International Journal of Refrigeration 30 (2007) 873e886

www.elsevier.com/locate/ijrefrig

Modelling of reciprocating and scroll compressors

Marie-Eve Duprez1, Eric Dumont1, Marc Frère*
Thermodynamics Department, Faculté Polytechnique de Mons, 31 bd Dolez, 7000 Mons, Belgium
Received 6 June 2006; received in revised form 14 November 2006; accepted 18 November 2006
Available online 22 December 2006

Abstract

This paper presents simple and thermodynamically realistic models of two types of compressors widely used in domestic
heat pumps (reciprocating and scroll compressors). These models calculate the mass flow rate of refrigerant and the power
consumption from the knowledge of operating conditions and parameters. Some of these parameters may be found in the tech-
nical datasheets of compressors whereas others are determined in such a way that the calculated mass flow rate and electrical
power match those given in these datasheets.
The two models have been tested on five reciprocating compressors and five scroll compressors. This study has been limited
to compressors with a maximum electrical power of 10 kW and for the following operating conditions: evaporating temperatures
ranging from 20 to 15  C and condensing temperatures ranging from 15 to 60  C.
The average discrepancies on mass flow rate and power for reciprocating compressors are 1.10 and 1.69% (for different
refrigerants: R134a, R404A, R22, R12 and R407C). For scroll compressors, the average discrepancies on mass flow rate and
power are 2.42 and 1.04% (for different refrigerants: R134a, R404A, R407C and R22).
Ó 2006 Elsevier Ltd and IIR. All rights reserved.

Keywords: Refrigeration; Air conditioning; Modelling; Performance; Reciprocating piston; Scroll compressor; R134a; R404A; R22; R407C

Modélisation des compresseurs à piston et à spirale

Mots clés : Réfrigération ; Conditionnement d’air ; Modélisation ; Performance ; Compresseur à piston ; Compresseur à spirale ; R134a ;
R404A ; R22 ; R407C

1. Introduction

Since the beginning of the 1990s, under the pressure of

different laws concerning the limitation of CO2 emissions,
the heat pumps market has known a new expansion in the
* Corresponding author. Tel.: þ32 65 37 42 06; fax: þ32 65 37
42 09.
domestic sector.
E-mail addresses: marie-eve.duprez@fpms.ac.be (M.-E. In the year 2000, there were 5750 heat pumps sold
Duprez), eric.dumont@fpms.ac.be (E. Dumont), marc.frere@ in Germany for house heating, 2750 in Finland, 3000 in
fpms.ac.be (M. Frère). Norway (for both water and space heating), 23,000 in
1
Tel.: þ32 65 37 42 08; fax: þ32 65 37 42 09. Sweden, 7200 in Switzerland and 7500 in France [1]. The
0140-7007/$35.00 Ó 2006 Elsevier Ltd and IIR. All rights reserved.
doi:10.1016/j.ijrefrig.2006.11.014
874 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

Nomenclature
d diameter (m) 3 ratio between the dead space and the swept
h specific enthalpy (J kg1) volume (e)
HP high pressure (Pa) h efficiency
IP intermediate pressure (Pa) r density (kg m3)
LP low pressure (Pa)
m mass (kg) Subscripts
N compressor rotation speed (t min1) c circulated
p pressure (Pa) calc calculated
P power (W) cond condensation
pvsat saturated vapour pressure (Pa) constr constructor data
qm mass flow rate (kg s1) d dead space
qv volume flow rate (m3 s1) el electrical
s specific entropy (J kg1 K1) evap evaporation
T temperature (K) ex exhaust
u specific internal energy (J kg1) i inlet of the compressor
UA global heat transfer coefficient (W K1) iso-s isentropic
v specific volume (m3 kg1) pseudo-iso-s
V volume (m3) pseudo-isentropic
W work (J) mecha mechanical
Dpsuc pressure drop in the suction valve (Pa) s swept
DTlog log-mean difference temperature (K) suc suction
DTsup superheating (K) w fictitious wall

