Desalination and Water Treatment 202 (2020) 27–37

October
www.deswater.com
doi: 10.5004/dwt.2020.26145

Optimal design and techno-economic evaluation of renewable energy powered

combined reverse osmosis desalination and brine treatment unit

Ewaoche John Okampo*, Nnamdi Nwulu

Department of Electrical and Electronic Engineering Science, Faculty of Engineering and Built Environment,
University of Johannesburg, South Africa, emails: [email protected] (E.J. Okampo), [email protected] (N. Nwulu)

Received 1 January 2020; Accepted 21 May 2020

abstracts
The reverse osmosis (RO) desalination technique has been identified as a viable means of fresh-
water production, but its high energy requirement, high cost, and waste (brine) remain serious
challenges. This study, therefore, explores efficient energy from renewable energy sources (RES)
and brine management in the production of freshwater using the integration of RO, electro-dialy-
sis (ED) and crystallization methods. The objective of this study is to minimize the levelized cost
of energy and brine production whilst maximizing freshwater and salt production. The proposed
design is such that the feed water (saline water) is passed through the RO unit for desalination;
the brine produced from the RO unit is further desalinated by the ED method, leaving a very
high concentration to be crystallized into soluble salts thereby achieving a zero brine production.
Furthermore, for energy-efficient management, an optimal sizing of energy sources which includes
grid power, wind power and solar power, was carried out considering mitigation of carbon emis-
sion and its cost and the intermittent limitation of the RES. This integrated design ensures that the
internal and external costs of desalination are evaluated and minimized.
Keywords: Optimization; Renewable energy sources; Reverse osmosis desalination; Brine

1. Introduction world still largely depend on fossil energy sources, espe-

cially convectional grid generators. The direct consequence
Continuous growth in population, constant water
of this is high greenhouse gas emission, which also has cost
pollution and other water stress have made freshwater
implications alongside its environmental impacts. The cost
scarcity a major problem around the world [1]. Treatments
implications of emitted carbon gas and brine management
such as water reuse, seawater desalination have become
are considered as the external costs of desalination [3].
prominent in recent years owing to the vast availability of
Therefore, the main challenges of seawater desalination are,
seawater around the world [2]. The seawater usually con-
hence, energy requirement and brine management. These
tains a large deposit of salt, making it difficult and unhealthy
challenges make seawater desalination quite expensive
for drinking. Desalination is, therefore, the separation of
both in terms of cost and environmental impacts of energy
freshwater from seawater (in other cases, brackish water),
supply source and brine disposal. Hence, efficient energy
leaving behind more concentrated saline water known as
supply and brine management are crucial to carbon emis-
the brine. Indiscriminate brine disposal is harmful to the
sion reduction [4] and cost evaluation of desalination.
environment. Also, most desalination plants around the

* Corresponding author.

1944-3994/1944-3986 © 2020 Desalination Publications. All rights reserved.

28 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37

To tackle these challenges, renewable energy sources system that include a 2-stage high-pressure reverse osmosis
(RES) have been highly exploited and integrated into desali- (HPRO) was made, and the results show that adding HPRO
nation systems to cut down the cost and gas emission of con- is uneconomical.
ventional fossil energy supply [5,6]. Hossam-Eldin et al. [5] For sustainability and environmental friendliness, when
investigate the economic viability of a desalination system considering seawater desalination, the cost and proper
powered by RES with the objective of minimizing excess management of energy supply and brine production must
energy using an optimization program developed to deter- be put into consideration. The contribution of this study is,
mine the best hybrid RESs. Koutroulis and Kolokotsa [6] therefore, (i) to evaluate the optimal cost of the freshwater
proposed an optimization methodology for sizing photovol- output considering carbon emission, demand response and
taic (PV) modules and wind generators to power a desalina- brine treatment cost, (ii) to minimize the impact of carbon
tion system for a small community of residential households emission, (iii) to maximized freshwater production, and (iv)
at minimal total system cost. to minimized brine discharged. Thus, the remainder of this
The thermal process of desalination is more energy paper is organized such that section II details system mod-
and cost-intensive than the membrane process; hence, the els; section III presents a case study; section IV presents and
predominant use of the reverse osmosis (RO) technique. discusses the results, while section V concludes the study.
One major limitation of the RO method is the volume of
brine (concentrate) produced during the desalination pro-
cess [7]. Different brine management approaches have been 2. System models
proposed, and some implemented [1,7,8]. Most available
2.1. System architecture
brine treatment technologies like desalination technolo-
gies are thermal-based or membrane-based. Most often, Fig. 1 describe the system design which has two basic
the same techniques are used in a specific combination for sections; (i) the power section which integrates RESs of wind
both desalination and brine treatment. In this study, RO is and PV with grid power and (ii) the desalination and brine
used for desalination while electro-dialysis (ED) and crys- treatment section which integrates RO, ED and CRY units.
tallizer (CRY) are used for further treatment of the concen- The power sources are optimally scheduled with time of
trate. ED being a membrane technology is cost-effective for use (TOU) demand response and considering carbon emis-
brine treatment, as it utilized ion selectivity to further sep- sion to achieve minimal levelized cost of energy (LCOE)
arate freshwater from the concentrate. This process does whereas, the desalination and brine treatment section is
not entirely convert the brine to potable water and salt. Its designed such that the RO unit desalinates 40% of the feed-
performance is limited by scaling soluble salt on the mem- water (seawater) to freshwater, leaving 60% as brine. This
brane. Therefore, further treatment of the highly concen- brine is passed into the ED which further desalinates 20%
trated brine is required. The crystallization of the remaining of the feedwater into freshwater (i.e., 33.3333% of the brine
volume of brine can lead to zero brine discharge as the result- passed into ED) leaving a more concentrated brine into the
ing product will be crystals of salt and evaporated freshwa- CRY unit. The crystallization unit evaporates 10% of the
ter, which can be condensed and collected for drinking. The feed water (16.6667% of the original brine form RO) which
feasibility of RO-ED-CRY was presented by Panagopoulos is condensed and collected as additional freshwater. The
et al. [9], and a framework for cost and energy needs model remaining 30% of the feedwater (which is 50% of the original
was established. A comparison of RO-ED-CRY and the same brine from RO) forms crystals of salt leaving zero discharge.

