Heat recovery potential and electrical performances in-field investigation on a hybrid PVT module

Applied Energy 205 (2017) 44–56

Contents lists available at ScienceDirect

Applied Energy journal homepage: www.elsevier.com/locate/apenergy

Heat recovery potential and electrical performances in-field investigation on a hybrid PVT module

MARK

⁎

Diego Vittorini , Nicola Castellucci, Roberto Cipollone University of L’Aquila, Department of Industrial and Information Engineering and Economy, L’Aquila, Italy

H I G H L I G H T S and hybrid PVT modules thermal behavior in-field characterization. • PV distribution in solar cells monitoring through infrared thermography. • Temperature efficiency/environmental conditions correlation. • Electrical cooled PVT design/implementation/validation (model-based/experimental). • Water • Heat recovery assessment in PV and PVT prototypes via bottoming ORC plant.

A R T I C L E I N F O

A B S T R A C T

Keywords: PV-thermal module model Conversion efficiency Solar cells cooling Solar-based micro-cogeneration

The aim of the present study is the characterization of PVT modules electrical performance in real operating conditions, as well as the investigation of thermal recovery via a cooling circuit integrated with a third generation PV module. The approach combines both theoretical and experimental tools: a MatLab® simulation model provides a reliable theoretical basis, whose validation is performed on experimental evidences from in-field PV module tests. The model represents the module energy balance, under unsteady operating conditions; a full set of measurements allowed to validate the theoretical approach, thus offering the possibility to evaluate the effects of both variable outdoor air temperature and pressure and wind speed. Water-cooled PV modules electrical performance increases by as much as 33%, with respect to the situation in which no cooling is performed and up to a 20% electric efficiency is achieved, with a 2.0 L/min water flow rate on the back. A major drawback is that thermal recovery for cogeneration purposes is not effective, due to a low thermal gradient (10 K maximum) on the water. When a 10 mm thick glass cover was integrated in the PV module along with a frame to reduce wind circulation over exposed surfaces, a 100–500 W thermal recovery on a day basis could be achieved. Furthermore, a 15–30 K increase in water temperature assures about the higher quality of the recoverable heat. The suitability of organic fluids instead of water to reduce the power absorption by the pump, is addressed as the most effective way to increase PVT electric output: the absorption with R236fa and R245fa is a 60% and 75% lower than with water, respectively. The novelty of the present study lies in the dual theoretical and experimental approach, leading to a validated non-steady-state mockup, easily adjustable to extend the analysis to PV modules arrays for residential applications. Furthermore, the use of an infrared camera, typically confined to post-manufacturing quality control, allows here a continuous monitoring of the module thermal field, key to evaluate the temperature effect on the module performances both in steady and unsteady conditions.

1. Introduction In the present industrial practice, PV modules are mainly based on either the sc-Si or the mc-Si cells technology, the advantage of the former being a higher efficiency in electrical conversion, whereas the

⁎

latter shows a higher durability. As the investment cost of mc-Si modules is lower, this technology, despite yielding up to 35% less electric energy per unit of surface area, is preferable when there are no strict space constraints for the installation and soft costs (design, installation and permits) represent an unimportant share of the total life-cycle cost.

Corresponding author. E-mail addresses: [email protected] (D. Vittorini), [email protected] (N. Castellucci), [email protected] (R. Cipollone).

http://dx.doi.org/10.1016/j.apenergy.2017.07.117 Received 9 May 2017; Received in revised form 18 July 2017; Accepted 26 July 2017 0306-2619/ © 2017 Elsevier Ltd. All rights reserved.

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

w Q̇ P Δ f M p u ε η μ ν ρ N s U air el m out pump sky v i

Nomenclature

h C H Nu Pr Ra k S T c g σ X V a PVT conv irr mod p rad th

convective heat transfer coefficient [W/m2 K ] thermal capacity [J/kg K ] pump hydraulic head [m] Nusselt number Prandtl number Rayleigh number thermal conductivity [W/m K ] surface [m2 ] temperature [K ] specific heat [J/kg K ] standard gravity acceleration [9.086 m/s2 ] Stefan-Boltzmann constant [5.67·10−8 W/m2 K4 ] measured value variance sensitivity PV thermal convection irradiation PV module constant pressure radiative thermal

water thermal power [W] power [W] variation flow rate [m3/s ] mass [kg ] pressure [mbar ] speed [m/s ] emissivity efficiency dynamic viscosity [kg/m s ] kinematic viscosity [m2 /s ] density [kg/m3] number of values standard deviation uncertainty air electrical motor output pump sky constant volume i-th measured value

configuration for a PVT system is features a thin flat copper sheet instead of Tedlar as the bottom layer and a single water channel, as in [18]: a 35.3% primary energy saving efficiency is appreciated on a 20.9% efficiency PV module. Kalogirou [19–21] has examined the effect of control parameters, such as water inlet and outlet temperatures, operating pressures and mass flow rate, on PVT performance, whereas the effect of temperature reduction on thermal and electric efficiency has been directly evaluated by Krauter and Ochs [22]. A similar sensitivity analysis, investigating both dependence of electric and thermal performance of a sheet-and-tube PVT collector for DHW production, on control parameters (e.g. pump operation, thermostat controller, flow rate), climate data and cells optical properties is performed in [23]. Ehsan Fadhil Abbas Al-Showany [24] recently investigated the impact of weather conditions on PV module performance by performing experiments on identical PV modules in a given location. The reduction in electric yield by unclean panel due to natural pollution deposition over a 3-month period was about 3.8% compared with the clean panel, and 13.8% compared to the water-cooled clean panel. Other studies assessing the performance of PVT systems are due to Tiwari and Soda [25,26], not to mention Chow studies focused on a PVT/thermal waterheating hybrid system [27,28] and Ben cheikh el Hocine et al. [29] who created a model for PVT collectors, though not explicitly providing a direct comparison between the model results and experimental measures. A comprehensive financial analysis of a PVT system for residential and commercial applications is in [30], where a sensitivity analysis clearly shows how both the net present value and the discounted payback period encourage the installation of a PVT system instead of a conventional side-by-side PV and solar thermal system. The environmental impact of hybrid PVT systems is discussed in [31], where a review of the published results aims at the definition of a common ground to address in the design phase and market

