See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330726999

Cost and energy requirements of hybrid RO and ED brine concentration

systems for salt production

Article  in  Desalination · January 2019

DOI: 10.1016/j.desal.2018.11.018

CITATIONS READS

6 911

7 authors, including:

Kishor Govind Nayar Jenifer Fernandes

GHD Pty Ltd Kuwait University
21 PUBLICATIONS   302 CITATIONS    4 PUBLICATIONS   12 CITATIONS   

SEE PROFILE SEE PROFILE

Bader Al-Anzi
Kuwait University / MIT
57 PUBLICATIONS   133 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Sanitation for 600 million in India: The Problem and Observations from a Service Project Exploring the Use of Meditation for Community Capacity Building View project

Next Generation Brine Desalination and Management for Efficiency, Reliability, and Sustainability View project

All content following this page was uploaded by Bader Al-Anzi on 04 February 2019.

The user has requested enhancement of the downloaded file.

 Desalination 456 (2019) 97–120

Contents lists available at ScienceDirect

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

Cost and energy requirements of hybrid RO and ED brine concentration T

systems for salt production
Kishor G. Nayara, Jenifer Fernandesb, Ronan K. McGoverna, Kyle P. Domingueza,
Adriene McCancec, Bader S. Al-Anzib, John H. Lienhard Va,*
a
Department of Mechanical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, United States
b
Environmental Technology and Management Department, College of Life Sciences, Kuwait University, Kuwait
c
Sloan School of Management, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, United States

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

Keywords: A new concept to concentrate seawater up to 200 g/kg for producing vacuum salt using a reverse osmosis (RO)
Reverse osmosis system hybridized with an electrodialysis (ED) system is presented. The RO system operates up to pressures of
Electrodialysis 120 bar and concentrates seawater up to 120 g/kg with the ED system concentrating RO brine to 200 g/kg. A
Hybrid parametric analysis to minimize the specific cost of brine concentration was conducted. Parameters varied were:
Salt production
the degree of RO-ED hybridization, ED current density, electricity prices and water prices. Optimal hybrid RO-
Brine concentration
ED designs reduced brine concentration costs by 33–70% over standalone ED systems, with revenue generated
Seawater
from water co-production further subsidizing costs by 1–6%. Optimizing ED current density reduced costs the
most. Including a crystallizer, the total reduction in production cost over a standalone ED-crystallizer system was
19–55%, with the production cost for a typical case being $111/tonne-salt. The proposed RO-ED-crystallizer
(REC) systems were found to be techno-economically feasible in Cyprus, Japan, Kuwait, Saudi Arabia, and the
USA. At a road transportation distance of 735 km, REC based seawater vacuum salt was competitive with
conventional vacuum salt. REC systems may open up the potential of small-scale decentralized salt production.

1. Introduction requires significant land areas and is used only in locations where land
prices are low. In conventional “vacuum salt” production [3], saturated
Globally, each year, more than 280 million tonnes of salt [1] is solutions of brine (S = 260 g/kg) extracted from solution mines are sent
produced from solar evaporation of seawater, from conventional rock to thermal or electrically driven evaporators and crystallizers (referred
mining and from the solution mining of saline brines [2]. Of this, 39% is in this paper hence simply as “crystallizer”) to evaporate off water and
used by the chloralkali industry for the production of chlorine and so- produce salt. Vacuum salt is more expensive than solar salt production
dium hydroxide, 22% for human consumption, 22% by the soda ash primarily because of the higher capital costs in equipment needed and
industry, 9% for de-icing of roads and 9% for other uses [2]. Salt is the energy costs since in the latter solar energy is used for free. Despite
important both as a crucial nutritional source and as a crucial raw the differing production costs, the global salt market can support both
material for global chemical production. Given the strategic importance methods of salt production partly because of high transportation costs.
of salt for the world, producers globally are looking at ways to make salt Transporting 1 tonne of any commodity per 100 km costs around
production more sustainable and reduce production costs. Salt pro- $10.34/tonne-100 km [5] by truck, 2.53 $/tonne-100 km by rail [5],
duction costs vary significantly with the method of production, with the and 1.14 $/tonne-100 km by barge [5]. Thus at a certain distance and
lowest production costs being for “solar salt” [3] (around $5-10/tonne between certain cities, “vacuum salt” becomes a more competitive
of salt [4]) and the highest production costs being for “vacuum salt” [3] option.
(around $30–50 /tonne-salt [4] for conventional vacuum salt produced One of the most expensive types of “vacuum salt” is “vacuum salt”
from saturated brines). In “solar salt” production [3], seawater is col- produced from seawater through the use of “brine concentration” and
lected in evaporation ponds and freely available solar insolation is used “crystallizer” systems. In such systems seawater (S = 35 g/kg) is first
to evaporate off water leaving behind salt. Solar salt production concentrated to near saturation levels (S = 180–260 g/kg) using a

*
Corresponding author.
E-mail addresses: [email protected] (K.G. Nayar), [email protected] (R.K. McGovern), [email protected] (K.P. Dominguez), [email protected] (A. McCance),
[email protected] (B.S. Al-Anzi), [email protected] (J.H. Lienhard).

https://doi.org/10.1016/j.desal.2018.11.018
Received 23 June 2018; Received in revised form 9 November 2018; Accepted 20 November 2018
0011-9164/ © 2018 Elsevier B.V. All rights reserved.
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 1. Simplified flow diagram of a conventional salt production plant with an ED system and crystallizer.

separate “brine concentration” system before being sent to a crystal- In this paper, our objective is the production of salt from seawater
lizer. Separate “brine concentration” systems are needed because and our analysis is restricted to MSED systems. For the convenience of
crystallizers are designed to be cost effective only when using nearly readers, henceforth in this paper, we will be referring to MSED systems
saturated brine. Significant energy is needed to concentrate seawater to simply as ED systems. The flow diagram of a conventional salt pro-
near saturation levels with the required volume reduction being around duction plant using ED is shown in Fig. 1. Seawater feed, typically at
90% [6]. When low cost land is not available for the construction of 35 g/kg, first flows in to both the diluate and concentrate channels of an
solar evaporation ponds, separate “brine concentration” systems such ED stack. In the concentrate channel, seawater is typically concentrated
as Mechanical Vapor Compression (MVC) [7-11] or Electrodialysis (ED) from 35 g/kg to 177–200 g/kg with the diluate discharged back in to
[12-18] needs to be used. To the best of our knowledge, the only the sea at a salinity much less than seawater. Typically the salinity
commercial “vacuum salt” production systems that produce salt from change happens along the length of a single membrane with the voltage
seawater through a combination of “brine concentration” systems and kept constant along the length [14]. Such a design using a single vol-
“crystallizers” are plants in Japan, Korea and Kuwait [14,17,19] that tage along the length of a stack is referred to in the literature as a
use ED to first concentrate seawater from 35 g/kg to around 180–200 g/ “single electric stage” design [28].
kg after which brine is sent to a crystallizer where salt is produced for While ED based brine concentration systems for salt production
either human consumption [19] or for chlor-alkali production [14]. A have been used in Japan for around 50 years, such systems have not
flow diagram of such a typical ED based “salt production” plant is been deployed widely outside of Japan and South Korea partly due to
shown in Fig. 1. the high costs involved in concentrating seawater. For reference,
ED is a versatile electric driven membrane based desalination Miyake et al. [29] had reported a production cost of 100 $/tonne-salt
technology, that was first conceptually conceived for demineralizing for an ED-crystallizer system with 65% of the costs being capital costs.
sugar syrup in 1890 [20], and developed for saline water desalination Another membrane based desalination technology that has become
in the 1940s and 1950s[20,21]. ED systems primarily consist of several very popular now is reverse osmosis (RO). RO is a pressure driven
pairs of anion exchange membranes (AEM), and cation exchange membrane based process that separates saline water into pure product
membranes (CEM) placed in between a cathode and anode (shown later water and a saltier brine stream [30]. Today, RO has become the most
in Fig. 5). Feed water is fed into channels created between CEMs and widely adopted seawater desalination technology [31]. Seawater RO is
AEMs. When a voltage is applied, cations move to the cathode and also now the most energy efficient desalination technology [9,32] for
anions move to the anode; however, the AEMs prevent the movement of seawater desalination. Thermodynamic analysis has also shown con-
cations and CEMs prevent the movement of anions resulting in one ceptually that RO can also be the most energy efficient brine con-
channel becoming concentrated in ions while the adjacent channel is centration technology for concentrating brine from 150 to 260 g/kg
depleted of ions. Consequently, these two adjacent channels are re- [33]. Conventional seawater RO operates at a pressure of 50–70 bar
ferred to as the ‘concentrate’ and ‘diluate’ channels. Depending on the recovering 40–50 % of seawater feed as pure product water with the RO
type of AEMs and CEMs used, ED can be designed for several applica- brine discharged back in to the sea at around a salinity of 60–70 g/kg.
tions including city-scale brackish water desalination [17,21], village- RO systems for concentrating brine to 260 g/kg do not currently exist in
scale water treatment [22], seawater brine concentration [14-19], de- operation due to the high operating pressures required for brine con-
nitrification of water for municipal water supply [21], demineralization centration (379.2 bar at NaCl saturation [33]). However, in the past
of wine, whey and sugar [23], in-home water treatment [24-26] and 5 years, advances in RO membrane technology have enabled the de-
wastewater treatment [17,27]. For salt production from seawater, velopment of higher pressure RO systems that can operate up to pres-
AEMs and CEMs that are additionally selective to monovalent ions are sures of 120 bar [34] corresponding to a RO brine salinity of 120 g/kg.
used. The resulting monovalent selective ED (MSED) systems con- These higher pressure RO systems are new and have not yet been
centrates sodium chloride preferentially over other ions in seawater adopted widely by the seawater desalination industry.
leading to the production of brines of salinity 180 g/kg to 200 g/kg that Some key research questions are: can the cost of seawater brine
are rich in sodium chloride. MSED systems for concentrating brine for concentration be reduced below that of current ED costs? If so, by how
salt production have been in commercial operation for more than much can the cost be reduced? Can hybridizing ED with RO reduce
50 years in Japan [19]. costs? And what are the economic implications of reducing brine

98
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 2. (a) Salt production plant with Reverse Osmosis, Electrodialysis and crystallizer sub-systems. (b) Flow diagram of plant with Reverse Osmosis, Electrodialysis
and crystallizer sub-systems.

concentration costs? Can a new salt production industry be created out of the ED system to a salinity of 200 g/kg. To make full use of the RO
of small-scale salt production plants that produce “vacuum salt” brine for salt production, all of the RO brine is sent to the concentrate
through “brine concentration” and crystallizer systems? Our concept, channel for further concentration, with only seawater feed at 35 g/kg
explained in Section 2, seeks to answers these questions and may in- used as input for the diluate channel. Simulations developed here were
spire further research and development work in this area. later used to confirm that RO-ED hybrid system costs were also mini-
mized when all of the RO brine was sent to the concentrate channel
[35]. The concentrate from the ED system is then sent to the crystallizer
2. Proposed concept: RO-ED hybrid brine concentration systems where it is separated in to pure salt and water streams. Because ions are
used with crystallizers for salt production transferred from the diluate stream to the concentrate stream, the ED
system discharges a stream more dilute than seawater back into the sea.
In this paper, we evaluate potential cost reduction for ED based
seawater brine concentration by using ED systems that are hybridized
with reverse osmosis (RO). Fig. 2a shows an illustration of our proposed 2.1. Why hybridize ED with RO for brine concentration?
RO-ED brine concentration system. Fig. 2b shows the detailed flow
diagram of the RO, ED and crystallizer systems. Seawater feed (35 g/kg) For increasing the economic feasibility of ED based brine con-
first flows into an RO system where pure product water and desalina- centration, we considered two paths: reduce system costs and increase
tion brine are produced. The brine produced by the RO system revenue. The capital cost of ED contributes significantly to the overall
(60–120 g/kg salinity) is then concentrated in the concentrate channel cost of salt production. From first principles, ED capital costs scale

99
K.G. Nayar et al. Desalination 456 (2019) 97–120

directly with membrane area and the amount of salt removed by the ED 3. Methodology
system. Thus, to reduce ED capital costs, the ED membrane area de-
ployed for the transfer of salt must be reduced. Holding the total salt In this section we discuss the techno-economic models used to si-
production capacity constant, ED membrane area can be reduced by: mulate the RO, ED and crystallizer systems.

(a) shifting the salt removal load from ED to a more efficient and cost- 3.1. Plant configuration
effective technology
(b) increasing the salt removed per square meter of ED membrane A standard seawater salinity of Ssw = 35 g/kg [40] is used as a
baseline for this study. Annual salt production is desired to be roughly
Both these approaches are evaluated in this paper. 100,000 tonnes/year. For simulation purposes, the seawater feed flow
For reducing ED capital costs by shifting some of the salt removal rate to the RO system was held constant at 50 m3/h, corresponding to
load from ED to an alternative technology, the best alternative today is an annual salt production capacity of 100,000 to 900,000 tonnes/year
arguably seawater RO and its variants. Hybridizing RO and ED systems depending on the degree of RO-ED hybridization. For a standalone ED
for brackish water desalination (S < 3 g/kg) has already been shown system, the concentrate inlet flow rate was taken to be 50 m3/h.
to be lower in cost than standalone ED systems [36]. Similarly, hybrid
RO-ED systems have been conceived to treat high salinity produced 3.2. Reverse osmosis process model
water from a salinity of 120 g/kg, producing pure water and ED con-
centrate at 167 g/kg and RO brine at 70 g/kg [37]. Previously, Tanaka The reverse osmosis model is an adaption of the model described by
et al. [38] have also shown that using RO brine salinity at 88 g/kg as Thiel et al. [33] with performance parameters adapted from Mistry
feed in to a ED brine concentration system reduced ED costs by 20% et al. [32]. Seawater feed enters the RO system at a pressure of 1 bar. A
when compared with using seawater feed. However, Tanaka et al. [38] circulation pump (CP) increases the feed pressure by ΔPCP = 1 bar. The
did not report on the combined cost of the RO-ED system or its eco- feed stream from the circulation pump, ṁ f,RO is then split in two
nomic feasibility and did not analyze the effect of varying the operating streams, a feed stream of mass flow rate ṁ p, RO goes to a high pressure
pressure of RO. pump (HP), and a feed stream having a mass flow rate equal to the
For reducing ED capital costs by increasing the salt removed per m2 brine flow rate ṁ b,RO goes to a pressure exchanger. The HP then pres-
of membrane, a direct approach is to increase the operating current surizes part of the feed to a pressure of PHP,RO, to overcome the osmotic
density. However, increasing the current density also increases the pressure difference and the pressure losses in the RO module (Fig. 3).
energy consumed and the operating cost. Tanka et al. [38] had reported The pressure exchanger meanwhile pressurizes the other feed stream to
that a current density of 300 A/m2 optimized the energy consumption Precover,RO pressure, which is then passed to a booster pump to attain the
of ED stacks using both seawater feed and desalination feed. However, pressure PHP,RO. The two feed streams then combine and enter the RO
the optimal current density optimizing ED energy consumption for ED module. The RO module separates the feed into a concentrated brine
feed salinities between 70 g/kg and 120 g/kg have not been reported stream and a pure product stream at a recovery ratio RRRO:
and the dependence of optimal current density on electricity prices has
ṁ p,RO
not been discussed. Furthermore, the current densities that optimized RRRO =
ṁ f,RO (1)
ED costs has not been reported in the literature.
For increasing revenue over standalone ED brine concentration For a given RO brine salinity (Sb,RO), the osmotic pressure at the end of
systems, a simple approach is by producing pure water from the RO the unit is:
sub-system of the RO-ED concept and selling the water. Water prices
ΠRO,b = SWOsm,Press (Sb,RO, t = 25 ∘C) (2)
vary significantly around the world, with water tariffs ranging from 0 to
7.54 $/m3 with a mean global water tariff of 1.21 $/m3 [39]. Further- Here, the osmotic pressure of seawater was determined using seawater
more, salt market prices also vary significantly in the world. To the best osmotic pressure correlations developed by Sharqawy et al. [41] and
of our knowledge, the literature has not addressed how salt and water Nayar et al. [42]. Eq. (3) gives the pressure corresponding to feed outlet
prices affect the design of RO-ED systems. from the circulation pump:
Several questions remain unanswered. If the primary application is
PCP,RO = Patm + ΔPCP (3)
salt production, does hybridizing ED with RO reduce costs and energy
consumption for salt production when compared with standalone ED while, the pressure from the high pressure pump is given by:
systems concentrating seawater? Does extending the use of RO to
PHP,RO = ΠRO,b + ΔPpinch, RO + ΔP loss, RO (4)
higher pressures and salinities (120 bar and 120 g/kg) increase the
economic feasibility of RO-ED systems for seawater brine concentra-
tion? How does energy consumption of RO-ED systems vary with in-
creasing hybridization (i.e., increasing the proportion of the salt re-
moval load taken by the RO system)? At what current densities should
ED in RO-ED systems be operated for reducing costs? How does the
system performance vary with water and electricity prices? Given the
varying salt prices around the world, in what locations and scenarios
would RO-ED based brine concentration be commercially feasible for
salt production? In this paper, we present an answer to these questions
through a techno-economic and parametric analysis of RO-ED systems
for concentrating seawater from 35 g/kg to 200 g/kg. Parameters stu-
died include: RO brine salinity (i.e., an indicator of amount of hy-
bridization), ED current density, electricity prices, water prices and salt
prices. We also discuss the implications of the proposed RO-ED concept
for salt production and highlight areas for future research work.

