Desalination 491 (2020) 114582

Contents lists available at ScienceDirect

Desalination
journal homepage: www.elsevier.com/locate/desal

Simulations of BWRO systems under different feedwater characteristics. T

Analysis of operation windows and optimal operating points
⁎
A. Ruiz-Garcíaa, , I. Nuezb, M.D. Carrascosa-Chisvertc, J.J. Santanac
a
Department of Mechanical Engineering, University of Las Palmas de Gran Canaria, Campus Universitario de Tafira, E-35017 Las Palmas de Gran Canaria, Gran Canaria,
Spain
b
Department of Electronic and Automatic Engineering, University of Las Palmas de Gran Canaria, Campus Universitario de Tafira, E-35017 Las Palmas de Gran Canaria,
Gran Canaria, Spain
c
Department of Process Engineering, University of Las Palmas de Gran Canaria, Campus Universitario de Tafira, E-35017 Las Palmas de Gran Canaria, Gran Canaria,
Spain

A R T I C LE I N FO A B S T R A C T

Keywords: Reverse osmosis (RO) is the leading technology for desalinating seawater and brackish water (BW). In general,
Desalination the desalination efficiency of BW is higher than that of seawater due to its lower total dissolved solids content. In
Brackish water addition to membrane fouling, fluctuations in the feedwater source are one of the main problems that affect
Optimal operation desalination systems. These variations can have a significant impact on the operating parameters and efficiency
Reverse osmosis
of BWRO systems. In this work, hydrochemical fluctuations in well groundwater were used to evaluate the
Operation window
different operation windows of two BWRO system configurations, as well as their optimal operating points in
terms of minimum specific energy consumption (SEC) and maximum flow recovery (R). Both configurations
comprised two stages and 6 BWRO spiral wound membrane elements per pressure vessel (PV). One configuration
had 2 and 1 PVs in the first and second stages respectively, and the other 3 and 2 PVs. Both systems were
simulated using a range of feed flow and feed pressure values in addition to the inorganic compositions of 24
feedwater samples taken over a 10-year period. The results showed wider operation windows for the 3:2 than the
2:1 configuration. A common operation window (able to operate with all 24 feedwater samples) with wide
operating margins was obtained for each configuration. For the 2:1 configuration, the 24-sample average SEC,
average maximum SEC and average minimum SEC of this common operation window were 0.760, 1.198 and
0.339 kWh m−3, respectively, while the corresponding R values were 33.47, 67.66 and 19.86%.

1. Introduction to improve the efficiency of RO processes. Work has been carried out,
for example, on improving the pretreatment stage to reduce the fouling
Reverse osmosis (RO) desalination has become a commonly adopted impact on RO membranes [3,4], on the use of RO membranes for brine
solution to deal with water scarcity problems [1]. Generally, RO pro- desalination [5], and on design of new RO membranes [6,7], etc. While
cesses are used to separate solutes from either seawater or brackish the use of energy recovery devices (ERDs) in SWRO systems is key to
water (BW) in order to obtain clean water suitable for, among other reducing specific energy consumption (SEC) [8], these devices are not
uses, human consumption and irrigation [1,2]. These processes can be commonly used in BWRO systems given their lower operating pressures
classified into two groups depending on the salinity of the feedwater. and brine flow rates [9]. Nonetheless, even without ERDs, BWRO is
Desalination plants which employ RO to treat brackish water (BWRO) generally more efficient than SWRO due to the lower TDSf content.
handle feedwaters with a total dissolved solids (TDSf) content that As well as the characteristics of the BWRO membranes [10–12], it is
ranges from 500 mg L−1 to 10,000 mg L−1, whereas the TDSf content of extremely important to take into consideration the characteristics of the
the feedwaters of seawater RO desalination plants (SWRO) tends to be feedwater in the design and operation of BWRO plants. As high flow
around 30,000 mg L−1. The efficiency of the RO process depends on recoveries (R) are achievable, the solubility of sparingly soluble mineral
several factors, including feedwater composition, RO membrane char- salts (e.g. silica (SiO2), calcium carbonate (CaCO3), calcium sulphate
acteristics, RO system design (stages, passes, etc.), and the operational dihydrate (denoted simply as CaSO4), barium sulphate (BaSO4), stron-
parameters. Significant efforts have been made in the last two decades tium sulphate (SrSO4) and calcium fluoride (CaF2)) could be exceeded.

⁎
Corresponding author.
E-mail addresses: [email protected] (A. Ruiz-García), [email protected] (I. Nuez), [email protected] (J.J. Santana).

https://doi.org/10.1016/j.desal.2020.114582
Received 10 April 2020; Received in revised form 10 May 2020; Accepted 4 June 2020
Available online 16 June 2020
0011-9164/ © 2020 Elsevier B.V. All rights reserved.
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Nomenclature T temperature (∘C)

TCF temperature correction factor
Acronyms TDS total dissolved solids (mg L−1)
Y fraction recovery
BWRO Brackish water reverse osmosis
ERD energy recovery device Greek letters
PV pressure vessel
RO reverse osmosis α water price (€ m−3)
SWRO seawater reverse osmosis β energy consumption price (€ kWh−1)
A average water permeability coefficient (m d−1 kg−1 cm2) Δπ osmotic pressure gradient (kg cm−2)
B average ion permeability coefficient (m d−1) Δp pressure drop (kg cm−2)
C concentration (mg L−1) π osmotic pressure (kg cm−2)
FF flow factor
J flux per unit area (m3 m−2 s−1 or kg m−2 s−1) Subscripts
Ksp solubility product
m molal concentration (mol kg−1) b brine
NDP net driven pressure (kg cm−2) f feed
p pressure (kg cm−2) i membrane element i
PF polarization factor j ion j
PR profit (€ h−1) max maximum
Q flow (m3 d−1) min minimum
R flow recovery (%) m membrane
Sm membrane surface (m2) p permeate
SDI silt density index s solute
SEC specific energy consumption (kW h m−3)