Less common, but still commercially available, are the PV modules based on a-Si cells, whose efficiency is the lowest among all available technologies, not to mention its sensitivity to ambient conditions. The characteristics of three main types of PV modules are in Table 1. Since, for any of them, efficiency is relatively low and not expected to significantly increase in the near future, the electrical yield can be maximized using add-on technologies, such as concentrating devices and cooling devices. The former increase the amount of solar energy collected by a given PV module, whereas the latter reduce the PV module operating temperature, hence increasing its efficiency and making heat available for cogeneration purposes as a side effect. In this paper the authors examine both a bare PV module and a PV module equipped with a tailored cooling system (PVT) and compare the efficiency of the two under same operating conditions. An extensive literature investigates all aspects related to PVT systems, from manufacturing to operation: the first attempt to define optimal configurations dates back to 1970 and it is due to Wolf [2] and Florschuetz [3], who aimed to provide a reliable modelling platform which could guide future research and development. Kern and Russel [4] and Garg and Agarwal [5] have investigated both air- and watercooled PVT systems, and Tripanagnostopoulos [6–8] has come to the conclusion that the former lead to an efficiency of 38–45% in c-Si and aSi modules, whereas the latter lead to a much higher efficiency of 55–60%. Tonoui and Tripanagnostopoulos [9] have studied the improvement in performance achievable by adding a suspended metal sheet in the middle of the air channel or by implementing finned surface to boost the heat exchange. Among the main findings of both theoretical and experimental studies is that water-cooled PVT systems have greater efficiency and also greater margins for improvement than air-cooled PVT ones. Nonetheless, major advantages of the latter are in the lower financial investment required and the more favorable life cycle cost, as shown in Raman and Tiwari [10]. Detailed studies by Chow et al. [11] aim to model PVT collectors behavior under a variety of environmental conditions, in addition to investigating the optimal configurations in terms of collector design [12,13], glazing [14,15] and fluid circulation [16]. Three types of uncovered PVT collectors are investigated in [17], with an extensive experimental campaign supporting both model calibration and validation, thus allowing a safe estimation of the annual yield for each specific PVT type. An enhanced

Table 1 PV modules main characteristics [1].

Cost [€/kW] Efficiency [%] Lifespan [years]

45

a-Si

mc-Si

sc-Si

1375 9 20

1562 15 20

1750 20 20

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

power lost to the surrounding environment by convection or re-irradiation. The paper also investigates the room for improvement in thermal recovery achievable through cooling fluids other than water, namely organic fluids. Further benefit associated to such an intervention is proved to be able to reduce power absorption by the pump, due to lower mass flow rates involved when the reduction on cells operating temperature is assigned. A similar approach to PVT systems characterization can be found in [42], where different collector fluids are tested and where water is presented as a better fit with respect to ethylene glycol, enabling a 25% efficiency enhancement.

development for the PVT technology. In this paper, the effect of both ambient conditions (i.e. solar irradiance, air temperature and speed) and cooling configuration adopted (i.e. water flowing in a pipe on the module back) on a commercial 1.7 m2 sc-Si PV module is investigated. Although the electrical performance dependence on solar cells operating temperature has been investigated, among the others, by Ravindra and Srivastava [32], Swapnil Dubey et al. [33], Skoplaki and Palyvos [34], and Ozgoren et al. [35], none of these studies provides an explicit mathematical correlation between electrical performance and ambient conditions, as done in the present paper. The role environmental conditions have in affecting PV module performances is further discussed in [36]: a different approach to evaluate the effect of weather and outdoor exposure on PV modules is based on the accelerated testing of the modules and a thorough investigation accounts for various types of PV module, both indoor and outdoor and for all main environmental variables impacting their performances. A further element to account for, in order to enable the comparison among results from various Authors, is the difference in the experimental conditions considered, case by case; for instance, despite the study by the Authors and the one by Ozgoren have similar goals (e.g. to correlate electrical efficiency and solar cells temperature) they slightly differ from each other, mainly because the two experimental campaigns examined PV modules with different electric efficiencies at design point and took place in different times of the year and in different climatic zones (July, Turkey vs November, Italy). Moreover, standard values for wind speed and solar radiation were considered by Ozgoren et al. (i.e. 1 m/s and 1000 W/m2 respectively), instead of actual values for all environmental parameters as done in the present study. Closer to the aim of the present paper, the studies by Ceylan et al. [37,38], present an effective approach to PVT module performance evaluation, based on process control automation, to allow the sweeping of a variety of operating conditions and to account for various cooling configurations, to select the best heat removal option and to extend system applicability to large-scale systems. A novel cooling technique, based on the heat removal from the PV modules achievable through water vaporization is presented in [39]: its adoption on both sc-Si and mc-Si 50 W nominal performance modules shows beneficial effects on the electric efficiency: the efficiency increase on a 9 h basis ranges between 3.0% and 8.8% in sc-Si modules. Same values for mc-Si modules are 4.2% and 9.4% in mc-Si. A further advantage is in the fact that the system operates without losses of flowing water. The financial dimension of such systems is investigated as well and the achievable financial benefit, with respect to the non-cooled configuration is discussed for different contexts: a minimum saving of 178 € is possible in Croatia, a maximum 2823 € in Australia. An overview of main findings research activity produced on PVT systems in both in-field and lab applications is in Table 2, where the efficiency increase is presented along with the cooling technique adopted. Once a model for the unit unsteady energy balance is defined and validated, based on data from an extensive experimental campaign, a clear indication of the electrical and thermal contributions to the overall performance comes out. Thermography happens to be particularly handy, since it allows at one time both the continuous thermal monitoring of the PV module and the detection of defects (e.g. hot spots, bus bar degradation), responsible for performance degradation [40,41]. Apart from dealing with the in-field energy characterization of a typical commercial PV module in a variety of possible scenarios, the study presents a combined experimental/numerical routine, that suits the energy characterization of any PV/PVT system of interest, once the model degrees of freedom and variables are accordingly set. The parameters, to which the system performance is sensitive the most, are detected; hence a guideline in the selection of a proper control strategy for the unit is provided. In particular, with regard to the possibility of enabling co-generation, the Authors examined the increase in thermal power recovered via refrigeration when the PV module is shielded from the wind and thermally insulated, so to drastically reduce the thermal

2. Materials and methods In order to run the simulation, it is necessary to measure air temperature, pressure and speed. Furthermore, in order to validate the results of the simulation model it is necessary to measure the PV module operating temperature via an infrared camera. The test facility, whose pictures and layout can be found in Figs. 1 and 2 respectively, was installed on the rooftop of a building within the premises of the University of L’Aquila. Main features for the PV module under investigation are in Table 3 and refer to performance evaluation at 318.2 K Nominal Operating Conditions Temperature (NOCT), a 800 W/ m2 irradiance, 293.15 K air temperature and 1 m/s wind speed. The full set of uncertainties associated to the measurement equipment, as provided in data sheet, is in Table 4. The layout in Fig. 2 shows also the cooling system used to refrigerate the panel, via an electric pump-driven water circulation, whose main components are: ● the PV module; ● the meteorological station (MS) and infrared camera (ID); ● the electrical circuit, with access hubs for PV electrical output monitoring and; ● the hydraulic circuit for coolant circulation. The cooling system was not embedded in the original test set, that aimed at characterizing the PV module as originally manufactured. Fig. 3 shows more details of the cooling system, with a view of the module back along with the cross section of the piping system, that has been obtained from a suitably bent coiled copper pipe. The serpentine pipe is mounted on the module back with the interposition of a conductive layer in order to guarantee that almost all heat flux is collected by water circulated in it, rather than being dispersed toward the external environment. An extensive experimental campaign was conducted on a 30-days Table 2 PVT systems - Efficiency increase for main cooling techniques in literature.

