Investigation of Solar Photovoltaic-Thermal (PVT) and Solar Photovoltaic (PV) Performance: A Case Study in Ghana

The main objective of this paper is to experimentally assess the real-life outdoor performance of a photovoltaic-thermal (PVT) module against a conventional photovoltaic (PV) system in a hot humid tropical climate in Ghana. An experimental setup comprising a water-based mono-crystalline silicon PVT and an ordinary mono-crystalline silicon PV was installed on a rooftop at the Kwame Nkrumah University of Science and Technology in Kumasi and results evaluated for the entire year of 2019. It was observed that the annual total output energy of PV module was 194.79 kWh/m2 whereas that of the PVT for electrical and thermal outputs were 149.92 kWh/m2 and 1087.79 kWh/m2, respectively. The yearly average daily electrical energy yield for the PV and PVT were 3.21 kWh/kWp/day and 2.72 kWh/kWp/day, respectively. The annual performance ratios for the PV and PVT (based on electrical energy output only) were 79.2% and 51.6%, respectively, whilst their capacity factors were, respectively, 13.4% and 11.3%. Whereas the highest monthly mean efficiency recorded for the PV was 12.7%, the highest combined measured monthly mean electrical/thermal efficiency of the PVT was 56.1%. It is also concluded that the PVT is a worthy prospective alternative energy source in off-grid situations.


Introduction
Solar energy is commonly collected as heat and electricity through thermal and photovoltaic (PV) technologies, respectively. A hybrid photovoltaic-thermal (PVT) integrates a solar thermal absorber and a PV into one unit. Whereas the PV cells generate electricity, the integrated thermal system absorbs residual heat energy from the cells and thus reduces their temperature in the process and also enhances their performance [1][2][3]. The two most cost-effective working fluids are water and air, with water type found to be more efficient [4]. Hybrid PVT collectors can reach net (electrical plus thermal) efficiencies of 70% or higher, with electrical efficiencies up to 15-20% and thermal efficiencies exceeding 50%, depending on the conditions [5]. The PVT technologies have the potential to reduce the use of materials, installation time, and the required space [6]. The advantage of PVTs in generating both electricity and thermal energy simultaneously makes them handy for domestic applications. However, despite  Copper -Special care was taken to ensure the PV and PVT modules were sourced from the same manufacturer for similarities in peripheral material composition and assembling techniques. The 200 W rated commercial PVT module had a layer of 72 mc-Si PV cells with a flat copper plate thermal system securely attached on its back. A thin adhesive layer, made up of ethylene-vinyl acetate (EVA) layer and the tedlar layer, was used to fix the PV module on to the thermal absorber plate. This compound adhesive layer also acted as a shock absorber to further strengthen the PV module. The heat conducted by the copper thermal plate from the PV cells was transferred by fluid (water in this case) flowing through 14 evenly distributed parallel copper pipes attached to the absorber plate and running from inlet to outlet manifolds. The thermal absorber was then covered with an insulator material and then finally with an aluminum foil to complete the thermal insulation on its back.
Both unshaded PVT and PV modules were oriented towards the south and inclined at a fixed angle of 8 o . Tilt angle allowed natural cleaning of the modules during rainfalls, which reduced soiling by dust settlement on the installations, and ensured optimum capture of the solar irradiation for the  Special care was taken to ensure the PV and PVT modules were sourced from the same manufacturer for similarities in peripheral material composition and assembling techniques. The 200 W rated commercial PVT module had a layer of 72 mc-Si PV cells with a flat copper plate thermal system securely attached on its back. A thin adhesive layer, made up of ethylene-vinyl acetate (EVA) layer and the tedlar layer, was used to fix the PV module on to the thermal absorber plate. This compound adhesive layer also acted as a shock absorber to further strengthen the PV module. The heat conducted by the copper thermal plate from the PV cells was transferred by fluid (water in this case) flowing through 14 evenly distributed parallel copper pipes attached to the absorber plate and running from inlet to outlet manifolds. The thermal absorber was then covered with an insulator material and then finally with an aluminum foil to complete the thermal insulation on its back.
Both unshaded PVT and PV modules were oriented towards the south and inclined at a fixed angle of 8 • . Tilt angle allowed natural cleaning of the modules during rainfalls, which reduced soiling by dust settlement on the installations, and ensured optimum capture of the solar irradiation for the location. In addition, the modules were manually cleaned on regular basis to reduce the effects Energies 2020, 13, 2701 4 of 17 of soiling. As shown in Figure 2, beneath the mounting frame for the PVT installations were two separated compartments for housing the main electrical/logging circuitry and water circulation system. location. In addition, the modules were manually cleaned on regular basis to reduce the effects of soiling. As shown in Figure 2, beneath the mounting frame for the PVT installations were two separated compartments for housing the main electrical/logging circuitry and water circulation system.  Figure 3 shows a schematic diagram of the experimental setup. The main heat transfer medium for the PVT was by force-circulated water in a closed-loop system. A flow jet direct current (DC) pump, with regulated input power, circulated the water at solar irradiance above 150 W/m 2 when the thermal absorber temperature was greater than its inlet water temperature. The solar irradiance was measured with a pyranometer mounted in-plane on the PVT module. Manual valves V1 and V2 were used to regulate the flow rate and monitored with a mechanical spring flow meter F (see Figure 3). The relatively warm water from the PVT went through a copper-coiled heat exchanger submerged in 70 L of water contained in an insulated tank. For the purpose of the experiment, the circulating water returning into the PVT inlet was further cooled with a water-to-air heat exchanger. This also reduced the problem of potential heat recirculation from the water tank back to the PVT, making it a better stand-alone system.

