CO2-NH3 Cascade System Technical Paper
CO2-NH3 Cascade System Technical Paper
CO2-NH3 Cascade System Technical Paper
Abstract
Until recently, fishing vessels could use R-22 to preserve the catch, but this refrigerant is no longer an
option in many areas. Ammonia has always been an efficient alternative, but had been perceived as
difficult and expensive to handle on the production deck of fishing vessels. Combining the strengths of
two environmentally friendly and efficient refrigerants, ammonia and CO2, and using standard industrial
refrigeration components, a cascade refrigeration solution was developed for this application. The
paper discusses the technical challenges, solutions, and benefits now and in the future for these types
of systems.
© IIAR 2003 1
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico, Copyright © 2003, IIAR
© IIAR 2003 2
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Introduction
CO2 refrigeration is not a new discipline, neither on land nor on ships. In fact, many
consider it to be one of the first refrigeration applications. The first compressor built
by our company back in the 19th century used CO2, and ships have been equipped
with CO2 refrigeration systems for at least 100 years.
Many things have changed since then. With the introduction of halocarbon
refrigerants, CO2 was largely abandoned. However, with increasing focus on the
negative environmental properties of halocarbon refrigerants, CO2 is being
“discovered” once more. However, re-introducing CO2 in refrigeration designs
involves much more than just finding the old drawings! Standards have changed,
and end users expect all the functionality normally found in state-of-the-art
refrigeration systems to be included with a new CO2 refrigeration plant. Using such a
high-pressure refrigerant posed quite a challenge to us, for example, to find
components strong enough to handle the pressure. However, it also spurred
innovation, because we had to fundamentally rethink our designs, for example, to
devise an effective defrost method.
This paper describes the development and proving of the system built for the
Norwegian-owned MS Kvannøy.
© IIAR 2003 3
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
• Two ammonia systems with the capacity to cool 300 m3 (10,600 ft3) of seawater
from +18°C (64°F) to +2°C (36°F) in five hours.
• An ammonia/CO2 cascade freezing system with 11 vertical plate freezers with
the capacity to process 210 metric tons (463,000 pounds) of fish to -18°C (0°F)
core temperature in 24 hours at an evaporating temperature of -49°C (-56°F).
Because air-cooling plates are installed in three of the nine RSW tanks, these can
also be used as -30°C (-22°F) holds. The cascade system also provides cooling
for a vertical ice machine capable of producing 30 tons of flake ice every 24
hours. All evaporators, except the interstage cooler, use CO2.
The eleven plate freezers are filled manually but empty automatically.
© IIAR 2003 4
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
increasing the temperature difference, we can also increase the heat transfer, which
decreased the freezing time, and thus increased the production capacity of the plate
freezer.
We decided that for a CO2 system, an evaporating temperature of -50°C (-58°F) was
realistic. For this system, freezing time would be about 80% of that in a typical
system operating at -40°C (-40°F) with the same size plate freezer. Thus, unit
production capacity would be about 125% that of the -40°C (-40°F) system—a 25%
increase. We initiated a project to verify this capacity gain, among other things, and
validated this 25% unit capacity increase.
Subsequently, we optimized the plate design specifically for CO2 use. These plates
had smaller channels and, as a consequence, were thinner. Thinner plates enabled us
to fit more plates in the same freezer space. Thus, a 39-compartment (40-plate) CO2
freezer could fit in the same space as a typical 37-compartment freezer, representing
an additional 5% increase in unit production capacity.
In tests aboard the Kvannøy, we found that the freezing time was even quicker than
the predicted 80% of a typical system. Although not verified, the authors’ opinion is
that using CO2 resulted in a lower equivalent temperature drop in the suction line and
plate freezer outlet hose than with other refrigerants. The high vapor density of CO2
yields a much lower volume flow in the return system. The lower volume flow of
CO2 produces a much lower velocity in the return system when compared to other
refrigerants, with substantially lower pressure loss as a result. The lower pressure
loss results in an evaporating temperature in the plate much closer to the low-
pressure receivers saturation temperature than seen before.
© IIAR 2003 5
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
agency inquiries focus on how many tons of fish can be frozen to a given core
temperature in one day, given a certain amount of deck space, rather than on
compressor horsepower or specific equipment.
Given this focus on unit production for each unit of deck space area, it is clear why
CO2 is an extremely promising option for the fishing industry. With the capacity
gained from lowering the evaporating temperature, the CO2 solution is extremely
attractive.
Finally, with CO2, the loss of product due to contamination from refrigerant leaks is
not a problem, as is the case with ammonia. Because the product in the processing
areas and in the cargo hold of large fishing vessels represents such a high value,
reducing risk of product damage results in reducing insurance costs.
