See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329802709

Investigation of an ejector powered double-effect absorption/recompression

refrigeration cycle

Article in International Journal of Refrigeration · December 2018

DOI: 10.1016/j.ijrefrig.2018.11.042

CITATIONS READS

17 119

3 authors:

Doniazed Sioud Mahmoud Bourouis

Ecole Nationale d'Ingénieurs de Monastir Universidad Rovira i Virgili
8 PUBLICATIONS 33 CITATIONS 182 PUBLICATIONS 2,629 CITATIONS

SEE PROFILE SEE PROFILE

Ahmed Bellagi
National Engineering School of Monastir - ENIM
171 PUBLICATIONS 1,894 CITATIONS

SEE PROFILE

All content following this page was uploaded by Doniazed Sioud on 22 January 2024.

The user has requested enhancement of the downloaded file.

 International Journal of Refrigeration 99 (2019) 453–468

Contents lists available at ScienceDirect

International Journal of Refrigeration

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

Investigation of an ejector powered double-effect

absorption/recompression refrigeration cycle
Doniazed Sioud a, Mahmoud Bourouis b,∗, Ahmed Bellagi a
a
U.R. Thermique & Thermodynamique des Procédés Industriels, Ecole Nationale d’Ingénieurs de Monastir-ENIM, University of Monastir, Monastir, Tunisia
b
Department of Mechanical Engineering, Universitat Rovira i Virgili, Av. Països Catalans No. 26, 43007 Tarragona, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The objective of the present work is to investigate the feasibility and the eventual improvement in perfor-
Received 9 August 2018 mance of an ejector powered water/lithium bromide double-effect absorption/recompression refrigeration
Revised 24 November 2018
cycle driven by high temperature heat sources.
Accepted 26 November 2018
The results show that the cycle performance parameters are significantly affected by the presence of
Available online 20 December 2018
the ejector and its characteristics. Further, the COP responses to variation of working conditions are differ-
Keywords: ent from those observed for the conventional double-effect absorption refrigeration cycles. The maximum
Double-effect absorption/recompression COP values of the ejector cycle occur at HP-generator temperatures lower by 20 °C to 25 °C than those of
cycle the conventional double-effect absorption cycle. The enhancement factor of the COP varies between 1.34
Water/lithium bromide and 1.70 at a driving steam temperature ranging from 240 °C to 340 °C. As regards the ejector design,
Ejector the ratio of HP-generator pressure and driving steam pressure should be kept as low as possible. The
High temperature heat sources
geometry of the design should allow for the maximum entrainment ratio feasible.
© 2018 Elsevier Ltd and IIR. All rights reserved.

Étude d’un cycle frigorifique à absorption/recompression double-effet alimenté

par un éjecteur

Mots-clés: Cycle à absorption/recompression double effet; Bromure de lithium/eau; Éjecteur; Sources de chaleur à haute température

1. Introduction several investigations in the open literature. The use of advanced

configurations such as double and triple-effect cycles is one way
Absorption chillers are usually used in large-scale air- to increase the efficiency of these cooling cycles.
conditioning systems for both industrial and commercial ap- In this context, some investigations proposed instead the
plications, and also in heat recovery from industrial power plants. integration of an ejector into absorption cooling systems. Two
The most common working fluids used in these systems are wa- alternative locations of the ejector in single-effect cycles were
ter/lithium bromide and ammonia/water. The former is used only discussed: between the generator and the evaporator and between
in air-conditioning applications, because water freezes at 0 °C, and the evaporator and the absorber. Several authors investigated the
the latter mainly in refrigeration applications that require cooling first arrangement. Sun et al. (1996) and Majdi (2016) theoretically
below 0 °C. The efficiency of absorption cycles is usually assessed explored the ejector water/lithium bromide single-effect absorp-
in terms of the coefficient of performance (COP), which is rather tion cycle using different geometries for the ejector. They con-
low for the single-effect configuration, namely around 0.7 for cluded that the COP of the ejector absorption cycle is higher than
water/LiBr and 0.5 for ammonia/water. Hence, the improvement of that of the conventional absorption cycle. Jiang et al. (2002) com-
the COP of absorption cooling cycles is an issue that is reported in pared the coefficient of performance of the ejector single-effect
absorption cycle with that of small capacity water/lithium bromide
double-effect absorption chillers. The authors reported that the
∗
Corresponding author. COP of the ejector absorption cycle was up to 0.9–1.0, just slightly
E-mail address: [email protected] (M. Bourouis). below that of the commercial double-effect absorption chillers.

https://doi.org/10.1016/j.ijrefrig.2018.11.042
0140-7007/© 2018 Elsevier Ltd and IIR. All rights reserved.
454 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Their results showed that the new design was simpler than that of
Nomenclature the two-stage heat transformer and double absorption heat trans-
former (where a pump and an absorber/evaporator are added to
Symbols a single-stage heat transformer). The delivered useful temperature
A Surface area, [m2 ] was higher than in the case of a single-stage heat transformer, and
COP Coefficient of performance, [–] the COP improved (Shi et al., 2001). Vereda et al. (2012) evaluated
F Circulation ratio, [–] the feasibility of an adjustable nozzle area in an ejector absorption
h Specific enthalpy, [kJ.kg-1 ] refrigerator using ammonia/lithium nitrate. They reported that the
m ˙ Mass flowrate, [g.s-1 , kg.s-1 ] activation temperature was reduced by about 9 °C on integrating
M Mach number, [–] the ejector, and the COP increased at moderate temperatures, par-
M∗ Modified Mach number, [–] ticularly when the temperature of the driving hot water inlet gen-
P Pressure, [kPa, bar] erator was lower than 81–92 °C. Garousi Farshi et al. (2014) com-
Q˙ Heat exchange rate, [kW] pared the results obtained by Vereda et al. (2012) with those of
R Gas constant, [J.kg-1 .K-1 ] the ammonia/NaSCN ejector absorption cycles. They concluded that
T Temperature, [°C, K] both COP and exergy efficiency were higher than in single-effect
W ˙ Pump power, [kW] absorption refrigeration cycles especially at low generator temper-
x, X Lithium bromide mass fraction in solution, [–] atures.
An advanced configuration, reported in the literature for ejec-
Greek symbols
tor absorption refrigeration cycles, consists of coupling the ejec-
α Mass ratio of regenerated refrigerant in MP and HP-
tor with the solution generator via an external loop. This in-
generators, [–]
volved a steam generator and a steam ejector which compressed
γ Specific heat ratio, (Cv/Cp), [–]
the vapour desorbed from the water/lithium bromide solution in
ε COP enhancement factor, [–]
order to re-heat the solution in the desorber. Thus, Eames and
η Efficiency (Appendix), [–]
Wu (20 0 0) described the configuration of an ejector powered wa-
ϱ Density, [kg.m-3 ]
ter/lithium bromide single-effect absorption/recompression refrig-
τ Temperature ratio, [–]
eration cycle and theoretically investigated its energy efficiency
ω Entrainment ratio, [–]
and performance characteristics. Their results showed that the COP
Indices of the cycle increased with the temperature of the driving heat.
ABS Absorber Eames and Wu (20 0 0) and Wu and Eames (1998) also investigated
CD Condenser experimentally a 5 kW re-compression refrigerator prototype. The
c Ejector constant area section coefficient of performance (COP) of this prototype was 1.03 at the
EVAP Evaporator design operating conditions, which was 14% lower than that of the
GEN Generator estimated theoretical value. However, it was higher than that of
HP High pressure the single-effect absorption chiller and close to the typical value of
in Input the conventional double-effect absorption cycle. The authors con-
HX Heat exchanger cluded that the ejector absorption refrigeration cycle was an effi-
MP Intermediate pressure cient configuration when a high temperature heat source is avail-
n Nozzle exit section able.
out Output The integration of ejectors into more elaborated configura-
P Pump tions of absorption cooling cycles was also considered. Hong et
sat Saturation state al. (2011) investigated a parallel-flow double-effect ejector absorp-
STM Steam tion cycle. Compared to the basic configuration of the single-effect
t Nozzle throat section absorption cycle, their results showed that the COP improved by
10%, 20% and 30% when the heat source temperature was set at
130 °C, 136 °C and 142 °C, respectively. Moreover, at heat source
Aphornratana and Eames (1998) carried out an experimental study temperatures higher than 122 °C and lower than 150 °C, the effi-
on a three-pressure ejector absorption refrigeration machine and ciency of the ejector double-effect absorption cycle was better than
showed that the COP of the ejector absorption machine was 30– that of the conventional double-effect absorption cycle. Farshi et al.
60% higher than that of the basic single-effect absorption cycle and (2012) investigated an ejector series-flow double-effect absorption
achieved the COP of small commercial double-effect absorption refrigeration cycle. The authors performed a thermodynamic anal-
refrigeration machines. In the literature on ammonia/water ejector ysis comparing their cycle with the single-effect and series-flow
absorption refrigeration systems, it is reported that higher values double-effect absorption cycles. They concluded that at a particu-
of the coefficient of performance can be achieved by adding a lar range of the heat source temperature between 92 °C and 137 °C,
flash tank between the ejector and the evaporator. Hence, Sirwan the proposed cycle operated more efficiently than the two other
et al. (2013) and Abed et al. (2015) carried out a theoretical study cycles.
to analyse the effect of the flash tank on the entertainment ratio For heat sources at temperatures over 170 °C, the triple-effect
and the cycle performances. Their results showed that this cycle configuration of the absorption cycle was investigated by several
configuration (ejector and flash tank) improved the coefficient researchers. Grossman and Wilk (1994, 1993) compared the the-
of performance. This was because the increase in the amount of oretical performances of parallel-flow and series-flow triple-effect
refrigerant delivered to the evaporator coming from the flash tank water/lithium bromide absorption chillers. In both cases, heat from
increased the cooling effect. external source was supplied solely to the HP-generator; each
Other researchers investigated the second configuration in intermediate-pressure generator was powered by heat rejected
which the ejector is located at the absorber entrance and replaced by the next high temperature condenser. In the configuration of
the solution expansion valve. Shi et al. (2001) explored this ar- series-connected generators, the aqueous salt solution exiting the
rangement for heat transformers using water/lithium bromide as absorber fed successively the low-pressure generator, then the
a working pair, and Sözen and Özalp (2003) using ammonia/water. intermediate-pressure generator and finally the HP-generator. In
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 455

