Вы находитесь на странице: 1из 10

International Journal of Refrigeration 28 (2005) 1092–1101

www.elsevier.com/locate/ijrefrig

CFD analysis of ejector in a combined ejector cooling system


E. Ruslya, Lu Ayea,*,1, W.W.S. Chartersb,1, A. Ooib
a
International Technologies Centre (IDTC), Department of Civil and Environmental Engineering, The University of Melbourne,
Melbourne, Vic. 3010, Australia
b
Department of Mechanical and Manufacturing Engineering, The University of Melbourne, Melbourne, Vic. 3010, Australia
Received 3 December 2003; received in revised form 10 February 2005; accepted 23 February 2005
Available online 23 May 2005

Abstract
One-dimensional ejector analyses often use coefficients derived from experimental data for a set of operating conditions with
limited functionality. In this study, several ejector designs were modelled using finite volume CFD techniques to resolve the
flow dynamics in the ejectors. The CFD results were validated with available experimental data. Flow field analyses and
predictions of ejector performance outside the experimental range were also carried out. During validation, data from CFD
predicted the entrainment ratios with greater accuracy on definite area ratios, although no shock was recorded in the ejector.
Predictions outside the experimental range—at operating conditions in a combined ejector–vapour compression system—and
flow conditions resulting from ejector geometry variations are discussed. It is found that the maximum entrainment ratio
happens in the ejector just before a shock occurs and that the position of the nozzle is an important ejector design parameter.
q 2005 Elsevier Ltd and IIR. All rights reserved.

Keywords: Refrigerating system; Ejector system; Modelling; CFD; Performance

Ejecteur dans un système de refroidissement: analyse à l’aide de la


dynamique des fluides
Mots clés : Système frigorifique ; Système à éjecteur ; Modélisation ; Dynamique numérique des fluides ; Performance

1. Introduction because of their simplicity, reliability, low installation and


operational costs. Their typically low COP of around 0.2 can
Ejector systems utilising natural or renewable thermal be improved substantially when combined with other
energy to run environment-friendly refrigerants can be a systems such as absorption or vapour-compression.
promising alternative to current systems. Compared to (Aphornratana and Eames [3], Sun [19]). A typical ejector
absorption systems, ejector systems have advantages construction comprises four distinct parts: a convergent–
divergent nozzle, a suction chamber attached to a constant
area duct and a diffuser. The typical processes inside an
* Corresponding author. Tel.: C61 3 8344 6879; fax: C61 3 8344 ejector begin with a high temperature and high pressure
6868. stream from the generator entering the ejector through the
E-mail address: lua@unimelb.edu.au (L. Aye). convergent–divergent nozzle. This stream is accelerated and
1
Member of commission E2 of IIR. expanded to supersonic speed at the nozzle exit where it
0140-7007/$35.00 q 2005 Elsevier Ltd and IIR. All rights reserved.
doi:10.1016/j.ijrefrig.2005.02.005
E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101 1093

