You are on page 1of 11

DES-12379; No of Pages 11

Desalination xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

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

Opportunities to reach economic sustainability in forward osmosis–reverse osmosis


hybrids for seawater desalination
Gaetan Blandin a,b, Arne R.D. Verliefde b, Chuyang Y. Tang c, Pierre Le-Clech a,⁎
a
UNESCO Centre for Membrane Science and Technology, School of Chemical Engineering, The University of New South Wales, Australia
b
Ghent University, Faculty of Bioscience Engineering, Department of Applied Analytical and Physical Chemistry, Particle and Interfacial Technology Group (PaInT), Coupure Links 653,
B-9000 Ghent, Belgium
c
Department of Civil Engineering, The University of Hong Kong, Hong Kong, China

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• A flux threshold of 30 L·m− 2·h− 1 was


defined for FO economic viability.
• Current membrane developments do not
match the fixed minimum flux threshold.
• Membranes with higher A
(N 5 L·m − 2·h− 1 ·bar − 1) and lower S
(b100 μm) are required.
• Pressure assisted osmosis could be an
economically favourable alternative to
FO.

a r t i c l e i n f o a b s t r a c t

Article history: Despite being energetically attractive, the economic sustainability of forward osmosis (FO) - reverse osmosis
Received 3 October 2014 (RO) hybrid process for seawater desalination has not yet been fully demonstrated. This study is the first to
Received in revised form 10 December 2014 carry out an extensive economic evaluation of FO-RO hybrid, benchmarked against stand-alone RO system.
Accepted 10 December 2014
This assessment clearly highlights that FO-RO hybrid can be beneficial, comparatively to RO, only for high energy
Available online xxxx
costs and/or substantial operational costs savings. It is also demonstrated that improvement in water permeation
Keywords:
flux, typically above 30 L·m−2·h−1 for classical water recoveries, is an absolute prerequisite to lower investment
Forward osmosis costs down to an economically acceptable level. Such fluxes are not achieved with current membrane developments
Economics when realistic feed and draw solutions are considered. In a second step, a sensitivity analysis was carried out to as-
Permeation flux sess the effects of process parameters on flux in FO, compared to the above-mentioned benchmark. Results indicate
Pressure assisted osmosis that for novel FO membranes, typically a coupling of higher water permeability (A N 5 L·m−2·h−1·bar−1) and lower
Hydraulic pressure structural parameter (S b 100 μm) is required. This study also shows that the concept of pressure assisted osmosis
(PAO) can help to reach higher fluxes with current commercial membranes, thereby demonstrating more
favourable economics compared to state-of-the-art FO membranes.
© 2014 Published by Elsevier B.V.

⁎ Corresponding author at: UNESCO Centre for Membrane Science & Technology, School of Chemical Engineering, Building F10, The University of New South Wales, Sydney, NSW 2052,
Australia.
E-mail address: p.le-clech@unsw.edu.au (P. Le-Clech).

http://dx.doi.org/10.1016/j.desal.2014.12.011
0011-9164/© 2014 Published by Elsevier B.V.

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
2 G. Blandin et al. / Desalination xxx (2014) xxx–xxx

1. Introduction water treatment streams, i.e., water reuse and desalination or simple
mixing of these streams [15]. One very recent study on economics con-
Within the last decade, a growing interest has been observed in firmed that FO–RO hybrids could offer an alternative to classical water
osmotically-driven membrane processes and more specifically in for- reuse and desalination scenarios, but only when considering a high
ward osmosis (FO), which has been evaluated as a promising alternative share of energy in the total desalination cost, assuming lower membrane
to reverse osmosis (RO) for seawater desalination to produce fresh costs (30$·m−2) and/or higher fluxes (15 L·m−2·h−1) than existing
water at low energy consumption [1]. Thanks to the recent develop- commercial membranes can obtain [8]. This demonstrates the need for
ment and commercialisation of novel semi-permeable membranes ded- a better consideration of economics in FO–RO hybrid within the desalina-
icated to this application, a significant amount of academic research has tion process and the necessity of sensitivity analysis. The need for perme-
been performed to better understand process limitations and proposed ation flux improvement is often mentioned [1,16], and is a key driver for
further improvement [1]. The first industrial FO desalination plant in research towards membrane development and FO process optimisation.
Oman has been recently inaugurated, testifying of the potential future However, despite the numerous reviews on the topic [1,2,17–23], and in
of this technology [2]. However, questions remain on the optimised comparison with pressure retarded osmosis (PRO) where a threshold of
configuration of such process, as well as its economic sustainability 5 W·m−2 has been set as breakeven point for sustainability [24,25], no
[3,4]. clear permeation flux threshold has been defined for FO sustainability
In the context of seawater desalination, FO is typically operated within for seawater desalination applications.
a two-step process. FO could be implemented as a pre-treatment of RO Water transport through FO membrane is limited by a number of
(Fig. 1). In this case, low salinity water is used as an FO feed solution to di- phenomena extensively reviewed elsewhere [26] — internal concentra-
lute seawater (draw). As a result of the seawater dilution, less energy tion polarisation (ICP), external concentration polarisation (ECP), foul-
(due to lower osmotic pressure of the feed) is required to produce fresh ing and reverse salt diffusion (RSD). Extended fundamental work was
water using the subsequent RO step. This type of scheme is referred to performed to model the impact of concentration polarisation on perme-
as FO-RO hybrid [5,6], and is typically of interest when impaired water ation flux through the use of the solution diffusion model [27–29].
is used as feed, as the scheme can uniquely combine water reuse and de- Membrane characteristics, i.e. pure water and salt permeability of the
salination [6,7], and has the advantages of double-barrier protection rejection layer (A & B) and the structural parameter (S), are key param-
against pollutants in the impaired water, low membrane fouling and sub- eters impacting permeation flux, and the ideal membrane for FO appli-
stantial energy savings. The FO-RO hybrid mentioned in this study, also cations should have a high water permeability and salt rejection (high
called osmotic dilution [8] has to be distinguished from other FO concepts A, low B), have minimal ICP (low S), and sufficient mechanical strength
also used for desalination. Those other concepts are combining a FO pro- to support industrial operation and possibly moderate pressure [17]. A
cess with a very high osmotic driving force using a closed loop draw solu- significant body of work has been performed in terms of FO membrane
tions that has to be recovered by nanofiltration (NF), RO or other technic optimisation since the introduction of the first FO commercial mem-
currently under study [9–11], yet are out of the scope of this manuscript. brane by HTI [16]; a recent review compiled an extended list of
One study specifically mentioned that the FO-RO hybrid de- all recent membrane developments [21] confirming that important
scribed above, using immersed FO membranes led to a specific ener- improvements have been obtained on laboratory scale. Water perme-
gy decrease for RO seawater desalination from 2.5 - 4 kWh·m− 3 ability enhancement by more than 20 in comparison with existing com-
down to 1.5 kWh·m− 3 [12]. However, these values were obtained mercial membrane was mentioned; however, it is difficult to translate
following dilution of seawater by a factor of 2.5 in the FO process. In this linearly into higher fluxes, as only a 6-fold increase in flux has
such conditions, this process relies mainly on water reuse and therefore been reached in FO operation using these membranes. Moreover, high
does not really qualify as desalination process. In general, lab-scale and permeation fluxes are often claimed in the lab, but the used test condi-
pilot operation of hybrid FO-RO have both confirmed technical process tions are usually different from industrial reality (i.e. pure water as feed
feasibility, by showing lower fouling tendency and claiming significant and a high salinity draw [30–34]), thus overestimating the practical
energy savings in comparison with current RO seawater desalination [1, values reachable on industrial scale. Standard test conditions have
7,13]. Despite these observations and intensive research, questions re- been proposed [35], but so far, most studies have been conducted
main on FO-RO hybrid economic sustainability for seawater desalination. using different conditions, such as type and concentration of draw and
Reasons for these questions are obvious: firstly, integration of FO will re- feed solutions, membrane orientations, type of spacers and cross-flow
quire additional investment costs that have to be ultimately compensated velocity (CFV). This further demonstrates the need for a clear, unambig-
by significant energy savings. Moreover, due to recent RO development, uous and comparative evaluation of existing membranes. Interestingly,
energy consumption in RO is getting so close to the thermodynamic the solution–diffusion model can be used as a modelling tool, integrat-
limit of desalination that additional energy savings may become marginal ing the impact of both membrane characteristics and process driving
[14]. Finally, the FO–RO hybrid process has been pointed out as not offer- forces on water and salt permeation flux. As such, it represents a predic-
ing advantages in comparison with two distinct, simpler and established tive and quantitative tool to (1) investigate optimisation of membrane

Fig. 1. Integration of FO in desalination process as FO-RO hybrid.

