Professional Documents
Culture Documents
Desalination
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 r t i c l e i n f o a b s t r a c t
Article history: Theoretical analysis of energy harvesting from concentrated brine of Sea Water Reverse Osmosis (SWRO) system
Received 14 October 2015 using Pressure Retarded Osmosis (PRO) is presented in this research. The mathematical model of SWRO–PRO hy-
Received in revised form 15 January 2016 brid system components such as SWRO unit, Energy Recovery Device (ERD), PRO unit and other auxiliary units
Accepted 21 January 2016
were discussed. The mathematical equations were solved adapting an object oriented “Modelica language”
Available online 28 January 2016
framework in Dymola software tool. The complex flowsheet models for six different SWRO–PRO hybrid config-
Keywords:
urations were created. The performance of the SWRO–PRO hybrid system configurations was studied. The pro-
Reverse osmosis cess and design parameters were optimized to reduce the Net Specific Energy Consumption (NSEC) of the
Desalination system. The optimization studies were performed using SQP technique that is available in optimization library
Pressure-Retarded Osmosis of Dymola. The possibility of using sea water (32,000 g/m3) and urban waste water (100–10,000 g/m3) as feed
SWRO–PRO hybrid system solution to the PRO for all the hybrid configurations were studied. Their performances were compared through
Modelling simulation and optimization studies. Among the six potentially viable SWRO–PRO configurations, the one
Optimization which does the direct mixing of diluted PRO draw outlet with feed water of SWRO aided to bring down the
NSEC by 49% in comparison with standard SWRO desalination system. This system does not require additional
ERD units and turbine at optimized process conditions, which are more expensive.
© 2016 Elsevier B.V. All rights reserved.
⁎ Corresponding author.
E-mail address: senthilmurugan@iitg.ernet.in (S. S.).
http://dx.doi.org/10.1016/j.desal.2016.01.027
0011-9164/© 2016 Elsevier B.V. All rights reserved.
S. S., S. S. / Desalination 389 (2016) 78–97 79
case IV, instead of sending the entire DS outlet through the turbine, part LP pump 1 by the 2nd ERD. In systems IV and V, the use of ERD-2 is ex-
of it (i.e. equal to the product water flow rate of SWRO) is fed to the 2nd pected to have a positive impact on reducing the NSEC. However, an ad-
ERD. It is used to exchange the energy between DS outlet stream to HP ditional ERD is used along with the one which was employed for energy
pump inlet stream. In case V, the DS outlet is recirculated to both HP and recovery from SWRO reject stream. In the remaining two cases (i.e. VI
S. S., S. S. / Desalination 389 (2016) 78–97 81
Table 1
Attributes of SWRO–PRO hybrid systems.
System type ➔ Case I Case II Case III Case IV Case V Case VI Case VII
and VII), the direct mixing of DS outlet with SWRO feed is enabled (sim- van't Hoff factor, universal gas constant, operating temperature and
ilar to close circuit PRO and desalination [15]) in order to reduce the sea molecular weight of NaCl, respectively.
water pre-treatment unit capacity. Thus, the energy consumption asso- The permeate concentration equation is derived by combining the
ciated with the pre-treatment of sea water can be minimized. Case VI is solute flux equation of SK model and the concentration polarization
a kind of modified version of case IV, where the ERD-2 is removed and a equation based on the film theory model [16]
fraction of the DS water equal to the flow rate of SWRO product water is 0 σ 1
directly fed to the HP pump. The remaining fraction of the DS outlet is 1þ 1−e
− Jv ð1−σ Þ
=B
B 1−σ C
depressurized in the turbine. In case VII, a large fraction of the DS outlet C f ¼ Cp @ A ð3Þ
eð k Þ
Jv
the membrane is given by Cfi, Cdi, and Cs are FS inlet concentration, DS inlet concentrations and
feed side concentration at support layer, respectively.
Pdrop Internal concentration polarization (ICP) is a strong function of sol-
ΔPRO ¼ Pf– −Patm ð6Þ
2 ute resistance of porous support layer and water flux [9]. The effective
concentration difference across the membrane with ICP effect is predict-
Feed side mass transfer coefficient of SWRO membrane is a function ed by equation given below
of fluid velocity. The empirical equation for mass transfer coefficient is
Cs
given by [18].
Jw eð Jw Kp Þ þ B eð Jw Kp Þ
−1
Cfm Cdm
¼ ð19Þ
Q f þ Qr y Cdm B eð Jw Kp Þ −1 þ J
k¼x ð7Þ w
2
where x and y are empirical constants. where Kp is solute resistance in porous support layer.
The mass balance equations are given below. The pressure drop across the feed and draw channel is estimated by
using Darcy's law [17]
Q f ¼ Qr þ Qp ð8Þ
ΔP f ¼ Pfi −Pfo
LPRO D f;PRO ðQ fi þ Q fo Þ ð20Þ
Q p ¼ Jv Sm ð9Þ ΔP f ¼
2
Q f C f ¼ Q r Cr þ Q p Cp ð10Þ
ΔPd ¼ Pdi −Pdo
LPRO Dd;PRO ðQ di þ Q do Þ ð21Þ
where Q p, Sm, and Cr are flow rate of permeate (i.e. product water), ΔPd ¼
2
total area of SWRO membrane per train, and concentration of brine
respectively. where Pfi, Pfo, Df,PRO, Q fi, Q fo, LPRO, Pdi, Pdo, Dd,PRO, Q di and Q do are FS inlet
SWRO recovery is defined as, pressure, FS outlet pressure, FS side Darcy's law constant, FS inlet flow
rate, FS outlet flow rate, length of one PRO membrane pressure vessel,
Qp DS inlet pressure, DS outlet pressure, DS side Darcy's law constant, DS
RecvRO ¼ 100 ð11Þ
Qf inlet flow rate and DS outlet flow rate, respectively.