creates an aerodynamic duct to entrain the low pressure and geometry could be modelled exactly as per the ejector
low temperature secondary stream into the mixing chamber. specifications in the experiments. Out of 13 cases tested, 10
The secondary stream is accelerated to sonic velocity and had entrainment ratios closer to the experimental data than
mixes with the primary stream in the constant area duct. The those predicted by the one-dimensional model. In the next
region of supersonic flow is terminated by a normal shock phase of study, a basic ejector design was derived to suit a
wave further down the duct or in the diffuser. Across the combined ejector–vapour compression system. This full-
shock, pressure increases but Mach number (velocity) scale ejector geometry was then modelled in CFD and
reduces to a subsonic value. The mixed stream then enters examined at the conditions outside the one-dimensional
the subsonic diffuser where it undergoes a re-compression model calibration range for further analysis of the ejector
process to reach the back-pressure (condenser pressure) at flow field. An attempt was also made to obtain a flow field as
near zero velocity. similar as possible to that modelled in the one-dimensional
Theoretical analyses of ejectors often use a one- analysis—with uniform mixing pressure, choked secondary
dimensional perspective and have mainly been used to flow and shock. The condition at which the maximum
compare different refrigerants and evaluate system entrainment ratio occurred was discovered. To achieve the
characteristics under varying operating conditions highest possible throughput, the constant area diameter and
(Eames et al. [4], Huang et al. [10], Sun [20]). When the nozzle exit distance relative to the entry of the constant
further details of the flow field are required, this method area duct were varied under several sets of operating
is often inadequate. conditions using CFD analysis.
Another approach to ejector modelling is to use
computational fluid dynamics (CFD) that is capable of 1.1. Combined ejector–vapour compression system
producing details of the flow field and fluid properties based
on numerical solutions of the flow domain. Specific types of The hybrid of the simple and reliable ejector system and
flow involving heat transfer, radiation, turbulence, etc. at the technologically mature vapour compression system can
any given operating conditions or model geometry can be be very beneficial. When solar energy or the cost-effective
simulated for extensive analysis of the flow related proper- waste heat can be introduced as the ejector’s heat source, the
ties. Data that is difficult to obtain in an experimental set up combined system can also realistically achieve a COP
can be easily analysed using CFD. improvement of around 20–50% (Sun [19], Sokolov and
Recent CFD studies on ejectors that modelled the flow as Hershgal [18]).
incompressible discounted the flow’s supersonic nature This combined system comprises an ejector cooling
(Smith et al. [17]); others treated it as compressible but sub-cycle operating at high suction (evaporating)
failed to capture some fundamental aspects of the flow such temperature of 25–32 8C range for high efficiency and
as the secondary flow choking and the shock wave (Riffat a vapour compression sub-cycle at 5–45 8C range for
and Everitt [13]). In this study, the experimental data of reduced compression work. The heat supply can have a
Huang et al. [9] was used to validate the results obtained temperature as low as 80 8C from cheap low-grade
from CFD modelling. Comparisons were made for CFD and waste heat or solar energy. At the intercooler the
one-dimensional predictions to the corresponding experi- refrigerant vapour from the vapour-compression side
mental data. In CFD, the area ratios as part of the ejector rejects heat to the liquid refrigerant from the ejector

Fig. 1. Schematic diagram of combined ejector–vapour compression system.