Instrumentations, System Control and Data Collection
A programmable data logger (CR300, Campbell Scientific, Logan, UT, USA) was used to record both meteorological and the modules' performance data. The measurements were sampled every 10s and then averaged over 15-min periods, from which hourly, daily and monthly data were determined. The meteorological measurements were the global solar irradiance in plane-of-array  Figure 3 shows a schematic diagram of the experimental setup. The main heat transfer medium for the PVT was by force-circulated water in a closed-loop system. A flow jet direct current (DC) pump, with regulated input power, circulated the water at solar irradiance above 150 W/m 2 when the thermal absorber temperature was greater than its inlet water temperature. The solar irradiance was measured with a pyranometer mounted in-plane on the PVT module. Manual valves V1 and V2 were used to regulate the flow rate and monitored with a mechanical spring flow meter F (see Figure 3). The relatively warm water from the PVT went through a copper-coiled heat exchanger submerged in 70 L of water contained in an insulated tank. For the purpose of the experiment, the circulating water returning into the PVT inlet was further cooled with a water-to-air heat exchanger. This also reduced the problem of potential heat recirculation from the water tank back to the PVT, making it a better stand-alone system. location. In addition, the modules were manually cleaned on regular basis to reduce the effects of soiling. As shown in Figure 2, beneath the mounting frame for the PVT installations were two separated compartments for housing the main electrical/logging circuitry and water circulation system.  Figure 3 shows a schematic diagram of the experimental setup. The main heat transfer medium for the PVT was by force-circulated water in a closed-loop system. A flow jet direct current (DC) pump, with regulated input power, circulated the water at solar irradiance above 150 W/m 2 when the thermal absorber temperature was greater than its inlet water temperature. The solar irradiance was measured with a pyranometer mounted in-plane on the PVT module. Manual valves V1 and V2 were used to regulate the flow rate and monitored with a mechanical spring flow meter F (see Figure 3). The relatively warm water from the PVT went through a copper-coiled heat exchanger submerged in 70 L of water contained in an insulated tank. For the purpose of the experiment, the circulating water returning into the PVT inlet was further cooled with a water-to-air heat exchanger. This also reduced the problem of potential heat recirculation from the water tank back to the PVT, making it a better stand-alone system.