The theoretical maximum power density is achieved when the hy-
draulic pressure difference is equal to half the value of the effective os-
2.5. PRO model equations motic pressure difference [17]. The DS inlet pressure equation is given
below
Water flux and salt flux for PRO membrane are written as below [16,
18]. ΔΠ eff −ΔPd þ Δp f
Pdi ¼ þ Pfi ð22Þ
2
Jw ¼ APRO ½Δπeff −ΔPPRO ð12Þ
where ΔPd and ΔPf are draw side pressure drop and feed side pressure
iRT C
Δπeff ¼ 1− fm Cdm ð13Þ drop, respectively.
M Cdm Pressure difference across the membrane is written as [17]
2.6. ERD model equations at LP inlet, velocity at HP outlet, velocity at LP outlet, friction loss coeffi-
cient and density of feed water, respectively.
The simplified Bernoulli's equation for calculating LP outlet pressure Average velocity at LP and HP inlets
of ERD unit is given below [24].
Vli ¼ Q li Ar
ð31Þ
ηERD ðQ hi Phi −Q ho Pho Þ þ Q li Pli −K ηERD Q ho ρf Vhi 2 −Q lo ρb Vli 2
Plo ¼ Vhi ¼ Q hi Ar
ð32Þ
Q lo
ð30Þ where Ar is the cross sectional area of ERD inlets.
Small fraction of HP stream shall leak through piston edge to LP
where ηERD, Q hi, Q li, Q ho, Q lo, Phi, Pli, Pho, Plo, Vhi, Vli, Vho, Vlo, K, and ρf are stream to lubricate ERD's part movement which can be written in the
ERD efficiency, flow rate at HP inlet, flow rate at LP inlet, flow rate at HP mathematical form as below
outlet, flow rate at LP outlet, pressure at HP inlet, pressure at LP inlet,
pressure at HP outlet, pressure at LP outlet, velocity at HP inlet, velocity Lub flow ¼ Q hi Lub flow ratio ð33Þ
Table 2
Industrial SWRO operating data.
Feed water flow rate of SWRO Operating pressure of SWRO Recovery Product water flow rate Product concentration Reject pressure
Qf [m3/s] Pf [bar] [%] Qp [m3/s] CP [ppm] PR [bar]
0.29101 50.47 33.1 0.09340 0.09699 −3.84 44 42.99 2.30 48.23 48.337 −0.22
0.29105 55.81 39.6 0.11603 0.11534 0.59 49 46.46 5.18 54.72 54.721 0.00
0.29110 60.28 44.5 0.12809 0.1295 −1.10 52 49.58 4.65 59.22 59.224 −0.01
Note 1: The model parameters such A, B, σ, x, y, and Df are estimated by tuning model with experimental data.
Note 2: The experimental is taken from sea water desalination plant data reported in [26].
Note 3: The sea water concentration = 32,000 g/m3, temperature = 298 K.
84 S. S., S. S. / Desalination 389 (2016) 78–97
Q lo ¼ Q hi −Lub flow ð34Þ Applying mass balance across inlets and outlets of ERD
Q ho ¼ Q hi þ Q li −Q lo ð35Þ
where Lub_flow is the lubrication flow from HP side to LP side and
Lub_flow_ratio is the ratio between lubrication flow to HP inlet Due to high and low pressure stream mixing, the concentra-
flow. tions of outlet streams will be varying from the inlet streams.
Based on solute mass balance, salinity at high pressure outlet of ERD is 2.8. Mixer and splitter model equations
given by
Mixer is used to mix and distribute multiple streams. Splitter is used
Clo ¼ ðChi −Cli ÞMix=100 þ Cli ð36Þ to split and distribute multiple streams. The pressure drop across the
mixer and splitter is assumed as negligible. Mass balance for both
mixer and splitter is written as
where Chi, Clo, Cli, Cho, and Mix are concentration of brine inlet (high-
pressure), concentration of brine outlet (low-pressure), concentration
Xm Xn
of feed inlet (low-pressure), concentration of feed outlet (high-pressure) Q out;m ¼ Q in;n ð40Þ
1 1
and mixing percentage, respectively. From solute balance, the salinity at
high pressure outlet of ERD is given by
where m and n are total number of outputs and total number of inputs
Chi Q hi þ Cli Q li −Clo Q lo respectively.
Cho ¼ ð37Þ
Q ho
2.9. Turbine model equations
where Pin and Pout, PHU are pressure of inlet stream, pressure of outlet
stream and pressure head utilized to generate power by turbine,
Pout ¼ Pin þ PHP respectively.
