Applied Energy 219 (2018) 240–255

Contents lists available at ScienceDirect

Applied Energy
journal homepage: www.elsevier.com/locate/apenergy

Mechanical compressor-driven thermochemical storage for cooling T

applications in tropical insular regions. Concept and efficiency analysis
⁎
Franco Ferruccia, , Driss Stitoub, Pascal Ortegaa, Franck Lucasa
a
GEPASUD Laboratory, University of French Polynesia, BP 6570, 98702 Tahiti, French Polynesia
b
CNRS-PROMES Laboratory, Tecnosud, Rambla de la thermodynamique, 66100 Perpignan, France

H I GH L IG H T S

• Combination of a thermochemical reactor and a mechanical vapor compression unit.

• The system can provide cooling and energy storage in the context of smart microgrids.
• Thermochemical method presents no self-discharge and long charge/discharge life cycle.
• The compressor aids the desorption and compensates for a low heat source temperature.
• With a waste heat source it attains a cooling capacity of 4 kWh /day/m of PV panel.
cold
2

A R T I C LE I N FO A B S T R A C T

Keywords: The energy situation in tropical insular regions, as is found in French Polynesia, presents a number of challenges,
Solar cooling including heavy dependence on imported fuel, high transport costs from the mainland and weak electricity grids.
Solid/gas sorption By contrast, these regions possess a variety of renewable energy resources, which are favorable for the ex-
Thermochemical ploitation of smart micro grids and energy storage technologies. With regards to electrical energy demand, the
Chemical adsorption
high temperatures commonly seen in these regions throughout the entire year implies that a large proportion of
Tropical and insular regions
electricity consumption (∼40%) is used for space cooling, even during evening hours. Framed within this
Microgrid
context, this paper presents an air conditioning system driven by photovoltaic electricity that combines a me-
chanical vapor refrigeration system and a thermochemical storage unit. Thermochemical processes enable the
storage of energy in the form of chemical potential with virtually no losses, which can be used to produce cold
during the evening hours without running a compressor. These processes are implemented using thermo-
chemical reactors, in which a reversible chemical reaction between a solid compound and a gas takes place. The
solid/gas pair used in this study is barium chloride salt (BaCl2) reacting with ammonia (NH3), which is also the
coolant fluid in the refrigeration circuit.
In the proposed system, the photovoltaic-driven electric compressor is used during the day either to run the
refrigeration circuit when a cooling demand occurs or to decompose the ammonia-charged salt and to remove
the gas from the thermochemical reactor when there is no cooling requirement. During the evening, when there
is no electricity from solar sources, the system changes its configuration and the reactor reabsorbs the ammonia
gas from the evaporator to produce cold. The efficiency of this hybrid system is evaluated in this work and
compared with alternative processes which utilize either electrochemical (lead-acid, lithium-ion batteries) or
thermal storage (ice, chilled water) for cold production.

1. Introduction
The energy situation in tropical insular regions is complex and
presents both challenges and opportunities. In such environments, there
is a strong dependence on foreign energy, mainly fossil fuels.
Furthermore, these regions often present weak or poorly interconnected
electricity grids, which make them prone to frequency and voltage
perturbation [1]. These regions typically exhibit climate patterns of
year-round high temperatures and humidity. Additional challenges in
certain tropical insular regions include geographic remoteness, high
transport costs from the mainland, low population density and the fact
that large-scale energy projects are not typically considered viable.

⁎
Corresponding author.
E-mail address: franco.ferrucci@doctorant.upf.pf (F. Ferrucci).

https://doi.org/10.1016/j.apenergy.2018.03.049
Received 31 December 2017; Received in revised form 7 March 2018; Accepted 18 March 2018
Available online 24 March 2018
0306-2619/ © 2018 Elsevier Ltd. All rights reserved.
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Nomenclature Greek letters

PV photovoltaic Δ difference
S/G solid/gas υ stoichiometric coefficient
TR thermochemical reactor η efficiency
MVC mechanical vapor compression τ mixture mass ratio
COP coefficient of performance ρ effective density
DOD depth of discharge θ Carnot coefficient
CFCs chlorofluorocarbons
HCFCs hydrofluorocarbons Subscripts
HP high pressure
LP low pressure 0 reference
BaCl2 barium chloride salt 1 operating point 1
ENG expanded natural graphite 2–3 operating mode 2 combined with mode 3
T temperature amb ambient
P pressure cold cold (heat source)
Ex exergy cond condenser
Q heat (J) comp composite
I solar irradiation (kWh/m2/day) des desorption
W compressor work (J) disch discharge
M molar mass (kg/mol) eq equilibrium
N number of moles evap evaporator
n number of moles per m3 ex exergy
Vswept compressor’s swept volume rate (m3/h) g gas (superheated)
X reaction global advancement (0…1) gas gas
qV,cold cooling capacity per volume of storage (Whcold/L) init initial
qS,cold cooling capacity per surface of solar collector (Whcold/m2/ is isentropic
day) liq-vap liquid-vapor
S0 anhydrous salt max maximum
S1 ammonia-charged salt r reactor
Pref reference pressure, 1 Pa ref reference
R universal gas constant, 8.3144 J/K/mol room room (heat source)
ΔHV0 NH3, standard enthalpy of vaporization, 23,366 J/mol s source (during desorption)
ΔSV0 NH3, standard entropy of vaporization, 193.3 J/K/mol sat saturated
ΔHR0 BaCl2–NH3, standard enthalpy of transformation, synt synthesis
38,250 J/mol vol volumetric
ΔSR0 BaCl2–NH3, standard entropy of transformation, 232.38 J/
K/mol Superscripts
cp0 heat capacity of the anhydrous salt, 75.1 J/K/mol
cp1 heat capacity of the ammonia-charged salt, 390 J/K/mol 0 standard (25 °C, 1 atm)
is isentropic

French Polynesia, the region of relevance to this work, is faced with just
such challenges.
French Polynesia is comprised of more than one hundred islands
and atolls that are distributed in a region approximately the size of
Western Europe [2]. Nearly half of these islands are uninhabited and
many others have a very low population (see Table 6 in Appendix A).
The ratio between primary energy imports and total energy consump-
tion in French Polynesia is above 85% [3], which puts the region in a
vulnerable position due to the unpredictability of oil prices. What is
more, the price of electricity for a small household (3000 kWh/year) is
about 0.30 USD/kWh [4], between two and three times the price found
in North America [5] and Europe [6]. A study conducted in 2011 on the
use and trends of electro-domestic equipment, including air condi-
tioners, in Tahiti and Moorea, French Polynesia’s most populated is-
lands, showed that there was a steady rise in energy usage in the home,
notably of air conditioning units, representing 36% of usage [7].
On the positive side, French Polynesia is rich in renewable re-
sources, particularly solar energy. For instance, the average global solar
horizontal irradiation (a key parameter that assesses solar energy po-
tential) for fifteen of the most inhabited islands of French Polynesia and
accounting for 94% of its total population (see Table 6 in Appendix A),
is 5.8 kWh/m2/day. Such a high value makes the exploitation of solar
energy in French Polynesia an interesting proposition. This is especially
the case for photovoltaic technology due to its maturity, declining costs
and adaptability for use in both large and small-scale energy projects.
Notwithstanding the foregoing, electricity produced by solar energy
presents some difficulties. For instance, due to its stochastic nature,
solar energy is non-dispatchable–that is, its production cannot be
regulated in order to meet electricity needs. Moreover, photovoltaic
technology, the most commonly employed electricity production
method using solar energy, does not possess an inherent energy reserve
(normally called inertia or spinning reserve), meaning that sudden var-
iations of solar irradiation are not attenuated at the output and strongly
perturb the electricity grid. These characteristics make the supply of
electricity using solar energy challenging, especially where the pro-
portion of its use over total electricity production is high [1]. To
overcome these difficulties, the combination of smart electricity grids
with energy storage systems has proven to be an effective solution [8].
The term smart grid refers to an electricity network that performs
intelligent management of resources, including production and storage,
in order to deliver a reliable and sustainable electricity supply.
Currently, there is a great deal of research in the field of control and
optimization strategies applied to smart grids. For instance, Iqbal et al.
[9] present a comprehensive review of publications concerning