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
G. Blandin et al. / Desalination xxx (2014) xxx–xxx 3

and/or process parameters, (2) evaluate current (i.e., state-of-the art) an energy recovery device to assess the impact of FO pre-dilution on RO
membranes in similar conditions to obtain a fair comparison and (3) de- energy consumption for three different scenarios with the goals to de-
fine opportunities for further improvement (i.e., carry out a sensitivity crease energy consumption and/or increase water production of existing
analysis). RO system:
As an alternative to new material developments or the use of high
osmotic driving forces, the pressure-assisted osmosis (PAO) concept (1) RO installation is maintained at its original feed flow and recov-
has been recently proposed. In this concept, hydraulic pressure applied ery, FO is mainly used to lower RO energy usage for the same
on the feed side is used as an additional driving force to enhance water final permeate production (Fig. 2a, continuous blue line).
permeation flux [36–38]. Recent studies confirmed that hydraulic pres- (2) RO feed flow is increased with the amount of water permeated in
sure could indeed increase fluxes substantially and even more than ex- the FO stage (Fig. 2a, dotted red line); RO operates at same recov-
pected due to an increase in membrane permeability with increasing ery but higher feed and permeate flow; i.e. higher production.
flux, as a consequence of membrane deformation [36,39]. In PAO oper- (3) Fixed 50% FO recovery and RO installation is maintained at its
ation, the FO membrane maintained good separation performances, original feed flow but recovery is increased; permeate produc-
leading to increased water permeation fluxes and limited RSD. This tion is increased (Fig. 2b).
mode of operation thus tackles the so-called “water–salt permeability
trade-off”, which currently limits FO membrane development and
therefore represents an interesting option for FO improvement [36]. The conditions used for the ROSA® calculation are described in
The objectives of this study are therefore to: Fig. S1 of supporting document, and the final energy consumption
values were calculated incorporating energy recovery devices.
• Establish the economic basis for cost calculation in FO–RO hybrid, to Fig. 2a showed that considering scenario 1 (keeping RO at similar
define the optimum FO process configuration for desalination, and feed flow and recovery), up to 36% decrease of the energy consumption
to set a threshold for FO minimum permeation flux required to is possible. In such scenario, in the RO step, part of the seawater is re-
reach economic viability. placed by pure water permeated in the FO step from impaired water,
• Use the solution–diffusion model to evaluate potential options to no production increase occurs, energy savings are observed as a direct
tackle current flux limitations of state-of-the-art FO membranes and benefit of osmotic dilution. Alternatively, using FO permeate to increase
configurations, i.e., by membrane development and implementation the existing RO feed flow (scenario 2) did not lead to any significant en-
of hydraulic pressure as additional driving force. ergy reduction, the benefit of osmotic dilution being counter-balanced
• Assess the relevance and extent of need for further membrane and/or by additional pressure needed to filtrate more water on a similar RO fil-
process development related to requirements for FO economic tration surface area. As a result, more production could occur but with-
sustainability. out energy savings in the RO part, making this configuration irrelevant.
In the third scenario (50% FO recovery, similar RO feed flow, and in-
creasing RO recovery Fig. 2b), energy consumption already equals
2. Economics of FO in desalination stand-alone RO already (2.2 kwh·m− 3) at 60% RO recovery. In fact,
the benefit of osmotic dilution is lost due to the operation of RO at
2.1. FO integration in RO desalination schemes high operating pressure to allow for higher permeation flux. As a result,
scenarios for which increased permeate production was considered did
Based on the process described in Fig. 1, FO integration in RO desali- not allow for significant energy savings, and a simple extension of RO
nation schemes decreases the salinity of the outlet draw solution (dilut- unit may be preferable to FO–RO hybrid implementation. Thus, only
ed seawater). In this study, similar flow for feed and draw solution and the implementation of FO–RO hybrid to decrease energy needs (scenar-
50% of the feed permeating to the draw (i.e. 50% recovery) were consid- io 1) will be considered in this study. In that case, the implementation of
ered in order to be in a combined water reuse and desalination scheme. FO will lead to additional investment costs that need to be compensated
Energy estimation was performed using ROSA® software and integrating by the obtained energy savings and potential operational expenditure

2.5
4.0
RO energy consumpon (kWh.m-3)

RO energy consumpon (kWh.m-3)

(b)
3.5
(a)
2.0
3.0

2.5
1.5

Constant RO feed flow (4000m3.h-1) 2.0

Increased RO feed flow with FO permeate


1.0 1.5
0% 20% 40% 60% 80% 100% 45% 55% 65% 75%
% FO recovery %RO recovery

Fig. 2. RO energy consumption considering different scenario of FO–RO hybrid implementations using ROSA software: (a) operating at constant RO feed flow or increasing the RO feed flow
with the amount of water permeated in FO stage, (b) operating at 50% FO recovery and constant RO feed flow but increasing RO recovery. FO operating as similar feed and draw solution
flows, FO recovery defined as % feed permeating through the draw. RO feed salinity was calculated using initial seawater salinity of 35 g·L−1 of red sea salt [36] and adjusted based on the
dilution occurring in the FO step depending on FO recovery. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
4 G. Blandin et al. / Desalination xxx (2014) xxx–xxx