Instrumentations, System Control and Data Collection
A programmable data logger (CR300, Campbell Scientific, Logan, UT, USA) was used to record both meteorological and the modules' performance data. The measurements were sampled every 10s and then averaged over 15-min periods, from which hourly, daily and monthly data were determined. The meteorological measurements were the global solar irradiance in plane-of-array

Instrumentations, System Control and Data Collection
A programmable data logger (CR300, Campbell Scientific, Logan, UT, USA) was used to record both meteorological and the modules' performance data. The measurements were sampled every 10s and then averaged over 15-min periods, from which hourly, daily and monthly data were determined. The meteorological measurements were the global solar irradiance in plane-of-array (W/m 2 ), ambient wind velocity (m/s), ambient relative humidity (%) and ambient air temperature ( • C). The electrical Batteries, radiator fans, data logger and circulation pump were powered from the battery charger, serving as external loads to the modules. Logged data from the battery chargers included output currents (A) and voltages (V) from modules.
Apart from ambient temperature, all other temperatures were separately measured with calibrated temperature sensors (PT100, Campbell Scientific, Logan, Utah, USA). As shown in Figure 3, measured temperatures included the PVT inlet water temperature (T 1 ), PVT module back temperature (T 2 ), PVT outlet water temperature (T 3 ), PV module back temperature (T 4 ) and water storage tank temperature (T 5 ). In addition to recording data, the logger was programed to control the functionalities of the mechanical components in the PVT setup based on real-time in-plane global irradiation. Active water circulation through the PVT was kept at a constant flowrate per desired set value. Table 2 shows the list and basic characteristics of the instrumentation used in this study.

Variable Instrumentation Measurement Accuracy Resolution
Flow Gauge (L/min) Mechanical spring flow meter not available Bim205 smart charger w/MPPT Not specified 0.1

Module Temperature
The peak or rated power of a PV module is determined under standard test conditions (STC), which are solar irradiance of 1 kW/m 2 , module temperature of 25 • C and air-mass ratio (AM) of 1.5 (AM = 1.5). However, in real life outdoor situations, the ambient conditions are different from these STC, and hence, the PV module power output will differ from the rated power. The cell temperature T cell ( • C) at any ambient temperature T amb ( • C) is given as: where NOCT is the nominal operating cell temperature ( • C), G P is the in-plane global irradiance (W/m 2 ), η cell is cell efficiency (%), (τα) is the effective transmittance-absorptance, U L is the loss coefficient (W/m 2 • C) and Z NOCT means parameter Z at NOCT. The loss coefficient can further be expressed as: where ω is wind speed (m/s). Additionally, at NOCT, T amb.NOCT is 20 • C, G P.NOCT is 800 W/m 2 and ω is 1 m/s, at no load operation (i.e., η cell = 0). Equation (1) can therefore be simplified as: Thus the effect of solar irradiance, ambient temperature and wind speed on solar PV module can be quantified by their impact on the module temperature as given in Equation (3).

PV Performance Indices-Energy Yield, Performance Ratio and Efficiency
The performance of a PV system is usually examined using a number of selected performance indices, including energy yield, performance ratio and efficiency. The energy yield is defined as output normalized by the PV system's rated capacity. It specifies how many hours in a day the PV system must operate at its rated power in order to produce the same amount of energy as was measured [40,41]. It is given as: where Y A is the array yield in kWh/kW p /day, E av.d is the average daily module DC energy output (kWh/day) and P rated is the rated kilowatt peak electrical power (kW p ) of the PV module at STC. The performance ratio (PR) measures the overall effect of losses on the rated output of the system and indicates how close its performance is to the ideal performance during real life operation. The PR is useful for the comparison of modules that receive different amounts of irradiation, especially due to geographical location and or PV inclination. It is given as [40,41]: where S h (h/day) is the plane-of-array average daily peak sun-hours, which is the same as the reference yield, Y R . The reference yield is the ratio of the total in-plane solar radiation to the array reference irradiance, H R (usually taken as 1 kW/m 2 ). It is a measure of the theoretical energy available at a specific location over a specified time period [41] given as: The PV module efficiency is given as: where A m is the module total surface area (m 2 ) and G p is the in-plane solar irradiance (kW/m 2 ). E dc is the DC power from the module in kW. Depending on the available data and desire level of resolution, the efficiencies can be determined on instantaneous, hourly, daily, monthly and annual bases [41].

