Artificial Sun—A Stand to Test New PVT Minimodules

Abstract

Hybrid photovoltaic thermal (PVT) modules have gained more attention because of their benefits of higher total efficiency and lower gross area of installation in comparison with photovoltaic (PV) or solar thermal collectors (T). Although international standards for separate panels, photovoltaics, or thermal collectors are available, the lack of testing procedures for PVT panels is a problem, especially if a high level of integration between the two parts is implemented. In the paper, a new stand to test new PVT minimodules is proposed and verified. It allows a reduction of the influence of environmental conditions on the tested T or PVT structures. Research conducted on lamp configurations confirms the possibility of achieving a high uniformity of light intensity, with values close to the AM1.5 spectrum standard (1049 ± 34 W/m²). The first measurements of new PVT minimodules have proven their usefulness, as well as the potential of a new hybrid solution.

1. Introduction

Solar energy is one of the most important sources of renewable energy. It can be converted into electricity with the aid of photovoltaic (PV) modules or into heat with the aid of solar thermal (T) collectors. The combination of both systems was proposed in the late 1970s [1,2], and since then, photovoltaic thermal (PVT) collectors have gained the attention of many researchers [3,4]. They offer higher efficiency compared to PV solutions due to the temperature reduction of the operating solar cells. Furthermore, PVT modules ensure a better use of land (similar to the case of semi-transparent modules used as architectural elements [5,6]) due to their smaller gross area in comparison to separate PV or solar thermal solutions.

Hybrid PVT panels contain two connected components: a photovoltaic panel and a solar thermal collector. The first is responsible for the conversion of solar energy into electrical energy, whereas the second is responsible for taking the heat dissipated in the photovoltaic part and transferring it to coolant. Next, the warm medium (usually liquid) is transported to an external installation. Therefore, PVT systems offer both electricity and heat that can be used. PVT solutions [3,4,7,8,9,10,11,12,13,14] can be distinguished with respect to the type of coolant used in the collector section, the type of solar cells used in the PV section, the design and layout of the system, or the approach to integrate the two components. Furthermore, criteria such as operating temperature or type of application can be used to design and optimise the system. Although the two components are connected and offered as an integrated solution, testing procedures and standards are supplied for separate solutions, such as only PV modules or solar thermal collectors [11,15,16,17,18].

The experimental tests of the PVT systems are not standardised. They cover indoor measurements [15,18,19] and outdoor measurements [7,10,15,16,20], and their final results may differ by a lot depending on the test conditions. All the approaches include monitoring coolant flow rate, inlet and outlet coolant temperature, ambient temperature, and solar irradiance. Furthermore, weather conditions are checked and gathered for outdoor measurements. The final efficiency is calculated with the aid of different models [17,21] and it may differ by about 50%.

Guaraccino et al. [16] showed and compared the results of measurements conducted on three different PVT modules under varying weather conditions. They used the standard testing procedure for thermal collectors. Hence, they kept solar irradiance above 700 W/m² with a variation of no more than 50 W/m², and they performed the measurements at the nominal coolant flow rate, or 0.02 kg/s ± 10%, as advised (if the value of the nominal flow rate was not given). The calculated efficiency of the modules was affected not only by the design properties (such as connection of thermal and PV parts or applied glazing), but also by temperature or wind speed. To provide a better comparison of the test reported in different studies, some normalisation procedures are recommended. For example, Sun et al. [17] proposed a novel formula to calculate the nominal operating cell temperature that involved the influence of a liquid mass flow rate on the temperature of the PV structures. In the case of indoor measurement systems, one of the crucial elements is the light source [15]. Its selection depends on the requirements of the designed stand. We may optimise it for indoor applications or imitate real outdoor conditions. Furthermore, the used setups are designed as straight [18] or upside down [19] versions. The latter ought to increase the amount of heat to reach the tested PVT solution, but in most cases, it also affects the coolant flow. According to authors’ knowledge, only open constructions are used.

The lack of testing standards for PVT panels is a significant problem, especially in the case where the integration of both components in PVT construction is so deep that the whole module must be treated as a single unit at the design and test stage. It takes place in the PVT construction proposed by the authors [22], where the same plate is both the base plate for the silicon photovoltaic cells and the top part of the thermal collector. To support this investigation, a dedicated test stand called the Artificial Sun has been designed and verified. It allowed us to simulate the real solar irradiance conditions with good radiation uniformity. Due to the close construction, it ensures a good energy balance and limits the influence of environmental conditions. The stand has been successfully used for PVT minimodule tests.

