Gould, Fluor, Aliso Viejo, California; and R. JOYNER and A. DESHMUKH, Fluor, London,
As the world continues along its energy transition trajectory, demand for blue
hydrogen (H2) and reduced carbon emissions will make high-purity
carbon dioxide (CO2) streams available in many process plants. This article
investigates how these streams could be utilized to add value to projects.
1 of this article explores the potential use of supercritical CO2 as
a refrigerant, replacing conventional refrigerants using hydrocarbons or hydrofluorocarbons
(HFCs) to provide chilling duty on large energy and chemicals projects. The global
warming potential (GWP) of CO2 is significantly lower than commonly
used HFC refrigerants; therefore, the use of CO2 as a replacement
for many HFC refrigerants would present a reduction in greenhouse gas (GHG) emissions
when measured in equivalent mass of CO2.
article considers how the operating conditions of CO2 refrigeration
systems can be optimized to minimize energy consumption and improve efficiency.
It also discusses the use of process simulation tools to compare a CO2
refrigerant system with a conventional propane refrigeration system regarding
capital and operating costs (CAPEX/OPEX), operability, safety, and the potential
for integration with carbon capture and sequestration (CCS) technology.
The history of CO2 as a refrigerant. The
use of CO2 as a refrigerant dates back to the 19th century, with the
first patent for a refrigerator using R-744 (CO2) being granted in
1850.1 The popularity of CO2 as a refrigerant increased
in the early 20th century due to its lower flammability and toxicity vs. the
two most common alternative refrigerants at the time: ammonia (NH3)
and sulfur dioxide (SO2). By the 1930s, 80% of marine vessels used CO2
for refrigeration and comfort air cooling,1 and it was used
universally by British ships into the 1940s.2 However, following the
introduction of refrigerants using chlorofluorocarbon (CFC) in the early 1930s,
CO2 began to be phased out and replaced because of the high versatility
and reduced system complexity offered by these new synthetic refrigerants. By
the 1950s, synthetic refrigerants had come to dominate industrial refrigeration
interest in the applications of CO2 as a refrigerant has persisted,
particularly following the phase-out of ozone-depleting CFC refrigerants in the
1980s and 1990s. CO2 has been effectively applied in automotive air
conditioning and commercial refrigeration systems,3 and its use in
the low-temperature stage of industrial cascade refrigeration systems was
common by the 2000s.4
further restrictions on refrigerants with a high GWP came into effect (e.g., the
2014 EU F-Gas Regulation5), the need for alternative refrigerants
intensified. In 2008, there were fewer than 150 transcritical CO2
refrigeration systems across Europe. By 2020, there were more than 29,000,
predominantly in commercial refrigeration applications.6 However, CO2
refrigerant has yet to take hold in large energy and chemicals plants,
where the use of synthetic refrigerants or hydrocarbons, such as propane and
butane, is still most common.
is forecast to contribute approximately 78% of the equivalent GHG emissions
between 2013 and 2050.7 However, the high impact of CO2 is
predominantly due to its abundance, rather than its potency, as a GHG. CO2’s
GWP is significantly lower than commonly used HFC refrigerants, as shown in TABLE 1. Overall, HFC
emissions are increasing at a rate of 8%/yr, and emissions are projected to
rise to 7%/yr–19%/yr of global equivalent CO2 emissions by 2050.8
In the Kigali Amendment to the Montreal Protocol, 197 countries committed to
cut the production and consumption of HFCs by more than 80% over the next 30 yr.9
The use of CO2 as a replacement for any of the HFC refrigerants
listed in TABLE 1
would therefore present a significant reduction in GHG emissions when measured
in terms of GWP, even if the CO2 used was eventually released into the
GWP of alternative hydrocarbon refrigerants (e.g., propane and butane) is lower
than CO2; therefore, to realize the benefits in terms of reducing
GHG emissions, the CO2 used as a refrigerant would have to be
eventually sequestered or otherwise utilized in a manner that did not result in
it being released into the atmosphere. As this article will detail, the
potential exists to integrate a CO2 refrigeration system with CCS
CO2 refrigerant system. The
basic refrigeration system considered in this article is a simple closed-loop refrigeration
cycle (FIG. 1).
This configuration is fundamentally the same, regardless of which refrigerant
is used. The principal difference is that, with a conventional refrigerant such
as propane, the compressor aftercooler is a total condenser; whereas in the CO2
configuration, it is a cooler with no phase change. This will be
discussed further in the following section.
refrigerant is a saturated vapor at the suction of the compressor. The
compressor increases the pressure of the refrigerant, which is then condensed
or cooled and routed to the accumulator. The accumulator acts as a buffer
vessel before the refrigerant moves to the chiller. The cool refrigerant from
the accumulator is flashed through the Joule-Thomson (J-T) valve to a low
pressure, resulting in a cold two-phase stream, which is routed to the chiller.
