Departamento de Ingeniera Mecnica, Facultad de Ingeniera, Universidad de Concepcin, Casilla 160-C, Concepcin, Chile
Thermodynamics Laboratory, University of Lige, Campus du Sart Tilman Btiment B49, Parking P33, B-4000 Lige, Belgium
a r t i c l e
i n f o
Article history:
Received 26 November 2008
Accepted 9 November 2009
Available online 13 November 2009
Keywords:
Scroll compressor
Testing
Modelling
High pressure
a b s t r a c t
Refrigeration and air-conditioning compressors are designed to work under well-dened conditions. In
some applications it is interesting to observe their performances beyond these conditions, for example
in the case of a high temperature two-stage heat pump or of a cooling system working at high temperature.
In this study a compressor is characterized experimentally with refrigerant R134a and through 118
tests at condensing pressures varying from 8.6 up to 40.4 bar (tsat = 33.9 C to tsat = 100.8 C) and evaporating pressures varying from 1.6 up to 17.8 bar (tsat = 15.6 C to tsat = 62.4 C). Under these conditions
the compressor motor was pushed at its maximal current in several tests.
This compressors performance is mainly characterized by its isentropic and volumetric efciencies. It
presents a maximal isentropic efciency of 72%, corresponding to a pressure ratio of around 2.52.6. The
volumetric efciency decreases linearly from almost 1.0 (for a pressure ratio of 1.3) to 0.83 (for a pressure
ratio of 9.7). A slight degradation of the isentropic and volumetric efciencies is observed when the compressor supply and exhaust pressures are increased for a given pressure ratio; this could be due to an
internal leakage.
The compressor tests are used to identify the six parameters of a semi-empirical simulation model.
After parameter identication, experimental and simulated results are in very good agreement, except
for some points at high compressor power where the compressor is pushed at its maximal current.
2009 Elsevier Ltd. All rights reserved.
1. Introduction
Characterization of thermal machines is not only restricted to
the experimental analysis, modelling is also very important to
study the behaviour of overall systems during the design or the
simulation step. Both experimental and modelling aspects are
applied here for a refrigeration scroll compressor.
With respect to the compressor modelling, the distinction can
be made between empirical, semi-empirical and deterministic simulation models.
Empirical models are based on polynomial regressions and are
commonly used by manufacturers to present the performance of
the compressor.
Deterministic models rely on a comprehensive description of
the compressor based on control volumes analysis (involving heat
and mass transfers). Generally, these models associate a geometrical description of the machine and a thermodynamic description of
the compression process and require the exact characteristics of
606
Nomenclature
AU
C
H_
h
_
M
N
n
P
Q_
Rp
rv,in
T
t
V
v
w
W
_
W
Z
Subscripts
amb
ambient
cal
calorimeter
cor
Coriolis
cp
compressor
el
electrical
eV
evaporator
ex
exhaust
they are easier to integrate into the modelling of the whole refrigeration system. Moreover, unlike empirical models, semi-empirical
models allow the extrapolation of the components performance
beyond the range of data used to identify their parameters.
Hiller and Glicksman [7] seem to be the rst ones to propose a
semi-empirical model of a compressor, in the special case of a
reciprocating machine. Sauls [13] proposed a simple description
of the under-and over-compression processes characteristics of
scroll machines. Haberschill et al. [5] developed a model of a hermetic scroll compressor that accounts for the under- and overcompression processes, the internal leakage and the heat transfer
to the ambient. This model was improved by Winandy et al. [18],
who introduced the description of the suction and exhaust heat
transfers and of the electro-mechanical losses.
In this study, the semi-empirical model proposed by Winandy
et al. [18] is considered to characterize the scroll compressor as
continuation of the validation and as verication of the hypothesis
of this model under extended conditions.
Regarding the experimental analysis of thermal machines, it is
currently restricted to one nominal point or two to three dimensional testing conditions. The system or its components are rarely
run at the limits of their operating range. Performance of the system or component under such extreme conditions can be estimated by means of simulation, but without any certainty that
the system would work reliably. This is the reason why, when a
test bench has been set up, it is convenient to exploit it at most
to address these questions at least once. For example, very wide
testing conditions, corresponding to different applications, could
be dened for one component.
