Content uploaded by Farshad Tajeddini

Author content

All content in this area was uploaded by Farshad Tajeddini on Mar 15, 2019

Content may be subject to copyright.

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Thermal analysis and optimization of a system for water harvesting from

humid air using thermoelectric coolers

M. Eslami

a,⁎

, F. Tajeddini

b

, N. Etaati

a

a

School of Mechanical Engineering, Shiraz University, Shiraz, Iran

b

School of Mechanical Engineering, Sharif University of Technology, Tehran, Iran

ARTICLE INFO

Keywords:

Thermoelectric cooler

Atmospheric water generation

Thermodynamic optimization

ABSTRACT

Condensation of water vapor available in atmospheric air can be considered as a solution for water scarcity

problem. In this paper, a comprehensive thermodynamic analysis of water production from humid air using

thermoelectric coolers (TECs) is presented. The system consists of a number of thermoelectric coolers, a fan to

supply the required air ﬂow circulation, two cold and hot air channels, heat sinks and solar cells for powering the

thermoelectric coolers and fan. Eﬀects of various design parameters are investigated and discussed. The pro-

posed design is optimized to get the maximum eﬀectiveness which is deﬁned as the amount of produced water

per unit of energy consumption. Sensitivity analysis is used to ﬁnd the optimum number of TECs, length of the

channels and performance of the system at diﬀerent temperatures. The resulting system is capable of producing

26 ml of water within 1 h from the air with 75% relative humidity and the temperature of 318 K by consuming

only 20 W of electrical power. In addition, the annual performance and optimization of this device in three

southern cities of Iran are presented based on hourly meteorological data. Finally, comparison of the present

system with other air water generators indicates that the proposed design is the most energy eﬃcient system

among similar devices especially in high relative humidity.

1. Introduction

Nowadays, water scarcity is one of the most serious issues in the

world. Approximately, around 97.5% of the water content of the earth

is salty seawater which means only 2.5% of the existing water is fresh.

Almost 70% of this amount is frozen at the polar ice caps, and around

30% exists in the form of moisture in the air or underground aquifers.

Therefore, it can be concluded that only less than 1% of the earth’s fresh

water is accessible for direct human use [1]. Mekonnen et al. [2] no-

tiﬁed that as many as four billion people all around the world face the

problem of water scarcity for at least one month per year. All these

factors have brought about the need to study solutions addressing the

water scarcity problem.

Among diﬀerent methods of desalination, atmospheric water gen-

eration (AWG) can be an easy method for fresh water production

especially for places with high relative humidity. In this approach,

ambient air is cooled down below the dew point temperature and the

condensed water is collected. Vapor compression refrigeration, ab-

sorption refrigeration and thermoelectric cooling (TEC) can be used for

this purpose. Thermoelectric coolers are devices which function on the

basis of Peltier eﬀect. By passing an electric current through them, they

produce a temperature diﬀerence resulting in a cooling eﬀect. In

comparison with vapor compression and absorption refrigeration, TEC

devices have no moving parts and require less maintenance. Therefore,

they are suitable for designing simple and portable AWG systems.

However, the designer must be very careful about the performance and

eﬃciency of TECs at various operating conditions.

There are diﬀerent approaches to study properties and modeling the

behavior of thermoelectric coolers [3–7]. Zhao and Tan [3] presented a

study of material, modeling, and application of thermoelectric coolers.

Fraisse et al. [4] compared diﬀerent methods of modeling TECs. Also,

Mani [5] studied the behavior of thermoelectric coolers numerically

and analytically and revealed that the results of these two approaches

are in good agreement.

The coeﬃcient of performance is among the most important topics

related to thermoelectric coolers. For this purpose, Enescu and Virjoghe

[6] provided a review of thermoelectric cooling parameters and per-

formance. In addition, Xuan [7] investigated the eﬀect of thermal and

contact resistance of thermoelectric coolers. Based on his studies, the

amount of COP depends on thermoelectric length. Also by increasing

the thermal contact resistance, this dependence increases signiﬁcantly.

The maximum COP of a TEC device in both cooling and heating

https://doi.org/10.1016/j.enconman.2018.08.045

Received 25 April 2018; Received in revised form 2 August 2018; Accepted 12 August 2018

⁎

Corresponding author.

E-mail address: meslami@shirazu.ac.ir (M. Eslami).

Energy Conversion and Management 174 (2018) 417–429

0196-8904/ © 2018 Elsevier Ltd. All rights reserved.

T

mode is one of the important issues that should be considered. Cosnier

et al. [8] examined the performance of thermoelectric coolers by ex-

perimental and numerical analysis and revealed that it is possible to

reach the coeﬃcient of performances above 1.5 for cooling mode, and 2

for heating mode. Also, Liu et al. [9] used thermoelectric coolers for

various air conditioning applications and showed that it is possible to

reach the COP of 2.59 for cooling mode and 3.01 for heating mode.

These results suggest that TEC devices can be a good choice for water

harvesting if they are used eﬃciently.

Reducing the hot side temperature of a thermoelectric cooler is an

approach to increase the coeﬃcient of performance. For example,

Sadighi Dizaji et al. [10] used water ﬂow for cooling the hot side of a

TEC instead of air and showed that it is possible to increase the cold side

performance of TEC signiﬁcantly. Seo et al. [11] studied the eﬀect of

diﬀerent heat sink's shapes on the performance of TECs, numerically

and showed the shape of heat sinks can change the operating perfor-

mance of thermoelectric coolers. Also Via'n and Astrain [12] designed a

heat sink for the cold side of a TEC and showed that by using this heat

sink, COP can increase up to 32%. In addition, Zhu et al. [13] studied

the eﬀect of diﬀerent heat exchanger sizes on the performance of TECs

theoretically. According to their studies, the highest amount of COP is

achieved by using the optimal heat sink size.

Another important parameter that signiﬁcantly aﬀects the perfor-

mance of a TEC is the electrical current. Tan et al. [14] applied the

second law of thermodynamics and showed that the amount of current

must be precisely determined to achieve the optimal cold side tem-

perature. Also Tan and Fok [15] presented an approach to analyze and

optimize a thermoelectric cooling system.

The application of TECs in water harvesting from air is reported in

several experimental studies [16–24]. Vian et al. [16] designed a device

which was able to condense 0.969 L of water from the air in each day.

