Energy and Exergy Analysis of Low-Global Warming Potential Refrigerants as Replacement for R410A in Two-Speed Heat Pumps for Cold Climates

: Heat pumps (HPs) are being developed with a new emphasis on cold climates. To lower the environmental impact of greenhouse gas (GHG) emissions, alternate low global warming potential (GWP) refrigerants must also replace the exclusive use of the refrigerant R410A, preferably without re-engineering the mechanical hardware. In this paper, we analyze the performance of four low-GWP alternative refrigerants (R32, R452B, R454B, and R466A) relative to the conventional R410A and draw conclusions on the relative performances for providing heating in cold climates based on the Air-Conditioning, Heating, and Refrigeration Institute (AHRI) 210 / 240 standard for two-speed heat pumps. The simulations are carried using the Department of Energy, Oak Ridge National Laboratory (DOE / ORNL) Heat Pump Design Model (HPDM), a well-known heating, ventilation, and air conditioning (HVAC) modeling and design tool in the public domain and the HVAC research and development community. The results of the simulation are further scrutinized using exergy analysis to identify sources of systemic ine ﬃ ciency, the root cause of lost work. This rigorous approach provides an exhaustive analysis of alternate low-GWP refrigerants to replace R410A using available compressors and system components, without compromising performance.


Introduction
The earth continues to experience record increases in surface temperatures, largely caused by anthropogenic activity. At the 21st Conference of the Parties in Paris 2015 (COP21), 188 countries reached an agreement to achieve global warming below 2 • C by the end of 2100. A global phasedown of hydrofluorocarbons (HFCs) has been estimated to reduce warming by 0.5 • C, by 2100. This paper addresses the research question of replacing currently used R410A with low global warming potential (low-GWP) refrigerants for cold climate heat pumps, in which there is a gap in current knowledge and performance assessment especially related to drop-in replacements. Exergy analysis of cold climate heat pumps at the component level provides greater insight into systemic inefficiency. Previous exergy-related studies have been limited to ordinary vapor compression heat pumps in conjunction with renewable energy sources, or incremental improvements through coupling, or using different refrigerants [1][2][3][4][5][6][7][8][9][10][11][12][13]. In this paper, the application in cold climate conditions, and operation at the temperature levels of the Air-Conditioning, Heating, and Refrigeration Institute (AHRI) 210/240 standard for two-speed heat pumps are new, relative to drop-in low-GWP refrigerants. To identify systemic inefficiency, a thermodynamic approach through exergy analysis is applied to heat pumps using renewable energy [14,15], to R22 substitution [16], to low-temperature heat sources [17], to sorption and ground-source or geothermal heat pumps [18,19], and to the general problem of heat pumps for simultaneous heating and cooling [20], but not to the specific issue of providing heating in cold climates based on the AHRI 210/240 standard for two-speed heat pumps by replacing R410A with low-GWP refrigerants.
Alternate, low-GWP replacements should have the following key characteristics: (1) a similar cooling/heating capacity per volumetric flow rate; (2) comparable or higher energy efficiency; (3) ability to work in the same temperature and pressure ranges; and (4) a resultant compressor discharge temperature that is similar or lower than the baseline refrigerant, and therefore, there is no need to change the compressor oil. Refrigerant fluid properties are significant for flow and wettability [21]. Table 1 depicts promising alternative refrigerants for R410A with much lower GWPs. The R410A replacements are mostly blends containing R32 and hydrofluoroolefin (HFO) R1234yf, as well as a pure refrigerant, R32. They have approximately 70% lower global warming potential (GWP) than R410A. Some compositions of R32 and R1234yf are mildly flammable and are classified as A2L. It should be noted that R466A is an A1 refrigerant, i.e., non-flammable [22]. These refrigerant blends have negligible temperature glides, i.e., deviation between the temperature starting condensation and starting evaporation at the same pressure. All the alternatives have higher critical temperatures than R410A, making them more suitable for high condensing temperature operation.