The power generated by turbine is given by
PH Q in ð39Þ
Wconsumed ¼
ηpump
Wgen ¼ PHU Q in ηturbine ð43Þ
where Qin, Qout, Pin, Pout, PHP, and ηpump are the inlet flow rate, outlet
flow rate, pressure of inlet stream, pressure of outlet stream, pressure
head developed in the pump, and efficiency of the pump, respectively. where ηturbine is the efficiency of turbine cum generator system.
Table 3
Important input parameters of the system.
PRO
Hydro dynamic permeability (APRO) 1.61389 × 10−6 m/bar·s
Salt permeability (BPRO) 2.44 × 10−7 m/s
Multiplying constant in mass transfer coefficient equation (x) 0.0038 m−0.5 s−0.5
Power constant in mass transfer coefficient equation (y) 0.5 –
Feed channel Darcy's law constant (Df, PRO) 200 bar·s/m4
Draw channel Darcy's law constant (Dd, PRO) 200 bar·s/m4
Area of one membrane (SPRO) 7.43224 m2
Length of one PRO membrane (lPRO) 0.9626 m
Total number of membranes per pressure vessel (mPRO) 5 –
Total number of pressure vessels per train (NoP) 291 –
Universal gas constant (R) 0.00008314 m4·bar/mol·K
Molecular weight of NaCl (M) 58.5 g/mol
SWRO
Reflection coefficient (σ) 0.999 –
Water permeability coefficient of membrane (A) 5.0815 × 10−7 m/bar·s
Salt permeability coefficient of membrane (B) 4.34 × 10−7 m/s
Area of one membrane (SRO) 7.43224 m2
No. of membrane per pressure vessel (mRO) 7 –
van't Hoff factor (i) 2 –
Membrane length (l) 0.9626 m
Darcy's law constant on feed side (Df) 200 bar·s/m4
Constant in mass transfer coefficient equation (x) 0.0038 m−0.5 s−0.5
Constant in mass transfer coefficient equation (y) 0.5 –
Permeate pressure (Pp) 1 bar
ERD
Volumetric mixing (Mix) 5.2 %
Lubrication flow ratio (Lub_flow_ratio) 0.01 –
Density (ρ) 997 kg/m3
Efficiency of ERD (ηERD) 96 %
Friction coefficient in ERD (K) 0.0005 –
86 S. S., S. S. / Desalination 389 (2016) 78–97
Feed to SWRO
3.1.1. Modelling framework Net flow rate m3/s 0.291 0.23–0.3
Modelling and simulation were done in an object oriented, declara- Pressure bar 55 30–70
tive and multi-domain modelling language “Modelica” with a help of a FS to PRO
software called Dymola. The mathematical model equations of each FS–DS inlet flow ratio – 1 0.8–2
process units were programmed in Modelica language to create individ- Concentration (except for Case-II) g/m3 100 100–10,000
ual process units (i.e. SWRO Module, PRO module and ERD) with user DS to PRO
designed icons and graphically connected to build integrated flowsheet DS inlet pressure bar Eq. (22) 5–15
model.
Flowsheet parameters
Ratio of brine recirculation – 1 0.8–1
3.1.2. Individual process unit simulation
For verification purpose, the individual process unit models were
simulated in appropriate operating range. So that the integrated
flowsheet model can be simulated for wide range of operating condi- increasing CP effect. Concentration of permeate water decreases till
tions. The simulation results are analysed and verified with literature. feed pressure ≈33 bar and then increases with inlet pressure (Fig. 2c)
as expected for SWRO membranes [27]. The increase in inlet flow rate
3.1.3. SWRO unit reduces the CP effect which results into lower product water concentra-
The seawater is pumped to SWRO unit by using HP pump (Fig. 2a). tion but SEC increases due to increase in pumping energy consumption
The SWRO model and design parameters were taken from literature (Fig. 2d).
[25]. As shown in Table 2, the SWRO model parameters such as A, B,
σ, x, y, and Df were estimated by tuning model with experimental 3.1.4. PRO unit
data used in reference [26]. For simulation study, the sea water concen- Flowsheet for PRO model is shown in Fig. 3a. Waste water (100 g/m3)
tration is fixed as 32,000 g/m3 and the effect of feed pressure and flow was used as FS and SWRO brine (52,000 g/m3) was used as DS. The DS
rate on SWRO performance were verified. While varying pressure and FS inlet flow rates were fixed as 0.1745 m3/s by assuming 40% re-
or flow rate, the other variables were fixed to their default values i.e. covery in SWRO unit. Due to unavailability of experimental plant data
0.291 m3/s and 55 bar respectively. As shown in Fig. 2b, the SWRO re- for PRO unit, the PRO membrane parameters were taken from literature
covery and reject concentration increases with inlet pressure as expect- [28] and design parameters were calculated based on the PRO design
ed. For example, 40% recovery and 53,300 g/m3 reject concentration, capacity requirements as described in SWRO–PRO trains section. For ex-
which is essential for PRO draw solution inlet was achieved at 56 bar ample, five number of PRO modules should be connected in series per
pressure. pressure vessel to achieve 67% recovery in PRO unit (for given value
The SEC decreases even though the work done by the pump in- of membrane permeability = 1.61389 × 10−6 m/bar·s, surface area
creases with increase in inlet pressure (Fig. 2c). However, at higher per module = 7.43224 m2 and DS inlet pressure = 10 bar). The
feed pressure, the rate of decrease of SEC is relatively small due to model parameters of PRO process are given in Table 3.