1094 E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101

side and evaporates it. A schematic diagram of the available in the literature (Keenan et al. [11], Henzler [7],
combined system is shown in Fig. 1. ESDU [5] and Al-Khalidy and Zayonia [1]).
Keenan et al. [11] depicted the length of the constant area
duct as a function of the pressure ratio between nozzle inlet
and suction chamber inlet in a chart. For the pressure ratio
2. Model description considered, an (LCy)/D (where L is length of constant-area
section; y is distance from primary nozzle exit to beginning
2.1. Refrigerant selection of constant-area portion of mixing tube and D is constant
area duct diameter) of 10 resulted. Half this length was to be
The working fluid in the combined cycle was selected the distance of the nozzle exit to the constant area duct entry
based on the criteria and simulation results in Rusly et al. to yield a fair approximation of constant pressure mixing.
[15]. Apart from being an HCFC, R141b is the most The other half was, therefore, the length of the constant area
favourable refrigerant for the given intercooler temperature duct. Hence, the two parameters were equal at 5 times the
range. Two other potential refrigerants identified were constant area duct diameter. Diffuser length was suggested
R245ca and R245fa but they are expensive and not readily to be 8 times the diameter of the constant area duct
available. according to Al-Khalidy and Zayonia [1] and Henzler [7].
Diffuser diverging angle was selected at 3.58 following
Henzler [7] and ESDU [5]. Nozzle converging and
2.2. Ejector design (base model)
diverging angles and suction chamber converging angle of
12, 7 and 108 were selected, respectively, as per ESDU [5].
The ejector design for the combined ejector–vapour
Fig. 2 shows the complete geometry of the ejector.
compression system was instigated from a one-dimensional
The ejector geometry obtained was modelled as a 2D
analysis. The low generator temperature at 80 8C enabled
axi-symmetric model in Fluent, a commercial CFD software
heat utilisation from various cheap sources such as industrial
package. The grid structure was triangular unstructured to
waste heat, flue gas, automobile exhaust or solar energy, etc.
conform to the curving nature of the ejector wall. The grid
It is well documented that high evaporating (or suction)
was initially made at about 40,000 elements and later
temperature increases ejector performance. This is greatly
adapted to about 120,000 elements to confirm that the
compromised in a conventional ejector refrigeration system
results are grid independent. The triangular mesh allows
because the meaning of refrigerating or air conditioning
flexible meshing on areas of high gradient such as the nozzle
diminishes with increasing evaporator temperature. In stark
exit and the constant area duct where a normal shock is
contrast, the combined system permits a much higher
expected. The turbulence model employed was realizable k–
evaporating (or intercooler) temperature to absorb the
3 (Shih et al. [16]) with compressible flow nature across the
condensation heat of the low-pressure compression cycle.
entire ejector. More detailed information regarding the CFD
In this study, 32 8C was selected as the ejector sub-cycle’s
modelling can be found in Rusly [14].
evaporating temperature, sufficiently high for the ejector to
achieve high efficiency but low enough to allow the
compression cycle to reject heat. The condensing tempera- 2.3. Effects of ejector geometry on performance
ture of the ejector sub-cycle at 40 8C was chosen to be as low
as practical for best system performance yet high enough to The constant area diameter was thought to be the most
allow heat rejection to the sink. influential parameter on ejector performance due to its direct
The constant-pressure-mixing one-dimensional model proportion with the secondary flow area. This parameter
based on Huang et al. [9] was developed for the analysis. manifests in the ejector area ratio defined as follows: Area
The model analysed the ejector at critical-mode operation ratioZarea of constant area duct/area of nozzle throat.
where the primary and secondary flows were both choked. A range of area ratios from 6.44 to 10.64 was selected
The geometrical output from the model was the constant and experimentally tested in Huang et al. [9]. They claimed
area section diameter having satisfied all the fluid dynamics an entrainment ratio of 0.7 at 100 8C generator temperature
constraints including constant pressure mixing and a normal could be achieved for an ejector area ratio of around 16.
shock in the ejector. Other ejector geometries were However, they did not explain why 16 was not used for all
determined from relevant ejector design guidelines the area ratios of the ejectors tested and what would happen

Fig. 2. Ejector geometry modelled in CFD and used as the base model.
E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101 1095

if the area ratio was increased beyond 16. Of equal suction chamber and constant area dimensions were
importance is the nozzle exit distance relative to the replicated precisely and all other supplementary dimensions
constant area duct entrance. This parameter was exper- were derived from the ejector design guidelines. Complete
imentally varied and a distance of 1.5 the constant area duct flow information was obtained from CFD analysis but
diameter was found to be the optimum distance (Huang et al. comparisons with the experimental data and one-dimen-
[9]). sional predictions could only be made with the published
In the present CFD study, these two parameters were entrainment ratios and area ratios.
investigated for their role in ejector performance supported The ejectors simulated in CFD were AG, AB, EC and EE
by extensive flow information and properties. The two (nozzle and constant-area section identification codes used
parameters were varied independently at constant operating in the experiments of Huang et al. [9]) at the operating
conditions and the flow field data was studied to provide conditions tested in Huang et al. [9]. Table 1 summarises
insights into the ejector geometry–performance relationship. this along with the entrainment ratios, area ratios and errors
Variations of constant area diameter involved diameter from the two methods compared with the experimental data.
change of C20, C40, C60, C80% up to flush with the Additionally, the primary and secondary mass flow rates
diffuser. The nozzle exit distance was varied from the base from CFD analysis are also shown.
model design by G20% with C indicating increasing Fig. 3 depicts the graphical comparison of the three
distance away from the constant area duct entry. The nozzle methods for entrainment ratio. It is shown that CFD
exit distance of 1.5 diameter of constant area duct was also produced entrainment ratio closer to the experimental
simulated for comparison. results than the one-dimensional analysis in 10 out of 13
cases tested. The errors from CFD analysis were very low,
lower than 10% compared to those from one-dimensional
3. Results and discussion predictions of up to 23%. Moreover, the CFD results showed
that in all of Huang et al.’s [9] experimental test cases
3.1. CFD results validation simulated, no shock was present in the ejector, which was
inconsistent with the assumption of the presence of a shock
The experimental data in Huang et al. [9] was used to in their one-dimensional model.
validate the CFD results from modelling the ejectors used in
their experiments with similar operating conditions. This 3.2. CFD analysis outside experimental range
data had a 5% uncertainty from the two rotameters used to
measure the primary and entrained flows of the ejector 3.2.1. Base model
(Huang and Chang [8]). The nozzle diameters and distance, The boundary conditions for the base model were set as

