Applied Thermal Engineering 50 (2013) 318e326

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering

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

Simplified steady-state modeling for variable speed compressor

Wenhua Li*
Modeling, Analysis, Simulation and Computation (MASC), Carrier Corporation, PO Box 4808/TR4, 6304 Carrier Parkway, Syracuse, NY 13057, USA

h i g h l i g h t s

< Physical-based modeling of variable speed.

< Three types of compressors: reciprocating, scroll and piston rotary are investigated.
< Characteristics of volumetric and isentropic efficiency for variable speed compressor are explored.

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 detailed analysis of semi-empirical methods to calculate mass flow rate, shaft
Received 30 May 2012 power and discharge temperature for three types of variable speed compressors: reciprocating, scroll and
Accepted 25 August 2012 piston rotary. The proposed methods are an integration of physical-based models for constant speed
Available online 31 August 2012
compressor and the physical characteristics of volumetric efficiency and isentropic efficiency between
different speeds. The physical-based models were first validated with good agreement with experimental
Keywords:
data from publication for the three types of constant speed compressors. The comparison between
Refrigeration
modeling results and experimental data from publication for the three types of variable speed
Compressor
Model
compressors shows the RMS errors are less than 3%, 3% and 3
C for refrigerant mass flow rate,
Volumetric efficiency compressor power input and discharge temperature, respectively. The model of variable speed
Variable speed compressor will allow the reduction of the number of experimental data required to characterize vari-
able speed compressor behavior in the modeling of refrigeration systems because of its physical
mechanisms.
Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction
Variable speed compressor is of great interest in air conditioning
and refrigeration system because it offers better capacity regulation
than the conventional on/off control. The modeling of this kind of
compressor plays a very important role in the simulation of air
conditioning and refrigeration system [1e3]. Whether in the
refrigeration system simulation of steady state or unsteady state, an
accurate compressor model which can provide reasonable and
relatively accurate extrapolation outside the compressor envelope
is also essential for the system modeling.
A lot of studies have been conducted on the semi-empirical
modeling of the constant speed compressor through its efficiency
[4e7]. In the current researches on the variable speed compressor,
constant compressor efficiency [8] or polynomial functions of
condensation temperature and evaporation temperature [9] or
neural networks [10] are used in their compressor models. Among
* Tel.: þ1 785 317 2624.
E-mail addresses: wenhuali73@gmail.com, wenhua.li@carrier.utc.com.
them, Yang et al. [10] proposed a model to calculate the volumetric
and isentropic efficiencies for variable speed compressor using
neural network. The kind of model will need a lot of data points to
train the neural network. Shao et al. [9] presented a model of
variable speed compressor based on the performance data
provided by compressor manufacturers for each speed. The model
is built at the basis frequency. Mass flow rate and power at the basic
frequency are second-order polynomial functions of condensation
temperature and evaporation temperature. These modeling
methods of variable speed compressor have no any physical
meaning, and therefore cannot represent the physical mechanisms
appearing in variable speed compressor behavior.
The Characterization of a variable speed compressor itself can be
analyzed through its volumetric efficiency and isentropic efficiency.
Koury et al. [8] considered the efficiencies irrelevant to the opera-
tion frequency in their numerical simulation of a variable speed
refrigeration system. Cuevas and Lebrun [11] presented that the
compressor efficiencies are not enormously influenced by the
compressor supply frequency between 35 and 75 Hz. However, as
Shao et al. [9] pointed out, when the inverter compressor operates
at low frequency, the lubrication worsens and friction loss increase,

1359-4311/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.applthermaleng.2012.08.041
 W. Li / Applied Thermal Engineering 50 (2013) 318e326
Nomenclature
a1, a2, a3 fitting parameters for shaft power model
b1, b2 fitting parameters for mass flow rate model
c1, c2, c3 fitting parameters for discharge temperature model
d1, d2, d3 fitting parameters for volumetric efficiency model
e1, e2, e3 fitting parameters for isentropic efficiency model
C clearance fraction
dp suction pressure drop [kPa]
h enthalpy [kJ kg1]
k refrigerant isentropic coefficient
m mass flow rate [kg s1]
n polytropic index
N compressor rotation speed [rpm]
P pressure [kPa]
Pr pressure ratio of discharge pressure with respect to
suction pressure
T temperature [
C]
y specific volume [m3 kg1]
V