Fig. 3. Diagram of RO system modeled.

100
K.G. Nayar et al. Desalination 456 (2019) 97–120

where a pinch pressure (ΔPPinch,RO) of 10 bar and pressure loss

(ΔPloss,RO) of 2 bar is assumed in the RO module. From Mistry et al.
[32], the pressure recovered from the pressure exchanger was:
ρf, RO
Precover,RO = PCP,RO + ηPX (PHP,RO − ΔP loss,RO − Patm )
ρ b, RO (5)
Efficiencies of the RO pumps (ηp,RO) and of the pressure exchanger
(ηPX,RO) were taken to be 0.85 and 0.96 respectively. From these pres-
sures, the work of the pumps could be calculated:

̇ ̇ ΔPCP 1
WCP,RO = Vf,RO
ηp,RO (6)

ṁ p,RO 1
ẆHP,RO = Vf,RO
̇ ΔPHP
ṁ f,RO ηp,RO (7)

ṁ b,RO 1
ẆBP, RO = Vf,RO
̇ ΔPBP
ṁ f,RO ηp,RO (8)
Fig. 4. Variation of specific capital cost of RO with capacity fitted to the
Hence, the total work input to the system is: equation SpCapExRO,p,day = 3619 − 201.3 × ln (VRO,p,day ) with an R2 value of
0.999996 and absolute percentage deviation from data of 0.06%.
ẆRO = WCP,RO
̇ + ẆHP,RO + ẆBP,RO (9) Source: Data source: Ref. [43].
The specific energy consumption (kWhe/m3-RO-product) is:
the rate of return (rreturn = 7%) was assumed to calculate the annuity
ẆRO
Etotal, RO = factor of the RO plant:
̇
1000[W/kW] Vp,RO,hr (10)
t life
̇
where Vp,RO,hr is the product water produced per hour. The specific
Annuityfactor =
1− ( 1
1 + rreturn )
energy consumption (kWhe/tonne-salt) of the RO system normalized to rreturn (16)
salt produced is:
Thus, the total RO capital cost after annualization was:
ẆRO
SpERO =
ṁ salt × 3600[s/h] (11) CapExRO
CapExRO,yr =
Annuityfactor (17)
where msalt is the salt produced per second. The RO energy costs were
obtained from the electricity cost (Costelec), and in our analysis an From the RO process model, we had previously calculated the an-
electricity cost of US $0.10/kWh was assumed: nual RO energy costs. However, operating expenses also include
ẆRO membrane replacement costs, maintenance costs and labor costs. From
OpExRO, energy, hour = Cost elec the capital cost data from DesalData [43], we know that RO membranes
1000[W/kW] (12)
only contribute to 6.5% of the total capital cost of an RO system:
Accounting for a capacity factor, Capfac = 0.9, the annual RO energy
CapExRO, mem. = 0.065 × CapExRO (18)
cost is:
OpExRO, energy, yr = OpExRO, energy, hour × 8760[h/year] × Capfac (13) Seawater RO membranes have been reported to last 3–7 years [44-47]
with an increase in membrane life observed in recent years with
membrane development. The most recent seawater RO plant data we
3.2.1. Cost model for conventional seawater RO obtained, from 2013 [47], reported an observed life of almost 7 years
DesalData [43] has extensive data for the specific capital cost of even when the plants only used conventional pretreatment consisting of
real-world seawater RO desalination plants. The capital cost estimator media filters. For our calculations, we conservatively chose a mem-
tool from DesalData accounts for a variety of factors such as the size of brane life of 5 years. Annualizing membrane replacement cost every
the plant, the type of pretreatment, the salinity of the seawater, etc. For 5 years we have the annual membrane replacement cost:
the purposes of this paper, we are interested in obtaining the specific
capital cost of RO in the 500–2500 m3/day capacity range. We observed CapExRO, mem.
OpExRO,mem., yr=
a strong dependence of the specific cost of RO on the capacity of the RO Annuityfactor
plant. To obtain an accurate estimate of the capital cost of RO, capacity × [(1 + rreturn )−5 + (1 + rreturn )−10 + (1 + rreturn )−15]
ranges from 250–250,000 m3/day were entered in to DesalData's ca-
(19)
pital cost estimator tool, with the input parameters being conventional
operating conditions and “standard” pre-treatment. Fig. 4 shows the which translated financially to an annual membrane element replace-
variation of the specific capital cost of RO with the capacity of the RO ment rate of 15%. DesalData [43] reported that the other operating
system. The specific capital cost of RO was found to vary logarith- costs of other maintenance work, chemicals and labor amounted to be
mically with product water capacity as: approximately constant for seawater RO systems at 0.03 $/m3, 0.07 $/
m3 and 0.08 $/m3 respectively per amount of RO product water pro-
SpCapExRO,p,day = 3619 − 201.3 × ln (VRO,p,day ) (14) duced. Together these other operating costs amounted to 0.18 $/m3 of
2
with an R value of 0.999996 and absolute percentage deviation from RO product water. This translated to an annual operating cost con-
data of 0.06%. tribution of:
Thus, the total capital cost for the RO system was: OpExRO, maint., chem., labor, yr = 0.18$/m3 × Vp,̇ RO, yr (20)
̇
CapExRO = SpCapExRO,p,day × Vp,RO,day (15)
where Vp, RO, yr is the annual amount of water produced by the RO
The capacity factor (Capfactor = 0.9), a project life (tlife = 20 years) and system in m3/year. Thus, the total operating cost of RO was:

101
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 5. Diagram showing an ED stack. The ED plant is assumed to be a single stack with Ncp identical parallel cell-pairs.

OpExRO, yr= OpExRO, energy, yr + OpExRO, mem., yr seawater RO capital cost came from pressure affected components.
+ OpExRO, maint., chem., labor, yr Components such as pretreatment and intake/outfall were not added to
(21)
the HPRO stage since they were already accounted for in the preceding
which brings us to the specific capital cost of RO water normalized to conventional seawater RO system. HPRO membrane replacement costs
the annual salt production as: were three times higher due to the higher cost of HPRO membrane
CapExRO,yr + OpExRO,yr elements. Other maintenance, labor and chemical costs were assumed
SpCostRO = to be the same on a per m3 feed basis as conventional seawater RO.
ṁ salt, yr (22) The costs and energy needs associated with the HPRO stage were
were, ṁ salt, yr , is the annual amount of salt produced from the RO-ED- added to the conventional SWRO stage and reported together in the
crystallizer based salt production plant in tonnes/year. results for the RO sub-system.
To validate our RO cost model, we compared it against water price
predictions from Desaldata for the same input conditions. For the 3.3.1. Scaling potential and need for antiscalants or further pretreatment
production capacity ranges 250–2500 m3/day of product, our model for HPRO stage
deviated from DesalData's estimates by only 4.3–5.5 %. The deviation We further evaluated the scaling potential when concentrating
arose from minor differences from DesalData in calculating energy seawater to 120 g/kg using a HPRO stage. The composition of 120 g/kg
consumption and the assumed membrane life. brine leaving a HPRO stage was estimated by neglecting salt passage
through the membrane and concentrating standard seawater composi-
3.3. High pressure (60–120 bar) RO stage process model tion [40]. We verified our predicted composition against that predicted
by Dow's ROSA [49] and WAVE [50] simulation software. They were on
Analysis of seawater RO plant operation data from the literature average within 5% of each other for all major ions. The scaling po-
[48] showed that, on average, seawater RO plants around the world tential was evaluated by calculating the saturation indices for possible
operated at a recovery ratio of 42% with an operating pressure near scale forming salts by running the HPRO brine composition against the
60 bar. Practically, extending RO operation beyond 60 bar to 120 bar Pitzer electrolyte models [51] in the software PHREEQC [52,53]. There
pressure involves the addition of a high pressure RO (HPRO) stage was a clear need for antiscalants due to the potential for both carbonate
designed around high pressure RO membranes [34]. In this paper, the and sulfate scaling. We worked with an antiscalant company [54] that
reference “RO” refers to the sub-system that includes a conventional had considerable operational experience with seawater desalination to
SWRO stage and, a HPRO stage when the operating pressure exceeds determine the correct antiscalant and dosage needed to prevent scale
60 bar. formation. The divalent ions in the 120 g/kg HPRO brine was de-
The energy required to operate the HPRO stage was calculated using termined to be well within the operational limits of basic antiscalants.
the same process model described for RO in Section 3.2. The associated chemical cost had been accounted for from the specific
For RO operating pressures higher than 60 bar, no datasets on chemical costs from DesalData. We ascertained that if the right anti-
equipment costs were publicly available. However, by looking at the scalants were used there was no operational need from a scaling point
individual cost contributions to a conventional seawater RO plant and of view, to carry out additional pretreatment such as using nanofiltra-
by using both engineering estimates and quotes from component tion (NF) to enable the addition of a HPRO stage. However, the addition
manufacturers, we were able to estimate the cost of an HPRO stage (see of NF pretreatment maybe considered by salt production plant de-
Appendix A for details). Overall we calculated that for the same re- signers for other reasons such as to produce salt of even higher purity
covery ratio, the specific capital cost of a HPRO stage was only 1.09 (> 99.8%), to reduce scaling and maintenance costs in the crystallizer
times that of conventional seawater RO. and to increase the plant capacity factor. For the benefit of readers
The main reason why the HPRO stage cost increase was minor de- interested in including an NF system, we have reported an NF process
spite the operating pressure being double is that only 29% of the model and the costs for NF pretreatment in Appendix B.

102
K.G. Nayar et al. Desalination 456 (2019) 97–120

3.4. Electrodialysis process model From data given by an ED manufacturer [58], we set the effective
cell-pair area for each membrane pair (Acp) to be 0.395 m2. Thus, we
The ED system was modeled by adapting a model developed by can determine the number of cell pairs as:
McGovern et al. [55], which was itself based on a transport model N
developed originally by Fidaleo and Moresi [56]. In this work, we ∑ j = 1 Acp, tot, j
Ncp =
adapted the model by McGovern et al. [55] for the concentration of Acp (27)
brine coming from seawater RO (see Fig. 2b), keeping both the con-
The transport model from McGovern et al. was then applied to each
centrate and diluate channels fully continuous, with the salinities of
individual cell to obtain the net salt (Js,j) and water flux (Jw,j) going
both channels varying along the length of the ED stack. The salinities at
from the diluate to the concentrate in each cell:
the inlet (Sc, i) and outlets (Sc, o) of the concentrate stream are known,
with the former being the salinity of the brine leaving the RO system ij
Js,j = Ts,j⋅ − Ls,j⋅(Cc,m,j − Cd,m,j)
and the latter being 200 g/kg. The value of 200 g/kg for the ED con- F (28)
centrate outlet was chosen to reflect the maximum salinity seen in in-
dustrial ED systems [57]. Seawater was directly fed in to the diluate ij
Jw,j = Tw,j⋅ + L w,j⋅(πc,m,j − πd,m,j)
channel of the ED system. All the brine leaving the RO system was F (29)
assumed to flow into the inlets of the ED concentrate channels. An ED where, Ts,j and Tw,j are the salt and water transport numbers for the ED
stack can be considered as a collection of Ncp identical parallel cell-pairs membrane and Ls,j and Lw,j are the salt and water permeabilities of the
which consist of an anion exchange membrane, a cation exchange membrane. Cc,m,j and Cd,m,j are the molar concentrations of salt in the
membrane, a diluate channel, and a concentrate channel. Currently concentrate and diluate at the surface of the membrane. πc,m,j and πd,m,j
operational industrial ED stacks are ‘single stage’ designs [57] where are the osmotic pressures of the diluate and concentrate at the surface
the change in salinity on the concentrate side from 35 g/kg to 200 g/kg of the membrane.
happens across a single ED cell-pair. We have assumed the same ‘single McGovern et al. [55] experimentally determined the transport and
stage’ operation for our ED system. For capturing the effect of transport permeability numbers for a NEOSEPTA AMX and CMX membranes at
variation along the length of an ED cell-pair, the full length of a single high salinities. McGovern et al. had expressed the transport and per-
ED cell-pair was discretized in to N computational cells such that each emeability values as simple correlations which are reproduced below:
cell sees the same change in salinity on the concentrate side:
Tscp = −4 × 10−6Sd2 + 4 × 10−5Sd + 0.96 ± 0.04 (30)
Sc, o − Sc, i
Sc, k + 1 = Sc, k +
N−1 (23) Twcp = −4 × 10−5Sc2 − 1.9 × 10−2Sd + 11.2 ± 0.6 (31)
where Sc, i and Sc, o are the salinities at the inlet and outlet of the
Lscp = min(2 × 10−12Sd2 − 3 × 10−10Sd + 6 × 10−8,
concentrate stream and k = 1 to N − 1. For ED simulations, N = 50 was
2 × 10−12Sc2 − 3 × 10−10Sc + 6 × 10−8) ± 6 × 10−9[m/s] (32)
used as it gave grid independent accurate results while keeping the
simulations computationally fast. The transfer of salt and water in an
L wcp = 5Sc−0.416 ± 2 × 10−5 [mol/m2s bar] (33)
ED computation cell is shown in Fig. 6.
The total molar flow rate of salt (Ns,̇ c, j + 1) and water (Ṅw, c, j + 1) in the The molar flow rate of salt and water in the diluate in ‘j + 1’th com-
concentrate channels are obtained from the known mass flow rate putational cell is calculated from:
flowing in to the concentrate channel and the salinity along the con-
Ns,̇ d, j + 1 − Ns,̇ d, j = −Acp, tot, j Js, j (34)
centrate channel:
ṁ c, j Sc, j ̇ d, j + 1 − Nw,
Nw, ̇ d, j = −Acp, tot, j Jw, j (35)
Ns,̇ c, j =
1000 × MWs (24)
The salt and water flux across the ED membrane is driven by the
The net salt flux (Js,j) and net water flux (Jw,j) from the diluate voltage applied across an ED cell-pair (Vcp). This voltage is related to
channel in to the concentrate channel can then be related to the total the current density by the following expression from McGovern et al.
concentrate molar flow rates as: [55]:

Ns,̇ c, j + 1 − Ns,̇ c, j = A cp, tot, j Js, j (25) hd hc ⎞

Vcp= ⎜⎛r¯am, j + r¯cm, j + + ⎟ ij
⎝ σ Λ d C d , j σ Λ c Cc, j ⎠
̇ c, j + 1 − Ṅw, c, j = A cp, tot, j Jw, j
Nw, (26) r¯cm, j 2hr
+ ij + i j + Eam, j + Ecm, j
where Acp, is the total cell-pair area discretized to each computa- Ncp σkr Ncp, j (36)
tot,j
tional cell.
The expression can be best understood by referring to Fig. 7 which
reproduces the equivalent circuit diagram of an ED cell-pair from
McGovern et al. [55]. Briefly, from left to right, the terms in Eq. (36)
represent the resistances of the anion and cation exchange membranes,
the diluate and concentrate channel resistances, the resistance of an
extra cation exchange membrane at the end of a stack, the resistance of
the rinse stream which may be present in stacks and the voltage drops
across the anion and cation exchange membranes arising from con-
centration polarization. The detailed description of each term with the
expressions for calculating concentration polarization are given in Ref.
[55]. The voltage across the each cell pair can be added together with
the voltage across the electrodes to get the total voltage across the ED
stack:
VED, stack = Ncp Vcp + Vel (37)
Fig. 6. Transport of salt and water in an ED computation cell. The stack power can then be calculated as:

103
K.G. Nayar et al. Desalination 456 (2019) 97–120

for membrane replacement, maintenance and labor. From the industry

[57], we know the membranes cost (SpCostED, mem., m) to be $ 222/m2
of membrane. Thus the capital cost for one set of ED membranes for an
ED system will be:
CapExED, mem. = SpCostED, mem., m × Amem, total (48)
Monovalent selective ED membranes have a reported life of 5–10 years.
From our conversations with ED manufacturers and plant operators, we
chose an average life of 7 years for our cost model. Annualizing this cost
of replacing ED membranes every 7 years we have the annual mem-
brane replacement cost:
CapExED, mem.
OpExED, mem., yr=
Annuityfactor
Fig. 7. Circuit diagram of an ED cell pair. × [(1 + rreturn )−7 + (1 + rreturn )−14] (49)
ED brine concentration systems for salt production typically have one
N
ẆED, stack = full-time operator. The average annual salary for a plant operator in the
∑ i j Acp, j (Ncp Vcp + Vel)
j=1 (38) USA is $50,000 [59,60]. This was assumed to be the annual labor cost
from the ED system (OpExED, labor, yr). Other maintenance costs and
The power required for pumping the concentrate and the diluate was chemical costs were obtained by converting data from brackish water
calculated using the expressions for friction factor and pressure drop ED plants in the literature [59] by the method previously used by
from McGovern et al. [36]: McGovern et al. [55]. Sajtar and Bagley had reported the specific cost of
̇ in
ΔPdil Vdil, ̇ in
ΔPcon Vcon, other maintenance and chemicals to be 0.0285 and 0.007 $/m3 of feed.
ẆED, pump = + Using the model from McGovern et al. [55] a brackish ED plant would
ηp, ED ηp, ED (39)
need 0.39 m2 of cell pair area per m3/day of feed capacity. Using a
The total power required to run the ED sub-system was thus: membrane effectiveness of 0.64 and accounting for 2 ED membranes
per cell pair, this translates to 1.22 m2 of total ED membrane area per
ẆED = ẆED, stack + ẆED, pump (40)
m3/day of feed capacity. Assuming 365 days of operation, the main-
Normalizing the total power consumed by the ED sub-system to the salt tenance and chemical costs can be converted from per $/m3 of feed to
produced, we can obtain the specific energy consumption (kWhe/tonne- per m2 of total ED membrane area per year. Thus, the specific annual
salt) of the ED system: cost of maintenance (SpCostED, maint., m) and chemicals (SpCostED, chem.,
2
m) was calculated to be 8.5 and 2.1 $/m -year. We can then get the
ẆED
SpEED = annual maintenance and chemicals costs for the ED system and the total
ṁ salt × 3600[s/h] (41) annual ED operating expenses:
The hourly energy costs of the ED system can also be obtained as: OpExED, maint., chem., yr= (SpCostED, maint., m + SpCostED, chem., m)
ẆED × Amem, total (50)
OpExED, energy, hour = Cost elec
1000[W/kW] (42)
OpExED, yr= OpExED, energy, yr + OpExED, mem., yr
Accounting for a capacity factor, Capfac = 0.9, the annual ED energy
+ OpExED, labor, yr + OpExED, maint., chem., yr (51)
cost was obtained as:
OpExED, energy, yr = OpExED, hour × 8760[h/year] × Capfac which brings us to the specific cost of ED normalized to salt production
(43)
as:
The capital costs for the ED system can be calculated from the total
CapExED,yr + OpExED,yr
membrane area required. SpCostED =
ṁ salt, yr (52)
N
Acp, total = Ncp ∑ Acp, j Values of inputs and constants used in the ED model are given in
j=1 (44) Table 1.
Acp, total
Amem, total = 2 3.4.1. Validation of electrodialysis model
Mem eff (45)
3.4.1.1. Laboratory scale validation. The ED model presented here was
Here, the factor 2 accounts for the fact that each cell pair has two validated against experiments carried out by McGovern et al. [55]
membranes and Memeff accounts for the fact that only a portion of the (Fig. 8). Each stage shown in Fig. 8 represents a pairing of diluate and
membrane area is used for transport due to the presence of spacers and concentrate salinities ranging from 0 to 225 g/kg. McGovern et al.
gaskets. From conversations with industry [57] we know that the spe- experimentally determined the specific process time, a proxy for the
cific capital cost of a high salinity ED plant (SpCostED, CapEx, m) is ap- membrane area required to treat a given quantity of brine and the
proximately $600/m2 of membrane. This leads to the total capital costs specific energy required for treatment. The model described in this
being: work was used to simulate each stage test (Fig. 9a and b). As can be
CapExED, total = SpCostED, CapEx, m × Amem, total seen from these figures, the model described here closely matched the
(46)
results from [55]. The average absolute percentage deviation between
Thus the total ED capital cost after annualization is: the model and the experimental data was 13% for specific process time
CapExED, total and 11% for specific energy.
CapExED,yr =
Annuityfactor (47)
3.4.1.2. Industrial stack scale validation. To further validate the ED
From the ED process model, we had calculated the annual ED energy model, we also compared the results predicted by our model against
costs. To obtain the total operating costs, we need to account for costs results from an industrial stack [57]. The variation of concentrate and

104
K.G. Nayar et al. Desalination 456 (2019) 97–120

Table 1
Electrodialysis model parameters.
a
Symbol Value Reference

System inputs
Sc, i, ED Sb, RO –
Sc, o, ED 200 g/kg –
Sd, i, ED Ssw = 35 g/kg –

Solution properties
D 1.61 × 10−9 m2/s [55]
tcu 0.5 [55]
v 8.9 × 10−7 m2/s [55]

Flow properties/geometry
h 0.5 mm –

Membrane parameters
σ, Memeff 0.64 ± 0.03 [55]
r̄m 3.5 × 10−4 ± 1 × 10−4 Ω m2 [55]
Tscp Eq. (30) [55]
Twcp Eq. (31) [55]
Lscp Eq. (32) [55]
L wcp Eq. (33) [55] b
Stack parameters
Vel 2.1 V [55]

Fig. 9. ED model presented in this work validated against experimental data

from [55] for (a) Specific process time and (b) Specific energy. Note that the
specific process time is a proxy for ED membrane area.

Fig. 8. Experimental data on the diluate and concentrate salinities corre-

sponding to stages from [55].

diluate salinities along the length of a single ED membrane is shown

along with the model inputs parameters and output area is shown in
Fig. 10. The model took in input conditions from data given by the ED
manufacturer, Asahi Glass and Chemicals Co., for an ED stack which
had an effective cell pair area of 24 m2. The industrial stack data used as
input conditions for the model were: concentrate salinity in the stack
varying from 39 g/kg to 177 g/kg (corresponding to a concentration
variation from 40 g/L to 200 g/L), an inlet concentrate flow rate of
59 L/h, a total stack inlet flow rate of 1000 L/h, a diluate to concentrate
flow rate ratio at the outlet of the stack (Dil-con-ratioo) of 16 and a
current density of 250 A/m2. For a diluate to concentrate flow rate ratio
at the stack inlet (Dil-con-ratioi) of 80, the model matched the reported
value for Dil-con-ratioo of 16. At this condition, the model predicted the
Fig. 10. Simulating the variation of diluate and concentrate salinities and the
required cell pair area to be 22.7 m2 matching the actual cell pair area
effective salinity from the solution flux along the length of a stack for the given
of the industrial stack to within 5.4%. We were also able to verify the
conditions of an industrial stack [57]. Simulated area closely matches the ef-
predicted membrane area against a full installation of ED stacks. For
fective area reported by the industry.
producing 100,000 tonne/year of salt, our predicted membrane area
matched that reported by the manufacturer [57] to within 11.8%.
Since the dataset from the manufacturer did not report energy [16,61,62]. Tanaka [16] had reported that an industrial ED stack op-
consumption, the energy consumption predictions of the model was erated at 266 A/m2 producing a concentrate at 174 g/L of NaCl (at 90%
separately verified against industrial stack data from the literature purity) had an energy consumption of 165 kWh/tonne. Simulating the

105
K.G. Nayar et al. Desalination 456 (2019) 97–120

for table salt production and 20–40 % for chloralkali grade vacuum salt
[64], reflecting the degree of purity needed for each application. Table
salt only needs to be at least 97% pure [65] while chloralkali grade
vacuum salt must be at least 99.9% pure [66]. For our calculations, we
sought a target salt purity of at least 99.8% and assumed that a crys-
tallizer taking in solution mined feed will have a purge ratio (PR250 g/kg)
of 0.2 or 20%.
The purge ratio of the crystallizer in a ED-crystallizer or RO-ED-
Fig. 11. Control volume showing the crystallizer.
crystallizer is however different from the purge ratio of a crystallizer
taking in solution mined brine as feed. The salinity of the ED con-
same conditions, our model reported an energy consumption of centrate (200 g/kg) is lower than that from solution mining (250 g/kg).
172 kWh/tonne, within 4.7%. Furthermore, the amount of potassium and sulfate in the ED con-
The ED process model was thus verified to be accurate against both centrate is lower than that in solution mined brine. The composition of
laboratory scale and industrial scale data. a typical ED concentrate stream was reported by an ED manufacturer to
have 190 g/L of sodium chloride, 1.2 g/L of calcium and magnesium
3.5. Crystallizer ions and 0.24 g/L of sulfate ions [58].
Since the monovalent selective ED membranes concentrate po-
The crystallizer component of the system separates out salt from tassium in a similar way as sodium, from the amount of sodium we
high salinity brine (Fig. 11). The crystallizer has the subcomponents: an estimated the potassium in the ED concentrate. By our estimates, so-
evaporator, a de-watering unit, and a dryer to produce salt ready for dium chloride represented 93.2% by weight of the salts in the ED
shipment. The critical subcomponent, which consumes the most energy concentrate. We then increased the concentration of the ED concentrate
and is most expensive, is the evaporator. Evaporators for “vacuum salt until the saturation limit of sodium chloride was reached and calculated
production” are either steam driven multi-effect evaporators or elec- the concentration of potassium and sulfate ions to be 4.8 g/L. This was
tricity driven mechanical vapor recompression (MVR) systems [63]. 40% lower than the impurity level in solution mined brine. Thus, the
Both types of evaporators require similar amounts of primary energy at purge ratio for a crystallizer using ED concentrated seawater can be
around 450 kWh-thermal/tonne-salt [63]. For this paper, we assume an 40% lower than that in conventional solution mining. For our concept
electric MVR evaporator for the crystallizer. From conversations with of using an RO-ED-crystallizer, the effective purge ratio was:
the industry [64], we obtained a generic capital cost figure of $15
200g/kg 4.8g/L
million for a MVR based crystallizer system that can produce PR = PR250 g/kg × × = 0.096
250g/kg 8g/L (58)
100,000 tonnes of salt, and an average energy consumption number of
150 kWhe/tonne of salt. These crystallizer systems typically take in The approach above to calculate PR was validated by a crystallizer
nearly saturated brine at 250 g/kg, as opposed to 177–200 g/kg brine manufacturer [64]. We know the capital cost of a crystallizer taking in
delivered by conventional industrial ED systems. Thus the capital cost conventional solution mined brine as feed. What governs the capital
and operating cost numbers need to be adjusted. cost is the evaporation rate of the crystallizer. The evaporation rate of a
In a crystallizer, the feed gets separated in to salt, pure product fixed crystallizer size is constant regardless of the salinity of the feed
water and a purge stream: entering the crystallizer.
ṁ f, Crys. = ṁ p, Crys. + ṁ salt + ṁ purge (53)
⎡ 1000 g/kg 1 − PR ⎤
Sf, Crys. Spurge ṁ evap, Crys. = ṁ p, Crys. = ṁ salt ⎢ ⎛⎜ ⎞×
⎟
Spurge
− 1⎥
ṁ f, Crys. = ṁ salt + ṁ purge ⎢ ⎝ Sf, Crys. ⎠ 1 − PR ⎥
1000 g/kg 1000 g/kg (54) ⎣ Sf, Crys.
⎦
= constant
where,
(59)
ṁ f, Crys. = ṁ c, o, ED (55)
The amount of salt produced by the crystallizer varies with the salinity
For our calculations we assumed that the purge stream is saturated with of the feed entering the crystallizer. This leads to the specific capital
a salinity of 250 g/kg: costs and operating costs increasing inversely with the salt production
rate. The increase in specific cost or decrease in salt production for a
Spurge = 250 g/kg (56) crystallizer using a feed different from that in solution mining is cap-
The ratio of the purge stream to the feed stream is given by the purge tured by the crystallizer salinity scale-up factor (CSSF):
ratio (PR),

PR =
ṁ purge
ṁ salt, 250 g/kg
( 1000 g/kg
Sf, Crys. )× 1 − PR
1 − PR
Spurge
Sf, Crys.
−1

ṁ f, Crys. (57) CSSF = =

which is determined by the composition of the feed and the required

ṁ salt
( 1000 g/kg
250 g/kg )× 1 − PR250 g/kg

1 − PR250 g/kg
Spurge −1
Sf, Crys. (60)
purity of the salt. For salt production from solution mining, the feed
coming in to the crystallizer is typically saturated with sodium chloride For feed salinity of 200 g/kg, PR250g/kg of 0.2 and PR of 0.096, CSSF was
and has about 8 g/L of impurities of potassium and sulfate ions. Lime 1.38. Accounting for the plant capacity factor, the annual salt produced
and soda ash are used to reduce the levels of calcium and magnesium to in tonnes is:
levels on the order of mg/L leaving only potassium and some residual 86400 s/day × 365 day/year
sulfates behind. ṁ salt,yr = ṁ salt × × Capfac
1000 kg/tonne (61)
At typical evaporator operating temperatures, potassium chloride is
more soluble than sodium chloride and the purge ratio is used to control Thus, the capital expense of the crystallizer is
the potassium concentration in the salt stream. A high purge ratio leads $15, 000, 000
to a lower build-up of potassium and sulfates in the purge stream which CapEx Crys. = × ṁ salt,yr × CSSF
100, 000 tonne/yr (62)
directly leads to a more pure salt stream. For conventional solution
mining based salt production, the purge ratio (PR250g/kg) is around 15% and the specific capital cost of the crystallizer is:

106
K.G. Nayar et al. Desalination 456 (2019) 97–120

CapEx Crys. 3.6.1. Overall cost validation

SpCapEx Crys. =
ṁ salt, yr Annuityfactor (63) The RO specific cost model was validated against DesalData. The
ED-crystallizer specific cost model was validated against production
The specific energy consumption of the crystallizer adjusted for the cost figures reported in the literature [29]. Miyake et al. [29] had re-
lower feed salinity is: ported the production cost from a 250,000 tonne/year ED-crystallizer
SpECrys. = 150 kWh e/tonne × CSSF plant to be 100 $/tonne-salt. The operating current density was given to
(64)
be 300 A/m2 while the electricity price reported was 5 cents/kWh e.
leading to a specific energy cost for the crystallizer of: Since project life and a rate of return was not reported, we assumed a
20 year project life and factored in the lower interest rates in Japan and
SpOpEx Crys., energy = SpECrys. × Cost elec (65) assumed a rate of return of 5%. For these inputs, our model reported a
Crystallizer systems of the capacity we are considering typically production cost of 109 $/tonne-salt coming within 9% of the figure
have two full-time plant operators and a process chemist to ensure the reported by Miyake et al. [29]. Thus, the overall RO-ED-crystallizer
quality of the salt being produced [64]. The average salary for a plant specific model was verified to be reasonably accurate.
operator in the USA is $50,000 [60] while that of a process chemist is
$65,000 [67] bringing the annual labor cost of the crystallizer system 3.7. Salt and brine purity
(OpExCrys., labor, yr) to $165,000. The cost of replacing parts, main-
tenance and chemicals for the crystallizer system was obtained from the In this paper, the composition of feed seawater we used for our
literature [68] from costs reported for a zero liquid discharge system for analysis was the reference composition of standard seawater from
treating 1 million gallons per day of feed (3785 m3/day). Normalizing Millero et al. [40]. The proportion of sodium chloride in seawater by
these costs per unit feed, the specific annual cost of parts, maintenance weight is 77.9%. Concentration of seawater by RO does not change the
and chemicals was 273, 164 and 73 $/m3 of feed. Together, the specific purity of brine significantly. However, the ED system with monovalent
annual cost of overall maintenance and chemicals for the crystallizer selective membranes increases the amount of sodium chloride in the
(SpCost Crys., maint., chem. m3/day ) was 510 $-day/m3 of feed to the crystal- brine during the concentration process. Based on public data reported
lizer, resulting in an annual cost of parts, maintenance and chemicals by an ED manufacturer [58], a standard ED brine concentrator with
of:. monovalent selective membranes concentrates sodium chloride such
that the final ED concentrate has 190.1 g/L of sodium chloride, 1.2 g/L
OpEx Crys., maint., chem., yr= SpCost Crys., maint., chem. m3/day of calcium ions, 1.2 g/L of magnesium ions and 0.24 g/L of sulfate ions.
× V̇ f, Crys., m3/day × 365 day/year × Capfactor The corresponding proportion of sodium chloride in the ED concentrate
was by our calculation was 93.2%. The crystallizer with its purge ratio
(66)
further separates sodium chloride from other constituents in the brine
Thus, the total specific operating cost of the crystallizer was: which leaves through the purge stream. Once the salt is dried, the
purity of the dry salt obtained from an ED-crystallizer is at least 99.8%
SpOpEx Crys., tot.= SpOpEx Crys., energy
pure. The salt produced by the RO-ED-crystallizer is also expected to
OpEx Crys., maint., chem., yr + OpEx Crys., labor, yr have the same degree of purity. It must be noted that the purity of the
+
ṁ salt, yr (67) salt produced is not sufficient for direct use in an electrolyzer for
chloralkali production. Further purification of the produced wet salt is
which brings us to the specific cost of the crystallizer normalized to salt needed for use in chloralkali production.
production as:
SpCostCrys. = SpCapEx Crys. + SpOpEx Crys., tot. (68) 4. Parametric design of hybrid RO-ED systems for brine
concentration for salt production
The cost predictions of the crystallizer model described above was
independently verified with an ED manufacturer [57]. This section discusses the effects of hybridizing RO with ED, ED
current density on RO-ED design, electricity prices and water prices.
3.6. Overall cost and energy required for RO-ED-crystallizer based salt
production system 4.1. Hybridizing RO with ED reduced costs and energy required for brine
concentration for salt production
From the specific cost expressions for the RO system, the ED system
and the crystallizer, we can obtain a total specific cost of the overall Hybridizing RO with ED all the way up to 120 g/kg RO brine sali-
system: nity (i.e., “fully hybridizing”) while operating the ED system at con-
ventional operating conditions (described later in the paragraph) re-
SpCostREC = SpCostRO + SpCostED + SpCost Crys. (69) duced the specific cost of brine concentrations by 7%. Fig. 12 shows the
specific cost of the RO-ED brine concentration system varying with RO
It must be noted that specific cost of the overall system does not include
hybridization. Hybridization with RO is represented on the x-axis by the
the cost of packaging, iodizing and storage facilities. From a ED-crys-
salinity of the concentrate inlet to the ED system (salinity of brine
tallizer plant operator, we know that the purity of the salt produced is
leaving RO). The leftmost column corresponding to a 35 g/kg ED con-
at least 99.8%. The overall specific cost reported in this paper is for
centrate inlet salinity represented a standalone ED system with the
producing 99.8% pure salt.
other columns representing RO-ED hybrid systems. For each of the RO-
The water production from the RO system reduces the overall cost
ED configurations, the ED current density was fixed at 300 A/m2, the
of production of salt. The revenue from the water produced by the RO
cost of electricity was 0.1$/kWhe, the Dil-con-ratioi selected was fixed
system can be normalized to obtain a specific revenue in terms of salt
at 80, and the concentrate salinity leaving the ED stack was fixed at
specific revenue per tonne of salt produced:
200 g/kg. The specific cost of concentrating brine from 35 g/kg to
̇
SpRevREC = Vp,RO,yr × Price water (70) 200 g/kg using a standalone ED system was $89/tonne-salt while an
RO-ED hybrid system with RO brine at 120 g/kg cost only $82/tonne-
The overall specific energy required for the RO-ED-crystallizer was: salt — a cost reduction of 7%. In comparison, partial hybridizing of RO
SpEREC = SpERO + SpEED + SpECrys. with ED by stopping at the limits of operation of conventional seawater
(71)
RO (i.e., RO brine salinity of 60 g/kg) cost $87/tonne-salt — a cost

107
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 12. Specific cost of RO-ED systems decreases with increasing RO hy- Fig. 14. Specific costs, of an RO-ED system with an RO brine salinity of 120 g/
bridization (i.e., with increasing ED concentrate inlet salinity). kg, broken down across ED and RO capital and operating costs.

reduction of only 2%. labeled in the figure as well. ED capital costs contributed the most at
Hybridizing RO with ED all the way up to 120 g/kg RO brine sali- $38/tonne-salt corresponding to 47% of the total cost followed by the
nity while operating the ED system at conventional operating condi- operating cost which was $40/tonne-salt corresponding to 49% of the
tions reduced the specific energy consumption for brine concentration total cost. The operating costs consisted of $18/tonne-salt in energy
by 13%. Fig. 13 shows the specific energy consumption of RO-ED brine costs, $14/tonne-salt in membrane replacement costs and $8/tonne-salt
concentration varying with RO hybridization. Just as in Fig. 12, the for other maintenance, chemical and labor expenses. If the ED mem-
degree of hybridization is captured in the x-axis by the salinity of the brane replacement costs are combined with the initial ED capital costs,
concentrate inlet to the ED system. The parameters held constant in they together amount to $53/tonne-salt or 64% of the total cost. RO as
each configuration in Fig. 13 are identical to the ones described in a whole contributed to only around $4/tonne-salt corresponding to 5%
Fig. 12. The leftmost column corresponded to a standalone ED system. of the total cost.
Each column shown in the Fig. 13 represents an RO-ED hybrid con-
figuration, with the parameters held constant being the same as in 4.1.2. RO was more cost-effective than ED at concentrating seawater
Fig. 12. While a standalone ED system needed 219 kWhe/tonne-salt to While RO contributed only 5% to the total specific cost of an RO-ED
concentrate brine from 35 g/kg to 200 g/kg, a RO-ED hybrid system system with 120 g/kg RO brine salinity, the RO sub-system was re-
with RO brine at 120 g/kg consumed only 191 kWhe/tonne-salt for the sponsible for 52% of the total salinity increase, concentrating seawater
same brine concentration load. The reduction in specific energy con- from 35 g/kg to 120 g/kg. However, since RO and ED systems work
sumption seen in RO-ED hybrid systems with increasing hybridization differently, the salinity change does not directly equate to salt trans-
with RO highlights that RO is more energy efficient than ED at con- ferred. In an RO system, strictly speaking no salt transfer occurs, only
centrating brine from 35 g/kg to 120 g/kg when the current density in water transfer; in an ED system, both salt and water are transferred
the ED system is 300 A/m2. from the diluate to the concentrate. To effectively compare the brine
concentration abilities of the two systems, we imagined replacing the
4.1.1. ED capital and membrane costs contributed the most to RO-ED RO system with an “ideal blackbox salt concentrator” that increases
specific costs salinity while keeping the total mass flow rate constant. The “ideal
The distribution of the specific cost of an RO-ED system with an RO blackbox salt concentrator” delivered the same output brine of salinity
brine salinity of 120 g/kg across the capital and operating costs of the Sb,RO and mass flow rate ṁ b,RO from seawater with the same feed sali-
RO-ED system is shown in Fig. 14. The individual cost contributions are nity Sf,RO and assumed unchanged mass flow rate ṁ b,RO . Thus, the
“effective salt transferred” by the “ideal blackbox salt concentrator”
that replaced the RO system was:
Δ Salteff, RO = ṁ b,RO (Sb,RO − Sf,RO) (72)

The “effective salt transferred” by the ED system was the same as the
actual salt transferred by it:
Δ Salteff, ED = Δ SaltED = ṁ c,o,ED Sc,o,ED − ṁ c,i,ED Sc,i,ED (73)

Adding the effective salt transferred by the RO and ED systems we have

the total effective salt transferred as:
Δ Salteff, RO-ED = ṁ c,o,ED Sc,o,ED − ṁ c,i,ED Sf,RO (74)

The percentage of the “effective salt transferred” in the RO and ED

systems along with the percentage salinity change from each system is
shown in Fig. 15. While RO contributed to 52% of the salinity change, it
only contributed to 9% of the effective salt transferred in the brine
concentration process. Even with the RO system providing brine at
Fig. 13. Specific energy consumption of RO-ED systems reducing with in- 120 g/kg for the ED system to concentrate to 120 g/kg, 91% of the salt
creasing RO hybridization represented by increasing ED concentrate inlet in the ED concentrate outlet was transferred by the ED system from the
salinity. ED diluate stream.

108
K.G. Nayar et al. Desalination 456 (2019) 97–120

consumption to the effective salt transferred:

ṁ salt
NormSpERO = SpERO ×
Δ Salteff, RO (77)

ṁ salt
NormSpEED = SpEED ×
Δ Salteff, ED (78)

For an RO-ED system operating with a RO brine salinity of 120 g/kg and
at a ED current density of 300 A/m2, the ED system had a normalized
specific energy consumption of 182 kWhe/tonne-salt while the RO
system had a normalized specific energy consumption of only 85 kWhe/
tonne-salt. Thus, on “effective salt transferred” basis, RO was slightly
more than twice as energy efficient as conventional ED (i = 300 A/m2)
at concentrating seawater. This is why hybridizing ED with RO led to
reductions in the specific energy consumption seen in Fig. 13.

4.2. Revenue from co-production of water increased cost-effectiveness of

Fig. 15. Percentage of salinity change and salt transferred in RO and ED sub- hybridizing RO with ED by 1–6%
systems for a RO-ED system with RO brine at 120 g/kg salinity.
Unlike a standalone ED system which only produces 200 g/kg brine
To effectively compare the cost-effectiveness of RO and ED sub- as the useful product, RO-ED based brine concentration systems also
systems, we normalized the specific cost of the RO and ED sub-systems produce pure water, a useful byproduct. Water prices and the cost of
to the effective salt transferred in each sub-system: water production have a wide distribution. Research conducted by IBM
ṁ salt [75] reported that the real cost of water production after factoring in
NormSpCostRO = SpCostRO × water wastage, varied from 0.34 $/m3 in Manila, Philippines to 2.26 $/
Δ Salteff, RO (75)
m3 in London, United Kingdom with a global weighted index cost of
ṁ salt 1.31 $/m3. Zetland and Gasson [39] analyzed water tariffs from 308
NormSpCostED = SpCostED ×
Δ Salteff, ED (76) cities around the world and reported that water tariffs ranged from 0 to
7.54 $/m3 with a global average water tariff of 1.21 $/m3. Table 2
Fig. 16 shows the salt transfer normalized specific cost of the RO further gives the cost of water produced at various desalination plants
and ED sub-systems for an RO-ED system operating with a RO brine around the world. To account for the large spread of water costs and
salinity of 120 g/kg. While the ED system had a normalized specific cost water prices around the world, we calculated the potential revenue that
of $78/tonne-salt, the RO system had a normalized specific cost of $37/ can be generated from co-production of water from an RO-ED brine
tonne-salt. This effectively meant that on a “effective salt transferred” concentration system at water prices ranging from 0.25 $/m3 to 2 $/m3.
basis, ED was twice as expensive as RO. To confirm this, we used our Fig. 17 shows the water revenue from a RO-ED brine concentration
models and simulated the cost of producing a fixed amount of 120 g/kg system normalized to the quantity of salt produced, with the water
brine from seawater, using first an RO system and then an ED system. revenue also represented as a percentage of standalone ED brine con-
Our simulations again showed that RO could produce 120 g/kg brine centration costs. The RO-ED configuration simulated here had an RO
from seawater at around half the cost of an ED system producing the brine salinity of 120 g/kg, with other parameters same as in Section 4.1.
same amount of brine. Thus, depending on the geographic location, co-production of water
makes an RO-ED system more cost-effective than a standalone ED
4.1.3. RO was more energy efficient than conventional ED (i = 300 A/m2) system by 1% to 6%. When the revenue from the RO system is coupled
at concentrating seawater with the cost reduction of 7% reported in Section 4.1, RO-ED hybrid
To effectively compare the effective energy use of the RO and ED systems can be more cost-effective than a standalone ED system by 8%
sub-systems, we used the same approach used to compare cost-effec- to 13%.
tiveness. The normalized specific energy consumption of the RO and ED
sub-systems was obtained by normalizing the specific energy
4.3. Optimizing ED current density further reduced RO-ED brine
concentration costs

From Fig. 14, we know that for a fully hybridized RO-ED system
(i.e., RO brine salinity at 120 g/kg), the biggest contributor to the cost is

Table 2
Production cost of water at various desalination plants across the globe.
Country, City Year commissioned Capacity (m3/ Cost Ref.
day) ($/m3)

Singapore 2005 136,000 0.45 [69]

Israel, Soreq 2013 410,000 0.52 [70]
Israel, Ashkelon 2005 296,000 0.51 [70]
USA, Tampa 2004 86,000 0.68 [70]
Australia, Perth 2006 126,000 0.75 [70]
Kuwait – 1128 0.84 [71]
Saudi Arabia, 2011 127,800 1.03 [72]
Fig. 16. Normalized specific cost of RO and ED sub-systems, in an RO-ED Yanbua
India, Chennai 2010 100,000 1.03 [73]
system with a RO brine salinity of 120 g/kg. The specific cost was normalized
Japan, Fukuoka 2005 50,000 1.9 [74]
per tonne of salt transferred in each sub-system.