the parallel-connected arrangement, the refrigerant-rich solution Lizarte and Marcos (2016) optimized the coefficient of performance
was split into three streams that fed simultaneously the three of a parallel flow water/lithium bromide triple-effect absorption
generators. The authors reported that the triple-effect parallel-flow cooling system both under design and off-design conditions (for
designs, similar to what they noticed in the case of double-effect a variable-temperature heat source). For off-design conditions and
cycles, always yielded a higher performance than the series-flow at high generator temperatures ranging from 142 °C to 227 °C the
configurations. The performances reached at the specific thermal optimal COP achieved was within the range of 1.05–2.13.
conditions (Tgen = 260 ◦ C; Tevap = 4.1 ◦ C; Tcond = 32 ◦ C) were In conclusion of this literature overview, it can be stated that:
COPPARALLEL =1.44 vs. COPSERIES =1.29. More elaborated triple-effect
configurations were also reported to reach even higher values of (i) The coefficient of performance of the conventional single
the COP (Grossman et al., 1993), COPPARALLEL =1.78; COPSERIES =1.40 and double-effect water/LiBr absorption chillers is rather
Erickson et al. (1996) studied a water/lithium bromide triple-effect low: 0.7–0.8 for the former and 1.0–1.2 for the latter.
absorption refrigerator. Their results showed that triple-effect (ii) The integration of ejectors into single and double-effect ar-
absorption systems can attain a COP 50% over double-effect cy- rangements of absorption cooling systems can enhance their
cles, almost equal to 1.8. Kim et al. (2002) theoretically analysed performance under specific working conditions.
four configurations of compressor-assisted water/lithium bromide (iii) The highest COPs are reached with triple-effect designs (up
series-flow triple-effect absorption cooling cycles. The aim of their to 2) but, besides the growing complexity of these designs,
study was to reduce the temperature of the driving heat by 40 °C temperatures over 180 °C are required for the heat source.
in order to avoid corrosion problems associated with generator At these working temperatures in the generator, the con-
temperatures over 180 °C. In the considered configurations the ventional working fluid water/LiBr suffers from serious prob-
compressor was placed alternatively posterior to the high-pressure, lems of corrosion and thermal decomposition.
the middle-pressure, the low-pressure generator or to the evap-
The objective of the present paper is to propose a novel
orator. The authors concluded that all four configurations were
ejector powered water/lithium bromide double-effect absorp-
suitable as the COP was found to have increased from 1.54 in the
tion/recompression refrigeration cycle driven by high temperature
case of the conventional triple-effect cycle to a COP of 1.70–1.74 for
heat sources. The ejector is placed, as proposed in reference Eames
the newly proposed schemes. The work required of the compressor
and Wu (20 0 0), in an external circuit to the absorption cycle,
was about a 3–5% of the cooling capacity of equivalent mechanical
and is itself driven by a steam generator. The investigations, per-
energies. Using the software ABSIM (Grossman and Wilk, 1994,
formed by numerical simulations, aim to assess the feasibility and
1993; Grossman et al., 1993), Kaita (2002) investigated three con-
the eventual improvement in performance of the proposed cycle
figurations of the triple-effect water/lithium bromide absorption
configuration at temperatures of the driving steam ranging from
cooling cycle, namely parallel-flow, series-flow and reverse-flow.
240 °C to 340 °C. The objective is to reach COP−values comparable
The author found that the parallel-flow configuration achieved the
to those of triple-effect cycles, maintaining the working tempera-
highest coefficient of performance (2.0); however the driving heat
ture lower than 170 °C in the high pressure generator to overcome
temperature exceeded 220 °C. Gebreslassie et al. (2010) estimated
the corrosive effect and thermal instability of water/LiBr at high
the coefficient of performance, the exergetic efficiency and the
temperatures.
exergy destruction rate for half- to triple-effect water/lithium
The operating principles of the absorption/recompression re-
bromide absorption cooling cycles. The authors concluded that the
frigeration cycle considered and the thermodynamic simulation
COP increased significantly from the double to the triple-effect
methodology are described. The effect of the driving heat source
cycles and decreased gradually at high temperature heat sources.
temperature, the generator temperature and the ejector design on
Similar trends were observed for the exergetic efficiency. Álvarez
the cycle performance is analysed.
et al. (2015) investigated a triple-effect absorption cooling cycle
driven by high temperature heat sources (up to 260 °C). The cycle
configuration consisted of a water/lithium bromide double-effect 2. Configuration of the water/lithium bromide double-effect
cycle coupled with a high pressure single-effect cycle in which an absorption/ recompression cycle
aqueous (lithium, potassium, sodium) nitrate solution was used
as a working pair. The authors reported that the coefficient of The ejector powered water/lithium bromide double-effect ab-
performance was 1.73 at a heat source temperature of 250 °C and a sorption/recompression cycle is shown schematically in Fig. 1. It
cooling water temperature of 30 °C. Azhar and Siddiqui (2013) car- comprises three fluid circuits: driving steam loop, refrigerant and
ried out a performance and economic analysis of water/lithium solution circuits.
triple-effect absorption cooling cycles. Depending on the config-
uration of the considered cycles, they found a COP in the range 2.1. Steam circuit
(0.70 − 0.86) for single-effect, (1.20 − 1.55) for double-effect and
up to 2.16 for the optimized triple-effect designs. The results also A high temperature heat source is used in a steam generator to
showed that the operating cost of the double-effect cycle was generate water vapour at high pressure (20). This is piped to the
about 68% of that of the single-effect cycle, the operating cost of ejector primary nozzle, where an expansion takes place producing
the triple-effect cycle was 60 to 75% of that of the double-effect a high velocity jet and generating a partial vacuum in the mixing
cycle and around 45% of that of the single-effect cycle. Shirazi et al. chamber of the ejector. A fraction (22) of the steam liberated in the
(2016) investigated five configurations of solar absorption cooling high-pressure generator (HP-generator) is sucked off in this cham-
and heating systems. The authors performed the analysis taking ber and entrained by the primary flow. The mixed stream then
into account energetic, economic, and environmental aspects at passes through a supersonic-subsonic diffuser that compresses it
different climate conditions. They reported that the COP of triple- to the pressure P23 . The exiting steam (23) condenses inside the
effect cycle driven by solar collectors was between 1.61 to 1.72, tubing coil immersed in the salt solution of the HP-generator, lib-
depending on the location, and that the required temperature in erating the heat needed for the desorption of water (17) from
the collector was about 210 °C. They concluded that the extra capi- the aqueous solution. Part (27) of the condensate water from the
tal cost compared to solar single-effect systems could be recovered heater coil (24) is pumped back into the steam generator via a wa-
in less than 10 years and that this configuration had the lowest ter pump (W – PMP) and the rest (25) flows into the high-pressure
carbon dioxide emissions of the systems taken into consideration. condenser (HP-condenser).
456 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Fig. 1. Schematics of the ejector powered double-effect absorption/recompression refrigeration cycle investigated.