Cooling Efficiency
Table 1 shows the coefficient of performance (COP) in BTUh/kW units for a number
of different system configurations and operating conditions. The table values
represent the best efficiencies found using our own computer program. In modeling
the different system options, we assumed that single-stage systems would use screw
compressors without economizing, and that two-stage systems would use
reciprocating compressors with a closed intercooling system. We modeled the
cascade system after the system used on the Kvannøy, which used reciprocating
© IIAR 2003 6
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Evaporating temperatures for brine systems are lower than for direct systems, and
thus have brine systems have lower COPs. Therefore, we did not model them. It is
clear that the ammonia/CO2 cascade system delivers a superior COP in both running
conditions. In fact, the cascade COP at -50°C (-58°F) is substantially equal to the
single stage COP for R-22 and ammonia at -40°C (-40°F) and not far off the two
stage values.
The unique defrost system used on the Kvannøy has an effect on the COP as well.
The defrost compressors’ suction flow does not have to be condensed by the
ammonia circuit, and consequently, the required capacity and power demand of the
ammonia circuit is reduced. Since the defrost compressor is running with a COP
roughly twice as good as the ammonia compressors, every kW (or BTU/hr) used in
the defrost compressor motor results in double that amount saved in the ammonia
compressor motors. The overall effect of running the defrost compressor is naturally
dependent on the system configuration, but on a system like the Kvannøy, the COP
at an evaporating temperature of -50°C (-58°F) can be increased to 2.09 during
defrosting.
The careful reader might object that the act of defrosting adds heat to both the
equipment and product, and that this heat would need to be removed, resulting in a
reduction in COP over time rather than an increase. This is true in most applications,
but not in vertical plate freezers. When the vertical plate freezer finishes a freezing
cycle, the processed product is virtually one big ice block. Under normal operation, it
is necessary to apply heat to the plates to release these ice blocks. As such, the
calculation of total capacity already accounts for the heat of defrosting. If we were
able to use some other heat source to defrost the plates, the COP would be 1.78, as
mentioned earlier, but when that heat is supplied by the defrosting compressor, the
© IIAR 2003 7
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Table 2 presents the total compressor suction volume flow of refrigeration capacity
at the running conditions given above and with the same compressor constellation.
All values are stated relative to the ammonia/CO2 cascade system operating at -
50°C/+25°C (-58°F/77°F) conditions. To be able to make an “apples to apples”
comparison of the installed swept volume, it was necessary to “scale” the systems so
that they delivered the same cooling capacity. This was not possible to achieve
exactly with the compressor sizes available. Thus, scaling the systems to achieve the
same cooling capacity required using a “fractional” compressor (e.g., using 2.7
compressors). The reader will note from Table 2 that the total compressor suction
volume required for the CO2/ammonia cascade is much less than for any other
system.
Product Quality
Product quality is all-important for getting a good price. Of the factors contributing
to better quality, shorter freezing time and lower end temperature are the most
important ones that relate to the refrigeration system. Shorter freezing time results in
smaller ice crystals and less cell damage to the product, and a lower end temperature
reduces the amount of free water in the product, resulting in a longer shelf life. Both
a shorter freezing time and a lower end temperature can be achieved with CO2,
because the lower evaporating temperature yields a lower temperature at the plates,
and the average temperature is therefore lower, too.
Very often, blocks frozen in typical vertical plate freezers are soft on top, usually
because the upper part of the plate can become very hot during defrost. Any product
softness increases the chance that the product will suffer damage from handling, and
such a loss in quality is deducted from the sales price. The combination of a low
© IIAR 2003 8
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
If the following factors are taken into consideration, the investment in CO2 can be
very attractive:
• Saved production area or more production capacity using the same area
• Improved freezing time (both theoretically and in practice)
• Improved quality of frozen products
• Better COP and thus, lower energy cost
© IIAR 2003 9
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Environment
To illustrate the impact of using CO2 in lieu of a more typical system, consider the
result of hypothetically replacing the existing refrigeration system in two of the
largest super trawlers that have vertical plate freezers with a CO2/ammonia cascade
freezing system of similar capacity:
• For an R-22 system, such as in the “Wilhelm Van Der Zwan”: the 55-tonne
(121,000-pound) R-22 refrigerant charge would be replaced with a 12-tonne
(26,000-pound) charge of CO2 and a 1-tonne (2,200-pound) charge of ammonia
• For an ammonia-brine system, such as in the “Atlantic Dawn”: the 5-tonne
(11,000-pound) ammonia charge and 100 tonnes (220,000 pounds) of calcium
chloride would be replaced by a 12.5-tonne (27,600-pound) charge of CO2 and a
1-tonne (2,200-pound) charge of ammonia
Since the Atlantic Dawn would no longer have to bear the weight of a brine system,
there might be an additional advantage in reduced fuel usage. When we compare the
weight of the ammonia/CO2 system versus the ammonia/brine system, the
machinery, vessels and refrigerant more or less would weigh the same for the two
© IIAR 2003 10
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
systems, but the big difference is the 100 tons of brine. With information provided
by ship design consultants Vik & Sandvik in Norway, we can estimate the difference
in fuel usage using the following assumptions:
• A large fishing vessel will average 16 knots every 24 hours for 100 days a year.
