i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

Available online at www.sciencedirect.com

ScienceDirect
w w w . i i fi i r . o r g

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

Optimization of a vapor compression heat pump

for satellite cooling
Ian H. Bell*, Vincent Lemort
University of Li
ege, Energy Systems Research Unit, Li
ege, Belgium

article info

abstract

Article history:

In recent years the heat fluxes that must be removed from aerospace electrical systems

Received 2 April 2015

have been steadily increasing, motivating the use of vapor compression heat pumps to cool

Received in revised form

the electrical components.The heat pump system considered here is a conventional four-

3 June 2015

component heat pump that uses an oil-free scroll compressor in place of the oil-lubricated

Accepted 22 June 2015

compressor that is more often employed for terrestrial applications.The first part of this

Available online 30 June 2015

study considers the fluid selection, and refrigerant R152a is found to be a good choice. This
study then delves into the detailed performance analysis of the oil-free scroll compressors

Keywords:

that are envisaged to be used in this 12 kW system, with predicted COP over 4.0. Finally, the

Heat pump

entire operating envelope of the heat pump system is considered, including variations in

Scroll compressor

electrical load and seasonal variations in the radiative environmental temperature.The

Satellite cooling

source code for the analysis presented here and that of PDSim are provided as supplemental information.
2015 Elsevier Ltd and IIR. All rights reserved.

chaleur a

compression de vapeur
Optimisation d'une pompe a
pour le refroidissement de satellite

chaleur ; Compresseur a

spirale ; Refroidissement de satellite
Mots cles : Pompe a

1.

Introduction

Satellites are deployed into low earth orbit (LEO) or geosynchronous orbit (GEO) in order to provide a platform for a wide
range of missions, including telecommunications systems,
surveillance systems, and astronomical systems, amongst
others. For terrestrial electronics systems, the last few

decades have brought with them a large increase in electrical

component power density. Aerospace applications have seen
a similar increase in electrical power density of their electrical
payloads, which results in a commensurate need in cooling
for the electronic components.
For many past and present satellites, cooling of components was achieved through traditional heat pipes connecting
electrical components and large radiators. As electrical

* Corresponding author.
E-mail addresses: ian.h.bell@gmail.com (I.H. Bell), vincent.lemort@ulg.ac.be (V. Lemort).
http://dx.doi.org/10.1016/j.ijrefrig.2015.06.017
0140-7007/ 2015 Elsevier Ltd and IIR. All rights reserved.

70

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

Nomenclature
Arad
f
hi
hi
l
L
m_
N
p
Q_
0

q
00
q
R
T
DT
a
u
ur
h
COP
wi
w
_
W

Radiator area (m2)

Rotational frequency (Hz)
Inlet enthalpy (J kg1)
Internal height of channel (m)
Segment length
Length (m)
Mass flow rate (kg s1)
Number ()
Pressure (Pa)
Heat transfer rate (W)
Heat transfer per unit length (W m1)
Heat transfer per area (W m2)
Thermal resistance (m2 K W1)
Temperature (K)
Temperature difference (K)
Heat transfer coefficient (W m2 K1)
Rotational speed (rad s1)
Reduced rotational speed ()
Efficiency ()
Coeff. of Performance ()
Internal width of channel (m)
Width (m)
Power (W)
Radiator emissivity ()

Subscripts

Ambient
1,2,3,4 State point index
c
Radiator area
c
Condensing
cr
Reduced condensing
crit
Critical
circuit
Circuit
channel Channel
e
Evaporation
evap
Evaporator
e,i
Evaporator inlet
e,o
Evaporator outlet
er
Reduced evaporation
i
Inlet
flow
Flow
r
Refrigerant
rad
Radiator
pass
Pass
o
Outlet
oi
Overall isentropic
v
Volumetric
mc
Microchannel
pass
Radiator pass
s
Surface
sh
Superheat
sc,target Subcooling target
triple
Triple point
total
Total
TIM
Thermal Interface Material

