Applied Thermal Engineering 30 (2010) 792–799

Contents lists available at ScienceDirect

Applied Thermal Engineering

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

Dynamic model of a hermetic reciprocating compressor in on–off cycling

operation (Abbreviation: Compressor dynamic model)
Demba Ndiaye, Michel Bernier *
Département de Génie Mécanique, École Polytechnique de Montréal, C.P. 6079, Succursale Centre-Ville, Montréal, Québec, Canada H3C 3A7

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

Article history: This paper presents a dynamic model of a hermetic reciprocating compressor in on–off cycling operation.
Received 28 May 2009 The proposed model is detailed enough to account for the important phenomena influencing the suction
Accepted 12 December 2009 and discharge mass flow rates and the electrical power drawn by the compressor, but simple enough to
Available online 29 December 2009
be usable for different reciprocating compressor designs with readily available data from manufacturers.
The experimental validation of the model under steady-state and transient conditions in both heating
Keywords: and cooling modes confirms the good performance of the model.
Hermetic reciprocating compressor
Ó 2010 Elsevier Ltd. All rights reserved.
Dynamic model
On–off cycling operation
Heat pump
Refrigeration system

1. Introduction
The compressor is often the most complex component in refrig-
eration systems. The compressor models found in the literature
may be grouped into either steady-state or dynamic models, each
of various degrees of complexity. The models of Cavallini et al. [1]
and Navarro et al. [2] are two examples of a simplified steady-state
model. More complex steady-state models will require very spe-
cific proprietary data only available from manufacturers. Further-
more, due to the complex mass, pressure and thermal
fluctuations inside the compressor, models will be specific to a par-
ticular compressor design. Steady-state compressor models cannot
capture the dynamic behavior of refrigeration systems. For exam-
ple, the suction and discharge mass flow rates may be different
during transient operations. Equally important is to have a model
that can account for the off-operation of the compressor to fully
capture the on–off cycling of the refrigeration system. In this case,
the evolution of the refrigerant in the shell is important with re-
gards to the interaction of the compressor with the evaporator,
the evolution of temperatures in the compressor, and compressor
heat losses during shutdown.
In simplified dynamic modeling of compressors, the shell is
usually treated as a single lumped element and the discharge mass
flow rate is derived from correlations [3,4] or from a thermody-
namic analysis of the compression-expansion process [5–7]. None
of these models addresses the off-operation of the compressor.
* Corresponding author. Tel.: +1 514 340 4711x4381; fax: +1 514 340 5917.
E-mail address: michel.bernier@polymtl.ca (M. Bernier).
Further, the model of Rossi and Braun [4] assumes equal suction
and discharge mass flow rates at all times. This assumption is only
valid in steady-state operation.
This paper presents a dynamic model of a hermetic reciprocat-
ing compressor in on–off cycling operation. The proposed model is
detailed enough to account for the important phenomena influenc-
ing the suction and discharge mass flow rates and the electrical
power used by the compressor, but simple enough to be usable
for different reciprocating compressor designs with readily avail-
able manufacturer data.
In the subsequent sections of this paper, the behavior of the
refrigerant in the shell and the resulting temperature evolution
of the different components such as the motor are investigated.
The discharge flow rate and the work of compression are deter-
mined via a thermodynamic analysis of the compression and
expansion processes, and the electrical power used by the com-
pressor is evaluated on the basis of manufacturer’s performance
data. Finally, a discussion of the experimental validation of the
model in both steady-state and transient conditions follows the
presentation of the numerical solution technique.
2. Suction and discharge
Fig. 1 presents a schematic view of a hermetic reciprocating
compressor. The refrigerant enters into the compressor through
the suction tube. The refrigerant cools the compressor by
absorbing heat coming from the motor, the mechanical parts,
the oil, and other metallic parts including the discharge tubing
and the wall of the shell. A pressure drop occurs when the

1359-4311/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.applthermaleng.2009.12.007
 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799 793