46

Module type

Efficiency increase (%)

Cooling technique

Ref.

a-Si mc-Si, a-Si mc-Si, mc-Si mc-Si sc-Si mc-Si, mc-Si, a-Si mc-Si mc-Si, mc-Si, sc-Si mc-Si, mc-Si sc-Si

39.0 26.0 54.0 43.0 30.0 28.0 14.2 14.5 5.3 30.7 25.3 12.0 10.0 5.6 3.2 7.1 6.4

Air cooling (prototype) Air cooling (prototype) Liquid cooling (prototype) Liquid cooling (prototype) Finned air channel (prototype) Thin metallic sheet (prototype) Glycol/water cooling Water + thin copper sheet Finned water channel Thermosiphon system Thermosiphon system Water + copper sheet, unglazed Water + copper sheet, glazed Water + copper tube Water Water vaporizing Water vaporizing

[7] [7] [7] [7] [9] [9] [17] [18] [19] [20] [20] [23] [23] [35] [37] [39] [39]

sc-Si sc-Si

sc-Si sc-Si

sc-Si sc-Si sc-Si

Applied Energy 205 (2017) 44–56

D. Vittorini et al. Fig. 1. Test facility.

safely assumed as the least favorable for heat recovery, as ambient conditions provide a negative contribution to the module heating-up, differently from what happens during warmer seasons. In other words, the only recovery is possible on the heat actually associated to the module thermal and electrical behavior and operation-related losses and no positive contribution to the thermal energy available to the water comes from the environment. Consequently, the paper provides a clear indication on the minimum heat recovery achievable on a system as the one considered here. With higher ambient temperatures and a higher solar radiation available, a higher recovery could be easily attained, but then an environment-related bias would affect the results. A 15 min time step has been selected for the acquisition of all relevant physical quantities, i.e. electric power, solar radiation and operating temperature, as a smaller step would have resulted in no significant benefits in term of analysis precision. The electrical parameters direct detection led to data in Fig. 4. The top part show the PV module electrical yield as a function of the solar irradiance, in both reference and real operating conditions (RTP and STP). The bottom part report the PV module front temperature at the same values of solar irradiance, both during RTP and STP. While real operating values are measured directly, the electrical output at design points relates to the solar irradiance through the module performance index, as declared by the manufacturer: this explains the significant gap between the curves, as real operating conditions almost always differ from normal operating ones. Comparing RTP to STP, it can also be noted that electrical yield and performance is always higher during the former, due to a lower thermal level of the solar cells in the module: this appears clear matching information from both the top part and the bottom part of Fig. 4: when the temperature is the same, the PV electrical performance has the same value, regardless of which phase is occurring between RTP and STP. In first assumption, the module temperature is considered uniform on the whole module surface. Given the PV module emissivity (ε = 0.85), the module timevarying operating temperature can be derived from the knowledge of ambient conditions and module geometrical features, by solving Eq. (1), which states the module overall energy balance:

Fig. 2. Test facility layout.

Table 3 PV module main characteristics. REC230PE [W] [V] [A] [V] [A] %

Pmax VOC iSC VMPP iMPP

ηmax =

VMPP·iMPP Pirr

170 26.8 6.3 33.6 6.8 17.0

base in November 2015, in both absence and presence of a dedicated embedded cooling system. Since the experimental activity has entirely been performed outdoor, the ambient conditions variability (i.e. solar irradiance, outdoor air temperature, wind speed and humidity) directly affects the measurements repeatability. In order to retain repeatability, a set of reference conditions has been selected and only those days complying with these conditions, have been taken into account in the analysis. The reference set is shown in Table 5 along with maximum variations of each physical quantity among the five days considered (“Diff” labeled column in Table 5). As known the results validity is affected by environmental conditions and ultimately by the season in which the experimental activity is carried out. The conditions the paper accounts for (i.e. experimental activity in the winter season) can be

̇ ̇ ) + mmod ·Cmod· Pirr = Pel,out + (Qconv + Qrad

dTmod dt

(1)

Direct measurement provided values for Pirr and Pel , whereas the values ̇ ̇ are a function of temperature, based on the convection and Qrad of Qconv heat transfer law and the emission law, whose expressions are shown in Eqs. (2) and (3), respectively:

Table 4 Measurement uncertainties. Quantities and related uncertainties Power Irradiance Electric output

Air and PV Module ± 4.0% ± 0.75%

Air temperature Air pressure

Water ± 2.0% ± 2.0%

Wind speed Module temperature

47

± 2.5% ± 2.0%

Water temperature Water flow rate

± 1.2% ± 1.5%

Applied Energy 205 (2017) 44–56

D. Vittorini et al. Fig. 3. Cooling circuit details.

a) PV module back

b) Cooling circuit cross section

Table 5 Reference conditions for the experimental campaign. Hour

Solar radiation

Diff.

W/m2 9 10 11 12 13 14 15

159.9 374.4 503.5 547.1 505.3 378.1 165.5

Ambient temperature

Diff.

K ± 2.5 ± 2.7 ± 2.1 ± 3.2 ± 4.8 ± 4.1 ± 3.9

Wind speed

Diff.

m/s

284.3 287.5 290.1 291.9 293.1 293.5 293.3

± 0.5 ± 0.7 ± 0.3 ± 0.2 ± 0.1 ± 0.3 ± 0.3

0.57 0.96 1.29 1.58 1.82 2.00 2.14

± 0.01 ± 0.02 ± 0.05 ± 0.05 ± 0.04 ± 0.01 ± 0.03

̇ Qconv = h·Smod·(Tmod−Tair )

(2)

4 4 ̇ = εmod·σ·Smod·(Tmod Qrad −Tsky )

(3)

under the assumption of clear-sky conditions, i.e.: 1.5 Tsky = 0.0552·Tair

(4)

The convective heat transfer coefficient (h ) is a function of measured ambient conditions, according to the Nusselt number expression for airdriven convection processes, involving flat plates, either vertical or inclined, as in Eq. (5): 2

0.387·Ra1/6 ⎫ Nu = ⎧0.825 + ⎨ [1 + (0.492/ Pr )9/16]8/27 ⎬ ⎩ ⎭

Fig. 4. Electrical performance and temperature effects.

