T solar air collector for building fenestration applications

Accepted Manuscript Title: Performance and optimization of a BIPV/T solar air collector for building fenestration applications Authors: A. Chialastri, M. Isaacson PII: DOI: Reference:

S0378-7788(16)31443-8 http://dx.doi.org/doi:10.1016/j.enbuild.2017.05.064 ENB 7645

To appear in:

ENB

Received date: Revised date: Accepted date:

9-11-2016 30-3-2017 21-5-2017

Please cite this article as: A.Chialastri, M.Isaacson, Performance and optimization of a BIPV/T solar air collector for building fenestration applications, Energy and Buildingshttp://dx.doi.org/10.1016/j.enbuild.2017.05.064 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Performance and optimization of a BIPV/T solar air collector for building fenestration applications A. Chialastri, M. Isaacson Baskin School of Engineering, Center for Sustainable Energy and Power Systems, Department of Electrical Engineering University of California at Santa Cruz, Santa Cruz, California, 95064, USA

Research Highlights     

A building-integrated hybrid photovoltaic-thermal air collector was tested. A maximum temperature rise of 31C was observed for lower angles during the winter. Thermal and electrical efficiencies of 25-40% and 6-8% respectively were recorded. A 2-D model was built in COMSOL Multiphysics to optimize the glazing system. Simulations showed that a 3-panes glazing could boost the temperature rise by 37%.

Abstract The different elements of the building envelope such as facades, roof and windows play a central role in its thermal behaviour, and new technologies that integrate their architectural functions with energy generation are emerging. A prototype of a buildingintegrated photovoltaic/thermal (BIPV/T) air collector was built, which is intended to perform the functions of thermal and electrical generation, light transmission and shading control. In this work, the prototype was tested under different conditions to investigate its thermal and electrical performances. The results showed a maximum temperature rise (from bottom to top) of 31°C and average thermal and electrical efficiencies of 31% and 7%, respectively. The experimental data were used to build a twodimensional model in COMSOL Multiphysics, in order to assist in the optimization of the various system components for the design of the next prototype. Simulations were performed on the glazing system to optimize the thermal output, through the use of coatings and additional glass panels. Different configurations were analyzed, and it was found that a 3-pane system with low-e coatings applied to the inside surfaces represents the best cost-effective solution, which results in a 64.7°C air temperature output and a 40% increase in temperature rise over the existing prototype.

1. Introduction Buildings constitute a remarkable fraction of the global energy demand, accounting for about 40% of the total consumption [1, 2], a large part of which is being used for space heating and cooling in both the residential and commercial sector. The efforts that nowadays many countries are putting in trying to reduce buildings energy consumptions, which are increasingly promoting initiatives aimed towards the development of net-zero energy buildings, are raising a growing interest in new integrated energy generation technologies such as building-integrated photovoltaic (BIPV) and building-integrated photovoltaic/thermal systems (BIPV/T) [3]. The advantage of integrated systems is that they replace parts of the building envelope, such as roofs and façades, performing the same structural, architectural and aesthetic functions but integrating these with the generation of electricity, heat, or both. BIPV/T systems are a more recent development of the photovoltaic/thermal (PV/T) technology, whose research dates back to the 1970s [4], when combined PV/T collectors started to be studied and tested [5-7]. The idea of solar cogeneration of electricity and heat with a single collector is related to the fact that photovoltaic modules convert only a portion of the incoming radiation into electricity, which is 12-18% for commercial silicon modules currently available, while the remaining 80% or more is reflected back or converted into heat, that is normally lost to the outdoor environment [8]. Since the PV conversion efficiency decreases as the module temperature increases, removing the excess heat by circulating a cooling fluid, such as air or water, behind the panel has a positive effect on the electrical performance and allows the collection of thermal energy that would be otherwise lost. Kumar and Rosen [9] found that a PV/T collector produces more energy per unit area than individual PV and thermal collectors, and Fujisawa and Tani [10] came to a similar conclusion from an exergy evaluation, showing the potential that this technology has when both electricity and heat are needed with a limited installation area, as it is the case for the building sector. The concept of BIPV/T was first introduced in 1996 by Clarke et al. [11], who performed laboratory tests and simulations of a PV ventilated façade with heat recovery, which resulted in a lower operating cell temperature and higher electrical efficiency with an added thermal generation. Since then, the BIPV/T has attracted increasing attention and much research has been done all over the world.