cooling demands have increased, radiator surface area has

not increased at the same pace. Therefore, to reject more heat
for a given radiator area, the surface temperature of the
radiator must increase. Alternatively, the heat rejection rate
can be further increased through the use of deployable
radiators.
If the desired radiator surface temperature is above the
working temperature of the electrical components to be
cooled, it is necessary to employ active cooling through the
use of a heat pump to boost the temperature of a working fluid
above the desired radiator surface temperature. Through
condensation of the working fluid, the heat can be rejected
through the radiator.
The heat pump systems required for aerospace applications share many commonalities with those of terrestrial applications. In the simplest case, both terrestrial and aerospace
heat pump units are composed of the conventional fourcomponent system e compressor, condenser, expansion device and evaporator.
In some ways, design and optimization of heat pumps for
aerospace applications differs from that of terrestrial
applications. Some of the most important considerations are:

Satellites must be delivered to orbit by rocket engines,
which costs approximately USD $25,000/kg1 (EARSC,
2011).

The satellite will operate in a low-gravity environment,
making oil-refrigerant separation difficult.

There is no possibility of servicing the system during its
lifetime.

The maximum working temperatures of the electrical
components is specified
There is a relative paucity of published literature on the
design of heat pumps for aerospace and low-gravity applications. Woolley (1993) patented an oil separator that it is
claimed would be well-suited for application in zero-gravity
environments. St. Pierre (1987) patented a heat pump system that can operate in zero-gravity environments.
Messaros and Verstracte (1994) describe the development
process for a multi-stage piston compressor for a JouleeThomson cooler. This compressor uses a forced oil lubrication system, and has a specified failure rate of less than
1.42 failures per million hours. It employs the oil separator
described in the patent of Woolley. Nikanpour et al. (1997)
investigated a range of system designs for an aerospace
heat pump system with cooling capacity between 300 W and
2 kW and evaporation and condensation temperatures of
276 K and 343 K respectively. They find an optimal radiator
temperature that minimizes the system mass. Domitrovic
et al. (2003) carried out an experimental campaign to characterize the performance of a heat pump prototype at
terrestrial gravity. Cole et al. (2006) described preliminary
work on a development project for a 5 kWe15 kW cooling
capacity heat pump that would achieve a COP of 1.7 with
evaporation and condensation temperatures of 4.44  C and
60  C respectively.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

2.

System and component models

In this section the envisaged heat pump is described,

including models for the components that comprise the system as well as detailed description of the compressor, which
forms the heart of the system. This general description is not
dependent on the additional constraints applied to the
system.

2.1.

Fig. 2 e Block diagram for compressor model.

Overview

The vapor compression heat pump of a satellite cooling system is composed of four components as seen in Fig. 1. A
compressor compresses the superheated working fluid from
evaporation to condensation pressure (state points 1 to 2), the
working fluid is condensed in the radiatively-cooled
condenser (state points 2 to 3), the working fluid is expanded
in the expansion valve (state points 3 to 4), and then the cool
working fluid boils in the evaporator which cools the electrical
load (state points 4 to 1). One of the particularities of this
system is that the capacity is directly fixed by the electrical
load in the heaters, rather than being partially governed by the
source temperature, as in an air-source heat pump. Here the
temperature of the electrical elements mechanically coupled
with the evaporator are obtained as an output of the simulation. In this case, both the radiator and evaporator employ
microchannel heat exchangers for weight considerations. In
the description that follows, the components will be described
in the order that the fluid passes through them.
The model employed for the compressor can range from
either a very simple model (fixed volumetric and isentropic
efficiency) to a very detailed model (fully mechanistic,
comprehensive analysis). In this paper, the full gamut of
compressor model types are used, the simplified model is
used for fluid selection, and the detailed model for system
characterization once the working fluid has been selected.
Finally, a correlation is used to reduce the data from the
detailed compressor model. In any case, the performance of
_ and mass flow rate m)
_ can be
the compressor (power input W
characterized as a function of the boundary conditions of the
compressor, as seen in Fig. 2.