The PRO system performance is analysed by calculating following polarization. As expected, the reverse trend is obtained for power
model outputs such as power density and recovery (Fig. 3). For a density and recovery with respect to FS flow rate (Fig. 3c)). As re-
fixed PRO membrane surface area, the dilution water flux was in- ported by many researchers [23,29–31], both recovery and power
creased with DS inlet flow rate that resulted into higher recovery density increased with respect to draw inlet concentration
(i.e., 1 − Q fo/Q fi). On the other hand the power density decreased (Fig. 3d). The concentration of FS and DS outlets were increasing
due to increased pressure drop in DS membrane channel (Fig. 3b). with DS inlet concentration and this is due to higher PRO recovery
This phenomenon was happening because the power density is high- (Fig. 3g). Recovery decreases and power density increases while in-
ly sensitive towards pressure drop than dilutive concentration creasing draw solution inlet pressure (Fig. 3e).
As shown in Fig. 3f, the FS inlet concentration is varied to study the above sea water concentration with FS inlet concentration greater
effect of an internal concentration polarization on recovery and power than 700 g/m3. Also the power density is reduced with reduced water
density. Due to severe ICP, the DS outlet concentration is increased flux.
3.1.5. ERD unit 3.1.6.1. Effect of SWRO feed pressure on SWRO–PRO system performance.
Energy Recovery Device (ERD) is an important equipment in The feed pressure to SWRO unit was varied from 30 to 70 bar and corre-
SWRO plant, to reduce the energy consumption. The efficiency of sponding variation in some of the important variables are plotted in Fig. 5.
ERD is a function of operating pressure and flow rate. As shown For cases I to V, the product water concentration decreases to a minimum
in Fig. 4, the HP stream is supplied from high pressure source value and then increased with feed pressure (Fig. 5a) and this is due to
(i.e. SWRO brine) and LP stream is supplied from low pressure the CP effect at membrane surface [20]. However, for the cases VI and
source (i.e. sea water feed). The energy between HP & LP inlet VII, the product water concentration increased with feed pressure. This
streams is exchanged with minimal mixing between inlet and outlet was occurring due to increase in SWRO feed concentration with direct
streams. mixing of diluted DS outlet and SWRO feed. For all cases, SWRO and
ERD model simulation results are presented in Fig. 4. In general, the PRO recoveries were increasing with membrane feed pressure (Fig. 5b
ERDs are designed such that the optimal efficiency may be achieved & c). The trend of SWRO and PRO recovery curves were almost same
when LP side inlet flow rate is equal to HP side inlet flow rate. Therefore, for cases I to V. However, for case II, PRO recovery was found to be too
the ERD model was simulated to verify the trend of HP outlet pressure low due to severe ICP. For cases VI and VII, both PRO and SWRO recoveries
and concentration with respect to HP inlet pressure and concentration were higher than the former cases at the pressure below 60 bar. Later
respectively. As expected, HP outlet pressure increases with HP inlet both recoveries were found to be lesser at above 60 bar. This phenome-
pressure (Fig. 4b) and HP outlet concentration increases with inlet con- non was occurring due to the PRO unit capacity limitation i.e. the DS con-
centration (Fig. 4c) [32]. centration can not to be reduced below the seawater concentration.
For the cases III to V, The power density was increasing till a maxi-
3.1.6. SWRO–PRO process flowsheet mum value attained and then decreased (Fig. 5d). To understand this
The performance of integrated flowsheet models (Fig. 1) was phenomenon in more detail, the effect of SWRO inlet pressure on effec-
analysed by simulating with respect to the operating conditions tive osmotic pressure difference across the PRO membrane and DECP
mentioned in Table 4. The efficiency of HP pump, LP pump and ERD were analysed for Case V and shown in Fig. 6. The PRO power density
were assumed as 80%, 83% and 96% respectively. FS inlet concentra- is proportional to water flux and DS outlet pressure. The water flux is
tion of PRO for the case II was fixed as 32,000 g/m3 (i.e. sea water). proportional to the osmotic pressure difference across the PRO mem-
The design parameter of individual units are given in Table 3. brane. As shown in Fig. 6a, the effective osmotic pressure difference
was found to be increasing and reached the maximum at 50 bar NSEC and product water concentration. As shown in Fig. 7, the cases
SWRO inlet pressure. The DS inlet flow rate was expected to decrease VI and VII were found to be the best by satisfying above definitions.
with SWRO recovery and this leads to severe DECP effect (Fig. 6b). For cases I to V, the trend of SWRO recovery, PRO recovery and power
The above described phenomenon was not happening for the cases density were found to be similar and this may be possible when
VI and VII, hence no maximum found within the simulation range. SWRO feed conditions are identical.
As shown in Fig. 5e, in case VII, the direct mixing of DS outlet with
SWRO feed aids to lowest NSEC among other cases. The optimum 3.1.6.3. Effect of PRO DS inlet pressure on SWRO–PRO system performance.
