Available online at www.sciencedirect.

com

ScienceDirect
Energy Procedia 78 (2015) 1901 – 1906

6th International Building Physics Conference, IBPC 2015

Validation of Computational Fluid Dynamics (CFD) Model of a

Building Integrated Photovoltaic/Thermal (BIPV/T) System
D. Roelevelda,*, G. Hailua, A.S. Funga, D. Naylora, T. Yangb, A.K. Athienitisb
a
Department of Mechanical & Industrial Engineering, Ryerson University, 350 Victoria Street, Toronto, Ontario, Canada M5B 2K3
b
Department of Building, Civil and Environmental Engineering, Concordia University, 1455 De Maisonneuve Blvd. W., Montreal, Quebec,
Canada H3G 1M8

Abstract

A building integrated photovoltaic/thermal (BIPV/T) system was studied using computational fluid dynamics (CFD). The
BIPV/T system was composed of solar photovoltaic (PV) panels mounted in front of a wall/roof with an air channel in between.
A numerical model was developed consisting of a PV layer, an air channel, and an insulation and plywood layer. Experiments
were conducted by the Solar Simulator Laboratory at Concordia University using three different orientations of 0°, 45° and 90° at
two different flow rates. A comparison of the temperatures profiles between the numerical predictions and the experimental data
is presented. Overall, the numerical results were in good agreement with the experimental data.
© 2015The
© 2015 TheAuthors.
Authors. Published
Published by by Elsevier
Elsevier Ltd.Ltd.
This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the CENTRO CONGRESSI INTERNAZIONALE SRL.
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the CENTRO CONGRESSI INTERNAZIONALE SRL
Keywords: BIPV/T; Numerical Modelling; Computational Fluid Dynamics; Mixed Convection;

1. Introduction

Computational fluid dynamics (CFD) was used to model building integrated photovoltaic/thermal (BIPV/T)
systems. The BIPV/T system is composed of solar photovoltaic panels mounted in front of the building wall/roof
with an air gap in between creating a channel. Air was used as a coolant in this channel in order to reduce the
temperature of the PV panels and increase their efficiency [1]. This heated air can then be used in a heat pump or for

* Corresponding author. Tel.: +1-416-979-5000; fax: +1-416-979-5265.

E-mail address: droeleve@ryerson.ca

1876-6102 © 2015 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the CENTRO CONGRESSI INTERNAZIONALE SRL
doi:10.1016/j.egypro.2015.11.359
1902 D. Roeleveld et al. / Energy Procedia 78 (2015) 1901 – 1906

thermal storage. The purpose of this study was to develop a numerical model and validate it against the experimental
data obtained from the Solar Simulator Laboratory at Concordia University.
There have been numerous studies on BIPV/T systems [2-6]. Most of these studies have involved conducting
experiments to analyze the performance of the BIPV/T system and then develop a simplified mathematical model.
Candanedo et al. [7] conducted an experimental study to develop heat transfer correlations for the top and bottom
surface inside a BIPV/T system with forced convection. The correlations were developed for a BIPV/T system at a
tilt angle of between 30° and 45° and a Reynolds number between approximately 250 and 7000. They found that the
heat transfer of the top surface was higher than predicted by existing forced convection correlations.
Liao et al. [8] conducted a numerical and experimental study of a BIPV/T system. A two-dimensional model was
developed using the realizable k-İ turbulence model. Utilizing experimental data, the measured temperatures of the
PV and insulation were used as the boundary conditions in the numerical model. The CFD predictions were in good
agreement with the experimental velocity profiles that were obtained using particle image velocimetry. The
convective coefficients were higher than expected due to leading edge effects and the turbulent nature of the flow.
Convective heat transfer coefficient correlations were developed. Naewngerndee et al. [9] conducted a CFD study to
investigate the effects of flow rate and PV panel configurations on BIPV/T systems. It was shown that a
configuration of 4 × 4 was better than 8 × 2, and 3 × 4 was better than 4 × 3 or 6 × 2. They also determined that the
optimized flow rate for a 4 × 4 configuration was approximately 1 – 3 L/min.
Getu et al. [10] conducted a CFD study of an air based BIPV/T system using two amorphous PV panels. A
similar numerical model to the current study was developed to predict the temperatures of the PV panel, air channel,
and insulation. Air velocities between 0.5 and 2.0 m/s were studied using a horizontal orientation. Experiments were
conducted at the Solar Simulator Laboratory at Concordia University and this experimental data was used to validate
the numerical model. A two-dimensional model was tested using both a k-İ and k-Ȧ turbulence model, where it was
determined that the k-Ȧ turbulence model yielded the best agreement with the PV temperature and the k-İ
turbulence model yielded the best agreement with the air channel and insulation temperatures. Also, at lower air
velocities, the numerical predictions were less accurate due to the low Reynolds number.
In this study, a CFD model was used to investigate the air flow conditions and thermal behavior inside a BIPV/T
system. Experimental data has been obtained from the Solar Simulator Laboratory at Concordia University for
various flow rates and orientations of the BIPV/T system using mono-crystalline PV panels with a transparent
backing. This experimental data was used to validate the numerical model. Mixed convection was used inside the air
channel, where the flow was highly turbulent. A comparison of the temperature profiles inside the BIPV/T system
was conducted between the numerical results and the experimental data. This included results from two different
mass flow rates of approximately 174 kg/h and 232 kg/h at three different orientations of 0°, 45° and 90°. The
numerical solutions were used to present various data, such as the average temperatures and heat transfer
coefficients of the BIPV/T system.