Fig. 1. Schematic diagram of combined desalination and brine treatment process.

 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37 29

2.2. Grid power and cost model AMCPV = APV × MCPV × r (9)
The hourly power supply from the grid serves as a
backup to satisfy the load demand of the desalination unit 2.3.2. Wind power and cost model
such that in the case of intermittency of RES, the desalina-
tion unit will depend on the grid for power supply. When The hourly power output of the wind generator (Wp(t)) is
there is excess energy as a result of high-power output from given as [10].
the RES or as a result of demand response, the excess energy
can be sold back to the grid. Therefore, the hourly power 1 3
Wp ( t ) = × η′w × ρair × C p × AWT × V ( t ) (10)
(GPi(t)) imported from the grid and the hourly power (GPe(t)) 2
exported to the grid are optimized variables ranging from
zero to maximum hourly load demand by the RO unit as where ήw is the efficiency of the wind generator, ρair is the air
expressed in Eqs. (1) and (2) respectively. density, Cp is the power coefficient, AWT is the swept area
of the wind turbine (WT) and V(t) is the hourly wind speed
given as [12]:
0 ≤ GPi ( t ) ≤ GPi max (1)
∝
h
V ( t ) = VR ×   (11)
0 ≤ GPe ( t ) ≤ GPe max (2)  hR 

The maximum transferable power to and from the grid is where V(t) is the hourly speed at projected height (h), VR is
assumed to equal: the hourly speed at reference height (hR) and α is the power-
law exponent equivalent to 1/7.
The economics of using wind power for desalination is
GPi max = GPemax (3)
similar to that of PV and in this study it is analyzed based
on the initial capital cost (ICWT) and total maintenance cost
The cost of grid power is dependent on the energy (TMCWT), depending on the area of the WT as follows:
price and the difference between imported power and
exported power to the grid. It is assumed that the unit cost
IC WT = AWT × C WT (12)
of purchase power is equal to the unit cost of selling power
back to the figures. Hence, excess energy production that is
not needed by the desalination unit is sold back to the grid
to compensate for the cost of imported power or at least AMC WT = AWT × MC WT × r (13)
reduce the importation cost of power from the grid.

CGPi ( t ) = GPi ( t ) × P ( t ) (4) 2.4. RO desalination plant power demand and cost model
The hourly power demand (PWD(t)) of the RO desalina-
CGPe ( t ) = GPe ( t ) × P ( t ) (5) tion unit depends on the specific energy consumption (SEC)
to produce 1 m3 of freshwater, which in this study is 3 kWh/
m3 and the actual volume of water (QW(t)) produced per
(
CGP = ∑ CGPi ( t ) × P ( t ) − GPe ( t ) × P ( t ) ) (6)
hour [13,14].
∀t = 1, 2 , 3 , 8760
PWD ( t ) = QWRO ( t ) × SEC (14)
where P(t) is the hourly unit price of transferable grid power
and CGP is the total annual cost of transferable grid power. The daily water production capacity is given as:

24
2.3. RES and cost model
DQW = ∑QWRO ( t ) (15)
2.3.1. PV power and cost model t =1

The hourly output power (PPV(t)) supply by the PV array The water dispensation model and network are not con-
is given as [10,11]: sidered in this study. Thus, water tank capacity (WTK) express
in m3 is assumed to be twice-daily water production capac-
PPV ( t ) = APV × η′ × SI ( t ) (7) ity to make enough space available, even if there is water
remaining from the previous day.
where APV is the area of the photovoltaic array in (m2),
SI(t) is hourly solar irradiation and ή is the PV efficiency. WTK = DQW × 2 (16)
Evaluation of PV cost is based on the area of PV array,
initial capital cost and annual maintenance cost expressed as: RO desalination cost model includes the initial capi-
tal cost (ICRO), annual maintenance and operational cost
ICPV = APV × CPV (8) (AMCRO), annual membrane replacement cost (ACMR),
30 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37

annual treatment chemical cost (ACCH) and water tank cost hour. The demand response load is an optimize variable
(CWTK) [15,16]. expressed as:

ICRO = CRO × DQW (17) − ∆Lmax ( t ) ≤ ∆L ( t ) ≤ ∆Lmax ( t ) (25)

where
ICWTK = CWTK × WTK (18)
∆Lmax ( t ) = 0.3 ×  PWD ( t ) + PB ( t )  (26)
2.5. Brine treatment power demand model
Also, for even distribution of load shift, cut of demand
The brine treatment section includes the ED and at certain hours of the day equals an increase of demand at
crystallization units. The volumetric quantity of brine pro- other hours of the day expressed as:
duced (QB) from RO plant depends its water production
24
capacity and water recovery ratio (RR) expressed as [2]:
∑∆L ( t ) = 0
t =1
(27)

DQW
DQB = × (1 − RR ) (19) The application of TOU demand response optimally
RR
shifts loads from peak hours when the price of demand
where RR is the percentage of volume freshwater produced is highest to either standard or off-peak hours yet, allow
by the RO desalination to the volume of saline feed water the demand at every hour to be met. The cost of demand
(QF) [8], it is assumed to be 40% in this study. response is the difference between the cost of power demand
before and after load variation due to demand responses.
It is expressed as [18].
QW
RR = × 100 (20)
QF
( )
 PWD ( t ) + PB ( t ) × λ ( t ) − 
∆LC = ∑   (28)
This implies that QB can also be calculated as [8]:
 WD(( B ) )
 P ( t ) + P ( t ) − ∆L ( t ) × λ ( t ) 

QB = QF − QW (21) t = 1, 2 , 3 ,…8760

Also, Eq. (19) is modified to consider hourly brine pro- 2.7. Carbon emission cost and global warming impact of
duction as: energy source
The global warming impact (GWI, in other words, car-
QWRO ( t )
QB ( t ) = × (1 − RR ) (22) bon emission of fossil-fuel energy source, could be due to
RR construction or operation of the plant, with the impact due
to operation exceeding that due to construction by multiple
Therefore, the hourly power demand (PB(t)) of the orders [4]. Thus, this study considered the impact due to
brine treatment unit depends on the sum of SEC by ED and operation as adopted by Baumgärtner et al. [4]. The authors
the crystallization unit, which in this study are adapted define GWI as the summation of input power Pj,t of every
from [9,17] as 3.7 and 4.5 kWh/m3, respectively. Hence; fossil-fuel units j in every time step t ∈ T multiplied by spe-
cific operational emission factor SEFj,t and the period Δtt
PB ( t ) = QB ( t ) × SEC (23) of time step represented as:

For a zero brine discharge, the brine from the RO unit,  

GWI g = ∑  ∆tt ∑SEFj ,t × Pj ,t  (29)
when passed through the ED unit, produces some quanti- t∈T 
 j 
ties of freshwater (33.3333% of the brine, in this study) and
a more concentrated brine, which is further passed into the where
crystallizer. The also produced some quantities of freshwater
(in this case, 16.6667% of the total brine from the RO unit), emission due to electrical energ genarated
and the remaining amounts of brine are crystallized salt and SEFj ,t = (30)
electrical energ genarated
other compounds. Thus, the total freshwater produced from
the combine RO-ED-CRY system is given as: In other to estimate carbon emission cost, the formulated
model given by Molinos-Senante and González [3] Kesieme
TQW ( t ) = QWRO ( t ) + QWED ( t ) + QWCRY ( t ) (24) et al. [19] was adopted as:

 $   kwh   kgCO 2 −ee 

CE  3  = Energy supply  3  × Emission factor  ×
2.6. TOU demand response load and cost model m   m   kwh 
 $ 
The maximum allowable demand variation (increase/ Carbon tax   (31)
decrease) per hour is assumed 30% of total demand at that  kgCO 2 − e 
 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37 31

The grid specific emission factor and carbon tax Objective function:
depend on the country of location of the plant. This study
adopted the calculated emission factor for South Africa min  w1 × ACS + w2 × ∑ GWI + w3 × ∑ QB ( t ) − w4 × ∑ TQW ( t ) 
by Brander et al. [20]. The value of the emission factor is
1.069 kg CO2/kWh. The new South Africa carbon tax rate is (38)
ranged from R6-R48 (0.40 $–3.17 $) [21]. The lowest value
of 0.41 $ is considered in this study. S.t.

2.8. Optimization problem formulation PPV ( t ) + Wp ( t ) + GPi ( t ) = PWD ( t ) + PB ( t ) + ∆L ( t ) + GPe ( t ) (39)

In this study, the annualized system cost matrix is used

for the economic evaluation of the RO desalination sys- TQW ( t ) ≥ WD ( t ) (40)
tem powered by a grid and RES to determine the cost of
freshwater and the LCOE. The annualized cost of the system
QW min ≤ QWRO ( t ) ≤ QW max (41)
(ACS) involves the capital recovery factor (CRF) and the
total system cost which includes, total initial cost (TICC),
total maintenance and operation cost (TMC) of all the com- min
PWD ≤ PWD ( t ) ≤ PWD
max
(42)
ponents that makes up the system [22–24]. Also included in
the ACS is demand response cost (ΔLC) and total carbon ∀t = 1, 2 , 3…8760
emission cost (EC).

APV ≥ 0 (43)
TICC = ICPV + IC WT + ICRO + ICED + ICCRY + ICWTK (32)

AWT ≥ 0 (44)
TMC = AMCPV + AMC WT + AMCRO + AMCED + AMCCRY (33)
Eq. (39) expresses the power balance that ensures power
supply from PV, wind generator and the grid at any time t
ACS = TICC × CRF + TMC + ACMR × CRF + ACCH +
equals the total demand by the RO desalination and brine
CGP + ∆LC + EC (34) treatment unit and the power exported to the grid at the
same time t. Eq. (40) is a water balance that ensures the total
where water produces at any hour t, equals, or in excess of water
demand at that hour t, while Eq. (41) ensures water produce
i (1 + i )
n
per hour remains in a boundary of the required limit. Eq. (42)
CRF = n = 1…19 (35) is the limit of power required by RO to produce water at
(1 + i )
n
−1 any time t, and Eq. (43) expresses the limit of the area of PV
while Eq. (44) limits the area of the WT. Other Eqs. (1)–(3)
n is the number of years in the lifetime of the system of and (25), which express the limit of grid imported power,
which interest rate i is considered. the limit of grid exported power, maximum allowable trans-
fer power and limit of demand response load respectively.
ACS Also, Eq. (27) ensures even distribution of load shift.
COP = + Saltp (36)
∑ TQW ( t )
3. Case study
In this study, the metrological data from Stellenbosch
ACS + PWD ( t ) + PB ( t )
LCOE = (37) University, Western Cape Province of South Africa is used.
∑ P ( t ) + W ( t ) + GP ( t )
PV p i The hourly solar irradiation and wind speed, as represented
P ( t ) + W ( t ) + GP ( t ) in Figs. 2 and 3, were collected from the Southern African
PV p i
Universities Radiometric (SAURAN). Also, the TOU energy
price (in US$) for South Africa gotten from the Eskom sched-
The optimization problem is formulated as a multi- ule of the standard price for Eskom tariffs 2019/2020 [25] is
objective linear programming problem as expressed in also implemented for the demand response program. Fig. 4
Eq. (34) with the objective of minimizing the ACS and carbon represents the TOU price per hour of a day.
emission while maximizing the quantity of freshwater pro- The choice of Western Cape Province of South Africa is
duction subject to constraints expressed by Eqs. (35)–(39). based on its proximity to the coast of the Atlantic Ocean and
The weighting factors (W1, W2, W3 and W4) were allocated the existing water scarcity problem in the area. The study
based on the preference of significant concern, with GWI of has shown that the RO desalination technology is the most
emission rank highest, then ACS and quantity of water pro- suited for South Africa for several reasons, as reflected by
duce rank lowest since meeting water demand is paramount Swartz et al. [26]. Nonetheless, the west-coast water is usually
to having excess water production. This multi-optimization frigid and high salinity in nature (on average 3.5 g/L), thus
problem is solved using COMPLEX solver of the advanced requires high operating pressure and low membrane fluxes
interactive multidimensional modeling system (AIMMS). and in turn, high energy requirement [26]. These challenges
32 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37

Fig. 4. Hourly price of grid power.