Brinkworth et al. [12] validated a model for naturally ventilated PV façades, and Gaillard et al, [13] carried out an experimental evaluation of a naturally ventilated BIPV façade under real operating conditions, and demonstrated that the system can contribute to meet the building heating and ventilation demand, in addition to providing electrical generation. PV façades were also integrated with more conventional passive solar heating systems, such as Trombe walls [14], consisting of an exterior glazing pane and an interior thermal absorbing wall, with vents at the top and bottom of the latter to allow air circulation between the air cavity and the indoor. A BIPV/T Trombe wall was modeled by Jie et al. [15], finding that a room temperature increase of 12.3 C is possible, compared with a conventional wall, while Sun et al. [16] obtained a higher indoor air temperature for a BIPV/T Trombe wall façade with a window, compared to a same size conventional Trombe wall. Koyunbaba et al. [17] carried out experimental tests and modeling that showed a maximum electrical and thermal efficiency of 4.52% and 27.2%, respectively. Many studies have been made on BIPV/T systems in forced ventilation. Nagano et al. [18] developed vertical exterior wallboards incorporating PV cells, and found that an increase in thermal efficiency from about 22% to 29% could be achieved with the addition of a glass cover in front of the wallboard. Athienitis et al. [19] constructed a prototype of BIPV/T collector integrated with an unglazed transpired collector (UTC), which was further applied to a full-scale office building in Montreal, Canada. Pantic et al. [20] analyzed three different BIPV/T configurations integrated with roof, showing that the addition of a vertical glazed solar air collector in series with the outlet of the roof system provides significant increase in the air temperature output, while adding a glazing cover on top of the PV would cause an increase of thermal generation but also a reduction in electrical performances, due to a lower solar radiation received and higher PV temperatures. Agrawal and Tiwari [8] designed roof-integrated BIPV/T air channels able to be connected in series or in parallel, and developed a one-dimensional transient model to select an appropriate system suitable for cold conditions in India. Aste et al. [21] designed and implemented a BIPV/T system on a tilted façade using semi-transparent PV modules, reporting thermal and electrical efficiency varying from 20 to 40% and from 9 to 10%, respectively. Charron and Athienitis [22] carried out a theoretical study to optimize the performances of a ventilated double façade with integrated photovoltaics and motorized blinds, and found that PV modules placed within the air cavity can improve the thermal efficiency by 25% but at the expense of a 21% reduction in electrical generation. The extensive interest in air-based BIPV/T systems is mainly driven by their flexibility and ease of integration with multiple building elements, due to the availability of air and its lightweight, as well as to the absence of leakage or freezing problems, resulting in low installation and maintenance costs [4]. However, there are some limitations and challenges that researchers and developers are currently facing, which are mainly related to the limited heat transfer coefficient of the air compared to water-based systems. This affects the ability to cool the PV panels, so that there is a conflict between generating thermal energy at high temperature and keeping the PV panels at low temperature for better electrical yields. Moreover, when the system operates at high temperatures, higher heat losses will occur and the thermal efficiency decreases, with a consequent reduction in electrical efficiency as well, because of the reduction of heat transferred to the air, which cause higher PV operating temperatures. Therefore, the design process will depend on the specific building requirements, the fraction of electrical and thermal energy needed and its intended use. Another constraint of BIPV/T collectors, when compared to conventional PV/T technology, is the fact that their installation, in terms of orientation and tilting angles, is strictly connected to the building itself and the design of its components. Nevertheless, there are a number of factors that affect the performances of air-based BIPV/T systems and that can be optimized to improve the overall efficiency, such as the number of glass covers and glazing material, the use of anti-reflective or low-emissivity coatings, the thermal absorptivity of the PV panels, the radiative properties of the non absorbing materials, the convective mode of the airflow, the flow regime and the mass flow rate. In this paper, the thermal and electrical performances of a BIPV/T air collector prototype for glazed façades applications are investigated. The solar window prototype has been developed in Palo Alto, CA, based on a patent by Dr. N.S. Kapany [23]. Some of the challenges and factors of influence on the performances mentioned above are addressed, with the scope of assessing the features that need to be optimized, which will guide the design of the next prototype. Results from experimental data are reported and were used to develop a two-dimensional model, in order to find the optimal number of glazing covers and low-emissivity coatings that would improve the thermal performances.

2. Experimental setup The constructed prototype consists of 2 double glazed window compartments with air cavities in between, held together by an aluminum frame, as shown in Fig. 1.

VMP × I MP FFPV = modules = 0.756 are placed inside the double-glazing, constituting the heat absorbing surfaces. In the bottom VOC × I SC which has an area of 1.3 m2, a BP Solar SX170B photovoltaic module (rated at 170 W with a nominal compartment, efficiency of 13.5%) has been installed, while the top part consists on a 6x10 array of Parallax XHHOO 1-4 PV modules, with a power rating of 1 W each, for a top glazing area of 0.658m2. The 60 modules are mounted on plastic rods, which can be manually tilted up to 25 with respect to the normal to the modules. The dimensions of a single module are 125x63 mm, and its voltage and current at maximum power are, respectively, VMP=6V and IMP=0.166A. The open-circuit voltage is VOC=7.2 V and the short-circuit current ISC=0.183A. From these values, the fill factor FF has been determined to be [24]: (1) The rated electrical efficiency of a single module, at nominal conditions of AM=1.5, 25C and 1000 W/m2, is given by [25]:

h=

VOC × I SC × FF =12.65% I solar × Apanel

(2)

where Isolar is the solar irradiance, equal to 1000 W/m2, and Apanel is the area of a single PV panel of 7.87x10 -3m2. Two vents were made in the frame, one at the bottom on the front part of the window that serves as inlet for the cold air and one at the top on the back side, which represents the hot air outlet. A metal grid and aluminum fins at the bottom vent prevent dust from entering the window. The transition between the frame and the air space is realized through the use of holes in the aluminum shell, in order to allow for air circulation and connect the two air gaps. A schematic of the cross sectional view of the window, with the airflow represented, is shown in Fig. 2. The airflow is controlled by 10 DC computer fans of 2.6 W each, that are installed inside the top part of the prototype along with a heat exchanger composed of a U-shaped copper pipe and aluminum fins, connected together by a metal plate. The pipe runs down along the right side of the collector, and ends in a storage tank containing glycol and a water pump to circulate it. This allows for the heat to be transferred to a fluid circuit, which could be used for different applications. Further improvements to the current prototype could lead to the integration with radiant floor heating systems or solar cooling, if high enough temperatures could be achieved to feed an absorption chiller and generate cold in warm climates. To test the thermal performance of the prototype, 2 K-type thermocouples were installed at the bottom vent, and 4 K-type at the top vent, measuring the input and output air temperature, respectively. An Extech 410 Multimeter and two EA10 EasyView Dual Input Thermometers were used to read them. The air velocity at the output has been tested with a Kanomax Anemomaster 6006 LITE hot wire anemometer, which reads the air speed in the direction perpendicular to the probe and has an accuracy of ±5%. To ensure this orthogonality condition, a small duct has been used to extend the output section and allow the air to come out horizontally. The measurements have been taken at 9 different positions on the output section, and for each of them the probe was placed at 3 different height levels, so as to have a grid of 27 data points. A pyranometer Ambient Weather TM-206, with an accuracy of ±10 W/m2, has been used to record the global irradiance incident perpendicularly to the window surface, as well as the global tilted irradiance (including beam, diffuse and reflected components) at local solar azimuth and elevation. In order to measure the power output of the 60 modules of the top array, a 24V DC system was set up, which includes the PV modules, a 40 Ah 12V DC battery and the 12V DC load represented by the fans, all connected to a Tracer-2210RN MPPT solar charge controller, as shown in Fig. 3. The latter controls the power flow between the PV array and the loads, depending on the battery charge level, and ensures that the array works at the maximum power point it could achieve at any given time. The PV array voltage, current and power output are measured by a wattmeter, which is placed between the array and the charge controller terminals. A remote display is connected to the charge controller, indicating the voltage and current levels of both the battery and the load, as well as the battery capacity percentage. A schematic of the overall BIPV/T system is shown in Fig. 4, where the two fluid paths are drawn, the first being the one for the airflow that enters from the air inlet at the bottom of the collector, rises to the top through the doubleglazing, the fans and heat exchanger and is released through the outlet at the top. The other is the fluid circuit of the glycol, which is stored in the thermal storage tank and is pumped to the top of the collector to be input to the heat exchanger and circulated back down to the tank.

