Solar Energy: Sciencedirect

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Solar Energy 181 (2019) 475–489

Contents lists available at ScienceDirect

Solar Energy
journal homepage: www.elsevier.com/locate/solener

Identifying optimal operating conditions of solar-driven silica gel based T


adsorption desalination cooling system via modern optimization
Hegazy Rezka,b, , Ahmed S. Alsamanc, Mujahed Al-Dhaifallahd, Ahmed A. Askalanyc,e,

Mohammad Ali Abdelkareemf,g,h, , Ahmed M. Nassefa,i


⁎⁎

a
College of Engineering at Wadi Addawaser, Prince Sattam Bin Abdulaziz University, Saudi Arabia
b
Electrical Engineering Dept., Faculty of Engineering, Minia University, Egypt
c
Department of Mechanical Engineering, Faculty of Industrial Education, Sohag University, Sohag 82524, Egypt
d
Systems Engineering Department, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia
e
The University of Edinburgh, School of Engineering, Institute for Materials and Processes, Sanderson Building, The King's Buildings, Mayfield Road, EH9 3BF Edinburgh,
Scotland, UK
f
Dept. of Sustainable and Renewable Energy Engineering, University of Sharjah, P.O. Box 27272, Sharjah, United Arab Emirates
g
Center for Advanced Materials Research, University of Sharjah, PO Box 27272, Sharjah, United Arab Emirates
h
Chemical Engineering Department, Faculty of Engineering, Minia University, Egypt
i
Department of Computers and Automatic Control Engineering, Faculty of Engineering, Tanta University, Egypt

ARTICLE INFO ABSTRACT

Keywords: The target of this study is to maximize the performance of solar-driven adsorption desalination cooling (SADC)
Adsorption system by defining the optimal operating conditions using a modern optimization algorithm. A mathematical
Optimization model for the SADC system employing silica gel has been proposed. Then, a robust, simple, and quick optimi-
Water desalination zation algorithm named radial movement optimizer is applied for determining the best operating parameters of
Cooling
the SADC system. The SADC’s decision variables used in the optimization process are cycle time, hot water inlet
Solar energy
temperature, cooling water inlet temperature, and flow rate. The performance of the SADC system is evaluated
Renewable energy
concerning the specific daily water production (SDWP), the coefficient of performance (COP) and specific
cooling power (SCP). The optimization process results are compared with their corresponding experimental
results. Several sets of the parameters’ constraints that represent different conditions are considered during the
optimization process. A 70% increase in SDWP and SCP is achieved by using the optimal operating conditions
with no change in the system design or the used materials. An amount of 6.9 m3/day/ton desalinated water,
191 W/kg cooling capacity and 0.961 COP are demonstrated as the possible outputs of the proposed SADC
system. This research shows the validity of this optimization technique in exploring all possibilities and showing
the best-operating conditions of the SADC system.

1. Introduction amount of energy that costs 0.26–0.54 USD/m3 in case of saline water,
and 0.2–0.4 USD/m3 in case of brackish water (Ghaffour et al., 2013;
Securing fresh water and energy resources become the most top Ng et al., 2017; Shahzad et al., 2017; Shahzad et al., 2018a; Shahzad
priorities of human beings life (Bait and Si–Ameur, 2018; Singh et al., et al., 2019). Fossil fuel is the conventional energy source for the de-
2018). Although two-thirds of the earth are full of water, but un- salination units that is not only expensive, but also it has a harmful
fortunately, around 97% of it is saline. For the survival of mammals and environmental impact. Renewable energy driven desalination units is
plants, fresh water should be available to satisfy their daily needs. In considered a solution for the water scarcity problem, energy shortage,
this regard, water desalination is considered as one of the proper ways and environmental concerns (Abdelkareem et al., 2018a,b). Among the
of securing fresh water. Although membrane-based desalination tech- different available renewable energy sources, solar energy is considered
niques such as reverse osmosis (RO) are lower energy consumption as the best as it is available especially in the water scarcity regions such
compared to thermal-based ones, they are still consume a considerable as the Gulf region, and it has no environmental impacts (Arabpour


Corresponding author at: College of Engineering at Wadi Addawaser, Prince Sattam Bin Abdulaziz University, Saudi Arabia.
⁎⁎
Corresponding author at: Dept. of Sustainable and Renewable Energy Engineering, University of Sharjah, P.O. Box 27272, Sharjah, United Arab Emirates.
E-mail addresses: [email protected] (H. Rezk), [email protected], [email protected] (M.A. Abdelkareem).

https://doi.org/10.1016/j.solener.2019.02.024
Received 10 November 2018; Received in revised form 11 January 2019; Accepted 11 February 2019
0038-092X/ © 2019 International Solar Energy Society. Published by Elsevier Ltd. All rights reserved.
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Nomenclature Subscripts/superscript

Symbols ads adsorption


al aluminum
A area [m2] b brine
Cp specific heat [J/kg⋅K] bed adsorption bed
Ds surface diffusion coefficient [m2/s] cond condenser
Dso pre-exponential coefficient [m2/s] ch chilled water
E characteristic energy [J/kg] cu copper
Ea activation energy [J/mol] des desorption
hfg water latent heat [J/kg] evap evaporator
Hst isosteric heat of adsorption [J/kg] hex heat exchanger
M mass [kg] hw heating water
m· mass flow rate [kg/s] i inlet
P pressure [kPa] o outlet
R universal gas constant [J/mol⋅K] s salt
Rp average radius of the particle [m] sat saturation
T temperature [°C] sg silica gel
t time [sec] sw sea water
w adsorption capacity [Kg/kg] v vapor
w0 maximum adsorption capacity [Kg/kg] w water
X concentration [ppm]

Roghabadi et al., 2018; Hansen and Vad Mathiesen, 2018; Kasaeian demonstrated that applying an internal heat recovery between the
et al., 2018), and thus it is used extensively in water desalination units evaporator and condenser in the AD cycle does not only increase the
(Siddiqui and Dincer, 2018; Takalkar and Bhosale, 2018; Wu et al., daily production rate of potable water, but also allows the cycle to
2017). operate at a low temperature of 50 °C which could be obtained from
Adsorption desalination (AD) is one of the desalination methods solar energy (Thu et al., 2017). In (Youssef et al., 2016), a theoretical
that is simple, easy in operation compared to the other methods, and study exhibited that the water productivity and/or the cooling load that
requires low energy intensity sources such as solar energy (Alsaman could be obtained in two absorber bed cycles, linked with an integrated
et al., 2016). AD is significantly developed during the last decades and evaporator/condenser, significantly increased where a daily water
is considered as a promising alternative to the traditional intensive production rate of 15.2 m3 per ton of adsorbent (AQSOA-Z02) and
energy consumption desalination methods (Alsaman et al., 2016; cooling load up to 46.6 Rton per ton of adsorbent at 10 °C evaporator
Askalany, 2016). A typical AD unit consists of three main equipment, temperature, were realized. Few reports discussed the effect of weather
i.e., evaporator, condenser, and one or more adsorption beds (Ng et al., conditions on the performance of solar driven AD units. A theoretical
2013). Simultaneously high-quality potable water and cooling effect are study on the effect of Mediterranean weather conditions on the per-
obtained from the AD cycle using one heat source input (Alsaman et al., formance of AD unit exhibited a maximum cooling capacity of 14.7 kW
2016). Solar-driven AD units have been reported by several researchers under Nicosia’s weather conditions (Koronaki et al., 2016). Another
demonstrating successful operation (Du et al., 2017; Mohamed et al., theoretical study performed by Balghouthiet et al. demonstrated that
2017). Several studies have been done to investigate different adsorbent the AD system was efficient under the Tunisian weather (Balghouthi
materials such as zeolite, silica gel, ferroaluminophosphate, activated et al., 2008).
alumina, and metal-organic frameworks (MOFs) (Elsayed et al., 2017; Although the overall performance of the AD system is improved
Mitra et al., 2015b). when it integrated with other desalination systems (Ng et al., 2015;
AD is applied for simultaneous desalination and cooling application Shahzad et al., 2018b; Shahzad and Ng, 2017; Shahzad et al., 2016a;
using CPO-27Ni MOF showing superior daily water productivity of Shahzad et al., 2015; Shahzad et al., 2016b), however the purpose of
22.8 m3 ton−1 of adsorbent at an evaporator temperature of 40 °C, a this study is to prove the validity of the modern optimization technique
condenser temperature of 5 °C, and desorption temperature of 95 °C. as an efficient method for increasing the efficiency of the standalone AD
The water productivity increased with increasing evaporator tempera- system driven by solar energy.
ture and decreasing condenser temperature. However, for space cooling Specific daily water production (SDWP), specific cooling power
using evaporator temperature lower than 20 °C, cooling load equivalent (SCP), and coefficient of performance (COP) are usually used to define
to 65 Rton/ton adsorbent was obtained (Youssef et al., 2017). Using the performance of hybrid adsorption desalination cooling (ADC) units.
copper sulfate (new adsorbent material) and water vapor pair, Ali et al. The experimental work done by Ng et al. demonstrated that the AD
(2018) demonstrated through a theoretical study using two bed con- cycle could produce chilled water at a temperature of 7–10 °C with a
figuration that AD could be effectively used for simultaneous water specific cooling capacity of 25–35 Rton/ton, and SDWP of 3–5 m3 per
desalination and cooling purposes using solar energy with a daily water ton of silica gel (Ng et al., 2009). The AD cycle showed a conversion
productivity of 8.2 m3 ton−1 of the adsorbent, specific cooling power performance ratio of 0.8–1.2. A similar SDWP of 4.7 ton per ton of silica
(SCP) of 227 W kg−1 of adsorbent and 0.57 CP. Ni MOF/sea water is also reported using four bed AD system (Wang and Ng, 2005). Four
working pair was used for simultaneous potable water production, and tons of water per ton of silica gel is also obtained using AD prototype
cooling that used for ice production (Dakkama et al., 2017). The results unit (Wang et al., 2007). With heat recovery, the SDWP and PR in-
demonstrated that using three cycles, switching time of 3 min, and creased by 15.7% and 42.5%, respectively. Waste heat was successfully
saline water of 35,000 ppm, a maximum daily ice production rate of 8.9 used to derive AD unit with SDWP of 3.6 m3 and 23 Rton per ton of
tones per ton of adsorbent at 95 °C generation (desorption) tempera- silica gel (Wang et al., 2012). Several theoretical studies have been
ture, −1°C chilled antifreeze, and an ambient temperature of 24 °C, was carried out on AD systems (Ali et al., 2016; Mitra et al., 2015a; Thu
achieved. Through experimental and theoretical study, Saha et al. et al., 2016; Youssef et al., 2015).

