pubs.acs.

org/est Article

Comparison of Energy Consumption of Osmotically Assisted

Reverse Osmosis and Low-Salt-Rejection Reverse Osmosis for Brine
Management
Zhangxin Wang, Dejun Feng, Yuanmiaoliang Chen, Di He,* and Menachem Elimelech*
Cite This: Environ. Sci. Technol. 2021, 55, 10714−10723 Read Online

ACCESS Metrics & More Article Recommendations *

sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
Downloaded via GUANGXI UNIV FOR NATIONALITIES on August 10, 2021 at 23:53:45 (UTC).

ABSTRACT: Minimum and zero liquid discharge (MLD/ZLD) are emerging

brine management strategies that attract heightened attention. Although
conventional reverse osmosis (RO) can improve the energy efficiency of MLD/
ZLD processes, its application is limited by the maximum hydraulic pressure
(ΔPmax) that can be applied in current membrane modules. To overcome such
limitation, novel RO-based technologies, including osmotically assisted RO
(OARO) and low-salt-rejection RO (LSRRO), have been proposed. Herein, we
utilize process modeling to systematically compare the energy consumption of
OARO and LSRRO for MLD/ZLD applications. Our modeling results show that
the specific energy consumption (SEC) of LSRRO is lower (by up to ∼30%) than that of OARO for concentrating moderately saline
feed waters (<∼35,000 mg/L TDS) to meet MLD/ZLD goals, whereas the SEC of OARO is lower (by up to ∼40%) than that of
LSSRO for concentrating higher salinity feed waters (>∼70,000 mg/L TDS). However, by implementing more stages and/or an
elevated ΔPmax, LSRRO has the potential to outperform OARO energetically for treating high-salinity feed waters. Notably, the SEC
of both OARO and LSRRO could be 50% lower than that of mechanical vapor compressor, the commonly used brine concentrator
in MLD/ZLD applications. We conclude with a discussion on the practicability of OARO and LSRRO based on membrane module
availability and capital cost, suggesting that LSRRO could potentially be more feasible than OARO.
KEYWORDS: brine management, reverse osmosis, brine concentration, energy efficiency, practicability, minimal liquid discharge,
zero liquid discharge

■ INTRODUCTION
Freshwater scarcity has become a critical global challenge.1−3 A
use of RO reduces the volume of brine treated using a thermal
brine concentrator, thereby improving the energy and cost
substantial amount of freshwater is turned to wastewater due efficiency of ZLD systems. Due to its high energy utilization
to the ever-growing public and industrial activities.4 Discharg- efficiency,14 RO is also commonly used in MLD systems.10,15
ing wastewaters without adequate treatment could cause severe However, to date, ZLD systems are still costly because
pollution of freshwater resources, exacerbating freshwater conventional RO cannot completely replace the thermal brine
scarcity.5,6 To address this global problem, a strict wastewater concentrator due to the restriction on the operating
management strategy, zero liquid discharge (ZLD), has been pressure.16,17 Specifically, with the current operating pressure
proposed.7,8 By recovering all the water in wastewater streams of conventional RO (<85 bar),16 its maximum brine
and leaving only solid waste, ZLD can obviate the risk of concentration cannot exceed 100,000 mg/L, whereas the
wastewater discharge and maximize freshwater usage efficiency. required brine concentration for a brine crystallizer is over
Because ZLD is very costly to achieve,9 minimum liquid 200,000 mg/L.7 Furthermore, the amount of freshwater that
discharge (MLD) that targets a water recovery of up to 95% can be recovered using current MLD systems is also limited
has been proposed.10,11 Unlike ZLD, MLD allows the balance because of the limit on the operating pressures in RO.17
of economic factors with environmental impacts. To develop MLD/ZLD systems with improved perform-
Early ZLD systems were built based on thermal processes, ance, novel staged RO technologies that can overcome the
where the saline wastewater was evaporated first in a brine
concentrator and then in a brine crystallizer or an evaporation
pond.7,12 The application of such systems was economically Received: March 12, 2021
prohibitive because of the high capital and operational costs of Revised: June 29, 2021
thermal processes. To improve the efficiency of brine Accepted: July 1, 2021
management, reverse osmosis (RO), a membrane-based Published: July 16, 2021
technology widely used in desalination, has been incorporated
in ZLD systems before the thermal brine concentrator.7,13 The

© 2021 American Chemical Society https://doi.org/10.1021/acs.est.1c01638

10714 Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

applied pressure limit of conventional RO are needed. RO, the allowable ΔP must be increased, which is currently
Specifically, osmotically assisted RO (OARO) and low-salt- not feasible.
rejection RO (LSRRO) have been proposed.16−22 Both OARO Osmotically Assisted RO. To boost Rw,max without
and LSRRO are able to highly concentrate wastewaters using elevating ΔP, a novel staged RO configuration named
moderate operating pressures, thereby advancing MLD/ZLD OARO has been developed (Figure 2A).18,19 An N-stage
systems. However, the “sweet spots” for these two technologies OARO system comprises one conventional RO module and N
(i.e., scenarios in which a specific technology is preferred) for − 1 bilateral RO modules, with the conventional RO module
MLD/ZLD applications have not been established. Compar- placed at the last stage. In each bilateral stage, two saline
ison of the performance of these technologies is therefore streams flow countercurrently with one being pressurized and
critical for guiding the development of future MLD/ZLD the other remaining unpressurized. The pressure-driven water
systems and for the ultimate goal of displacing thermal permeation results in concentration of the pressurized stream
evaporators in MLD/ZLD systems. and dilution of the unpressurized stream. To maintain steady
In this study, we use process modeling to systematically concentrations of the saline streams, the saline streams loop
compare the two novel RO-based technologies, OARO and between adjacent stages, with the water permeation volume at
LSRRO, for brine management and MLD/ZLD applications. each stage being fixed.16 Specifically, after being concentrated
We start by introducing the working principles of OARO and in one stage, the pressurized stream is depressurized and cycled
LSRRO, demonstrating that both technologies can achieve back to the previous stage for dilution; after being diluted in
higher water recovery rates (i.e., discharge brines with higher that stage, the unpressurized stream is pressurized and sent to
concentrations and smaller volumes) than conventional RO the next stage for concentration. Freshwater is produced from
with moderate operating pressures. Using specific energy the last conventional RO stage, and the final brine is
consumption (SEC) as a performance metric, we systemati- discharged from the first bilateral stage after recovering the
cally compare OARO and LSRRO in MLD/ZLD applications. energy embedded in the brine stream.
By analyzing the impact of operating parameters, including With the use of bilateral stages in OARO, the osmotic
feed salt concentration, number of stages employed, and pressure of the unpressurized saline stream (πS) is able to
maximum operating pressure on SEC, we identify the sweet reduce Δπ from πB to πB − πS. Therefore, following the
spots of feed water salinities for each technology. We conclude working principle of RO (i.e., ΔP ≥ Δπ), the maximum πB in
with a discussion of practical considerations of the two OARO (πB,max) can exceed ΔP (i.e., πB,max = ΔP + πS).
technologies, highlighting potential advantages of LSRRO over According to van’t Hoff’s relationship, a larger π B,max
OARO for MLD/ZLD applications. corresponds to a higher cB,max, and Rw,max increases with

