Applied Energy 239 (2019) 1371–1384

Contents lists available at ScienceDirect

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

Modelling and experimental validation of a small-scale trigenerative T

compressed air energy storage system
⁎
Mohamad Cheayba,b, Mylène Marin Gallegoa, Mohand Tazerouta, Sébastien Poncetb,
a
CNRS Joint Unit GEPEA, Department of Energy Systems and Environment, IMT Atlantique, Nantes, France
b
Mechanical Engineering Department, Université de Sherbrooke, Sherbrooke, QC, Canada

H I GH L IG H T S

• AThethermodynamic model of a simple trigenerative-CAES is developed.

• The model is validated experimentally.
• Technological
performances of the simple configuration are assessed.
• issues on the trigenerative CAES are highlighted.

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

Keywords: New advances in compressed air energy storage systems have been recently made especially regarding the use of
Compressed air energy storage heat generated from compression. On this basis, the concept of the trigenerative compressed air energy storage
Trigeneration (T-CAES) has recently been proposed. Many studies highlighted the feasibility and the benefits of this system to
Thermodynamic modelling be placed close to the energy demand.
Industrial pilot unit
The aim of this study is to examine a simple configuration of this system by a coupled experimental/mod-
Air motor
elling approach. This paper presents a detailed thermodynamic model of both the main components and the
whole system. An experimental bench is used to validate the model and to investigate the effect of the operating
parameters on the system efficiency and the model accuracy.
The model predictions are consistent with the experimental measurements during the charge, storage and
discharge phases. It has been found that the temperature drop across the pressure regulator should not be
ignored and is governed by the Joule-Thomson effect. Besides, it has been observed that the input temperature of
the air motor must be accounted for in the assessment of future improved configurations.
It was noted that the system efficiency increases significantly by adding the cooling and heating potentials.
However, the round trip efficiency remains low at 15.6%. Output values of the model are in good agreement
with the experimental results with an error less than 13.2%. The model can be applied as a basis for the per-
formance assessment of prospective configurations and improvements of trigenerative compressed air energy
storage.

1. Introduction Electrical energy storage (EES) is becoming a vital aspect to ensure

the balance between energy production and demand [2–5] and to deal
Growth in energy demand around the world would require an in- with the intermittent nature of solar or wind energy sources. Moreover,
crease in the consumption of fossil fuels [1]. In order to deal with in off-grid sites, EES appears as a key component to achieve a sufficient
economic and environmental problems related to fossil fuels, the major penetration of renewable energy resources [6,7]. As an example, in
current research focuses on integrating renewable energy resources Québec, where electricity is cheap (7 cents/kWh in average) and pro-
(RES) [2] and promoting a new energy transition [1]. However, the duced at 97% from hydroelectric dams [8], the major issues to be faced
inherent intermittent renewable energy generation is a central chal- are firstly the shift between production and demand for both cooling
lenging issue to be solved [2]. and heating and secondly the presence of remote communities in the

⁎
Corresponding author.
E-mail address: [email protected] (S. Poncet).

https://doi.org/10.1016/j.apenergy.2019.01.222
Received 3 August 2018; Received in revised form 15 November 2018; Accepted 26 January 2019
Available online 14 February 2019
0306-2619/ © 2019 Elsevier Ltd. All rights reserved.
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Nomenclature Subscripts

T temperature (°C) c compression

P pressure (bar) e expansion
Cp heat capacity (kJ kg−1 °C−1) i stage number i or heat exchanger i
r ideal gas constant (J K−1 kg−1) out output
n polytropic coefficient in input
βc compression ratio d expansion valve
βe expansion ratio el electrical
Ẇ power (kW) m mechanical
ṁ mass flow rate (kg s−1) th thermodynamic
ε heat exchanger effectiveness th,m thermodynamic to mechanical conversion
ms masse stored (kg) l,tes cold thermal energy storage
V volume (m3) h,tes hot thermal energy storage
t time (s) ch charge
Nc number of compression stages dis discharge
Nres number of storage tanks a air
Nu Nusselt number w heat transfer medium
Ra Rayleigh number res air reservoir
h heat convection coefficient (W m−2 K−1) max maximum pressure
H height (m) min minimum pressure
e thickness (m) CAES compressed air energy storage
D internal diameter (m) A-CAES adiabatic CAES
Rth thermal resistance (W−1 m K) T-CAES trigenerative CAES
λ thermal conductivity (W m−1 K−1) TES thermal energy storage
μT Joule Thomson coefficient (°C/bar) AM compressed air motor
η efficiency I.DIS.S initial discharge state
COP coefficient of performance
ηg comprehensive efficiency

North of the province, which do not have access to the main electrical compressed air energy storage (A-CAES) has been proposed. The main
network. In these communities, replacing diesel power station by EES feature of A-CAES is to recover heat produced during compression to
appears as a particularly interesting solution, due to the large amount of preheat the air during expansion, hence it eliminates the need for fossil
available wind or solar. fuels and promotes a cleaner technology. The main demonstration
Several EES such as pumped hydroelectric (PHS), compressed air projects are the ongoing European project “ADELE-ING” [14], the pilot
energy storage (CAES), thermal energy storage (TES), chemical or plant “TICC-500” designed by the Chinese Academy of Sciences [10]
electrochemical batteries, flywheel, capacitor and supercapacitor have and recently the pilot scale of underground A-CAES developed in
been developed so far. Chen et al. [4] and Luo et al. [5] carried out Switzerland [15,16].
detailed reviews of the main characteristics of each technology. Each CAES is mainly applied to large-scale applications (tens or hundreds
system has certain advantages and limitations based on efficiency, en- of MW). Recently, CAES using overground pressure vessel reservoir has
ergy density, power range, time of response and investment cost [5]. been proposed [7,17] which enables to apply CAES at a smaller scale
The suitable technology depends mainly on its application. and closer to energy demand [18]. Besides, it has been demonstrated
Among EES technologies, PHS is the most widely implemented. It is that CAES enables the cogeneration of heat and cooling demand so that
technically mature with good efficiency (70–85%), long storage time the concept of T-CAES has been introduced [18–30].
and a more than 40 years lifetime [5]. Besides PHS, compressed air For small scale applications, electrochemical batteries remain the
energy storage (CAES) has a strong potential due to its high reliability most popular technology for EES and cogeneration plants such as ab-
[9,10], low capital and maintenance costs and good part-load perfor- sorption chiller for heating/cooling production. By contrast, those
mance [9,11]. In their recent review, Wang et al. [12] stated that PHS technologies entail using toxic materials (such as LiBr in absorption
and CAES technologies exhibit rather similar characteristics in terms of chiller), chemical wastes (Li in batteries) that remain for a long time
scale and cost, while CAES is not so restricted to suitable geological [4,19] and batteries have been suffering from low lifetime [4]. Faced
locations. with this situation, CAES is considered as a very promising alternative
The thermodynamic process in a conventional CAES (or Diabatic solution to batteries for energy storage in remote areas [7,20] and
CAES) may be described as follows: during the period of excess energy especially small scale Trigenerative-CAES (T-CAES) to replace diesel
production, ambient air is compressed via a multi-stage compressor power stations used by small housing groups. The literature on T-CAES
with intercooling to increase the compression efficiency and the energy is scarce [18–30] and mainly limited to large scale systems. Thus, it is
density [7,9]. Compressed air is stored in underground reservoirs such worthwhile to put forward more studies on small scale T-CAES. The
as saline dome caverns, porous rock formation or old mines [11,13]. present work is besides part of a joined research project undergoing in
During the period of high energy demand, the mechanical energy is IMT Atlantique (France) and Université de Sherbrooke (Canada) dedi-
converted again into electrical energy by expansion. Stored air is re- cated to the implementation of small scale T-CAES in a remote com-
leased, heated by means of combustion before expanding into the gas munity in the North of Québec.
turbine to generate electricity. The main drawback of D-CAES is the low Despite the difference between A-CAES and T-CAES in terms of
round-trip efficiency which is around 42% for the Huntorf’s plant preheating energy, technical and technological aspects, the thermo-
(Germany) and 54% for the McIntosh’s plant (USA) [9]. dynamic and thermal modelling and analysis approach are similar.
In order to overcome this limitation, the concept of adiabatic Start with A-CAES, extensive studies have been conducted in recent