(5) work, a trigger value for the PV module temperature must be provided. In addition to the assumption of a uniform temperature field on the module surface, a major approximation is introduced here: the PV module temperature is considered uniform on the module thickness (i.e. perpendicularly to the surface) too. In fact, due to direct exposure to the sunlight, the front part of the PV module, heats up more than the back part does, hence the balance equation should account for a conductive 1-D flux between the two surfaces. Typical temperature values for a normal day of operation are reported in Fig. 6 for both the module front and back surfaces: the initial 295.2 K difference shrinks at about 9 K already after 6 h of operation. Fig. 6 also features two lines representing the average real

Moreover, Eq. (5) fixes the Nusselt number dependency on Prandtl (Pr ) and Rayleigh (Ra) Numbers, with the latter being greater than 109. The flow diagram in Fig. 5 highlight the main steps of the calculation, needed to solve Eq. (1). A three-main-branches architecture applies to the flow diagram: the upper branch calculates air properties from ambient conditions and subsequently derives Nusselt number and the convective flux; the middle branch calculates the radiative flux starting from the PV module properties (emissivity and surface area); the lower branch receives measured values of solar irradiance and electric output. From a merge of all terms, the thermal power stored within the module comes out and the PV module temperature at the next step of calculation, comes out as well. Of course, for this method to 48

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

Fig. 5. Simulation model – flow chart.

transmitted to the environment is represented: radiation accounts for a slightly greater contribution, ranging from 55% to 65%, which increases during the day as solar irradiance increases. Looking at Fig. 8 the following conclusions can be drawn. Firstly, the model underestimates the heat loss during RTP and overestimates it during STP, in full agreement with what already noted for the temperature: this comes expected as the heat flux depends directly on the temperature, and more specifically the convective portion depends on the first power of T whereas the radiative portion depends on the fourth power of T . Secondly, the radiative heat exchange always prevails on the convective heat exchange, at any given time of the day (and, hence, at any given solar irradiance), particularly during the STP, when the PV module hits its highest operating temperature. These two aspects are also in agreement with the increase of radiative contribution against the convective contribution over time, as the Sun moves along its apparent path in the sky. It is worth observing that most part of energy entering the system is lost to the environment through convection and radiation, as clearly shown in Fig. 7. If this loss term is significantly reduced, preventing reirradiation with a glass cover and shielding the PV module from wind gusts, the operating temperature increases. If on the one hand, this leads to a decrease in the module electric efficiency, on the other hand the lower thermal loss to the environment assures that a significantly greater amount of heat is available for recovery. Moreover, higher temperatures shall be expected: the two effects, combined, end up favoring cogeneration.

temperature between the front and the back of the PV module, and the calculated temperature resulting from the simulation model: it is evident how the gap between these two curves is negligible, with a shift that never exceeds 1%. It can also be observed that during the first part of the day, from 9:00 to 12:00, while the sun is ascending along its path across the sky, the model tends to underestimate the PV module operating temperature. On the contrary, from 12:00 to 15:00, while the sun is descending along its path across the sky, the model tends to overestimate the operating temperature. Furthermore, due to direct exposure to sunlight, the front part of the panel, differently from the back part, heats up much more during early hours of the day than during late hours: from here on, the first part of the day will be referred to as RTP (Rising Temperature Phase), whereas the second part of the day will be referred to as STP (Stabilized Temperature Phase). The power balance from the simulation model, lead to Fig. 7. Bars show the trend of each term appearing in Eq. (1), and it is especially important to notice that the storage term Cmod·Mmod ·ΔT /Δt cannot be neglected when compared with the electric power generated, i.e. the target quantity of this study. This evidence justifies the non-steady-state approach adopted, as opposed to the quasi-steady-state approximation which would result from Eq. (1) when the storage term was ruled out. The total heat loss to the environment, through either convection and irradiance, was calculated from Eqs. (2) and (3), both inputting the actual operating temperature and the simulated temperature resulting from the energy balance expressed in Eq. (1). These losses account for the most significant part of the input solar energy. In Fig. 8 the percent weight of each heat transfer mechanism to the overall thermal power

Fig. 6. PV module characterization: temperatures at back and front.

49

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

Fig. 7. Energy balance – simulated power fluxes.

the possible choice to few, or possibly one, selected options. First off, a proper insulation is needed to avoid that part of the heat to be recovered is transferred and lost to the surroundings. In the design phase, among all major related issues to address, particular attention deserves the cells shadowing. As a consequence, the cooling system should be conveniently installed on the module back. With respect to the cooling, the choice is limited by constraints related to environmental impact, leakage-related health risks and financial feasibility, especially when the application is intended for the residential field. These requirements, along with the prototypal nature of the application, led to water as the cooling medium. Thermal information from infrared camera and thermistors allowed to shrink even more the field of viable solutions eligible for heat removal. Experimental evidence is in general agreement with the assumptions made at the model level and in particular the temperature distribution is uniform on both the front and the back of the module. As a matter of fact, the heat transfer only takes place perpendicularly to the module’s surface (mono-dimensional flux) and no heat flux occurs between back and front surface. The cooling system consists of a sheet-and-tube configuration, with a copper coil pipe used to circulate water to the module back. Conductive Silicon guarantees the pipe/module physical and thermal contact and an insulation layer on the pipe surface prevents the heat dispersion to the outer environment. Fig. 2 reports the schematics for the test bench used to investigate the water cooled PV module: pump speed modulation allowed the mass flow rate regulation. Pressure increase and flow rate measurements were needed to estimate the work done on the fluid and, through the pump and electric motor efficiencies, the electric power absorption by the pump. The system monitoring occurred for four different water mass flow-rates: for each of them, the module electric power, the thermal power removal by water and the cell operating temperature have been measured, thus estimating the achievable electrical benefit. The electrical and thermal efficiencies are defined as follows:

9

Fig. 8. Relative weight of heat exchange mechanisms and model error.

3. Results and discussion The technological integration between a standard PV module and a cooling system is possible in many different ways, among which some are preferable, according to the goal to achieve, be it the maximization of either the electrical output or the total output of the system, i.e. the sum of electrical power and recovered heat. Some considerations along with a continuous thermal monitoring of both the front and rear surface of the PV module, the former through the use of the infrared camera and the latter making use of thermistors, allowed for narrowing down Table 6 Water cooled PV module – Main indicators at various coolant flow rates. Measured values

Uncertainty

Flow rate

ΔTw

Δpw

Thermal power recovered

Efficiency

l/min

K

bar

W

Electrical %

0 0.5 1.0 1.5 2.0

– 10.4 6.1 3.8 1.9

– 0.5 0.5 0.5 0.5

– 49 98 115 84

15.1 17.2 18.7 19.6 20

50

Power

Efficiency

Thermal %

%

Electrical %

Thermal %

– 5.3 12.7 15.1 11.8

– 5.5 32.1 31.9 51.8

3.4 9.2 2.8 1.7 0.9

– 2.8 2.9 2.1 4.2

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

ηel,PVT =

Pel,out −Pel,pump

ηth,PVT =

(6)

Pirr

Q̇ w Pirr

(7)

Main results obtained are in Table 6, along with the uncertainty Fig. 9. Water cooled PV module – main indicators at various coolant flow rates.