The rest of the cycle is a conventional double-effect lithium then preheated (3) in the solution heat exchanger (S–HX1). Partic-
bromide absorption cycle including the two remaining fluid cir- ular to this double effect cycle is the fact that the MP-generator
cuits, solution and refrigerant loops. is fed by two solution streams. As illustrated in the box to the
right of Fig. 1, the splitter (S – SPL) divides the refrigerant-rich
stream 3 into two parts: one part (3a), together with the partially
2.2. Refrigerant circuit
cooled concentrated salt solution (16), feeds this generator, while
the other part (11) flows further via the pump (S–MP2) and heat
The HP-condenser is alimented by two streams. The first one
exchanger (S–HX2) to the HP-generator. The two mounted heat ex-
(26) comes from the liquid water splitter (L–SPL) mounted at
changers (S–HX1 and S–HX2) are used to recuperate heat from the
the exit of the HP-generator, after decompression to the appropri-
salt-rich solutions (14 and 4) exiting the generators and to preheat
ate pressure in the liquid expansion valve (W – VLV) takes place.
the refrigerant-rich solution flowing counter-currently.
The second stream is the rest (21) of the water vapour (17) liber-
Two liquid expansion valves (S–VLV2 and S–VLV1) are mounted
ated in the HP-generator. The released condensation heat Q˙ HP−CD
on the salt-rich solution line in order to reduce the pressure of the
is used to drive the water desorption in the intermediate-pressure
solution 14 gradually to that of the MP-generator (16) and finally
generator (MP-generator). The formed condensate (18) passes
to that of the absorber (6). The splitting ratio in the splitter (S–SPL)
through the valve R – VLV2 where its pressure is reduced to that
is set by the condition that the heat duty of the HP-condenser is
of MP-condenser (19). The steam liberated in the MP-generator (7)
large enough to produce as much vapour from the MP-desorber as
also feeds this condenser. The exiting liquid water (8) undergoes a
is necessary to make the composition of the salt-rich solution (4)
last adiabatic pressure reduction (R – VLV1), which causes a large
equal to that of the solution leaving the HP-generator (14), i.e.
decrease in temperature as a result of partial evaporation. It then
flows (9) into the machine evaporator where it totally evaporates, x4 = x14
making available the cooling capacity Q˙ EVAP . Formed refrigerant
Fig. 2 presents the proposed cycle on the Dühring–diagram for
vapours (10) are dissolved in the absorber by the salt-rich solu-
a particular set of operating conditions. The solution circulation
tion (6) returning from the generators. The heat released by this
loop (1 − 11 − 13 − 14 − 4 − 6) is similar to that of a conventional
exothermic process Q˙ ABS , like the condensation heat Q˙ MP−CD from
double effect absorption cycle: Water is taken up at the absorber
the MP-condenser, is rejected to the cooling medium.
(6 − 1 ), released as vapor at the generators (17 and 7), and con-
densed at 18 and 8. In the considered cycle however, a fraction of
2.3. Solution circuit the formed vapor 17 is compressed in the ejector and contributes
to heat the solution 13 by condensation. Thus, heat otherwise re-
The salt-weak (refrigerant-rich) solution leaving the absorber jected to the environment is recovered in order to improve the ef-
(1) is pumped (S–MP1) into the MP-generator pressure (2) and ficiency of the system. The ejector process is shown in Fig. 3 on
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 457

Fig. 2. Dühring–diagram of the ejector powered double-effect absorption/recompression refrigeration cycle.

Fig. 3. Ejector powered double-effect absorption/recompression refrigeration cycle on (P − h )–diagram of water.

(P − h )−chart. The external ejector loop processes are represented formulated for each machine component. Hypotheses and assump-
by the line (20 − Nozzle exit − 23 − 24 − 28). In the mixing cham- tions for the calculations are summarized in the following:
ber, after expansion in the nozzle, stream 20 mixes with stream 22,
and the resulting gas mixture is compressed to backpressure P23.
The line through points (18 − 19 − 8 − 9 − 10) represents a process • Simulations and analyses are performed under steady state con-
series in the condensers and evaporator. ditions.
• Heat losses to surroundings at generators, condensers, absorber
3. Cycle simulation model and evaporator are negligible.
• Pressure losses in ducts and machine components are negligi-
3.1. Mathematical modelling ble.
• Refrigerant exiting condensers is in saturated liquid state.
In order to develop a simulation model for the configuration • Refrigerant exiting evaporator is in saturated vapour state.
of the proposed absorption chiller, mass and energy balances are • Solution and refrigerant expansion valves are isenthalpic.
458 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

• Refrigerant vapour and liquid solution leaving generators are at Combining with the energy balance of the high-pressure
same temperature. generator leads finally to
• Liquid solutions leaving absorber and generators are saturated.
(h17 − h13 ) + F2 (h14 − h13 )
• Heat from HP-condenser covers intermediate generator heat ˙ 20 = m
m ˙ 17 (12)
needs with 5 °C temperature pinch.
(1 + ω )(h23 − h24 )
• Solution liquid compositions leaving generators are identical. where F2 =
˙ 14
m
=
x13
is the circulation factor in the new
˙ 17
m x14 −x13
• Absorption and condensation temperatures are equal, T8 = T1 .
cycle configuration.
• Evaporation temperature, T10 , is set to a fixed value. • Energy balance
• Effectiveness of heat exchangers is set to ηS−HX1 = ηS−HX2 =
50%. h20 + ωh22
h20 + ωh22 = (1 + ω )h23 ⇒ h23 = (13)
• Liquid pumping is isentropic, ηS−PMP1 = ηS−PMP2 = ηW−PMP = 1+ω
100%. c. Water pump
• Mass balance
Neglecting kinetic and potential energy changes, the balance
equations are as follows: ˙ 28 = m
m ˙ 27 = m
˙ 20 (14)
Total mass balance
  • Energy balance
˙ in =
m ˙ out
m (1)
m ˙ w−PMP = m
˙ 20 h27 + W ˙ 20 h28 (15)
Species mass balances
  with
(xi m˙ )in = (xi m˙ )out i = H2 O, LiBr (2) P28 − P27
˙ w−PMP = m
W ˙ 20 (16)
Energy balance ρ27
 
(m˙ h )in − (m˙ h )out + W
˙ + Q˙ = 0 (3) d. Steam generator
• Mass balance

Q˙ and W
˙ are heat and work transfer rates, respectively.
˙ 20 = m
m ˙ 28 (17)
The system coefficient of performance COP is per definition:
• Energy balance, with QSTMG the driving heat supplied to the
Q˙ EVAP Q˙ EVAP
COPED =  ≈ (4) chiller at the steam generator
Q˙ STM + W ˙ P Q˙ STM
˙ 20 (h20 − h24 )
QSTMG = m (18)
As an illustration of the modelling procedure, the particular
case of the steam generator-loop is treated. e. Expansion valve
• Mass balance