476
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Optimizing SADC operating parameters’ is one of the main strate- stage flash (MSF) desalination unit. Their proposed objective function
gies that could improve system output performance and hence reduces included exergy efficiency, total cost, and environmental pollutants. An
the cost of the obtained permeate water. SADC’s performance depends exergetic efficiency of 46.7%, and a total cost of 3.76 million USD/year
on many operating parameters. They can be enumerated as cycle time, are reported using the selected optimal design (Najafi et al., 2014). A
hot water inlet temperature, cooling water inlet temperature and flow multi-objective GA optimization is used to design a combined gas tur-
rate. Experimental optimization of such operating parameters is not an bine and the MSF desalination plant (Hosseini et al., 2012). The opti-
easy task because it requires not only huge time, cost, and efforts, but mization process resulted in reducing the cost of products and en-
also it is limited by a restricted number of data points. Mathematical vironmental cost impact by 13.4% and 53.4%, respectively, whereas
modeling is considered as an effective and efficient way to overcome the total exergy efficiency increased by 14.8% (Hosseini et al., 2012).
this problem by applying different modeling strategies to describe the Mokhtari et al. (2014) used a GA algorithm to optimize a solar desali-
SADC performance and hence parameter optimization can be carried nation plant composed of solar parabolic collectors, steam generators,
out. The idea of using meta-heuristic (evolutional) optimization algo- and multiple effect distillation (MED) unit. The objective function
rithms with different desalination system conditions in order to obtain comprised 17 parameters including the diameter and the width of the
the optimal operating parameters that have been discussed in previous collector, steam generator pinch, approach temperatures, and MED
literature. The performance optimization of a capacitive desalination number of effects (Mokhtari et al., 2014). Based on economic and
system (CDIS) are conducted by Wajid Saleema and Woo-Seung (Saleem thermodynamic point of views, Shakib (Shakib et al., 2012) did the
and Kim, 2018). They employed single- and multi-objective genetic optimization of MED desalination with thermal vapor compression.
algorithms (GA) as an optimization tool. The performance of a CDIS is They used four objective functions for different optimization cases to
evaluated in terms of fresh water recovered, lowest concentration point, maximize exergy efficiency, minimizing specific heat transfer area, and
specific energy consumption, energy consumption per unit mass of maximizing performance ratio (PR). Asgari (Sanaye and Asgari, 2013)
adsorbed salt, accumulated water concentration (average), and amount optimized the performance of combined cycle power plants integrated
of salt (g) adsorbed in capacitive electrodes. Five operating parameters with an MSF desalination unit using multi-objective GA algorithm. The
(spacer volume, electrode capacitance, applied voltage, inlet feed flow design parameters that have been used in the optimization process are
rate, and cell dead volume) have been studied in the optimization drum pressure, heat recovery steam generator (HRSG), pinch point
process (Saleem and Kim, 2018). Najafi et al. (2014) used GA to opti- temperature in HRSG, temperature of top brine in MSF, temperature of
mize solid oxide fuel cell gas turbine hybrid cycle combined with multi- last stage and number of stages of the MSF. Two objective functions are

Fig. 1. Layout of the SADC system.

477
H. Rezk, et al. Solar Energy 181 (2019) 475–489

evaporator to produce a cooling effect. Both of the adsorption and


desorption processes occur in two sequential steps, isobaric and isos-
teric processes as shown in Fig. 2.

3. Mathematical model

This section presents the equations used in the modeling of the


SADC system. The modeling process includes mass and energy balances
of the adsorption beds, the evaporator, and the condenser. The equa-
tions of the performance criteria that formulate SDWP, SCP and COP
are also presented. It is worth mentioning that all parameters’ values,
used in this study, are considered from a previous investigation in
which the model was validated (Alsaman et al., 2017).

Fig. 2. PTC diagram (1–2 isosteric heating, 2–3 isobaric desorption, 3–4 isos-
The governing equations of the SADC system are presented in
teric cooling, and 4–1 isobaric adsorption). this section.
Equilibrium water uptake
D-A model estimated for silica gel, adsorption isotherms
considered; the first one is the summation of investment and opera-
(Eq. (1)) (Askalany et al., 2012)
tional costs in addition to the penalty for producing NOx emissions,
n
while the second objective function is the cycle total amount of exergy RT P
w = w exp ln sat
destruction. The optimizing results lead to total exergy destruction from E P (1)
465 to 484 MWs, and a total annual costs of 192–214 million USD.
Although the overall performance of the AD system is improved when it LDF equation estimated for silica gel, adsorption kinetics
integrated with other desalination systems (Ng et al., 2015; Shahzad (Kim et al., 2014; Saha and Kashiwagi, 1997)
et al., 2018b; Shahzad and Ng, 2017; Shahzad et al., 2016a; Shahzad dC
= ks a v (wo w)
et al., 2015; Shahzad et al., 2016b), the purpose of this study is to prove dt (2)
the validity of the modern optimization technique as an efficient
where,
method for increasing the efficiency ADC system driven by solar energy.
In our previous study (Alsaman et al., 2017), we have designed and Ds
ks a v = Fo
tested ADC driven by solar energy for simultaneous water desalination Rp2 (3)
and cooling effect under Egypt's climate. The system successfully op-
Ea
erated with a SCP of 112 W/kg of silica gel, and a SDWP of 4 m3 per ton Ds = Dso exp
RT (4)
silica gel. The essential goal of this study is to maximize the perfor-
mance of solar-driven adsorption desalination cooling (SADC) system Mass balance equations
(Alsaman et al., 2017) by defining the optimal operating conditions Salt water and condensate water mass balance;
using a modern optimization algorithm. The performance of the SADC
dMsw, evap
system was assessed in terms of largest SDWP, COP and SCP for the .
= msw mp. , cond mb.
dt
, in
(5)
different operating parameters of cycle time, hot water inlet tempera-
ture, cooling water inlet temperature and flow rate. Two optimization Evaporator and salt mass balance;
methods were applied according to the SADC system requirements. dMsw, evap dCads
.
Single- and multi-objective radial movement optimizer (RMO) were = msw , in mb. Msg
dt dt (6)
used to optimize the performance of the SADC system subject to con-
strained decision variables. The obtained results, after optimization, are dXsw, eva . dCads
Msw, eva = Xsw, in msw Xsw, in mb. XD Msg
compared with their corresponding results that obtained experimen- dt
, in
dt (7)
tally.
Energy balance equations
The following sections in the manuscript are organized as follow: (i)
Adsorption bed energy balance;
Section 2 presents the operating principles of the SADC system, (ii)
Section 3 introduces the mathematical model of the SADC system, (iii) dTbed
[(Mcp )cu + (Mcp )al + (Mcp )sg + Msg cp v C ]bed
the optimal parameter determination based radial movement optimizer dt
is explained in Section 4, (IV) Section 5 discusses the obtained results, dCbed
= Msg Hst m w· cp w (Tw, out Tw, in ) bed
and (V) finally the main findings are highlighted in Section 6. dt (8)
Heat of adsorption [38];
2. Operating principles of the SADC system Co
1
n ET Co
1 n
n
Hst = hfg + E ln + ln
C n C (9)
Fig. 1 illustrates the SADC system which is mainly composed of
three main parts, an evaporator, two adsorption beds working alter- Condenser energy balance;
natively and a condenser in addition to a collecting tank, pumps, and
dTcond dMd
valves. As it can be seen in the figure, bed 1 starts as a desorption bed, [(Mcp )cu + (Mcp )iron + (Mcp ) w ]cond = hf (Tcond )
dt dt
while bed 2 starts as an adsorption bed. After adsorption/desorption
dCdes
time (half cycle time), the adsorption and desorption beds are running + hfg (Tcond ) Msg
dt
in the opposite way. During the desorption process, the vapor is re-
jected to the condenser where the condensation process takes place + m w· cp w (Tcond )(Tw, in Tw, out ) condmb·
producing the purified water. In meanwhile the adsorption process is (10)
running in the other bed where the vapor is generating from the