2.2.

71

Radiator

The radiator in this case is taken to be a square with length per

side of Hrad as seen in Fig. 3. There are then a given number of
passes Npass, which yields a width of radiator per pass of
wpass Hrad/Npass. The total length of the circuit is equal to

Fig. 3 e Overall schematic of radiator.

Lcircuit NpassHrad. While in principle it is possible to split the

circuit into multiple parallel paths (to increase mass flux and
heat transfer coefficient), that effect is not studied here.
The pass is formed of a microchannel condenser connected to a radiator panel through a thermal interface material (TIM) as seen in Fig. 4. The thermal interface material
causes a significant thermal gradient due to its relatively poor
thermal conductance.
The local surface temperature can be obtained from the
energy balance at the radiator surface which is at a temperature of Ts. Here we assume that the radiator surface has
infinitely high thermal conductance resulting in no spreading
losses. Thus the energy balance can be given by


Tr  Ts
wmc
q0 s T4s  T4 wpass
Rtotal
0

where q is the heat transfer per unit length in W m1, is the

emissivity (here taken as 0.8), s is the Stefan-Boltzmann
constant in W m2 K4, Ts is the surface temperature in K, T
is the ambient temperature in K, Tr is the working fluid temperature in K, and Rtotal is the combined thermal resistance of
the TIM and microchannel in m2 K W1. This equation must be
solved numerically for Ts. It is known that Ts is between T
and Tr, so a bounded solver like Brent's method can be used to
solve, guaranteeing a solution. The total thermal resistance
between the surface temperature and the refrigerant temperature is given by
Rtotal Rflow wmc RTIM :

Fig. 1 e System schematic.

(1)

(2)

The thermal interface resistance RTIM is given by the thermal interface material selected, which in this case is an
aerospace-grade TIM compatible with the vacuum of

72

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

Fig. 4 e Radiator pass cross-section.

interstellar space, with a thermal impedance of approximately 1.0 cm2 K W1. The flow thermal resistance is given by
Rflow h0 awmc 1

(3)

where a is the heat transfer coefficient in the channel in

W m2 K1, h0 is the overall surface effectiveness, and wmc is
the total width of the microchannel condenser in m. The
surface effectiveness is based on the heat transfer coefficient
in the channel and the microchannel condenser dimensions.
The circuit of the radiator is divided into a number of segments in the flow direction. The heat transfer correlation is
selected based on the local phase of the fluid, and is given in
Table 1.
ho hi 

q0 l
m_

&

po pi

dp
l
dz

(4)

where l is the length (in m) of an element of the circuit. The

total pressure drop is obtained as the summation of the
pressure drops from each element, and the heat transfer rate
is obtained as the summation of the heat transfer in each
element.

2.3.

Expansion valve

The expansion valve is taken to have idealized behavior; the

expansion valve is assumed to be always able to maintain the
desired superheat at the outlet of the evaporator for all operating conditions.

2.4.

Evaporator

The evaporator considered in this analysis is a best-case

evaporator with a uniform heat load along the entire length
of the evaporator, as shown in Fig. 5. In reality the load profile
in the evaporator is much more uneven, with a mixture of
different maximum unit temperatures and heat fluxes. The
surface temperature can be obtained directly from the heat
flux applied by the electrical units from
00

Ts q Rtotal Tr :

(5)

Table 1 e Correlations used in microchannel condenser.

Single-phase
Two-phase

Heat transfer

Pressure drop

Gnielinski (1975)
Shah (1979)

Churchill (1977)
Kim and Mudawar (2012)

Fig. 5 e Evaporator cross-section.