average growth in terms of the number of installed heat On the one hand, there are models of reciprocating com-
pumps between 1997 and 2000 is about 15% a year. pressors in which the compressor is divided in several vol-
It is important to generalize the use of domestic heat umes (elements such as compression chamber, valves.).
pumps in order to decrease the primary energy consump- Those models require input data very difficult to obtain or
tion in the dwelling sector. For each heat pump project, known only by the constructor and non-available in the data-
the type of heat pump must be correctly chosen; the de- sheets. The volumes of the different elements and the effec-
sign and installation steps must be carried out carefully. tive area of valves are also required. The transient fluid
For optimization purpose, it would be interesting to dis- conservation equations (continuity, momentum and energy)
pose a calculation tool able to simulate the behaviour of are integrated in the whole compressor domain and the en-
the heat pump integrated to the residence so that the ergy balance for the refrigerant inside the cylinder is com-
energy consumption could be predicted and the design puted for each time step during the operating cycle [2e7].
of the heat pump (and/or of the house) could be adapted On the other hand, models in which thermodynamic as-
in such a way that the environmental impact is minimized. sumptions are made are also found. In those models, data
This simulation tool should remain as simple as possible are not very difficult to obtain: e.g. refrigerant inlet state,
so that its use could be widespread. outlet refrigerant pressure, clearance volume, motor speed.
In this aim of modelling such a complete system, it is im- In Ref. [8] eight input data are sufficient to determine
portant to have the simplest and the most accurate models of mass flow rate and required compressor power. In Refs.
its components. The compressor is one of the main part of [9e11] (model derived from Ref. [8]) the refrigerant mass
a heat pump as it sets its mass flow rate which governs the flow rate is affected by the clearance volume re-expansion,
heat flows. It is thus the first component of the heat pump by a pressure drop in the suction valve and by a heat transfer
to be modelled. from a fictitious isothermal wall. The friction power loss is
The types of compressors used in domestic heat pumps composed of a constant contribution and another one propor-
are usually the reciprocating and scroll ones. Reciprocating tional to the isentropic power. Ref. [12] presents a simple
compressors are mainly used as far as low thermal power is thermodynamic model for reciprocating compressors used
concerned (heating water) whereas scroll compressors are in domestic appliances. It required the knowledge of five pa-
widespread for space heating. rameters easy to determine. The main difference between
Many different models of those two types of compressors this model and the one presented in Ref. [9] is the fact that
with different degrees of complexity are found in the the heat transfer phenomena in the compressor are not con-
literature. sidered in Ref. [12].
 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 875

The same two categories may be defined for scroll log p

compressors.
Models [13e16] require the knowledge of pocket vol- HP = pex 3
umes and perimeters for every six degrees of rotation, the
height, thickness and pitch of the scrolls that are quite
Δpsuc
difficult to obtain. In those models, the whole compressor iso-s
is divided in several chambers and the compression i
LP 0
process is simulated for every gas pockets. It requires
the evaluation of areas, volumes, pressures, temperatures psuc 1 2
and specific volumes for every crank angle. Mass and h
energy conservation equations were developed for each
chamber. Fig. 1. Diagram (log p, h) of the compression.
Model [17] is exclusively thermodynamic and has the
same philosophy as [9e11] used for reciprocating compres-
sors. The refrigerant mass flow rate is affected by a suction 2.1.1. Prediction of the refrigerant mass flow rate
temperature increase and the compression process is consid- The data are:
ered isentropic to the ‘‘adapted’’ pressure and isochoric until
the discharge pressure.  Evaporation temperature Tevap.
The ARI Standard 540 [18] for positive displacement  Condensation temperature Tcond.
compressors recommends the use of third-degree-equations  Temperature at the compressor inlet (point i, Fig. 1)
of 10 coefficients for the calculation of power input, mass Ti or superheating DTsup.
flow rate of refrigerant, current or compressor efficiency.
Those coefficients have no physical meaning so that the The parameters of the model are:
extrapolation of the performances outside the operating
range used for the fitting leads to unrealistic performances  The suction line diameter dsuc.
values.  The heat transfer coefficient multiplied by the heat
The purpose of this study is not the development of transfer surface during the isobaric heating process
a complex model that can relate the performances of the in the suction line UAsuc.
compressor to its detailed geometry and that could be used  The compressor rotation speed N.
for technological developments. A simple and thermo-  The swept volume Vs.
dynamically realistic model is needed. It should be accu-  The ratio between the dead space Vd and swept
rate enough to give mass flow rates and power values volumes 3.
that can be used in a global model of a heat pump.  The temperature of the fictitious wall Tw.
Parameters appearing in such a model should be found
in the technical datasheets of the compressors or should Both high and low pressures (HP and LP) can be calcu-
be determined in such a way that the calculated mass lated from the phase change temperatures using Ref. [19]
flow rate and electrical power match those given in these (Refprop 7.0Ò).
datasheets.  
LP ¼ pvsat Tevap ð1Þ

2. Reciprocating compressors HP ¼ pvsat ðTcond Þ ð2Þ

in which pvsat(T) is the saturation vapour pressure law of the

2.1. Modelling of reciprocating compressors
refrigerant.
Point 0, corresponding to the saturated vapour at low
The model developed by Lebrun and coworkers [9e11] pressure (Fig. 1), is thus determined.
has been adapted given the particular requirements men- Temperature Ti at the inlet of the compressor (point i) is
tioned above. either directly given or calculated by Eq. (3).
The evolution of the thermodynamic state of the refriger-
ant through the compressor is presented in Fig. 1. Ti ¼ Tevap þ DTsup ð3Þ
The compression process is divided in three steps.
Ti and LP allow the calculation of hi and ri respectively, the
 Isenthalpic pressure drop in the suction valve (ie1). specific enthalpy and density of the refrigerant at the com-
 Isobaric heating up in the suction pipe due to a heat pressor inlet using Ref. [19].
transfer with a fictitious wall at temperature Tw The suction pressure psuc (point 1) is given by Eq. (4).
(1e2).
 Isentropic compression (2e3). psuc ¼ LP  Dpsuc ð4Þ
876 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

in which Dpsuc is the pressure drop in the suction valve; it p

can be deduced from Eq. (5).
pex 3' 3
pd 2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qm ¼ suc 2Dpsuc ri ð5Þ
4
in which:

 qm is the mass flow rate of refrigerant.