Fig. 2. Hourly solar irradiance.
(m/s)

Fig. 3. Hourly wind speed.

Fig. 5. Hourly water demand.

make freshwater production using desalination quite expen- 4. Results, discussion and sensitive analysis
sive, albeit still the preferred option for freshwater sus-
4.1. Results
tainability. There are several already existing desalination
plants operating in South Africa, with the largest been sit- Fig. 6 shows the hourly power supply and sold back to
uated in Mossel Bay with a capacity of 15,000,000 L/d [27]. the grid, the power from the wind generator and the power
Despite the several existing temporally desalination plants output of the PV generator. An average day simulation is
in operation, recent studies shown that the region is still in used for purposes of simplicity. Fig. 7 depicts the hourly
dare need of about 150,000,000 L/d of freshwater to meet volume of freshwater produced from the three units, RO,
demand and a continuous increase of over 30,000,000 L ED and CRY. Fig. 8 represents the hourly volume of brine,
per annual to achieve about 350,000,000 L/d to guarantee freshwater and salt produced, while Table 1 is the summary
sustainability in the nearest future [28]. While there is a results of the other optimized parameter.
good propose solution of building a more permanent sea-
water desalination plant of capacity above 150,000,000 L/d, 4.2. Discussion and sensitivity analysis
this study proposes the optimization of existing modu-
lar desalination plants as well, by incorporating RES and The result represented in Fig. 6 shows the impact of
demand response program to increase freshwater produc- TOU demand response in shifting load from peak hours
tion capacity at reduce LCOE and hence minimize the cost to off-peak and standard hours for energy cost-effective
of freshwater. An assumed hourly water demand curve management. The selection of the energy sources and the
based on expected behavioral water usage at different hours subsequent power output depending on the cost and the car-
of the day, as represented in Fig. 5, is used in this study. bon emission of the energy source, hence the resulting low
Other data used especially for components cost evaluation power output from the grid, which depends mainly on fossil
are taken from Abdelshafy et al. [15] and Wu et al. [16]. fuel generators. The carbon emitted from the grid generator
 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37 33

Fig. 6. Hourly energy dispensed.

Fig. 7. Hourly freshwater dispensed.

Table 1
Input parameters [15,16]

Parameters Value Parameters Value Parameters Value

3
Project time 20 y CPV 285 $/m WT rated power 1 kW
Interest rate 5% ήw 85% CRO 532 $/m3/d
Discount rate 3% ρair 1.23 kg m3 Plant maintenance cost 0.2 $/m3
Emission factor 1.07 Cp 0.59 Chemical cost 0.06 $/m3
Emission tax 0.41 $ MCPV 35 $/y CMR 0.06 $/m3
PV rated power 1 kW CWT 1,804 $ CWTK 255 $/m3
PV efficiency (ή) 16% MCPV 100 $/y

comes with a cost (261,460 $/y), therefore increasing the cost into it has a high concentration of salt. The crystallization
of grid power supply. This gives the advantage to the RES process is, therefore, the main brine treatment unit produc-
(wind and PV generators), hence the high power output ing soluble salts and a small volume of freshwater, leaving a
from the two sources. zero brine discharge.
The hourly volume of freshwater produce from the The total potable water produced from RO-ED-CRY at
three units (RO, ED and CRY) as depicted in Fig. 7 shows every hour of the day alongside the hourly brine produc-
that the RO unit, which is the main desalination unit, pro- tion from the RO unit, as well as the salt produced from
duces the highest quantity of freshwater and then the ED the brine, is shown in Fig. 8. This result indicates that a
unit. The crystallization unit produces only a small vol- large volume of the feed water is converted into freshwa-
ume of freshwater since the larger volume of brine passed ter while the remainder crystallized into salts. Furthermore,
34 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37

Fig. 8. Hourly freshwater, brine and salt produced.