• The energy required each hour to transport 100 tonnes (220,000 pounds) at this
speed is approx 50 kWh (170,000 BTU), equal to the energy from burning 13
liters (3.43 gallons) of fuel.
For land-based installations, many laws around the world limit the quantity of
ammonia allowed in refrigeration installations, especially in densely populated areas.
This trend is showing in the marine sector as well. This makes it very difficult to
make large efficient freezing production installations with ammonia. The
ammonia/CO2 cascade system seems to be a very attractive alternative, especially
when it can be designed to perform with the same or better efficiency than a
traditional ammonia installation, and made with standard refrigeration components,
and with the same features as typical halocarbon installations (e.g., oil rectifiers,
miscible oil, hot gas defrost)
© IIAR 2003 11
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
In the system described herein, ammonia is used as the high temperature refrigerant,
but halocarbons can also be used, if there is a specific demand to avoid the use of
ammonia, and where local laws do not prohibit their use.
The CO2 refrigeration circuit controls restrict its operation to a set pressure range to
permit the use of standard industrial refrigeration components. Thus, the system can
be operated, maintained, serviced, and understood by the existing staff of operating
personnel and service engineers. Further, spare parts are readily available with short
delivery times anywhere in the world, and can be installed by local service staff. The
system piping design confines the ammonia to the machinery room, where safety
precautions are relatively easy to implement. CO2 is fed to production areas, cold
stores, freezing tunnels, because it presents less risk to both people and products.
For managing potential ammonia releases, CO2 can be mixed with ammonia, which
quickly react with each other to form ammonium carbamate. In the Danish Fire-
fighting School in Esbjerg, part of the training for handling an ammonia release
includes using CO2. It would be very convenient to do this in the machinery room,
where CO2 is available in good quantities. If this is done, personnel must take proper
precautions to avoid exposure to high concentrations of ammonia, and prolonged
exposure to the CO2. Because it is a heavy gas, CO2 displaces air in poorly
ventilated spaces, which lowers the oxygen concentration. The American
Conference of Governmental Industrial Hygienists has noted that 30-minute
exposures to concentrations of CO2 of 5% or greater have resulted in intoxication.
(ACGIH, 1971) The US Navy permits emergency exposures to this concentration of
up to one hour. (US Navy, 1962) A concentration of 10% or higher can cause
unconsciousness and eventual death by asphyxiation over the course of hours of
exposure. (OSHA, 1989)
© IIAR 2003 12
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Evaporators
Due to the high saturation pressure of CO2, the vapor density is also high. At -40°C
(-40°F) the saturated vapor density of CO2 is 26.2 kg/m3 (1.636 lb/ft3), while for
ammonia and R-22 it is 0.6447 kg/m3 (0.04 lb/ft3) and 4.9 kg/m3 (0.306 lb/ft3)
respectively. While this has a positive effect on compressor performance, it can, if
not taken into consideration, have an adverse effect on evaporator performance. As
an example, consider an evaporator operating at a temperature of -40°C (-40°F) that
produces a given amount of refrigeration. If we calculate the exit velocity of the
suction vapor using different refrigerants, we would note that the value for CO2 is
nearly an order of magnitude lower than both ammonia and R-22, for circulation
rates of 1 to 8. At -50°C (-58°F) the differences in exit velocity become even more
pronounced. While this property of CO2 naturally leads to lower pressure loss in the
suction line and evaporator, the flow is likely to be a “lazy river” in the bottom of the
evaporator channel. Insufficient agitation of the liquid leads to inadequate wetting of
the channel wall and, as a consequence, a poor heat transfer coefficient.
As discussed earlier, the solution to this problem for CO2 is to use smaller
channels/pipes and longer circuits. As CO2 has a much better saturation pressure–
temperature relation than the above-mentioned refrigerants, it is entirely possible to
find a solution that yields a good heat transfer while keeping the pressure loss and
equivalent temperature penalty within acceptable limits.
From the outset, our CO2 refrigeration efforts have focused on fisheries, and fisheries
commonly use vertical plate freezers (VPFs). However, the usual suppliers of plate
freezers were not in a position to examine the flow and thermodynamic behavior of
CO2 within their freezers. Thus, we needed to take the initiative to develop a
program together with a plate freezer supplier if we wanted to obtain an optimized
product.