3. Methodology The main objective of this research is to investigate the performances of the current prototype design, in order to identify what improvements need to be made. For this purpose, experimental data were taken and used to build a twodimensional model, which will be used to perform simulations of different designs to optimize the various system components. As a first step, the optimization of the glazing system is presented in this paper. The results from field measurements have been used to characterize the system thermal performances, whose parameters of interest are the temperature rise from bottom to top of the unit, the thermal power generated and the thermal efficiency. The thermal power is represented by the net heat transfer rate absorbed by the air and carried out of the collector, which is considered a control volume with one inlet and one outlet, consisting of the bottom and top vents. The net rate of heat transfer to the fluid Q [W] is given by [26]:

Q = G × cp × (To u t -Ti n )

(3)

where G [kg/s] is the mass flow rate, cp [kJ/kg·K] is the air specific heat at constant pressure and ∆T=Tout–Tin [ºC] is the temperature difference between the output temperature Tout and the input temperature Tin. The specific heat of air between 20ºC and 50ºC, which is a common operating temperature range for the current prototype, does not change noticeably. It only varies from 1.005 kJ/kg·K at 300K to 1.007 kJ/kg·K at 330K, so the calculations have been made using an average constant value of 1.006 kJ/kg·K. The mass flow rate is calculated as follows [27]:

G = r ×v× A

(4)

where  [kg/m3] is the air density, v [m/s] is the average air speed at the top vent, and A [m2] is the cross sectional area of the vent, which is equal to 0.02484 m2. The air density is computed for each set of data using the following relationship [27]:

r=

pa Rair ×Tabs

(5)

where pa [Pa] is the atmospheric pressure, Rair=287.058 J/kg·K is the air gas constant, and Tabs [K] is the absolute temperature of the airflow. Since the prototype has a similar operation to that of a solar air collector, its thermal efficiency th can be expressed as [26]:

ht h =

Q I g l × Ag l

(6)

where Igl [W/m2] is the solar irradiance incident perpendicularly to the glass panels and Agl [m2] is the total frontal glazing area, equal to 1.958 m2, which represents the useful effective area that captures and absorbs the solar radiation. Similarly, the actual operating electrical efficiency of the PV array can be calculated as follows [24]:

hel =

Pel I gl × APV

(7)

with APV being the total area of the 60 PV modules, which is equal to 0.4725m2 and Pel is the electrical power output. The experimental data served as inputs to develop a two-dimensional CFD model in COMSOL Multiphysics 5.2 software [28], which couples the heat transfer and fluid flow across the prototype. A simplified cross sectional geometry has been modeled, which does not include the heat exchanger, the fans and the solar array rods. Other assumptions include the modeling of the PV modules as a single domain made of silicon and simplifications for the airflow path such as the replacements of holes with open passages and the repositioning of the vents to allow a vertical flow. Fig. 5 shows the top and bottom parts of the modeled geometry, as well as the meshing of the inlet, outlet and the passage between the two double-glazed cavities, where refined meshes and boundary layers were used around the corners and along the walls for CFD computations. The model accounts for the solar radiation incident on the window, the conduction within the solids parts, the radiation exchange between surfaces, the convective cooling by the outside environment and the convection that takes place inside the window, between the solid objects and the airflow.