The liquid portion of the refrigerant from the J-T valve is fully vaporized in
the chiller, and the resulting vapor is routed to the suction of the compressor
via a knockout drum, completing the cycle. The chilling duty is provided by the
latent heat of the liquid refrigerant as it vaporizes in the chiller. The gas
portion has minimal impact on the heat duty.
The impact of operating pressure. In
a conventional, sub-critical refrigeration loop, the compressor’s discharge
pressure is determined by the vapor pressure of the refrigerant at the
condensing temperature, typically based on the minimum approach to the cooling
medium. The refrigerant would therefore be a saturated liquid when it is
supplied to the J-T valve. The refrigerant may be a subcooled liquid at the J-T
valve inlet should the system include an economizer.
the critical temperature of CO2 is 31.1°C. Consequently, liquid CO2
cannot be formed above this temperature regardless of the pressure. Many plants
utilize atmospheric air as a cooling medium. Given that an approach in the region
of 10°C to the maximum design ambient air temperature is typically required,
this means that condensing CO2 at 31°C is infeasible by air cooling
in all but the coldest of climates. As the formation of a liquid CO2
refrigerant stream is infeasible, the alternative is to produce a supercritical
CO2 stream at conditions such that a low-temperature, two-phase
stream is produced when it is flashed through the J-T valve.
that the CO2 stream will be supercritical at the compressor
discharge conditions, it will not condense through the compressor aftercooler,
as there is no observable phase change between liquid and supercritical or gas
and supercritical phases. Without the limitation of operating at the vapor
pressure of the refrigerant, there is a greater degree of freedom in the selection
of the compressor discharge pressure. This prompts the following question: Is
there an optimum pressure at which to operate the discharge of the compressor?
FIG. 2 shows
a pressure-enthalpy diagram of CO2. As an example, consider a plant
that requires a process stream to be chilled to 5°C—and thus a CO2 refrigerant
stream at 0°C—to allow a reasonable temperature approach in the chiller. Air
cooling is available with a design ambient air temperature of 35°C. A refrigerant-side
outlet temperature of 45°C from the air cooler is therefore considered. The CO2
refrigerant will leave the chiller shell as a saturated vapor. The boiling
refrigerant temperature will be maintained by controlling the compressor’s
suction pressure. Conditions in the shell are plotted in FIG. 2 at the
intersection of the dewpoint curve with the 0°C isotherm, which occurs at 33.5
compressor discharge conditions can be found through process simulation. The
line representing the path through the compressor in FIG. 2 corresponds to a compressor
adiabatic efficiency of 75%. The gradient of the line is a function of the
compressor’s efficiency. The steeper the line, the higher the efficiency, thus,
the lower the discharge temperature. Considering constant efficiency, any
compressor discharge pressure will fall on this line. Four points have been
shown in FIG. 2,
corresponding to compressor discharge pressures of 100 barg, 120 barg, 140 barg
and 160 barg.
horizontal line can be drawn from the compressor discharge until it intersects
with the 45°C isotherm. This represents the compressor aftercooler, with the
end of the line representing the inlet to the J-T valve. Given that the J-T
valve is adiabatic, the path through the valve can be drawn vertically down
from the point representing the outlet of the aftercooler until it intersects
the 0°C isotherm. This point is in the two-phase region for all four of the
cycles drawn in FIG. 2.
chilling duty is provided by the latent heat of the CO2 as it vaporizes;
therefore, for all four cases, the liquid flowrate will be the same. However,
the vapor fraction at the outlet of the J-T valve will be different. As the
vapor fraction increases, the vapor flow associated with the liquid also increases.
The consequence of this is an increased circulation rate; hence, increased
compressor size and power consumption. Alternatively, consider that for all
four cycles, the J-T valve outlet is at the same temperature and pressure but
with different molar enthalpies. The lower the compressor discharge pressure,
the higher the molar enthalpy at the outlet of the J-T valve. The chiller duty
is equal to the difference in molar enthalpy between the inlet and outlet of
the exchanger multiplied by the molar flowrate of the refrigerant. Therefore, a
higher inlet molar enthalpy results in a lower molar enthalpy change across the
exchanger; thus, a higher flowrate is required to achieve the same duty.