NSEC with respect to SWRO feed pressure for cases I to V were esti- In all the previous sensitivity analysis, the DS inlet pressure was found
mated around 38 to 45 bar. For cases VI and VII, the NSEC decreases as a function of effective osmotic pressure difference across the PRO
with decrease in SWRO feed pressure. Power generation was expect- membrane. The theoretical maximum power density may be achieved
ed to increase with power density and reduction in NSEC. But, this when the hydraulic pressure difference among DS and FS were equal
behaviour was not observed while comparing power density and to half the value of effective osmotic pressure difference [22]. But, this
NSEC of systems grouped under “with” and “without” brine recircu- is not always true for a system when pressure and concentration vary
lation (i.e. II–V versus VI & VII). This was due to decreased SWRO unit along the length of the membrane module [19]. Therefore, the effect
power consumption by minimizing SWRO feed concentration for of DS inlet pressure on the system performance was studied (Fig. 8).
cases VI & VII. As expected, the DS inlet pressure does not have any impact on either
product water concentration or SWRO recovery for cases II and III,
3.1.6.2. Effect of net SWRO feed flow rate on SWRO–PRO system perfor- since there is no recirculation flow from PRO to SWRO. However, in
mance. As shown in Fig. 7, while increasing SWRO feed flow rate from other cases, recirculation flow from PRO to SWRO was enabled, which
0.23 to 0.3 m3/s, the SWRO recovery and PRO recovery are found to be leads to variation in product water concentration and SWRO recovery
decreasing due to pressure drop in membrane channel (Fig. 7a & b). with respect to DS inlet pressure. PRO recovery and NSEC were decreas-
On the other hand, power density was increasing due to reduced recov- ing and power density was increasing with DS inlet pressure for cases
ery in SWRO followed by increase in DS flow rate. However, for the case III–VII. SWRO and PRO recovery of cases VI and VII were higher than
II, the power density was decreasing due to severe ICP. The best SWRO– all other cases due to reduced CP effect. Choosing seawater as FS for
PRO hybrid system should provide high SWRO & PRO recovery, low the case II resulted into severe ICP that in turn leads to decreased
power density and increased NSEC with DS inlet pressure. It is evident 3.1.6.5. Effect of PRO FS concentration on SWRO–PRO system performance.
from the Fig. 5 that if the same study was done at the pressure above As reported by Altaee et al. [34], the PRO power density and recovery
60 bar, then the trend of SWRO recovery, PRO recovery, power density can be influenced by FS concentration. In this simulation, the FS inlet
and NSEC would have been reversed. concentration was varied from 100 g/m3 to 10,000 g/m3 and this may
be the case when waste water is considered as a source. As shown in
3.1.6.4. Effect of PRO FS/DS inlet ratio on SWRO–PRO system performance. Fig. 10b & c, both PRO recovery and power density were decreasing
The ratio between FS and DS inlet flow rate is very important. Altaee with FS inlet concentration and may reach zero asymptotically when
et al. reported that the DS flow rate has more impact on power density both FS and DS concentrations were equal. Therefore, NSEC was increas-
than the feed flow rate [33]. The relationship between DS/FS flow rate ing while FS inlet concentration was increased and reaches asymptotic
and power density are not linear. Increasing FS/DS ratio will decrease value for cases III to V (Fig. 10d). But for cases VI and VII, the NSEC
the effect of concentration polarization but cannot eliminate it, because was expected to increase due to the interaction between PRO and
very high flow rates would increase the overall power consumption SWRO streams by direct mixing. For example, in case VII the NSEC is
instead of improving power density. Therefore, optimal ratio is very lesser than other systems with FS inlet concentration lesser than
important to achieve maximum power output from PRO. In this 1300 g/m3, but for higher FS inlet concentration NSEC increased rapidly.
simulation study, since DS flow rate was fixed based on SWRO unit op-
erating conditions, the FS/DS ratio was varied from 0.8 to 2 and corre- 3.1.6.6. Effect of back mixing ratio on SWRO–PRO system performance. As
sponding hybrid system output variables are plotted in Fig. 9. As shown in Fig. 11, back mixing ratio is applicable for cases VI and VII
expected, FS/DS ratio does not have much impact on product water only. The effect of back mixing was studied for two different SWRO
concentration for all the cases. For cases IV to VII, SWRO recovery inlet pressures (i.e. 55 bar and 65 bar) by keeping all other operating
was increasing steadily, whereas PRO recovery was decreasing and condition as in Table 4. The flow ratio between DS flow to SWRO and
power density was increasing with FS/DS ratio as reported by Altaee DS flow at point P1 is defined as back mixing ratio (refer Fig. 1f &
et al. [33]. For the case II, both power density and PRO recovery were g) and will vary from 0 to 1. For the case VI, DS recirculation flow to
almost constant compared to other cases. Therefore, the NSEC in- HP pump was kept equal to the product water flow rate.
creased with FS/DS ratio for case II. For other cases, NSEC reaches to As shown in Fig. 11a, the DS outlet concentration was decreasing
minimum value then again increasing with FS/DS ratio. with back mixing ratio at 55 bar, and it was vice versa at 65 bar SWRO
feed pressure. When DS outlet concentration is realized below sea water respect to back mixing ratio. NSEC was found to be decreasing while in-
concentration, the SWRO feed is diluted and that leads to decreased DS creasing back mixing ratio for case VII (i.e. independent of SWRO feed
inlet concentration. Similarly, recovery of both PRO and SWRO units pressure variation). On the other hand, for case VI, the NSEC was mar-
was increasing at 55 bar and decreased at 65 bar of SWRO feed pressure. ginally decreasing at 55 bar and it was contrast at 65 bar.