3.1.1. Steam generator loop model

˙ 26 (= m
˙ 25 = m
m ˙ 22 ) (19)
The generator loop encompasses the machine components
steam generator, ejector, high-pressure desorber, expansion valve, • Energy balance
stream-splitter, and water pump.
h25 = h26 (= h27 = h24 ) (20)
a. High-pressure desorber
f. Further, setting T24 for the temperature of condensing steam in
• Total mass balance
the heating coil and P17 for the secondary flow pressure (high-
˙ 13 + m
m ˙ 23 = m
˙ 14 + m
˙ 17 + m
˙ 24 (5) pressure in the HP-generator), the properties of the salt-rich
solution and the water-vapour exiting this generator can be
˙ 23 = m
Since m ˙ 24 , this equation reduces to evaluated based on the equilibrium data of the water-lithium
bromide system: T14 =T24 − TG ( TG is set to 5 K in simula-
˙ 13 = m
m ˙ 14 + m
˙ 17 (6)
tions);
• LiBr mass balance
x14 = xsat (T14 , P17 ); h14 = hsat−liq (T14 , x14 ); h17 = hsat−vap (T14 , P17 )
˙ 13 x13 = m
m ˙ 14 x14 (7)
The properties of saturated pure liquid water exiting the heat-
Combining both the latter equations leads to ing coil are evaluated using steam tables
x13 x14
˙ 14 = m
m ˙ 17 ˙ 13 = m
m ˙ 17 (8) P24 = Psat−steam (T24 ); h24 = hsat−liq−water (T24 )
x14 − x13 x14 − x13
Similarly for the properties of the saturated steam leaving the
• Energy balance
steam generator at fixed temperature T20
x14 x13
˙ 23 (h23 − h24 ) + m
m ˙ 17 h13 = m
˙ 17 h17 + m
˙ 17 h14 P20 = Psat−steam (T20 ); h20 = hsat−steam (T20 )
x14 − x13 x14 − x13
(9) 3.2. Theoretical cooling efficiency
b. Ejector
˙ 22 (= m
• The entrained vapour flow rate m ˙ 26 ) is writ-
˙ 25 = m The heat released by the condensation of the mixed vapour
ten as stream m˙ 23 in the high-pressure generator can be evaluated using
the energy balance of this machine component (Eq. 8)
˙ 22 = ω m
m ˙ 20 (10)
(h17 − h13 ) + F2 (h14 − h13 )
where ω is the ejector entrainment ratio and m˙ 20 , the mo- ˙ 20 = m
m ˙ 17 (21)
(1 + ω )(h23 − h24 )
tive steam flow rate. The mixed gas flow rate m˙ 23 (= m˙ 24 ) is
then: This equation gives the relation between the motive steam
flow-rate m ˙ 20 , the desorbed refrigerant flowrate m
˙ 17 , the entrain-
˙ 23 = (1 + ω )m
m ˙ 20 (11) ment ratio ω, the thermodynamic state of the salt weak and strong
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 459

solutions, (h13 , h14 ), the circulation ratio F2 and the specific con- The simulation program consists of two subprograms corre-
densation enthalpy (h23 − h24 ). sponding to the actual refrigerator configuration: a subprogram
The useful cold Q˙ EVAP is produced by the evaporation of the re- for the water-lithium bromide circuit and another for the driving
frigerant in the machine evaporator, steam circuit with the ejector as the main component.
Before running the simulations for the ejector powered double-
Q˙ EVAP = (m ˙ 7 )(h10 − h9 )
˙ 17 + m (22)
effect absorption/recompression cycle, the simulation program was
Neglecting the energy supplied to the three pumps, it fol- first tested and validated by comparing the calculated results with
lows that for the coefficient of performance of the proposed cycle, the literature data whenever available. For the validation, three
COPER : tests were performed. First, the ejector subprogram was validated.
Then, the simulation model was tested for an ejector driven single-
Q˙ EVAP (m˙ 17 + m˙ 7 )(h10 − h9 ) effect absorption cycle. Finally, the main program was validated for
COPER = = (23)
Q˙ STM m˙ 20 (h20 − h28 ) a conventional double-effect absorption chiller without ejector.

(h10 − h9 )(h23 − h24 ) 4.1. Ejector model validation

COPER = (1 + ω )
( (h17 − h13 ) + F2 (h14 − h13 ) )(h20 − h28 )
 
m˙ 7 (h10 − h9 )(h23 − h24 ) Detailed mathematical formulations of the ejector 1D flow
+ (1 + ω ) model are described in the Appendix. Given the operating con-
˙ 17 ( (h17 − h13 ) + F2 (h14 − h13 ) )(h20 − h28 )
m
ditions (primary, secondary and backpressure) as well as the
(24)
geometry of the ejector, an iteration procedure was applied to
determine the entrainment ratio ω. For the validation of the
Setting α for the ratio of the refrigerant amount generated in
ejector model experimental bibliographical data were used (Chen
the intermediate-pressure and high-pressure generators, α = m˙ 7 ,
m˙
17 and Sun, 1997; Huang et al., 1999). Three adjustable parameters
the final expression for the coefficient of performance becomes were introduced to account for the irreversibility of gas expansion
  in nozzle (ηn ) and diffuser (ηd ), and for mixing of primary and
Q˙ EVAP (h10 − h9 )
COPER = = (α + 1 ) secondary streams (ηm ). Eleven ejectors with an ejector area ratio
˙QSTM ( 17
h − h 13 ) + F2 (h14 − h13 )
  ( AAct ) ranging from 6.44 to 10.64 and a nozzle throat diameter be-
(h23 − h24 ) tween 2.62 mm and 2.82 mm were tested using R141b as a working
× (1 + ω ) (25)
(h20 − h28 ) fluid. As illustration of the obtained results, Fig. 4a and 4b show
that with the set of values (ηn = 0.95, ηd = 0.85, ηm = 0.88), the
The expression of COPER is thus the product of two terms. The experimental data reported in reference Huang et al. (1999) were
former on the left side of Eq. (25) represents the theoretical ex- fairly reproduced. The difference observed in Fig. 4b is probably
pression of the performance of a conventional double-effect cycle due to the parameter ηm set in our calculations as constant while
and the latter, an enhancement factor. It describes the performance in reference Huang et al. (1999) its value was varied with the
improvement as a result of the use of the steam ejector. Therefore, ejector area ratio ( AAc ).
t
the considered cycle is virtually a double effect cycle with an en-
The average absolute deviation between experimental and cal-
hanced performance due to the steam ejector
culated entrainment ratio is 8%. This value is acceptable consid-
Q˙ EVAP ering the large number of algebraic equations in the theoretical
COPER = = ∈ COPconv (26) ejector model and the adoption of perfect gas assumption in these
Q˙ STM
equations.
(h −h )
The ratio (h23 −h24 ) in the enhancement factor indicates the frac-
20 28
tion of the energy supplied to the motive stream (h20 − h28 ) used 4.2. Validation using the ejector powered single-effect
to heat the salt solution in the HP-generator. Because h28 ࣃ h24 , it absorption/recompression cycle
follows that:
(h23 − h24 ) (h23 − h24 ) To our knowledge, only one reference Eames and Wu (20 0 0) re-
 ports on experimental and theoretical investigations of re-
(h20 − h28 ) (h20 − h24 )
compression cycles with an ejector in the external loop, which
(h20 − h24 ) (h20 − h23 ) (h20 − h23 )
= − =1− (27) however deals solely with single effect configurations. The data
(h20 − h24 ) (h20 − h24 ) (h20 − h24 ) presented in this paper are used for the partial validation of our
(h20 −h23 ) program after modification to accommodate the configuration con-
(h20 −h24 ) is the fraction of supplied heat (h20 − h24 ) used to en-
sidered.
train the vapour by the ejector from the solution chamber into
The simulations were performed for similar operating con-
the coil. So theoretically to get the best energy performance, one
ditions: equal absorption and condensation temperatures
has to optimize the enhancement factor ∈ by minimizing the ra-
(h −h ) (Tcond = Tabs = 30 ◦ C), evaporator temperature (Tevap = 5 ◦ C),
tio (h20 −h23 ) and maximizing the entrainment factor ω. This result driving steam temperature of 198.3 °C (corresponding to a satura-
20 24
emphasizes the huge influence of the ejector operation on the per- tion pressure of 15 bar), 54% salt concentration in liquid solution
formance of the novel cycle. leaving the absorber, 50 mbar secondary pressure, and cooling
capacity of 5 kW.
4. Model validation Fig. 5 shows a comparison of predicted COP with the exper-
imental data reported in reference (Eames and Wu (20 0 0). As
Using the software Engineering Equation Solver, EES (Klein and can be noticed, the concordance is good, with a maximum de-
Alvarado, 2005), a computer code simulating the steady-state be- viation of 4% between both sets of data. In view of the scatter-
havior of the refrigerator and an analysis of its performances under ing of the experimental data and the simplifying assumptions in
various operating conditions was developed. The thermo-physical the ejector model, the deviation can be considered acceptable. Fur-
properties of the working fluid system H2 O/LiBr are available as ther, it is worth note that the COP of this single-effect absorp-
internal functions in the software library. tion/recompression cycle is almost 1, which is higher than the
460 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Fig. 4. Validation of the ejector model based on experimental data of reference (Chen and Sun, 1997).

Table 1
Comparison of results calculated with data from reference (Somers, 2009).