The measured data from Feb. 24, 2015, at 11:00 were used for the input solar irradiance and the boundary conditions for the inlet air temperature, output air velocity and wind speed. The sun is modeled as an infinite distance source emitting radiation as a blackbody at 5780 K, and the radiation direction was set to the local solar elevation. The incident irradiance has been set to the observed value of 1140 W/m2, and in order to account for the transmission losses through the glass, the measured transmission coefficient of 0.937 has been applied. Ibrahim et al. [29] reported determinations of the global tilted irradiance for various months in a close latitude location (30.78°N), with a maximum value of 1152.97 W/m2 in March of 2009, that would suggest that a similar value for the global tilted irradiance at solar azimuth and elevation could be possible for 36° latitude in late February. The measured value of the global tilted irradiance might also indicate a larger contribution of the reflected component, due to the light colored concrete pavement of recent installation on the site (albedo values around 0.4 - 0.5), as well as the proximity to several metallic surfaces, which could cause extra reflections. The very clear sky and dry conditions on February 24, 2015, with a relative humidity of less than 30% at 11am, as well as a solar constant of about 1394 W/m2 might be other contributing factors. The radiation was divided into 3 spectral bands: the first includes the UV and visible spectrum of wavelengths until 780nm, the second comprises the solar near infrared (NIR) from 780nm to 2.5m, and the third covers the longwave infrared portion for wavelengths greater than 2.5m. This enables us to specify the opacity of materials for each individual spectral band, so that glass can be considered to be transparent to the visible and NIR bands, but opaque to the infrared portion, as well as transparent to the first band alone, allowing us to simulate the use of NIR selective glasses. The main limitation of the software is that domains are either considered to be fully opaque or fully transparent to a specific spectral band, so the results differ somewhat from the real situation, but they rather serve as a relative comparison between the different configurations in order to guide the design of the next generation prototype. The model solves the conjugate heat transfer problem where both conduction in solids and convection in fluids are involved, in addition to surface-to-surface radiation, and couples the heat equations with the Navier-Stokes and continuity equations for the fluid flow. The boundary conditions were based on the experimental data: the input air temperature was set to 21°C at the inlet, and a pressure equal to the atmospheric one was applied to the same boundary, while at the outlet the output velocity was set at a value so as to have the same measured mass flow rate of 0.0156 kg/s. Initial conditions are set to 20°C for the temperature, a zero pressure with respect to the atmospheric pressure and 0.2 m/s along the y direction for the air speed. A convective heat flux boundary condition was applied to all exterior surfaces in order to account for convective cooling by the wind, and the heat transfer coefficient for exterior forced convection is calculated during the computation from the external ambient temperature, set to 20°C, and the wind speed. A value of 4 m/s was used for the latter, which is the average speed reported by meteorological data between 10:00 and 14:00 for that particular day and location [30]. Diffuse surface boundary conditions have been applied to all boundaries participating in radiation heat transfer. The emissivity of the different materials are reported in Table 1, along with the other material properties used (except for the air properties that depend on the temperature), while those for the glass surfaces vary depending on the type of simulation, and the value of 0.84 refers to the emissivity in the infrared spectrum of uncoated glass. The first model was aimed to reproduce the current prototype design, so no glass coatings were applied, and an emissivity of 0.84 was used for all the 4 glass surfaces of the 2 panels, which are opaque to the infrared radiation and transparent to visible and NIR.

4. Results and discussion 4.1 Experimental measurements Field measurements were performed in Salinas, CA, for the summer, fall and winter seasons, with the prototype placed in the outdoor environment and oriented to the south. The results for the days of July 10, 2014, November 25, 2014 and February 24, 2015 are reported in Fig. 6, where the hourly change of global solar irradiance on the window surface, average air temperature output, air temperature rise between the output and input vents and heat transfer rate are shown. As it can be seen in the figure, the solar irradiance for July 10 th ranges from 90 W/m2 at 10:00 to 247.67 W/m2 at noon, and then drops in the afternoon to values ranging from 120-200 W/m2. The low solar input observed in the summer season is due to the high solar elevation, which was 75º at noon for July 10th, which causes a smaller horizontal component reaching the collector vertical surface. On November 25th, the solar altitude was instead 32.5º

at noon, and much higher values were recorded: 650 W/m2 is the radiation at 10:00 and 860 W/m2 was the peak irradiance at noon, with an afternoon decrease from of 680 W/m2 at 14:00 to 490 W/m2 at 15:00. Very similar values were measured on February 24th, with a peak of 765 W/m2 at noon, corresponding to a solar elevation of around 44º, and values in the 500-700 W/m2 range from 13:30 to 15:00. It can be seen that the different solar inputs are reflected on the temperature and heat transfer rate profiles in the results. The output temperature reaches a maximum of 33.3ºC at 11:30 on July 10th, with values ranging from 30 to 32ºC for most of the day, while the maxima are 50.5ºC for November 25 th and 53.2ºC for February 24th, with average values from 11:00 to 15:00 of 46.7ºC and 48.4ºC, respectively. The average air temperature rise between 11:00 and 15:00 are 11.2ºC for the summer, 25.4ºC for the fall and 26ºC for the winter results, with peaks of 13.3ºC, 28.6ºC and 31.2ºC, respectively. The heat generated on July 10 th is most of the time within a narrow range of 100-160 W, with a peak of 188W at noon, while the increased temperature rise in the fall and winter seasons makes these results more than double for November 25th and February 24th, with average values from 11:00 to 15:00 of 476.7W and 412.9W, respectively, and peaks of 553W and 492.6W, respectively. Due to the emphasis on the thermal properties, the PV system was setup to investigate the electrical efficiency during the conditions of highest operating PV temperatures, and it was tested during the fall and winter seasons, when the PV temperatures were expected to raise the most, with a corresponding higher drop in the performances. Fig. 7 shows the profiles of the electrical power generated and the average temperature on the back of the modules for November 25th and February 24th. The results are given for a PV array tilting of 90° with respect to the ground, that is the normal to the window and the normal to the PV array are parallel. In fall the lower solar altitude and lower modules temperature provide a higher photovoltaic generation, with an average of 24.5W from 9:30 to 14:30 and a peak of 29.5W. PV temperatures ranges from 50°C to 67°C, with an average of 64.5°C from 10:30 to 14:30. The electrical generation for 24.02.2015 is instead 22.8W on average from 11:00 to 14:15, with a peak of 24W at 11:45, and the average temperature in the same time range is 67.5°C, with a peak temperature of 70°C. The high temperature levels to which the PV modules are subjected, ranging from 50 to 70°C, is one of the main cause of electrical efficiency losses. Considering the case of 25.11.2014 as an example, the peak power was 29.5 W at 11:30, when the modules average temperature was 65°C. The measured irradiance normal to the array plane at that time is 820 W/m2, which results in a total input power on the whole array of 387.45 W. The actual operating efficiency can then be calculated by the ratio of power generated to the input power, which is equal to (29.5 / 387.45) x 100 = 7.6%, a result that includes optical losses through the glass, thermal losses due to the high operating modules temperatures, as well as system losses such as mismatch losses between the 60 PV modules, transmission losses through the conductors, and electronic converter losses (MPPT). Fig. 8 shows the thermal and electrical efficiency of the prototype for the different tested seasons. On July 10 the thermal efficiency remains very close to the 37-40% range, also due to lower operating temperatures that causes lower heat losses, while greater differences occur in fall and winter, where values between 25% and 43% for November 25th and between 26% and 35% for February 24th are achieved. The minimum and maximum values for the electrical efficiency are 6.16% and 7.96% for November 25 th and 6.2% and 7.4% for February 24th, respectively, while the averages are 7.2% in fall and 6.7% in winter. Other tests were done for different tilting angles and with the collector tracking the sun, which showed some improvements, especially in the summer seasons when lower incident angles could be achieved by reducing the tilting of the window with respect to the ground, that resulted in doubling the performances and having output profiles much closer to those for winter and fall. However, tracking is not a practical solution for incorporating such collectors into a building façade, that for the majority of residential and commercial buildings are installed vertically, so that these testing configurations are not very suited for common applications. In order to improve the efficiency and achieve higher temperatures, the features of the current prototype need to be optimized. This includes designing a better absorber, which means enhancing the radiation captured by the PV array and improving the heat transfer coefficient between the array and the air, in order to obtain higher heat collected by the airflow and lower PV temperatures, which is expected to increase both the thermal and electrical efficiency. The efficiency of the PV modules used has to be chosen according to the specific thermal and electrical requirements, since more efficient modules would convert a higher portion of radiation into electricity, but would also generate less heat and vice versa. The frame of the current design is made of aluminum, which is one of the main causes of heat losses, and therefore it requires further improvements with a better design and the use of different materials. Another component requiring optimization is the glazing system, which affects the solar radiation transmitted to the inside as well as the convective and radiative heat losses from the interior to the outside environment. The installation of additional glass panels and the use of low-emissivity or spectrally selective coatings may contribute to the system performances.