a) Electrical efficiency

b) Module temperature

c) Water temperature 51

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

gain, when R236fa and R245fa are used, respectively. The viability of organic fluids as alternative to water in module cooling opens the way to a further development of the system, e.g. the implementation of measures to reduce heat losses to the environment, hence increasing the amount of recoverable heat. On the same level, the coupling of a properly designed section for thermal storage to be integrated to the PV module, would allow a higher thermal energy available at a reasonably higher temperature levels [43–45]. When a target heat removal is set, the increased amount of thermal energy at the module, due to a lower exchange to the surroundings, would call for a more intense cooling and, as a consequence, for either a significantly higher water flowrate, leading to a higher power absorption by the pump, or the adoption of a phase-change coolant, to keep lower values for the flowrate (especially in large scale plants) and to prevent the electric gain from being offset by the power absorption at the pump. It is worth observing that, regardless of the flow rate considered, the recovered thermal power is pretty much unusable for cogeneration purposes, as the water temperature experiences very low increases: based on experimental evidences, 298 K is the upper threshold for water temperature, which corresponds to a 10 K increase in water temperature. Furthermore, the highest temperature increase occurs at the lowest flow rate, when the thermal power recovered is at its lowest, hence requiring a much larger absorbing area for the PV system, possibly coupled to solar concentration devices, in order to even consider the possibility of cogeneration. Heat availability for further uses and the need to have a negligible power absorption by the pump along with the system automation (i.e. the adoption of automated process control devices) are the keys to system development and scalability to residential sector and building applications. The model consistency was verified with reference to the water cooled situation, by implementing the energy balance according to Eq. (16), in which the need to account for heat removal by water requires an additional term with respect to those in Eq. (1):

analysis results, as presented in [31]. Main parameters to uncertainty evaluation are defined in Eqs. (8)–(12):

X=

1 N

V=

1 N −1

∑

Xi

(8)

∑ Xi2−X 2

(9)

s=

V

(10)

a=

1 N

(11)

U=

∑i =1 ai2 ·Si2

R

(12)

Fig. 9 shows the variation of ηel , Tmod and Tw with the coolant flow rate. It is worth observing that a common element to all cooling conditions is that higher values for electrical efficiency always occur when solar irradiance is in the 450–920 W/m2 range, with the peak efficiency typically reached at a 550 W/m2 irradiance. The maximum scatter around it, within the abovementioned irradiance range is 1.5% in the non-water cooled scenario, whereas the lowest one happens with a 2.0 L/min cooling rate and ranks at 0.4% (i.e. four times lower than the maximum one achievable), which suggests how the higher the coolant flow rate is the higher are the performances control and repeatability. Nonetheless, the more the cooling regime drifts toward a medium-to-low level, the lower the gain in the achievable electrical yield becomes (Fig. 9a). In particular, when the coolant flow rate increases from 1.5 L/min to 2.0 L/min, the absolute gain in the electric power generated is only 3.65 W, this figure being the product of the percent increase of electrical efficiency (0.4%, from 19.6% to 20%), the solar irradiance (550 W/m2), and the actual absorbing area of the PV module itself (1.7 m2). As a consequence, higher flow rates are not worth being investigated, given the negligible increase of electric performance. Moreover, an increased water flow rate would lead to a higher power absorption by the pump: even though that contribution hardly exceeds 2 W in all swept situations (Table 7), even the slightest increase in power absorption has to be avoided, given the very low power output for the whole system. The power the pump requires is in Eq. (13):

Pel,pump =

̇ ̇ ) + Mmod ·Cmod· dTmod + Q̇ w Pirr = Pel,out + (Qconv + Qrad dt

The panel front temperature was picked as the reference term, the simulated temperature could be compared to, for each considered water flow rate. Main reason for this was the scarce data availability on the panel back temperature and consequently the impossibility to estimate an average value, representative for the thermal level of the whole system. Anyway, this lack of data did not represent a shortcoming, since the thermal gap between PV module front and back, already negligible in the non-water cooled case, was expected to further reduce in presence of a cooling system. Fig. 10 reports the PV module front temperature for different water flow rates, as already shown in Fig. 9b, with the addition of a dashed line for the simulated temperature. Among the most noteworthy findings, Fig. 10 suggests that:

ρw ·g ·H ·fw ηm ·ηpump

(13)

where a 0.9 efficiency is assumed for the electric motor (ηm ) and a 0.8 efficiency applies to the pump efficiency (ηpump ), and the hydraulic head, or the energy provided by the pump, is calculated as:

H=

1 1 2 2 ·(p −p ) + ·(vout −vinl ) + (z out −z inl ) ρw g out inl 2g

(14)

By inputting the actual values into Eq. (13), the electrical power absorbed by the pump becomes:

Pel,pump = 0.84·105·fw [W]

(16)

(15)

● the difference between simulated and measured temperature never exceeds 275.2 K, which in relative terms translates into an absolute error always lower than 1%; this confirm how close module front and back thermal profiles are one other; ● the higher the water flow rate, the flatter looks the temperature curve, this meaning that RTP and STP are no more as distinct as they were in the case of a non-water cooled module; ● the model tends to underestimate the temperature in the first hours

3

with fw in [m /s]. In order to evaluate system applicability, the pump power absorption needs to be compared to the increase in electric yield with respect to the case of non-water cooled PV module (i.e. a 25-to55 W power increase). As in Table 7, the power absorption by the pump accounts for a 2.5–4.5% the gain in the electric output for the module. It is worth observing that a different selection of the cooling fluid, could easily lead to a lower power absorption by all devices employed to allow coolant circulation on the back of the module. Such a reduction effect is mostly due to the higher thermal capacity some fluids (e.g. organic fluids) have with respect to water, and thus the lower mass flow rate the pump or the compressor have to process, for a given useful effect in terms of removed heat. Table 8 reports the power absorption by the pump, when a 115 W heat removal is considered and a 0.5 bar pressure difference applies: it represents a 1.3% and a 0.9% the electric

Table 7 Electric power absorbed by the pump for different flowrates. Water flowrate [l/min] Pump electric power [W] Electric gain [W]

52

0.5 0.58 22.5

1.0 1.16 38.4

1.5 1.74 47.8

2.0 2.31 51.6

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

Table 8 Electric power absorbed by the pump for different fluids. Fluid

Water

R236fa

R245fa

Flowrate [l/min] Pump electric power [W] Electric gain [W]

1.5 1.74 47.8

0.56 0.7 48.8

0.38 0.44 49.1

of operation and to overestimate it in the following hours, in full agreement with the model behavior already observed in the case of non-water cooled module. A difference, though, is in the fact that the higher the flow rate, the sooner does the transition from underestimation to overestimation occurs, with the shift due to the RTP shrinking; ● when the extracted thermal power increases from 49 W to 115 W (see Table 6), the measured and simulated thermal profiles match better; this assures about the temperature uniformity on the panel and, consequently, confirms the validity of the lumped parameter model in Eq. (14).