241
F. Ferrucci et al.
optimization algorithms and tools used in the context of renewable
energy sources. Kyriakarakos et al. [10] propose an intelligent man-
agement algorithm with prediction capabilities applied to autonomous
micro grids with multiple energy generating sources. In another review,
Guerrero et al. [11,12] present advanced control techniques used in
order to optimize the management of micro grids.
Smart grids don’t necessarily have a size restriction, which makes
their application potentially widespread: from a grid in a small village
of a few kW (also called smart micro grid) to an electricity network that
manages hundreds of MW. For instance, Salas et al. [13] present a re-
view of smart hybrid systems combining diesel generators, photovoltaic
panels and batteries for application in remote villages and isolated is-
lands. They also describe several examples of such existing systems. On
a larger scale, Colak et al. [14] discuss the status of existing smart grid
projects across Europe while Hatziargyriou et al. [15] present examples
of research and development smart grid projects around the world.
Energy storage systems are key assets in smart electricity grids that
use stochastic renewable energy sources. They have two fundamental
roles: (a) ensuring power quality by providing voltage and frequency
stabilization and (b) equalizing the overall daily electricity production
with consumption- for instance, by returning the surplus electricity
produced during the day to the grid at night. The most common energy
storage systems fall into one of the following categories: electro-
chemical (e.g. lead-acid, lithium-ion batteries), electrical (e.g. super-
capacitors, superconducting magnetics), chemical (e.g. fuel cells, elec-
trolyzers), mechanical (e.g. pumped hydro, compressed air, flywheels)
and thermal (e.g. sensible, latent or sorption heat). Palizban [16] and
Luo [17] review many of the energy storage systems found in modern
electricity grids and give a comprehensive list of the services that they
provide.
Within the category of thermal storage systems resides what is re-
ferred to as thermochemical sorption, a phenomenon that allows for the
storage of energy in the form of chemical potential which can later be
utilized to produce a cooling effect. A sorption process exploits the
physical or chemical bonds between a sorbent material (liquid or solid)
that has the ability to absorb an active gas in order to produce a thermal
or cooling effect. When the sorbent is liquid, the process is referred to as
liquid/gas absorption. When solid, the process is called physical or
chemical solid/gas adsorption–this latter is also known as thermo-
chemical sorption [18]. Compared with sensible and latent heat sto-
rage, an advantage of thermochemical sorption is that it has virtually
no energy losses. Moreover, compared with electrochemical storage
technologies, thermochemical sorption has a longer charge/discharge
life cycle. Sorption technology has found its footing in solar cooling
applications, where it is mainly utilized in conjunction with solar
thermal collectors. Kim [19] and Ferreira [20] present comprehensive
reviews of such systems.
Given the importance of energy storage in modern electricity grids
in tropical insular regions like French Polynesia, where a significant
proportion of energy is used for space cooling, the present work pro-
poses and analyzes a system that integrates an energy storage method
based on a thermochemical sorption process that is coupled with a con-
ventional mechanical vapor compression cooling machine powered by
photovoltaic electricity. The combination of these elements presents an
innovative solution that offers energy storage capabilities for cooling
applications. Additionally, whereas existing systems using sorption
technology for cooling applications require the use of heating sources
with temperatures higher than 55 °C, such as the thermochemical
cooling systems experimented by Stitou [18] and Wang [21], the pro-
posed system can operate with much lower heating source tempera-
ture–as low as 35 °C. This allows for the revalorization of a wider range
of low-grade heat sources that are currently not exploited due to their
low exergy value. The compressor–driven method for energy storage
and deferred cooling has not yet been explored in literature and re-
presents an innovative concept, which is described and developed in
this work.
242
Applied Energy 219 (2018) 240–255
This paper is organized as follows: Section 2 introduces the principle
of the solid/gas thermochemical sorption process. Section 3 presents
the system architecture that combines a thermochemical reactor with a
conventional mechanical vapor compression cooling system. The sec-
tion also explains the various operating modes. Section 4 outlines the
system’s operating conditions and admissible range of operation. Sec-
tion 5 assesses the system performance by analyzing relevant efficiency
indicators. Section 6 discusses the results. Section 7 provides detail
regarding a conceptualized prototype which is planned to soon be
under development. Finally, Section 8 presents the conclusions of the
article.
This work elaborates upon the paper presented at ICAE2017, Aug
21–24, Cardiff, UK by (a) further developing the equations related to
the operating conditions of the proposed system, (b) performing a
parametric analysis in order to ascertain the system’s admissible oper-
ating range, (c) conducting a study of the compressor’s operating point,
(d) expanding upon the system’s performance by adding an exergy
analysis and (e) providing detail regarding the prototype under devel-
opment.
2. Principle of solid–gas thermochemical sorption process
The principle of solid/gas (S/G) thermochemical sorption processes
is based on the thermal effect of a reversible reaction between a solid
(S) and a reactive gas (G). The reaction follows the general equation:
Synthesis(cold production)
{S0} + υ ·G ⇄ {S1} + υ ·ΔHr
desorption(energy storage) (1)
where S1 is salt rich in gas, S0 is salt poor in gas, υ is the stoichiometric
coefficient and ΔHr is the enthalpy of the reaction per mole of gas [18].
The strong chemical bonding involved in these reactions translates into
high values of ΔHr. The solid/gas reaction from (1) can be coupled with
a liquid/gas phase change process of the same working gas in order to
implement a cold production machine:
evaporation
G (liq. ) + ΔHliq − vap ⇄ G (gas)
condensation (2)
where ΔHliq-vap corresponds to the enthalpy of vaporization.
A cold production process that is based on thermochemical reac-
tions needs three components for its most basic implementation: a re-
actor that contains the salt, a two-phase heat exchanger in which eva-
poration or condensation of the reactive fluid occurs and a reservoir to
contain the condensed fluid. The process has two distinct phases: energy
storage and cold production, which are related to the direction of the
processes in (1) and (2).
From right to left, the reaction (1) is called desorption. In this phase,
thermal energy is transformed into chemical potential for a deferred
cold production. The solid S1, charged with gas, absorbs an amount of
heat (i.e. endothermic process), releases the gas and transforms into S0.
The gas flows out of the reactor to the heat exchanger, where it con-
densates and continues to the reservoir, where it is stored in liquid
form. This phase corresponds to the right-to-left process in (2).
From left to right, the reaction (1) is called synthesis and it is when
cold is produced. The solid S0 absorbs and reacts in an exothermic
manner with the gas coming from the heat exchanger, which acts as an
evaporator, forming the salt S1 and releasing the heat of reaction to the
environment. This phase corresponds to the left-to-right process in (2).
The S/G pair used in this study is barium chloride salt, BaCl2, which
reacts with eight moles of ammonia (NH3) to form BaCl2·8NH3. This salt
has been successfully used for solar ice-making and solar air con-
ditioning processes using simple flat plate solar collectors [18]. This S/
G pair is suitable for such an application as it can operate (i.e. react)
with relatively low temperatures and moderate pressure. The use of
ammonia has the advantage that it is readily available, inexpensive, not
a contributor to ozone depletion, greenhouse effect or global warming
and is biodegradable [22]. However, its toxicity requires taking unique
F. Ferrucci et al.
safety measures. Since ammonia reacts with copper in the presence of
common contaminants such as air and water, ammonia systems are
constructed using aluminum, carbon-steel and stainless-steel compo-
nents [23].
The two processes, i.e. the chemical reaction (1) and the phase-
change (2), are mono-variant and their thermodynamic equilibrium
conditions follow the Clausius–Clapeyron relation. The equilibrium
conditions for the two processes are determined by only one variable,
either pressure or temperature:
ΔHi0 ΔSi0
ln ( )=−
Pi,eq
Pref R·Ti,eq
+ R
, The system has three operating modes: (a) Mode 1, conventional
r :reaction
with i = ⎧
⎩ liq-vap : liquid−vapor
⎨ (3)
where Peq and Teq correspond to the equilibrium pressure and tem-
perature of each process, ΔH0 and ΔS0 are the standard enthalpies and
entropies of the two processes and Pref is a reference pressure [18]. The
nomenclature section provides the numerical values of all the men-
tioned parameters. Fig. 1 depicts the Clausius–Clapeyron diagram of the
processes, which is a graphical representation of (3).
2.1. Reaction global advancement X
At any given moment of time and in any reaction phase (i.e.
synthesis, desorption or no reaction), there will be elements of both
sides of (1) in the reactor- that is, salt rich in reacting gas S1 as well as
anhydrous salt S0. The reaction global advancement X expresses the
fraction of S1 with respect to the total quantity of salt:
NS1 NS1
X= =
NS NS1 + NS0
In (4), NS1 and NS0 represent the number of moles of the rich and
anhydrous salt, respectively, while NS is the total number of moles of
salt contained in the reactor. The reaction global advancement goes
from X ≈ 0 to X ≈ 1 during synthesis and from X ≈ 1 to X ≈ 0 during
desorption. Note that, unlike physical adsorption, the reaction equili-
brium condition does not depend on X for chemical adsorption pro-
cesses (i.e. the variable X does not appear in (3)).
3. System overview
The architecture of the proposed hybrid system, shown in Fig. 2,
comprises the elements of a typical mechanical vapor compression re-
frigeration cycle (MVC) (i.e. evaporator, compressor, condenser, re-
servoir and throttling valve), a grid-connected photovoltaic (PV) in-
stallation (i.e. solar panels and inverter) and a thermochemical reactor
(TR) for energy storage. The MVC cycle uses ammonia as refrigerant,
(a)
Fig. 1. Clausius–Clapeyron diagram of the BaCl2–NH3 thermochemical sorption process (dashed) and the NH3 phase-change process (solid); (a) pressure and temperature in linear scale;
(b) pressure in log scale and abscissa linear in −1/T (-K−1) (a second x axis shows the corresponding values in °C).
Applied Energy 219 (2018) 240–255
which also corresponds to the reactive gas of the TR.
Both the PV installation and the grid are used to meet electricity
requirements of the building and, in particular, to power the com-
pressor of the MVC cycle for cold production. In addition, each time
that there is a surplus of energy from the PV panels and no cooling
needs, the system switches its configuration mode and the extra power
is used to run the compressor in order to drive the desorption of the TR,
thus storing electricity in the form of chemical potential energy. The
stored energy is later used to produce a cooling effect for the building,
without the need for the compressor.
MVC cycle, (b) Mode 2, storage phase (desorption) and (c) Mode 3,
discharge phase (synthesis).
Fig. 3 depicts the three operating modes and includes a schema of
each system configuration with the flow of NH3 through the circuit
(inactive elements are greyed out), and its corresponding Clausiu-
s–Clapeyron and P–h charts. The Clausius–Clapeyron diagram shows
the mono–variant equilibriums of the liquid/gas phase–change of NH3
and the solid/gas reaction occurring in the TR. The P–h chart gives a
graphical representation of the path of the refrigerant and allows for
direct measurement of the evaporator’s cold production and com-
pressor-consumed work. Table 1 gives a description of the operating
points indicated on the Clausius–Clapeyron and P-h charts for mode 2
(desorption) and mode 3 (synthesis).
The paragraphs below further describe the three operating modes.
Mode 1: Conventional MVC cycle
This operating mode is shown in Fig. 3(a), in which a cooling de-
mand is occurring. The reactor is isolated from the rest of the circuit
(the reactor's valve is closed) and the system behaves as a conventional
MVC refrigeration circuit with the following elements in closed-loop:
(4) evaporator–compressor–condenser–reservoir–throttling valve or capil-
lary tube. The electrical energy necessary to run the compressor and
produce the cooling effect comes either from the solar panels and/or
from the electric grid. The isolated reactor is at ambient temperature
and its pressure is fixed by the S/G equilibrium condition (see Fig. 1).
Mode 2: Energy storage (desorption)
In this mode, shown in Fig. 3(b), it is assumed that there is no
cooling demand and there is a surplus of energy from the PV installa-
tion. This extra energy is therefore used to power the compressor and
desorb the TR. The evaporator is disconnected from the circuit and the
remaining elements are connected in the following order: re-
actor–compressor–condenser–reservoir. The compressor pumps the gas
out of the TR, decomposing the ammonia–charged salt, thus storing
chemical potential energy for later use. The heat of desorption is pro-
vided either by the environment at ambient temperature or by a low-
grade waste heat source, resulting in reduced power consumption by
the compressor.
Fig. 3(b) shows two distinct cases- one in which the heat source is
id is
liqu es ium
led nth ibr
coo Sy uil
Sub rium eq
quil
ib
S/G ion
e rpt
G/L gas so
ted De
erhea
Sup
(b)
243
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

