DOI: 10.18462/iir.gl2022.

0010

AIR-AIR SPLIT AC WITH MINIMUM CHARGE OF PROPANE

Klas ANDERSSON*(a), Gunilla LUNDAHL(a), Eric GRANRYD(b)
(a)
Klas Andersson Engineering,
Lidingö, 18161, Sweden, [email protected]
(b) Professor emeritus, The Royal Institute of Technology,

Stockholm, 10044, Sweden, [email protected]

ABSTRACT
This paper describes technology for an environmentally friendly split type air conditioner, characterized by
using safe amounts of propane as refrigerant. The objective was to demonstrate a system, using less than
150 g of propane, providing minimum 3.5 kW cooling capacity with a COP2≥3.5 at air flows that allow
acceptable noise levels.

The condenser and evaporator were of standard microchannel type. The evaporator headers were rebuilt to
reduce charge and different techniques were tested for even refrigerant distribution. A DC-motor scroll
compressor for AC in electric vehicles was used, characterized by small internal volumes, small oil charge and
wide capacity range (800-8500 rpm). The system also included a specially built minichannel liquid/suction
line heat exchanger and a standard electronic expansion valve. The connection lines in-between the indoor
and outdoor unit were 6 m long.

The project target was reached with a charge of 143 g propane.

Keywords: Air conditioner, Propane, Minimum charge, Heat exchangers.

1. INTRODUCTION
This project is based on experiences from earlier studies at the Royal Institute of Technology (Fernando et al,
2008) and a study of ground source heat pump with minimum charge of propane (Andersson, 2018). This
time the same approach was applied on a split air conditioner.

Since a couple of years there is a shift going on from R410A (GWP 2088, typical charge>1000 g) to R32 (GWP
675, typical charge>800 g) for air conditioners of the type studied in this paper. R32 is not a long term solution
according to the Kigali Amendment to the Montreal Protocol. The substitutes at hand are natural or HFO:s.
The HFO:s are synthetic, non natural refrigerants that decompose into so called “For ever chemicals” in the
atmosphere and questioned for their environmental impact. Many natural alternatives, like propane, are
flammable. As pointed out by many authors (IIR 25th info. note, 2014) reducing refrigerant charge will be
important for most systems. Beside the low density, propane has shown to be a good refrigerant in many
aspects (Pavel, 2014). In parallel to the shift to R32, large numbers (hundreds of thousands) of propane
systems have been placed on the Asian markets. Typically they have a charge of about 350 g. This can not be
accepted in western countries for safety reasons. According to EN378 it is recommended to have maximum
150 g of propane to avoid aggravating complications for installation. Most potential hazardous situations
can then be taken care of within the system boundaries according to IEC60335-2-40. The project goal was to
use less than 150 g of propane, providing minimum 3.5 kW cooling capacity with a COP2≥3.5 at a maximum
air flow of 15 m3/min and 30 m3/min for the indoor and outdoor unit respectively. Isobutane (R600a) is an
interesting alternative but has not been studied in this project.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
 2. THE SYSTEM
A number of versions of the system were built and tested throughout the project, all with the same basic
layout as illustrated in Figure 1, including measured quantities. The indoor unit was connected to the outdoor
unit with 6 m long, flexible polymer hoses, approved for refrigerants. The outdoor unit was built to fit in to a
standard architecture whiles the indoor unit was quite different from the standard appearance, having a
quadratic shape with an axial fan. See Figure 2.

p→ p→
flow p p flow
L3

p T p
T
E
rpm
L1 Tw,d Tw,d
T E E
p
rpm A
A rpm Tw,d
p L2

p: Pressure Indoor
p: Pressure drop T T
p unit
A: Air flow T p p T L7
T
T: Temperature w,d: wet and dry
E, rpm: Electric power and speed L4 Heat
L5 L6
Figure 1: System layout and measurement positions.

