Comprehensive experimental performance study on a small-capacity transcritical R744 vapour-compression refrigeration unit equipped with an innovative ejector Etude expérimentale complète

Ejector-equipped transcritical R744 condensing units are believed to lead to a low-to-zero commercial refrigeration sector. In order to overcome the persisting barrier to their wider adoption represented by the lack of an affordable ejector control technique, the novel pulse-width modulation (PWM) ejector, being low cost, simple and invulnerable to clogging was recently implemented. However, additional experimental evaluations are needed. Therefore, in this experimental work the performance of two PWM ejector-equipped transcritical R744 condensing units, i.e. with and without overfed evaporator, was carried out. The experimental assessment was implemented at the medium temperature (MT) of about -5 ◦ C, heat sink temperatures from 30 ◦ C to 40 ◦ C and compressor speeds from 40 Hz to 60 Hz. The outcomes obtained revealed that the PWM ejector can effectively control the high pressure in transcritical operating conditions, regardless of the selected heat sink temperature and compressor speed. In addition, at the same cooling capacity, the PWM ejector-equipped R744 system was found to permit energy savings between 7.0% and 11.1% without overfed evaporator and between 11.5% and 16.3% with overfed evaporator compared to the standard R744 unit (i.e. with vapour by-pass valve and without ejector), respectively. Finally, higher values of coefficient of performance (COP) were found to be offered by the PWM ejector compared with its today ’ s available competitors.


