Control of CO2 evaporation in an integrated photovoltaic module: experiments and modelling

The use of environmentally friendly refrigerants and the improvement of the system performance are two main topics in the research on heat pump technology. Given the recent limitations imposed on the usage of high global warming potential refrigerants, there is a growing demand in the market for the adoption of natural refrigerants, like CO2. Nevertheless, CO2 heat pumps operate under transcritical conditions, which can adversely affect its efficiency. Therefore, incorporating a solar heat source into a CO2 heat pump can offer a solution to mitigate low system performance issues. In the present study, a photovoltaic module integrated with the evaporator of a CO2 heat pump is experimentally studied. The PV-T evaporator exploits solar radiation to both generate electricity in the PV cells and ensure the evaporation of CO2 in the collector’s tubes. Simultaneously, the PV conversion efficiency is improved by the cooling effect of the evaporation. The present evaporator works in dry expansion mode, thus the refrigerant flow after the expansion device is sent to the solar collectors, where it evaporates before returning to the compressor. When using a PV-T evaporator, it is necessary to prevent superheating in the evaporator to keep a uniform and efficient cooling of the PV cells. The current heat pump design prevents the occurrence of superheated vapor at the PV-T evaporator’s outlet. Beside the experimental activity a dynamic numerical model of the system has been developed and validated.


Introduction
The use of CO2 as refrigerant in heat pump systems is growing due to the recent restrictions imposed by many countries and organizations such as the Regulation No 517 of the European Union of 2014 [1] and the Kigali Amendment of the United Nations of 2016 [2], that force a phasedown of hydrofluorocarbons to reduce the emissions of greenhouse gases.CO2 is an environmentally friendly refrigerant with a low global warming potential (GWP100yr=1) and zero ozone-depleting potential (ODP=0), it is non-toxic and non-flammable (ASHRAE class A1).However, due to its low critical point and higher operating pressure compared to synthetic refrigerants, the design of CO2 systems should account for these issues [3].An interesting solution to increase the efficiency of CO2 heat pumps is to exploit solar radiation as a low-temperature thermal source [4].In direct solar-assisted heat pumps (DX-SAHPs) the evaporator is a solar collector, which converts the incident radiation into heat, which is then exchanged with the refrigerant.In the literature, few experimental studies on direct solar-assisted heat pumps (DX-SAHPs) working with CO2 as the refrigerant are available.Duarte et al. [5] analyzed the performance of a CO2 heat pump with a sheet-and-tube solar thermal collector as evaporator.The surface temperature of the solar evaporator was experimentally measured with an infrared camera.The results showed that when the solar radiation is high, a high degree of superheat is observed at the collector's outlet and the performance of the heat pump is penalized by reduction in the evaporating area [6].In a previous numerical work on the same heat pump system, Paulino et al. [7] suggested the use of an electronic expansion valve to control the refrigerant mass flow rate at the evaporator inlet, and so the value of the superheating at the collector's outlet.
Less works investigate CO2 heat pumps with photovoltaic-thermal (PV-T) collector as direct expansion evaporator.Zanetti et al. [8] demonstrate that a solar assisted heat pump with a direct PV-T evaporator can reach higher performance in terms of COP compared to the air source.Paradis et al. [9] numerically analyzed the performance of a new geometry of PV/T solar collector composed by monocrystalline PV cells coupled on the back by an absorber plate with a stainless steel serpentine tube.The authors showed that in this geometry, used as evaporator in a CO2 heat pump, the electrical efficiency can increase of about 2% compared to a standard solar PV.
The presence of superheated vapor in the solar evaporator is much more disadvantageous in a PV-T collector.In this case, both a decrease in the collector efficiency and photovoltaic efficiency is detected, because the refrigerant cannot effectively cool down the solar cells.In the literature, the problem of superheating control in PV-T evaporators of DX-SAHP is limitedly addressed.The present paper investigates the control of the evaporation pressure in a solar-assisted heat pump with CO2 as refrigerant and PV-T evaporators.The layout of the refrigerant loop has been designed to avoid superheated vapor at the collectors' outlet and thus to operate at the optimal possible evaporation temperature.The heat pump has been studied both experimentally and numerically, through a dynamic model of the system components.After the validation, the dynamic model has been used to evaluate the transient performance and control of the heat pump.