PVT Performance Indices-Heat Gained, Thermal Energy Yield and Efficiency
The overall performance of a PVT system is a combination of both PV (electricity) and its thermal (heat energy) components. The thermal gain of the system is given as: where M w is the water mass flow rate (kg/s), c p is the specific heat of water (kJ/kg • C) and ∆T ( • C) is the temperature difference, expressed as: where T 1 and T 3 are the inlet and outlet water temperature, respectively (see Figure 3). The mass flow rate M w can also be expressed in volumetric terms as: where V w is the volumetric flow rate in m 3 /s and ρ(T) is the density of water (kg/m 3 ) at temperature T. Both c p and ρ(T) were assumed to be constant (c p = 4.18 J/kg • C, ρ(T) = 1000 kg/m 3 ) throughout the analysis presented in the study. The thermal efficiency of the PVT is given as: Combining Equations (7) and (11), the overall efficiency of a PVT system is given as:

Clearness Index
The clearness index is the fraction of the solar radiation reaching the top of the atmosphere that makes it through the atmosphere to reach the Earth's surface. It is normally calculated as a ratio of the monthly averaged daily global solar radiation on horizontal surface (H av ) to the monthly averaged daily extraterrestrial solar radiation (H o.av ) at a given site.
The monthly daily average clearness index (K T ): The extraterrestrial irradiance H o can be determined using the mathematical expression: where n is the day of the year, L is the latitude of the site, δ is the declination angle of the sun, G sc is the extraterrestrial solar constant 1.37 kW/m 2 and H SR is the sunrise hour given as: H o.av can then be calculated as: where R is the number of days in the month. For this study, monthly daily average clearness indexes were generated for the site using HOMER Pro energy simulation software (Version 3.13.6, Homer Energy, Boulder, CO, USA) [42] which employs existing global data sources and libraries in its predictions.

Ambient Conditions
The performance of a solar collector is influenced by a number of interactive factors including, weather conditions. There is however, a very high level of variability and uncertainty in predicting meteorological variables. Nonetheless, PV cell temperature, for instance, is a function of ambient temperature, wind speed and global irradiance (Equation (3)). respectively. The average ambient temperature is dependent on the time or season of the year. Ghana has two main climatic seasons namely, the wet and dry seasons. Typically, the wet season in Kumasi starts from April to October. It is normally characterized by cloudy weather conditions, relatively higher mean monthly precipitation and relatively lower mean monthly ambient temperatures. The dry season (which predominantly spans November to March) records low-to-no precipitation and higher ambient temperatures.
Energies 2020, 13, x FOR PEER REVIEW 8 of 17 higher mean monthly precipitation and relatively lower mean monthly ambient temperatures. The dry season (which predominantly spans November to March) records low-to-no precipitation and higher ambient temperatures. The lowest monthly average daily irradiations were in July and August (with clearness index values of 0.42 and 0.43, respectively) due to dense cloudy sky conditions. Although the general sky conditions for January and December were not cloudy, they signified the harmattan season when lots of dust and smoke prevailed in the atmosphere and led to high levels of solar attenuation. Annual lowest and highest daily insolation of 0.91 kWh/m 2 /day and 6.10 kWh/m 2 /day were observed in the months of June and March, respectively. Furthermore, the annual average daily solar energy for the site was 4.05 kWh/m 2 /day.  The lowest monthly average daily irradiations were in July and August (with clearness index values of 0.42 and 0.43, respectively) due to dense cloudy sky conditions. Although the general sky conditions for January and December were not cloudy, they signified the harmattan season when lots of dust and smoke prevailed in the atmosphere and led to high levels of solar attenuation. Annual lowest and highest daily insolation of 0.91 kWh/m 2 /day and 6.10 kWh/m 2 /day were observed in the months of June and March, respectively. Furthermore, the annual average daily solar energy for the site was 4.05 kWh/m 2 /day. Figure 6 shows the monthly average and maximum wind speed within the vicinity of the installation. For the period of the study, the monthly average wind speed was less than 1.00 m/s, which prospectively indicated that wind speed effects on solar photovoltaic performance could be negligible. However, observed maximum monthly wind speeds ranged from 3 m/s in February to 11 m/s in October.