478
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Evaporator energy balance; vector of particles. Analogous to most of heuristic algorithms, the
dT particles’ locations in the search space is defined with an (nPop*nDim)
[Ms, eva cp s (Teva, Xs, eva ) + Mcu, eva cpcu, eva ] eva = hf (Teva, Xs, eva ) ms·, in matrix. Where nPop denotes the solutions’ number which is usually
dt
dC selected based on the user, and nDim denotes the number of dimensions
hfg (Teva ) des Msg which represent the number of variables to be optimized. The initial
dt
·
particles in the matrix are given in equation (19) (Seyedmahmoudian
+ mch cpch (Tch, in Tch, out )
et al., 2016a,b). Fig. 3 shows the flowchart of the RMO algorithm.
hf (Teva, Xs, eva ) mb· (11) x1,1 x1, nDim
The outlet temperatures for heat exchangers is expressed x ij =
xnPop,1 xnPop, nDim (19)
by;

UAhex where i = 1; 2; 3; … ; nPop and j = 1; 2; 3; …; nDim


Tw, out = Thex + (Tw, in Thex ) exp
(m·cp ) w (12) x ij = Lb + r (Ub Lb) (20)

The heat of evaporation, desorption, and condensation where Lb is the lower limit in search area; Ub is the upper limit of
energy are given by; the search area, and r is a random value that taken from normal dis-
tcycle
tribution between range [0 1].
.
Qeva =
0
mch cpch (Tch, in Tch, out ) dt (13) The nPop and nDim are considered during every trail as constant
values and cannot change (Seyedmahmoudian et al., 2016a,b). The
tcycle
Qdes = ·
mhw cp w (Thw, in Thw, out ) dt particles are scattered radially in straight lines from the central point
(14)
0
over the radius according to the velocity vector Vi,j that obtained from
tcycle Eq. (21) (Vanithasri et al., 2016).
Qcond = m w· cp w (Tcw, in Tcw, out ) dt (15)
0
xmax(j ) xmin(j )
Vmax (j) =
Cycle performance parameters; k
Vij = rVmax (j ) i = 1, 2, , nPop
tcycle m w. cp w (Tcw, out Tcw, in )
SDWP = dt j = 1, 2, , nDim (21)
0 hfg Msg (16)
where Xmax(j) , Xmin(j ) : are the limits of jth dimension.
.
tcycle mch cpch (Tch, in Tch, out ) In RMO, the inertia weight is linearly decreased based on the
SCP = dt
0 Msg (17) iteration number and it is denoted by W as illustrated with equation
(22) (Vanithasri et al., 2017).
.
tcycle mchcp ch (Tch, in Tch, out )
COP = ·
dt Wmax Wmin
0 mhwcp w (Thw, in Thw, out ) (18) Wk = Wmax Iterationk
Iterationmax (22)
where Wmax and Wmin, are the max and min values of the inertia weight
and are equal to 1 and 0 respectively. Iterationmax indicates the max-
4. Optimal parameter determination based radial movement imum number of iterations.
optimizer This velocity vector, calculated in Eq. (23), indicates how far the
particles will move around the center. The movement of particles
Radial movement optimization (RMO) is one of swarm-based sto- should be controlled efficiently to balance between the exploration and
chastic optimization methods which has many similarities to the pop- exploitation phases. This balancing is necessary to prevent the optimi-
ular evolutionary techniques such as PSO and DE. The RMO algorithm zation process from stochastic jumps. The random and long jumps are
has a major distinction than the evolutionary techniques such that the the two factors that may inhibit the optimization process from reaching
proposed solutions are distributed radially around a central point in the the global minimum which is its main target (Vanithasri et al., 2016)
search space and this point is updated at each iteration Vijk = rWk Vmax(j) (23)
(Seyedmahmoudian et al., 2016a,b). This technique offers number of
advantages that positively affects the speed of convergence which Fig. 4 explains how the particles run away from the center-point.
makes it very high as well as the efficiency becomes also very high. The center is shown in a blue color, and the particles, scattered around
Moreover, compared to meta-heuristic techniques, the implementation the center, appear in red. The limits of Vmax in the sphere are demon-
of the algorithm and its complexity are very less and tracking perfor- strated with dash lines (Seyedmahmoudian et al., 2016a,b). The fitness
mance is stable (Seyedmahmoudian et al., 2016a,b; Vanithasri et al., value of every proposed solution in the swarm is evaluated iteratively.
2016). Another more advantage that differentiates this algorithm over The radial best (Rbest) vector is the particle that gives the minimum
other evolutional ones is that it requires less size of computation error of the current scattering swarm. Equation (24) is applied for
memory to process the RMO algorithm. This comes from the idea that, calculating the new center-point (Seyedmahmoudian et al., 2016a,b)
for all particles (solutions), the position and velocity are not needed to
Centre New = CentreOld + C1 (Gbest Center Old) + C2 (Rbest Center Old )
be transferred from one iteration to another. Additionally, in utilizing
RMO, the random movement prevents the algorithm from being (24)
trapped at a local optimum while searching for the global best Where: C1 and C2 are two constants that effect Rbest (Radial Best)
(Seyedmahmoudian et al., 2016a,b). As most of the optimizers, the first and Gbest (Global Best) respectively (see Fig. 5).
step of the RMO optimization process is to initialize the proposed C1 and C2 are the two constants that have direct effect on the values
particles throughout the search area, where these particles represent of Gbest and Rbest, respectively. It is preferred to assign a larger value for
the suggested solutions to the optimization problem. These solutions C2 than C1. However, both values should be limited to be less than one.
are then evaluated via the cost (objective) function that defines the The values of C1 and C2 larger than one increase the length of the
fitness values for every particle in each iteration. The movement of the update vector and hence slow down the convergence and weaken the
central point vector which the generated particles are scattered around ability to find the global optimum that are considered the main lim-
relies on the Rbest (Radial Best), Gbest (Global Best) and the random itation during the optimization process. For a fast possible solution, the

479
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Start

Set the RMO algorithm parameters and


Initialize the places ofnPop solutions
randomly

Evaluate the fitness value of each


solution

Choose the best as a Center

Scatter the solutions around the center


along the radii

Evaluate the fitness value of each


solution
no yes
K=1

Centerne w=Centerold +C2Rbk Centerne w=Centerold +C1(Gb-Centerold )+C2(Rbk-Centerold )

Yes No
Gb > Rb

Gb = Rb

K+1

No
K = max
Yes
Optimal solution

End

Fig. 3. The RMO algorithm flowchart.

Gb

k
Update Centerk+1
Center

Rbk
Fig. 4. Scattering the solutions along the radii; Vmax is the largest radius of the
sphere (Al-Dhaifallah et al., 2018). Fig. 5. Center-point movement via the update vector.

480
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Fig. 6. Center’s position update of a three-dimension optimization problem (Al-Dhaifallah et al., 2018).