(OPEX) reduction (less fouling, less membrane replacement due to packing density) down to 0.5 (the hypothetical optimised FO situation
lower operating pressure) to make the FO–RO hybrid economically whereby high performance modules and cheaper FO materials are pos-
sustainable. sible due to low pressure operation). The sensitivity analysis of the po-
tential CAPEX savings of the FO–RO hybrid versus conventional RO is
2.2. FO–RO hybrid costs — sensitivity analysis plotted in Fig. 3, as a function of the average FO permeation flux.
As expected, the FO–RO hybrid process was found to be always more
At this stage, it is difficult to balance the energy savings in the FO–RO expensive than the RO baseline (0% savings). This is due to the addition-
hybrid and the increased capital expenditure (CAPEX). Due to the very al FO installation, increasing the investment cost. In the most favourable
limited amount of data available on FO CAPEX and OPEX in the litera- case (high FO flux, low FO cost), the FO unit is expected to add 20%
ture, assumptions have to be made. The sense and non-sense of these CAPEX to the whole desalination plant. This demonstrates that, despite
assumptions will be assessed in this section through a sensitivity being promoted as a potentially better process, the economic viability of
analysis. the FO–RO hybrid cannot be established from the CAPEX.
Conventional seawater RO is the only system that is rather similar to This presents a much more challenging case to reach economic
FO, and will here be used as a benchmark for assessment of FO cost cal- viability of FO-based systems only based on energy gain. This is in
culations. Existing CAPEX data from RO literature were adapted to give contrast to benefit–cost ratio analyses reported in previous studies [7,
an approximation of FO CAPEX. Pre-treatment and intake costs were 12], for which only membrane (12–30$·m−2 [7,24]) or module costs
considered similar for stand-alone RO and FO–RO hybrid in this study. (100$·m−2 [12]) were considered.
In order to estimate the cost of FO systems and the potential interest Incorporating more costs than just filtration surface area-dependent
of FO–RO hybrid, a number of assumptions have been made regarding costs in the CAPEX, as proposed in this study, can be considered as more
water production costs and the CAPEX distribution within the overall realistic since the costs calculated like this include all related equipment
FO–RO hybrid desalination plant. All these parameters will be discussed and auxiliary investments. Moreover, accounting only for membrane
in the following paragraphs, together with the impact and importance and module as CAPEX surface dependent (i.e. 7%) is not relevant: other-
of permeation flux on FO costs and FO–RO hybrid costs (and thus FO wise increasing permeation flux would not lead to any economics
sustainability). benefits. As such, considering 50% to 100% surface dependent CAPEX ap-
pears more realistic and better translates the need for higher perme-
2.2.1. FO and FO–RO CAPEX ation flux in FO. According to first rough estimation (Fig. S3), 75% flux
Considering values obtained from desaldata.com [40] and according dependant seems to be an appropriate assumption and will be consid-
to the latest RO technology, an average cost value of 0.76$·m−3 of water ered in further simulation.
produced can be considered as the benchmark for RO in this study. It is also estimated that current state-of-the-art for FO using spiral-
Overall, RO water desalination cost can be split in energy costs, CAPEX wound modules with relatively expensive membrane and low packing
and maintenance/labour (OPEX) at 32, 38 and 30%, respectively [40]. density are likely to result in a FO/RO CAPEX ratio even higher than 2
CAPEX represents a significant amount of the produced water cost, i.e. (Fig. 3b). This, in combination with the low permeation fluxes reported
0.29$·m−3 for the overall desalination plant (38% if 0.76$·m−3, in the literature (5 to 10 L·m−2·h−1), results in doubles the CAPEX for
Fig. S2, [40]). It could be reasonably assumed that 50% of the CAPEX is an FO–RO hybrid desalination plant compared to a classical RO CAPEX
dedicated to the RO filtration step (Fig. S3, [40]), i.e. 0.14$·m−3, and and makes FO–RO economically unattractive. The need for cheaper ma-
the rest related to the pre- and post-treatment. Additionally, benchmark terials, higher packing density modules and higher permeation fluxes is
RO was assumed with 20 years plant lifetime, a working load of thus clearly demonstrated here. Even with the recent improvements
8000 h·y−1 and 20 L·m−2·h−1 as average permeation flux for RO oper- achieved in FO membrane materials, reaching an FO/RO CAPEX ratio
ation [41–43]. of 0.5 is probably a very optimistic hypothesis, as it requires break-
When translating CAPEX of RO to FO, some assumptions are re- through long-term developments such as cheap hollow fibre modules
quired, based on the differences in operating regime. Considering with extremely high packing densities that still need to prove their
spiral-wound modules, FO operates with two solutions that need to be suitability for treatment of wastewater with high concentrations of
pumped on both sides on the membrane, leading to lower intrinsic foulants. Assumptions of a future FO/RO CAPEX ratio of 1 may therefore
packing density and thus higher prices of FO modules compared to RO appear more realistic in the near future (Porifera already proposes sys-
modules. In addition, FO membranes in their current state of develop- tems with similar packing density as RO modules [44], Toyobo men-
ment are also inherently more expensive than existing RO membranes. tioned PRO pilot tests with hollow fibre elements [45]). As such, an
Furthermore, current commercial FO performance is lower than RO, es- FO/RO CAPEX ratio = 1 will be considered in the further assessment
pecially in terms of average permeation flux. On the other hand, FO of the potential economic viability of FO–RO hybrids.
operates without the need for high pressure, and as a consequence, sig-
nificant savings could be obtained in terms of material requirements for 2.2.2. Energy costs
the pressure vessels, modules and pumps for an FO installation. As demonstrated in Section 2.1, energy needs for the RO step can be
As such, not only membrane and module costs should be considered lowered by 30% as a result of the osmotic dilution in FO. However it is
for FO and RO economics comparison, but also all the other CAPEX need- still unclear if those significant energy cost savings are sufficient to com-
ed for the filtration, such as pumps, piping, civil engineering and moni- pensate the irremediable additional CAPEX related to the FO. Assuming
toring need to be considered. To clarify the influence of assumptions energy costs of 0.24$.m−3 [40] and considering specific energy con-
made on the CAPEX for FO compared to RO, some sensitivity studies sumption of RO about 2.2 kWh·m−3 (Fig. 2a), the average energy cost
were performed on 2 key parameters: 1) the filtration surface depen- could be estimated to 0.1$·kWh− 1. However, energy cost is highly
dency of FO CAPEX investment, i.e., how much of the CAPEX is directly dependent on the location and global market prices. Therefore, a
related to required membrane surface area and 2) the comparison be- sensitivity analysis was carried out on the impact of energy cost
tween FO and RO CAPEX, expressed as ratio FO/RO CAPEX investment. (and consequently the contribution of energy to total costs within
Surface dependency of CAPEX was varied from 7% (where only the the desalination plant) whereby the energy price was varied in the
membranes and pressure vessels CAPEX is assumed to be dependent range of 0.05 to 0.2$·kWh− 1 (Fig. 4).
on filtration surface area), to 100% (where the whole CAPEX is assumed Fig. 4 confirms that the higher the energy cost (and thus its contribu-
to be dependent on filtration surface area) (Fig. S3). As first estimate, tion to the overall desalination cost), the easier the economic justifica-
FO/RO CAPEX ratios were varied from 2 (which is more or less the cur- tion of FO–RO hybrids becomes. It is clear that a breakeven point for
rent situation, due to the expensive FO membranes and the lower FO–RO hybrids compared to RO desalination is only reached when

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
G. Blandin et al. / Desalination xxx (2014) xxx–xxx 5

0% 0%
0 10 20 30 40 50 0 10 20 30 40 50
(a) (b)

CAPEX savings (%)

CAPEX savings (%)


-50% -50%

FO 100% surface dependent


FO/RO CAPEX rao=0.5
FO 75% surface dependent
FO/RO CAPEX rao=1
FO 50% surface dependent
FO/RO CAPEX rao=2
FO 7% surface dependent
-100% Jw(L.m-2.h-1) -100% Jw (L.m-2.h-1)

Fig. 3. CAPEX savings (in %) of FO–RO hybrid vs. RO with permeation flux depending on the surface dependence of the capex (for FO/RO capex ratio = 1) (a) and the FO/RO CAPEX ratio (for
75% surface dependence) (b).