It is especially the case with the results presented in this study,
where a very large experimental program was carried out on a
xed speed hermetic scroll compressor. For instance, severe operating conditions, in terms of pressure and pressure ratio, were
dened.
f
i
in
loss
m
meas
mix
nom
oh
r
ref
s
sat
sd
sim
su
v
w
nal
initial
internal
loss
motor
measured
mixture
nominal
overheating
refrigerant
reference
isentropic or swept
saturation
slip
simulated
supply
volumetric or isovolumetric
water
Greek symbols
D
differential
a
power loss coefcient
e
efciency
e
error
c
isentropic coefcient
h
error function
r
standard deviation
1
3000 rpm
19:2 A
607
608
Table 1
Measurements ranges and uncertainties.
supply temperature :
1825 C
exhaust temperature :
70120 C
supply pressure :
37 bar
exhaust pressure :
1225 bar
refrigerant mass flow rate : 0:10:25 kg s1
electrical power :
611 kW
Variable
Measurement range
Uncertainty
Pressures
Pr,su,cp
Pr,ex,cp
1.5517.78 bar
8.6040.44 bar
Flows
_ mix;cor
M
58.2398.7 g s1
Electrical power
_ cp
W
412216,808 W
Table 2
Test conditions.
Supply pressure
Exhaust pressure
Supply temperature
Exhaust temperature
Refrigerant mass ow rate
Compressor electric power
1.617.8 bar
8.640.4 bar
10.5 to 70.6 C
53.5136 C
0.0550.637 kg s1
4.1216.81 kW
_ lamps W
_ fan H_ w;cal Q_ cal;amb Q_ cp;amb 0
W
609
Wlamps
Heating lamps
Calorimeter
.
Qcp ,amb
H w ,cal
Cooling coil
Compressor
Qcal ,amb
Load cell
Fan
Reservoir
W fan
Fig. 4. Compressor calorimeter thermal balance.
P V
Pr;su V s c1
r;ex
s
rv ;in c
c1
rv ;in
rv,in is the built-in volume ratio, the ratio between the suction and
discharge volume, which is a geometrical characteristic of the
compressor.
610
Fig. 10. Isentropic and volumetric efciencies for pressure ratios between 2.0 and
3.0.
According to this result, the supply pressure effect can be cancelled for a compressor volume ratio given by:
1
r v ;in cc1
In that case the compression power depends linearly on the exhaust pressure.
According to test conditions carried out in this study, c for
R134a varies between 0.86 and 1.06 (according to Eq. (4)). By considering an isentropic compression and the R134a as an ideal gas,
the corresponding internal pressure ratio is of the order of 2.5
3.5. As the maximum of the isentropic efciency is reached when
the internal pressure ratio is equal to the external one, these results can be compared with the results presented in Fig. 9, where
the maximal isentropic efciency is reached for an external pressure ratio of the order of 2.5, which explains the surprising result
mentioned previously and shown in Fig. 8.
Fig. 11. Isentropic and volumetric efciencies for pressure ratios between 2.0 and
3.0.
isentropic efciency decreases also, but the effect of the unadapted built-in volume ratio losses is lower.
The volumetric efciency decreases linearly when the pressure
ratio increases. At low pressure ratios, it approaches unity. In fact,
the volumetric efciency is calculated here by using the manufacturers swept volume. Perhaps it is slightly underestimated, which
can explain the values higher than unity. This could be also explained by the passive supercharging effect described by Nieter
[11].
In Fig. 9, a large scattering in the compressor efciency is observed for compressor ratios between 2.0 and 3.0. To explain this
scattering, Figs. 10 and 11 are added. They show the compressor
efciencies against the compressor exhaust pressure and compressor power respectively, for pressure ratios between 2.0 and 3.0.
According to Fig. 10, both isentropic and volumetric efciencies
decrease when the compressor exhaust pressure (or supply pressure) increases. This could be attributed to the compressors internal leakages, which increase when the compressor works at high
pressures. In Fig. 11, an important degradation on the isentropic
efciency is observed. This can be attributed to some additional
electro-mechanical losses that could appear at this high power
and pressures: it may be due to a lack of lubrication or to some
deformations of the scroll than can increase frictions.
611
4. Compressor modelling
4.1. Compressor model
The compressor modelling is carried out as proposed by Winandy et al. [18], who decomposed the evolution of the refrigerant inside the compressor in several steps, as shown in Fig. 12: supply
heating-up, isentropic compression, adiabatic compression at constant volume and exhaust cooling-down. This model is able to predict the compressor exhaust temperature, the refrigerant ow rate,
the ambient loss and the compressor electrical power, by using the
condensing and evaporating pressures, supply temperature, surrounding temperature and compressor speed as inputs. It uses four
predened parameters and six identied.
The scroll compressor is characterized by a xed built-in volume ratio rv,in imposed by its geometry. For a given refrigerant
and operating conditions, there is an internal pressure ratio corresponding to this volume ratio. If the external pressure ratio is different to the internal pressure ratio, the compressor is not adapted
The compressor electrical (or shaft) power can be split into two
_ in and the compressor
terms: the internal compression power W
_
electromechanical losses W cp;loss , which in turn can be split into
_ loss;0
other two terms: constant electro-mechanical losses term W
and mechanical losses proportional to the internal compression
_ in .
power: a W
_ cp W
_ in W
_ cp;loss W
_ in W
_ loss;0 a W
_ in
W
2in
Pin
2in
Pex,1
Psu,1
Vf
Vi
Psu,1
1
Vf
(a)
Vi
(b)
P
Pex,1
Pin
Pex,1 = Pin
2
2in
Psu,1
1
Vf
Vi
(c)
Fig. 13. (a) Adapted Pex = Pin, (b) over-compression Pex < Pin and (c) under-compression Pex > Pin.