Table 1
Lower and upper bounds of operating parameters for different studied cases.
Case Half cycle Hot temp Cold temp fhw fcw fch

1 [250 1200] [55 95] [20 40] [0.02 0.4] [0.02 0.4] [0.002 0.04]
2 [250 1200] [55 95] [30 40] [0.02 0.4] [0.02 0.4] [0.002 0.04]
3 [250 1200] [55 95] [20 40] [0.02 0.2] [0.02 0.2] [0.002 0.04]
4 [250 1200] [55 95] [30 40] [0.02 0.2] [0.02 0.2] [0.002 0.04]
5 [250 1200] [55 92.5] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]
6 [250 1200] [55 88.9] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]
7 [250 1200] [55 85.5] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]
8 [250 1200] [55 82.3] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]
9 [250 1200] [55 79] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]
10 [250 1200] [55 76.2] [30 40] [0.02 0.2] [0.02 0.3] [0.002 0.025]

Table 2 optimization model are the operating parameters which were also used
Parameters of the RMO optimizer and their corresponding values. for constraints development, according to the limitations of the SADC
Parameter Value Parameter Value
system. In this work, the SADC’s mathematical model is used to predict
the system performance that has four operating parameters. These
K 5 Population size 10 parameters can be appropriately adjusted to optimize the SADC per-
C1 0.7 Number of iterations 50 formance. Therefore, the four decision variables were chosen based on
C2 0.8 No of Runs 30
the operating parameters.
Table 1 shows the parameters’ constraints of the optimization model
higher values of C1 and C2 can help, but for an unknown function low that derived from the operating parameters which are based on system
values are recommended (Vanithasri et al., 2016) limitations. The witching time is considered in the optimization as the
As soon as the location of the new center-point has been defined, the half cycle time. It has been changed from 250 s up to 1200 s. The chilled
swarm’s particles are then redistributed, and the process is repeated. water temperature is also taken into account as it has been considered
The optimization algorithm is continued up to a stopping criterion is equals the cooling water temperature. In the mathematical model, it has
satisfied. Fig. 6 illustrates the process of updating the new center-point been considered that the cooling water and the inlet water to the eva-
via the update vector (Seyedmahmoudian et al., 2016a,b). porator are taken from the same source where we prefer to make it
The optimization problem can be stated as in the following ques- more realistic. Cooling and chilled water temperatures are considered
tion: what are the optimal operating parameters’ values of the four to be in range of 20–30 °C since this is the range of ambient temperature
variables; cycle time, hot water inlet temperature, cooling water inlet in the Middle East region. Lowering temperature of the water source
temperature and flow rate that maximize the SADC system perfor- under 20 °C, means cooling the water before using it that consumes
mance? energy and would decrease the overall efficiency of the system.
SADC process performance is optimized under constrained oper- In single objective optimization with constrained parameters, every
ating parameters using radial movement optimizer. The cost function was treated separately, and the lower and upper values of
SADC’performance equations are chosen as the cost functions that are the four decision variables are set to their constraints values based on
required to be maximized. Furthermore, the decision variables in the the operating conditions. The goal of using single-objective function

481
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Cost Function SDWP, COP and SCP


Sum
Cycle time

Hot water SADC


mathematical
Cooling water model SDWP
COP
SCP
Flow rate

RMO Optimizer Model of SADC


Fig. 7. The configuration of the optimization process.

Table 3 optimization is to improve the specific performance of the system re-


Optimized output parameters under each studied condition with changing the gardless of the other performance criteria. Multi-objective optimization
cost function. is also used for simultaneous optimization of all performance functions
No SCMOF 1 (Optimized COP only) of the SADC. In multi-objective optimization, the generated solution is
feasible and satisfies all the objective functions at an acceptable level.
COP Change, % SDWP Change, % SCP Change, % Multi-objective optimization will be divided into single constrained, or
multi-constrained based on the number of constraints that will be used
1 0.961 82.7 4.21 −38.45 116.561 −38.63
2 0.608 16.7 2.88 −27.93 79.95 −27.92 for the optimization process. In single constrained multi-objective
3 0.925 67.57 4.043 −36.49 109.439 −38.07 (SCMOF), the objective functions were optimized based on only one
4 0.597 19.64 3.19 −17.4 88.543 −17.34 decision variable, while the others were kept constants. On the other
5 0.606 19.76 2.903 −26.73 80.622 −26.62 hand, in multi-constrained multi-objective (MCMOF) many decision
6 0.606 21.2 2.815 −29.45 78.171 −29.4
variables are involved in the optimization process. When the perfor-
7 0.607 19.72 2.878 −27.32 79.93 −27.29
8 0.607 16.73 2.912 −25.79 80.881 −25.71 mance of the system must be improved with a single operating para-
9 0.607 13.04 2.975 –22.47 82.61 –22.38 meter, SCMOF was used. However, when there is an option to alter
10 0.607 11.79 2.997 −20.33 83.218 −20.28 multiple operating parameters to enhance the SADC performance, then
No SCMOF 2 (Optimized SDWP only) MCMOF was used. In this work, two optimization techniques of RMO
were used to optimize the performance of the SADC process; single
1 0.486 −7.6 6.909 1.01 191.322 0.73
objective optimization with multiple constraints, and multi-objective
2 0.433 −16.89 4.199 5.08 115.131 3.79
3 0.486 −11.96 6.483 1.84 179.701 1.69
optimization with multiple constraints.
4 0.404 −19.04 4.001 3.6 108.526 1.32 Therefore, the Objective function (OF) can be formulated as in
5 0.451 −10.87 4.108 3.69 112.654 2.54 equations (25)-(28). Subject to lower and upper bounds of operating
6 0.415 −17 4.086 2.41 112.106 1.25 parameters for different studied cases based on table 1.
7 0.461 −9.07 4.058 2.47 111.827 1.73
8 0.468 −10 4 1.94 110.794 1.77 SCMOF1 = max(SDWP) (25)
9 0.477 −11.17 3.929 2.4 108.714 2.15
10 0.489 −9.94 3.845 2.21 106.116 1.66 SCMOF2 = max(COP) (26)
No SCMOF 3 (Optimized SCP only)
SCMOF3 = max(SCP) (27)
1 0.48 −8.75 6.897 0.83 191.384 0.77
2 0.412 −20.92 4.183 4.68 116.079 4.65 SDWP COP SCP
3 0.49 −11.23 6.477 1.74 179.786 1.73 MCMOF4 = max + +
4 0.424 −15.03 3.957 2.46 109.73 2.44 1 2 1 (28)
5 0.427 −15.61 4.093 3.31 113.492 3.3
where ξ1 is the optimal value of SDWP using first OF, ξ2 is the op-
6 0.447 −10.6 4.072 2.06 112.895 1.96
7 0.453 −10.65 4.042 2.07 112.145 2.01 timal value of COP using second OF and ξ3 is the optimal value of SCP
8 0.475 −8.65 3.994 1.78 110.836 1.81 using third OF. The purpose of using ξ1, ξ2, ξ3 in equation (28) is to
9 0.493 −8.19 3.92 2.16 108.714 2.15 normalize (same weight) the values of SDWP, SCP, and COP.
10 0.488 −10.13 3.833 1.89 106.319 1.85
The values of the operating parameters of the RMO optimizer are
No MCMOF (All Optimized output (reference case) shown in table 2. The Optimization process configuration is illustrated
in Fig. 7.
1 0.526 – 6.84 – 189.929 –
2 0.521 – 3.996 – 110.926 –
3 0.552 – 6.366 – 176.72 – 5. Results and discussions
4 0.499 – 3.862 – 107.116 –
5 0.506 – 3.962 – 109.862 –
The adsorption desalination system has been modeled and opti-
6 0.5 – 3.99 – 110.727 –
7 0.507 – 3.96 – 109.93 – mized for 10 cases as shown in Table 3. In each case, 30 runs are
8 0.52 – 3.924 – 108.869 – executed to determine the best operating parameters that could achieve
9 0.537 – 3.837 – 106.427 – the highest values of the three required outputs (SDWP, SCP, and COP).
10 0.543 – 3.762 – 104.388 – Heating and cooling water inlet temperature with half cycle time are
considered the operating parameters that needed to be optimized.
Figs. 8–10 show the effect of the heating, cooling water tempera-
tures and half cycle time on the SDWP, SCP, and COP. The minimum

482
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Fig. 8. Effect of hot water inlet temperature with half cycle time on COP and SDWP.

Fig. 9. Effect of hot water inlet temperature with half cycle time on COP and SCP.

half cycle time that could enable a complete cycle has been studied and from the other side is expected where raising the heating water tem-
reported to be 250 s. Fig. 8 shows the increase of SDWP with increasing perature will increase the desorbed water vapor as it can be seen in the
the half cycle time up to 600 s. While after that time, SDWP begins to PTC diagram (Fig. 2).
decrease with increasing cycle time. This indicates that no more vapor Figs. 9 and 10 show the effect of cooling water inlet temperature on
is desorbed and hence its amount is neglected compared to the con- SDWP, SCP, and COP. It can be seen in that the three parameters that
sumed time. Moreover, the effect of half cycle time on COP is also measure the performance of AD system increase with decreasing the
presented in Fig. 8 where the COP always increases with the increase of cooling water inlet temperature. Raising the temperature difference
the cycle time. between the cooling temperature and the bed temperature decreases
The figure also indicated that by raising the hot water inlet tem- the time of the isosteric cooling giving more time for the isobaric
perature the SDWP is increased where maximum SDWP (6.9 m3/day/ cooling since the half cycle time is constant.
ton of silica gel) achieved at the highest hot water inlet temperature of Table 3 shows the optimized values of every single objective con-
90 °C. On the contrary, hot water inlet temperature does not show any strained function based on operating parameters. From the table, it can
significant effect on the COP as by raising the heating temperatures, due be noticed that the objective performance functions of the ASDC system
to the consumed energy by the metal parts of the bed with no sig- were optimized within the range of operating parameters. Three ob-
nificant increase in the desorbed water vapor. jective functions of SDWP, COP and SCP, were maximized using the
Fig. 9 shows the effect of hot water inlet temperature on the SCP optimization tool. The reference case for the comparison is multiple
with changing of half cycle half-cycle time. The maximum SCP is re- constraints, multi-objective function.
corded at the highest studied hot water inlet temperature. The relation Fig. 11 shows a comparison between a single-and multi-objective
between the hot water inlet temperature from a side and SCP and SDWP optimization in terms of percentage performance improvement. As it

483
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Fig. 10. Effect of cooling water inlet temperature with half cycle time SDWP and SCP.

Fig. 11. Percentage change of each output parameter under each case study; (a) optimized COP (b) optimized SDWP (c) optimized SCP.

can be seen from the figure, multi-objective optimization resulted in In order to ensure the validation and robustness of the RMO opti-
smaller maximum improvements compared to those obtained in case of mizer, the algorithm was executed for 30 runs. The best, worst, median,
single-objective optimization. These findings are related to the fact that mean and standard deviation (StD) values of the obtained results are
in multi-objective, a feasible solution sacrifices a certain amount of estimated. Table 4 illustrates the statistical analysis of the RMO opti-
each objective function to get an overall optimal solution. mizer performance for each case study with different objective func-
Fig. 10(a) shows the results of optimizing of COP as an optimized tions. From the table, it can be noticed that the RMO optimizer is ef-
parameter. The figure shows the access to the high values of the COP, ficient in obtaining the optimal parameter that improves the SADC’s
that increases from 11.8 to 82.7%, while a decline of both SDWP and system. This appeared in the low values of the standard deviation (StD)
SCP from −17.4 to −38.45% and from −17.34 to −38.63% respec- of all parameters. The maximum and minimum StD is found to be 0.003
tively. Fig. 10(b) shows the results of optimizing of SDWP as an opti- and 0.115, respectively. As an example, the details of 30 runs for the
mized parameter to access to the high values, which increases from first and ninth cases are shown in Tables 5 and 6, respectively.
1.01 to 5.08%. As can be seen in the figure, the SCP values increase Fig. 12 presents the results of the optimization process for two
from 0.73 to 3.79%, while decline of COP from −7.6 to −19.04%. different cases (case 1 and case 9) with more than one run for all per-
Fig. 10 (c) shows the results of optimizing of SCP as an optimized formance results of the AD system. The figure shows the optimum
parameter to access to the high values, which increases from 0.77 to conditions that maximize each of COP, SDWP, and SCP either alone or
4.65%. As it can be seen in the figure, the SDWP values increase from together. As it can be seen in Fig. 11, turns out that each parameter
0.83 to 4.68%, while decline of COP from −8.19 to −20.92%. could reach its highest value apart, however, there is no case that three

484
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Table 4
Statistical analysis of RMO optimizer for each case study.
Case COP SDWP SCP All no COP SDWP SCP All

1 Best 0.961 6.909 191.38 2.527 6 Best 0.607 4.086 112.89 2.781
Worst 0.653 6.651 187.19 2.473 Worst 0.597 3.855 109.01 2.749
Median 0.921 6.869 190.56 2.518 Mean 0.602 4.047 111.503 2.774
Mean 0.858 6.851 190.48 2.516 Median 0.602 4.034 111.598 2.772
Std. dev. 0.115 0.054 0.823 0.011 Std. dev. 0.003 0.047 0.969 0.008

2 Best 0.608 4.199 116.07 2.765 7 Best 0.607 4.058 112.145 2.796
Worst 0.597 4.034 111.00 2.694 Worst 0.591 3.934 106.476 2.724
Mean 0.605 4.176 115.58 2.748 Mean 0.603 4.029 111.301 2.789
Median 0.604 4.157 115.07 2.746 Median 0.601 4.019 110.741 2.781
Std. dev. 0.003 0.04 1.186 0.015 Std. dev. 0.005 0.033 1.472 0.018

3 Best 0.925 6.483 179.78 2.561 8 Best 0.607 4 110.836 2.819


Worst 0.648 5.92 174.51 2.488 Worst 0.596 3.84 105.26 2.736
Mean 0.873 6.446 179.374 2.549 Mean 0.603 3.964 110.288 2.807
Median 0.833 6.408 178.802 2.543 Median 0.602 3.948 110.085 2.8
Std. dev. 0.096 0.111 1.206 0.017 Std. dev. 0.003 0.05 1.018 0.021

4 Best 0.603 4.001 109.73 2.769 9 Best 0.607 3.929 108.714 2.84
Worst 0.581 3.8 103.097 2.727 Worst 0.594 3.713 103.83 2.801
Mean 0.597 3.93 109.241 2.747 Mean 0.603 3.892 107.506 2.832
Median 0.596 3.927 108.742 2.746 Median 0.602 3.878 107.326 2.827
Std. dev. 0.005 0.04 1.487 0.013 Std. dev. 0.003 0.054 1.167 0.011

5 Best 0.606 4.108 113.492 2.769 10 Best 0.607 3.845 106.319 2.855
Worst 0.594 4.009 109.916 2.699 Worst 0.595 3.645 102.006 2.778
Mean 0.601 4.085 112.918 2.753 Mean 0.604 3.823 105.245 2.845
Median 0.601 4.077 112.642 2.749 Median 0.603 3.802 104.947 2.836
Std. dev. 0.003 0.021 0.834 0.017 Std. dev. 0.003 0.046 1.201 0.019

Table 5
The details of 30 Runs for the first case.
Runs Optimized COP only Optimized SDWP only Optimized SCP only All Optimized output

COP SDWP SCP COP SDWP SCP COP SDWP SCP OF COP SDWP SCP

1 0.78 2.967 83.168 0.496 6.842 189.816 0.494 6.905 190.236 2.473 0.606 6.368 176.575
2 0.937 4.208 113.696 0.493 6.888 191.327 0.52 6.836 189.802 2.519 0.511 6.866 190.567
3 0.657 5.188 144.102 0.474 6.883 190.934 0.496 6.942 190.417 2.519 0.535 6.778 188.129
4 0.9 3.34 94.995 0.503 6.905 190.889 0.515 6.838 189.798 2.51 0.548 6.711 185.754
5 0.95 4.052 112.238 0.521 6.835 189.827 0.507 6.878 190.88 2.515 0.54 6.748 187.311
6 0.905 3.748 109.082 0.5 6.886 191.117 0.512 6.869 190.479 2.52 0.527 6.809 189.126
7 0.946 4.174 114.758 0.507 6.88 190.83 0.493 6.856 190.397 2.527 0.539 6.797 188.581
8 0.867 3.73 104.911 0.48 6.874 190.6 0.489 6.895 191.367 2.518 0.541 6.753 187.485
9 0.926 4.205 112.353 0.511 6.87 190.65 0.467 6.881 191.085 2.519 0.578 6.623 183.795
10 0.872 3.762 105.944 0.455 6.805 188.684 0.491 6.882 190.992 2.523 0.545 6.766 187.314
11 0.872 3.768 105.97 0.465 6.882 191.145 0.495 6.885 191.108 2.505 0.509 6.823 189.401
12 0.655 4.998 138.802 0.451 6.853 190.036 0.466 6.865 190.258 2.52 0.514 6.855 190.369
13 0.716 2.438 68.277 0.514 6.853 190.298 0.516 6.853 190.217 2.51 0.533 6.763 187.417
14 0.938 4.21 113.804 0.475 6.893 191.359 0.491 6.87 190.833 2.527 0.526 6.84 189.929
15 0.938 4.21 113.705 0.447 6.794 188.503 0.488 6.867 190.583 2.515 0.503 6.876 191.01
16 0.947 3.814 107.881 0.516 6.833 189.672 0.451 6.868 190.555 2.518 0.536 6.773 187.948
17 0.926 4.205 112.351 0.494 6.89 191.3 0.495 6.89 191.203 2.516 0.524 6.809 189.052
18 0.656 4.863 135.074 0.492 6.9 190.73 0.518 6.852 190.028 2.527 0.569 6.701 185.206
19 0.653 5.213 144.67 0.541 6.744 187.037 0.48 6.897 191.384 2.524 0.538 6.789 188.357
20 0.656 4.677 129.883 0.51 6.869 190.629 0.503 6.879 191.016 2.526 0.538 6.79 188.511
21 0.948 4.21 115.008 0.498 6.886 191.197 0.505 6.904 189.576 2.522 0.516 6.861 190.368
22 0.872 3.769 105.978 0.503 6.857 190.31 0.499 6.889 191.136 2.521 0.532 6.805 188.43
23 0.656 5.198 144.338 0.5 6.885 191.145 0.55 6.748 187.192 2.519 0.538 6.779 187.646
24 0.937 4.203 113.6 0.438 6.83 189.434 0.511 6.865 190.596 2.501 0.504 6.837 189.408
25 0.946 4.21 114.783 0.492 6.895 191.336 0.476 6.95 190.935 2.51 0.563 6.647 184.564
26 0.961 4.21 116.561 0.469 6.802 188.878 0.496 6.853 190.28 2.506 0.504 6.845 190.14
27 0.917 3.443 98.337 0.511 6.8 188.739 0.504 6.878 190.953 2.517 0.524 6.81 189.131
28 0.936 4.163 112.694 0.486 6.909 191.322 0.489 6.898 191.295 2.508 0.547 6.697 185.946
29 0.926 3.932 112.302 0.429 6.651 180.834 0.526 6.817 189.305 2.517 0.527 6.802 188.764
30 0.928 4.205 112.596 0.513 6.844 189.904 0.511 6.868 190.493 2.523 0.526 6.837 189.146