Table 1
Summary of test conditions, performance results and errors from CFD and one-dimensional analyses compared with experimental results of
Huang et al. [9]

Tg 8C Tc 8C Te 8C 1D Exper- CFD 1D Exper- CFD 1D error CFD CFD pri CFD


AR iment AR model iment ER ER (%) error (%) flow sec flow
AR ER (kg/s) (kg/s)
AG ejector
78 29.2 12 8.26 7.73 7.73 0.6659 0.6132 0.5775 8.59 K5.82 0.008663 0.005003
84 32.4 12 8.17 7.73 7.73 0.4769 0.4790 0.4835 K0.44 0.94 0.009973 0.004822
90 36.0 12 8.00 7.73 7.73 0.4142 0.4034 0.3831 2.68 K5.03 0.011421 0.004375
95 38.7 12 7.96 7.73 7.73 0.3434 0.3503 0.3548 K1.97 1.30 0.012744 0.004522
78 29.1 8 7.64 7.73 7.73 0.4609 0.4393 0.4395 4.92 0.05 0.008676 0.003813
84 32.3 8 7.68 7.73 7.73 0.3704 0.3883 0.3977 K4.61 2.41 0.009972 0.003966
90 36.7 8 7.03 7.73 7.73 0.2395 0.3040 0.3343 K21.22 9.96 0.011422 0.003818
95 38.6 8 7.38 7.73 7.73 0.2144 0.2552 0.2480 K15.99 K2.83 0.013323 0.003304
AB ejector
78 29.5 8 7.48 6.99 6.99 0.4422 0.3922 0.4097 12.75 4.45 0.008660 0.003548
84 33.6 8 6.99 6.99 6.99 0.3042 0.3117 0.3370 K2.41 8.13 0.009972 0.003361
90 37.5 8 6.65 6.99 6.99 0.2093 0.2718 0.2749 K22.99 1.13 0.011942 0.003283
EC ejector
95 39.3 12 7.69 7.26 7.26 0.3235 0.3040 0.3154 6.41 3.76 0.014456 0.004560
EE ejector
95 34.2 8 8.89 9.17 9.17 0.3253 0.3505 0.3684 K7.19 5.11 0.014456 0.005326

Tg, generator temperature; Tc, condenser temperature; Te, evaporator temperature; AR, area ratio; ER, entrainment ratio; pri flow, primary flow
rate; sec flow, secondary flow rate.
1096 E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101

in Section 2.2, i.e. 80, 40 and 32 8C for the generator,


condenser and evaporator temperatures, respectively, for
R141b. Other properties such as specific heat, thermal
conductivity, viscosity and molecular weight were derived
from R141b’s real fluid thermodynamic properties available
in REFPROP (McLinden et al. [12]). Density was obtained
using the ideal gas relationship; a convention for compres-
sible flow formulation in Fluent (Fluent 6 Manual [6]). Since
the operating conditions were well below the critical points,
the compressibility factor for such flows tended to be near
one making this approach justifiable.
A Mach-number contour plot of the solution shows a shock
Fig. 3. Comparison between CFD, one-dimensional model predic- at the nozzle tip (Fig. 4(a)). The flow downstream of the shock
tions of entrainment ratios of ejectors AG, AB, EC and EE with the was all sub-sonic with reversed flow to the secondary flow
experimental results of Huang et al. [9]. suction inlet. This is called a back-flow or malfunction mode
and the ejector at this condition does not function (ejector
operational modes in Huang and Chang [8]).

