CO2 System Performance

Produced by Ian Wilson

Bitzer Australia
134-136 Dunheved Circuit
St Marys 2760
(Sydney) Australia
Tel.: +61 (2) 88019300


IIR 29 - Carbon dioxide for use as a refrigerant



To date there has been a large volume of theoretical data on how this refrigerant performs, but very little hard evidence exists that documents the actual performance of R744 (CO2).  How it compares to more conventional refrigerants, in a controlled test environment, on a full scale freezer store application.


This presentation deals with the, equipment and  installation costs, the operating performance, the actual operating costs and the service and maintenance issues.  All of which have been measured for each of two separate refrigeration systems which are installed into the one freezer store.


Each of the two systems (both 72kW cooling capacity at -31SST / 35degC ambient), the first a single stage R404A system and the second a R744 / R404A cascade system operate at the same room and ambient conditions. They have been running, week on week off, so an accurate comparison can be drawn between the two.



Room data

The freezer store is held at or near to -25degC.  It is 28m x 14m x 7m (high) and is used for the storage of frozen food for a packing and distribution center in western Sydney (Australia).

There are two plants.

Plant  “A” consists of one multiple compressor rack (single stage ) rack 1

Plant  “B” consists of two multiple compressor racks (connected together as a cascade system ) rack 2 high stage and rack 3 low stage.


Plant “A” (Rack 1) - Multiple (4) single stage reciprocating piston compressors on R4O4A

The system is a rack of four, six-cylinder semi hermetic piston compressors (model Bitzer 6H-35.2Y) with a common discharge, suction and oil system. Each compressor has two capacity control heads and head cooling fans fitted.

The east wall of the freezer has three Buffalo Trident (model EBBL 323) electric defrost evaporators, which are feed by Danfoss (electronic pulse) TX valves, and drawing liquid refrigerant from a single horizontal receiver vessel, fitted to the rack.

The suction lines are collected in a gas header/accumulator, while the discharge is collected in a common header and directed into a single oil separator.

The operating conditions are -31degC SST and 45degC SCT.

Figure 1 Cascade system pipe schematic


Plant “B” (Rack 2) High stage - Multiple single stage reciprocating piston compressors on R404A

Three four-cylinder semi hermetic piston compressors (model Bitzer 4NC-20.2Y) with a common discharge, suction and oil system. Each compressor has one capacity control head fitted.

The operating conditions are -15degC SST and 45deg C SCT.


Plant “B” (Rack 3) Low stage - Multiple single stage reciprocating piston compressors on R744

Three two-cylinder semi hermetic piston compressors (model 2EC-4.2K) with a common suction, discharge and oil system. One compressor is fitted with a variable speed drive to achieve capacity control.

The west wall of the freezer has three Buffalo Trident (model EBBL 255 ) electric defrost evaporators, which are feed by Danfoss (electronic pulse) TX valves, and drawing liquid refrigerant from a single vertical receiver vessel, which is located within the freezer store.

The operating conditions are -31degC SST and -10degC SCT.

The inter-stage heat exchangers are two Alfa Laval ACH120 plate type units which are mounted on the low stage rack. The heat generated in the low stage discharge is transfused into the high stage suction via a direct expansion valves and operate at a design superheat of 6K.  The design TD of the heat exchangers is 5K.

The condenser is a roof mounted air cooled type Buffalo Trident (model LDV 454B) and is split 50/50 between the two plants. It has been fitted with six 800mm EC-motor (3 per system). The two half’s of the condenser are each dedicated to one or the other plant, so the two systems have exactly the same ambient conditions, and the same heat rejection capacity.


Figure 2 General arrangement



The design temperature of the freezer store is -25deg C.

The control system is configured so only one of the two systems can operate at any one time.

Each individual evaporator has its own control system so if one area of the room achieves its desired  room temp, that evaporator liquid supply is stopped and the other coils run on as before. This will reduce the load on the rack, and as a result the rack controller type Danfoss (model ACCESS 55) allows the compressor capacity to be reduced accordingly.

The condenser fan motors are EC-type motors and are controlled by the rack controller, which monitors the R404A discharge pressure in the systems and increases the fan speed on a rise in pressure.