109
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 17. Variation of specific revenue from water production at different water Fig. 19. Variation of specific energy of ED and RO-ED systems with current
prices. density.

the capital cost and the membrane replacement cost of the ED sub- energy consumption was around 300 A/m2. This difference is high-
system. From Fig. 14 it can be further seen that the energy cost was only lighted in Fig. 19 which captures the variation of specific energy with
34% of the capital and membrane replacement cost, implying that there ED current density for the systems shown in Fig. 18. Tanaka et al. [38]
was room for optimizing the trade-off between capital and energy costs. had previously reported that for both a standalone ED system and an
As discussed in Section 2, the capital cost can be reduced by increasing RO-ED system where the RO system was a conventional seawater RO
ED current density which increases the salt removed per m2 membrane system with a brine outlet of around 60 g/kg, the current density that
and reduces the membrane area required for a fixed salt production reduced the specific energy consumption was 300 A/m2. While the RO-
capacity. ED configurations in this paper used seawater as a diluate input and
Fig. 18 shows how ED current density affects the specific cost of a Tanaka et al. used RO brine as diluate input, our findings on optimal
standalone ED system and two RO-ED systems, where one RO system energy consumption match. We found that 300 A/m2 is the ED current
operates like conventional seawater RO producing brine at 60 g/kg and density that minimized specific energy consumption for both a stan-
the other RO system is fully hybridized operating at the highest pres- dalone ED system and an RO-ED system with RO brine salinity of 60 g/
sure of around 120 bar corresponding to a RO brine salinity of 120 g/ kg (i.e. conventional seawater RO). For a fully hybridized RO-ED
kg. The other parameters had the same values reported in Section 4.1. system with a RO brine salinity of 120 g/kg, the ED current density that
For reference, the electricity cost used for this result was $0.1/kWhe. It minimized specific energy consumption was slightly less, at 280 A/m2.
can be seen that from a specific cost point of view, the optimal current
density of operation is 600 A/m2. For an RO-ED system with RO brine
salinity of 120 g/kg, the specific cost reduced by 27% when the current 4.4. Effect of electricity prices on optimal ED current density
density was increased from 300 A/m2 to 600 A/m2. From 300 A/m2 to
600 A/m2, such a sharp reduction in specific cost was seen because the The optimal ED current density that minimizes the specific cost of
reduction in membrane area reduced the ED capital and membrane RO-ED sub-systems is also dependent on the electricity prices. In this
replacement costs far more than the associated increase in energy costs. paper, we have selected an electricity driven crystallizer which is also
Beyond 600 A/m2, the specific costs start increasing as the energy cost affected by electricity prices. Fig. 20 shows the variation of specific cost
begins to dominate overall costs. for the RO-ED-crystallizer system with ED current density and elec-
Although the optimal current density from a specific cost point of tricity prices. The RO-ED configuration represented here is a fully hy-
view was 600 A/m2, the current density that minimized the specific bridized RO-ED system with RO brine salinity of 120 g/kg. The lowest
electricity price shown here of 1.6 US cents/kWh e corresponds to the
industrial electricity price in Kuwait, where electricity is subsidized.

Fig. 18. Variation of specific cost of ED and RO-ED systems with current den- Fig. 20. Variation of specific cost with current density at different electricity
sity. prices.

110
K.G. Nayar et al. Desalination 456 (2019) 97–120

brine salinity is 120 g/kg reduced specific costs another 7%. At lower
electricity prices of 5 US cents/kWhe and 1.6 US cents/kWhe, opti-
mizing ED current density and the reduced RO-ED operating expenses,
led to specific cost reductions of 19% and 37% respectively. The cu-
mulative effects of each cost reduction step on the specific cost of brine
concentration is highlighted in the last column in Table 3. The revenue
generated by water is also accounted for in the last rows of the table
with water prices of 0.5 $/m3, 1 $/m3 1.5 $/m3 and 2 $/m3, leading to
the effective specific cost reducing further by 2%, 3%, 5% and 6% re-
spectively. At a fixed electricity price, the effective specific cost re-
ductions in order of magnitude of contribution were from: optimizing
ED current density (26%), fully hybridizing ED with RO up to a RO
brine salinity of 120 g/kg (7%) and producing water (2–6 %).
Fig. 22 adds the specific cost of the crystallizer component to the
brine concentration systems shown in Fig. 21, showing the total specific
costs of an RO-ED-crystallizer system for various cases along with the
Fig. 21. Specific costs of RO-ED systems optimized for various global cases with
percentage reduction in the specific costs when compared with a
percentage reduction in costs compared to standalone ED case. standalone ED-crystallizer configuration. With the crystallizer, a stan-
dalone ED system operated for a global average case with an electricity
price of 10 US cents/kWhe produced salt at a cost of $137/tonne-salt. A
From Fig. 20, we can see that the optimal ED current density that
fully hybridized and current density optimized RO-ED system with a
minimizes the total specific cost decreases with increasing electricity
crystallizer produced salt at a cost of $111/tonne-salt, corresponding to
price. For electricity prices of 1.6, 5, 10, 15 and 20 US cents/kWhe, the
a production cost decrease of 19%. For the special case of Kuwait, when
optimal ED current densities correspondingly were 1350, 800, 600, 500
the subsidized electricity prices are included in the system design and
and 450 A/m2.
cost calculations, the total cost of producing salt using an optimized RO-
ED-crystallizer was $61/tonne-salt. A conventional standalone ED-
5. Optimal RO-ED hybrid designs significantly reduced the cost of crystallizer when operated in Kuwait produced salt at a cost of $101/
brine concentration and salt production tonne-salt. Thus, for the special case of Kuwait, the optimized RO-ED-
Crystallizer produced salt at a cost 55% lower than a conventional
When all the parameters discussed in Section 4 are considered to- standalone ED-crystallizer system operated at 10 US cents/kWhe and
gether, fully hybridized and current density optimized RO-ED systems 40% lower than one operated at Kuwaiti electricity prices. Thus, hy-
can cost 33–70 % less than standalone ED brine concentrators for salt bridizing RO with ED up to 120 g/kg, RO brine salinity and optimizing
production application. When including the crystallizer, optimized RO- ED current density for electricity prices reduced the cost of salt pro-
ED-crystallizers can cost 19–55 % less than standalone ED-crystallizer duction significantly.
based seawater salt production systems.
Fig. 21 shows the specific costs of brine concentration for standa- 6. Global feasibility of RO-ED hybrid systems for concentrating
lone ED and optimal RO-ED designs, along with the percentage re- seawater for salt production
duction in costs of RO-ED systems from that of standalone ED systems.
For all brine concentration systems shown, the feed was seawater at Based on production costs discussed in Section 5 and market prices
35 g/kg and the final concentration was 200 g/kg. All the RO-ED sys- of salt listed in Table 4 obtained from salt manufacturers and national
tems shown are fully hybridized with the RO brine being at the highest government agencies, we believe that RO-ED hybrid brine concentra-
possible salinity of 120 g/kg. The cases shown are: a global average case tion systems for salt production are technically and economically fea-
corresponding to an electricity price of 10 US cents/kWhe, a low elec- sible in parts of Japan, the Middle East, Europe and the United States.
tricity price case of 5 US cents/kWhe and the special case of Kuwait The economic feasibility of each region depends on its unique cir-
where the electricity price is 1.6 US cents/kWhe. A standalone ED cumstances that arise from the following conditions:
system designed for a global case, concentrated seawater at a cost of
$89/tonne-salt, while a fully hybridized RO-ED system cost 33% less at (a) High market demand for edible salt or industrial salt by-products
$60/tonne-salt. In Kuwait, an optimal RO-ED system can take ad- such as chlorine and sodium hydroxide in the region
vantage of the subsidized electricity price to concentrate brine for just (b) High transportation costs from available resource to point of con-
$27/tonne-salt. A conventional standalone ED system designed for sumption (road and shipping costs)
higher electricity prices even with the lower electricity costs in Kuwait (c) Few natural salt deposits
still had a cost of $70/tonne-salt (capital cost contribution was $43/ (d) Lack of land availability
tonne-salt). Thus, the optimal RO-ED system cost 70% less than a (e) Strategic national security reasons
conventional standalone ED system operated at a electricity price of 10 (f) Scarcity of drinking water
US cents/kWhe and 62% less than one operated at a electricity price of
1.6 US cents/kWhe. Thus, hybridizing RO with ED up to 120 g/kg RO The presence of one or more of these indicators substantially in-
brine salinity and optimizing ED current density for electricity prices creases the price for vacuum salt in the region. Fig. 23 illustrates the
reduced the cost of brine concentration significantly. different salt prices found throughout the world. To evaluate the eco-
Table 3 highlights the critical cost reducing steps and the associated nomic feasibility of the proposed RO-ED-crystallizer concept in a
specific cost reductions that led to the outcome shown in Fig. 21. The country, we compared the production cost of an optimal RO-ED-crys-
baseline case was a standalone ED system operated in a conventional tallizer, $111/tonne-salt reported in Section 5, against the market price
case at a current density of 300 A/m2 and electricity price of 10 US of salt. For Kuwait, we used a production cost of $61/tonne-salt ac-
cents/kWhe, with the specific cost of $89/tonne-salt. Optimizing the counting for the lower electricity price in Kuwait. We reiterate here that
current density of a standalone ED system for an electricity price of 10 salts depending on their production method have different production
US cents/kWhe reduced the specific cost of brine concentration 26%. costs with vacuum salt being significantly more expensive than solar
Fully hybridizing an ED system with an RO system such that the RO salt, and this fact is reflected in their market prices as well. The market

111
K.G. Nayar et al. Desalination 456 (2019) 97–120

Table 3
Cost reducing steps and the associated specific cost reductions in RO-ED systems corresponding to cases in Fig. 21. The last column shows the cumulative percentage
reduction from the baseline case caused with each cost reducing step.
System Cost reducing step SpCost ΔSpCost ΔSpCost ΔSpCost
($/tonne) ($/tonne) (%) total (%)

Standalone ED (global) Baseline case $89 $– 0% 0%

Standalone ED (global) Optimizing idensity $66 $23 26% 26%
RO-ED (global) Fully hybridizing with RO to 120 g/kg $60 $6 7% 32%
RO-ED (5 ¢/kWh e) Optim. for 5 ¢/kWhe $43 $17 19% 52%
RO-ED (Kuwait) Optim. for 1.6 ¢/kWhe $27 $16 18% 70%
Including water revenue
RO-ED (Kuwait) Water price at $0.5/m3 $25 $1 2% 71%
RO-ED (Kuwait) Water price at $1/m3 $24 $3 3% 73%
RO-ED (Kuwait) Water price at $1.5/m3 $23 $4 5% 74%
RO-ED (Kuwait) Water price at $2/m3 $21 $5 6% 76%

is given in units of US cents per tonne per km, as well as in units of

$/tonne for transporting across 500 km, 1000 km and 1500 km. We
observed from market reports in the literature that there was a wide
range of transportation costs especially for road transportation costs.
However, market reports in the literature did not specify the metho-
dology and sample size used to obtain the transportation figures, and
made it difficult to obtain an accurate average estimate of the cost of
transportation. The most reliable averaged transportation data we
found was from the US DOT. For reliability of our predictions, we re-
stricted our transport cost estimation to this dataset. Transportation by
road at 10.34 US cents/tonne-km, was the most expensive mode of
transport, with transportation by rail and barge being 1.87 and 1.14 US
cents/tonne-km respectively.
For understanding the impact of transportation costs on the feasi-
bility of RO-ED-crystallizer systems for salt production, the production
costs for conventional-solution-mining based vacuum salt and solar salt
Fig. 22. Specific costs of RO-ED-crystallizer systems optimized for various need to be discussed. Solar salt production costs vary between $5–10
global cases with percentage reduction in costs compared to standalone ED- /tonne-salt. For our analysis work in this section, we have assumed a
crystallizer case producing 99.8% pure salt. production cost of $10/tonne-salt. In conventional-solution- mining
based vacuum salt, near saturated brine is sent to a conventional
crystallizer. From Section 3.5 on crystallizer costs, we have calculated
prices of vacuum salt are well above the salt production cost from the
the production cost of conventional solution mining based vacuum salt
proposed RO-ED-crystallizer concept in Cyprus, Japan, Saudi Arabia
to be $35/tonne-salt. From Section 5, we know that the cost of pro-
and the USA. The RO-ED-crystallizer concept may likely be competitive
ducing salt from seawater through an optimized RO-ED-crystallizer
in these countries. It must be also noted that solar salt prices are far
system for a typical global case is $111/tonne-salt. Given this in-
lower than the production cost of the RO-ED-crystallizer concept.
formation and the transport costs mentioned in Table 5, we calculated
In this section, we further discuss some interesting dynamics related
the break-even distance at which the seawater based RO-ED-crystallizer
to salt pricing and transportation costs, that highlights potential op-
system we have proposed becomes cost competitive with current
portunities for the RO-ED-crystallizer concept and new technologies for
methods. The break-even distances are also shown in Table 5. It is quite
producing salt from seawater. We further discuss the specific feasibility
clear that the seawater-based optimized RO-ED-crystallizer systems
of the RO-ED-crystallizer concept for the USA and Kuwait.
become truly competitive when salt is transported by road as compared
to rail and barge, with these systems being competitive at a distance of
6.1. High transportation costs can create spot market opportunities
735 km from conventional solution mining and at a distance of 977 km
from existing solar salt production locations.
The economic competitiveness of a salt production method is in-
Another aspect related to transportation costs is the cost of shipping
fluenced heavily by the cost of transporting salt between the manu-
salt between countries by sea, which is different from “barge” transport.
facturing facility and the customer, especially the cost of transporting
Shipping costs between countries don’t scale directly with distance. The
salt by road. Table 5 shows the average cost for transporting a tonne by
volume of trade between countries and whether containers return
road, rail and barge. The average cost of transportation obtained from
empty or full affects costs more than distance. The limited relationship
data published by the US Department of Transportation (US DOT) [5],

Table 4
Salt prices from around the world.
Country Purity % Type of salt, Degree of processing Price ($/tonne) Min. order, Packaging, Distrib. level Ref. Is RO-ED Feasible ?