INVERTER

CONDENSER SOLAR
PANELS
HEAT
RELEASED
TO THE CONTROLLER
AMBIENT
COMPRESSOR
MICRO
GRID
TO SUPPLY
ELECTRICITY
REFRIGERANT REQUIREMENTS

HEAT
RESERVOIR EXTRACTED
FROM THE
BUILDING
HEAT EXCHANGED WITH
THE AMBIENT OR WITH A EVAPORATOR
LOW-GRADE HEAT SOURCE

THERMO-CHEMICAL
REACTOR

Fig. 2. System architecture combining a thermochemical storage unit (thermochemical reactor) and a typical mechanical vapor compression refrigeration cycle powered by photovoltaic
electricity.

the external environment (solid curve) and another with a heat source
of 50 °C (dashed curve). The two cases consider a 10 K temperature
difference between the reactor and the heat source.
The system can alternate between mode 1 and mode 2 until the re-
actor is completely discharged of gas (i.e. the reaction global ad-
vancement X ≈ 0). At that point, the maximum amount of energy has
been stored.
Mode 3: Cold production by using stored energy (synthesis)
In this operating mode, Fig. 3(c), the potential energy stored in the
reactor is expended to produce cold. This process does not require the
electric compressor and by consequence is well suited for cold pro-
duction during the evening hours, when there is no solar energy
available but ambient temperature is still high. The compressor and
condenser are disconnected from the circuit and the remaining ele-
ments are arranged in the following order: reservoir–evapor-
ator–reactor. The liquid ammonia flows from the reservoir to the eva-
porator, where it evaporates –producing a cooling effect– and continues
to the TR, where the synthesis reaction occurs. Given that synthesis is
an exothermic process, the reactor rejects heat to the external en-
vironment, which acts as a heat sink. This operating mode can continue
until the TR is completely charged with gas (i.e. the reaction global
advancement is close to 1).
The system is managed by a smart controller that decides which
operating mode to activate based on criteria such as user input (e.g.
room temperature setting), energy produced by the photovoltaic pa-
nels, electricity consumed by the building and energy stored in the
thermochemical reactor.
environment at ambient temperature. As will be shown later in this
Section, in this scenario the compressor work per mole of gas is max-
imal. In the second scenario, Fig. 4(b), the heat of desorption is pro-
vided at a temperature higher than ambient temperature T0, which
reduces the compressor work with respect to the previous scenario.
Finally, in Fig. 4(c), the heat source temperature has risen sufficiently
so that desorption is produced with no mechanical work required. This
situation occurs when the desorption temperature is such that the
thermochemical reactor pressure is equal to that of the condenser.
Fig. 5 shows the Clausius–Clapeyron diagram of the system during
the synthesis phase, where the reactor is at temperature Tsynt, pressure
Psynt and exchanges heat with the environment that is at ambient
temperature T0. The difference between these two temperatures is
ΔTsynt. The evaporator is at temperature Tevap and pressure Pevap, coin-
ciding with the TR pressure Psynt, and exchanges heat with a cold source
that is at temperature Tcold, with a temperature difference ΔTevap. This
cold source corresponds, for example, with the temperature of the room
to be cooled.
Next, the formulas associated with the operating modes of the
system will be described.
Desorption process (mode 2):
The reactor and condenser temperatures and pressures are:
Tdes = Ts−ΔTdes (5)
Pdes = Pr ,eq (Tdes ) (6)
Tcond = T0 + ΔTcond (7)

Pcond = Pliq − vap,eq (Tcond ) (8)

4. Operating conditions
Eq. (6) and (8) correspond to the S/G and L/G Clausius–Clapeyron re-
This section describes the equations corresponding to the desorption lation, respectively, shown in (3).
(storage) and synthesis (cold production) processes, which will be later The compressor intake and discharge pressures coincide, respec-
applied to a performance analysis of the system. Prior to introducing tively, with the reactor and condenser pressures. The specific enthalpy
the equations, it is necessary to first outline Figs. 4 and 5. and entropy of the gas at the suction and discharge lines are given by:
Fig. 4 shows the Clausius–Clapeyron diagram of the system during
hsuct = h gas (Tdes,Pdes ) (9)
desorption phase, in three distinct scenarios. In each case, the TR is at
temperature Tdes, pressure Pdes and exchanges heat with a source that
ssuct = sgas (Tdes,Pdes ) (10)
has a temperature of Ts. The difference between these two temperatures
is ΔTdes. The condenser is at temperature Tcond, pressure Pcond and ex- is
hdisch = hgas (ssuct ,Pcond ) (11)
changes heat with the environment that is at a temperature of T0, with a
temperature difference of ΔTcond. The value of hisdisch
refers to the isentropic specific gas enthalpy at the
In the first scenario, Fig. 4(a), the heat source for desorption is the discharge line, which, together with the compressor isentropic

244
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Fig. 3. Operating modes of the hybrid system; (a) conventional MVC cycle; (b) storage phase; (c) discharge phase and cold production. The figure shows the system configuration for each
mode together with the Clausius–Clapeyron and P–h charts.

Table 1
Description of the operating points indicated on the Clausius–Clapeyron and P–h charts for mode 2 (desorption) and mode 3 (synthesis), corresponding to Fig. 3b and Fig. 3c.

Mode 1: compressor–driven desorption process Mode 2: synthesis process and cold production
➀ → ➁ → ➂ → ➃ (Fig. 3b) ➄ → ➅ → ➆ → ⑧ (Fig. 3c)

Point Description Point Description

➀ Reactor operating point during gas desorption.
➁ Compressor’s discharge line.
➂ Condenser; vapor fully condensed.
➃ Liquid reservoir; subcooled liquid is stored at ambient temperature.
➀→➁ Mechanical compression process. Gas is pumped from the reactor to the
condenser. The enthalpy difference between the two points is equal to the
compression mechanical work per mole of gas. An isentropic efficiency is taken
into account.
➁→➂ Constant pressure condensation process. Heat of condensation is released to
the environment.
➂→➃ Sub-cooling process. The fluid is in the reservoir and heat is released until it
reaches ambient temperature.
➄ Reservoir; fluid in saturated liquid state.
➅ Liquid-vapor mixture after an isenthalpic throttling in the evaporator.
➆ Evaporator. Fluid fully evaporated.
⑧ Reactor operating point during synthesis process.
➄→➅ Isenthalpic throttling in the evaporator.
➅→➆ Constant-pressure evaporation process. Heat is absorbed from the room
enabling the cold production. The enthalpy difference between the two points
is equal to the evaporator cooling effect per mole of fluid.
➆→⑧ Constant-pressure superheating of the fluid which is heated and absorbed by
the reactor in synthesis phase.

245
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Compressor‐driven reactor desorption (storage phase) – Clausius–Clapeyron chart

gas
Qcond Qcond Qcond Qdes
Pcond Tcond Pcond Tcond Pcond Tcond
=Pdes W=0 Tdes
W=Wmax 0<W<Wmax
gas gas
Qdes
Pdes Tdes

Qdes
Ts,max
Pdes Tdes

T0 = T s T0 Ts T0 Ts
Heat of desorption provided at Heat of desorption provided at a Purely thermal desorption
ambient temperature temp. higher than ambient temp. (no compressor work)

(a ) (b) (c)
Fig. 4. Clausius–Clapeyron diagram of the system during desorption phase (i.e. mode 2) for three different cases (a) Qdes provided at ambient temperature T0, (b) Qdes provided at Ts > T0,
(c) purely thermal desorption process (no compressor work).