The above phenomenon was in compliance with Fig. 5b. For a fixed FS
inlet concentration, the power density was proportional to DS concen- 3.2. Optimization
tration. Therefore, power density was also expected to perform similar
to DS outlet concentration profile (Fig. 11d). The power density and Reducing the SWRO–PRO system NSEC is one of the main objectives
SWRO recovery profiles were found to be opposite in nature with of this work. As presented in the SWRO–PRO process flowsheet section,
S. S., S. S. / Desalination 389 (2016) 78–97 93
the NSEC is a function of process parameters. The relative impact of pro- NSEC of all the hybrid systems were found to be higher than the
cess parameters on NSEC are presented in Table 5. The relative impact SWRO system.
on NSEC is defined based on the gain between NSEC and process param- The lowest possible NSEC was achieved for the case VII when the FS
eter and linearity between NSEC and process parameter. The impact inlet concentration was 100 g/m3. The calculated optimal NSEC was
levels are defined as follows: (i) high means gain is larger and nonlinear, 1.120 and 0.842 kWh/m3 for the production rate of 0.1164 (i.e. target)
(ii) medium means gain is larger and linear and (iii) low means gain and 0.054 m3/s. The corresponding energy saving with respect to case I
lesser and linear. The target production for SWRO was fixed as were 41.7% and 49.33% respectively. The optimized process operating
0.1164 m 3 /s. The NSEC was optimized for two scenarios i.e. with conditions for best cases are reported in Table 6. The optimized decision
and without target production at five different FS inlet concentrations variables such as FS/DS ratio, DS inlet pressure and DS back mixing ratio
(i.e. 100 g/m3, 500 g/m3, 1000 g/m3, 5000 g/m3 and 32,000 g/m3). The were found to be at their bounds for case VII. The SWRO–PRO system
performance of cases III–VII were compared with case I (i.e. existing does not require turbine for energy generation since all the brine water
SWRO system). The case II was also reported under case III with FS was recirculated at optimized process condition. The PRO operating pres-
inlet concentration of 32,000 g/m3, since both cases were similar with sure was hitting to lower bound i.e. PRO operation conditions are moving
respect to optimization problem formulation. towards Forward osmosis process operating condition. Similarly, for
higher FS concentration, it was found that the existing SWRO system
3.3. Objective function (case I) is energy efficient compared to hybrid SWRO–PRO configurations.
The optimization results with relaxed target product water flow were
The NSEC is defined as the amount of energy consumed by all pumps presented in Fig. 13. For all the cases, the calculated optimum NSEC with-
minus the energy generated in turbine per unit quantity of pure water out target production was lesser than NSEC with target production. None
produced from SWRO. of the system was able to meet the target production (0.1164 m3/s) at op-
timal NSEC and their optimal product water flow rate was not equal. The
Xj X1 ,
optimal product water flow rate for SWRO system was 0.045 m3/s. For
WPump − WTurb
minimize NSEC ¼ 1 0
ðQ p 3600Þ ð44Þ benchmarking purpose, the hybrid systems having production rate great-
er than 0.045 m3/s were compared among them with respect to NSEC.
The comparisons of SWRO–PRO hybrid systems from lowest to highest
where NSEC, j, WPump , W Turb , and Q P are net specific energy con- NSEC for different FS inlet concentration were given below.
sumption (kWh/m3), number of pumps, power consumption in 100 g/m3 is given as VII N VI N IV N III N I
pump (kW), power generation by turbine (kW) and product water 500 g/m3 is given as VII N VI N IV N I N III
flow rate (m3/s) respectively. 1000 g/m3 is given as VII N VI N IV N I N III
5000 g/m3 is given as V N I N IV N III
3.4. Constraints 32,000 g/m3 is given as I N V N IV N VII N III N VI
Above observations shows that, if low concentrated waste water
Based on the target production rate and other operational limita- is fed as a FS to PRO, then the hybrid SWRO–PRO system which
tions, constraints and bounds for decision variable are given in does the back mixing was found to be energy efficient solution for
Table 5. The decision variables which are not applicable for given hybrid desalination. For higher FS concentration, it was found that the
system were marked as not applicable (NA). The significance of decision existing SWRO system (case I) is energy efficient compared to hybrid
variable on NSEC was marked with three levels (i.e. high, medium, low). SWRO–PRO configurations. The optimum operating conditions for
If required, the variable with low significance may be excluded from op- the best performing case at five different FS inlet concentrations
timization decision variable for reducing complexity in optimization. were given in Table 6. The estimated optimal operating condition
However, in this study, all the applicable decision variables were consid- of PRO system concluded that the PRO was operated at high pressure
ered for optimizing NSEC. (i.e. PRO mode) for all configurations while target water production
constraint was relaxed.