1372
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

years. A comprehensible and detailed literature review on the different coefficient COP of 0.62. Facci et al. [18] proposed and studied a con-
concepts of CAES is conducted by Budt et al. [21]. Kim et al. [22] figuration of T-CAES to fulfill the needs of a power peak shaving for
carried out an energy and exergy analysis of different configurations of energy users with an electric power of 18 kW. The results proved the
CAES to characterize each system. Hartman et al. [23] concluded that adaptability of the system for civil applications with an efficiency of
the efficiency of the polytropic configuration is at 60% higher than the 30%, besides they dissected the effect of design and operating para-
adiabatic one of 70%. Budt and Wolf [24] pointed out that the tem- meters via a sensitivity analysis, and addressed also the technological
perature level of TES does not affect the ideal round-trip electrical ef- issues and technical constraints. Lv et al. [29] applied a thermodynamic
ficiency and they proposed a novel A-CAES with low-temperature model to evaluate the energy and economic performance on a monthly
thermal storage which is lower than 200 °C. Following this study, Luo basis of T-CAES based on electrical energy peak load shifting for a hotel.
et al. [11] modelled the components of the system and mainly in- The results, which are not confirmed experimentally, show that the
vestigated on the effect of the efficiency and number of stages of trigeneration system works efficiently at relatively low pressure, and
compressors and turbines and the heat transfer rates of heat exchangers. the efficiency can reach 76.3% at 15 bars. The T-CAES technology can
The authors found a maximum value of 60% for the electric efficiency. represent an annual monetary cost saving of about 53.9%. Liu et al.
Grazzini and Milazzo [25] identified a set of criteria optimizing the [30] introduced a configuration of T-CAES and focused on discharge
design of A-CAES with focus on heat exchanger parameters. The first process formed by a scroll expander. They measured its polytropic ex-
experiment on A-CAES was published by Wang et al. [10], their system ponent as a function of the ambient temperature and discussed the ef-
is at medium scale of 500 kW input and 315 kW power output, it was fect of expansion ratio and maximum storage pressure on the perfor-
designed with 5 compression stages, 3 expansion stages, pressurized mances of the system. Li et al. [19] presented a novel tri-regenerative
water as TES at a temperature of 120 °C and an overground steel air system able to meet the end’s user demands in terms of electricity,
reservoir as compressed air storage. Their experimental results showed cooling and heating of a small office building in Chicago with a max-
an electric efficiency of 22.6% which is lower than the designed of imum power of 52 kW. A global storage electric efficiency of 50% was
40.7%. In addition, the dynamic variation of the most relevant oper- achieved.
ating parameters were reported. A focus on the underground pilot is The different values of electric efficiency ranging from 30% to 57%
reported in Refs. [14,15]. Geissbühler et al. [15] experimentally eval- in the above studies relies on how the heat stored is used and the choice
uated the effect of cycle of air charging/discharging on an unused of the design parameters such as the number of compression stage, the
tunnel as a compressed air reservoir. Becattini et al. [16] studied nu- compression and expansion ratio, heat exchanger effectiveness…etc.
merically and experimentally a combined sensible/latent TES system Those results show that this generation of CAES has a good efficiencies
composed of stainless-steel tubes encapsulating an Al–Cu–Si alloy. and has a competitive performance index compared to conventional tri-
With regard to trigenerative-CAES, T-CAES has been studied at large generation systems [19].
scale with the intent to provide district heating and cooling. In this Most of the literature mentioned above related to CAES focuses on
scope, Arabkooshar et al. [26] applied this concept to a wind farm of theoretical modelling, parameter optimization and system configura-
300 MW power capacity. It was proven the potential of their proposed tions of T-CAES or A-CAES (see the review of Wang et al. [12]).
configuration to support district heating and cooling networks. The Available experimental studies [10,14,15] concentrate on validating
values of power to power, power-to-cooling and power-to-heat effi- numerical models of sub-systems or showing the real behavior of the
ciencies of this system were 30.6%, 32.3% and 92.4%, respectively. system. As an illustration, Geissbühler et al. [15] and Becattini et al.
Han and Guo [27] derived a configuration from A-CAES (with 80 MW [16] focused on the experimental validation of numerical models de-
power input and 100 MW power output) enabling to produce cooling veloped for compressed air in cavern and thermal energy storage. In the
energy from the last stage and distribute the excess of heat as heating second place, experimental data reported by Wang et al. [10] are in-
energy. A variable expansion ratio was proposed to increase the electric sufficiently correlated with the developed models. Moreover, these
efficiency which attained 44.5%. experimental data concerns medium scale CAES, whereas very few
Furthermore, Jannelli et al. [19] proposed a T-CAES for a small experimental studies are focused on micro or small scale CAES. The
scale stand-alone photovoltaic power plant with 3.7 kW electric com- present paper is an attempt to fill this gap.
pressor input and 1.7 kW expander electric output. The system was able In this work, an accurate causal steady-state model (input/output
to satisfy the energy demand in terms of electricity and cooling for a modelling) of a simple configuration of T-CAES is developed with a
radio base station with an electric efficiency of 57% and a performance focus on airside components. Experimental small-scale pilot (4 kW

Fig. 1. Scheme of the charge phase.