parameters could be maximized together. Accordingly, the COP, SDWP value of each parameter in this case equal about 95% of result opti-
and SCP values are 0.65, 6.5 m3/ton silica gel and 180 W/kg silica gel, mization of each parameter separately as an optimized parameter. The
respectively at considered each parameter an optimized parameter. objective function variation with each iteration for case 1 and case 9 as
While at optimization of all parameters of AD system together, the an example is shown in Fig. 12.
optimized parameter result is more than 2.8, which meaning that the Fig. 13 shows the values of the optimized output parameters of the

485
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Table 6
The details of 30 runs for ninth case.
Runs Optimized COP only Optimized SDWP only Optimized SCP only All Optimized output

COP SDWP SCP COP SDWP SCP COP SDWP SCP CF COP SDWP SCP

1 0.604 2.926 81.224 0.472 3.854 106.155 0.431 3.885 107.558 2.834 0.535 3.831 106.255
2 0.603 2.729 75.763 0.495 3.92 108.671 0.521 3.748 103.83 2.84 0.537 3.837 106.427
3 0.601 3.07 85.235 0.464 3.91 108.371 0.529 3.813 105.736 2.835 0.548 3.792 105.089
4 0.602 2.961 82.16 0.503 3.909 108.525 0.515 3.813 105.734 2.828 0.508 3.908 108.297
5 0.597 3.068 85.113 0.433 3.794 104.728 0.46 3.914 108.593 2.838 0.541 3.817 105.965
6 0.605 3.109 86.315 0.473 3.887 107.756 0.493 3.92 108.714 2.832 0.525 3.86 107.115
7 0.599 2.889 80.164 0.505 3.878 107.539 0.519 3.887 107.794 2.817 0.546 3.762 104.329
8 0.602 3.122 86.662 0.475 3.929 108.689 0.497 3.891 107.913 2.827 0.555 3.751 104.047
9 0.603 2.771 76.902 0.521 3.826 106.107 0.45 3.905 108.333 2.836 0.543 3.807 105.643
10 0.601 3.022 83.894 0.519 3.888 107.819 0.464 3.846 106.63 2.839 0.524 3.881 107.43
11 0.604 3.096 85.938 0.433 3.886 107.666 0.541 3.819 105.928 2.82 0.535 3.802 105.477
12 0.604 2.742 76.117 0.469 3.923 108.702 0.503 3.899 108.226 2.82 0.519 3.858 106.841
13 0.604 3.177 88.214 0.528 3.862 107.198 0.533 3.849 106.748 2.825 0.556 3.745 103.829
14 0.606 3.045 84.561 0.443 3.902 108.059 0.49 3.869 107.332 2.835 0.517 3.891 107.921
15 0.604 2.885 80.09 0.481 3.923 108.814 0.527 3.865 107.273 2.817 0.561 3.714 103.085
16 0.594 2.532 70.225 0.495 3.905 108.349 0.474 3.867 107.318 2.819 0.559 3.723 103.34
17 0.606 2.895 80.4 0.474 3.921 108.779 0.502 3.824 105.945 2.84 0.528 3.865 107.189
18 0.599 3.183 88.329 0.464 3.897 108.034 0.489 3.909 108.208 2.833 0.52 3.881 107.456
19 0.596 3.197 88.663 0.518 3.896 107.78 0.474 3.942 108.442 2.801 0.546 3.73 103.488
20 0.604 2.742 76.139 0.503 3.91 108.512 0.505 3.91 108.446 2.826 0.512 3.891 107.817
21 0.598 3.174 88.126 0.504 3.726 103.291 0.516 3.869 107.318 2.803 0.544 3.739 103.776
22 0.603 3.029 84.088 0.494 3.851 106.814 0.513 3.869 107.332 2.839 0.541 3.82 105.991
23 0.601 2.849 79.085 0.421 3.866 107.04 0.498 3.891 107.926 2.835 0.539 3.824 105.828
24 0.604 2.87 79.663 0.477 3.929 108.714 0.505 3.912 108.399 2.819 0.5 3.913 108.591
25 0.597 2.773 76.963 0.489 3.834 106.384 0.508 3.906 108.355 2.836 0.537 3.827 106.139
26 0.603 2.696 74.853 0.472 3.918 108.673 0.447 3.894 107.865 2.837 0.543 3.81 105.747
27 0.599 2.988 82.946 0.501 3.913 108.561 0.546 3.795 105.324 2.831 0.512 3.899 108.182
28 0.605 3.118 86.558 0.519 3.881 107.655 0.52 3.874 107.454 2.836 0.532 3.846 106.599
29 0.597 3.217 89.275 0.515 3.713 102.668 0.524 3.85 106.748 2.806 0.566 3.673 101.878
30 0.607 2.975 82.61 0.449 3.887 107.741 0.451 3.907 108.363 2.816 0.534 3.799 105.348

Fig. 12. Variation of the selected objective function with each iteration for case 1 and case 9.

SADC system (COP, SDWP, and SCP) under the considered objective COP obtained using the proposed optimized model and the experi-
functions. mental data obtained in our previous study (Alsaman et al., 2017). The
The optimal input parameters of SADC system of each studied case, plots show that the optimization process gives better results (70% in-
with varying the objective function, are shown in Table 7. crease in terms of SDWP and SCP compared to those obtained experi-
Figs. 14 and 15 present a comparison between the SDWP, SCP and mentally using the same design and materials (Alsaman et al., 2017)) at

486
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Fig. 13. Optimized performance under different cost functions.

Table 7
The optimal operating parameters under each case study.
Case Optimum Run tcycle Tcwc Twi Twic Tchi Tcond fhw fcw fch fcwc Te Tcwco

1 COP 26 250 20 55 20 20 40 0.02 0.4 0.003 0.069 20 22.19


DWP 28 570 38.84 95 20 20 20 0.4 0.4 0.0131 0.2814 20 30.67
SCP 19 552 33.39 95 20 20 20 0.4 0.4 0.04 0.02 20 26.86
All 18 773 33.0 83.6 20 40 40 0.39 0.4 0.005 0.302 28.60 30.52

2 COP 8 1200 30 64.2 30 40 30.79 0.34 0.4 0.04 0.02 32.42 30


DWP 27 637 40 95 30 35.30 32.77 0.24 0.4 0.003 0.39 30 30
SCP 7 571 30 95 30 40 30 0.4 0.4 0.04 0.26 30 40
All 1 817 35.03 82.5 30 32.37 34.88 0.36 0.4 0.027 0.02 33.36 39.54

3 COP 11 250 40 55 20 36.52 40 0.02 0.2 0.003 0.12 20.05 40


DWP 18 593 26.79 95 20 20 40 0.2 0.2 0.02 0.2 29.2 20
SCP 20 608 20 95 20 40 20 0.2 0.2 0.02 0.2 26.72 20
All 14 795 37.74 89.26 20 33.68 24.22 0.2 0.2 0.02 0.2 29.90 29.13