China 99.2% Vacuum salt (Domestic), refined 90 25 tonnes, 50 kg bags, ex-works [76] No
Cyprus 99.5% Solar salt (Domestic), refined 200 n.a. [77] Yes
India 99.3% Solar salt (Industrial), refined 80 28 tonnes, 50 kg bags, ex-works [78] No
Kuwait 97–99% Raw salt (mix of solar and vacuum) 100 Not defined [79] Yes
Saudi Arabia 99.8% Vacuum salt (Industrial), refined 190 25 tonnes, 1000 kg bags, ex-works [80] Yes
USA n.a. Vacuum salt 190 avg. f.o.b. plant [81] Yes

112
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. 23. Prices for salt around the world (includes both vacuum and solar salt).

with distance is clear from Table 6 which shows the cost of shipping 20 Table 6
foot containers between select ports around the world. It must be noted Container shipping costs between ports and countries, with the costs translated
here that the routes reported in Table 6 are not in anyway re- into cost per tonne of salt. Routes and associated costs shown are not indicative
presentative of popular salt shipment routes and are only shown to of popular salt shipping routes and are shown only to convey general variation
highlight that distance does not affect shipping costs as much as in in shipping costs.
other transportation modes. For example, despite the vastly different Origin to destination Cost Ref. Cost per tonne
distances, shipping a container from New York, USA to Izmir, Turkey
costs almost as much as shipping from Barcelona, Spain to Izmir, City to city Country to ($/container) ($/tonne)
country
Turkey. Furthermore, shipping from New York, USA to Shanghai, China
is still cheaper than the examples mentioned previously. For con- China to USA $400 [82] $16
textualizing the container costs, a full 20 foot container can hold India to USA $800 [82] $32
25 tonnes of salt [84]. The cost of shipping from one port to another in New York to Haifa USA to Israel $1478 [83] $59
Miami to Haifa USA to Israel $1757 [83] $70
units of USD per tonne of salt is also given in Table 6. In general, the New York to Izmir USA to Turkey $1025 [83] $41
cost to ship salt between countries can contribute $10–80 /tonne-salt to Barcelona to Izmir Spain to Turkey $978 [83] $39
the market price of salt, depending on the ports considered. New york to USA to China $748 [83] $30
From Tables 5 and 6 it can be seen that the difference in production Shanghai
Valencia to Shanghai Spain to China $657 [83] $26
costs between the salt produced from an optimized RO-ED-crystallizer
system and solar salt and conventional solution mined vacuum salt, can
be made up in specific circumstances when salt needs to be transported
that the RO-ED-crystallizer concept can create for the small-scale de-
over long distances by road and between specific ports where shipping
centralized production of salt. Both RO and ED systems can be scaled
costs are high.
down to installations that can produce 10,000–50,000 tonnes per year
of salt close to customers without significant changes to the brine
concentration costs discussed in Section 5. However, the specific cost of
6.2. Small scale decentralized production of salt has economic potential
crystallizers do increase as production capacity reduces and de-
termining the minimum practical capacity of a crystallizer is beyond the
Table 4 showed that the RO-ED-crystallizer concept can be feasible
scope of this work. Nevertheless, we believe that small RO-ED-crystal-
in certain countries based on market price of salt. From the discussion
lizer salt production plants by the sea could generate significant amount
in the previous section, the RO-ED-crystallizer concept for producing
of savings for large customers of salt as a result of significant savings on
salt in certain regions is high due to high salt transportation costs. What
transportation and shipping costs.
follows naturally from these observations is the potential opportunity

Table 5
Average cost of transportation by road, rail and barge, with distance where RO-ED-crystallizer costs break-even with the production cost of solution mined [2]
vacuum salt ($ 35/tonne-salt assumed) and solar salt ($ 10/tonne-salt assumed).
Transportation cost Break-even distance (km)

Mode Transportation cost Ref. 500 km 1000 km 1500 km Vacuum salt Solar salt
(Solution mined)

US ¢/tonne-km $/tonne $/tonne $/tonne km km

By road 10.34 [5] 52 103 155 735 977

By rail 1.87 [5] 9 19 28 4064 5401
By ship 1.14 [5] 6 11 17 6667 8860

113
K.G. Nayar et al. Desalination 456 (2019) 97–120

6.3. Case studies • Our analysis showed that hybridizing ED with RO up to RO brine
salinities of 120 g/kg will reduce costs. The cost advantage in our
6.3.1. USA opinion justifies a pilot trial of the proposed configuration. For pilot
The average market price for vacuum salt in the USA is $190/tonne- testing, achieving higher current densities in the ED system should
salt [81] which is far above the production cost of $111/tonne-salt from be prioritized over hybridizing with RO.
the RO-ED-crystallizer concept. The lack of available natural salt supply • While the focus of our analysis was on concentrating seawater and
in some parts of the USA along with the high transportation costs producing salt, the RO-ED-crystallizer concept could also be used for
present an opportunity for RO-ED along the west coast and the south- producing salt from saline inland lakes. However, saline inland lakes
east coast of the USA where there are few resources to produce vacuum have compositions very different from seawater. Hence, the feasi-
salt. The dearth of resources results in tonnes of salt being trucked and bility of RO-ED-crystallizers for producing salt from saline inland
sent via railroad. Transporting salt by road, the most common method, lakes should be examined on a case to case basis.
adds $52/tonne for 500 km of travel. The creation of a couple of
10,000–50,000 tonne plants could lower the purchasing price of salt in 8. Conclusions
this part of the USA and change the market dynamic within those re-
gions. This suggests a reasonable commercial potential for the RO-ED- 8.1. Brine concentration
crystallizer concept in the USA.
RO systems are more cost effective at concentrating seawater than
6.3.2. Kuwait ED systems, and hybridizing ED with RO reduced brine concentration
Kuwait, like other countries in the Middle East, has abundant costs. However, the ED systems should be fully hybridized with the
sources of hydrocarbons needed for production of plastics, petro- emerging high pressure RO technology, up to operational limits of an
chemicals, organic intermediates, etc., which require chlorine for or- RO brine salinity of 120 g/kg for significantly reducing costs (7%).
ganic synthesis. Vacuum salt is the raw material used for chlorine Partial hybridizing of RO and ED systems, with the RO brine salinity
production, and thus Kuwait and other Middle Eastern countries have a restricted to the operational limit of conventional seawater RO of 70 g/
high local demand for salt. Kuwait already has a standalone ED plant kg brine salinity, maybe considered only if water prices are sufficiently
that concentrates brine for salt production for chloralkali production high as the cost reductions alone over standalone ED are small (2–3 %).
[14] and which is a good candidate for adopting the proposed RO-ED- ED capital costs are the main contributor to the cost of brine con-
crystallizer concept for producing salt from seawater. From our litera- centration and increasing ED current density can reduce ED capital
ture review, we were able to obtain an import price of “raw salt” of costs significantly with there being a cost optimal value for ED current
$100/tonne-salt [79]. This aggregated value includes both high purity density. Optimizing ED current density to electricity prices can reduce
vacuum salt as well as less pure solar salt. We expect that the import specific costs (∼20%) more than fully hybridizing with RO alone
price of high purity vacuum salt will be higher. From Section 5, when (∼10%).
the low electricity prices in Kuwait are taken advantage of, the cost of Fully hybridizing RO with ED up to an RO brine salinity of 120 g/kg,
producing salt using an optimized RO-ED-crystallizer was $61/tonne- optimizing ED current densities and customizing current density for
salt. Thus, even at a market price of $100/tonne-salt, the RO-ED-crys- local electricity prices reduced the brine concentration cost by 33–70 %
tallizer concept can be feasible in Kuwait. Another factor that enhances when compared with standalone ED systems with the variation coming
the feasibility of the RO-ED-crystallizer concept in Kuwait, and the from varying electricity prices globally.
Middle-East in general, is the capability to co-produce drinking water.
Drinking water in the Middle East is obtained primarily from seawater 8.2. Salt production
desalination plants with the actual price of water being $1/m3 or more.
From Section 4, co-production of water could generate a secondary Hybrid RO-ED brine concentration systems coupled with a crystal-
revenue stream greater than $2.7/tonne-salt in Kuwait and the Middle lizer are technically and economically feasible for producing salt from
East. Thus, we strongly believe that there is a reasonable commercial seawater in parts of Japan, Kuwait, the Middle East, Europe and the
potential for the RO-ED-crystallizer concept in Kuwait and by extension USA.
many other parts of the Middle East. The total production cost of salt from a RO-ED-crystallizer was
lower that from conventional standalone ED-crystallizer systems by
7. Limitations and areas for future research 19–55 %. The production cost of salt from an optimal RO-ED-crystal-
lizer system varied from $61–111 /tonne-salt. For a global case corre-
As we arrive at our conclusions, we felt it necessary to highlight sponding to an electricity price of 10 US cents/kWhe, the salt produc-
some of the limitations of our work and areas for future research. tion cost from the proposed RO-ED-crystallizer was $111/tonne-salt —
sufficiently below the market price of vacuum salt in Cyprus, Kuwait,
• The final salt purity produced from the RO-ED-crystallizer concept is Saudi Arabia, and the USA.
expected to be only at least 99.8% pure and cannot be used directly Co-production of water from RO-ED brine concentration systems
in an electrolyzer for chloralkali production. Further purification of can further subsidize salt costs 1–6 % improving the economic feasi-
the salt is needed for use in an electrolyzer. This could be done ei- bility in these markets.
ther before the ED concentrate enters the crystallizer or on the salt
produced. Future work can look in to the economics of how this Nomenclature
adds to cost and changes the price that the salt could command.
• The ED transport for salts other than sodium chloride has not been Acronyms
characterized at high salinity. Thus, the composition of the ED
concentrate would be best determined through experiments. Future CapEx capital expense, $
research work should look into developing an ED model that can Capfac capacity factor, –
accurately predict the composition of ED concentrate. Costelec electricity cost, $/kWhe
• While we have shown that operating ED at higher current densities Crys crystallizer
than 300 A/m2 reduces costs, further development work needs to be CSSF crystallizer salinity scale-up factor, –
carried out to properly design such a system with appropriate ED electrodialysis
choices of channel height, stack length etc. OpEx operating expense, $

114
K.G. Nayar et al. Desalination 456 (2019) 97–120

Pricewater water price, $/m3 Vcp cell pair voltage, V

RO reverse osmosis V̇ volume flow rate, m3/s
SpCapEx specific capital expense, $/tonne-salt Ẇ Power, W
SpCost specific cost, $/tonne-salt
SpE specific energy, kWhe/tonne-salt Greek Symbols
SpOpEx specific operating expense, $/tonne-salt
Δ difference
Roman Symbols ν kinematic viscosity, m2/s
π osmotic pressure, bar
A area, m2 ρ density, kg/m3
C concentration, mol/m3
D diffusion coefficient, m2/s
E potential, V Subscripts
E specific energy consumption, kWh/m3
F Faraday's constant, C/mol am anion exchange membrane
h channel height, m c concentrate
i current density, A/m2 cm cation exchange membrane
J molar flux, mol/m2 ⋅s cp cell pair
Ls salt permeability, m2/s d diluate
Lw permeability to water, mol/m2 ⋅s⋅bar ED electrodialysis
Memeff Membrane effectiveness, – f feed
ṁ mass flow rate, kg/s m at membrane surface
ṁ salt, yr salt mass flow rate, tonne/year i counting index, inlet
Ṅ molar flow rate, mol/s o outlet
P pressure, bar pump pump
Π osmotic pressure, bar p product
r rate of return on capital, – RO reverse osmosis
Re Reynolds number, – s salt
r̄ area resistance, Ω⋅m2 w water
S salinity, kg salt/kg solution
Sc Schmidt number, – Acknowledgment
Sh Sherwood number, –
t solution transport number, – The authors would like to thank Kuwait Foundation for the
Ts membrane salt transport number, – Advancement of Sciences (KFAS) for their financial support through
Tw membrane water transport number, – Project No. P31475EC01. The authors would like to thank various in-
T̄ dustry representatives cited in this work for sharing their expertise and
v flow velocity, m/s perspectives with us.

Appendix A. Cost model for high pressure reverse osmosis

In this section, we evaluate the capital costs of adding a HPRO stage to a seawater RO system producing 550 m3/day of product water. The brine
from the seawater RO system goes to the HPRO stage at a rate of 726 m3/day where pure product water and 120 g/kg brine is produced. To obtain
the cost of HPRO, we first evaluated the capital costs of a seawater RO system taking in the same amount of feed (726 m3/day) producing 305 m3/
day at 42% recovery. For this purpose we used DesalData's seawater RO capital cost estimator [43]. DesalData had collected the cost information
from numerous plants and created an accurate cost estimator. The input data used for the calculator is shown in Table A.7.

Table A.7
Input variables used on DesalData [43] to obtain capital costs for a conventional seawater RO plant.

Plant feature Data

Capacity 305 m3/day

Seawater salinity 30,000 mg/L
Seawater min. temp. 15 °C
Seawater max. temp. 24 °C
Pretreatment Standard
Second pass 0%
Remineralization No
Intake/Outfall Typical
Permitting Typical
Country Any

115
K.G. Nayar et al. Desalination 456 (2019) 97–120

Fig. A.24. Breakdown of capital costs in a 305 m3/day capacity seawater RO plant with the data sourced from Desaldata [43]. The components affected by pressure
account for 29 % of the capital costs.

Table A.8
Break down of capital cost of RO and HPRO per m3/day production capacity for a 305 m3/day system. RO capital costs obtained from Desaldata [43].

CapEx contributor RO CapEx ($-day/m3) HPRO-RO CapEx ratio HPRO CapEx ($-day/m3-prod)

Equipment and materials $ 530 1 $ 530

Civil costs $ 387 1 $ 387
Installation services $ 221 1 $ 221
Intake/Outfall $ 166 0 $ 0
Pretreatment $ 173 0 $ 0
Design costs $ 249 1 $ 249
Legal and professional $ 16 1 $ 16
Piping, high-grade alloy $ 334 1.24 $ 414
Pumps $ 169 1.67 $ 282
Membranes $ 161 3 $ 484
Pressure vessels $ 35 1.67 $ 58
Energy recovery devices $ 27 1.67 $ 45
Total $ 2468 $ 2686

Fig. A.24 lists the different components contributing to the total capital costs of a seawater RO plant. Of all the components listed, only five terms
are affected by pressure: piping, pumps, membranes, pressure vessels and energy recovery devices. This is highlighted in Fig. A.24. These com-
ponents together contribute 29% to the overall capital costs for a seawater RO plant. Individually: piping, pumps, membranes, pressure vessels and
energy recovery devices, each account for 14%, 7%, 7%, 1% and 1% of overall capital costs. The cost of operating seawater RO at 120 bar can be
accurately estimated if the cost of these four components at 120 bar operation are estimated.
Table A.8 lists the specific capital cost of each component of a seawater RO plant, a conversion factor to convert the component cost from RO to
HPRO and the final estimated cost of each component of a HPRO stage of the same size. The total specific capital cost of a 305 m3/day seawater RO
plant was 2468$-day/m3 of product water. A HPRO stage added to a seawater RO plant can be assumed to have the same specific costs of the RO
plant for equipment and materials (i.e. refers to control systems and other peripheral equipment [48]), civil costs, installation services, design costs,
and legal and professional costs. The costs associated with the intake/outfall and pre-treatment (i.e. refers to coagulation and media filters etc. to
remove particulates) can be neglected for the HPRO stage since these components have already been accounted for in the RO plant. For the piping,
pumps, membranes, pressure vessels and energy recovery devices, we estimated conversion factors of 1.24, 1.67, 3, 1.67 and 1.67. The justifications
for these are described below.

A.1. Pumps

We were able to find in the literature pump costs at the pilot scale of around 250 m3/day flow rates [85,86]. At the pilot scale, the 120 bar pump
[85] cost 1.67 times more than the 60 bar pump [86] for the same amount of liquid being pumped. We believe that at larger capacities and with
economies of scale coming in with wider adoption, cost differences between a 60 bar and 120 bar pump will decrease significantly. We applied this
1.67 factor to the data from DesalData and estimated the cost of pumps for a 120 bar RO plant. This is shown in Table A.8.