1373
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

nc, i − 1
compressor’s power and 2 kW expander’s power) is described. Then, the nc , i
̇ , c, i = γ − 1
Wth .ṁ c .Cp, a. Tin, c, i. ⎡
nc, i
− 1⎤
model is validated experimentally by paying a careful attention to each γ nc , i − 1 ⎢βc ⎥
⎣ ⎦ (3)
component. To the best of the author’s knowledge, the present work is
the first combined thermodynamic/experimental approach for a small Input temperature of each stage depends only on the previous
scale CAES, which is a necessary step towards the implementation of compression stage output temperature and previous heat exchanger.
CAES in remote communities. Heat exchanger effectiveness is defined by:
Tout . c, i − Tin, c, i + 1
εi =
2. Modelling of the compressed air energy storage process Tout , c, i − Tamb (4)

By reformulating Eq. (4):

CAES principle is divided into three phases: (1) charge or com-
pression phase, (2) storage phase and (3) discharge or expansion pro- Tin, c, i + 1 = Tamb + ΔTi (5)
cess. Since the optimal system configuration is beyond the scope of this
where ΔTi = (1 − εi )(Tout , c, i − Tamb) (6)
article and we focus on the thermodynamic modelling and its experi-
mental validation, the proposed configuration remains relatively Consequently, the input temperatures of the compressors are equal
simple. It is based on the early commercialized configuration proposed to the ambient temperature plus a value ΔTi , which expresses the effect
by Energetix group (UK) [31] and that studied by Liu and Wang [30]. of heat exchanger effectiveness.
Heat removal from the compression is expected to fulfil the heating Adding the electrical efficiency and mechanical efficiency, and re-
demand rather than reheat the expander during the discharge phase. placing Eq. (6) in (3), the total electric power consumption forNc stages
Figs. 1 and 2 show the scheme of the proposed configuration during the is calculated as:
charge and discharge phases, respectively.
Nc
The six main components of T-CAES are compressors, heat ex- 1 γ − 1 nc , i nc, i − 1
Wel̇ = . C . ṁ c . ∑ ⎛ (Tamb + ΔTi − 1).βc,i nci − 1⎞
changers, storage tank, regulation valve, expander and thermal energy ηe . ηm . p, a i=1
γ nc , i − 1 ⎝ ⎠
storage. In this section, we present a detailed model of each component
(7)
(except for TES).
A steady state approach is considered during the charge and dis- However, the electric power is delivered by renewable energy re-
charge phases. The perspective of this study is to use this model for sources and it is imposed as an input parameter, so that the mass flow
assessment of performances of prospective configurations and im- rate is the output value. It is derived from Eq. (7) as:
provements of T-CAES. A steady state approach allows a fast calculation ηe . ηm Wel̇
time so it can be easily use in an optimization procedure. Furthermore, ṁ c = nc, i − 1
Nc γ − 1 n
the next section will show that results are in good agreement with the Cp, a. ∑i = 1 γ n c−, i 1 ⎛ (Tamb + ΔTi − 1). (βc,i nci − 1) ⎞
c, i ⎝ ⎠ (8)
current experimental performances. In order to reduce the modelling
complexity, the following assumptions are introduced:
2.1.2. Air tank
1. Pressure losses at the admission, through the discharge valve, and in The minimum pressure of the air storage is limited by the operating
the heat exchangers are not considered. pressure of the air motor, thus a residual amount of air remains in the
2. Compressed air is considered as a perfect gas, except in the reg- reservoir:
ulating valve.
Pmin Vres
3. Constant global compression ratio is considered. mresidual = Nres
r . Tres (9)

The air mass that could be stored is limited by the maximum pres-
2.1. Modelling of the charge phase sure allowed in the storage tank and calculated by Eq. (10) using the
ideal gas law. Furthermore, during the charge phase it is assumed that
2.1.1. Compressors the temperature into the tanks is uniform. In other words, the tem-
Volumetric compressors are suitable for limited mass flow rates and perature into a tank is the same as at the entrance of the tank:
high-pressure ratios [32,33], and so are the most suitable compressors Tres = Tin, res .
for our system. As mentioned above, multi-stage compression is needed
to reduce the specific energy consumption for a certain air mass to be (Pmax − Pmin ).Vres
ms = Nres
stored. r . Tin, res (10)
Though dynamic compressors whose efficiency reflects the internal
The time required to completely charge the reservoir is calculated
heat and the isentropic term are the most commonly used, volumetric
by:
compressors are characterised by some amount of heat lost to the en-
vironment which is implicitly expressed by a polytropic coefficient
nc < γ .Thus, the output temperature of each stage is written as:
nc, i − 1
Tout , c, i = Tin, c, i .βc,i nc, i
(1)

where βc is the compression ratio.

Thermodynamic work of one compression stage is calculated by
integrating d (P. V ) throughout the thermodynamic cycle which gives:
nc , i
Wth, c, i = .mc . r . [Tout , c, i − .Tin, c, i]
nc , i − 1 (2)

The thermodynamic power of one compression stage is deduced

from Eq. (2) by introducing Eq. (1) and expressing the ideal gas con-
stant as a function of heat capacity: Fig. 2. Scheme of the discharge phase.

1374
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

ms by the ideal gas law:

tch =
ṁ c (11)
−Nres .t
T (t ) = Tamb + (T(t = 0) − Tamb ) e ms . Cpa. Rth (15.a)
2.1.3. Heat exchangers However, the ambient temperature may vary with time. In this case, by
It is known that countercurrent heat exchangers have a better per- discretizing time, the new solution gets:
formance than concurrent heat exchangers so this type of heat ex-
−Nres .t
changers has been selected for this study. The thermal energy generated T(t(j) ) = Tamb (t(j − 1) ) + (T(t(j − 1) ) − Tamb (t(j − 1) ) ) e ms . Cpa. Rth (j − 1) (15.b)
from compression is transmitted from the heat source (compressed air)
to the heat sink, which is the heat transfer medium. The energy balance The average thermal resistance of the natural boundary layer is
equation for each heat exchanger i writes: defined as:
̇ , i = ṁ c . Cp, a. (Tout , c, i, −Tin, c, i + 1) = ṁ h, w . Cp, w. (Tout , ch, i − Tltes )
Qch (12) 1
Rtha =
hπ (D + 2d ) H (16)
where w designates the heat transfer medium and the right side of the
equation represents the heat absorbed by it. Tltes denotes the tempera- The average heat convection coefficient for a cylinder is defined as:
ture of the incoming cooling fluid considered at the ambient tempera-
Nu
ture. h = λa
H (17)
Eq. (12) indicates that the thermal power can be computed from the
air side. The total thermal energy that could be stored is calculated by: where H is the height of the cylinder and Nu is the Nusselt number
i = Nc
expressed by the empirical formula reported by Bejan [34].
Qs = ∑ ṁ c . Cp, a. (Tout , c, i − Tin, c, i + 1). tch
4 ⎡ 7. RaH . Pr ⎤
0.25
4. (272 + 315. Pr )H
i=1 (13) Nu = +
3⎢⎣ 5. (20 + 21. Pr ) ⎥
⎦ 35. (64 + 63. Pr ). D (18)

2.2. Modelling of the storage phase To sum up, Fig. 3 presents the block diagram of the mathematical
model of the charge phase.
It should be noted that the reservoir insulation is not expected to
account for the increasing energy density for a given storage pressure. 2.3. Modelling of the discharge phase
Heat transfer between the environment and the compressed air occurs
for any phase. Consequently, the air storage temperature and pressure In the same way, as for the charge phase, the discharge time can be
variation are accounted for only during the storage phase. The fol- calculated by:
lowing assumptions are made: ms
tdis =
ṁ e (19)
1. The heat capacity of the compressed air is constant, independently
of the pressure variation.
2. The temperature gradient inside the storage tank is negligible. 2.3.1. Pressure regulator
3. The thermal resistance of the wall thickness is negligible compared The physical law that governs the pressure reduction (throttling of
to that due to natural convection. compressed air) is the Joule-Thomson Law on which the enthalpy re-
mains constant and the internal energy decreases as:
Heat transfer to the environment induces a decrease of the air sto-
ΔU = ΔPV (20)
rage temperature, which results in:
ms Cpa dT The majority of CAES modelling has considered compressed air as
(Tamb − T )
= an ideal gas throughout this component which leads to constant tem-
Nres dt Rth (14)
perature. However, this assumption is no longer true when the pressure
The first order differential Eq. (14) is solved. The time-temperature reduction magnitude is important. Temperature variation may be ex-
variation is expressed by Eq. (15.a) and the pressure is computed simply pressed by the Joule-Thomson coefficient:

Fig. 3. Block diagram of the model for the charge phase.