Refrigerant Properties
A candidate refrigerant should present the following two key characteristics: the enthalpy difference between saturated vapor line and the saturated liquid line (i.e., latent heat of vaporization) and the critical refrigerant temperature (to separate supercritical and subcritical working regions). In addition, volumetric heat of vaporization quantifies the cooling capacity per unit volumetric flow. Bigger volumetric capacity refrigerants lead to bigger cooling capacities if using the same compressor size.
Four refrigerants to replace R410A are compared in Figure 1 by temperature-enthalpy diagrams. All the replacements (except R466A) can tolerate higher critical temperatures and have wider domes, which means they are more suitable for high ambient operations. Figure 2 shows the volumetric heat of vaporization versus the average saturation temperature. R32 has noticeably larger volumetric heat of vaporization than R410A, and R466A is slightly larger. R452B and R454B have volumetric heats of vaporization slightly lower than that of R410A. When approaching the critical temperature from either side of the dome, the heat of vaporization drops until it vanishes at the critical point.
Saturation temperature versus density of R410A and the four alternative refrigerants are shown in Figure 3. All the alternatives, except for R466A, have lower suction vapor density and less refrigerant mass flow rate than R410A for a fixed compressor displacement volume. Lower volumetric capacity results in smaller capacity with the drop-in application and the same compressor. The saturation temperature is impacted by the heat transfer characteristics and refrigerant pressure drop. The smaller cooling capacity is also due to the lower saturation temperature in a drop-in application.

Model Description, Features, and Assumptions
The Department of Energy, Oak Ridge National Laboratory (DOE/ORNL) Heat Pump Design Model [23] was used to model a baseline heat pump with R410A as the refrigerant. The heat pump system consists of a condenser, an evaporator, a thermo-expansion valve, and a compressor. Capabilities of the DOE/ORNL Heat Pump Design Model (HPDM), are described extensively in [23] and the salient features pertaining to this paper are described concisely below.

Compressor Model
The compressor power consumption and refrigerant mass flow rate were modeled using AHRI 10-coefficient compressor maps [24]. A compressor shell loss ratio, to the compressor power consumption is a user input to quantify the refrigerant energy gain in the compressor flow path. The predicted refrigerant flow rate is corrected by the real suction density, as described in [25]. The compressor map, originally developed for R410A, was used to model the alternative refrigerants. It was assumed that the compressor would maintain the same volumetric and isentropic efficiencies at the same suction and discharge pressures. Thus, the efficiencies were defined from the original R410A map as a function of the suction and discharge pressures, adopted by the alternative refrigerants in the same pressure range. The volumetric, and the isentropic efficiency are defined in Equations (1) and (2), respectively, as: where m r is compressor mass flow rate, kg·s −1 ; P is compressor power, W; η vol is the compressor volumetric efficiency; Sp rotation is the compressor rotational speed, rotations per second; D suction is the suction refrigerant density, kg·m −3 ; η isentropic is compressor isentropic efficiency; H suction is the compressor suction enthalpy, J·kg −1 ; and H discharge,s is the enthalpy, J·kg −1 , obtained at the compressor discharge pressure and suction entropy. The approach of converting the compressor map of a baseline refrigerant used by its drop-in replacements has been extensively validated in [17]. The map conversion method for the alternative refrigerants reached similar accuracy as the compressor map developed for the original refrigerant.

Heat Exchanger Model
The fin-and-tube condenser is modeled using a segment-to-segment modeling approach, i.e., dividing one tube into multiple mini control volumes; each control volume considers air-side and refrigerant-side heat transfer and energy balance with possible phase change. Heat transfer within each control volume is calculated using an ε-NTU method. Pressure drops are considered in both the air side and refrigerant side. Figure 4 shows the diagram. The kth segment indicates a single segmented-control volume having energy transfer between the air side and refrigerant side. In addition, . m a,k is the air flow rate across the segment, h a,k,i and h a,k,o are the air side inlet and outlet enthalpies, . m r,k is the refrigerant side mass flow rate, P r,k,i and P r,k,o are the inlet and outlet refrigerant pressures, and h r,k,i and h r,k,o are the refrigerant side inlet and outlet enthalpies. Using the same modeling approach as the segment-to-segment fin-tube condenser (Figure 4), the evaporator model additionally considers dehumidification. The method introduced by Braun et al. [26] was adopted to deal with water condensing on a cooling coil, where the heat and mass transfer was driven by the difference between specific enthalpies of the inlet air and saturated air at the refrigerant temperature. As noted, the mini segmental modeling approach could reveal the glide of a zeotropic refrigerant (refrigerant with a temperature glide between the onset of boiling and onset of condensation), since the temperature increment was accounted for by each individual segment along the refrigerant flow path.
Through an extensive literature survey, we identified heat transfer and pressure drop correlations suitable for modeling and optimizing air conditioners and heat pumps, and valid for various low GWP refrigerants. The references are listed in Table 2. Table 2. Selected references describing correlations for two-phase heat transfer and pressure drop.