Reactor synthesis (cold production) necessary energy (sensible heat) to bring the initial fractions of salt S0
Clausius–Clapeyron chart and S1 (given by the initial reaction global advancement Xdes,init) from
their initial state at ambient temperature T0 to desorption temperature
Tdes.
Now that the equations corresponding to the desorption phase have
been introduced, it is possible to clarify why the compressor work is
maximal when the TR is desorbed at ambient temperature, as shown in
Fig. 4(a), and why it diminishes with the increase of the TR heat source
temperature, as in Fig. 4(b), until becoming zero, as seen in Fig. 4(c).
Qevap gas Qsynt
One way to explain this phenomenon is with the help of a pressure-
Pevap
=Psynt Tsynt enthalpy (P-h) chart corresponding to the processes of Fig. 4. Although
Tevap such a chart is not available in Fig. 4, there is one shown in Fig. 3(b),
which refers to the same desorption process. In the P-h chart of
Fig. 3(b), paths 1 → 2 and 1′→2′ each describes compression processes
(explained in Table 1). Fig. 3(b) shows that the enthalpy difference
between the compressor discharge and suction lines (hisdisch − hsuct) in-
Tcold T0 creases (abscissa of P-h chart in Fig. 3b), when the pressure difference
augments (ordinate of P-h chart in Fig. 3b). Additionally, since (12)
Fig. 5. Clausius–Clapeyron diagram of the system during synthesis phase (i.e. mode 3). states that this enthalpy difference is proportional to the compressor
work, it can be seen that, the greater the pressure difference, the greater
efficiency ηis, serves to compute the compressor specific irreversible the compression work. At the limit, when the desorption temperature is
work: sufficiently high, both the pressure difference and compression work
is become zero. In the P-h chart of Fig. 3(b), this would happen if point ➀
hdisch −hsuct
Wdes = rose and point ➁ moved to the left until they overlap each other.
ηis (12) Synthesis process (mode 3):
The maximum desorption temperature is defined as the temperature The reactor and evaporator temperatures and pressures are given
at which the TR reaches the same pressure as the condenser. At that by:
point, as shown in Fig. 4(c), the compressor work becomes zero: Tsynt = T0 + ΔTsynt (16)
Tdes,max = Tr ,eq (Pliq − vap,eq (Tcond )) (13)
Psynt = Pr ,eq (Tsynt ) (17)
Ts,max = Tdes,max + ΔTdes (14)
Tevap = Tliq − vap,eq (Pevap) = Tcold−ΔTevap (18)
Eq. (13) involves both the S/G and the L/G Clausius–Clapeyron rela-
tions from (3). Pevap = Psynt (19)
The heat of desorption Qdes is equal to the enthalpy of reaction, plus
As shown in (17) and (19), during synthesis, the evaporator and
the necessary energy required for the reactants to reach the operating
reactor have the same pressure, corresponding to the S/G equilibrium
conditions of desorption, that is, Tdes and Pdes. The value of Qdes, ex-
pressure at temperature Tsynt.
pressed per mole of gas and considered positive when heat flows toward
The heat of synthesis per mole of reacting gas, Qsynt, is given in (20)
the reactor, is given by:
and is considered positive when heat leaves the reactor.
1
Qdes = ΔHr0 + (cP,s1 Xdes,init + cP,s0 (1−Xdes,init ))(Tdes−T0 ) 1
υ (15) Qsynt = |ΔHr0 |−(hg (Tsynt ,Psynt )−h vap,sat (Tevap))− (c PS1 Xsynt ,init
υ
The first term of the second member in (15) accounts for the en- + c PS 0 (1−Xsynt ,init ))(Tsynt −T0) (20)
thalpy of reaction (endothermic). The second term represents the

246
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

The first term on the right side of (20) refers to the enthalpy of
reaction. The absolute value bars are necessary as the reaction is, in this
case, exothermic, so ΔHr0 is negative. The second term refers to the
necessary energy to bring the input gas to reacting conditions. This gas
leaves the evaporator as 100% saturated vapor with temperature Tevap
and enthalpy hvap,sat(Tevap), and reacts at (Tsynt,Psynt) with a specific
enthalpy of hg(Tsynt,Pr). The last term in (20) represents the necessary
energy (sensible heat) to bring the initial proportions of salt S0 and S1
from ambient temperature T0 to synthesis temperature. The value
Xsynt,init represents the reactor global advancement at the start of the
synthesis process.
The cold production per mole of gas is given by:
Qevap = h vap (Tevap)−hliq (T0)
(21)
with hvap and hliq as the specific enthalpy of the fluid fully evaporated
and condensed, respectively.
5. System performance
This section presents the mathematical formulation of three per-
formance indicators that are normally used in storage and cold pro-
duction systems. These are:
This formula of COP1 will be utilized in the section below, which
compares the system being studied with other energy storage methods.
Although this COP will not be further elaborated upon in this work, it
should be noted that ammonia has a high latent heat of vaporization
(eight times greater than R-12 and six times greater than R-134a, per
mass of refrigerant) and operates with pressure levels comparable to
those of other refrigerants. These characteristics result in a highly en-
ergy-efficient MVC cycle with minimal environmental impact [22].
The second, COP2–3, relates the compressor work during desorption
phase in mode 2 to the amount of cold produced during synthesis phase
in mode 3:
cold production in mode 2 Qevap
COP2 − 3 = =
compressor work for reactor desorption in mode 3 Wdes
(23)
By inserting (12) and (15) into (23), the function of COP2–3 is ob-
tained, which is a function of the following list of arguments:
COP2 − 3 = COP2 − 3 (T0,Ts,ηis ,ΔTdes,ΔTsynt ,ΔTcond,ΔTevap) (24)
In the results and discussion section, a parametric study of this equation
is performed.

(a) Coefficient of performance, COP. In the context of the system
analyzed here, this parameter indicates the relationship between
cold production and the electric consumption of the compressor.
(b) Exergy efficiency, ηex. This parameter not only considers the com-
pressor work, but also takes into account the heat used to desorb the
TR. It can be said that ηex assigns an economic value (cost) to the
heat of desorption.
(c) Cooling capacity, which will be evaluated by two parameters: i) the
capacity of cold production per unit of volume and mass of the TR
and ii) the daily cold production per area of solar panel. The eva-
luation of the second parameter is relevant due to the fact that the
system is intended to store energy coming from PV panels.
Below, further detail is given with regards to each indicator men-
tioned above.
5.1. Coefficient of performance (COP)
For the system presented in this study, two distinct coefficients of
performance need to be defined. The first, COP1, is associated with
operating mode 1, when the system works as a traditional ammonia
MVC cycle and there is no energy storage:
cold production in mode 1
COP1 =
compressor work in mode 1
5.2. Exergy efficiency
Fig. 6 shows heat and exergy flows of the system during the storage
and cold production phases. An amount of exergy Exs flows towards the
system during the storage phase and another quantity, Excold, flows out
of the system during the cold production phase. The heat of con-
densation and synthesis are rejected at the dead-state temperature T0,
and thus do not contribute to the exergy flow.
The exergy efficiency is defined in general terms as the ratio be-
tween the useful exergy and the exergy expended (or paying exergy). In
the system under study, the useful exergy corresponds to the exergy
|Excold| of the heat transferred from the low-temperature medium at
Tcold to the system (which occurs in the evaporator). The absolute value
is needed as the exergy is negative (i.e. it leaves the system). The exergy
expended is the sum of the compressor work Wdes (100% exergy) and
the exergy Exs of the heat transferred from the source at Ts to the system
(which happens in the reactor). This is expressed in the following eq.:
useful exergy |Ex cold | |Qevap θcold |
ηex = = =
exergy expended Wdes + Exs Wdes + Qdes θs (25)
The Carnot coefficients for desorption phase θs and for cold pro-
duction θcold are defined as follows:
T0
θs = 1−
(22) Ts (26)

: exergy
Qdes : heat
Ts Reactor
Exs COLD
PRODUCTION
Compressor (SYNTHESIS)
Wdes
Qcond Qsynt
T0 Condenser Reactor
Excond=0 Exsynt=0
ENERGY
STORAGE
(DESORPTION)
Qevap
Tcold Evaporator
Excold
Fig. 6. Heat and exergy flow of the storage/cold production system. The light gray blocks represent the system during storage phase or mode 2 (left), and cold production or mode 3
(right).