This can cause scaling, one of the biggest membrane fouling problems rates of 16, 12, 6 and 3 g L−1 were also considered. The objective
in BWRO desalination [13,14]. To counter this problem, antiscalant has function was formulated as a mixed integer nonlinear programming for
to be used in the pretreatment or intermediate demineralization treat- minimizing the total cost or energy consumption subject to thermo-
ments [15] to increase R [16]. Usually, TDSf and temperature (T) var- dynamic, technical and flexibility constraints. They established that RO
iations are more common in brackish water than seawater feedwater systems with 2 stages were the best choice when TDSf content was
sources [17]. Such variations, when they occur, complicate the efficient below 28 g L−1. The calculated BWRO systems for TDSf from 3 to 16 g
operation of BWRO systems [18–20]. Yan-Yue Lu et al. [21] studied the L−1 had ERDs and interstage pumps with different arrangements for
optimum design of RO systems under different TDSf and product spe- each TDSf. A sensitivity analysis was also carried out, but only for a
cifications. They focused on optimizing an equation related to the ca- SWRO case. Mingheng Li conducted a further work on the optimization
pital and operating costs, and considered the incorporation of ERDs and of SEC in RO systems [29]. This time, a unified mathematical model to
interstage pumps in the BWRO system design. Different TDSf values (16, describe both BWRO and SWRO desalination was developed. SEC was
12, 6 and 3 g L−1 NaCl) were used. Their analysis involved use of the optimized considering different RO configurations, R, ERDs and inter-
FilmtecTMBW30-400 membrane. They obtained different BWRO con- stage pumps. M.A. Al-Obaidi et al. [30] carried out a performance
figurations (stages, interstage pumps, number of pressure vessels (PVs), analysis of a full-scale BWRO plant with two passes. The impact of
BWRO elements per PV) for each TDSf, and all the BWRO systems had variations in TDSf (1.098–1.318 g L−1), Qf (74–88.8 m3 h−1), pf
an ERD. K. M. Sassi and I. M. Mujtaba [22] studied the optimal design (1–1.5 MPa) and T (25–30 ∘C) on permeate concentration and R was
and operation of a BWRO system considering membrane fouling. A si- evaluated. The impact was assessed separately in that when one para-
mulation was carried out to validate the code with experimental data meter was varied the rest were considered constant in the existing
from another published work [23], and a sensitivity analysis was also BWRO system. In a later work, and using the same full-scale BWRO
performed. They also used the FilmtecTMBW30–400 membrane. The plant, A. A. Alsarayreh et al. [31] undertook a simulation-based per-
BWRO system had two stages, each with two PVs of three membrane formance evaluation which considered different recycled ratios of
elements. The sensitivity analysis that they performed considered a feed brine. They concluded that it was possible to increase product capacity
pressure (pf) from 0.6 to 2.5 MPa, increasing the number of membrane by around 3% with 100% recycle percentage of the high salinity brine
elements per PV (up to 7), TDSf (2.5 and 5 g L−1) and feed spacers. The stream. The same research group continued their work with the same
feed flow (Qf) was considered constant (20.4 m3 h−1), and the fouling plant, evaluating and minimizing the energy consumption [32]. They
effect was evaluated through a previously proposed model for water found it was possible to reduce total energy consumption by between
permeability coefficient decline [24]. Their analysis was done without 47% and 53.8% when compared with the calculations for the original
limiting R due to scaling or varying the Qf. M. Li and B. Noh [25] design without an ERD. Y. D. Ahdab et al. [33] carried out a compre-
studied the validation of a model-based optimization of BWRO opera- hensive investigation of brackish groundwater composition (in the
tion. They carried out their study on the basis of constant Qf (350 m3 United States) in relation to minimum desalination energy costs. They
h−1) and TDSf (0.95 g L−1). As the BWRO plant was already built, the consider the desalination unit as a black box with inputs and outputs.
study focused on its optimization. The same plant was used by M. Li They based the study on calculation of least work of separation which
[26] in a later work to optimize the SEC, as well as in another study represents a baseline for SEC of desalination systems.
related to the use of multitrains that allow different operating condi- Most of the aforementioned works consider BWRO systems with
tions [27]. Y. Du et al. [28] analyzed the optimization of an RO systems ERDs, which is not usual. The use of interstage pumps is also uncertain
network involving both SWRO and BWRO. The analysis included use of given the low efficiency of these pumps and the added complexity in
the FilmtecTMBW30–400 membrane, as well as ERD and interstage terms of operation and maintenance due to variations in the operating
pumps. Constant T (20 ∘C) and permeate flow (120 m3 h−1), and TDSf parameters. Most of these BWRO system configurations were designed

2
A. Ruiz-García, et al. Desalination 491 (2020) 114582

considering constant TDSf and/or Qf, but variation of these parameters 34 m3 h−1 [36]. With these flow rates and the aforementioned R, the
can result in the BWRO system being unable to operate within the limits use of ERDs would be unusual. The installation of interstage pumps
recommended by the membrane manufacturer or far from the optimal implies one more element to be considered in terms of operation and
operating point. This paper aims to evaluate different operation win- maintenance. While their deployment may be interesting from a theo-
dows (ranges of Qf and pf) of 2-stage BWRO systems under variable TDSf retical point of view, from a practical perspective their use would be
and T. For this purpose, a total of 24 well groundwater samples taken complex under variable operating conditions and consideration also
over the course of a 10 year period were analyzed and used, and a needs to be given to the generally low efficiency of these pumps. Even
computational tool for the design and simulation of BWRO systems was without incorporating ERDs and interstage pumps, the number of
employed which has been previously published and validated by the possible solutions remains high in terms of the number of PVs per stage
authors [34]. and the number of BWRO membrane elements per PV. With respect to
the aforementioned Qf, it seemed coherent to consider 3:2 (3 PVs in the
2. Material and methods first stage and 2 in the second stage) and 2:1 configurations. The
number of simulations needed to evaluate the operation windows
2.1. Feedwater (range of possible pf and Qf) for the 24 samples (Table 1) would be
numerous. Installing too many BWRO membrane elements in series
The groundwater well is located in the south of Gran Canaria could cause operation and maintenance problems as the membranes get
(Canary Islands, Spain). Its coordinates are latitude 27∘50′52.04” N, fouled, and the pressure drop limit recommended by the membrane
longitude 15∘29′00.20”W, and an elevation of 160 m above mean sea manufacturer (0.345 MPa) in a stage could be exceeded. The last sim-
level. The hydrochemistry of this well has previously been described by plification was to consider 6 BWRO membrane elements per PV in both
the authors, as well as how the samples were taken and processed for stages, i.e. 3(6):2(6) and 2(6):1(6). As the algorithm used was pre-
the determination of the different parameters [17]. The samples were viously validated with four full-scale BWRO desalination plants with
taken over the course of a 10 year period, two or three analyses per the FilmtecTMBW30-400 membrane element [34], this was the element
year. Table 1 shows the pH, T and inorganic composition of the selected for the simulations.
groundwater. The TDS considered were the sum of the analyzed ions.
The highest TDS were found in sample 11 (5815.20 mg L−1) and the 2.3. Calculation algorithm
lowest in sample 20 (1218.79 mg L−1). The silt density index (SDI) was
assumed to be between 2 and 3, as is usual for this type of water after a 2.3.1. Estimation of maximum R
microfiltration stage [35]. One of the main limiting factors when designing BWRO desalination
plants is the maximum R (Rmax) for scale prevention. This value de-
2.2. BWRO systems considered pends on the feedwater characteristics and the antiscalant used in the
pretreatment. The antiscalants considered and their theoretical power
The number of possible BWRO configurations to desalinate parti- of inhibition of measured salts in the samples are shown in Table 2. The
cular types of brackish water depends on the desired operating range Rmax for each antiscalant was calculated according to a previously
(SEC, R, permeate concentration). Usually, an RO system with two published algorithm [16]. The highest value was selected as a con-
stages is appropriate for BW desalination, as has been demonstrated in straint in the simulation algorithm. It should be noted that the calcu-
previous works [28,34]. Even if we consider only two stages, the lated values are theoretical. In practice, it is not usual to reach the limit
number of possible configurations is still high. Another factor to be but to leave a safety margin.
considered is the potential use of ERDs. Normally, the R ratios are be-
tween 60 and 85% and the Qf is limited by the source. In Gran Canaria 2.3.2. Process modeling
island, the average Qf of BWRO desalination plants is between 29 and The solution-diffusion [1,37] transport model was implemented in

