RO membrane are shown in Figures 4a-c.
Compared to the pristine NF membrane (Figure 3.2a), the pristine RO membrane (Figure 3.4a)
appears denser with less void space. Similar to the NF membrane, the air pocket that is
observed clearly and continuously in the pristine membrane (Figure 3.4a) becomes disjointed
and less visible – especially at 300 psi (Figure 3.4c). And the PE fibrils that can be clearly
distinguished in the pristine membrane (Figure 3.4a) are no longer visible in the compacted
membranes (Figures 3.4b and 3.4c). Values of layer thickness corresponding to the SEM images
are shown in Figures 3.4d-f. Compared to the pristine NF membrane (Figure 3.2d), the pristine
RO membrane (Figure 3.4d) has a smaller air pocket and less PE void space. All
layers/components of the RO membrane undergo more compaction at 300 psi (Figure 3.4c)
than 80 psi (Figure 3.4b). Total membrane thickness decreases by 15 µm (13%) and 29 µm
(21%) at 80- and 300-psi applied pressures.

48
Figure 3.4: SEM images of the cross-sectional morphology of the (a) pristine RO membrane, (b) RO membrane
tested under 80-psi pressure, and (c) RO membrane tested under 300-psi pressure. Corresponding values of
layer/component and total membrane thickness for (d) the pristine RO membrane, (e) the RO membrane
tested under 80-psi pressure, and (f) the RO membrane tested under 300-psi pressure. Compared to the
pristine NF membrane, the pristine RO membrane appears denser with less void space and all layers undergo
more compaction at 21 bar (300 psi) than 80 psi.
In Figure 3.5a, results are similar to those for the NF membrane in that the thicknesses of
the PSf layer and PE solid material decrease relatively less than the entire membrane and the
thicknesses of the air pocket and PE void space decrease relatively more. In Figure 3.5b, it can
be seen that the greatest relative compaction (up to 40%) occurs in the PE solid material, which
also accounts for 47% of the total pristine membrane thickness. The ability of the solid material
to compact more, and counterbalance the applied pressure more, at 300 than 80 psi was
unexpected. Also, at 300 psi, more relative compaction occurs in the PE void space than in the
PSf support layer (28% compared to 21%) while the PSf support layer accounts for 41% of total
compacted membrane thickness and the PE void space only accounts for 5%. The decrease in
the relative compaction of the air pocket (from 14 to 11%) suggests that the air pocket is likely
Pristine 80 psi applied pressure 300 psi applied pressure
PE backing layer
PE solid material
PE void space
air pocket
PSf support layer 50 μm
8 μm 12 μm
63 μm
46 μm
6 μm 8 μm
58 μm
43 μm
5 μm 5 μm
51 μm
(a) (b) (c)
(d) (e) (f)
PSf support layer
PA selective layer
133 ± 2 μm 118 ± 3 μm 104 ± 4 μm
continuous
air pocket
disjointed air pockets
PE fibril aggregated PE fibrils void space void space
becomes smaller
PE backing layer

49
fully compacted at 80 psi and does not undergo significant further compaction at 300 psi. At the
higher applied pressure of 300 psi, there is essentially no additional relative compaction of the
combined air pocket and PE void space even though there is 8% greater total membrane
compaction.
Figure 3.5: (a) Relative thickness and (b) relative compaction of RO membrane layers/components. The
greatest compaction occurs in the PE solid material, which further compacts at 300-psi to counterbalance the
applied pressure.

50
3.4.2 Membrane creep-and-recovery behavior
3.4.2.1 Creep-and-recovery experiments with constant pressure
The data in Figure 3.6 show NF and RO membrane creep-and-recovery over seven cycles of
operation. For the NF membrane (Figure 3.6a) membrane compacts 15.0% and recovers 10.8%
in the first cycle; the result is 4.2% irrevocable compaction. Because the RO membrane was
tested under a higher pressure (200 psi), greater membrane compaction in the first cycle
(20.9%) was observed compared to that in the NF membrane. Over the subsequent cycles for
both membranes, irrevocable compaction shows a gradually increasing trend, indicating that
the membranes experience greater permanent deformation and possible morphological
changes. This trend also points to the gradual development of membrane fatigue.

51
a) NF membrane

52
Figure 3.6: Normalized impedance across seven creep and recovery cycles of (a) the NF membrane under a
constant pressure of 80 psi and (b) the RO membrane under a constant pressure of 200 psi. The gradual
occurrence of creep is an indication of the development of membrane fatigue, which leads to a final
irrevocable compaction of 5.1% for the NF membrane and 5.0% for the RO membrane.
Due to equipment limitations in holding precise forces over longer time periods, the test
duration was limited to seven cycles (equivalent to 168 h) for each membrane sample. To gain
insight into the effect of pressure on membrane compaction over longer durations,
compressive experiments were also performed on age-accelerated RO membrane samples. As
shown in Figure 3.7, the pristine RO membrane undergoes 20.6% average compaction, which is
consistent with the results shown in Figure 3.6b (20.9% average compaction). For the aged
b) RO membrane

53
membrane samples, compaction increases with the duration for which the samples were
exposed to UV radiation. Thus, aging the membrane samples makes them more susceptible to
membrane fatigue.
Figure 3.7: Normalized impedance and corresponding compaction of pristine and aged RO membrane
samples under a constant pressure of 200 psi. Membrane samples were aged by exposing them to UV
radiation for 4, 6, 24, 48, and 72 h. RO membrane deformation (compaction) increases with the duration for
which the samples were exposed to UV radiation likely due to the development of membrane fatigue.

54
3.4.2.2 Creep-and-recovery experiments with incrementally increasing pressure
Figure 3.8a shows how instantaneous compaction and creep contribute to NF membrane
compaction during incremental pressure increases. The initial high decrease in impedance
indicates instantaneous compaction for step 1 that accounts for 93% of the compaction in this
step. In subsequent steps, the contribution of instantaneous compaction decreases (reaching
51% in step 8). The initial high instantaneous compaction is likely because the free-standing
membrane fibrils have more degrees of freedom and are more susceptible to deformation (i.e.,
compaction) when pressure is first applied to the membrane. Then, when pressure is held
constant, impedance decreases slowly due to creep, which accounts for 7% of the compaction
in the first step. As the pressure increases in subsequent steps, the membrane fibrils likely
break and aggregate to counterbalance the applied pressure, which can lead to membrane
collapse [75, 79, 82]. As the packed/collapsed membrane becomes denser, membrane stiffness
likely increases and can provide greater resistance to pressure than the pristine membrane; this
is referred to as “the stiffening effect” [90]. The contribution of creep in subsequent steps,
increases from 7 to 49% as shown in Table 3.1. The increasing contribution of creep and
decreasing contribution of instant compaction results in the creep/instant compaction ratio
increasing from 8% to 94% over the eight step increases. The time-dependent deformation
associated with creep indicates the development of membrane fatigue.

55
a) NF membrane

56
Figure 3.8: Creep-and-recovery behavior with incrementally increasing pressure for (a) the NF membrane with
stepwise pressure increases from 10 to 80 psi. The role of creep increases with each step, while the
instantaneous compaction decreases with each and (b) the RO membrane with stepwise pressure increases
from 80 to 150 psi from steps 1-8; two additional steps (steps 9 and 10) were applied at a higher-pressure
(200 and 250 psi). For both membranes, the role of instantaneous compaction decreases, and the role of
creep increases with each step.
Table 3.1: Percentages of NF membrane compaction contribution from instantaneous compaction and creep.
Compaction percentage (%) Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 Step 7 Step 8
Instantaneous 93 75 70 70 69 55 54 51
Creep 7 25 30 30 31 45 46 49
φ 8 34 42 44 45 80 85 94
b) RO membrane

57
Similar results were seen for the RO membrane in Figure 3.8b which includes two
additional steps (steps 9 and 10) at a higher pressure-interval (50 psi/step). Similar to the NF
membrane, the initial instantaneous compaction is much higher than subsequent ones. Then,
over steps 1 to 8, instantaneous compaction decreases from 95 to 59% of total compaction. At
the same time, creep increases from 5 to 41% (shown in Table 3.2.). Consequently, the
creep/instantaneous compaction ratio increases from 5 to 68% over steps 1 to 8. Additionally,
the average irrevocable compaction of 10% is higher than the 3% irrevocable compaction
observed for the NF membrane. This indicates that the RO membrane operated under a typical
operating pressure of 300 psi experiences greater permanent deformation (and possible
greater morphological changes) than the NF membrane operating under a typical operating
pressure of 80 psi.
Table 3.2: Percentages of RO membrane compaction contribution from instantaneous compaction and creep.
Compaction
percentage (%)
Step
1
Step
2
Step
3
Step
4
Step
5
Step
6
Step
7
Step
8
Step
9
Step
10
Instantaneous 95 85 82 77 71 57 59 59 88 83
Creep 5 15 18 23 29 44 41 41 12 17
φ 5 18 22 30 41 76 69 68 13 20

58
3.4.3 Membrane hysteresis and method validation
3.4.3.1 Membrane hysteresis and in-situ validation
The strain hysteresis behavior of the NF membrane detected by the mechanical analyzer is
shown in Figure 3.9a in the Supporting Information. After repeated loading and unloading of 80
psi stress, the average maximum strain is 2.7% and the average irrevocable strain is 2.2%. For
the same experiments, the impedance hysteresis detected by the potentiostat is shown in
Figure 3.9b. By subtracting the average minimum impedance from the average initial
impedance, the average maximum impedance change is 7.8%. Furthermore, when the average
final impedance is subtracted from the average initial impedance, the average irrevocable
impedance is 2.6%. Comparison of the average irrevocable strain (2.2%) with the average
irrevocable impedance difference (2.6%) provides in-situ method validation. Also, the gradual
increases in the average maximum strain and average irrevocable strain for each ramping and
buckling cycle indicate the development of membrane fatigue.