An operational parameter that can be tuned according to the requirements of the integration between the collector and the building HVAC system is the air mass flow rate. The effects of a change in air velocity have been tested during the winter measurements by modifying the input voltage of the fans, which were operated at 6V, 7.5V, 9V and 12V, with a corresponding linear increase in the average output air speed, equals to 0.5 m/s, 0.58 m/s, 0.64 m/s and 0.8 m/s, respectively. The results of the average values of temperature rise (∆T), electrical power (P el), heat transfer rate (Q), input and output air temperature (Tin and Tout), and power consumption of the fans are reported in Table 2. It can be noticed that lower air speed values result in higher temperature rises, which are close to 26°C for the 6 V and 7.5 V cases, but also in lower thermal outputs, equal to 340 and 413 W, respectively. This is due to a lower convective heat transfer coefficient between the absorber surfaces (the PV modules) and the air, which results in lower heat transferred to the air and higher surface temperature. At the highest voltage level of 12 V, corresponding to an air speed of 0.8 m/s, the ∆T drops to 24.4°C, but the heat transfer rate absorbed by the air rises to 535W. Therefore, if thermal energy at lower temperature can be used, a higher mass flow rate is recommended, as it would improve the heat generated and the system thermal efficiency, even though the increased electrical demand for ventilation should be considered as well. For applications with a higher temperature requirement, such as space heating, a lower mass flow rate can be used, so that an air velocity of 0.58 m/s represents the optimal configuration, as it provides the highest temperature rise, and a relatively high heat transfer rate, as well as low power consumption by the fans, equal to only 8 W, resulting in a positive net electrical generation. In order to allow the integration of this BIPV/T system with solar cooling applications, where both high temperatures and high mass flow rates are required, further improvements and optimizations of the current prototype are needed.

4.2 Simulation results Fig. 9 shows the simulation results of the two-dimensional model in steady state, where the air speed and temperature fields are displayed. It can be seen that the air enters from the bottom at 21°C and its temperature progressively increases as it rises to the top and enters in contact with the absorbers (the PV module first at the bottom, and the array on the second compartment at the top). The bottom module reaches about 67°C at its highest point, where the air temperature is close to 48°C and the velocity around 0.7 m/s. The fluid then enters the larger region where the PV array is located, where the velocity drops to 0.2-0.3 m/s and so the temperature field is spread over a larger cross section, resulting in a slightly lower average temperature of 46°C at the bottom of the array, which increases to about 55°C at the top of the array. A comparison between the experimental and the simulated results is given in Table 3. The simulated output air temperature is 52.56°C, while the maximum measured experimentally on the reference day was 53.2°C at 12:30. The other results of the simulation are 0.015609 kg/s for the mass flow rate, 1.0816 kg/m3 for the air density, 31.56°C for the temperature rise, 77.18°C for the top PV array temperature, 495.92 W for the heat transfer rate, and 32.51% for the thermal efficiency, which compare to 0.015603 kg/s, 1.0842 kg/m3, 31.56°C, 76°C, 492.64 W and 32.68%, respectively, for the measured data, with percent errors of less than 1.55%. The model was used to simulate the use of low-emissivity (low-e) coatings and additional glass panes in order to optimize the glazing system. The placing of low (low-e) coatings on the interior glass surfaces provides a reflection of the long-wave infrared radiation emitted from the interior surfaces (PV array and some parts of the frame), resulting in a larger portion of heat trapped inside the double-glazing. The results from several configurations are reported in Table 4 in terms of output air temperature and thermal efficiency, which shows that a low-e coating on surface 3 of the glass panes brings the output temperature from 52.56°C to 57°C and the thermal efficiency from 32.51% to 36.6%, while a low-e coating applied to both interior surfaces (2 and 3) gives a temperature of 58.66°C and a 38.1% efficiency. The value of emissivity of the low-e surfaces used for all the simulations is 0.173.

By keeping the coatings on surfaces 2 and 3, the addition of a third pane of glass (Fig.10) on the right side (representing the possible interior room), in order to allow the same solar radiation from the front but limit the heat losses from the back side, further improves the performances, and the temperature and thermal efficiency reach 62.6°C and 41.58%, respectively. The width of the gap used in the model has been set to 9.5 mm in order to keep the same geometry and have a comparison with the previous results. An additional low-e coating on surface 5 provides a gain of other 2°C, for an average output temperature of about 64.73°C. The efficiency in this configuration sets around 43.44%.