■
increasing cB,max (eq 1). Thus, OARO can achieve a higher
NOVEL RO-BASED PROCESSES FOR MLD AND Rw,max than conventional RO with the same ΔP.
ZLD Low-Salt-Rejection RO. Another staged RO technology
LSRROthat can achieve a high Rw,max with moderate ΔP has
Limitations of Conventional RO. In order to recover
recently been proposed (Figure 2B).17 In an LSRRO system, a
freshwater from the saline feed, the applied hydraulic pressure
conventional RO module is used in the first stage, while
(ΔP) in RO needs to be larger than the transmembrane
LSRRO modules are used in the subsequent stages. During
osmotic pressure difference (Δπ). In conventional RO with
operation, in each low-salt-rejection stage, the pressurized
near perfect salt rejection, Δπ is equal to the osmotic pressure
concentrated brine is directly sent to the next stage as feed,
of the brine (πB) because the osmotic pressure of the product
while the diluted permeate is repressurized and cycled back to
water is negligible (Figure 1A). According to van’t Hoff’s
the previous stage as additional feed (i.e., ΔP at each stage is
approximation, πB increases with the brine concentration (cB).
constant). Freshwater is produced from the first conventional
During RO operation, the feed becomes more concentrated as
RO stage, while the final brine is discharged from the last low-
more freshwater is being recovered, leading to an increased πB.
salt-rejection stage after energy recovery.
Once πB reaches ΔP, the water recovery ceases and the feed
Like OARO, LSRRO employs saline solutions to mediate
cannot be further concentrated. In other words, the maximum
Δπ, albeit through a different pathway. Specifically, in LSSRO,
brine concentration (cB,max) of conventional RO is achieved as
the use of LSRRO membranes results in saline permeates in
πB = ΔP (Figure 1B). Following simple mass balance
the low-salt-rejection stages. The osmotic pressure of the saline
equations, the maximum water recovery (Rw,max) can be
permeate (πP) can reduce Δπ from πB to πB − πp, thereby
calculated, knowing cB,max and the initial feed salt concentration
resulting in a πB,max larger than ΔP (i.e., πB,max = ΔP + πP). In
(c0) (detailed derivations in the Supporting Information)
other words, LSRRO can achieve a larger πB,max than
c0 conventional RO with the same ΔP. Because a larger πB,max
R w,max = 1 − leads to a higher Rw,max for given ΔP, LSRRO can attain a
c B,max (1)
higher Rw,max compared to conventional RO.
For a given ΔP, cB,max is fixed. According to eq 1, with a fixed
cB,max, Rw,max decreases with increasing c0. To avoid detrimental
effects of high hydraulic pressure on membranes and
■ PERFORMANCE OF OARO AND LSRRO IN MLD
APPLICATIONS
modules,23,24 ΔP is typically <85 bar.16 In most RO Water Recovery in Two-Stage OARO and LSRRO for
operations, ΔP does not exceed 80 bar, resulting in a cB,max MLD. Based on the working principles of OARO and LSRRO,
of ∼94,000 mg/L TDS (Figure 1B). Such cB,max leads to Rw,max the theoretical maximum brine concentration (cB,max) and
of ∼93, 63, and 26% for feed solutions with c0 of 10,000, maximum water recovery rate (Rw,max) in an N-stage OARO/
35,000, and 70,000 mg/L TDS, respectively. As discussed LSRRO system can be calculated as a function of the stage
previously, for MLD/ZLD applications, a higher Rw,max is number (N) and maximum applied pressure (ΔPmax) (detailed
desired. However, to further enhance Rw,max in conventional derivations in the Supporting Information)
10715 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

Figure 1. Conventional RO operation and performance limits. (A) Schematic of a conventional RO system. The color darkness qualitatively
represents the salinity of the solution, and the white arrow denotes the direction of transmembrane water flux. (B) Maximum brine concentration
(cB,max, orange left vertical axis) and maximum water recovery (Rw,max, blue right vertical axis) as a function of applied hydraulic pressure (ΔP). The
dotted, dashed, and solid blue curves depict the maximum water recovery at a given applied hydraulic pressure for feed waters with initial feed salt
concentrations (c0) of 10,000, 35,000, and 70,000 mg/L, respectively. The solid linear line (orange) represents the maximum possible brine
concentration at each applied hydraulic pressure. The green region represents the operation regime of conventional seawater RO (i.e., ΔP ≤ ∼80
bar). The water temperature in the calculations was set at 298 K.

Figure 2. Working principles of novel RO-based brine concentration technologies. (A) Schematic of an N-stage OARO. The stages with double
lines in the middle represent the bilateral RO modules, and the stage with a solid line in the middle refers to a conventional RO module. In each
stage, fresh water permeates from the pressurized and more saline stream to the unpressurized and less saline stream, with the streams being looped
between adjacent stages using high-pressure pumps and energy recovery devices (ERDs). (B) Schematic of an N-stage LSRRO. The stage with a
solid line in the middle denotes a conventional RO module, while the stages with dotted lines in the middle represent the LSRRO modules. In each
stage, the pressurized brine is directed to the next stage, serving as feed, and the permeate is repressurized and sent back to the previous stage as
additional feed. In both panels, the color darkness qualitatively represents the salinity of the solution, and the white arrows denote the direction of
the transmembrane water flux.