00

where Ts is the surface temperature of the unit in K, q is the

heat flux applied at the top of the microchannel in W m2,
Rtotal is the combined thermal resistance of the TIM and
microchannel in m2 K W1, and Tr is the refrigerant temperature in K. The pressure drop in the evaporator also tends to
decrease the refrigerant temperature along the length of the
evaporator in the flow direction.
As in the radiator, the evaporator is subdivided into a
number of elements in the flow direction, and in each
element, the inlet state of the element is used to determine
the heat transfer coefficient and pressure gradient for the
element. The correlations used are as given in Table 2.
In current high-power electrical components, heat fluxes
of 100 W cm2 are typical (Herwig, 2013); this flux is put
through a 20 thermal spreader to reduce the heat flux at the
microchannel evaporator to 5 W cm2. The same TIM material
used for the condenser is selected for the junction between
the microchannel and the electrical units. The thermal
impedance is approximately 1 cm2 K W1, which will result in
a temperature rise over the TIM of 5 K. The thermal resistance
from the flow boiling and surface efficiency in the evaporator
will contribute to an additional increase in the unit
temperature.

2.5.

Integration

Finally the entire system analysis can be carried out by solving

a 3 input/3 residual set of equations. The inputs are the inlet
pressure to the compressor, the outlet pressure of the
compressor and the inlet pressure to the evaporator, and the
residuals to be driven to zero by the nonlinear system solver
are.

Table 2 e Correlations used in microchannel evaporators.

Single-phase
Two-phase

Heat transfer

Pressure drop

Gnielinski (1975)
Bertsch et al. (2009)

Churchill (1977)
Kim and Mudawar (2013)

73

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

Capacity of the evaporator must be equal to electrical load

_ 1  h4  Q_ load
r0 mh

(6)

Actual subcooling1 must equal target subcooling

r1 DTsc  DTsc;target

(7)

Pressure drop of the evaporator must equal difference in

pressures over evaporator
r2 pe;i  pe;o  Dpe

(8)

Newton's method is used to solve the system of non-linear

equations. The additional constraints are that the subcooling
and the superheating are both 5 K.

2.6.

Model implementation

The overall model is written in C, and uses some numerical

routines developed in CoolProp for solving equations and
systems of equations. In addition, the correlations are
implemented in the ThermalCorr library. The source code
used to generate all the figures in this paper is attached as an
electronic appendix.

3.

Working fluid selection

So far, the analysis has been working fluid agnostic, but not all
working fluids are well suited to this application. As a first cut,
the fluid must have a triple point lower than the expected
evaporation temperature and a critical temperature greater
than the expected condensing temperature. Furthermore, the
fluid must have acceptable environmental characteristics
(toxicity and flammability) which will be described below.
Based on preliminary analysis and the constraints given
above, the evaporation and condensation saturated temperatures can be estimated. This analysis predicts that the
evaporation and condensation temperatures should be
approximately 40  C and 90  C respectively. In order to carry
out the first cut of fluid selection, the entire fluid library of
CoolProp 5.0.6 is employed, which includes 112 pure and
pseudo-pure fluids (Bell et al., 2014). The first thermodynamic
filtering is to keep all the fluids for which Ttriple < 40  C and
Tcrit > 90  C; when the first filtering is applied, 78 fluids
remain.
The next step is to consider flammability, toxicity and
environmental compatibility. For this application, very flammable and very toxic fluids are deemed not acceptable. Based
on the known flammability of certain classes of fluids (alkanes, alkenes, alkynes, fatty-acid methyl esters, alcohols)
and/or toxicity (ASHRAE Standard 34 classes B1 and B2), the
list of candidate fluids can be reduced to 33.
Finally, using simplified analysis, the COP and required
compressor displacement for the different fluids can be
1

A pseudo subcooling is used if the outlet of the radiator is

two-phase at an intermediate point in iteration.

calculated. Table 3 summarizes the performance of the cycles based on the assumption of equivalent isentropic efficiencies of 0.6 for all fluids, evaporation temperature of
40  C and condensation temperature of 90  C. A large COP
and a small (but not too small) displacement are desired,
and a good tradeoff between efficiency and displacement
can be found with the working fluid R152a, which will be
used in the further studies. Compressor sizes below
approximately 20 cm3/rev are problematic because the ratio
of leakage area to compressor displacement becomes large,
resulting in poor compressor performance. A rotational
speed of 60 Hz was selected in order to mirror terrestrial
applications.