energy costs exceed 0.15$·kWh−1 and when a high permeation flux Results confirm that current FO performance (5–10 L·m−2·h−1 [5,6,
(30 L·m−2·h−1) is used. Above 0.15 kWh−1, the energy costs savings 12]) does not allow economic savings for seawater desalination com-
become predominant and additional CAPEX is easier to justify, allowing pared to RO, even when integrating the optimistic hypothesis of 30%
for lower permeation flux. When the energy cost is thus equal to or OPEX savings (Fig. 5a and b). However, from Fig. 5a it is clear that
below 0.1 kWh−1, energy savings are insufficient to cover the CAPEX in- (when considering 0.1$·kWh−1 as energy cost) 30% OPEX savings and
crease. In that case, potential OPEX costs savings should be considered high permeation flux (N20 L·m−2·h−1) will result in more economical
to justify the economic interest for FO–RO hybrid. operation of FO–RO hybrids than simple RO. Overall, savings become
even more significant when higher energy cost are considered (see
2.2.3. Operational expenditures (OPEX) Fig. 5b). This demonstrates the importance of OPEX in the evaluation
Implementing FO before RO could lead to potential OPEX savings of FO–RO hybrid and RO economics.
(maintenance, membrane replacement, chemical cleaning) thanks to
the lower pressure operation in RO and significantly less fouling in FO 2.3. Importance of permeation flux
[13]. This results in a more continuous process, less need for chemicals
and potentially a longer membrane lifetime. In the analysis below, the It is thus more than clear that permeation flux will play a key role in
FO wastewater feed is considered pre-treated and therefore no addi- FO–RO hybrid economical sustainability. From the realistic scenarios
tional treatment is deemed necessary. No clear values for reduction in developed in Fig. 5a and b, the conclusion can be drawn that an average
OPEX for FO, compared to RO, are given by the literature, and a sensitiv- flux 30 L·m−2·h−1 constitutes a breakeven point, above which FO–RO
ity analysis is carried out. As a first approximation, reductions in opera- hybrids become more economical than RO.
tional costs of 15 and 30% were considered for two energy cost scenarios It is however also important to differentiate between the initial and
(0.1 and 0.15 kWh−1) in Fig. 5a and b. the average fluxes in the FO installation for process design. Indeed, in

20%

10%

0%
0 5 10 15 20 25 30 35 40 45 50
Overall savings (%)

-10%

-20%

0.05$.kWh-1
-30% 0.1$.kWh-1
0.15$.kWh-1
0.2$.kWh-1
-40%

-50%
Jw (L.m-2.h-1)

Fig. 4. Impact of energy costs on the overall savings of FO–RO hybrid vs. RO (CAPEX FO assumptions: 75% flux dependent, FO/RO ratio = 1; similar OPEX for FO and RO).

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
6 G. Blandin et al. / Desalination xxx (2014) xxx–xxx

20% 20%
(a) (b)
10% 10%

Overall savings (%)


Overall savings (%) 0% 0%
0 10 20 30 40 50 0 10 20 30 40 50
-10% -10%

-20% -20%
0% OPEX savings 0% OPEX savings
-30% -30%
15% OPEX savings 15% OPEX savings
-40% 30% OPEX savings -40% 30% OPEX savings

-50% -50%
Jw(L.m-2.h-1) Jw (L.m-2.h-1)

Fig. 5. Impact of OPEX savings hypotheses of FO–RO hybrid on the overall savings of FO–RO hybrid vs. RO for energy cost of 0.1$·kWh−1 (a) and 0.15$·kWh−1 (b) (CAPEX FO assumptions:
75% flux dependent, FO/RO ratio = 1).

the literature, FO membrane performance is usually reported using the this study [6]. The actual osmotic pressure at the membrane surface
optimum driving force at the initial stage of filtration (i.e., at extremely (πFm) is dependent on the permeation flux and, considering concentra-
low feed water recovery). In practice, as a result of water permeation, tive ECP, could be written as in Eq. (3).
RSD, and increasing recovery, a dilution of the draw solution and an
JW
increase in feed concentration will occur, decreasing the driving force π Fm ¼ πFbulk e k ð3Þ
and thus the permeation flux. Typically, considering a reasonable FO re-
covery of 50%, the aforementioned objective of an average flux of
with πFbulk the osmotic pressure of the feed solution and k the mass
30 L·m−2·h−1 can only be obtained if the FO membranes reach an ini-
transfer coefficient in the feed channel, as defined in Eq. (4).
tial permeation flux of 40 L·m−2·h−1 (Fig. S4).
In the following section, based on the solution–diffusion model, the Sh  D
potential to reach the benchmark flux of 40 L·m−2·h−1 in FO will be k¼ ð4Þ
dh
critically discussed. More specifically, a sensitivity analysis will be per-
formed on two approaches that can be used to increase permeation
with Sh the Sherwood number, D the self-diffusion coefficient of the
flux: (1) further membrane development and (2) the use of hydraulic
solute and dh the hydraulic parameter of the flow channel [48–50]. On
pressure on the feed water as additional driving force (PAO).
the draw side, dilutive ICP occurs in the support layer of the membrane,
decreasing the osmotic pressure at the active layer of the membrane:
3. Flux modelling
− Jw K
3.1. Theoretical background πDm ¼ πDbulk e ð5Þ

In osmotic membrane processes, water flux is generally dependent with πDbulk the osmotic pressure of the draw solution and K the resis-
of osmotic and hydraulic pressures, and the specific properties of the tance to solute diffusion (s.m−1) defined as in Eq. (6):
membrane. In such processes, the well-known and most used equation
ts  τ S
to calculate water flux is based on the solution–diffusion theory [16,28] K¼ ¼ ð6Þ
Dε D
(Eq. (1)).
where ts is the thickness (m), τ the tortuosity and ε the porosity of the
JW ¼ A  ðΔP−Δπm Þ ð1Þ
support layer. These latter parameters are characteristics of the mem-
brane, and the S factor in which they are combined is commonly re-
where Jw is the water flux (L·m−2·h−1), A is the membrane-specific
ferred to as the structural parameter [51]. By integrating Eqs. (5) and
pure water permeability (L·m− 2·h−1·bar− 1), ΔP is the applied hy-
(6) into Eqs. (1), (7) for the flux in FO, can be derived [38]:
draulic pressure differential (bar) between the feed (PF) and the draw
(PD) and Δπ is the osmotic pressure differential (bar) between the  JW

− J K
feed (πDm) and the draw (πFm) across the active layer of the membrane. JW ¼ A  πDbulk e w −πFbulk e k þ P F −PD : ð7Þ
As a result of concentration polarisation and to better account for ICP,
ECP and RSD, a reflection coefficient can be introduced (σ) into
Eq. (7) expresses the flux in FO (with or without applied pressure),
Eq. (1). This parameter refers to the ratio of the effective (Δπm) to the
assuming that there is no RSD. As a result of RSD, a further reduction
bulk (Δπbulk) osmotic pressure difference (Eq. (2)).
in osmotic driving force is observed. Following existing models [28],
JW ¼ A  ðΔP−σ  Δπbulk Þ ð2Þ the impact of RSD on flux has been incorporated [52], and advanced so-
lution–diffusion models considering ICP, ECP and RSD have been obtain-
ed [48]. In the case of PAO integration of the RSD effect results in Eq. (8):
In FO operation, especially when wastewater is used as feed, the
membrane orientation with the active layer facing the feed solution is 0 1
preferred to avoid irreversible fouling and has proven to be more effi- JW
BπDbulk  e− Jw K −πFbulk  e k C
cient on long term operation [46,47] despite higher ICP on the draw JW ¼AB
@   þ P F −PD C
A ð8Þ
side [27]. In such configuration, the impact of fouling on the flux is B − Jw K
JW
1− : e −e k
very limited and will thus not be considered, as a first approach, in JW