Fig. 11. Electrical performance comparison between PVT and PV.

higher recovery, still keeping the same PV cells operating temperature. This could be done by reducing both the module re-irradiation and the wind speed around exposed surfaces. A 10 mm thick glass cover with 97% and 25% transmittance for visible light and infrared radiation, respectively, reduces the infrared radiation to up 83%, which corresponds to a raw 50% decrease of the overall thermal losses. This energy could be recovered by cooling the panel and restoring the cells temperature to usual conditions. During the day, 70-to-340 W could be additionally recovered and, at a 0.5 L/min flow rate up to 20 K increase in water temperature can be attained. Furthermore, also convection losses could be reduced, as they represent a 50-to-75% share of the radiation losses (Fig. 8): the recoverable power could be further increased of 50-to-150 W, by encapsulating the module in a frame, which decreases the wind speed on the surfaces. Finally, a relation between the PV module electric performance and ambient conditions, more precisely outdoor air temperature and wind speed, has been derived by interpolating the

The percent electrical efficiency increase has been mapped for the water cooled module in all different operating conditions, i.e. with different values of solar irradiance and coolant flow rate, as in Fig. 11. At any given value of coolant flow rate, an advantage in terms of electrical efficiency of the water cooled panel over the non-water cooled one is always detected even if with slight differences across the full range of solar irradiance values: it reaches its peak at low values of solar irradiance, especially at the end of STP, whereas its minimum occurs with higher solar irradiance values, typically at noon. For any given solar irradiance, the bigger the coolant flow rate is, the bigger is PVT advantage over PV, although the marginal gain keeps getting smaller at each flow rate increment. It can be concluded that the PV panel greatly benefits of the cooling at any level of solar irradiance. When the goal is to improve both amount and quality (in terms of water temperature) of the thermal power recoverable from the module, Fig. 7 gives an useful finding: thermal losses account for a big share of the energy compart (i.e. about 85%) and reducing it would allow a

Fig. 10. PVT temperature shift at different water flow rates.

a) 0.5l/min

b) 1.0 l/min

c) 1.5l/min

d) 2.0 l/min 53

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

only depend on the water mass flow rate considered. The Tukey’s post hoc test results are in Table 10. A 0.875 HSD value applies to the case at hand and it appears clear that the PV module cooling affects the electric performance in every case considered. The difference between means approaches the critical value when a 1.0 L/ min and a 1.5 L/min configurations are compared, whereas the difference is lower than the critical value moving from a 1.5 L/min to a 2.0 L/min set-up, i.e. the difference on the electric efficiency in the two cases are negligible and the cooling benefit is smaller than in other cases.

experimental data plotted in Fig. 12. Based on the experimental data and the regression model considered, factors accounting for PVT electric performance variability with both environmental temperature and wind speed are −0.7%/1 °C and +2.4%/1 m/s, respectively. Trendlines equations are:

ηel [%] = −0.16·Tair [°C] + 22.51

(17)

ηel [%] = 0.38·ws [m/s] + 18.92

(18)

The electric performance decreases with temperature, as expected, whereas it increases with wind speed, thanks to the increased effectiveness of convection, that keeps the PV module at lower operating temperatures. A high variability in values clusters can be appreciated from Fig. 12: a 20.4% average electric performance that matches an 284.7 K outdoor air temperature, with a 2.2% and 2.9% upper and lower scatter, respectively. Same indicators for a 288.7 K temperature rank at 19.7%, 0.5% and 0.2% (see Fig. 12a). Same sensitivity analysis is performed, having the wind speed as parameter. A 2.5 m/s wind speed exhibits a 18.8% average electric performance, with a 1.4% and 3.9% upper and lower scatter, respectively. As the wind speed drifts toward 5.5 m/s, a symmetrical 0.2% scatter is detected, on a 20.9% average efficiency (see Fig. 12b). As a matter of fact, the lower ambient temperature and wind speed are the higher the scatter on the electric performance is: this evidence, combined with the slope of trend-lines on electrical efficiency, shown in Fig. 12, suggests that the best operating condition in terms of balance between the requirements of results repeatability and electric performances optimization, is achievable with the mid-range values of ambient temperature and wind speed, i.e. close to 288.7 K and 3.5 m/s respectively. Trends shown in Fig. 12 take into account the PV module dynamic thermal behavior, i.e. the storage term Cmod·Mmod ·ΔT /Δt which, as already shown in Fig. 7, proved to play a significant role in real word operating conditions. The experimental data collected by the Authors during the test campaign conducted in November 2015 also led to the following correlation between the PV electric performance and the solar cells operating temperature:

ηel [%] = −0.114·Tcell [°C] + 17.408

4. Conclusions A mathematical model has been developed to describe the unsteady thermal balance of a PV module in real operating conditions and its validation performed by direct comparison of simulated values with the measured ones for all main physical quantities involved in the heat exchange process:

(19)

a) Sensitivity to outdoor air temperature

Ozgoren et al. [30] provide instead the following correlation:

ηel [%] = −0.263·Tcell [°C] + 22.004

(20)

As a matter of fact, for the system at hand, experimental data correlation show a lower sensitivity of electric efficiency to cells operating temperature with respect to similar studies: a –0.85%/1 °C temperature factor comes out, i.e. 60% lower than −2.05%/1 °C [35]. The final datum provided by the Authors is aligned to the most recent literature in the topic: in [46] the temperature factor for a m-Si PVT system (sprinkling-based cooling) ranges between −0.8%/1 °C and −1.3%/ 1 °C. Similar studies deal with c-Si PVT modules, featuring different cooling techniques and a safe estimation of the temperature factor ranges between −0.35%/1 °C and −0.8%/1 °C [34,47–50], with −0.45%/1 °C [51] and −0.5%/1 °C [52] the most recurring values. Lower values are proved to characterize a-Si PVT: −0.2%/1 °C and −0.4%/1 °C are the maximum and minimum values as provided in [49,53–55]. The difference comes from the different electric efficiency at design point for the PV modules investigated and to the difference in ambient conditions under which the experimental campaigns were performed. An effective way to investigate the dependence between the PVT electric performances and the conditions under which the cooling is performed, namely the water mass flow rate, is the statistical analysis of data from experimental campaign, as in [46]. Table 9 summarizes results from the analysis of variance (ANOVA) for the system at hand and suggests that the findings obtained with each cooling configuration

b) Sensitivity to wind speed Fig. 12. PV electric efficiency dependence on ambient conditions.

54

Applied Energy 205 (2017) 44–56

D. Vittorini et al.

convection to the surroundings). The Authors estimate that the adoption of a 10 mm thick glass cover along with a frame to reduce wind circulation over exposed surfaces leads to:

Table 9 One way ANOVA on electric efficiency in five different PVT configurations. Denomination

N

Mean

Std. dev.

Std. error

No cooling Cooling – 0.5 L/min Cooling – 1.0 L/min Cooling – 1.5 L/min Cooling – 2.0 L/min

239 239

15.1 17.2

2.814 2.262

0.015 0.012

239

18.7

2.378

0.014

239

19.6

2.443

0.013

239

20.0

2.432

0.012

Sum of squares

Deg. of freedom

Mean square

F-test value

Significance

3827.77

4

956.942

155.883

0.05

7305.25 11133.02

1190 1194

6.139

Between groups Within groups Total

● a 100–500 W thermal recovery on a daily basis; ● a 15–30 K increase in water temperature, i.e. a higher quality for the recoverable heat. The benefits achieved with PV module refrigeration are partially counterbalanced by the power absorption of the circulation pump, which increases with the flow rate and ultimately, once a target temperature is fixed for solar cells, with the level of solar radiation hitting the module. Such an electric power could be significantly reduced by employing a phase-change fluid as coolant, due to a higher heat removal rate: ● by switching to R236fa, the pump power absorption drowns to a 40% the value it has with water; ● pump power absorption further reduces to a 25% when R245fa is circulated.