Given that increasing the compressor discharge pressure
will reduce the circulation rate of the refrigerant required to meet the same
chilling duty, this should reduce the power demand for the compressor. However,
increasing the compression ratio of the compressor should conversely result in
increased power demand. Another notable aspect of the cycles plotted in FIG. 2 is that there
are diminishing returns in reducing the enthalpy at the inlet to the chiller as
the compressor discharge pressure is increased—i.e., the gap between the J-T
valve outlet points for the 100 barg and 120 barg cycles is larger than from
120 barg to 140 barg and so on. This is because the 45°C isotherm becomes
increasingly steep as the pressure increases beyond ~90 barg. These factors
imply that there may be an optimum compressor discharge pressure that minimizes
the required power of the compressor.
relationship is revealed by plotting
compressor power vs. compressor discharge pressure (FIG. 3) in the range of 110 barg–160
barg. An optimum compressor discharge pressure exists at approximately 121
barg. Below this pressure, the increasing circulation rate dominates,
increasing the power demand. Above this pressure, the increasing compression
ratio becomes the dominant factor due to the diminishing returns in reducing
the circulation rate. Note: The calculated compressor power is the hydraulic
power (i.e., power at 100% adiabatic efficiency).
question that follows is whether this optimum point is always at 121 barg or
whether it is dependent on other variables, such as the chiller temperature or
the air cooler discharge temperature.
FIGS. 4a–4f show
plots of compressor discharge pressure against power for a range of air cooler
temperatures (32°C–60°C), with fixed refrigerant temperatures ranging from 0°C–50°C.
Red points indicate the minimum compressor power on each curve. These figures
show the following correlations between chiller temperature, aftercooler
temperature, compressor discharge pressure and compressor power:
is also a correlation between the chiller’s temperature and the compressor’s
power, although, in FIGS.
4a–4f, this relationship is not immediately apparent. FIG. 5 shows the optimum
compressor discharge pressure curves for all chiller temperatures plotted against
the aftercooler discharge temperature.
is important to remember that the definition of optimum in this context
is the pressure corresponding to the minimum power consumption of the
compressor. As the discharge pressure of the compressor increases, limitations
on compressor design may necessitate a multi-stage machine, and design pressure
limitations may require changes in pipe class that can incur step changes in
~120 barg, there are only small reductions in compressor power for further
increases in discharge pressure. The true optimum pressure should therefore be
assessed on a case-by-case basis, considering the capital cost of the
equipment. In a closed-loop system, it is unlikely that increasing the
discharge pressure beyond ~120 barg will yield a capital cost reduction.
However, in an open-loop system where the refrigeration loop is integrated with
a multi-stage injection compressor, higher pressure stages may be available, in
which case FIG. 5 provides a useful guideline as to which is the most
efficient stage to remove the refrigerant.
Alternative refrigerants. Refrigerant
selection is often constrained by the temperature required in the chiller,
which is limited to the atmospheric boiling point of the refrigerant, given
that it is undesirable to operate the suction of the refrigerant compressor at
vacuum conditions. It is often preferrable to select a refrigerant with the
highest boiling point that can still meet the chilling requirement, as this
will enable the system to operate at a lower pressure. TABLE 2 shows the
atmospheric boiling points of some commonly used industrial refrigerants.
FIG. 6 shows a pressure-temperature diagram with boiling point curves for
CO2 and for the refrigerants listed in TABLE 2. Each curve terminates at the
component’s critical point. For any temperature, the saturation pressure of CO2
is higher than that of all the alternative refrigerants shown. Given that the
refrigerant will leave the chiller as a saturated vapor, this means that a CO2
refrigeration system will always operate at a higher pressure than any of these
alternative refrigerants at the same temperature.
The boiling curve of CO2 does not extend down to
atmospheric pressure (0 barg). Instead, it reaches its triple point at ~4.2
barg and –56.6°C. Below this point, the equilibrium is solid-vapor
(sublimation) rather than liquid-vapor. If the J-T valve were to decrease CO2
to below this pressure, solid CO2 (i.e., dry ice) would form in the J-T
valve and downstream pipework. Therefore, the limitation on the minimum
achievable temperature for CO2 refrigerant is not the atmospheric
boiling point (which, for CO2, does not exist), but the triple
comparison, the triple points of the alternative refrigerants are shown in TABLE 3. All the
alternative refrigerants listed have triple points that occur at deep vacuum; therefore,
the formation of solid-phase refrigerant is of no concern.