Point T (°C) P (kPa) ˙ (kg.s-1 )

m X (% LiBr)

Ref. (Somers, 2009) Present work Ref. (Somers, 2009) Present work Ref. (Somers, 2009) Present work Ref. (Somers, 2009) Present work

1 29.9 29.9 0.881 0.88 1 1.0 0 0 52.7 52.5

2 29.9 29.9 4.178 4.173 1 1.0 0 0 52.7 52.5
3 46.9 47.6 4.178 4.173 1 1.0 0 0 52.7 52.5
4 74.9 75.8 4.178 4.173 0.856 0.857 61.6 61.2
5 52.4 52.9 4.178 4.173 0.856 0.857 61.6 61.2
6 45.5 47.5 0.881 0.88 0.856 0.857 61.6 61.2
7 57.7 57.8 4.178 4.173 0.066 0.064 0 0
8 29.7 29.7 4.178 4.173 0.144 0.143 0 0
9 5.1 5.1 0.881 0.88 0.144 0.143 0 0
10 5.1 5.1 0.881 0.88 0.144 0.143 0 0
11 57.7 57.8 4.178 4.173 0.541 0.549 52.7 52.5
12 57.8 57.8 64.37 66.73 0.541 0.549 52.7 52.5
13 91.1 91.6 64.37 66.73 0.541 0.549 52.7 52.5
14 144.8 144.8 64.37 66.73 0.463 0.471 61.6 61.2
15 101.3 101.3 64.37 66.73 0.463 0.471 61.6 61.2
16 77.1 78 4.178 4.173 0.463 0.471 61.6 61.2
17 123.3 124.7 64.37 66.73 0.078 0.078 0 0
18 87.8 88.7 64.37 66.73 0.078 0.078 0 0
19 29.7 29.7 4.178 4.173 0.078 0.078 0 0

values ( ∼ 0.7) usually achieved with conventional water/lithium imum deviation of 0.4%. Pressure deviations for low, medium
bromide single-effect absorption cycles. and high pressure are 0.3%, 0.1%, and 4%, respectively. The gap
in LiBr-concentration is approximately 0.6%. Finally the absolute
4.3. Validation using a double-effect absorption refrigeration cycle average deviation in flow rate is 0.9%, an acceptable value as the
without an ejector flow rate is calculated according to mass and energy balances,
so it accumulates all the previous deviations. In conclusion, this
The third test of our program was based on the simulation subprogram of our simulation tool can be considered validated
results reported by Somers (2009), obtained using an ASPEN as the calculated cycle data are well in agreement with those
PLUS model for a double-effect absorption chiller. The simulations reported in reference Somers, (2009).
were performed under following operating conditions: Evaporator
temperature, 5.1 °C; condenser temperature, 29.7 °C; absorber 5. Thermodynamic analysis of the ejector driven double-effect
temperature, 29.9 °C; heat exchanger effectiveness, 0.5 and flow cycle: simulation results and discussion
rate of refrigerant-rich solution leaving the absorber, 1 kg/s. Table 1
shows a detailed comparison of our simulation results with the The first step in our investigations of the proposed cycle was
data reported in reference Huang et al. (1999). As can be noticed, an extension of the last test. For similar operating conditions, a
the obtained results are well in agreement with the published steam driven ejector was integrated in the double-effect absorp-
data. Calculated temperatures are almost identical, with a max- tion/recompression cycle (Fig. 1) and the effect of the primary flow
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 461

Table 3
Operating conditions investigated.

Case Thermal Pinch at HX-inlet, T = Tout − Tin

# conditions TPinch
◦
1 T8 = T1 = 35 C TPinch, evap = 3 ◦C TCooling Medium = 5 ◦ C
T10 = 4 ◦ C TPinch, abs (cond ) =3 ◦
C TChilled Water = −5 ◦ C
◦ ◦
2 T8 = T1 = 40 C TPinch, evap = 3 C TCooling Medium = 5 ◦ C
T10 = 4 ◦ C TPinch, abs (cond ) = 3
◦
C TChilled Water = −5 ◦ C
◦ ◦
3 T8 = T1 = 35 C TPinch, evap = 3 C TCooling Medium = 5 ◦ C
T10 = 12 ◦ C TPinch, abs (cond ) = 3
◦
C TChilled Water = −5 ◦ C
◦ ◦
4 T8 = T1 = 40 C TPinch, evap = 3 C TCooling Medium = 5 ◦ C
T10 = 12 ◦ C TPinch, abs (cond ) = 3
◦
C TChilled Water = −5 ◦ C

Table 4
Nozzle area ratio and primary flow temperature for cases dealt with.

Nozzle area ratio ( AAnt )

Primary flow temperature,

T20 (Saturated steam) Case #1 Case #3 Case #2 Case #4

200 °C 4 5 4 4
240 °C 7 8 5 6
280 °C 9 11 7 8
Fig. 5. Model predicted COP vs. LiBr-concentration in the generator in comparison 320 °C 13 17 10 12
with experimental data from reference (Eames and Wu, 20 0 0).

Table 2
Base case study conditions.
• For ejector driving steam temperatures of 200 °C and larger, the
COP increases, reaches a maximum and decreases progressively
T20 [°C] T14 [°C] Nozzle area ratio ( An
)
At to join the COP-curve of the double-effect absorption cycle.
200 [103–127] 5 • The COP-maximum occurs at HP-generator temperatures in the
240 [103–145] 8 range of 120 °C–125 °C, thus it is lower by 20 °C to 25 °C than
280 [103–151] 12
that of the maximum COP for the conventional double-effect
320 [103–155] 18
340 [103–158] 20 absorption cycle.
• The maximum COP is higher as the driving steam tempera-
ture increases: 1.38 for T20 = 240 ◦ C, 1.40 for T20 = 280 ◦ C, 1.43
for T20 = 320 ◦ C and 1.44 for T20 = 340 ◦ C. The calculations
show that the enhancement factor ∈, and consequently the COP
increases with increasing T20 and HP-generator temperature:
from 1.0 (no enhancement) to 1.24 for T14 = 105 ◦ C and from
1.34 to 1.7 for T14 = 135 °C, i.e., ∈= 0.24 for the lower HP-
generator temperature and ∈= 0.36 for the higher tempera-
ture.

In conclusion for these base-case simulations, we note that for

the cycle considered, the predicted COP generally increases with
the heat source temperature. So driving steam temperatures should
be kept as high as possible to achieve larger enhancement factors
and higher performances. Similar effects will be observed in the
results of the following investigations.
The object of the following simulations was to carry out a sen-
sitivity analysis of the proposed chiller configuration with respect
to the variations in thermal conditions of the chilled water (cooling
effect), the cooling medium used to dissipate the heat released in
the absorber and MP−condenser and to the operating conditions
of the ejector. Chilled water was set either at 7 °C or 15 °C, corre-
Fig. 6. COP vs. HP-generator temperature T14 for various driving steam tempera- sponding to the evaporator temperatures of 4 °C and 12 °C, respec-
tures T20 . tively. The MP−condenser and absorber were supposedly working
alternatively at 35 °C and 40 °C. Table 3 summarizes these condi-
tions. Table 4 gives the ejector geometry and the primary stream
pressure P20 , fixed by the generated steam temperature T20 , was
conditions considered for the various cases. The nozzle area ratios
analysed. Specific conditions for this base case study are given in
were selected so that for a fixed steam generator temperature, i.e.
Table 2. The simulation results are depicted in Fig. 6 representing
primary flow pressure P20 , the entrainment ratio ω was at or near
the COP evolution with HP-generator temperature T14 for various
maximum. Further, it was tested so that in each case the cycle rep-
driving steam temperatures T20 and in comparison with the COP
resentation on the Dühring−diagram was far from the crystalliza-
behaviour of a conventional double-effect absorption refrigeration
tion line, so avoiding any risk of crystallization.
cycle.
It should be noted further that the simulations are performed
In view of these results we notice that:
by assuming a value of 0.85 for nozzle and diffuser efficiencies,
• For the conventional double-effect cycle, the highest COP=1.32 (ηn = ηd = 0.85 ). Primary and secondary mixing is supposed to
is reached by an HP-generator temperature of 150 °C. provoke no notable irreversibility (ηd = 1).
462 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Fig. 7. COP vs. HP-generator temperature T14 for the cases #1 (a), #3 (b), #2 (c) and #4 (d) and various driving steam temperatures T20 .