© IIAR 2003 13
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Since the product side dominates the heat transfer in a VPF, we reasoned that the
heat transfer coefficient on the refrigerant side had little influence unless it was
extremely low. Thus, our design work focused on creating a plate with a channel that
gave an acceptable flow pattern and pressure loss throughout most of the freezing
period. The concept was tested in our lab, where the performance was verified.
Defrosting
A special design for defrosting CO2 vertical plate freezers systems with hot gas is
essential, where no other means of defrost is possible. Figure 1 shows a schematic of
the defrost system we developed. The system consists mainly of a special version of
a standard high-pressure compressor suited for 45-bar (653 psi) use and a hot gas
bypass valve. The special compressor is connected to the discharge side of the CO2
cooling compressors through a desuperheater, enabling it to produce hot gas at 45
bar (653 psi) corresponding to a condensing temperature of +10°C (50°F).
© IIAR 2003 14
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
• The system saves energy when defrosting, as the COP of the defrost compressor
is roughly double that of the ammonia compressors. This means, in fact, that if it
is possible to utilize the savings on the ammonia compressors by reducing their
capacity or stopping the compressors, we will save 200 kW (682 MBH) on the
ammonia compressors each time we use 100 kW (341 MBH) on the defrost
compressor. This creates energy saving possibilities when cold water or other
means of condensing at +10°C (+50°F) are present.
• The design offers a large defrost capacity (approximately 800 to 1000 kW
heating capacity (2,729 to 3,412 MBH) from this small reciprocating compressor.
We learned a lot from our experience with the system on the Kvannøy:
• The design defrosts the whole plate, hose connections and manifolds very well.
• The plates do not get hot at the top while they are still frozen at the bottom, as is
often seen on typical systems, which increases the quality of the frozen blocks.
• It can be difficult to release the frozen blocks from evaporator parts that do not
take an active part in the defrost, e.g., end plates, bottom plates, so heat must be
actively transferred to these areas by metallic contact with the defrosting plates.
• The system offers a tremendously fast defrost, as long as there the CO2 cooling
compressors are sufficiently loaded. This is similar to a conventional system,
where we need to have a minimum cooling loads to generate sufficient hot gas
for a decent defrost.
• The system demands a more complicated and very fast-reacting regulating and
control system to work well. Both the defrost compressor and ammonia system
must be regulated and the ammonia compressors must be able to decrease
capacity very quickly, as the defrost compressor “steals” approximately 800 kW
(2,729 MBH) of capacity from them in a matter of seconds.
• After adjusting the regulator, the system works very well and performs an even
better defrost than conventional systems.
© IIAR 2003 15
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Capacity Regulation
Each step in capacity for CO2 compressors produces very large changes in cooling
capacity for the system. At -50°C (-58°F) for a given capacity, the swept volume
ratio between CO2 and ammonia is about 1 to 35. This means that at -50°C (-58°F)
suction temperature, a unit increase in volumetric capacity on the CO2 compressors
yields approximately 35 times more additional capacity than a similar increase in an
ammonia system. Thus, capacity and system regulation for CO2 systems must be
capable of reacting much faster than those used with ammonia systems.
The need to handle large and rapid changes in capacity is even more important on the
ammonia side of the system when the CO2 hot gas defrost system is used. The
changes are so rapid when the defrost compressor starts up, that a hot gas bypass
system on the ammonia side is necessary to keep pressures and temperatures in the
system balanced while the ammonia compressors are reducing their capacity. Even
with the variable speed drives on the ammonia screws, the hot gas bypass feature is
necessary. Reciprocating compressors on the ammonia side will have a clear
© IIAR 2003 16
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
advantage, with their capability of cutting in and out all the capacity in a matter of
seconds and even running fully unloaded with a minimum of power consumption for
some time. This can turn out to be a useful feature on ammonia reciprocating
compressors in these types of systems. The clear advantage of the screw compressors
on the ammonia side is the large sizes of the available compressors to match the
capacity needed.
To illustrate how fast the changes in the system are during normal operation,
consider the Kvannøy system. Two plate freezers are filled with whole round
herring at approximately +1°C (34°F) and started nearly at the same time. The initial
load on the freezer is huge, so three reciprocating compressors start up to handle the
load. After 10 minutes, though, only one compressor remains running at 25%
capacity at -50°C (-58°F) suction temperature. The time required to pull the core
temperature of the fish blocks down to -25°C (-13°F) was 1 hour and 45 minutes.
Our experience with the system on the Kvannøy shows that compressor controls,
system control strategies, and everything else, have to be adapted to this new type of
system. We cannot do just as we used to do on conventional systems, and must
rethink our control systems and make them fit to the behavior of these systems.