Nomenclature

a regression coefficient a state of the refrigerant in the cylinder after suction and
A area (m2) before compression
C clearance ratio b state of the refrigerant in the cylinder after compression
C friction factor and before discharge
Cp specific heat (J kg1 K1) c compressor
Cst constant c refrigerant in the shell of the compressor
D diameter (m) c state of the refrigerant in the cylinder dead volume be-
h enthalpy (J kg1) fore expansion
h heat transfer coefficient (W m2 K1) C convective
L length (m) ch crankcase heater
m mass (kg) d state of the refrigerant in the cylinder dead volume after
m_ mass flow rate (kg s1) expansion
n polytropic exponent dis discharge
N frequency (Hz) dt discharge tubing
P pressure (Pa) dv discharge valve
Q_ heat transfer rate (W) env environment
T temperature (°C) f free volume in the shell
t time (s) f friction
V volume (m3) Hz hertz
w specific work (J kg1) m compressor motor
W _ power (W) o lubricating oil
W work (J) r refrigerant
rc refrigerant to metallic parts of the compressor
Greek letters R radiative
D differential suc suction
l viscosity (kg m1 s1) sv suction valve
g efficiency sw swept
q density (kg m3) th theoretical
v volumetric
Subscripts w wall
1 suction side of the compressor
2 upstream piston suction valve Superscript
3 downstream piston suction valve 0 initial
4 upstream piston discharge valve
5 downstream piston discharge valve
6 discharge side of the compressor

Tenv
The refrigerant in the discharge tubing will lose part of its heat
Tw Tw to the compressor.
Tc , hc , ρ c The lubricating oil can mix with the refrigerant and be en-
trained into the system. Given that there is no viable procedure
5 to model the oil-refrigerant interaction and considering that refrig-
4
Tw eration systems are conceived to minimize oil entrainment in the
Tenv 3 Tenv system, the effects of the oil will be neglected in the modeling of
2 the compressor.

2.1. Suction side

discharge
6 The analysis on the suction side is based on a few assumptions.
(dis) 1 suction
Tw First, it is assumed that any refrigerant entering into the compres-
(suc)
sor will first mix with the refrigerant already present in the shell
before entering in the cylinders. It is also assumed that this mix-
OIL
ture is thermally homogeneous. Finally, the effects of suction and
discharge mufflers on the heat exchange and pressure drop are
not accounted for.
Fig. 1. Schematic representation of a hermetic reciprocating compressor. With these assumptions, the following mass and energy conser-
vations equations apply for the refrigerant in the shell:

dqc
_ suc  m
Vf ¼ m _2 ð1Þ
refrigerant passes through the suction valve. A pressure drop dt
may also occur in the suction tubing. After compression in the
cylinders, the refrigerant is discharged through the discharge dðqc hc Þ dP
Vf ¼ m _ 2 h2 þ Q_ c þ Q_ ch þ suc V f
_ suc hsuc  m ð2Þ
valve and tubing, where it experiences other pressure drops. dt dt
794 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799
The last term of the right hand side of Eq. (2) is the energy asso-
ciated with pressure variation with time. The other terms are de-
scribed in the nomenclature. The heat Q_ c exchanged between the
refrigerant and the motor, the mechanical parts, the oil, and the
other metallic parts is extremely difficult to determine with preci-
sion due to the complexity of the heat exchanges inside the shell.
To better capture these heat exchanges, it would be necessary to
find the temperatures of each of the involved parts and to take into
account radiation (with the necessary view factors), convection
and conduction. However, since the metallic parts have a high
thermal conductivity and the oil has a tight contact with these
metallic parts, Q_ c may be more easily determined by assuming that
all parts in contact with the refrigerant are at the same tempera-
ture T w [3].
Consequently, the value of Q_ c can be determined with:
Q_ c ¼ hrc Ac ðT w  T c Þ
Meyer [8] evaluates hrc at about 50 W/m2 K.
An energy balance on the oil and the other metallic parts of the
compressor can be used to determine T w :
 X  dT
w
mo Cpo þ mw Cpw ¼ Q_ m þ Q_ dis  Q_ c  Q_ env
dt
The heat coming from the electromechanical losses of the motor
is the difference between the electric power used by the motor and
the theoretical work of compression:
Q_ m ¼ W
_ c m
_ 2 wth
The refrigerant in the discharge tubing is at a high temperature.
Before leaving the compressor, this superheated refrigerant loses
heat to the other parts of the compressor. This heat loss, Q_ dis , will
be evaluated in the next subsection.
The heat exchanged with the environment is:
Q_ env ¼ Aenv henv ðT w  T env Þ
The heat exchange coefficient henv comprises a convective part
henv C and a radiative part henv R :
henv ¼ henv C þ henv R
The coefficient henv C is evaluated by considering natural convec-
tion heat exchange between a vertical cylinder and the surround-
ing air (see [9]). The coefficient henv R is evaluated by assuming
that the surrounding surfaces are at the temperature T env and con-
sidering the radiative heat exchange as a one between a grey body
and the black enclosure encompassing it (see [10]).
2.2. Discharge side
The walls of the discharge tubing are assumed to be at the tem-
perature T w , and the refrigerant in the tubing loses heat (Q_ dis ) to
the other parts of the compressor. Assuming that the flow in the
discharge tubing reaches almost immediately the steady-state re-
gime, and that the specific heat and density of the refrigerant along
with the convection heat transfer coefficient (between the refriger-
ant and the walls of the tubing) are constant along the tubing, re-
sults in the following (see Fig. 1):
 