1375
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

∂T follows:
μT = ⎛ ⎞
⎝ ∂P ⎠h (21) 1 − ne
Tout , e = Tin, e. βd ne (26)
In order to find the isenthalpic curves and evaluate their slopes
̇ , e = γ − 1 ne .ṁ e .Cp, a. T . ⎡1 − βe ne e ⎤
1−n
given by μT , experiments were conducted by Hoxton [35]. This last Wth in, e ⎣ ⎦
author covered a wide range of pressure (from 25 atm to 150 atm). His γ ne − 1 (27)
semi-empirical approximation is given by [35]: Pin, e
where βe = Pa
.
ΔT 50.1 + 0, 0297. P 14830 − 1, 674. P 366000 − 19093. P In order to account for the deviation between real and ideal ther-
= + +
ΔP T T2 T3 modynamic cycles and also mechanical losses, the thermodynamic or
− (0.122 − 0.0000157. P ) (22) pneumatic mechanical efficiency is introduced:

Eq. (22) can be written as: Wm ̇ ,e

ηth, m = ηm . ηth =
̇
Wth, e (28)
ΔT
= 2. a. P + b
ΔP (23)
Mechanical shaft power and mass flow rate variations versus shaft
where rotation speed are generally provided for each operating pressure by
the manufacturer. Thus, shaft power curves versus mass flow rate can
1 ⎛ 0, 0297 1, 674 19093 be deduced. By polynomial interpolation of the function Wm ̇ , e (ṁ , Pin ),
a= ⎜ − − + 0.0000157⎞⎟
2 ⎝ Tin, d Tin, d 2 Tin, d3 ⎠ the thermodynamic-mechanical efficiency is obtained via Eq. (28).
Finally, accounting for the electrical efficiency, the electrical power
50.1 14830 366000 can be calculated as:
b= + + − 0.122
Tin, d Tin, d 2 Tin, d3
Wel̇ , e = ηel, e .ηth, m . Ẇ th, e (29)
Temperature variation versus pressure is obtained by integration of
Eq. (23). Fig. 4 shows a block diagram of the discharge phase model. Input
pressure and temperature of the expansion valve are equal to those of
T (P ) = a. P2 + b. P + c (24)
the air tank and decrease with time. However, as mentioned above, the
where c accounts for the family of isenthalpic curves and is calculated present study is limited to a time-independent model. Therefore the
by: model of each component is validated separately by experimental tests
and an average value of the input temperature of the air motor is taken
c = Tin, d − a. Pin, d 2 − b. Pin, d (25)
into account to validate the whole model for the discharge phase.

2.3.2. Air motor (AM) 2.4. Performance criteria

For small-scale applications, the choice of the best expander tech-
nology among the volumetric (piston, scroll) [36,37] or the even re- T-CAES is not a simple technology, it has several roles such as an
cently developed high-speed axial turbines [38] still remains an open electrical storage technology and meets heating and cooling demands.
research issue. Scroll machines exhibit higher performances [9,36], In order to assess this system, the round trip electrical efficiency or
whereas they are restricted to small pressure ratios without handling simply the electrical efficiency and other performance index have been
high values of pressure and temperature [38]. On the other hand, piston used. They are defined as follows:
expanders may be advantageous for CAES application where high-
Eout , el Wel̇ , e tdis
pressure input would be expected thanks to their low prices. In this ηel = =
Ein, el Wel̇ , c tch (30)
perspective, the present study tested piston expander regardless of the
fact that commercially available piston air motor (AM) are designed for Qs + Qcool + Eout , el
maximum pressure input of 8 bar. COPg =
Ein, el (31)
By analogy with compression, the output temperature and the
thermodynamic power of the ideal expansion cycle are obtained as where Qcool is the cooling energy that can be produced by the exhaust

Fig. 4. Block diagram of the model for the discharge phase.

1376
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

air of the air motor. For air conditioning, the cold air flow is mixed with
the air flow at the ambient temperature so that the resulting air flow
reaches the comfort temperature of the room. Thus, cooling energy is
calculated by:
Qcool = ṁ e . Cp, a. (Tcomfort − Tout , e ). tdis (32)

with Tcomfort = 21°C .

However, the coefficient of performance defined above over-
estimates the efficiency of our trigenerative system [19]. Thus the
comprehensive efficiency defined by Li et al. [19] is also adopted.
Qheat Qcool
COPref , heat
+ COPref , cool
+ Eout , el
ηg =
Ein, el (33)

COPref , heat and COPref , cool are the performance coefficients of con-
Fig. 6. PID of the CAES pilot unit - IMT Atlantique, France.
ventional heat pumps for heating and cooling, respectively. As a first
approximation, they are considered equal to 4 and 3 respectively as
proposed by [28]. expanded through an air motor (5) coupled to a DC generator of 2 kW
The first two terms of the numerator of Eq. (33) show the equivalent maximum power. The admission air motor pressure could vary in the
electrical energy needed to produce the heating and cooling energy by a range [2–8] bars (absolute values). The pressure at the entrance of the
conventional heat pump. Hence, the index named by Li et al [18] as air motor is controlled manually via a pressure regulator (4). The pilot
comprehensive efficiency reflects the ratio between the two following is equipped with sensors for continuous monitoring of power, pressure,
terms: (1) the sum of the output electrical energy and the equivalent temperature and mass flowrate (5).
electrical energy for heating and cooling, (2) the electrical energy input. Experimental results are illustrated in Figs. 7–12. Experimental
We prefer to employ the term “global equivalent electrical efficiency” uncertainties associated with the measurements have been determined
instead of “comprehensive efficiency”. We put the emphasis on this from the manufacturer’s data of the temperature captors ( ± 1.6 °C),
efficiency, which is the most important in order to compare the T-CAES pressure captors (between 0.05% and 0.1% of the measure range), and
technology with other storage technologies such as electrochemical airflow captors (0.08% of the measure range).
batteries.
3.2. Compressor chain
3. Material and methods
The piston reciprocating compressor is a three-stage compressor
This section begins with a general layout of the experimental bench. arranged in a W form, cooled by air-air heat exchanger between two
Then, a detailed description of each component and the experimental consecutive stages. The cooling air flow is forced by one ventilator. The
procedure and operating parameters are presented. pressure ratios of the first 2 stages are 7.6 and 5.6 respectively but the
pressure ratio of the third stage varies from 4.08 to 7.2 according to the
3.1. Experimental setup state of charge. It should be noticed that for a fixed global pressure
ratio, the optimal distribution is symmetrical [39], which is different
To study the characteristics of the CAES system, an industrial pilot from the manufactured compressor. This is due to a variable global
unit has been built in GEPEA laboratory at the IMT Atlantique, France. pressure ratio. Air humidity is ignored because the compressor is
The process instrumentation diagram and photograph are shown in equipped with a dehumidifier. Table 1 shows the main characteristics of
Figs. 5 and 6. the compressor chain.
The rated power of the compressor driven-motor (1) is 4 kW with a During the charge phase, the values of pressure, temperature, power
maximum global compression ratio of 325. Compressed air is stored in and mass flow rate are recorded every 20 min, which is accurate enough
six reservoirs (2) with a capacity of 50 L each. The air stored is regarding their temporal variations.