Introduction
The refrigeration sector is of vital importance to modern society.However, it is responsible for 7.8% of global greenhouse gas (GHG) emissions, a figure bound to increase due to the growing overall refrigeration demand (IIR, 2021).About 63% of its global GHG releases are owing to the leaks of fluorocarbons (i.e.chlorofluorocarbons, hydrochlorofluorocarbons and hydrofluorocarbons) from the refrigeration systems (IIR, 2021).A recent report highlighted the need for immediate, rapid and large-scale actions to decrease GHG emissions to avoid crossing the global warming level of 1.5 • C in the next years (IPCC, 2021).Therefore, the EU F-Gas Regulation 517/2014 (European Parliament, 2014) and the Kigali Amendment to the Montreal Protocol (United Nations Environment Program, 2017), among others, are leading to a massive phase-out of hydrofluorocarbons (HFCs).The refrigeration systems (i.e.condensing units) operating in small-format retail outlets (e.g.grocery, discount and convenience stores) are being particularly affected by HFC gradual abandonment due to the need for an ultra-low global warming potential (GWP) and zero ozone depletion potential (ODP) refrigerant.Also, since the public can easily access the condensing units and their refrigerant charge can be relatively significant, the focus is on safe (i.e.non-flammable and non-toxic) working fluids.
Carbon dioxide is a safe (ASHRAE classification = A1), eco-friendly (GWP = 1, ODP = 0) and inexpensive refrigerant (indicated as R744), which presents favourable thermo-physical properties (Lorentzen, 1994;Kim et al., 2004) and energy efficiencies in heating mode (Dai et al., 2020;Smitt et al., 2021).However, the low critical temperature of R744 results in considerable energy consumptions on the part of CO 2 refrigeration systems at high heat sink temperatures, i.e. in transcritical regime (Cavallini, A., Zilio, C., 2007;Sawalha et al., 2017).The need for highly performing transcritical R744 condensing units is particularly needed in small-format stores, as their refrigeration systems are accountable for 30% to 60% of the total required electricity (Tassou et al., 2011).It is worth mentioning that the annual electricity demand (per sales area) of grocery, discount and convenience retail outlets is approximately three times higher than that of supermarkets and hypermarkets (Tassou et al., 2011).
Two-phase ejectors for expansion work recovery purposes have been proven to be the key to implement highly performing transcritical R744 vapour-compression cooling systems (Elbel and Lawrence, 2016;Gullo et al., 2018;Ringstad et al., 2020).The experimental work by Lee et al. (2014) revealed that the COP and the cooling capacity are enhanced by 6% and 11% in transcritical CO 2 air conditioners, respectively.Lucas and Koehler (2011) and Nakagawa et al. (2011) measured COP increases of up to 17% and 27% in a CO 2 refrigeration system, respectively.Ozaki et al. (2004) and Haida et al. (2016) experimentally evaluated a COP improvement of up to 20% in a mobile CO 2 air conditioning system and 7% in a CO 2 refrigeration unit, respectively.The experimental study by Haida et al. (2021) revealed that the most effective control strategy of the cooling load is obtained with a pressure lift (pressure difference between diffuser outlet and suction inlet) of 4 bar in subcritical regime and 6 bar at transcritical unsteady operating conditions.The experimental data collected by Elbarghthi et al. (2021) proved that up to 36.9% of the available work rate can be recovered as well as a maximum pressure lift of 9.51 bar can be attained in supermarket applications.As for food retail stores, Hafner et al. (2014) estimated that the use of ejectors implies COP improvements of 17% in Athens (Greece), 16% in Frankfurt (Germany) and 5% in Trondheim (Norway).Compared to parallel compression, energy savings between 15% and 30% at outdoor temperatures from 22 • C to 35 • C were measured in a supermarket located in Italy (Hafner et al., 2016).Expósito-Carrillo et al. (2021) evaluated COP enhancements of up to 13% thanks to the ejector in a supermarket refrigeration system operating at the outdoor temperature of 25 • C. Azzolin et al. (2021) proved with the aid of field measurements that the annual average COP can be augmented by up to 18% by coupling overfed evaporators, ejectors and evaporative gas cooler in a food retail store located in Tunis (Tunisia) and Milan (Italy).The advanced exergy analysis carried out by Gullo (2021) highlighted the need for an ejector in a transcritical R744 refrigeration supermarket system.The recent experimental work implemented by Singh et al. (2020Singh et al. ( , S. 2021) demonstrated the advantageous use of ejector-equipped CO 2 refrigeration supermarket systems in the Indian climate context.Bodys et al. (2018Bodys et al. ( , J. 2021) ) proposed the adoption of a modulated opening of the suction nozzle bypass duct to enhance the performance of R744 ejectors installed in large-capacity applications.Ameur and Aidoun (2021) showed that the use of a two-phase ejector allows a transcritical CO 2 heat pump unit to increase the heating capacity and the COP by 14% and 9% at the evaporator temperature of − 20 • C, respectively.However, the current lack of a cost-effective methodology for an appropriate ejector capacity control is substantially breaking the diffusion of ejector-equipped transcritical R744 refrigeration systems in small-capacity applications (e.g.condensing units) (Gullo et al., 2020).To the best of the authors' knowledge, at present three ejector capacity control methodologies can be adopted in small-capacity solutions.
The first methodology involves the needle-based ejector in which the refrigerant flow is suitably modulated by modifying the throat crosssectional area of the ejector motive nozzle with the aid of a needle (Elbel and Hrnjak, 2008;Liu and Groll, 2008).Elbel and Hrnjak (2008)  ρ P. Gullo et al. experimentally showed that the use of such an ejector allows increasing the COP and the cooling capacity by 7% and 8%, respectively.Xu et al. (2012) measured ejector efficiencies from 0.2 to 0.3 in an ejector-equipped CO 2 heat pump unit.The results obtained in the numerical work by He et al. (2019) revealed 5% to 11% lower mass entrainment ratios (ratio of suction nozzle mass flow rate to motive nozzle mass flow rate) and similar exergy efficiency as contrasted to the passive ejector (i.e.without flow modulation mechanism).Liu and Groll (2013) experimentally experienced ejector motive nozzle, suction nozzle and mixing section efficiencies of 0.50-0.93,0.37-0.90and 0.50-1.00,respectively.As for simultaneous cooling and heating applications, Liu et al. (2016) measured an enhancement in COP of 71.4% as well as a decrease in system capacity of 21.3%.Liu et al. (2012) evaluated COP and cooling capacity improvements by about 30.7% and 32.1% over the basic CO 2 system, respectively.However, the needle-based ejector features considerable vulnerability to clogging (Lawrence and Elbel, 2019;Zhu and Elbel, 2020) as well as a complicated and expensive design (Lawrence and Elbel, 2019;Zhu and Elbel, 2020).
The second methodology includes the vortex-based ejector in which the refrigerant flow enters the ejector motive nozzle both axially ( ṁR744,axial ) and tangentially ( ṁR744,tangential ) to generate a vortex.Therefore, the ejector capacity is suitably controlled by varying the vortex strength (i.e.vortex strength = ṁR744,tangential /( ṁR744,axial + ṁR744,tangential )) (Zhu and Elbel, 2020).On the one hand, the vortex-based ejector is less complicated, expensive and vulnerable to clogging compared to the needle-based ejector solution.On the other hand, at present the vortex ejectors perform similarly to or slightly worse than the latter (Zhu and Elbel, 2020).
The third methodology involves a solenoid valve installed upstream of the ejector motive nozzle, which is employed for implementing the pulse-width modulation (PWM) effect to effectively control the refrigerant flow through the ejector (Gullo et al., 2021).At the water temperature at the gas cooler inlet of 35 • C, R744 evaporating temperature of approximately − 5 • C (typical value for condensing units) and compressor speed of 50 Hz, Gullo et al. (2021) experimentally proved the effectiveness of the proposed technique by showing that the high pressure can be increased from around 87 bar to 112 bar.Furthermore, at the optimal running modes COP improvements by more than 5% in comparison with the passive ejector and by more than 10% in comparison with the basic R744 unit (i.e. with flash vapour by-pass valve and without ejector) were assessed.However, Gullo et al. (2021) highlighted the need for a comprehensive experimental performance study in transcritical regime involving: the effect of the compressor speed and water temperature at the gas cooler inlet (i.e.heat sink temperature) on the performance of both the PWM ejector and the overall system; the assessment of the energy benefits provided by overfeeding the evaporator.
Therefore, the goal of this work is to bridge the aforementioned two knowledge gaps by exhaustively investigating the influence of water temperatures at the gas cooler inlet (t water,gc in ) of 30 • C, 35 • C and 40 • C and that of compressor speed values of 40 Hz, 50 Hz and 60 Hz on the performance of both the PWM ejector and the overall system.In addition, the benefits from the evaporator overfeeding have also been evaluated.The whole investigation has been carried out by setting the ethylene glycol-water (35/65%) mixture (EG) temperature at the evaporator inlet (t EG,evap in ) to 5 • C, leading the evaporator (at dryexpansion conditions) to perform at about − 5 • C (typical value for condensing units).In addition, the performance of the R744 system outfitted with the PWM ejector has been contrasted to that of the R744 unit using the flash vapour by-pass valve (i.e.no ejector) as well as to that of the R744 system relying on the passive ejector.
The main characteristics, the advantages and the disadvantages of the available capacity control strategies for two-phase ejectors installed in small-capacity vapour-compression systems are summarized in

Table 1
Summary of the available capacity control strategies for two-phase ejectors installed in small-capacity applications (Gullo et al., 2020).Table 1.
The manuscript is arranged as follows: the test facility, the PWM concept, the test methods and the experimental uncertainty are presented in Section 2, while the results are described and discussed in Section 3. Finally, the conclusions and future developments are summarized in Section 4.

Test facility
The schematic of the test facility in which the whole experimental campaign was carried out is presented in Fig. 1.The experimental apparatus, which can be run in either ejector mode or standard mode (i.e. with flash vapour by-pass valve and without ejector), was outfitted with: • an R744 transcritical semi-hermetic reciprocating compressor with a displacement of 1.12 m 3 ⋅h − 1 at 1450 rpm; • brazed plate heat exchangers (0.492 m 2 of evaporator heat transfer area and 0.328 m 2 of gas cooler heat transfer area); • a suction line heat exchanger (SLHX), which was 50 cm stainless steel tubes tin-soldered together; • stepper-motor expansion valves; • a 2 kW low pressure lift two-phase ejector for expansion work recovery (Table 2); • a liquid level sensor and a 3 L receiver, featuring a sight glass at the bottom and one at the top; • distilled water was used for cooling down R744 discharged by the compressor, whereas EG was employed for removing the heat in the evaporator.
During the whole experimental campaign it was ensured that the refrigerant level was at least covering the bottom sight glass of the receiver (Gullo et al., 2021).The polyalkylene glycol (PAG) oil was employed and a metering valve was used for favouring the return of the lubricant from the liquid outlet of the separator to the vapour outlet (compressor suction line).The experimental facility was also outfitted with a power analyser, nineteen temperatures sensors (fifteen for R744, two for distilled water, two for EG), ten pressure gauges (all for R744), one EG volume flow meter, one water volume flow meter and two mass flow meters for R744.The accuracies and calibration range of the experimental equipment are summarized in Table 3.An internal heat exchanger (IHX), being 100 cm hair-pinned stainless steel tubes tin-soldered together, was installed to avoid the presence of vapour at the inlet of the R744 mass flow meter on the intermediate pressure side.Pressure sensors related to the motive nozzle, suction nozzle and diffuser pressures were mounted directly on the ejector housing.Also, pressures and flow rates were acquired at 1000 Hz and every 0.01 s, whereas temperatures were logged every 0.10 s, respectively.Further details regarding the test facility can be found in Kaern et al. (2018) and Gullo et al. (2021).

Innovative flow modulation technique for R744 ejectors
The innovative flow modulation technique for R744 ejectors implemented by Gullo et al. (2021) was based on the control of the refrigerant flow through the ejector involving the PWM effect, being widely employed for the same purpose in the evaporator expansion valves.The PWM technique applied to an evaporator expansion valve behaves as follows: • within a time period (called either PWM period or cycle time) generally equal to 3-6 s, the valve coil opens as the voltage signal is transmitted (i.e.refrigerant flows through the expansion valve) and closed as the voltage signal is not transmitted (i.e.refrigerant does not flow through the expansion valve); • the required R744 mass flow rate is then given by changing the relation between the opening and closing time of the valve coil.
The PWM methodology was then applied to solenoid valve installed upstream of the ejector motive nozzle (indicated as MSV in Fig. 1) with the aim of varying the on/off state of the ejector and thus controlling the R744 mass flow rate through the motive nozzle.
According to the results presented by Gullo et al. (2021), the OD represented the ratio of the motive nozzle solenoid valve opening time to the PWM period (in percentage).It is worth remarking that Gullo et al. ( 2021) investigated three different PWM periods, i.e. 1 s, 2 s and 3 s.The PWM period was selected equal to 2 s in this study (as recommended by Gullo et al., 2021), as: • a delay in MSV closing was observed during the experimental campaign and it was found to be considerably detrimental to the case relying on PWM period = 1 s (Gullo et al., 2021); • the values of COP and ejector efficiency for the case involving PWM period = 2 s were similar to those based on PWM period = 3 s.However, the former featured lower pressure fluctuations on the high pressure side, resulting in higher ejector and system lifetime (Gullo et al., 2021).
The benefits of PWM ejector can be summarized as follows (Gullo et al., 2021): • the PWM technique is a well-known and reliable flow modulation technique, being widely used in conventional expansion valves thanks to its high simplicity, accuracy and affordability; • the PWM methodology does not involve the use of a needle, resulting in negligible clog issues; • the PWM ejector offers no constraints in terms of manufacturing sizes.

Test methods
The work was carried out by considering three different temperatures of the water entering the gas cooler (t water,gc in ) (i.e.30.0 • C ± 0.5 • C, 35.0 • C ± 0.5 • C and 40 • C ± 0.5 • C) and three compressor speed values (i.e.40 Hz, 50 Hz and 60 Hz).The operating conditions of a condensing unit were reproduced by setting the ethylene glycol-water (35/65%) mixture (EG) temperature at the evaporator inlet (t EG,evap in ) to 5.0 • C ± 0.5 • C (i.e.t evap of about − 5 • C) and the degree of superheating (ΔT superheating ) to 8 K ± 0.5 K during dry-expansion evaporator conditions.During the whole experimental campaign the water mass flow rate was equal to 0.0918 kg⋅s − 1 ± 0.0044 kg⋅s − 1 , while the EG mass flow rate was selected as 0.1152 kg⋅s − 1 ± 0.0004 kg⋅s − 1 , respectively.Steady-state conditions were considered to be attained as all the temperatures varied within 0.2 K over 60 s.Thus, the experimental measurements were acquired for 5 min and averaged over the collection period.The experimental apparatus was controlled and monitored with the aid of Minilog and the experimental measurements were collected via LabView 2016 and National Instruments cDAQ modules.Also, the evaporator overfeeding was implemented by increasing the opening degree (in Minilog) of the expansion valve installed upstream of the heat exchanger (EVOD).In particular, lower values of OVED meant lower R744 mass flow rates through the evaporator and thus higher values of superheating degree (or higher values of quality of R744 leaving the evaporator), whereas higher values of OVED meant higher R744 mass flow rates through the evaporator and thus lower values of superheating degree (or lower values of quality of R744 leaving the evaporator).

Experimental uncertainty
The uncertainty propagation analysis was performed by employing Engineering Equation Solver (EES) (F-Chart, 2020).REFPROP (Lemmon et al., 2018) was used for assessing the thermo-physical properties of R744, EG and water during the data reduction process.The outcomes of the uncertainty propagation assessment are listed in Table 4. Repeatability of data points was shown to be within 5%.The maximum discrepancy in terms of heat balance between R744 and the secondary fluids was found to be within 3% in the gas cooler and within 5% in the evaporator.

Influence of heat sink temperature (t water,gc in )
In this Subsection, the influence of the heat sink temperature (t water,gc in ), high pressure and OD of the motive nozzle solenoid valve (MSV) on the COP is investigated.Three values of t water,gc in , i.e. 30 • C, 35 • C and 40 • C, were selected, while the compressor speed was kept at 50 Hz.In Fig. 2 to Fig. 7 the point at the lowest value of the high pressure was measured at OD of MSV = 100%, whereas that at the highest value of the high pressure was measured at OD of MSV = 60%.In this study OD of MSV was set to 100%, 90%, 80%, 70% and 60% in Fig. 2 to Fig. 7.The results presented in Fig. 2 highlight that the PWM methodology was capable of increasing the high pressure by reducing the OD of the MSV at each selected t water,gc in .In fact, the two-phase ejector was employed as the replacement for a conventional high pressure expansion valve, which is able to increase the high pressure by reducing its OD.Therefore, similarly to the conventional high pressure expansion valve in a standard transcritical R744 system, at a given heat sink temperature the reduction of the OD of the MSV allowed for the decrement in the refrigerant flow rate though the MSV and thus for an increment in the high pressure (due to the lower enthalpy difference through the gas cooler).As an example, at t water,gc in equal to 30 • C the gas cooler pressure could be increased from 87.43 bar (OD = 100%) to 112.36 bar (OD = 60%).The increment in the high pressure resulted in an increase in the compressor power input as well as in the cooling capacity, leading to an optimal high pressure maximizing the COP at each t water,gc in .It is worth remaking that at very high values of the gas cooler pressure the cooling capacity tended to decrease owing to a decrement in the refrigerant mass flow rate as a consequence of a reduction in the compressor volumetric efficiency.In particular, the results obtained revealed that at t water,gc in = 30 • C, the maximum COP value is 1.95 (obtained at P gc = 80.62 bar and OD = 90%), at t water,gc in = 35 • C, the maximum COP value is 1.67 (evaluated at P gc = 89.39bar and OD = 80%) and at t water,gc in = 40 • C, the maximum COP value is 1.41 (assessed at P gc = 96.50bar and OD = 80%).The increase in temperature of R744 leaving the gas cooler with a rise in the heat sink temperatures caused a rise in both high pressure and R744 density.This, in turn, increased the R744 mass flow rate drown by the compressor, while the R744 enthalpy difference between suction and discharge of the compressor rose too.Also, as the water temperature at the gas cooler inlet increased, the heat absorbed as well as R744 mass flow rate through the evaporator decreased.Thus, COP decreased with a rise in water temperature in the gas cooler, as also observed by Lee et al. (2014) and Lawrence and Elbel (2019).
The values of P lift and η ejector are presented in Fig. 3 and Fig. 4 with respect to t water,gc in , high pressure and OD of the MSV, respectively.The results depicted in Fig. 3 can be justified by considering that: • as the OD of the MSV was reduced, the R744 mass flow rate through the ejector motive nozzle decreased and thus thus Φ m increased.Therefore, P lift was found to decrease, as any ejector is able to lift more refrigerant to a lower (diffuser) pressure; • as the heat sink temperatures were raised, the high pressure (and thus the motive nozzle pressure) went up, promoting higher values of available recoverable expansion work and thus higher values of P lift .
The outcomes showed in Fig. 4  Furthermore, the PWM ejector presented increases in η ejector values by 1.9 % at t water,gc in = 30 • C, 0.2 % at t water,gc in = 35 • C and 2.2 % at t water, gc in = 40 • C compared to the passive ejector, respectively.It is worth mentioning that the PWM ejector efficiencies were found to be lower than those measured in a laboratory (Fredslund et al., 2016).Nevertheless, the values represented in Fig. 4 were actually underestimated, as they were averaged over the whole data collection period, i.e. the period  P. Gullo et al. during which MSV was closed over the selected PWM period was included (Gullo et al., 2021).

Influence of compressor speed
In this Subsection, the influence of the compressor speed, high pressure and the OD of the MSV on the COP is studied.Three values of compressor speed, i.e. 40 Hz, 50 Hz and 60 Hz, were taken into account, while t water,gc in was maintained at 35 • C. The outcomes shown in Fig. 5 suggest that, regardless of the selected compressor speed, the PWM ejector could control the high pressure in transcritical running modes effectively by varying the OD of the MSV.In addition, it was found that at the compressor speed of 40 Hz the maximum COP value is 1.77 (obtained at P gc = 86.94bar and OD = 70%), at the compressor speed of 50 Hz the maximum COP value is 1.67 (evaluated at P gc = 89.39bar and OD = 80%) and at the compressor speed of 60 Hz the maximum COP value is 1.59 (assessed at P gc = 91.46bar and OD = 90%).The lower values of COP with respect to the compressor speed was due to the higher gas cooler pressure values at 60 Hz compared to 40 Hz, resulting in larger enthalpy differences between the compressor inlet and outlet.Furthermore, R744 mass flow rate in the compressor rose as a consequence of the increase in speed, which resulted in a growth in the compressor power input.On the one hand, the mass entrainment ratio increased, i.e.R744 mass flow rate through the evaporator grew more than R744 mass flow rate in the compressor.This, in combination with a rise in enthalpy difference through the evaporator, led to an increase in the cooling capacity.On the other hand, the compressor power input increases "faster" than the cooling capacity.As a consequence, COP  P. Gullo et al. decreased with respect to the compressor speed, as also shown by Lee et al. (2014) and Lawrence and Elbel (2019).
The values of P lift and η ejector are presented in Fig. 6 and Fig. 7 as a function of the compressor speed, high pressure and OD of the MSV, respectively.The results presented in Fig. 6 were a consequence of the fact that a reduction in OD meant a decrease in R744 mass flow rate flowing through the motive nozzle and thus an increase in Φ m , resulting in P lift degradation (as explained in Subsection 3.1).As for the outcomes presented in Fig. 7, these were a compromise between the behaviour of P lift and that related to Φ m (as explained in Subsection 3.1).At the optimal high pressure, the values of P lift , Φ m and η ejector were equal to 3.48 bar, 0.546 and 0.202 at the compressor speed of 40 Hz, 4.00 bar, 0.559 and 0.236 at the compressor speed of 50 Hz and 4.46 bar, 0.558 and 0.260 at the compressor speed of 60 Hz.The higher values of P lift , Φ m and η ejector at 60 Hz compared to 40 Hz were due to the fact that the increase in compressor frequency allowed for higher gas cooler pressure values and thus for higher available recoverable expansion work.
To have a clearer picture, the experimental campaign was extended to the scenarios involving compressor speed of 40 Hz and t water,gc in = 30 • C as well as compressor speed of 60 Hz and t water,gc in = 40 • C. At the former scenario a maximum COP value of 2.12 (obtained at P gc = 77.54bar and OD = 80 %) was observed, while at the compressor speed of 60 Hz and t water,gc in = 40   Finally, at all the investigated conditions, the maximum value of COP was obtained as a consequence of the increase in high pressure, which suggested that the ejector without PWM-based control was oversized.

Effect of evaporator overfeeding
In this Subsection, the energy benefits from combining the PWM ejector with the evaporator overfeeding were assessed at the heat sink temperatures (t water,gc in ) of 30 • C, 35 • C and 40 • C and compressor speeds of 40 Hz, 50 Hz and 60 Hz.Firstly, it was observed that the overfeeding of the evaporator did not influence the value of the OD of the MSV leading to the optimal high pressure maximizing the COP.Therefore, all the evaluations involving the evaporator overfeeding were carried out at the same optimal OD of the MSV as for the corresponding cases with dryexpansion evaporator (ΔT superheating = 8 K±0.5 K).As shown in Table 5, in fact, the highest COP was observed at OD of MSV of 80 % consistently with the results involving the dry-expansion evaporator.In addition, the maximum COP for all three scenarios summarized in Table 5 involved the value of EVOD of 40 %.Therefore, ΔT superheating = 8.0 K ± 0.5 K observed respectively at EVOD equal to 31 % at OD of the MSV equal to 90 %, EVOD equal to 32 % at OD of the MSV equal to 80 % and EVOD equal to 35 % at OD of the MSV equal to 70 % disappeared as the EVOD was set to 40 % (i.e.ΔT superheating = 0.0 K ± 0.5 K).This led to a value of the quality of the R744 leaving the evaporator equal to about 1 in all the investigated scenarios, whereas it further decreased by setting EVOD to values lower than 40 %.In particular, an optimal value of EVOD resulting in the highest value of Φ m and thus maximizing the COP at each investigated operating condition was observed.All these results were experienced in the whole experimental campaign involving the evaporator overfeeding, being consistent with the findings by Lawrence and Elbel (2019).As proved by Lawrence and Elbel (2019), in fact, as the dry region was experienced within the evaporator (EVOD around 30 % in this study), lower R744 mass flow rate was going into the evaporator, causing lower values of Φ m (and thus higher values of P lift ) and poorer heat transfer performance.The lower overall heat transfer in the evaporator was due to the dry region, involving a much lower heat transfer coefficient compared to the vaporization region.On the contrary, the evaporator overfeeding (EVOD = 40 % in this study) led to higher values of R744 mass flow rate through the evaporator and thus higher values of Φ m (resulting in lower values of P lift ) and greater heat transfer performance.The better overall heat transfer coefficient of the evaporator was due to the disappearance of the dry region, translating into a further increase in COP.For values of EVOD above 40 % (i.e.values of quality of R744 leaving the evaporator below 1) the COP was found to be lower due to the high R744 mass flow rate within the evaporator, leading to a decrement in P lift while the overall heat transfer of the overfed evaporator did not vary considerably.
In general, the evaporator overfeeding caused an increase in COP, cooling capacity and entrainment ratio at the expanse of the pressure lift, as shown in Table 6 at the operating conditions maximizing the COP.The lack of evaporator overfeeding at t water,gc in = 30 • C and compressor speed of 40 Hz resulted in a COP penalization by 4%, while the COP penalization at t water,gc in = 40 • C and compressor speed of 60 Hz was equal to 2%.

PWM ejector vs. passive ejector vs. standard mode
In this Subsection, the performance of the standard refrigeration unit (i.e. with vapour by-pass valve and without ejector) and that of the PWM ejector-equipped refrigeration system with and without overfed evaporator are compared.The results of this comparison are summarized in Table 7 at the operating conditions maximizing the COP (except for the passive ejector).At the investigated operating conditions compared to the standard system it was observed that:

Table 5
Effect of opening degree of evaporator expansion valve (EVOD) and opening degree (OD) of the motive solenoid valve (MSV) on COP (compressor speed = 50 Hz, t water,gc in = 35 • C, t eg,evap in = 5 • C, PWM period = 2 s).• the passive ejector featured COP improvements between 1.2% and 3.6% at t water,gc in = 30  2020) revealed that at present the needle-based ejector and the vortex ejector (including evaporator overfeeding) can increase the COP by up to 4% compared to the passive ejector.Therefore, since the PWM ejector was found to offer COP enhancements by up to 11.8% without overfed evaporator and 15.1% with overfed evaporator over the passive ejector, the PWM ejector seems to offer higher energy advantages of compared to its today's competitors.In addition, the results listed in Table 7 also highlight that the cooling capacity could be increased from 0.5% to 14.3% by the passive ejector and by from 5.8% to 16.3% by the PWM ejector compared to the standard system.In addition, the evaporator overfeeding allowed for improvements in cooling capacity from 7.6% to 20.3% in comparison with the standard system.It is worth mentioning that the value of cooling capacity related to the passive ejector operating at t water,gc in = 35 • C and 40 Hz (Table 7) derived from the experimental uncertainty propagation as well as from the R744 system being not operating at optimum running modes.Finally, it is worth mentioning that at the same cooling capacity the PWM ejector-equipped R744 system without overfed evaporator offered energy savings between 7.0% (at t water,gc in = 30 • C) and 11.1% (at t water, gc in = 40 • C) compared to the standard R744 unit.As the evaporator overfeeding was considered, the PWM ejector-equipped R744 system with overfed evaporator was found to decrease the energy consumption by 11.5% (at t water,gc in = 30 • C) and 16.3% (at t water,gc in = 40 • C).In addition, at t water,gc in = 35 • C and same cooling capacity a reduction in the volume flow rate drawn by the compressor of the PWM ejector-equipped R744 system with overfed evaporator by about 9% was evaluated.From a cost perspective, this compressor size reduction and the aforementioned growth in cooling capacity can promote the adoption of commercial transcritical R744 refrigeration systems in warm and hot climates substantially.

Conclusions and future developments
Transcritical R744 condensing units are expected to offer a substantial decrease in the carbon footprint of the commercial refrigeration industry.Currently, the lack of affordable technologies to decrease their energy consumption (while preserving overall cost-effectiveness and simplicity) is perceived.This technological gap could be effectively filled by employing the new pulse-width modulation (PWM) ejector, being low-cost, simple and invulnerable to clogging.In addition, although the PWM working principle has been proved for small-capacity vapourcompression systems, it has no practical size or application constraints.
In this work, an experimental campaign involving two PWM ejectorequipped transcritical R744 condensing units, i.e. with and without overfed evaporator, has been conducted.The experimental analysis has been performed at the medium temperature (MT) of about − 5 • C, heat sink temperatures from 30 • C to 40 • C and compressor speeds from 40 Hz to 60 Hz.The results obtained suggest that: • regardless of the selected heat sink temperature and compressor speed, the PWM ejector can control the high pressure in transcritical regime effectively; • the PWM ejector leads to higher COP improvements than its today's available competitors;

Table 7
Performance comparison among all the investigated small-capacity transcritical R744 vapour-compression refrigeration solutions (t eg,evap in = 5 • C, PWM period = 2 s, ΔT superheating = 8 K ± 0.5 K for Std, Pass ej and PWM ej, ΔT superheating = 0 K for PWM ej with OVEV).• at the same cooling capacity the PWM ejector-equipped R744 system leads to energy savings between 7.0% and 11.1% without overfed evaporator and between 11.5% and 16.3% with overfed evaporator compared to the standard R744 unit (i.e. with vapour by-pass valve and without ejector), respectively.
The following future work needs to be carried out: • investigation on the potential energy benefits of the PWM ejector in subcritical and transition regime (with and without overfed evaporator); • assessment of the annual energy savings related to the PWM ejector in different locations worldwide; • study on the potential energy benefits of the PWM ejector in air conditioning and heat pump units; • thorough assessment of the dynamic performance of the PWM ejector.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
associated with η ejector represented a compromise between the decreasing trend of P lift and growing trend of Φ m with the high pressure for each selected t water,gc in .At the optimal high pressure the values of P lift , Φ m and η ejector were equal to 3.45 bar, 0.580 and 0.252 at t water,gc in = 30 • C, 4.00 bar, 0.559 and 0.236 at t water, gc in = 35 • C and 4.81 bar, 0.509 and 0.214 at t water,gc in = 40 • C.

Table 2
Primary geometry parameters of the used two-phase ejector.

Table 3
Accuracies and calibration range of the measurement equipment.

Table 4
Experimental uncertainties of the calculated parameters.
• C, between 1.5% and 5.5% at t water,gc in = 35 • C and between 8.2% and 8.6% at t water,gc in = 40 • C; • the PWM ejector offered COP enhancements between 4.4% and 9.0% at t water,gc in = 30 • C, between 5.6% and 11.8% at t water,gc in = 35 • C and of 9.8% at t water,gc in = 40 • C; • the PWM ejector coupled with the overfed evaporator allowed for COP improvements between 6.9% and 13.3% at t water,gc in = 30 • C, between 8.8% and 15.1% at t water,gc in = 35 • C and between 12.1% and 13.9% at t water,gc in = 40 • C.

Table 6
Summary of the main findings for the PWM ejector-equipped small-capacity transcritical R744 vapour-compression refrigeration unit (t eg,evap in = 5 • C, PWM period = 2 s).