59
Figure 3.9: Hysteresis of the NF membrane is demonstrated through (a) strain hysteresis detected by dynamic
mechanical analysis and (b) impedance hysteresis detected by a potentiostat. Comparison of the average
irrevocable strain (2.2%) with the average irrevocable impedance difference (2.6%) provides in-situ method
validation.
The strain and impedance hysteresis behaviors of the RO membrane are shown in Figures
3.10a and 3.10b. Comparison of the average irrevocable strain (7.8%) with the average
irrevocable impedance difference (5.6%) provides in-situ method validation. The difference
between these two values is likely due to measurement noise under high pressure. Also, the
gradual increases in the development of membrane fatigue is indicated.
(a) Strain hysteresis behavior (b) Impedance hysteresis behavior

60
Figure 3.10: Hysteresis of the NF membrane is demonstrated through (a) strain hysteresis detected by
dynamic mechanical analysis and (b) impedance hysteresis detected by a potentiostat. Comparison of the
average irrevocable strain (2.2%) with the average irrevocable impedance difference (2.6%) provides in-situ
method validation.
3.4.3.2 Ex-situ method validation
From SEM images (Figure 3.11), the thickness percent difference between the compacted
and pristine NF membrane is 3.4%, and the thickness percent difference between the
compacted and pristine RO membranes is 5.2%. These values are comparable with the average
irrevocable strain of the NF membrane (2.2%) in Figure 3.9a and the average irrevocable strain
of the RO membrane (7.8%) in Figure 3.10a.
(a) Strain hysteresis behavior (b) Impedance hysteresis behavior

61
Figure 3.11: SEM images for (a) the pristine NF membrane, (b) the compacted NF membrane imaged in a fully
recovered state after hysteresis experiments with 80 psi applied pressure; the thickness percent difference of
the compacted membrane compared with the pristine membrane is 3.4%, (c) the pristine RO membrane, and
(d) the compacted RO membrane imaged in a fully recovered state after hysteresis experiments with 200 psi
applied pressure; the thickness percent difference of the compacted membrane compared with the pristine
membrane is 5.2%.
(a) Pristine NF membrane (b) Compacted NF membrane
average thickness: 138 ± 5 μm
average thickness: 134 ± 1 μm
thickness decreases: 3.4%
(c) Pristine RO membrane (d) Compacted RO membrane
average thickness: 133 ± 2 μm average thickness: 126 ± 3 μm
thickness decreases: 5.2%

62
3.4.4 Difference between compressive and tensile tests for water desalination membranes
Figures 3.12a and 3.12b illustrate the compaction and strain data for multi-layer,
heterogeneous membranes (i.e., NF and RO membranes). As expected, tensile tests for both NF
(tested under 80 psi applied pressure) and RO membranes (tested under 300 psi pressure)
exhibit an underestimation of average compaction, resulting in a 13 % and 19% difference in
membrane deformation between tensile and compressive tests. Therefore, using tensile test
strain results to estimate the compressive compaction results for both multi-layer,
heterogeneous membranes is inaccurate. Surprisingly, when tested under a typical operating
pressure of 5 psi, the single-layer, homogeneous low-pressure membrane (used in Ding et al.
[12]) also exhibited a significant underestimation of the average compaction in tensile tests,
with a 28% difference in membrane deformation (shown in Figure 3.12c). The “stiffening effect”
theory, discussed in Section 3.4.2.2, can be one of the reasons for this underestimation. The
paper reveals, for the first time, the critical importance of prioritizing compressive testing over
tensile testing for studying membrane compaction and accurately characterizing its mechanical
properties.

63
(a) NF membrane

64
(b) RO membrane

65
Figure 3.12: Tensile and compressive deformation for (a) low-pressure membranes, (b) NF membranes, and
(c) RO membranes. Average compaction measured by compressive tests is different from the average strain
measured by tensile tests for all three types of membranes.
(c) Low-pressure membrane

66
3.5 Implications
In this study, we successfully extended the capabilities (and validation) of our novel
method to quantify real-time compaction and recovery of water treatment membranes under
higher pressures. We validated this method up to pressures of 330 psi, providing a robust
correlation between irrevocable impedance difference and irrevocable strain, as well as
between real-time compaction and SEM observations. This advancement enables accurate
quantification of compaction and recovery for NF and RO membranes used in brackish water
desalination and potable water reuse applications.
Our findings demonstrate that membrane compaction is not solely attributable to
support/backing layer compaction. The air pockets in both NF and RO membranes, often
overlooked in previous studies, play a crucial role in membrane compaction and should not be
ignored. We also revealed that the void space in the PE backing layer contributes most to NF
membrane compaction, while the solid material in the PE backing layer contributes most to RO
membrane compaction.
We also found that instantaneous compaction is likely influenced by the stiffness of the
material fibrils, while longer-term creep is influenced by the increased density of the
pack/collapsed membrane. We propose that the development of membrane fatigue during
typical operation (i.e., induced solely by TMP) can be indicated by several metrics. These
include the increased average irrevocable compaction values in the creep-and-recovery
experiments, the increased creep/instantaneous compaction ratio (φ), and the increased

67
average maximum strain and average irrevocable strain in the hysteresis experiment.
Furthermore, susceptibility to membrane fatigue increases with membrane age.

68
Chapter 4 Analysis of Anthropogenic Waste Heat Emission from an
Academic Data Center
4.1 Abstract
The rapid growth in computing and data transmission has significant energy and
environmental implications. While there is considerable interest in waste heat emission and
reuse in commercial data centers, opportunities in academic data centers remain largely
unexplored. In this study, real-time onsite waste heat data were collected from a typical
academic data center and an analysis framework was developed to determine the quality and
quantity of waste heat that can be contained for reuse. In the absence of a comprehensive
computer room monitoring system, real-time thermal data were collected from the data center
using two arrays of thermometers and thermo-anemometers in the server room. Additionally, a
computational fluid dynamics model was used to simulate temperature distribution and
identify “hot spots” in the server room. By simulating modification of the server room with a
hot air containment system, the return air temperature increased from 23 to 46 ◦C and the
annual waste heat energy increased from 377 to 2004 MWh. Our study emphasizes the
importance of containing waste heat so that it can be available for reuse, and also, that reusing
the waste heat has value in not releasing it to the environment.

69
4.2 Introduction
The world's continued and strong demand for access to digital and virtual media has led to
rapid growth in computing, data transmission, and storage over the last decade [91]. Because
of this, data centers have increased in prevalence and size. Internet data transmission for video
streaming, virtual conferencing, and online shopping surged even more during the COVID-19
pandemic [92] and has continued well after. The proliferation of emerging computational tools
(e.g., machine learning, data mining, hyper-scale cloud computing, and blockchain distribution)
has intensified the demand for computing resources [92]. Most recently, the advent of
sophisticated artificial intelligence and advanced deep learning technologies, represented by
ChatGPT, will further increase computing demand due to data-intensive architectures and
learning models [93].
Tighter connections between the digital world and the physical world are intensifying
concerns associated with data center energy and resource consumption [91, 92, 94]. Electricity
consumption by data centers around the world has increased dramatically every year from
2010 to 2020 and is currently uncapped [92]. In 2021, it was estimated that the worldwide
electricity consumption of data centers was between 220 and 320 TWh, which corresponded to
0.9-1.3% of global total electricity demand [95].
The major approach that has been taken to improve data center energy efficiency is to
upgrade information and communication technology (ICT) equipment. This has kept the growth
in world electricity demand relatively low (only increasing from 1 to 1.1%) while global Internet
users have doubled and global Internet data transmission has increased 15-fold [95]. However,

70
increasing energy efficiency by improving efficiency of ICT equipment is approaching theoretical
limits [92, 95]. Until new computing technologies, such as quantum computing, become
available, further advances in energy efficiency can only be expected from improvements in
cooling and ventilation equipment, optimization of room configuration and flow, and reuse of
waste heat [92]. Use of renewable energy and reuse of waste heat can also contribute to
building decarbonization, which is a key sustainability goal [92, 95].
Waste heat, which is currently considered to be an undesired by-product of data center
operation, can be a resource if the decision is made to reuse the waste heat. In more traditional
industries (e.g., petroleum/steel, cement, and chemical industries), waste heat has been
quantified and reused for many decades; in the ICT industry (and especially data centers),
waste heat emission and its reuse have only become of interest in the recent decade to
improve overall energy efficiency.
Power usage effectiveness (PUE) has long been the most popular index used to represent
energy efficiency of data centers [91, 96, 97]. PUE is the ratio of total facility power to ICT
equipment power; the closer PUE is to 1, the more efficient the data center is considered to be
[96]. In 2020, the average PUE value for typical data centers in the US was 1.6 [98]. In
comparison, the best hyper-scale data centers run by Google, Microsoft, and Amazon achieved
PUE values of 1.1 [92, 95, 99], 1.25 [99], and 1.14 [99]. Large-scale data centers in the private
sector following LEED design concepts reported an average PUE value of 1.38 in 2022 [98].
Academic data centers are typically known to be less efficient. In 2014 Choo et al. [100]
reported a PUE of 2.73 and in 2020, Dvorak et al. [101] reported a PUE of 1.78 for the academic
data centers they studied.

71
At the theoretical limit for PUE (i.e., at a PUE of 1.0), all server room power is used by the
ICT equipment; none is used for facility infrastructure. For hyper-scale data centers, this means
that the power for facility infrastructure is provided entirely by renewable energy or waste heat
from the ICT equipment. It does not necessarily mean the data center energy consumption has
been reduced only that waste heat or renewable energy is used to power the auxiliary
equipment [96, 102]. One way to reduce data center energy consumption is to reduce cooling
system energy consumption [103, 104]. Reducing cooling system energy consumption improves
cooling system efficiency (CSE), with a CSE less than 1 representing efficient cooling [105-107].
Achieving a CSE “good practice” benchmark of 0.8 kW per ton of cooling load (KW/ton) [105-
107] along with a PUE close to 1 would indicate a more sustainable data center [108].
Amazon, Google, and Facebook are now the top three purchasers of renewable power in
the United States. Because the only power these companies purchase is renewable power, they
have achieved zero net carbon emissions [95]. Although more than 97% of inlet energy to data
centers becomes low-grade waste heat (less than 3% is applied to useful work) [103, 109, 110],
these companies do not reuse waste heat to power their ICT equipment. Instead, they upgrade
the waste heat quality and then use it for district heating or other auxiliary equipment [92].
Thus, additional energy input is required (e.g., from a heat pump) for the large quantity of
waste heat to be reused, but not only is the waste heat reused, the cooling load is decreased
because less heat has to be cooled by computer room air-conditioner (CRAC) units.
District heating is the heating of commercial and/or residential buildings using waste heat
from industrial processes and power generation or using renewable sources. It is the most
widely used method to reuse waste heat in commercial data centers [109, 111-118]. Stockholm

72
Data Parks (Stockholm, Sweden) is a program that brings waste heat from data centers to the
Stockholm district heating network; in 2022, over 100 GWh of waste heat was used to meet
Stockholm’s heating needs [117]. Amazon’s Headquarters (Seattle, Washington, USA) also
operates a waste heat reuse system for the Westin Building Exchange (a data center) [118].
Because the exhausted temperature can only heat water to around 18 °C, instead of direct
usage, Amazon passes the water through five heat-reclaiming chillers to concentrate the heat
and raise the temperature to 54 °C to meet local district heating requirements. Another Nordic
case study is a 10-MW Yandex data center in Mäntsälä, Finland, which uses a heat exchanger to
warm water to approximately 30-45 °C for use in district heating [111]. Also in Finland,
Microsoft and Fortum are planning to build data centers in the Helsinki metropolitan area; the
goal is for the waste heat from the data centers to eventually provide 40% of heating needs
[119]. Other examples for how waste heat from data centers can be reused include preheating
a swimming pool [120, 121], warming a greenhouse [122], and providing warm water for
aquaculture [123, 124].
In older data centers, lack of a computer room monitoring system (CRMS) is a barrier to
accurately assessing cooling and energy efficiencies. For example, a case study from a 2019
report of the Data Center Optimization Initiative [125] shows that five of 24 data centers failed
to prove they were meeting energy-efficiency goals because they did not have effective
monitoring systems to track their energy efficiency [125, 126]. Not only do academic data
centers often lack advanced CRMS, they often have ICT equipment that is older with a slower
pace of upgrade/replacement, and are mostly equipped with outdated air-cooling systems. This
is probably a major contributing factor to why case studies focusing on academic data centers