Fig. 5. Photograph of the CAES pilot unit - IMT Atlantique, France.

1377
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Fig. 7. Pressure variations during the charge phase.

The effectiveness of the heat exchanger is calculated using Eq. (4), 3.3. Storage reservoir
and the polytropic coefficient, after algebraic manipulation, is calcu-
lated as: The total capacity of the six tanks is 0.3 m3, and the maximum al-
lowed pressure is 350 bar. Nevertheless, for security purpose, the
1 compressor stops automatically as soon as the air pressure reaches
nc =
Tout , i ⎞
ln ⎛
⎜ ⎟ 315 bar. A high value of the storage pressure is chosen to account for
⎝ Tin, i ⎠
1− the fact that to store a given air mass, the cost consequences of oper-
ln(βc ) (34)
ating at low maximum pressure is more severe than at too high pressure
Unfortunately, we do not know the exact mechanical efficiency ηm [17]. The minimum operating pressure of the air storage is chosen at
and electrical efficiency ηe. The latter depends on the nominal power 5 bar to help and facilitate the control of the pressure manually.
and standard efficiency of the motor. The electrical efficiency usually During the storage phase, the air temperature and pressure inside
ranges between 0.75 and 0.83 for a 4 kW motor as proposed by [40]. An the tank are collected to validate the model.
average value of 0.8 is then considered here. Similarly, a value of 0.9 is
taken for the mechanical efficiency which is estimated in the range of 3.4. Air motor and pressure regulator
[0.88–0.95] by [33]. It is important to mention that the different
parameters displayed in Table 1 are obtained for a fixed pressure ratio. Three types of experiments were carried out:

– Firstly, the pressure input of the air motor (AM) was manually

Fig. 8. Input and output temperatures of each component during the charge phase.

1378
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Fig. 9. Air mass flow and compressor power during the charge phase.

monitored and regulated to 5 bar throughout the discharge phase. the air motor is determined by:
The procedure for obtaining the physical variables is the same as for
1
the charge phase with a record frequency of 1 min, which is ac- ne = ln(βe )
ceptable compared to the discharge time (around 2 h). 1+ Tout , e
ln( )
– Secondly, we focused on the temperature change across the throt- Tin,e (35)
tling valve. It was rigorously accounted and acquired once the
pressure of the air tanks varies, allowing a validation of the model.
– Finally, the impact of pressure, temperature and mass flow are at the 4. Results and discussions
inlet of the AM on its efficiency was investigated.
Model and experiment results are compared and summarised in
For these two later experiments, the operating conditions in terms of Tables 3 and 4. It can be seen that the reservoir is completely charged at
pressure and temperature are obtained according to the state of dis- a maximum pressure of 181 bar after 4 h and 20 min, compressors
charge of the stored air while the valve is constantly regulated. The consume 13.72 kWh with a heating energy that can be stored of 5.72
mass flow rate or in other words the speed of rotation of the air motor kWh. The air motor provides 0.437 kW for one hour and 8 min while
shaft is controlled by changing the load electric resistance. the reservoir is completely discharged. The electrical round trip effi-
It is noticeable that a significant unavoidable amount of heat ciency is very low (3.6%), whereas the comprehensive efficiency or the
transfer occurs from the environment to the air flowing in the flow global equivalent electrical efficiency is higher and equal to 15.16%
meter, inducing an unpredicted value of the air temperature at the owning the values of heating energy and cooling energy (0.8 kWh).
entrance of the air motor. For validation purpose, an average value of The parameters predicted by the model such as the air mass flow
this temperature is taken into account. rate in the compressor, the charge and discharge times and also the
The electrical efficiency of the generator used is 0.83. Table 2 shows electrical and comprehensive efficiencies are in good agreement with
the characteristics of the discharge phase. The polytropic coefficient of the experimental results. The maximum error is 13.1%, which is cal-
culated by:

Fig. 10. Pressure and temperature of the stored air during the storage phase.

1379
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Fig. 11. Pressure variations during the discharge phase.

|model result − experimental result| instabilities when opening and closing the exhaust valve.
error (%) =
Maximum (model result , experimental result ) (36) According to Fig. 8, the temperatures of each component and of the
stored air increase, then stabilize. This transitional period for the
Finally, a detailed analysis for the charge, storage and discharge compressor and heat exchangers lasts around 40 min and is due to the
phases was conducted to study and illustrate the behaviour of each heat exchange taking place with the environment.
component. The stabilization duration of the compressed air temperature is
longer (around two hours) due to the fact that the initial amount of air
4.1. Charge phase (at 8.8 bar and ambient temperature) acts to cool the stored air. The
constant value of the temperature after 2 h proves that the heat ex-
The charge phase was conducted at the room temperature of around change with the environment could be neglected compared to the mass
22 °C. Figs. 7 and 8 present respectively the input and output pressure heat transfer through the charge air flow, as proposed in the model.
and temperature variations during the charge phase. Two operation Similar behaviours for the temperature and pressure profiles can be
periods can be identified according to the global pressure ratio varia- found in the experimental results reported by Wang et al. [10] though
tions. the stabilization of the air temperature inside the reservoir was not
During the first period, the pressure output of each stage is constant achieved in their case. This can be explained by a more important initial
with a small fluctuation at the outlet of the third stage due to

Fig. 12. Input and output temperatures of each component during the discharge phase.

1380
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Table 1 Table 3
Experimental characteristics of compression chain. (*) refers to calculated va- Experimental and model results of the charge phase.
lues.
F.CH.S @ 181 bar F.CH.S @ 315 bar
Parameter 1-stage 2-stage 3-stage Pipe Storage tank First period Second period
connection
Parameters Experimental Model Error (%) Experimental
Inlet pressure [bar] 1.013 7.9 44.4 181–315
Outlet pressure [bar] 7.7 44.4 181–315 Air flow rate [kg/h] 14.4 15.09 4.6 14.18
Pressure ratio (*) 7.6 5.6 4.08–7.2 Charge time [h] 4.3 3.79 11.9 7
Inlet temperature [°C] 22 62.0 58.0 43.3 33 Consumed electric 13.72 12.09 11.9 22.86
Outlet temperature 118.0 198.7 152.0 33 energy [kWh]
[°C] Potential heating energy 5.27 4.87 7.6 8.61
Mechanical efficiency 0.88–0.95 (0.9) [kWh]
Electric efficiency 0.75–0.83 (0.8)
Shaft Power [kW] 3.19
Polytropic coefficient 1.16 1.25 1.22 Table 4
(*)
Experimental and model results of the discharge phase.