Figure 2: The out- and indoor units on wheels, connected with flexible hoses.
2.1. Condenser
The condenser was a standard so called “microchannel” heat exchanger with two passes. The microchannel
technique provides high heat transfer performance in relation to the internal volumes. The condenser was
selected from an estimated balance of these properties. The geometry is given in Figure 3 and Table A1 in
the Appendix.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
 530

Section 1

Pass 1
Inlet

Section 2

Header
500
Section 3

Liquid
pools Pass 2

Outet Channels Tube

16

Figure 3: Condenser dimensions (left) and design principle for condenser and evaporator (right).

2.2. Evaporator
The evaporators were of the same type as the condenser.
Three different evaporators were tested with different
configurations as described in Table 1. The evaporators
originally had insertion of the tubes into the headers. In
evaporator B and C these insertions were cut using wire
sparking technique. One header end was thereafter
provided with a screw connection with O-ring sealing in
order to allow easy exchange of distributor. The header
before and after this process is shown in Figure 4.

Table 1. Evaporator configurations tested

Evaporator A A B B C C
Inlet position bot bot bot top bot top
Header orientation vert horiz horiz horiz horiz horiz
No. of passes 2 2 1 1 1 1
No. of distributers tested non non 1 2 1 1
inserted tubes yes yes no no yes no Figure 4: Rebuilt of evaporator headers

The conclusions from the tests of different evaporator configurations were that

- the headers should be horizontal. Drainage of condensed water with vertical headers (horizontal tubes)
was insufficient and resulted in poor heat transfer and blockage of the air flow.

- two passes gives high pressure drop and complicates distribution since there are two entrances to
parallel tubes.

- the insertion of the tubes into the headers creates stagnant sections where separated liquid is
accumulated.

- with the insertion of the tubes into the header removed, the refrigerant inlet should be at the top.
Injection via the bottom header accumulates separated liquid at the bottom, increasing the charge.

- even distribution is crucial, not only for the performance but also for the charge.

The evaporator dimensions are given in Table A2. Evaporator A was the same as the condenser and was
rejected due to its size, high refrigerant side pressure drop and uneven distribution. Evaporator C was

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
specially built for the project, with deeper tubes and fins in the direction of air flow and with more tubes and
channels compared to B. It has yet only been evaluated with some initial tests that do not have the quality
to be published. The results given in this paper are therefore for evaporator B, which is a rebuilt standard
heat exchanger, with removed insertion of the tubes and with a distributer developed within the project.

2.3. Fans
The same axial fan models were used for the evaporator and condenser. At the nominal air flows for the
project, 15 m3/min and 30 m3/min for the evaporator and condenser respectively, the total fan efficiency,
including motor, is about 30%. The operation point for the evaporator fan differs between dry and wet
conditions. At wet conditions the pressure drop in the evaporator is higher due to water condensation. This
is a very interesting but complicated optimization issue that was out of the scope of this study.

The nominal air flows were defined by a bench mark study of air flows related to acceptable noise levels. A
difference though is that the bench marked indoor units have a different type of fan (cross flow / tangential).
The noise levels were not measured but the subjective impression by listening was that they were low
enough.

2.4. Compressor
The compressor is a major contributor to the project target by its small interior volumes and small oil charge.
It is of semi-hermetic scroll type with permanent magnet DC motor with a speed range of 800-8500 rpm,
giving the system a wide capacity range. The drive electronic is integrated in the compressor and cooled by
the suction gas. It has ball bearings why oil mainly is required for lubrication and sealing of the scroll wheels.
The compressor is made for climatization in electric vehicles using R134a or R1234yf as refrigerant. An
alternative oil adopted for R290 with this compressor was used. Data and estimated refrigerant charge hold
up is given in Table A3.