73
are rarely reported in the literature. And, the older ICT equipment and air-cooling systems
disperse the waste heat to the server room, which is counter to the primary task of collecting
waste heat if it is going to be reused.
In reviewing the literature over the last five years, we found eleven case studies on
academic data centers; of these, only five (i.e., [101, 127-130]) analyzed reuse of waste heat. In
all five studies, the waste heat that can be reused from the academic data center was
determined using analytical methods with waste heat temperatures that were calculated based
on consumer use of the data center rather than measured in real time. Calculated
temperatures ranging from 25 to 45 °C for air- and liquid-cooling scenarios [101, 127] resulted
in annual waste heat quantities ranging from 3156 – 11045 MWh. District (campus) heating was
the only application discussed. None of the studies confirmed their calculated average return
air temperatures with actual temperature measurements.
In the current study, we collected onsite waste heat data and developed an analysis
framework to estimate energy efficiency and quantify the waste heat from a typical academic
data center. In the absence of a comprehensive CRMS, real-time thermal data were collected
from the data center using two arrays of thermometers and thermo-anemometers in the server
room. Using this data, cooling and energy efficiencies for the server room were calculated. We
also evaluated data center operating trends. A CFD model was then developed to simulate and
predict the global temperature distribution of the server room. The results of the simulations
were used to identify regions of minimum cooling efficiency (“hot spots”) and estimate the
quality and quantity of anthropogenic waste heat emission from the data center. We also

74
simulated a widely applied modification, a hot air containment (HAC) system; HAC systems
have been shown to improve cooling efficiency, but this study shows how containing waste
heat is critical to increase waste heat quality. To our knowledge, we are the first study to
present measured real-time data of waste heat generated and collected in an academic data
center. Making this data available and analyzing it for key metrics (waste heat data quality and
quantity) is an important first step in improving the energy efficiency of similar facilities.
4.3 Materials and Methods
4.3.1 Server room selected for simulation
In this study, the High-Performance Computing (HPC) server room of the Center for
Advanced Research Computing (CARC) at the University of Southern California (USC) (Los
Angeles, CA) was used to represent a typical academic data center. The plan area of the server
room is 288 m2
, with dimensions of 18, 16, and 4 m in length, width, and height. The room
contains 2,500 computer servers and several storage systems as well as high-speed network
equipment in a five-row (A to E) configuration. The server room is cooled by nine CRAC units
(seven in operation and two idle) that operate with air-based ventilation systems. Further
details of the server room are provided in Table 4.1. Unless otherwise stated, “server room” in
this paper refers to the HPC server room at USC.

75
Table 4.1: Rack group and CRAC information.
Rack Group Servers per Rack group CRAC number Operating Status
A 268 CRAC-1 ON
CRAC-2 ON
B 592 CRAC-3 OFF
CRAC-4 ON
C 588 CRAC-5 ON
CRAC-6 ON
D 436 CRAC-7 OFF
CRAC-8 ON
E 594 CRAC-9 ON
4.3.2 Air-cooling system in the server room
Fig. 4.1a is a plan view of the server room showing the air-cooling system. The CRAC units
cool air to 18 °C and emit the chilled air into the cold aisles of the server room via ceiling and
ground vents. The server room has 72 ceiling cold vents (24 for each cold aisle) and 72 floor
cold vents (24 for each cold aisle); each vent has length and width dimensions of 0.25 m. The
servers in the rack groups draw the chilled air into their central processing units and graphic
processing units, and at the same time, emit hot exhaust that first mixes with the air in the hot
aisle and is then drawn to the CRAC units via ceiling hot vents. The server room has 15 ceiling
hot vents (5 for each hot aisle) with length and width dimensions of 0.58 m. The server room
air-cooling system keeps the whole server room temperature relatively constant at
approximately 21 °C. The cooling process is closed-loop; the CRAC units do not ventilate
external air into the computer room, and thus, the airflow rate of warm return air entering the
CRAC units, and the airflow rate of chilled air emitted from the CRAC units are equal.
The energy required by the CRAC units is based on the temperature difference (∆T) shown
in Fig. 4.1b. ∆T is the temperature difference between the inlet (i.e., return air shown by the

76
red line) and outlet (i.e., chilled air shown by the blue line). The temperature difference
describes the quality of the potential free energy that will be emitted as waste heat. The
quantity of the waste heat can be estimated based on the heat from the warm return air. If that
waste heat is then transferred from the data center via a heat sink or heat exchange device, the
waste heat can be reused. However, even if the heat is just emitted to the ambient
environment, the cooling load on the CRAC units gets reduced because taking the waste heat
out of the server room (before it enters the CRAC units) decreases the return hot air
temperature.

77
Figure 4.1: (a) Section view of the server room. Air is circulated by computer room air-conditioning (CRAC)
units that draw hot exhaust generated by the servers into the warm return air channel and blow chilled air
back into the server room via ceiling and ground vents. (b) Schematic diagram of server room indicating the
temperature difference between the warm returned air and chilled air in the CRAC units (∆T).
a)
b)

78
4.3.3 Analysis framework and collection of primary data
The analysis framework developed for typical academic data centers is shown in Figure 4.2
and detailed in the subsequent sections.
Figure 4.2: Analysis framework to estimate energy efficiency and anthropogenic waste heat generated by a
representative academic server room.

79
Because this server room was built in 2010 and does not utilize CRMS technology, primary
data collection is the first step in the analysis framework. Primary data collected for the analysis
includes a) row-wise temperature (Tri in ℃) and flow rate (qri in m3
/s), b) random
temperature (Txi in ℃) and flow rate (qxi in m3
/s), where i represents one of the A to E rack
groups in the server room, c) hot and cold vent temperature (Tvi in ℃) and flow rate (qvi in
m3
/s), d) electricity consumption over data collection time, t (Qt in kWh), and e) computing load
(%; indicating mean CPU occupation as % of total capacity).
Row-wise thermal data was collected every ten minutes over one month using an array of
15 thermometers and 10 fabricated thermo-anemometers that was placed in the rack groups of
the server room. The automated thermo-anemometers were each fabricated with a single-chip
microcomputer (Raspberry Pi 4.0, UK), two temperature probes (Raspberry Pi Foundation, UK),
one anemometer, and two long-lasting batteries (PKCell, China). The thermometers and
thermo-anemometers were coded with Python 3.0 to transmit and store the thermal data. The
devices were installed following the “4-Points Rules” [131], which say that thermal data should
be monitored at both low and high rack positions on both cold and hot aisles. Placement of the
thermometers and anemometers in one of the hot-aisle racks is shown in Fig. 3a. A
thermogram of the rack taken using FLIR ONE Pro by Teledyne FLIR (Wilsonville, OR) shows the
temperature distribution of the rack (Fig 3b).

80
Figure 4.3: (a) Photograph of fabricated thermo-anemometer installed in a rack facing a hot aisle; (b) infrared
photograph of the rack taken by FLIR ONE Pro, which shows the highest temperature as 41.5 ℃.
Random and hot- and cold-vent thermal data were measured intermittently over one
month using hand-held, self-powered, thermo-anemometers (Digi-Sense Hot-Wire Thermoanemometer UX-20250-16, Switzerland). The accuracy of temperature measurement by the
thermometers and thermo-anemometers is reported as 0.1 °C by the manufacturers. The
accuracy of air velocity is reported as 0.01 m/s by the manufacturer. Systematic errors of ±
1.5 °C for the air temperature and ± 0.04 m/s for the air velocity (calibrated with nominal
airflow velocity at 4.00 m/s) provide an indication of uncertainty in the measured data. The
instruments were calibrated prior to use by the manufacturers at 23 °C under 54% RH and were
tested in the data center for two weeks prior to data collection. The hand-held thermoanemometers were used to collect thermal data at random locations within the hot aisle of
each rack group (Table 4.2.) to measure the potential waste heat quality and quantity emitted
Long
lasting
batterie
s
Anemometer
Single-chip
microcomputer
and on-board
temperature
probe in
high position
Long-range
temperature
probe in
low position
41.5 ℃

81
by the servers. Average values of temperature and flow rate for the hot exhaust leaving
through ceiling vents and chilled air entering through cold vents on the floor and ceiling are
given in Table 4.3.
Table 4.2: Average temperatures and airflow rates at random locations in the hot aisles of the server
room.
Hot aisle temperature
(°C)
Airflow rate
(m3
/s)
Location 1 (in rack group A) 22.6 7.61
Location 2 (in rack group B) 31.0 16.4
Location 3 (in rack group C) 30.0 6.54
Location 4 (in rack group D) 45.4 3.66
Location 5 (in rack group E) 31.4 11.2
Average 32.1 9.07
Table 4.3: Average temperatures and flowrates entering the hot vents and exiting the cold vents.
Temperature
(°C)
Airflow rate
(m3
/s)
Hot vents ceiling 24.1 0.93
Cold vents floor 13.0 0.91
Cold vents ceiling 15.0 0.74
Electricity consumption for the server room was obtained from the facility’s monthly
electricity bill. The average monthly electricity consumption for the server room is 354,050
kWh/month with an average electricity consumption of 177,025 kWh/month for the CRAC units
specifically.
The computing load was obtained from the server room's operating log that is documented
on Grafana (version 7.4.3, Grafana Labs, New York, NY). Representative durations (including
one day, one week, and one month) were selected to represent peak scenarios (i.e., weekdays

82
and mid-semester) and off-peak scenarios (i.e., weekends and off-semester). The data were
used to identify operating trends for the data center. a) b)

83
4.3.4 Calculation of cooling and energy efficiency indices
The first question in the framework (Figure 4.1) (“Is the server room cooling and energy
efficient?”) can be answered by calculating CSE and PUE using electricity consumption and
global and row-wise thermal data.
CSE is given by [96]:
CSE = -./012/ 3445672 89:;/< =4>/0
-./012/ 3445672 ?41) (4.1)
where average cooling system power is considered as the CRAC cooling power (kW), and
average cooling load (ton of chilled air) is the cooling required to maintain a chilled-air
temperature of 18 °C.
PUE is given by [96]:
PUE = @4;15 A1'656;9 =4>/0
B3@ CDE6(/0 (4.2)
where ICT equipment power is the power required by server, storage, and network equipment
and total facility power (i.e., the power required to run the server room) is comprised of both
ICT equipment power and auxiliary equipment power (e.g., lighting and HVAC). If CSE and PUE
are better than or equal to the current industrial benchmarks (i.e., 0.8 KW/ton for CSE and 1.6
for PUE) [105, 132], the system is considered to be cooling and energy efficient and a waste
heat analysis is performed (discussed later). If the system is considered to not be cooling and
energy efficient, then the next step is to analyze operating trends and thermal conditions.

84
4.3.5 Characterization of operating trends and thermal conditions
Mean CPU occupation was used to characterize the server room operating trends for the
daily, weekly, and monthly scenarios. For each scenario, peak and off-peak cases are
categorized according to: (a) mid-semester (weekday), (b) mid-semester (weekend), (c) offsemester (weekday), and (d) off-semester (weekend). Row-wise temperature data collected
from the hot and cold aisles for each rack group over a 30-day period was used to characterize
the server room thermal conditions.
4.3.6 Development of CFD model
A CFD model to simulate server room airflow and temperature was developed using ANSYS
Workbench (Canonsburg, PA); the room geometry was created, a computational mesh was
generated, and CFD equations were solved. Figure 4.4 shows the HPC room geometry created
using the SpaceClaim tool. The average temperature and velocity for the cold air coming from
the ceiling vents were initially set at 18.0 °C and 0.35 m/s. The average temperature and air
velocity for the cold air coming from the floor vents were initially set at 18.0 °C and 0.42 m/s.
Figure 4.4: Server room geometry with the hot rack faces and outlet vents colored red and the cold vents
colored blue.