247
F. Ferrucci et al.
T0
θcold = 1−
Tcold
The expression of ηex is derived by combining (12), (15), (21), (25), (26)
and (27). Its value is a function of several parameters, namely:
ηex = ηex (T0,Ts,ηis ,ΔTdes,ΔTsynt ,ΔTcond,ΔTevap)
As in the case of the COP2–3, the results and discussion section shows
a parametric study of this equation.
5.3. Cooling capacity
The cooling capacity of a thermochemical reactor, that is, the cold
production per volume or mass of TR, depends on the nature and the
amount of salt in the reactor. Generally, the salt is mixed with an inert
binder (expanded natural graphite) to enhance heat and mass transfers
in the reactor. The mixture, characterized by a mass ratio τ (generally
0.8 kg of salt per kg of mixture), is compressed in situ to form a con-
solidated reactive medium with high thermal conductivity, high por-
osity and mechanical elasticity [24]. The consolidation of the salt-
graphite mixture is carried out in the reactor until a given effective
composite density ρcomp of reactive medium is obtained. Practical den-
sities range from 300 to 500 kg/m3 of reactor in order to obtain ac-
ceptable gas permeability.
These characteristic parameters of the reactive salt implementation
allow for the evaluation of the number of moles of reactive gas (am-
monia) per m3 of reactor, namely ngas, which is absorbed by the reactor
connected to the evaporator during the synthesis phase:
τ . ρcomp
ngas = υ|ΔX |
Msalt
In (29), Msalt is the molar mass of the anhydrous salt (208.23 g/mol)
and |ΔX| is the reaction global advancement difference between the
start and the end of the synthesis process.
Multiplying the value of ngas with the cold production per mole of
gas shown in (21), the cold production per volume of reactor (or vo-
lumetric cooling capacity) is obtained:
qV ,cold = Qevap Ngas
Considering a value of Qevap of 20 kJ/mol for ammonia, a composite
effective density ρcomp of 400 kg/m3, Msalt of 0.208 kg/mol and ΔX of
0.9, the reactor volumetric energy density for cooling production qV,cold
equals 220 kJ per liter of reactor (or 60 Whcold/L approximately). This
last value, plus the information regarding the reactor’s shape and the
(a)
Fig. 7. Compressor pressure ratio during desorption process (mode 2) as a function of source temperature Ts, for three scenarios of ambient temperature T0. (a) ΔTdes and ΔTcond of 5 K; (b)
ΔTdes and ΔTcond of 10 K.
Applied Energy 219 (2018) 240–255
materials of its walls, allows for the computation of the reactor’s mass
(27) energy density, resulting in an approximate value of 200 kJ per kg of
reactor.
Keeping in mind that the system being studied is intended to store
energy coming from PV panels, another criterion worth analyzing is the
(28) cooling capacity per surface of solar collector, namely qS,cold, which
assesses how efficiently the system takes advantage of the surface ex-
posed to sunlight. The parameter qS,cold specifies the daily cold pro-
duction per square meter of solar panel, using the following formula:
qS,cold = Isolar ·ηPV + conv ·COP2 − 3, (31)
with Isolar as the solar irradiation (typically given in units of kWh/m2/
day) and ηPV+conv as the solar efficiency of the PV panels and
converter.The results and discussion section compares the values of
qV,cold and qS,cold with other energy storage methods.
6. Results and discussion
This section presents and analyzes the results of the equations for-
mulated thus far, in order to draw conclusions about the system being
studied. These results include: compressor operating condition, tem-
perature levels of cold production, coefficient of performance, exergy
efficiency and a comparison with other energy storage technologies.For
the study of the coefficient of performance and exergy efficiency, the
following parameters have been chosen:
(i) COP2 − 3 = COP2 − 3 (arg), with arg
(29) =( T0 , Ts , ηis ,ΔTdes,ΔTsynt ,ΔTcond,ΔTevap )
⏟ °C T …⏟ 
25,30,35 T 0⏟ s,max 0.7 … 0.8 5K,10K
(ii) ηex = ηex (arg) (32)
The selected values of ambient temperature are in accordance with
typical weather patterns in French Polynesia. The temperature differ-
ences of 5 K and 10 K permit reasonable rates of heat transfer. The
source temperature Ts varies continuously in most of the cases ana-
(30)
lyzed, ranging from ambient temperature to Ts,max. The compressor
isentropic efficiency varies continuously from 0.7 to 0.8.
All of the computations were performed with MATLABTM Software
and the CoolProp library [25], the latter of which was used to retrieve
the thermodynamic properties of ammonia.
(b)
248
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

6.1. Compressor operating conditions

During the desorption process, in mode 2, the compressor does not

work at an operating point normally found in a MVC refrigeration cycle,
since the compressor’s intake line is not connected to an evaporator, but
rather to a thermochemical reactor. For this reason, it is important to
ensure that the compressor’s recommended operating range is respected
while the reactor is desorbed. To that effect, this section analyzes two
key parameters: the compression ratio and the gas temperature at the
compressor discharge line (also called ‘discharge temperature’).
With regards to the compression ratio, defined as the quotient be-
tween the discharge and intake pressures, there is a maximum value
that should be respected. This value varies depending on the type of
compressor technology utilized. For example, for reciprocating com-
pressors, the maximum allowable compression ratios are usually be-
tween 5 and 6 [26].
With reference to gas temperature at the discharge line, manu-
facturers recommend that it stay below 185 °C so as to not degrade the
Fig. 9. Compressor-driven TR, regions of admissible desorption temperatures Tdes, as a
oil that lubricates the compressor’s internal mechanisms [26].
function of the condenser temperature Tcond, that allow the compressor to operate within
In this study, it will be considered that the maximum allowable the recommended operating range (i.e. compression ratio inferior to 5 and discharge gas
compression ratio is 5 and that the maximum permissible discharge temperature less than 180 °C). Striped area: ηis = 0.7, Solid & striped area: ηis = 0.8.
temperature is 180 °C.
Figs. 7 and 8 show the compression ratio and discharge tempera-
discharge temperature of 180 °C, the source temperature Ts must be
ture, respectively, of the system being studied, as a function of the
greater than 30–45 °C (depending on the values of ΔTdes, ΔTcond, ηis and
source temperature Ts, for scenarios of the parameters T0, ΔTdes, ΔTcond
T0). This restricts the circumstances under which the system can op-
and ηis.
erate with only one heat source at ambient temperature.
For the analysis of the compression ratio, Fig. 7 shows that low
To continue with the discussion, one might ask the following
values of Ts result in high compression ratios. This means that the
question: for a given condenser temperature Tcond, what is the range of
maximum compression ratio permitted imposes a lower limit on the
desorption temperature Tdes that ensures the operation of the com-
source temperature Ts. Additionally, by comparing graphs (a) and (b) of
pressor within recommended limits? This question is answered in
Fig. 7, it can be seen that, by reducing the temperature differences ΔTdes
Fig. 9, which shows the admissible range of Tdes as a function of con-
and ΔTcond from 10 K to 5 K, the compressor pressure ratio is reduced by
denser temperature, for two scenarios (namely 0.7 and 0.8) of com-
a factor of up to 1.5, especially for lower values of Ts. The reduction of
pressor isentropic efficiency ηis.
these temperature differences can be achieved, for instance, by im-
The two areas shown in Fig. 9, one for each value of ηis, have the
proving the thermal transfer of the TR and condenser.
same upper limit, corresponding to the value Tdes,max (see (13)), where
With regards to discharge temperature, Fig. 8 shows the same be-
the compressor work is zero. The lower limit, on the other hand, is
havior as seen in the analysis of the compression ratio (that is, the lower
different for each scenario. When ηis = 0.8 (region shown in grey), the
the Ts, the higher the discharge temperature). Again, this effect imposes
lower limit is defined by two curves. The first defines the desorption
a different lower limit on the source temperature Ts. Fig. 8(a) and (b)
temperature where the compressor discharge temperature is equal to
also shows that a reduction in ΔTdes and ΔTcond from 10 K to 5 K pro-
180 °C: values of Tdes above this curve indicate a lower discharge tem-
duces a drop in discharge temperature of approximately 40 °C, which is
perature. The second represents the curve of desorption temperature
a considerable amount.
which produces a compression ratio of 5; values of Tdes above this curve
An important point derived from the analysis of Figs. 7 and 8 is that,
indicate a compression ratio inferior to 5.
in order to respect the maximal allowable pressure ratio of 5 and gas

(a) (b)
Fig. 8. Gas temperature at the compressor discharge line during desorption process (mode 2), as a function of source temperature Ts, for three scenarios of ambient temperature T0. (a)
ΔTdes and ΔTcond of 5 K; (b) ΔTdes and ΔTcond of 10 K. The shaded areas represent the variation of ηis from 0.7 to 0.8 (note that a higher ηis results in a lower Tdisch).