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
G. Blandin et al. / Desalination xxx (2014) xxx–xxx 7

with B the membrane-specific solute permeability. This model con- model. These parameters have already been pointed out as being highly
siders important limitations to the mass transfer in FO and PAO and variable and dependent on operational conditions during the measure-
the influence of intrinsic membrane characteristics (A, B and S through ment [59].
K) on these limitations, in addition to the effect of hydrodynamics Despite intensive research, the threshold value for initial flux (J0) of
(k) and driving forces (ΔΠ and ΔP). Thus, through Eq. (8), a sensitivity 40 L·m−2·h−1 is not reached experimentally for any of the membranes.
analysis of the impact of one or several membrane parameters on the Moreover, most of the high flux values reported in the literature were
overall efficiency of FO systems (in terms of flux) can be performed. It obtained using model feed and draw solutions with a high osmotic pres-
is therefore possible to assess which membrane parameters need to sure on the draw side (N45 bar osmotic) and pure water on the feed
be altered (independently) to obtain the initial flux benchmark of (except Oasys TFC) to increase the driving force and avoid ECP. This is
40 L.m−2.h−1 as set out above. not representative of any current industrial configuration. The modelled
fluxes for the WW–SW systems show significantly lower permeation
3.2. Operating conditions fluxes than those observed in the lab-scale testing: all fluxes in the
WW–SW system were modelled to be below 26.6 L·m−2·h−1. It was
Based on membrane characteristics and operation parameter values also observed that fluxes above 10 L·m−2·h−1 were generally obtained
found in the literature, FO fluxes for different commercial and lab-scale with membranes featuring high A (N1.5 L·m−2·h− 1) and low S
FO membranes were calculated using Eq. (8), and compared to experi- (b 67 μm). Some examples confirmed that either too high S (e.g., SW30
mental results [21]. Then, to better mimic wastewater (WW [7,12]) without fabric) or a too low A (e.g. HTI CTA, TFC FO solution 3) are prohib-
and seawater (SW) salinity, model calculations were conducted consid- itive to allow high permeation fluxes [37,56].
ering a WW–SW system, featuring initial salinities of feed and draw so- Interestingly, among the commercial membranes and according to
lutions of 1.2 g.L−1 and 35 g.L−1 red sea salt (Red Sea Inc.) respectively. experimental literature values, the Oasys TFC membrane demonstrated
CTA (HTI, Albany, OR) membrane characteristics [37] were used as ref- the best performance. Most likely, this is due to the combination of a
erence and for model calculations unless otherwise specified. Specific high A (4.72 L·m− 2·h−1) and relatively low S (365 μm). The main
and combined impacts of membrane intrinsic parameters (A, B, S), drawback of a high A value for FO membranes, remains the accompany-
and hydraulic pressure on permeation flux were evaluated. k was con- ing increase of B value, leading to higher RSD [62]. This is partly
sidered constant and similar to our former study [36]. Diffusion coeffi- confirmed here, since the Oasys membrane also features one of the
cients for NaCl and MgCl2 were obtained from [48,53]. The boundary highest B values (1.2 × 10−7 m·s−1). The recently synthesised nanofi-
conditions described in Table 1 were considered with regards to mem- ber support layers (PI-FO, NC PVA PA, NC PAN PA in Table 2 [57,
brane characteristics and process conditions found in [21]. In PAO con- 60–62]) appears to be very promising. The example of NC PVA PA dem-
ditions, membrane deformation was considered as described in our onstrates a good fitting of the model with experimental values and also
former work [36]. shows the highest modelled flux in industrial conditions (i.e.
26.6 L·m−2·h− 1). The NC PVA PA membrane manages to lower the
4. Current flux and potential for improvement structural parameter of the membrane below 100 μm, using a less tortu-
ous support and, as a result, demonstrates significantly decreased ICP.
4.1. Current membrane performance As an intermediate conclusion, it is clear that all the existing mem-
branes considered in this study remain below our defined flux criterion
In order to assess and contrast accurately the hydraulic perfor- for FO sustainability (i.e., J0 = 40 L·m−2·h−1). A detailed sensitivity
mances of a number of FO membranes (both commercially available analysis will be conducted in the next paragraph to evaluate if improve-
and under small-scale development), a number of flux values have ment of membrane characteristics has the potential to ever reach the
been calculated and reported in Table 2. At first, reported flux perfor- flux threshold.
mances from literature (Jlit) were compared with fluxes modelled
using Eq. (8), based on the operating conditions mentioned in the liter- 4.2. Improving membrane properties
ature (Jmod) to validate the model relevance. Then, flux values were all
normalised under the WW–SW model and calculated based on Eq. (8). Both membrane characteristics of active layer (A,B) and support
Despite the assumption of similar hydrodynamic conditions for all layer (S) will ultimately impact the permeation flux. As such, previous
membranes (constant k), a good fitting of the model with the reported studies [51,63] have focussed on improving both membrane permeabil-
experimental values is observed. Indeed, differences were lower than ity and selectivity (by decreasing the B/A ratio) as well as decreasing the
17% (or 2 L·m− 2·h− 1) in most of the cases and for a wide range of membrane structural parameter [57,60–62]. Sensitivity analysis on the
membrane types and permeation fluxes. This shows the validity of respective impact of A, B and S values on the initial water flux in FO
Eq. (8) to model fluxes in FO. There are two exceptions to this trend: was performed (Fig. 6).
the Oasys TFC membrane exhibits a much higher flux than expected A non-linear increase of flux is observed with the membrane perme-
by the model, while the opposite is observed with the NC PAN PA. One ability (A) as a consequence of exponentially enhanced CP (both ICP and
of the reasons for this discrepancy could be the difficulty to measure ECP) with the permeation flux. Typically, in the defined conditions and
the support layer characteristics, i.e. K and S, which were used in the even considering extremely high A values (up to 10 L·m−2·h−1·bar−1),
flux remains limited below 25 L·m−2·h−1 (Fig. 6a, based on HTI CTA
membrane S value). It is thus clear that only improving membrane per-
Table 1 meability is not sufficient to reach threshold fluxes of 40 L·m−2·h−1
Model boundaries and HTI CTA characteristics and experimental conditions used in [36]. when seawater is used as draw solution. Results also show that salt per-
Model boundaries HTI CTA meability has a relatively lower influence on the permeation flux; in the
extreme condition of very poor salt rejection (B = 1 × 10−6 m·s−1), the
Lower range Higher range
permeation flux is impacted by less than 10%. Although not directly af-
A L·m−2·h−1·bar−1 0.5 10 0.6
fecting flux, poorer salt rejection may limit the potential to reach high
B ×10−7 m·s−1 0.01 10 0.5
S ×10−6 m 100 1000 678 recovery as a direct consequence of a huge decrease in the osmotic driv-
k ×10−5 m·s−1 1.1 1.1 ing force by an increasing salt concentration in the feed (Fig. S5 of
Pfeed Bar 0 10 0–10 supporting document). In the specific context of FO–RO hybrid operated
πFbulk Bar 0.9 0.9 with WW as feed and SW as draw, very high recovery (N95%) is gener-
πDbulk Bar 24.5 24.5
ally not a requirement, therefore further decrease in salt permeability

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
8 G. Blandin et al. / Desalination xxx (2014) xxx–xxx

Table 2
Reported membrane characteristics and initial permeation flux (Jlit), compared with flux values obtained with solution diffusion model obtained using initial study conditions (Jmod) and
conditions used in this study (Jmod WW–SW).

Membrane reference A B S Jlit Jmod Jmod Ref.


Ww–Sw

L·m−2·h−1·bar−1 ×10−7 m·s−1 ×10−6 m L·m−2·h−1 L·m−2·h−1 L·m−2·h−1

Commercial membranes
HTI CTA 0.55 0.48 463 10.1 9.2 6.8 [37]
HTI CTA-W 0.32 0.04 1000 5.0 3.6 3.6 [54]
HTI CTA-NW 0.47 0.27 1000 4.4 4.4 4.4 [54]
HTI TFC 1.63 0.83 690 10.0 11.3 9.6 [37]
Oasys TFC 4.72 1.20 365 30.0 22.6 20.7 [37]