 dsuc is the valve coefficient considered as a pseudo
psuc 3" 2
dimension.
Vc
The thermodynamic state of the refrigerant at point 1 is
determined by Eq. (6) using Ref. [19]. Vd Vd + Vs V
Vs
hi ðTi ; LPÞ ¼ h1 ðT1 ; psuc Þ ð6Þ
Fig. 2. Crank diagram.
The temperature at the end of the isobaric heating pro-
cess (transformation 1e2) T2 may be deduced from Eq. (7).
s300 ¼ s30 ð14Þ
UAsuc DTlog suc ¼ qm ðh2  h1 Þ ð7Þ
s30 is calculated knowing the pressure p30 and the specific
with the log-mean difference temperature defined by Eq. (8). volume v30 at point 30 .
ðTw  T1 Þ  ðTw  T2 Þ p30 ¼ HP ð15Þ
DTlog suc ¼ ð8Þ
ln TTww T
T2
1
v30 ¼ v3 ð16Þ
In Eqs. (5) and (7), the refrigerant mass flow rate is In the same way, considering the compression step
calculated by Eq. (9). as isentropic, v3 may be calculated from pressure p3 and
1 specific entropy s3 by:
qm ¼ qvc ð9Þ
v2 p3 ¼ HP ð17Þ
in which: s3 ¼ s2 ð18Þ

 v2 is the specific volume of refrigerant at point 2. m300 in Eq. (12) is calculated by:
 qvc is the circulated volume flow rate. Vd
m300 ¼ ð19Þ
v30
qvc, is determined by Eq. (10).
Assuming that Tevap, Tcond and Ti or DTsup are given, Eqs.
N (1)e(19) lead to the determination of qm once the parame-
qvc ¼ Vc ð10Þ
60 ters of the model are available.
The circulated volume, Vc, is the difference between The parameters are:
volumes V2 and V300 defined in the Crank diagram (Fig. 2).
V2 is given by Eq. (11):  The temperature of the fictitious wall Tw.
 The heat transfer coefficient multiplied by the heat
V2 ¼ Vd þ Vs ¼ 3Vs þ Vs ð11Þ transfer surface during the isobaric heating process
in the suction line UAsuc.
The determination of V300 requires the mass of gas (m300 )  The ratio between dead space and swept volumes 3.
and the specific volume of refrigerant (v300 ) at point 300  The diameter of the suction pipe dsuc.
(Eq. (12)).  The rotation speed of the compressor N.
 The swept volume Vs.
V300 ¼ v300 m300 ð12Þ
The last two ones are given by the constructor.
Considering the expansion 30 300 as isentropic, the spe- The calculation procedure is given in Fig. 3.
cific volume v300 can be calculated knowing the pressure
p300 and the specific entropy s300 at point 300 .
2.1.2. Prediction of the electrical power
Once the mass flow rate is calculated, each point in Fig. 1
p300 ¼ psuc ð13Þ is known.
 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 877

Tevap

(1)

LP

Point 0 (saturated vapour at LP) Tsup ou Ti

(3)

psuc init Point i

(4)

T2 init psuc hi

(6)

Tcond Point 2 Point 1

(2)
HP s2

(17), (18)
Point 3

v3 Vs

(15), (16)

Point 3' Vd

(11)
s3'
m3' V2
(13), (14)
Point 3"

v3"

(12)
V3"

Vc N

(10)
qvc
(9)
qm

Fig. 3. Mass flow rate calculation.

In the previous section, compression (2e3) was supposed factor (helhiso-s), which is the product of the electrical and
to be isentropic. The mechanical power is then written isentropic efficiencies.
(Eq. (20)). This factor allows us to take into account both the
electrical losses and the losses due to the non-isentropic
Pmecha ¼ qm ðh3  h2 Þ ð20Þ compression.
The calculation of the actual electric power consumption This term was considered to be a polynomial function of
provided to the compressor requires the knowledge of the the compression ratio (HP/LP) (Eq. (21)).
878 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

 6  5  4  3 The evolution of the thermodynamic state of the refriger-

HP HP HP HP
hel hiso-s ¼ a þb þc þd ant through the compressor is presented in Fig. 4.
LP LP LP LP
 2   The compression process is divided in three steps.
HP HP
þe þf þg ð21Þ
LP LP  Isobaric heating up in the suction pipe due to a heat
transfer with a fictitious wall at temperature Tw (1e2).
Knowing the coefficients of the polynomial function it is  First part of the compression (assumed isentropic,
possible to calculate the electrical power (Eq. (22)). 2e3): the gas is compressed until the volume created
by the scrolls matches exhaust volume, Vex. The pres-
Pmecha
Pelcalc ¼ ð22Þ sure at point 3 is called ‘‘intermediate pressure, IP’’; it
hiso-s hel can be lower or greater than the high pressure.
The prediction of this power consumption requires new  End of the compression at Vex (constant volume) by
parameters (a, b, c, d, e, f, g) but no new operating data: mass accumulation in the exhaust chamber until the
HP and LP were needed for the mass flow rate calculation. pressure is equal to the exhaust one (3e4).