Reference for Correlation
Fin-and-tube heat exchanger (FTC) evaporation heat transfer Thome [21] FTC condensation heat transfer Cavallini et al. [27] FTC evaporation pressure drop Choi et al. [28] FTC condensation pressure drop Choi et al. [28] The two-phase heat transfer correlations [21,27] account for local flow patterns, considering refrigerant properties. The idea of the flow-pattern dependent evaporation model is to first estimate the wetted surface inside a horizontal smooth tube using a flow map prediction. The model covers stratified flow; stratified wavy, annular wavy, intermittent, and annular flow; annular flow with partial dry out; and mist flow in evaporation. The general expression for local evaporating heat transfer coefficient is as follows: In Equation (3), θ dry is the dry angle corresponding to the dry circumference of the tube, which is determined from the flow pattern and void fraction. The void fraction model for this correlation is described in [29]. In addition, h wet is the heat transfer coefficient at the wet circumference of the tube, which is composed of a nucleate boiling term h nb and a convective boiling term h cb ; h v is the heat transfer coefficient at the dry circumference of the tube.
The local evaporation heat transfer coefficient calculated with the correlation of Thome [21] is shown in Figure 5. The correlation reasonably predicts degradation of heat transfer coefficient at high quality due to dry out of the liquid film, while most of the other correlations are not able to do so. This model can also reveal the effect of heat flux.
The local condensation heat transfer coefficient according to Cavallini [27] is given in Equation (5) as: where θ f ilm is the falling film angle around the perimeter of the tube, which is dependent on the local flow pattern and void fraction; h f is the Nusselt film condensation coefficient; and h c is the convective condensation heat transfer coefficient. The Cavallini [27] condensation correlation also starts with the flow pattern prediction, i.e., applying different heat transfer mechanisms specific to local flow patterns. It was assumed that azeotropic refrigerant (with negligible temperature glide) mixtures had similar heat transfer performances to those of pure refrigerants. Thus, pure refrigerant correlations can work for an azeotropic refrigerant mixture such as R410A. However, zeotropic refrigerant mixtures with large temperature glides, have different behaviors.
Heat transfer correlations developed for pure refrigerants must be corrected if they are to be used for zeotropic refrigerants. Stephan [30] proposed a correction method for condensation as well as evaporation. In this method, the mixture heat transfer coefficient h m is defined as: where h f is the heat transfer coefficient computed from a pure refrigerant heat transfer model; h g is the heat transfer coefficient of the vapor phase, which can be calculated with the Dittus-Boelter equation; and δQ SV /δQ T is the ratio between the sensible heat transfer rate and the total heat transfer rate. Bell [31] suggested that if the total isobaric temperature glide was around 7-8 • C, then the ratio could be approximated as: where ∆T is the temperature glide, ∆h m is the enthalpy of latent heat of the mixture, x is the quality (mass fraction of vapor in an equilibrium mixture of vapor and liquid), and Cp g is the vapor specific heat. We adopted air-side heat transfer correlations specific to individual fin types for dry and wet surfaces. To simplify the models for the thermo-expansion valve, it was assumed that the liquid enthalpy ahead of the expansion device was equal to that at the evaporator inlet. The evaporator superheat degree and the condenser subcooling degree were direct inputs. A heat exchangers' airflow rate and corresponding fan power consumption were direct inputs from the laboratory measurements or from the manufacturer's data. Temperature changes and pressure drops of refrigerant lines, e.g., liquid, suction, and discharge lines were specified using the measured data from experiments. REFPROP 10.0 from Lemmon et al. [32] was used to calculate the refrigerant properties. New refrigerants can be modeled in REFPROP by creating a refrigerant definition file.