Heat exchangers I.DIS.S @ 181 bar I.DISCH.S @

Parameter 1-stage 2-stage 3-stage Experiment n°1 300 bar
Effectiveness, ε (*) 0.583 0.796 0.836 Experiment n°2
Parameters Experimental Model Error (%) Experimental

air mass in the study of Wang et al. [10] (around 38% of the total mass Average power [kW] 0.437 0.432 1.1 0.435
stored). Discharge time [h] 1.13 1.16 2.6 1.66
Produced electric energy 0.49 0.50 1.5 0.72
As it can be seen on Fig. 9, the absorbed electric power is constant as [kWh]
well as the air mass flow (apart from weak fluctuations). The air Cooling energy [kWh] 0.80 0.78 1.9 1.14
pressure in the reservoir increases linearly from 8.8 bar to 181 bar in 4 h Round trip electric 3.6 4.1 13.1 3.2
and 20 min, which demonstrates the adequacy of the perfect gas rela- efficiency (%)
COP 0.48 0.51 6.0 0.46
tion to model the stored air.
Comprehensive 15.16 16.4 7.5 14.24
According to Table 3, the values of the air flow and charge time efficiency (%)
predicted by the model are accurate with an error which does not ex-
ceed 12%. These errors are caused by the pressure losses not being
accounted, uncertainties regarding the mechanical and electric effi- representing an error of 1.6%.
ciencies and mostly by the transitional regime which is not considered It should be noticed that the change of heat capacity with pressure
by the model. was introduced and investigated and it is found that it reduces insig-
During the second period, the pressure ratio of the first stage re- nificantly the percentage of error. Hence, potential error sources could
mains constant and that of the second stage slowly increases. At the be identified as:
same time, this ratio increases according to the reservoir pressure for
the third stage (Fig. 7). – Storage tanks are closely placed and natural boundary layers may be
As shown in Fig. 8, the outlet temperatures of the last two com- mixed contrary to the model assumptions.
pression stages raise which implies an increase of the outlet tempera- – Real air properties and humidity (even if partial humidification
ture of all HEX since the efficiency of each HEX remains almost con- occurs) are not taken into consideration.
stant.
The driven motor is asynchronous, the increase in global pressure
4.3. Discharge phase
ratio means an increase of the shaft torque and a slow decrease in the
rotation speed. Thus, as it can be observed in Fig. 9, the air mass flow
The discharge phase is examined at the room temperature of
slightly decreases until 13.6 kg/h and the input power increases
reaching a maximum value of 3.6 kW for an air pressure of 315 bar. 26.4 °C. As outlined previously, the input pressure of the air motor is
regulated continuously at 4 bar (gauge pressure), its fluctuation can be
observed in Fig. 11. In practice, the air pressure does not expand to the
4.2. Storage phase atmospheric pressure. Consequently, the pressure output of the AM
remains always above this level (Fig. 11). The air reservoir pressure
The experimental and modeling results are illustrated in Fig. 10 for decreases almost linearly from 301 bar to around 9 bar during this
the storage phase. The predicted temperature and pressure variations phase which takes 1 h and 40 min.
are in line with the measured ones. The stored air temperature de- Fig. 12 displays the temperature variations with time. Two types of
creases exponentially and achieved the ambient temperature after a profiles can be observed. During the first 10 min, the outlet tempera-
storage duration of around 5 h. The pressure drop is similar and reaches tures of the expansion valve and consequently of the AM decrease
296 bar which is slightly lower than the predicted one (301 bars), which is inconsistent with the Joule-Thomson law, saying that the

Table 2
Experimental characteristics of the discharge phase. (*) refers to calculated values.
Parameter Expansion valve Flow meter Air Motor

Inlet pressure [bar] 300–9 (max 300) 5 5

Outlet pressure [bar] 5 5 1.031
Inlet temperature [°C] 24 –22 < T < 14 −5 < T < 24 (10 on average)
Outlet temperature [°C] –22 < T < 14 −5 < T < 24 (10 on average)
Mass flow rate [kg/h] 49.28 49.28
Polytropic coefficient 1.1 (*)

1381
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

temperature should rise when the pressure input increases. This var- 4.4.2. Air motor
iation can be explained by the fact that the exhaust air serves to cool the Since the polytropic coefficient is associated with the amount of
stored air as it can be noticed from its temperature profile. Afterwards, heat transfer from the environment to the expanding air, it depends on
the temperature rises as expected. several factors such as the ambient temperature and the input pressure
Concerning the first experiment, the initial pressure was set to181 and temperature. The effect of the ambient temperature is not explored
bar, despite the fact that the average air mass flow rate equals to in this study since it did not substantially change during the experi-
49.28 kg/h is higher than that of compression. The power output was mental campaign. By comparison of the different values of the poly-
equal to 0.437 kW (see Table 4). It is even lower than the power input tropic factor, it is found that ne is independent of the operating condi-
since the electric efficiency is very low (3.6%). The poor performance of tions. For instance, it is equal to 1.12 for (Pin, e = 4bar; Tin, e=1 °C) and
the system (ηel =0.032; COP = 0.48) is justified by the difference of the 1.1 for (Pin, e = 6bar; Tin, e=13 °C) . Consequently, the value of 1.1 was
operating pressures during the compression and expansion processes selected.
and is clearly apparent in the exergy losses or the irreversibilities in the In order to examine the effects of the operating conditions of the air
throttling valve. It should be noted that the real performance is higher motor on the output power provided by the manufacturer, these curves
since the cooling potential between the expansion valve and the air are compared to the experimental results in Fig. 14 for different mass
motor input is not evaluated. flows, input pressures and temperatures. The experimental values are
Errors in terms of the discharge phase and the power output are very consistent with the power curves with a difference varying between
low (maximum 2.6%) caused by the average value of the input air 0.7% and 16.8% according to the input temperature. As detailed in
temperature of the air motor introduced in the model and the fluctua- Table 5, when the temperature approaches the ambient temperature,
tion of the pressure input due to the manual regulation. Owning the the error is reduced. This is because the manufacturer curves of the air
error during the charge phase, the maximum error was for the round motor are made for industrial applications for which the maximum air
trip efficiency (13.1%), which remains quite acceptable. storage pressure does not exceed 30 bar and therefore the input tem-
Finally, when the discharge occurs from the initial state of 300 bar perature is close to the room temperature.
(experiment n°2), the electric efficiency and comprehensive efficiency It is of great importance to derive the thermodynamic to electrical
are slightly lower (3.2% and 14.24% respectively). These values result efficiency conversion independently of the temperature input, which is
from two effects: a higher value of compression ratio lowers the per- done in Fig. 14 using Eq. (28). Regardless the operating pressure, the
formance meanwhile, the variable compression ratio contributes to an efficiency is optimal for a certain value of RPM, denoted RPMoptimal but
increase of the efficiency. still not high enough (around 25%) and lower than the value achieved
with a scroll expander (reported as higher than 70% [36]). Similar
conclusions and values can be found in the works of Yu et al. [41] and
4.4. Focus on the throttling valve and the air motor Lemoufet-Gasti [42]. The low-efficiency values can be attributed
mainly to the air leaks at small RPM [43] and to mechanical losses at
In this section, we investigate more in-depth the effect of the op- high RPM.
eration characteristics of these two components. Low operation pressures are linked to an efficient thermodynamic
electrical conversion due to fact that the displacement volume is no
4.4.1. Throttling valve longer sufficient to expand the air to the atmospheric pressure. For
Using the approximation of Noel [32] expressed in Eqs. (24) and instance, the air expands to 1.05 bar when the input pressure is 3 bar
(25), isenthalpic curves are plotted and compared to the experimental whereas it expands to 1.12 bar for 4 bar pressure input (see Table 5).
results in Fig. 13. By throttling the stored air to 5 bar, higher the initial
pressure is, larger the temperature drops. The maximum error of the 5. Conclusions and perspectives
calculated values is around 11% and is recorded when the initial
pressure at 245 bar is out of the range of this approximation [25 atm; In this paper, we presented a steady state input/output model of a
150 atm]. Nonetheless, this error means an insignificant difference of simple configuration of the recently proposed concept of compressed
1 °C or 2 °C. As a result, the approximation of Noel remains accurate air energy storage system named as trigenerative compressed air energy
enough to model the pressure regulator for the pressure range con- storage system (T-CAES). An industrial pilot unit was described. Several
sidered here. experimental tests were carried out in order to investigate the real