The two stage system has additional control logic. The high stage rack load is determined by the low stage discharge pressure. On a rise in low stage discharge pressure, the rack controller increases the capacity of the high stage rack so the low stage discharge conditions are maintained within a narrow band.



The cost of the three racks the six heat exchangers and the condenser, where all tracked so a comparison could be drawn.

So the cost to the contractor could be separated from the cost to the end user. The rack system costs were calculated separately. This has been done as the contractor supplies the interconnecting pipe work, plus the pipe insulation, between the various items.

The two racks that make up the cascade system where found to be 19.5% more expensive than the single stage rack.

This higher cost is largely due to the additional safety equipment that the CO2 requires under the Australian / New Zealand occupational heath and safety codes, and the fact that a reasonable amount of the components were special builds, which had to be air freighted from Europe.

As CO2 gains in popularity and more CO2 equipment becomes available this additional cost will be reduced.


When the system heat exchangers, pipe work, insulation, bracketing and so on was added the additional expense was reduced to 9.1%.

The main factors at work here were the large reduction in the size of the pipe work and insulation.

The suction lines for the single stage plant were 3x 2-1/8”copper lines, while the CO2 suction lines were 1x 1-3/8”copper line. Due to the larger diameter pipe work on the single stage R404A system, each evaporator had to be piped back to the rack individually. The smaller line size required by the CO2 system allowed for the three evaporators to be run in a single main suction line.

In addition, the CO2 evaporators were physically smaller, and less expensive, due to the increased specific cooling capacity of the refrigerant. It was found that the R404A evaporators need approximately 20% more surface area to achieve the same performance as the CO2 evaporators, in a direct expansion low temperature application based on the same temperature difference between evaporating temp and room temperature.

The refrigerant in the two systems also have an influence in the total cost. Both systems each had similar total refrigerant charge in them, with the single stage system having 220kg of R404A, and the cascade system having 116kg of R404A and an additional 132kg of CO2, (116+132= 248kg).


The R404A having a cost (at the time of purchase) of AUD$ 22.33 per kg, while the CO2 had a cost of AUD$ 2.40 per kg. The cascade system had a AUD$ 2005.00- advantage over the single stage system. 

The direct global warming potential, or G.W.P. of the two systems, due to direct emissions, in the event of a total loss of the entire refrigerant charge is also of great importance.

Because CO2 is used as the base unit for measuring G.W.P. this comparison is relatively simple. A kg of R404A. has a G.W.P. of 3260 units, while a kg CO2 is equal to 1 unit.


Rack  “A”                220 kg  x  3260                 = 717 200 units

Rack  “B”                116 kg  x  3260                 = 378 160 units +

Rack  “C”                132 kg   x   1                     =        132

Cascade system  B + C                                   =  378292

 717 200  -  378292  =  338 908 units difference between the two systems can be shown.



The plants have now been in operation since November 2004, and both systems have undergone regular service. It is not expected that either of the systems will have a cost advantage as the life expectance for the equipment is identical.

For example the relatively more expensive oil used by the CO2 Compressors is off set by the reduced amount required.

One area where there may be a difference is in the event of a gas loss, the less expensive CO2 used in the cascade system, will be cheaper to replace. While the relatively smaller R404A charge in the cascade system, will also be less expensive to replace. To off set this, it is more likely that the CO2 gas charge be lost due to the potential increase in gas pressure in the event of a total power failure.



The design of the CO2 rack has some unusual features which are required to maintain compressor temperatures at an expectable level. In earlier prototypes, it was found that CO2 compressors suffered from very low operating temperatures, which if left uncheck, would result in a high concentration of refrigerant in the oil within the compressor sump, causing premature compressor failure. Superheats of 25K at the TX valve were required to maintain acceptable sump temperatures in our prototype test freezer.


To prevent this, additional heat exchangers were added between the CO2 suction line and the R404A high stage liquid line. As a result the CO2 suction gas temperature is maintained at the compressor between +5 and +15 deg C.  As a result the CO2 system COP fell by 10% due to the increased suction gas temperature, and subsequent drop in gas density, which resulted in lower mass flow. To balance the equation, liquid sub cooling of the R404A was found to be similar in its net effect. The liquid R404A is sub-cooled from 38degC to around 30degC (depending on flow rates and the number of heat exchangers in operation). The sub cooling effect in the R404A resulted in an improvement in the high stage cooling capacity from 24.9kWQ per compressor to 27.4kWQ and a rise in COP from 2.28 to 2.51, an increase of 10%. To maintain equal room conditions one of the CO2 compressors operating speed was altered by the control system via a variable speed drive which was fitted to provide capacity regulation.