Heat pump prototype and sensors
The prototype presented in this work is a solar-assisted heat pump working with CO2 as the refrigerant.The unit is installed at the Department of Industrial Engineering, University of Padova, Italy (45° 24′ 23′′ N, 11° 52 40′′E).A picture of the prototype is reported in Figure 1a.The heat pump is equipped with three photovoltaic-thermal (PV-T) solar collectors which work as the evaporator of the system.The schematic of the refrigerant loop is reported in Figure 1b.The refrigerant operates a transcritical cycle, therefore after the inverter driven compressor (COMP), the CO2 is sent to a brazed plate gas-cooler (GC), where hot water on the secondary loop is produced.The cooled refrigerant is then sent to an internal heat exchanger (INT) where its temperature is further lowered, and then it enters an electronic expansion valve (EEV).After the expansion device, the refrigerant is sent to the three PV-T evaporators.The vaporized CO2 obtained at the outlet of the evaporator enters a receiver where the liquid and vapor phases are separated.The vapor phase is then extracted at the top of the receiver and sent back to the internal heat exchanger, where it is heated up before entering the compressor.
Each PV-T collector is composed by a sheet-and-tube heat exchanger coupled to a polycrystalline module (270 W nominal power).The absorber of the sheet-and-tube heat exchanger is an aluminum plate and it has been coupled to a fifteen tubes copper serpentine (pitch 8 mm).The internal diameter of the copper tubes is equal to 6 mm and the thickness is equal to 1 mm. Figure 1b) also reports the main sensors installed on the heat pump.Various temperature and pressure sensors have been positioned at the inlet/outlet of the above-mentioned components (on the refrigerant side).The temperature and mass flow rate of the water at the inlet/outlet of the gas-cooler are also measured.In addition, the air temperature and solar irradiance (positioned on the same plane of the solar evaporators) are recorded.Finally, the power consumption of the compressor is measured by means of a power analyzer.

Data reduction
Experimental tests have been performed both in steady-state and dynamic conditions.The collected data have been used to calculate the main heat pump performance indicators.On the water side, the heat flow rate at the gas-cooler is evaluated considering the water temperature increase (∆)  : a) b) Figure 1.Schematic of the solar-assisted heat pump, including the main sensors ("T" temperature sensor, "P" pressure sensor, "CFM" Coriolis flow meter) installed on the refrigerant and water loops.
On the refrigerant side, the values of the specific enthalpy ℎ at the inlet/outlet of the main components where the refrigerant is found in single-phase conditions (ℎ 1 , ℎ 2 , ℎ 3 , ℎ 4 ) are calculated with Refprop [10] using the measured values of pressure and temperature.
In addition, the evaporation temperature   can be determined with Refprop, by using the measured value of the low-pressure ( 6 ).The superheating at the outlet of the evaporator can be evaluated as:  =  6 −   (2) Finally, the coefficient of performance of the heat pump  is calculated with the ratio between the heating capacity   and compressor power consumption :  =   / (3) All the measured parameters are acquired with a sampling rate of 10 s through a data logger Agilent 34970-A and saved through LABView software.Each steady-state point is calculated as an average value considering a time interval equal to 5 minutes.

Low-pressure control
The control system of the present heat pump manages to fix the high-pressure level to the desired value by automatically varying the aperture of the expansion valve.Therefore, there is no active device in charge of the control of the low-pressure.However, from Figure 1b, it is possible to understand that, if the liquid level inside the low-pressure receiver is constant, the refrigerant at the outlet of the receiver is always found as saturated vapor (specific enthalpy equal to ℎ  ), since it is extracted from the top: The hypothesis of constant level has been verified experimentally by checking the position of the liquid level inside the receiver through sight glasses.Furthermore, the value of the specific enthalpy in 7 (ℎ 7 ) can be indirectly calculated by applying the energy balance to the refrigerant loop: where the two terms on the left-hand side represent the specific energy fluxes at the gas-cooler and compressor respectively, whereas the term on the right-hand side represents the specific heat flux absorbed at the low-pressure side.From eq. ( 5): The vapor quality  7 can be calculated through Refprop, at the corresponding pressure.The calculated values of  7 are reported in Figure 2 for all the experimental tests conducted under steadystate conditions at various compressor speeds and environmental conditions.It can be observed that as hypothesized in eq (4), the vapor at the outlet of the receiver is always between 0.94 and 1.