hrw Adt
T 6 ¼ T w þ ðT 5  T w Þ  exp 
_ 6 Cpdt
m
Q_ dis ¼ m
_ 6 Cpdt ðT 5  T 6 Þ
Q_ dis
h6 ¼ h5 
m_6
The pressure drop in the discharge tubing is estimated to be [9]:
_ 26
32C f Ldt m
DPdt ¼ ð11Þ
p2 qdt D5dt
The friction coefficient is given by the Blasius formula [11]:
 1=4
4m_6
C f ¼ 0:079 ð12Þ
pldt Ddt
The mean convection heat transfer coefficient hrw is given by the
correlation of Dittus and Boelter (see [9]).
The mean values (specific heat, density, viscosity, thermal con-
ductivity) of the refrigerant in the discharge tubing are evaluated
at the mean temperature ðT 5 þ T 6 Þ=2 and at the mean pressure
ðP 5 þ P 6 Þ=2.
3. Compression in the cylinder
ð3Þ
Fig. 2 gives a schematic representation of the compression–
expansion cycle in the cylinder. The refrigerant enters the inlet at
a pressure P2 assumed to be equal to the suction pressure P suc .
The refrigerant experiences a pressure drop DP sv when passing
ð4Þ through the suction valve, and its pressure becomes P 3 . At the exit
of the cylinder, the refrigerant will also experience a pressure drop
DPdv through the discharge valve. Its pressure then goes from P4 to
P5 .
Beginning the analysis of the compression–expansion cycle
with point ‘‘a” in Fig. 2, where the piston is at the bottom dead cen-
ð5Þ ter: the cylinder is full of refrigerant at pressure P3 , and as the pis-
ton heads to the top dead center, compression, assumed polytropic
PV n ¼ Cst, takes place up to point ‘‘b”. At this point, the pressure
(P4 ) is slightly higher than the pressure on the other side of the dis-
charge valve, and thus this valve opens and the refrigerant is dis-
charged. Heat transfer in the discharge process is neglected [12];
and then T b ¼ T c . The top dead center is reached in ‘‘c”. A defined
ð6Þ volume of refrigerant at pressure P 4 remains in the cylinder: it is
the clearance volume or dead volume. Clearance volume is inevita-
ble and is due to mechanical constraints. The piston then begins a
return travel to the bottom. The refrigerant in the dead volume
then expands. This expansion is also assumed polytropic. At point
ð7Þ ‘‘d”, the pressure in the cylinder (P3 ) becomes slightly lower than
the pressure on the other side of the suction valve; the latter then
opens. Refrigerant enters in the cylinder and mixes with the refrig-
erant that occupied the dead volume. This mixing will induce some
refrigerant property changes. However, it is assumed that the tem-
perature of the refrigerant at the end of the suction process is the
same as the temperature of the sucked refrigerant: T d ¼ T a . In ‘‘a”,
the piston reaches the bottom dead center, and is ready to head
back to the top dead center. A new cycle is beginning.
ð8Þ
ð9Þ
ð10Þ
Fig. 2. Compression–expansion cycle in the cylinder.
 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799 795

The following additional hypotheses for the modeling of the 3.2. Specific theoretical work
compressor are considered:
The theoretical work of compression W th (J) is obtained by inte-
 the refrigerant behaves as an ideal gas grating the PV curve in Fig. 2 relatively to the volume:
 there is no fluctuation of the mass flow rate I
 the compression and expansion of the vapor in the cylinder are W th ¼ PðVÞdV ð18Þ
polytropic processes with the same polytropic exponent
 the pressure drops through the suction and discharge valves are It can be shown that:
isenthalpic processes " n1 #
 the oil in the compressor has negligible effects on the n P4 n
W th ¼ P3 ðV a  V d Þ 1 ð19Þ
refrigerant. n1 P3