As the defrost compressor has a maximum differential pressure of 25 bar (363 psi), a
maximum suction pressure of 25 bar (363 psi), and must be able to deliver at hot gas
defrost pressure of 45 bar (653 psi), we need to control the system such that the
compressor’s operation is restricted to its allowed working conditions. This is done
by combining a quick-reacting hot gas bypass valve with a suction pressure
regulating valve for the defrost compressor. The challenge is to maintain the correct
working conditions for the defrost compressor when the compressor begins a defrost
cycle, and within seconds after removing approximately 800 kW (2,729 MBH) of
capacity from the cascade interstage heat exchanger. This demands an immediate
© IIAR 2003 17
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
removal of capacity from the ammonia compressors to keep the CO2 pressure in the
cascade interstage heat exchanger above 20 bar (290 psi). This is necessary for the
defrost compressors to be capable of producing the 45 bar (653 psi) defrost pressure
and continue to work within the 25 bar (363 psi) differential pressure allowed.
If the CO2 cooling compressors are providing insufficient capacity to feed the defrost
compressor, the CO2 pressure in the interstage heat exchanger could drop below the
required 20 bar (290 psi). To avoid this, we also use the defrost compressors’
capacity regulation and the hot gas bypass valve to keep the compressors’
differential pressure under control. This feature secures the balance of the system.
When the load on the CO2 cooling compressors is insufficient, the defrost takes a
longer time, just as with conventional systems.
When too little cooling load is available, it means that not all plate freezers are
running, and thus the time needed to defrost, empty and fill the plate freezer is of less
importance. With sufficient cooling load on the plate freezers, the defrost compressor
will give us a defrost capacity of approximately 800 to 1000 kW (2,729 to 3,412
MBH), which is sufficient to do a complete defrost on the plates, hoses and
manifolds, within the maximum 4-5 minutes desired by the fishermen.
We faced a problem with variations in the defrost time, but this had nothing to do
with the ammonia/CO2 system. The problem was in communicating to the automatic
emptying robot what the defrost time will be, so the automatic emptying process can
begin at the proper time. The defrost time needs to be the shortest possible time
when all plate freezers are in use to ensure the maximum production capacity.
However when insufficient suction gas is available to the defrost compressor, the
compressor regulation reduces capacity to keep the compressor within its permitted
operating range. With the capacity reduction the maximum defrost capacity cannot
be achieved thus increasing the defrosting time. The reason for insufficient suction
gas for the defrost compressor is that there are not enough fish to fill all freezers and
© IIAR 2003 18
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
thus the overall cooling capacity is low. In this situation the longer defrost time has
no effect on production capacity since other freezers are available. The necessary
defrost time for a plate freezer was thus signaled to the robot by the load on the
cooling compressors, and it could then begin emptying the freezer at the appropriate
time.
As discussed under capacity regulation, the changes in pressures and capacity on the
system are much faster than in conventional systems because of the high efficiency
on the evaporators and the high volumetric efficiency on the compressors. This can
make it difficult to control the pressure very accurately with the compressors, as it
will be necessary if we want to run a system close to -55°C (-67°F) or even just -
50°C to -52°C (-58°F to -62°F) without unintentionally risking the creation of dry
ice, which happens at -56.6°C (-69.9°F). We prevent this occurrence with the very
quick-reacting electronically controlled hot gas bypass that will not allow the
pressure to drop under a -55°C (-67°F) limit put into the system. This feature ensures
we will not run into problems with dry ice in the system, no matter how the operating
personnel should run the system.
System Lubricant
The lubricant used in the CO2 system is a polyol ester (POE), and is fully miscible
with the CO2 over the whole temperature range in this application. POE has been
extensively tested and approved by the compressor manufacturer. Thus, we were
able to design the piping and oil return system in nearly the same way as with a
typical R-22 system, i.e., with an oil rectifier. Without this feature, it would be much
more difficult and expensive to install a CO2 system in a marine application. The
limited available space for large oil separators, and the fact that a fishing vessel is
not always stationary, but is often subject to wave motion, makes return of
immiscible oil very problematic. Also, because POE is not as dense as liquid CO2, it
would float on top of the CO2 liquid, making oil draining or recovery very difficult.
© IIAR 2003 19
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
We were unable to measure any negative effect on the evaporators’ performance due
to the presence of small amounts of oil in the refrigerant. As mentioned before, the
freezing time was even better than anticipated.
The presence of oil in the CO2 may actually turn out to be an advantage, lubricating
components such as valves and regulators, but this has yet to be proven. We do know
that the strong cleaning and poor lubricating characteristics of CO2 can cause
problems in oil free systems, such as valves and threads seizing.