4.

Compressor analysis

The compressor style selected for this aerospace heat pump

application is a scroll compressor. This style of compressor is
a well-proven technology for terrestrial air conditioning and
heat pump applications. The scroll compressor is characterized by having few moving parts, no valves, and generally high
performance over a wide range of operating conditions. In
addition, oil-free scroll compressors are already available on
the market in limited quantities for terrestrial heat pumps,
which will reduce the new engineering required for this
application.
The free and open-source mechanistic code Positive
Displacement Simulation (PDSim) has been developed by the
authors and is used to simulate the compressor performance.
The source code of the simulation framework used to generate
the compressor results in this paper is available in the
Supplemental information.
Selection of the dimensions and other parameters of the
compressor are highly dependent on the working fluid
selected for the system. Furthermore, due to the zero-gravity
application, it has been decided to use a non-lubricated machine; this decision will have significant consequences on the
machine's performance as the leakage gap widths must be
properly selected to avoid scroll/scroll contact. From discussions with experienced engineers, it seems radial and axial
gap widths of 20 mm are about the best that could be achieved
for oil-free scroll machines. The most important compressor

Table 3 e Candidate working fluids (Te 40  C, Tc 90  C,

5 K superheat, 0 K subcooling, 12 kW evaporator capacity;
displacement at 60 Hz rotation).
Name

Ttriple

C

Tcrit

C

Vdisp
cm3

COP


R152a
R134a
HFE143m
R1234ze(E)
R1234yf
R236FA
R227EA
R236EA
R1234ze(Z)

118.6
103.3
33.1
104.5
53.1
93.5
126.8
103.1
0.1

113.3
101.1
104.8
109.4
94.7
124.9
101.8
139.3
150.1

45.5
50.4
59.4
66.5
71.7
96.3
97.0
110.8
110.9

2.5
2.1
2.2
2.1
1.6
2.4
1.7
2.5
2.8

74

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

design parameters are summarized in Table 4. The built-in

volume ratio was selected to well match the expected
imposed pressure ratio. The other input parameters are given
in the PDSim input file in the Supplemental information.

Table 4 e Inputs for PDSim.

Parameter

Value
3

Displacement volume [cm /rev]

Built-in volume ratio []
Leakage gap widths [mm]
Motor Drive efficiency [%]
Refrigerant

45
2.7
20
95
R152a

4.1.

Compressor parametric study

Fig. 6 gives the results from the simulation of the nonlubricated compressor for this heat pump system. The evaporation saturation temperatures, condensation saturation
temperature and rotational frequencies were linearly varied
in the following ranges:

Te: suction saturation temperature: 30  Ce50  C

Tc: discharge saturation temperature: 70  Ce90  C

f: rotational frequency: 30 Hze80 Hz
From these ninety data points it can be seen that the
volumetric efficiency is a very strong function of the rotational
speed, and a weak function of the saturated temperatures.

Fig. 6 e Results of the compressor simulation in PDSim.

75

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

The volumetric efficiency increases as the difference TcTe

decreases. A somewhat similar trend can be seen for the
overall isentropic efficiency. In general, the isentropic efficiency increases monotonically with the rotational speed.
There is a ridge in the overall isentropic efficiency surface for a
given rotational speed. The optimal overall isentropic efficiency ridge is obtained for a temperature difference TcTe of
approximately 45 K. This optimal efficiency ridge represents
the points for which the built-in volume ratio best matches
the imposed pressure ratio.

4.2.