3.1. Refrigerant flow rate The specific theoretical work (wth in J/kg) is the compression
work (W th ) required to compress a unit mass of refrigerant:
Due to the presence of the dead volume, the compressed refrig-
erant is not entirely discharged, and the state of the refrigerant at W th W th
wth ¼ ¼
the beginning of the compression is different than the state of the ðV a  V d Þqa ðV a  V d Þqsuc
incoming refrigerant. To account for this effect, the volumetric effi-
" n1 #
n P3 P4 n
ciency is defined. It is the ratio of the sucked refrigerant mass on ¼ 1 ð20Þ
n  1 qsuc P3
the mass that would have been sucked if there was no dead
volume:

ðV a  V d Þqa Va  Vd Va  Vd 3.3. Electrical power used

gV ¼ ¼ ¼ ð13Þ
ðV a  V c Þqsuc V a  V c V sw
The electrical power W _ c required by the compressor depends
In the same manner, the clearance ratio C is defined: mainly on the suction and discharge pressures. The power required
may be directly estimated using the performance curves of the
Vc
C¼ ð14Þ compressor manufacturer.
V sw
As the power required is not much sensitive to the degree of
Given that P3 V na ¼ P4 V nb and P 3 V nd ¼ P 4 V nc , the substitution of Eq. superheat of the refrigerant, the following equation inspired from
(14) into Eq. (13) leads to: the ANSI/ARI 540-1999 standard [15] may be used:
"  #
1=n _ c ¼ a0 þ a1 P1 þ a2 P6 þ a3 P2 þ a4 P1 P6 þ a5 P2 þ a6 P3
P4 W
gV ¼ 1  C 1 ð15Þ 1 6 1
P3
þ a7 P21 P6 þ a8 P1 P26 þ a9 P36 ð21Þ
Eq. (13) may also be rearranged as follows, giving:
where the ‘‘a” coefficients are determined by regression using the
ðV  V d Þqa NHz m_r m_r performance curves of the compressor.
gV ¼ a ¼ ¼ ð16Þ
ðV a  V c Þqsuc NHz ðV a  V c Þqsuc NHz V sw qsuc NHz
4. Numerical solution
where m_r ¼m_2¼m _5¼m_ 6 (kg/s) is the refrigerant mass flow rate
discharged by the compressor. It is then given by: 4.1. Compressor on
"  #!
1=n
P4
_ r ¼ gV qsuc V sw N Hz ¼
m 1C  1 qsuc V sw NHz ð17Þ In the numerical solution, the problem is separated into two
P3
parts: (1) the shell and its content and (2) the refrigerant flow rate.
The determination of the refrigerant flow rate requires the
knowledge of four values that are not usually readily available 4.1.1. Shell
The suction flow rate m _ suc is assumed known. The heat coming
from manufacturers: the clearance ratio, the pressure drop through
the suction valve, the pressure drop through the discharge valve, from the electromechanical losses of the motor Q_ m and the heat
and the polytropic exponent. The latter may be deemed close to coming from the refrigerant in the discharge tubing Q_ dis are known
the isentropic exponent [12]. The mean isentropic exponent for from the preceding iteration in the solution of the refrigerant flow
the compression and the expansion given by Bourdouxhe et al. rate problem.
[13] is used here. The value is n ¼ 1:114 for R-22 (it is an average The steps below are followed in each time step:
over the temperature range from 20 °C to +60 °C). From available
data in the open literature (see for instance [12]), it can be con- 1. calculate the flow rate at the suction of the cylinder by integrat-
cluded that, generally: ing Eq. (1):

_r¼
q0c  qc _ suc
 the clearance ratio varies from 3% to 7%: a good average value is m Vf þ m ð22Þ
5%
Dt
 the pressure drop through the suction valve is around 4% 2. calculate the enthalpy of the refrigerant in the shell by integrat-
 the pressure drop through the discharge valve is around 2%. ing Eq. (2):
0 0 0