A.2. Membranes

From conversations with membrane suppliers, we found that the current price of the high pressure RO membrane, DOW XUS180808 Reverse
Osmosis Element [34], was 3 times that of conventional seawater RO membranes.

A.3. Pressure vessels

The pressure vessels in an RO plant must surround the membranes to ensure that they do not burst when the high pressure feed enters them. The
DOW XUS180808 element has an 8-inch diameter and needs a pressure vessel that can withstand the full 120 bar that the pumps will provide. The
ASME (American Society of Mechanical Engineers) publishes an equation on pipe thickness relative to many factors when choosing piping for plants

116
K.G. Nayar et al. Desalination 456 (2019) 97–120

[87]. The equation is as follows:

P *(d + 2c )
t=
2(SE − P (1 − Y )) (A.1)
where t is pipe thickness, P is the design pressure, d is the pipe diameter, c is the mechanical allowances including corrosion, S is the allowable
strength, E is the quality factor, and Y is the material factor. The quality factor E is influenced by the type of pipe. For seamless pipes, E is 1. The
Material Factor Y is influenced by the operating temperature of the pipes. Below 900 °F, the factor is taken to be 0.4. For steel, the allowable stress is
20,000 psi. The mechanical allowance typically chosen to allow for threading and corrosion is 0.125 in.
Given this equation, and a pressure vessel internal diameter of 8 in., we can calculate that at 60 bar (870 psi), the thickness of material required is
0.309 in. or 7.85 mm. This means that the cross section area of the vessel was 8.07 in. 2 or 52.08 cm2. However, at 120 bar, the thickness of material
required is 0.504 in. or 12.80 mm. This leads to a cross section area of 13.45 in. 2 or 87.80 cm2. The cross-sectional area increased by 1.67 times.
Assuming that the only cost change comes from requiring thicker material, the pressure vessels in an HPRO system will cost 1.67 times that in a
conventional seawater RO system.

A.4. Piping

Similar to pressure vessels, using Eq. (A.1), piping also needs an upgrade. The piping for the DOW XUS180808 element [34] must have an
internal diameter of 1.125 in. or 28.575 mm. This leads to a 60 bar (870 psi) requirement of 0.158 in. or 3.955 mm thickness corresponding to a cross
section area of 0.63 in. 2 or 4.04 cm2. At 120 bar (1740 psi), the thickness of material required is 0.188 in. or 4.78 mm corresponding to a cross
section area of 0.78 in.2 or 5.01 cm2. Thus, the cross-sectional area and the volume of material required increased 1.24 times. Assuming that the
material cost is the main price determinant, we can estimate the capital cost of piping in an HPRO system to be 1.24 times that in a conventional
seawater RO system.

A.5. Energy recovery devices

While we could not get direct quotes for how much an energy recovery device (ERD) operating at 120 bar could cost, we estimated that at scale,
the cost increase would be similar to that for pumps. Thus conservatively we assumed that ERDs in a HPRO system would cost 1.67 times that in a
conventional seawater RO system.

A.6. Specific capital cost of HPRO

The factors used estimated that the specific cost of a HPRO stage was 2686$-day/m3 of product water at 42% recovery or 1.09 times the cost of
conventional seawater RO. It must be noted here that the HPRO skids itself (pressure vessels, piping, pumps, membranes and ERD) cost 1.77 times
that of conventional seawater RO. However, the skids only amounted to 29% of the total specific cost leading to only a 9% increase in the stage
specific cost:
SpCapExHPRO,p,day = 1.09 × SpCapExHPRO,p,day (A.2)
Thus, the total capital cost for the HPRO stage was:
̇
CapExHPRO = SpCapExHPRO,p,day × VHPRO,p,day (A.3)
The HPRO stage capital expenses was then annualized to obtain:
CapExHPRO
CapExHPRO,yr =
Annuityfactor (A.4)
HPRO membranes contributed to 18% of the cost of the HPRO stage:
CapExHPRO, mem. = 0.18 × CapExHPRO (A.5)
Assuming a membrane life of 5 years and annualizing we have a HPRO membrane replacement cost of:
CapExHPRO, mem.
OpExHPRO, mem., yr=
Annuityfactor
× [(1 + rreturn )−5 + (1 + rreturn )−10 + (1 + rreturn )−15] (A.6)
The maintenance, chemical and labor cost for seawater RO was 0.18$/m of RO product water. Multiplying this with the annual product water flow
3

from the HPRO stage we have the annual operating cost of maintenance, chemicals and labor:
OpExHPRO, maint., chem., labor, yr = 0.18 $/m3 × Vp,̇ HPRO, yr (A.7)
The total annual operating expense for the HPRO stage was:
OpExHPRO, yr= OpExHPRO, energy, yr + OpExHPRO, mem., yr
+ OpExHPRO, maint., chem., labor, yr (A.8)
where OpExHPRO, energy, yr was obtained be applying the RO process model to the HPRO feed and brine salinities. Thus, the total specific capital cost
of the HPRO stage normalized to the annual salt production was:
CapExHPRO,yr + OpExHPRO,yr
SpCostHPRO =
ṁ salt, yr (A.9)

117
K.G. Nayar et al. Desalination 456 (2019) 97–120

Appendix B. Cost model for nanofiltration

While an NF system may not be needed for the operation of a HPRO stage due, it may be implemented for other reasons such as increasing the
purity of the salt and for reducing maintenance costs on the ED and crystallizer systems. For the benefit of salt production plant designers, we have
provided a preliminary analysis of NF here.
To the best of our knowledge, there is some operational experience in using NF systems with seawater as feed [88]. We used Dow Chemical's
ROSA program [49] to evaluate the divalent rejections of Dow's NF90 and NF270 membranes and determined that rejection was poor if RO brine was
used. Seawater feed yielded much higher rejections with NF270-4040 showing the highest rejection of divalent ions. The NF system configuration
and composition of the permeate was determined using Dow's ROSA software [49]. A configuration using two stages with the first stage operated at
22.4 bar (ΔPp,NF = 21.4 bar), with the first stage having roughly 3 times as many pressure vessels as the second stage, with each pressure vessel
having 6 elements was found to operate at 70% recovery with seawater as feed. A booster pump was not used between stages. ROSA estimated that
the reduction in calcium, magnesium and sulfate ions from the feed to the permeate was 56, 59 and 97% respectively. We worked with an antiscalant
company [54] and ascertained the type of and dosage of antiscalant needed to prevent scale formation in the NF system on the concentrate side.
A pump efficiency of 0.85 was assumed and the work consumed by the NF pump was calculated as follows:

̇ ΔPp,NF 1
ẆNF = Vf,NF
ηp,NF (B.1)
The hourly energy costs of the NF system can be obtained as:
ẆNF
OpExNF, energy, hour = Cost elec
1000[W/kW] (B.2)
Accounting for a capacity factor, Capfac = 0.9, the annual NF energy cost was obtained as:
OpExNF, energy, yr = OpExNF, energy, hour × 8760[h/year] × Capfac (B.3)
The specific cost of maintenance, chemicals and labor of RO on a per feed basis was assumed for NF. Assuming an average RO recovery of 42%, this is
0.0756$/m3-feed. Multiplying this with the annual feed flow to the NF stage we have the annual operating cost of maintenance, chemicals and labor:
OpExNF, maint., chem., labor, yr = 0.0756 $/m3 × Vf,̇ NF, yr (B.4)
3
The capital cost for the seawater NF system was sourced from an actual installation of a 2-stage NF system treating 4000 m /h of seawater at a
recovery ratio of 70% [89]. The specific capital cost of the system complete with process equipment, controls, civil construction, membranes and
installation costs was 948 $-day/m3 of NF permeate [89]. A seawater RO plant taking in the same amount of feed would be able to produce
40,320 m3 /day of product water. The specific capital cost of such a plant from Eq. (14) is 1484 $-day/m3 of RO product [43]. The specific capital
cost of the NF system for the capacity we were interested in was obtained by scaling with Eq. (14) as:
947.9
SpCapExNF,p,day = SpCapExRO,p,day ×
1484 (B.5)
Nanofiltration membranes accounted for only 5% of the specific capital costs. A membrane life of 5 years was assumed. Based on the same method
previously used for RO, we obtained the annualized the capital costs and membrane replacement costs. The energy and maintenance cost was added
to obtain an NF specific cost.
For a NF system that treated 100% of the RO feed, for an optimized RO-ED-crystallizer operated at 600 A/m2 and 120 g/kg RO brine salinity, the
specific cost of the NF system was 5.1$/tonne-salt. This could be considered as an upper bound in costs if NF was added to the RO-ED-crystallizer
configuration.
Pilot tests would have to be conducted to ascertain and weight the benefits and cost of using NF. It is possible that the use of NF may lead to
reasonable reductions in maintenance costs. Furthermore, NF pretreatment may not be needed on 100% of the RO feed depending on the final
objective. By our estimates, even if only 55% of the RO feed was treated using NF, the saturation limit of calcium sulfate at 120 g/kg brine salinity
could be avoided. The specific cost of such an NF system was only 2.8$/tonne-salt. We report these numbers here to encourage future research work
into hybridization of NF with RO-ED systems.

References 10.1016/j.desal.2016.11.022.
[7] H.T. El-Dessouky, H.M. Ettouney, Fundamentals of Salt Water Desalination,
Elsevier, Amsterdam, 2002.
[1] U.S. Geological Survey, Mineral Commodity Summaries, Tech. Rep. 5 2018 https:// [8] H. Ettouney, Design of single-effect mechanical vapor compression, Desalination
minerals.usgs.gov/minerals/pubs/mcs/2018/mcs2018.pdfhttps://doi.org/10. 190 (1-3) (2006) 1–15 http://linkinghub.elsevier.com/retrieve/pii/
3133/70194932. S0011916406001196https://doi.org/10.1016/j.desal.2005.08.003.
[2] V.M. Sedivy, Economy of salt in chloralkali manufacture, National Salt Conference, [9] D.M. Warsinger, K.H. Mistry, K.G. Nayar, H.W. Chung, J.H. Lienhard, Entropy
Gandidham, 2008. generation of desalination powered by variable temperature waste heat, Entropy 17
[3] G. Westphal, G. Kristen, W. Wegener, P. Ambatiello, H. Geyer, B. Epron, C. Bonal, (11) (2015) 7530–7566 http://www.mdpi.com/1099-4300/17/11/7530/https://
G. Steinhauser, F. Götzfried, Sodium chloride, Ullmann's Encyclopedia of Industrial doi.org/10.3390/e17117530.
Chemistry, vol. 33, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, [10] J. Swaminathan, K.G. Nayar, J.H. Lienhard, Mechanical vapor compression–mem-
2010, pp. 9–12, , https://doi.org/10.1002/14356007.a24_317.pub4. brane distillation hybrids for reduced specific energy consumption, Desalin. Water
[4] V.M. Sedivy, Salt Partners Ltd., Personal Communication to Kishor Nayar, (2018). Treat. (2016) 1–11 http://www.tandfonline.com/doi/full/10.1080/19443994.
[5] Bureau of Transportation Statistics, National Transportation Statistics, Tech. rep. 2016.1168579https://doi.org/10.1080/19443994.2016.1168579.
U.S. Department of Transportation, 2013, http://www.bts.gov/publications/ [11] M.A. Jamil, S.M. Zubair, On thermoeconomic analysis of a single-effect mechanical
national%7B_%7Dtransportation%7B_%7Dstatistics%2F%7B%25%7D0ANTS. vapor compression desalination system, Desalination 420 (February) (2017)
[6] H.W. Chung, K.G. Nayar, J. Swaminathan, K.M. Chehayeb, J.H. Lienhard, 292–307, https://doi.org/10.1016/j.desal.2017.07.024.
Thermodynamic analysis of brine management methods: zero-discharge desalina- [12] R. Yamane, M. Ichikawa, Y. Mizutani, Y. Onoue, Concentrated brine production
tion and salinity-gradient power production, Desalination 404 (2017) 291–303 from sea water by electrodialysis using exchange membranes, Ind. Eng. Chem.
http://linkinghub.elsevier.com/retrieve/pii/S0011916416310396https://doi.org/ Process Des. Dev. 8 (2) (1969) 159–165, https://doi.org/10.1021/i260030a003.