Compressor map

Over the ninety data points presented above, the mean

elapsed time for one simulation run was approximately four
minutes. While this computational speed is excellent when
considering the capabilities of the simulation code, it is much
too slow to use the detailed compressor model directly in the
integrated cycle model. Therefore it is necessary to either
obtain a correlation for compressor performance or to carry
out a multi-dimensional interpolation. In this case it was
decided to employ a correlative method where the overall
isentropic and volumetric efficiencies are fit as a function of
reduced rotational speed and reduced saturation temperatures. The form of the fit (obtained by manual iteration) is
given by




hv B0 Ter B1 T2er B2 T3er uBr 3 B4 ur

(9)

hoi C0 Tcr C1 ur C2 T2cr C3 T2er C4 Ter Tcr C5 u2r Ter C6 T2cr u2r

(10)
Tcr Tc K=353:13 K

(11)

Ter Te K=313:15 K

(12)





ur u rad$s1 345:575 rad$s1

(13)

where the coefficients Bi and Ci are given in Table 5.

Fig. 7 shows the accuracy of this correlative formulation.
The average absolute error of the volumetric and overall
isentropic efficiency correlations are 0.66% and 1.07%
respectively, and all points fall within 5%.

5.

Integrated analysis

In this section, results are presented for the heat pump under
study. This analysis covers a description of the components in
the system, results from the component models, a parametric
study based on altering the rotational speed of the compressor

Fig. 7 e Accuracy of correlations used for the compressor

model (dashed lines show 5%).
and the effective ambient temperature, and optimization of
the COP of the system based on setting the maximum surface
temperature in the evaporator.

5.1.

System and constraints

The heat pump system under study has a nominal cooling

capacity of 12 kW. It has the overall configuration described
above, and the details of the heat exchangers are described
here. Details of the compressor model are described above.
The evaporator is formed of 10 channels in parallel, each
with internal dimensions of 3.1 mm in width and 6.2 mm in
height.
The dimensions of the microchannel heat exchanger are
given in Table 6. These dimensions correspond to a ratio wmc/
wpass of approximately 4:1.

5.2.

Detailed results

The analysis here focuses on one operational point in detail in

order to investigate the performance of the heat exchangers,
compressor, and demonstrate some features of the system.
The most important values are summarized in Table 7.
Fig. 8 shows the temperature profiles in the condenser.
Here we see the three zones in the heat exchanger, first a desuperheating zone, then a two-phase zone, and finally the
subcooled zone. The difference in temperature between the
radiator surface and the refrigerant is always less than 2 K and
is therefore not plotted. Thus we can say that the limiting
thermal resistance is that between the radiator surface temperature Ts and the ambient temperature T. The decrease in
saturation temperature caused by the pressure drop is quite

Table 5 e Coefficients for efficiency correlations.

B0
B1
B2
B3
B4
C0

4.1890229
1.24550014
1.65814515
0.851259913
0.788119616
0.744036625

C1
C2
C3
C4
C5
C6

0.472016327
23.4915664
23.2113301
46.3226046
0.524246134
0.348140304

Table 6 e Microchannel condenser dimension.

Parameter

Value

Parameter

Value

Hrad
wi
wmc

5m
3.1 mm
60 mm

Nchannel
hi

10
12.4 mm

76

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

Table 7 e Output parameters for selected point.

Parameter

Value
38.5  C
77.1  C
4.40
58.7 g s1
65.0 kPa
36.1 kPa
0.665
0.818

Te
Tc
COP
m_
Dprad
Dpevap
hoi
hv

Fig. 10 e Heat transfer coefficient in the evaporator.

Fig. 8 e Temperatures in the radiator.

small, as evidenced by the nearly flat saturation temperature

of the refrigerant in the two-phase zone.
Fig. 9 shows the temperature profiles in the evaporator.
The refrigerant temperature decreases in the two-phase region (by approximately 1.5 K) due to the impact of the pressure
drop (total pressure drop is around 40 kPa), while the refrigerant temperature increases in the superheated zone. The
surface (component) temperature increases near the end of
the evaporator due to the strong decrease in heat transfer
coefficient caused by dryout in the transition from two-phase
to superheated vapor, as shown in Fig. 10.