Table 10 Post-hoc Tukey’s test.

The viability of organic fluids as alternative to water in module cooling provides a further boost to the system development and opens the way to system configurations with an increased heat recovery capability. As previously stated, heat availability for further uses and the need to have a negligible power absorption by the pump along with the system automation (i.e. the adoption of automated process control devices) are the keys to system development and scalability to residential sector and building applications. As for the percent increase of electric efficiency when switching from the plain PV module to the PVT system, the Authors here obtained values in the 13.9–32.4% range dependently on the coolant flowrate, whereas Ozgoren et al. report a significantly lower 18.3%. It is worth noticing that Ozgoren et al. consider only two values, 0.03 kg/s and 0.077 kg/s, whereas the Authors here investigate more scenarios in terms of coolant flowrates and fluid type.

Cooling

No cooling 0.5 L/min 1.0 L/min 1.5 L/min

0.5 L/min

1.0 L/min

1.5 L/min

2.0 L/min

2.1 – – –

3.6 1.5 – –

4.5 2.4 0.9 –

4.9 2.8 1.3 0.4

Each cooling set-up (i.e. each L/min configuration) has its mean (third column in Table 9). Values in this table are the differences (in absolute value) among the mean for the L/min on the column and the L/min on the row. These values are the test values for the post-hoc Tukey’s test. In bold, those witnessing that the difference among the rowconfiguration and the column-configuration is not negligible.

● the PV temperature is assumed spatially uniform and the relative error on it never exceeds 10% (underestimation during RTP, overestimation during STP); ● the error on simulated PV temperature gets lower as the solar irradiance approaches the maximum efficiency region (i.e. 800 W/m2); ● the PV module electric performance decreases with outdoor air temperature and increases with wind speed, as these two parameters directly affect the convective heat exchange mechanism, and ultimately the steady-state solar cells operating temperature.

Acknowledgements Prof. Domenica Paoletti and her research group are acknowledged for the use of the infrared camera and the digital image processing. References [1] http://www.altenergy.org/renewables/solar-energy-efficiency-milestone.html. [2] Wolf M. Performance analysis of combined heating and photovoltaic power systems for residences. Energy Convers 1976;16:79–90. [3] Florschuetz LW. Extension of the Hottel-Whiller model to the analysis of combined photovoltaic/thermal flat plate collectors. Sol Energy 1979;22:361–6. [4] Kern JEC, Russell MC. Combined photovoltaic and thermal hybrid collector systems. In: Proceedings of the 13th IEEE photovoltaic specialists. Washington (DC, USA); 1978. p. 1153–7. [5] Garg HP, Agarwal RK. Some aspects of a PV/T collector/forced circulation flat plate solar water heater with solar cells. Energy Convers Manage 1995;36(2):87–99. [6] Tripanagnostopoulos Y, et al. Energy, cost and LCA results of PV and hybrid PV/T solar systems. Prog Photovolt: Res Appl 2005;13:235–50. [7] Tripanagnostopoulos Y, et al. Hybrid photovoltaic thermal. Sol Energy 2002;72(3):217–34. [8] Tripanagnostopoulos Y. Aspects and improvements of hybrid photovoltaic thermal solar energy systems. Sol Energy 2007;81(9):1117–31. [9] Tonui JK, Tripanagnostopoulos Y. Air cooled PV/T solar collectors with low cost performance improvements. Sol Energy 2007;81:498–511. [10] Raman V, Tiwari GN. A comparison study of energy and exergy performance of a hybrid photovoltaic double-pass and single-pass air collector. Int J Energy Res 2009;33:605–17. [11] Chow TT. Performance analysis of photovoltaic-thermal collector by explicit dynamic model. Sol Energy 2003;75(2):143–52. [12] Chow TT. Hybrid photovoltaic and thermal solar collector designed for natural circulation of water. Appl Energy 2006;83(3):199–210. [13] Chow TT. Effect of flow channel dimensions on the performance of a box-frame photovoltaic/thermal collector. J Power Energy Proc IMechE Part A 2006:681–8. [14] Chow TT, et al. A comparative study of PV glazing performance in warm climate.

The 1.67 m2 surface PV module features a cooling system, to allow at one time, the enhancement of conversion efficiency of solar cells and the recovery of low-grade heat. As a guideline, when evaluating the opportunity to integrate existing systems with cooling devices (especially with bigger plant scales and higher power levels), it should be observed that: ● heat removal from the module pushes the electrical efficiency up to 20% when a 2.0 L/min water flow rate applies, i.e. a 33% relative gain with respect to the non-water cooled PV module; ● for the system at hand, a trade-off can be found on the coolant flowrate, as the efficiency gain becomes negligible above the 2.0 L/ min threshold: indeed, switching from a no-cooling condition to 0.5 L/min circulation results in a 2% electric efficiency net increase, whereas the same flowrate increment, from 1.5 to 2.0 L/min, results in a 0.4% net increase only. The system viability for thermal uses has to face limitations associated with the low thermal content and the negligible water temperature increase, which calls for the adoption of dedicated module configurations (e.g. front glass cover to limit re-irradiation and/or 55