Fig. 13. Isenthalpic curves of throttling in the pressure regulator.

1382
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

Fig. 14. Power versus RPM of the air motor (left) and thermodynamic to electrical conversion (right).

Table 5 practical applications. It is the first reliable model validated with ex-
Power and pressure output of the air motor for different operating conditions. perimental data for small-scale compressed air energy storage system,
Pin (bar) Tin (°C) RPM PowerEXP Powermodel(kW) Pout (bar) Error (%)
which can be applied for prospective trigenerative system studies.
(kW) On the other hand, such a steady-state modelling is limited and is
not sufficient when the compression ratio is variable. In addition, the
3 19 897 0.3 0.34 1.05 11.8 transient nature of the discharge process should be accounted to assess
3 5.6 892 0.283 0.34 1.05 16.8
4 18.5 898 0.447 0.45 1.12 0.7
properly the cooling potential and the output power.
4 −2 881 0.39 0.45 1.12 13.3 As the efficiency of such a compressed air energy storage system
remains low, future work will be devoted to the optimisation in terms of
system configuration and design parameters taking into account op-
behaviour of our system, focusing on the discharge phase. Experimental erational and technological aspects.
results were found to be in good agreement with the model results, with Finally, three different potential strategies would be applied also in
an error which does not exceed 13.2%. a future work to reduce losses in the throttling valve and improve the
Since a significant difference between the charge pressure at 181 performance of the system:
bars and discharge pressure at 5 bars takes place, the round trip elec-
trical efficiency is very low at 3.6% and the comprehensive efficiency 1 Increasing the expansion ratio until 25 by introducing the recently
which represents the equivalent electrical storage efficiency reaches developed micro-turbines [44].
15.6% underlying the importance of the trigeneration concept. 2 Replacing the throttling valve by a Ranque-Hilsch vortex tube or a
During the charge phase, the consumed electric energy of 13.72 cascade of vortex tubes. These devices have indeed the capability to
kWh was 12% higher than the model of 12.1 kWh. The error was caused relax the air to reduce the maximum inlet pressure allowed by
mainly by the transitional regime required to stabilize compressed air current microturbines, while producing a hot and a cold stream. The
temperature. produced cooling power would be then reinjected within some heat
The transitional regime during the charge phase takes about 40 min. exchangers of the compressed air energy storage system.
Pressure and temperature in the air tank during this phase were ac- 3 Adding a supersonic ejector after the throttling valve to mix the air
counted, the temperature stabilises at a particular level which differs from tanks at different pressures (which would reduce the cost of the
from the results of Wang et al. [10]. It highlights in particular the prime storage tanks) and guarantee a more constant pressure at the inlet of
importance of the initial mass of stored air. the turbine. It has been successfully applied by Chen et al. [45] for a
During the storage phase, the air temperature as well as the pressure large-scale adiabatic compressed air energy storage system.
were modelled and found in good agreement with the experimental
measurements. Unfortunately, it has not been possible to provide those Acknowledgement
variations during the discharge phase because of the influence of the
cold exhaust air of the air motor flowing around our system. M.C. and S.P. acknowledge the financial support of the Natural
During the discharge process, the output electric energy of 0.49 Sciences and Engineering Research Council of Canada (NSERC) and the
kWh was 1.5% less than the model of 0.5 kWh. A particular attention company Sigma Energy Storage Inc. through a Collaborative Research
was paid to the behavior of each system component during this phase. and Development grant (386141936).
It has been demonstrated that the Joule-Thomson effect governs the
temperature change across the pressure regulator and the assumption of References
constant temperature is no longer true as stated in previous models of
compressed air energy storage system. [1] Zou C, Zhao Q, Zhang G, Xiong B. Energy revolution: from a fossil energy era to a
A comprehensive analysis of the air motor efficiency was presented new energy era. Nat Gas Ind B 2016;3(1):1–11.
[2] Weitemeyer S, Kleinhans D, Vogt T, Agert C. Integration of renewable energy
to clarify its performance curves. The reported high values of the power sources in future power systems: the role of storage. Renew Energy 2015;75:14–20.
output associated with higher input temperatures provide a first evi- [3] Beaudin M, Zareipour H, Schellenberglabe A, Rosehart W. Energy storage for mi-
dence of the benefit of preheating. tigating the variability of renewable electricity sources: an updated review. Energy
Sustain Dev 2010;14(4):302–14.
The model developed is expressed directly in terms of design and [4] Chen H, Cong TN, Yang W, Tan C, Li Y, Ding Y. Progress in electrical energy storage
thermodynamic parameters and enables an accurate prediction of the system: a critical review. Prog Nat Sci 2009;19(3):291–312.
system performance and provides relevant output parameters for [5] Luo X, Wang J, Dooner M, Clarke J. Overview of current development in electrical
energy storage technologies and the application potential in power system