Furthermore, Jradi et al. [17] theoretically and experimentally studied

a system including 5 channels with 20 thermoelectric coolers in each

powered by solar cells. This device is combined with a solar distiller

humidifying ambient air to increase distillate output of water produc-

tion. They showed that it is possible to produce 10 L of water during a

summer day in Beirut. In another study, Yao et al. [18] produced

33.1 g/h of water by using a dehumidiﬁcation device having more heat

sinks on the two sides of thermoelectric coolers. In addition, Atta [19]

designed a prototype including three TEC elements and a photovoltaic

cell. He applied this system in Yanbu climate conditions and could

produce almost 1 Liter of condensed water per hour. Besides, Joshi

et al. [20] installed 10 TEC in a channel with the length of 70 cm and

tested it in several diﬀerent climate conditions. Based on this design,

they harvested 240 ml of water in 10 h at a relative humidity of 90%

and mass ﬂow rate of 25 g/s. Tan and Fok [21] designed an AWG

system and investigated the eﬀect of input power to TECs and inlet

mass ﬂow rate on the amount of produced water. They revealed that it

is possible to produce 50 ml of water in 3 h in an average relative hu-

midity of 77%. Also, Liu et al. [22] built a portable water generator

with two thermoelectric coolers and investigated the eﬀect of inlet air

relative humidity and air ﬂow rates and showed that the maximum

amount of generated water is 25.1 g per hour with 58.2 W input power.

Munoz-Garcia et al. [23] designed a similar system for irrigation of

young trees. Based on this design, they could harvest 35 ml water per

hour from the air. Moreover, Pontious et al. [24] could harvest 0.21 L of

water in a day with 0.33 kWh of energy consumption.

Recently, Shourideh et al. [25] performed a theoretical and ex-

perimental analysis of a Peltier AWG by optimizing the cold side ex-

tended surface and the cooling system. But they didn't investigate the

Nomenclatures

Ac

cross section area of a ﬁn(

m)

2

Afﬁns area (m)

2

A

t

total area (m)

2

C

p

speciﬁc heat (kJ/kg K)

COP coeﬃcient of performance

D diameter (m)

Eﬀeﬀectiveness (L/J)

ffraction factor

henthalpy (J/kg)

h

conv

convection heat transfer coeﬃcient (W/m

2

K)

H height of each channel’s hole (m)

Icurrent (A)

kfthermal conductivity of base air ﬂow (W/m K)

ksthermal conductivity of base plate material (W/m K)

K

mTEC module thermal conductance (W/K)

l length (m)

l

c

characteristic length (m)

N number

Nu Nusselt number

pperimeter (m)

Ppower (W)

Pr Prandtl number

Qtransferred heat (W)

Rresistance (K/W)

R″thermal resistance (m2K/W)

R

m

TEC module electrical resistance (ohm)

Re Reynolds number

S

m

TEC module Seebeck coeﬃcient (V/K)

T thickness (m)

Ttemperature (K)

uair velocity

(

m/s

)

Vvoltage (V)

V

̇

volume ﬂow rate of water (L/s)

w width (m)

Greek symbols

η

0

overall surface eﬃciency

η

feﬃciency of ﬁn with an adiabatic tip

ν

kinematic viscosity (m/s

)

2

ΔPpressure drop (Pa)

T

Δ

temperature diﬀerence (K)

ρdensity (kg/m

3

)

ϕ

relative humidity

ω

humidity ratio

Subscripts

a ambient

c cold side

equ equivalent resistance

h hot side

hyd hydraulic

LMTD log mean temperature diﬀerence

max maximum

opt optimum

t

b,

resistance of the un-ﬁnned part of the heat sink

t

bas

e

,

thermal resistance of the base surface

t

c,

contact resistance

resistance of the extended surfaces

TEC, m thermoelectric cooler

w water

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

418

eﬀect of thermal resistances and other parameters like the number of

TECs in the optimization. They also compared the energy consumption

of their design with some other AWG systems and showed that their

system has a better performance. Besides, Salek et al. [26] performed a

thermodynamic optimization for a solar driven ammonia absorption

refrigeration cycle used for air dehumidiﬁcation combined with a saline

water desalination cycle.

The above literature review shows that most of the researches on

thermoelectric AWG systems published so far are experimental and no

comprehensive analytical solution and optimization for water har-

vesting by TECs has been provided. Therefore, this article tries to pre-

sent a complete thermodynamic analysis of water production from

humid air using TECs by considering eﬀects of various design para-

meters, including thermal resistances and ﬁns geometry. The resulting

solution provides the necessary information to ﬁnd the optimum

number of TECs, length of the channel, electrical current and air mass

ﬂow rate. The objective is to maximize the amount of water production

per unit power consumption of the fan and thermoelectric coolers in

diﬀerent operating conditions. Besides, the possibility of water pro-

duction at diﬀerent atmospheric conditions (relative humidity and

temperature) can be predicted. Hence, the idea of using a controller to

turn the device on and oﬀis also investigated to decrease the power

consumption while producing the same amount of water. As case stu-

dies, the annual performance of the device is investigated for three

southern cities in Iran. These locations are typical examples of places

with high relative humidity but very low annual rainfall.

2. System description

As shown in Figs. 1 and 2, the system considered in this article

consists of a number of thermoelectric coolers placed in series. Air ﬂows

through two channels on the hot and cold sides and heat sinks increase

the surface of heat transfer. A fan supplies the required air ﬂow circu-

lation and solar cells power the thermoelectric coolers and the fan.

The temperature distribution on the surface of the channels on both

sides of the thermoelectric cooler is assumed to be uniform. The dis-

tance between two neighboring thermoelectric coolers is considered to

be 1.5 cm to make this assumption reasonable. The entering air stream

ﬁrst passes through the channel on which the cold side of the ther-

moelectric coolers are placed, and after being cooled and dehumidiﬁed,

goes through the warm channel and cools the hot side of thermoelectric

cooler. Fig. 2 shows a schematic representation of the TECs inside

channels.

Also, Table 1 presents the speciﬁcations of the thermoelectric cooler

in this study.