2.5. Connection lines

In Table A4, the dimensions of the connection tubes defined in Figure 1 are given together with calculated
pressure drop, expressed as saturated temperature difference and estimated refrigerant hold up. The two-
phase line, L7, needs some extra attentions since it together with the distributer adds quite some pressure
drop that reduces the operation range for the expansion valve, limiting the maximum capacity (mass flow)
of the system. Balancing capacity range (pressure drop) against charge is an interesting challenge. For this
line and the suction line, L1, flexible polymer hoses were used. This make the system into a flexible, more or
less portable unit which is handy for experimental work but could also be of interest for some commercial
applications. The suction line dimensions are also a trade off between charge and pressure drop. The length,
6 m, was the result of the bench mark study as recommended for related market requirements. However, it
is a risky concept since liquid refrigerant and oil can be captured in lower bends. It is always a good idea with
a continuous downward slope of the suction line. As a consequence of short connecting tubes, the
temperatures measured can be influenced by heat conduction from adjacent components. To reduce that
problem stainless steel tubes where used for lines L4 and L6.

2.6. Internal heat exchanger

The internal heat exchanger was made by pressing copper tubes into a flat shape and soldering them
together, including connection headers. The dimensions and calculated characteristics are given in Table A5.
The heat balance over the internal heat exchanger was used to calculate the vapor quality at the evaporator
outlet. It was experienced that at a measured superheat at the evaporator outlet of 5 K, the vapor quality
could be about 0.95. The relation between the superheat and outlet quality is an indirect measure of the
distribution in the evaporator. The large diameter of the outlet header seems to provide conditions where
superheated gas flows together with a fraction of liquid that is not in thermodynamic equilibrium. Remaining
liquid enters the long, wide suction line to the outdoor unit and increases charge. In order to avoid liquid at
the compressor inlet the internal heat exchanger is important.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
2.7. Expansion valve and filter/dryer
The expansion valve was of standard electronic type and the filter dryer was of household refrigerator type,
rebuilt to minimize the voids.

2.8. Charge considerations

It is likely that liquid is separated in the wide headers. In
the condenser the liquid can be assumed to accumulate
in the bottom (illustrated in Figure 3), filling the bottom
tube channels. In the evaporators the liquid accumulates
along the horizontal bottom of the headers as illustrated
Liquid
in Figure 5. At the top inlet, with most of the insertion of
pool
the tube removed (about 1 mm is left of the solder joint
after the wire sparking illustrated in Figure 4) the liquid
level is low. With original insertion it can, in worst case, Liquid
be in level with the tube endings. At the bottom outlet, pool
there will be a liquid pool proportional to any inclination
downwards from the outlet. It is therefore important to
ensure that the header is horizontal or preferably, slightly Figure 5: Liquid pools in headers
inclined downwards towards the outlet. The evaporator
outlet connection needs to be arranged for continuous inclination downwards with no edges to overcome.
For both the evaporator and condenser, a liquid film along all inner surfaces also has to be considered. All
together this makes estimate of expected charge difficult.

The wide, flexible suction hose can easily have up and down coming bends where liquid and oil can be
accumulated. A heat balance over the internal heat exchanger indicates the presence of remaining liquid at
its vapor side inlet. This liquid adds charge in the suction line. Also here, a thin film of oil and refrigerant
mixture along the hose walls can be assumed. For the two-phase line the charge is estimated by stepwise
calculation along its length, taking the pressure drop and corresponding increase of the vapor fraction into
account (in a similar way as for capillary tube calculations). The calculation was iteratively balanced with the
pressure drop over the expansion valve. A complication is the added pressure drop over the distributer which
is difficult to calculate. 60

The charge in the compressor is calculated from 50

known volumes and states on the respective low and 40
Charge [gram]

high pressure side. The amount of refrigerant 30

dissolved in the oil can be estimated from solubility 20
charts. The distribution of the oil in the system and
10
the corresponding states are unknown though,
which makes the estimate uncertain. 0

Despite these uncertainties an attempt to estimate

the charge in the respective components is given in
Figure 6. The assumptions are adopted to sum up to
Figure 6: Estimated charge distribution
the total charge, 143 g, used in the system.

3. TEST SETUP

3.1. System control and performance measurement

The components and the measurement system used different communication protocols. To control and
acquire data from the different parts, a special data system was developed allowing for setting fan and
compressor speeds, expansion valve opening ratio and real time plotting of data from these components,

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
merged with output from the measurement system. The data system used CoolProp for calculations and
EasyView for real-time presentation.