3.5. Optimization results and discussion As concluded by Straub et al. [35], the above analysis was also
identified energy saving prospective for hybrid SWRO–PRO system
The optimum NSEC for case I was obtained to be 2.079 and after including energy losses in membrane, pumps and energy recovery
1.661 kWh/m3 for the production rate of 0.1164 (i.e. target) and devices. However, it will be feasible only when large amount of pre-
0.0448 m3/s respectively. For the fixed production rate, the optimum treated waste water is available at locations nearby SWRO desalination
NSEC for the cases III–VII are presented in the bar chat with respect to plant. The following important practical constraints has to be addressed
five different FS inlet concentrations Fig. 12. The energy saving with re- before implementation of such a hybrid SWRO–PRO system: (i) bio
spect to the existing SWRO system is also reported. The positive values fouling will be the likely biggest technical challenge for the combined
of energy saving represent that the SWRO–PRO hybrid system is effi- SWRO–PRO system due to use of waste water as FS and this issue may
cient than SWRO system. As shown in Fig. 12, for lower FS inlet concen- be resolved by implementing proper pre-treatment system; (ii) inte-
tration, all the SWRO–PRO hybrid configurations were found to be grated operation of the hybrid SWRO–PRO system requires large size
energy efficient than SWRO system. But at higher FS concentration, storage facility for waste water storage; and (iii) validation of overall
Table 5
Constraints for optimization and their relative impact on NSEC.
Effect on NSEC
Sl. no. Variable Bounds
Case I Case II Case III Case IV Case V Case VI Case VII
1 Product water flow rate Qp ≥ 0.1164 m3/s High High High High High High High
2 SWRO feed pressure 30 bar ≤ Pf ≥ 70 bar High High High High High High High
3 SWRO feed flow rate 0.234 m3/s ≤ Qf ≤ 0.291 m3/s Medium Medium Medium Medium Medium Low Medium
6 DS inlet pressure 5 bar ≤ Pdi ≤ 15 bar NA High Medium High High High High
4 FS/DS flow ratio 0.8 ≤ FS/DS ≤ 2 NA High Low Medium Medium High High
5 Concentration of waste water 100 g/m3 ≤ Cfi ≤ 10,000 g/m3 NA NA Medium Medium Medium High High
7 Back mixing ratio 0 ≤ nmix ≤ 1 NA NA NA NA NA High High
94 S. S., S. S. / Desalination 389 (2016) 78–97
Fig. 12. Optimum NSEC and the percentage energy saving (with target).
Table 6
Optimal process parameters.
FS inlet Best Optimum value of decision variables Values of other important variables at optimum point
concentration case
Feed water Operating FS/DS ratio PRO DS inlet Brine back Product water Flow rate Flow rate at Flow rate at Flow rate at Flow rate
(g/m3)
flow rate pressure of PRO inlets pressure mixing ratio flow ratea at BP pump LP pump-1 LP pump-2 LP pump-3 at FS outlet
of SWRO of SWRO Pdi [bar] nmix Q p [m3/s] Q HPP [m3/s] Q LPP-1 Q LPP-2 Q LPP-3 Q fo [m3/s]
Qf [m3/s] Pf [bar] [m3/s] [m3/s] [m3/s]
Fig. 13. Optimum NSEC and the percentage energy saving (without target).
economic benefit considering both capital and operating expenditure of • Among the six SWRO–PRO configurations presented in this work,
PRO unit in place of SWRO brine post treatment unit. the one (i.e. case VII) which does the direct mixing of diluted DS
outlet with SWRO feed was found to be energy efficient solution
4. Conclusion for desalination.
• For optimal case VII, the calculated NSEC is reduce by 41% and 16%
Mathematical model for SWRO–PRO hybrid system was developed with respect to SWRO desalination system for FS inlet concentration
and implemented in Modelica language using Dymola software tool. of 100 g/m3 and 1000 g/m3 respectively.
The verified individual process unit models are used for SWRO–PRO • The identified best configuration (i.e. case VII) in this work does
flowsheet simulation studies and the following observations were made: not require additional ERD and turbine at optimized process
conditions.
• When sea water was used as feed solution for PRO, the SWRO–PRO
hybrid system is found to be an unsustainable system with respect
to SWRO system.
• This is mainly due to severe ICP in PRO unit. Unless the ICP is nullified 5. Future work
with help of innovative PRO membranes, the self-sustainable SWRO–
PRO configuration may not be viable if sea water is used as a FS solution. Further studies on improved return on investment (ROI) for hybrid
• When water waster was used as FS for PRO, the SWRO–PRO hybrid SWRO–PRO system may help us to verify whether the proposed hybrid
system was found to be energy efficient solution for desalination system can provide improved ROI than the SWRO with directly
application. reclaiming wastewater.
96 S. S., S. S. / Desalination 389 (2016) 78–97
[14] B.A. Qureshi, S.M. Zubair, Exergetic analysis of a brackish water reverse osmosis de- [25] S. Senthilmurugan, A. Ahluwalia, S.K. Gupta, Modeling of a Spiral-wound Module
salination unit with various energy recovery systems, Energy 93 (2015) 256–265, and Estimation of Model Parameters Using Numerical Techniques, 2005.
http://dx.doi.org/10.1016/j.energy.2015.09.003. [26] N. Bhutani, A. Purohit, V.K. Mantravadi, S. Mekapati, S. Subbaiah, Energy efficiency
[15] A. Efraty, Pressure retarded osmosis in closed circuit: a new technology for clean solutions for RO desalination plant, Desalin. Water Treat. 51 (2013) 5049–5055,
power generation without need of energy recovery, Desalin. Water Treat. 51 http://dx.doi.org/10.1080/19443994.2013.795258.