Laboratory scale development


SUB 90 0.97 0.53 498 8.1 9.7 6.2 [53]
TFC-FO 1.14 1.30 492 18.1 16.6 9.8 [34]
TFC 9PSf-100NMP 1.63 2.30 389 20.5 18.5 12.9 [32]
TFC-1 1.15 0.47 710 9.5 8.1 8.2 [54]
TFC-2 1.80 0.94 670 12.0 10.1 10.1 [54]
TFC 1.44 0.39 782 7.0 6.0 8.5 [55]
TFC PES/SPSF 0.77 0.31 238 26.0 24.1 10.8 [33]
TFC FO solution 3 0.73 0.69 324 21.0 19.8 9.3 [56]
PI-FO 1.25 0.30 450 13.4 11.1 10.8 [57]
CAP-II-TFC 1.42 0.37 695 16.7 16.1 9.1 [30]
A-FO 0.94 0.81 1370 5.0 5.3 5.1 [58]
B-FO 2.23 0.56 595 14.0 12.5 11.9 [58]
#C-FO 3.50 0.62 550 19.0 15.7 14.6 [59]
TFC-PES 1.64 0.78 252 32.1 32.4 16.3 [31]
NC PVA PA 1.69 0.67 66 27.2 27.2 26.6 [60]
NC PAN PA 2.04 4.40 109 34.9 47.9 24.8 [61]

RO membrane
SW30 NW 1.28 0.80 9583 2.2 2.1 1.4 [34]
SW30 (no fabric) 1.46 1.10 2155 7.3 6.6 4.4 [34]

(tighter membrane) is not expected to lead to any substantial improve- efficiently impact permeation flux, as a result of low ICP, up until the
ment to the process, and could be detrimental to the water permeability. point where ECP becomes limiting (which might be the case at high
As described in Eq. (5), ICP is correlated to the structural parameter feed water recoveries). In this study, very distinct A-driven and S-
(S) of the membrane support layer. As such, an optimised S value is ex- driven regions could not be recognized, in contrast to what was men-
pected to have a significant effect on flux, by limiting ICP and thus in- tioned in the literature [57], indicating that A and S are both key param-
creasing osmotic pressure driving force efficiency. Permeation flux as eters that need to change simultaneously to obtain high permeation
a function of different structural parameter (S) values for different fluxes. Typically, a membrane with a combination of a high A value of
membrane pure water permeabilities (A) is shown in Fig. 6b. As expect- 5 L·m−2·h−1 and a small S value of 100 μm would allow to reach the
ed, decrease of the S value leads to substantial improvement in the per- desired initial flux. According to the current state-of-the-art, mem-
meation flux. Typical values for S of 300 to 500 μm (as observed with branes with such low S-value [57,60,61] were synthesised recently
current commercial membranes) are a limiting factor to reach accept- thanks to nanofiber support materials, but to our knowledge, none of
able fluxes. Reaching the threshold flux of 40 L·m−2·h−1 is only possi- the developed FO membranes have reached the required level in terms
ble for S around 100 μm. In that case, each increase of A value will also of A (N5 L·m−2·h−1·bar−1). Moreover, the mechanical resistance of

60 (a) 60
S=100μm (b)
50 B=1x10-6 m.s-1 50 S=300μm
B=1x10-7 m.s-1 S=500μm
40 B=1x10-8 m.s-1 40
J0 (L.m-2.h-1)
J0 (L.m-2.h-1)

B=1x10-9 m.s-1
30 30

20 20

10 10

0 0
0 5 10 0 5 10
A (L.m-2.h-1.bar-1) A (L.m-2.h-1.bar-1)

Fig. 6. Respective and combined impact of (a) water and solute permeability (A and B) and (b) structural parameter (S) on initial permeation flux (J0) (using the WW–SW model).

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
G. Blandin et al. / Desalination xxx (2014) xxx–xxx 9

very low S-value membranes is questionable. A recent study [60] dem- mechanical resistance and, as a result, enhanced deformation and per-
onstrated good mechanical stability of low S-value membranes during meability. Moreover, improvement of membrane characteristics (A
formulation, but the need for further improvement of membrane me- and S) will be further beneficial to PAO, limiting the hydraulic pressure
chanical strength was pointed out. to be applied. In addition to the beneficial impact on the permeation
flux, hydraulic pressure also limits RSD [36], allowing not only for higher
4.3. Implementing hydraulic pressure as additional driving force permeation flux, but also a higher recovery (Fig. 8).
One major drawback of PAO is the additional energy needed for feed
In PAO operation, hydraulic pressure is added to the feed water to pressurisation. To allow for fair comparison between FO and PAO, a
assist the osmotic driving force and enhance permeation flux. It was sensitivity analysis has been conducted to consider the trade-off be-
also reported that the HTI CTA membrane suffers from deformation tween savings in required membrane surface area and thus CAPEX
due to the compression of the membrane on the spacers [36] leading costs when using pressure, and the increased energy costs required
to increasing membrane permeability. The model was first validated for feed pressurisation. Impact of hydraulic pressure on recovery and as-
by comparison and fitting with experimental data from 0 to 10 bar sociated membrane surface area was calculated using HTI CTA mem-
(Fig. S6 from supporting document). Model calculations were then car- brane characteristics (Fig. 8a). FO and PAO energy consumptions were
ried out at 0 and 10 bar of applied hydraulic pressure on the feed side to calculated based on the specific energy consumption of the feed and
estimate the potential for PAO within the FO-RO hybrid process. The draw pumps present in the system. An additional hydraulic pressure
modelling was carried out with and without incorporation of mem- of 0.5 bar on both feed and draw sides of the process was integrated
brane deformation and its impact on the pure water permeability into the model to consider frictional losses in the system. Required
(Fig. 7). pumping energy power was calculated using Eq. (9).
In the FO–RO hybrid process configuration (Fig. 7), a strong response
of flux to hydraulic pressure is observed, especially for higher water Q  ðPoutlet −Pinlet Þ
Wpump ¼ ð9Þ
permeability membranes. This is the result of the (almost) linear corre- η
lation of permeation flux with water permeability and hydraulic
pressure (Eq. (1)), since hydraulic pressure is not affected by ICP. Inter- with Wpump the pump power, Q the flow entering the pump, Pinlet
estingly, a discrepancy between the fluxes reached by low and high S the pressure of the flow inlet stream, Poutlet the pressure of the outlet
value membranes remains apparent and demonstrates that the efficien- stream and η the pump efficiency, fixed at 90% in this study. Pumping
cy by which the osmotic pressure is used, is still very important in costs per m−3 of FO/PAO permeate were then calculated considering
determining flux. For higher S membranes, the osmotic pressure is inef- 50% FO recovery and energy costs estimated at 0.1$·kWh−1.
ficiently used due to the high ICP and flux increase is mainly due to the When operating in PAO mode, owing to the additional driving
pressure effect, for lower S membranes, ICP is low and the contribution force and membrane deformation, the filtration surface area can be
of osmotic pressure is still clear. When incorporating the effect of mem- decreased by more than 3 times to reach the same recovery (i.e. 90
brane deformation, the same trends with increasing pressure are seen, vs 27 m2·m− 3·h− 1 for 0 and 10 bar respectively) (Fig. 8a). As a re-
but they are even more pronounced and even more flux improvement sult, CAPEX is reduced correspondingly (Fig. 8b). It can be observed
is observed with increasing pressure. Due to the applied hydraulic pres- that PAO operation leads to an overall significant cost decrease,
sure, the apparent water permeability increases. even with its higher energy requirements. Above an applied pressure
PAO is thus confirmed to be a potential strategy to overcome current of 6 bar, pumping energy costs become too detrimental, resulting in
limitations of FO, and may help to reach the 40 L·m−2·h−1 permeation only marginal additional savings. This confirms that the application
flux threshold. The threshold flux is already theoretically obtained for of moderate additional pressure on the feed side could be beneficial
membranes with a relatively thick support layer (500 μm) at limited hy- to overall FO–RO hybrid process economics. Higher operating pres-
draulic pressure applied (10 bar). When membrane deformation is sures (above 6 bar) are not recommended since these pressures
taken into account, the 40 L·m−2·h− 1 threshold is reached at even may lead to higher materials costs to withstand pressure and poten-
lower operating pressure (8.5 bar). As such, PAO offers an alternative tially more severe fouling (in addition to the observed economic
to development of very thin membranes with low S values and poten- limitation).
tially also low mechanical resistance, to reach the high fluxes needed.
Importantly also, the extent of membrane deformation has been consid- 5. Conclusion
ered only according to existing data on HTI CTA behaviour in this study.
Most likely the development of thinner support layer may result in less FO can present a sustainable opportunity for future fresh water pro-
duction, but important improvements are still needed to overcome cur-
80 rent limitations. The initial economic study in this paper demonstrated
70
the need for a better knowledge of FO economics, since the additional
A= 0.6L.m-2.h-1.bar-1 & S=500μm
CAPEX of an FO–RO hybrid system can be detrimental in comparison
60 A= 5L.m-2.h-1.bar-1 & S=500μm with conventional RO. FO–RO hybrid integration has been shown to
A= 5L.m-2.h-1.bar-1 & S=100μm be beneficial only if substantial energy and OPEX costs savings are ob-
50
tained by using the FO. A first economic evaluation proposed in this
J0: L.m-2.h-1