4 COP 16 1200 32.88 66.85 30 40 38.10 0.2 0.2 0.018 0.069 40 34.57
DWP 3 580 37.04 95 30 39.10 30 0.2 0.2 0.003 0.103 40 37.49
SCP 20 632 33.82 95 30 30 35.36 0.2 0.2 0.02 0.02 30 40
All 4 785 32.79 85.21 30 33.16 35.17 0.2 0.2 0.015 0.124 39.81 36.86

5 COP 6 1200 39.80 65.89 30 39.96 30 0.12 0.3 0.024 0.167 40 36.43
DWP 19 679 34.19 92.5 30 31.75 31.41 0.2 0.3 0.003 0.074 35.30 37.49
SCP 21 597 40 92.5 30 30 30 0.2 0.3 0.025 0.046 30 30
All 16 832 39.53 87.26 30 36.09 31.45 0.15 0.3 0.009 0.232 37.24 38.21

6 COP 9 1200 37.30 65.29 30 40 30.3 0.18 0.3 0.025 0.054 30 40


DWP 13 532 30 88.9 30 36.29 40 0.2 0.3 0.005 0.3 34.33 30
SCP 25 612 32.31 88.5 30 31.25 37.85 0.2 0.3 0.024 0.111 31.42 36.56
All 21 787 40 86.39 30 33.51 39.82 0.2 0.3 0.025 0.3 40 30

7 COP 8 1200 35.06 65.08 30 40 33.84 0.2 0.3 0.025 0.07 32.33 40
DWP 13 624 30 85.5 30 40 30 0.2 0.3 0.005 0.060 30 30
SCP 14 597 40 85.5 30 37.73 36.29 0.2 0.3 0.025 0.26 36.71 40
All 13 834 33.51 85.5 30 37.61 37.13 0.2 0.3 0.008 0.25 36.54 32.69

8 COP 14 1200 38.32 65.40 30 40 40 0.2 0.3 0.025 0.13 30 31.22


DWP 6 610 30 82.3 30 30 35.17 0.2 0.3 0.015 0.3 40 30
SCP 17 630 36.08 82.3 30 40 33.80 0.19 0.3 0.025 0.247 39.136 30
All 24 813 32.94 82.3 30 36.85 40 0.2 0.3 0.0159 0.1353 31.21 30

9 COP 30 1200 40 66.20 30 40 31.952 0.2 0.3 0.025 0.1967 40 40


DWP 24 604 30 79 30 30 40 0.2 0.3 0.0125 0.3 30.40 35.36
SCP 6 655 30 79 30 40 40 0.2 0.3 0.0182 0.3 35.57 34.87
All 2 847 33.42 79 30 30 32.33 0.2 0.3 0.0165 0.21 40 40

10 COP 19 1200 35.96 66.49 30 40 39.82 0.2 0.3 0.025 0.2323 39.56 37.88
DWP 28 616 40 76.2 30 40 30 0.2 0.3 0.0068 0.02 37.50 30
SCP 16 610 38.30 76.2 30 40 30 0.2 0.3 0.017 0.02 32.17 31.35
All 27 834 30 76.2 30 40 30 0.2 0.3 0.025 0.3 32.18 40

the same regeneration temperatures. Where the SDWP increased from in performance would be related to the decreasing of the cooling
4 m3/ton/day to 6.9 m3/ton/day, and the SCP increased from 112 W/kg temperature from 25 °C to 15 °C, which duplicates the working capacity
to 191 W/kg, at the same regenerating temperature. Such improvement (Fig. 2).

487
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Fig. 14. Optimized and experimental results of the ADS at different regeneration temperatures at cooling temperature of 20 °C, experimental data are used after
permission from (Alsaman et al., 2017).

Fig. 15. Optimized and experimental results of the SADC at different regeneration temperatures at cooling temperature of 20 °C, experimental data are used after
permission from (Alsaman et al., 2017).

6. Conclusion the possible outputs of the proposed SADC system. A 70% increase, in
both SDWP and SCP, is reported just by using the right operating
Improving the performance of the silica gel SADC system is the main conditions with no change in the system design or the used materials. It
goal of this research. A proofed and validated lumped mathematical is worth mentioning that the modern optimization tool, used in this
model based on a previous experimental study are proposed and em- research, improved the performance of the SADC system by obtaining
ployed in the modeling of the system. Optimizing the performance of a the most appropriate driving conditions.
silica gel SADC system is accomplished by using a modern optimizer to
obtain the optimum operating conditions of heating, cooling, cycle time References
and flow rate of heating water. Simple and fast optimization algorithm
named radial movement optimizer is used for determining the values of Abdelkareem, M.A., El Haj Assad, M., Sayed, E.T., Soudan, B., 2018a. Corrigendum to
these optimal operating parameters. Several sets of the parameters’ “Recent progress in the use of renewable energy sources to power water desalination
plants. Desalination 435, 97–113.
constraints representing different conditions are considered during the Abdelkareem, M.A., El Haj Assad, M., Sayed, E.T., Soudan, B., 2018b. Recent progress in
optimization process. To ensure robustness and validity of the opti- the use of renewable energy sources to power water desalination plants. Desalination
mizer, the algorithm is executed for 30 runs. 435, 97–113.
Al-Dhaifallah, M., Nassef, A.M., Rezk, H., Nisar, K.S., 2018. Optimal parameter design of
The results, obtained from the optimization process, are compared fractional order control based INC-MPPT for PV system. Sol. Energy 159, 650–664.
with the experimental results. The results shows that the SADC system Ali, E.S., Askalany, A.A., Harby, K., Diab, M.R., Alsaman, A.S., 2018. Adsorption desali-
could present higher performance if it has been worked at the de- nation-cooling system employing copper sulfate driven by low grade heat sources.
Appl. Therm. Eng. 136, 169–176.
termined operating conditions. An amount of 6.9 m3/day/ton desali-
Ali, S.M., Haider, P., Sidhu, D.S., Chakraborty, A., 2016. Thermally driven adsorption
nated water, 191 W/kg cooling capacity and 0.961 COP are reported as cooling and desalination employing multi-bed dual-evaporator system. Appl. Therm.