Wind Speed (ω)
index values of 0.42 and 0.43, respectively) due to dense cloudy sky conditions. Although the general sky conditions for January and December were not cloudy, they signified the harmattan season when lots of dust and smoke prevailed in the atmosphere and led to high levels of solar attenuation. Annual lowest and highest daily insolation of 0.91 kWh/m 2 /day and 6.10 kWh/m 2 /day were observed in the months of June and March, respectively. Furthermore, the annual average daily solar energy for the site was 4.05 kWh/m 2 /day.   Figure 6 shows the monthly average and maximum wind speed within the vicinity of the installation. For the period of the study, the monthly average wind speed was less than 1.00 m/s, which prospectively indicated that wind speed effects on solar photovoltaic performance could be negligible. However, observed maximum monthly wind speeds ranged from 3 m/s in February to 11 m/s in October.

Module Temperature ( )
The variations in PV and PVT module temperatures at daytime are shown in Figure 7. The recorded average PV module temperatures at daytime were relatively higher than average PVT module temperatures, ranging from 1.3% higher in the month of August to 6.9% higher in March. This could translate into a relatively higher net energy yield per rated wattage for PVT module, due to the lower cell temperature and higher heat recoveries.

Module Temperature (T cell )
The variations in PV and PVT module temperatures at daytime are shown in Figure 7. The recorded average PV module temperatures at daytime were relatively higher than average PVT module temperatures, ranging from 1.3% higher in the month of August to 6.9% higher in March. This could translate into a relatively higher net energy yield per rated wattage for PVT module, due to the lower cell temperature and higher heat recoveries. Figure 7b,c show the contrasting effects of solar irradiance fluxes on the modules temperatures. Cloudy and unstable weather conditions, smoke and dust in the atmosphere intermittently attenuated incidental solar irradiation resulting in fluctuations and generally lowered ambient temperatures with telling effects on solar module temperature. Nonetheless, the highest recorded operating cell temperatures for the modules for the period under study were 70.6 • C and 60.5 • C for PV module and PVT module, respectively, recorded on the 17 October 2019 (Figure 7c). The high module temperatures were consistent with an earlier study in the same environment [30] where similar  Table 3 presents the summary installations' site ambient conditions and PV and PVT module temperatures.

Module Temperature ( )
The variations in PV and PVT module temperatures at daytime are shown in Figure 7. The recorded average PV module temperatures at daytime were relatively higher than average PVT module temperatures, ranging from 1.3% higher in the month of August to 6.9% higher in March. This could translate into a relatively higher net energy yield per rated wattage for PVT module, due to the lower cell temperature and higher heat recoveries.

Dec.
Feb. Apr. Jun. Aug. Oct. Dec.   (Figure 7c). The high module temperatures were consistent with an earlier study in the same environment [30] where similar observations were made for different PV technologies. Table 3 presents the summary installations' site ambient conditions and PV and PVT module temperatures.