612
addition, this wall transfers heat with the refrigerant and the compressor surrounding.
The adapted conditions (called here in) is determined by the
isentropic process from su1 to in and by considering the following
constraint:
r v ;in
v r;su1
v r;in
_ r;cp Ncp V s
M
v r;su1
_ cp
Ncp Nnom;cp 1 C sd;m W
AU su;cp
_ r;cp
M
AU su;cp;nom
_ ref
M
!0:8
AU ex;cp
_ r;cp
M
AU ex;cp;nom
_ ref
M
!0:8
Nnom 50 Hz
_ ref 0:25 kg s1
M
Slip coefficient :
C sd;m 0:004
AU ex;cp 35
_ r;cp
M
_ ref
M
!0:8
W K1
10
where AUsu,cp,nom and AUex,cp,nom are parameters of the model.
4.2. Model validation
Table 2 gives an idea of the extreme values for the 118 tests carried out to identify the compressor parameters. It can be observed
that the compressor was tested on a wide range of power, pressures and temperatures. Undoubtedly, the compressor supply temperature and the electrical power are the most constraining
variables; because both are limited to ensure a safe operation of
the compressor electric motor. Here, these two variables were
pushed to their limits.
In the experimental analysis it was mentioned that some scattering in the compressor efciency could be explained by a compressor internal leakage. This effect was analyzed in the
modelling and no-remarkable inuence was obtained. In fact,
internal leakage modelling is complex because the equivalent leakage area is inuenced by the oil circulation, thermal dilatation,
internal pressures, compliance on the orbiting scroll. . . and here
it is considered as a constant equivalent area. Moreover, the actual
model without internal leakage gives reasonable agreement with
the experimental measurement. Thus, it will not be considered
for this compressor.
To identify the compressor parameters, a function h is dened.
It depends on the relative errors of the following variables: compressor electric power, refrigerant mass ow rate, compressor wall
and exhaust temperatures. This function is dened as follows:
v
u X
2
n
u1 m X
V j;i;sim V j;i;meas
ht
n j1 i1
V j;i;meas
11
613
e
n
n
1X
1X
V i;mes V i;sim
ei
n i1
n i1
"
n
1X
ei ei 2
n i1
#0:5
12
where Vi,mes is the measured variable and Vi,sim is the simulated one.
When the sample (or test number) is higher than 20, the condence
limits are dened by the following equation:
Zr
e p
n
13
Average error
Standard deviation
Minimal deviation
Maximal deviation
Condence limits
tr,ex,cp
0.0 K
1.4 K
4.0 K
+6.1 K
0.3 K
+0.2 K
0.001 kg s1
+0.002 kg s1
0.224 kW
0.034 kW
1
Mr,cp
0.001 kg s
0.008 kg s
Wcp
0.129 kW
0.517 kW
1
1
0.008 kg s
(7.2%)
2.459 kW (4.6%)
0.030 kg s
1
(+4.8%)
+0.375 kW (+15.3%)
614
Fig. 19. Effect of the compressor overheating on the isentropic and volumetric
efciencies.
inputs of the model: supply overheating, supply and exhaust pressure and pressure ratio. This simulation is carried out by considering the model ow diagram given in Fig. 18, where are indicated
the inputs and outputs of the model.
Fig. 19 shows the variation of the isentropic and volumetric efciencies against the compressor pressure ratio at two different
refrigerant overheating points. Only a slight degradation on these
efciencies is observed when the overheating decreases.
On the other hand, supply pressure seems to have a very strong
effect on the compressor efciencies, as shown in Fig. 20. This effect is stronger at low supply pressures (lower than 5 bar), where
the optimal external pressure ratio increases from around 2.0 to
3.0. In this zone, the isentropic efciency can vary between 10%
and 15% and volumetric efciency around 5%.
A last simulation is carried out to determine the effect of the exhaust pressure on the efciencies in Fig. 21. Here the model reproduces the same trend observed in the experimental results, but it is
not so pronounced at high compressor exhaust pressures. It conrms the hypothesis of the internal leakage, which is not considered here in the modelling, but which appears in the
experimental results and increases the electro-mechanical losses.
5. Conclusions
Fig. 20. Effect of the compressor supply pressure on the isentropic and volumetric
efciencies.
Fig. 21. Effect of the compressor exhaust pressure on the isentropic and volumetric
efciencies.
615