488
H. Rezk, et al. Solar Energy 181 (2019) 475–489

Eng. 106, 1136–1147. cycle power plants integrated with Multi-stage Flash (MSF) desalination unit.
Alsaman, A.S., Askalany, A.A., Harby, K., Ahmed, M.S., 2016. A state of the art of hybrid Desalination 320, 105–117.
adsorption desalination–cooling systems. Renew. Sustain. Energy Rev. 58, 692–703. Seyedmahmoudian, M., Horan, B., Rahmani, R., Maung Than Oo, A., Stojcevski, A.,
Alsaman, A.S., Askalany, A.A., Harby, K., Ahmed, M.S., 2017. Performance evaluation of 2016a. Efficient photovoltaic system maximum power point tracking using a new
a solar-driven adsorption desalination-cooling system. Energy 128, 196–207. technique. Energies 9 (3), 147.
Arabpour Roghabadi, F., Ahmadi, N., Ahmadi, V., Di Carlo, A., Oniy Aghmiuni, K., Seyedmahmoudian, M., Horan, B., Soon, T.K., Rahmani, R., Oo, A.M.T., Mekhilef, S.,
Shokrolahzadeh Tehrani, A., Ghoreishi, F.S., Payandeh, M., Mansour Rezaei Fumani, Stojcevski, A., 2016b. State of the art artificial intelligence-based MPPT techniques
N., 2018. Bulk heterojunction polymer solar cell and perovskite solar cell: concepts, for mitigating partial shading effects on PV systems – a review. Renew. Sustain.
materials, current status, and opto-electronic properties. Sol. Energy 173, 407–424. Energy Rev. 64, 435–455.
Askalany, A.A., 2016. Innovative mechanical vapor compression adsorption desalination Shahzad, M.W., Burhan, M., Ang, L., Ng, K.C., 2017. Energy-water-environment nexus
(MVC-AD) system. Appl. Therm. Eng. 106, 286–292. underpinning future desalination sustainability. Desalination 413, 52–64.
Askalany, A.A., Salem, M., Ismail, I., Ali, A.H.H., Morsy, M., 2012. Experimental study on Shahzad, M.W., Burhan, M., Ghaffour, N., Ng, K.C., 2018a. A multi evaporator desali-
adsorption–desorption characteristics of granular activated carbon/R134a pair. Int. J. nation system operated with thermocline energy for future sustainability.
Refrig. 35 (3), 494–498. Desalination 435, 268–277.
Bait, O., Si–Ameur, M., 2018. Enhanced heat and mass transfer in solar stills using na- Shahzad, M.W., Burhan, M., Ng, K.C., 2018b. Energy storage & desalination. Int. J.
nofluids: a review. Sol. Energy 170, 694–722. Comput. Phys. Ser. 1 (2), 52–60.
Balghouthi, M., Chahbani, M.H., Guizani, A., 2008. Feasibility of solar absorption air Shahzad, M.W., Burhan, M., Ng, K.C., 2019. A standard primary energy approach for
conditioning in Tunisia. Build. Environ. 43 (9), 1459–1470. comparing desalination processes. npj Clean Water 2 (1), 1.
Dakkama, H.J., Youssef, P.G., Al-Dadah, R.K., Mahmoud, S., 2017. Adsorption ice making Shahzad, M.W., Ng, K.C., 2017. An improved multievaporator adsorption desalination
and water desalination system using metal organic frameworks/water pair. Energy cycle for gulf cooperation council countries. Energy Technol. 5 (9), 1663–1669.
Convers. Manage. 142, 53–61. Shahzad, M.W., Ng, K.C., Thu, K., 2016a. Future sustainable desalination using waste
Du, B., Gao, J., Zeng, L., Su, X., Zhang, X., Yu, S., Ma, H., 2017. Area optimization of solar heat: kudos to thermodynamic synergy. Environ. Sci. Water Res. Technol. 2 (1),
collectors for adsorption desalination. Sol. Energy 157, 298–308. 206–212.
Elsayed, E., Al-Dadah, R., Mahmoud, S., Anderson, P.A., Elsayed, A., Youssef, P.G., 2017. Shahzad, M.W., Thu, K., Kim, Y.-D., Ng, K.C., 2015. An experimental investigation on
CPO-27(Ni), aluminium fumarate and MIL-101(Cr) MOF materials for adsorption MEDAD hybrid desalination cycle. Appl. Energy 148, 273–281.
water desalination. Desalination 406, 25–36. Shahzad, M.W., Thu, K., Ng, K.C., WonGee, C., 2016b. Recent development in thermally
Ghaffour, N., Missimer, T.M., Amy, G.L., 2013. Technical review and evaluation of the activated desalination methods: achieving an energy efficiency less than 2.5
economics of water desalination: current and future challenges for better water kWhelec/m3. Desalin. Water Treat. 57 (16), 7396–7405.
supply sustainability. Desalination 309, 197–207. Shakib, S.E., Amidpour, M., Aghanajafi, C., 2012. A new approach for process optimi-
Hansen, K., Vad Mathiesen, B., 2018. Comprehensive assessment of the role and potential zation of a METVC desalination system. Desalin. Water Treat. 37 (1–3), 84–96.
for solar thermal in future energy systems. Sol. Energy 169, 144–152. Siddiqui, O., Dincer, I., 2018. Examination of a new solar-based integrated system for
Hosseini, S.R., Amidpour, M., Shakib, S.E., 2012. Cost optimization of a combined power desalination, electricity generation and hydrogen production. Sol. Energy 163,
and water desalination plant with exergetic, environment and reliability considera- 224–234.
tion. Desalination 285, 123–130. Singh, A.K., Singh, D.B., Mallick, A., Kumar, N., 2018. Energy matrices and efficiency
Kasaeian, A., Sharifi, S., Yan, W.-M., 2018. Novel achievements in the development of analyses of solar distiller units: a review. Sol. Energy 173, 53–75.
solar ponds: a review. Sol. Energy 174, 189–206. Takalkar, G.D., Bhosale, R.R., 2018. Mathematical modeling, simulation and optimization
Kim, Y.-D., Thu, K., Masry, M.E., Ng, K.C., 2014. Water quality assessment of solar-as- of solar thermal powered Encontech engine for desalination. Sol. Energy 172,
sisted adsorption desalination cycle. Desalination 344, 144–151. 104–115.
Koronaki, I.P., Papoutsis, E.G., Papaefthimiou, V.D., 2016. Thermodynamic modeling and Thu, K., Saha, B.B., Chua, K.J., Ng, K.C., 2016. Performance investigation of a waste heat-
exergy analysis of a solar adsorption cooling system with cooling tower in driven 3-bed 2-evaporator adsorption cycle for cooling and desalination. Int. J. Heat
Mediterranean conditions. Appl. Therm. Eng. 99, 1027–1038. Mass Transf. 101, 1111–1122.
Mitra, S., Kumar, P., Srinivasan, K., Dutta, P., 2015a. Performance evaluation of a two- Thu, K., Yanagi, H., Saha, B.B., Ng, K.C., 2017. Performance investigation on a 4-bed
stage silica gel+ water adsorption based cooling-cum-desalination system. Int. J. adsorption desalination cycle with internal heat recovery scheme. Desalination 402,
Refrig. 58, 186–198. 88–96.
Mitra, S., Kumar, P., Srinivasan, K., Dutta, P., 2015b. Performance evaluation of a two- Vanithasri, M., Balamurugan, R., Lakshminarasimman, L., 2016. Modified radial move-
stage silica gel + water adsorption based cooling-cum-desalination system. Int. J. ment optimization (MRMO) technique for estimating the parameters of fuel cost
Refrig. 58, 186–198. function in thermal power plants. Eng. Sci. Technol. Int. J. 19 (4), 2035–2042.
Mohamed, M.H., William, G.E., Fatouh, M., 2017. Solar energy utilization in water pro- Vanithasri, M., Balamurugan, R., Lakshminarasimman, L., 2017. Radial movement opti-
duction from humid air. Sol. Energy 148, 98–109. mization (RMO) technique for solving unit commitment problem in power systems. J.
Mokhtari, H., Bidi, M., Gholinejad, M., 2014. Thermoeconomic analysis and multi- Electr. Syst. Inform. Technol.
objective optimization of a solar desalination plant. J. Solar Energy 2014. Wang, D., Yan, W., Vijapur, S.H., Botte, G.G., 2012. Enhanced electrocatalytic oxidation
Najafi, B., Shirazi, A., Aminyavari, M., Rinaldi, F., Taylor, R.A., 2014. Exergetic, eco- of urea based on nickel hydroxide nanoribbons. J. Power Sources 217, 498–502.
nomic and environmental analyses and multi-objective optimization of an SOFC-gas Wang, X., Ng, K.C., 2005. Experimental investigation of an adsorption desalination plant
turbine hybrid cycle coupled with an MSF desalination system. Desalination 334 (1), using low-temperature waste heat. Appl. Therm. Eng. 25 (17), 2780–2789.
46–59. Wang, X., Ng, K.C., Chakarborty, A., Saha, B.B., 2007. How heat and mass recovery
Ng, K.C., Shahzad, M.W., Son, H.S., Hamed, O.A., 2017. An exergy approach to efficiency strategies impact the performance of adsorption desalination plant: theory and ex-
evaluation of desalination. Appl. Phys. Lett. 110 (18), 184101. periments. Heat Transfer Eng. 28 (2), 147–153.
Ng, K.C., Thu, K., Chakraborty, A., Saha, B.B., Chun, W.G., 2009. Solar-assisted dual- Wu, G., Kutlu, C., Zheng, H., Su, Y., Cui, D., 2017. A study on the maximum gained output
effect adsorption cycle for the production of cooling effect and potable water. Int. J. ratio of single-effect solar humidification-dehumidification desalination. Sol. Energy
Low-Carbon Technol. 4 (2), 61–67. 157, 1–9.
Ng, K.C., Thu, K., Kim, Y., Chakraborty, A., Amy, G., 2013. Adsorption desalination: an Youssef, P., Al-Dadah, R., Mahmoud, S., Dakkama, H., Elsayed, A., 2015. Effect of eva-
emerging low-cost thermal desalination method. Desalination 308, 161–179. porator and condenser temperatures on the performance of adsorption desalination
Ng, K.C., Thu, K., Oh, S.J., Ang, L., Shahzad, M.W., Ismail, A.B., 2015. Recent develop- cooling cycle. Energy Procedia 75, 1464–1469.
ments in thermally-driven seawater desalination: energy efficiency improvement by Youssef, P.G., Dakkama, H., Mahmoud, S.M., Al-Dadah, R.K., 2017. Experimental in-
hybridization of the MED and AD cycles. Desalination 356, 255–270. vestigation of adsorption water desalination/cooling system using CPO-27Ni MOF.
Saha, B.B., Kashiwagi, T., 1997. Experimental investigation of an advanced adsorption Desalination 404, 192–199.
refrigeration cycle. Ashrae Trans. 103, 50. Youssef, P.G., Mahmoud, S.M., Al-Dadah, R.K., 2016. Numerical simulation of combined
Saleem, M.W., Kim, W.-S., 2018. Parameter-based performance evaluation and optimi- adsorption desalination and cooling cycles with integrated evaporator/condenser.
zation of a capacitive deionization desalination process. Desalination 437, 133–143. Desalination 392, 14–24.
Sanaye, S., Asgari, S., 2013. Four E analysis and multi-objective optimization of combined

489

You might also like