Electrical Energy Outputs for both PV and PVT
The electrical energy outputs from both PVT and PV were measured at maximum power point (MPP). Both the PV and the PVT outputs followed the same trend as in-plane solar radiation (Figures 5  and 8a). As shown in Figure 8a, in comparison with the other months, the DC outputs from both PV and PVT modules were generally low in the months of June to September. For the PV, the monthly average daily electrical energy outputs varied from 0.39 kW/m 2 /day in August to 0.62 kWh/m 2 /day in April and November. In the case of the PVT however, it varied from 0.29 kW/m 2 /day in August to 0.54 kWh/m 2 /day in April. Unlike the PV, the PVT performance dependency factors go beyond environmental factors to include PV parking factor, type and flow rate of the thermal fluid, type and design of thermal absorber [7,10,11] etc. which were not covered in the scope of this work.
Energies 2020, 13, x FOR PEER REVIEW 11 of 17 environmental factors to include PV parking factor, type and flow rate of the thermal fluid, type and design of thermal absorber [7,10,11] etc. which were not covered in the scope of this work. The annual total electrical energy measurements, per unit area, assessed from PV and PVT modules were 194.79 kWh/m 2 and 149.92 kWh/m 2 , respectively (Figure 8b). In addition to the electrical energy, the PVT also provided an added incentive of 601.57 kWh/m 2 of heat energy over the same period.

Efficiency
The efficiency was based on only DC power output, since the installations were not connected to inverters. It was calculated for every 15 min (data resolution) for daily sun-hour periods (with filtered irradiance ≥ 150 W/m 2 ) and then averaged monthly. Nameplate rated efficiencies (DC) for PV and PVT are 16.60% and 15.08%, respectively (see Table 1). From the results, the monthly averaged electrical efficiency for PV varied from 11.6% to 12.7% over the data collection period, indicating a deviation range of 23.1% to 30.1%, relative to rated efficiency. In the case of the PVT, the electrical efficiency varied from 9.9% to 11.5%, representing a deviation of 28% to 38% from STC rating. In addition to that, the thermal efficiency for the PVT had a wider variation of 29.44% to 44.84%. This represented a deviation of 5.6% to 38% from its zero-loss collector efficiency of 47.5% (see Table 1). This could be attributed to the slow thermal response (time constant of up to 8 min [8]) of the PVT's thermal absorber (copper sheet with pipes in this case) to erratic spectral changes, typical of the study environment [30,43]. It can also be deduced from Table 4 that the average thermal efficiency of the commercial PVT was predominantly lower than those reported in most studies in the literature. This could be attributed to many factors, including poor contact between absorber and PV. A similar observation was made about a commercial water PVT by Guarracino et al. [8] in another study. Notwithstanding this observation, the highest recorded monthly mean net efficiency was 56.14%.  The annual total electrical energy measurements, per unit area, assessed from PV and PVT modules were 194.79 kWh/m 2 and 149.92 kWh/m 2 , respectively (Figure 8b). In addition to the electrical energy, the PVT also provided an added incentive of 601.57 kWh/m 2 of heat energy over the same period.