249
F. Ferrucci et al.
The region that refers to ηis = 0.7 (with a diagonal stripe pattern)
has a lower limit marked by a single curve, corresponding to the dis-
charge temperature of 180 °C.
Note that the values of temperature Ts and T0 can be derived from
Tdes and Tcond by adding and subtracting ΔTdes and ΔTcond, respectively.
6.2. Cold production, temperature levels
An interesting aspect to analyze is the relationship between eva-
porator temperature Tevap and the temperature T0 of the TR heat sink
during cold production (synthesis process, mode 3). Fig. 10 shows this
relationship, which indicates a linear growth of Tevap with T0. Put an-
other way, the system will produce cold at lower temperatures when
the ambient temperature is low. This means that the system is well
suited for cold production during the evening hours, when no solar
energy is available, ambient temperature is no longer at its maximum
and the building still has a cooling demand.
This does not mean, however, that the system becomes ineffective at
higher ambient temperature levels. For instance, even for a value of
ambient temperature as high as 35 °C, Tevap will be 8 °C (with ΔTsynt =
5 K). This value of evaporating temperature is sufficient to provide cold
to a room at Tcold of 24 °C through an air-chiller unit connected to the
evaporator (thus Troom − Tevap = 16 °C).
For moderate values of ambient temperature, the system can ef-
fectively produce water at 0–5 °C in a chilled water loop in contact with
the evaporator and then provide cold to the building through a fan-coil
unit. See [18] for experimental results of this scheme.
6.3. Coefficient of performance
Fig. 11 summarizes the analysis of the coefficient of performance
expressed in (23) and (24). The input parameters are set according to
(32). The figure shows the COP2–3 as a function of source temperature
Ts, for three scenarios of ambient temperature T0. The shaded areas
represent the variations of the compressor isentropic efficiency from 0.7
to 0.8. The graphs (a) and (b) differ in their values of ΔT mentioned in
(32) (that is, 5 K and 10 K).
The rising trend of the COP2–3 is due to the fact that, as Ts rises, the
compressor work decreases. These curves show a vertical asymptote at
Ts = Ts,max, where the compressor work approaches zero. Additionally,
the higher the isentropic efficiency, the higher the COP2–3. In Fig. 11,
the values of ηis = 0.7 and ηis = 0.8 correspond, respectively, to the
lower and upper limits of each shaded area.
Based on Fig. 11, it can also be surmised that, for a given source
temperature Ts, the COP2–3 decreases when ambient temperature rises.
This is because, with increases of T0, the condenser pressure also aug-
ments, which elevates the compression ratio and results in an increase
of the compressor work.
It should be reiterated, nevertheless, that since the heat of deso-
rption is not included in the definition of COP2–3, it is not considered as
a cost- this can only be true if either the heat is taken from the en-
vironment at ambient temperature, or it is a waste heat from another
process.
The comparison of graphs (a) and (b) of Fig. 11 shows that a re-
duction in the heat transfer temperature differences ΔTdes, ΔTcond and
ΔTsynt from 10 K to 5 K causes an important increase in the coefficient of
performance- by 30% approximately for the lower range of Ts and
larger for higher values of Ts. For example, for values of Ts of 50 °C and
T0 of 25 °C, the increase of COP2–3 due to the reduction of the afore-
mentioned parameters, is of approximately 100%. This sensitivity
analysis plays an important role in the economic study of the system.
Section 6.5 will use these results to compare the system under study
with other energy storage methods.
250
Applied Energy 219 (2018) 240–255
6.4. Exergy efficiency
The analysis of the exergy efficiency, formulated in (25), is sum-
marized in Fig. 12, and shows ηex as a function of the source tem-
perature Ts, with the rest of the parameters varying according to (32).
The figure shows a monotonous growth of ηex with Ts, which implies
that the maximum exergy efficiency is reached when the reactor is
desorbed without the assistance of the compressor. However, this last
situation can only take place if the temperature Ts of the heat source is
sufficiently high (otherwise the desorption process will not occur).
It must be noted that this result is not inherently obvious. Before
performing the analysis, the authors supposed that the maximum value
of ηex might be attained for an intermediate temperature Ts, between
ambient temperature T0 and maximum temperature Ts,max. In fact, that
situation is indeed the case, but only with higher compressor isentropic
efficiencies and smaller temperature differences of ΔTdes, ΔTcond, ΔTsynt
and ΔTevap, which is technologically unattainable.
As seen thus far, a main advantage of the system being studied lies
in its capability to take advantage of a heat source to desorb the TR,
when the temperature of the given heat source is insufficient to produce
a purely thermic desorption (alternatively said, when the TR pressure is
lower than the condenser pressure). When such a scenario takes place,
the compressor will assist the desorption process by providing the ne-
cessary pressure lift.
6.5. Comparison with other energy storage methods
This section compares the following systems, as shown in Table 2:
Pb and Li-ion batteries, ice and chilled water thermal storage and
thermochemical storage. The criteria for the comparison are: cold
production per volume of storage qV,cold, daily cold production per
surface of solar collector qS,cold and life span of the storage technology.
Table 2 and Fig. 13 show the results of the comparison. For the
system studied in this work, the values of qV,cold and qS,cold are computed
according to (30) and (31). The comparison assumes a daily solar ra-
diation of 5.8 kWh/m2/day (see Table 6), a solar efficiency of 14% for
PV panels/inverters and 60% for thermal flat-plate solar collectors.
Note that for the batteries, qV,cold is computed by multiplying their net
energy densities by the COP of the corresponding MVC cycle and qS,cold
is computed according to (31), though applying the COP definition (22)
of a traditional MVC cycle.
As indicated in Table 2 and Fig. 13, the electrochemical technolo-
gies have a considerably shorter life span with respect to the other
Fig. 10. Evaporator temperature during cold production in mode 3, as a function of
ambient temperature T0 (which acts as the heat sink of the TR), for two different values of
ΔTsynt: 5 K and 10 K. By reducing ΔTsynt from 10 K to 5 K, the evaporator temperature
drops by approximately 6.5 K.
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

(a) (b)
Fig. 11. Coefficient of performance COP2–3 of the compressor-driven thermochemical storage for cold production as a function of source temperature Ts, for three scenarios of ambient
temperature T0. (a) ΔTdes, ΔTcond and ΔTsynt of 5 K; (b) ΔTdes, ΔTcond and ΔTsynt of 10 K. The shaded areas represent the variation of the compressor isentropic efficiency ηis from 0.7 to 0.8.

storage methods. The life span of thermal and thermo-chemical storage
is limited to the life span of the installation (usually considered to be
20 years), and not the storage itself. At PROMES Laboratory, France,
thermo-chemical reactors with more than 10 × 103 cycles have been
tested, without showing a decrease in performance.
By analyzing Fig. 13, some interesting observations can be made
about the system under study:
• It has a cooling capacity six times larger than chilled water systems,
and a cooling capacity comparable to that of ice storage systems.
Moreover, while thermal energy storage methods have important
self-discharge rates (sensible heat losses), thermochemical tech-
nology has virtually no losses (provided that the valve that separates
the thermochemical reactor from the rest of the system has no
leakage).
• Although the systems with electrochemical batteries present higher
cooling capacity values, they suffer from much shorter life spans.
• By revalorizing the wasted heat rejected by another process, the
studied system can reach important values of daily cold production
per surface of solar panel, up to 4kWhcold/day/m2.
6.5.1. Techno-economic consideration
Any attempt of an economic analysis of the system presented in this
work would be incomplete for two main reasons: first, there are no
known commercially available thermochemical reactors, since most of
the developments using such technology were conducted in research
laboratories. However, once the technique to produce the composite is
mastered, the cost to obtain the raw materials should be reasonable.
This assumption is based on estimated costs of the raw materials
(barium chloride, expanded natural graphite). Other auxiliary compo-
nents of the reactor (e.g. heat exchanger in contact with the reactor)
would also need to be considered in the economic analysis.
Second, since ammonia refrigeration machines are currently em-
ployed almost exclusively in large systems (ranging from hundreds to
thousands of kW of refrigerating power), it is difficult to find compo-
nents for small size implementations. However, with the continuous
ban or phasing-out of current refrigerants (i.e. CFCs and HCFCs) and
with no definitive environmental assessment on the latest generation of
refrigerants, natural refrigerants offer an interesting alternative. It is
expected that ammonia will be established in the market for small and
medium refrigeration systems [23]. This situation would lead to an
increased availability of ammonia-compatible components with

(a) (b)
Fig. 12. Exergy efficiency of the compressor-driven thermochemical storage for cold production as a function of source temperature Ts, for three scenarios of ambient temperature T0. (a)
ΔTdes, ΔTcond, ΔTsynt and ΔTevap of 5 K; (b) ΔTdes, ΔTcond, ΔTsynt and ΔTevap of 10 K. The shaded areas represent the variation of ηis from 0.7 to 0.8.

251
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Table 2
Comparison of storage technologies for cold production.

Storage technology Energy density DOD allowed Net energy Round-trip COP qV,cold (Whcold/ qS,cold (Whcold/ Storage cycle
(Wh/L) (%) density (Wh/L) efficiency (%) L) m2/day) life

Pb battery & MVC cycle [27–29] 90 50 45 75 4.21 190 2600 1500

Li-ion battery & MVC cycle [30] 100 100 100 90 4.21 420 3100 3500
MVC cycle & ice [31] 50 100 50 95 2.52 50 1900 Very long
MVC cycle & chilled water [31,32] 10 100 10 95 4.21 10 3200 Very long
TC + PV + heat source at 67 90 605 100 3.23 60 2600 Very long
Ts = 40 °C
TC + PV + heat source at 67 90 605 100 4.83 60 3900 Very long
Ts = 50 °C
TC + thermal solar collector, 67 90 605 100 0.44 60 1400 Very long
Ts = 73 °C [18]

1
Tcond = 40 °C, Tevap = 0°C, ηis = 0.75.
2
Tcond = 40 °C, Tevap = −15 °C, ηis = 0.70.
3
Tcond = 40 °C, Fig. 11b.
4
Thermal COP.
5
See Section 5.3.

scale projects. Nonetheless, a few manufacturers offer compressors with

Fig. 13. Comparison of storage technologies for cold production. Graphical representa-
tion of Table 2.
competitive prices, which, in turn, has the potential to make the system
analyzed in this work a viable solution from an economic perspective.
7. Prototype development
A prototype of the system presented in this study is currently being
developed at GEPASUD Laboratory (French Polynesia), which will
allow for a comparison and contrast of the results presented in this
study with experimental data.
A critical component of the prototype is the electrical compressor,
since its swept volume rate directly influences the size of the reactor.
This can be explained from (33) and (34), which show that for a given
desorption time Δtdesorp, the number of moles of salt Nsalt (and, by
consequence, the reactor size) is proportional to the gas flow rate, as
well as to the compressor swept volume rate. In (34), ηvol is the com-
pressor’s volumetric efficiency and vgas is the gas’s specific volume.
Δtdesorp
Nsalt ≈ ̇ (desorption)
Ngas
υ ·ΔXdesorp (33)
small capacities (e.g. for a few kW of cooling capacity). Table 3 lists the
smallest commercially available ammonia-compatible reciprocating
compressors for cooling applications identified by the authors of this
work. Oil-less and diaphragm reciprocating compressors have also been
considered for the prototype. In particular, diaphragm compressors
were initially good contenders, since they have a flexible membrane in
place of a piston which ensures that the compressed gas does not come
into contact with the compressor’s lubricant. However, these types of
compressors were cost-prohibitive for the project.
The compressor bought for the construction of the prototype in this
study corresponds to the model O121DK100/NH3 from German man-
ufacturer HKT Huber-Kälte-Technik, as it presents the lowest swept
volume rate of all the models analyzed (see Table 3). The compressor
has a nominal swept volume rate of 6 m3/h and is equipped with a 3 kW
motor and a frequency converter that allows for speed control to a
minimum of 60% of its nominal value, which produces a swept volume
rate of 3.6 m3/h. This limit is imposed by the manufacturer in order to
avoid lubrication problems. The compressor also includes a water-
cooled cylinder head that will only be utilized if the gas at the discharge
line surpasses 165 °C (manufacturer recommendation [26]). For lu-
brication, the compressor utilizes a synthetic type of oil that is im-
miscible with ammonia, so an oil-return circuit (i.e. oil separator and
additional filters) will be installed. Other filters will be mounted in
order to minimize the presence of oil in the thermochemical reactor.
The heat source at temperature Ts for the desorption process will be
implemented with an electrical heater in direct contact with the TR,
which will allow for a straightforward regulation of the thermo-
chemical reactor operating point. During the synthesis process, the TR
will be cooled-down using forced air convection.
Fig. 14 shows a diagram of the prototype under development and
Table 4 specifies the relationship between the operating mode (i.e.
mode 1: conventional MVC cycle, mode 2: desorption, mode 3:
Table 3
Examples of small-capacity, ammonia-compatible reciprocating compressors available in
the market.
Brand Model Vswept min (m3/h)