N ΔPmaxc0 conventional RO and two-stage OARO/LSRRO can achieve

c B,max =
π0 (2a) Rw,max of 89 and 95%, respectively, resulting in an Rw,max
difference of only 6%. In contrast, with a c0 of 70,000 mg/L,
π0 conventional RO and two-stage OARO/LSRRO can achieve
R w,max = 1 − Rw,max of 26 and 63%, respectively, leading to an Rw,max
N ΔPmax (2b)
difference of 37%.
where π0 is the osmotic pressure of the feed solution. We note that, despite the dependence of Rw,max on c0, (1 −
In practical applications, the more stages an OARO/LSRRO Rw,max) in OARO/LSRRO is always half of that in conventional
system uses, the more complicated the system construction is. RO, suggesting that two-stage OARO/LSRRO can reduce the
Therefore, for the MLD applications discussed in this section, minimum brine flow rate (QB,min) from conventional RO by
OARO and LSRRO systems with two stages were investigated. half. Reducing QB,min by a factor of two in two-stage OARO/
Comparison of Rw,max between conventional RO and two-stage
LSRRO is directly related to cB,max being two times of that in
OARO/LSRRO is shown in Figure 3A. For all three feed
solutions shown, two-stage OARO/LSRRO can achieve higher conventional RO (eq 2a). Based on mass balance of salts in
Rw,max than conventional RO. The difference in Rw,max between RO operation (i.e., QBcB = Q0c0, where Q0 is the initial feed
two-stage OARO/LSRRO and conventional RO varies with flow rate), a 100% increase in cB,max results in a 50% reduction
the feed salt concentration (c0). With a c0 of 10,000 mg/L, of QB,min. Therefore, two-stage OARO/LSRRO can signifi-
10716 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

Figure 3. Comparison between OARO and LSRRO in MLD applications. (A) Maximum water recovery (Rw,max) from conventional RO (gray
columns) and two-stage OARO/LSRRO (blue columns) with varying feed salt concentrations (c0). (B) SEC of two-stage OARO (orange
columns) and LSRRO (green columns) for treating feed waters of varying salinities in MLD applications. The targeted brine concentration (cB)
was set at 140,000 mg/L, corresponding to water recoveries (Rw) of 93, 75, and 50% for c0 of 10,000, 35,000, and 70,000 mg/L, respectively. SEC
in each stage (SECi) of two-stage OARO for concentrating feed solutions of (C) 10,000 and (D) 70,000−140,000 mg/L. SECi of two-stage
LSRRO for concentrating feed solutions of (E) 10,000 and (F) 70,000−140,000 mg/L. In panels (C−F), the x-axis refers to the permeate flow rate
in each stage (QPi) normalized by the flow rate of the produced fresh water (QP) and the dashed blue lines denote the osmotic pressure of the
brine. In all calculations, the maximum hydraulic pressure (ΔP) was assumed to be 80 bar, the over-pressurization factor (k) was set to 1.1, and the
temperature was kept at 298 K. Detailed operating conditions (ΔP) and SEC in each stage of OARO are presented in Table S1 of the Supporting
Information. The SEC of LSRRO is the minimum SEC after optimizing the observed salt rejection rates of the stages and ΔP; detailed information
is shown in Table S4 of the Supporting Information.

cantly facilitate MLD applications because it can further To maintain adequate water flux along the membrane
minimize the volume of brines compared to conventional RO. module in practical RO operation, ΔP is in excess of Δπ
Energy Consumption in Two-Stage OARO and LSRRO everywhere inside the module (i.e., Δπ < ΔP).16 The ratio of
for MLD. Because both OARO and LSRRO can achieve high ΔP to Δπ at the brine outlet is the over-pressurization factor
Rw,max with moderate ΔP, the SEC for water recovery becomes denoted by k (i.e., ΔP = kΔπ). Because cB,max and Rw,max in eqs
a critical performance metric for the comparison between the 2a and 2b and Figure 3A are obtained at the limit of Δπ = ΔP,
two technologies in MLD applications. By definition, SEC in they are theoretical performance limits that can hardly be
an N-stage OARO/LSRRO system is the total energy achieved in practical applications because in practical
consumption normalized by the volume of the recovered or applications, an over-pressure is applied to compensate for
produced freshwater. Neglecting the inefficiencies of the ERDs, concentration polarization and the pressure drop along the
SEC can be calculated as module. Therefore, for the comparison of SEC in two-stage
OARO and LSRRO for practical MLD applications, with a
N
∑i = 1 Q P, iΔPi ΔPmax of 80 bar, a practical cB of 140,000 mg/L was set as the
SEC = target brine concentration and k was assumed to be 1.1.16,25
QP (3) For a comprehensive comparison, three feed solutions with
varying c0 (i.e., 10,000, 35,000, and 70,000 mg/L) were
where QP is the flow rate of the produced fresh water, QP,i is investigated. With a target cB of 140,000 mg/L, the water
the permeate flow rate in the ith stage, and ΔPi is the operating recoveries (Rw) for the feed solutions with c0 of 10,000, 35,000,
pressure in the ith stage. and 70,000 mg/L are 93, 75, and 50%, respectively. The
10717 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