2. Numerical Model of BIPV/T

A numerical model was developed of a BIPV/T system based on the experiments carried out in the Solar
Simulator Laboratory at Concordia University [11]. The BIPV/T consisted of two PV panels mounted above a layer
of insulation and plywood separated by an air channel. A schematic of the BIPV/T is shown in Figure 1(a). The
BIPV/T was 2.039 m long by 0.529 m wide by approximately 87 mm thick. A schematic of the PV panel is shown
in Figure 1(b). The PV panel consisted of 5 layers of various materials, which included tempered glass, mono-
crystalline silicon PV cells, two adhesive layers and the transparent backing material. The thickness of the PV panel,
tPV, was 4 mm. The air channel thickness, b, was 45 mm. On the other side of the air channel, typical construction
materials made up the wall. In the experiments, there were extruded polystyrene insulation and plywood, with
thicknesses of tINS = 25.4 mm (1 in) and tply = 12.7 mm (0.5 in).
Utilizing these dimensions, a numerical model was created in the commercial multi-physics finite element
analysis software COMSOL Multiphysics [12]. The model was two-dimensional with temperature dependent air
properties. The flow was turbulent inside the channel, so a k-Ȧ turbulence model was used. The boundary conditions
on the outside of the PV panel included the measured solar radiation as a heat flux, surface to ambient radiation, and
convective heat transfer due to the wind. The measured solar radiation from the experiments was qsolar = 1030 W/m2.
 D. Roeleveld et al. / Energy Procedia 78 (2015) 1901 – 1906 1903

a b

Fig. 1: (a) Schematic of the BIPV/T system and (b) schematic of the PV panel

The PV panel produced electricity, which reduced the total amount of heat transfer through the BIPV/T system. In
order to account for this, volumetric heat absorption was applied to the mono-crystalline silicon PV cells layer. This
was calculated from the measured amperage (I) and voltage (V) that the PV panel produced in each case. For the
surface to ambient radiation, qrad, the glass had an emissivity of 0.89 and the temperature of the simulated sky, Tsky,
from the solar simulator experiments was used. It varied for each experiment, but Tsky was approximately 14.3 °C.
The convective heat transfer due to the wind, hw, was calculated from Duffie and Beckman [13]:

1 kf
h w = 0.86 Re L0.5 Pr 3
L (2)

where kf is the thermal conductivity of the air, Pr is the Prandtl number of the air, and ReL is the Reynolds number
based on the wind speed. It should be noted that L refers to the hydraulic diameter of the PV panel (i.e. L = 4 ×
(plate area) / (plate circumference)). The air properties were evaluated at the ambient temperature of the outside air,
T’, and the wind speed was 2.2 m/s for the experiments. On the backside of the BIPV/T, a natural convective heat
transfer coefficient was used (hnat = 2.8 W/(m2 K)).
For the air channel, forced convection was used to pump outside air through the channel in order to cool the PV
panel. The mass flow rate of the air, ী, and the inlet air temperature, Tinlet, were measured in the experiments and
used as inputs for the air inlet in the numerical model. Because a turbulence model was used, the turbulence
intensity and turbulence length scale needed to be defined. A high turbulence intensity (lT = 15%) was used in order
to increase the accuracy of the results. The turbulence length scale was calculated to be 0.07 × hydraulic diameter of
the air channel. At the outlet of the model, a pressure outlet and outflow were used. The direct integration surface to
surface radiation model was also used between the back of the PV panel and the front of the insulation inside the air
channel. The emissivity of the back of the PV layer was 0.89 and the emissivity of the insulation was 0.5. A volume
force was also added to the air channel in the numerical model to account for buoyancy forces.
A grid study was conducted to ensure that the numerical solution was grid independent. Three grid sizes of
68,500, 33,335 and 15,225 elements were used with the vertical orientation and ী = 232 kg/h. A Richardson
extrapolation [14] was performed using the PV heat transfer coefficient (hPV) as the critical variable to be tested. The
numerical uncertainty for the grid size of 68,500 elements was 1.5%. The results are shown in Table 1 and the
difference between the coarse and fine grid was 1.6%. Therefore, the grid size of 68,500 elements was determined to
be sufficient for this study.