ηvol Frigopol TYP 14PI-DLRC-3 10 @ 35 Hz*

̇ =
Ngas ̇
·Vswept HKT O 121 DK100/NH3 3.6 @ 30 Hz*,x
vgas (T ,P ) (34) GEA F2 NH3 6.95 @ 960 rpm**
Bitzer W2TA-S190 15.7 @ 1450 rmp
With this consideration, and with the idea of obtaining a small-scale
prototype, an investigation into the availability of small ammonia- * Nominal frequency: 50 Hz.
compatible compressors in the market was conducted. As mentioned in ** Nominal speed: 1450 rpm.
x
Section 6.5.1, ammonia-based refrigeration systems are used for large- Purchased for the prototype implementation.

252
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Needle valve
T
(for capacity control)

V2
Thermochemical
Reactor Compressor Oil
T,P T Pre-cooler T,P T,P De-superheater T T T
V1 V4 separator
M
Condenser
Filter
Fan T T T
Electric V5
cylinder
heater
V3 head cooler

L
T
T
Pump
Liquid
T Thermocouple prove
Municipal water ammonia
P Pressure meter (15-20°C) reservoir

L Level meter V6

T T Expansion
valve
Evaporator Ammonia
charge/drain

Cold chamber

Fig. 14. Diagram of the prototype under development at GEPASUD Laboratory, University of French Polynesia.

Table 4 control method’ in order to reduce the ammonia flow rate coming out of
Correspondence between position of valves from Fig. 14 and operating mode of proto- the reactor during desorption phase. This is necessary since the com-
type. O: open, C: closed.
pressor minimum swept volume rate of 3.6 m3/h would still be too high
Operating mode Valve Heater Fan for the size of reactor in consideration. The implementation of such
capacity control consists of a feedback loop that connects the com-
V1 V2 V3 V4 V5 V6 pressor discharge line with the intake line, a ten-turn needle valve to
regulate the fraction of gas that returns to the intake line and two heat
Mode 1 C O O O O O OFF OFF
Mode 2 O O C O O C ON OFF exchangers- namely ‘pre-cooler’ and ‘de-superheater’, which avoid any
Mode 3 O C O C C O OFF ON gas temperature build-up effects. Although this type of capacity control
is energy inefficient, it allows for a simple flow rate control.
Table 5 shows a list of relevant design parameters and operating
Table 5 points of the prototype. The parameters that were imposed by design
Prototype: selected design parameters (marked with an asterisk) and simulated oper-
are shown with an asterisk. They were selected in order to keep the size
ating conditions.
of the reactor reasonably small, while maintaining the compressor
Reactor, general within safe operating conditions. The other information contained in
Effective length (m) 1.5* Table 5 was obtained by simulating the system shown in Fig. 14 em-
External diameter (m) 0.11* ploying MATLABTM Software and the CoolProp library [25].
ρENG (kg/m3) 120*
The thermochemical reactor under consideration consists of a
τsalt (0…1) 0.75*
Composite (BaCl2/ENG) mass (kg) 6.3 stainless steel tube filled with 6.3 kg of BaCl2/ENG composite. When
Ammonia mass (kg) 2.5 fully charged, it will contain about 2.5 kg of ammonia. This mass of
Reactor, desorption phase refrigerant can produce 0.82 kWh of cooling effect during synthesis.
Heat source temperature (°C) 45*
Another important parameter selected by design was a desorption time
Reactor gas temperature (°C) 33
Reactor pressure (bar) 4.1
of two hours. This value allows for a sufficiently slow desorption phase,
Heat of desorption (from electric heater) (W) 800 which, in turns reduces the power requirements of the heat exchangers.
Outward gas flow rate (kg/h) 1.23 As a final note, it should be pointed out that the compressor ac-
ΔX, global advancement difference (0…1) 0.8 quired for the construction of the prototype allows for a production of
Desorption time (minute) 120*
up to 7 kW of refrigerating power, which is sufficient to cool a space of
Compressor, desorption phase
Swept volume rate (m3/h) 3.6* 45 m2 (considering a cooling load of 150 W/m2). This means that, by
Temperature at intake line (°C) 26 properly resizing the thermochemical reactor and its auxiliary compo-
Pressure at intake line (bar) 4.1 nents, a real-scale system using the same compressor could be im-
Temperature at discharge line (°C) 170
plemented, for example, in order to provide cooling to a room and not
Condensing temperature (°C) 40
Condensing pressure (bar) 15.8
only to a small cold chamber (as in the case of the prototype).
Pressure ratio (HP/LP) 3.9
Isentropic efficiency (0…1) [26] 0.7*
Control capacity, mass return fraction (%) 87
8. Conclusions
Reactor, synthesis phase
Total cooling capacity (kWh) 0.82 This study demonstrates that a thermochemical reactor coupled
with a photovoltaic-driven mechanical compressor is an innovative
solution that offers energy storage capabilities for deferred cooling
synthesis) and the position of the valves, the state of the fan and the applications. Normally, the analysis and application of thermochemical
electric heater. reactors is limited to solar cooling contexts. The present work shows
The prototype will implement what is referred to as ‘capacity that such technology can provide energy storage services in modern

253
F. Ferrucci et al. Applied Energy 219 (2018) 240–255

Table 6
List of the fifteen most inhabited islands in French Polynesia [33] (which account for 94% of the total population), with their corresponding areas and average global horizontal
irradiation (GHI) [34].

Index Island Population (2017) Area (km2) Solar GHI (kWh/m2/day)

1 Tahiti 189,517 1045 5.73

2 Moorea 17,816 134 5.73
3 Raiatea 12,249 175 5.74
4 Bora bora 10,549 38 5.74
5 Huahine 6075 74 5.13
6 Tahaa 5234 88 5.74
7 Nuku Hiva 2951 387 6.61
8 Rangiroa 2709 79 5.93
9 Rurutu 2466 33 5.21
10 Hiva Oa 2243 316 6.57
11 Tubuai 2217 45 5.13
12 Ua Pou 2213 106 6.55
13 Mangareva 1384 16 5.34
14 Maupiti 1286 14 5.71
15 Raivavae 903 16 5.14

electricity grids and is particularly interesting for regions with rising
energy demands for space cooling. Thermochemical technology has
important features which are sought out in any storage system: com-
petitive storage capacity, no self-discharge and long charge/discharge
life cycle.
In contrast with 100%-solar-heat-driven thermochemical systems,
in which desorption process cannot occur if the heat source tempera-
ture is not sufficiently high, the system presented in this work utilizes
an electrical compressor to aid the process and compensates for a lower
heat source temperature. More concretely, this study shows that for the
pair barium chloride/ammonia (BaCl2/NH3), with the use of a com-
pressor, the thermochemical reactor can reduce the reacting tempera-
ture by up to 20 K (allowing, for instance, a desorption process with a
heat source at 35 °C).
The reduction in the reacting temperature permits the revaloriza-
tion of a wider variety of low-grade heat sources that are currently not
exploited due to their low exergy content; these include waste heat
from photovoltaic installations, low-temperature fuel cell stacks and
even from within the system itself (i.e. by being able to recover the heat
of condensation).
Another important result is that the system can reach a cooling
capacity value of 4 kWhcold/day per m2 of photovoltaic panel, which is
higher than the values of the other systems analyzed in this work (i.e.
electrochemical batteries, chilled water, ice and 100%-solar-heat-
driven thermochemical systems).
An additional significant point of this work relates to the com-
pressor operating conditions. The study shows that during the energy
storage phase, the compressor can work within a normal working
range, maintaining parameters such as the pressure ratio and discharge
gas temperature under safe limits. This means that the system can use a
standard ammonia reciprocating compressor designed for operating a
mechanical vapor compression refrigeration cycle.
From a techno-economic perspective, it is expected that ammonia
will be established in the market for small and medium refrigeration
systems [23]. This situation would lead to an increased availability of
ammonia-compatible components with competitive prices, which, in
turn, would have the potential to make the system analyzed in this
work a viable solution from an economic perspective. Moreover, the use
of the natural refrigerant ammonia reduces the environmental impact
[23], an increasingly important factor in any energy related system.
Acknowledgements
The corresponding author would like to acknowledge the Service de
l'énergie du gouvernement de la Polynésie française (SDE) for the financial
support of his PhD thesis and Miss Colleen Eleanor Leminski, for
proofreading the article.