85
The CFD surface mesh was generated using triangular elements and the volume mesh was
generated from the surface mesh with tetrahedral cells. Sizing limitations were applied to
different regions of the domain to increase accuracy in a reasonable computational time.
Limiting the mesh size to 25 cm in the whole domain and to 8 cm in the vicinity of the hot rack
faces, hot vents, and cold vents resulted in 2,572,433 cells. Figure 5.5 depicts the computational
mesh.
Figure 4.5: Computational mesh with finer cells in the critical regions of the hot rack faces, hot vents, and cold
vents.
The CFD equations were solved using the realizable k-ε model, a one-phase model in Fluent
with air (1.225 kg/m3 density) as its only material. This model was chosen because it allows
turbulence when describing air mass- and heat-transfer properties. The near-wall treatment in
the k-ε model was considered as a non-equilibrium wall function. The temporal and spatial
discretization schemes that were used is listed in Table 4.4. The k-ε model is chosen to solve the
CFD equations in ANSYS Fluent, and the first-order implicit is used as the temporal

86
discretization with a simple pressure-velocity coupling. The global temperature distribution was
then determined by the heat map generated in Fluent.
Table 4.4: Computational mesh with finer cells in the critical regions of the hot rack faces, hot vents, and cold
vents.
Pressure Momentum Turbulent Kinetic
Energy (k)
Turbulent Dissipation
Rate (ε) Energy
Second order Second order First order upwind First order upwind Order upwind
4.3.7 Estimation of server room waste heat quality and quantity
The annual collectable potential energy due to changes in the internal energy of air with
temperature difference of the CRAC units in the server room (∆U in MWh) was calculated using:
∆U = ∆W × t (4.3)
where t is time (h) and ∆W is the heat power (kW) that can be calculated from the potential
waste heat collected from the server room and is given by:
∆W = ṁ × C160,@% × ∆T (4.4)
where ṁ is the mass flowrate (kg/s) of the hot exhaust that enters each ceiling hot vent (see
below). C160,@% is the specific heat capacity (kJ·kg-1
·K-1
) for air at the temperature of the returned
hot exhaust, which is equal to 1.0061 kJ·kg-1
·K-1 (at 1 atm pressure). ∆T (in °C) is the
temperature difference of the CRAC units:
∆T = T> − T' (4.5)

87
where Tw is warm return air temperature from the ceiling hot vents and Tc is chilled air
temperature blowing into the server room from the CRAC units. For the server room, the chilled
air enters at a stable temperature of approximately 18 °C.
ṁ from equation 4.4 is given by:
ṁ = q̇ × ρ (4.6)
where � is the air density (kg/m3
) determined based on the average temperature of the
returned hot exhaust. q̇ (m3
·s-1
) is the air flow rate for the returned air entering the 15 hot
vents and is given by:
q̇ = v1.2 × A × 15 (4.7)
where v1.2 is average hot vent airflow velocity (m/s) and A is area of each hot vent (0.34 m2
).
This server room has five hot vents distributed evenly in the ceiling of the three hot aisles (15
hot vents in total); all hot vents have the same dimension.

88
4.4 Results and Discussion
4.4.1 Server room CSE and PUE
The data in Fig. 4.6 show CSE and PUE indices calculated for the server room over one
month. In Fig 4.6a, the daily CSE values range from 1.25 to 1.37 KW/ton; the 30-day average is
1.28 kW/ton. This average value is higher than the good practice benchmark of 0.8 kW/ton
[106, 107]. In Fig. 4.6b, the daily PUE value ranges from 1.7 to 5.2; the 30-day average is 2.5.
This average value is higher than the US average value of 1.6 [98].
The average CSE and PUE values in this study indicate cooling and energy inefficiencies that are
not unexpected in academic data centers. Cooling system inefficiencies may be due to chilled air
bypassing and short-circuiting in the open-aisle configuration (Figure 4.2a). Cooling system
inefficiencies may also result from slow responses of the cooling system to changes in
computing load, which are discussed in the next section.

89
Figure 4.6: (a) Cooling efficiency of the server room given as cooling system efficiency (CSE). The server room
has an average CSE of 1.28 kW/ton, which is higher (less efficient) than the good practice benchmark of 0.8
kW/ton. (b) Energy efficiency of the server room given as power usage effectiveness (PUE). The server room
has an average PUE value of 2.5, which is higher (less efficient) than the US average PUE of 1.6.
Server room average CSE 1.28 kW/ton
Good practice benchmark 0.8kW/ton
0 5 10 15 20 25 30
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
CSE (kW/ton)
Duration (days)
Server room average PUE value of 2.5
US average PUE value of 1.6 (in 2020)
b) Server room power usage effectiveness (PUE)
0 5 10 15 20 25 30
0
1
2
3
4
5
6
PUE
Durations (Days)
Server room average CSE 1.28 kW/ton
Good practice benchmark 0.8kW/ton
a) Server room cooling system efficiency (CSE)
0 5 10 15 20 25 30
0.6
0.7
0.8
0.9
1.0
1.1
1.2
1.3
1.4
CSE (kW/ton)
Duration (days)

90
4.4.2 Server room operating trends and temperature distribution
Mean CPU occupation data (Fig. 4.7) shows the variability of the workload on the servers.
Looking at the daily scenario (Fig. 4.7a), mean CPU occupation starts increasing after 8 am and
reaches the highest value in the afternoon (except weekends in off-semester). In all cases, CPU
occupation ranges from 19 to 37%. In the weekly scenario (Fig. 4.7b), mean CPU occupation
ranges from 17 to 39% with the average off-semester mean CPU occupation (23%) less than the
mid-semester mean CPU occupation (34%). The inset graph in Fig. 4.7b shows raw hourly data.
This data has high variability that is not observed in Fig. 4.7a where the hourly data has been
averaged. A typical monthly scenario is shown in Fig. 4.7c. Mean CPU occupation ranges from
17 to 53% (excluding the period of server-room maintenance) and as expected, off-semester
mean CPU occupation is generally less than mid-semester mean CPU occupation. Server room
CPU occupation trends can be concluded as (a) the average server room mean CPU occupation
is less than 50% for the time period considered; (b) CPU occupation is almost always greater in
mid-semester than off-semester and is generally less on weekends than weekdays; and (c) the
daily variability of mean CPU occupation can range from 20 to 50% in mid-semester. If this large
variability cannot be mitigated by more-advanced cooling systems with faster response time,
the result is decreased cooling system efficiency.

91

92

93
Figure 4.7: Computing load data collected from the server room for three scenarios. a) A daily scenario for
mean CPU occupation during weekdays and weekends in mid- and off-semester. b) A weekly scenario for
average mean CPU occupation during mid- and off-semester. The inset graphs show the variance in a typical
daily scenario. c) A typical monthly scenario for average mean CPU occupation during mid- and off-semester.
The data circled in red were collected during a maintenance period.
The data in Fig. 4.8 show temperature data for each rack group collected over one month.
The average hot-aisle temperatures have a wider range (24 to 34 °C) than the average cold-aisle
temperatures (16 to 20 °C). The hot-aisle variability for rack groups B and D is greater than it is
for rack groups A, C, and E. The servers in rack group B and D have mixed ownership, with some
of the servers belonging to individual research groups. Use of the servers depends on each
group’s research agenda, which results in surges between full operation and cool-down periods

94
depending on project due dates and publication schedules. Also, the hot aisles (especially for
rack groups B and D) have greater vertical temperature variability in that sometimes the high
position is warmer than the low position and sometimes the high position is cooler. The greater
stability and fewer fluctuations in cold aisle temperature are likely because the cold aisle has
vents in both the ceiling and floor whereas the hot aisle only has vents in the ceiling. Other
likely reasons for uneven hot-aisle temperature distribution are the nonuniform cooling loads
and unevenly distributed rack server density that is common in academic data centers.

95
Figure 4.8: Row-wise temperature data and average row-wise temperature for (a) rack group A, (b) rack
group B, (c) rack group C, (d) rack group D, and (e) rack group E.
0 5 10 15 20 25 30
50
45
40
35
30
25
20
15
10
Temperature (°C)
Duration (Days)
a) Temperature for rack group A
Hot aisle average temperature at 25 °C
Cold aisle average temperature at 20 °C
0 5 10 15 20 25 30
50
45
40
35
30
25
20
15
10
) C°( Temperature
Duration (Days)
b) Temperature for rack group B
Hot aisle average temperature at 34 °C
Cold aisle average temperature at 18 °C
0 5 10 15 20 25 30
50
45
40
35
30
25
20
15
10
) C°( Temperature
Duration (Days)
c) Temperature for rack group C
Hot aisle average temperature at 24 °C
Cold aisle average temperature at 16 °C
Hot aisle average temperature at 34 °C
Cold aisle average temperature at 16 °C
0 5 10 15 20 25 30
50
45
40
35
30
25
20
15
10
) C°( Temperature
Duration (Days)
d) Temperature for rack group D
Hot aisle average temperature at 26 °C
Cold aisle average temperature at 19 °C
0 5 10 15 20 25 30
50
45
40
35
30
25
20
15
10
) C°( Temperature
Duration (Days)
e) Temperature for rack group E

96
4.4.3 Server room waste heat estimation and reuse analysis from CFD model results
4.4.3.1 Current server room configuration with open aisles
CFD simulation results are shown as a temperature contour plot in Figure 4.9. Figures 4.9ag show temperature distribution at heights of 0.5 to 3.5 m in 0.5 m increments; Figure 4.9h
provides a reminder of the server room aisle and rack group configuration. In Figures 4.9e-g,
only outlines of the five rack groups are shown because the rack height is shorter than 2.5 m.
Looking at Figures 4.9a-d, hot aisles B/C and D/E are hotter than hot aisle A. Looking at Figures
4.9a-g, hot exhaust temperatures decrease with distance from the floor (i.e., temperatures
decrease as the hot exhaust rises). Clearly, there are some regions in the server room where
the temperature is significantly higher than in adjacent regions; these hot spots are indicated
by red coloring. Hot spots can be mitigated by three strategies: replacing or upgrading the ICT
equipment, shifting the computing load, or improving the room cooling configuration. In this
study, a hot-air containment system (HAC) system is simulated to mitigate hot spots (discussed
in section 4.4.3.2).
Figures. 4.9a-d also show that the hot air from the hot aisles diffuses into open space as it
rises. And when it enters the space above rack groups (Figures. 4.9e-g), chilled air has chilled it
prior to it entering the return air channel. Some chilled air by-passes the server inlets where it
should be cooling the servers; this will be discussed more with Figure. 4.10.