3. Governing equations

Each thermoelectric cooler is identiﬁed by four basic characteristics

including

I

max

,

V

ma

x

,T

Δ

max and

Q

ma

x

. Along with the hot side tempera-

ture of a TEC (

T

h

), the parameters required for modeling thermoelectric

coolers are deﬁned, as follows [10,28]:

=SV

T

mmax

h(1)

=−

R

TTV

TI

(Δ )

mhmaxmax

hmax

(2)

=−

K

TTVI

TT

(Δ )

2Δ

mh max max max

hmax (3)

where

S

m

is the Seebeck coeﬃcient,

R

m

is electrical resistance and

K

mis

thermal conductivity.

By applying energy balance for a thermoelectric cooler, the cooling

power and the heat released from the hot side of the thermoelectric

cooler can be calculated [10,28]:

=−−

Q

SIT IR KT

2Δ

cmc mm

2

(4)

=+−

Q

SIT IR K

T

2Δ

hmh mm

2

(5)

In these equations, Iis the electric current,

T

c

is the cold side tem-

perature of TEC,

T

h

is the hot side temperature of TEC,

Q

c

is the cooling

power,

Qh

is the amount of heat dissipated from the hot side of TEC and

ΔT is the temperature diﬀerence between hot and cold side of ther-

moelectric cooler:

=−TTT

Δ

hc

(6)

On the other hand, the ﬁrst law of thermodynamic gives:

=−

P

QQ

TEC h c (7)

where

P

TEC is thermoelectric power consumption. Combining Eqs.

(4)–(7):

=+

P

SIT IRΔ

TEC m m

2

(8)

Also, the coeﬃcient of performance of the thermoelectric cooler is

deﬁned as:

Fig. 1. A schematic of the system under study.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

419

=Q

P

C

OP

c

TEC (9)

The heat transfer between the air and the TECs in the channels is

clearly related to the enthalpy change as follows:

=−

Q

mh h

̇()

c c ai r in c air out c__ _ _ (10)

=−

Q

mC T T

̇(

)

h h p h air out h air in h___ __ (11)

where C

p

is the speciﬁc heat of air across the hot channel and

h

is the

enthalpy of humid air, in J/kg, which itself is a function of speciﬁc

humidity

ω

[29]:

=−+ + −∗hCT ω T( 273) (2501.3 1.86 ( 273)) 100

0

p

(12)

As no water condensation occurs in the hot channel, the speciﬁc

humidity of the air does not change so the enthalpy changes, Δh, can be

replaced by C

T

Δ

p

to calculate heat dissipated by the air in Eq. (11).In

addition, since the hot side of TEC is cooled by the air discharge from

the cold channel, then

=TT

air in h air out c__ _ _

.

The heat transfer in the channels can also be related to the tem-

perature diﬀerence between the air stream and the hot and cold sur-

faces by LMTD method [30]:

=

Q

T

R

Δ

cLMTD coldside

c

_

(13)

=

Q

T

R

Δ

hLMTD hotside

h

_

(14)

=−−−

−

−

()

TTTTT

Δ

[( ) ( )]

ln

LMTD coldside air out c c a c

TT

TT

___

air out c c

ac

__

(15)

=−−−

−

−

()

TTT TT

Δ

[( ) ( )]

ln

LMTD hotside h air in h h air out h

TT

TT

___ _ _

hairinh

hairouth

__

__ (16)

In Eqs. (13) and (14), parameters

R

c

and

R

h

are the total thermal

resistance between the TEC surface and air ﬂow in the cold and hot

channels respectively. They can be calculated by adding diﬀerent

thermal resistances as shown in Fig. 3:

In Fig. 3,

R

tc,is the contact resistance between thermoelectric

coolers and the heat sink attached to the cold channel. It can be mod-

eled using the following expression:

=′′

R

R

Nlw

tc

tc

TEC TEC TEC

,,

(17)

N

TE

C

is the number of thermoelectric coolers and

l

TEC

and

w

TE

C

are

the length and width of each TEC, respectively.

R

tbas

e

,is the thermal

resistance of the base surface on which thermoelectric coolers are in-

stalled and is calculated as follows:

=

R

t

kl w

tbase base

s channel channel

,

(18)

where

t

bas

e

,

l

channe

l

and

w

channe

l

are illustrated in Figs. 1 and 2 and ksis the

thermal conductivity of base plate material.

The convection heat transfer resistance of the un-ﬁnned part of the

heat sink

R

t

b

,can be combined with resistance of the extended surfaces

R

tf N,(

)

as an equivalent resistance

R

equ. Assuming the lateral surfaces of

the channel to be insulated [30]:

=

R

ηAh

1

equ

tconv

0(19)

In this equation,

η

0

is the overall eﬃciency of the heat sink, given by

[30]:

=− −

η

NA

Aη1(1

)

fin f

tf0(20)

=Awl2

ffinc (21)

=+ll t

2

cfin

fin

(22)

=+−All N N Hl(2 2)

t hole channel hole hole hole channel (23)

where

N

fin,Af,

A

t

,

l

c

, and

N

hol

e

are the number of ﬁns, the ﬁn area, the

total area of heat transfer, equivalent ﬁn length and the number of air

passages, respectively. Also, Hhol

e

and

l

hol

e

are shown in Fig. 1 and lfi

n

,

Fig. 2. The layout of thermoelectric coolers and the air passages over them.

Table 1

Thermoelectric cooler speciﬁcations KRYOTHERM TB-127-2,0-1,05(62)

[27].

Parameter Value Unit

I

ma

x

17.6 A

V

ma

x

15.7 V

Q

ma

x

171.0 W

T

Δ

ma

x

69.0 K

l

TEC 62.0 mm

w

TE

C

62.0 mm

Fig. 3. Thermal resistances between the cold side of TEC and air ﬂow in the channel.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

420

w

fin and

t

fin are illustrated in Fig. 4.

Assuming that the ﬁn tips are adiabatic,

η

fis given by [30]:

=

η

ml

ml

tanh

f

c

c(24)

in which,

=mhp

kA

conv

sc

(25)

=+

p

wt2(

)

fin fin (26)

=Awt

cfinfi

n

(27)

where pis the perimeter and

Ac

is the cross section area of a rectangular

ﬁn.