Fig. 4. Mach number plots of the ejectors at 80, 40 and 32 8C for generator, condenser and evaporator temperatures at different nozzle exit
distance from the constant area duct entry.
E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101 1097

If the nozzle exit was moved to a distance 1.5 its experiments—the shock became a diamond shape convex
diameter from the constant area duct entrance and the shock. Downstream, the primary flow formed a short sonic
suction chamber converging angle was increased from 10 to convergent duct creating suction for the secondary flow. The
308 (Fig. 4(b))—as practised in Huang et al.’s [9] ejector operated with an entrainment ratio of about 0.67. The

Fig. 5. Mach number contour plots of the three ejector flow field scenarios and flow field magnification at nozzle exit and constant area duct
entry.
1098 E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101

secondary flow was sub-sonic throughout and at this condition and reduced the primary flow velocity. At the diffuser, the
the ejector operated in a single (primary flow) choking or sub- mixed flow was subsonic and decelerated further to reach
critical mode. This condition is unstable in that a slight the back pressure. The flow field at the nozzle exit contained
increase of the back pressure will upset the system balance and a series of oblique shocks and expansion waves that formed
drop the entrainment ratio rapidly towards zero. a converging duct characteristic of nozzle over-expansion to
pre-compress the primary flow before mixing took place
(Fig. 5(a)). Increasing the generator temperature caused this
3.2.2. Three stages of ejector flow field
In an attempt to obtain the flow field assumed in one- over-expansion to slowly subside and the flow field was
dimensional analysis, the boundary conditions of the ejector nearly shock free at 118 8C (scenario 2). This near isentropic
were changed. The evaporating and condensing temperatures condition had a flow signature close to the assumptions of
were fixed to maintain the system balance. Hence the generator constant pressure mixing and shock in Huang et al.’s [9]
was the main variable to alter. The pressure difference between one-dimensional analysis. The primary flow’s supersonic
the ejector primary inlet and the outlet governs the secondary duct (of around Mach 2.35) choked the secondary flow at the
flow entrainment as reflected in the state of nozzle exit flow. A constant area duct entrance and formed an oblique shock in
normal shock at the nozzle exit such as in Fig. 4(a) indicates the diffuser. Mixing appeared at a relatively constant
that this pressure difference is not high enough for the ejector pressure and the ejector was now operating in the double-
to function. To increase it, the restricted condenser pressure choking or critical mode (Fig. 5(b)). Increasing the
cannot be lowered but the generator pressure can be increased generator temperature further led the ejector to enter the
to obtain the same effect. under expansion state (scenario 3). The shock took place
As the generator temperature was increased to 85 8C further into the diffuser and at 150 8C it was totally normal
(scenario 1), the shock disappeared from the nozzle exit and to the flow (Fig. 5(c)). In all cases, the mixed flow across the
was replaced by a supersonic converging duct extending to shock was brought to sub-sonic velocity with consequent
the constant area duct exit with velocity as high as Mach
1.75. Mixing of the two flows choked the secondary flow

Fig. 6. Static pressure distributions at 85, 118 and 150 8C generator Fig. 7. Mach number distributions at 85, 118 and 150 8C generator
temperatures across entire ejector for primary and secondary flows. temperatures across entire ejector for primary and secondary flows.
E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101 1099