5.1. Effect of evaporator and condenser/absorber temperatures sorber/condenser temperature, T1 = 30 ◦ C, and higher evaporator
temperature, T10 = 12 ◦ C. Following are some specific remarks.
The results of the simulations of the cases #1 to #4 are de-
picted graphically in Fig. 7. Table 5 summarizes the most impor-
tant results. • As observed earlier in the discussion of Fig. 6, the maximum
The COP of the conventional double-effect configuration ex- COP is higher as the driving steam temperature increases. In
pectedly increases with an increase in HP-generator tempera- fact, for the lowest steam temperature, T20 = 200 ◦ C, no or in-
ture T14 , it reaches a maximum and then decreases slowly. The significant effect is noticed in the case studies #1, #2 and #4,
maximum performance depends also on the evaporator and con- as the COP −curves for ejector driven and conventional cycles
denser/absorber temperature, T10 and T8 (or T1 ) respectively. The cover each other. This is also true for steam temperature T20 =
lowest COP is seen in the cycle with the lowest evaporator tem- 240 ◦ C in case #2.
perature (T10 = 4 ◦ C ) and highest condenser and absorber tem- • For cases #1 and #2, the enhancement effect of the ejector re-
perature (T1 = 40 ◦ C ): COP=1.17 at T14 = 200 ◦ C. Conversely, a compression on the maximum COP is insignificant; however,
COP=1.35 produced by an HP-generator temperature T14 = 155 ◦ C this maximum is now reached at lower HP-generator temper-
is obtained in a thermodynamically favourable case (lower ab- ature (by 25 °C less).
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 463

Table 5
Summary of simulation results.

Case # Conventional double-effect Ejector recompression double-effect T20 = 320 °C Temperature range for COP–Enhancement

1: Max COP = 1.25 Max COP = 1.25 T14 ≈ 45 °C

T8 = T1 = 35 °C @T14 = 175 °C @T14 = 150 °C
T10 = 4 °C
2: Max COP = 1.17 Max COP = 1.11 T14 ≈ 35 °C
T8 = T1 = 40 °C @T14 = 200 °C @T14 = 175 °C
T10 = 4 °C
3: Max COP = 1.35 Max COP = 1.57 T14 ≈ 55 °C
T8 = T1 = 35 °C @T14 = 155 °C @T14 = 120 °C
T10 = 12 ◦ C
4: Max COP = 1.27 Max COP = 1.35 T14 ≈ 48 °C
T8 = T1 = 40 °C @T14 = 170 °C @T14 = 145 °C
T10 = 12 °C

• The maximum enhancement of the COP is observed in case #3 The ejector model presented in the Appendix will help inter-
(35 °C at the absorber/condenser and 12 °C at the evaporator). pret the numerical simulation results and apprehend the beneficial
Here a maximum COP of 1.57 is reached at T14 = 120 ◦ C. The effects –and limits–of the integration of ejectors into an absorption
performance improvement is lower in the case of a higher cool- chiller. According to this model, the entrainment ratio depends on
ing temperature (case #4). For the highest steam temperature, nozzle area ratio, primary flow and secondary flow stagnant prop-
a COP of 1.35 is reached at T14 = 145 ◦ C. erties, and backpressure, i.e.
• The effect of the ejector recompression is limited to a specific  
An
HP-generator temperature range T14 depending on the driv- ω =f , P20 , T20 , P22 , T22 , P23 (28)
At
ing steam temperature. For higher T20 the range T14 is more
extended. The last column of Table 5 gives T14 for cases #1 to However, and as mentioned earlier, in an ejector driven chiller
#4 at the highest driving steam temperature T20 = 320 ◦ C. It is the secondary flow properties depend on the processes taking
noteworthy also that this range is dependent on the operation place in the HP-generator. These are determined by the evapo-
conditions and increases with increasing COP maximum. rator and absorber/condenser temperatures (salt concentration of
incoming solution) as well as by the temperature of the heat-
On concluding this part of the investigations, it is noted that ing coil, i.e., by the backpressure properties. For fixed driving flow
the ejector recompression technique affects the performance of the conditions and absorber/condenser temperature, the entrainment
double-effect absorption cycle in two aspects: improvement of the ratio thus depends solely on nozzle geometry (An /At ) and back
COP as in the cases #3 and #4 and the shift of maximum COP at pressure P23 , ω = f(An /At , P23 ). As an illustration of this depen-
lower HP-generator temperatures in all the cases considered. dency, Fig. 9(a) depicts the evolution of ω with nozzle area ra-
tio and backpressure for case #3 with a primary flow temperature
T20 = 240 ◦ C (P20 = 33.4 bar). It shows that for fixed values of P23
5.2. Effect of ejector operation the entrainment ratio first grows with increasing (An /At ), reaches a
maximum, then decreases and finally vanishes. Conversely, for a
We now focus attention on the ejector operation to inter- constant nozzle area ratio, ω increases with falling backpressure,
pret some of the simulation results previously presented. As an passes through a maximum and decreases to zero. Fig. 9(a) reveals
illustration, Fig. 8(a) and 8(b) depict the evolution of the entrain- also that the extent of the feasible variation range for both vari-
ment ratio for cases #1 and #2 with HP-generator temperature T14 ables reduces continuously for higher entrainment ratios, culmi-
for various primary flow temperatures T20 . One notices that ω de- nating in one single point. Further, larger values of ω are obtained
creases monotonously, quasi-linearly, with increasing T14 and van- for lower values of the variables (An /At ) and P23 . The intersection
ishes at a certain temperature. The same behaviour is observed of a horizontal plane with the 3D-surface in Fig. 9(a) (illustrated
when varying the T20 . However, when the primary flow temper- in light green for ω = 0.3) is a curve representing the set of vari-
ature is increased, the ω vs. T14 line is shifted up to higher values able couples generating the same entrainment ratio. The projec-
of the entrainment ratio, and the limit temperature T14 is pushed tion of such intersection lines on the basis horizontal plane consti-
farther away. For case #1 the maximum value of ω (0.5) is reached tutes a so-called contour plot sketch of the 3D–surface, depicted in
at the lowest possible temperature T14 = 120 ◦ C and the high- Fig. 9(b).
est considered T20 = 320 ◦ C. For the thermodynamically less favor- The 3D−surface ω = f(An /At , P23 ) as well as its contour plot
able case #2 (Fig. 8b), a maximum ω = 0.37 is obtained at T14 = representation are bounded. In Fig. 9(a) the feasible ω region is
140 ◦ C and the highest T20 . Further the range of T14 in Figs. 7 (and limited by two curves plotted in red corresponding to ω = 0. Be-
Table 5), where an enhancement of the COP is observed yond these limits the ejector is in off-design conditions and will
corresponds to the feasible, non-vanishing range of ω in Fig. 8(a) not work.
and 8(b). Consider first the limit of the feasibility domain for lower noz-
The behaviour of ω vs. backpressure depicted in Fig. 8(c) and zle area ratios (“Lower Limit” curve in Fig. 9(b)). This curve sets
(d) matches up with its evolution vs. T14 , since the flow 23 ex- a geometrical lower limit to the nozzle design. In fact, with de-
iting the ejector (Fig. 1) constitutes the heat source for the HP- creasing nozzle area ratio (Ai /At ), the nozzle exit pressure Pn ex-
generator. A rise (or drop) of backpressure P23 –and consequently pectedly rises and comes closer to the secondary flow pressure P22 .
of T23 – results in an increase (or decrease) of the HP-generator The suction of the secondary flow into the mixing chamber, driven
temperature T14 . In fact, the secondary flow pressure and salt so- by the pressure difference (P22 − Pn ), declines gradually and even-
lution in the HP-generator are monotonously increasing functions tually vanishes for Pn = P22 . Consequently, at this limit ω falls to
of the backpressure for fixed primary flow conditions and constant zero. For higher backpressure–and therefore higher secondary flow
absorber/condenser temperature. pressure –, the lower limit of (Ai /At ) is accordingly further reduced.
464 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Fig. 8. Effect of HP-generator temperature T14 (a & b) and backpressure (c & d) on entrainment ratio ω for cases #1 and #2.