Figure 2.
Vapor quality at the outlet of the receiver (state 7 in Figure 1), for the tests realized in steady-state conditions.This result is of particular interest because, under the above hypothesis (null variation of the liquid level inside the receiver), it can also be concluded that, in steady-state conditions, the enthalpy at the outlet of the evaporator ℎ 6 is equal to ℎ 7 (thus,  6 =  7 ≈ 1).This means that the PV-T evaporators are always forced to completely evaporate the CO2 mass flow rate until saturated conditions: the lowpressure of the system is thus passively regulated to complete the evaporation process.This pressure control is very efficient, since it provides the highest evaporation temperature at each operative condition and allows the refrigerant to flow at constant temperature in the collectors, providing a uniform cooling of the cells.In fact, a higher value of evaporation temperature would cause an incomplete evaporation process due to the higher heat losses towards the environment, whereas a lower evaporation temperature would cause the presence of superheated vapor in the evaporator tubes.This finding (no presence of superheating at the outlet of the evaporators) has been confirmed also through the temperature measurements obtained with the thermocouples placed on the outlet tube of the PV-T collectors.Based on the data collected by these thermocouples, the difference between the measured temperatures and the saturation temperature was always within ± 1.0 K.

Heat pump performance
In this section, the experimental performance of the heat pump operating in steady-state conditions is presented.The efficiency of a heat pump is strictly related to the environmental conditions and in the case of a solar-assisted heat pump, solar radiation plays a significant role.Figure 3 reports the variation of the evaporation temperature for different values of solar irradiance.Experimental data have been recorded when the compressor was working at 50% of the maximum speed and the high pressure was equal to 80 bar.Furthermore, the temperature difference of the secondary water across the gas-cooler (∆)  was maintained equal to 5 K and the inlet temperature equal to 30 °C.All the experimental points have been collected when the air temperature was between 7 °C and 7.5 °C, therefore the effect of air temperature is negligible when comparing the performance at different irradiance values.Figure 3  that solar irradiance has a positive effect on the evaporation temperature, which increases linearly with the radiation showing a growth rate equal to 0.8 K each 100 W m -2 .Figure 4 reports the variation of the COP with the evaporation temperature obtained in steady-state conditions.The experimental points are referred to data collected when the high pressure was fixed at 80 bar and the water inlet temperature at the gas-cooler was equal to 30 °C, at different environmental conditions.It can be observed that the COP displays an increasing trend with the evaporation temperature.This effect is due to the lower pressure ratio, which decreases the compression work.The effect of the different compressor speeds is also reported: the coefficient of performance and evaporation temperature are lower at high compressor velocities.In fact, when increasing the compressor speed, the refrigerant mass flow rate increases and, since the energy flux incident on the panels is fixed, the evaporation of the CO2 is guaranteed by decreasing the evaporation temperature.

Model description
A dynamic model of the solar assisted heat pump has been developed in Matlab environment.The model is capable to solve the CO2 transcritical cycle when the velocity of the compressor, the high-pressure, the inlet condition of the water and the environmental conditions are selected.For each time step, the  model outputs are the main energy fluxes and the thermodynamic conditions of the refrigerant along the loop.More details about the model can be found in [11].In this work, the equation used for the modelling of the PV-T evaporators and of the receiver are presented, since they represent the primary components for regulating the system.
The PV-T evaporator is discretized into the three main layers of the PV sandwich, namely the glass ("g"), the PV cells ("pv") and the absorber plate ("abs").For each layer of thickness  and density , the energy conservation equation can be written considering the energy fluxes between the various layers for an infinitesimal area  =  (see Figure 5).
In eq. ( 7) to ( 9), the last term represents heat flux due to conduction with the adjacent discretization.In eq. ( 8),   is the specific electrical power generated by the PV cell.In eq. ( 9),   is the heat flux provided to the fluid flowing in the evaporator tubes: Where  is the global heat transfer coefficient between the plate and the fluid bulk (see [11]), ̇ is the mass flow rate processed by the compressor and ℎ  is the effective enthalpy drop of the refrigerant occurring in the tubes positioned under the considered discretization.
Where the pedix "l" refers to the liquid phase,   is the refrigerant mass inside the receiver,   is the receiver volume and  is the total enthalpy.
It can be noted that, as reported in Section 3.1, under steady-state conditions, eq (12) yields ℎ 6 = ℎ  (i.e., saturated vapor at the outlet of the evaporator).On the contrary, during transient conditions, the presence of a liquid/vapor mixture or superheated vapor can be found.