the volume of freshwater produced (3,005 m3/d) from this Table 2

combined model of desalination and brine treatment, as Summary results of optimized parameters
shown in Table 2, is higher than the volume produced by
standalone RO unit previously presented by Okampo and ACS ($) 1,401,367
Nwulu [29] with similar input parameters which are also Emission (kgCO2–e) 637,708
the expected desalination capacity to meet the baseline daily Emission cost ($) 261,460
water demand (1,250 m3/d). Also, this study presents the Daily water produced (m3) 3,005
unit cost of freshwater and salt as a single unit cost of pro- Daily salt produced (m3) 1,288
duction, since it is difficult to separate the cost of production
LCOE ($/kW) 0.89
of freshwater and salt because the system is design as a sin-
gle unit with the same components cost, energy sources and Unit cost of production of freshwater and salt ($/m3) 1.056
their costs. The unit cost of production of freshwater and APV (m2) 9,431
salt (0.89 $) is within the range of cost of water for stand- AWT (m2) 50,891
alone desalination units presented in literature which range
from 0.5–2.39 $ [15,16,29]. Furthermore, most standalone
RO desalination systems do not account for carbon emis-
5. Conclusion
sion cost and brine treatment, therefore, use less energy, less
treatment chemicals and required fewer membrane replace- This study presents efficient energy and brine man-
ment than a combined model that produces freshwater and agement in the production of freshwater using the integra-
treats brine with zero discharge. Hence, the unit cost of pro- tion of RO, ED and crystallization methods. The objective
duction of this combined model is relatively low consider- of this study is to minimize LCOE and brine production
ing the economic and environmental factors. On the other whilst maximizing freshwater and salt production at a min-
hand, the LCOE of this model (1.06 $) is within the average imal cost. The proposed design is such that the feed water
value of those of standalone RO desalination systems which (saline water) is passed through the RO unit for desalina-
usually range between 0.5–1.2 $, suggesting a similar cost tion; the brine produced from the RO unit is further desali-
of energy for a standalone desalination unit and a combine nated by the ED method, leaving a very high concentration
desalination-brine treatment unit. This is because the com- to be crystallized into soluble salts thereby achieving a zero
ponents cost of similar energy sources used for a standalone brine production. Furthermore, for energy-efficient man-
unit is the same for a combined unit and depends largely agement, optimal sizing of energy sources, which include
on the size of the system and the volume of production. grid power, wind power and solar power, was carried out
Sensitivity analysis was performed to evaluate the considering mitigation of carbon emission and its cost and
impacts of the percentage increase of water demand on three the intermittent limitation of the RES. This integrated design
cost matrices (ACS, LCOE and cost of products, COP) as ensures that the internal and external costs of desalination
depicted in Figs. 9–11, respectively. are evaluated and minimized. The results show the impact
The results of the sensitivity analysis show a propor- of emission and its cost on the energy cost, increasing the
tional increase in ACS, a moderate increase in LCOE and cost of grid energy supply, making RES more cost-effective
a decrease in the cost of products against the percentage as well as environmentally friendly. Also, the LCOE is within
increase in water demand. This implies that an increase in the average value of those of standalone desalination units
water demand results in a rise in ACS and LCOE but the suggesting a similar cost of energy for a standalone desali-
unit cost of production of water and salt is reduced with nation unit and a combined desalination-brine treatment
more freshwater production to meet the increased water unit. This is because the components cost of similar energy
demand. sources used for a standalone unit is the same for a combined
 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37 35

Fig. 9. Impacts of percentage increased on the annual cost of the system.

Fig. 10. Impacts of percentage increased on levelized cost of energy.

Fig. 11. Impacts of percentage increased on the cost of water and salt.

unit and depends largely on the size of the system and the of use. Also, the annual carbon emission (637,708 kgCO2–e) is
volume of production. Further study can investigate the cost still very high, a further reduction can be achieved with the
of salt-based on location as it is expected that the salt pro- integration of the storage system to supplement the incon-
duced if further treated can be added advantage of the com- sistency of the RESs. This will reduce the dependence on
bined model of desalination and brine treatment as it can be grid power but might increase the cost of production.
36 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37