Also, the convection heat transfer coeﬃcient h, in Eqs. (19) and

(25), is calculated from the Nusselt number:

=hNuk

D

conv

f

hyd (28)

In which, kfis the thermal conductivity for air. Assuming turbulent

ﬂow, for both cooling and heating channels [30]:

=

N

u Re Pr for cooling0.023 0.8 0.3 (29)

=

N

u Re Pr for heating0.023 0.8 0.4 (30)

where

Pr

is the Prandtl number and

Re

is the Reynolds number deﬁned

as follows:

=

R

euD

ν

average hyd

(31)

=

u

mN

ρl H

̇/

average ahole

hole hole

(32)

Also, in Eqs. (28) and (31), the characteristic length is hydraulic

diameter Dhy

d

that is given by:

==

+

DA

p

lH

lH

44

2( )

hyd hole

hole

hole hole

hole hole (33)

Finally, the total resistance is obtained:

=+ +

R

RR R

tc tbase equ,,

(34)

With the same procedure for hot channel,

R

h

is calculated. Besides,

the rate of water production in

L/

s

is obtained using the following mass

balance equation for water:

=−

V

mω ω ρ

̇̇()

1000

waairair

average

incoutc(35)

in which,

ρ

average is the density of water at the temperature of

+TT

2

air out c air in c__ __

.

To calculate the required fan power, pressure drop across the air

passage is required [30]:

=pf

ρu

Dl

Δ

average

hyd

channe

l

2

(36)

=−

−

f

Re(0.790ln( ) 1.64) 2(37)

Dimensions of the channel are chosen to have a turbulent ﬂow for

better heat transfer. Having

pΔ

, the fan power consumption is given by

[30]:

=

P

m

ρ

p

̇Δ

fan (38)

4. Solution procedure

By considering the relative humidity of the outlet air from the cold

channel to be equal to 100% and solving Eqs. (1)–(38), the values of

power for the fan and thermoelectric coolers and also the amount of

produced water are obtained. It should be noted that if the amount of

produced water becomes negative, it means that the assumption of

saturated air at the outlet is incorrect. In this case, instead of using this

assumption, the condition =

ω

ω

air air

incoutcis used, in order to obtain

correct results. Fig. 5 shows a ﬂow chart of diﬀerent steps in the solu-

tion.

The geometrical and thermophysical properties of the present

system are given in Table 2.

4.1. Validation

At ﬁrst, the performance of the TEC modeled using Eqs. (1)–(9) is

compared with results of the software provided by the thermoelectric

manufacturer (KRYOTHERM) [27]. For

T

h

= 300 K, and T

Δ

= 20 K and

40 K, changes of COP and

Q

c

relative to the electric current is presented

in Figs. 6 and 7. These results approve the accuracy of analytical

equations used for modeling the thermoelectric cooler.

In the next step, the present model is used to simulate the experi-

ments of Jradi et al. [17].Fig. 8 illustrates the amount of water pro-

duced at diﬀerent air ﬂow rates for the electrical current of 2.6 A. Re-

sults show the maximum error of 6% compared to the experimental

data and simulation results of Jradi et al. [17]. Also, eﬀects of changing

the electrical current for the ﬁxed air ﬂow rate of 0.0155 kg/s is shown

in Fig. 9 which validates the accuracy of the present simulations.

Fig. 4. A schematic of ﬁn sizing.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

421

5. Results and discussion

5.1. Parametric study and optimization

In this section, the eﬀect of diﬀerent parameters on performance of

the AWG system is studied. In order to optimize the design, an objective

function called eﬀectiveness is deﬁned as follows:

=+

E

ff m

PP

̇w

TEC fan (39)

This indicator represents the amount of produced water relative to

the energy consumption of the system. Higher values of this quantity

mean that the system consumes less energy to produce the same

amount of water. This is an important feature for the systems supplied

by solar energy. A system with higher eﬀectiveness needs smaller PV

and battery system.

A sensitivity analysis can provide the path for optimization. For

instance, Fig. 10 shows the eﬀect of changing electrical current on ef-

fectiveness of the system at air ﬂow rate of 0.015 kg/s, ambient tem-

perature of 308 K, relative humidity of 75% and 10 TEC modules placed

in a 77 cm long channel. It is observed that by increasing the current,

Eﬀﬁrst rises and then falls. Therefore, it is very important to supply an

Fig. 5. The solution procedure.

Table 2

Fixed parameters needed to solve the problem.

Parameter Value Unit

l

hole 4mm

H

hol

e

30 mm

N

hol

e

8–

t

bas

e

1mm

t

fin 3mm

N

fin

7–

k

f

c

_

0.0257 W/m K

k

fh_

0.0271 W/m K

k

s229 W/m K

ν

c

15.11E−06 m

2

/s

νh16.97E−06 m

2

/s

Pr

c0.713 –

Pr

h

0.703 –

R″0.001 m

2

K/W

Fig. 6. Change of COP vs Current for =

Δ

T20

K

and =

Δ

T40

K

for the con-

sidered TEC (KRYOTHERM) [27].

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

422

optimum amount of current to the system. However, the optimum value

of the current is a function of operating conditions and also the number

of thermoelectric coolers.

To ﬁnd the optimum number of thermoelectric coolers, 15–20 TECs

are considered. The length of the channel is proportional to the number

of thermoelectric coolers. The ambient condition is assumed to be 308 K

with relative humidity of 75%. By changing the electrical currents from

zero to I

max

for ṁ

a

between 0.01 and 0.02 kg/s, the amount of

E

ff

ma

x

is

calculated for all cases. Results reported in Table 3 reveal that the

maximum eﬀectiveness occurs at COP = 1.9 in all cases. The system

with 18 thermoelectric coolers and inlet air ﬂow rate of 0.0117 kg/s has

the maximum value of Eﬀwhich is equal to 1.638E−07 L/J. Therefore,

the optimum length of the channel is 1.386 m with 18 TECs placed in

series. This design is used from now on for the study of other para-

meters.

Eﬀects of changing the electricity current for a channel with the

optimum length is shown in Fig. 11. The air ﬂowrate of 0.0117 kg/s, the

ambient temperature of 308 K and the relative humidity of 75% are

considered. Results reveal that the maximum amount of water pro-

duction is 42 ml in one hour and it occurs when I= 2.315 A. However,

the maximum Eﬀdoesn't occur at this point. In fact,

=E

ff

max

1.638E−07L/J is 1.65 times higher than the Eﬀat maximum water

production and occurs at I = 1.349 A.