5.3.

Fig. 11 e Overall results at T 220 K.

General results

During the course of an orbit, the satellite will be exposed to a

wide range of effective surrounding temperatures, and the
variation in ambient temperature will consequently impact
the working temperatures of the system. Figs. 11 and 12 show
the results for the system performance for two ambient
temperatures of 220 K and 130 K for fixed capacity and
displacement per revolution of the compressor. At T 220 K,

Fig. 9 e Temperatures in the evaporator.

Fig. 12 e Overall results at T 130 K.

as the rotational speed of the compressor increases, the

evaporation temperature decreases and the condensing temperature increases. The increase in temperature lift results in
a monotonic decrease in COP with rotational speed. When the
ambient temperature decreases, the evaporation and
condensation pressures also decrease (as does their difference), and as a result, the COP of the system at a T of 130 K is
better than that of the system at a T of 220 K.
The two most important parameters are the COP of the
system and the maximum temperatures of the electrical
components Tend. From a consideration of the COP, the slower
the rotational speed, the better. The optimal rotational speed
is that which yields exactly the required component temperatures as this rotational speed will maximize the COP.

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

5.4.

77

Optimization for imposed surface temperature

The electronics that form the primary payload of satellites

typically have a maximum working temperature, which can
vary widely depending on the particular payload. For illustrative purposes we have selected a 70  C maximum working
temperature of the electrical components. As shown in Fig. 9,
the maximum surface temperature in the evaporator is in the
superheated vapor section. Furthermore, Figs. 11 and 12 show
that the COP will be maximized for the rotational speed which
yields the highest working temperature of the components.
Thus it can be said that for a given payload power output and
sink temperature, the highest COP will be realized for the
rotational speed that yields exactly the maximum surface
temperature.
In practical systems, the electrical payload and the effective ambient temperature are not constant and vary
throughout the mission. In this analysis, this temporal variation has been realized by varying the evaporator load between 9 kW and 14 kW, and varying the effective ambient
temperature between 150 K and 220 K.
An additional numerical routine has been constructed that
nests the conventional cycle solver described above inside
another solver loop. This additional solver minimizes the
difference between the maximum surface temperature in the
evaporator and the specified maximum surface temperature
in the evaporator by adjusting the rotational speed of the
compressor. In this way, the solver will simultaneously
enforce the constraint of Ts,end  Ts,end,spec as well as maximizing the COP.
Figs. 13 through 16 show the results of the optimization for
a fixed evaporator maximum surface temperature of 70  C.
Fig. 13 shows that the required rotational speed only varies
between 52 Hz and 67 Hz, a relatively small range, and the
resulting COP values are greater than 3.2 (see Fig. 14. It should
be noted that the compressor map is only valid for evaporation temperatures between 30  C and 50  C, and condensation
temperatures between 70  C and 90  C. Figs. 15 and 16 show
that some of the values at capacities less than approximately
10 kW are outside the range of the compressor map, and
should be viewed with suspicion.
In general, increasing the effective ambient temperature
T or increasing the evaporator load Q_ evap has a similar impact
on the cycle performance. Increasing either of them:

Fig. 14 e Coefficient of performance as a function of T and

Q_ evap for maximum surface temperature of 70  C.

Fig. 15 e Evaporation temperature as a function of T and

Q_ evap for maximum surface temperature of 70  C.

Fig. 16 e Condensation temperature as a function of T and

Q_ evap for maximum surface temperature of 70  C.

reduces the COP of the system

increases the evaporation temperature

increases the condensation temperature

increases the required rotational speed of the compressor

6.

Conclusions

In this paper, component and system models have been

developed and used to simulate the performance of a heat
pump used to cool the electrical components of a satellite. The
primary conclusions of this study are that:

Fig. 13 e Compressor rotational frequency as a function of

T and Q_ evap for maximum surface temperature of 70  C.