The cooling capacity was determined by the air flow and the enthalpy Table 2. Air test conditions
change of the air over the evaporator, defined by wet and dry bulb Tdry,bulb Twet,bulb RH Dew point
temperatures. With known cooling capacity, the refrigerant mass °C °C % °C
Outdoor 35 24 40 19.5
flow was determined from the states measured at the evaporator in-
Indoor 27 19.4 49 15.5
and outlet. The mass flow and measured states along the system (see
Figure 1) defined the performance of the other system components. All tests were performed at the
conditions given in Table 2.

During measurement the evaporator and fan unit was connected to a measurement unit, including an extra
fan, compensating for the pressure drop in the measurement system. The condenser fan was replaced with
a fan-nozzle unit to be able to measure the air flow. The fan power and condenser performance with the real
fan was then confirmed in a separate measurement using fan data equations for the air flow. This adds some
uncertainty for condenser air flow.

3.2. Charging
The system was filled from a small, 400 g, bottle connected to the filling station via a capillary tube winded
into spiral spring to avoid any forces to influence the scale reading. The capillary tube also lowers the filling
flow rate, facilitating accurate filling. The scale had a resolution of 0.1 g.

The charge was determined by filling during operation at nominal conditions (Q2~3.5 kW T1~43°C, T2~10°C)
until a combination of about 4 K subcooling and superheat was achieved. At higher capacities, when T1
increases and T2 decreases, less charge is required and the opposite for lower capacities. There are two
reasons for that. 1) The internal volumes of the low pressure side (mainly suction lines and compressor) are
much larger than on the high pressure side. 2) Higher T1 and lower T2 increase the vapor fraction in the long
two-phase line and at the evaporator inlet. It is an interesting challenge to design a minimum charge system
for a wide range of operation conditions. In this project mainly the nominal conditions were considered.

4. RESULTS
All results given in this chapter are with a charge of 143 g R290 and at the test conditions given in Table 2.
The evaporator was type B with removed inserted tubes and injection at the top. The connection lines
between the in- and outdoor units were 6 m long and had an inner diameter of 10.5 mm and 5.2 mm for the
suction and two-phase hose respectively.

4.1. System performance 5500 4,5

Q2
Figure 7 shows the cooling capacity and COP2 variation 5000 COP2
with compressor speed at 30 m3/min condenser and 11- 4,0
4500
Cooling capacity [W]

15 m3/min evaporator air flow. The target capacity, 3.5

4000 3,5
kW, was reached at 1900 rpm compressor speed and
COP2

3500
then the COP2, including fan powers, was just above the
3,0
target, 3.5. At that point the evaporator air flow was app. 3000
12 m3/min. The evaporator fan was operated at constant 2500
2,5
setting. As the evaporating temperature decreased with 2000
increased compressor speed, the rate of water
1500 2,0
condensation (drain rate) increased in proportion, 0 1000 2000 3000 4000 5000
blocking the air flow. This is illustrated in Figures 8 and
Compressor speed [rpm]
9. The jump in air flow and pressure drop in Figure 9 is
related to the passage of the dew point, 15.5°C, at the Figure 7. System performance with compressor
evaporator outer surfaces, somewhat above the speed. Superheat 5 K, Evap. air flow 11-15 m3/min,
Cond. air flow 30 m3/min.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
evaporating temperature. The water drainage from the fins varied in an intermittent way illustrated by the
pressure drop at 3.5 kW, which is not a minimum.

3 20 16 28
Drain rate
18

Evaporating temperature, T2 [°C]

T2 15
2,5 27
16

Air side pressure drop [Pa]

14
14
2 26
Drain rate [kg/h]

13

Air flow [m3/min]

12
1,5 10 12 25
8
1 11
6 24
10
4
0,5 Air flow 23
2 9
Pressure drop
0 0 8 22
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Compressor speed [rpm] Compressor speed [rpm]

Figure 9. Drain rate and evaporating temperature Figure 8. Evaporator air flow and air side pressure
with compressor speed. Superheat 5 K, Evap. air drop with compressor speed. Superheat 5 K,
flow 11-15 m3/min, Cond. air flow 30 m3/min. Cond. air flow 30 m3/min.

