Charge Reduction and Performance Analysis of a Heat Pump Water Heater Using R290 as a Refrigerant—A Field Study ^†

Abstract

Heat pump water heaters (HPWHs) are a proven technology for water heating that has been commercialized. The adoption of HPWHs for domestic and commercial water heating is growing rapidly because of their superior performance compared with alternative water heating methods. Whereas most existing systems use R-134a as a working refrigerant, R290 has gained major attention owing to its superior thermodynamic properties. The goal of the current study is to assess the performance of residential HPWH with R290 as a direct refrigerant replacement for R134a. Two units of a 50 gal HPWH were used in this experimental study. A baseline unit contained R134a refrigerant, and a prototype unit contained R290 refrigerant. The prototype unit was developed through the modification of a commercially available HPWH unit to achieve a low charge of R290 refrigerant. Another major modification was the replacement of the baseline compressor with a compressor designed for R290. Tests were conducted in a field environment (a research and demonstration house) using programmed drawn profiles daily. The prototype that reduced the charge by 43–47% provided displayed performance comparable to the baseline unit regarding first-hour rating (FHR) and the uniform energy factor (UEF).

1. Introduction

Most residential water heaters are conventional natural gas-fired or electric resistance-based. Such systems are relatively simple and have low equipment costs, but their efficiency is significantly lower than that of HPWHs. Since its early development, HPWH technology has held great appeal because of its higher thermal efficiency, which can lead to substantial energy savings over other systems [1]. Furthermore, HPWHs are attractive for use in northern regions of the world due to having a high coefficient of performance (COP) in colder environments [2]. Despite having few components, the traditional heat pump system is quite complex because selecting components (e.g., evaporator, compressor, condenser) is critical to its overall efficiency [3,4]. The addition of a storage tank and relatively different condenser configurations add to the system’s complexity. Several studies focused on design and efficiency improvement of HPWHs have targeted the performance of individual components [5], overall system configuration [6,7,8,9,10], and refrigerant selection [11,12,13,14], while other studies have investigated the feasibility of HPWHs in cold climates [15]. Nawaz et al. [12,13] conducted a study to analyze two possible configurations for condenser deployment. The configurations were counter flow and parallel counter flow, based on the general orientation of the flow of the refrigerants compared with the flow of water within the tank. This study concluded that thermal stratification heavily depends on the condenser design, and orientation can impact the use of the water heating process.

R290 can work at high evaporating temperatures, making it a good solution for waste heat recovery [16]. Studies have shown that R290 has better characteristics compared to other refrigerants [17]. For example, it was reported that for a window air conditioning system, the energy consumption and cooling capacity of R290 were lower than those of R22 by 12.4–13.5% and 6.6–9.7%, respectively [18]. This refrigerant has also been found to exhibit good performance in heat pump systems. One study [19] involving R290 in a heat pump system under subcooling conditions showed that it displayed a nominal COP and heating capacity (with auxiliary consumption) of 5.61 and 47.1 kW, respectively. In another study which examined a solar assisted microchannel HPWH system, it was found that for a solar radiation intensity of 120–614 Wm⁻² the COP ranged from 3.08 to 3.92 [20].

Despite the positive aspects of R290, it has been classified as an A3 refrigerant by the American Society of Heating, Refrigerating, and Air-Conditioning Engineers, which means that it is highly flammable [21,22]. However, the safe use of R290 is possible with proper precautions such as a seal tight system, charge minimization, and proper ventilation [19]. As such, considerable research and development is currently dedicated to reducing the R290 charge amount to mitigate issues with its high flammability.

This work aims to test market-ready HPWHs with R134a as a baseline refrigerant and R290 as a replacement. The test campaign objective is to minimize the R290 charge amount to run the baseline unit with comparable performance. The campaign is designed for the seasonal testing of the HPWH prototype using R290 as a refrigerant. Furthermore, the baseline HPWH unit, which uses the R134a refrigerant, is also tested simultaneously and in the same ambient conditions for comparison. The units are installed in the garage of a medium-sized, single-family detached house. Different draw profiles are considered in the study to provide a comprehensive analysis of the prototype with R290.

2. Experimental Setup

A medium-sized, single-family house (2400 ft²) dedicated to research is used to run the tests. The home is located in Knoxville, TN, USA (climate zone #4). The original HPWH and the prototype are installed in the garage, which is not air-conditioned. The HPWH unit selected for testing is currently available on the market, has a 50 gal tank, and requires 240 V connectivity. This model information is not provided according to the Nondisclosure agreement between ORNL and the manufacturer. Furthermore, the baseline R134a charge is 1.59 lb. The unit has two 3000 W electric heaters based on the selected running mode, of which there are four: hybrid, efficiency, electric, and away. The efficiency mode turns off the electric heaters, whereas the hybrid mode uses both heat pumps and electric heaters. The electric mode is for heating by electric heaters only, and the away mode is used when the unit will not be used for an extended period. The units were set up for this investigation to run in hybrid mode during testing. The hot water temperature was 51.7 °C. Figure 1 displays the tested unit, the data acquisition (DAQ) system, and the instrumentation inside the garage.

Some modifications have been applied to the baseline unit. Because of a nondisclosure agreement (NDA) with an industrial partner, details of the prototype cannot be disseminated. The condenser coil is the main development component in the modified system. Two propane compressor prototypes of different sizes have been used in the prototype. The compressors are of a reciprocating type and are not yet available on the market. The other baseline unit components remain in their original condition.

Figure 2 presents a schematic diagram of the HPWH instrumentation system, and it is apparent from the figure that the unit is fully instrumented on both the refrigerant and water sides. Thermocouples (TCs) are placed on the refrigerant tubing before and after the compressor and before and after the expansion valve. Similarly, pressure transducers are used to monitor the refrigerant pressures at these points to find the refrigerant-associated properties at each state. On the water side, temperatures of cold and hot outlet water are observed. In addition, the water temperature inside the tank is monitored in relation to the tank height using a TC tree of eight TCs, distributed vertically at six equal volume increments. The thermocouples labeled TC1 to TC6 are used to build a thermocouple tree to measure the tank water temperature vertically.

The remaining two TCs are placed at heights at the exact location of the temperature sensors that control the electric resistance heaters on the water heater. In addition, the inlet and outlet ambient air conditions at the evaporator coil are monitored. The water flow rate is measured at the outlet of the water heater. The DAQ system controls each draw’s flow rate and total volume. Finally, the power consumption is measured using watt transducers with a nominal accuracy of 0.5%. A Campbell instrumentation system was used for data acquisition during the experiment in which measurements were recorded every 15 s, 1 min, and 60 min.

Pressure transducers (Omega PX309) sourced from Omega Engineering, Michigan City, IN, USA with an accuracy of 0.25% were installed in the refrigerant line to monitor the pressure at the compressor, expansion valve inlet, and outlet locations. The TCs were attached to the vapor compression coil using thermal paste at the exact locations. The measurement location is thermally insulated to reduce any measurement uncertainty by reducing heat transfer from the surrounding ambient. A flow controller “Alicat” with 0.6% accuracy was installed downstream of the tank outlet to control and measure the draw flow rate.

The prototype is a modified version of a commercially available R134a HPWH baseline unit. The aim behind making the prototype is to reduce the amount of R290 charge used compared with R290 drop-in-replacement in the baseline unit, which is ~1 lb. Both the baseline unit and the prototype are installed in the garage space of a single-family home. The two HPWHs are set to work daily with a programmed hot-water draw pattern, with daily tests from December 2022 until February 2023. The testing ambient conditions are listed in

Table 1. Ambient conditions.

Day	Ambient Temp. (°C)	Inlet Water Temp. (°C)
21 December 2022	9.51	11.67
2 January 2023	10.61	10.76
4 January 2023	10.88	11.37
30 January 2023	13.03	10.84
1 February 2023	13.09	11.00
8 February 2023	14.21	11.12

.

3. Uncertainty Analysis

The data processing stage included randomly selecting multiple days to obtain average seasonal performance data. The ambient conditions for the selected days were similar for consistency. The electric energy consumption and added thermal energy for water heating are summed for each selected day and then averaged for the season. The water heating thermal energy is calculated based on the water side properties according to the following equation [23]:

$${\mathrm{Q}}_{total}={\int }_0^{t^{\ast }}\left[{{(\dot{m}}_{out}\mathrm{C}}_{\mathrm{P}}{\mathrm{T}}_{\mathrm{o}\mathrm{u}\mathrm{t}}\right)-{{{(\dot{m}}_{in}\mathrm{C}}_{\mathrm{P}}\mathrm{T}}_{\mathrm{i}\mathrm{n}}\left)\right]dt+\left(m_{tank}{\mathrm{C}}_{\mathrm{P}}\right({{\overline{\mathrm{T}}}_{\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{k}}}_{\mathrm{f}}-{{\overline{\mathrm{T}}}_{\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{k}}}_{\mathrm{i}}\left)\right)$$

(1)

where $\mathrm{Q}$ is the thermal heating energy for a whole day of measurement; $\dot{m}$ is the mass flowrate (kg/s); ${\mathrm{C}}_{\mathrm{P}}$ is the water thermal capacity (J/kg. k); ${\mathrm{T}}_{\mathrm{o}\mathrm{u}\mathrm{t}}$ is the outlet hot water temperature; ${\mathrm{T}}_{\mathrm{i}\mathrm{n}}$ is the supply cold water temperature; $\mathrm{t}$ is the sample recorded time; $t^{\ast }$ is the total sampling time; and ${{\overline{\mathrm{T}}}_{\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{k}}}_{\mathrm{f}}$ and ${{\overline{\mathrm{T}}}_{\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{k}}}_{\mathrm{i}}$ are the final and initial average temperatures of the water inside the tank, respectively. The thermal energy uncertainty is calculated using uncertainty propagation methods.

The thermal energy equation is split into ${\mathrm{Q}}_{integral}$ and ${\mathrm{Q}}_{tank},$ where

$${\mathrm{Q}}_{integral}=\left[{{(\dot{m}}_{out}\mathrm{C}}_{\mathrm{P}}{\mathrm{T}}_{\mathrm{o}\mathrm{u}\mathrm{t}}\right)-{{{(\dot{m}}_{in}\mathrm{C}}_{\mathrm{P}}\mathrm{T}}_{\mathrm{i}\mathrm{n}}\left)\right]\times \mathrm{t}$$

(2)

and

$${\mathrm{Q}}_{tank}=\left(m_{tank}{\mathrm{C}}_{\mathrm{P}}\left({\overline{\mathrm{T}}}_{{tank}_f}-{\overline{\mathrm{T}}}_{{tank}_i}\right)\right)$$

(3)

The uncertainty ($\delta )$ for each equation is calculated, where ${\delta Q}_{integral}$ is calculated using the following equations:

$${\delta Q}_{integral}=\sqrt{\left({\left(\delta r\times t\right)}^2+{\left(r\times \delta t\right)}^2\right)}$$

(4)

where

$$\delta r=\sqrt{{\left({\delta \dot{m}}_{out}{\mathrm{C}}_{\mathrm{P}}{T}_{out}\right)}^2+{\left({\delta \dot{m}}_{in}{\mathrm{C}}_{\mathrm{P}}{T}_{in}\right)}^2+{(\left({\dot{m}}_{out}{T}_{out}-{\dot{m}}_{in}{T}_{in})\delta {\mathrm{C}}_{\mathrm{P}}\right)}^2+{{\dot{(m}}_{out}{{\mathrm{C}}_{\mathrm{P}}\delta T}_{out})}^2+{{\dot{(m}}_{in}{{\mathrm{C}}_{\mathrm{P}}\delta T}_{in})}^2}$$

(5)

The uncertainty in the tank thermal energy and ${\delta Q}_{tank}$ is as follows:

$${\delta Q}_{tank}=\sqrt{{\left({\delta m}_{tank}{\mathrm{C}}_{\mathrm{P}}∆T_{tank}\right)}^2+{\left(m_{tank}{\delta \mathrm{C}}_{\mathrm{P}}∆T_{tank}\right)}^2+{\left({\delta m}_{tank}{\mathrm{C}}_{\mathrm{P}}\delta ∆T_{tank}\right)}^2}$$

(6)

Accordingly, the total thermal energy uncertainty is calculated as follows:

$${\delta Q}_{total}=\sqrt{{\left({\delta Q}_{integral}\right)}^2+{\left({\delta Q}_{tank}\right)}^2}$$

(7)

The uncertainty in the thermal energy measurement was found to be ±4%.

To calculate the uncertainty in COP,

$$COP={\displaystyle \frac{Q_{total}}{P}}$$

(8)

where $P$ is the electric energy consumption in kWh. The same methodology is used and the uncertainty in the COP is calculated according to the following equation:

$$\delta COP=COP.\sqrt{{\left({\displaystyle \frac{{\delta Q}_{total}}{Q_{total}}}\right)}^2+{\left({\displaystyle \frac{\delta P}{P}}\right)}^2}$$

(9)

where the power consumption measurement uncertainty $\delta P$ was determined to be $\pm 5\%$ Finally, the COP uncertainty $\delta COP$ was found to be $\pm 6\%$

4. Results and Discussion

A prototype is built based on the baseline unit to run with R290. The condenser coil has been modified in the prototype. In addition, two R290 compressor prototypes of different sizes have been used in the prototype. Because of the NDA mentioned earlier, the modified condenser and the two compressor prototypes cannot be described here. Figure 3 presents the performance of the four tested units (i.e., baseline with R134a, baseline with R290 drop-in replacement, prototype with small compressor, and prototype with large compressor). The compressor is modified for all four units. The figure shows that the small compressor prototype can reach a UEF of 3.7 at 275 g of propane. The smaller propane compressor prototype with about 275 g of R290 outperforms the baseline unit with R134a. After modifying the condenser coil design connected to the small prototype compressor, the prototype reached a UEF of 3.5 at 220 g of R290 charge. Prototype development is discussed further in Nawaz et al. [24].

To better understand the performance variation between R134a and R290, a comparison of the thermodynamic performance of both refrigerants is displayed on the pressure–enthalpy diagram in Figure 4. The vapor compression thermodynamic cycle is plotted to show the cycle’s four states. The pressure of both refrigerants is comparable at the evaporator coil, but the R290 pressure is slightly higher in the condenser. Furthermore, the slight difference between the two refrigerants does not affect the overall performance. A distinction between the two gases can be observed when we look at the enthalpy of both refrigerants. The thermal capacity of R290 is almost double the thermal capacity of R134a, which reduces the charge amount of R290 to around half the charge of R134a. The superiority of R290 is an advantage that helps with charge amount reduction.

The thermodynamic advantage of R290 enables the charge reduction in the baseline drop-in replacement and the prototype.

Table 2. System characteristics for different prototype designs.

Prototype	Grams	Charge (Oz)	UEF	Discharge Superheat (K)	Energy Usage (kW)	Running Time (h)
A (Large Comp)	396.89	14	3.75	11.66	2.12	5.23
B (Small Comp)	311.84	11	3.68	18.04	2.17	5.01
C (Small Comp& modified condenser)	201.28	7.1	3.56	19.62	2.38	5.66

presents system characteristics for three versions of the prototype, large compressor (prototype A), small compressor (prototype B), and small compressor with modified condenser design (prototype C). Hereinafter, prototype C will be referred to as MaxTECH. The system-optimized charge amounts are 14, 11, and 7.1 Oz, respectively, due to the design differences. The first observation is seen between prototypes A and B. That is, the difference in UEF, energy consumption, and running time is less than 5%. This is due to the R290 high heat capacity (e.g., $C_p=2926{\displaystyle \frac{\mathrm{J}}{\mathrm{k}\mathrm{g}.\mathrm{K}}}\mathrm{a}\mathrm{n}\mathrm{d}{C}_v=1744{\displaystyle \frac{\mathrm{J}}{\mathrm{k}\mathrm{g}.\mathrm{K}}}\mathrm{a}\mathrm{t}2\mathrm{M}\mathrm{P}\mathrm{a}\mathrm{a}\mathrm{n}\mathrm{d}43°\mathrm{C}$). The main difference is observed in the compressor discharge superheat temperature which increases from 11.66 to 18.04 $\mathrm{R}$ This behavior is justified as the compressor is transferring energy to a lesser amount of refrigerant. The volumetric performance of R290 enables further modification in the prototype to reduce the charge. The baseline wrapped-around condenser is modified in MaxTECH accordingly. It is found that the charge amount is reduced by about 36% with a loss of about 3% in performance. This was evident by the limited increase in power consumption by about 10%. The charge reduction enforces an increase in the discharge superheat degree between prototypes B and MaxTECH similar to the behavior for prototypes A and B. Remarkably, the superheat increase is about 1.5 $\mathrm{R}$ which does not impose notable concerns. Hence, reducing the charge while keeping the system unchanged results mainly in a runtime increase (about a 13% increase between prototypes B and MaxTECH). This clear observation for the tested prototypes invites further effort in system optimization to further reduce the system charge without compromising the performance.

Two draw profiles have been selected to assess the units’ performance.

Table 3. Draw times, duration, and flow rates used in the test.

Time	Low Draw (LPM)	Draw Duration (Minutes)	Time	Medium Draw (LPM)	Draw Duration (Minutes)
06:30	5.68	11	06:30	5.68	11
16:24	3.79	20	10:36	3.79	20
			17:54	17.03	3
			18:54	17.03	3
Total		31			37

lists the draw pattern characteristics over 24 h. The draw pattern is selected to simulate the usage pattern of small- and medium-sized families.

The unit total running time and the accumulated draw volume per day were calculated and presented in Figure 5. The accumulated low usage draws about 37 gal per day, whereas, for medium use, the total drawn volume is ~59–63 gal. Medium-draw patterns were emulated on 2 and 30 January 2023, whereas low-draw patterns occurred for the other presented days. The running time of the two units was approximately similar during the low-draw pattern except for 8 February 2023, where MaxTECH ran for 37 min longer than the baseline unit. On 2 January 2023, the running time for the medium-draw-pattern was approximately 517 min for both units. However, on 30 January 2023, the baseline unit ran for 29 min longer than the MaxTECH unit.

To further understand the performance of both systems, the total daily power consumption and the thermal energy added to the water tank are calculated and plotted in Figure 6. The power consumption is comparable daily for both units except on 8 February 2023, when MaxTECH consumed 0.56 kWh more than the baseline unit consumed. Interestingly, the total added heat for both units is similar. For the added water heating thermal energy, the baseline unit showed slightly higher added heat except for 4 January and 30 January 2023, when the baseline showed noticeably higher added heat by 0.56 and 1.0 kWh, respectively, as seen in Figure 6. The corresponding added heat for both units is comparable.

2 January 2023, is the only day the MaxTECH unit added more heat (0.22 kWh). The daily system performance is evaluated by calculating the coefficient of performance (COP), which is the ratio between the added thermal heat and the consumed energy of the system. Figure 7 presents the COP values for both units. COP uncertainty was quantified using an uncertainty propagation method as explained earlier, and the Engineering Equation Solver (EES) [24] was utilized for the calculations. The baseline unit performs better than the MaxTECH unit performs for all selected days, a logical result based on the measurements shown in Figure 5 and Figure 6. The maximum COP is 3.16 for the medium-draw pattern and 2.65 for the low-draw pattern. This finding is justifiable because these units have 50 gal storage tanks, and the design is optimized to deliver an FHR of 66 gal. For the MaxTECH unit, the maximum COP is 3.1 for a medium-draw-pattern and 2.34 for a low-draw-pattern. Conversely, the lowest COP is 2.3 and 1.9 for the baseline and MaxTECH units, respectively; this performance occurred during the low-draw-pattern. Similar results were observed in a study by Ghoubali et al. [23], where the COP of an HPWH with a 52.8 gal (200 L) tank varied from ~2–3 for approximately 198 g of refrigerant charge. The largest performance deviation, corresponding to a difference of 0.56, occurred on 8 February 2023, during a low-draw-pattern day. The average COPs of the MaxTECH unit are 2.08 and 2.88 for the low- and medium-draw-pattern days, respectively. On the other hand, the average COPs of the baseline unit are 2.49 and 3.13 for the low- and medium-draw patterns, respectively.

The system’s internal performance is analyzed to understand the difference in the water heating process. The suction and discharge temperature profiles for both units are plotted in Figure 8 for 30 January. The maximum discharge temperature difference is around 6 °F, and the two-system temperature profiles are comparable, as seen in the figure. Furthermore, the profiles illustrate two running stints. For the first one, the MaxTECH system started running about 1 h before the baseline system started running and stopped running about 80 min before the baseline was disconnected, as seen in Figure 8. Moreover, as shown in the figure, the start and stop time stamps for the second stint are almost identical for both units. The trend is similar to the temperature profiles for the suction and discharge pressures of both units.

The power consumption measured for both units is plotted in Figure 9. The power consumption profiles followed the exact behavior of the suction and discharge temperatures in Figure 8. The MaxTECH unit used more energy than the baseline unit used in both stints of operation. This difference is reflected in the discharge temperatures, as seen in Figure 8.

The comparative test between the baseline and MaxTECH units is conducted to assess the prototype’s performance, which is operating with 275 g of R290. The significant advantage of the prototype is the ability to perform satisfactorily with the reduced charge. The unit used in the study has a 50 gal storage tank, an appropriate size for a family of four or fewer members. Therefore, small to medium hot water usage is selected for the test. The average COPs for the baseline and MaxTECH are 2.67 and 2.31, respectively. The COP is found to be lower than the baseline unit with R134a. A plausible explanation for this discrepancy is that the baseline is designed and optimized for a 50 gal tank serving a family with medium hot water usage. Performance is affected because of the system running with a lower heating load. This is evident for individual COPs for the tested days; the superiority of the baseline unit with R134a over the prototype is apparent. It is important to remember that the prototype has been built by only directly replacing the compressor and adjusting the condenser design. The evaporator coil and the control module are unchanged. The electronic expansion valve setup is changed to keep the superheat around 10 °F like the baseline. In addition, system optimization can have a noticeable effect on system performance enhancement. Finally, the MaxTECH unit is not optimized as a whole system, which indicates the potential for higher COP.

5. Conclusions

This study’s MaxTECH prototype is tested against the baseline system. As explained earlier, the modifications are performed in the condenser coils, and the compressor is replaced with an R290 compressor for reduced charge and optimum performance. No changes are made to the other system components, including the controls. Further system optimization is possible; however, the focus was on proving the concept. The prototype’s performance is promising, although it underperforms compared with the baseline unit by a small margin, which was expected by the refrigerant flow in this wrap direction. Prototype performance can be further improved by optimizing the unchanged system components, such as the refrigerant flow direction through the condenser and fan size, in addition to the control’s optimization. Modifications can further reduce the charge to the liquid line and evaporator coil in addition to the control’s optimization. The test objective is to present the potential of deploying R290 in an HPWH with low charge so that performance is still acceptable. The two systems suffered a decline in performance for low-draw patterns because these 50 gal units are considered oversized for a low-draw pattern (which typically requires only 32–33 Gallon/day). The two units have similar suction and discharge temperatures. The running time is comparable overall for both units for most days. However, there is a noticeable time difference in fewer days without an evident trend. In subsequent testing stages, the prototype will be further optimized for R290 refrigerants, including using a different evaporator coil design and customized controls. The findings clearly highlight the potential of using R290 as a refrigerant and the need for system optimization to harness the full performance enhancement. The optimization should include component design and system controls.