compressor displacement [m3]
V volume flow rate [m3 s1]
resulting in worse operation performance of inverter-driven
compressor. Shao et al. [9] also mentioned that the volumetric
efficiency will increase when the inverter-driven compressor
operation at high frequency due to the influence of leakage and
heating on suction gas. Tassou and Qureshi [12] investigated the
performance of compressors including open-type reciprocating,
a semi-hermetic reciprocating and an open-type rotary vane
through experiment and found out that the volumetric efficiency of
all three compressors increases with the increasing of speed and
reach maximum at the design speed and all three compressors
exhibit a rising isentropic efficiency with a reduction in speed.
Therefore, the study of compressor efficiency is still an efficient
way to characterize the performance of variable speed compressor.
In this work, the author tries to present a semi-empirical method to
calculate mass flow rate, power input and discharge temperature
for three types of variable speed compressors: reciprocating, scroll
and piston rotary. The method requires the integration of physical-
based models for constant speed compressor and the physical
characteristics of compressor efficiency between different speeds.
2. Analysis of variable speed compressor
Characterization of a compressor itself can be analyzed through
its volumetric efficiency and isentropic efficiency, which are
computed by the following equations:
V_ suc
hv ¼
Ncomp Vdisp
V_ suc ðhis hsuc Þ
his ¼
ysuc W
where, Ncomp is compressor speed, Vdisp is compressor displace-
ment, Vsuc is volumetric flow rate, W is actual compressor work, his
is the function of discharge pressure and suction entropy, hsuc is
suction enthalpy and ysuc is suction specific volume.
Volumetric efficiency and isentropic efficiency of a variable speed
compressor vary with the pressure ratio of discharge pressure to
suction pressure, as well with the compressor speed. One may
expect that the volumetric efficiency and isentropic efficacy have
a dependence on the compressor speed due to fluid acceleration
effects and cross correlations to system pressure ratio and valve
319
W compressor power [W]
Wloss constant part of power loss [W]
Wt polytropic compression power
UA overall heat transfer coefficient [J K1]
a ratio of electromechanical losses to Wt
h efficiency
Subscripts
amb ambient
cal calculation
comp compressor
cond condensing
dis discharge
disp displacement
evap evaporation
exp experiment
is isentropic
ref reference
shell compressor shell
suc suction
v volumetric
pressure drops. In general, a variable speed compressor is designed
with maximum volumetric efficiency and isentropic efficiency at the
design speed. The two efficiencies will decrease when actual oper-
ational speed deviates from the design speed. These are observed in
the experiment of variable speed compressor [12].
Fig. 1 is the illustrations of the two efficiencies for a rolling piston
type rotary inverter compressor. The compressor [9] has piston
displacement of 20.7 cm3. The experimental data from the literature
is given at four different speeds of 30, 60, 90 and 120 Hz with the
evaporating temperature from 10 to 15
C and condensing
temperature of 40, 50 and 60
C. In Fig. 1, it can be seen that the
compressor has maximum volumetric efficiency and isentropic effi-
ciency at the speed of 90 Hz. Further observance in Fig. 1 shows that
the variation tendency of the two efficiencies with the pressure ratio
is similar at different compressor speed, indicating that the relations
between different speeds for the two efficiencies may be indepen-
dent of pressure ratio. By taking 60 Hz as reference speed, volumetric
efficiency and isentropic efficiency are normalized with respect to the
data at 60 Hz and are plotted as a function of normalized speed in
Fig. 2. Evidently, the normalized volumetric efficiency and isentropic
efficiency are determined only by the normalized speed and is
independent of the pressure ratio. Consequently, the two efficiencies
as function of normalized speed can be represented by a second-
order polynomial function, as shown in Fig. 2.
As a result from above analysis, one can see that an accurate
physical model of compressor at the reference speed is a necessary
(1)
step for predicting the performance of a variable speed compressor. In
reality, providing plenty of experimental data for all speeds is time-
consuming work, specifically for the development of a new
(2)
compressor. However, building a physical-based model is able to
efficiently reduce the number of experimental data and can provide
accurate predictions cover the extensive range of pressure ratio at
different speed. Therefore, the following section will explore the semi-
physical modeling of constant speed compressor so as to provide
strong basis for further modeling of a variable speed compressor.
3. Simplified modeling of constant speed compressor and
validation
Some literature [4,7,13] developed the simplified model of
a scroll compressor based on a lot of detailed information of
320 W. Li / Applied Thermal Engineering 50 (2013) 318e326

Fig. 2. (a) Normalized volumetric efficiency as a function of normalized speed; (b)

normalized isentropic efficiency as a function of normalized speed.

Fig. 1. (a) Volumetric efficiency as a function of Pr at different speed; (b) isentropic Threlkeld [14] viewed the compression and expansion processes
efficiency as a function of Pr at different speed.
in a reciprocating compressor as a polytropic process instead of
isentropic process and proposed the following to calculate the
actual volumetric efficiency,
experimental data such as machine suction temperature, scroll

suction temperature, scroll discharge temperature, machine hv ¼ 1  C Pr1=n  1 (3)
discharge temperature, scroll suction pressure drop, scroll
discharge pressure difference, suction pressure, discharge pressure, where n is polytropic component.
ambient temperature and shaft power. Here we’d like to develop
the semi-empirical models only based on experimental data as
shown in Fig. 3. Such models can be used to predict mass flow rate,
electric power and discharge temperature when inputs of suction Tamb
temperature, suction pressure and discharge pressure are given.