Clearly, the amount of water production is proportional to the re-

lative humidity of the ambient air. Fig. 12 projects this eﬀect for the air

ﬂow rate of 0.0117 kg/s and I = 1.349 A. Therefore, it is possible to

produce 106 ml of fresh water at high relative humidity within one

hour.

To ﬁnd the optimal performance of the system at diﬀerent ambient

temperatures, the relative humidity of the entering air is considered to

be constant and equal to 75%, while the input current to thermoelectric

coolers is varied from 0 to I

max

, also air ﬂow rate changes between

0.001 and 0.02 kg/s. The optimum operating mode of the system is

calculated for each temperature. These results are reported in Table 4.

Based on the above results:

(1) At the optimal performance, with the increase in entering air

temperature, the input electric current to the system decreases.

Because when entering air temperature goes higher, the perfor-

mance range of thermoelectric becomes more limited. If the current

exceeds a speciﬁc value, the warm side of thermoelectric cooler gets

too hot which has a negative eﬀect on its performance. So, for an

increase in the entering air temperature, the input current to the

system should be decreased.

(2) With the increase in air temperature, the optimal air ﬂow rate is

Fig. 7. Change of

Qc

vs Current for =

Δ

T20

K

and =

Δ

T40

K

for considered TEC

(KRYOTHERM) [27].

Fig. 8. The amount of produced water as a function of air ﬂow.

Fig. 9. The amount of produced water as a function of electrical current.

Fig. 10. The eﬀect of increasing electric current on the value of eﬀectiveness.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

423

decreased as well, which is justiﬁed with the previously discussed

observation: by decreasing input current to the system, and there-

fore decreasing input power to thermoelectric cooler, the generated

cooling power also decreases; thus the optimal air ﬂow rate to the

system decreases as well, since the cooling power won’tbesuﬃ-

cient for higher amounts of ﬂow rate

(3) By increasing the air temperature,

E

ff

ma

x

and the COP at which

E

ff

ma

x

occurs increase. This is because, with the increase of entering

air temperature, the power consumption of thermoelectric cooler

also decreases. In addition, the optimum air ﬂow rate and fan

power consumption decrease, so the total amount of power con-

sumption decreases, and since warmer air contains more moisture

(at the same relative humidity), water production is easier.

Therefore, with the increase in temperature,

E

ff

ma

x

and COP also

increase.

Figs. 13–21 indicate the eﬀect of changing current on water con-

densation, Eﬀ,

T

h

,

T

c

,T

Δ

,

P

TEC,

Qh

,

Q

c

and

C

O

P

at the optimum air ﬂow

rate of each temperature. The eﬀect of current on water production of

the system with optimum air ﬂow rate at each temperature is reported

in Fig. 13 for the relative humidity of 75%. Based on this ﬁgure, the

maximum amount of water produced in an hour occurs at 313 K and is

equal to 44 ml.

Fig. 14 is also presented to investigate eﬀects of the change in

current on the amount of eﬀectiveness. Comparing with Fig. 13,itis

clear that the maximum amount of Eﬀdoesn't occur at the current

which maximum amount of water is harvested for all cases. It is in-

teresting to note that 26 ml of water can be harvested from the air with

75% relative humidity and 318 K by using only 20 W electrical power

within 1 h.

Fig. 15 indicates that by increasing the electrical current,

T

h

always

rises. Also at a ﬁxed current, by increasing the inlet air temperature, the

hot side temperature rises and since it's a limiting factor of the system,

Table 3

Eﬀect of change of number of TECs and length of channel on optimization of the device.

NO TEC I

channel

(m) Eﬀ

max

(L/J) Current (A) COP ṁa(kg/s)

Re

h

Re

c

15 1.155 1.587E−07 1.396 1.9 0.0109 4689 4175

16 1.232 1.615E−07 1.375 1.9 0.0111 4776 4252

17 1.309 1.632E−07 1.369 1.9 0.0115 4950 4408

18 1.386 1.638E−07 1.349 1.9 0.0117 5038 4485

19 1.463 1.637E−07 1.344 1.9 0.0120 5212 4641

20 1.54 1.629E−07 1.336 1.9 0.0124 5343 4757

Fig. 11. The amount of produced water (in ml/h) as a function of the electric

current, at 308 K and 90% relative humidity and air ﬂow rate of 0.0117 kg/s.

Fig. 12. The amount of produced water (in ml/h) as a function of relative

humidity, air ﬂow rate of 0.0117 kg/s and I = 1.349 A.

Table 4

System optimization results at 75% relative humidity and diﬀerent inlet tem-

peratures.

T

a

(K)

ṁ

a optimu

m

_

(kg/s) Current (A) COP Eﬀ

max

(L/J)

298 0.0135 1.606 1.7 1.102E−07

303 0.0127 1.485 1.8 1.299E−07

308 0.0117 1.349 1.9 1.638E−07

313 0.0105 1.081 2.4 2.284E−07

318 0.0084 0.758 3.2 3.684E−07 Fig. 13. Produced water (in ml/h) as a function of the electric current, at 75%

relative humidity for optimum air ﬂow rate at diﬀerent temperatures.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

424

one should be careful about the hot side temperature of TEC in order to

prevent damage.

As shown in Fig. 16, by increasing electric current,

T

c

ﬁrst falls then

rises slightly. The reason is that by increasing current,

T

h

increases. At

ﬁrst, the optimal ﬂow rate is able to cool the hot side suﬃciently and so

T

h

has a negligible eﬀect on the cold side temperature and

T

c

decreases

as expected. But after a certain current, this optimal ﬂow rate is not able

to cool the hot side enough and an increase in the hot side is observed

which also results in an increase in

T

c

. However, the range of the change

in cold side temperature is not wide, unlike the hot side temperature.

The temperature diﬀerence between the hot and cold side ( T

Δ

) has

an ascending trend with increasing current as shown in Fig. 17. This

behavior is consistent with the previously mentioned change of

T

h

and

T

c

with the current. It should be noted that the air inlet temperature has

little eﬀect on optimum T

Δ

.

Fig. 18 indicates that the input power to thermoelectric cooler is

approximately the same for all inlet air temperatures and only the

performance range of each case is diﬀerent; it shows that optimum

mode of the system occurs at a same thermoelectric cooler's power for

diﬀerent ambient temperatures.

