The Refrigerant Is Part of the Machine
When a homeowner or specifier compares air-source heat pumps, the refrigerant is usually reduced to a line on the data sheet and a number for global warming potential (GWP). In engineering terms it is much more than that. The working fluid defines the pressures at which the machine runs, the mass and volume of vapour the compressor must move for each kilowatt of heat delivered, the temperature at which the compressed gas leaves the compressor, the oil with which it must be compatible, the safety classification of the installation and the highest water temperature that the unit can deliver on the coldest day. The two fluids at the centre of the present transition in the UK domestic market, R32 (difluoromethane, CH2F2) and R290 (propane, C3H8), differ on every one of these dimensions.
The regulatory driver is the phase-down of hydrofluorocarbons (HFCs) under the F-gas regime, which uses GWP as the principal measure of climate impact. R410A, long the standard in domestic and light commercial heat pumps, has a GWP of 2,088 on the 100-year basis used in the regulations. R32 has a GWP of 675, about a third of that, and R290 has a GWP of 3, which is effectively zero. The values differ between the successive assessment reports of the IPCC, with R32 at 771 in the most recent, and the specifier should confirm which basis a regulation or tender uses, as well as the current state of the UK F-gas rules, which have diverged from those of the EU. This audit looks beneath the GWP figure at the thermodynamics. It traces the vapour-compression cycle on the pressure-enthalpy diagram, quantifies a worked comparison at a UK winter design condition, examines temperature glide and why it matters, analyses the behaviour of scroll compressors and their lubricants, and discusses performance at low ambient temperatures. All the cycle figures were computed from the real-fluid equations of state, not from rules of thumb, and they are stated with their assumptions.
Pressure-Enthalpy Cycle Calculus
The vapour-compression heat pump cycle is conveniently represented on a diagram of pressure against specific enthalpy. The refrigerant leaves the evaporator as a slightly superheated vapour (state 1), is compressed to the condensing pressure (state 2), is cooled and condensed in the condenser, leaving as a slightly sub-cooled liquid (state 3), and is expanded through the expansion valve at constant enthalpy to the evaporating pressure (state 4). Four quantities follow from the enthalpies. The evaporator duty per kilogram of refrigerant, the refrigerating effect, is h1 − h4, where h4 = h3. The condenser duty per kilogram, the heating effect, is h2 − h3. The specific compressor work is h2 − h1. The heating coefficient of performance is COP = (h2 − h3) / (h2 − h1).
The compressor work is related to the isentropic work by the isentropic efficiency, ηis, as Wc = m-dot × Δhis / ηis, where Δhis = h2s − h1 is the enthalpy rise along the isentrope from the suction state to the discharge pressure, and the actual enthalpy at discharge is h2 = h1 + Δhis / ηis. The mass flow rate required for a heating capacity Q is m-dot = Q / (h2 − h3), and the volume flow rate that the compressor must sweep at the suction is V-dot = m-dot / ρ1. This last relationship is the origin of the volumetric heating capacity, (h2 − h3) × ρ1, in kJ per cubic metre of suction vapour, which determines how large a compressor is needed for a given duty. For an ideal gas with constant isentropic exponent k, the isentropic work is the familiar expression k / (k − 1) × p1v1 × [(p2 / p1)(k−1)/k − 1], and the discharge temperature of an isentropic compression is T2s = T1 × (p2 / p1)(k−1)/k. Real refrigerant vapours depart considerably from the ideal gas at the pressures involved, so the real-fluid enthalpy method is used for numbers, but the ideal-gas expression explains why fluids with a high isentropic exponent run hot.
A Worked Comparison at a UK Design Day
Consider an air-source heat pump on a UK winter design day with an outdoor air temperature of −7 degrees Celsius, delivering a flow temperature of 55 degrees Celsius. An evaporating temperature of −15 degrees Celsius, allowing about 8 K between the air and the refrigerant, and a condensing temperature of 60 degrees Celsius, allowing 5 K for the condenser approach, are representative. Take a single-stage cycle with 5 K of suction superheat, 3 K of liquid sub-cooling and an isentropic efficiency of 0.70, with the compression treated as adiabatic. The results from the real-fluid properties are as follows.
For R290, the evaporating pressure is 2.92 bar and the condensing pressure 21.2 bar, a pressure ratio of 7.3. The latent heat of vaporisation at −15 degrees Celsius is 395 kJ/kg, the refrigerating effect is 208 kJ/kg, the specific compressor work is 137 kJ/kg, the heating effect is 345 kJ/kg and the cycle COP is 2.52. The isentropic discharge temperature is 73 degrees Celsius and, with the 0.70 efficiency and no heat loss from the compressor, the discharge temperature is 90 degrees Celsius. The suction vapour density is 6.3 kg/m3, and the volumetric heating capacity is approximately 2,190 kJ/m3.