40 study has determined a 30 L·m−2·h− 1 as average permeation flux


(equivalent to 40 L·m−2·h−1 for initial permeation flux) as threshold
30 flux value to guarantee FO economic sustainability. Despite the high
fluxes claimed by the latest membrane developers (both commercial
20
and in lab settings), none of them currently allows to reach this thresh-
10 old FO flux value, especially when real operational conditions are con-
sidered (i.e., real feed and draw solutions, not synthetic solutions; and
0 higher recoveries with associated osmotic driving force decrease). To
FO (0bar) PAO (10bar) PAO (10bar & mbr
deformaon)
reach the FO economically sustainable flux, further membrane improve-
ments are therefore needed; typically, a combination of higher water
Fig. 7. Impact of pure water permeability, structural parameter, osmotic and hydraulic permeability (A N 5 L·m−2·h−1·bar−1) and lower structural parameter
pressure applied on the feed side on the permeation flux for FO–RO hybrid process. (S b 100 μm) are pertinent targets. Membrane mechanical resistance

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
10 G. Blandin et al. / Desalination xxx (2014) xxx–xxx

100%
0.4
(a) (b)
FO/PAO CAPEX
80% 0.3 FO/PAO energy costs
Recovery (%)

$.m-3
60%
0.2
FO
40% PAO 2bar
PAO 4bar 0.1
20% PAO 6bar
PAO 8bar
PAO 10bar 0.0
0% FO PAO PAO PAO PAO PAO
0 50 100 150 200 2bar 4bar 6bar 8bar 10bar
Filtraon surface area (m2.m-3.h)

Fig. 8. Impact of applied hydraulic pressure on (a) recovery and (b) FO/PAO costs per m3 of permeate (HTI CTA membrane characteristics, with deformation, WW-SW system, CAPEX FO
assumptions: 75% flux dependent, FO/RO ratio = 1).

will be a major challenge, since deformation has been already observed [13] S. Lee, et al., Comparison of fouling behavior in forward osmosis (FO) and reverse
osmosis (RO), J. Membr. Sci. 365 (1–2) (2010) 34–39.
in PAO mode with commercial high-S FO membrane. Alternatively, PAO [14] R. Semiat, J. Sapoznik, D. Hasson, Energy aspects in osmotic processes, Desalin.
proved to achieve higher flux with current commercial membranes, Water Treat. 15 (1–3) (2010) 228–235.
while demonstrating more favourable economics. Further work will [15] H. Takabatake, et al., More than 30% energy saving seawater desalination system by
combining with sewage reclamation, Desalin. Water Treat. 51 (4–6) (2012)
be needed to refine the economics especially considering fouling behav- 733–741.
iour in FO/PAO when higher flux of hydraulic pressure is applied; possi- [16] S. Zhao, et al., Recent developments in forward osmosis: opportunities and chal-
bly requiring more stringent cleaning procedure and further limiting lenges, J. Membr. Sci. 396 (2012) 1–21.
[17] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applications, and
the interest of FO–RO hybrid. recent developments, J. Membr. Sci. 281 (1–2) (2006) 70–87.
[18] T.S. Chung, et al., Emerging forward osmosis (FO) technologies and challenges
Acknowledgements ahead for clean water and clean energy applications, Curr. Opin. Chem. Eng. 1
(2012) 246–257.
[19] T.-S. Chung, et al., Forward osmosis processes: yesterday, today and tomorrow, De-
The authors acknowledge the financial support of the National Centre salination 287 (2012) 78–81.
of Excellence in Desalination Australia, which is funded by the Australian [20] Ge, Q., M. Ling, and T.-S. Chung, Draw Solutions for Forward Osmosis Processes: Devel-
opments, Challenges, and Prospects for the Future. J. Membr. Sci., (0).
Government through the National Urban Water and Desalination Plan. [21] C. Klaysom, et al., Forward and pressure retarded osmosis: potential solutions for
global challenges in energy and water supply, Chem. Soc. Rev. 42 (16) (2013)
Appendix A. Supplementary data 6959–6989.
[22] J. Su, et al., Forward osmosis: an emerging technology for sustainable supply of clean
water, Clean Techn. Environ. Policy (2012) 1–5.
Supplementary data to this article can be found online at http://dx. [23] N.A. Thompson, P.G. Nicoll, Forward osmosis desalination: a commercial reality, IDA
doi.org/10.1016/j.desal.2014.12.011. World Congress — Perth Convention and Exhibition Centre (PCEC), Perth, Western
Australia September 4–9, 2011, 2011.
[24] Helfer, F., C. Lemckert, and Y.G. Anissimov, Osmotic Power with Pressure Retarded
References Osmosis: Theory, Performance and Trends – A Review. J. Membr. Sci., (0).
[25] S.E. Skilhagen, Osmotic power — a new, renewable energy source, Desalin. Water
[1] J.J. Qin, W.C.L. Lay, K.A. Kekre, Recent developments and future challenges of for- Treat. 15 (1–3) (2010) 271–278.
ward osmosis for desalination: a review, Desalin. Water Treat. 39 (1–3) (2012) [26] C.Y. Tang, et al., Coupled effects of internal concentration polarization and fouling on
123–136. flux behavior of forward osmosis membranes during humic acid filtration, J. Membr.
[2] P. Nicoll, Forward osmosis is not to be ignored, Desalin. Water reuse 22 (4) (2013). Sci. 354 (1–2) (2010) 123–133.
[3] R.K. McGovern, J.H. Lienhard V, On the potential of forward osmosis to energetically [27] G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polarization in for-
outperform reverse osmosis desalination, J. Membr. Sci. 469 (0) (2014) 245–250. ward osmosis: role of membrane orientation, Desalination 197 (1–3) (2006) 1–8.
[4] R.W. Field, J.J. Wu, Mass transfer limitations in forward osmosis: are some potential [28] K.L. Lee, R.W. Baker, H.K. Lonsdale, Membranes for power generation by pressure-
applications overhyped? Desalination 318 (2013) 118–124. retarded osmosis, J. Membr. Sci. 8 (2) (1981) 141–171.
[5] O.A. Bamaga, et al., Hybrid FO/RO desalination system: preliminary assessment of [29] S. Loeb, et al., Effect of porous support fabric on osmosis through a Loeb-Sourirajan
osmotic energy recovery and designs of new FO membrane module configurations, type asymmetric membrane, J. Membr. Sci. 129 (2) (1997) 243–249.
Desalination 268 (1–3) (2011) 163–169. [30] X. Li, et al., Thin-film composite membranes and formation mechanism of thin-film
[6] T.Y. Cath, J.E. Drewes, C.D. Lundin, A Novel Hybrid Forward Osmosis Process for layers on hydrophilic cellulose acetate propionate substrates for forward osmosis
Drinking Water Augmentation using Impaired Water and Saline Water Sources, processes, Ind. Eng. Chem. Res. 51 (30) (2012) 10039–10050.
WERC, 2009. [31] P. Sukitpaneenit, T.S. Chung, High performance thin-film composite forward osmo-
[7] T.Y. Cath, et al., A multi-barrier osmotic dilution process for simultaneous desalina- sis hollow fiber membranes with macrovoid-free and highly porous structure for
tion and purification of impaired water, J. Membr. Sci. 362 (1–2) (2010) 417–426. sustainable water production, Environ. Sci. Technol. 46 (13) (2012) 7358–7365.
[8] V. Yangali-Quintanilla, et al., Lowering desalination costs by alternative desalination [32] A. Tiraferri, et al., Relating performance of thin-film composite forward osmosis
and water reuse scenarios, Desalin. Water Treat. (2014) (in press). membranes to support layer formation and structure, J. Membr. Sci. 367 (1–2)
[9] C.H. Tan, H.Y. Ng, A novel hybrid forward osmosis–nanofiltration (FO-NF) process (2011) 340–352.
for seawater desalination: draw solution selection and system configuration, [33] K.Y. Wang, T.S. Chung, G. Amy, Developing thin-film-composite forward osmosis
Desalin. Water Treat. 13 (1–3) (2010) 356–361. membranes on the PES/SPSf substrate through interfacial polymerization, AIChE J
[10] S.M. Park, et al., Optimization of hybrid system consisting of forward osmosis and 58 (3) (2012) 770–781.
reverse osmosis: a Monte Carlo simulation approach, Desalin. Water Treat. 43 [34] N.Y. Yip, et al., High performance thin-film composite forward osmosis membrane,
(1–3) (2012) 274–280. Environ. Sci. Technol. 44 (10) (2010) 3812–3818.
[11] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, Desalination by ammonia–carbon di- [35] T.Y. Cath, et al., Standard methodology for evaluating membrane performance in os-
oxide forward osmosis: influence of draw and feed solution concentrations on pro- motically driven membrane processes, Desalination 312 (2013) 31–38.
cess performance, J. Membr. Sci. 278 (1–2) (2006) 114–123. [36] G. Blandin, et al., Validation of assisted forward osmosis (AFO) process: impact of
[12] V. Yangali-Quintanilla, et al., Indirect desalination of Red Sea water with forward os- hydraulic pressure, J. Membr. Sci. 447 (2013) 1–11.
mosis and low pressure reverse osmosis for water reuse, Desalination 280 (1–3) [37] B.D. Coday, et al., Effects of transmembrane hydraulic pressure on performance of
(2011) 160–166. forward osmosis membranes, Environ. Sci. Technol. 47 (5) (2013) 2386–2393.