Table 1
Feed water inorganic composition in mg L−1.
Sample pH HCO3− Cl− SO4= NO3− Na+ K+ Ca2+ Mg2+ Fe2+ SiO2 TDS T

1 7.05 175 2620 165 7.9 400 29 474 475 0.6 60.5 4407.00 25.20
2 6.94 155 2500 180 5.3 422 38 532 383 0.2 36 4251.50 25.60
3 7.37 175 2650 168 5.3 450 30 561 406 0.15 55.7 4501.15 25.00
4 7.48 100 2420 192 5.7 307 30 512 409 0.11 53.4 4029.21 25.10
5 7.24 122 1715 150 6.9 257 28 368 292 0.09 52.5 2991.49 24.90
6 7.27 216 2230 323 9.6 369 30 480 399 0.1 65 4121.70 25.50
7 7.05 190 3180 306 3 458 17 783 483 0.35 57 5477.35 25.30
8 7.05 167 2418 175 6 451 29 660 264 0.5 57 4227.50 25.00
9 7.03 92 2680 166 4.3 339 35 605 428 0.13 30 4379.43 24.80
10 7.46 287 2684 196 1 720 48 432 370 0.09 59 4797.09 25.40
11 7.10 304 3362 180 2.2 830 85 566 453 1 32 5815.20 25.70
12 7.10 305 3360 180 2 828 83 570 450 0.8 30 5808.80 25.00
13 7.40 184 2420 182 1.4 324 32 570 390 0.27 56.3 4159.97 25.20
14 7.10 185 2872 200 5 383 28.9 594 505 0.48 55.8 4829.18 25.10
15 7.80 155 2610 209 0.5 399 38.1 645 405 0.21 92.9 4554.71 25.00
16 7.40 152 2966 273 12.9 469 33.6 600 504 0.19 54.5 5065.19 24.80
17 6.90 260 3023 218 5 425 36.9 632 552 0.08 54.7 5206.68 24.60
18 7.70 173 2930 253 6.8 410 17 620 504 0.14 46 4959.94 25.70
19 7.00 170 2758 232 7.2 363 38.2 669 546 0.15 50.8 4834.35 25.50
20 7.60 215 484 85.6 13.6 208 11.7 76.8 78.1 0.094 45.9 1218.79 25.00
21 8.04 193 1831 150 8.36 468 22.5 395 323 0.17 52.1 3443.13 25.50
22 8.19 197 1715 148 8.1 622 28.6 423 308 0.11 34.2 3484.01 25.20
23 7.72 227 654 108 6.73 214 15.6 120 88.5 1.21 66.6 1501.64 25.10
24 7.58 196 2259 152 74.8 354 35.1 465 418 1.22 62.3 4017.42 25.00

3
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Table 2 1 1
TCF = exp ⎡3,020⋅⎛ − ⎞⎤
Theoretical power of inhibition. ⎢ ⎝ 298 273 + T ⎠ ⎥ (11)
⎣ ⎦
Salt Antiscalant
FF (Flow factor) is the coefficient used for considering operating
Genesys LF Genesys SI Vitec 3000 Vitec 4000 time and fouling. A value of 1 was used to estimate the performance as
new FilmtecTMBW30-400 elements were considered. Next in the de-
CaCO3 LSI ≤ 2.6 LSI ≤ 2.4 LSI ≤ 3 LSI ≤ 2.5 velopment of Eq. (1) is the expression of the NDP, which depends on pfi,
CaSO4 3.5⋅Ksp 3.5⋅Ksp 3.5⋅Ksp 3.5⋅Ksp
Δpfbi, ppi, πmi and the average osmotic pressure of the permeate (πpi):
SiO2 210 mg L−1 375 mg L−1 120 mg L−1 2⋅saturation
Δp fbi
NDP = (Δp − Δπ ) = p fi − − ppi − πmi + πpi
2 (12)
the algorithm [34] as it usually provides results close to the real be-
havior of these systems. The transport equations used mean membrane Δpfbi depends on two parameters obtained experimentally (a and b)
element values, and the pressure drop in the permeate as well as T and the average feed-brine flow:
changes along the RO systems were disregarded. The calculation al-
Q f + Q ri ⎞ b ⎞
gorithm considers some simplifications that are detailed in a previous Δp fbi = ⎜⎛a⋅⎛ i ⎟⋅0.07
⎝ ⎝ 2 ⎠ ⎠ (13)
work [34]. It additionally allows the use of various equations for de-
termination of the polarization factor (PFi) and the pressure drop along Eq. (13) is multiplied by 0.07 due to unit conversion, from psi to kg
the BWRO membrane elements in the feed-brine side (Δpfbi). The fol- cm−2. In order to calculate πpi, the average ionic permeability coeffi-
lowing equations were used: cient (B) has to be multiplied by TCF (Eq. (14)), since B depends on T.
Qp = A⋅(Δp − Δπ )⋅Sm (1) This enables calculation of the ion concentration of the permeate (Cpj):