Set CEDG — DG carbon emission cost, $

CEg — Grid carbon emission cost, $
t — Time, h
COP — Cost of products (water and salt), $/m3
GPi(t) — Grid power imported, kW
Parameters GPe(t) — Grid power exported, kW
I — Interest rate GWI — Global warming impact, kgCO2–e
r — Discount rate LCOE — Levelized cost of energy, $
ή — PV efficiency ΔL(t) — Demand response load curtail, kW
ήw — Wind turbine efficiency PB(t) — Hourly power demand for brine treatment,
ρair — Air density kW
Cp — Power coefficient PPV(t) — Hourly power output of PV, kW
α — Power law exponent PWD(t) — Hourly power demand by RO Plant, kW
λ(t) — Marginal price of energy, $/kWh QWRO(t) — RO hourly volume of water produce, m3
ACCH — Annual cost of RO treatment chemicals, $ Saltp — Volume of salt produce, m3
ACMR Annual cost of RO membrane replacement, $ SEFDG — DG specific emission factor, kg/L
—
AMCPV — Annual maintenance cost of PV, $ SEFj,t — Grid specific emission factor, kgCO2–e/kWh
AMCRO — RO annual maintenance cost, $ TEC — Total carbon emission cost, $
AMC WT — Annual maintenance cost of the wind Wp(t) — Wind generator power output, kW
generator, $
CRF — Capital recovery factor Abbreviations
CGPi(t) — Cost of grid power imported, $/kW
CGPe(t) — Cost of grid power exported, $/kW DR — Demand response
CGP — Total cost of grid transferable power, $ PV — Photovoltaic
CWT Wind turbine cost, $ RES — Renewable energy sources
—
CPV — PV panel cost, $/m2
CRO — RO cost, $/m3 References
CWTK — Water tank cost, $/m3
[1] A. Giwa, V. Dufour, F. Al Marzooqi, M. Al Kaabi, S.W. Hasan,
GPimax — Maximum imported grid power, kW Brine management methods: recent innovations and current
GPemax — Maximum exported grid power, kW status, Desalination, 407 (2017) 1–23.
h — Projected height of wind turbine, m [2] E. Jones, M. Qadir, M.T.H. van Vliet, V. Smakhtin, S.-m. Kang,
hR — Reference height of wind turbine, m The state of desalination and brine production: a global outlook,
Sci. Total Environ., 657 (2019) 1343–1356.
ICPV — Initial capital cost of PV, $
[3] M. Molinos-Senante, D. González, Evaluation of the economics
ICRO — Initial capital cost of RO Plant, $/m3/d of desalination by integrating greenhouse gas emission
ICWTK — Initial capital cost of the water tank, $ costs: an empirical application for Chile, Renewable Energy,
ICWT — Initial capital cost of the wind generator, $ 133 (2019) 1327–1337.
ΔLmax — Maximum allowable change in demand, kW [4] N. Baumgärtner, R. Delorme, M. Hennen, A. Bardow, Design
of low-carbon utility systems: exploiting time-dependent grid
ΔLC — Cost of demand response load curtails, $
emissions for climate-friendly demand-side management,
MCWT —
Cost of maintenance of a wind turbine, $ Appl. Energy, 247 (2019) 755–765.
MCPV — Cost of maintenance of a PV panel, $ [5] A. Hossam-Eldin, A.M. El-Nashar, A. Ismaiel, Investigation into
P(t) — Price of grid power, $/kWh economical desalination using optimized hybrid renewable
Pωmin — Minimum power demand, kW energy system, Int. J. Electr. Power Energy Syst., 43 (2012)
1393–1400.
Pmax — Maximum power demand, kW
WD [6] E. Koutroulis, D. Kolokotsa, Design optimization of desalination
QB — Quantity of brine produced, m3 systems power-supplied by PV and W/G energy sources,
QF — Quantity of feed water, m3 Desalination, 258 (2010) 171–181.
Qw — Quantity of freshwater produced, m3 [7] A. Subramani, J.G. Jacangelo, Treatment technologies for
QWmin — Minimum RO water produce, m3 reverse osmosis concentrate volume minimization: a review,
Sep. Purif. Technol., 122 (2014) 472–489.
QWmax — Maximum RO water produce, m3
[8] A. Panagopoulos, K.-J. Haralambous, M. Loizidou, Desalination
RR — Freshwater recovery ratio brine disposal methods and treatment technologies – a review,
SEC — Specific energy consumption, kWh/m3 Sci. Total Environ., 693 (2019) 133545.
SI(t) — Solar irradiation, W/m2 [9] K.G. Nayar, J. Fernandes, R.K. McGovern, B.S. Al-Anzi,
TICC — Total initial capital cost of system, $ J.H. Lienhard V, Cost and energy needs of RO-ED-crystallizer
systems for zero brine discharge seawater desalination,
TMC — Total maintenance and operational cost, $
Desalination, 457 (2019) 115–132.
V(t) — Wind speed, m/s [10] N.I. Nwulu, X.H. Xia, Optimal dispatch for a microgrid
VR — Reference wind speed, m/s incorporating renewables and demand response, Renewable
WD(t) — Hourly water demand, m3 Energy, 101 (2017) 16–28.
[11] U. Damisa, N.I. Nwulu, Y.X. Sun, Microgrid energy and reserve
management incorporating prosumer behind-the-meter
Variables resources, IET Renewable Power Gener., 12 (2018) 910–919.
[12] B.S. Borowy, Z.M. Salameh, Methodology for optimally sizing
ACS — Annualized cost of system, $ the combination of a battery bank and PV array in a wind/
APV — Area of photovoltaic, m PV hybrid system, IEEE Trans. Energy Convers., 11 (1996)
AWT — Area of wind turbine, m 367–375.
 E.J. Okampo, N. Nwulu / Desalination and Water Treatment 202 (2020) 27–37 37