The bottom border of the 3D-surface forms its intersection with the mixing chamber (shrinkage of the Mach number of primary
the ω = 0 plane and corresponds in Fig. 9(b) to the red “Upper gas from M20n to M22n ) corresponds to a maximum entrainment of
Limit” curve. The calculations show that for a fixed value of back- secondary flow.
pressure (along a horizontal line in Fig. 9(b)), the Mach number When, by applying fixed nozzle area ratio, backpressure is de-
Ma of the mixed stream a, first decreases, passes through a mini- creased, the evolution of the entrainment ratio follows a verti-
mum, then increases and becomes at last equal to that of the pri- cal line segment in Fig. 9(b), beginning on the red “Upper Limit”
mary flow at the nozzle exit. That means, the mixed gas mass flow curve with ω = 0 and ending on the “Lower Limit” curve with
rate reduces to that of the primary flow and hence, no secondary ω = 0 again. Downwards along this line segment, ω increases first,
gas is entrained, i.e. ω = 0. Concomitantly, the ejector area ratio reaches a maximum and then abruptly falls to zero. In Fig. 8 just
(Ac /At ) declines and reaches its minimum value at that limit. For those branches of the ω vs. backpressure curves are depicted
ideal flow conditions (isentropic flow and ideal mixing of primary where the entrainment ratio is falling with a rising P23 (or T14 ).
and secondary gas streams: ηn = ηm = ηd = 1), the ejector area ra- More generally, on increasing the ejector backpressure, and conse-
tio reduces, at that limit, to the nozzle area ratio, quently the secondary flow pressure, by fixing ejector geometry, a
gradual reduction in entrainment ratio is caused.
(Ac /At )min = An /At The COP evolution vs. backpressure (or alternatively HP-
generator temperature T14 ) represented in Fig. 7 exhibits rather
The maximum value of ω on this line corresponds to the low-
complex behaviour as it is conditioned by both the functioning of
est Mach number Ma , Ma = M22n . The deceleration of the gas in
 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 465

Fig. 9. (a) 3D-representation of entrainment ratio vs. nozzle area ratio and backpressure for case #3 at driving steam temperature T20 = 240 ◦ C (P20 = 33.4 bar); (b) Contour
plot representation of the same. (For interpretation of the references to colour in this figure, the reader is referred to the web version of this article.)

should be compensated by a proportional increase of ejector area

ratio to maintain constant the entrainment ratio.

6. Conclusions

An ejector driven double-effect absorption refrigeration cycle

was presented and its energy performance discussed theoreti-
cally. In the water/lithium-bromide double-effect absorption cycle
a steam ejector was incorporated in order to re-compress a frac-
tion of the vapor leaving the HP-generator for re-heating the salt
solution. Two alternative evaporation temperatures were consid-
ered (4 °C or 12 °C) and two scenarios for the cooling of con-
denser/absorber were discussed 35 °C and 40 °C. To drive the cycle,
a high temperature heat source is required for generation of high-
pressure steam.
Before running the simulations for the cycle configuration in-
vestigated in the present paper, the simulation tool was first tested
Fig. 10. Entrainment ratio vs. nozzle area ratio and ejector ratio for case #3 at driv-
and validated by comparing the calculated results with literature
ing steam temperature T20 = 240 ◦ C (P20 = 33.4 bar).
data whenever available. For the validation three tests were per-
formed. First, the ejector sub-program was validated. Then, the
simulation model was tested for an ejector driven single-effect ab-
the double-effect chiller and the ejector behaviour. By increasing sorption cycle. Finally, the main program was validated for a con-
the backpressure P23 −and concomitantly the HP-generator temper- ventional double-effect absorption chiller without an ejector. The
ature T14 − the COP tends first to increase even for a conventional results predicted by the different parts of the simulation program
double-effect absorption chiller however, the entrainment ratio in are in good concordance with the data reported in the literature.
the considered pressure range (Fig. 8) falls. The global effect is thus As regards the thermodynamic analysis of the double-effect ab-
first an increase of COP and then a decrease after passing a maxi- sorption/recompression cycle, it is worthy of note that the cy-
mum corresponding to equal but antagonistic effects. cle performance parameters are significantly affected by the pres-
Finally, Fig. 10 represents the dependency of the entrainment ence and characteristics of the ejector, as the improvement of the
ratio ω on the geometrical parameters of the ejector (An /At ) and COP largely depends on the entrainment ratio. Further, the COP
(Ac /At ) in case #3 with driving steam temperature T20 = 240 ◦ C responses to variation of working conditions were different from
(P20 = 33.4 bar). It shows the limits of geometrical feasibility for those observed for the conventional double-effect absorption re-
the specified conditions and the evolution of the performance ω frigeration cycles. The most significant results are discussed in the
in this feasible geometric domain. It is worth mention that the following.
feasible domain of the ejector design becomes narrow for higher
entrainment ratios. The intersection of the light-green horizontal • The coefficient of performance (COP) of the ejector powered
plane with the 3D-surface representing ω = f(An /At , Ac /At ) gives double-effect absorption/recompression cycle improved signif-
the locus of design of a set of values for the geometrical parame- icantly as the driving steam temperature increased. At the op-
ters yielding the same entrainment ratio (ω = 0.3 in Fig. 10). This erating conditions of the base case study, the COP-maximum
curve is nearly linear; meaning that an increase of nozzle area ratio occurred at HP-generator temperatures in the range of
466 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

120 °C–125 °C, lower by 20 °C to 25 °C than those for a max- as low as possible. The ejector geometry should be chosen for
imum COP of the conventional double-effect absorption cycle. a maximum, practical, and feasible entrainment ratio.
This maximum COP increased from 1.38 to 1.44 when the driv-
ing steam temperature was varied between 240 °C and 340 °C.
At an HP-generator temperature of 135 °C, the enhancement Acknowledgements
factor of the COP was in the range 1.34 – 1.70 for the above
mentioned interval of the driving steam temperature. Doniazed Sioud gratefully acknowledges the Tunisian Ministry
• COP values responding to changes in HP-generator temperature of Higher Education and Scientific Research for funding her intern-
were found to differ from those of the conventional double- ships at Rovira i Virgili University of Tarragona (Spain).
effect absorption cycle. In particular, the COP increased with
the heat source temperature T20 , even when the HP-generator
temperature T14 remained unchanged. Moreover, at a constant Appendix
temperature of the driving steam, the COP decreased just after
reaching its maximum value when the HP-generator tempera- Simplified 1D mathematical ejector model
ture increased.
• The ejector recompression technique affects the performance of The ejector performance is correlated to the entrainment ra-
the double-effect absorption refrigeration cycle in two ways: (i) tio ω. To predict the ejector performance, the complicated flow and
eventual improvement of the COP and (ii) shift of the maximum mixing problems within the ejector must be solved mathemati-
COP to lower HP-generator temperatures in each case consid- cally. The following simplified 1D–analysis is based on the assump-
ered. tions:
• The ejector powered double-effect absorption/recompression
• Primary and secondary fluids have comparable molecular
refrigeration cycle is a good configuration to make effective use
weights and ratios of specific heat and are correctly described
of hot heat sources at a specific temperature range in an HP-
using the ideal gas model with constant specific heat.
generator. For instance, at an evaporator temperature of 12 °C
• Primary fluid expands adiabatically in the nozzle, and mixture
and a steam generator temperature of 320 °C, the maximum
of primary and secondary fluid compresses adiabatically in the
COP of the ejector powered cycle is 1.57, i.e. 0.22 higher than
diffuser.
that of the corresponding value for the double-effect absorp-
• Isentropic efficiencies, ηn and ηd , take into account friction
tion cycle at a condenser temperature of 35 °C and 0.08 higher
losses in nozzle and diffuser.
at a condenser temperature of 40 °C with a COP= 1.35.
• Inlet velocities of primary and entrained fluids are insignifi-
• An efficient steam ejector is essential to obtain a high COP from
cant; velocity of the compressed mixture at ejector outlet is ne-
the ejector powered cycle. The enhancement of the cycle per-
glected.
formance due the presence of the ejector is however limited to
• Mixing of primary and secondary fluids in the suction chamber
a specific range of operating conditions and ejector geometry.
occurs at constant pressure.
Outside this range, the ejector is off-design. Furthermore, and
this may be of practical importance, the larger the entrainment Governing equations
ratio desired, the narrower this feasible range of operation and
geometrical parameters will be. • The primary vapour (20) expands irreversibly through the
• Generally, the following requirements should be met to make nozzle. During this process, a portion of available energy is
the ejector powered cycle run efficiently. The ratio between the transformed into kinetic energy. Assuming an isentropic effi-
solution generator pressure (secondary flow of the ejector) and ciency ηn , the velocity of the flow at the nozzle outlet (n) can
the steam pressure (primary flow of the ejector) should be kept be expressed in terms of the Mach number M20n by applying

Fig. A1. Schematic diagram of the ejector.