3.1. Model of mass flow rate Tsuc, Psuc Tdis, Pdis

For a compressor model, the volumetric efficiency is used to
Compressor
calculate the suction volume flow rate. The actual volumetric effi- m
ciency is defined as the ratio of actual volume of refrigerant
entering compressor to the geometric displacement of the
compressor, as described in eq. (1).
Physically, since the suction valve throttling, suction gas heating, Shaft power, W
the clearance volume, and internal leakage, the actual volumetric
efficiency is less than 1. Fig. 3. Conceptual schema of the compressor model.
 W. Li / Applied Thermal Engineering 50 (2013) 318e326 321

Jähnig [5] and Cabello [15] introduced suction pressure drop for Here we have four parameters a1, a2, a3 and Wloss to be calcu-
compressor mass flow model because of its significant effect on lated by fitting experimental data.
mass flow rate although small. Therefore, the actual volume effi-
ciency in the paper of Jähnig [5] has the formulation as 3.3. Model of discharge temperature

 1=n 
Pdis Generally, the energy balance for the entire compressor is
hv ¼ 1  C 1 (4) depicted by
Psuc ð1  dpÞ

_ dis  hsuc Þ þ Q_ amb

W ¼ mðh (9)
Jähnig [5] thought of the relation of dependence between C
and n, so he took n as isentropic coefficient k to look for best C
where, Q_ amb is the heat transfer rate to the ambient, which can be
when experimental data is used for fitting. With the suction
calculated as:
pressure drop dp and the internal leakage, we reformulated
eq. (4) as Q_ amb ¼ UAshell ðTshell  Tamb Þ (10)

 1=k  The most difficult part is how to calculate Tshell, which may be
Pdis
hv ¼ b1 þ b2 (5) related to a lot of factors such as pressure ratio, suction pressure,
Psuc ð1  dpÞ discharge pressure, or discharge temperature. Kim and Bullard [6]
suggested a liner relationship with the discharge temperature as
Tshell ¼ a þ bTdis. Duprez et al. [17] considered shell temperature is
where we have three unknown parameters b1, b2 and dp in eq. (5),
related to condensing saturation temperature and evaporating
which need experimental data for fitting.
saturation temperature and assumed Tshell ¼ a þ bTevap þ cTcond.
A good correlation should be a good compromise between
3.2. Model of electric power simplicity, reduced number of unknown parameters and results
accuracy, extrapolation capability. Our observance and analysis of
In ASHRAE Toolkit [16], the calculation of compressor shaft experimental data from publications which over a wide range of
power is proposed as follows: operating conditions suggests the calculation of shell temperature
as
W ¼ ð1 þ aÞWt þ Wloss (6) pdis
Tshell ¼ c1 þ c2 þ c3 =pdis (11)
psuc
where Wloss is the constant loss due to electro-mechanical loss, a is
Which tells the dependence of the shell temperature on the
the ratio of electromechanical losses to Wt. Wt is the polytropic
pressure ratio is not enough to capture the physical, an individual
compression power defined as
discharge pressure is also needed to do better prediction of shell
temperature.
 n1 
n Pdis n In short, although models (eqs. (5), (8) and (11)) proposed in this
Wt ¼ Psuc V_ suc 1 (7) article are derived from reciprocating compressor, the validation
n1 Psuc
and exploration for scroll and rotary compressor is also performed
It is interesting to notice Negrão et al. [4] and Jähnig [5] assume hereafter by aid of experimental data. The parameters are deter-
the polytropic component n as isentropic coefficient k in their mined by minimizing the following objectives function to
calculation of power consumption. However, by observing and obtain optimized agreement with the measured data at given data
investigating the experimental data of reciprocating, scroll and points:
rotary compressor, we believe that a simple isentropic coefficient k sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
cannot fully reflect the compression power since the polytropic Xmexp mcal 2 XWexp Wcal 2
dm ¼ min ; dW ¼ min ;
process is very complicate. Instead, a constant number needs to be mexp Wexp
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
XTexp Tcal 2
added with the isentropic coefficient k to consider the shifting of
actual polytropic process, compared to isentropic process. This dT ¼ min ; T is discharge temperature ð12Þ
leads to the following proposed calculation of compressor shaft Texp
power as
where Levenberg Marquardt method is used to solve the least
 a2 þk1  square problem and estimate the parameters for models of mass
Pdis k
W ¼ Psuc V_ suc a1 þa3 =Pdis þ Wloss (8) flow rate, electric power and discharge temperature.
Psuc

Table 1
List of sources of experimental data for single speed compressor.