At a ﬁxed current, the rate of heat removal from the hot side of

thermoelectric cooler

Qh

decreases by increasing inlet air temperature

as illustrated in Fig. 19. Because with the increase of inlet air tem-

perature, the optimum inlet air ﬂow rate decreases, so the rate of heat

removal from the hot side of thermoelectric cooler decreases.

Also, Fig. 20 demonstrates that the rate of heat transfer from the

cold side has an ascending trend at ﬁrst; and at a speciﬁc current, the

maximum cooling eﬀect occurs. On the other hand,

Q

c

decreases with

increasing inlet air temperature. The reason is that

Q

c

depends on

Qh

and thermoelectric power and as it was explained before, the optimum

amount of input power to TECs is the same in all cases and

Qh

decreases

with increasing inlet air temperature at a ﬁxed current. So

Q

c

must be

reduced with warmer inlet air at a ﬁxed current.

Finally, Fig. 21 shows that at a ﬁxed current, COP falls with the

increase of inlet air temperature. The reason is that

Q

c

decreases while

the power consumption

P

doesn’t change signiﬁcantly with increasing

inlet air temperature.

Fig. 14. Eﬀas a function of the electric current, at 75% relative humidity for

optimum air ﬂow rate of system at diﬀerent temperatures.

Fig. 15. Hot side temperature of TECs as a function of the electric current, at

75% relative humidity for optimum air ﬂow rate at diﬀerent temperatures.

Fig. 16. Cold side temperature of TECs as a function of the electric current, at

75% relative humidity for optimum air ﬂow rate of system at diﬀerent tem-

peratures.

Fig. 17. Changes in temperature diﬀerence between hot and cold side of TECs

as a function of the electric current, at 75% relative humidity for optimum air

ﬂow rate of system at diﬀerent temperatures.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

425

5.2. Annual performance in three southern cities of Iran

In this section, the annual amount of water condensation and energy

consumption is calculated for three southern cities of Iran. All these

three cities are faced with the problem of water scarcity. Also, they are

proper candidates to evaluate the proposed system, since they are in the

vicinity of the sea and therefore, have high values of relative humidity.

For this purpose, hourly weather data for Kish island, Bandar-e-Abbas

and Bandar-e-chabahar are obtained via Iran meteorological organiza-

tion [31] for the duration of one year.

At ﬁrst, it is assumed that the system is always on, with a constant

electrical current and air ﬂow rate supplied. Although the optimum

values of these two controlling parameters were found in the previous

section for diﬀerent inlet temperatures, the ambient condition is vari-

able during a year. Therefore, now the question is that what electrical

current and air ﬂow rate results in the best yearly performance?

To ﬁnd the answer, the eﬀect of changing electrical current on

yearly eﬀectiveness is demonstrated in Fig. 22 for Kish island. The

optimum electrical current at each ﬂow rate is lower than what was

found in the previous section for the corresponding ﬁxed inlet

temperature. Again, higher annual Eﬀcan be achieved by reducing the

air ﬂow rate. However, less water is produced if a lower air ﬂow rate is

supplied by the fan as illustrated in Fig. 23. It shows the eﬀect of cur-

rent on water condensation during 1 year for diﬀerent air ﬂow rates.

The interesting point is that the optimum current for maximum pro-

duction in a year is almost the same as the value found in the previous

section for a ﬁxed inlet temperature. Therefore, the variable ambient

condition has a little eﬀect on the optimum current at the desired ﬂow

rate for maximum production.

Taking a closer look at the results, it is found that the system with

constant electrical current is not capable of producing water in almost

half of the year. Therefore, a wise decision is to turn oﬀthe system in

these times for energy saving purposes. In fact, a fundamental ad-

vantage of the present analysis is that it can predict the possibility of

producing water for diﬀerent values of relative humidity and tem-

perature. If a controller is used to turn the system oﬀand on, then the

same amount of water is gained with higher eﬀectiveness.

To study the behavior of the on/oﬀsystem, water condensation,

power consumption (kWh) and eﬀectiveness is calculated during one

Fig. 18. Input power to TECs as a function of the electric current, at 75% re-

lative humidity for optimum air ﬂow at diﬀerent temperatures.

Fig. 19. Heat removal from hot side of TECs as a function of the electric cur-

rent, at 75% relative humidity for optimum air ﬂow rate of system at diﬀerent

temperatures.

Fig. 20. Heat removal from the cold side of TECs as a function of the electric

current, at 75% relative humidity for optimum air ﬂow rate of system at dif-

ferent temperatures.

Fig. 21. Changes in coeﬃcient of performance of TECs as a function of the

electric current, at 75% relative humidity for optimum air ﬂow rate of system at

diﬀerent temperatures.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

426

year for Kish island at diﬀerent electrical currents and air ﬂow rates.

Fig. 24 shows that the eﬀectiveness of the on/oﬀsystem falls as the

supplied electrical current increases. Unlike the always-on case, one

cannot ﬁnd an optimum current for maximum eﬀectiveness. The reason

is that during a year, there are times that the ambient relative humidity

is very high. In these periods, it is possible to produce water with very

small values of electrical current. Such a system is shut down most of

the time and produces a small amount of water very eﬃciently.

Therefore, the objective function for optimization of the yearly

performance of the on/oﬀsystem can be the total amount of water

production which is the same as the maximum value found in Fig. 23.

As expected, more water can be harvested with higher air ﬂow rates,

although with lower eﬀectiveness. Table 5 summarizes the annual

performance of the system. It is interesting that the on/oﬀsystem can

produce larger amounts of fresh water than the always-on device with

approximately the same eﬀectiveness. Also, the same amount of water

can be harvested much more eﬃciently if the system can be switched

oﬀwhile there is no water production.

Similarly, the amount of water production, power consumption and

eﬀectiveness for two other cities of Bandar-e-Abbas and Bandar-e-cha-

bahr are investigated for

=ṁ0. 0117 kg/

s

a

and I = 0.9 A and I = 2.1 A.

Results are compared with Kish islands in Figs. 25 and 26.

According to Fig. 25, power consumption for all three cities is ap-

proximately the same if the system is always on. However, the amount

of water condensation in Bandar-e-chabahar is much higher and

thereby the highest Eﬀamong the three cities occurs for this location

which is equal to 0.533 L/kWh. Also, when the system is on/oﬀ, the

highest Eﬀis for Kish island. The reason is that the electrical current of

I = 0.9 A is the optimum value obtained based on the annual weather

data of this city. Also, Fig. 26 shows that the maximum amount of water

condensation during 1 year (312 L) is produced in Bandar-e-chabahar

which has higher relative humidity compared to two other cities.