strong compression. Through the remainder of the diffuser, 3.3. Nozzle exit distance and constant area diameter
the flow underwent further gradual compression to a near variations
zero velocity at the outlet. Fig. 5 shows Mach number
contour plots of the three different flow fields along the 3.3.1. Position of nozzle exit
ejector and into the constant area duct. Figs. 6 and 7 depict Aphornratana and Eames [2] reported that moving the
the static pressure and Mach number distributions for the nozzle into the mixing chamber—at constant generator
primary and secondary flows across the ejector for the three input—reduced the system COP or cooling capacity with a
scenarios above; (a) for scenario 1, (b) for scenario 2 and (c) resultant higher critical condenser pressure and vice versa.
for scenario 3. Varying the nozzle exit position by 20% with all other
Entrainment ratio increased to a maximum of 1.04 at conditions unchanged did not affect the entrainment ratio
85 8C generator temperature when the primary flow exiting significantly. However, the results seemed to support the
the nozzle was in the over-expansion state with no shock in characteristics above with lower entrainment ratio resulting
the ejector. Further increase reduced it to less than 0.5 at the when the nozzle exit was moved closer to the constant area
near isentropic condition. In the under-expansion state, the duct entry (Table 3). At a nozzle exit distance 1.5 the
entrainment ratio decreased further and the shock took place constant area duct diameter, the ejector produced better
further in the diffuser. Table 2 below summarises flow performance than the base model—with its nozzle exit
information as a result of increasing the generator distance derived from Keenan et al.’s [11] prescription—as
temperature from 80 to 150 8C. The ejector model used in shown in Fig. 4.
this part of the study had a nozzle exit distance from the
constant area duct entry 1.5 the constant area duct diameter 3.3.2. Constant area diameter
for better performance. Changing the constant area duct diameter affects ejector
It can be seen that the maximum entrainment ratio performance dramatically. The diameter was increased from
occurred at 85 8C generating temperature that produced a the base model in 20% increments until flush with the
pressure difference between the ejector primary inlet and diffuser exit. Two sets of operating conditions were used for
back pressure which choked the secondary flow in over- this investigation, i.e. 85 and 118 8C generator temperatures
expansion state (scenario 1) just before a shock would take with 40 8C condenser temperature and 32 8C evaporator
place at a higher generator temperature. temperature to represent ejector operation at maximum

Table 2
Flow information as a result of generator temperature increase

Tg, Tc, Te Pri. flow rate Sec. flow rate ER Pri. flow state Sec. flow Shock Shock position
choking
80, 40, 32 0.009083 0.006050 0.6661 Over- No No –
expanded
85, 40, 32 0.010670 0.011108 1.0411 Over- Entrance of No –
expanded CAS
87, 40, 32 0.011166 0.011028 0.9876 Over- Entrance of No –
expanded CAS
90, 40, 32 0.011943 0.010911 0.9136 Over- Entrance of Oblique 5% Inside
expanded CAS diffuser
100, 40, 32 0.014804 0.010678 0.7213 Over- 12% Inside Oblique 15% Inside
expanded CAS diffuser
110, 40, 32 0.018132 0.010381 0.5725 Over- 12% Inside Oblique 25% Inside
expanded CAS diffuser
115, 40, 32 0.019997 0.009985 0.4993 Over- 15% Inside Oblique 33% Inside
expanded CAS diffuser
118, 40, 32 0.021171 0.009783 0.4621 Near isentro- Entrance of Oblique 40% Inside
pic CAS diffuser
120, 40, 32 0.021977 0.009672 0.4401 Near isentro- Entrance of Oblique 45% Inside
pic CAS diffuser
130, 40, 32 0.026342 0.009347 0.3548 Under- 12% Inside Oblique 55% Inside
expanded CAS diffuser
150, 40, 32 0.036851 0.009672 0.2625 Under- 12% Inside Normal 80% Inside
expanded CAS diffuser

CAS, constant area section; Tg, generator temperature (8C); Tc, condenser temperature (8C); Te, evaporator temperature (8C); pri. flow rate and
sec. flow rate in kg/s, ‘% inside’ indicates distance from the entrance of the section relative to the whole section (CAS or diffuser).
1100 E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101

Table 3
CFD simulation results of nozzle exit distance variation

Ejector model Mass flow rate (kg/s) Entrainment ratio


Primary Secondary
NEDK20% 0.027243 0.020797 0.7634
Base 0.02724 0.02225 0.8165
NEDC20% 0.027244 0.022275 0.8176

Tg, 130 8C; Tc, 40 8C; Te, 32 8C; NED, nozzle exit distance.