2.2. Determination of the parameters

3.1.1. Prediction of the refrigerant mass flow rate
2.2.1. Determination of the parameters: mass The data are:
flow rate calculation
The four unknown parameters required for the calcula-  Evaporation temperature Tevap.
tion of the mass flow rate are: Tw, UAsuc, 3 and dsuc. The  Condensation temperature Tcond.
assumption that they are constant for a given compressor  Temperature at the compressor inlet (point 1, Fig. 4)
and a given refrigerant is made. T1 or superheating DTsup.
Their values are found by minimizing the discrepancies
between the calculated mass flow rates and the ones pro- The parameters of the model are:
vided in the datasheets for given operating conditions (LP,
HP, Ti, DTsup).  The temperature of the fictitious wall Tw.
The fictitious wall temperature, Tw, has not a significant  The heat transfer coefficient multiplied by the heat
influence on the mass flow rate so its value was set to 50  C. transfer surface during the isobaric heating process
in the suction line UAsuc.
 The swept volume Vs.
2.2.2. Determination of parameters: electrical
 The compressor rotation speed N.
power calculation
The parameters needed for the determination of the elec-
Both high and low pressures (HP and LP) can be calcu-
trical power are the coefficients of the polynomial equation
lated from the phase change temperatures using Ref. [19].
linking the compression ratio HP/LP to the product of effi-
 
ciencies (helhiso-s). First the isentropic mechanical power LP ¼ pvsat Tevap ð24Þ
is calculated (Eq. (20)). The ratio Pmecha/Pelectrical is then
deduced from the electrical power provided by the manufac- HP ¼ pvsat ðTcond Þ ð25Þ
turer in the same operating conditions. in which pvsat(T) is the saturation vapour pressure law of the
This ratio is the product of the electrical and isentropic refrigerant.
efficiencies (helhiso-s) (Eq. (23)). Point 0, corresponding to the saturated vapour at low
Pmecha pressure (see Fig. 4), is thus determined.
¼ hiso-s hel ð23Þ
Pelectrical
A sixth degree polynomial law is used to correlate helhiso-s
log p
to the compression ratio. The values of the parameters are
determined by an optimization procedure. HP = pex 4

Iso-V
IP 3
3. Scroll compressors
LP = psuc Iso-s
3.1. Modelling of scroll compressors
0 12
h
The model is an adaptation of the work by Lebrun
et al. [17]. Fig. 4. Diagram (log p, h) of the thermodynamic process.
 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 879

Temperature T1 at the inlet of the compressor (point 1) is Mass flow rate model
either directly given or calculated by Eq. (26).
T1 ¼ Tevap þ DTsup ð26Þ Point 2 qm

T1 and LP allow the calculation of h1, specific enthalpy of

the refrigerant at the compressor inlet using Ref. [19]. h2 s2 v2 Vs Vs/Vex p2= LP
The temperature at the end of the isobaric heating pro-
cess (transformation 1e2) T2 may be deduced from Eq. (27). (41)
s3
msuc Vex
UAsuc DTlog suc ¼ qm ðh2  h1 Þ ð27Þ

in which qm is the mass flow rate of refrigerant. (40)

With the log-mean difference temperature defined by v3
Eq. (28).
ðTw  T1 Þ  ðTw  T2 Þ
DTlog suc ¼ ð28Þ Point 3 Tcond
ln TTww T
T2
1

(25)
In Eq. (27), the refrigerant mass flow rate is calculated by s4 h3 p3= IP HP = pvsat (Tcond)
Eq. (29).
1
qm ¼ Vs N ð29Þ
v2 IP/LP a b

in which v2 is the specific volume of refrigerant at point 2.

Assuming Tevap, Tcond and T1 or DTsup are given, Point 4
(44)
Eqs. (24)e(29) lead to the determination of qm once the pseudo-iso-s
parameters of the model are available.
The parameters are:
(43)
 The temperature of the fictitious wall Tw. Pelcalc
 The heat transfer coefficient multiplied by the heat
transfer surface during the isobaric heating process Fig. 6. Determination of power consumption, diagram.
in the suction line UAsuc.
 The swept volume Vs. pressure, IP, can be either higher or lower than the high
 The rotation speed of the compressor N. pressure.
The mechanical work W23 corresponding to this com-
The last two ones are given by the constructor. pression step is given by Eq. (30).
The calculation procedure is given in Fig. 5. W23 ¼ msuc ðu3  u2 Þ ð30Þ

3.1.2. Prediction of the electrical power in which:

The compression 2e3 is supposed to be isentropic. It oc-
curs between the low pressure and the intermediate pressure  u is the specific internal energy of the gas.
for which the exhaust volume is obtained. This intermediate  msuc is the suctioned mass.