5.3. Comparison between the present design and other AWG systems

Finally, the speciﬁc energy consumption (which is the energy con-

sumption (kWh) for condensation of 1 m

3

of water) for diﬀerent AWG

systems proposed in the literature is compared in Table 6.

It is found that the present design is the most energy eﬃcient system

among similar devices proposed in the literature. The reason is that

diﬀerent operating parameters are considered simultaneously in the

present optimization. However, the eﬀectiveness of the present system

is not competitive at lower values of relative humidity. Because the

optimizations are performed for 75% relative humidity, if the system is

intended to work at lower relative humidity, one should optimize the

system again to have better performance. Also, it is important to note

that the ambient relative humidity greatly aﬀects the speciﬁc energy

consumption. For example, a 5% increase in relative humidity can de-

crease the speciﬁc energy consumption by about 50%. Also by com-

parison between present work and the commercial system using a vapor

compression refrigeration cooling [32], it can be seen that in high re-

lative humidity, the present design is even marginally performing

better.

6. Conclusion

In this study, thermodynamic analysis and optimization of a device

for water harvesting from humid air is presented. Air cooling is

achieved by a number of thermoelectric coolers. All of the necessary

geometrical, meteorological, technical and thermophysical parameters

are taken into account to have an accurate estimate of the actual system

behavior, including thermal resistances and ﬁns geometry.

Optimizations were performed to obtain fresh water with minimum

possible energy consumption. According to the results, 18 TECs in a

channel with the length of 1.386 m has the best Eﬀamong the cases

investigated. Optimal electrical current and air ﬂow rate were

Fig. 22. Eﬀ(in L/kWh) as a function of the electric current, at diﬀerent mass

ﬂow rate for Kish island during 1 year; The system is always on.

Fig. 23. Water condensation as a function of the electric current, at diﬀerent

mass ﬂow rate for Kish island during 1 year.

Fig. 24. Eﬀas a function of the electric current, at diﬀerent mass ﬂow rate for

Kish island during 1 year. The system is on/oﬀ.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

427

calculated for diﬀerent inlet air temperature. It was found that these

two controlling parameters should be decreased at higher air tem-

peratures. The system is capable of harvesting 26 ml of water from the

air with 75% relative humidity and 318 K by using only 20 W power

within 1 h.

Also, the annual performance of this device in three southern cities

in Iran (Kish island, Bandar-e-Abbas and Bandar-e-chabahar) were in-

vestigated for two modes of operation: the always-on and on/oﬀ

Table 5

Annual performance of the system in Kish island for diﬀerent air mass ﬂow rates.

The system is always on The system is on/oﬀ

ṁa(kg/s) ṁa(kg/s)

0.0135 0.0127 0.0117 0.0105 0.0084 0.0135 0.0127 0.0117 0.0105 0.0084

Maximum water condensation (L) 265.4 241.2 209.7 173.8 112.4 265.4 241.2 209.7 173.8 112.4

Water condensation at Eﬀ

max

(L) 153.59 147.57 98.03 91.12 46.98 n/a n/a n/a n/a n/a

Iat maximum water condensation (A) 2.4 2.2 2.1 1.9 1.6 2.4 2.2 2.1 1.9 1.6

Iat Eﬀ

max

(A) 1.2 1.2 0.9 0.9 0.6 0 0 0 0 0

Power consumption at maximum water

condensation (kWh)

1289.91 1086.61 1020.02 793.75 560.83 784.94 652.02 572.72 450.95 281.77

Power consumption at Eﬀ

max

(kWh) 431.05 409.40 260.52 235.37 114.01 0 0 0 0 0

Power consumption at Eﬀ

max

of system when is

always on (kWh)

The same as Power consumption at Eﬀ

max

(kWh) for each air ﬂow rate 201.21 192.97 99.35 92.07 33.61

Eﬀat Maximum water condensation (L/kWh) 0.206 0.222 0.206 0.219 0.200 0.338 0.370 0.366 0.385 0.400

Eﬀ

max

(L/kWh) 0.356 0.360 0.376 0.387 0.412 ∞∞∞∞∞

Eﬀfor Iat

E

ff

max

of the always on (L/kWh) The same as Eﬀ

max

for each air ﬂow rate 0.763 0.765 0.987 0.990 1.400

Fig. 25. Annual water production, power consumption and eﬀectiveness for three locations for duration of 1 year at air ﬂow rate of 0.0117 kg/s and I = 0.9 A.

Fig. 26. Annual water production, power consumption and eﬀectiveness for three locations for duration of 1 year at air ﬂow rate of 0.0117 kg/s and I = 2.1 A.

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

428

system. Results reveal that there is no optimum electrical current for

the eﬀectiveness of the on/oﬀsystem. Among the three locations con-

sidered, Bandar-e-chabahar has the highest yearly yield because of

higher ambient air relative humidity. In the end, a comparison between

this device and other air water generators indicates that present design

is the most energy eﬃcient system among similar devices; especially in

high relative humidity.

References

[1] Claire S. The Last Drop. Mazaya, Summer Issue; 2002. p. 22–5.

[2] Mekonnen MM, Hoekstra AY. Four billion people facing severe water scarcity. Sci

Adv 2016;2(2):e1500323.

[3] Zhao D, Tan G. A review of thermoelectric cooling: materials, modeling and ap-

plications. Appl Therm Eng 2014;66(1–2):15–24.

[4] Fraisse G, Ramousse J, Sgorlon D, Goupil C. Comparison of diﬀerent modeling

approaches for thermoelectric elements. Energy Convers Manage 2013;65:351–6.

[5] Mani PI. Design, modeling and simulation of a thermoelectric cooling system (TEC);

2016.

[6] Enescu D, Virjoghe EO. A review on thermoelectric cooling parameters and per-

formance. Renew Sustain Energy Rev 2014;38:903–16.

[7] Xuan XC. Investigation of thermal contact eﬀect on thermoelectric coolers. Energy

Convers Manage 2003;44(3):399–410.

[8] Cosnier M, Fraisse G, Luo L. An experimental and numerical study of a thermo-

electric air-cooling and air-heating system. Int J Refrig 2008;31(6):1051–62.