A quadruple glazing was also simulated, with the low-e coatings kept on the same surfaces. The additional glass on the front reduces the heat losses to the outside on the front side, but also allows less radiation to reach the PV absorbers, so that the ultimate effect is a thermal efficiency of 43.41%, with an average output temperature of 64.7°C, values that are practically the same as in 3 pane configuration. A 4-pane window is not therefore a cost-effective solution. Lastly, the effects of NIR absorbing glass and argon filled gap has been studied for the 3-pane system. The middle glass can be made of NIR absorbing material, which is typical of a tinted glass, in order to block that part of the spectrum that would otherwise be transmitted to the inside of the building, and as a result the second glass would see its temperature increased, thus enhancing the heat transferred to the air. The NIR absorption has been simulated by making the glass opaque to the corresponding spectral band, and by setting an emissivity of 0.7, which means that 70% of NIR radiation is absorbed and 30% reflected (the emissivity is equal to the absorptivity for an opaque surface at the same temperature and wavelength). As it can be seen in Table 4, this configuration was found to be very close to the previous configuration (triple pane with low-e on surfaces 2, 3 and 5), bringing the temperature up to only 65.42°C, while the thermal efficiency is increased to 44%. This relatively small improvement can be explained by the fact that with the PV modules in vertical position and such highly packed, only a small amount of radiation reaches the glass. In a situation where more space between the rows is made, as well as under different tilting conditions, the use of a NIR absorbing glass could provide larger improvements with respect to a regular low-e glass. The filling of the inner gap with Argon (right plot of Fig. 14) instead of air, provides slightly better results than the previous case, and the average output temperature rises to 66.43°C, bringing the efficiency close to 45%. The triple-glazing configuration, with low-e coatings on surfaces 2, 3 and 5 can be considered to be the optimal setup in terms of cost-effectiveness, as the other solutions involving a 4-pane glazing, tinted glass and argon filling do not provide enough improvements to justify the additional costs. Fig. 11 shows a comparison between the simulated output temperature profiles, which are those of typical duct flows, at the output section before and after the optimization, along with the average values. This arrangement provides an increase in temperature rise from 31.56°C to 43.73°C, while enhancing the thermal efficiency from 32.51% to 43.44%, corresponding to a percentage increase of 38.56% and 33.31%, respectively.

5. Conclusions The thermal and electrical performances of a prototype of a BIPV/T airflow window, with solar modules inside of a double-glazing, have been tested. The results from experimental measurements showed that the collector performed better at smaller solar elevation conditions, and the measured output temperature reached a maximum of 53.2°C during the winter with an average of 48.4°C, corresponding to a temperature rise of around 26°C on average and 31°C maximum. The thermal efficiency ranges between 25 and 40%, while the electrical one remains quite constant to around 68%, with a generation of 20-25W for the top PV array. The drop in electrical performance is mainly affected by the high operating module temperatures, which reach 70°C. Higher electrical performances could be achieved through the use of more efficient PV modules, though the effect that an increased electrical generation would have on the thermal output should be further investigated. Another way to improve the modules efficiency would be to place them on the bottom part of the window, where the temperature levels are lower, and hence they would operate more efficiently. Tests performed at different air velocities showed the possibility to change the mass flow rate depending on whether a higher air temperature or a higher thermal output is needed. The opportunity of integration with applications like solar cooling, requiring both high temperatures and flow rates, suggests the need for further improvements and optimizations of the various system components, such as the PV absorbers, the frame and the glazing setup. A two-dimensional model was constructed in COMSOL Multiphysics based on the experimental data, and was used to carry out the optimization of the glazing system. Simulations have shown significant improvements through the use of low emittance coatings, and have established an optimal configuration of the glazing system, consisting of a triple glazing with low-e coatings applied on surfaces # 2, 3 and 5, which represents the best cost-effective solution. This set up provided an air temperature of 64.7°C and a rise of 43.7°C, as well as a thermal efficiency of 43.4%, corresponding to an increase of 21.6%, 40%, and 32.8%, respectively, with respect to the experimental results. A few more degrees could be reached with the use of NIR absorbing glass and Argon filling, and temperatures up to

66.43°C were reached. However, the cost-effectiveness of these last solutions should be evaluated to determine whether the little benefits could be worth the additional cost. Future research will investigate the potential improvements that might result from replacing the aluminum frame of the window with construction materials that are more suited for limiting heat losses such as wood, vinyl or fiberglass, as well as a better design of the PV array.

Acknowledgments This research was supported by the US NSF PIRE program, grant#1243536, the University of California Advanced Solar Technologies Institute, and by N.S. Kapany through SolarPath Inc. The authors wish to thank J. Tarter, L. Tarter, Mark Hintzke and N.S. Kapany for the construction of the prototype and helpful discussions.