[13] A. Maleki, M.G. Khajeh, M.A. Rosen, Weather forecasting for [23] M.A. Al-Obaidi, G. Filippini, F. Manenti, I.M. Mujtaba, Cost
optimization of a hybrid solar-wind–powered reverse osmosis evaluation and optimisation of hybrid multi effect distillation
water desalination system using a novel optimizer approach, and reverse osmosis system for seawater desalination,
Energy, 114 (2016) 1120–1134. Desalination, 456 (2019) 136–149.
[14] W.X. Peng, A. Maleki, M.A. Rosen, P. Azarikhah, Optimization [24] C. Koroneos, A. Dompros, G. Roumbas, Renewable energy
of a hybrid system for solar-wind-based water desalination driven desalination systems modelling, J. Cleaner Prod.,
by reverse osmosis: comparison of approaches, Desalination, 15 (2007) 449–464.
442 (2018) 16–31. [25] Eskom, Schedule of Standard Prices for Eskom Tariffs 1 April
[15] A.M. Abdelshafy, H. Hassan, J. Jurasz, Optimal design of a 2019 to 31 March 2020 for Non-Local Authority Supplies, and
grid-connected desalination plant powered by renewable 1 July 2019 to 30 June 2020 for Local Authority Supplies, 2019,
energy resources using a hybrid PSO–GWO approach, Energy pp. 1–48. Available at: http://www.eskom.co.za/CustomerCare/
Convers. Manage., 173 (2018) 331–347. TariffsAndCharges/Documents/Eskom%20schedule%20of%20
[16] B.J. Wu, A. Maleki, F. Pourfayaz, M.A. Rosen, Optimal design standard%20prices%202019_20%20Rev00%20%28full%20
of stand-alone reverse osmosis desalination driven by a version%20excl%20Transflex%29.pdf
photovoltaic and diesel generator hybrid system, Sol. Energy, [26] C.D. Swartz, J.A. Du Plessis, A.J. Burger, G. Offringa,
163 (2018) 91–103. A desalination guide for South African municipal engineers,
[17] K.G. Nayar, J. Fernandes, R.K. McGovern, K.P. Dominguez, Water SA, 3 (2006) 641–647.
A. McCance, B.S. Al-Anzi, J.H. Lienhard V, Cost and energy [27] T. Goga, E. Friedrich, C.A. Buckley, A LCA (Life Cycle
requirements of hybrid RO and ED brine concentration systems Assessment) Comparison of Wastewater Reclamation and
for salt production, Desalination, 456 (2019) 97–120. Desalination for the Ethekwini Municipality – A Theoretical
[18] E. Dehnavi, H. Abdi, Optimal pricing in time of use demand Study, 2015. Available at: http://prg.ukzn.ac.za/docs/default-
response by integrating with dynamic economic dispatch source/projects/wisa_2016-paper_taahira.pdf?sfvrsn=2
problem, Energy, 109 (2016) 1086–1094. [28] Water and Sanitation Department City of Cape Town, Water
[19] U.K. Kesieme, N. Milne, H. Aral, C.Y. Cheng, M. Duke, Outlook 2018 Report, 2018, pp. 1–5. Available at: https://resource.
Economic analysis of desalination technologies in the context capetown.gov.za/documentcentre/Documents/City%20
of carbon pricing, and opportunities for membrane distillation, research%20reports%20and%20review/Water%20Outlook%20
Desalination, 323 (2013) 66–74. 2018%20-%20Summary.pdf
[20] M. Brander, A. Sood, C. Wylie, A. Haughton, J. Lovell, Electricity- [29] E.J. Okampo, N.I. Nwulu, Optimal energy mix for a reverse
Specific Emission Factors for Grid Electricity, Ecometrica, 2011, osmosis desalination unit considering demand response, J. Eng.
pp. 1–22. Available at: https://ecometrica.com/assets/Electricity- Des. Technol., (2020) 1726–0531, doi: 10.1108/JEDT-01-2020-0025.
specific-emission-factors-for-grid-electricity.pdf
[21] National Treasury, Republic of South Africa: Media Statement
Publication of the 2019 Carbon Tax Act, 2019, pp. 10–13
Available at: http://www.treasury.gov.za.
[22] L.G. Acuña, M. Lake, R.V. Padilla, Y.Y. Lim, E.G. Ponzón,
Y.C. Soo Too, Modelling autonomous hybrid photovoltaic-
wind energy systems under a new reliability approach, Energy
Convers. Manage., 172 (2018) 357–369.