The discharge pressure drop has a negligible effect on the mass

_ suc hsuc þ Q_ c þ PsucDP
m t
suc
V f þ qDc ht c V f
hc ¼ qc ð23Þ
flow rate [14] and may be set to 0. Ultimately, the suction pressure V þm
Dt f
_r
drop may also be set to 0. The clearance ratio may be derived from
the nominal performance data published by the manufacturer of 3. calculate qc and T c according to Psuc and hc by using the thermo-
the compressor. dynamic relations between the refrigerant properties
796 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799

4. calculate T w by combining Eqs. (4), (3) and (6): 5. Model validation and discussions

0 P The validation of the model of the compressor was part of a val-

Q_ m þ Q_ dis þ hrc Ac T c þ Aenv henv T env þ TDwt ðmo Cpo þ mw Cpw Þ idation effort of a detailed dynamic model of a commercial 3 tons
Tw ¼ 1
P
Dt
ðmo Cpo þ mw Cpw Þ þ hrc Ac þ Aenv henv (10.6 kW) water-to-air heat pump. The experiments were carried
ð24Þ out in an environmental chamber to control air conditions. The
heat pump was instrumented with: a thermocouple and a pressure
sensor at the refrigerant side entrance and exit of the four main
5. calculate Q_ c by using Eq. (3) components (the two heat exchangers, the compressor, and the
6. calculate Q_ env by using Eq. (6) expansion valve), a gas Coriolis flow meter at the discharge of
7. compare the actual values of m _ r and hc with their values at the the compressor, a liquid Coriolis flow meter at the entrance of
preceding iteration; if convergence is not obtained, go back to the expansion valve, two wattmeters to measure the electricity
step 1 for another iteration. consumed by the fan and the compressor. A data acquisition sys-
tem was coupled with the measuring devices. The absolute uncer-
4.1.2. Refrigerant flow rate tainties on the measured values were determined to be: ±5.2 kPa
The temperature T w of the parts in contact with the refrigerant on the pressure, ±0.35 °C on the temperature, ±7 W and ±13 W
is known from the numerical solution of the shell side. The follow- on the electric power used respectively by the fan and the com-
ing steps are used to obtain the refrigerant flow rate: pressor. The Coriolis flow meters have uncertainties of ±0.3% (of
reading) and ±1% (of reading), respectively for gas and liquid. The
1. calculate m_ r by using Eq. (17) refrigerant in the experiments is R-22. A complete description of
2. estimate W _ c from the compressor manufacturer data the experimental set-up is provided by Ndiaye [16].
3. calculate Q_ m by using Eq. (5)
4. estimate the pressure drop through the suction valve 5.1. Steady-state conditions
5. estimate the pressure drop through the discharge valve
6. calculate the enthalpy of the refrigerant at the discharge ori- The compressor model was submitted to an experimental vali-
fice as: dation in steady-state conditions for the heating and cooling
modes. The data used in the model are presented in Table 1; the
h5 ¼ hc þ wth ð25Þ
pressure drops through the suction and discharge valves are ne-
glected. The real value of the heat from the crankcase heater is un-
7. calculate T 5 according to P 5 and h5 by using the thermody- known; its value is set to the measured power when the
namic relations between the refrigerant properties compressor is off.
8. calculate T dis by using Eq. (8) Table 2 presents the results of the model validation. The mea-
9. calculate Q_ dis by using Eq. (9) sured refrigerant pressures at the suction and discharge sides,
10. calculate hdis by using Eq. (10) along with the measured refrigerant temperature at the suction
11. calculate DPdt by using Eq. (11) side and the temperature of the surrounding air were used to com-
12. compare the actual values of m _ r and hdis with their values at pute the exit conditions: refrigerant flow rate, refrigerant enthalpy,
the preceding iteration; if there is no convergence, go back and electrical power used. These computed values were then com-
to step 1 for another iteration. pared with the measured exit conditions. The ‘‘measured” exit en-
thalpy is obtained from the measure of the pressure and the
temperature of the refrigerant at the discharge of the compressor.
4.2. Compressor off
The last three rows of the table give the deviations between mea-
sured and computed values. The model under predicts the refriger-
In this case, N Hz ¼ 0 and m_ r ¼ 0. The electrical power used is the
_ c ¼ Q_ ch . There still ant flow rate by 7.4% and 8.5% for the heating and cooling modes,
power corresponding to the crankcase heater: W
respectively. The enthalpy difference is the difference between the
may be some flow at the suction side. We assume that
exit and entrance values. It is predicted within 1.7% in heating
T 5 ¼ T 6 ¼ T dis ¼ T w and thus Q_ dis ¼ 0 .
mode, and 12.4% in cooling mode. Finally, the electrical power used
The following procedure holds:
is under predicted by 2.2% in heating and over predicted by 3.6% in
cooling mode. These results show a reasonably good agreement be-
1. calculate the enthalpy of the refrigerant in the shell by integrat-
tween the model and the experimental data.
ing Eq. (2):