2. Artificial Sun

A general outline of the Artificial Sun stand is presented in Figure 1. It consists of a dome with holes for halogen lamps on its top and a test cold plate designed as a base of the photovoltaic thermal (PVT) module [22]. The carried out investigations focus on the minimodule [23], the dimensions of which are reduced to one third of the conventional photovoltaic or collector panels; hence, they are 300 × 500 mm. The distance between the halogen lamps and the test module is equal to 500 mm.

The main requirements for designing the measuring stand were to ensure the highest possible energy transfer to the bottom plate (reduction of the energy losses through the dome) and to achieve solar irradiance at the level of 1000 W/m², with a high uniformity at the surface of the PVT module. In order to meet the defined demands, the problem was investigated numerically to analyse the different factors that influence PVT solutions. Consequently, the proposed stand was manufactured and tested.

3. Numerical Analysis

3.1. Numerical Model

The conducted numerical analyses focused on the reduction of energy losses to the surrounding area. The modelled structure is presented in Figure 2. It consists of an upper part, a dome with four holes (140 × 180 mm) at the top corresponding to the halogen lamps, and a 2 mm thick aluminium plate at the bottom that represents the top layer of the test cold plate of a PVT module. The enclosure is filled with air. Figure 2b illustrates the construction of the dome walls.

The conducted study covered the investigation of a material for the dome (metal (steel) or isolator (polystyrene)), an emissivity of the surface (rough layers of isolators or reflective layers), and a double layer design of the dome walls (a metal wall with an internal isolation of differing thicknesses). Therefore, the following cases were distinguished:

a.

Insulator with a rough surface;

b.

Steel with a reflective surface;

c.

Double wall of 5 mm with a rough surface and a steel wall with a 5 mm thick isolator with an untreated surface;

d.

Double wall of 5 mm with a reflective surface and a steel wall with a 5 mm thick isolator with an additional reflective layer;

e.

Double wall of 10 mm with a reflective surface and a steel wall with a 5 mm thick isolator with an additional reflective layer; and

f.

Double wall of 20 mm with a reflective surface and a steel wall with a 5 mm thick isolator with an additional reflective layer.

The mesh was generated for every geometry (single and double layer), and it contained about 350,000 elements with 250,000 inside the air enclosure. The grid sensitivity was checked. The model was prepared with the aid of ANSYS CFX commercial software based on the finite-volume method.

A series of simulations has been run for the following assumptions and boundary conditions:

• The heat flux at every rectangle representing a lamp equals 168,651 W/m², which corresponds to 170 W in total;
• The temperature at the bottom of the aluminium plate is 32 °C (efficient cooling with the aid of 30 °C liquid coolant is assumed);
• Ambient heat exchange is modelled with the aid of a convection boundary condition with the reference temperature of 25 °C and a heat transfer coefficient of 10 W/m²K, and 2 W/m²K at the side walls and the top wall of the dome, respectively;
• The heat exchange in the enclosure involves thermal radiation (the differential approximation model [24] is used; hence no heat absorption or scattering within the air is analysed), the solid surfaces are opaque, and the emissivity equals:

  ○

  0.9 for the painted metal surfaces and for the rough layers of the isolator; and

  ○

  0.1 for the metal surfaces and for the isolators covered with reflective layers.

• The buoyancy model is turned on (free convection with laminar flow) and gravity acceleration is directed perpendicularly to the bottom plate; and
• The properties of the materials used are summarised in

  Table 1. Properties of the materials used for the simulations.

  	Density ρ [kg/m3]	Thermal Conductivityk [W/(mK)]	Specific Heat cp [J/(kgK)]	Dynamic Viscosity µ [Pa·s]	Expansion Coef. β [1/K]
  Aluminium	2702	237	903
  Steel	7854	60.5	434
  Polystyrene	55	0.027	1210
  Air (25 °C)	1.185	0.026	1004	1.83 × 10−5	3.36 × 103

  [24].

3.2. Results of Simulations

The numerical steady-state analysis focused on six different designs of the stand dome (see cases a–f in Section 3.1). The main goal was to investigate the thermal performance of the stand. The sample temperature distributions within the cross section of the modelled structure for the dome walls made of nonconductive (case a) and conductive (case b) materials are presented in Figure 3. Higher temperature values dominate in close proximity to the lamps. The temperature of air under the dome is characterized by high uniformity, and the values are higher in the case of insulators with rough surfaces and high emissivity values.

To compare the designs of the six investigated domes, the results that cover the power transferred to the plate (P _{cold plate}) in absolute and relative units are gathered in

Table 2. Results of simulations for the different dome designs.

Material/ Emissivity	P (Cold Plate) [W]	P (Cold Plate)/P Lamps [%]	TEave (Dome External Wall) [°C]	TIave (Dome Internal Wall) [°C]	Tmax (Lamp) [°C]	Tave (Air) [°C]
a. insulator with e = 0.9	54.4	32	44.0		125.8	49.1
b. steel with e = 0.1	101.2	60	34.1		135.4	36.9
c. double wall 5 mm with e = 0.9	70.9	42	36.5	46.4	131.1	57.8
d. double wall 5 mm with e = 0.1	118.2	70	31.5	37.3	143.0	44.7
e. double wall 10 mm with e = 0.1	124.3	73	30.7	40.2	147.0	51.6
f. double wall 20 mm with e = 0.1	132.2	78	29.6	45.3	152.3	62.6

. They are accompanied by the average temperature of the dome walls (T_Eave—external in all the cases, T_Iave—internal in all double wall cases), the average temperature of the air under the dome, and the maximum temperature T_max on the surfaces of the lamps.

The use of double walls with the insulation layer on the stand allows a reduction of heat losses to the surrounding area. The highest power transfer to the cold plate is equal to 132.2 W and it is obtained for the thickest tested wall. An increase of 18 percent points is noticed in comparison to the single layer metal dome. On the other hand, the average air temperature under the dome also increases with the increase of the thickness of the insulator. The presented results also depict the influence of the internal surface emissivity on the thermal properties of the stand. The cold plate absorbs more irradiance if the double wall with 5 mm insulation is covered with a reflective layer (case c vs. case d). Furthermore, surface emissivity and additional insulation thickness influence the temperature of the dome wall (both internal and external, once). It should be noted that in case f, the average wall temperature is below the cold plate temperature due to the fact that the ambient temperature is 25 °C.

The conducted analysis allowed the identification of the construction to be manufactured. The single conductive wall design with reflective inner surfaces allows the transfer of up to 60% of irradiated power to the cold plate while keeping the air temperature at a slightly higher level in comparison to the ambient values. Too high an ambient temperature under the dome is not a desired phenomenon due to the fact that it changes the operating conditions of the tested PVT modules.

4. Experimental Verification

4.1. Experimental Setup

In order to verify the results of the numerical analysis, the experimental setup of the Artificial Sun and the entire measurement system has been manufactured. Figure 4 depicts photos of the stand.

The experimental setup has differed with respect to the modelled design in the maximum number of lamps foreseen to be mounted on the dome. In the final version, their number was limited to three. In order to accurately simulate solar radiation, halogen lamps of 160 W and 230 W have been used. A test PVT structure, which is based on the Artificial Sun setup, contained a minimodule with four parallel mini-channels, each 2 mm in height, covered with a 2 mm thick aluminium plate. The plate, as indicated in [22], is the base on which the silicon solar cells have been mounted.

In the experiment, water was used as a working fluid. A Lytron RC030 recirculating chiller operating in a closed-loop system guaranteed flow rates of 0.25 to 6.0 L/min, with a constant inlet temperature of 20 °C. During the tests, water flow rate, the inlet and outlet temperature of the liquid, the temperature in the middle of the minipanel top surface (between the solar cells), and the current–voltage characteristic of the PV module were collected. Three Pt100 sensors with LI-24 temperature transmitters were used to monitor the temperature values, and the water flow rate was registered with a PEM 1000ALW electromagnetic flowmeter. Furthermore, electrical power and I–V characteristics were measured with a Keithley 2401 computer-controlled current-voltage source.

4.2. Results and Discussion

The experimental measurements started with testing different configurations of the light sources. The aim was to measure the incident irradiation intensity of the halogen lamps and to check its uniformity in order to select the setup that ensured the expected average values across the entire plate/front of the PVT minimodule. The experiment with various lamp settings was to obtain two weather conditions: STC (standard test conditions) and NOCT (normal operating cell temperature).

The light intensity was measured by placing a set of light intensity meter probes on the surface of the aluminum plate at nine evenly spaced points (Figure 5, left). Various configurations of lamps with powers of 160 W and 230 W were tested. For each configuration, several measurements were conducted to check the reliable recurrence of the results. The obtained results were collected, and they were presented as color maps, histograms, and comparative tables. The sample results of the irradiation energy values, depending on the lamp power and configuration, are given in the graph shown in Figure 5 (right).

On the basis of the obtained results, configurations with a single lamp were rejected due to the low values of light intensity and the high unevenness of lighting. The 3 × 160 W configuration was characterized by very good uniformity of illumination, which was better than the configurations with two lamps (2 × 160 W and 2 × 230 W). However, if better regulation to lower the value of light intensity is demanded (e.g., 800 W/m²), setting a uniform illumination is much more difficult for a configuration with three lamps than in the case of a configuration with two lamps. For this reason, three or more lamp concepts were abandoned. According to the obtained results, to simulate the STC conditions, two lamps of 230 W need to be used, and for NOCT conditions, two lamps of 160 W need to be used.

Additional improvements for two-lamp configurations that were assumed to improve the tested parameters have also been examined. As a result, two of the implemented changes had a relevant influence on light distribution and intensity:

• changing the angle of lamp setup to direct the light into the centre of the plate; and
• removing the covering/protecting glass to increase the incident irradiation value.

Figure 6 shows the irradiation energy graph for the configuration of 2 × 160 W lamps and the selected modifications, while the calculated parameters are shown in

Table 3. Parameters of irradiation of the 2 × 160 W lamp configuration with modifications.

Modification	Average Irradiation [W/m2]	Standard Deviation [W/m2]	Relative Standard Deviation [%]
None	713	43.6	6.11
Angle of 15°	702	29.6	4.22
No glass	1045	68.2	6.53
Both of the above	1049	33.8	3.23

.

The presented results led to the conclusion that setting the lamps towards the center of the measuring plate at an angle of approximately 15° does not affect the average value of the light intensity; however, it significantly improves the uniformity of the light intensity. On the other hand, the results clearly show that removing the protective glass from the headlamps significantly increases the irradiation intensity values. Applying the variant without the protective glass allowed us to achieve light intensity values close to the AM1.5 spectrum standard (the fixed value of the intensity of solar radiation reaching the Earth, which is equal to 1000 W/m²). Therefore, further experiments using the designed Artificial Sun stand were conducted with the chosen selection of lamp setup.

To verify the energy balance of the stand, the heat transfer rate to the water with respect to its flow rate was tested. The obtained results are presented in Figure 7. For the lower velocities of the liquid where heat exchange was poorer, the total power transferred to the cold plate exceeded approximately 60% of the irradiation energy, while the value reached 92% for the highest tested water flow rate. It was in good agreement with the simulations presented in Section 3.1.

Next, the measurements were conducted for the test PVT minimodule. Current–voltage characteristics and electrical power on the PVT minimodule were collected, along with simultaneous temperature measurements. After the module reached the temperature of approximately 80 °C (the value monitored in the middle of the top aluminum plate between the solar cells), the cooling system was turned on (with a water flow rate of 6 L/min), which resulted in improved photovoltaic performance. Figure 8 presents the photovoltaic module’s maximum power value as a function of its surface temperature, with the cooling system start indicated.

The Artificial Sun stand for the PVT minimodules testing was designed in detail and manufactured with several modifications implemented. The first results obtained for the laboratory-made, mini-demonstrator of the PVT system proved the suitability of the setup and the usability of the device for future developments in the photovoltaic thermal area.

5. Conclusions and Discussion

The paper focuses on a new testing setup for T and PVT modules. This is a close design, indoor solution with compact dimensions that is prepared to conduct experimental verification of the minimodules with the reduced length and width in comparison to conventional panels. The stand was designed with the support of numerical analysis that was oriented on thermal properties. Therefore, the simulations covered different dome wall designs and various materials and surface treatments to depict the influence of these factors on the energy balance and the temperature field in the air area under the dome. Even though a higher amount of lamp energy transfer can be obtained for double-layer walls, the decision to manufacture metal single-layer walls was made. The main reason was the possibility to keep the air temperature under the dome at reasonable level, while keeping a good energy balance (60% of lamp power). The simulations also indicated that there was no need for four lamps, and thus, at the manufacturing stage, the number of mounting sockets was reduced to three. To obtain the demanded solar irradiance conditions, different configurations of halogen lamp settings were tested. Two 160 W lamps and an inclined angle assembly were selected. Thermal verification of the final stand confirmed the results of the simulation (57% of the lamp power was transferred to the cold plate for the highest tested water flow rate, which was above 90% of energy irradiation).

To sum up, the Artificial Sun stand has been designed, manufactured, and tested. It is an indoor solution that allows conducting experimental verifications with a minimal influence of environmental conditions. This is the first reported closed-design measurement setup for T and PVT modules. As depicted in the paper, it ensures uniform energy irradiation (a value of 1049 ± 34 W/m² was achieved), with light intensity close to the AM1.5 spectrum standard. The conducted thermal measurements illustrate the influence of the coolant flow rate on the operation of the PVT module, as well as an increase in the amount of electricity produced as a result of the lower operating temperature of the solar cells. Subsequently, the stand is planned to be used for new minimodule [22] tests. The uniqueness of the novel PVT modules results from their level of integration and the manufacturing techniques used. Furthermore, different standardisation models used for solar thermal collectors are going to be applied to verify the possibility of avoiding the influence of coolant flow rates on the calculated efficiency values.