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011
G. Blandin et al. / Desalination xxx (2014) xxx–xxx 11

[38] T. Yun, et al., Flux behavior and membrane fouling in pressure-assisted forward os- [52] W.A. Phillip, J.S. Yong, M. Elimelech, Reverse draw solute permeation in forward
mosis, Desalin. Water Treat. 52 (4-6) (2013) 564–569. osmosis: modeling and experiments, Environ. Sci. Technol. 44 (13) (2010)
[39] Q. She, et al., Effect of feed spacer induced membrane deformation on the perfor- 5170–5176.
mance of pressure retarded osmosis (PRO): implications for PRO process operation, [53] J. Su, T.S. Chung, Sublayer structure and reflection coefficient and their effects on
J. Membr. Sci. 445 (2013) 170–182. concentration polarization and membrane performance in FO processes, J. Membr.
[40] Desaldata.com, Website desaldata, www.desaldata.com. Sci. 376 (1–2) (2011) 214–224.
[41] G. Codemo, et al., Hamriyah SWRO desalination plant—largest sea water IMS plant, [54] J. Wei, et al., Synthesis and characterization of flat-sheet thin film composite for-
IDA World Congress–Atlantis, The Palm–Dubai, UAE November, 2009. ward osmosis membranes, J. Membr. Sci. 372 (1–2) (2011) 292–302.
[42] lenntech, Reverse Osmosis Plant, http://www.lenntech.com/systems/reverse-osmosis/ [55] N. Ma, et al., Zeolite-polyamide thin film nanocomposite membranes: towards en-
ro/rosmosis.htm2014. hanced performance for forward osmosis, J. Membr. Sci. 405–406 (2012) 149–157.
[43] nanoH2O, NanoH2O Case Studies, http://www.nanoh2o.com/case-studies. [56] N. Widjojo, et al., The role of sulphonated polymer and macrovoid-free structure in
[44] Poriferanano, Element Packing Density Comparison, http://www.poriferanano.com/ the support layer for thin-film composite (TFC) forward osmosis (FO) membranes, J.
technology/forward-osmosis-membrane/2013. Membr. Sci. 383 (1–2) (2011) 214–223.
[45] K. Saito, et al., Power generation with salinity gradient by pressure retarded osmosis [57] X. Song, Z. Liu, D.D. Sun, Nano gives the answer: breaking the bottleneck of internal
using concentrated brine from SWRO system and treated sewage as pure water, concentration polarization with a nanofiber composite forward osmosis membrane
Desalin. Water Treat. 41 (1–3) (2012) 114–121. for a high water production rate, Adv. Mater. 23 (29) (2011) 3256–3260.
[46] S. Zhao, L. Zou, D. Mulcahy, Effects of membrane orientation on process [58] R. Wang, et al., Characterization of novel forward osmosis hollow fiber membranes,
performance in forward osmosis applications, J. Membr. Sci. 382 (1–2) J. Membr. Sci. 355 (1–2) (2010) 158–167.
(2011) 308–315. [59] L. Shi, et al., Effect of substrate structure on the performance of thin-film composite for-
[47] S. Zou, et al., The role of physical and chemical parameters on forward osmosis mem- ward osmosis hollow fiber membranes, J. Membr. Sci. 382 (1–2) (2011) 116–123.
brane fouling during algae separation, J. Membr. Sci. 366 (1–2) (2011) 356–362. [60] J.M.C. Puguan, et al., Low internal concentration polarization in forward osmosis
[48] A. Achilli, T.Y. Cath, A.E. Childress, Power generation with pressure retarded osmosis: an membranes with hydrophilic crosslinked PVA nanofibers as porous support layer,
experimental and theoretical investigation, J. Membr. Sci. 343 (1–2) (2009) 42–52. Desalination 336 (2014) 24–31.
[49] G. Schock, A. Miquel, Mass transfer and pressure loss in spiral wound modules, De- [61] N.-N. Bui, et al., Electrospun nanofiber supported thin film composite membranes
salination 64 (C) (1987) 339–352. for engineered osmosis, J. Membr. Sci. 385–386 (2011) 10–19.
[50] A.R.D. Verliefde, et al., Influence of solute-membrane affinity on rejection of un- [62] N.Y. Yip, M. Elimelech, Performance limiting effects in power generation from salin-
charged organic solutes by nanofiltration membranes, Environ. Sci. Technol. 43 ity gradients by pressure retarded osmosis, Environ. Sci. Technol. 45 (23) (2011)
(7) (2009) 2400–2406. 10273–10282.
[51] A. Tiraferri, et al., A method for the simultaneous determination of transport and struc- [63] G.M. Geise, et al., Water permeability and water/salt selectivity tradeoff in polymers
tural parameters of forward osmosis membranes, J. Membr. Sci. 444 (2013) 523–538. for desalination, J. Membr. Sci. 369 (1–2) (2011) 130–138.

Please cite this article as: G. Blandin, et al., Opportunities to reach economic sustainability in forward osmosis–reverse osmosis hybrids for sea-
water desalination, Desalination (2014), http://dx.doi.org/10.1016/j.desal.2014.12.011

You might also like