This solves another problem for us, because we can use PAO, which is not
hygroscopic, for lubricating seals, gaskets, threads, etc. during construction and
assembly of the CO2 system.
Proper cleaning of pipes, vessels, and components seems to be even more important
for CO2 systems than for typical ammonia systems. The cleaning properties of the
CO2 will remove dirt and particles very quickly and transport these contaminants to
the filters and compressor. We observed that the lubricant initially charged into the
compressors seems to very quickly become discolored by dirt particles, grease, or oil
from the valves, compressors, and other components. However, it does not pose a
problem for the compressors as long as the lubricant is properly filtered, as it is on
compressors today.
© IIAR 2003 20
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Stand-alone Units
Since the saturation pressure of CO2 at even moderate ambient temperatures is too
high for convenient storage, a stand-alone unit is employed to keep the pressure
down in the low-pressure receiver. With intelligent management of the system, a
complete drainage of evaporators, etc., can be achieved, reducing the stand-alone
pressure problem to cooling of the low-pressure receiver.
In the Kvannøy project, we considered an additional task for the stand-alone unit. If
properly insulated, the transmission heat load in the cargo holds should be relatively
small. Since the minimum capacity of a CO2 refrigeration compressor is fairly large,
it is very likely that the compressor will experience many start/stop situations, with
excessive wear as a result. Usually the cargo holds do not need a –50°C (-58°F)
temperature, so it was suggested that we transfer the hold-keeping capacity to one or
two stand-alone units during periods where the cargo holds were the only duty for
the system. This equipment uses small, relatively inexpensive, standard air-cooled
condensing units. Naturally, the greatest benefit is for small cargo holds, such as in
our case.
© IIAR 2003 21
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
We chose reciprocating compressors for this application because they can operate at
the low capacities often required, and offer excellent COPs at both full and part load.
If we had chosen a screw compressor, its minimum capacity of 15-20 % would be
too much for low load situations, and would require a hot gas bypass to run most of
the time, even if fitted with a variable speed drive.
Three of the six reciprocating compressors are specially modified with a maximum
discharge pressure of 45 bar (653 psi), which enables each of them to serve as the
defrost compressor. Since only one of these will act as the defrost compressor at a
time, it means there are always two compressors in the system that can be used as a
backup unit during service, or if a problem should occur. This redundancy ensures a
high degree of reliability in our system, and provides a service and spare part benefit
to a ship that depends completely on the functionality of the refrigeration system.
The CO2 condensing circuit is also built with standard refrigeration components,
with the only demand being that they have a MOP of 40 bar (580 psi). This is not a
problem today, as most common stop, regulating and safety valves meet this
standard.
In this circuit, the 50 bar (725 psi) MOP makes it more difficult to use standard
refrigeration products. Stop valves and safety relief valves are available with the
higher pressure rating as the only difference. Solenoid valves are not available, but
pneumatic or motor-activated ball valves widely used in the refrigeration industry
are, and have no problem in handling both CO2 and the 50 bar (725 psi) pressure.
Normal refrigeration regulating valves are a problem, but with the special defrost
system used, the pressure setting and regulation are managed by the defrost
© IIAR 2003 22
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
compressor and its bypass valve system, so no further pressure regulation devices are
needed to keep the pressure in the right range for the defrost.
As the compressor manages the pressure, we only need to worry about what is
leaving the evaporator. If it is liquid, we have delivered the latent heat of the
refrigerant. If it is vapor, we must close the line to make sure that the vapor deliver
its latent heat. This calls for a float valve, but standard refrigerating valves with
MOP of 50 bar (725 psi) are not available. Other available float valves with the
necessary capacity for defrosting production equipment in a sufficiently short time,
will be extremely expensive, difficult to get with short notice and therefore not the
kind of product we want to have in our system. Luckily, the steam industry has run
into the same problem and has developed an elegant solution: float and thermostatic
steam traps. Their only problem is their limited capacity, but using several of these in
parallel gives a perfect solution to the problem with a suitable regulation
performance.
Another great advantage with the defrost compressor system is that the only vessel
necessary in the MOP 50 bar (725 psi) system is the defrost compressor oil separator.
The MOP of the R-404A high side is 28 bar (~400 psi). The MOP in this circuit is
determined by the refrigerant’s properties, but R-404A was a good choice for the
task. Condensing units from the commercial refrigeration area for R-404A with the
needed capacity are easily available, and easy to service.
Ammonia Circuit
All components in this circuit are normal industrial refrigeration components for
ammonia, and are rated at an MOP of 24 bar (~350 psi).