97
Figure 4.9: Temperature contour plots showing the horizontal temperature distribution for several rack
heights: (a) Z = 0.5m,(b) Z = 1.0m, (c) Z = 1.5m, (d) Z = 2.0m, (e) Z = 2.5m,(f) Z = 3.0m, (g) Z = 3.5m, and (h)
configuration of server room rack groups and aisles; hot spots were identified horizontally in the middle and
at the ends of rack groups.

98
The airflow vectors in Fig. 4.10a show that the hot exhaust emitted by the servers into
aisles B/C and D/E is trapped by chilled air coming from the cold aisles. This chilled air is
bypassing the server inlets where it should be cooling the servers. This short-circuiting leads to
less cooling of the servers and hotter exhaust entering the hot aisles. Fig. 4.10b, shows the
resulting vertical temperature distribution in a sectional slice of the server room, which clearly
indicates the vertical distribution of the hot spots. And the average hot/cold aisles temperature
is shown in Fig. 4.11.
Figure 4.10: (a) Air flow distribution given by vectors. Chilled air can be seen flowing from the ceiling cold
vents directly to the hot vents due to bypassing and short-circuiting. (b) Vertical temperature distribution in
the server room. Hot spots are shown in red on the faces of the rack groups in the hot aisles.
a) Server room vertical air flow vector
distribution
b) Server room vertical temperature
distribution
A A/B B/C C/D D/E E
A A/B B/C C/D D/E E

99
Figure 4.11: Vertical average temperature distribution for the server room with open-aisle configuration (a)
hot aisle and (b) cold aisle.
In this study, cold-aisle temperatures increase from 18.0 (the initial chilled air temperature
entering from the vents) to 18.5 °C in aisle C/D. Although the bypass effect causes the air
entering the CRAC units from the hot aisles to be cooler than it would otherwise be, the CRAC
units operate based on the temperature of the cold aisles. Thus, more energy is required to
maintain the target cooling temperature (18.0 °C) and satisfy server cooling needs. The
increased energy due to the bypass effect contributes to the higher PUE and CSE values for the
data center. Furthermore, data center managers are often more concerned with cold-aisle
temperatures, which are often seen as an indicator of overall cooling efficiency [133].
According to the CFD simulation results (shown in Table 4.5), the current server room
configuration with open aisles has an average return air temperature of 23 °C with the dry air
density of 1.19 kg/m3 at 1 atm. If reuse of the waste heat is being considered, 23 °C is low
0 10 20 30 40 50 60 70
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Height (m)
Aisle average temperature (°C)
a) Hot aisle
0 10 20 30 40 50 60 70
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Height (m)
Aisle average temperature (°C)
b) Cold aisle

100
quality compared to the district heating target of 40 °C reported in the literature [106, 120,
121]. The theoretical ∆T that can be obtained is only approximately 5 °C. Using equations 3-7
and a warm air velocity of 1.41 m/s, a heat power (∆W) of 43.1 kW of waste heat is emitted via
a cooling tower on the data center roof. For one year’s continuous operation, this calculates as
377 MWh of emitted waste heat.
Table 4.5: Average temperature for hot and cold aisles in the server room with open-aisle configuration
Height (m) Aisle Temperature (°C)
Aisle A Aisle A/B Aisle B/C Aisle C/D Aisle D/E Aisle E
0.5 25.1 18.3 51.4 18.3 64.5 18.0
1.0 25.8 18.6 49.0 18.5 56.4 18.0
1.5 26.7 18.8 41.8 18.6 47.2 18.0
2.0 27.9 18.8 29.1 18.6 34.1 18.0
2.5 27.5 18.5 22.4 18.4 23.4 18.0
3.0 24.4 18.2 19.9 18.1 20.1 18.0
3.5 21.5 18.1 19.0 18.0 19.0 18.0

101
4.4.3.2 Example of improved server room configuration with HAC system
CFD simulation results for the improved server room with an HAC system are shown in Fig.
4.12. As can be seen in Fig. 4.12a, the HAC system confines the hot exhaust and prohibits it
from mixing with the chilled air after leaving the hot aisles. The hot exhaust in aisles B/C and
D/E is contained and the aisle temperatures are much higher than they are in Fig. 4.10a. Fig.
4.12b shows that the HAC system prevents chilled air from bypassing the server inlets; instead,
the chilled air is available to cool the servers.

102
Figure 4.12: (a) Air flow distribution given by vectors. Chilled air can be seen flowing from the ceiling and floor
cold vents to the servers without bypassing occurring. (b) Vertical temperature distribution in the server room.
Temperature in the hot aisles is higher than open-aisle configuration.
According to the CFD simulation results (shown in Table 4.6) and the average hot/cold aisle
temperatures shown in Fig. 4.13, the improved server room configuration with an HAC system
has an average return air temperature of 46 °C (with the dry air density of 1.11 kg/m3 at 1 atm),
which is twice that of the open-aisle configuration (23 °C). The improved return air temperature
exceeds the district heating target of 40 °C reported in the literature [120, 121, 134]. The
theoretical ∆T that can be obtained is approximately 28 °C with a warm air velocity of 1.44 m/s.
A heat power (∆W) of 288.8 kW of waste heat is emitted via the cooling tower on the data
b) Server room vertical temperature distribution with HAC
system
A A/B B/C C/D D/E E
a) Server room vertical air flow vector distribution with HAC
system
A A/B B/C C/D D/E E

103
center roof. The quality of this waste heat is much higher than in the open-aisle configuration
and represents waste heat that may be worthwhile to reuse for auxiliary equipment, similar to
the leading commercial data centers (e.g., Amazon, Google, and Microsoft). With an average
return air temperature of 46 °C, for one year’s continuous operation, this calculates as 2004
MWh of waste heat. Compared to the open-aisle configuration, the configuration with the HAC
system has returned air that is two times warmer and can produce five times more energy.
Also, it can be seen that this annual waste heat quantity (2004 MWh) is just below the range of
those from the literature (3156 –11045 MWh in [101, 127-130]).
Table 4.6: Average temperature for hot and cold aisles in the server room with HAC system
Height (m) Aisle Temperature (°C)
Aisle A Aisle A/B Aisle B/C Aisle C/D Aisle D/E Aisle E
0.5 31.1 18.0 53.1 18.0 70.9 18.0
1.0 28.1 18.0 50.8 18.0 59.7 18.0
1.5 27.3 18.0 53.0 18.0 55.7 18.0
2.0 27.2 18.0 55.7 18.0 53.4 18.0
2.5 26.2 18.0 56.4 18.0 52.6 18.0
3.0 25.2 18.0 53.7 18.0 49.0 18.0
3.5 24.7 18.0 47.5 18.0 44.9 18.0

104
Figure 4.13: Vertical average temperature distribution for the server room with a basic hot air containment
system: (a) hot aisle and (b) cold aisle.
0 10 20 30 40 50 60 70
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Height (m)
Aisle average temperature (°C)
a) Hot aisle
0 10 20 30 40 50 60 70
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
Height (m)
Aisle average temperature (°C)
b) Cold aisle

105
4.5 Conclusions and Implications
In this study, real-time onsite waste heat data was collected, and an analysis framework
was developed to characterize the quality and quantity of waste heat from a typical academic
data center. CFD simulations found that waste heat quality for the server room was low (23 °C)
and less than the district heating target of 40 °C reported in the literature. However, by
implementing a basic hot air management approach (an HAC system), cooling efficiency
increased, and the temperature doubled (to 46 °C). Five times more energy is produced in the
HAC configuration than in the open-aisle configuration (2004 vs 377 MWh annually). Our study
emphasizes that containing waste heat is the first step in making it available for reuse, and that
containment can significantly increase its quality. It should be noted that liquid-cooling systems
inherently contain waste heat and, given the renewed interest in these systems [135],
opportunities for waste heat reuse are only expected to grow. Further, opportunities to
upgrade the waste heat to a higher quality using heat concentration technology (e.g., a heat
pump system) will be critical in analyzing the mechanics of reusing waste heat. And in general,
the robust growth in the data center cooling market (from $12.7 billion today to 29.6 billion by
2028), bodes well for further improvements to system efficiency, especially since the research
and academic sector is projected to have the highest growth rate [136].
In addition to the value of recovering waste heat as a resource, there may be value in not
releasing the waste heat to the ambient environment. This is particularly true if the data center
is located in an urban setting where surface heating contributes to the urban heat island effect
or disturbance of ambient environment is a concern in cold areas [9]. For urban environments,

106
use of a waste heat data set in a broader analysis of the environmental impact of
anthropogenic heat emitted by industrial and residential buildings would be beneficial. And
specifically for academic data centers, as universities look to be more sustainable and operate
more efficiently [100], opportunities to operate with minimal unintended consequences and a
lifecycle mindset may be valued.

107
Chapter 5 Conclusions
5.1 Research Synopsis
This dissertation investigates real-time characterization of water treatment membrane
compaction from low-pressure (e.g., MD and MF) to high-pressure (e.g., NF and RO) and
explores the integration of a novel low-grade waste heat in MD systems.
5.2 Summary of Real-time Water Treatment Membrane Characterization
A novel method combining EIS with DMA was developed to characterize membrane
compaction in real-time under varying temperatures and pressures. This novel method provides
the capability of detecting even subtle deformation caused by membrane creep under longterm operation. Our findings demonstrate that membrane compaction is not solely attributable
to support/backing layer compaction. The air pockets in both NF and RO membranes, often
overlooked in previous studies, play a crucial role in membrane compaction and should not be
ignored. We also revealed that the void space in the PE backing layer contributes most to NF
membrane compaction, while the solid material in the PE backing layer contributes most to RO
membrane compaction.
We also found that stepwise pressure increases that induced membrane compaction due
to both instant compaction and creep, enabled differentiation of the contribution of creep to
total compaction. The initial instant compaction contributed more to total compaction than the
subsequent instant compaction. Instantaneous compaction is likely influenced by the stiffness
of the material fibrils, while longer-term creep is influenced by the increased density of the

108
pack/collapsed membrane. We propose that the development of membrane fatigue during
typical operation (i.e., induced solely by TMP) can be indicated by several metrics. These
include the increased average irrevocable compaction values in the creep-and-recovery
experiments, the increased creep/instantaneous compaction ratio (φ), and the increased
average maximum strain and average irrevocable strain in the hysteresis experiment.
Furthermore, susceptibility to membrane fatigue increases with membrane age.
5.3 Summary of Waste Heat Characterization and Integration in MD Systems
An analysis framework was developed to characterize the quality and quantity of a novel
waste heat from a typical academic data center, and real-time onsite waste heat data was
collected. CFD simulations found that waste heat quality for the server room was low (23 °C)
and less than the district heating target of 40 °C reported in the literature. However, by
implementing a basic hot air management approach (an HAC system), cooling efficiency
increased and the temperature doubled (to 46 °C). Five times more energy is produced in the
HAC configuration than in the open-aisle configuration (2004 vs 377 MWh annually). This waste
heat quality and quantity advancement emphasizes that containing waste heat is the first step
in making it available for reuse, and opportunities for waste heat reuse are only expected to
grow. Further, opportunities to upgrade the waste heat to a higher quality using heat
concentration technology (e.g., a heat pump system) will be critical in analyzing the mechanics
of reusing waste heat, and it can be a potential “free” heating source for MD systems.
In addition to the value of recovering waste heat as a resource, there may be value in not
releasing the waste heat to the ambient environment. This is particularly true if the data center

109
is located in an urban setting where surface heating contributes to the urban heat island effect
or disturbance of ambient environment is a concern in cold areas. For urban environments, use
of a waste heat data set in a broader analysis of the environmental impact of anthropogenic
heat emitted by industrial and residential buildings would be beneficial.

110
References
[1] A. Alkhudhiri, N. Darwish, and N. Hilal, "Membrane distillation: A comprehensive review,"
Desalination, vol. 287, pp. 2-18, 2012.
[2] Y. Choi, G. Naidu, L. D. Nghiem, S. Lee, and S. Vigneswaran, "Membrane distillation
crystallization for brine mining and zero liquid discharge: opportunities, challenges, and recent
progress," Environmental Science: Water Research & Technology, vol. 5, no. 7, pp. 1202-1221,
2019.
[3] A. Deshmukh et al., "Membrane distillation at the water-energy nexus: limits, opportunities, and
challenges," Energy & Environmental Science, vol. 11, no. 5, pp. 1177-1196, 2018.
[4] M. Elimelech and W. A. Phillip, "The future of seawater desalination: energy, technology, and
the environment," science, vol. 333, no. 6043, pp. 712-717, 2011.
[5] R. D. Gustafson, S. R. Hiibel, and A. E. Childress, "Membrane distillation driven by intermittent
and variable-temperature waste heat: System arrangements for water production and heat
storage," Desalination, vol. 448, pp. 49-59, 2018.
[6] R. D. Gustafson, A. L. McGaughey, W. Ding, S. C. McVety, and A. E. Childress, "Morphological
changes and creep recovery behavior of expanded polytetrafluoroethylene (ePTFE) membranes
used for membrane distillation," Journal of Membrane Science, vol. 584, pp. 236-245, 2019.
[7] A. Alkhudhiri and N. Hilal, "Membrane distillation—Principles, applications, configurations,
design, and implementation," in Emerging Technologies for Sustainable Desalination Handbook:
Elsevier, 2018, pp. 55-106.
[8] X. Wei, K. T. Sanders, and A. E. Childress, "Reclaiming wastewater with increasing salinity for
potable water reuse: water recovery and energy consumption during reverse osmosis
desalination," Desalination, vol. 520, p. 115316, 2021.
[9] S. L. Plata et al., "Zero liquid discharge and water reuse in recirculating cooling towers at power
facilities: review and case study analysis," ACS ES&T Engineering, vol. 2, no. 3, pp. 508-525,
2022.
[10] C. P. Morrow et al., "Integrating an aerobic/anoxic osmotic membrane bioreactor with
membrane distillation for potable reuse," Desalination, vol. 432, pp. 46-54, 2018.
[11] S. Hube et al., "Direct membrane filtration for wastewater treatment and resource recovery: A
review," Science of the total environment, vol. 710, p. 136375, 2020.
[12] W. Ding, K. Ma, and A. E. Childress, "A Novel Method to Quantify Real-Time Compaction of
Water Treatment Membranes," Environmental Science & Technology Letters, 2023/09/29 2023,
doi: 10.1021/acs.estlett.3c00656.

111
[13] L. B. Freund and S. Suresh, Thin film materials: stress, defect formation and surface evolution.
Cambridge university press, 2004.
[14] A. L. McGaughey and A. E. Childress, "Wetting indicators, modes, and trade-offs in membrane
distillation," Journal of Membrane Science, vol. 642, p. 119947, 2022.
[15] B. Van der Bruggen, C. Vandecasteele, T. Van Gestel, W. Doyen, and R. Leysen, "A review of
pressure-driven membrane processes in wastewater treatment and drinking water production,"
Environmental progress, vol. 22, no. 1, pp. 46-56, 2003.
[16] N. Joseph, P. Ahmadiannamini, R. Hoogenboom, and I. F. Vankelecom, "Layer-by-layer
preparation of polyelectrolyte multilayer membranes for separation," Polymer Chemistry, vol. 5,
no. 6, pp. 1817-1831, 2014.
[17] S. S. Shenvi, A. M. Isloor, and A. Ismail, "A review on RO membrane technology: Developments
and challenges," Desalination, vol. 368, pp. 10-26, 2015.
[18] J. Zhang and S. Gray, "Effect of applied pressure on performance of PTFE membrane in DCMD,"
Journal of Membrane Science, vol. 369, no. 1-2, pp. 514-525, 2011.
[19] J. Zhang and S. Gray, "Modelling heat and mass transfers in DCMD using compressible
membranes," Journal of Membrane Science, vol. 387, pp. 7-16, 2012.
[20] K. W. Lawson, M. S. Hall, and D. R. Lloyd, "Compaction of microporous membranes used in
membrane distillation. I. Effect on gas permeability," Journal of membrane science, vol. 101, no.
1-2, pp. 99-108, 1995.
[21] A. L. McGaughey, P. Karandikar, M. Gupta, and A. E. Childress, "Hydrophobicity versus pore size:
polymer coatings to improve membrane wetting resistance for membrane distillation," ACS
Applied Polymer Materials, vol. 2, no. 3, pp. 1256-1267, 2020.
[22] A. McGaughey, R. Gustafson, and A. Childress, "Effect of long-term operation on membrane
surface characteristics and performance in membrane distillation," Journal of Membrane
Science, vol. 543, pp. 143-150, 2017.
[23] G. Rao, S. R. Hiibel, A. Achilli, and A. E. Childress, "Factors contributing to flux improvement in
vacuum-enhanced direct contact membrane distillation," Desalination, vol. 367, pp. 197-205,
2015.
[24] R. D. Gustafson, J. R. Murphy, and A. Achilli, "A stepwise model of direct contact membrane
distillation for application to large-scale systems: Experimental results and model predictions,"
Desalination, vol. 378, pp. 14-27, 2016.
[25] A. Najib, J. Orfi, E. Ali, and J. Saleh, "Thermodynamics analysis of a direct contact membrane
distillation with/without heat recovery based on experimental data," Desalination, vol. 466, pp.
52-67, 2019.

112
[26] P. Termpiyakul, R. Jiraratananon, and S. Srisurichan, "Heat and mass transfer characteristics of a
direct contact membrane distillation process for desalination," Desalination, vol. 177, no. 1-3,
pp. 133-141, 2005.
[27] M. El-Bourawi, Z. Ding, R. Ma, and M. Khayet, "A framework for better understanding
membrane distillation separation process," Journal of membrane science, vol. 285, no. 1-2, pp.
4-29, 2006.
[28] S. Lin, N. Y. Yip, and M. Elimelech, "Direct contact membrane distillation with heat recovery:
Thermodynamic insights from module scale modeling," Journal of membrane science, vol. 453,
pp. 498-515, 2014.
[29] K. S. Christie, T. Horseman, and S. Lin, "Energy efficiency of membrane distillation: Simplified
analysis, heat recovery, and the use of waste-heat," Environment international, vol. 138, p.
105588, 2020.
[30] E. Curcio and E. Drioli, "Membrane distillation and related operations—a review," Separation
and Purification Reviews, vol. 34, no. 1, pp. 35-86, 2005.
[31] F. Banat, N. Jwaied, M. Rommel, J. Koschikowski, and M. Wieghaus, "Desalination by a “compact
SMADES” autonomous solarpowered membrane distillation unit," Desalination, vol. 217, no. 1-
3, pp. 29-37, 2007.
[32] A. Cipollina, M. Di Sparti, A. Tamburini, and G. Micale, "Development of a membrane distillation
module for solar energy seawater desalination," Chemical engineering research and design, vol.
90, no. 12, pp. 2101-2121, 2012.
[33] J. B. Gálvez, L. García-Rodríguez, and I. Martín-Mateos, "Seawater desalination by an innovative
solar-powered membrane distillation system: the MEDESOL project," Desalination, vol. 246, no.
1-3, pp. 567-576, 2009.
[34] J. Koschikowski, M. Wieghaus, and M. Rommel, "Solar thermal-driven desalination plants based
on membrane distillation," Desalination, vol. 156, no. 1-3, pp. 295-304, 2003.
[35] R. Schwantes et al., "Membrane distillation: Solar and waste heat driven demonstration plants
for desalination," Desalination, vol. 323, pp. 93-106, 2013.
[36] A. Jansen, J. Assink, J. Hanemaaijer, J. Van Medevoort, and E. Van Sonsbeek, "Development and
pilot testing of full-scale membrane distillation modules for deployment of waste heat,"
Desalination, vol. 323, pp. 55-65, 2013.
[37] N. Dow et al., "Pilot trial of membrane distillation driven by low grade waste heat: Membrane
fouling and energy assessment," Desalination, vol. 391, pp. 30-42, 2016.
[38] A. Hagedorn, G. Fieg, D. Winter, J. Koschikowski, and T. Mann, "Methodical design and
operation of membrane distillation plants for desalination," Chemical Engineering Research and
Design, vol. 125, pp. 265-281, 2017.

113
[39] H. Lee, F. He, L. Song, J. Gilron, and K. K. Sirkar, "Desalination with a cascade of cross-flow hollow
fiber membrane distillation devices integrated with a heat exchanger," AIChE Journal, vol. 57,
no. 7, pp. 1780-1795, 2011.
[40] J. Gilron, L. Song, and K. K. Sirkar, "Design for cascade of crossflow direct contact membrane
distillation," Industrial & Engineering Chemistry Research, vol. 46, no. 8, pp. 2324-2334, 2007.
[41] Y.-D. Kim, K. Thu, and S.-H. Choi, "Solar-assisted multi-stage vacuum membrane distillation
system with heat recovery unit," Desalination, vol. 367, pp. 161-171, 2015.
[42] J. Swaminathan, H. W. Chung, and D. M. Warsinger, "Energy efficiency of membrane distillation
up to high salinity: Evaluating critical system size and optimal membrane thickness," Applied
Energy, vol. 211, pp. 715-734, 2018.
[43] L. Cheng, Y. Zhao, P. Li, W. Li, and F. Wang, "Comparative study of air gap and permeate gap
membrane distillation using internal heat recovery hollow fiber membrane module,"
Desalination, vol. 426, pp. 42-49, 2018.
[44] E. Ali, J. Orfi, and A. Najib, "Assessing the thermal efficiency of brackish water desalination by
membrane distillation using exergy analysis," Arabian Journal for Science and Engineering, vol.
43, no. 5, pp. 2413-2424, 2018.
[45] J. Andrés-Mañas, A. Ruiz-Aguirre, F. Acién, and G. Zaragoza, "Assessment of a pilot system for
seawater desalination based on vacuum multi-effect membrane distillation with enhanced heat
recovery," Desalination, vol. 443, pp. 110-121, 2018.
[46] D. Y. F. Ng, Y. Chen, Z. Dong, and R. Wang, "Membrane compaction in forward osmosis process,"
Desalination, vol. 468, p. 114067, 2019.
[47] K. M. Persson, V. Gekas, and G. Trägårdh, "Study of membrane compaction and its influence on
ultrafiltration water permeability," Journal of membrane science, vol. 100, no. 2, pp. 155-162,
1995.
[48] D. M. Davenport et al., "Thin film composite membrane compaction in high-pressure reverse
osmosis," Journal of membrane science, vol. 610, p. 118268, 2020.
[49] A. E. Childress, P. Le-Clech, J. L. Daugherty, C. Chen, and G. L. Leslie, "Mechanical analysis of
hollow fiber membrane integrity in water reuse applications," Desalination, vol. 180, no. 1-3, pp.
5-14, 2005.
[50] M. T. M. Pendergast, J. M. Nygaard, A. K. Ghosh, and E. M. Hoek, "Using nanocomposite
materials technology to understand and control reverse osmosis membrane compaction,"
Desalination, vol. 261, no. 3, pp. 255-263, 2010.
[51] J. Lee and Y. J. Lim, "On the control strategy to improve the salt rejection of a thin-film
composite reverse osmosis membrane," Applied Sciences, vol. 11, no. 16, p. 7619, 2021.

114
[52] K. Ebert, D. Fritsch, J. Koll, and C. Tjahjawiguna, "Influence of inorganic fillers on the compaction
behaviour of porous polymer based membranes," Journal of Membrane Science, vol. 233, no. 1-
2, pp. 71-78, 2004.
[53] S. Stade, M. Kallioinen, A. Mikkola, T. Tuuva, and M. Mänttäri, "Reversible and irreversible
compaction of ultrafiltration membranes," Separation and Purification Technology, vol. 118, pp.
127-134, 2013.
[54] A. Volkov, "Membrane compaction," Encyclopedia of Membranes, vol. 18, pp. 35-60, 2014.
[55] A. Grekhov, A. Belogorlov, A. Yushkin, and A. Volkov, "New express dynamic technique for liquid
permeation measurements in a wide range of trans-membrane pressures," Journal of
membrane science, vol. 390, pp. 160-163, 2012.
[56] S. Lee, E. Lee, M. Elimelech, and S. Hong, "Membrane characterization by dynamic hysteresis:
Measurements, mechanisms, and implications for membrane fouling," Journal of Membrane
Science, vol. 366, no. 1-2, pp. 17-24, 2011.
[57] Z.-X. Low and J. Shen, "Determining stability of organic solvent nanofiltration membranes by
cross-flow aging," Separation and Purification Technology, vol. 256, p. 117840, 2021.
[58] T. Hoang, G. Stevens, and S. Kentish, "The effect of feed pH on the performance of a reverse
osmosis membrane," Desalination, vol. 261, no. 1-2, pp. 99-103, 2010.
[59] Y. Zhao, G. S. Lai, J. Y. Chong, and R. Wang, "Dissecting the structure-compaction-performance
relationship of thin-film composite polyamide membranes with different structure features,"
Journal of Membrane Science, vol. 654, p. 120553, 2022.
[60] X. Li, S. Zhang, F. Fu, and T.-S. Chung, "Deformation and reinforcement of thin-film composite
(TFC) polyamide-imide (PAI) membranes for osmotic power generation," Journal of membrane
science, vol. 434, pp. 204-217, 2013.
[61] J. Wu et al., "Reverse osmosis membrane compaction and embossing at ultra-high pressure
operation," Desalination, vol. 537, p. 115875, 2022.
[62] M. Aghajani et al., "Relationship between permeation and deformation for porous membranes,"
Journal of Membrane Science, vol. 526, pp. 293-300, 2017.
[63] M. Aghajani, M. Wang, L. M. Cox, J. P. Killgore, A. R. Greenberg, and Y. Ding, "Influence of
support-layer deformation on the intrinsic resistance of thin film composite membranes,"
Journal of membrane science, vol. 567, pp. 49-57, 2018.
[64] "Dynamic Mechanical Analysis Basic Theory & Applications Training Day 1," 2020: TA
Instruments.
[65] V. E. Reinsch, A. R. Greenberg, S. S. Kelley, R. Peterson, and L. J. Bond, "A new technique for the
simultaneous, real-time measurement of membrane compaction and performance during

115
exposure to high-pressure gas," Journal of Membrane Science, vol. 171, no. 2, pp. 217-228,
2000.
[66] J. Kavanagh, S. Hussain, T. Chilcott, and H. Coster, "Fouling of reverse osmosis membranes using
electrical impedance spectroscopy: measurements and simulations," Desalination, vol. 236, no.
1-3, pp. 187-193, 2009.
[67] L. D. Tijing, Y. C. Woo, J.-S. Choi, S. Lee, S.-H. Kim, and H. K. Shon, "Fouling and its control in
membrane distillation—A review," Journal of Membrane Science, vol. 475, pp. 215-244, 2015.
[68] A. Antony, T. Chilcott, H. Coster, and G. Leslie, "In situ structural and functional characterization
of reverse osmosis membranes using electrical impedance spectroscopy," Journal of Membrane
Science, vol. 425, pp. 89-97, 2013.
[69] T. C. Chilcott, J. Cen, and J. M. Kavanagh, "In situ characterization of compaction, ionic barrier
and hydrodynamics of polyamide reverse osmosis membranes using electrical impedance
spectroscopy," Journal of Membrane Science, vol. 477, pp. 25-40, 2015.
[70] B. Hinderliter, S. Croll, D. Tallman, Q. Su, and G. Bierwagen, "Interpretation of EIS data from
accelerated exposure of coated metals based on modeling of coating physical properties,"
Electrochimica acta, vol. 51, no. 21, pp. 4505-4515, 2006.
[71] M. Simić, A. K. Stavrakis, and G. M. Stojanović, "A low-complexity method for parameter
estimation of the simplified Randles circuit with experimental verification," IEEE Sensors Journal,
vol. 21, no. 21, pp. 24209-24217, 2021.
[72] J. C. Tuberquia, N. Nizamidin, and G. K. Jennings, "Effect of superhydrophobicity on the barrier
properties of polymethylene films," Langmuir, vol. 26, no. 17, pp. 14039-14046, 2010.
[73] Y. Chen, Z. Wang, G. K. Jennings, and S. Lin, "Probing pore wetting in membrane distillation using
impedance: early detection and mechanism of surfactant-induced wetting," Environmental
Science & Technology Letters, vol. 4, no. 11, pp. 505-510, 2017.
[74] P. D. Sutrisna, J. Hou, H. Li, Y. Zhang, and V. Chen, "Improved operational stability of Pebaxbased gas separation membranes with ZIF-8: A comparative study of flat sheet and composite
hollow fibre membranes," Journal of membrane science, vol. 524, pp. 266-279, 2017.
[75] R. Young and S. Eichhorn, "Deformation mechanisms in polymer fibres and nanocomposites,"
Polymer, vol. 48, no. 1, pp. 2-18, 2007.
[76] C. R. Martin, "Template synthesis of electronically conductive polymer nanostructures,"
Accounts of chemical research, vol. 28, no. 2, pp. 61-68, 1995.
[77] A. E139-11, "Standard test methods for conducting creep, creep-rupture, and stress-rupture
tests of metallic materials," ed: ASTM International West Conshohocken, 2018.

116
[78] R. R. Ferlita, D. Phipps, J. Safarik, and D. H. Yeh, "Cryo-snap: A simple modified freeze-fracture
method for SEM imaging of membrane cross-sections," Environmental Progress, vol. 27, no. 2,
pp. 204-209, 2008.
[79] M. Brownell and A. K. Nair, "Deformation mechanisms of polytetrafluoroethylene at the nanoand microscales," Physical Chemistry Chemical Physics, vol. 21, no. 1, pp. 490-503, 2019.
[80] C. Luo, J. Pei, W. Zhuo, Y. Niu, and G. Li, "Phase transition behavior and deformation mechanism
of polytetrafluoroethylene under stretching," RSC advances, vol. 11, no. 63, pp. 39813-39820,
2021.
[81] R. J. Ohlund, B. H. Dahdah, G. R. Guillen, and A. E. Childress, "Augmenting ocean water
desalination with potable reuse: Concept feasibility in terms of cost and environmental
impacts," Desalination, vol. 569, p. 116941, 2024.
[82] R. J. Young and P. A. Lovell, Introduction to polymers. CRC press, 2011.
[83] J. Wu et al., "Polyamide Reverse Osmosis Membrane Compaction and Relaxation: Mechanisms
and Implications for Desalination Performance," Journal of Membrane Science, p. 122893, 2024.
[84] A. W. Hendry, "Compression Failure Theories," in Structural Masonry. London: Macmillan
Education UK, 1998, pp. 56-70.
[85] Z. Liu, H. Wang, Y. Li, X. Wang, and A. P. S. Selvadurai, "Triaxial Compressive Strength, Failure,
and Rockburst Potential of Granite Under High-Stress and Ground-Temperature Coupled
Conditions," Rock Mechanics and Rock Engineering, vol. 56, no. 2, pp. 911-932, 2023/02/01
2023, doi: 10.1007/s00603-022-03066-5.
[86] ASTM, "Standard Practice for Fluorescent Ultraviolet (UV) Lamp Apparatus Exposure of Plastics,"
2021, doi: 10.1520/. D4329-21. 2.
[87] H. Gao, L. Yuan, G. Liu, and J. Gao, "An ultraviolet light-responsive polyacrylate film with high
fatigue resistance," Journal of Polymer Research, vol. 29, no. 2, p. 46, 2022.
[88] Standard Practice for Exposing Nonmetallic Materials in Accelerated Test Devices that Use
Laboratory Light Sources, ASTM, 2019.
[89] M. Frigione and A. Rodríguez-Prieto, "Can accelerated aging procedures predict the long term
behavior of polymers exposed to different environments?," Polymers, vol. 13, no. 16, p. 2688,
2021.
[90] Z. Zhang, P. Bai, Y. Xiao, Y. Guo, and Y. Wang, "Conformation-Induced stiffening effect of
crosslinked polymer thin films," Communications Physics, vol. 6, no. 1, p. 332, 2023.
[91] M. Dayarathna, Y. Wen, and R. Fan, "Data center energy consumption modeling: A survey," IEEE
Communications Surveys & Tutorials, vol. 18, no. 1, pp. 732-794, 2015.