Name Source of data Refrigerant Displacement[cm3] Type Psuc[kPa] Pdis[kPa]

Dataset 1 Chen [18] R410A 8.11 Scroll 570e1100 1650e2700
Dataset 2 Winandy et al. [7] R22 175 Scroll 240e730 1190e2530
Dataset 3 Cuevas and Lebrun [11] R134a 54.25 Scroll 510e2040 1000e3910
Dataset 4 Winandy and Lebrun [19] R22 100.6 Scroll 200e520 1680e2900
Dataset 5 Compressor X, Negrão et al. [4] R134a 11 Recip 66e164 1100e1700
Dataset 6 Compressor Y, Negrão et al. [4] R134a 13.1 Recip 66e164 1100e1700
Dataset 7 Dataset B9A, Jähnig et al. [5] R12 8.095 Recip 100e170 900e1200
Dataset 8 Dataset A4, Jähnig et al. [5] R134a 7 Recip 90e180 950e1500
Dataset 9 Jovane [20] R410A 9.68 Rotary 600e1000 1510e3360
322 W. Li / Applied Thermal Engineering 50 (2013) 318e326

3.4. Validation of the models for constant speed compressor

Experimental data which are used for model validation are lis-
A
ted in Table 1. They are all from public literature and from seven
different journal papers. These first hand test data were performed
in different conditions and covered four different refrigerant types
of R12, R134a, R22 and R410A. Dataset 1e4 were for scroll
compressor while dataset 5 to 8 were for reciprocating compressor
and dataset 9 was for piston rotary compressor.
The simulation results are depicted in Fig. 4. The relative error
on the mass flow rates, shaft power is within 5% and absolute error
on discharge temperature is within 4
C for the three types of
compressors: scroll, reciprocating and rotary. The small error
establishes the calculated mass flow rates, shaft power and
discharge temperature are in good agreement with measured
values for the three kinds of compressors. It also tells that the
models are able to be employed to simulate the other type of
compressor such as scroll and rotary although they are derived
originally from reciprocating compressor.
As further evidence, Table 2 is provided to check the simulation
performance of the relative error of mass flow rate, electric power
and absolute error on the discharge temperature. Clearly, from the
table, the relative error on the mass flow rate varies B
between 5.27% and 5.36%. The relative error on the shaft power
varies between 4.45% and 4.14%. The absolute error on the
discharge temperature varies between 2.28
C and þ3.78
C. Max.
RMS errors for mass flow rate, shaft power and discharge temper-
ature are 2.96%, 2.29% and 1.68
C, respectively.

4. Modeling analysis of variable speed compressor

As described in Section 2, the normalized volumetric efficiency,

isentropic efficiency can be represented using a simple second-
order polynomial function of normalized speed and can be
formulated by the following equations


2
hv =hv;ref ¼ d1 þ d2  N=Nref þ d3  N=Nref (13)


2
his;ref =his ¼ e1 þ e2  N=Nref þ e3  N=Nref (14)

Then, according to eqs. (1) and (13), the volumetric flow rate at
any speed is calculated as C
N h
Vsuc ¼ Vsuc;ref  v
Nref hv;ref
( !2 !3 )
N N N
¼ Vsuc;ref  d1 þ d2 þd3 (15)
Nref Nref Nref

Similarly, the compressor input power at any speed is derived

from eqs. (2) and (14) and is obtained as

Vsuc his;ref
W ¼ Wref 
Vsuc;ref his
Vsuc n

2 o
¼ Wref  e1 þ e2  N=Nref þ e3  N=Nref (16)
Vsuc;ref

In above eqs. (13)e(16), sub-index ref is referred to compressor

model at a single speed. The calculation of volumetric flow rate and
power input at the referred single speed can be completed by using
eqs. (5) and (8) in Section 3.
Fig. 4. Comparison of measured and computed data for single speed compressors.
Model of calculating discharge temperature for constant speed
compressor illustrated in Section 3.3 can be directly applied to
a variable speed compressor because the model is developed based
 W. Li / Applied Thermal Engineering 50 (2013) 318e326 323

Table 2
Fitting error for the models.