Table 1
Tevap
Characteristics of the studied reciprocating compressors
(24)
T2 init Compressor Volumic Number Bore Stroke P (W)
LP
flow of (mm) (mm) (R134a,
rate cylinders Tcond ¼ 40  C,
Point 0 (saturated vapour at LP) Tsup or T1
Point 2 (m3 h1) Tevap ¼ 0  C)
(26) R1 39.36 4 60 40 5080
2 Vs N Point 1 R2 73.6 4 70 55 9460
R3 55.99 4 63.5 50.8 7600
(29)
R4 70.9 4 68.3 55.6 9150
qm R5 14.9 2 Not Not 2162
available available
Fig. 5. Mass flow rate calculation.
880 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

Table 2
Reciprocating compressors: results

Compressor Fluid UAsuc (W K1) dsuc (cm) 3 (%) Discrepancy Discrepancy

qm (%) P (%)
Compressor R1: Vs ¼ 452.414 cm3, R134a 43.53 2.918 2.50 1.04 3.45
N ¼ 1450 t min1 R404A 48.46 1.772 2.36 0.99 2.37
R22 48.91 2.275 4.24 1.02 2.71
R12 50 2.457 4.22 1.11 1.66
Compressor R2: Vs ¼ 845.977 cm3, R134a 15.66 2.781 3.41 0.53 1.30
N ¼ 1450 t min1 R404A 19.01 2.043 2.97 0.34 1.16
R22 19.77 2.405 4.28 0.50 2.30
R12 20.35 2.567 4.18 1.42 0.53
Compressor R3: Vs ¼ 643.563 cm3, R134a 67.93 2.610 4.73 0.80 1.98
N ¼ 1450 t min1 R404A 40.09 2.105 4.59 0.75 1.45
R22 5.35 1.962 5.91 0.37 2.24
Compressor R4: Vs ¼ 814.943 cm3, R134a 104.39 3.417 5.11 1.22 1.24
N ¼ 1450 t min1 R404A 95.44 2.678 4.79 1.09 0.86
R22 43.97 2.432 5.88 0.39 0.82
Compressor R5: Vs ¼ 85.64 cm3, R134a 13.77 1.221 8.25 1.86 2.79
N ¼ 2900 t min1 R404A 19.38 1.017 6.76 3.06 1.65
R407C 22.18 1.132 8.55 1.59 1.23

The mechanical work during suction, Wsuc, is given by Vsuc ¼ msuc v2 ð33Þ
Eq. (31).
Vex ¼ msuc v3 ð34Þ
Wsuc ¼ Vsuc LP ð31Þ
which leads to Eq. (35).
The work at the exhaust, Wex, is given by Eq. (32).
W ¼ msuc ðu3  u2 Þ  LPmsuc v2 þ HPmsuc v3 ð35Þ
Wex ¼ Vex HP ð32Þ
given
The total work between points 2 and 4 is obtained by the
u2 ¼ h2  LPv2 ð36Þ
summation of the three works (relations (30)e(32)).
Volumes Vsuc and Vex can also be written as Eqs. (33) u3 ¼ h3  IPv3 ð37Þ
and (34).

Fig. 7. Compressor R2, R134a  qm ¼ f(Tevap).

 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 881

Fig. 8. Compressor R2, R134a  qmcalc/qmconstr ¼ f(Tevap).

Eq. (35) becomes Eq. (38). in which:

W ¼ msuc ðh3  h2 Þ þ ðHP  IPÞmsuc v3 ð38Þ  The exhaust volume Vex is determined from the suc-
which leads to the expression of the power (Eq. (39)). tion volume (given by the constructor) and from the
ratio between this volume and the exhaust volume
P ¼ qm ðh3  h2 Þ þ ðHP  IPÞqm v3 ð39Þ (Vsuc/Vex) which is a parameter of the model.
 The suctioned mass msuc is calculated by:
In Eq. (39) qm is determined by the procedure described
in Section 3.1.1; h2 is calculated by Ref. [19] knowing the
temperature and pressure conditions at point 2; the high Vsuc
msuc ¼ ð41Þ
pressure HP is known via relation (24). v2
The specific volume v3 is calculated by:
with v2 calculated by Ref. [19].
In Eq. (39), point 3 (h3 and IP) is known because both the
Vex specific volume v3 and the specific entropy s3 are known
v3 ¼ ð40Þ
msuc (Eqs. (40) and (42)).

Fig. 9. Compressor R2, R134a  P ¼ f(Tevap).