Baseline Heat Pump and Rating Conditions
The DOE/ORNL Heat Pump Design Model (HPDM) was used for analytical evaluations for a baseline pump having a two-speed compressor. The two-speed heat pump has the total cooling capacity of 5 ton/17.6 kW at 35 • C ambient temperature/26.7 • C indoor dry bulb temperature (DB) and 19.4 • C indoor wet bulb temperature (WB). The high and low speeds of the scroll compressor provide 100%/67% capacity. The indoor and outdoor heat exchangers are described in Table 3. For the system modeling in, heating mode, the evaporator exit was assumed to have a constant superheat degree of 10 R (5.6 K). The AHRI 210/240 [33] standard was used for two-speed heat pumps. In heating mode, the heat pump should be rated at the indoor dry bulb temperature of 21.1 • C. At the low speed, the ambient temperature should vary at 16 Since HPDM is a steady-state system model, it does not predict cyclic performance and frost/defrost penalty. We adopted a typical degradation coefficient of 0.1 for the high speed, and 0.15 for the low speed. The AHRI 210/240 standard quantifies frosting/defrost loss at the outdoor condition of 1.7 • C DB/0.6 • C WB. If no frost effect is correlated by an analytical model, the standard recommends a 98% factor to scale the power consumption and a 91% factor to scale the heating capacity. The AHRI 210/240 standard assumes no frost formation at ambient temperatures above 8.  Table 4 predicts heating performances using R410A at the low (_L) and high speed (_H), including heating capacities (kW), heat pump heating coefficient of performances (COPs), compressor discharge temperatures, T( • F/ • C) and the compressor isentropic efficiency Equation (2). Comparing the high speed with the low speed at low ambient temperatures, the efficiency degradation is more pronounced because the loss factors ascend with the compressor pressure ratio. At −8.3 • C, the isentropic efficiency decreases from 64% to 50%. A typical heat pump sized to meet the building cooling load cannot satisfy heating load at low ambient temperatures.  Figure 6 compares the heat pump capacities as a function of the ambient temperature, versus the building load line of DHR min defined by the AHRI 210/240 standard for the ASHRAE climate zone IV. DHR min indicates a well-insulated building. Even in an adequately insulated home, the heat pump total capacity still cannot provide enough heating capacity. The high compressor speed will be called at an ambient temperature below −3 • C. The systemic COPs, based on the first law of thermodynamics, present a partial picture of energy efficiency. To investigate the sources of systemic inefficiency and where the opportunity of further efficiency improvements may be found, we perform an exergy analysis based on both the first and second laws of thermodynamics.

General Mass, Energy, and Exergy Balances
The general mass and energy balances equations are given by Equations (8) and (9), respectively, out m r − in m r = 0 (8) where Q o is the rate of thermal energy exchanged between the control volume (CV) around a specific component of the heat pump and the surroundings at T o ; W is the rate of work input to the compressor; h, v 2 /2, and gz represent specific enthalpy, kinetic energy, and potential energy, respectively; and m r is the refrigerant mass flow rate Equations (8) and (9) apply to steady-state steady flow conditions. The general exergy balance equation [34][35][36] applied across each component under evaluation is, where σ Total (W·K −1 ) is the sum of the entropy generation rates due to internal irreversibility and those from irreversible thermal energy transfer between the control volume surrounding each component and its surroundings. This method of exergy analysis of heat pumps has been validated and applied in low-GWP heat pumps, and in more complex cases of ground source heat pumps due to interaction with both the ground and the ambient air [37]. The irreversibility, which is the thermodynamic measure of the lost opportunity to do useful work, is the product of the ambient temperature, T o (K), and the rate of entropy generation, σ (W·K −1 ), is given by: The

Energy, Entropy Generation, and Irreversibility Calculations
The work input to the compressor obtained from the HPDM model for each outdoor temperature is the algebraic sum of W Re f rig.. and Q Comp which equals W Comp. , as shown in Tables 5-8. The negative sign for Q Comp is because heat is flowing out of the compressor CV, as per the International Union of Pure and Applied Chemistry (IUPAC) convention (energy inflow is positive, outflow is negative). From entropy generation rate, σ Comp. , the total compressor irreversibility, I Comp. = T o σ Comp. , where T o is the ambient temperature specified at the top of Tables 5-8 which must be converted to absolute temperature for use in Equations (10) and (11).
For a rigorous analysis, we must show consistency between the compressor work input (W Comp. ) from the HPDM model and that calculated from irreversibility analysis. The total actual work input to the compressor must be the sum of the Carnot work and the irreversibility losses. The Carnot work for an ideal heat pump with no irreversibility is given by: where T h is the set temperature (21.1 • C) of the conditioned space during heating and T L is the outdoor temperature indicated on the header of Tables 5-8. The total cyclic irreversibility, I Total , is the sum of the irreversibility of each component, i.e., the compressor, condenser, expansion valve (EXV), and line loss between the evaporator and the compressor inlet (I 6−1 ) and the line loss from the compressor outlet to the evaporator inlet, (I 2−3 ). The actual work input must be the sum of the ideal work (Carnot work) and the lost work due to irreversibility. Thus, W a = W Carnot + I Total and is in reasonable agreement with the work input to the heat pump, W Comp. derived from the model. This calculation is performed for each ambient temperature as per the AHRI 210/240 standard. We emphasize that the work computed from irreversibility analysis for the heat pump cycle and that obtained from the HPDM model are consistent, complementary, and validate our analyses based on both the first and second laws of thermodynamics.