1383
M. Cheayb, et al. Applied Energy 239 (2019) 1371–1384

operation. Appl Energy 2015;137:511–36. [24] Wolf D, Budt M. LTA-CAES – a low-temperature approach to adiabatic compressed
[6] Papathanassiou SA, Boulaxis NG. Power limitations and energy yield evaluation for air energy storage. Appl Energy 2014;125:158–64.
wind farms operating in island systems. Renew Energy 2006;31(4):457–79. [25] Grazzini G, Milazzo A. A thermodynamic analysis of multistage adiabatic CAES. In:
[7] Ibrahim H, Bourji A, Ghandour M, Merabet A. Optimization of compressed air Proceedings of the IEEE, vol. 100, no. 2; 2012. p. 461–72.
storage’s volume for a stand-alone wind-diesel hybrid system. In: Electrical Power & [26] Arabkoohsar A, Dremark-Larsen M, Lorentzen R, Andresen GB. Subcooled com-
Energy Conference (EPEC), Halifax; 2013. p. 1–7. pressed air energy storage system for coproduction of heat, cooling and electricity.
[8] Anjos MF, Gabriel SA, Guerra C. “L’énergie au Québec et au Canada : un document Appl Energy 2017;205:602–14.
pour engager la conversation“, Report, Institut de l’énergie Trottier, Polytechnique [27] Han Z, Guo S. Investigation of discharge characteristics of a tri-generative system
Montréal; 2015. based on advanced adiabatic compressed air energy storage. Energy Convers
[9] Luo X, Wang J, Dooner M, Clarke J, Krupke C. Overview of current development in Manage 2018;176:110–22.
compressed air energy storage technology. Energy Proc 2014;62:603–11. [28] Jannelli E, Minutillo M, Lubrano Lavadera A, Falcucci G. A small-scale CAES
[10] Wang S, Zhang X, Yang L, Zhou Y, Wang J. Experimental study of compressed air (compressed air energy storage) system for stand-alone renewable energy power
energy storage system with thermal energy storage. Energy 2016;103:182–91. plant for a radio base station: a sizing-design methodology. Energy
[11] Luo X, Wang J, Dooner M, Clarke J, Krupke C. Modelling study, efficiency analysis 2014;78:313–22.
and optimisation of large-scale adiabatic compressed air energy storage systems [29] Lv S, He W, Zhang A, Li G, Luo B, Liu X. Modelling and analysis of a novel com-
with low-temperature thermal storage. Appl Energy 2016;162:589–600. pressed air energy storage system for trigeneration based on electrical energy peak
[12] Wang J, Lu K, Ma L, Wang J, Dooner M, Miao S, et al. Overview of compressed air load shifting. Energy Convers Manage 2017;135:394–401.
energy storage and technology development. Energies 2017;10(991):1–22. [30] Liu J-L, Wang J-H. Thermodynamic analysis of a novel tri-generation system based
[13] Wolf D. Methods for design and application of adiabatic compressed air energy: on compressed air energy storage and pneumatic motor. Energy 2015;91:420–9.
storage based on dynamic modeling. Laufen: Oberhausen; 2011. [31] Goodwin JDA, Derby J. Compressed air batteries. Energize 2011:26–8.
[14] Sciacovelli A, Li Y, Chen H, Wu Y, Wang J, Garvey S, et al. Dynamic simulation of [32] Robison DH, Beaty PJ. Compressor types, classifications, and applications, In:
adiabatic compressed air energy storage (A-CAES) plant with integrated thermal Proceedings of the 21st Turbomachinery Symposium, Texas A&M University; 1992.
storage – link between components performance and plant performance. Appl [33] Bloch HP. A practical guide to compressor technology. 2nd ed. Hoboken, N.J:
Energy 2017;185:16–28. Wiley-Interscience; 2006.
[15] Geissbühler L, Becattini V, Zanganeh G, Zavattoni S, Barbato M, Haselbacher A, [34] Bejan A. Convection heat transfer. fourth ed. Hoboken, New Jersey: Wiley; 2013.
et al. Pilot-scale demonstration of advanced adiabatic compressed air energy sto- [35] Hoxton LG. The Joule-Thompson effect for air at moderate temperatures and
rage, Part 1: Plant description and tests with sensible thermal-energy storage. J pressures. Phys Rev 1919;13(6):438.
Storage Mater 2018;17:129–39. [36] Lemort V, Guillaume L, Legros A, Declaye S, Quoilin S. A comparison of piston,
[16] Becattini V, Geissbühler L, Zanganeh G, Haselbacher A, Steinfeld A. Pilot-scale screw and scroll expanders for small scale Rankine cycle systems. In: Proceedings of
demonstration of advanced adiabatic compressed air energy storage, Part 2: Tests the 3rd international conference on microgeneration and related technologies in
with combined sensible/latent thermal-energy storage. J Energy Storage J buildings, Naples; 2013.
2018;17:140–52. [37] He W, Wang J. Optimal selection of air expansion machine in compressed air energy
[17] Proczka JJ, Muralidharan K, Villela D, Simmons JH, Frantziskonis G. Guidelines for storage: a review. Renew Sustain Energy Rev 2018;87:77–95.
the pressure and efficient sizing of pressure vessels for compressed air energy sto- [38] WeiBs AP. Volumetric expander versus turbine–which is the better choice for small
rage. Energy Convers Manage 2013;65:597–605. ORC plants. In: 3rd ASME ORC conference, Brussels (Belgium); 2015.
[18] Facci AL, Sánchez D, Jannelli E, Ubertini S. Trigenerative micro compressed air [39] Vadasz P, Weiner D. The optimal intercooling of compressors by a finite number of
energy storage: concept and thermodynamic assessment. Appl Energy intercoolers. J Energy Res Technol 1992;114(3):255–60.
2015;158:243–54. [40] Burt CM, Piao X, Gaudi F, Busch B, Taufik NF. Electric motor efficiency under
[19] Li Y, Wang X, Li D, Ding Y. A trigeneration system based on compressed air and variable frequencies and loads. J Irrig Drain Eng 2008;134(2):129–36.
thermal energy storage. Appl Energy 2012;99:316–23. [41] Yu Q, Cai M. Experimental analysis of a compressed air engine. J Flow Control,
[20] Zafirakis D, Kaldellis JK. Autonomous dual-mode CAES systems for maximum wind Measur Visual 2015;03(04):144–53.
energy contribution in remote island networks. Energy Convers Manage [42] Lemofouet-Gatsi S. Investigation and optimisation of hybrid electricity storage
2010;51(11):2150–61. systems based on compressed air and supercapacitors. PhD thesis, EPFL, Lausanne;
[21] Budt M, Wolf D, Span R, Yan J. A review on compressed air energy storage: basic 2006.
principles, past milestones and recent developments. Applied Energy [43] Beater P. Pneumatic drives: system design, modelling and control. Berlin, London:
2016;170:250–68. Springer; 2007.
[22] Kim YM, Shin DG, Favrat D. Operating characteristics of constant-pressure com- [44] Weiss Andreas P, Zinn Gerd. Micro turbine generators for waste heat recovery and
pressed air energy storage (CAES) system combined with pumped hydro storage compressed air energy storage. In: no. 15th conference on power system en-
based on energy and exergy analysis. Energy 2011;36(10):6220–33. gineering, thermodynamics & fluid flow-ES 2016, Pilsen, Czech Republic; 2016.
[23] Hartmann N, Vöhringer O, Kruck C, Eltrop L. Simulation and analysis of different [45] Chen LX, Hu P, Zhao PP, Xie MN, Wang DX, Wang FX. A novel throttling strategy
adiabatic compressed air energy storage plant configurations. Appl Energy for adiabatic compressed air energy storage system based on an ejector. Energy
2012;93:541–8. Convers Manage 2018;158:50–9.

1384