882 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

s3 ¼ s2 ð42Þ Their values are found by minimizing the discrepancies

between the calculated mass flow rates and the ones pro-
Knowing each term in relation (39), it is possible to calcu- vided in the datasheets for given operating conditions.
late P which is the thermodynamic mechanical power (with The fictitious wall temperature, Tw, has no significant
the first step considered as isentropic). The calculation of influence on the mass flow rate so that its value was set to
the electrical power Pelcalc would require the introduction 50  C.
of an isentropic efficiency in the first term of the right-hand
member of Eq. (39) as well as an electrical efficiency applied 3.2.2. Determination of the parameters:
to each step of the compression. Due to the fact that these two electrical power calculation
efficiencies are not applied to the same terms it is not possible The influent parameters in the power prediction are
to gather them in one term (helhiso-s) as it was the case for re- the ratio between the suction and the exhaust volumes
ciprocating compressors. It appeared that the introduction of (Vsuc/Vex) and the coefficients of the linear law of the
two new parameters to be dealt with in the optimization pro- pseudo-isentropic efficiency (hpseudo-iso-s) versus the com-
cedure led to unrealistic parameter values and to too good pression ratio IP/LP.
performances compared to the experimental accuracy of The isentropic efficiency is expressed by Eq. (44).
the data. As a consequence, given the low influence of the
second term of the right-hand term of Eq. (39), we only intro- IP
hpseudo-iso-s ¼ a þb ð44Þ
duced one parameter (applied to the isentropic step) called LP
the pseudo-isentropic efficiency and which takes into
These three parameters are determined by minimizing
account both the irreversibilities and the electrical losses.
the discrepancies between the calculated electrical power
qm ðh3  h2 Þ and the ones provided in the datasheets for given operating
Pelcalc ¼ þ ðHP  IPÞqm v3real ð43Þ conditions.
hpseudo-iso-s

For a correct representation of the electrical power, it 4. Results

appeared that this pseudo-isentropic efficiency should be
considered as a linear function of the ratio IP/LP. v3real is In this section, simulation results of mass flow rate and
the specific volume at the ‘‘real’’ point 3 (non-isentropic). power for both types of compressors are presented.
The power calculation procedure is presented in Fig. 6.
4.1. Reciprocating compressors
3.2. Determination of the parameters
Five reciprocating compressors were studied. Some of
3.2.1. Determination of the parameters: mass their characteristics are presented in Table 1 [20e22].
flow rate calculation Table 2 presents values of the different parameters
The two unknown parameters in the mass flow rate pre- (UAsuc, dsuc and 3) and the average discrepancies, respec-
diction are Tw and UAsuc. The assumption that they are con- tively, on the mass flow rate and on the electrical power
stant for a given compressor and a given refrigerant is made. for each compressor and each studied fluid.

Fig. 10. Compressor R2, R134a  Pcalc/Pconstr ¼ f(Tevap).

 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 883

Table 3 For evaporation temperatures higher than 20  C the

Characteristics of the studied scroll compressors discrepancy on the mass flow rate is lower than 2.5%.
Compressor Volumic flow P (W) (R134a,
rate (m3 h1) Tcond ¼ 40  C, Tevap ¼ 0  C) 4.2. Scroll compressors
S1 25 3770
S2 36 5420 Five reciprocating compressors were studied. Some of
S3 8.6 1340 their characteristics are presented in Table 3 [20e22].
S4 35.56 5500 Table 4 presents the values of the different parameters
S5 43.5 6866 (UAsuc, Vsuc/Vex, a and b) and the average discrepancies,
respectively, on the mass flow rates and on the electrical
The model is very good at predicting mass flow rate and powers for each compressor and each fluid.
power consumption of reciprocating compressors: average The model is very good at predicting mass flow rate and
discrepancies are lower than 3%. In most cases, the points power of scroll compressors. The mean discrepancies are
for which the discrepancy is higher than 5% correspond to lower than 3.5%. In most cases, points for which discrep-
evaporation temperatures lower than 20  C and/or conden- ancy is higher than 5% correspond to evaporation tempera-
sation temperature higher than 60  C. In residential heat tures lower than 20  C and/or condensation temperature
pumps, those ranges of temperatures are rare. higher than 60  C. In the residential heat pumps, those
Figs. 7 and 8 present, respectively, the evolution of the ranges of temperatures are rare.
mass flow rate (calculated and given by the manufacturer) Figs. 11 and 12 present, respectively, the evolution of
and the related discrepancy as a function of the evaporation mass flow rate and the discrepancy on this mass flow rate
temperature for different condensation temperatures (30, 40 as a function of the evaporation temperature for different
and 50  C). condensation temperatures.
For evaporation temperatures higher than 20  C the The discrepancy on the mass flow rate is generally lower
discrepancy on the mass flow rate is lower than 1%. than 5%. It can reach nearly 10% for high condensation tem-
Figs. 9 and 10 present, respectively, the evolution of the peratures (55 and 60  C).
electrical power (calculated and given by the manufacturer) Figs. 13 and 14 present, respectively, the evolution of
and the related discrepancy as a function of the evaporation power and the discrepancy on this power as a function of
temperature for different condensation temperatures (30, 40 the evaporation temperature for different condensation
and 50  C). temperatures.