[9] Liu ZB, Zhang L, Gong G, Luo Y, Meng F. Experimental study and performance

analysis of a solar thermoelectric air conditioner with hot water supply. Energy

Build 2015;86:619–25.

[10] Dizaji HS, Jafarmadar S, Khalilarya S, Moosavi A. An exhaustive experimental study

of a novel air-water based thermoelectric cooling unit. Appl Energy

2016;181:357–66.

[11] Seo YM, Ha MY, Park SH, Lee GH, Kim YS, Park YG. A numerical study on the

performance of the thermoelectric module with diﬀerent heat sink shapes. Appl

Therm Eng 2018;128:1082–94.

[12] Vian JG, Astrain D. Development of a heat exchanger for the cold side of a ther-

moelectric module. Appl Therm Eng 2008;28(11–12):1514–21.

[13] Zhu L, Tan H, Yu J. Analysis on optimal heat exchanger size of thermoelectric cooler

for electronic cooling applications. Energy Convers Manage 2013;76:685–90.

[14] Tan H, Fu H, Yu J. Evaluating optimal cooling temperature of a single-stage ther-

moelectric cooler using thermodynamic second law. Appl Therm Eng

2017;123:845–51.

[15] Tan FL, Fok SC. Methodology on sizing and selecting thermoelectric cooler from

diﬀerent TEC manufacturers in cooling system design. Energy Convers Manage

2008;49(6):1715–23.

[16] Vián JG, Astrain D, Domınguez M. Numerical modelling and a design of a ther-

moelectric dehumidiﬁer. Appl Therm Eng 2002;22(4):407–22.

[17] Jradi M, Ghaddar N, Ghali K. Experimental and theoretical study of an integrated

thermoelectric–photovoltaic system for air dehumidiﬁcation and fresh water pro-

duction. Int J Energy Res 2012;36(9):963–74.

[18] Yao Y, Sun Y, Sun D, Sang C, Sun M, Shen L, et al. Optimization design and ex-

perimental study of thermoelectric dehumidiﬁer. Appl Therm Eng 2017;123:820–9.

[19] Atta RM. Solar water condensation using thermoelectric coolers. Int J Water Resour

Arid Environ 2011;1(2):142–5.

[20] Joshi VP, Joshi VS, Kothari HA, Mahajan MD, Chaudhari MB, Sant KD.

Experimental investigations on a portable fresh water generator using a thermo-

electric cooler. Energy Proc 2017;109:161–6.

[21] Tan FL, Fok SC. Experimental testing and evaluation of parameters on the extrac-

tion of water from air using thermoelectric coolers. J Test Eval 2012;41(1):96–103.

[22] Liu S, He W, Hu D, Lv S, Chen D, Wu X, et al. Experimental analysis of a portable

atmospheric water generator by thermoelectric cooling method. Energy Proc

2017;142:1609–14.

[23] Muñoz-García MA, Moreda GP, Raga-Arroyo MP, Marín-González O. Water har-

vesting for young trees using Peltier modules powered by photovoltaic solar energy.

Comput Electron Agric 2013;93:60–7.

[24] Pontious K, Weidner B, Guerin N, Pierrakos O, Altaii K. Design of an atmospheric

water generator: Harvesting water out of thin air. In: Systems and information

engineering design symposium (SIEDS); 2016. p. 6–11.

[25] Shourideh AH, Ajram WB, Al Lami J, Haggag S, Mansouri A. A comprehensive study

of an atmospheric water generator using Peltier eﬀect. Therm Sci Eng Prog

2018;6:14–26.

[26] Salek F, Moghaddam AN, Naserian MM. Thermodynamic analysis and improvement

of a novel solar driven atmospheric water generator. Energy Convers Manage

2018;161:104–11.

[27] Informational booklet. Kryotherm, http://kryothermtec.com/catalogs.html/; 2017

[accessed 2 July 2017].

[28] Liu Z, Zhang L, Gong G. Experimental evaluation of a solar thermoelectric cooled

ceiling combined with displacement ventilation system. Energy Convers Manage

2014;87:559–65.

[29] Cengel YA, Boles MA. Thermodynamics: an engineering approach. 8th ed. New

York: McGraw-Hill; 2015.

[30] Bergman TL, Incropera FP, DeWitt DP, Lavine AS. Fundamentals of heat and mass

transfer. 7th ed. New York: John Wiley & Sons; 2011.

[31] IRIMO REPORT Server; 2018. http://reports.irimo.ir/jasperserver/login.html/

[accessed 15 March 2018].

[32] Ambient Water Systems; 2018. http://www.ambientwater.com/en/systems/ [ac-

cessed 28 March 2018].

[33] EcoloBlue Water from Air; 2018. http://ecoloblue.com/ecoloblue-1000/47-

ecoloblue-1000.html/[accessed 28 March 2018].

Table 6

Speciﬁc energy consumption (in kWh/

m

3

) for the present work and other AWG systems.

System Speciﬁc energy consumption (kWh/

m

3

)

T

a

(K) Relative humidity % Comments

Present work 252.78 303 90 –

Present work 489.95 306 85 –

Present work 911.57 306 80 –

Present work 769.23 318 75 –

Present work 22336.38 318 60 –

Present work 1013.47 Kish island climate Kish island climate The system is on/oﬀ

Present work 1876.95 Bandar-e-chabahar

climate

Bandar-e-chabahar

climate

The system is always on

Tan and Fok [21] 7294.11 302 79 –

Shanshan Liu et al. [22] 2318.72 296.6 92.7 –

Pontious et al. [24] (Peltier) 1571.43 ––Exact data for climate condition is not available

Shourideh et al. [25] 2002.00 303 60 At the ﬁrst hour of test (unsteady)

Shourideh et al. [25] 1870.00 303 60 At the second hour of test (steady)

Shourideh et al. [25] 936.00 306 80 At the ﬁrst hour of test (unsteady)

Shourideh et al. [25] 922.00 306 80 At the second hour of test (steady)

Commercial AWG system [32] 257.18 303 90 Ambient water group (vapor compression

refrigeration)

Commercial AWG system [33] 410 303 80 EcoloBlue group (vapor compression

refrigeration)

M. Eslami et al. Energy Conversion and Management 174 (2018) 417–429

429