5.2. Transient conditions

0
q h 0 0
_ suc hsuc þ Q_ c þ Q_ ch þ PsucDP
m t
suc
V f þ Dc t c V f
hc ¼ qc ð26Þ Validation tests were also performed in transient conditions. In
V þm
Dt f
_r
each test, the heat pump was run for 10 min and then turned off
for 10 min, producing a 20 min cycle. In each mode of operation, a
2. calculate qc and T c according to Psuc and hc by using the thermo- series of five cycles was run. The representative cycle of each
dynamic relations between the refrigerant properties mode of operation (usually the fifth or the fourth) was visually
3. calculate T w by combining Eqs. (4), (3) and (6) (i.e. use Eq. (24)): determined. The results presented below detail the specific per-
4. calculate Q_ c by using Eq. (3) formance of the compressor transient model relatively to the
5. calculate hdis according to P 6 ¼ P dis and T w ¼ T dis by using the measured refrigerant discharge flow rate and power use. Two
thermodynamic relations between the refrigerant properties modes of operation of the heat pump are studied: heating and
6. calculate Q_ env by using Eq. (6) cooling with the fan remaining on at compressor shutdown. In
7. compare the actual value of hc with its value at the preceding both cases, the thermal expansion valve without bleed port is
iteration; if convergence is not obtained, go back to step 1 for used. For convenience, pressures in the evaporator and condenser
another iteration. are also given.
 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799 797

Table 1
Data used in the validation of the compressor model.

Unit Value
Refrigerant – R-22
Clearance ratio – 0.037
Total swept volume m3 38.1E6
Speed Hz 58.33
Mean compression–expansion isentropic exponent – 1.114
Free volume in the shell m3 6.07E3
Height of the compressor seen as a cylinder m 0.38
Diameter of the compressor seen as a cylinder m 0.21
Total area of the parts in contact with the refrigerant m2 0.01
Heat exchange coefficient refrigerant – parts in contact W/ 50
m2 K
Oil charge kg 0.98
Specific heat of the lubricating oil J/ 2000
kg K
Sum of the heat capacities of the metallic parts in contact J/K 5000
with the refrigerant
Exchange area compressor – environment m 2
0.32 Fig. 3. Comparison between measured compressor discharge flow rate and model
Temperature of the surrounding air in heating mode °C 20 prediction in the heating mode in transient conditions.
Temperature of the surrounding air in cooling mode °C 27
Thermal conductivity of the surrounding air W/ 0.026
mK
Thermal dilatation coefficient of the surrounding air K1 3.26E3
Thermal diffusivity of the surrounding air m2/s 0.21E4
Cinematic viscosity of the surrounding air m2/s 0.16E4
Emissivity of the exterior surface of the compressor – 0.25
Interior area of the discharge tubing m2 0.012
Length of the discharge tubing m 0.4
Mean diameter of the discharge tubing m 0.0095
Pressure drop through the suction valve % 0
Pressure drop through the discharge valve % 0
Heat from the crankcase heater W 6

Table 2
Results of the validation of the compressor model in the steady-state regime.

Heating Cooling
mode mode
Entrance Suction pressure (MPa) 0.427 0.488
Suction temperature (°C) 3.4 25.8 Fig. 4. Comparison between measured compressor power use and model prediction
Discharge pressure (MPa) 1.433 0.881 in the heating mode in transient conditions.
Surrounding air temperature (°C) 19 27
Measured exit Refrigerant flow rate (kg/s) 0.0364 0.0413
Refrigerant enthalpy (kJ/kg) 456.2 453.0
Electric power used (kW) 1.85 1.38
Computed exit Refrigerant flow rate (kg/s) 0.0337 0.0378
Refrigerant enthalpy (kJ/kg) 457.0 456.6
Electric power used (kW) 1.81 1.43
Deviations (%) On the flow rate 7.4 8.5
On the enthalpy difference 1.7 12.4
On the power used 2.2 3.6