For R32, the evaporating pressure is 4.88 bar and the condensing pressure 39.3 bar, a pressure ratio of 8.1. The latent heat is 337 kJ/kg, the refrigerating effect is 204 kJ/kg, the specific work is 135 kJ/kg, the heating effect is 338 kJ/kg and the cycle COP is 2.51. The isentropic discharge temperature is 126 degrees Celsius and the adiabatic discharge temperature at 0.70 efficiency is 158 degrees Celsius. The suction vapour density is 13.0 kg/m3, and the volumetric heating capacity is approximately 4,380 kJ/m3.
Three conclusions follow. First, at the same isentropic efficiency the two fluids deliver almost the same cycle COP, 2.52 and 2.51. The differences that appear in practice therefore derive from the compressor and heat exchangers rather than from an inherent thermodynamic advantage of either fluid. Second, R32 has about twice the volumetric heating capacity, so an R290 machine needs a compressor of roughly double the swept volume for the same duty, and larger suction lines. Third, the discharge temperature of R32 is dramatically higher, by about 53 K isentropically, because of its high isentropic exponent, and the figure of 158 degrees Celsius for an adiabatic compressor is an upper bound: in a real scroll compressor, part of the heat of compression is lost through the shell and carried away by the suction gas, and measured discharge temperatures are lower, but they remain above 110 degrees Celsius and the difference from R290 is large. At a mild condition, evaporating at 0 degrees Celsius and condensing at 45 degrees Celsius, the pressure ratios are 3.2 and 3.4, the COPs are 4.35 and 4.26 and the adiabatic discharge temperatures are 65 and 103 degrees Celsius, which shows how the divergence widens with the pressure ratio.
The statement that R290 has a superior isentropic compression efficiency is therefore best treated as qualified. The fluid does not by itself confer a higher ηis, but its lower pressures, with a condensing pressure of 21 bar against 39 bar, reduce the pressure difference that drives internal leakage and the mechanical loading of the compressor, and its lower discharge temperature allows the unit to operate over a wider envelope without special measures. The efficiency of a given compressor is also a function of how closely its geometry is matched to the pressure ratio, which is discussed below.
Zeotropic Blends versus Single-Component Phase Dynamics
A pure fluid boils and condenses at a constant temperature for a given pressure. A mixture of fluids with different volatilities does not. In a zeotropic, or non-azeotropic, blend the more volatile component boils off first, so the composition of the liquid and of the vapour changes as the phase change proceeds, and the temperature increases from the bubble point, at which boiling begins, to the dew point, at which it ends. The difference is the temperature glide. R407C, a blend of R32, R125 and R134a, is the standard example: the calculation gives a glide of 6.6 K at an evaporating dew point of −15 degrees Celsius and 4.2 K at a condensing dew point of 60 degrees Celsius. R410A is a blend of R32 and R125 whose composition is close to azeotropic, and its glide is only about 0.1 K, so that it behaves in practice as a pure fluid. R32 and R290 are single-component fluids with zero glide, and their boiling and condensation lines are isothermal.
The Heat-Transfer Penalty
Glide has three effects on a plate heat exchanger. The first is a thermodynamic one: for a given dew-point temperature at the evaporator outlet, the mean evaporating temperature is lower by about half the glide, so the cycle runs with a larger lift. The second is on the heat-transfer coefficient. In a mixture, the lighter component evaporates preferentially at the liquid-vapour interface and the remaining liquid becomes enriched in the heavier component, which has a higher boiling point. The additional mass-transfer resistance in diffusing the species through the liquid film reduces the effective boiling heat transfer coefficient, and measured degradation relative to the ideal mixing value is commonly reported in the range of 10% to 30%, depending on the mixture and the flow. The third is a practical one concerning leakage: when a blend leaks as a vapour, the more volatile component is lost preferentially and the composition shifts, so that a topped-up charge no longer has the original properties, and the manufacturers of R407C systems typically require the charge to be added as a liquid, and in the case of a significant leak to be replaced entirely.