calculated SEC for the different feed solutions are presented in the required brine concentration (cB) at the inlet of a thermal
Figure 3B (detailed calculation in the Supporting Information). brine crystallizer is over 200,000 mg/L.7 Based on eq 2a, to
As shown in Figure 3B, the SEC of two-stage LSRRO achieve such concentrated brine in OARO or LSRRO, at least
increases with increasing c0, whereas the SEC of two-stage three stages are needed. Throughout this study, a target cB of
OARO remains constant. To elucidate the difference in SEC of 234,000 mg/L was set for ZLD applications. With a ΔPmax of
two-stage OARO and LSRRO with different c0, the total SEC 80 bar and over-pressurization factor k of 1.1, the SEC of
is broken into the SEC in each stage. By definition, the SEC in three-stage OARO and LSRRO in ZLD applications with
the ith stage (SECi) can be calculated as different c0 has been calculated, as shown in Figure 4A
(detailed calculation in the Supporting Information).
Q P, iΔPi Like MLD applications (Figure 3B), for ZLD applications
SECi =
QP (4) with increasing c0, the SEC of OARO remains constant,
whereas the SEC of LSRRO increases (Figure 4A). As
Following the working principle of OARO (Figure 2A), the discussed previously, the SEC of OARO is only determined
permeate flow rate from each stage is constant (i.e., QP,i = QP). from cB, which is independent of c0. In contrast, the SEC of
Thus, SECi is equal to ΔPi (eq 4), and the total SEC is equal to LSRRO increases with c0 because a higher c0 results in larger
the sum of ΔPi (Figure 3C,D). As ΔPi = kΔπi, the total SEC is permeate flow rates in the low-salt-rejection stages, requiring
equivalent to the sum of kΔπi. In the bilateral stages, Δπi is the more energy for repressurization and recirculation. Based on
osmotic pressure difference between the concentrated streams the SEC comparison (Figure 4A), with three stages, LSRRO is
at the outlets of the ith stage and the i + 1th stage (i.e., Δπi = more competitive than OARO only when treating feed water
πB,i − πB,i+1); in the last conventional RO stage, ΔπN = πB,N. with a relatively low c0 (i.e., 10,000 mg/L), while for treating
Thus, the total SEC can be calculated as feed waters with higher c0 (i.e., 35,000 and 70,000 mg/L),
N N−1 OARO is substantially more energy efficient than LSRRO
SEC = ∑ kΔπi = ∑ k(πB,i − πB,i+ 1) + kπB,N = kπB,1 (Figure 4A). We note that mechanical vapor compression
i i
(MVC), the most widely used thermal brine concentrator in
(5) ZLD applications, has an SEC of 20−25 kW h m−3.26,27
Because the SEC in three-stage OARO is less than 30% of that
For OARO, because the final concentrated brine is in MVC, three-stage OARO is energetically superior to MVC.
discharged from the first stage, πB,1 is indeed the osmotic In contrast, three-stage LSRRO is more energy efficient than
pressure of the concentrated brine from OARO, πB. Therefore, MVC for treating moderately saline feeds, while for treating
the SEC of OARO is equivalent to kπB (Figure 3C,D). In other hypersaline feeds, the two technologies are energetically
words, for a given cB in ORAO, the SEC is determined. comparable.
Therefore, the SEC of two-stage OARO is independent of c0. Energy Consumption in Four-Stage OARO and
Following the working principle of LSRRO, ΔPi in each LSRRO for ZLD. The SEC of four-stage OARO and LSRRO
stage is constant (i.e., ΔPi = ΔP). For two-stage LSRRO with in ZLD applications with different c0 has also been calculated
different c0, SEC1 is approximately the same (Figure 3E,F). (Figure 4B), assuming the same operating conditions as three-
The similar SEC1 is because SEC1 is equal to ΔP, and to stage OARO and LSRRO (i.e., ΔPmax of 80 bar and k of 1.1).
achieve the minimum SEC in two-stage LSRRO, ΔP is always Detailed calculations are presented in the Supporting
close to ΔPmax (Table S4 in the Supporting Information). The Information. For OARO, the SEC remains constant with the
difference in the overall SEC in two-stage LSRRO is attributed stage number. To elucidate the independence of SEC on the
to SEC2. For a given ΔP, the flow rate of the brine from the 1st stage number, the SEC of each stage in three-stage and four-
conventional RO stage (QB,1) increases with c0. With a larger stage OARO (SECi) for a c0 of 35,000 mg/L was calculated
QB,1 sent to the 2nd stage, the permeate flow rate from the 2nd and compared. As shown in Figure 4C,D, for a given cB in
stage (QP,2) becomes larger. A larger QP,2 requires a larger ORAO, although the required hydraulic pressure in each stage
SEC2 for repressurization and recirculation, thereby leading to (ΔPi) can be reduced with more stages being used, the sum of
a larger SEC, as seen when comparing SEC2 in Figure 3E,F. ΔPi remains the same based on our previous discussion (i.e.,
As shown in Figure 3B, the SEC of two-stage LSRRO is ∑Ni=1ΔPi = kπB). Because the SEC of OARO is equal to the
∼70% of that of two-stage OARO for a c0 of 10,000 mg/L, sum of ΔPi (eq 3), it is independent of the stage number (i.e.,
while LSRRO and OARO have comparable SEC for a c0 of SEC = kπB).
35,000 mg/L. As c0 is increased to 70,000 mg/L, the SEC of Likewise, to understand the different SEC of LSRRO with
two-stage LSRRO increases and is ∼65% greater than that of different stage numbers, the SECi in three-stage and four-stage
two-stage OARO. This comparison suggests that in MLD LSRRO for a c0 of 35,000 mg/L was calculated and shown in
applications, two-stage LSRRO is more energy efficient than Figure 4E,F. The SEC of the first conventional RO stage
two-stage OARO for treating low-salinity feeds, while two- (SEC1) is similar in three-stage and four-stage LSRRO. As
stage OARO is more energy efficient than two-stage LSRRO previously discussed, SEC1 in LSRRO is always equal to ΔP. In
for treating hypersaline feeds.

■
both three-stage and four-stage LSRRO, to achieve the
minimum SEC, ΔP should be close to ΔPmax (Tables S2 in
PERFORMANCE OF OARO AND LSRRO IN ZLD the Supporting Information), thereby resulting in similar SEC1.
APPLICATIONS The difference in SEC of three-stage and four-stage LSRRO is
Energy Consumption in Three-Stage OARO and attributed to the different SECi in the low-salt-rejection stages.
LSRRO for ZLD. The goal of OARO/LSRRO in ZLD For a given cB, the total feed volume reduction in LSRRO is
applications is to produce a highly concentrated brine that can fixed. When four stages are used, the permeate flow rates (QP,i)
be directly sent to the thermal brine crystallizer, thus in the low-salt-rejection stages substantially decrease (Figure
eliminating the use of thermal brine concentrator. Typically, 4E,F, Table S5 in the Supporting Information). As smaller QP,i
10718 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

Figure 4. Comparison of energy consumption in multi-stage OARO and LSRRO in ZLD applications. (A) SEC of three-stage OARO (orange
columns) and LSRRO (green columns) with varying feed salt concentrations (c0). SEC in each stage (SECi) of (C) three-stage and (D) four-stage
OARO with a c0 of 35,000 mg/L. SECi of (E) three-stage and (F) four-stage LSRRO with a c0 of 35,000 mg/L. The target brine concentration (cB)
was set at 234,000 mg/L (feed for crystallizer in ZLD); the blue dashed lines in panels (C−F) denote the corresponding osmotic pressure of the
target brine concentration. In all calculations, the maximum operating pressure (ΔPmax) was assumed to be 80 bar with an over-pressurization factor
(k) of 1.1, and the temperature was kept at 298 K. Detailed operating conditions (ΔP) and SEC for each stage in OARO are presented in Table S2
of the Supporting Information. The SEC in LSRRO is the minimum SEC after optimizing the observed salt-rejection rates of the stages and ΔP;
detailed information is presented in Table S5 in Supporting Information.

demands smaller energy consumption for repressurization and recently produced high-pressure RO elements that can
recirculation, the SEC of four-stage LSRRO becomes notably withstand an operating pressure of up to 120 bar.26 With
smaller than that of three-stage LSRRO. this technical development, it is necessary to compare the
With four stages, the SEC advantage of LSRRO toward future performances of OARO and LSRRO at elevated ΔPmax.
OARO in treating feed waters with relatively low c0 (i.e., Assuming a ΔPmax of 120 bar (i.e., the maximum operating
10,000 mg/L) is amplified, and for treating feed waters with pressure of current high-pressure RO elements), the SEC of
moderate c0 (i.e., 35,000 mg/L), LSRRO becomes more three-stage OARO and LSRRO in ZLD application with
energy efficient than OARO. Furthermore, when treating feed different feed solutions was calculated and shown in Figure 5A
waters with high c0 (i.e., 70,000 mg/L), the SEC advantage of (detailed calculations in the Supporting Information).
four-stage OARO over four-stage LSRRO decreases substan- Comparing Figures 4A and 5A, the SEC of three-stage
tially compared to the case with three-stage OARO and OARO with ΔPmax of 80 and 120 bar is the same, because
LSRRO. Following such a trend, LSRRO could be energeti- ΔPmax is not a limiting factor in three-stage OARO for ZLD
cally more favorable than OARO with more stages, even for applications. To achieve a cB of 234,000 mg/L with a k of 1.1,
treating hypersaline feed waters. Notably, four-stage LSRRO is three-stage OARO only requires a ΔP of ∼73 bar. Once ΔPmax
much more energy efficient than MVC, with SEC of LSRRO exceeds 73 bar, three-stage OARO can be used in ZLD
being 2.5−10 kW h m−3 compared to 20−25 kW h m−3 of applications. A larger ΔPmax does not benefit SEC because SEC
MVC. is only determined from cB. However, for three-stage LSRRO,
Performance of OARO and LSRRO with Elevated the SEC with a ΔPmax of 120 bar (Figure 5A) is substantially
Operating Pressure for ZLD. Although current ΔPmax of smaller than that with a ΔPmax of 80 bar (Figure 4A). Such
conventional RO modules is limited to ∼85 bar, it could be difference can be explained by SECi (Figure 5B). With a larger
increased in the future if a demand for such modules arises. ΔPmax, despite an increased SEC1, the brine volume from the
Several RO membrane manufacturers (e.g., Dow) have first conventional stage RO decreases. A decreased brine
10719 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

Figure 5. Comparison of energy consumption in OARO and LSRRO operating at an elevated hydraulic pressure. (A) SEC of three-stage OARO
and LSRRO with varying feed salt concentrations (c0) and a maximum operating pressure (ΔPmax) of 120 bar. (B) SEC in each stage (SECi) of
three-stage LSRRO with a feed salt concentration of 35,000 mg/L and ΔPmax of 120 bar. In all calculations, the target brine concentration (cB) was
set at 234,000 mg/L (feed for crystallizer in ZLD), the over-pressurization factor (k) was fixed at 1.1, and the temperature was kept at 298 K. The
blue dashed line in panel B denotes the corresponding osmotic pressure of the target brine concentration. Detailed operating conditions (ΔP) and
SEC in each stage of OARO are presented in Table S3 of the Supporting Information. The SEC in LSRRO is the minimum SEC after optimizing
the observed salt rejection rates of the stages and ΔP; detailed information is presented in Table S6 of the Supporting Information.

volume substantially reduces the permeate flow rates in the

low-salt-rejection stages (QPi) for repressurization and
recirculation, thereby leading to a notably smaller SEC.
At an elevated ΔPmax of 120 bar, three-stage LSRRO is more
energy efficient than three-stage OARO for treating feed waters
with relatively low and moderate c0 (i.e., 10,000 and 35,000
mg/L). For treating hypersaline feed waters (i.e., c0 = 70,000
mg/L), three-stage LSRRO is comparable to three-stage
OARO with a slightly larger SEC. Based on the obtained
SEC with a ΔPmax of 120 bar, we conclude that LSRRO will be
energetically more advantageous than OARO with the
availability of high pressure RO modules. Notably, with a
ΔPmax of 120 bar, the SEC in three-stage LSRRO would be
substantially lower than that in MVC (i.e., 3−6 kW h m−3 in
LSRRO versus 20−25 kW h m−3 in MVC).

■ IMPLICATIONS AND OUTLOOK

To overcome the operating pressure limitation of conventional
Figure 6. Radar chart for comparing the performance of OARO and
LSRRO in MLD/ZLD applications. The green and orange regimes
denote the performance of OARO and LSRRO, respectively. The
RO in MLD/ZLD applications, novel staged RO technologies dashed green curves refer to the potential future performance of
including OARO and LSRRO have been proposed. In this LSRRO.
study, we performed the first direct comparison of the energy
consumption in OARO and LSRRO in MLD/ZLD applica- OARO for treating high-salinity feed waters by implementing
tions using process modeling. For a thorough comparison of more stages and/or elevating the operating pressure, which is
these two technologies, the practicability and potential illustrated by the dashed lines in Figure 6. Based on our
feasibility of the two technologies should be considered in analysis, the SEC of OARO and LSRRO could be less than
addition to energy efficiency. However, because both OARO 50% of that of MVC,25−27 suggesting that both technologies
and LSRRO are emerging technologies, data regarding have the potential to replace MVC in MLD/ZLD applications.
membrane module availability and capital costs are limited. The practicability of OARO and LSRRO was further
Thus, we qualitatively compare OARO and LSRRO using a compared using two performance metrics: membrane module
radar chart approach (Figure 6), followed by a discussion of availability and capital cost. Because membrane modules are
the various components depicted in the radar chart. essential to the practical implementation of OARO/LSRRO
Based on our analysis, both OARO and LSRRO can highly systems, their availability becomes a critical metric for
concentrate saline feed waters using moderate operating evaluating the practicability of OARO/LSRRO. In addition,
pressure. Hence, OARO and LSRRO are equally rated in like other desalination systems, the practical implementation of
this category of MLD/ZLD capability in Figure 6. In MLD/ OARO/LSRRO system is heavily determined by their overall
ZLD applications, LSRRO is more energy efficient than OARO costs.19,28−30 Thus, capital cost is another important perform-
for concentrating low-salinity feed waters, whereas OARO is ance metric for evaluating the practicability of OARO/LSRRO.
energetically more competitive than LSRRO while treating Distinctive membrane modules are used in OARO and
hypersaline feed waters. Therefore, as shown in Figure 6, in LSRRO systems. Specifically, OARO uses bilateral RO
terms of energy efficiency, LSRRO outperforms OARO for membrane modules, in which spacers are needed on both
concentrating low-salinity feed waters, while OARO is better sides of the membrane to simultaneously maintain channel
than LSRRO for treating high-salinity feed waters. However, structures and enhance mass transfer coefficients.31,32 For
we have shown that LSRRO could potentially outperform mitigating the parasitic energy consumption resulting from the
10720 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