5.2.1. Heating mode

At startup, pressures are equal over the entire system. Since the
refrigerant is less superheated than in the steady-state regime, the
compressor discharge flow rate is higher than at steady-state con-
ditions as shown in Fig. 3. It should be noted that the experimental
uncertainties are small (±0.3% reading) and are not shown in Fig. 3.
Very quickly, refrigerant builds up in the condenser to the det-
riment of the evaporator. The condensation pressure rises while
the evaporator pressure decreases (see Fig. 5). The compressor Fig. 5. Measured mean condensation and evaporation pressures in the heating
mode in transient conditions.
flow rate rapidly reaches a peak and then begins to decrease with
the rise of the condensation pressure and the beginning of
superheating in the evaporator. After about one minute and a half
(see Fig. 3), the whole system is tending towards equilibrium. The peak in the power use at startup (see Fig. 4) corresponds to
Meanwhile, the discharge flow rate has almost reached its steady- the peak in the condensation pressure seen in Fig. 5 (power use in-
state value. The model fully captures these tendencies of the creases with compression ratio). This first peak is well captured by
measured flow rate. the model. Another peak on the power use is noted at around
798 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799
Fig. 6. Comparison between measured compressor discharge flow rate and model
prediction in the cooling mode in transient conditions.
Fig. 7. Comparison between measured compressor power use and model prediction
in the cooling mode in transient conditions.
t ¼ 450 s. This second peak is abnormal and cannot be explained.
Except for this situation, although the simulated data are not en-
tirely within the limits defined by the experimental uncertainties,
the model reproduces relatively well the experimental data. The
evaluation of the power by regression of manufacturer data might
also be the cause of the observed discrepancy.
5.2.2. Cooling mode
In the present case, at compressor shutdown, what constitutes
the evaporator (the refrigerant-to-air heat exchanger) has the
highest pressure (in the experiments, the air temperature is higher
than the water temperature). Thus, in the shutoff phase, the refrig-
erant in the evaporator migrates to lower pressure portions of the
system. At the next startup, as seen in Fig. 8, the evaporator pres-
sure decreases well beyond its steady-state value, before going up.
This pronounced decrease in the evaporator pressure is caused by
the lack of refrigerant in the evaporator. It is seen in Fig. 6 that at
startup, the compressor discharge flow rate very quickly increases,
before decreasing and finally gradually going up again. The ob-
served decrease in the flow rate is due to the fact that, as there is
a lack of refrigerant in the evaporator, superheating of the refriger-
ant present in the evaporator is very fast (lower qsuc ), which re-
duces the flow rate (see Eq. (17)). The model captures this
phenomenon, and does a relatively good prediction of the experi-
mental flow rate. Fig. 7 also shows that the compressor reaches
its steady-state value very quickly and it remains relatively stable.
Again, the model prediction of the compressor power use is good.
Fig. 8. Measured mean condensation and evaporation pressures in the cooling
mode in transient conditions.
This validation exercise shows that the proposed model is rela-
tively accurate in predicting the steady-state and transient behav-
iors of hermetic reciprocating compressors. Possible causes of
discrepancies are as follows.
The inability to access some critical manufacturer data is a lim-
iting factor. For instance, in the present study, the clearance ratio
was unknown. This value could however be evaluated using, in
the model, the nominal performance data of the manufacturer of
the compressor.
In the model, the effects of the oil in the compressor and in the
refrigerant flow rate were neglected. More research is needed in
this area to find a reliable procedure for modeling the oil effects.
It was assumed that the refrigerant admitted into the shell
mixes entirely with the refrigerant already present in the shell.
However, Meyer [8] has shown that a variable proportion of the
refrigerant will be directly sucked into the cylinders. Provided that
the proportion of refrigerant that is directly sucked into the com-
pressor may be known with sufficient accuracy, the model might
benefit in incorporating that fact.
Finally, the degree of superheat of the refrigerant has some
influence on the electrical power used [17] and may be accounted
for in computing the electrical power for a better accuracy of the
model.
6. Conclusion
A dynamic model of a hermetic reciprocating compressor was
proposed. The model accounts for the off-operation of the com-
pressor. The model was validated as part of a validation effort of
a detailed dynamic model of a water-to-air heat pump. The valida-
tion was undertaken under steady-state and transient conditions
in both heating and cooling modes.
Under steady-state conditions, compressor discharge flow rate
is under predicted by 7.4% and 8.5% for the heating and cooling
modes, respectively. Enthalpy variation between the entrance
and the exit of the compressor is predicted within 1.7% in heating
mode, and 12.4% in cooling mode. Power use is under predicted by
2.2% in heating and over predicted by 3.6% in cooling mode. These
results show a reasonably good agreement between the model and
the experimental data.
Under transient conditions, measured evolutions of discharge
flow rate and power use were compared with model results using
10 min on–10 min off cycles. In both heating and cooling modes, as
shown in Figs. 3, 4, 6 and 7, the agreement between the model and
the experiments is good. For example, the model is able to capture
the peak in power usage at startup.
 D. Ndiaye, M. Bernier / Applied Thermal Engineering 30 (2010) 792–799
Acknowledgements
This research received financial support from the National Sci-
ences and Engineering Research Council of Canada and from the
CANMET Energy Technology Centre of Natural Resources Canada.
The Quebec-based heat pump manufacturer ThermoPlus provided
logistic support.
References
[1] A. Cavallini, L. Doretti, G.A. Longo, L. Rossetto, B. Bella, A. Zannerio, Thermal
analysis of a hermetic reciprocating compressor, in: International Compressor
Engineering Conference, Purdue University, IN USA, 1996.
[2] E. Navarro, E. Granryd, J.F. Urchueguia, J.M. Corberan, A phenomenological
model for analyzing reciprocating compressors, International Journal of
Refrigeration 30 (2007) 1254–1265.
[3] X. Xu, Modélisation dynamique d’un système frigorifique domestique à
compression de vapeur, Ph.D. Dissertation, École des mines de Paris, Paris,
France, 1996.
[4] T. Rossi, J.E. Braun, A real-time transient model for air conditioners, in: 20th
International Congress of Refrigeration, Sydney, Australia, 1999.
[5] J.W. MacArthur, Analytical representation of the transient energy interactions
in vapour compression heat pumps, ASHRAE Transactions 90 (1) (1984) 982–
996.
799
[6] Z. Chen, W. Lin, Dynamic simulation and optimal matching of a small-scale
refrigeration system, International Journal of Refrigeration 14 (1991) 329–335.
[7] X. Yuan, Y. Chen, D. Xu, Y. Gu, X. Liu, A computer simulation and experimental
investigation of the working process of a domestic refrigerator, in: 18th
International Congress of Refrigeration, Quebec, Canada, 1991.
[8] W.A. Meyer, An investigation into heat transfer process in a small hermetic
refrigeration compressor, Master’s thesis, Purdue University, IN, USA, 1987.
[9] A. Bejan, Convection Heat Transfer, Wiley, New York, 1995.
[10] F. Kreith, Principles of Heat Transfer, International Textbook Company, USA,
1966.
[11] B.R. Munson, D.F. Young, T.H. Okiishi, Fundamentals of Fluid Mechanics, Wiley,
New York, 2002.
[12] F.C. McQuiston, J.D. Parker, J.D. Spitler, Heating, Ventilating, and Air
Conditioning Analysis and Design, Wiley, Hoboken, NJ, 2005.
[13] J.-P.H. Bourdouxhe, C. Saavedra, M. Grodent, K.L. Silva, J.J. Lebrun, Toolkit for
primary HVAC system energy calculation – part 2: reciprocating chiller
models, ASHRAE Transactions 100 (1) (1994) 774–786.
[14] P. Popovic, H.N. Shapiro, A semi-empirical method for modeling a
reciprocating compressor in refrigeration systems, ASHRAE Transactions 101
(2) (1995) 367–382.
[15] ANSI/ARI, ANSI/ARI Standard 540-1999: Positive Displacement Refrigerant
Compressors and Compressor Units, Arlington, VA, USA, 1999.
[16] D. Ndiaye, Étude numérique et expérimentale de la performance en régime
transitoire de pompes à chaleur eau – air en cyclage, Ph.D. Dissertation, École
Polytechnique de Montréal, Montréal, Canada, 2007.
[17] A.E. Dabiri, C.K. Rice, A compressor simulation model with correction for the
level of suction gas superheat, ASHRAE Transactions 87 (2) (1981) 771–782.