3. Results and Discussion

Experiments were carried out at the Solar Simulator Laboratory at Concordia University. These experiments used
three different orientations of horizontal, 45°, and vertical and two different mass flow rates of approximately 174
kg/h and 232 kg/h inside the air channel. Thermocouples were used to measure the temperature profiles of the
backside of the PV panel, the surface of the insulation layer inside the air channel, and the centerline of the air
.
1904 D. Roeleveld et al. / Energy Procedia 78 (2015) 1901 – 1906

Table 1: Grid sensitivity study for ী = 232 kg/h at a vertical orientation.

Heat Transfer
Grid Size Number of Elements Coefficient of PV
hPV (W/m2)
Fine 68,500 15.55
Medium 33,335 15.46
Coarse 12,225 15.31

channel. This experimental data was used to validate the numerical model. Figures 2-4 show a comparison between
the numerical results and the experimental data for the temperature profiles of the a) backside of the PV panel, b)
insulation surface, and c) the centerline of the air channel. The average air velocity in the flow direction inside the
air channel was approximately 1.7 m/s for ী = 174 kg/h, 2.3 m/s for ী = 232 kg/h, and 2.4 m/s for ী = 247 kg/h.
Overall the numerical predictions showed good agreement with the experiment data.
Figure 2 shows both ী = 174 kg/h and ী = 232 kg/h at a horizontal orientation. In Figure 2, both cases show that
the numerically predicted PV and insulation temperature profiles were higher than the experimental data. For the air
channel centerline, the numerical results compared favorably with the experimental data. It should be noted that the
temperature drops at the center of the PV panel in the experimental data. This was due to the packing factor of the
PV cells between the two PV panels. There was a gap in the PV cells at the joint between the two PV panels that
allows the solar radiation to transmit through the PV layer. Figure 3 shows a 45° orientation for ী = 174 kg/h and ী
= 247 kg/h inside the air channel. In Figure 3(a), both the PV and air channel numerical solutions showed good
agreement with the experimental data, but the numerical results were higher than the experimental data for the
insulation surface temperature. At the higher flow rate, Figure 3(b) shows the numerically predicted PV and
insulation temperature profiles were higher than the experimental data, but the air channel centerline showed good
agreement. Figure 4 shows the vertical orientation. The temperature profile of the air channel centerline showed
good agreement and the insulation was higher for both flow rates. In Figure 4(a) the numerically predicted PV
temperature was lower and in Figure 4(b) the PV temperature was higher than the experimental data.
Table 2 shows the results from the numerical model for all cases studied. It should be noted that due to the
radiation exchange between the PV and insulation, both surfaces were heating the air in the channel. As expected,
the PV temperatures were higher and the heat transfer coefficients were lower for the 174 kg/h cases compared to
the higher flow rates. The PV temperature was approximately 2.5 °C cooler at the higher flow rate, which increased
the PV panels’ electrical efficiency. When comparing a mass flow rate of 174 kg/h, the heat transfer coefficients of
the PV and insulation were similar for all three orientations. For the vertical and horizontal orientations at a flow
rate of 232 kg/h, the heat transfer coefficients were also similar. As expected, the 45° orientation had higher heat
transfer coefficients because of the higher mass flow rate of 247 kg/h.

4. Conclusion

In this study, a numerical model was developed to predict the temperature profiles inside a BIPV/T system. This
model consisted of a PV panel, air channel, and insulation and plywood. The air channel utilized mixed convection
and a k-Ȧ turbulence model for the air flow. Various boundary conditions were used that are typical of a BIPV/T
system, such as a solar heat flux, radiation exchange with the ambient, and a convective heat flux due to wind on the
PV layer. A natural convective heat flux was used on the backside of BIPV/T system. The numerical results were
compared with experiments conducted by the Solar Simulator Laboratory at Concordia University. Multiple
orientations of 0°, 45° and 90° were tested at two different mass flow rates of approximately 174 kg/h and 232 kg/h.
The predicted temperature profiles inside the BIPV/T system were validated against the experimental data and
showed good agreement. Various results, such as the average temperatures and heat transfer coefficients, were also
presented.
 D. Roeleveld et al. / Energy Procedia 78 (2015) 1901 – 1906 1905

a b

Figure 2: Comparison of the temperature profiles between the numerical results and experimental data for a horizontal orientation and mass flow
rates of (a) 174 kg/h and (b) 232 kg/h

a b

Figure 3: Comparison of the temperature profiles between the numerical results and the experimental data for a 45° orientation and mass flow
rates of (a) 174 kg/h and (b) 247 kg/h.

a b

Figure 4: Comparison of the temperature profiles between the numerical results and the experimental data for a vertical orientation and a mass
flow rate of (a) 174 kg/h and (b) 232 kg/h.
1906 D. Roeleveld et al. / Energy Procedia 78 (2015) 1901 – 1906

Table 2: Comparison of the numerical results for all cases.