 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468 467

Fig. A2. Solution procedure for ejector performance analysis.

the energy conservation, The relation between the Mach numbers upstream and down-
  stream of this shock wave is given by:
2 ηn P20
γ −1
γ

2
γ −1 + Ma
2
M20n = −1 (A1)
γ −1 Pn Mb = 2γ
(A5)
γ −1 Ma − 1
2

• The secondary vapour (22) expands reversibly in the suction • The pressure increase across the shock wave is then
chamber. Its velocity at the nozzle exit plane is expressed in 
terms of the Mach number M22n of the entrained fluid
Pb Ma 1 + γ 2−1 M2a
   = (A6)
2 P22 (γ −1)/γ
Pa Mb 1 + γ 2−1 M2b
M22n = −1 (A2)
γ −1 Pn The constant pressure assumption implies Pa = Pn .

• The mixing process is modelled by 1D–continuity, momentum • The pressure lift in the diffuser is given by the relation
and energy equations. These equations are combined to de- P23 γ − 1 2 γ /(γ −1)
fine the critical Mach number M∗a of the mixture at plane (a) = 1 + ηd Mb (A7)
Pb 2
in terms of the critical Mach number for the primary and en-
trained fluids at section (n), • Further, the nozzle throat and outlet area ratio is
√   (γ +1)/(γ −1)
M∗20n +ω τ
M∗22n
M∗a = ηm  (A3) An 1 2 γ −1
(1 + ωτ )(1 + ω ) = (1 + M220n (A8)
At M220n γ +1 2

M∗ denotes the ratio of the local fluid velocity and the velocity And the ratio of the nozzle throat area to the constant section
of sound at critical conditions and τ = T22 . The relation between
T
20
area,
both Mach numbers writes,   γ −1 1/(γ −1)  P23  Pb 1/γ
 At Pb (γ −1)/γ
= 1−  2

P20 P23
(A9)
2M∗ 2 Ac P23 (1 + ωτ )(1 + ω )
M= (A4) 1 − γ 2+1
(γ + 1 ) − (γ − 1 )M∗ 2
• The Mach number, Ma , of the mixed stream can thus easily be References
calculated directly from M∗a . In the constant area section (a)-
Abed, A.M., Alghoul, M.A., Sirawn, R., Al Shamani, A.N., Sopian, K., 2015. Perfor-
(b) a sudden change of flow occurs from supersonic to subsonic mance enhancement of ejector-absorption cooling cycle by re-arrangement of
conditions that simultaneously produce a rise in static pressure. solution streamlines and adding RHE. Appl. Therm. Eng. 77, 65–75.
468 D. Sioud, M. Bourouis and A. Bellagi / International Journal of Refrigeration 99 (2019) 453–468

Álvarez, M.E., Esteve, X., Bourouis, M., 2015. Performance analysis of a triple-effect Jiang, L., Gu, Z., Feng, X., Li, Y, 2002. Study of new absorption-ejector hybrid refrig-
absorption cooling cycle using aqueous (lithium, potassium, sodium) nitrate so- eration system. In: Proceedings of the International Refrigeration and Air Con-
lution as a working pair. Appl. Therm. Eng. 79, 27–36. ditioning Conference.
Aphornratana, S., Eames, I.W., 1998. Experimental investigation of a combined ejec- Kaita, Y., 2002. Simulation results of triple-effect absorption cycles. Int. J. Refrig. 25
tor-absorption refrigerator. Int. J. Energy Res. 22 (3), 195–207. (7), 999–1007.
Azhar, Md., Siddiqui, M.A., 2013. thermodynamic analysis of a gas operated triple ef- Kim, J.S., Ziegler, F., Lee, H., 2002. Simulation of the compressor-assisted triple-effect
fect absorption cycle. Int. J. Innov. Res. Sci. Eng. Technol. (ISSN) 2 (5), 1610–1616. H2 O/LiBr absorption cooling cycles. Appl. Therm. Eng. 22 (3), 295–308.
Chen, Y.M., Sun, C.Y., 1997. Experimental study of the performance characteristics of Klein S.A., Alvarado F., 2005. Engineering Equation Solver, Version 7.441. Middleton,
a steam-ejector refrigeration system. Exp. Therm. Fluid Sci. 15 (4), 384–394. WI: F–chart software.
Eames, I.W., Wu, S., 20 0 0. A theoretical study of an innovative ejector powered ab- Lizarte, R., Marcos, J.D., 2016. COP optimisation of a triple-effect H2 O/LiBr absorption
sorption-recompression cycle refrigerator. Int. J. Refrig. 23 (6), 475–484. cycle under off-design conditions. Appl. Therm. Eng. 99, 195–205.
Eames, I.W., Wu, S., 20 0 0. Experimental proof-of-concept testing of an innovative Majdi, H.Sh., 2016. Performance evaluation of combined ejector LiBr/H2O absorption
heat-powered vapour recompression-absorption refrigerator cycle. Appl. Therm. cooling cycle. Case Stud. Therm. Eng., 7, 25–35.
Eng. 20 (8), 721–736. Sözen, A., Özalp, M, 2003. Performance improvement of absorption refrigeration
Erickson, D.C., Potnis, S.V., Tang, J., 1996. Triple effect absorption cycles–refrigera- system using triple-pressure-level. Appl. Therm. Eng. 23 (13), 1577–1593.
tion heat pumps. In: Proceedings of the 31st Intersociety. Energy Conversion Shi, L., Yin, J., Wang, X., Zhu, M.S., 2001. Study on a new ejection-absorption heat
Engineering Conference (IECEC), pp. 1072–1077. 2. transformer. Appl. Energy 68 (2), 161–171.
Farshi, L.G., Mahmoudi, S.M.S., Rosen, M.A., Yari, M., 2012. Use of low-grade heat Shirazi, A., Taylor, R.A., White, S.D., Morrison, G.L., 2016. A systematic paramet-
sources in combined ejector–double effect absorption refrigeration systems. J. ric study and feasibility assessment of solar-assisted single-effect, double-effect,
Power Energy 226 (5), 607–622. and triple-effect absorption chillers for heating and cooling applications. Energy
Garousi Farshi, L., Mosaffa, A.H., Infante Ferreira, C.A., Rosen, M.A., 2014. Thermody- Convers. Manag. 114, 258–277.
namic analysis and comparison of combined ejector–absorption and single-ef- Sirwan, R., Alghoul, M.A., Sopian, K., Ali, Y., Abdulateef, J., 2013. Evaluation of adding
fect absorption refrigeration systems. Appl. Energy 133, 335–346. flash tank to solar combined ejector–absorption refrigeration system. Solar En-
Gebreslassie, B.H., Medrano, M., Boer, D., 2010. Exergy analysis of multi-effect wa- ergy 91, 283–296.
ter–LiBr absorption systems: from half to triple effect. Renew. Energy 35 (8), Somers, C., 2009. Modelling Absorption Chillers in aspen, “Simulation of Absorption
1773–1782. Cycles For Integration Into Refining processes”. University of Maryland, College
Grossman, G., Wilk, M., 1994. Advanced modular simulation of absorption systems. Park, MD, USA Masters of Science.
Int. J. Refrig. 17 (4), 231–244. Sun, D.W., Eames, I.W., Aphornratana, S., 1996. Evaluation of a novel combined ejec-
Grossman, G., Wilk, M., 1993. Enhanced Absorption Cycle Computer Model. Oak tor-absorption refrigeration cycle I: computer simulation. Int. J. Refrig. 19 (3),
Ridge National Laboratory, Oak Ridge-Tennessee, p. 37831. 172–180.
Grossman, G., Wilk, M., Devault, R.C., 1993. Simulation and performances analysis of Vereda, C., Ventas, R., Lecuona, A., Venegas, M., 2012. Study of an ejector-absorp-
triple-effect absorption cycle. In: Proceedings of the ASHRAE Winter Meeting. tion refrigeration cycle with an adaptable ejector nozzle for different working
Hong, D., Chen, G., Tang, L., He, Y., 2011. A novel ejector-absorption combined re- conditions. Appl. Energy 97, 305–312.
frigeration cycle. Int. J. Refrig. 34 (7), 1596–1603. Wu, S., Eames, I.W., 1998. A novel absorption-recompression refrigeration cycle.
Huang, B.J., Chang, J.M., Wang, C.P., Petrenko, V.A., 1999. A 1-D analysis of ejector Appl. Therm. Eng. 18 (11), 1149–1157.
performance. Int. J. Refrig. 22 (5), 354–364.

View publication stats