Table 4
Scroll compressors: results

Compressor Fluid UAsuc Vsuc/Vex a (e) b (e) Discrepancy Discrepancy

(W K1) (e) qm (%) P (%)
Compressor S1: Vs ¼ 143.678 cm3, R134a 19.01 2.379 0.777 2.585 1.70 1.12
N ¼ 2900 t min1 R404A 48.36 2.165 0.010 0.682 1.57 0.19
R407C 42.06 2.219 0.312 1.463 2.31 0.46
R22 60.33 2.187 0.671 2.329 2.21 0.59
Compressor S2: Vs ¼ 206.897 cm3, R134a 26.99 2.378 0.772 2.570 1.74 1.12
N ¼ 2900 t min1 R404A 69.59 2.166 0.003 0.698 1.59 0.16
R407C 59.49 2.220 0.316 1.473 2.31 0.46
R22 87.09 2.189 0.677 2.345 2.18 0.59
Compressor S3: Vs ¼ 47.778 cm3, R134a 5.39 2.302 0.473 1.766 1.47 0.85
N ¼ 3000 t min1 R404A 18.97 2.089 0.320 1.361 2.95 1.02
R22 22.67 2.128 0.642 2.142 3.15 2.26
R407C 17.44 2.156 1.152 3.302 1.97 1.21
Compressor S4: Vs ¼ 197.556 cm3, R134a 107.02 2.637 0.134 1.022 2.32 0.53
N ¼ 3000 t min1 R404A 44.93 2.667 0.113 0.954 2.95 1.49
R22 131.05 2.937 0.215 1.422 2.78 1.24
R407C 146.36 2.799 0.194 1.273 2.95 1.08
Compressor S5: Vs ¼ 24.99 cm3, R134a 34.64 2.466 0.826 2.752 2.02 0.66
N ¼ 2900 t min1 R22 62.43 2.293 0.579 2.148 1.84 0.58
R407C 99.14 2.239 0.465 1.787 2.23 0.55
884 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

Fig. 11. Compressor S3, R404A  qm ¼ f(Tevap).

The discrepancy on the power is also generally lower The reciprocating compressor model requires the knowl-
than 5%. It can be higher than 5% for low condensation tem- edge of six parameters. Two of them are given in the data-
peratures (10 and 15  C). sheets: the swept volume (Vs) and the compressor rotation
speed (N). The last four are: the temperature of the fictitious
5. Conclusion wall (Tw) which is actually set to a constant value (50  C),
the global heat transfer coefficient at suction (UAsuc), the
In this paper, simple and thermodynamically realistic diameter of the suction pipe (dsuc) and the ratio between
models of the two types of compressors found in the residen- the dead space and the swept volume (3). Those three param-
tial heat pumps (reciprocating and scroll compressors) have eters have to be fitted in order that the calculated mass flow
been presented. Those models calculate the mass flow rate of rates match the ones given in the datasheets. The determina-
refrigerant and the power consumption from the knowledge tion of the powers requires the knowledge of the seven coef-
of operating conditions (phase change temperatures and ficients of the polynomial law relating the product of the
refrigerant temperature at the inlet of the compressor or electrical and isentropic efficiencies to the compression
superheating) and different parameters. ratio.
Parameters appearing in the models are found in the tech- This model has been tested on five compressors. The
nical datasheets of the compressors or they are fitted in such mean discrepancies values on the determination of,
a way that the calculated mass flow rate and electrical power respectively, mass flow rate and power consumption are
match those given in these datasheets. 1.15 and 2.04% (R134a), 1.38 and 1.43% (R404A),

Fig. 12. Compressor S3, R404A  qmcalc/qmconstr ¼ f(Tevap).

 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886 885

Fig. 13. Compressor S3, R404A  P ¼ f(Tevap).

0.51 and 1.86% (R22), 1.27 and 1.09% (R12), 1.59 and This model has been tested on five compressors. The
1.23% (R407C). mean discrepancy values on mass flow rate and power are
The scroll compressor model requires the knowledge of 1.89 and 0.76% (R134a), 2.76 and 1.12% (R404A), 2.39
seven parameters. Two of them are given in the datasheets: and 0.91% (R407C), 2.59 and 1.31% (R22).
the swept volume (Vs) and the compressor rotation speed Although, it was not hinted in this paper those models fail
(N). The last five are: the temperature of the fictitious wall at predicting correct outlet temperatures (up to 50  C differ-
(Tw) which is actually set to a constant value (50  C), the ence depending on the compressor and fluid, usually the dif-
global heat transfer coefficient at suction (UAsuc), the ratio ference is around 20e25  C). So the same philosophy at the
between the swept volume and the exhaust volume (Vsuc/ outlet than at the inlet (taking into account a heat transfer
Vex) and the two parameters a and b of the linear law of with the fictitious wall) should be applied. The determination
the pseudo-isentropic efficiency. Those four parameters of the global heat transfer coefficient at the exhaust (UAex)
have to be fitted in order that the calculated mass flow rate requires values of exhaust temperature or values of heat
and power matches those given in the datasheets. UAsuc is flow rate at the condenser. As temperature Tw has a significant
fitted using the mass flow rate data while the three other influence on the outlet temperature of the compressor, it
parameters (Vsuc/Vex, a, and b) are fitted using power seems interesting to make it vary with the phase change
consumption data. temperatures and power consumption of the compressor.