The refrigerant R152a is found to be a good working fluid

for this system due to its high efficiency and reasonable
compressor sizing.

The COP that can be achieved are quite good over a wide
range of operating conditions.

78

i n t e r n a t i o n a l j o u r n a l o f r e f r i g e r a t i o n 5 8 ( 2 0 1 5 ) 6 9 e7 8

The system coefficient of performance can be maximized

by setting the maximum evaporator surface temperature
equal to the maximum allowed temperature.

When the maximum electrical component temperature is
imposed, the best performance is found at low ambient
temperature and low evaporator load.

Acknowledgments
The authors would like to thank the CNES (Centre National
d'Etudes Spatiales) and Thales Alenia Space for the financial
funding of this work. The scientific and technical support of
Thales Alenia Space is also gratefully acknowledged.

Appendix A. Supplementary data

Supplementary data related to this article can be found at
http://dx.doi.org/10.1016/j.ijrefrig.2015.06.017.

references

Bell, I.H., Wronski, J., Quoilin, S., Lemort, V., 2014. Pure and
pseudo-pure fluid thermophysical property evaluation and
the open-source thermophysical property library CoolProp.
Ind. Eng. Chem. Res. 53 (6), 2498e2508. URL. http://pubs.acs.
org/doi/abs/10.1021/ie4033999.
Bertsch, S.S., Groll, E.A., Garimella, S.V., 2009. A composite heat
transfer correlation for saturated flow boiling in small
channels. Int. J. Heat Mass Transf. 52, 2110e2118.

Churchill, S., 1977. Friction factor equation spans all fluid flow
regimes. Chem. Eng. 84 (24), 91e92.
Cole, G.C., Scaringe, R.P., Grzyll, L.R., Ewert, M.K., 2006.
Development of a gravity-insensitive heat pump for lunar
applications. In: Space Technology and Applications (STAIF);
11e15 Feb. 2007, vol. 2006.
Domitrovic, R., Chen, F., Mei, V., Spezia III, A., 2003. Microgravity
heat pump for space station thermal management. Habitation
9, 79e88.
EARSC, May 2011. ISRO to Launch French Satellite in 2012.
rme-und den
Gnielinski, V., 1975. Neue Gleichungen fur den Wa
mten Rohren und
Stoffubergang in turbulent durchstro
len. Forsch. Im. Ingenieurwes. 41 (1), 8e16.
Kana
Herwig, H., 2013. High heat flux cooling of electronics: the need
for a paradigm shift. J. Heat Transf. 135, 111013, 1:2.
Kim, S.-M., Mudawar, I., 2012. Universal approach to predicting
two-phase frictional pressure drop for adiabatic and
condensing mini/micro-channel flows. Int. J. Heat Mass
Transf. 55, 3246e3261.
Kim, S.-M., Mudawar, I., 2013. Universal approach to predicting
two-phase frictional pressure drop for mini/micro-channel
saturated flow boiling. Int. J. Heat Mass Transf. 58, 718e734.
Messaros, M.C., Verstracte, J.L., 1994. Design and development of
a high reliability oil lubricated compressor for a space Borne
Joule-Thompson cryocooler. In: International Compressor
Engineering Conference at Purdue University.
Nikanpour, D., Aidoun, Z., De-Parolis, L., Lebru, A., 1997.
Advanced heat pumps for interplanetary spacecraft/lander
thermal control. In: Proceedings of the Sixth European
Symposium on Space Environmental Control Systems.
Shah, M., 1979. A general correlation for heat transfer during film
condensation inside pipes. Int. J. Heat Mass Transf. 22,
547e556.
St. Pierre, M., 1987. Zero Gravity (Position Insensitive) LowTemperature Multi-Component Refrigerator.
Woolley, R.A., 1993. Separation Method and Apparatus for a
Liquid and Gas Mixture.