Mass flow rate Electric power Discharge temperature

Min rel. error % Max. rel. error % RMS error % Min rel. error % Max. rel. error % RMS error % Min abs. error [
C] Max. abs. error [
C] RMS error [
C]
Dataset 1 4.30 5.36 2.96 4.45 3.71 2.29 1.62 2.94 1.12
Dataset 2 3.26 4.38 1.65 3.36 2.29 1.23 1.87 2.52 1.15
Data set 3 1.80 2.66 1.13 4.05 4.14 2.26 0.42 2.58 0.74
Data set 4 2.48 3.24 1.87 0.80 0.49 0.41 1.40 1.50 0.86
Data set 5 4.37 2.61 2.07 1.72 1.80 1.02 1.80 2.82 1.59
Data set 6 5.27 3.39 2.73 0.84 1.09 0.56 2.27 3.78 1.68
Data set 7 1.87 1.69 1.18 0.50 0.67 0.38 2.28 2.92 1.59
Data set 8 2.81 2.37 1.28 1.51 1.01 0.77 N/A N/A N/A
Data set 9 1.14 0.98 0.62 1.36 2.41 1.0 0.33 0.48 0.22

on energy balance with the consideration of the heat transfer loss
to the ambient. Therefore, any modification of the model for vari-
able speed is not necessary.
It should be noted, compared to the models for constant speed
compressor, we introduce additionally six parameters d1, d2, d3 and
e1, e2, e3 to consider the effect due to variable speed. However,
actual calculation should follow the limits at N ¼ Nref:
results for each data set. The parameter estimates were obtained
using the complete datasets. Fig. 5 shows the relative error of mass
flow rate is within 2% and the relative error of compressor power is
within 3.5% for dataset A. Figs. 6 and 7 shows the relative error of
mass flow rate and that of compressor power for dataset B and
dataset C are all within 5%. The overall calculation results of the

d1 þ d2 þ d3 ¼ 1 (17)
a
e1 þ e2 þ e3 ¼ 1 (18)

which tell us that actually we only need four additional parameters

to compute mass flow rate and compressor power for a variable
speed compressor.
Compressor datasets used in the study for model validation are
listed in Table 3. Three different types of compressor are considered
and they are scroll, reciprocating and rotary compressor. Dataset A
is data from manufacturer. Dataset B and dataset C are first hand
experimental data from publication. Each dataset provides data
with at least four different speeds and much broad range of suction
pressure and discharge pressure. It should be emphasized here that
the model validation for the three types of compressors under
constant speed has been performed in Section 3. Since these
models presented in Section 3 are built on the basis of empirical
and simplified characterized relations from the fundamental
physical laws, their integration with eqs. (15) and (16) for variable
speed compressor will greatly improve the modeling accuracy and
modeling prediction.
The simulation results are depicted in Figs. 5e7 and Table 4 b
show the estimated parameters and RMS errors of simulation

Table 3
Datasets for modeling analysis of variable speed compressor.

Dataset A Dataset B Dataset C

Source of data Shao et al. [9] Cuevas and Darr and
Lebrun [11] Crawford [21]
Displacement [cm3] 20.7 54.25 170
Speed range 30, 60, 90, 35, 40, 50, 1000, 1500, 2000,
120 Hz 65, 75 Hz 2500, 3000, 3500,
4000 rpm
Refrigerant R22 R134a R134a
Compressor type Rolling piston Scroll Reciprocating
rotary
Psuc [kPa] 355e790 480e2040 167e372
Pdis [kPa] 1533e2427 1000e4030 1075e2718
Points of data 312 48 61
Reference speed 60 Hz 50 Hz 2000 rpm
Application Inverter air Inverter air Automotive air
conditioner and conditioning conditioning
heat pumps Fig. 5. Comparison of measured and computed data for variable speed compressor
(dataset A): (a) mass flow rate; (b) power input.
324 W. Li / Applied Thermal Engineering 50 (2013) 318e326

Fig. 7. Comparison of measured and computed data for variable speed compressor
(dataset C).
Fig. 6. Comparison of measured and computed data for variable speed compressor
(dataset B).
Table 4
Estimated parameters and RMS error for datasets of variable speed compressor.
three datasets are in good agreement with measured data within
Dataset A Dataset B Dataset C
a reasonable accuracy as shown in Table 4: the RMS errors for
b1 1.083 1.047 0.779
calculated mass flow rates and power inputs are within 2.54% and
b2 0.077 0.051 0.036
2.78%, and the difference between the calculated and measured dp 0.000 0.000 0.000
discharge temperatures is below 2.92
C. a1 0.527 0.333 1.140
Further analysis on the modeling of variable speed compressor a2 0.852 1.197 0.295
is performed on dataset C, which was recorded from an automotive a3 19.146 440.393 447.050
Wloss 115.000 214.000 350.000
compressor and spans seven different speeds of 1000, 1500, 2000, c1 N/A 344.934 293.487
2500, 3000, 3500 and 4000 rpm. Detailed experimental data and c2 N/A 1.488 34.940
simulation results are listed in Table 5. From the table, one can see c3 N/A 33,202.623 24,202.367
that there are only a few data points for each speed and the pres- UAshell N/A 9.993 2.786
d1 0.709 0.693 1.223
sure ratio increases from low speed to high speed. The pressure
d2 0.416 0.543 0.206
ratio ranges from 3.9 to 10.4. Modeling for one speed is not
 W. Li / Applied Thermal Engineering 50 (2013) 318e326 325

Table 4 (continued ) appropriate because there are only a few data points and the range
Dataset A Dataset B Dataset C of pressure ratio is very short for each speed.
d3 0.125 0.236 0.017
In the study of dataset C, we use the semi-physical models
e1 1.870 1.599 0.507 presented for constant speed compressor in Section 3, as well as
e2 1.418 1.060 0.606 the relations of volumetric efficiency and isentropic efficiency
e3 0.548 0.461 0.113 between different speeds shown in eqs. (15) and (16) to determine
RMS error in m [%] 1.0 1.72 2.54
the parameters by minimizing the objectives function so as to
RMS error in W [%] 1.47 2.78 2.32
RMS error in Tdis [
C] N/A 0.329 2.92 obtain optimized agreement with the entire 61 data points.
Simulation results in Table 5 shows the relative error of mass flow
rate and power are within 4% except one point which has the error

Table 5
Relative error for dataset C of variable speed compressor.

N Tamb Ts Ps Pd Pr mexp Rel. error Wexp Rel. error Tdis,exp Abs. error





RPM C C kPa kPa kg/h % W % C C
984 53.3 9.2 372.3 1837.5 4.9 132.9 2.6 1713 2.7 74.1 0
992.7 50.1 6.7 344.7 1698.9 4.9 124.6 2.7 1635 3.5 70.1 0.5
997.3 50.4 8.6 324.1 1379.0 4.3 119.57 3.7 1462 1.2 65.8 0.1
997.9 50.9 5.1 317.9 1374.8 4.3 118.12 3 1432 2.4 62.7 0
999.5 47.2 3.9 295.8 1245.9 4.2 103.33 3.5 1371 0.9 59.4 3.5
1000.1 40.9 0.3 272.3 1075.6 3.9 94.21 7 1274 2.1 50.7 7.1
1000.1 43.4 2.5 292.3 1320.4 4.5 101.11 4.1 1413 2.4 60.4 3.5
1000.5 72.5 0.2 282.0 1146.6 4.1 98.34 5.9 1268 5.5 58.4 1.2
1001.5 53.8 8.3 341.3 1503.1 4.4 125.96 3.1 1544 1 68.1 0.5
1002.2 73.7 0.8 286.8 1229.3 4.3 104.6 0.5 1315 3.5 60.5 0.5
1007.2 46 11.7 249.6 1095.6 4.4 83.28 2.5 1214 0.5 68.9 0.7
1009.9 41.7 2.4 273.7 1170.0 4.3 97.93 1.7 1336 1.3 57.6 3.5
1484.4 54.1 6.6 332.3 2187.7 6.6 157.12 1.2 2633 3.3 89 1.5
1492.8 51.1 3.4 306.8 2016.7 6.6 148.91 2.5 2479 3.4 83.3 0.9
1495.6 53.9 5.7 293.7 1721.6 5.9 146.37 4.1 2305 1.3 81.2 2.5
1499.3 51.7 5.9 277.2 1551.3 5.6 138.48 4 2191 1.3 79.1 2.1
1504.3 47.6 0.2 249.6 1396.9 5.6 121.65 0.6 2059 2.6 71.5 1.4
1513.1 49.4 0.7 266.8 1535.5 5.8 136.94 4.1 2155 0.7 74.2 2.1
1990.2 54.3 5.4 311.6 2457.3 7.9 171.19 2.1 3423 0.1 97.4 1
1999.9 53.8 4.2 268.9 1881.6 7.0 157.03 1.5 2976 0.4 91.4 2.9
2000.1 49.7 2.2 242.0 1656.1 6.8 147.01 2.5 2741 0.6 81.8 2
2002.4 51.8 1.3 285.4 2262.9 7.9 162.75 0.2 3217 0.5 90.5 1.8
2003.6 73.8 6.1 228.2 1544.4 6.8 140.02 1.5 2597 1.9 76.4 0.2
2004.9 47.7 2.5 226.2 1496.2 6.6 140.07 3.9 2661 2 80.2 1.5