117
[92] E. Masanet, A. Shehabi, N. Lei, S. Smith, and J. Koomey, "Recalibrating global data center energyuse estimates," Science, vol. 367, no. 6481, pp. 984-986, 2020.
[93] OpenAI, "GPT-4 Technical Report," p. arXiv:2303.08774doi: 10.48550/arXiv.2303.08774.
[94] M. Poess and R. O. Nambiar, "Energy cost, the key challenge of today's data centers: a power
consumption analysis of TPC-C results," Proceedings of the VLDB Endowment, vol. 1, no. 2, pp.
1229-1240, 2008.
[95] G. Kamiya. "Data Centres and Data Transmission Networks." https://www.iea.org/reports/datacentres-and-data-transmission-networks
[96] A. Capozzoli, M. Chinnici, M. Perino, and G. Serale, "Review on performance metrics for energy
efficiency in data center: The role of thermal management," in International Workshop on
Energy Efficient Data Centers, 2014: Springer, pp. 135-151.
[97] S. Long, Y. Li, J. Huang, Z. Li, and Y. Li, "A review of energy efficiency evaluation technologies in
cloud data centers," Energy and Buildings, p. 111848, 2022.
[98] Y.-T. Lee, C.-Y. Wen, Y.-C. Shih, Z. Li, and A.-S. Yang, "Numerical and experimental investigations
on thermal management for data center with cold aisle containment configuration," Applied
Energy, vol. 307, p. 118213, 2022.
[99] H. Baniata, S. Mahmood, and A. Kertesz, "Assessing anthropogenic heat flux of public cloud data
centers: current and future trends," PeerJ Computer Science, vol. 7, p. e478, 2021.
[100] K. Choo, R. M. Galante, and M. M. Ohadi, "Energy consumption analysis of a medium-size
primary data center in an academic campus," Energy and Buildings, vol. 76, pp. 414-421, 2014.
[101] V. Dvorak, V. Zavrel, J. Torrens Galdiz, and J. L. Hensen, "Simulation-based assessment of data
center waste heat utilization using aquifer thermal energy storage of a university campus," in
Building Simulation, 2020, vol. 13, no. 4: Springer, pp. 823-836.
[102] E. Ali, "Energy efficient configuration of membrane distillation units for brackish water
desalination using exergy analysis," Chemical Engineering Research and Design, vol. 125, pp.
245-256, 2017.
[103] J. Cho, T. Lim, and B. S. Kim, "Viability of datacenter cooling systems for energy efficiency in
temperate or subtropical regions: Case study," Energy and buildings, vol. 55, pp. 189-197, 2012.
[104] L. Karimi et al., "Water-energy tradeoffs in data centers: A case study in hot-arid climates,"
Resources, Conservation and Recycling, vol. 181, p. 106194, 2022.
[105] S. Greenberg, E. Mills, B. Tschudi, P. Rumsey, and B. Myatt, "Best practices for data centers:
Lessons learned from benchmarking 22 data centers," Proceedings of the ACEEE Summer Study
on Energy Efficiency in Buildings in Asilomar, CA. ACEEE, August, vol. 3, pp. 76-87, 2006.

118
[106] L. Wang and S. U. Khan, "Review of performance metrics for green data centers: a taxonomy
study," The journal of supercomputing, vol. 63, no. 3, pp. 639-656, 2013.
[107] P. Mathew, S. Ganguly, S. Greenberg, and D. Sartor, "Self-benchmarking guide for data centers:
Metrics, benchmarks, actions," Lawrence Berkeley National Lab.(LBNL), Berkeley, CA (United
States), 2009.
[108] M. Pärssinen, M. Wahlroos, J. Manner, and S. Syri, "Waste heat from data centers: An
investment analysis," Sustainable Cities and Society, vol. 44, pp. 428-444, 2019.
[109] T. Lu, X. Lü, M. Remes, and M. Viljanen, "Investigation of air management and energy
performance in a data center in Finland: Case study," Energy and Buildings, vol. 43, no. 12, pp.
3360-3372, 2011.
[110] M. Monroe, "How to Reuse Waste Heat from Data Centers Intelligently," 2016.
[111] M. Wahlroos, M. Pärssinen, S. Rinne, S. Syri, and J. Manner, "Future views on waste heat
utilization–Case of data centers in Northern Europe," Renewable and Sustainable Energy
Reviews, vol. 82, pp. 1749-1764, 2018.
[112] J. Velkova, "Thermopolitics of data: cloud infrastructures and energy futures," Cultural studies,
vol. 35, no. 4-5, pp. 663-683, 2021.
[113] J. Vela, "Helsinki data centre to heat homes," The Guardian, vol. 20, 2010.
[114] M. Fontecchio, "Companies reuse data center waste heat to improve energy efficiency,"
TechTarget SearchDataCenter, 2008.
[115] R. Miller, "Data center used to heat swimming pool," Data Center Knowledge, vol. 2, 2008.
[116] E. Oró, P. Taddeo, and J. Salom, "Waste heat recovery from urban air cooled data centres to
increase energy efficiency of district heating networks," Sustainable cities and society, vol. 45,
pp. 522-542, 2019.
[117] J. Nerell. "Heat recovery from data centres." https://eumayors.ec.europa.eu/sites/default/files/2023-10/2023_CoMo_CaseStudy_Stockholm_EN.pdf
[118] "Westinbldg.com, 2018." https://www.westinbldg.com/Content/PDF/WBX_Fact_Sheet.pdf
[119] T. Nieminen. "Fortum and Microsoft’s data centre project advances climate targets."
https://www.fortum.com/data-centres-helsinki-region
[120] P. Huang et al., "A review of data centers as prosumers in district energy systems: Renewable
energy integration and waste heat reuse for district heating," Applied Energy, vol. 258, p.
114109, 2020.

119
[121] K. Ebrahimi, G. F. Jones, and A. S. Fleischer, "A review of data center cooling technology,
operating conditions and the corresponding low-grade waste heat recovery opportunities,"
Renewable and Sustainable Energy Reviews, vol. 31, pp. 622-638, 2014.
[122] M. Sandberg, M. Risberg, A.-L. Ljung, D. Varagnolo, D. Xiong, and M. Nilsson, "A modelling
methodology for assessing use of datacenter waste heat in greenhouses," in Third International
Conference on Environmental Science and Technology, ICOEST, Budapest, 19-23 October 2017,
2017.
[123] G. Mountain. "Land-based lobster farming will use waste heat from data center."
https://greenmountain.no/data-center-heat-reuse/
[124] W. D. Center. "The Commencement of Japanese Eel Farming."
https://corp.wdc.co.jp/en/news/2024/01/240/
[125] "Data Center Optimization Initiative (DCOI)." https://viz.ogp-mgmt.fcs.gsa.gov/data-center-v2
[126] S. Moss. https://www.datacenterdynamics.com/en/news/giant-us-federal-spending-billincludes-data-center-energy-efficiency-measures/
[127] J. Oltmanns, D. Sauerwein, F. Dammel, P. Stephan, and C. Kuhn, "Potential for waste heat
utilization of hot-water-cooled data centers: A case study," Energy Science & Engineering, vol. 8,
no. 5, pp. 1793-1810, 2020.
[128] J. Hou, H. Li, N. Nord, and G. Huang, "Model predictive control for a university heat prosumer
with data centre waste heat and thermal energy storage," Energy, vol. 267, p. 126579, 2023.
[129] C. Montagud-Montalvá, E. Navarro-Peris, T. Gómez-Navarro, X. Masip-Sanchis, and C. Prades-Gil,
"Recovery of waste heat from data centres for decarbonisation of university campuses in a
Mediterranean climate," Energy Conversion and Management, vol. 290, p. 117212, 2023.
[130] H. Li, J. Hou, T. Hong, Y. Ding, and N. Nord, "Energy, economic, and environmental analysis of
integration of thermal energy storage into district heating systems using waste heat from data
centres," Energy, vol. 219, p. 119582, 2021.
[131] ISO/IEC 30134-2:2016, 2018. [Online]. Available: https://www.iso.org/standard/63451.html
[132] C. Nadjahi, H. Louahlia, and S. Lemasson, "A review of thermal management and innovative
cooling strategies for data center," Sustainable Computing: Informatics and Systems, vol. 19, pp.
14-28, 2018.
[133] H. Alissa, K. Nemati, B. Sammakia, K. Ghose, M. Seymour, and R. Schmidt, "Innovative
approaches of experimentally guided CFD modeling for data centers," in 2015 31st Thermal
Measurement, Modeling & Management Symposium (SEMI-THERM), 2015: IEEE, pp. 176-184.
[134] M. Wahlroos, M. Pärssinen, J. Manner, and S. Syri, "Utilizing data center waste heat in district
heating–Impacts on energy efficiency and prospects for low-temperature district heating
networks," Energy, vol. 140, pp. 1228-1238, 2017.

120
[135] P. Li, J. Yang, M. A. Islam, and S. Ren, "Making ai less" thirsty": Uncovering and addressing the
secret water footprint of ai models," arXiv preprint arXiv:2304.03271, 2023.
[136] "Data Center Cooling Market by Solution (Air Conditioning, Chilling Unit, Cooling Tower,
Economizer System, liquid Cooling System, Control System), Service, Type of Cooling, Data
Center Type, Industry, & Geography - Global Forecast to 2030," 2024.

Abstract (if available)

Abstract Membrane distillation (MD) is a thermally driven water desalination process gaining interest for its ability to treat challenging saline waters, achieve high rejection of non-volatile contaminants, and contribute to zero-liquid-discharge (ZLD) desalination. However, MD is often criticized for its high thermal energy demand and inefficiency compared to pressure-driven processes like reverse osmosis (RO). Additionally, MD performance can be significantly degraded due to membrane compaction, scaling, and biofouling. This work aims to advance the MD process by achieving deeper insights into membrane compaction and integrating waste heat to improved energy efficiency. A novel method combining electrical impedance spectroscopy (EIS) with dynamic mechanical analysis (DMA) is developed to characterize membrane compaction in real-time under varying temperatures and pressures. Additionally, an analysis framework is created to identify and utilize low-grade waste heat from an academic data center. A CFD model simulates the quality and quantity of waste heat emissions, which can be used for MD process. This study provides valuable insights into understanding real-time membrane compaction and the development of membrane fatigue, as well as strategies to enhance MD system efficiency through innovative heat reuse techniques.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

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

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

PDF

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

PDF

Fate of antibiotic resistance in anaerobic membrane bioreactors

PDF

Catalytic oxidation and precursor identification of disinfection byproducts in recycled water

PDF

Expanding sputtering synthesis domains for the discovery of novel film microstructures and morphologies

PDF

Design, dynamics, and control of miniature catalytic combustion engines and direct propane PEM fuel cells

PDF

Assessing indoor and outdoor air quality by characterizing the physicochemical and toxicological properties of particulate matter using…

PDF

Optimizing biomembrane reactor systems for water reclamation and reuse applications

PDF

Physical principles of membrane mechanics, membrane domain formation, and cellular signal transduction

PDF

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

PDF

Predictive real-time assessment for power grid monitoring and situational awareness using synchrophasors with partial PMU observability

PDF

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

PDF

The physics of emergent membrane phenomena

PDF

Mesoscale SOFC-based power generator system: modeling and experiments

PDF

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

PDF

Process intensification in hydrogen production via membrane-based reactive separations

PDF

Wastewater-based epidemiology for emerging biological contaminants

Asset Metadata

Creator Ding, Weijian (author)

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

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Environmental Engineering

Degree Conferral Date 2024-08

Publication Date 10/14/2024

Defense Date 06/12/2024

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

Tag dynamic mechanical analysis,electrochemical impedance spectroscopy,fatigue,membrane distillation,OAI-PMH Harvest,real-time compaction monitoring,water treatment membranes

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Childress, Amy (committee chair), Hodge, Andrea (committee member), McCurry, Daniel (committee member), Smith, Adam (committee member)

Creator Email dingw@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC113996WTH

Unique identifier UC113996WTH

Identifier etd-DingWeijia-13129.pdf (filename)

Legacy Identifier etd-DingWeijia-13129

Document Type Dissertation

Format theses (aat)

Rights Ding, Weijian

Internet Media Type application/pdf

Type texts

Source 20240620-usctheses-batch-1172 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

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

Repository Email cisadmin@lib.usc.edu