Approvals
© IIAR 2003 23
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
DNV has produced some preliminary guidelines that have been sent to some relevant
industrial refrigeration manufacturers and contractors. In these guidelines, the rules
for the CO2 part of the system are much gentler than those that apply to the ammonia
part of the system, in spite of the much higher operating pressure on the CO2 parts of
the system.
The alarm and warning levels for ammonia and/or CO2 concentrations in the areas
around the refrigerating system are, on this first ship, specially approved by DNV
and to be set at the following levels.
CO2
Ammonia
© IIAR 2003 24
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
• Be careful not to vent CO2 to an area where the concentration can build to an
asphyxiating level. NOTE: CO2 is heavier than air so its concentration will be
greatest in low areas, e.g., at the floor, in the basement.
• When venting refrigerant before evacuating a system or part of a system, take
care to release only the vapor phase, because any liquid that comes out will form
dry ice, which could block the release valve, line or hose. This could lead to an
incorrect observation that the system pressure has equalized with atmospheric
pressure. If the part of the system is subsequently dismantled where pressure is
still present, serious injury, or worse, could result.
• Changing or cleaning filters is very easy. If a small amount of liquid is still
present in the filter housing when it is equalized to atmospheric pressure, only a
little dry ice will form in the filter. This dry ice can easily be removed, and the
filter changed or cleaned, refitted and put back into operation.
• Properly handled, dry ice will not present a problem, and can actually be an
advantage. Unlike “water ice,” dry ice has a higher density (i.e., is more
compact) than the liquid phase, so it will contract rather than expand. Thus, the
creation of dry ice will not damage pipes, filters, vessels, or other components.
This characteristic can be very useful in service work. If a liquid-containing pipe
or vessel needs to be opened for service or other reasons, it can be vented to
atmospheric pressure and kept vented, and the liquid turned to dry ice. The
service work can be completed and the system closed again. When the pressure is
subsequently raised above CO2’s triple point, the dry ice will disappear and
change back to liquid. Not much refrigerant would be lost, and it makes it
© IIAR 2003 25
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
possible to service CO2 systems in situations where it would be very difficult and
expensive to service a conventional system.
• In low-pressure receivers, the high density of dry ice must be considered. Dry
ice will sink below the liquid, and could block the liquid inlet to the pumps, or
even mechanically damage the pumps. This is a good reason for paying close
attention to the minimum capacity on the cooling CO2 compressors, to ensure
that the low-pressure receiver pressure is always kept above the triple point.
• When CO2 systems are built or opened for service it is essential that proper
vacuum procedures are followed to get all moisture out from the system, to avoid
the serious problems that can occur in connection with water contamination of
CO2 systems.
We always assume that CO2 would enter the ammonia system in the event of a leak,
because the pressure on the CO2 side of the interstage heat exchanger is much higher
than on the ammonia side.
Approximately 0.03 % of the natural atmosphere is CO2. Thus, CO2 inevitably finds
its way into ammonia plants, especially low temperature plants operating below
atmospheric pressure. Over time, quite a large quantity of CO2 can be sucked into
© IIAR 2003 26
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
these systems through leaks. This phenomenon plays an active role in creating some
of the sludge found in leaking ammonia systems.
By using a shell and tube interstage heat exchanger, we can repair a leak on the spot.
This is very important on a ship, where the removal and exchange of such a large
piece of equipment would be extremely difficult, time consuming, and expensive. On
our system, we used a shell and tube interstage heat exchanger with double endplates
that vents to atmosphere between these plates. But on future systems we do not
regard this as a precaution we need to take. We made contact with a Danish company
who has delivered equipment for production of CO2 for more than 30 years. In these
systems, shell and tube ammonia evaporators from our company with single
endplates have been used for condensing CO2 without having any problems with
leaks between CO2 and ammonia.
Water Contamination
CO2 refrigerant is sensitive to water contamination. Liquid phase CO2 can dissolve
much more water than the gas phase at the same temperature and pressure, resulting
in the same problem we have with R-22: freezing up of expansion valves.
© IIAR 2003 27
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Fortunately, the acceptable limits for water content in a fresh CO2 charge are very
low, which avoids these problems.
The water and CO2 solution is actually carbonic acid. Carbonic acid promotes
galvanic corrosion, and will attack materials inside the system. We have heard
reports of heavy acid attacks on steel pipes and on screw compressor ball bearings in
old CO2 systems with water contamination. Good evacuation procedures will keep
the systems free of water, and must be followed before charging CO2 into the
system, and also when it is opened for service and maintenance.
Filter-driers used for R-22, R-404A, R-507 etc. can also be used for CO2 as well, and
have shown to be very effective. We suggest mounting filter-driers in a bypass line
for the cold liquid line where they will be most efficient.