2005.5 40.6 7.7 210.3 1313.5 6.2 125.78 3.1 2567 3.5 69.3 6.9
2007 52.2 4.1 252.4 1680.9 6.7 151.5 3.2 2813 0.7 88.6 3.4
2008.3 72.1 6.7 213.1 1365.2 6.4 126.51 2.9 2463 1.7 76.6 0.4
2013.5 46.1 8 187.5 1285.2 6.9 103.19 0.3 2260 2.3 92.3 3
2495.5 54.6 4.8 298.5 2717.9 9.1 182.71 2.5 4152 1.5 105.3 2.5
2497.4 52.3 3.1 237.9 1801.6 7.6 159.85 2.6 3379 1.4 92.9 0.1
2498.9 53.5 2.7 253.0 2019.5 8.0 166.83 1.4 3594 0.9 94.7 1.4
2502.1 51.8 0.6 277.2 2504.2 9.0 174.36 1.6 3912 1.8 98.8 2.7
2503.3 47.7 4.5 211.7 1595.5 7.5 147.74 3.2 3189 1.2 83.2 2.5
2507.4 49.4 3.8 228.2 1770.6 7.8 154.99 0.8 3289 2.4 86.6 0.7
2990.2 46.6 0.2 204.8 1927.1 9.4 137.67 0.9 3761 5 103.7 2.1
2997.3 53.3 1.6 244.8 2128.4 8.7 172.41 0.5 4145 1.5 103.9 0.5
2999.2 48.6 5.5 203.4 1652.0 8.1 152.45 1.5 3689 1.5 92.4 1
3000.7 54.8 1.9 264.8 2422.8 9.2 181.89 2 4436 2 108.4 0.7
3001.7 52.6 0.1 271.7 2671.7 9.8 181.57 3.7 4525 3 107.6 1.8
3002 47.3 4.6 195.8 1625.8 8.3 140.34 1.4 3617 3 95.3 2
3004.3 49.9 5.3 219.3 1856.1 8.5 161.39 0.2 3790 2.5 94.7 1.2
3004.6 70.6 9.7 195.1 1516.2 7.8 143.15 4.2 3471 2.3 87.8 2.2
3004.7 52.9 1.9 228.2 1899.5 8.3 166.47 2.3 3907 1.3 101.7 1.8
3006.1 40.4 11.1 191.0 1472.0 7.7 143.38 2.7 3638 3.8 81.3 9.4
3008.1 47.3 8.8 167.5 1410.0 8.4 108.95 2.7 3121 3.2 109.5 1
3014.3 72.3 8.8 208.9 1754.7 8.4 154.68 1.1 3659 2.9 90.3 0.9
3495.5 54.8 1.1 255.1 2558.6 10.0 187.38 0.8 4985 0.9 112.8 1.4
3502.7 52.8 0.9 221.3 1949.2 8.8 170.01 1.8 4281 2 107.1 2.5
3503.7 53.9 0.4 237.9 2213.2 9.3 177.67 0.1 4605 1.1 108.4 0.4
3513.1 49.8 5.9 215.8 1949.8 9.0 166.79 0.4 4226 2.9 98.9 0.6
3517 48.1 6.8 198.6 1712.0 8.6 156.94 0.5 4088 0.8 96.6 2
3538.4 52.6 3.7 237.9 2337.3 9.8 177.72 0.8 4625 2.4 105.3 1.4
3997.9 54.1 0.5 234.4 2276.7 9.7 181.3 0.2 5004 0.4 111.8 1.7
3998.1 52.7 4.3 234.4 2409.7 10.3 181.16 0.1 5001 0.5 107.8 3.5
4000.3 55.1 1.1 254.4 2643.5 10.4 191.51 0.1 5407 1.2 116.6 3
4002.5 40.8 12.6 183.4 1558.2 8.5 156.49 1 4422 6 89.2 10.8
(continued on next page)
326 W. Li / Applied Thermal Engineering 50 (2013) 318e326

Table 5 (continued )

N Tamb Ts Ps Pd Pr mexp Rel. error Wexp Rel. error Tdis,exp Abs. error





RPM C C kPa kPa kg/h % W % C C
4003.6 48.3 7.4 194.4 1754.7 9.0 159.48 1.4 4404 2.2 100.6 2.4
4004.1 71.8 10.1 202.0 1879.5 9.3 162.52 1.4 4380 2.3 97.3 3.9
4004.6 71.1 11.4 184.1 1607.2 8.7 151.41 1.3 4179 0.5 96.7 2.9
4010.6 53.2 0.3 217.2 2000.2 9.2 173.14 2.4 4652 0.2 111.4 1.8
4011.3 49.7 6.4 211.7 2034.0 9.6 170.1 1.3 4606 0.4 103.2 1.6