cross-flow streams in the module, loose spacers are required. 6). According to the comparison of membrane module
However, loose spacers would cause severe membrane availability and capital cost, LSRRO should be a more practical
deformation under the applied hydraulic pressure that technology than OARO in ZLD/MLD applications.
significantly compromises membrane performance.33,34 Be- We emphasize that this study is a theoretical investigation
cause of the absence of robust bilateral spiral wound RO for providing guidelines for future technology development.
membrane modules, hollow fiber membrane modules have The primary goal of this study is to compare the energy
been used in pilot studies.35,36 However, these cellulose efficiency of OARO and LSRRO, and the whole analysis is
triacetate (CTA) hollow fiber membranes have very low water based on process modeling. For a more detailed and
flux because of the inherent low permeability of the CTA industrially relevant comparison of these two technologies,
active layer and the large structural parameter of the membrane the levelized cost of water (LCOW) needs to be assessed.16,43
which induces significant internal concentration polarization.37 LCOW of a water treatment technology is defined as the cost
The low water flux of such modules will require a very large per unit volume of product water. To estimate LCOW, a
membrane area, which will significantly increase the capital techno-economic analysis should be carried out.44,45 We
cost. Furthermore, with such membrane modules, there is a further note that although the process modeling approach is
significant pressure drop of the stream in the hollow fibers, able to evaluate the energy efficiency of the OARO/LSRRO
which would compromise the energy efficiency of the OARO systems, it cannot provide the specific system design
system. parameters (e.g., membrane properties, membrane module
In contrast, LSRRO uses conventional RO modules with sizes, and feed flow rates) that are critical for techno-economic
low-salt-rejection membranes. Such membrane modules can be analysis. To acquire these design parameters, future work
commercial nanofiltration modules (albeit their operating should carry out detailed module-scale analysis of OARO/
pressure will be limited to less than ∼40 bar) or commercial LSRRO systems.41,46
RO modules after chemically treating the RO membranes with In addition, several practical issues during system operation,
an oxidant such as chlorine.38 The latter is a very appealing such as membrane fouling and scaling, have not been
option as RO modules at their end-of-life can be chemically considered. For practical MLD/ZLD applications, with the
treated to be used in LSRRO, instead of being discarded and feed solutions being highly concentrated, fouling and scaling of
creating solid waste. Therefore, regarding the availability of the membranes could be significant challenges.47,48 Thus, to
membrane modules, LSRRO is advantageous over OARO practically implement OARO/LSRRO, effective pretreatment
(Figure 6), suggesting that LSRRO is more ready to be of feed waters (e.g., the addition of antiscalants, coagulation
implemented in practice than OARO. and softening, and nanofiltration) and/or other fouling and
In fact, even with commercially available bilateral RO scaling mitigation strategies (e.g., development of novel
modules, to avoid rupture of current thin-film composite membrane materials and operational modes) should be
membranes supported by a spacer, the applied ΔP cannot implemented.49−52
exceed 55 bar,24,39 suggesting that the transmembrane water
flux in OARO is limited. Additionally, the water permeation in
the bilateral RO modules could cause severe concentration
■
*
ASSOCIATED CONTENT
sı Supporting Information
polarization on both sides of the membrane, further reducing
The Supporting Information is available free of charge at
the transmembrane water flux.35,40−42 For practical applica-
https://pubs.acs.org/doi/10.1021/acs.est.1c01638.
tions, a smaller transmembrane water flux requires a larger
membrane area, corresponding to an increased capital cost of Derivation of eqs 1, 2a, and 2b; calculation of SEC in N-
membrane modules. In comparison, the LSRRO membrane stage OARO; calculation of SEC in two-stage LSRRO;
modules can withstand high ΔP because densely woven mesh calculation of SEC in three-stage LSRRO; calculation of
spacers can be used on the permeate side to support the thin- SEC in four-stage LSRRO; operating conditions and
film composite membranes. Additionally, with the use of low- SEC of OARO; and optimization results of SEC in
salt-rejection membranes, a large portion of salts permeates LSRRO (PDF)

■
through the membranes, consequently mitigating concen-
tration polarization effects. Thus, LSRRO can maintain
relatively high transmembrane water fluxes and therefore AUTHOR INFORMATION
requires a smaller membrane area than OARO. In other words, Corresponding Authors
the expense for membrane modules in LSRRO is potentially Di He − Institute of Environmental and Ecological Engineering,
lower than that in OARO. Guangdong University of Technology, Guangzhou,
The capital cost of an OARO/LSRRO system includes not Guangdong 510006, China; Southern Marine Science and
only the expense for membrane modules but also the expenses Engineering Guangdong Laboratory (Guangzhou),
for high-pressure pumps and ERDs. Based on the working Guangzhou 511458, China; Email: di.he@gdut.edu.cn
principles of OARO/LSRRO (Figure 2), an N-stage OARO Menachem Elimelech − Department of Chemical and
system comprises N high-pressure pumps, N ERDs, and N Environmental Engineering, Yale University, New Haven,
membrane modules/stages, whereas an N-stage LSRRO Connecticut 06511, United States; orcid.org/0000-0003-
system contains N − 1 high-pressure pumps, only one ERD, 4186-1563; Phone: +1 (203) 432-2789;
and N membrane modules/stages. Obviously, with a similar Email: menachem.elimelech@yale.edu
number of stages, an LSRRO system requires fewer high-
pressure pumps and ERDs than an OARO system. Thus, in Authors
terms of capital cost, LSRRO should be more competitive than Zhangxin Wang − Institute of Environmental and Ecological
OARO as fewer membrane modules, high-pressure pumps, and Engineering, Guangdong University of Technology,
ERDs are needed (i.e., LSRRO outperforms OARO in Figure Guangzhou, Guangdong 510006, China; Southern Marine
10721 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

Science and Engineering Guangdong Laboratory (13) Elimelech, M.; Phillip, W. A. The Future of Seawater
(Guangzhou), Guangzhou 511458, China Desalination: Energy, Technology, and the Environment. Science
Dejun Feng − Institute of Environmental and Ecological 2011, 333, 712−717.
Engineering, Guangdong University of Technology, (14) Park, K.; Kim, J.; Yang, D. R.; Hong, S. Towards a low-energy
Guangzhou, Guangdong 510006, China seawater reverse osmosis desalination plant: A review and theoretical
Yuanmiaoliang Chen − NUS Graduate School for Integrative analysis for future directions. J. Membr. Sci. 2020, 595, 117607.
(15) Dow Minimal Liquid Discharge: A water management
Science and Engineering, National University of Singapore,
approach that can help you increase recovery and reduce costs.
Singapore 117456, Singapore
https://www.dupont.com/content/dam/Dupont2.0/Products/
Complete contact information is available at: water/literature/609-50241.pdf (accessed 07, 21).
https://pubs.acs.org/10.1021/acs.est.1c01638 (16) Chen, X.; Yip, N. Y. Unlocking High-Salinity Desalination with
Cascading Osmotically Mediated Reverse Osmosis: Energy and
Notes Operating Pressure Analysis. Environ. Sci. Technol. 2018, 52, 2242−
The authors declare no competing financial interest. 2250.

■
(17) Wang, Z.; Deshmukh, A.; Du, Y.; Elimelech, M. Minimal and
ACKNOWLEDGMENTS zero liquid discharge with reverse osmosis using low-salt-rejection
membranes. Water Res. 2020, 170, 115317.
This work was supported by the Program for Guangdong (18) Bartholomew, T. V.; Mey, L.; Arena, J. T.; Siefert, N. S.;
Introducing Innovative and Entrepreneurial Teams Mauter, M. S. Osmotically assisted reverse osmosis for high salinity
(2019ZT08L213) and the Israel-U.S. Collaborative Water- brine treatment. Desalination 2017, 421, 3−11.
Energy Research Center (CoWERC) via the Binational (19) Bartholomew, T. V.; Siefert, N. S.; Mauter, M. S. Cost
Industrial Research and Development Foundation (BIRD) Optimization of Osmotically Assisted Reverse Osmosis. Environ. Sci.
Energy Center grant EC-15. Technol. 2018, 52, 11813−11821.

■
(20) Ahunbay, M. G. Achieving high water recovery at low pressure
REFERENCES in reverse osmosis processes for seawater desalination. Desalination
2019, 465, 58−68.
(1) Hoekstra, A. Y. Water scarcity challenges to business. Nat. Clim. (21) Chong, T. H.; Loo, S.-L.; Krantz, W. B. Energy-efficient reverse
Change 2014, 4, 318−320. osmosis desalination process. J. Membr. Sci. 2015, 473, 177−188.
(2) Vörösmarty, C. J.; McIntyre, P. B.; Gessner, M. O.; Dudgeon,
(22) Chang, H.; Li, T.; Liu, B.; Vidic, R. D.; Elimelech, M.;
D.; Prusevich, A.; Green, P.; Glidden, S.; Bunn, S. E.; Sullivan, C. A.;
Crittenden, J. C. Potential and implemented membrane-based
Liermann, C. R.; Davies, P. M. Global threats to human water security
and river biodiversity. Nature 2010, 467, 555−561. technologies for the treatment and reuse of flowback and produced
(3) Grant, S. B.; Saphores, J.-D.; Feldman, D. L.; Hamilton, A. J.; water from shale gas and oil plays: A review. Desalination 2019, 455,
Fletcher, T. D.; Cook, P. L. M.; Stewardson, M.; Sanders, B. F.; Levin, 34−57.
L. A.; Ambrose, R. F.; Deletic, A.; Brown, R.; Jiang, S. C.; Rosso, D.; (23) Coday, B. D.; Heil, D. M.; Xu, P.; Cath, T. Y. Effects of
Cooper, W. J.; Marusic, I. Taking the ″Waste″ Out of ″Wastewater″ Transmembrane Hydraulic Pressure on Performance of Forward
for Human Water Security and Ecosystem Sustainability. Science Osmosis Membranes. Environ. Sci. Technol. 2013, 47, 2386−2393.
2012, 337, 681−686. (24) Straub, A. P.; Osuji, C. O.; Cath, T. Y.; Elimelech, M.
(4) Urban, J. J. Emerging Scientific and Engineering Opportunities Selectivity and Mass Transfer Limitations in Pressure-Retarded
within the Water-Energy Nexus. Joule 2017, 1, 665−688. Osmosis at High Concentrations and Increased Operating Pressures.
(5) Shaffer, D. L.; Arias Chavez, L. H.; Ben-Sasson, M.; Romero- Environ. Sci. Technol. 2015, 49, 12551−12559.
Vargas Castrillón, S.; Yip, N. Y.; Elimelech, M. Desalination and (25) Wei, Q. J.; McGovern, R. K.; Lienhard, V. J. H. Saving energy
Reuse of High-Salinity Shale Gas Produced Water: Drivers, with an optimized two-stage reverse osmosis system. Environ. Sci.:
Technologies, and Future Directions. Environ. Sci. Technol. 2013, Water Res. Technol. 2017, 3, 659−670.
47, 9569−9583. (26) Davenport, D. M.; Deshmukh, A.; Werber, J. R.; Elimelech, M.
(6) Vidic, R. D.; Brantley, S. L.; Vandenbossche, J. M.; Yoxtheimer, High-Pressure Reverse Osmosis for Energy-Efficient Hypersaline
D.; Abad, J. D. Impact of Shale Gas Development on Regional Water Brine Desalination: Current Status, Design Considerations, and
Quality. Science 2013, 340, 1235009. Research Needs. Environ. Sci. Technol. Lett. 2018, 5, 467−475.
(7) Tong, T.; Elimelech, M. The Global Rise of Zero Liquid (27) Schwantes, R.; Chavan, K.; Winter, D.; Felsmann, C.; Pfafferott,
Discharge for Wastewater Management: Drivers, Technologies, and J. Techno-economic comparison of membrane distillation and MVC
Future Directions. Environ. Sci. Technol. 2016, 50, 6846−6855. in a zero liquid discharge application. Desalination 2018, 428, 50−68.
(8) Zhang, C.; Shi, Y.; Shi, L.; Li, H. X.; Li, R. Y.; Hong, S.; Zhuo, S. (28) Greenlee, L. F.; Lawler, D. F.; Freeman, B. D.; Marrot, B.;
F.; Zhang, T. J.; Wang, P. Designing a next generation solar Moulin, P. Reverse osmosis desalination: Water sources, technology,
crystallizer for real seawater brine treatment with zero liquid and today’s challenges. Water Res. 2009, 43, 2317−2348.
discharge. Nat. Commun. 2021, 12, 998. (29) Liu, X.; Shanbhag, S.; Bartholomew, T. V.; Whitacre, J. F.;
(9) Shi, Y.; Zhang, C.; Li, R.; Zhuo, S.; Jin, Y.; Shi, L.; Hong, S.;
Mauter, M. S. Cost Comparison of Capacitive Deionization and
Chang, J.; Ong, C.; Wang, P. Solar Evaporator with Controlled Salt
Precipitation for Zero Liquid Discharge Desalination. Environ. Sci. Reverse Osmosis for Brackish Water Desalination. ACS ES&T Engg
Technol. 2018, 52, 11822−11830. 2021, 1, 261−273.
(10) Hermsen, N.; Rosenberg, S.; Gorenflo, A. MLD Approach (30) Dudchenko, A. V.; Bartholomew, T. V.; Mauter, M. S. Cost
yields significant opportunity. https://www.watertechonline.com/ optimization of multi-stage gap membrane distillation. J. Membr. Sci.
minimum-liquid-discharge-opportunity/ (accesssed 07, 21). 2021, 627, 119228.
(11) Hotchkies, J. W. Future of Water Resource Recovery. In (31) Haidari, A. H.; Heijman, S. G. J.; van der Meer, W. G. J.
Assessing Global Water Megatrends; Biswas, A., Tortajada, C., Rohner, Optimal design of spacers in reverse osmosis. Sep. Purif. Technol.
P., Eds.; Springer: Singapore, 2018; pp 77−85. 2018, 192, 441−456.
(12) Onishi, V. C.; Carrero-Parreño, A.; Reyes-Labarta, J. A.; Fraga, (32) Rice, D.; Barrios, A. C.; Xiao, Z.; Bogler, A.; Bar-Zeev, E.;
E. S.; Caballero, J. A. Desalination of shale gas produced water: A Perreault, F. Development of anti-biofouling feed spacers to improve
rigorous design approach for zero-liquid discharge evaporation performance of reverse osmosis modules. Water Res. 2018, 145, 599−
systems. J. Cleaner Prod. 2017, 140, 1399−1414. 607.