that some margin to the triple point would be required to mitigate the risk of
forming dry ice in the system, CO2 could practically be used as a
refrigerant for services down to around –50°C. This is comparable to the
minimum temperature achievable using propane, R-134a, R-125 and R-32. Of these,
propane is by far the most common refrigerant used in large energy and oil and
gas projects, and as a result, can be simulated most reliably in common process
simulation software. There is also increasing pressure to phase out HFC
refrigerants due to their high GWP. Propane is a likely replacement for many of
these HFC refrigerant systems given its low GWP and low boiling point. Therefore,
propane will be the refrigerant to which CO2 will be compared in the
System depressuring and material
a system contains a liquid that has a vapor pressure greater than atmospheric
pressure, the potential for low temperatures to be generated by
auto-refrigeration during system depressuring must be considered. Refrigeration
systems are particularly prone to this phenomenon, as they often contain pure
or nearly pure component liquids that are at their boiling points at elevated
a pure component system, a singular defined boiling temperature corresponds to
every pressure. If the system pressure falls below the fluid’s vapor pressure
at the system temperature, this will result in vapor generation. The latent
heat required to generate this vapor is lost as sensible heat, resulting in the
system becoming colder. Vapor will be generated until the system pressure is
equal to the vapor pressure at the new, lower temperature.
the pressure continues to reduce due to mass being lost from the system—as is
the case during system depressuring—this phenomenon will continue. As the
pressure drops and liquid boils off, it will remain on its boiling point curve
until it either reaches its atmospheric boiling point or until all liquid has
been boiled off and all that remains is saturated vapor. If the former is the
case, the system will remain with an inventory of liquid at its atmospheric
boiling point until heat input from the environment can boil off all remaining
liquid. Therefore, it is good practice to set the minimum design temperature of
the system to be no greater than the atmospheric boiling point of the
For propane, the atmospheric boiling point is –42°C. This is fortuitously
a small margin above the minimum design temperature that is typically specified
for low-temperature carbon steel (LTCS), which is –46°C.12 As previously
discussed, CO2 does not have an atmospheric boiling point; rather,
it has an atmospheric sublimation point—however, the principle is the
same as described above.
means that during rapid depressuring, it is highly likely that solid-phase CO2
(dry ice) will form in the system at the atmospheric sublimation temperature of
CO2 (–78.5°C). This design temperature cannot be accommodated by
LTCS; rather, it requires alternative metallurgy suited to such low
temperatures, such as stainless steel. Therefore, all equipment in a CO2
refrigeration system would require a step change in capital costs compared to
an equivalent propane refrigeration system due to the more expensive material
requirement to set the minimum design temperature to the atmospheric boiling
(or sublimation) point can be alleviated if it can be justifiably argued that
the system will not reach this temperature in practice. Consider that, as the
system cools down, the temperature will fall below that of the ambient air, and
it will begin to gain heat from the surrounding environment. If the system is
depressurized sufficiently slowly, an equilibrium can be reached where the heat
gain from the environment is sufficient to replace the heat lost in vaporizing
the refrigerant. The slower the rate of depressurization, the higher the
temperature will be at which this equilibrium can occur.
hydrocarbon systems, there is often a code-imposed requirement to depressurize
a system to a certain pressure within a defined timeframe. However, this requirement
may not apply to CO2 systems as the fluid is non-flammable. It may
be possible to depressurize the system sufficiently slowly so that the system
never drops below the design temperature of LTCS; however, this must be proven
on a case-by-case basis through rigorous dynamic simulation.
CASE STUDY: A CARBON CAPTURE PLANT WITH
INTEGRATED CO2 REFRIGERATION
CO2 refrigeration system would perhaps be best integrated with a
carbon capture and injection plant where there is a significant refrigeration
duty required. One such system would be a propylene carbonate carbon capture
plant. Propylene carbonate is a physical solvent that is regenerated through
multi-stage flashes with no heat input. These plants have no process heat demand
but can have a significant refrigeration duty associated with chilling the lean
solvent for improved rich-solvent loading in the absorber.13 The
chilled solvent temperature is typically in the range of –25°C to –30°C with a
very tight approach in the chiller. For this case study, a refrigerant
temperature of –30°C was selected with a solvent temperature of –29°C.
case study was performed, comparing a conventional propane refrigeration system
to a CO2 refrigeration system to provide chilling duty to a
propylene carbonate CO2 capture plant. The plant will capture CO2
on the basis shown in TABLE
4 and compress the captured CO2 for sequestration.
configurations for the CO2 refrigeration system were considered. The
first was a closed-loop CO2 system as described in earlier sections.
The second one was an open-loop system (FIG. 7) in which the refrigeration system was
integrated with the CO2 injection compressor.
CO2 injection compressor considered was based on a seven-stage machine
that would increase pressure from 0.85 barg to 180 barg for injection, with
intercooling to 45°C between each stage. The interstage pressures are shown in TABLE 5 and were based
on a typical pressure profile.