more than 5%. The absolute error of discharge temperature from
Table 5 is within 4
C except that there are four data point not
within the range. These large errors may be caused by data
measurement.
5. Conclusion
It is observed from performance data for variable speed
compressor that normalized volumetric efficiency and isentropic
efficiency can be represented by a second-order polynomial func-
tion of normalized speed. The physical-based models for constant
speed compressor are presented as reference to calculate the
volumetric efficiency and isentropic efficiency at any speed. The
simulation results for three types of variable speed compressors e
reciprocating, scroll and piston rotary show the RMS errors are less
than 3%, 3% and 3
C for refrigerant mass flow rate, compressor
power input and discharge temperature, respectively.
One of Experimental data for variable speed compressor is from
compressor manufacturers and others are from first hand lab data
in publications. Although some experimental data has a few data
points for each speed, the modeling results show good accuracy
due to the physical-based methods. This will allow the reduction of
experimental data required to characterize variable speed
compressor behavior in the modeling of refrigeration systems
because of its physical mechanisms.
References
[1] W. Chen, X. Zhou, S. Deng, Development of control method and dynamic
model for multi-evaporator air conditioners (MEAC), Energy Conversion and
Management 46 (2005) 451e465.
[2] Y.C. Park, Y.C. Kim, M.-K. Min, Performance analysis on a multi-type inverter
air conditioner, Energy Conversion and Management 42 (2001) 1607e1621.
[3] S. Shao, H. Xu, C. Tian, Dynamic simulation of multi-unit air conditioners
based on two-phase fluid network model, Applied Thermal Engineering 40
(2012) 378e388.
[4] C.O.R. Negrão, R.H. Erthal, D.E.V. Andrade, L.W. Silva, A semi-empirical model
for the unsteady-state simulation of reciprocating compressors for household
refrigeration applications, Applied Thermal Engineering 31 (6e7) (2011)
1114e1124.
[5] D.I. Jähnig, D.T. Reindl, S.A. Klein, A semi-empirical method for representing
domestic refrigerator/freezer compressor calorimeter test data, ASHRAE
Transaction 106 (2000).
[6] M.-H. Kim, C.W. Bullard, Thermal performance analysis of small hermetic
refrigeration and air-conditioning compressors, JSME International Journal
Series B 45 (4) (2002).
[7] E. Winandy, O.C.S. Saavedra, J. Lebrun, Experimental analysis and simplified
modeling of a hermetic scroll refrigeration compressor, Applied Thermal
Engineering 22 (2) (2002) 107e120.
[8] R.N.N. Koury, L. Machado, K.A.R. Ismail, Numerical simulation of a variable
speed refrigeration system, International Journal of Refrigeration 24 (2001)
192e200.
[9] S. Shao, W. Shi, X. Li, H. Chen, Performance representation of variable-speed
compressor for inverter air conditioner based on experimental data, Inter-
national Journal of Refrigeration 27 (2004) 805e815.
[10] L. Yang, L. Zhao, C. Zhang, B. Gu, Loss-efficiency model of single and variable-
speed compressors using neural networks, International Journal of Refriger-
ation 32 (2009) 1423e1432.
[11] C. Cuevas, J. Lebrun, Test and modeling of a variable speed scroll compressor,
Applied Thermal Engineering 29 (2e3) (2009) 469e478.
[12] S.A. Tassou, T.Q. Qureshi, Comparative performance evaluation of positive
displacement compressors in variable-speed refrigeration applications,
International Journal of Refrigeration 21 (1998) 29e41.
[13] C. Cuevas, J. Lebrun, V. Lemort, E. Winandy, Characterization of a scroll
compressor under extended operating conditions, Applied Thermal Engi-
neering 30 (2010) 605e615.
[14] J.L. Threlkeld, Thermal Environmental Engineering, Prentice-Hall, Englewood
Cliffs, NJ, 1962.
[15] R. Cabello, Navarro, E. Torrella, Simplified steady-state modeling of a single
stage vapor compression plant: model development and validation, Applied
Thermal Engineering 25 (2005) 1740e1752.
[16] J.P. Bourdhouxhe, M. Grodent, J. Lebrun, HVACTOOLKIT: A Toolkit for Primary
HVAC System Energy Calculation, ASHRAE, 1999, ISBN 1-883413-73-7.
[17] M.E. Duprez, E. Dumont, M. Fre`re, Modeling of scroll compressors e
improvements, International Journal of Refrigeration 33 (2010) 721e728.
[18] Y. Chen, Mathematical modeling of scroll compressors, Ph.D. thesis, 2000, Ray
W. Herrick Laboratories, Purdue University.
[19] E.L. Winandy, J. Lebrun, Scroll compressors using gas and liquid injection:
experimental analysis and modeling, International Journal of Refrigeration 25
(8) (2002) 1143e1156.
[20] M. Jovane, Modeling and Analysis for a Novel Rotary Compressor, Ph.D.
dissertation, 2007, Purdue University.
[21] J.H. Darr, R.R. Crawford, Modeling of an Automotive Air Conditioning
Compressor Based on Experimental Data. ACRC TR-14, Air Conditioning and
Refrigeration Center, University of Illinois, 1992.