10722 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723
Environmental Science & Technology pubs.acs.org/est Article

(33) Kim, Y. C.; Elimelech, M. Adverse Impact of Feed Channel Membranes Using Induced Electrophoretic Mixing. Environ. Sci.
Spacers on the Performance of Pressure Retarded Osmosis. Environ. Technol. 2020, 54, 3678−3690.
Sci. Technol. 2012, 46, 4673−4681. (52) Zhao, S.; Liao, Z. P.; Fane, A.; Li, J. S.; Tang, C. Y.; Zheng, C.
(34) She, Q.; Hou, D.; Liu, J.; Tan, K. H.; Tang, C. Y. Effect of feed M.; Lin, J. Y.; Kong, L. X. Engineering antifouling reverse osmosis
spacer induced membrane deformation on the performance of membranes: A review. Desalination 2021, 499, 114857.
pressure retarded osmosis (PRO): Implications for PRO process
operation. J. Membr. Sci. 2013, 445, 170−182.
(35) Togo, N.; Nakagawa, K.; Shintani, T.; Yoshioka, T.; Takahashi,
T.; Kamio, E.; Matsuyama, H. Osmotically Assisted Reverse Osmosis
Utilizing Hollow Fiber Membrane Module for Concentration Process.
Ind. Eng. Chem. Res. 2019, 58, 6721−6729.
(36) Nakagawa, K.; Togo, N.; Takagi, R.; Shintani, T.; Yoshioka, T.;
Kamio, E.; Matsuyama, H. Multistage osmotically assisted reverse
osmosis process for concentrating solutions using hollow fiber
membrane modules. Chem. Eng. Res. Des. 2020, 162, 117−124.
(37) Kakihana, Y.; Jullok, N.; Shibuya, M.; Ikebe, Y.; Higa, M.
Comparison of Pressure-Retarded Osmosis Performance between
Pilot-Scale Cellulose Triacetate Hollow-fiber and Polyamide Spiral-
Wound Membrane Modules. Membranes 2021, 11, 177.
(38) Yip, N. Y.; Elimelech, M. Performance Limiting Effects in
Power Generation from Salinity Gradients by Pressure Retarded
Osmosis. Environ. Sci. Technol. 2011, 45, 10273−10282.
(39) Straub, A. P.; Yip, N. Y.; Elimelech, M. Raising the Bar:
Increased Hydraulic Pressure Allows Unprecedented High Power
Densities in Pressure-Retarded Osmosis. Environ. Sci. Technol. Lett.
2014, 1, 55−59.
(40) Chang Kim, Y.; Min, T. Influence of osmotic mediation on
permeation of water in reverse osmosis: Experimental and numerical
analysis. J. Membr. Sci. 2020, 595, 117574.
(41) Deshmukh, A.; Yip, N. Y.; Lin, S.; Elimelech, M. Desalination
by forward osmosis: Identifying performance limiting parameters
through module-scale modeling. J. Membr. Sci. 2015, 491, 159−167.
(42) Tiraferri, A.; Yip, N. Y.; Straub, A. P.; Romero-Vargas
Castrillon, S.; Elimelech, M. A method for the simultaneous
determination of transport and structural parameters of forward
osmosis membranes. J. Membr. Sci. 2013, 444, 523−538.
(43) Qin, M.; Deshmukh, A.; Epsztein, R.; Patel, S. K.; Owoseni, O.
M.; Walker, W. S.; Elimelech, M. Comparison of energy consumption
in desalination by capacitive deionization and reverse osmosis.
Desalination 2019, 455, 100−114.
(44) Wan, C. F.; Chung, T.-S. Techno-economic evaluation of
various RO plus PRO and RO plus FO integrated processes. Appl.
Energy 2018, 212, 1038−1050.
(45) Atia, A. A.; Yip, N. Y.; Fthenakis, V. Pathways for minimal and
zero liquid discharge with enhanced reverse osmosis technologies:
Module-scale modeling and techno-economic assessment. Desalina-
tion 2021, 509, 115069.
(46) Straub, A. P.; Lin, S.; Elimelech, M. Module-Scale Analysis of
Pressure Retarded Osmosis: Performance Limitations and Implica-
tions for Full-Scale Operation. Environ. Sci. Technol. 2014, 48, 12435−
12444.
(47) Jiang, S.; Li, Y.; Ladewig, B. P. A review of reverse osmosis
membrane fouling and control strategies. Sci. Total Environ. 2017,
595, 567−583.
(48) Matin, A.; Rahman, F.; Shafi, H. Z.; Zubair, S. M. Scaling of
reverse osmosis membranes used in water desalination: Phenomena,
impact, and control; future directions. Desalination 2019, 455, 135−
157.
(49) Horseman, T.; Yin, Y.; Christie, K. S.; Wang, Z.; Tong, T.; Lin,
S. Wetting, Scaling, and Fouling in Membrane Distillation: State-of-
the-Art Insights on Fundamental Mechanisms and Mitigation
Strategies. ACS ES&T Engg 2021, 1, 117−140.
(50) Guan, Y. F.; Boo, C.; Lu, X.; Zhou, X.; Yu, H. Q.; Elimelech, M.
Surface functionalization of reverse osmosis membranes with sulfonic
groups for simultaneous mitigation of silica scaling and organic
fouling. Water Res. 2020, 185, 116203.
(51) Rao, U.; Iddya, A.; Jung, B.; Khor, C. M.; Hendren, Z.; Turchi,
C.; Cath, T.; Hoek, E. M. V.; Ramon, G. Z.; Jassby, D. Mineral Scale
Prevention on Electrically Conducting Membrane Distillation

10723 https://doi.org/10.1021/acs.est.1c01638
Environ. Sci. Technol. 2021, 55, 10714−10723