where Qp is the permeate flow, A is the water permeability coefficient Sm ⎛ Cf j⋅(1 + FCi ) ⎞
Cpj = Bj⋅PFi⋅TCF ⋅ ⋅⎜ ⎟
of the membrane, (Δp − Δπ) is the net driven pressure (NDP) and Sm is Q pi ⎝ 2 ⎠ (14)
the membrane area.
Once Cpj is obtained, πpi is calculated according to Eq. (5). First, Qpi
Solute transport equation:
is calculated, and with that, Yi, and then the result is compared to the
Qs = B⋅ΔC⋅Sm (2) estimation. Therefore, the function to minimize is the difference be-
tween both element recoveries. Knowing the interval of the variable
where Qs is the solute flow through the membrane, B is the solute
(0–1), the MATLAB® optimizing function [40,41] was used to find the
permeability coefficient of the membrane, and ΔC is the concentration
solution.
gradient of solute on either side of the membrane.
The coefficient A (Eq. (1)) depends on three variables: osmotic
2.4. Simulation considerations
pressure on the membrane surface (πmi), T, and the flow factor (FF)
related to fouling and operating time [38].
The constraints considered were those established by the membrane
A = A (A0 , πmi )⋅TCF ⋅FF (3) manufacturer for the FilmtecTMBW30-400 element: minimum rejection
flow (Qr−min=3 m3 h−1), maximum permeate flow (Qp−max=1.43 m3
where A0 is the initial water permeability coefficient, and TCF is the h−1), maximum feed flow (Qa−max=17 m3 h−1) and maximum re-
temperature correction factor. covery per membrane element (Rmax−i=19%), all according to the
Cfbi origin and SDI of the feedwater (well water and SDI < 3) [38]. The feed
πmi = πfi⋅ ⋅PFi pressure range considered was between 7 and 20 bar in steps of 0.5 bar,
Cfi (4)
and the feed flow range between Qr−min and Qa−max in steps of 10 m3
where πfi is the osmotic pressure of the feedwater, and PFi is the con- d−1. The SEC was calculated considering a 100% efficiency of the high
centration polarization factor of the membrane element i. pressure pump and electrical engine. A profit simulation was also car-
ried out assuming an energy consumption cost range of 0.09–0.16 €
πfi = 0.0787⋅(273 + T )⋅Σm j (5) kWh−1 (β) and a water sale price range of 0.2–0.55 € m−1 (α). These
ranges were determined in accordance with established prices in the
Cmi = Cfbi⋅PFi (6)
Canary archipelago. A simple profit (PR) function was used for the
bi C assessment (Eq. (15)).
⎛ 1 + Cfi ⎞
Cfbi = Cfi⋅⎜ PR = Qp⋅(α − β⋅SEC ) (15)
⎜ 2 ⎟⎟
⎝ ⎠ (7)

PFi = e0.7 ⋅ Yi (8) 3. Results and discussion

Yi is the element recovery as a fraction. An additional module was Table 3 shows the Rmax obtained for each antiscalant. The limiting
developed to calculate PFi using Eq. (9). compounds were CaCO3 and SiO2. In most cases, the highest values
Cm were obtained with the Vitec 4000 (specific silica inhibitor) antiscalant.
PFi = The range of Rmax for each antiscalant along the 24 samples is quite
Cfb (9)
pronounced; from 21.25 to 76.47, from 7.25 to 76.47, from 31.82 to
76.47 and from 70.81 to 89.07 for the Genesys LF, Genesys SI, Vitec
• The TCF is calculated as follows [39]: 3000 and Vitec 4000 antiscalants, respectively.
Figs. 1 and 2 show the ideal SEC in the operation windows for both
If T ⩾ 25∘C: BWRO systems considered with sample 1. A wider operation window
1 1 can be seen for the 3:2 than the 2:1 configuration. This is because the
TCF = exp ⎡2,640⋅⎛ − ⎞⎤ 3:2 configuration has one more PV in the second stage, allowing the
⎢
⎣ ⎝ 298 273 + T ⎠ ⎥
⎦ (10)
system to work with higher Qf per PV in the first stage without ex-
∘
If T ⩽ 25 C: ceeding the constraints established by the RO membrane manufacturer.

4
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Table 3
Rmax for each antiscalant.
Sample Genesys LF Genesys SI Vitec 3000 Vitec 4000

1 52.70 52.70 52.70 76.35

2 72.04 72.04 72.04 86.02
3 56.31 56.31 56.31 80.48
4 59.37 59.37 59.37 82.53
5 58.76 58.76 58.76 80.59
6 49.44 49.44 49.44 75.53
7 55.51 55.51 55.51 77.76
8 55.29 55.29 55.29 77.65
9 76.39 76.39 76.39 89.07
10 54.03 54.03 54.03 79.16
11 75.19 75.19 75.19 84.78
12 76.47 76.47 76.47 84.48
13 55.99 55.99 55.99 80.60
Fig. 3. R in the operation window for the configuration 2(6):1(6) and sample 1.
14 56.31 56.31 56.31 78.15
15 31.82 31.82 31.82 70.81
16 57.11 57.11 57.11 81.21
17 56.81 56.81 56.81 78.41
18 70.00 70.00 70.00 79.30
19 60.48 60.48 60.48 80.24
20 68.05 68.05 68.05 82.40
21 32.86 20.86 70.86 73.84
22 21.25 7.25 42.25 72.53
23 52.75 52.75 52.75 73.84
24 53.24 53.24 53.24 78.93

Fig. 4. R in the operation window for the configuration 3(6):2(6) and sample 1.

If we compare Figs. 1 and 2 with Figs. 3 and 4, which show the

corresponding R values, it can be seen that the areas with low SEC
values do not necessarily correspond to areas with high R values, and
that a high R value does not always imply a high SEC value. The highest
R are found in the interval between 1 and 1.5 MPa and, in terms of Qf
per PV in the first stage, in the interval between 7 and 12 m3 h−1.
Relatively high R can be achieved with the ideal SEC of between ap-
Fig. 1. SEC in the operation window for the configuration 2(6):1(6) and sample proximately 0.5 and 0.6 kWh m−3 for both configurations and this
1. inorganic composition (sample 1). It should be mentioned that the R
adopted depends on the antiscalant used. Three of the antiscalant
products considered would allow operation with R values of up to 52%.
Only the Vitec 4000 would allow operation with R close to 72%.
However, the highest R is not necessarily the best option, as specific
silica antiscalant usually costs around twice as much as other anti-
scalants [42].
Figs. 5 and 6 show the trend of TDSp in the operation window for
both BWRO systems. In general, TDSp decreases as pf increases. Higher
TDSp values are found at low pressure for the entire range of Qf per PV
in the first stage. This trend is in accordance with the data provided by
the membrane manufacturer (Dupont®), calculated using its WAVE
software program.
Feedwater fluctuations in terms of T, inorganic composition and
TDSf can have a significant impact on the operation window of a BWRO
system. The effect of TDS can be seen if we compare the SEC values
shown in Figs. 1 and 2 for sample 1 (with a TDSf value of 1.2 g L−1)
Fig. 2. SEC in the operation window for the configuration 3(6):2(6) and sample
1. with those shown in Figs. 7 and 8 for sample 20 (with a TDSf value of
4.4 g L−1). It can be seen how the operation window is smaller and SEC
is lower for sample 20 in both configurations. The lower TDSf of sample
As can be seen, the color distribution is more or less similar, with the
20 translates into higher Qp in the first stage in comparison with sample
lowest SEC values in both figures found in the interval between 0.7 and
1, making it more likely that the constraints are not met in terms of
1.2 MPa and, in terms of Qf per PV in the first stage, in the interval
Qr−min and Rmax−i.
between 5 and 8 m3 h−1. It is possible to obtain minimum SEC values
With the characteristics of sample 20, some operating points are
below 0.5 kWh m−3 with both configurations.
reached that would be not appropriate with other samples according to

5
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Fig. 5. TDSp in the operation window for the configuration 2(6):1(6) and Fig. 8. SEC in the operation window for the configuration 3(6):2(6) and sample
sample 1. 20.

Fig. 6. TDSp in the operation window for the configuration 3(6):2(6) and Fig. 9. Pump curves and average SEC in the common operation window for the
sample 1. configuration 2(6):1(6) and considering the 24 samples.

Fig. 7. SEC in the operation window for the configuration 2(6):1(6) and sample Fig. 10. Pump curves and average R in the common operation window for the
20. configuration 2(6):1(6) and considering the 24 samples.

the constraints imposed by the membrane manufacturer. Figs. 9 and 10 the first stage, a larger pump (Ebara EVMS45 6-0F5/22) was required to
show respectively the average SEC and R in the common operation operate with this configuration. The values obtained are noticeably
window for all 24 samples considering the 2:1 configuration. The Ebara different to those obtained with the 2:1 configuration. For the 3:2
EVMS20 16/18.5 pump was selected to show the possible operating configuration, the 24-sample average SEC, average maximum SEC and
curves when working with a variable frequency drive. The pump was average minimum SEC of this common operation window were 0.933,
oversized to take into account future effects of fouling on the BWRO 1.494 and 0.410 kWh m−3, respectively, while the corresponding R
systems. For the 2:1 configuration, the 24-sample average SEC, average values were 30.52, 61.52 and 18.17%.
maximum SEC and average minimum SEC of this common operation Tables 4 and 5 show the operating points that correspond to the
window were 0.760, 1.198 and 0.339 kWh m−3, respectively, while the minimum SEC value for each of the 24 samples with the 2:1 and 3:2
corresponding R values were 33.47, 67.66 and 19.86%. Figs. 11 and 12 configurations, respectively. For the 2:1 configuration (Table 4), pf
show respectively the average SEC and R in the common operation ranges between 0.74 and 0.98 MPa and shows a general tendency to
window for all 24 samples considering the 3:2 configuration. In this increase with TDSf, while Qf ranges between 5.08 and 7.17 m3 h−1 in
case, the operating window was wider and, as 3 PVs were deployed in the first stage and per PV and shows a general tendency to increase as

6
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Table 5
Operating points of minimum SEC for configuration 3(6):2(6).
Sample pf Qf R TDSp SEC
(MPa) (m3 h−1) (%) (mg L−1) (kWh m−3)

1 0.83 5.50 54.40 145.62 0.426

2 0.74 5.50 47.41 187.04 0.410
3 0.83 5.50 53.51 149.48 0.433
4 0.78 5.50 54.37 145.79 0.401
5 0.74 5.92 60.78 116.24 0.336
6 0.78 5.50 55.06 147.48 0.396
7 0.93 5.50 52.49 149.93 0.493
8 0.74 5.08 50.56 179.39 0.404
9 0.83 5.50 54.67 141.97 0.424
10 0.88 5.50 53.85 154.07 0.455
11 0.98 5.50 51.36 51.36 0.530
12 0.98 5.50 51.17 163.54 0.532
Fig. 11. Pump curves and average SEC in the common operation window for 13 0.78 5.50 53.75 149.33 0.405
the configuration 3(6):2(6) and considering the 24 samples. 14 0.88 5.50 54.31 143.31 0.451
15 0.83 5.50 54.63 144.19 0.424
16 0.88 5.50 51.98 152.92 0.472
17 0.83 5.08 49.21 170.08 0.471
18 0.88 5.50 53.23 149.82 0.461
19 0.83 5.50 53.79 147.22 0.430
20 0.69 7.58 71.22 56.10 0.268
21 0.69 5.50 55.58 148.18 0.343
22 0.74 5.92 60.91 117.79 0.335
23 0.69 7.17 69.32 69.33 0.275
24 0.78 5.50 55.98 161.47 0.389

Table 6
Operating points of Rmax for configuration 2(6):1(6).
Sample pf Qf SEC TDSp R
(MPa) (m3 h−1) (kWh m−3) (mg L−1) (%)

1 1.52 8.83 0.564 40.99 74.82

Fig. 12. Pump curves and average R in the common operation window for the 2 1.47 8.83 0.547 41.70 74.72
configuration 3(6):2(6) and considering the 24 samples. 3 1.52 8.83 0.569 41.52 74.22
4 1.47 9.25 0.545 37.83 74.94
5 1.37 9.67 0.482 30.84 79.06
Table 4 6 1.47 9.25 0.542 38.73 75.44
Operating points of minimum SEC for configuration 2(6):1(6). 7 1.62 8.83 0.629 45.17 71.41
8 1.47 8.83 0.545 40.99 75.04
Sample pf Qf R TDSp SEC
9 1.52 8.83 0.563 40.01 74.95
(MPa) (m3 h−1) (%) (mg L−1) (kWh m−3)
10 1.57 8.83 0.594 45.22 73.33
1 0.83 5.50 51.95 71.01 0.446 11 1.67 8.42 0.656 52.79 70.61
2 0.74 5.50 44.68 84.86 0.432 12 1.72 8.42 0.665 50.72 71.74
3 0.83 5.50 51.04 72.99 0.454 13 1.47 9.25 0.548 38.44 74.55
4 0.78 5.50 51.74 68.51 0.421 14 1.57 8.83 0.592 42.25 73.64
5 0.74 5.92 57.63 51.95 0.355 15 1.32 8.00 0.521 44.36 70.53
6 0.78 5.50 52.46 69.21 0.415 16 1.62 8.83 0.612 43.27 73.47
7 0.93 5.50 50.31 77.71 0.514 17 1.62 8.83 0.616 42.64 72.94
8 0.74 5.08 47.99 80.77 0.426 18 1.57 8.83 0.597 43.70 73.00
9 0.83 5.50 52.19 69.22 0.444 19 1.52 8.83 0.567 41.18 74.48
10 0.88 5.50 51.54 77.58 0.476 20 1.13 10.92 0.383 15.74 81.72
11 0.98 5.50 49.28 49.28 0.553 21 1.18 8.42 0.445 36.91 73.42
12 0.98 5.50 49.05 86.98 0.555 22 1.32 10.92 0.508 29.03 72.35
13 0.78 5.50 51.12 70.25 0.426 23 1.23 12.58 0.462 16.36 73.70
14 0.88 5.50 52.01 72.15 0.471 24 1.47 9.25 0.538 42.79 75.94
15 0.83 5.50 52.17 70.30 0.444
16 0.88 5.50 49.64 77.22 0.494
17 0.93 5.50 52.68 71.34 0.491 are quite similar to those for the 3:2 configuration in terms of pf
18 0.88 5.50 50.95 75.46 0.481 (0.69–0.98 MPa), Qf (5.08–7.58 m3 h−1), R (47.41–71.22%), TDSp
19 0.83 5.50 51.36 71.82 0.451
(51.36–187.04 mg L−1) and SEC (0.27–0.53 kWh m−3). The R in the
20 0.74 7.17 75.69 22.97 0.270
21 0.74 5.92 54.89 57.82 0.372 first stage was in a similar interval for both configurations of between
22 0.74 5.92 57.79 52.61 0.354 42.64 and 50.19% (data not shown).
23 0.74 7.17 70.28 27.52 0.291 Tables 6 and 7 show the operating points that correspond to the
24 0.78 5.50 53.34 75.75 0.409 maximum Rmax for the two BWRO configurations considered. In this
case, the ranges of pf for both configurations are higher, 1.13–1.72 and
0.88–1.72 MPa for the 2:1 and 3:2 configuration, respectively. The Qf
TDSf decreases. R (ranging between 44.68 and 75.69%) shows the same
ranges are also found to be higher for the operating points of Rmax than
general tendency with respect to TDSf as Qf. TDSp ranges between 22.97
for those of minimum SEC. The TDSp are lower and the ranges of SEC
and 86.98 mg L−1, and minimum SEC between 0.27 and 0.56 kWh
higher, from 0.38 to 0.67 and 0.33–0.65 kWh m−3 for the 2:1 and 3:2
m−3. SEC shows a tendency to decrease with TDSf. The results obtained
configurations, respectively. The highest Rmax value was obtained for

7
A. Ruiz-García, et al. Desalination 491 (2020) 114582

Table 7 CRediT authorship contribution statement

Operating points of Rmax for configuration 3(6):2(6).
Sample pf Qf SEC TDSp R (%) A. Ruiz-García:Conceptualization, Methodology, Writing - ori-
(MPa) (m3 h−1) (kWh m−3) (mg L−1) ginal draft, Visualization, Investigation, Validation, Writing - review
& editing.I. Nuez:Conceptualization, Resources, Supervision.M.D.
1 1.52 8.83 0.553 68.42 76.33
Carrascosa-Chisvert:Data curation, Investigation, Formal
2 1.47 8.83 0.536 70.20 76.29
3 1.52 8.83 0.557 69.25 75.78
analysis.J.J. Santana:Data curation, Investigation, Supervision,
4 1.47 9.25 0.533 63.74 76.69 Formal analysis.
5 1.37 10.08 0.477 51.69 79.97
6 1.47 10.08 0.543 62.01 75.21 Declaration of competing interest
7 1.62 8.83 0.616 73.87 72.96
8 1.47 8.83 0.533 69.05 76.64
9 1.52 8.83 0.552 66.82 76.48 The authors declare that they have no known competing financial
10 1.57 8.83 0.582 74.69 74.85 interests or personal relationships that could have appeared to influ-
11 1.67 8.42 0.643 85.63 72.03
ence the work reported in this paper.
12 1.72 8.42 0.652 81.86 73.10
13 1.47 9.25 0.535 64.73 76.32
14 1.57 8.83 0.580 69.84 75.15 References
15 1.47 10.50 0.577 63.31 70.76
16 1.62 8.83 0.600 70.97 74.94 [1] M. Qasim, M. Badrelzaman, N.N. Darwish, N.A. Darwish, N. Hilal, Reverse osmosis
17 1.62 8.83 0.604 69.89 74.45 desalination: a state-of-the-art review, Desalination 459 (2019) 59–104, https://
18 1.57 8.83 0.585 72.16 74.50 doi.org/10.1016/j.desal.2019.02.008 URL http://www.sciencedirect.com/science/
19 1.52 8.83 0.556 68.71 75.97 article/pii/S0011916418325037.
20 1.18 12.17 0.397 25.92 82.39 [2] S. Burn, M. Hoang, D. Zarzo, F. Olewniak, E. Campos, B. Bolto, O. Barron,
21 1.23 9.67 0.462 60.02 73.64 Desalination techniques—a review of the opportunities for desalination in agri-
22 1.23 10.50 0.470 53.80 72.41 culture, Desalination 364 (2015) 2–16, https://doi.org/10.1016/j.desal.2015.01.
23 0.88 9.25 0.333 45.21 73.56 041 desalination for Agriculture, URL http://www.sciencedirect.com/science/
24 1.47 9.25 0.526 72.21 77.64 article/pii/S0011916415000600.
[3] S.F. Anis, R. Hashaikeh, N. Hilal, Reverse osmosis pretreatment technologies and
future trends: a comprehensive review, Desalination 452 (2019) 159–195, https://
doi.org/10.1016/j.desal.2018.11.006 URL http://www.sciencedirect.com/science/
sample 20 and the lowest for sample 15, reflecting the different in- article/pii/S0011916418320411.
organic compositions of the samples (Tables 1 and 3). [4] J. Kavitha, M. Rajalakshmi, A.R. Phani, M. Padaki, Pretreatment processes for
seawater reverse osmosis desalination systems-a review, J. Water Process Eng. 32
The profit analysis that was made of the common operating win- (2019) 100926, https://doi.org/10.1016/j.jwpe.2019.100926 URL http://www.
dows showed that the most profitable operating points were those with sciencedirect.com/science/article/pii/S2214714419303277.
the highest R, pf = 1.03 MPa and Qf = 10.5 m3 h−1 for the 2:1 con- [5] D.M. Davenport, A. Deshmukh, J.R. Werber, M. Elimelech, High-pressure reverse
figuration, and pf = 1.18 MPa and Qf = 13 m3 h−1 for the 3:2 config- osmosis for energy-efficient hypersaline brine desalination: current status, design
considerations, and research needs, Environ. Sci. Technol. Lett. 5 (8) (2018)
uration. The weight of α was higher than β as the Canary Archipelago is 467–475, https://doi.org/10.1021/acs.estlett.8b00274 (URL doi:10.1021/acs.
a region with subsidized electrical energy costs and desalinated water estlett.8b00274).
has a high value due to water scarcity in the region. The calculated [6] D.L. Zhao, S. Japip, Y. Zhang, M. Weber, C. Maletzko, T.-S. Chung, Emerging thin-
film nanocomposite (TFN) membranes for reverse osmosis: a review, Water Res.
profit values ranged between 0.79 and 4.4 € h−1 and between 1.33 and 173 (2020) 115557, , https://doi.org/10.1016/j.watres.2020.115557 URL http://
8.36 € h−1 for the 2:1 and 3:2 configurations, respectively. It should be www.sciencedirect.com/science/article/pii/S0043135420300932.
noted that the corresponding profit values are affected by the fact that [7] H. Saleem, S.J. Zaidi, Nanoparticles in reverse osmosis membranes for desalination:
a state of the art review, Desalination 475 (2020) 114171, , https://doi.org/10.
Qp is higher for the 3:2 than the 2:1 configuration due to its higher Qf. It 1016/j.desal.2019.114171 URL http://www.sciencedirect.com/science/article/
should be mentioned that the capital costs would be higher for the pii/S0011916419312974.
configuration 3:2 as more PVs and membrane elements would be in- [8] S.A. Urrea, F.D. Reyes, B.P. Suárez, J.A. de la Fuente Bencomo, Technical review,
evaluation and efficiency of energy recovery devices installed in the Canary Islands
stalled. desalination plants, Desalination 450 (2019) 54–63, https://doi.org/10.1016/j.
desal.2018.07.013 URL http://www.sciencedirect.com/science/article/pii/
S0011916417314510.
[9] A. Drak, M. Adato, Energy recovery consideration in brackish water desalination,
4. Conclusions Desalination 339 (2014) 34–39, https://doi.org/10.1016/j.desal.2014.02.008 URL
http://www.sciencedirect.com/science/article/pii/S001191641400071X.
In this study, the operation windows of two BWRO system config- [10] Y. Okamoto, J.H. Lienhard, How RO membrane permeability and other perfor-
mance factors affect process cost and energy use: a review, Desalination 470 (2019)
urations (2(6):1(6) and 3(6):2(6)) were evaluated using the fluctuations
114064, , https://doi.org/10.1016/j.desal.2019.07.004 URL http://www.
of the characteristics of the feedwater from a real groundwater well. sciencedirect.com/science/article/pii/S0011916419305752.
The optimal operating points for each sample were determined in terms [11] A. Ruiz-García, I. de la Nuez Pestana, Feed spacer geometries and permeability
of minimum SEC and maximum R. The operating margins were found to coefficients. Effect on the performance in BWRO Spriral-wound membrane mod-
ules, Water 11 (1) (2019) 1–13, https://doi.org/10.3390/w11010152 URL https://
be quite wide. The operating parameters of the BWRO systems were www.mdpi.com/2073-4441/11/1/152.
significantly influenced by fluctuations in the feedwater characteristics. [12] P.M. Biesheuvel, L. Zhang, P. Gasquet, B. Blankert, M. Elimelech, W.G.J. van der
In general, the operating windows were wider for the 3:2 than the 2:1 Meer, Ion selectivity in brackish water desalination by reverse osmosis: theory,
measurements, and implications, Environ. Sci. Technol. Lett. 7 (1) (2020) 42–47,
configuration. With such variability, it would be appropriate to design https://doi.org/10.1021/acs.estlett.9b00686 (URL doi:10.1021/acs.estlett.
flexible BWRO systems which are capable of working at least in the 9b00686).
common operation window. In these cases, it could be interesting to [13] A. Karabelas, S.T. Mitrouli, M. Kostoglou, Scaling in reverse osmosis desalination
plants: a perspective focusing on development of comprehensive simulation tools,
implement real-time control strategies in order to maximally optimize Desalination 474 (2020) 114193, , https://doi.org/10.1016/j.desal.2019.114193
the operation through observation of fluctuations in the inorganic URL http://www.sciencedirect.com/science/article/pii/S0011916419313165.
composition of the feedwater. From the economic point of view, the [14] A. Matin, F. Rahman, H.Z. Shafi, S.M. Zubair, Scaling of reverse osmosis membranes
used in water desalination: phenomena, impact, and control; future directions,
operating points of the common operation windows with the highest R Desalination 455 (2019) 135–157, https://doi.org/10.1016/j.desal.2018.12.009
were found to be the most profitable according to the prices established URL http://www.sciencedirect.com/science/article/pii/S0011916418318897.
in the region were the groundwater is located. In future works, changes [15] X. Li, D. Hasson, R. Semiat, H. Shemer, Intermediate concentrate demineralization
techniques for enhanced brackish water reverse osmosis water recovery – a review,
in the operation windows and optimal operating points as the mem-
Desalination 466 (2019) 24–35, https://doi.org/10.1016/j.desal.2019.05.004 URL
branes get fouled should be evaluated. http://www.sciencedirect.com/science/article/pii/S0011916418310816.

8
A. Ruiz-García, et al. Desalination 491 (2020) 114582

[16] A. Ruiz-García, J. Feo-García, Estimation of maximum water recovery in ro desa- [30] M.A. Al-Obaidi, A.A. Alsarayreh, A.M. Al-Hroub, S. Alsadaie, I.M. Mujtaba,
lination for different feedwater inorganic compositions, Desalin. Water Treat. 70 Performance analysis of a medium-sized industrial reverse osmosis brackish water
(2017) 34–45, https://doi.org/10.5004/dwt.2017.20476 URL http://www. desalination plant, Desalination 443 (2018) 272–284, https://doi.org/10.1016/j.
deswater.com/DWT_abstracts/vol_70/70_2017_34.pdf. desal.2018.06.010 URL http://www.sciencedirect.com/science/article/pii/
[17] A. Ruiz-García, M.D. Carrascosa-Chisvert, V. Mena, R.M. Souto, J.J. Santana, S0011916418310415.
I. Nuez, Groundwater quality assessment in a volcanic mountain range (South of [31] A.A. Alsarayreh, M.A. Al-Obaidi, A.M. Al-Hroub, R. Patel, I.M. Mujtaba,
Gran Canaria Island, Spain), Water 11 (4) (2019) 1–20, https://doi.org/10.3390/ Performance evaluation of reverse osmosis brackish water desalination plant with
w11040754 URL https://www.mdpi.com/2073-4441/11/4/754. different recycled ratios of retentate, Comput. Chem. Eng. 135 (2020) 106729, ,
[18] A. Zhu, P.D. Christofides, Y. Cohen, Energy consumption optimization of reverse https://doi.org/10.1016/j.compchemeng.2020.106729 URL http://www.
osmosis membrane water desalination subject to feed salinity fluctuation, Ind. Eng. sciencedirect.com/science/article/pii/S0098135419310348.
Chem. Res. 48 (21) (2009) 9581–9589, https://doi.org/10.1021/ie900729x (URL [32] A.A. Alsarayreh, M.A. Al-Obaidi, A.M. Al-Hroub, R. Patel, I.M. Mujtaba, Evaluation
doi:10.1021/ie900729x). and minimisation of energy consumption in a medium-scale reverse osmosis
[19] M. Li, Minimization of energy in reverse osmosis water desalination using con- brackish water desalination plant, J. Clean. Prod. 248 (2020) 119220, , https://doi.
strained nonlinear optimization, Ind. Eng. Chem. Res. 49 (4) (2010) 1822–1831, org/10.1016/j.jclepro.2019.119220 URL http://www.sciencedirect.com/science/
https://doi.org/10.1021/ie9012826 (URL doi:10.1021/ie9012826). article/pii/S0959652619340909.
[20] M. Li, Reducing specific energy consumption in reverse osmosis (RO) water desa- [33] Y.D. Ahdab, G.P. Thiel, J. Böhlke, J. Stanton, J.H. Lienhard, Minimum energy re-
lination: an analysis from first principles, Desalination 276 (1) (2011) 128–135, quirements for desalination of brackish groundwater in the United States with
https://doi.org/10.1016/j.desal.2011.03.031 URL http://www.sciencedirect.com/ comparison to international datasets, Water Res. 141 (2018) 387–404, https://doi.
science/article/pii/S0011916411002499. org/10.1016/j.watres.2018.04.015 URL http://www.sciencedirect.com/science/
[21] Y.-Y. Lu, Y.-D. Hu, X.-L. Zhang, L.-Y. Wu, Q.-Z. Liu, Optimum design of reverse article/pii/S0043135418302999.
osmosis system under different feed concentration and product specification, J. [34] A. Ruiz-García, I. de la Nuez-Pestana, A computational tool for designing BWRO
Membr. Sci. 287 (2) (2007) 219–229, https://doi.org/10.1016/j.memsci.2006.10. systems with spiral wound modules, Desalination 426 (2018) 69–77, https://doi.
037 URL http://www.sciencedirect.com/science/article/pii/S0376738806007071. org/10.1016/j.desal.2017.10.040 URL http://www.sciencedirect.com/science/
[22] K.M. Sassi, I.M. Mujtaba, Optimal design and operation of reverse osmosis desali- article/pii/S0011916417310950.
nation process with membrane fouling, Chem. Eng. J. 171 (2) (2011) 582–593, [35] A. Ruiz-García, E. Ruiz-Saavedra, S.O.P. Báez, Evaluation of the first seven years
https://doi.org/10.1016/j.cej.2011.04.034 URL http://www.sciencedirect.com/ operating data of a RO brackish water desalination plant in Las Palmas, Canary
science/article/pii/S1385894711004864. Islands, Spain, Desalin. Water Treat. 54 (12) (2015) 3193–3199, https://doi.org/
[23] A. Abbas, Simulation and analysis of an industrial water desalination plant, Chem. 10.1080/19443994.2014.908145 (URL doi:10.1080/19443994.2014.908145).
Eng. Process. 44 (9) (2005) 999–1004, https://doi.org/10.1016/j.cep.2004.12.001 [36] C.I. de Aguas de Gran Canaria, Hydrological plan of Gran Canaria, Canary gov-
URL http://www.sciencedirect.com/science/article/pii/S0255270105000097. ernment, Gran Canaria (Spain), URL, 2014. http://www.aguasgrancanaria.com/
[24] A. Abbas, N. Al-Bastaki, Performance decline in brackish water film tec spiral pdfs/PlanHidro/PHart47_MemInfo.pdf.
wound RO membranes, Desalination 136 (1) (2001) 281–286, https://doi.org/10. [37] M.A. Al-Obaidi, C. Kara-Zaitri, I.M. Mujtaba, Scope and limitations of the irrever-
1016/S0011-9164(01)00191-6 URL http://www.sciencedirect.com/science/ sible thermodynamics and the solution diffusion models for the separation of binary
article/pii/S0011916401001916. and multi-component systems in reverse osmosis process, Comput. Chem. Eng. 100
[25] M. Li, B. Noh, Validation of model-based optimization of brackish water reverse (2017) 48–79, https://doi.org/10.1016/j.compchemeng.2017.02.001 URL http://
osmosis (BWRO) plant operation, Desalination 304 (2012) 20–24, https://doi.org/ www.sciencedirect.com/science/article/pii/S0098135417300571.
10.1016/j.desal.2012.07.029 URL http://www.sciencedirect.com/science/article/ [38] D.W. Solutions, Filmtec Reverse Osmosis Membranes Technical Manual, Dupont
pii/S0011916412004006. Water Solutions, 2020 URL https://www.dupont.com/content/dam/dupont/amer/
[26] M. Li, Optimal plant operation of brackish water reverse osmosis (BWRO) desali- us/en/water-solutions/public/documents/en/45-D01504-en.pdf.
nation, Desalination 293 (2012) 61–68, https://doi.org/10.1016/j.desal.2012.02. [39] A.R. Marsh, P.K. Eriksson, Projecting RO desalination system performance with
024 URL http://www.sciencedirect.com/science/article/pii/S0011916412001257. filmtec spiral-wound elements, Proc. of the Seminar on Membrane Processes, 1988.
[27] M. Li, Optimization of multitrain brackish water reverse osmosis (BWRO) desali- [40] R.P. Brent, Algorithms for Minimization Without Derivatives, Courier Corporation,
nation, Ind. Eng. Chem. Res. 51 (9) (2012) 3732–3739, https://doi.org/10.1021/ 2013.
ie202796u (URL doi:10.1021/ie202796u). [41] G.E. Forsythe, M.A. Malcolm, C.B. Moler, Computer Methods for Mathematical
[28] Y. Du, L. Xie, Y. Wang, Y. Xu, S. Wang, Optimization of reverse osmosis networks Computations, Prentice-Hall, 1977.
with spiral-wound modules, Ind. Eng. Chem. Res. 51 (36) (2012) 11764–11777, [42] A. Ruiz-García, J. Feo-García, Antiscalant cost and maximum water recovery in
https://doi.org/10.1021/ie300650b (URL doi:10.1021/ie300650b). reverse osmosis for different inorganic composition of groundwater, Desalin. Water
[29] M. Li, A unified model-based analysis and optimization of specific energy con- Treat. 73 (2017) 46–53, https://doi.org/10.5004/dwt.2017.20806 URL https://
sumption in BWRO and SWRO, Ind. Eng. Chem. Res. 52 (48) (2013) 17241–17248, www.deswater.com/DWT_abstracts/vol_73/73_2017_46.pdf.
https://doi.org/10.1021/ie4031083 (URL doi:10.1021/ie4031083).