There is a large potential for improved performance if higher air flows could be accepted, which can be seen
in Figure 10. This is though limited by acceptable noise levels. The maximum for both cooling capacity and
efficiency is at flow rates above 25 m3/min. Figure 11 shows the performance variation with the superheat.
The dependence is weak but shows a continuous increase with reduced superheat. Figure 12 shows the
system performance variation with the condenser air flow. The influence on the capacity is as expected small.
The COP2 has an optimum at about 35 m3/min, not far from the nominal 30 m3/min

3600 3,60
3900 3,9
3700 Cooling capacity 3550 3,55
COP2 3,7
3500
Cooling capacity [W]

3500 3,50
Cooling capacity [W]

3,5
COP2

3300
3450 3,45

COP2
3100
3,3
3400 3,40
2900
3,1
2700 3350 Q2 3,35
COP2
2500 2,9 3300 3,30
0 5 10 15 20 25 0 5 10 15
Evaporator air flow [m3/min] Superheat in evaporator [K]

Figure 10. Performance with evaporator air flow. Figure 11. System performance with superheat in
Superheat 5 K, Cond. air flow 30 m3/min. Comp. the evaporator. Evap. air flow 12 m3/min, Cond. air
speed 1900 rpm. flow 30 m3/min, Comp speed 1900 rpm.
4.2. Evaporator and condenser performance
The evaporator and condenser heat transfer performance (UA2-value) variation with the air flow and
corresponding fan powers are illustrated in Figure 13. The evaporator fan power increases with the flow rate
until the largest air flow, where the evaporator surface temperature passes the dew point and the pressure
drop decreases. The UA2 decreases linearly with increasing superheat by 3.44 (W/K)/K (not illustrated). The
pressure drop of the refrigerant side of the evaporator at 3.5 kW cooling capacity is 25 mbar corresponding
to a saturation temperature difference of 0.14°C (not illustrated).

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
The change in superheat has a direct influence on the 3600 3,70
subcooling in the condenser as illustrated in Figure 14.
Figure 15 shows infrared images of the evaporator at 4.3 3500
3,60
and 12.6 K superheat. The right hand side seems to be
3400
starved and the tendency worsens at the higher

Cooling capacity [W]

3,50

COP2
superheat. For the 4.3 K superheat there are cold spots 3300
at the bottom outlet indicating a possible combination 3,40
of remaining liquid and superheated vapor. The image 3200

only sees the evaporator front surface. Maldistribution 3100

3,30
Cooling capacity
inwards, in-between cannels along the tubes, is hidden. COP2
There is obviously a potential to improve the evaporator 3000 3,20
performance by better distribution. Filling it more even 10 20 30 40 50
Condenser air flow [m3/min]
is an advantage also in terms of charge compared to let
liquid out in the voluminous suction line Figure 12. System performance with condenser air
flow. Evap. air flow 12 m3/min, Superheat 5 K,
Comp speed 1900 rpm.
600 140
6
UA1
500 UA2 120
5
Cond fan power
100 Subcooling in condenser [K]
Fan power [W]

400 Evap fan power

4
UA-T inlet [W/K]

80
300 3
60
200 2
40
100 20 1

0 0 0
0 10 20 30 40 50 0 5 10 15
Air flow [m3/min] Superheat in evaporator [K]

Figure 13. Evaporator and condenser performance Figure 14. Subcooling variation with superheat.
with air flow. Superheat 5 K, Subcooling 4-5 K, Evap. air flow 12 m3/min, Cond. air flow 20 m3/min,
Comp speed 1900 rpm. Comp speed 1900 rpm.

23°C

Sup.=4.3 K, Tair,in=27°C, T2sat=9.5°C 5°C Sup.=12.6 K, Tair,in=27°C, T2sat=7.6°C

Figure 15. Thermal imaging of the air outlet side of the evaporator. The surface
was painted matt black.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
4.3. Internal heat exchanger and compressor efficiencies
The measured temperature effectiveness, IntHex, of the internal heat exchanger at the operation point at
which the project target was met was 0.39 and the corresponding UA-value was 11.3 W/K, referred to
logarithmic mean temperature difference.

𝑇𝑔𝑎𝑠,𝑜𝑢𝑡 − 𝑇𝑔𝑎𝑠,𝑖𝑛
𝜂𝐼𝑛𝑡𝐻𝑒𝑥 =
𝑇𝑙𝑖𝑞,𝑖𝑛 − 𝑇𝑔𝑎𝑠,𝑖𝑛

The compressor isentropic efficiency, including losses in the AC to DC rectifier, was 0.61 at project target
conditions. This is lower than expected. A reason could be that oil is captured in the suction line which
reduces lubrication and sealing of the scroll wheels. Also the volumetric efficiency was lower than expected,
0.92, probably for the same reason.

5. DISCUSSION

5.1. Charge considerations

A major topic of the project was the charge. During the project a lot of configuration combinations were
tested and some interesting differences in the charge was experienced. The learning from this quite extensive
trial and error process is summarized in Table 3 in which the charge and performance are given for the
respective variants. All are using the same condenser and operated at similar conditions. This does not tell
the full story but gives a base for discussion of charge considerations.

The system with evaporator A, Table 3. System configuration variants and their performance
being the same as the condenser, Evaporator In-Outdoor lines Charge Q2 /COP2
took 150 g despite the smaller Evap
Header
Inlet
Tube Distri- Two-phase Suction
gram kW / -
lines. The performance with orient. Insert butor L / di L / di
A Vert. Bot yes non 3m/5.2mm 3m/8.4mm 150 3.5 / < 3
vertical header was very poor due
A Horiz. Bot yes non 3m/5.2mm 3m/8.4mm 150 3.5 / 3.3
to the insufficient water drainage.
With vertical headers the
B Horiz. Bot no B1 3m/5.2mm 3m/8.4mm 140
performance improved, but
B Horiz. Top no B1 3m/5.2mm 3m/8.4mm 122
despite its large dimensions it
B Horiz. Top no B2 3m/5.2mm 3m/8.4mm 120
didn’t reach the efficiency target,
B Horiz. Top no B2 6m/5.2mm 6m/10.5mm 143 3.5 / 3.5
mainly because of the pressure
drop and uneven refrigerant
C Horiz. Bot yes C1 3m/5.2mm 3m/8.4mm 180
distribution.
C Horiz. Top no C1 6m/5.2mm 6m/10.5mm 155 3.5 / 3.4

The tests with evaporator B resulted in a charge decrease of about 20 g by injecting at the top instead of the
bottom. The switch to the larger connection lines added about 20 g.

Evaporator C with remaining tube insert, the small lines and injection at the bottom took 180 g and was
therefore not further tested. The test with removed tube insert, injection from the top and the larger lines
almost reached the target but took 155 g. The reason for the charge being higher than for evaporator B is
partly the larger headers (30 vs. 20 mm diameter). The main explanation is though poor distribution resulting
in large amounts of liquid entering the suction line and even passing the internal heat exchanger vapor side
into the compressor. This also reduces the system performance. It is likely that evaporator C will perform
better than B, within 150 g of charge, with improved distribution.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
5.2. Refrigerant considerations
Tests for annual efficiency declarations include lower capacity requirements at lower ambient air
temperatures than used in this paper. For the lowest capacity in such a test series this system has too high
capacity at the lowest compressor speed, 800 rpm. The result is low efficiency for this point. An interesting
alternative would be to use isobutane (R600a) instead of propane. The volumetric cooling capacity is about
58% lower for R600a than for R290 which would improve the annual efficiency considerably. The wide speed
range of the compressor would cover the maximum capacity required with R600a. Another benefit with
R600a is the low discharge temperature which allows high pressure ratios and high capacity even at high
outdoor temperatures with low risk of oil decomposition or compressor damages. The efficiency is higher
with R600a for a basic, ideal cycle. With an internal heat exchanger R600a comes out even better due to the
positive influence of internal superheat on the performance. However, the higher volume flow rates for
R600a increases pressure drop which has to be considered in the design of the heat exchangers and
connection lines. In terms of charge, R600a has slightly higher liquid density but less than half the vapor
density than R290. The system charge can therefore be expected to be more or less the same. A summary of
properties is given in Table A6 where also the R32 is included.

5.3. Control
It was most of the time possible to get stable operation with very low superheat. At some situations though,
oscillations of the superheat, and corresponding sub cooling was experienced. It was not systematically
investigated, but it might be related to conditions with remaining liquid at the inlet to the internal heat
exchanger vapor side. A hypothesis is that evaporation in the internal heat exchanger vapor side amplifies
the reconnection to the liquid side by additional subcooling, reducing the restriction over the expansion
valve. If so, even distribution in the evaporator, so that low superheat can be combined with high vapor
fraction at the evaporator outlet, becomes extra important.

At normal situations, with no or very small oscillations, changes in the system (compressor speed, fan speed
or expansion valve setting) could be made after which a new stable operation point was established more or
less instantly. The high refrigerant mass flow rate, about 12 g/s, in relation to the total charge causes
maldistribution of the refrigerant in the system to develop in seconds. This puts heavy demands on the
control system. It is important to avoid that gas bubbles follow the liquid at the condenser outlet which
potentialy introduce a risk of short-circuiting heat transfer between condenser and evaporator side. Despite
the potential short circuiting over the internal heat exchanger, it provides extra control margin against
harmful liquid entering the compressor.

6. CONCLUSIONS
It is possible to build an AC split unit with less than 150 g propane, providing 3.5 kW cooling capacity with
acceptable efficiency, noise levels and tube length between the in- and outdoor units.

It has been demonstrated that a wide range of capacity can be provided and that the drain rate (capacity to
dry humid air), an important property at many markets, is in level with products on the market. The drainage
of water from the evaporator surfaces is an important property for the performance that needs to be further
studied.

The minimum charge required for the system is defined at the lowest capacity. At higer capacities the optimal
charge is even smaller.

The compact vehicle DC compressor and the microchannel type of heat exchangers are key technologies for
minimum charge. The refrigerant distribution in the evaporator and the geometry of its headers are major
areas for further reduction of the charge and improvement of the performance. Injection at the top of the
evaporator reduces charge considerably.

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
It is recommended to use an internal heat exchanger in the type of system presented in this paper. It
increases control margins, allows low superheat in the evaporator and provides a small but positive
improvement of the cooling capacity and efficiency.

The control of the type of system presented in this paper requires new control strategies.

ACKNOWLEDGEMENTS
The project achievements are the result of a teamwork with academic experts, small local Swedish companies
and large international enterprises. It has been a pleasure to see all competences come together, in good
spirit, to take a small step towards a more sustainable world. I would like to thank all for the contributions.

The project depended on the foundation laid by patient, foreseeing research work, in the field of small
charges of natural refrigerants, by Professor Björn Palm and his colleagues at the division of Applied
Thermodynamics and Refrigeration at the Royal Institute of Technology in Stockholm.

My gratitude for the courage shown by the Swedish Energy Agency by approving the application for this
project, coming from a private company.

NOMENCLATURE
Temperature effectiveness of internal heat
T Temperature int,hex
exchanger Tvapor/Tmax
p Pressure sat Saturated conditions
Q2=3.5kW, T1=43°C, T2=10°C,
Nominal
Subcooling=4K,  Density
conditions
Superheat=4K, int,hex=0.4
1 State at condenser inlet 2 State at evaporator outlet
UAi Qi/(Ti-Tair,inlet) Q capacity
COP2 Q2/(Compressor + Fan powers)

REFERENCES
Andersson, K. 2018. Water to water heat pump with minimum charge of propane, 13 th Gustav Lorentzen
Conference on natural refrigerants. Valencia, Spain.
Fernando, W. P., Palm, B., Ameel, T., Lundqvist, P. & Granryd, E. (2008). A Minichannel Aluminium Tube Heat
Exchanger - Part I-III: Evaluation of Single-Phase Heat Transfer Coefficients by the Wilson Plot Method.
International journal of refrigeration, 31(4), 669-680, 681-695, 696-708.
Granryd, et. Al., 2009. Refrigerating Engineering. Textbook. The department of Energy Technology,
Stockholm.
Pavel Makhnatch, Gunda Mader, 2014. Refrigerants with low GWP and cost and energy efficiency
optimization of vapor compression systems, Effsys Report
Refrigerant Charge Reduction, International Institute of Refrigeration, 25th Informatory Note on Refrigeration
Technology

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway
 7. APENDIX – TABLES

Table A1. Condenser geometry Table A2. Evaporator geometry

Heat exchanging length [mm] Evaporator B C
500
Heat exchanging height [mm] Heat exchanging length [mm] 410 420
495
Fin distance [mm] Heat exchanging height [mm] 412 385
1.1
Fin height [mm] Fin distance [mm] 1.1
8.1
Fin depth [mm] Fin height [mm] 8.1 6.5
16
Tube height [mm] Fin depth [mm] 16 20.6
1.3
Tube height [mm] 1.3 1.3
Channels internal volume [cm3] 197.8
3
Channel volume [cm ] 141.1 206.7
Header internal volume [cm3] 202.2
3
Header volume* [cm ] 159/172 415/448
No. of flat tubes 52
No. of channels in each tube No. of flat tubes 43 52
16
Channel dimensions [mm] No. of channels/tube 16 20
0.63 x 0.74
Channel dim. [mm] 0.63 x 0.74 0.675 x 0.74
*Both headers. Inserted tubes / No inserted tubes
Change in channel volume is neglected
Table A3. Compressor characteristics
Swept volume [cm3/turn] 33 Table A4. Connection lines
Speed range [rpm] 800-8500 Di Length Tsat(p) Charge
Suction side volume [cm3] 500 mm m °C gram
Discharge side volume [cm3] 260 L1 10.5 6 0.06 7.0
Oil type PAG II L2 11 1.2 0.06 1.5
Oil amount [ml] 120 L3 7.83 1.6 0.1 2.1
Weight [kg] 6.85 L4 3.0 0.20 0.11 0.7
Charge Suction/Discharge/Oil [g] 4.9/7.1/5.8 L5 4.65 0.10 0.01 0.8
L6 3.0 0.20 0.11 0.7
L7 5.2 6 0.11 9.7

Table A5. Int. Hex. characteristics

Q2=3.5kW T1=43°C Vapor Liquid Table A6. Refrigerant comparison
T2=10°C Tsup=4K Tsub=4K side side Given operation conditions
Height [mm] 1.2 0.6
T1,sat [°C] 50
Width [mm] 10 10
T2,sat [°C] 10
Length [mm] 300 260
Subcooling in condenser [K] 4
Number of channels 16 3
Connection diameter [mm] 11 4.65 Superheat in evaporator [K] 4
Compressor isentropic efficiency 0.65
Charge hold up [g] 1.5 4.1
Internal hex effectiveness 0.4
Tsat(p) [K] 0.03 0.1
Resulting states
Refrigerant Propane Isobutane R32
p1,sat [bar] 17.1 6.8 31.4
p2,sat [bar] 6.4 2.2 11.1
1,vapor [kg/m3] 39 18 99
2,vapor [kg/m3] 14 6 30
1,liquid [kg/m3] 449 517 839
2,liquid [kg/m3] 515 569 1020
T-discharge [°C] 81 73 117
T-suction gas [°C] 27 27 27
Refrigerant performance for basic, ideal cycle (Granryd)
Volumetric cooling
3414 1439 6626
effect [kJ/m3]
COP2 for ideal cycle
5.39 5.70 5.30
and no internal hex
% Increase of Q2 per K
0.187 0.238 -0.137
internal hex superheat
% Increase of COP2 per K
0.132 0.171 -0.162
internal hex superheat

15th IIR-Gustav Lorentzen conference on Natural Refrigerants | June 13-15 | Trondheim, Norway