Efficiency
The efficiency was based on only DC power output, since the installations were not connected to inverters. It was calculated for every 15 min (data resolution) for daily sun-hour periods (with filtered irradiance ≥ 150 W/m 2 ) and then averaged monthly. Nameplate rated efficiencies (DC) for PV and PVT are 16.60% and 15.08%, respectively (see Table 1). From the results, the monthly averaged electrical efficiency for PV varied from 11.6% to 12.7% over the data collection period, indicating a deviation range of 23.1% to 30.1%, relative to rated efficiency. In the case of the PVT, the electrical efficiency varied from 9.9% to 11.5%, representing a deviation of 28% to 38% from STC rating. In addition to that, the thermal efficiency for the PVT had a wider variation of 29.44% to 44.84%. This represented a deviation of 5.6% to 38% from its zero-loss collector efficiency of 47.5% (see Table 1). This could be attributed to the slow thermal response (time constant of up to 8 min [8]) of the PVT's thermal absorber (copper sheet with pipes in this case) to erratic spectral changes, typical of the study environment [30,43]. It can also be deduced from Table 4 that the average thermal efficiency of the commercial PVT was predominantly lower than those reported in most studies in the literature. This could be attributed to many factors, including poor contact between absorber and PV. A similar observation was made about a commercial water PVT by Guarracino et al. [8] in another study. Notwithstanding this observation, the highest recorded monthly mean net efficiency was 56.14%.  Figure 9 shows that the highest and lowest monthly average daily reference yields were realized in the months of November (4.90 kWh/kW/day) and August (2.81 kWh/kW/day), respectively. The reference yield represents the number of peak sun hours. It is a function of the location, orientation and inclination of the solar PV array. As a result, the monthly daily average reference yield followed the same trend as the irradiance recorded in-plane as shown in Figure 9. The observed low reference yields in the months of July to September could be attributed to low in-plane solar irradiation.
Energies 2020, 13, x FOR PEER REVIEW 12 of 17 3.3.3. Reference Yield Figure 9 shows that the highest and lowest monthly average daily reference yields were realized in the months of November (4.90 kWh/kW/day) and August (2.81 kWh/kW/day), respectively. The reference yield represents the number of peak sun hours. It is a function of the location, orientation and inclination of the solar PV array. As a result, the monthly daily average reference yield followed the same trend as the irradiance recorded in-plane as shown in Figure 9. The observed low reference yields in the months of July to September could be attributed to low in-plane solar irradiation.

Array or Module Yield
The monthly average daily electrical energy yields for the PV and PVT are presented in Figure  10. The energy yield of the PV varied from a low of 2.5 kWh/kWp/day in the month of August to a high of 3.85 kWh/kWp/day in the months of April and November. In the case of the PVT, its electrical energy yield varied from a low of 1.83 kWh/kWp/day in the same month of August to a high of 3.56 kWh/kWp/day in April. As indicated earlier, the observed lower reference yields in the months of July to September were due to low in-plane solar irradiation. However, during the same period, the PV yields increased compared to the reference yield due to lower ambient temperatures on average translating into lower module temperatures (Figure 7a) and improved performance. The yearly daily average electrical energy yields for the period of the study were observed to be 3.21 kWh/kWp/day for the PV module and 2.72 kWh/kWp/day for the PVT module. The variation in monthly energy yield was similar to a pattern in an earlier study [30] in the same environment on assorted model technologies.

Array or Module Yield
The monthly average daily electrical energy yields for the PV and PVT are presented in Figure 10. The energy yield of the PV varied from a low of 2.5 kWh/kW p /day in the month of August to a high of 3.85 kWh/kW p /day in the months of April and November. In the case of the PVT, its electrical energy yield varied from a low of 1.83 kWh/kW p /day in the same month of August to a high of 3.56 kWh/kW p /day in April. As indicated earlier, the observed lower reference yields in the months of July to September were due to low in-plane solar irradiation. However, during the same period, the PV yields increased compared to the reference yield due to lower ambient temperatures on average translating into lower module temperatures (Figure 7a) and improved performance. The yearly daily 3.3.6. Capacity Factor (CF) The annual capacity factors (CF) of 13.35% and 11.30% were recorded for the PV and PVT, respectively. The CFs show the average of fraction of time in a year when the PV system is available to generate energy at its rated power output. Thus, the PV and PVT systems considered in this study can be said to have generated their rated electrical power only in about 48 days and 41 days, respectively.  Table 6 presents the summary of key electrical parameters for PV and PVT installations. Yearly average daily array yields of 3.21 kWh/kWp/day for the PV module, as against 2.72 kWh/kWp/day for the PVT module, seemed to suggest the PV module outperformed the PVT in electrical energy production. This was consistent with earlier studies [7]. The electrical energy yield from the PV in this study was on average 25.55% higher than what was generated from the PVT. However, with the additional incentive of thermal energy harvest, the PVT provided a better utilization of solar energy resources than the ordinary PV. Due to the continuous extraction of heat from the modules, the PVT