(2013) 7420–7430, http://dx.doi.org/10.1080/19443994.2013.793499. [27] A.R. Bartman, A. Zhu, P.D. Christofides, Y. Cohen, Minimizing energy consumption in
[16] Z.V.P. Murthy, S.K. Gupta, Estimation of mass transfer coefficient using a combined reverse osmosis membrane desalination using optimization-based control, J. Process
nonlinear membrane transport and film theory model, Desalination 109 (1997) Control 20 (2010) 1261–1269, http://dx.doi.org/10.1016/j.jprocont.2010.09.004.
39–49, http://dx.doi.org/10.1016/S0011-9164(97)00051-9. [28] N.Y. Yip, A. Tiraferri, W.A. Phillip, J.D. Schiffman, L.A. Hoover, Y.C. Kim, et al., Thin-
[17] S.S. Hong, W. Ryoo, M.S. Chun, S.O. Lee, G.Y. Chung, Numerical studies on the film composite pressure retarded osmosis membranes for sustainable power gener-
pressure-retarded osmosis (PRO) system with the spiral wound module for ation from salinity gradients, Environ. Sci. Technol. 45 (2011) 4360–4369, http://dx.
power generation, Desalin. Water Treat. 52 (2014) 6333–6341, http://dx.doi.org/ doi.org/10.1021/es104325z.
10.1080/19443994.2013.821041. [29] Y. Xu, X. Peng, C.Y. Tang, Q.S. Fu, S. Nie, Effect of draw solution concentration and op-
[18] A. Achilli, T.Y. Cath, A.E. Childress, Power generation with pressure retarded osmo- erating conditions on forward osmosis and pressure retarded osmosis performance
sis: an experimental and theoretical investigation, J. Membr. Sci. 343 (2009) in a spiral wound module, J. Membr. Sci. 348 (2010) 298–309, http://dx.doi.org/10.
42–52, http://dx.doi.org/10.1016/j.memsci.2009.07.006. 1016/j.memsci.2009.11.013.
[19] K.L. Lee, R.W. Baker, H.K. Lonsdale, Membrane for power generation by pressure re- [30] Q. She, D. Hou, J. Liu, K.H. Tan, C.Y. Tang, Effect of feed spacer induced membrane de-
tarded osmosis, J. Membr. Sci. 8 (1981) 141–171, http://dx.doi.org/10.1016/S0376- formation on the performance of pressure retarded osmosis (PRO): implications for
7388(00)82088–8. PRO process operation, J. Membr. Sci. 445 (2013) 170–182, http://dx.doi.org/10.
[20] S. Sundaramoorthy, G. Srinivasan, D.V.R. Murthy, An analytical model for spiral 1016/j.memsci.2013.05.061.
wound Reverse Osmosis membrane modules: part I — model development and pa- [31] M.F. Naguib, J. Maisonneuve, C.B. Laflamme, P. Pillay, Modeling pressure-retarded
rameter estimation, Desalination 280 (2011) 403–411, http://dx.doi.org/10.1016/j. osmotic power in commercial length membranes, Renew. Energy 76 (2015)
desal.2011.03.047. 619–627, http://dx.doi.org/10.1016/j.renene.2014.11.048.
[21] J.R. Mccutcheon, M. Elimelech, Influence of concentrative and dilutive internal con- [32] A.M. Farooque, A.M. Jamaluddin, A.R. Al-reweli, Comparative Study of Various Ener-
centration polarization on flux behavior in forward osmosis, 284 (2006) 237–247, gy Recovery Devices Used in SWRO Process 1, 2004.
http://dx.doi.org/10.1016/j.memsci.2006.07.049. [33] A. Altaee, G. Zaragoza, A. Sharif, Pressure retarded osmosis for power generation and
[22] C. Klaysom, T.Y. Cath, T. Depuydt, I.F.J. Vankelecom, Forward and pressure retarded seawater desalination: performance analysis, Desalination 344 (2014) 108–115,
osmosis: potential solutions for global challenges in energy and water supply, Chem. http://dx.doi.org/10.1016/j.desal.2014.03.022.
Soc. Rev. 42 (2013) 6959, http://dx.doi.org/10.1039/c3cs60051c. [34] A. Altaee, A. Sharif, G. Zaragoza, A.F. Ismail, Evaluation of FO–RO and PRO–RO de-
[23] F. Helfer, C. Lemckert, Y.G. Anissimov, Osmotic power with pressure retarded osmo- signs for power generation and seawater desalination using impaired water feeds,
sis: theory, performance and trends — a review, J. Membr. Sci. 453 (2014) 337–358, Desalination 368 (2014) 27–35, http://dx.doi.org/10.1016/j.desal.2014.06.022.
http://dx.doi.org/10.1016/j.memsci.2013.10.053. [35] A.P. Straub, A. Deshmukh, M. Elimelech, A.P. Straub, Pressure-retarded osmosis for
[24] M.J. Guirguis, Energy Recovery Devices in Seawater Reverse Osmosis Desalination power generation from salinity gradients : is it viable ? Energy Environ. Sci. 9
Plants with Emphasis on Efficiency and Economical Analysis of Isobaric Versus Cen- (2015) 31–48, http://dx.doi.org/10.1039/C5EE02985F.
trifugal Devices, University of South Florida, 2011.