Fig. 14. Compressor S3, R404A  Pcalc/Pconstr ¼ f(Tevap).

886 M.-E. Duprez et al. / International Journal of Refrigeration 30 (2007) 873e886

It is also possible to determine one set of parameters for [9] E. Winandy, O.C. Saavedra, J. Lebrun, Simplified modelling
all the refrigerants and for a given compressor. of a open-type reciprocating compressor, International Jour-
It must also be noticed that the isentropic and electrical nal of Thermal Science 41 (2002) 183e192.
efficiencies that are used for the determination of the power [10] M. Grodent, J. Lebrun, E. Winandy, Simplified modelling of
an open-type reciprocating compressor using refrigerants
have no real physical sense (the use of parameters with phys-
R22 et R410A. 2nd part: model, in: Proceedings of the 20th
ical sense leads to an over-parameterised model). This lack International Congress of Refrigeration, Sidney, vol. 3, 1999
of physical sense can explain the higher discrepancies at (paper 800).
high temperatures. [11] C. Hannay, J. Lebrun, D. Negoiu, E. Winandy, Simplified
modelling of an open-type reciprocating compressor using
refrigerants R22 et R410A. 1st part: experimental analysis,
References in: Proceedings of the 20th International Congress of Refrig-
eration, Sidney, vol. 3, 1999 (paper 657).
[1] H. Rivoalen, Heat Pump Market Overview in Europe, Hands- [12] D.I. Jähnig, D.T. Reindl, S.A. Klein, A semi-empirical
on Experience with Heat Pumps in Buildings, Workshop Re- method for representing domestic refrigerator/freezer com-
port HPC-WR-23, Arnhem, The Netherlands, October 2001. pressor calorimeter test data, ASHRAE Transactions 106
[2] C.D. Pérez-Segarra, J. Rigola, A. Oliva, Modelling and (2000) 122e130.
numerical simulation of the thermal and fluid dynamic behav- [13] J.-L. Caillat, S. Ni, M. Daniels, A computer model for scroll
iour of hermetic reciprocating compressors. Part 1: theoretical compressors, in: Proceedings of the International Compressor
basis, HVAC & R Research 9 (2) (2003) 215e235. Engineering Conference at Purdue University, USA, 1988,
[3] J. Rigola, C.D. Pérez-Segarra, A. Oliva, Modelling and pp. 47e55.
numerical simulation of the thermal and fluid dynamic behav- [14] Y. Chen, N.P. Halm, E.A. Groll, J.E. Braun, Mathematical
iour of hermetic reciprocating compressors. Part 2: experi- modelling of scroll compressors. Part I: compression process
mental investigation, HVAC & R Research 9 (2) (2003) modelling, International Journal of Refrigeration 25 (2002)
237e249. 731e750.
[4] J. Rigola, C.D. Pérez-Segarra, A. Oliva, Parametric studies on [15] Y. Chen, N.P. Halm, E.A. Groll, J.E. Braun, Mathematical
hermetic reciprocating compressors, International Journal of modelling of scroll compressors. Part II: overall scroll com-
Refrigeration 28 (2005) 253e266. pressor modelling, International Journal of Refrigeration 25
[5] J. Rigola, C.D. Pérez-Segarra, A. Oliva, J.M. Serra, M. (2002) 751e764.
Escribà, J. Pons, Advanced numerical simulation model of [16] C. Shein, R. Radermacher, Scroll compressor simulation
hermetic reciprocating compressors, Parametric study and model, Journal of Engineering for Gas Turbines and Power
detailed experimental validation, in: Proceedings of the 15th 123 (2001) 217e225.
International Compressor Engineering Conference at Purdue [17] E. Winandy, O.C. Saavedra, J. Lebrun, Experimental analysis
University, USA, 2000, pp. 23e30. and simplified modelling of a hermetic scroll refrigeration
[6] J.M. Corberan, J. Gonzalvez, J. Urchueguı́a, A. Calas, Model- compressor, Applied Thermal Engineering 22 (2002)
ling of refrigeration compressors, in: Proceedings of the 15th 107e120.
International Compressor Engineering Conference at Purdue [18] Performance Rating of Positive Displacement Refrigerant
University, USA, 2000, pp. 571e578. Compressors and Compressor Units, ARI Standard 504,
[7] M.L. Todescat, F. Fagotti, Thermal energy analysis in Air-Conditioning and Refrigeration Institute, Arlington,
reciprocating hermetic compressors, in: Proceedings of the 2004.
International Compressor Engineering Conference at Purdue [19] REFPROP, NIST Standard Reference Database 23, Version 7.0.
University, USA, 1992, pp. 1419e1428. [20] Bitzer e Software, <http://www.bitzer.de>.
[8] P. Popovic, H.N. Shapiro, A semi-empirical method for mod- [21] Copeland Selection Software Select, <http://www.ecopeland.
elling a reciprocating compressor in refrigeration systems, com>.
ASHRAE Transactions 101 (1995) 367e382. [22] Products datasheets, <http://danfoss.com>.