An additional heat exchanger has been fitted to the CO2 suction line as a safety device This heat exchanger is in the form of a coil in shell, and directs part of  the CO2 liquid line (at -10degC) through the coil so that in the event of over heating of the suction line the increased TD between the cold liquid and warm suction will provide greater heat exchange and there fore increased emergency cooling, while the relatively inefficient steel coil provides only slight heat exchange under normal conditions. In the event of a lower than normal suction temperature (below 0degC), the control system injects hot discharge gas. This has been found to be a very rare event.



Each of the two competing refrigeration systems is fitted with watt nods, which are able to capture the total power consumption of the entire system. Power is recorded at 30min intervals for the plants in operation, and includes all aspects of the system. Compressor motors and sump heaters, fan motors defrost heaters, and so on. As the trial is ongoing, and a full year has not elapsed, some assumptions have been made. The first is that spring and autumn share the same ambient conditions, where if fact there are minor variations. The second is that each day in a given season is exactly the same, which is clearly not the case. This has been done as is it not possable to collect data from week to week that is exactly consistent in all areas. Weekends were excluded as the room load is reduced due to the minimal air infiltration load.


X axis  =  Time line for 3 typical days  1 per season

Figure 3 Seasonal power consumption in kWh + freezer & ambient temperatures


As a result the most consistent method is to compare two days (one for each plant) that have a very close temperature profile, and call this a typical day for that season. Then multiply the results by the number of days per season. The graphs in figure 3 show three seasons summer, spring/autumn and winter. In each case the single stage system was found to have used more power. The trend line indicates that the power saving achieved by the cascade system increased as the ambient temperature dropped. This is largely due to the relative capacity of the inter-stage heat exchanger which is selected at 5K TD at full load. At lower ambient temperatures the system requires less capacity to maintain room conditions. At 50% capacity the inter-stage heat exchanger will operate at 2.5K TD. This allows the high stage suction pressure to rise by 2.5K, in turn making the high stage more efficient. The plant can turn down to 16.6% of full load capacity when the load is suitably reduced.








Power Consumption CO2-System [kWh/day]





Power Consumption R404A [kWh/day]





Power Consumption per Season CO2 [kWh]





Power Consumption per Season R404A [kWh]





Difference in Percentage [%]










Power Consumption per Year CO2 [kWh]





Power Consumption per Year R404A [kWh]





Difference in Power Consumption [kWh]





Difference in Percentage [%]






Figure 4 Power usage data by season and total usage


When the seasonal results are averaged, the cascade system was found to be in the order of 20.8% more energy efficient than the conventional system.

It is most likely that a good proportion of the energy savings can be attributed to the sub cooling of the high stage liquid, by the low stage suction gas. The 10% extra energy that was needed to drive the low stage system due to the heat exchange with the R404A liquid line has had a negative effect on the overall system, but  it has been more than offset by the gains which effect the power consumption of the much larger high stage motors.



Given the rapidly changing cost of refrigerants and the expected reduction in the cost of CO2 compatible components, plus the enormous variation in the cost of power around the world, it is not possible to provide the exact pay back period that is required to off set the more expensive cascade system. But it is safe to say that the larger the plant the more attractive CO2  becomes.

In the case detailed here, the payback period for the additional expense of the cascade system over the conventional system was less than 12 months.

This project has proven that superior performance, and a more environmentally friendly process can be applied to reduce the effects of direct and indirect global warming, while achieving long term cost reduction for the plant operator.

Clearly there are numerous advantages which will ensure that carbon dioxide cascade systems have a place in future refrigeration systems.





  1. Australian Government Bureau of Meteorology -
  2. Bitzer International  software version    4.0.1 -
  3. Coolpack software  version  1.46 -
  4. Bitzer Refrigerant Report    12th Edition  ( A-501-12 ) -
  5. BOC  Gases  Australia - 20Charles St   Parramatta  NSW  2150
  6. Danfoss Australia - Unit 4/ 7-11 South St  Rydalmere  NSW  2116

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer