Air-to-water heat pump assessment: Part 1 – Experimental setup

This paper is part 1 of the investigation on the exergetic and exergoeconomic parameters of an existing system with an air-to-water heat pump unit as a heat source. Part 1 presents the used experimental setup. The main aim of the conducted experimental tests is to develop a model of produced heat transfer rate and energetic COP at different ambient conditions. The obtained data is used in Part 2 of the study where the exergetic and exergoeconomic assessment is carried out. The performance of the considered system is evaluated using Seasonal Exergy Efficiency. Moreover, Part 2 of the study has presented the formulation of the cost of the product, and cost allocation within the heat pump unit based on exergy.


Introduction
In the reference literature, there is a wide discussion about the performance parameters of air-to-water heat pumps (AWHP). In many studies is conducted experimental and theoretical research on the factors that affected the coefficient of performance (COP) of an aerothermal heat pump and consumed electrical energy. For example, Zhao et al. [1] conduct the experimental study of an aerothermal heat pump and they establish the influence of the water temperature in the tank on the observed COP of the heat pump when the external air temperature remains almost constant.
Morrison et al. [2] determined experimentally the consumed electrical energy and instantaneous heat capacity for the considered in cited investigation types of the heat pump system. The experiment is carried out for a range of ambient temperature and relative humidity conditions and the aim of the study is to demonstrate the procedure for seasonal performance rating of air-source heat pump water heaters. Ji et al. [3] conduct a laboratory test of an AWHP with the utilization of the rejected heat in cooling mode. The experiment is conducted to validate the developed model of the system's performance. Another part of studies in reference literature deals with experimental and theoretical research of novel airto-water heat pump systems [4][5][6].
While many authors are conducting experiments to establish specific dependencies, which characterized air-to-water heat pumps, little is known about necessary output data for the conduction of dynamic exergy and exergoeconomic analysis of a real aerothermal heat pump system. Ozturk et al. [7] investigated experimentally an AWHP from the exergy point of view. The objective of the experimental test is to collect the data for the thermodynamic properties of the working fluids of the system at specific states and to evaluate the exergy destruction of the system's components at varying ambient conditions.
In this regard, the current paper suggests the simplified method for experimental research of an existing AWHP system, considering the system as a black-box with defined inputs and outputs. Thus, the used equipment, instruments, and realized measurements are minimized, because the parameters at the internal state of the refrigerant are not necessary to know. The study aims to establish the dependencies between specific system performance parameters (produced heat transfer rate and energetic COP) and dynamic changing ambient conditions. The obtained data is used in Part 2 of the study where the exergetinc and exergoeconomic assessment is carried out.

Air-to-water heat pump experimental setup
To determine the electrical energy input to the AWHP unit and produced heat transfer rate at different ambient conditions, an air-to-water heat pump experimental platform is developed. The experimental setup consists of instruments for the AWHP operating parameters data acquisition and displaying in real-time. The measurements were implemented in the period 09.02-22.02.2020. The general view of the system and schemes of the AWHP experimental setup are shown in Figure 1, 2 and 3, respectively.

Description of the equipment and instruments for the AWHP operating parameters data acquisition and displaying in real-time
The AWHP system ( Figure 2) is an existing system located in Varna Technical University, Bulgaria -at the laboratory of the Heat Technology Department. The AWHP unit is split type, model PUHZ-SW75VHA(-BS)\ERSC-MEC (Mitsubishi). The heat pump unit is the inverter-driven, reversible type and with a variable outlet (according to the classification of EN 14825:2019). The outdoor (OU) and the ERSC-MEC indoor (IU) units are mounted on both sides of the laboratory's eastern facade. The refrigerant is R410A. Тhe heat transfer medium is a 30% water solution of propylene glycol.  In addition, the AWHP system consists of four fan coil units with a nominal heat transfer rate of 5.7 kW (i.e., at the nominal value of logarithmic meant temperature difference ΔTnom = 50 K or at 75ºC/65ºC/20ºC). Figure 2 presented the location of the all of the used digital thermometers and equipment. The wire digital thermometers provide from 9-bit to 12-bit Celsius temperature measurements. T1 and T2 measure the temperature of the glycol solution inlet and outlet to the indoor unit (i.e., they measure the temperatures tw,e and tw,i, respectively). The digital thermometers Т1 and Т2 and are mounted on the supply and return pipe -before the buffer tank (BT). The digital thermometers T6 and T7 quantify the temperature of refrigerant flows. The scheme of the observed AWHP unit and location of the digital thermometers T1, T2, T6, and T7 is presented in Figure 3. The indoor air temperature is measured in two points: near to the wall (T5) and into the room -through combined humidity and temperature sensor (T4, H4). The outdoor air digital thermometer (T8 = te) is located on the eastern facade of the room, and it is protected from direct solar radiation and wind.
The flow rate of the glycol solution is measured by a single jet, dry dial liquids flow meter with a Hall effect sensor. The flow meter is velocity type, with the impeller mounted in radial position within the body of the flowmeter, whereas the incidence of the liquid flow is tangential. The liquids flow meter is mounted in the circuit of the indoor unit and buffer tank, on the supply water pipe of the system -in a horizontal position. The maximum pressure loss of the liquids flow meter is 0.63 bar (i.e., Δp 63) at volumetric flow rate with value within the interval of the minimal (Q1 =40 l/h) and permanent flow rate (Q3 =4 m 3 /h). Therefore, the measuring range of the liquids flow meter is Q3 / Q1 = 100 (R100-H). The declared maximum permissible error (MPE) on volumes between the minimum flow rate (Q1) and the transitional flow rate (Q2 = 64 l/h) is ±5%, whereas on volumes between the transitional flow rate (Q2 =64 l/h) and the overload flow rate (Q4 = 5 m 3 /h) the declared MPE is ±3%. The temperature range of the flow meter is 30÷90ºC.
The consumed electrical energy from the AWHP unit and the auxiliary elements is measured through the static electric meter. It is designed for the measurement and registration of active energy in single-phase two-wire networks for direct connection. The rated voltage of the electric meter is 230V and the basic (reference) current is Ib = 10 A. The electric meter has a built-in LED indicator, reproduces a pulse with a width of 40ms when it read 1Wh electrical energy. The electrical pulse enters the instruments for data acquisition and displays in real-time.
The accuracy and tolerance of the used types of equipment are summarized in Table 1. The system for data acquisition and displaying in real-time is based on the Arduino UNO platform, consisting of an 8-bit AVR microcontroller. A schematic diagram of the system is shown in Figure 4. The system consists of the following additional elements: -SD and Real-Time Clock -this module can record data on an SD card, while the clock acquisition data about time and date; -16x2 LCD display and keyboard -the module is used for displaying the data in realtime. It allows the selection of visualization of another operation parameter using the keys on the keyboard. The measured parameters were recorded every 1 minute. The ultimate operating parameters are the average statistics for the 30 minutes.

Identifying the outliers
In the current paper, for outliers filtering is used Chauvenet's criterion. To identifying the outliers, first, the data is grouped according to the outside air temperature, te. Within the i-th external air temperature bin, te±0.5ᵒC, it is observed n measurements of xi (i = 1, 2,..., n) that have a mean, x , and sample standard deviation, S. From point of view of Chauvenet's criterion, the measurements that should be rejected are those that have the probability of obtaining their deviation from the mean is less than 1/(2n). This statement is expressed by the equation [9]: The maximum acceptable normalized deviation, τmax, for various values of the number of measurements (n) is determined according to the look-up table [10].
In equation (1), the terms mean, x , and sample standard deviation, S, are determined as follow [10]:

Uncertainty analysis
After the outliers were identified and rejected, the uncertainty analysis of the obtained mean values was conducted. The current experimental measurements aim to establish the mathematical models of the produced heat power and energetic COP of the observed AWHP. Consequently, the objective of this analysis is to assess the translation of the uncertainties from the individual measurement into the uncertainty in the final results for Q and COP at an external air temperature bin, te±0.5ᵒC. In the current paper, the data from individual measured variables x1, x2, …, xn are combined to determine the final results for Q and COP. Therefore, the systematic uncertainty of the mean values of the parameter R is determined through the conceptual basis of the root of the sum of the squares [10]: According to the equation (4), the systematic uncertainty of the mean values of the Q and COP is: In equations (5a) and (5b) the partial derivatives / involved because they are not considered as a constant when the temperature of the waterglycol solution vary. The similar approach is used in [11]. It is important to note, that the volumetric flow rate, V, temperatures of the heat transfer medium, Twe and Twi, and electrical power, Wel, were determined from the measurements. The physical properties of the water-glycol solution, such as density, ρ, and specific heat capacity, cp, were taken from a table. Moreover, each partial derivative / i R x   was calculated by using the average values of the measured parameters. The systematic uncertainties of the measurements of the volumetric flow rate (BV), temperature (BTwi and BTwe), and electrical power (BWel) are presented in Table 1. The accuracies of the readings of the cp and ρ are ±1 kJ/(kgK) and ±0.01 kg/m 3 , respectively [11].
The results from the systematic uncertainty estimation of the results for Q and COP are presented in Table 2. The BCOP decreases with the outdoor temperature increasing. Moreover, the consumed electrical power decreasing gives rise to a decline in the systematic uncertainty of COP. A similar result is reported in [11]. Therefore, the partial derivative / el COP W   and accuracy of the electrical energy measurements have a major contribution to the results for BCOP. Contrarily, there is no clear tendency in the variation of BQ. It can be concluded that the systematic uncertainty in the Q is around 12% when the external air temperature is in the range of 2.03ºC to 11.13ºC. The noticeable slight increases in BQ can be explained by the fact that at certain values of the outside air temperatures (for example, te =3.07ºC, te =5.06ºC, te =5.99ºC, and te =7.9ºC) the difference in the temperature of the heat transfer medium in the supply and return pipe decreases due to the slight increase in the flow rate.

MATHEMATICAL MODELS OF THE PRODUCED HEAT POWER AND ENERGETIC COP
As a result of collecting experimental data of the AWHP system, the experimental modeling of the produced heat transfer rate and energetic COP was carried out. These models are used as input data in Part 2 of the current investigation where the exergy and exergoeconomic assessment of the system is conducted. The prediction of the produced heat transfer rate by the AWHP was implemented using linear regression of the experimental data. The measured mean values of the heat output rate were plotted as a function of external air temperature ( Figure 5) and the obtained linear relationship is: To define the operating modes of the considered AWHP system, it is assumed that up to a certain external air temperature, the heat pump unit produces heat output that fully satisfies the requirements for the heating of the building. At this external air temperature (called bivalent temperature), the partial load ratio, PLR = Q/Qmax, of the AWHP unit is equal to 1. Otherwise, at outdoor air temperature equal to the balance temperature, PLR is zero. Then, at PLR = 0 the heat loss of the building is equal to the internal and external heat gains. 1.4 n  -exponent to characterize the type of heat emission system (radiators, convectors, floor heating, etc.) [13]. Figure 6 illustrates the difference between the measured mean values of the temperature of the produced heat transfer medium and the calculated by equation (9) data for tw,e. It was observed 11 test points were defined by the mean value of the external air temperature within a temperature bin te±0.5ᵒC. The average difference between the measured tw,e and the predicted values is around 3.6%. The larger deviations can be noticed at lower outdoor air temperatures. In order to predict accurately energetic COP of the AWHP at partial load operating mode, the following mathematical model is used: where: PLRi = Qi / Qmax,i -partial load ratio at i-th external temperature bin; Z = Wel,i / Wel,max,i -the ratio of the consumed electrical power at part load operating mode and at full load (at the same outlet temperatures of the heat transfer medium, tw,e); COPmax,i -energetic coefficient of performance at i-th external temperature bin and j-th temperature of outlet heat transfer medium. The conducted experimental measurements indicate a linear relationship between PLRi = Qi / Qmax,i and Z = Wel,i / Wel,max,i. The Z was correlated in terms of PLR as shown in equation (11): 0.8332 0.1086 Z P L R    (11) A graphical representation of relationship from equation (11) is presented in Figure 7.