References [1] U. Eicker, Solar technologies for buildings. Chichester: Wiley, 2003, pp. 3-13. [2] "Buildings Energy Data Book", Buildingsdatabook.eren.doe.gov, 2012. [Online]. Available: http://buildingsdatabook.eren.doe.gov/ChapterIntro1.aspx. [3] M. Debbarma, K. Sudhakar, P. Baredar, "Comparison of BIPV and BIPVT: A review", Resource-Efficient Technologies, 2016. [4] T. Yang and A. Athienitis, "A review of research and developments of building-integrated photovoltaic/thermal (BIPV/T) systems", Renewable and Sustainable Energy Reviews, vol. 66, pp. 886-912, 2016. [5] M. Wolf, "Performance analyses of combined heating and photovoltaic power systems for residences", Energy Conversion, vol. 16, no. 1-2, pp. 79-90, 1976. [6] Kern E. C., & Russell M. C., 1978, "Combined photovoltaic and thermal hybrid collector systems", The 13th IEEE photovoltaic specialists' conference, pp. 1153-1157. [7] Florschuetz L. W., 1979, "Extension of the Hottel- Whillier Model to the Analysis of Combined Photovoltaic/Thermal Flat Plate Collectors", Solar Energy, 22, PP. 361-366. [8] B. Agrawal and G. Tiwari, "Optimizing the energy and exergy of building integrated photovoltaic thermal (BIPVT) systems under cold climatic conditions", Applied Energy, vol. 87, no. 2, pp. 417-426, 2010. [9] R. Kumar and M. Rosen, "A critical review of photovoltaic–thermal solar collectors for air heating", Applied Energy, vol. 88, no. 11, pp. 3603-3614, 2011. [10] T. Fujisawa and T. Tani, "Annual exergy evaluation on photovoltaic-thermal hybrid collector", Solar Energy Materials and Solar Cells, vol. 47, no. 1-4, pp. 135-148, 1997. [11] J. Clarke, J. Hand, C. Johnstone, N. Kelly and P. Strachan, "Photovoltaic-integrated building facades", Renewable Energy, vol. 8, no. 1-4, pp. 475-479, 1996. [12] B. Brinkworth, R. Marshall and Z. Ibarahim, "A validated model of naturally ventilated PV cladding", Solar Energy, vol. 69, no. 1, pp. 67-81, 2000. [13] L. Gaillard, S. Giroux-Julien, C. Ménézo and H. Pabiou, "Experimental evaluation of a naturally ventilated PV double-skin building envelope in real operating conditions", Solar Energy, vol. 103, pp. 223-241, 2014. [14] E.S. Morse, "Warming and ventilating apartments by the sun s rays." U.S. Patent No. 246,626. 6 Sep. 1881. [15] J. Jie, Y. Hua, H. Wei, P. Gang, L. Jianping and J. Bin, "Modeling of a novel Trombe wall with PV cells", Building and Environment, vol. 42, no. 3, pp. 1544-1552, 2007. [16] W. Sun, J. Ji, C. Luo and W. He, "Performance of PV-Trombe wall in winter correlated with south façade design", Applied Energy, vol. 88, no. 1, pp. 224-231, 2011. [17] B. Koyunbaba, Z. Yilmaz and K. Ulgen, "An approach for energy modeling of a building integrated photovoltaic (BIPV) Trombe wall system", Energy and Buildings, vol. 67, pp. 680-688, 2013. [18] K. Nagano, T. Mochida, K. Shimakura, K. Murashita and S. Takeda, "Development of thermal-photovoltaic hybrid exterior wallboards incorporating PV cells in and their winter performances", Solar Energy Materials and Solar Cells, vol. 77, no. 3, pp. 265-282, 2003. [19] A. Athienitis, J. Bambara, B. O’Neill and J. Faille, "A prototype photovoltaic/thermal system integrated with transpired collector", Solar Energy, vol. 85, no. 1, pp. 139-153, 2011. [20] S. Pantic, L. Candanedo and A. Athienitis, "Modeling of energy performance of a house with three configurations of building-integrated photovoltaic/thermal systems", Energy and Buildings, vol. 42, no. 10, pp. 1779-1789, 2010. [21] N. Aste, G. Chiesa and F. Verri, "Design, development and performance monitoring of a photovoltaic-thermal (PVT) air collector", Renewable Energy, vol. 33, no. 5, pp. 914-927, 2008. [22] R. Charron and A. Athienitis, "Optimization of the performance of double-façades with integrated photovoltaic panels and motorized blinds", Solar Energy, vol. 80, no. 5, pp. 482-491, 2006. [23] N. Singh Kapany, "Solar Window apparatus and method", US 8,046,960 B1, 2011. [24] J. Dunlop, Photovoltaic systems, 3rd ed. Orland Park, Ill.: American Technical Publishers, Inc., 2012, pp. 134-136. [25] G. Tiwari and S. Dubey, Fundamentals of photovoltaic modules and their applications. Cambridge: Royal Society of Chemistry, 2010, p. 99. [26] J. Duffie and W. Beckman, Solar engineering of thermal processes, 4th ed. Hoboken, N.Y.: Wiley, 2013, pp. 238, 290.

[27] Y. Çengel, Heat transfer. Boston, Mass.: WBC McGraw-Hill, 1998, pp. 6-20. [28] COMSOL Multiphysics Reference Manual, COMSOL AB, 2015. [29] A. Ibrahim, A. A. El-Sebaii, M. R. I. Ramadan, S. M. El-Broullesy, “Estimation of solar irradiance on inclined surfaces facing south in Tanta, Egypt.” International Journal of Renewable Energy Research, vol. 1, No. 1, pp. 18-25, 2011. [30] "Weather History for Salinas, CA | Weather Underground", Wunderground.com, 2016. [Online]. Available: https://www.wunderground.com/history/airport/KSNS/2009/5/10/DailyHistory.html.

  Fig 1: Photograph of the prototype.

Heat exchanger

Fans

PV array

PV module

Fig 2: Cross section and airflow schematics.

Fig 3: Photograph of the charge controller, the battery and the connections with the PV array (left) and DC fans (right).

Fig 4: Schematic of the BIPV/T system.

y

x

Fig 5: Top and bottom parts of the 2D model geometry (left), and meshes close-ups for the middle part (center), inlet (bottom right) and outlet (top right).

