Study on Energy Efficiency of an Off-Grid Vending Machine with Compact Heat Exchangers and Low GWP Refrigerant Powered by Solar Energy

Abstract

The purpose of this research is to design an energy-efficient off-grid vending machine that has a refrigeration system powered by an environmentally friendly refrigerant. The paper presents details from the equipment’s design phase in terms of the resistance structure, photovoltaic panels area dimensioning, but also in terms of the necessary cooling load to be provided by the refrigeration system. The refrigerant chosen was R290 following a comparative analysis, which showed that it has an efficiency 6.6% higher than R134a. During the monitoring periods, the outdoors temperature varied between −4–28 °C, the relative humidity between 22–100% and yet only a slight indoor temperature variation of ±1.5 °C compared to the set value was recorded. The battery module managed to store enough energy from the photovoltaic panels to keep the vending machine functional through three consecutive days with adverse environmental conditions, as evidenced by the low variation of indoor temperature

1. Introduction

Nowadays, the labor market is in a state of continuous development and people’s activities in office spaces and industry are becoming more and more intense. People tend not to give importance to feeding, although this has a negative impact on their health over the long run. In this context, vending machines have become important sources from which people can quickly get food and drink for their lunch break. It is a good alternative to fast food because healthy food can be stored and delivered by this means. The healthfulness aspect was comprehensively reviewed by Matthews M. and Horacek T. [1]. Even though the majority of food and beverage products sold in vending machines are considered to be of low nutritional value, a change of mentality can be made, and it must be considered that any type of refrigerated food can be stored inside them. A study focused on the nutritional profile has been made by Martinez-Perez N. et al. [2] and concluded that policies targeted at improving the nutritional characteristics of the products sold in automated vending systems is highly necessary. Despite the fact that the preliminary testing made on the vending machine, which is the subject of this article study, was developed with packaged food like pretzels, wafers, biscuits, and aluminum cans with juice and mineral water, in the later stages of the study and after highlighting the energy efficiency of the equipment, the authors intend to introduce inside food boxes with increased nutritional intake such as vegetables, fruits, or ready-to-cook food.

According to Ref. [3] there are currently around 20 million vending machines in the world, in which 4.2 million are connected to the IoT (Internet of Things), and the market is expected to reach 8.9 million connected units by 2024. In Europe there were approximately 3.8 million vending machines in 2019, as the JRC Technical Report [4] underlined, of which 2.36 million units have the function to deliver hot drinks and the rest of them being intended for food.

In this context and following the global trend, a thorough assessment and research of the possibilities for providing high energy efficiency for such equipment should be carried out. The best scenario would be when energy independence from the electricity grid can be provided.

After an evaluation of the scientific research carried out so far on this subject, we conclude that this topic is of particular importance, but we have not identified a detailed assessment of a solar powered vending machine for food and beverages. Even though, especially in the field of soft drinks, there are solar-powered equipment on the market [5], no detailed specifics are given on the choice of the type of refrigerant, the refrigerant load, energy efficiency of the cooling system or the operating time interval that can be covered only by using stored energy from the batteries.

To meet the current market requirements, a vending machine should be “smart” and well-integrated with the Internet of Things (IoT). Solano et al. [6] pointed out that a novel equipment should be accessed through a phone app, by which the user can make the product selection and the payment over the cloud using a secured proximity payment model. Sibanda V. et al [7] highlighted that people can save important time if the ability to control remotely a high-tech vending machine is well integrated. Ratnasri N. et al. [8], after a wide study, also stated that IoT technologies is a must regarding the satisfaction of customer preferences. Following the up-to-date trend stressed above, the vending machine analyzed in the current study is completely controlled via Wi-Fi and the interaction with the users is made through a mobile application. The available products and real-time quantities can be accessed by scanning a QR code.

Two of the vending machine’s essential features are energy efficiency and the fact that it must be environmentally friendly. These two aspects are closely related to the refrigeration system that maintains the inside temperature. The aspect of environmental protection is covered by using a refrigerant with low GWP (Global Warming Potential) and the best solution by an efficiency point of view is given by the highest value of Energy Efficiency Ratio (EER). The performance of low GWP refrigerants in this system was evaluated by Sethi et al. in Ref. [9]. Both refrigerants that were assessed, meaning R1234yf and R1234ze, proved to be good substitutes for R134a regarding small refrigeration systems, with R1234yf being able to show even better performance than R134a, providing that minor design changes are made. The research is the direct effect of the fact that R134a, one of the most used refrigerants in vending machines to present, needs to be replaced, as the GWP is much higher than the limits set by the F-Gas Regulation [10]. It must be specified that, starting from 1 January 2022, the refrigerant used in new refrigerators and freezers for commercial use, must have a GWP lower than 150. In the present study, the authors have also made a comparison between different refrigerants, following which the refrigerant with the lowest GWP value and best energy performance was chosen.

Regarding energy efficiency, Wongsuwan et al. [11] made a study on the beverage vending machine refrigeration system EER. The cooling load of the experimental stand consisting of 594 aluminum cans and Polyethylene terephthalate (PET) bottles was evaluated when replacing R134a with the R407C refrigerant. Considering that, after the replacement, EER diminished, the researchers concluded that for the next stage of the study the compressor replacing, and a careful component’s monitoring is necessary. Thus, by only replacing the refrigerant, not all the prerequisites necessary for obtaining energy efficiency are created, and this aspect can be identified only after a comparative analysis, similar to the one carried out by the authors in the present study.

However, although some studies, like the one developed by Harnanan K. et al. [12] intended to increase the vending machines efficiency by using energy-saving devices, the energy efficiency can only significantly increase by means of using renewable energy, more precisely solar energy, for fully covering the equipment’s energy demand. In addition to solar energy, one can also use wind energy or mechanical energy produced by people motion.

By using compact solar panels and internal storage batteries, the system can be made entirely off-grid and can operate even in cloudy or rainy days based on the surplus energy stored in the batteries. Kumar P. et al. in [13] investigated the performance of an automated vending system for medical products equipped with solar panels on top, which has the main advantage of being able to be used in remote locations and can give access to people who need medical care to medicines.

The vending machine analyzed in this article was completely designed, built, and installed at the headquarters of the Technical University of Civil Engineering Bucharest and was subsequently monitored in operation for two months of Spring 2022.

The present paper investigates the possibility of designing an off-grid vending machine system with high energy efficiency and includes investigations of energy efficient refrigerants coupled with the evaluation of an integrated PV system aimed at entirely satisfying the energy demands of the off-grid vending machine. The system’s performance is evaluated through temperature measurements and by PV system monitoring. Finally, the article offers a series of recommendations for the design of energy-efficient off-grid vending machines.

2. Materials and Methods

2.1. Mechanical Part Setup

The first step in the realization of an efficient energy independent vending machine was the design of an experimental model. The main subsystems are highlighted in Figure 1 as follows: Resistance structure—1; thermally insulated enclosure—2; outer casing—3; product delivery—4; refrigeration system—5; electricity production and storage—6; command and control—7.

The role of each subsystem is emphasized in the following: the resistance structure provides support for the component elements and also takes over the loads from exploitation and transfers them to the ground level; the thermally insulated enclosure is designed by using materials with low thermal conductivity, and it has the purpose to store the products before delivery; the outer casing is a thermally insulated metal enclosure that protects the mechanical and electronic components; the products’ delivery is a complex system consisting of two main components, the products’ tray in which the products are stored for delivery and the collecting and transport system, respectively; the refrigeration system produces cold in the thermally insulated room; the electrical system produces and stores electricity by mean of photovoltaic panels and accumulators; the command and control automation system controls the entire system by managing all the data received from the sensor systems and at the same time serves as an interface between the users and the automatic vending machine. The enclosure’s loading is made from the top by lifting the thermally insulated cover and inserting the trays with the products, as illustrated in Figure 2.

The delivery system is designed to use the electricity produced by the photovoltaic panels as efficiently as possible.

The vending machine experimental model was modeled in the first stage in a 3D format using the SolidWorks 2021 software [14]. Modeling in a virtual environment has allowed the optimization of certain components since the conception stage. In the second stage, some subsystems have been optimized using tools from the SolidWorks 2021 soft suite. In the third stage, subsystems were manufactured using technologies adapted to materials’ type.

The resistance structure subsystem is made of carbon steel metallic profiles. To reduce the material consumption, we achieved a resistance structure optimization using the FEM solver within SolidWorks 2021. The material consumption reduction for a metal frame design contributes to the reduction of gas emissions during manufacturing, both because of the smaller steel quantity and because of the manufacturing time reduction. A carbon steel metal structure is in accordance with the European regulations in force, being easily recyclable in the post-use stage.

The input data for the optimization of the resistance structure by using the FEM solver are briefly defined in

Table 1. Resistance structure components.

No. crt.	Name	Quantity/ Volume [cm3]	Density [g/cm3]	Total Mass [g]
1	Thermal insulation Komacel thick 10 mm—refrigeration enclosure [15]	101,542	0.5	50,771
2	Thermal insulation Termoconfort thick 100 mm—refrigeration enclosure [16]	955,166	0.034	32,475
3	Thermal insulation BOND thick 3 mm—refrigerated enclosure [17]	14,444	1.55	22,388
4	Product delivery system—product tray	6 trays	-	66,000
5	Metal structure made of aluminum profiles—refrigeration enclosure	1	-	20,619
6	Refrigeration installation (compressor, evaporator, condenser, fan, support)	1	-	15,672
Total storage space products TSP 187308
1	Outer casing—3 mm thick steel sheet	1	-	80,440
2	Outer casing—40 mm thick polystyrene thermal insulation	1	-	14,460
3	Product delivery system—pick-up and transport	1	-	18,452
4	Command and control system	1	-	4215
Total technical space and product delivery TTS 117567
1	Rechargeable batteries	4	-	152,000
2	Solar panel	4	-	55,200

.

For the 2.5 m height support frame optimization of the solar panels the variable loading was considered due to the wind action. Thus, according to the Design Code Evaluation of the wind action on the constructions [18], resulting in a wind dynamic pressure of 0.2 kPa, the authors decided to avoid the use of glass when making the outer shell of the equipment so as to limit the influence of direct solar radiation. Moreover, using this material, this will not be necessary as the user will interact with the equipment directly by means of a mobile application through which they will be able to see in real time the type and number of the available products The vending machine was designed without transparent surfaces in order to improve thermal insulation. Because the goal of the study is to improve the energy efficiency below a certain requirement, as specified in the previous chapter, it is important to determine the optimal design of the vending machine from a thermal transfer point of view. If the further results indicate that adding transparent surfaces is feasible from an energy efficiency point of view, this will be done at a later date

The manufacturing process started from the thermally insulated enclosure 3D modeling in Solid Works, followed by going through the actual manufacturing stages. These steps were emphasized in Figure 3.

2.2. Refrigeration Unit Dimensioning

A mathematical model was developed in order to predict the heat loads that have to be covered by the refrigeration system.

The inputs included:

-

Vending machine wall material structure and type together with the thermal properties.

-

Heat loads from products inserted inside the vending machine, storage cartridges, and transport system materials.

-

Heat load from ambient air when the cover door is opened for the new product’s loading.

-

Heat loads from the evaporator and condenser fans.

In order to increase the energy efficiency, the vending machine will not benefit from interior lighting and thus the heat input through lighting is zero.

The vending machine is placed at the headquarters of the Technical University of Civil Engineering Bucharest, thus the exterior conventional maximum design temperature $\mathrm{T}=35 °\mathrm{C}$ and outdoor relative humidity, $\phi =35\%$ were selected from the Romanian national regulation for Bucharest location [19].

The preliminary testing made on the vending machine, which is the subject of this article, was developed with packaged food like pretzels, wafers, biscuits, and aluminum cans with soda and mineral water, thus in the heat loads evaluation only a sensible load is considered regarding the products.

The heat load through the vending machine walls was calculated using Equation (1). The refrigeration system is designed in order to maintain an interior temperature ${\mathrm{T}}_{\mathrm{int}.}=6 °\mathrm{C}$.

$${\dot{\mathrm{Q}}}_{\mathrm{walls}}=\mathrm{U}·\mathrm{A}·\left({\mathrm{T}}_{\mathrm{ext}}-{\mathrm{T}}_{\mathrm{int}}\right)\left[\mathrm{W}\right].$$

(1)

The overall heat transfer coefficient, $\mathrm{U}$ was calculated with Equation (2). For the convection heat transfer coefficients, the following values were considered: ${\mathrm{h}}_{\mathrm{int}.}=12 \mathrm{W}·{\mathrm{m}}^{-2}·{\mathrm{K}}^{-1}$ and ${\mathrm{h}}_{\mathrm{ext}.}=24 \mathrm{W}·{\mathrm{m}}^{-2}·{\mathrm{K}}^{-1}$according to Romanian national regulation [20]. Thermal conductivity and material thickness were extracted from the sandwich panel technical sheet [21].

The effect of uneven solar radiation on the vending machine exterior walls was taken into account by increasing the outdoor temperature value with a range that takes into account the geographical positioning of the equipment. Thus $\Delta \mathrm{T}=15 \mathrm{K}$ for the walls facing south and top cover, $\Delta \mathrm{T}=5 \mathrm{K}$ for the walls facing east and west, and $\Delta \mathrm{T}=0 \mathrm{K}$ for the walls facing the north and the bottom cover [22,23].

$$1/\mathrm{U}=1/{\mathrm{h}}_{\mathrm{ext}.}+1/{\mathrm{h}}_{\mathrm{int}.}+{\displaystyle \sum }\left(\mathrm{L}/\mathrm{k}\right) \left[{\mathrm{m}}^2·\mathrm{K}·{\mathrm{W}}^{-1}\right].$$

(2)

The heat load through the cover door opening was calculated using the volumetric air flow during door opening, air specific enthalpies, and the time interval required for outdoor air cooling (8 h), with Equation (3). The air volume that needs to be cooled is equal to the volume of air inside the vending machine, since at each equipment supply there is a complete change of air that enters the ambient temperature.

$${\dot{\mathrm{Q}}}_{\mathrm{door}}=\left[{\mathsf{\rho }}_{\mathrm{air}}·{\mathrm{V}}_{\mathrm{air}}·\left({\mathrm{i}}_{\mathrm{ext}}-{\mathrm{i}}_{\mathrm{int}}\right)\right]/\left(3600·\mathrm{n}\right)\left[\mathrm{W}\right].$$

(3)

The heat load brought by products when introduced into the vending machine is given by Equation (4). The most disadvantaged situation was estimated, namely when all the products are replaced simultaneously and are at the maximum temperature at which they can be delivered with a transport vehicle without refrigeration, ${\mathrm{t}}_{\mathrm{prod}.}=30 °C$ respectively. Every type of product was weighted, and the specific heat values were taken from Refs. [24,25].

$${\dot{\mathrm{Q}}}_{\mathrm{prod}.}={\displaystyle \sum }\left({\mathrm{m}}_{\mathrm{prod}.}·{\mathrm{c}}_{\mathrm{p},\mathrm{prod}.}·\left({ \mathrm{T}}_{\mathrm{prod}.}-{\mathrm{T}}_{\mathrm{int}}\right)/\mathrm{t}\right)\left[\mathrm{W}\right].$$

(4)

The heat load of the evaporator fan motors is dependent on its efficiency given in the following Equation (5).

$${\dot{\mathrm{Q}}}_{\mathrm{fan}}={\mathrm{P}}_{\mathrm{el}.}/\mathsf{\eta } \left[\mathrm{W}\right].$$

(5)

All in all, the vending machine total heat load is presented in Equation (6).

$$\begin{gathered} {\dot{\mathrm{Q}}}_{\mathrm{TOT}.}={\dot{\mathrm{Q}}}_{\mathrm{walls}}+{\dot{\mathrm{Q}}}_{\mathrm{door}}+{\dot{\mathrm{Q}}}_{\mathrm{prod}.}+{\dot{\mathrm{Q}}}_{\mathrm{fan}}\left[\mathrm{W}\right] \\ {\dot{\mathrm{Q}}}_{\mathrm{TOT}.}=93+13+278+58=442 \mathrm{W}. \end{gathered}$$

(6)

The next step after determining the heat load is to establish which refrigerant will be used in the cooling system. The authors conducted a theoretical analysis in which nine refrigerants were considered. In the end, based on energy efficiency criteria and environmental protection, the refrigerant with the best performances is going to be selected.

The environmental, safety and physical properties of the 10 refrigerants R134a, R455A, R290, R1234yf, R448A, R449A, R452A, R513A, and R507 are plotted in

Table 2. Refrigerant properties.

Refrigerant	R134a	R507	R513A	R452A	R290	R448A	R449A	R1234yf	R455A	R744
ASHRAE safety classif.	A1	A1	A1	A1	A3	A1	A1	A2L	A2L	A1
ODP	0	0	0	0	0	0	0	0	0	0
100-year GWP	1430	3985	631	2140	3	1387	1397	4	148	1
Critical temp. [°C]	101.1	70.61	94.91	75.05	96.74	82.68	82.07	94.7	85.61	30.98
Critical pressure [bar]	40.59	37.05	36.48	40.14	42.51	45.95	45	33.82	46.54	73.77
Normal boiling point [°C]	−26.1	−46.74	−29.47	−46.93	+42.11	−46.12	−45.72	−29.49	−52.02	−62.89
Class	HFC	HFC	HFC/HFO	HFC/HFO	HC	HFC/HFO	HFC/HFO	HFO	HFC/HFO	Natural refrige-rant
Molecular mass [kg/kmol]	102	98.86	108.4	103.5	44.1	86.28	87.21	114	87.45	44.01

. Nowadays, according to Refs. [10,26] the refrigerant used in new cooling systems could be a natural substance (R290), a freon (HFC), a hydrofluoroolefin (HFO), or a mixture between an azeotropic (R4xx) and zeotropic (R5xx) substance. In the vending machine sector, the most common refrigerants are R134a, R290, and R744. The refrigerants’ performance was evaluated on a one-stage type vapor refrigerating system, with an air cooled condenser and air chilled evaporator, under the following assumptions:

-

The outdoor temperature calculation value is $35.3 °C$ and the relative humidity 35%, as mentioned above.

-

The condensing temperature is $49 °C$.

-

The condensing temperature value is $-4 °C$.

-

The evaporator outlet superheat $5 °C$.

-

The compressor isentropic efficiency ${\mathsf{\eta }}_{\mathrm{is}}=0.7$.

-

The compressor volumetric efficiency ${\mathsf{\eta }}_{\mathrm{C}}=0.95$.

-

The condenser’s subcooling $\Delta {\mathrm{t}}_{\mathrm{sub}.}=5\mathrm{K}$.

-

The superheat on the suction line $\Delta {\mathrm{t}}_{\sup .}=10\mathrm{K}$.

-

The effects of the pressure drop is negligible.

-

The system operates in a steady state.

-

14 K temperature difference between air and refrigerant at the condenser side and 10 K on the evaporator side were assumed.

The coefficients GWP and ODP values reported in the table are based on the latest AR4 numbers reported in IPCC (Climate change 2013) [27].

The thermodynamic properties and heat fluxes corresponding to each heat exchanger were obtained by using dedicated software: Coolpack [28] and Chemours Refrigerant Expert (CRE) 1.0 [29] and are presented in

Table 3. Comparative analysis for 10 refrigerants.

Refrigerant	R134a	R507	R513A	R452A	R290	R448A	R449A	R1234yf	R455A	R744
Parameter
Evaporating temperature [°C]	−4	−4	−4	−4	−4	−4	−4	−4	−4	−4
Evaporating pressure [bar]	2.53	5.49	2.79	4.92	4.16	4.82	4.469	2.75	4.12	31.30
Condensing temperature [°C]	49	49	49	49	49	49	49	49	49	49
Condensing pressure [bar]	12.85	23.06	13.41	21.83	16.74	22.07	20.77	12.72	21.73	75.00
Discharge temperature [°C]	83.19	78.23	75.37	79.77	81.63	87.5	90.83	69.88	75.37	87.5
Suction volume [m3/h]	0.09028	0.04012	0.07733	0.0437	0.11140	0.01749	0.0573	0.066	0.06915	0.01340
Refrigerant mass flow rate [kg/s]	0.0032	0.0045	0.0037	0.0042	0.00158	0.00314	0.0032	0.0042	0.0037	0.0077
Cooling capacity [kW]	0.442	0.442	0.442	0.442	0.442	0.442	0.442	0.442	0.442	0.442
Heating capacity [kW]	0.640	0.680	0.650	0.660	0.602	0.610	0.650	0.660	0.612	0.842
Compressor power [kW]	0.170	0.190	0.170	0.180	0.160	0.168	0.180	0.180	0.170	0.572
EER [-]	2.600	2.326	2.600	2.456	2.771	2.627	2.456	2.456	2.550	0.773

.

Assuming the same evaporating and condensing temperatures, R290 has slightly higher suction pressure and slightly lower discharge pressures when compared to R134a, R513A, and R1234yf.

At −4 °C, the suction temperature of R290 latent heat is 89.3% higher than R134a, 47.3% higher than R513A, and 43.3% higher than R1234yf.

R290 has a lower critical temperature than R134a, and R1234yf has lower critical temperature than R290.

The thermodynamic cycles’ comparative analysis was performed considering an isentropic efficiency of 0.7 based on the compressor type. A value of 0.05 bar for the pressure drops inside the evaporator and condenser was considered.

Assuming the same evaporating and condensing temperatures, R290 has slightly higher suction pressure and slightly lower discharge pressures when compared to R134a, R513A, and R1234yf.

At −4 °C suction temperature, R290 latent heat is 89.3% higher than R134a, 47.3% higher than R513A, and 43.3% higher than R1234yf.

R290 has a lower critical temperature than R134a, and R1234yf has lower critical temperature than R290.

By investigating the cooling performance, cycle efficiency, low GWP target, and the refrigerants thermodynamic proprieties, the best solution for this application was identified, namely R290. The main characteristics that should be emphasized are: the highest COP for cooling, that is, $2.771,$ directly related to the lowest compressor power consumption; low suction volume, which gives small compressor dimensions; and the lowest refrigerant mass flow rate, namely $0.00158 \mathrm{kg}/\mathrm{s}$, with an effect in refrigeration system compactness and reduced risk to the environment in the event of accidental leakiness. When directly compared with R134a, which is the most widespread refrigerant used in vending machines to date, R290 gives $6.6\%$ higher energy efficiency.

The propane’s only drawback is related to safety. It is a flammable refrigerant (A3 Class according to Table 2); thus, the vending machine must be protected from electric sparks to prevent explosions in case of leaks. This can be done by rigorously sealing electrical contacts and relays.

In the light of all the above, R290 was chosen as a refrigerant in the vending machine cooling system. Each refrigeration system component was selected further according to Table 3 specifications.

2.3. Electrical System

An off-grid vending machine requires a sustainable source of energy sufficiency to cover the electrical load. In the present case, the vending machine’s electrical load is represented by the refrigeration system (mostly by its compressor) and by the mechanical setup, which moves the products from the refrigerated area to the door from which the customers will retrieve them.

One important aspect to note is that neither the refrigeration system nor the mechanical system will be running at constant peak loads. The refrigeration system is expected to intermittently increase its load to maintain the required temperature inside the vending machine, while the mechanical system will operate only as required (when customers buy a product).

In order to ensure that the vending machine can function off-grid, a renewable energy supply is required so to cover the electrical load generated by the refrigeration and mechanical systems. For this purpose, the photovoltaic (PV) panels were envisaged as the appropriate solution for supplying the vending machine with sufficient power in order to maintain operational parameters.

The specialized software PVSyst [30] was used to evaluate different PV panel array configurations. The results indicated that in order to cover the electrical load, a PV array with a rated power of at least $750 \mathrm{W}$ was required, along with a battery storage system with a rated power capacity of at least $500 \mathrm{Ah}$, coupled to an inverter with a rated power of at least $1000 \mathrm{W}$.

In accordance with the above results, the following equipment was selected for the electrical power supply system: 3 Polycrystalline PV panels produced by Victron energy, model BlueSolar SP042702000 [31] ($1640\times 992\times 35 \mathrm{mm}$) with a nominal power of $270 \mathrm{W}$ (Figure 4a); and a battery system comprised of four $12 \mathrm{V}$ sealed gel batteries with a rated power capacity of $130 \mathrm{Ah}$ produced by Victron energy, model BAT412121104 [32] (Figure 4b). In order to convert the DC current produced by the PV array, the system was connected to an inverter with a rated power of $1000 \mathrm{W}$ produced by Mean Well, model A301-1K0-F3 [33] (Figure 4c) with an efficiency of 82%. The output of the PV array as well as the charging of the batteries was controlled via a Maximum Power Point Tracking (MPPT) solar charger produced by Victron energy, model BlueSolar MPPT 150|70–Tr [34] (Figure 4d) with 98% energy conversion efficiency.

The schematic of the electrical system is presented in Figure 5. The PV array is composed of 3 PV panels linked in a serial connection, which brings the combined power of the PV array to $810 \mathrm{W}$ ($3\times 270 \mathrm{W}$), above the required value of $750 \mathrm{W}$ recommended by the specialized PV software. The PV array is linked to the MPPT solar charger where the voltage of the electrical PV charge is lowered to a suitable value for charging the battery array while maintaining maximum power. The battery array is comprised of the four batteries linked in a parallel connection, supplied from the MPPT solar charger. In a parallel connection to the battery array, the inverter where the DC current forms the PV, and the battery arrays are converted to AC to be used by the refrigeration and mechanical systems can be found.

For controlling the flow of current from the PV array to the batteries and the inverter, a MPPT charge controller was selected [34]. MPPT controllers are known for their efficiency in converting the input voltage from PV arrays into a lower voltage appropriate for charging batteries, all while providing close to maximum power [35]. The efficiency in such systems is well documented [36,37,38]. The BlueSolar MPPT solar charge controller has a rated DC/DC conversion efficiency of 98% [34]. The technical documentation of this MPPT solar charger [34,35] does not go into detail regarding the algorithm it uses for tracking the maximum power point due to the algorithm’s proprietary nature.

The battery array undergoes a three-stage charging cycle, going through the following charging phases: (1) bulk, (2) absorption, and (3) float. The charging phases are controlled by the MPPT solar charger [35]. During the bulk stage (1), the MPPT solar charger introduces the maximum available current into the batteries in order to increase the electrical charge of the battery array. Once the battery array voltage reaches a value of $14.5 \mathrm{V}$, the MPPT controller passes into the absorption stage (2) where the battery array voltage is kept constant. As the batteries near their fully charged state, the current flowing into the batteries steadily diminishes and once it falls below the value of $9 \mathrm{A}$, the MPPT charger passes into the float stage (3). In this final stage, the battery array voltage is kept constant at a lower value of $13.5 \mathrm{V}$, and current is only allowed to enter the batteries so as to maintain this lower voltage [35].

In the case of photovoltaic panels, the experimental stand was equipped with a passive system for adjusting the lifting angle (adjustment system with two screw mechanisms mounted at the level of the fixing frame) of the panels. This system allows the adjustment by several degrees (+/−5 degrees) of the elevation angle. The purpose is to position the photovoltaic panels module in relation to the month of the year, so that the solar energy acquired can be the maximum. In addition, the angle depends on the geographical location where the equipment is positioned.

The PV Modules were installed at a 40° mounting angle, optimized for the climatic zone (Bucharest). The entire vending machine system was placed so that the PV array is facing South, in order to maximize the available solar energy

3. Results

3.1. Refrigeration unit

In the initial stage, the heat load calculation was performed. The following value resulted: ${\dot{\mathrm{Q}}}_{\mathrm{TOT}}=442 \mathrm{W}.$

Based on this value, the performance of a refrigeration system with mechanical compression of vapors with different refrigerants was analyzed, focusing on the ecological component (carbon dioxide footprint on the environment) and on the lowest possible energy consumption.

Due to the fact that the vending machine is placed outside, it is not necessary to follow [26] the regulation that provides the maximum refrigeration system charge.

Several refrigeration system components are presented in

Table 4. Refrigeration system components.

Equipment	Technical Data	Type
Evaporator coil	Length L = 50 cm; Width W = 10.5 cm; Height H = 14 cm Inlet/outlet connection $\mathsf{\Phi }$ 8.5 mm, Copper	KARYER VEGA
Evaporator fan motor	$\mathrm{Electrical} \mathrm{power} 36 \mathrm{W}$; Length L = 32 cm; Width W = 9 cm Height H = 11.5 cm	TU14
Condenser coil	Length L = 29 cm; Width W = 14 cm; Height H = 25 cm	LU-VE contorno
Condenser fan motor	$\mathrm{Electrical} \mathrm{power} 38 \mathrm{W}$	-
Hermetical single speed compressor	Displacement 4.5 cm3; Voltage/frequency 220–240 V/50 Hz Supply connection $\mathsf{\Phi }$ 8.1 mm, copper Discharge connection $\mathsf{\Phi }$ 6.55 mm, copper	EMBRACO EMX6141U

and Figure 6.

Due to the fact that the vending machine functioning is an off-grid type, based on the intensity of the solar radiation, the outdoor weather conditions—outdoor temperature and the intensity of the solar radiation—must be constantly monitored during the equipment’s operation. The accuracy of the measuring instruments is presented in

Table 5. Measuring instruments.

Equipment	Measured Value	Accuracy	Measured Interval
Pyranometer	Intensity of solar radiation	±3%	0–1500 W/m2
USB type humidity sensor	Temperature (air)	±0.5 °C	−20–70 °C
USB type	Relative humidity	±3%	0–100%
FLIR Thermal Camera	Temperature (surface)	±0.3 °C	−40–400 °C

.

A pyranometer connected to a data acquisition unit, presented in Figure 7a, was used to monitor the intensity of the solar radiation. The measuring range is from 0 to 1500 W/m², with a resolution of 0.1 W/m², operative range from −40 to 60 °C, and the accuracy range is placed between ±3%.

Temperatures and the relative humidity inside the vending machine and outside were monitored with USB type temperature sensors and their positioning can be noticed in Figure 7b. The temperature sensors have an operative range from −20 to 70 °C, an accuracy range between ±0.5 °C, and were set to monitor temperature and relative humidity at one minute time steps.

In order to have a clearer picture of how the equipment works and to demonstrate the off-grid operation, two distinct time intervals from March and April of 2022 were monitored, namely 7–25 March and 4–21 April. The lid of the vending machine was opened once every day in order to simulate the supply periods (cycles) with new products at ambient temperature. The product’s temperature before replacement and the temperature of the products that were newly introduced in the vending machine was monitored with a thermal imaging camera (Figure 8) to highlight how the indoor temperature is maintained. The newly introduced products were stored for a few days inside the Research Laboratory before every cycle and the temperature at which they were introduced was between $16 °\mathrm{C}$ and $23 °\mathrm{C}$. At the same time, products were delivered periodically through the delivery window to stimulate user’s product demand.

The outdoors temperature variation is presented in Figure 9 separately for the two analyzed time intervals.

In order to assess how solar radiation intensity varies in the analyzed period of the year in Romania at the Research Laboratory location, the authors selected a certain number of days in which to present the variation profile. The intensity of the solar radiation for a selected number of days, part of the analyzed time intervals, is provided below, in Figure 10a,b. The two intervals were chosen to highlight the recorded values in a period when the sky was clear (Figure 10a) and for a rainy period in which the sky was covered by clouds most of the time, respectively (Figure 10b).

The temperature variation inside the vending machine recorded by T1, T2, and T3 sensors is presented in Figure 11.

3.2. Electrical System

The following parameters of the electrical system were monitored for the duration of the experiment: the power of the PV array ($P_{PV}\left[\mathrm{W}\right]$); the battery array’s voltage ($V_{BAT}\left[\mathrm{V}\right])$ and current $\left(I_{BAT}\left[\mathrm{A}\right]\right)$; the MPPT solar charge controller’s state ($Charge State$); and the power consumed by the refrigeration system ($P_{REF}\left[\mathrm{W}\right]$).

Between the 5 and the 27 of April, these parameters were recorded with a time step of 1 min. The evolution of the above-mentioned parameters over the course of this time period is presented in Figure 12 and Figure 13. Additionally, one of the temperature sensors inside the food storage compartment of the vending machine (thermocouple T₃) was also included in Figure 12 and Figure 13, so that the impact of the electrical system on the refrigeration could be evaluated.

Of great interest in our study is the behavior of the MPPT solar charge controller and of the battery array, notably its performance in different environmental conditions (days with high/low solar radiation). To this end, Figure 14 highlights the behavior of the MPPT solar charge controller in two such cases. Figure 14a shows the MPPT controller functioning under normal, favorable conditions (on the 6 April 2022) and the battery array can be seen going through all of its charging stages. Figure 14b, on the other hand, highlights the system’s behavior under unfavorable conditions (on the 11 April 2022). In this latter case, the battery does not pass through all of its charging stages.

Naturally, since the results of Figure 14 indicate that the battery array charging system is susceptible to environmental conditions, the impact of a prolonged period of unfavorable conditions would be of interest in order to assess whether the refrigeration system is significantly impacted.

A time period of 5 days was selected for the investigation, between 17 and 21 April 2022, a period during which in the first three days (from 17 to 20) there was little solar irradiation. Thus, Figure 15 shows the system’s performance during this time period when due to the unfavorable conditions the performance of both the PV and the battery arrays was sub-par.

Finally, during normal operation conditions, the vending machine’s door does not remain open for long periods of time. It would be of interest however, to see how the system performed if the door was left open for a longer period of time (for example during a particularly lengthy loading process).

On 26 April 2022, the door of the vending machine was left open for 30 min between 08:30 and 09:00. The door was then subsequently closed, and the vending machine was left to operate as usual for the rest of the day. The parameters of the PV and battery arrays during this test are presented in Figure 15b.

4. Discussion

4.1. Refrigeration System

The graphs presented in the previous section showed in Figure 11a–d and the thermographic images shows that the designed refrigeration system was able to maintain a 6 °C average temperature of the products stored indoors, regardless of the external weather conditions, temperature, relative humidity, or solar radiation intensity.

The outdoor temperature in the two monitored time intervals varied in the range −4– 28 °C and the relative humidity in the range 22–100%, according to Figure 7a,b, with an average of around 50%.

The temperature measuring sensors were positioned in the areas most disadvantaged by the cold air flow distributed by the evaporator fan, respectively, in the two side corners (sensors T2 and T3) and on the fan opposite side at the bottom side of the vending machine (sensor T1).

Following the evaluation of the temperature profiles at the equipment’s interior, a slight temperature variation can be noticed, but which falls within the initial set interval namely, a maximum of ±1.5 °C to the set value. The time periods when the supply periods have been simulated had a direct impact on the indoor temperature. The refrigeration system was able to compensate for the heat gains. However, the supply periods had a much more pronounced effect on the interior relative humidity. On rainy days, the inside relative humidity reached a maximum value of 97% in comparison with sunny days when an average value of 50% was registered. Regarding this aspect, it should be mentioned that, in a situation where the stored products are sensitive to relative humidity, in the later testing stages, a dehumidification system should be mandatory in order to maintain it in a specific range.

According to Figure 10a,b, the intensity of the solar radiation reaches maximum values between 11–15 h, with values between 290–810 W/m² when the sky is clear and, in comparison, in the most unfavorable conditions, when the sky was covered by clouds the entire day (21 April 2022), an average value of 100 W/m² was recorded during the same hourly interval.

Therefore, a conclusion can be drawn, namely the refrigeration system that equips the experimental vending machine located in the headquarters of the Technical University of Civil Engineering ensures the set indoor temperature for storing the products. No changes in the appearance, consistency, or quality of the products were reported during the monitoring periods.

4.2. Electrical System

The results presented in Figure 12 and Figure 13 highlight the behavior of the vending machine. Firstly, over the course of the 22 days presented, the machine worked almost continuously with only two interruptions in the data: the first for two hours on 18 April due to an unforeseen interruption of the data acquisition server and the second due to planned server maintenance on 23 April. These data-side interruptions had little effect on the vending machine, as can be seen by the temperature indicated by thermocouple T₃, which remains almost constant before and after these interruptions, showing no sign of significant perturbation.

Secondly, the temperature indicated by thermocouple T₃ has remained almost constant over the course of the selected time interval. The electrical load of the refrigeration system ($P_{REF}\left[W\right]$) is seen in the graphs of Figure 12 and Figure 13 as a series of spikes, which correspond with the activation of the refrigeration system’s compressor. These spikes manifest themselves over the selected time interval at a frequency of about 3 per 10 min, indicating that the compressor switches on and stays active for 3 min, with an average interval of 7 min between activations.

On days with low solar irradiation, the battery array was seen to not complete its three charging stages, remaining in the bulk stage where the MPPT solar charge controller aims to charge the batteries with as much current as possible. The fact that during these intervals thermocouple T₃’s readings are not adversely affected, one can surmise that the battery array was well capable of covering the electrical load of the refrigeration system. The process was aided by the fact that in the days with low solar irradiation (corresponding generally to lower outside temperatures), Figure 12 and Figure 13 show a reduced electrical load density $P_{REF}\left[\mathrm{W}\right],$ signifying fewer activations of the refrigeration system’s compressor and consequently lower power consumption.

The MPPT solar charge controller can be seen working as expected in Figure 14a. during normal operating conditions. In this case, at around 07:00 in the morning, the battery arrays enter the bulk stage of charging whereas as much current is introduced in the batteries as is available from the PV array. The current flowing towards the battery array ($I_{BAT}\left[\mathrm{A}\right]$) can be seen steadily increasing up until the battery array voltage reaches a value of $14.5 \mathrm{V}$, at around 11:00, just before noon. At this point the MPPT controller switches to the absorption stage, keeping the voltage constant at this value of $14.5 \mathrm{V}$, as evidenced by the plateau seen in the battery voltage ($V_{BAT}\left[\mathrm{V}\right]$) in Figure 14a. between 11:00 and 13:00. This limitation of the voltage throttles the current flowing into the battery array, which starts diminishing from the moment the MPPT controller switches to the absorption stage. Once the current flowing to the batteries drops below $9 \mathrm{A}$, the MPPT controller then passes into the float stage. The battery voltage is kept at a steady value of $13.8 \mathrm{V}$ and very little current is supplied to the battery array, only enough to maintain the aforementioned voltage. Occasionally, the MPPT controller re-enters the bulk stage for brief moments, usually corresponding to the activation of the refrigeration system. This is the typical behavior of the MPPT controller and the battery array during normal conditions.

When conditions are unfavorable, as highlighted in Figure 14b. the MPPT controller enters the bulk stage in the early morning as usual, but never manages to exit this stage, since, as can be seen in Figure 14b, the battery array voltage $V_{BAT}\left[\mathrm{V}\right]$, never reaches the upper value of $14.5 \mathrm{V}$. Consequently, the absorption and float stages are never triggered by the MPPT controller. This behavior becomes even clearer when looking at Figure 15a, where, over the course of three days with low solar radiation, the battery array is never seen to enter the absorption or the float stages. During this time, the quasi-constant temperature indicated by thermocouple T₃ signifies that the electrical load was covered by the battery array. On 20 April, when the solar radiation rises, the battery array is seen entering the absorption stage but not quite managing to reach the float stage, as evidenced by the frequency with which it re-enters the bulk stage. Finally, on 21 April, the battery array resumes it normal three-stage charging cycle. These results indicate that even after 3 days of unfavorable conditions, the battery array managed to reach full capacity within 2 days of normal operation.

Leaving the door of the vending machine open for 30 min, besides producing an expected increase in interior temperature and electrical load, does not appear to hinder nor prolong the charging process of the battery array as evidenced by Figure 15b. Further research is however required, in order to ascertain if this conclusion holds true for lower values of solar radiation. Leaving the door open for longer periods of time would only be relevant for pushing the system’s limits, as it is not the typical operating mode of the vending machine.

Overall, the PV and the battery arrays have proved capable of maintaining the vending machine functional and monitoring its parameters over a period of 22 days with nearly no interruptions. The system proved resilient even when faced with multiple days of low solar radiation, when results indicate that a large part of the electrical load was covered by the battery array. The system has fulfilled its designed purpose of keeping the vending machine functional, although further research and potential optimization can be envisaged regarding to performance in other seasons with worse weather conditions.

Additional research is following in order to monitor the interaction between the refrigeration system and photovoltaic panels during the hot season. This research will be done by the team that conducted the present study and will be published in subsequent papers.

The refrigeration plant dimensioning was done considering the exterior parameters corresponding to Romania. In addition, when establishing the angle of inclination of photovoltaic panels, the optimal angle of inclination to the south corresponding to the geographical location was taken into account. However, the input data can be easily modified, as well as the mechanical structure, so that the equipment can be implemented in any geographical region.

The metallic structure modeled and optimized using the Solid Works software, behaved appropriately in operation. By resistance structure optimization we achieved a reduction in material consumption together with a gas emission reduction during the manufacturing process, which is in line with current trends.

5. Conclusions

By designing the equipment that is the object of this work, the authors intended to offer an efficient alternative for feeding a society that is always on the move.

After analyzing the measured data, we can conclude that the goal has been achieved, namely the realization of an energy-efficient off-grid vending machine whose operation is based only on solar energy absorbed by photovoltaic panels. The temperature inside the equipment has been maintained around the set value of $6 °C$ regardless of external conditions. However, the inside relative humidity is greatly influenced by the fresh air introduced with the products’ loading. As a consequence, in the future the equipment must be improved with a dehumidification system. To complete the study, the operation in the hottest months of the year—July and August—is going to be evaluated, and also the opportunity to cover the equipment outdoors with different types of material that reflect solar energy.

Data from the photovoltaic and battery modules indicates that the vending machine was kept within operational parameters for the entire experimental period. The system has proved robust even when faced with three consecutive days of adverse environmental conditions when the power generated by the photovoltaic array was low and the vending machine was mostly powered by the battery array. Immediately after this period, when environmental conditions returned to normal, the battery array was powered back to full after 1 day. Further investigation is required, particularly during the hottest months of the year when the system will be under its highest load during periods with clouds or rain (causing low PV power generation) and high temperatures. If the system proves capable of handling these conditions, future research could envisage further optimizing the battery or photovoltaic arrays and investigating the possibility of reducing the number of PV panels or batteries in the module.

This project was a proven success by allowing different users to access a vending machine, even in remote areas. The final product constitutes an innovative solution which aims at energy saving, coefficient of performance, and lowering the CO₂ footprint of refrigerating equipment.