Validation and verification of the obtained mathematical models
The assessment of the adequacy and accuracy of all presented models was performed according to the rules of regression analysis. First, the validity of the obtained regression model proved the high value of the coefficient of determination, R 2 (R 2 = 0.97 for the linear regression of Q and R 2 = 0.98 for Z). On the other hand, the adjusted R 2 values are 96.7% and 97.8% in the case of the relationships of Q and Z, respectively. Therefore, over 96% of the variance of Q and Z can be explained by the variance of the te and PLR, respectively. In conclusion, the obtained mathematical models can be considered statistically significant.
In addition, using the tools of MS Excel is analyzed the variance of the suggested models. The obtained results are presented in Table 3 and Table 4. The results from Tables 3 and 4 designate that the intercepts in equations (6) and (11) are statistically significant at a 95% confidence level. This is evidenced by the results of Pvalues. In both cases, they are significantly lower than the assumed level of significance (2.8 . 10 -12 <<0.05 and 3.9 . 10 -5 <<0.05). Moreover, the slopes of the equations (6) and (11) are also statistically significant at a 0.05 level of significance.
As verification of the obtained models, a comparison of the calculated values for the Q and COP through the measured data with those published by the manufacturer [8] is carried out. The results are shown in Figures 8 and 9. There is a little difference between the values of Q obtained by the experiment and those declared by the manufacturer. These differences can be explained in part by the distinction in conditions at which the heat power was measured by the manufacturer [8]. Another possible explanation for this result is differences in the physical properties of the used heat transfer medium. In [8] manufacturer presents the COP values characterized only by the AWHP, whereas in the current paper are presented the values for COP describe the whole system. Due to the lack of exact information about consumed electricity power by auxiliary equipment (fans, pumps, etc.) in the various operation modes of the system, in Figure 9 the values of COP obtained by the experiment are compared the data for COP of the unit at an outlet temperature of heat transfer medium equal to 45ºC and minimum inverter frequency (15 Hz). Thus, it is possible to assess the convergence of the obtained results. The experimental and declared data are approximated by a linear function. It can be concluded that the COP values determined using the measurement data have good convergence with declared in [8] values of the COP.

Conclusions
In Part 1 of the current investigation, an AWHP system was modeled and the obtained experimental results were evaluated. The objective of the paper was to establish the dependencies between produced power and energetic COP and dynamic changing ambient conditions. The developed mathematical models of Q and COP are used are Part 2 of the investigation.
This study has identified that the system uncertainty value of COP is in interval ± (7.1÷16.4) %, whereas the system uncertainty of Q varying in interval ± (10.98÷12.5) %. However, the conducted regression analyses confirmed the statistical significance of the regression models and their coefficients. In this case, it is not necessary to consider a larger number of measurements. It is proved that it is not necessary to add other parameters to the obtained models.
Moreover, an analytical procedure for the determination of the outlet temperature of the produced heat transfer medium is presented. It has proved that the obtained results about tw,e describe with high accuracy the actual behavior of the AWHP units with variable outlet. Therefore, the COP and tw,e models obtained in the current study for a particular application of the analyzed model AWHP unit can be used again in different applications of the same model heat pump.
The question raised by this study is how to achieve the maximum exergy content of the product of the AWHP system and is there a significant difference in the exergetic and exergoeconomic performance parameters of the AWHP system in different operations modes of the AWHP unit and backup heater. An attempt at a comprehensive answer to these questions is presented in Part 2 of the current investigation.