Solar Irradiance (W/m^2)

750 650 550 450 350 250

550 450 350

650 550 450 350 250

250

150

150

50

50

50

850

Solar Irradiance (W/m^2)

650

9:00 10:00 10:30 11:00 11:30 12:00 13:00 13:30 14:00 14:30 15:00 15:30 Time 14:00 14:30 15:00 15:30 9:00 10:00 10:30 11:00 11:30 12:00 13:00 13:30 Time 9:00 10:00 10:30 11:00 11:30 12:00 13:00 13:30 15:00 15:30 10-Jul14:00 14:30 25-Nov 24-Feb Time 10-Jul 25-Nov 24-Feb

150 950

750

Solar Irradiance (W/m

Solar Irradiance (W/m^2)

850

750

10-Jul

25-Nov

24-Feb

650 550 450 350 250 150 50

10-Jul

10-Jul

34 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4

25-Nov

24-Feb 200 180 160 140 120

Q (W)

T (°C)

9:00 10:00 10:30 11:00 11:30 12:00 13:00 13:30 14:00 14:30 15:00 15:30 Time

100 80 60 40 10:00 10:30 11:00 11:30 12:00 13:00 13:30 14:00 14:30 15:00 15:30 Time

55

600

25-Nov

ΔT (C)

550

450

35

400

30

350

25

300

55

20

250

600

50

15

200

550

10:00

11:00

12:00

13:00

14:00

15:00

Q (W)

500

40

T (º C)

45

9:00

15:45

Time

45 55

40

24-Feb

500

ΔT(ºC)

T_out_Avg

550

Q (W)

450

50

35

25 20

400

40 35

450

350

400

30

Q (W)

30

500 45

T (°C)

T (º C)

T_out_Avg

Q (W)

50

300 250

350

15

25

9:00 20

200 10:00 11:00

11:00 11:45

12:00

13:00

12:30

13:30

14:15

ΔT(ºC) ΔT

Q Q (W)

Time

14:00

15:00

15:45 300

15:00

Time

T_out_Avg T_out_Avg

Fig 6: Hourly data of the solar radiation, temperature output, temperature rise and generated heat for the days of July 10, November 25 and February 24.

25 20

30

65

28

60

25

55 50

20

45 40

T panels (°C)

PV Power (W)

30

70

35

15 15

30

10 10 9:30

10:30

11:30

12:30 Time

PV Power

9:30

10:30

13:30

14:30

T panels

11:30 15:30

55 50

24

45

22

40

20

35 30

18

25 16

2012:30

14 13:30

PV Power

24-Feb

60

26

25

Time

65

T panels (°C)

30

PV Power (W)

25-Nov PV Power (W)

35

70

20 14:30 15:30 11.00 11.45 12.30 13.30 Time

14.15

15.00

T panels PV power (W)

T panels (C)

Fig 7: Hourly variation of average modules temperature and PV power output for November 25 (left) and February 24 (right).

75 70 65 60 55 50 45 40 35 30 25 20

T panels (°C)

35

0.45

0.45

0.40

0.40

0.080

0.35

0.35

0.075 η_el

0.085

η_th

η_th

Fig 8: Thermal efficiency (left) and electrical efficiency (right) for the tested seasons.

0.30

0.30

0.25

0.25

0.20

0.20

0.070 0.065 0.060

0.055 9:00 10:30 11:00 11:30 12:30 13:15 13:35 14:00 14:30 15:00 15:30 15:50 9:00 10:30 11:00 11:30 12:30 13:15 13:35 14:00 14:30 15:00 15:30 15:50 9:30 10:30 11:30 12:30 Time Time Time

10-Jul

25-Nov

24-Feb

10-Jul

25-Nov

24-Feb

25-Nov

13:30

24-Feb

14:30

15:30

v (m/s)

T (°C)

0.8

0.7

70

0.6 60

0.5

50 0.4

0.3 40

0.2

30 0.1

Fig 9: Air velocity (left) and temperature field (right).

#6

#1 #2

#3

#5 #4

Fig 10: Glazing surfaces definition for a 3-pane system.

Fig 11: Output temperature profile and average values before (left) and after (right) the optimization.

Table 1 Material properties. Material

Density [kg/m3]

Thermal conductivity [W/mK]

Specific heat at constant pressure [J/kgK]

Emissivity visible band

Emissivity NIR band

Emissivity far infrared band

Aluminum

2700

238

900

0.37

0.37

0.03

Glass

2210

1.4

730





0.84

Silicon

2329

130

700

0.85

0.85

0.9

Table 2 Average parameters under different voltage configurations: Air speed, temperature difference, electrical and thermal powers, input and output temperatures and power consumed by the fans.

Input voltage of the fans (V) 6 7.5 9 12

Average air speed at the outlet (m/s) 0.5 0.58 0.64 0.8

ΔT (°C)

Pel (W)

Q (W)

Tin (°C)

Tout (°C)

25.5 26 22.4 24.4

19.6 21.7 20.7 20.4

340 413 400 535

27.4 22.3 20 22.8

52.9 48.4 42.3 47.2

Fans power consumption (W) 5 8.25 10.5 19.5

Table 3 Comparison between experimental data and simulation results.

Mass flow rate (kg/s) 0.015603

T_out (°C) 53.2

ΔT (°C)

Experiment

Output air density (kg/m3) 1.0816

Simulation

1.0842

0.015609

52.56

% Error

0.24 %

0.038 %

-1.2 %

Table 4 Simulation results for different glazing configurations.

Configuration

Glazing

T_out

th

(°C)

(%)

Uncoated

Double

52.56

32.51

Low-e on surface 3

Double

57

36.6

Low-e on surface 2, 3

Double

58.66

38.1

Low-e on surface 2, 3

Triple

62.6

41.58

Low-e on surface 2, 3, 5

Triple

64.73

43.44

Low-e on surface 4, 5, 7

Quadruple

64.7

43.41

Low-e on surface 2, 3, 5, 70% NIR absorbance on surface 3

Triple

65.42

44.04

Low-e on surface 2, 3, 5, 70% NIR absorbance on surface 3, Argon filling

Triple

66.43

44.91

31.2

T_pv_top (°C) 76

Heat transfer rate (W) 492.64

Thermal efficiency (%) 32.68

31.56

77.18

495.92

32.51

1.15 %

1.55 %

0.66 %

-0.51 %