Performance Ratio (PR)
As indicated earlier PR shows the overall effect of losses on PV ratings. For this study, the PRs for the PV and PVT were 79.2% and 51.6% respectively. These values were not very different from what were reported in other studies as shown in Table 5. The value of the PR is an important way of identifying problems emanating from system component failures and a good guide for improving system performance. 3.3.6. Capacity Factor (CF) The annual capacity factors (CF) of 13.35% and 11.30% were recorded for the PV and PVT, respectively. The CFs show the average of fraction of time in a year when the PV system is available to Energies 2020, 13, 2701 14 of 17 generate energy at its rated power output. Thus, the PV and PVT systems considered in this study can be said to have generated their rated electrical power only in about 48 days and 41 days, respectively. Table 6 presents the summary of key electrical parameters for PV and PVT installations. Yearly average daily array yields of 3.21 kWh/kW p /day for the PV module, as against 2.72 kWh/kW p /day for the PVT module, seemed to suggest the PV module outperformed the PVT in electrical energy production. This was consistent with earlier studies [7]. The electrical energy yield from the PV in this study was on average 25.55% higher than what was generated from the PVT. However, with the additional incentive of thermal energy harvest, the PVT provided a better utilization of solar energy resources than the ordinary PV. Due to the continuous extraction of heat from the modules, the PVT module generally operated at lower module temperatures than the PV module. This was however contradicted by Fuentes et al. [7] who attributed lower electrical performance of PVT to higher module temperatures compared to the PV module. Despite this, the thermal gains of the PVT made it better than PV in terms of total energy output per unit area. The monthly mean efficiency variations of the PVT module in this study also showed that, based on both electrical and thermal energy outputs, its overall efficiency could reach 56.34%, which is significantly higher than the maximum monthly mean value of 12.7% for the PV module. This meant that, in terms of physical installation and space, the PVT gave a better energy output per unit area than the PV. Hence, for better space utilization per energy output, the PVT was a better alternative than the PV. For off-grid rural settlements, the PVT could come in handy in the provision of both electricity and heating. Additionally, this could be very useful in clinics, schools, and for camping.

Conclusions
In this paper, a comparative performance valuation was conducted on water-based PVT and PV modules made of mc-Si cell technology in a dynamic environment for 2019.

•
The highest recorded instantaneous module temperatures were 70.6 • C and 60.5 • C for the PV module and PVT module, respectively, recorded in October. On the average, the PV module temperature remained relatively higher than that of PVT by 1.3% to 6.9%.

•
The annual total energy output for the PV module was 194.79 kWh/m 2 while that of the PVT for electrical and thermal outputs was 149.92 kWh/m 2 and 1087.79 kWh/m 2 , respectively.

•
The annual daily mean electrical energy yield for the PV and PVT were 3.21 kWh/kW p /day and 2.72 kWh/kW p /day, respectively.

•
The annual performance ratios based on only electrical energy for the PV and PVT were 79.2% and 51.6%, respectively, whereas their capacity factors were, respectively, 13.35% and 11.3%.

•
The monthly average electrical efficiency values for PV and PVT were 11.6-12.7% and 9.9-11.5% respectively. The thermal efficiency of the PVT had a wider variation from 29.44% to 44.84%. There is however the need to improve the thermal efficiency of commercial PVTs.
This study has shown that the flat plate water PVT application is feasible in environments with similar weather conditions to that of Kumasi. It could also be concluded that, based on the general performance of the two technologies, the PV is a better choice for very large-scale grid-connected systems, where the interest is mostly in electrical energy production. However, for domestic applications and small scale grid systems with provision for thermal energy use, the PVT is a better option. The study could not however cover the exergy analysis, economic evaluations and life cycle assessment of the current PVT/PV setup. These should be carried out so that the actual cost of PVT setup, the net cost of produced energy and their environmental impact could be determined. This information could be useful to stakeholders in Ghana in making informed decisions in energy systems.