Applied Energy 205 (2017) 44–56

D. Vittorini et al. Indoor Built Environ 2009;18(1):32–40. [15] Chow TT, et al. Energy and exergy analysis of photovoltaic-thermal collector with and without glass cover. Appl Energy 2009;86(3):310–6. [16] Chow TT. A sensitivity study of a hybrid photovoltaic/thermal water-heating system with natural circulation. Appl Energy 2007;84(2):222–37. [17] de Kezier C, et al. Performance analysis and yield assessment of several uncovered photovoltaic-thermal collectors: results of field measurements and system simulations. In: 32nd European photovoltaic solar energy conference and exhibition; 2016, doi:10.4229/EUPVSEC20162016-6AV.4.16. [18] Michael JJ, et al. Fabrication experimental study and testing of a novel photovoltaic module for photovoltaic thermal applications. Renew Energy 2016;90:95–104. http://dx.doi.org/10.1016/j.renene.2015.12.064. [19] Kalogirou SA. Use of TRNSYS for modeling and simulation of a hybrid PV-thermal solar system for cyprus. Renew Energy 2001;23:602–12. [20] Kalogirou SA, Tripanagnostopoulos Y. Hybrid PV/T solar systems for domestic hot water and electricity production. Energy Convers Manage 2006;47(18–19):3368–82. [21] Kalogirou SA, Tripanagnostopoulos Y. Industrial application of PV/T solar energy systems. Appl Therm Eng 2007;27:1259–70. [22] Krauter S, Ochs F. Integrated solar home system. Renew Energy 2004;29:153–64. [23] Guarracino I, et al. Dynamic coupled thermal and electrical modelling of sheet and tube hybrid photovoltaic/thermal (PVT) collectors. Appl Therm Eng 2016;101:778–95. http://dx.doi.org/10.1016/j.applthermaleng.2016.02.056. [24] Al-Showany Ehsan Fadhil Abbas. The impact of the environmental condition on the performance of the photovoltaic cell. Am J Energy Eng. 2016;4(1):1–7. http://dx. doi.org/10.11648/j.ajee.20160401.11. [25] Tiwari A, Sodha MS. Performance evaluation of solar PV/T system: an experimental validation. Solar Energy 2006;80:751–9. [26] Tiwari A, Sodha MS. Performance evaluation of a solar PV/T system: a parametric study. Renew Energy 2006;31:2460–74. [27] Ji J, Lu JP, Chow TT, He W, Pei G. A sensitivity study of a hybrid photovoltaic/ thermal water-heating system with natural circulation. Appl Energy 2007;84:222–37. [28] Chow TT, He W, Ji J. An experimental study of facade-integrated photovoltaic/ water-heating system. Appl Therm Eng 2007;27:37–45. [29] Ben cheikh, Touafek, Kerrour, Haloui, Khelifa. Model validation of an empirical photovoltaic thermal (PV/T) collector. Energy Proc 2015;74:0. http://dx.doi.org/ 10.1016/j.egypro.2015.07.749. [30] Tse KK, et al. Performance evaluation and economic analysis of a full scale waterbased photovoltaic/thermal (PV/T) system in an office building. Energy Build 2016;122:42–52. http://dx.doi.org/10.1016/j.enbuild.2016.04.014. [31] Good C. Environmental impact assessment of hybrid photovoltaic-thermal (PV/T) systems – a review. Renew Sustain Energy Rev 2016;55:234–9. http://dx.doi.org/ 10.1016/j.rser.2015.10.156. [32] Ravindra NM, Srivastava VK. Temperature dependence of the maximum theoretical efficiency in solar cells. Solar Cells 1979/80;1:107–9. [33] Dubey Swapnil, Sarvaiya Jatin Narotam, Seshadri Bharath. Temperature dependent photovoltaic (PV) efficiency and its effect on PV production in the world - PV Asia Pacific conference 2012. Energy Proc 2013;33:311–21. [34] Skoplaki E, Palyvos JA. On the temperature dependence of photovoltaic module electrical performance: a review of efficiency/power correlations. Solar Energy 2009;83:614–24. http://dx.doi.org/10.1051/epjconf/20134501106. [35] Ozgoren M, Aksoy MH, Bakir C, Dogan S. Experimental performance investigation of photovoltaic/thermal (PV-T) system. EPJ Web Conf 2013;45. [36] Rajput P, Sastry OS, Tiwari GN. Effect of irradiance temperature exposure and an

[37] [38] [39] [40]

[41]

[42]

[43]

[44] [45] [46]

[47]

[48] [49] [50]

[51]

[52]

[53] [54]

[55]

56

arrhenius approach to estimating weathering acceleration factor of glass, EVA and Tedlar in a composite climate of India. Solar Energy 2017;144:267–77. http://dx. doi.org/10.1016/j.solener.2017.01.027. Ceylan I, Ergun A. Thermodynamic analysis of a new design of temperature controlled parabolic trough collector. Energy Convers Manage 2013;74:505–10. Ceylan I, Gürel AE, Demircan H, Aksu B. Cooling of a photovoltaic module with temperature controlled solar collector. Energy Build 2014;72:96–101. Zsiboràcs H, et al. Technical-economic study of cooled crystalline solar modules. Solar Energy 2016;140:227–35. http://dx.doi.org/10.1016/j.solener.2016.11.009. Rajput P, Tiwari GN, Sastry OS, Bora B, Sharma V. Degradation of mono-crystalline photovoltaic modules after 22 years of outdoor exposure in the composite climate of India. Solar Energy 2016;135:786–95. http://dx.doi.org/10.1016/j.solener. 2016.06.047. Rajput P, Tiwari GN, Sastry OS. Thermal modelling and experimental validation of hot spot in crystalline silicon photovoltaic modules for real outdoor condition. Solar Energy 2016;139:569–80. http://dx.doi.org/10.1016/j.solener.2016.10.016. Blessy J, et al. Investigations on serpentine tube type solar photovoltaic/themal collector with different heat transfer fluids: experiment and numerical analysis. Solar Energy 2016;140:12–20. http://dx.doi.org/10.1016/j.solener.2016.10.045. Antonelli M, Baccioli A, Francesconi M, Desideri U, Martorano L. Electric production of a small size concentrated solar power plant with compound parabolic collectors. Renew Energy 2015;83:1110–8. Ricci M, Bocci E, Michelangeli E, Micangeli A, Villarini M, Naso M. Experimental tests of solar collectors prototypes systems. Energy Proc 2015;82:744–51. Colangelo G, Romano D, Tina GM. Performance evaluation of a new type of combined photovoltaic-thermal solar collector. J Solar Energy Eng 2015;137. Bai A, et al. Technical and economic effects of cooling of monocrystalline photovoltaic modules under Hungarian conditions. Renew Sustain Energy Rev 2016;60:1086–99. http://dx.doi.org/10.1016/j.rser.2016.02.003. Krauter S, Araujo RG, Schroer S, Hanitsch R, Salhi MJ, Triebel C, et al. Combined photovoltaic and solar thermal systems for facade integration and building insulation. Sol Energy 1999;67:239–48. Moshfegh B, Sandberg M. Flow and heat transfer in the air gap behind photovoltaic panels. Renew Sustain Energy Rev 1998;2:287–301. Radziemska E, Klugmann E. Thermally affected parameters of the current–voltage characteristics of silicon photocell. Energy Convers Manage 2002;43:1889–900. Mattei M, Notton G, Cristofari C, Muselli M, Poggi P. Calculation of the polycrystalline PV module temperature using a simple method of energy balance. Renew Energy 2006;31:553–67. Zaoui F, Titaouine A, Becherif M, Emziane M, Aboubou A. A combined experimental and simulation study on the effects of irradiance and temperature on photovoltaic modules clean, efficient and affordable energy for a sustainable future. In: The 7th International conference on applied energy (ICAE2015), EnergyProcedia75; 2015. p 373–80. Chandrasekar M, Rajkumar S, Valavan D. A review on the thermal regulation techniques for non integrated flat PV modules mounted on building top. Energy Build 2015;86:692–7. Chow TT. A review on photovoltaic/thermal hybrid solar technology. Appl Energy 2010;87:365–79. Han XY, Wang YP, Zhu L. The performance and long-term stability of silicon concentrator solar cells immersed in dielectric liquids. Energy Convers Manage 2013;66:189–98. Boer KW. Cadmium sulfide enhances solar cell efficiency. Energy Convers Manage 2011;52:426–30.