Appendix A

Table 6 shows a list of the fifteen most inhabited islands in French Polynesia, which account for 94% of the total population. The table includes
population, surface area and solar global horizontal irradiation (GHI), this last being an important parameter used in solar energy potential as-
sessments. The average GHI for these fifteen islands of French Polynesia is 5.8 kWh/m2/day.
Table 6 was included in this work to combine three points (i.e. population, area, GHI) that were not found together in literature. Besides, it
emphasizes the fact that the islands found in French Polynesia range in population from nearly 200,000 inhabitants to a few hundred (and even
fewer inhabitants in the islands not shown on this table). This diversity, in conjunction with the high values of solar energy resources, could be
exploited for the development of smart electricity grids of the future.

References polynesie-francaise.html [accessed December 2017].

[1] Erdinc Ozan, Paterakis Nikolaos G, Catalão João PS. Overview of insular power
systems under increasing penetration of renewable energy sources: opportunities
and challenges. Renew Sustain Energy Rev 2015;52:333–46. http://dx.doi.org/10.
1016/j.rser.2015.07.104.
[2] Institut de la statistique de la Polynésie française, La Polynésie en bref 2016 (French
Polynesia at a glance 2016). [in French and English]. Available online at: http://
www.ispf.pf/docs/default-source/publi-pr/polybref-2016.pd [accessed December
2017].
[3] “Les enjeux énergétiques de la Polynésie française”, presentation given by the
“Service de l’énergie et des mines” of French Polynesia in 2012 [in French].
Available online at: http://docplayer.fr/19043030-Les-enjeux-energetiques-de-la-
[4] Électricité de Tahiti (EDT). “Les tarifs”. Available online at: https://agence.edt.pf/
faq.
[5] Energy Information Administration, U.S. Department of Energy, Electric Power
Monthly with Data for December 2017, Table 5.3. Average price of electricity to
ultimate customers. Available online at: https://www.eia.gov/electricity/monthly/
current_month/epm.pdf [accessed February 2018].
[6] Statistical office of the European Union, Eurostat, Electricity price statistics, Table 1:
Electricity prices, first half of year, 2015-2017. Available online at: http://ec.
europa.eu/eurostat/statistics-explained/images/6/6e/Electricity_prices_2017s1_
update_28_11_2017-final.xlsx [accessed February 2018].
[7] Agence de l'Environnement et de la Maîtrise de l'Énergie de la Polynésie française,
Plan climat-énergie; 2015, p. 106 [in French]. Available online at: http://www.
polynesie-francaise.ademe.fr/sites/default/files/files/mediatheque/maquette_pce_

254
F. Ferrucci et al.
2015-bdef_dble_page.pdf [accessed December 2017].
[8] Clastres Cédric. Smart grids: Another step towards competition, energy security and
climate change objectives. Energy Policy 2011;39(9):5399–408. http://dx.doi.org/
10.1016/j.enpol.2011.05.024.
[9] Iqbal M, Azam M, Naeem M, Khwaja AS, Anpalagan A. Optimization classification,
algorithms and tools for renewable energy: a review. Renew Sustain Energy Rev
2014;39:640–54. http://dx.doi.org/10.1016/j.rser.2014.07.120.
[10] Kyriakarakos George, Piromalis Dimitrios D, Dounis Anastasios I, Arvanitis
Konstantinos G, Papadakis George. Intelligent demand side energy management
system for autonomous polygeneration microgrids. Appl Energy 2013;103:39–51.
http://dx.doi.org/10.1016/j.apenergy.2012.10.011.
[11] Guerrero JM, Chandorkar M, Lee TL, Loh PC. Advanced control architectures for
intelligent microgrids—Part I: Decentralized and hierarchical control. IEEE Trans
Ind Electron April 2013;60(4):1254–62. http://dx.doi.org/10.1109/TIE.2012.
2194969.
[12] Guerrero JM, Loh PC, Lee TL, Chandorkar M. Advanced control architectures for
intelligent microgrids—Part II: Power quality, energy storage, and AC/DC micro-
grids. IEEE Trans Ind Electron April 2013;60(4):1263–70. http://dx.doi.org/10.
1109/TIE.2012.2196889.
[13] Salas V, Suponthana W, Salas RA. Overview of the off-grid photovoltaic diesel
batteries systems with AC loads. Appl Energy 2015;157:195–216. http://dx.doi.
org/10.1016/j.apenergy.2015.07.073.
[14] Colak Ilhami, Fulli Gianluca, Sagiroglu Seref, Yesilbudak Mehmet, Covrig Catalin-
Felix. Smart grid projects in Europe: Current status, maturity and future scenarios.
Appl Energy 2015;152:58–70. http://dx.doi.org/10.1016/j.apenergy.2015.04.098.
[15] Hatziargyriou N, Asano H, Iravani R, Marnay C. Microgrids. IEEE Power Energ Mag
2007;5(4):78–94. http://dx.doi.org/10.1109/MPAE.2007.376583.
[16] Palizban Omid, Kauhaniemi Kimmo. Energy storage systems in modern
grids—Matrix of technologies and applications. J Storage Mater 2016;6:248–59.
http://dx.doi.org/10.1016/j.est.2016.02.001.
[17] Luo Xing, Wang Jihong, Dooner Mark, Clarke Jonathan. Overview of current de-
velopment in electrical energy storage technologies and the application potential in
power system operation. Appl Energy 2015;137:511–36. http://dx.doi.org/10.
1016/j.apenergy.2014.09.081.
[18] Stitou Driss, Mazet Nathalie, Mauran Sylvain. Experimental investigation of a solid/
gas thermochemical storage process for solar air-conditioning. Energy
2012;41(1):261–70. http://dx.doi.org/10.1016/j.energy.2011.07.029.
[19] Kim DS, Infante Ferreira CA. Solar refrigeration options – a state-of-the-art review.
Int J Refrig 2008;31(1):3–15. http://dx.doi.org/10.1016/j.ijrefrig.2007.07.011.
[20] Ferreira Infante, Kim Dong-Seon. Techno-economic review of solar cooling tech-
nologies based on location-specific data. Int J Refrig 2014;39:23–37. http://dx.doi.
org/10.1016/j.ijrefrig.2013.09.033.
[21] Wang RZ, Xu ZY, Pan QW, Du S, Xia ZZ. Solar driven air conditioning and
255
Applied Energy 219 (2018) 240–255
refrigeration systems corresponding to various heating source temperatures. Appl
Energy 2016;169:846–56. http://dx.doi.org/10.1016/j.apenergy.2016.02.049.
[22] ASHRAE Position Document on Ammonia as a Refrigerant; February 1, 2017.
Available online at: https://www.ashrae.org/File%20Library/About/Position
%20Documents/Ammonia-as-a-Refrigerant-PD-2017.pdf [accessed February 25,
2018].
[23] Ciconkov Risto. Refrigerants: There is still no vision for sustainable solutions. Int J
Refrig 2018;86:441–8. http://dx.doi.org/10.1016/j.ijrefrig.2017.12.006.
[24] Mauran S, Prades P, L’Haridon F. Heat and mass transfer in consolidated reacting
beds for thermochemical systems. Heat Recov Syst CHP 1993;13(4):315–9.
[25] Bell Ian H, Wronski Jorrit, Quoilin Sylvain, Lemort Vincent. Pure and pseudo-pure
fluid thermophysical property evaluation and the open-source thermophysical
property library cool-prop. Ind Eng Chem Res 2014;53(6):2498–508.
[26] HKT Huber-Kälte-Technik software package, version 9/2009. Available online at:
http://www.hkt-goeldner.de/wp-content/uploads/2011/08/HKTSoftware09_
2009.exe [accessed December 2017].
[27] Rand David AJ, Moseley Patrick T. Chapter 13 – energy storage with lead–acid
batteries. In: Electrochemical energy storage for renewable sources and grid bal-
ancing, Elsevier, Amsterdam; 2015. p. 201–22. ISBN 9780444626165. http://doi.
org/10.1016/B978-0-444-62616-5.00013-9.
[28] ROLLSTM Pb flooded deep cycle batteries, renewable energy line. Datasheet
available online at http://rollsbattery.com/wp-content/plugins/rollsbatteries/
pdfs/Renewable.pdf [accessed December 22, 2017].
[29] TROJANTM Pb deep-cycle flooded battery. Model J200RE with bayonet cap.
Datasheet version J200-RE DS 2016_0921, available online at http://www.
trojanbattery.com/pdf/J200-RE_Trojan_Data_Sheets.pdf [accessed December 22,
2017].
[30] TESLA PowerWall 2 AC, 13.2 kWh. Datasheet available online at http://www.
energymatters.com.au/wp-content/uploads/2016/11/tesla-powerwall-2-datasheet.
pdf [accessed December 22, 2017].
[31] Bruno Arcuri, Catalina Spataru, Mark Barrett, Evaluation of ice thermal energy
storage (ITES) for commercial buildings in cities in Brazil, Sustainable Cities and
Society, vol. 29; February 2017. p. 178–92. ISSN 2210-6707. http://doi.org/10.
1016/j.scs.2016.12.011.
[32] Sebzali MJ, Rubini PA. The impact of using chilled water storage systems on the
performance of air cooled chillers in Kuwait. Energy Build 2007;39(8):975–84.
http://dx.doi.org/10.1016/j.enbuild.2006.11.004.
[33] Institut de la statistique de la Polynésie française, Répartition de la population par
île; 2017 [in French]. Available online at: http://www.ispf.pf/docs/default-source/
rp2017/repart_poplegale_iles_2017_v3.pdf [accessed December 2017].
[34] NASA Atmospheric science data center. Surface meteorology and solar energy.
HOMER data access. Insolation on horizontal surface; Jul 1983–Jun 2005. https://
eosweb.larc.nasa.gov/cgi-bin/sse/homer.cgi [accessed December 2017].