Reynolds Average PV Heat Transfer Heat Transfer
Orientation Mass Flow Rate
Number Temperature Coefficient of PV Coefficient of Insulation
ী Re TPV hPV hINS
kg/h °C W/(m2 K) W/(m2 K)
Horizontal 174 9115 47.7 12.92 4.57
Horizontal 232 12165 45.8 15.65 5.80
45° 174 9182 46.4 12.84 4.50
45° 247 13056 43.8 16.25 5.90
Vertical 174 9151 47.4 12.84 4.57
Vertical 232 12237 44.7 15.55 5.73

Acknowledgements

This research was supported by Connect Canada through the Ontario Centres of Excellence (OCE), NSERC
Engage through the Natural Sciences and Engineering Research Council of Canada (NSERC), and s2e Technologies
Inc.

References

[1] Zondag HA. Flat-Plate PV-Thermal Collectors and Systems: A Review. Renewable and Sustainable Energy Reviews 2008; 12:891-959.
[2] Nagano K, Mochida T, Shimakura K, Murashita K, Takeda S. Development of Thermal-Photovoltaic Hybrid Exterior Wallboards
Incrporating PV Cells in and Their Winter Performance. Solar Energy Materials & Solar Cells 2003; 77:265-82.
[3] Anderson TN, Duke M, Morrison GL, Carson JK. Performance of a Building Integrated Photovoltaic/Thermal (BIPVT) Solar Collector. Solar
Energy 2009; 83:445-55.
[4] Agrawal B, Tiwari GN. Optimizing the Energy and Exergy of Building Integrated Photovoltaic Thermal (BIPVT) Systems under Cold
Climatic Conditions. Applied Energy 2010; 87:417-26.
[5] Pantic S, Candanedo L, Athienitis AK. Modeling of Energy Performance of a House with Three Configurations of Building-Integrate
Photovoltaic/Thermal Systems. Energy and Buildings 2010; 42:1779-89.
[6] Chen Y, Athienitis AK, Galal K. Modeling Design and Thermal Performance of a BIPV/T System Thermally Coupled with a Ventilated
Concrete Slab in a Low Energy Solar House: Part 1, BIPV/T System and House Energy Concept. Solar Energy 2010; 84;1892-907.
[7] Candanedo L, Athienitis AK, Park KW. Convective Heat Transfer Coefficients in a Building-Integrated Photovoltaic/Thermal System. J
Solar Eng 2000; 133:021002-1-14.
[8] Liao L, Athienitis AK, Candanedo L, Park KW, Poissant Y, Collins M. Numerical and Experimental Study of Heat Transfer in a BIPV-
Thermal System. J Solar Eng 2007; 129:423-30.
[9] Naewngerndee R, Hattha E, Chumpolrat K, Sangkapes T, Phonsitong J, Jaikla S. Finite Element Method for Computational Fluid Dynamics
to Design Photovoltaic Thermal (PV/T) System Configuration. Solar Energy & Solar Cells 2011; 95:390-3.
[10] Getu H, Yang T, Athienitis AK, Fung A. Computational Fluid Dynamics (CFD) Analysis of Air Based Building Integrated Photovoltaic
Thermal (BIPV/T) Systems for Efficient Performance. In eSIM 2014 Conference Proceedings, Ottawa, Canada. 18 pages.
[11] Yang T, Athienitis AK. A Study of Design Options for a Building Integrated Photovoltaic/Thermal (BIPV/T) System with Glazed Air
Collector and Multiple Inlets. Solar Energy 2014; 104:82-92.
[12] COMSOL Multiphysics 4.3b (Version 4.3.2.152) [software]. 2013.
[13] Duffie JA, Beckman WA. Solar Engineering of Thermal Processes, 4th ed. New Jersey: John Wiley & Sons, Inc; 2013.
[14] Celik IB, Ghia U, Roache PJ, Freitas CJ, Coleman H, Raad PE. Procedure for Estimation and Reporting of Uncertainty Due to Discretization
in CFD Applications. J Fluids Eng 2008; 130:1-4.