Propane refrigerant. The
vapor pressure of propane at 45°C is 14.4 barg. Allowing a 0.5-bar pressure
drop for the condenser provides a compressor discharge pressure of 14.9 barg.
The suction pressure is the vapor pressure of propane at –30°C, which equals 0.66
Closed-loop CO2 refrigerant. Using
FIG. 5, the
optimum compressor discharge pressure with a –30°C chiller and a 45°C air
cooler is 142 barg; however, this would result in a high compression ratio and a
high discharge temperature. From FIG. 4d, the power curve is quite flat beyond ~120 barg. In
fact, the compressor power at 120 barg is less than 3% higher than that at 142
barg. Therefore, 120 barg was selected as the compressor discharge pressure, as
this results in a discharge temperature of 170°C, which is high but within the
range of a centrifugal compressor. The suction pressure is the vapor pressure
of CO2 at –30°C, which equals 13.3 barg.
Open-loop CO2 refrigerant. In
the open-loop configuration, the refrigeration loop was positioned in a recycle
loop around several stages of the CO2 injection compressor. This has
the benefit of eliminating the separate refrigeration compressor and
aftercooler by incorporating these duties into the injection compressor and
disadvantage of this configuration is that the interstage pressures are
unlikely to be optimum for running the refrigeration loop. This means that the
additional compressor power will be greater than for the independent closed-loop
refrigeration unit, which can be optimized for chiller requirements.
closest pressure to the optimum of 142 barg is the discharge of Stage 6 at 120
barg; therefore, this will be the take-off location for the refrigerant supply
(downstream of the interstage cooler). It is beneficial to return the
refrigerant at the highest possible pressure that can meet the chiller
temperature of –30°C (vapor pressure of 13.3 barg). The best option is to
return the refrigerant to the suction of Stage 3 at 9.5 barg. TABLE 6 shows the
results of simulations for each refrigeration configuration.
propane refrigeration system operates at much lower pressures than an
equivalent CO2 refrigeration system. The propane system can be
accommodated by 300# piping (or even 150#) depending on the achievable
condensing pressure. A CO2 system will likely require 900# or 1500#
piping due to the high design pressure required. The issues with the depressurizing
temperatures discussed mean that stainless-steel metallurgy will usually be
required for the CO2 system, whereas the propane system can be
manufactured from LTCS.
The equipment for the CO2 system will typically be
physically smaller than the equivalent propane equipment. This is largely due
to the higher operating pressure and reduced system volumes. It may also be
possible to reduce the size and number of chillers required by using CO2,
which can yield a higher heat transfer coefficient than propane.
To provide the same chilling duty, a CO2 system
requires more power than a propane system. A closed loop is easier to optimize
and can have lower power requirements than an open-loop system. If an open-loop
system is used, it is highly beneficial to set the interstage pressures such
that stages exist close to the optimum refrigeration supply and return pressures.
On balance, it is unlikely that a closed-loop CO2
refrigeration system will provide a cost-competitive solution vs. a
conventional propane refrigeration system. However, if an open-loop system is
possible, in which the refrigeration loop is integrated with a multi-stage CO2
injection or export compressor, this may present a lower capital cost solution
than a standalone propane refrigeration system. This is especially true if the
additional compressor duty is a relatively small percentage of the total
injection compressor duty.
In addition, a CO2 refrigeration system has the following
The authors would like to thank Adish Jain, Process Technology
Director, Fluor, and Samantha Nicholson, Process Fellow, Fluor, for their
valuable input to this article.
CHARLIE GOULD is a Process Engineer at
Fluor. He has 8 yr of experience on a variety of upstream oil and gas,
refining, chemicals, heat and power, carbon capture and hydrogen projects. He
specializes in steady-state and dynamic process simulation, overpressure
protection and process safety time analysis. Gould earned an MEng degree in chemical
engineering from the University of Surrey in the UK and is a chartered engineer
ROBERT JOYNER is a Principal Process
Engineer at Fluor. He has more than 20 yr of experience in onshore and offshore
oil and gas, refining and chemicals projects, specializing in overpressure
protection and flare system design. He earned an MEng degree in chemical engineering
from Imperial College London and is a chartered engineer with IChemE.
ADYA DESHMUKH is a Senior Process Engineer
at Fluor. She has more than 20 yr of experience, specializing in CO2
processing, including Econamine FG PlusSM, direct air capture and CO2
compression/conditioning process design. She graduated from the University of
Surrey in chemical engineering and is a registered chartered engineer.