We use a model of sightglass that has a moisture indicator, that has been tested by
the manufacture with CO2, and that has proven to work well. The sightglass needs to
be selected to match the maximum expected water content of the CO2 at the
temperature and pressure where the indicator is placed in the system.
© IIAR 2003 28
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
CO2’s strong cleaning properties and low liquid viscosity make it a poor lubricant.
This should be considered when selecting components such as pumps and valves.
We use miscible oil and do not expect lubrication problems. So far, our experience is
very good, and no problems have been reported.
When designing systems with CO2, special care must be taken to prevent liquid from
being trapped in the system. If these parts of the system heat up, the resulting
hydrostatic pressure can easily exceed the MOP. A lockout, tag-out procedure for
service valves can be a good precaution.
On the CO2 side, it may be cheapest to change the complete system. Production
equipment like plate freezers and evaporators that need to handle hot gas defrost at
50 bar (725 psi) must be changed to specially designed evaporators for CO2 use
and/or specially approved for 50 bar (725 psi) MOP.
Valves in the existing system in many cases will not provide the necessary pressure
rating. Additionally, liquid lines must be slightly larger. The suction pipes will not
© IIAR 2003 29
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
be correctly sized to return oil back to the compressors, as the velocity will be too
low to lift two-phase flow in risers. The oversized suction lines and evaporators will
require an unnecessarily large CO2 charge. A larger charge will require a larger low-
pressure receiver and larger pumps. In our system, the low-pressure receiver must
also be equipped with a cooling coil for the stand-alone unit, so a new low-pressure
receiver specially designed for CO2 use is necessary.
By changing the whole system on the CO2 side we avoid problems related to mixing
old oil, sludge, and other contaminants from the old system with CO2. Problems of
this nature need to be considered, which we learned by accident in our research and
development test center. Old oil that we had been unable to remove formed a thick
sludge that clogged filters and orifices in our test rig.
© IIAR 2003 30
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
• The price of the “new refrigerants” such as R-134a, R-404A, R-507, R-410A,
etc., both now and in the future, speaks for itself, when considerable charge is
needed.
© IIAR 2003 31
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
The experience we gained during the design of the Kvannøy showed that it was not a
problem to get components that could handle pressures above what is normally
encountered in the refrigeration industry, but instead, to get them at a price where the
solution could still be economically competitive. Very often, we found that we had
crossed the line into the domain of products used in the chemical industry, with an
enormous price increase as a result.
The solution was in many cases to ‘push’ the traditional suppliers to tell us why their
components could not handle the increased pressures. Most would admit that while
most components could physically handle the pressure, the original classification
was performed at a pressure, which, at the time, was deemed to be sufficient for the
refrigeration industry. Individual pressure tests of these components ensured their
fitness for use in our CO2 system. We benefited from being a major consumer for
many suppliers, and were able to make the project happen as a result. It is our
opinion, however, that as more of the industry fully grasps the importance of CO2,
this problems will go away. It seems that for regulating equipment (e.g., stop valves),
the “magic line” is at 40 bar (580 psi), and 50 bar (725 psi) for non-regulating
equipment.
Careful design of the system to limit the number of components exposed to high
pressures greatly simplifies the task of sourcing the correct components. The system
we developed uses a minimum of regulating equipment in the high pressure defrost
cycle.
© IIAR 2003 32
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
Conclusion
CO2 is no longer the refrigerant of the future; it is a refrigerant of the present. This
project demonstrated that it is possible to make CO2 systems with all the features we
normally expect in a traditional system, and most of these features are realized in the
way we are used to doing things. Furthermore, with proper design, the system can
use components that are standard throughout the refrigeration industry.
In a very short time CO2 has become a most attractive refrigerant when it comes to
freezing applications. Once we had realized that the apparent problems with CO2
refrigeration could easily be solved, we realized its substantial benefits:
With the ammonia/CO2 cascade now being a viable option, there is no longer any
reason for large halocarbon systems and there is no longer a reason to have ammonia
in the production areas. The ammonia/CO2 cascade can do the job more efficiently,
quicker, and safer – both to the worker and to the planet.
© IIAR 2003 33
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
References
ACGIH. Carbon dioxide. In: Documentation of the threshold limit values for
substances in workroom air. 3rd ed. Cincinnati, Ohio. American Conference of
Governmental Industrial Hygienists. 1971.
OSHA. Carbon Dioxide, Industrial Exposure and Control Technologies for OSHA
Regulated Hazardous Substances, Volume I of II, Substance A - I. Occupational
Safety and Health Administration. Washington, DC: U.S. Department of Labor.
1989.
© IIAR 2003 34
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
© IIAR 2003 35
2003 IIAR ammonia Refrigeration Conference, Albuquerque, New Mexico
© IIAR 2003 36