118
K.G. Nayar et al. Desalination 456 (2019) 97–120

[13] W.E. Katz, The electrodialysis reversal (EDR) process, Desalination 28 (1) (1979) [39] D. Zetland, C. Gasson, A global survey of urban water tariffs: are they sustainable,
31–40, https://doi.org/10.1016/S0011-9164(00)88124-2. efficient and fair? Int. J. Water Resour. Dev. 29 (3) (2013) 327–342, https://doi.
[14] Y. Kobuchi, Y. Terada, Y. Tani, The first salt plant in the Middle East using elec- org/10.1080/07900627.2012.721672.
trodialysis and ion exchange membranes, Sixth Int. Symp. Salt ll (1983) 541–555. [40] F.J. Millero, R. Feistel, D.G. Wright, T.J. McDougall, The composition of standard
[15] Y. Tanaka, Mass transport and energy consumption in ion-exchange membrane seawater and the definition of the reference-composition salinity scale, Deep-Sea
electrodialysis of seawater, J. Membr. Sci. 215 (1-2) (2003) 265–279, https://doi. Res. I Oceanogr. Res. Pap. 55 (1) (2008) 50–72 http://linkinghub.elsevier.com/
org/10.1016/S0376-7388(03)00020-6. retrieve/pii/S0967063707002282https://doi.org/10.1016/j.dsr.2007.10.001.
[16] Y. Tanaka, A computer simulation of feed and bleed ion exchange membrane [41] M.H. Sharqawy, J.H. Lienhard V, S.M. Zubair, Thermophysical properties of sea-
electrodialysis for desalination of saline water, Desalination 254 (1-3) (2010) water: a review of existing correlations and data, Desalin. Water Treat. 16 (1-3)
99–107, https://doi.org/10.1016/j.desal.2009.12.008. (2010) 354–380, https://doi.org/10.5004/dwt.2010.1079.
[17] H. Strathmann, Electrodialysis, a mature technology with a multitude of new ap- [42] K.G. Nayar, M.H. Sharqawy, L.D. Banchik, J.H. Lienhard, Thermophysical proper-
plications, Desalination 264 (3) (2010) 268–288, https://doi.org/10.1016/j.desal. ties of seawater: a review and new correlations that include pressure dependence,
2010.04.069. Desalination 390 (2016) 1–24 http://linkinghub.elsevier.com/retrieve/pii/
[18] K.M. Chehayeb, D.M. Farhat, K.G. Nayar, J.H. Lienhard, Optimal design and op- S0011916416300807https://doi.org/10.1016/j.desal.2016.02.024.
eration of electrodialysis for brackish-water desalination and for high-salinity brine [43] DesalData, DesalData Cost Estimator, https://www.desaldata.com/cost_estimator,
concentration, Desalination 420 (2017) 167–182 http://linkinghub.elsevier.com/ (2017).
retrieve/pii/S0011916417305763https://doi.org/10.1016/j.desal.2017.07.003. [44] S. Avlonitis, K. Kouroumbas, N. Vlachakis, Energy consumption and membrane
[19] H. Kawate, K. Miyaso, M. Takiguchi, Energy savings in salt manufacture by ion replacement cost for seawater RO desalination plants, Desalination 157 (1-3)
exchange membrane electrodialysis, Sixth Int. Symp. Salt 2 (1983) 471–479. (2003) 151–158 http://linkinghub.elsevier.com/retrieve/pii/
[20] V. Shaposhnik, K. Kesore, An early history of electrodialysis with permselective S0011916403003953https://doi.org/10.1016/S0011-9164(03)00395-3.
membranes, J. Membr. Sci. 136 (1-2) (1997) 35–39 http://linkinghub.elsevier. [45] D. Akgul, M. Çakmakçi, N. Kayaalp, I. Koyuncu, Cost analysis of seawater desali-
com/retrieve/pii/S037673889700149Xhttps://doi.org/10.1016/S0376-7388(97) nation with reverse osmosis in Turkey, Desalination 220 (1-3) (2008) 123–131
00149-X. http://linkinghub.elsevier.com/retrieve/pii/S001191640700611Xhttps://doi.org/
[21] A.E.R. Reahl, Half a century of desalination with electrodialysis, Tech. rep. 10.1016/j.desal.2007.01.027.
TP1038EN0603 General Electric Company, 2006. [46] S.S. Mitra, M.K. Sharma, S. Rybar, C. Bartels, L. Pelegrin, Fujairah SWRO – man-
[22] N.C. Wright, A.G. Winter V., Justification for community-scale photovoltaic-pow- agement of membrane replacement, Desalin. Water Treat. 10 (1-3) (2009) 255–264,
ered electrodialysis desalination systems for inland rural villages in India, https://doi.org/10.5004/dwt.2009.784.
Desalination 352 (0) (2014) 82–91 http://linkinghub.elsevier.com/retrieve/pii/ [47] B. Salgado, E. Gasia-Bruch, V. García-Molina, A. Casa nas, S. Coker, Long-term
S0011916414004160 http://www.sciencedirect.com/science/article/pii/ benefits of DOW FILMTEC RO membranes, Desalin. Water Treat. 51 (1-3) (2013)
S0011916414004160https://doi.org/10.1016/j.desal.2014.07.035. 333–342, https://doi.org/10.1080/19443994.2012.703803.
[23] H. Strathmann, Electrodialysis and its application in the chemical process industry, [48] T. Pankratz, Water Desalination Report, 31 July 2017, (2017).
Sep. Purif. Rev. 14 (1) (1985) 41–66 http://www.tandfonline.com/doi/full/10. [49] Dow Water and Process Solutions, ROSA9 (Reverse Osmosis System Analysis),
1517/17425255.3.2.169https://doi.org/10.1080/03602548508068411. https://www.dow.com/en-us/water-and-process-solutions/resources/design-
[24] B. Pilat, Water of high quality for household conditions, Desalination 153 (1-3) software, (2018).
(2003) 405–407, https://doi.org/10.1016/S0011-9164(02)01135-9. [50] Dow Water and Process Solutions, WAVE (Water Application Value Engine), http://
[25] K.G. Nayar, P. Sundararaman, J.D. Schacherl, C.L. O’Connor, M. Heath, M. Gabriel, www.dupont.com/water/design-software.html, (2018).
N.C. Wright, A.G. Winter V, Feasibility study of an electrodialysis system for in- [51] K.S. Pitzer, Thermodynamics of electrolytes. I. Theoretical basis and general
home water desalination and purification in urban India, Proceedings of the ASME equations, J. Phys. Chem. 77 (2) (1973) 268–277, https://doi.org/10.1021/
2015 International Design Engineering Technical Conference, Boston, USA, 2015. j100621a026.
[26] K.G. Nayar, P. Sundararaman, C.L. O’Connor, J.D. Schacherl, M.L. Heath, [52] S.R. Charlton, D.L. Parkhurst, Modules based on the geochemical model PHREEQC
M.O. Gabriel, S.R. Shah, N.C. Wright, A.G. Winter V, Feasibility study of an elec- for use in scripting and programming languages, Comput. Geosci. 37 (2011)
trodialysis system for in-home water desalination in urban India, Dev. Eng. 2 (2017) 1653–1663, https://doi.org/10.1016/j.cageo.2011.02.005.
38–46 http://linkinghub.elsevier.com/retrieve/pii/S2352728516300045https:// [53] D.L. Parkhurst, C.A.J. Appelo, Description of input and examples for PHREEQC
doi.org/10.1016/j.deveng.2016.12.001. version 3-A computer program for speciation, batch-reaction, one-dimensional
[27] A. Moura Bernardes, M.A. Rodrigues, J.Z. Ferreira, Electrodialysis and Water Reuse, transport, and inverse geochemical calculations, Techniques and Methods, vol.
(2014), https://doi.org/10.1007/978-3-642-40249-4. book 6, U.S. Geological Survey, 2013, http://pubs.usgs.gov/tm/06/a43.
[28] K.M. Chehayeb, K.G. Nayar, J.H. Lienhard, On the merits of using multi-stage and [54] Avista Technologies Inc., Personal Communication to Kishor Nayar, https://www.
counterflow electrodialysis for reduced energy consumption, Desalination 439 avistatech.com, (2018).
(2018) 1–16 http://linkinghub.elsevier.com/retrieve/pii/ [55] R.K. McGovern, A.M. Weiner, L. Sun, C.G. Chambers, S.M. Zubair, J.H. Lienhard,
S0011916417326413https://doi.org/10.1016/j.desal.2018.03.026. On the cost of electrodialysis for the desalination of high salinity feeds, Appl.
[29] T. Miyake, M. Yoshida, T. Tanabe, Design and performance of an economical brine Energy 136 (2014) 649–661 http://linkinghub.elsevier.com/retrieve/pii/
electrodialysis process, Seventh Symposium on Salt, Amsterdam, vol. II, 1993, pp. S0306261914009982https://doi.org/10.1016/j.apenergy.2014.09.050.
79–84 http://www.worldsaltsymposium.org/download/design-and-performance- [56] M. Fidaleo, M. Moresi, Optimal strategy to model the electrodialytic recovery of a
of-an-economical-brine-electrodialysis-process/. strong electrolyte, J. Membr. Sci. 260 (1-2) (2005) 90–111 http://www.
[30] C. Fritzmann, J. Löwenberg, T. Wintgens, T. Melin, State-of-the-art of reverse os- sciencedirect.com/science/article/pii/S0376738805002486 http://linkinghub.
mosis desalination, Desalination 216 (1-3) (2007) 1–76 http://linkinghub.elsevier. elsevier.com/retrieve/pii/S0376738805002486https://doi.org/10.1016/j.memsci.
com/retrieve/pii/S0011916407004250https://doi.org/10.1016/j.desal.2006.12. 2005.01.048.
009. [57] Asahi Glass and Chemicals Co., Personal Communication to Kishor Nayar, (2016).
[31] International Desalination Association, IDA Desalination Yearbook 2015–2016, [58] Asahi Glass and Chemicals Co., Selemion Ion Exchange Membranes DW Type 3E,
Tech. rep. International Desalination Association, 2015. Tech. Rep. January, 2013 http://www.amp-ionex.com/products/selemion/pdf/
[32] K.H. Mistry, R.K. McGovern, G.P. Thiel, E.K. Summers, S.M. Zubair, J.H. Lienhard, selemion.pdf Chiba, Japan.
Entropy generation analysis of desalination technologies, Entropy 13 (12) (2011) [59] E.T. Sajtar, D.M. Bagley, Electrodialysis reversal: process and cost approximations
1829–1864 http://www.mdpi.com/1099-4300/13/10/1829/https://doi.org/10. for treating coal-bed methane waters, Desalin. Water Treat. 2 (1-3) (2009)
3390/e13101829. 284–294, https://doi.org/10.5004/dwt.2009.259.
[33] G.P. Thiel, E.W. Tow, L.D. Banchik, H.W. Chung, J.H. Lienhard, Energy consump- [60] Glassdoor, Chemical operator salary, https://www.glassdoor.com/Salaries/
tion in desalinating produced water from shale oil and gas extraction, Desalination chemical-operator-salary-SRCH_KO0,17.htm, (2018).
366 (2015) 94–112 http://linkinghub.elsevier.com/retrieve/pii/ [61] M. Reig, S. Casas, C. Aladjem, C. Valderrama, O. Gibert, F. Valero, C.M. Centeno,
S0011916414006857https://doi.org/10.1016/j.desal.2014.12.038. E. Larrotcha, J.L. Cortina, Concentration of NaCl from seawater reverse osmosis
[34] DOW FILMTEC Membranes, DOW XUS180808 Reverse Osmosis Element, Tech. rep. brines for the chlor-alkali industry by electrodialysis, Desalination 342 (2014)
The Dow Chemical Company, 2016, http://msdssearch.dow.com/ 107–117, https://doi.org/10.1016/j.desal.2013.12.021.
PublishedLiteratureDOWCOM/dh_0965/0901b8038096535c.pdf?filepath= [62] Y. Tanaka, Development of a computer simulation program of batch ion-exchange
liquidseps/pdfs/noreg/609-50257.pdf&fromPage=GetDoc. membrane electrodialysis for saline water desalination, Desalination 320 (2013)
[35] K.G. Nayar, J. Fernandes, R.K. McGovern, B. Al-Anzi, J.H. Lienhard, Costs and 118–133 http://linkinghub.elsevier.com/retrieve/pii/
Energy Needs of RO-ED-crystallizer Systems for Zero Brine Discharge Seawater S0011916413001896https://doi.org/10.1016/j.desal.2013.04.022.
Desalination, Desalination (2019), https://doi.org/10.1016/j.desal.2018.11.018 [63] V.M. Sedivy, Environmental balance of salt production speaks in favour of solar
under review. saltworks, Glob. NEST J. 11 (1) (2009) 41–48.
[36] R.K. McGovern, S.M. Zubair, J.H. Lienhard, The benefits of hybridising electro- [64] J. Niederberger, Titan Salt, Personal Communication to Kishor Nayar, (2018).
dialysis with reverse osmosis, J. Membr. Sci. 469 (2014) 326–335, https://doi.org/ [65] Codex Alimentarius International Food Standards, Standard for food grade salt,
10.1016/j.memsci.2014.06.040. Tech. rep. Joint FAO/WHO Food Standards Programme, Rome, Italy, 2001, http://
[37] R.K. McGovern, S.M. Zubair, J.H. Lienhard, Design and optimization of hybrid ED- www.fao.org/fao-who-codexalimentarius/sh-proxy/en/?lnk=1&url=https%5C%
RO systems for the treatment of highly saline brines, International Desalination 253A%5C%252F%5C%252Fworkspace.fao.org%5C%252Fsites%5C%252Fcodex
Association World Congress 2013, Tianjin, China, 2013. %5C%252FStandards%5C%252FCODEX%5C%2BSTAN%5C%2B150-1985%5C%
[38] Y. Tanaka, R. Ehara, S. Itoi, T. Goto, Ion-exchange membrane electrodialytic salt 252FCXS_150e.pdf.
production using brine discharged from a reverse osmosis seawater desalination [66] V.M. Sedivy, Purification of salt for chemical and human consumption, Proceedings
plant, J. Membr. Sci. 222 (1-2) (2003) 71–86, https://doi.org/10.1016/S0376- of the International Conference on Salt 2006, no. January, Ahemdabad, India, 2006,
7388(03)00217-5. pp. 78–90.

119
K.G. Nayar et al. Desalination 456 (2019) 97–120

[67] Glassdoor, Process chemist salary, https://www.glassdoor.com/Salaries/chemical- Communicationto Jenifer Fernandes (2018).
operator-salary-SRCH_KO0,17.htm, (2018). [78] Hitesh Mishra, Transworld Overseas, Gujarat, India, Personal Communication to
[68] U.S. Bureau of Reclamation, Zero Liquid Discharge, Southern California Regional Jenifer Fernandes (2016).
Brine-Concentrate Management Study – Phase I Brine-Concentrate Treatment and [79] Business Analytical Center, Kuwait Trade of Salt, Export Import and Market
Disposal Options Report, U.S. Department of the Interior, 2018, https://www.usbr. Prospects, 2015.
gov/lc/socal/reports/brineconcentrate/6TreatmentandDisposal_part2.pdf Ch. 3. [80] Malaz A. Mizo, Gulf Salt Company, Saudi Arabia, Personal communication to
[69] World Bank Report, Seawater and Brackish Water Desalination in the Middle East, Jenifer Fernandes (October 2016).
North Africa and Central Asia, A review of Key Issues and Experience in Six [81] U.S. Geological Survey, Salt, Mineral Commodity Summaries 2017, USA, January
countries, http://documents.worldbank.org/curated/en/890811468002102369/ 2017, pp. 140–141.
text/335150v10Seawater0mainreport1Final2.txt, (December 2004). [82] World Bank, Transport Costs and Specialization, World Development Report 2009,
[70] A. Tenne, Sea Water Desalination in Israel: Planning, coping with difficulties, and The World Bank, 2008, pp. 170–196 https://siteresources.worldbank.org/
economic aspects, http://www.water.gov.il/hebrew/planning-and-development/ INTWDRS/Resources/477365-1327525347307/8392086-1327528510568/
desalination/documents/desalination-in-israel.pdf, (2010). WDR09_12_Ch06web.pdf%5C%0A https://doi.org/10.1596/9780821376072_ch6
[71] M. Al-Sahali, S. Al-Qattan, Performance analysis of small capacity reverse osmosis Ch. 6.
desalination plants, Kuwait J. Sci. Eng. 39 (2012) 231–253. [83] IContainers.com, iContainers, https://www.icontainers.com, (2018).
[72] Y. Tokui, H. Moriguchi, Y. Nishi, Comprehensive environmental assessment of [84] M. Yonar, ANTGIDA, Istanbul, Turkey. Personal Communication to Kishor Nayar
seawater desalination plants: Multistage flash distillation and reverse osmosis (2018).
membrane types in Saudi Arabia, http://www.water.gov.il/hebrew/planning-and- [85] Cat Pumps, Cat 6841 Stainless Steel Pump, Tech. rep. Cat Pumps, Minneapolis, MN,
development/desalination/documents/desalination-in-israel.pdf, (2014). 2018, http://www.etscompany.com/catpump/pdfs/6841.pdf.
[73] Water Technology, Minjur Desalination Plant, Tamil Nadu, India, https://www. [86] Cat Pumps, Cat 3841 Stainless Steel Pump, Tech. rep. Cat Pumps, Minneapolis, MN,
water-technology.net/projects/minjurdesalination/, (2010). 2018, http://www.catpumps.com/products/pdfs/3841_E.pdf.
[74] Mitsuyoshi Hirai, Technology of Reverse Osmosis Desalination, Water Reuse [87] ASME, ASME B31.3 Process Piping Guide, Tech. rep. 2016 https://www.asme.org/
Promotion Centre; WDR/GWI, DesalData, 2014, https://www.jccp.or.jp/ products/codes-standards/b313-2016-process-piping.
international/conference/docs/%5C%2528Hirai%5C%252920141216_ [88] L. Awerbuch, M. Parker, I. Agha-mourad, Integrated Upgrading of Thermal
TECHNOLOGY_RO_Desalination.pdf". Processes and Nanofiltration Experience of SEWA, IDA World Congress-
[75] IBMWaterfund, Rickards Real Cost Water Index calculated by IBM, https:// Maspalomas, Gran Canaria, Spain, 2007 https://www.academia.edu/7818468/
researcher.watson.ibm.com/researcher/view_group.php?id=5047, (2018). Integrated_Upgrading_of_Thermal_Processes_and_Nanofiltration_Experience_of_
[76] Sunny Sang, Shougang Dinghao Trading Company, Shouguang, China, Personal SEWA_Project Maspalomas, Gran Canria, Spain.
Communication toJenifer Fernandes (2018). [89] L. Awerbuch, Personal Communication to Kishor Nayar, (2018).
[77] P. Americanos, M.P. Theodorou Salt Industry & Co. Ltd, Larnaca, Cyprus, Personal

120

View publication stats