Glide can also be exploited. In a counter-flow heat exchanger, where the water temperature changes by 5 K, a refrigerant with a matching glide keeps the temperature difference approximately constant along the exchanger, which reduces the entropy generation. The ideal limit is the Lorenz cycle, with the COP calculated using logarithmic mean temperatures of the heat sink and the heat source. Heating water from 40 to 45 degrees Celsius and cooling air from −7 to −12 degrees Celsius, the logarithmic mean temperatures are 315.6 K and 263.7 K, giving a Lorenz COP of 6.07. A Carnot cycle between the extremes of 45 and −12 degrees Celsius has a COP of 5.58. The potential gain of about 9% is real in theory, but it can be realised only if the heat exchangers are designed with matched counter-flow, and the degradation of the boiling heat transfer coefficient in mixtures offsets some or all of it. The pure fluids avoid this trade-off by presenting an isothermal boiling and condensing line, which gives predictable heat transfer and unchanging composition, at the price of a pinch point at the water outlet of the condenser.
Scroll Compressor Isentropic Efficiency and Lubricant Behaviour
Most domestic heat pumps use scroll compressors, in which two interleaved spiral scrolls, one fixed and one orbiting, form crescent-shaped pockets that move inward and shrink in volume, compressing the trapped gas. The built-in volume ratio Vi of the scroll, fixed by its geometry, determines the internal pressure ratio at which the pocket opens to the discharge port, which is approximately Vik. For a built-in volume ratio of around 2.8 and an exponent of 1.1 for propane, the internal pressure ratio is about 3.1, and for R32 with an effective exponent of 1.3 about 3.8. The system pressure ratio is set not by the compressor but by the evaporating and condensing temperatures. When it exceeds the internal ratio, as it does in the worked example with values of 7.3 and 8.1, the gas in the discharge plenum is at a higher pressure than the pocket when the port opens and flows back into the pocket, an irreversible under-compression loss. When the system ratio is below the internal ratio, as in mild weather, over-compression occurs, and the gas is compressed beyond the necessary pressure and then expands. The isentropic efficiency of a scroll compressor, which might be 0.75 to 0.80 near its design ratio, falls to something like 0.60 to 0.65 at the high ratios of a cold day, though the figures vary between designs.
Leakage is the other principal loss. The scroll relies on small clearances, the radial gap between the tips and the opposing base plate and the flank gaps between the walls, which are sealed by oil films and gas dynamics, not by contact. The leakage flow increases with the pressure difference across the clearance. In the example the difference between the condensing and evaporating pressures is 18.3 bar for R290 and 34.4 bar for R32, so the R32 compressor works against a driving pressure almost twice as large. Friction in the bearings and the Oldham coupling, motor losses and inverter losses add to the total, and the overall efficiency is the product of these.
Discharge Temperature and the Oil
Discharge temperature matters because it determines the thermal stress on the lubricant and the compressor materials. R32 compressors are lubricated with polyolester (POE) oil. POE is stable at normal temperatures but hygroscopic, and in the presence of moisture it hydrolyses to form organic acids and alcohols, a reaction whose rate rises steeply with temperature. At the discharge port, where the oil meets gas above 110 to 120 degrees Celsius, the local oil film temperature is higher still, and the oil viscosity falls sharply with temperature. The viscosity of the oil in the bearings must remain high enough to maintain a hydrodynamic film, and the problem is compounded by the refrigerant that dissolves in the oil and reduces the viscosity further. The result of excessive discharge temperature is accelerated oil degradation, carbonised deposits at the discharge valve and bearing wear. Manufacturers therefore specify a maximum discharge temperature, commonly in the range of 110 to 130 degrees Celsius for R32 compressors, and the controller reduces the compressor speed or injects liquid or vapour to stay within it.
R290 has the opposite problem in the lubrication system. Propane is highly soluble in the oils used with it, which include polyalkylene glycols, POE and hydrocarbon-compatible mineral or synthetic oils, and a large mass fraction can dissolve in the sump oil, especially when the compressor is cold, which reduces the viscosity and the lubricating capability at start-up. The designer controls this with a crankcase heater, a limit on the charge and a choice of oil of higher base viscosity. The low discharge temperature of R290 is a benefit for the oil and the motor insulation, but the oil dilution must be addressed in the design. In both cases it is necessary to maintain oil return to the compressor, which depends on the velocity of the refrigerant in the suction line, and this is an important consideration at low capacity.