Model validation and results
The numerical model has been used to simulate the dynamic behavior of the solar assisted heat pump.
Simulations have been run considering real operating conditions as reference.Figure 6 reports the main parameters recorded during an experimental test performed in dynamic conditions.The heat pump was working at 40% of the maximum compressor speed and high-pressure equal to 80 bar for several hours.Figure 6a shows the environmental conditions during 2 hours of that experimental test.It can be observed that the air temperature was between 14 °C and 17 °C whereas the solar irradiance was characterized by high variations due to the passage of clouds.The evaporation temperature at the beginning of the test was around 11 °C since before 12:50 there were clear sky conditions.At 13:15 the solar irradiance decreased from 1000 W m -2 to 300 W m -2 and the evaporation temperature reached 4°C.At 13:30 the sky was completely covered by clouds and the solar irradiance was below 200 W m -2 until 14:15.During this time interval, the evaporation temperature decreased, and the minimum value was equal to -1.9 °C.When the solar irradiance rose, the evaporation temperature increased accordingly.The response of the heat pump is not immediate, due to the thermal inertia of the PV-T evaporator.This is also confirmed by the numerical model, which is capable to reproduce the evolution of evaporation temperature with a mean deviation with respect to the experimental data equal to 0.63 K. Figure 6b displays the vapor superheating at the evaporator's outlet during the same test.It can be observed that the experimental superheating was below 1 K in the time intervals characterized by a decrease in the evaporation temperature (low irradiance).On the contrary, when the solar irradiance increased sharply, some superheating was detected.However, the superheating conditions at the outlet of the PV-T collectors were reached only for limited time intervals, until the regulation of the The numerical model is capable to reproduce when superheating occurs, but it overestimates its value, with a maximum deviation respect to the experimental data equal to 8 K.

Conclusions
In the present work, the performance and control of a dry expansion solar evaporator of a solar-assisted heat pump are presented.The heat pump prototype works with CO2 as refrigerant and is equipped with three PV-T solar collectors that work as the evaporator.The system operates in dry expansion mode, meaning that the flow of CO2 to the PV-T evaporators is controlled by the throttling valve.The experimental data show that, under steady-state conditions, the vapor quality at the evaporators' outlet was in a range between 0.94 and 1.The present configuration efficiently maximizes the evaporation temperature by fully exploiting all the available surface area for the evaporation of the CO2 flow rate and providing a uniform cooling of the PV cells.The evaporation temperature has been found to increase linearly with the solar irradiance when all parameters are constant, and the increase rate, in the present operative conditions, is equal to 0.8 K per 100 W m -2 .The numerical model presented here can simulate the dynamic operation of the system.The model was found to predict with good accuracy the response of the heat pump during a partly cloudy day: it was further demonstrated that superheating at the outlet of the evaporator is found only in transient conditions when solar irradiance varies sharply.

Figure 3 .
Figure 3. Evaporation temperature versus solar irradiance.Operating conditions: compressor speed equal to 50%, high pressure equal to 80 bar and air temperature between 7 °C and 7.5 °C.

Figure 4 .
Figure 4. Experimental values of the coefficient of performance versus evaporation temperature obtained during steady-state conditions.The points refer to tests realized at fixed high pressure equal to 80 bar and water inlet temperature at the gas-cooler equal to 30 °C.

Figure 5 .
Figure 5. Schematic of the main energy fluxes considered in the modelling of the PV-T evaporator.

Figure 6 .
Comparison between the experimental and numerical results obtained in dynamic conditions during a partly cloudy day: (a) air temperature, solar irradiance and evaporation temperature versus time; (b) superheating at the evaporator's outlet. shows