entrainment ratio and at ‘close to’ the one-dimensional- choking and the shock. This indicates the ejector was
model-assumed-condition. operating at the unbalance-single-choking mode although
At 85 8C generator temperature, increasing the diameter the entrainment ratio was still rising to 1.2. Further increase
by 20% resulted in around 25% entrainment ratio decrease of the constant area duct diameter resulted in entrainment
from 1.04 to 0.785 (Table 4). The secondary flow ratio reduction in the unstable sub-critical mode.
choking disappeared and the ejector was operating in the
over-expanded state with single choking. At 40% diameter
increase, the entrainment ratio dropped very rapidly to 4. Conclusions
around 0.069. The next increment at 60% caused the ejector
to go into the malfunction mode where entrainment changed From the investigation, it can be concluded that the
to flow reversal towards the suction chamber inlet. constant area diameter increase will only improve ejector
At 118 8C generator temperature, the flow field was performance when it is operating in critical mode with a
characterised by double choking and the presence of a shock in the diffuser. It also follows that in order to obtain
shock. Increasing the constant area diameter shifted the maximum output, the constant area diameter can be
secondary flow choking position towards the diffuser with a continuously increased until the shock disappears but the
resultant increase in the entrainment ratio (Table 4). At 40% secondary flow choking remains. The optimum area ratio for
diameter increase, the secondary flow choking occurred at a the 85, 40, 32 C case is 7.27 and for the 118, 40, 32 C case is
sixth of the way into the constant area duct and the 16.4.
entrainment ratio increased to 1.18. Enlarging the diameter Several other concluding remarks can be presented from
further to 60% saw the ejector lose the secondary flow the present study.

Table 4
CFD simulation results of constant area diameter increase

Operating conditions Area ratio ER Change (%) Pri. flow state Sec. flow Shock
choking
85, 40, 32 Base model 7.27 1.0411 0 Over- Entrance of No
expanded CAD
C20% 10.47 0.785 K24.6 Over- No No
expanded
C40% 14.26 0.0693 K91.2 Over- No No
expanded
C60% 18.62 Reversed – – – –
C80% 23.57 Reversed – – – –
118, 40, 32 Base model 7.27 0.4621 0 Near isentro- Entrance of Oblique
pic CAD
C20% 10.47 0.8108 75.5 Near isentro- Entrance of Oblique
pic CAD
C40% 14.26 1.1753 45.0 Near isentro- 15% Inside Oblique
pic CAD
C60% 18.62 1.2002 2.1 Over- No No
expanded
C80% 23.57 0.9875 K17.7 Over- No No
expanded
Flush with 28.44 0.9455 K4.3 Over- No No
diffuser expanded

CAD, constant area duct; ER, entrainment ratio; pri., primary; sec., secondary. ‘Operating conditions’ indicates Tg, Tc, Te in 8C. Change (%) is
calculated from current and previous results.
E. Rusly et al. / International Journal of Refrigeration 28 (2005) 1092–1101 1101