Low-Ambient Performance Boundaries on Sub-Zero Design Days
As the outdoor temperature falls, the evaporating temperature falls with it, the suction pressure and vapour density fall, and the mass flow through a compressor of fixed swept volume falls in proportion. In the worked example, the volumetric heating capacity of R290 falls from 3,540 kJ/m3 at an evaporating temperature of 0 degrees Celsius to 2,190 at −15 degrees Celsius, which is 62% of the original value, and that of R32 falls from 6,750 to 4,380, or 65%. A fixed-speed machine would lose about a third of its capacity between these conditions, while the building's heat demand rises, which is the fundamental mismatch of the air-source heat pump. An inverter-driven compressor compensates by running faster, up to its maximum speed, at which the capacity at −7 degrees Celsius air may be 70% to 90% of the nominal value depending on the design, and the efficiency declines as the pressure ratio and the speed increase.
Frost and Defrost
When the evaporator surface is below 0 degrees Celsius and below the dew point of the air, moisture deposits as frost, which blocks the airflow through the fins and insulates the surface. The frosting rate depends on the absolute humidity of the air as well as on the surface temperature, and the worst conditions are often not the coldest. At −7 degrees Celsius the air holds much less moisture than at +2 degrees Celsius at the same relative humidity, so frost accumulation per hour is frequently greatest in the range of about 0 to 5 degrees Celsius in damp weather, and defrost cycles are most frequent there. The defrost, usually by reversing the cycle to send hot gas to the outdoor coil, temporarily takes heat from the water circuit, and the energy lost, together with the reduced capacity of a frosted coil, accounts for a loss of seasonal efficiency which can be a few percent and in the worst conditions approach 10%. A system with insufficient water volume suffers a dip in flow temperature during each defrost, which is one of the reasons for adequate buffering.
Vapour Injection
Vapour injection, also called enhanced vapour injection (EVI) or economised operation, addresses the high-pressure-ratio problem. A portion of the condensed liquid is expanded to an intermediate pressure, roughly the geometric mean of the evaporating and condensing pressures, which is 7.9 bar for R290 and 13.9 bar for R32 in the example, and is vaporised in a heat exchanger or flash tank that sub-cools the main liquid flow. The resulting vapour is injected through a port in the scroll into the compression pocket part-way through the compression stroke. The effects are threefold. The sub-cooled liquid has a larger refrigerating effect, so the evaporator duty per kilogram increases. The injected mass raises the flow through the condenser and so the heating capacity, typically by a larger proportion at low ambient than at mild conditions. And the cooler injected vapour mixes with the gas in the pocket, reducing the discharge temperature and also raising the effective internal pressure ratio, which reduces the under-compression loss. The cost is the additional heat exchanger, valve and control complexity, and the improvement in efficiency, though real, is moderate compared with the improvement in capacity and in discharge temperature.
For R32, vapour or liquid injection is often necessary to keep the discharge temperature within the allowable limit when delivering 55 degrees Celsius or more at −7 degrees Celsius and below, and it is the principal means of extending the envelope. For R290, whose discharge temperature is some 35 to 50 K lower at comparable conditions, a single-stage cycle can often meet the same duty without injection, which is the basis of the statement that R290 can deliver 55 degrees Celsius at −7 degrees Celsius without elaborate measures. That statement should be read as a statement about the envelope and the thermal stress on the compressor and not as a claim that R290 has a better COP at the point. The worked example gives equal cycle COPs, and the manufacturers' declared COPs at A−7/W55 for both fluids typically lie in the range of about 2.2 to 2.7. Both fluids can also deliver higher water temperatures when needed, and R290, with a critical temperature of 96.7 degrees Celsius against 78.1 degrees Celsius for R32, retains a larger margin from the critical point at high condensing temperatures, which is why R290 is favoured for high-temperature heat pumps.