Component-Wise Irreversibility
More detailed information on the systemic irreversibility is derived from the component-level irreversibility for each temperature level in the AHRI 210/240 standard, as shown in Figure 7. The structure of component-level irreversibility between the high and low speeds at 8.3 • C ambient temperature is markedly different, i.e., condenser irreversibility is greater than compressor irreversibility for all refrigerants. The total irreversibility is also higher for the high speed vs. the low speed, as might be expected (rapid processes have higher entropy losses). The irreversibility structure for 1.7 • C high and low speeds are similar in structure with the overall irreversibility at the high speed greater than at the low speed. For the coldest ambient case of −8.3 • C, again the structure of the irreversibility remains similar for the high and low speeds but notably, the compressor irreversibility is of the same magnitude, but the condenser and evaporator irreversibility is smaller for the lower speed versus the higher speed due to smaller condenser and evaporator loads. The comparative irreversibility (Figure 8, left) and the heating COP relative to Carnot (Figure 8, right) of all refrigerants at each AHRI 210/240 standard rated condition is even more telling. Although all refrigerants perform relatively similar on a comparative basis, the irreversibility of R452B and R454B are consistently lower than that of the others. Regarding performance relative to Carnot (Figure 8, right), the pattern is unescapable when going from low temperature (−8.3 • C) to the high temperature (16.7 • C). At 16.7 • C, the Carnot heat pump efficiency is large, making the relative efficiency small as compared with the Carnot efficiency. The opposite is true at a low temperature where both the Carnot and the heat pump efficiency are smaller, making their ratio larger.

Results and Discussion
Our analyses based on the AHRI 210/240 standard rated conditions indicate that low-GWP refrigerants, listed in Table 1 are valid replacements for R410A based on better efficiencies and lower exergy losses. R32 has the best heat transfer performance and requires the smallest compressor for delivering the same capacity, supported by previous studies [22]. The condenser duty of R466A is the greatest among all other refrigerants (Tables 5-8). The higher condenser duty is reflected in slightly higher condenser irreversibility losses. Compressor irreversibility losses in high speed are generally greater than at low speed at the same rating conditions (Figure 7). Depending on the ambient conditions, the most irreversible component may be the compressor or the condenser unit ( Figure 7). Irreversibility losses in both, evaporator and EXV are low. Irreversibility losses in the connecting lines are negligible and mainly due to thermal losses. The losses in the connecting lines between the compressor and the indoor coil are greater than the losses between the evaporator and the compressor inlet due to the relative distances. These losses, though small, could be further mitigated with better thermal insulation. Opportunities for reducing the irreversibility losses in the condenser could involve improved design and lowering of the condenser temperature, albeit at the expense of reducing some condenser duty. Differences in COPs of the heat pump for the refrigerants at the respective rating conditions were small, but the Carnot efficiency trended higher when the temperature lift was smaller. That is why the COP relative to the Carnot COP ( Figure 8) gradually trended lower with rising ambient temperatures (corresponding to smaller temperature lifts). The calculation of the actual compressor work and that obtained from the HPDM model are in very good agreement. The agreement in the energy balances obtained from the model and by using the state points with corresponding values of state variables using REFPROP 10 are in excellent agreement, ranging from being exact to a maximum discrepancy of~0.6% (see %∆E values in Tables 5-8). The discrepancy is due to the convergence criteria in the HPDM program when it needs to be relaxed somewhat to get convergence.

Conclusions
It is technically feasible to replace R410A with low-GWP refrigerants without changing the heat pump hardware. The candidate low-GWP refrigerants have better efficiency and performance characteristics relative to R410A. The energy input to the compressor, systemic irreversibility, and the condenser duty are important considerations before selecting a refrigerant. R32 and R454B have the lowest irreversibility. R466A has slightly greater irreversibility but a larger condenser duty. The component-wise irreversibility, shown in Figure 7, is crucial for understanding the sources of systemic irreversibility within the cycle. It is here that improvements should be made to make the heat pump even more efficient. Changeover to low-GWP refrigerants is a practical pathway towards mitigating GHG releases into the atmosphere without sacrificing energy efficiency. Heat exchanger design modifications to reduce irreversibility can be made based on exergy analysis for additional improvements.