Comparison of one-dimensional and CFD analyses: [2] S. Aphornratana, I.W. Eames, A small capacity steam-ejector
refrigerator: experimental investigation of a system using
1. CFD analysis offers more complete flow field infor- ejector with movable primary nozzle, Int J Refrigeration 20 (5)
mation for ejector simulations and produces a better (1997) 352–358.
correlation coefficient to experimental results than the [3] S. Aphornratana, I.W. Eames, Experimental investigation of a
one-dimensional model. combined ejector–absorption refrigerator, Int J Energy Res 22
2. At the test conditions used in Huang et al. [9], the ejector (1998) 195–207.
flow fields produced from CFD simulations do not [4] I.W. Eames, S. Aphornratana, H. Haider, A theoretical and
indicate a shock as assumed in their one-dimensional experimental study of a small-scale steam jet refrigerator, Int J
Refrigeration 18 (6) (1995) 378–386.
model.
[5] ESDU, Ejectors and jet pumps, data item: 92042, ESDU
CFD prediction outside experimental range: International Ltd, London, UK, 1992.
[6] Fluent 6 manual, Lebanon, NH, USA: Fluent Inc.
3. CFD results indicate three distinct scenarios existing in [7] H.J. Henzler, Design of ejectors for single-phase material
ejector flow field, i.e. over-expansion, near isentropic systems, German Chem Eng 6 (1983) 292–300.
[8] B.J. Huang, J.M. Chang, Empirical correlation for ejector
and under-expansion pertinent to ejector operating
design, Int J Refrigeration 22 (1999) 379–388.
conditions from the constant pressure mixing ejector
[9] B.J. Huang, J.M. Chang, C.P. Wang, V.A. Petrenko, A 1-D
simulations.
analysis of ejector performance, Int J Refrigeration 22 (1999)
4. The condition closest to the assumptions of constant 354–364.
pressure mixing and shock presence in the one-dimen- [10] B.J. Huang, V.A. Petrenko, I.Y.A. Samofatov,
sional model does not provide the maximum entrainment N.A. Shchetinina, Collector selection for solar ejector cooling
ratio. system, Sol Energy 71 (4) (2001) 269–274.
5. Maximum entrainment ratio occurs at a pressure ratio [11] J.H. Keenan, E.P. Neumann, F. Lustwerk, An investigation of
between the ejector primary inlet and diffuser outlet just ejector design by analysis and experiment, ASME J Appl
adequate to choke the secondary flow but avoid a shock Mech 1950; 299–309.
in the diffuser. [12] M.O. McLinden, E.W. Lemmon, S.A. Klein, A.P. Peskin,
NIST thermodynamics properties of refrigerants and refriger-
General design considerations: ant mixtures database (REFPROP) version 6.0. Boulder, CO,
USA: National Institute of Standards and Technology; 1998.
6. At the combined system operating condition (80 8C [13] S.B. Riffat, P. Everitt, Experimental and CFD modelling of an
generating, 40 8C condensing and 32 8C evaporating ejector system for vehicle air conditioning, J Inst Energy 72
temperatures), nozzle exit position placed at 1.5 the (1999) 41–47.
diameter of constant area duct produces better perform- [14] E. Rusly, Ejector cooling with reference to combined ejector–
ance than that prescribed in Keenan et al.’s [11] vapour compression system. PhD Thesis. Australia: Depart-
experiments. Generally, 20% variation upstream or ment of Civil and Environmental Engineering, The University
downstream does not affect performance significantly. of Melbourne.
7. Area ratio increase improves ejector performance only [15] E. Rusly, W.W.S. Charters, A. Ooi, L. Aye, Combined solar
and electric system for space cooling. Heat pumps—better by
when it operates in the critical (double-choking) mode.
nature. 7th International Energy Agency Heat Pump Con-
8. Area ratio can be increased to allow the shock to
ference, May 19–22, Beijing, China; 2002. p. 301–312.
disappear from the diffuser but the choking of the
[16] T.H. Shih, W.W. Liou, A. Shabbir, Z. Yang, J. Zhu, A new k–3
secondary flow to remain for a higher entrainment ratio. eddy viscosity model for high Reynolds number turbulent
flows, Comput Fluids 24 (3) (1995) 227–238.
[17] S.J. Smith, S.B. Riffat, S. Wu, I. Eames, Low-pressure
Acknowledgements ejectors: prediction of performance by computational fluid
dynamics, Building Serv Eng Res Technol 18 (3) (1997) 179–
Eddy Rusly wishes to acknowledge the support of The 182.
University of Melbourne through the provision of a [18] M. Sokolov, D. Hershgal, Solar powered compression-
Melbourne Research Scholarship. enhanced ejector air conditioner, Sol Energy 51 (3) (1993)
183–194.
[19] D.W. Sun, Solar powered combined ejector–vapour com-
pression cycle for air conditioning and refrigeration, Energy
References
Convers Manage 38 (5) (1997) 479–491.
[20] D.W. Sun, Comparative study of the performance of an ejector
[1] N. AL-Khalidy, A. Zayonia, Design and experimental refrigeration cycle operating with various refrigerants, Energy
investigation of an ejector in an air-conditioning and Convers Manage 40 (1999) 873–884.
refrigeration system, ASHRAE Trans 1995; 383–391.

Вам также может понравиться