| Refrigerant Chemical Designation | Molecular Phase Behavior and Temperature Glide | Vapor Pressure and Volumetric Capacity | Isentropic Compression and Discharge Temperature Vector | Low-Ambient (−7°C) Heat Delivery Profile |
|---|---|---|---|---|
| R290 (propane, C3H8); GWP 3; safety class A3 (highly flammable) | Single component; zero glide; isothermal boiling and condensation; critical temperature 96.7°C | 2.92 bar at −15°C; 21.2 bar at 60°C; volumetric heating capacity approx. 2,190 kJ/m3 (about half that of R32) | Pressure ratio 7.3; isentropic discharge 73°C; adiabatic discharge approx. 90°C; cycle COP 2.52 | Delivers 55°C flow in a single-stage cycle with a wide margin on discharge temperature; needs about twice the compressor displacement; low mass charge but flammability controls apply |
| R32 (difluoromethane, CH2F2); GWP 675 (AR4 basis); safety class A2L (mildly flammable) | Single component; zero glide; isothermal phase change; critical temperature 78.1°C | 4.88 bar at −15°C; 39.3 bar at 60°C; volumetric heating capacity approx. 4,380 kJ/m3 | Pressure ratio 8.1; isentropic discharge 126°C; adiabatic discharge approx. 158°C (real units lower but above 110°C); cycle COP 2.51 | Compact compressor but needs vapour or liquid injection and discharge-temperature limiting to hold 55°C or more at −7°C; higher stress on POE oil |
| R410A (50% R32 / 50% R125); GWP 2,088 (reference) | Near-azeotropic blend; glide approx. 0.1 K, behaves as a pure fluid | 4.80 bar at −15°C; 38.4 bar at 60°C; capacity similar to R32 | Discharge temperature lower than R32 at the same conditions; the incumbent refrigerant being phased down | Established performance; being superseded because of GWP |
| R407C (R32 / R125 / R134a); GWP approx. 1,800 (reference) | Zeotropic blend; glide 6.6 K at −15°C and 4.2 K at 60°C; composition shifts on leakage | 2.63 bar at −15°C; 25.3 bar at 60°C (dew point); lower pressures than R32 | Lower discharge temperature than R32; heat-transfer coefficients reduced by mixture effects | Glide can be exploited in counter-flow exchangers but the heat-transfer penalty and service constraints limit its use |
The GWP values are those of the 100-year basis used in the F-gas regime, and the pressures and cycle results in the matrix come from real-fluid property calculations for the stated conditions. Real machines differ in the isentropic efficiency, the heat exchanger approach temperatures and the compressor heat loss, and the figures should be treated as a consistent basis for comparison and not as the performance of any product.
Safety, Charge and Regulatory Considerations
The thermodynamic advantages of R290 are accompanied by a significant safety constraint. Propane is classified A3 under ISO 817 and ASHRAE 34, meaning low toxicity and high flammability, with a lower flammability limit of about 2.1% by volume in air, equivalent to roughly 38 g/m3. R32 is A2L, with low flammability, a high lower flammability limit and a low burning velocity, which makes ignition much less likely. Product safety standards such as the IEC 60335-2-40 series set limits on the refrigerant charge as a function of room volume and the arrangement of the equipment, with requirements for ventilation, ignition-source exclusion and leak management. A propane monobloc unit installed outdoors, with the refrigerant circuit entirely outside the building, is the configuration which most readily satisfies these provisions, and it is why propane heat pumps are overwhelmingly packaged as outdoor monoblocs with a water connection to the dwelling. The refrigerant charge for the same duty is considerably lower by mass for propane than for R32, often less than half, owing to the low density of the liquid and the compact charge of modern microchannel exchangers, which assists compliance.
The installer's competence requirements are also affected. Work on hydrocarbon systems requires training and tools appropriate to flammable refrigerants, for example spark-free equipment and appropriate recovery procedures, and the installer should not assume that qualifications for fluorinated gases cover them.
Specification Guidance for Practitioners
Several points follow. First, do not select a refrigerant on the basis of a single parameter. The GWP matters for compliance and for the long-term availability of service refrigerant, but the choice of refrigerant affects the compressor envelope, the oil, the safety classification and the installation configuration. Second, compare products at identical declared conditions, for example the capacity and COP at A−7/W55 and at A7/W35, and verify the maximum flow temperature at the lowest ambient. A unit that reaches 55 degrees Celsius at −7 degrees Celsius only by operating close to its discharge temperature limit will have less margin and more wear than one that does so comfortably. Third, ask how the manufacturer protects the compressor: the discharge-temperature limit, the use of vapour or liquid injection and the controller's behaviour when the limit is approached. Fourth, for R290 units, review the installation requirements, including the clearance around the unit, the position of openings in the dwelling relative to the unit, the ventilation of any enclosure and the manufacturer's rules on the charge and the location. Fifth, confirm the position under the UK F-gas regulations and future restrictions, which may affect the cost and availability of the fluorinated refrigerants over the life of the product.
The central lesson of the comparison is that the two fluids are not simply better and worse. R32 offers a compact, high-capacity compressor in a mildly flammable fluid, at the cost of high discharge temperatures and a reliance on injection and control strategy at low ambient. R290 offers a benign discharge temperature, low pressures and a negligible climate impact, at the cost of a larger compressor and the discipline of managing a flammable refrigerant. The cycle efficiency is, to a first approximation, the same. What differs is the margin by which each machine reaches its operating point on the coldest day of the year, and the engineering question for the specifier is how much of that margin they need.